Forest Diversity and Management

TOPICS IN BIODIVERSITY AND CONSERVATION

Volume 2

The titles published in this series are listed at the end of this volume.
Forest Diversity and Management

Edited by

David L. Hawksworth
and
Alan T. Bull

Reprinted from Biodiversity and Conservation, volume 15:4 (2006)

123
A C.I.P. Catalogue record for this book is available from the library of
Congress.

ISBN-10 1-4020-5207-3 (HB)

ISBN-13 978-1-4020-5207-1 (HB)
ISBN-10 1-4020-5208-1 (e-book)
ISBN-13 978-1-4020-5208-8 (e-book)

Published by Springer,
P.O. Box 17, 3300 AA Dordrecht, The Netherlands.

www.springer.com

Printed on acid-free paper

All Rights Reserved

2006 Springer
No part of this work may be reproduced, stored in a retrieval system, or
transmitted in any form or by any means, electronic, mechanical, photo-
copying, microfilming, recording or otherwise, without written permission
from the Publisher, with the exception of any material supplied specifically for
the purpose of being entered and executed on a computer system, for exclusive
use by the purchaser of the work.

Printed in the Netherlands.

Contents
Forest Diversity and Management
Introduction 1
D. CLOSSET-KOPP, A. SCHNITZLER and D. ARAN / Dynamics in natural
mixed-beech forest of the Upper Vosges 3–33
OSWALDO TÉLLEZ-VALDÉS, PATRICIA DÁVILA-ARANDA and RAFAEL
LIRA-SAADE / The effects of climate change on the long-term conser-
vation of Fagus grandifolia var. mexicana, an important species of the
Cloud Forest in Eastern Mexico 35–47
OLIVARIMBOLA ANDRIANOELINA, HERY RAKOTONDRAOELINA,
LOLONA RAMAMONJISOA, JEAN MALEY, PASCAL DANTHU and
JEAN-MARC BOUVET / Genetic diversity of Dalbergia monticola
(Fabaceae) an endangered tree species in the fragmented oriental forest of
Madagascar 49–68
HÉLÈNE GONARD, FRANÇOIS ROMANE, IGNACIO SANTA REGINA and
SALVATORE LEONARDI / Forest management and plant species diver-
sity in chestnut stands of three Mediterranean areas 69–82
FRIEDRICH PATRICK GRAZ / Spatial diversity of dry savanna woodlands
Assessing the spatial diversity of a dry savanna woodland stand in north-
ern Namibia using neighbourhood-based measures 83–97
BÄRBEL BLEHER, DANA USTER and THOMAS BERGSDORF / Assessment
of threat status and management effectiveness in Kakamega Forest, Kenya 99–117
MASASHI OHSAWA and TAKUO NAGAIKE / Influence of forest types and
effects of forestry activities on species richness and composition of
Chrysomelidae in the central mountainous region of Japan 119–131
ANDREAS HEMP / The banana forests of Kilimanjaro: biodiversity and conser-
vation of the Chagga homegardens 133–157
M.G.P. TCHOUTO, M. YEMEFACK, W.F. DE BOER, J.J.F.E. DE WILDE,
L.J.G. VAN DER MAESEN and A.M. CLEEF / Biodiversity hotspots
and conservation priorities in the Campo-Ma‘an rain forests, Cameroon 159–192
R. KINDT, P. VAN DAMME and A.J. SIMONS / Tree diversity in western
Kenya: using profiles to characterise richness and evenness 193–210
GERHARD LANGENBERGER, KONRAD MARTIN and JOACHIM
SAUERBORN / Vascular plant species inventory of a Philippine lowland
rain forest and its conservation value 211–241
WEIBANG SUN, YUAN ZHOU, CHUNYUAN HAN, CHUNXIA ZENG,
XIAODONG SHI, QIBAI XIANG and ALLEN COOMBES / Status and
conservation of Trigonobalanus doichangensis (Fagaceae) 243–258
NIALL G. BURNSIDE, DAN J. METCALFE, ROGER F. SMITH and STEVE
WAITE / Ghyll woodlands of the Weald: characterisation and conservation 259–278
AI-LIAN ZHAO, XIAO-YONG CHEN, XIN ZHANG and DONG ZHANG /
Effects of fragmentation of evergreen broad-leaved forests on genetic
diversity of Ardisia crenata var. bicolor (Myrsinaceae) 279–291
M.G.P. TCHOUTO, W.F. DE BOER, J.J.F.E. DE WILDE and L.J.G. VAN DER
MAESEN / Diversity patterns in the flora of the Campo-Ma’an rain
forest, Cameroon: do tree species tell it all? 293–314
JEAN-REMY MAKANA and SEAN C. THOMAS / Impacts of selective logging
and agricultural clearing on forest structure, floristic composition and
diversity, and timber tree regeneration in the Ituri Forest, Democratic
Republic of Congo 315–337
ALFONSO GARMENDIA, SUSANA CÁRCAMO and OSCAR SCHWENDTNER /
Forest management considerations for conservation of Black Woodpecker
Dryocopus martius and White-backed Woodpecker Dendrocopos leucotos
populations in Quinto Real (Spanish Western Pyrenees) 339–355
STEVEN M. VAMOSI / A reconsideration of the reproductive biology of the
Atlantic forest in the Volta Velha Reserve 357–364
ALESSIO MORTELLITI and LUIGI BOITANI / Patterns of rodent species diver-
sity and abundance in a Kenyan relict tropical rainforest 365–380
J. LUIS HERNANDEZ-STEFANONI / The role of landscape patterns of habitat
types on plant species diversity of a tropical forest in Mexico 381–397
JANE HERBERT / Distribution, habitat and Red List status of the New Caledonian
endemic tree Canacomyrica monticola (Myricaceae) 399–406
GRACE NANGENDO, HANS TER STEEGE and FRANS BONGERS /
Composition of woody species in a dynamic forest–woodland–savannah
mosaic in Uganda: implications for conservation and management 407–435
JÖRN THEUERKAUF and SOPHIE ROUYS / Do Orthoptera need human land
use in Central Europe? The role of habitat patch size and linear corridors
in the Bialowiez·a Forest, Poland 437–448
BENIGNO GONZÁLEZ-RIVAS, MULUALEM TIGABU, KARIN GER-
HARDT, GUILLERMO CASTRO-MARÍN and PER CHRISTER ODÉN
/ Species Composition, diversity and local uses of tropical dry deciduous
and gallery forests in Nicaragua 449–467
B. BALAGURU, S. JOHN BRITTO, S.J., N. NAGAMURUGAN, D. NATARA-
JAN and S. SOOSAIRAJ / Identifying conservation priority zones for
effective management of tropical forests in Eastern Ghats of India 469–483
SWEN C. RENNER, MATTHIAS WALTERT and MICHAEL MÜHLENBERG
/ Comparison of bird communities in primary vs. young secondary tropi-
cal montane cloud forest in Guatemala 485–515
T.R. SHANKAR RAMAN / Effects of habitat structure and adjacent habitats on
birds in tropical rainforest fragments and shaded plantations in the
Western Ghats, India 517–547
Biodiversity and Conservation (2006) 15:1061–1061
Springer 2006
DOI 10.1007/s10531-006-0009-7

Introduction

Topics in Biodiversity and Conservation: Forest diversity and management

Natural forests, with a history of ecological continuity extending back for

thousands of years, are unrivalled as the treasure store of terrestrial biodi-
versity on Earth. Yet to date there is no fully comprehensive inventory of the
above- and below-ground biota of any forest available, even in western Eur-
ope. However, in conserving natural forests, it is reasonable to assume that the
myriads of unnamed bacteria, fungi, insects, mites and nematodes present will
be safeguarded along with the trees, provided that the forest structure is
maintained. But forests are also a key player in the global carbon cycle, and so
in the maintenance of the composition of the atmosphere crucial to Life as we
know it. To endanger and clear forests is at the peril of future generations, but
as so many peoples depend on forests for food and wood, the issue of how
forests can be used sustainably, in a way that protects the full spectrum of
organisms they contain, has to be addressed. Sadly, the compensatory planting
of new forests, whether for exploitation or conservation, does not fully address
the need; the trees, other plants and vertebrates may be secured by such ap-
proaches, but the full soil biota, and complete spectrum of organisms asso-
ciated with a natural forest are unlikely ever to be regained.
This compilation of peer-reviewed papers, drawn from researchers around
the world, examines many different aspects of forest diversity and manage-
ment. They consider forests in diverse locations, including Australia, Camer-
oon, China, France, Kenya, the Philippines, Poland, Uganda, and the UK. The
forest types considered vary from banana forests, savannah forests, and tro-
pical rainforest to the much revered ancient oak forest of Bialowieza in Poland.
The emphasis is on the trees themselves, including effects of logging, changes in
management practices, and climate change. In some cases the consequences of
forest disturbance or destruction on other plants, birds, or vertebrates are
reported for particular forests.
Given the wide range of topics brought together here, this collection should
be of particular interest to those involved in teaching forest conservation and
management, and requiring a cross-section of current work in the field.

DAVID L. HAWKSWORTH
The Yellow House, Calle Aguila 12,
Colonia La Maliciosa, Mataelpino,
Madrid 28492, Spain.
E-mail: [email protected]

[1]
Biodiversity and Conservation (2006) 15:1063–1093
Springer 2006
DOI 10.1007/s10531-004-1874-6

-1

Dynamics in natural mixed-beech forest

of the Upper Vosges

D. CLOSSET-KOPP*, A. SCHNITZLER and D. ARAN

LBFE, University of Metz, Campus Bridoux, rue du Ge´ne´ral Delestraint, 57070 Metz-Borny, France;
*Author for correspondence (e-mail: [email protected])

Received: 10 February 2004; accepted in Revised form 22 July 2004

Key words: Mixed beech forests, Age-structure, Architecture, Soil, Light regime, Stand history

Abstract. Forest dynamics were analysed in the Upper Vosges mountains of north-eastern France
in two reserve areas, Frankenthal-Missheimle (FM) and Grand Ventron (GV), located in the
Ballons des Vosges Natural Regional Park (Parc Naturel Régional des Ballons des Vosges). Two
plots of 3000 m2 each were established in mixed beech woodlands located just below sub-alpine
beech forests for long-term monitoring. The main aim of the study was to interpret how the
different species populations in mixed-beech woodlands in the Vosges grow and interact over the
long term, and to determine the disturbance history. The study combined vegetation description,
dendrological and structural data, architectural descriptions and drawings and light distribution
and soil analysis. Historical information was also taken into consideration. Soils in the two plots
showed available phosphate P values > 0.14 g kg
1, indicating good levels of phosphorus supply
for plants, except for A1/C horizon (1Va soil) which corresponds to a medium-fertility soil.
However, soils were found to be shallow because of the slope, a factor that may limit water
availability for adult trees and seedlings. As the canopy (composed of existing trees) consists of
shade trees, the growth rates for seedlings and saplings (potential trees) depends on the canopy
architecture: when growing in sunlit gaps, saplings reach full daylight (canopy height) in less than
100 years. When developing in shade (suppressed state), saplings may need up to 150 years before
reaching full daylight. Alternating periods of rapid and slow growth explain why some trees present
a wide range of stem diameters and ages in the area leading up to the canopy (some trees are more
than 300-years-old), in contrast with the relatively homogeneous height classes distribution, indi-
cating suppression periods. Trees in the FM and GV plots were found to have different growth
rates. Both study plots developed with similar past disturbance events, the two most important
being at the beginning of the 18th century. In addition, the forests were regularly affected by smaller
disturbances until present. Because of the spatial heterogeneity and large range of ages represented,
the forest stands within the two natural reserve areas are presently considered to be the best-
preserved sites in the upper Vosges, but their situation near the timber line prevents them from
becoming models for forest management at lower altitudes.

Introduction

Unmanaged forests are the last representatives of the pristine landscapes of

Europe. Unfortunately, they have practically disappeared from European
forest panels, with a total of only 3 millions hectares (i.e. 1.7% of the total
forest area; COST Action E4 1999). These forest relics are located primarily in
remote, inaccessible areas, in unproductive regions, hunting reserves or along
frontier borders (Peterken 1996; COST Action 4 1999; Motta et al. 2002;

[3]
1064

Schnitzler 2002). Apart from Russia, where old-growth forests may reach up to
20,000 km2 (Sittler et al. 2000), they are small patches (a mean range of 20–
100 ha) within a larger landscape patchwork of managed forests and various
land uses. For example, France, with 15 million hectares of metropolitan forest
cover, possesses the third largest woody domain in Europe, but only 0.2%
(300 km2) of this total are natural forests (Vallauri and Poncet 2003). Most of
these are concentrated in mountain regions (the Vosges, the Jura, the Pyrenees
and the Alps). Studies have demonstrated that these vestiges cannot remain
completely independent of their managed surroundings, and are unable to
preserve their potential biodiversity (Helle and Järvinen 1986).
Pollen diagrams have, however, demonstrated that even small ‘virgin’ forests
remain stable in composition over hundred of years (Bradshaw and Holmqvist
1999). The main causes of such stability and resilience are the high complexity
in structure and architecture associated with the complexity of biotic interac-
tions, which lead to remarkable resistance to climatological events (White
1978; Franks and McNaughton 1991).
All remaining natural forests urgently need protection in order to preserve
their cultural and scientific value, to protect their wildlife and genetic diversity
and to ensure sites for basic research in ecology. They also provide the nec-
essary reference data for applied research in forest management and environ-
mental monitoring (Leibendguth 1959; Peterken 1996).
The present study looked at several aspects of the forest dynamics and
biodiversity of mixed-beech woodlands of the ‘Parc Naturel Régional des
Ballons des Vosges’ (PNRBV) (Upper Vosges, north-eastern France). The
PNRBV has two natural reserves within its boundaries: the Grand Ventron
(GV), created in 1995 covering over 1647 ha, and the Frankenthal–Missheimle
(FM), created in 1989 with over 746 ha. Their respective elevations (GV: 720–
1204 m; FM : 690–1363 m) correspond to the submontane (400–800) and
montane belts (800–1100 altitude). Both reserves harbour typical, often
endangered, plant communities (Schwoehrer and Despert 1999; Schwoehrer
1999; Untereiner et al. 2002). The impact of human activity has been significant
since the French Revolution and includes logging, local fires and the planting
of non-indigenous spruce since the 1850s (Garnier 1994, 1998). Ungulate
densities are also closely linked to human activity and their populations have
increased considerably during the 20th century (ONF 2000; Heuzé 2002).
One of the main objectives of the creation of the PNRBV, within the
framework of which the present research was carried out, was to contribute to
the conservation and sustainable management of the forests through basic and
applied research and the development of innovative methodologies. For this
reason, our research focused on long-term, forest-monitoring plots situated in
strictly protected woodlands of the FM and GV natural reserves which still
include small stands of nearly natural woodlands (Gilg 1997).
The aim of the study is:
(i) to propose and innovate a sampling protocol for long-term studies,

[4]
 1065

(ii) to interpret how different species populations of Vosgian mixed-beech

woodlands grow and interact over the long-term in light of their distur-
bance history,
(iii) to propose a diagnostic of forest naturalness.

Materials and methods

Study sites

The Vosges form a long ridge with a continuous crest line, linked to the west
and to the Rhine valley to the east by steep slopes. In the southern Vosges, the
crest oscillates from 1000 to 1425 m. The climate is oceanic (1600 to
3000 + mm rainfall; mean annual temperature of 4 C above 1000 m). The
hercynian bedrock is mainly composed of granite and metamorphic rock,
partially covered with morainic material. They include biotitic granite rock,
locally porphyroid, with acid plagioclase (An10-20), K-Feldspar, quartz and
some apatite (Mansuy 1992). Soils range from acidic browns to podzols
(Souchier 1971; Bonneau et al. 1978).
The mixed-beech woods found between 600 and 1000 m are part of the
Fagion alliance. The main plant community is the Luzulo–Fagetum (Meusel
37) which includes two sub-associations (group Vaccinium myrtillus; group
Festuca altissima, Oberdorfer 1992) typical of Central European mountains
and hills north of the Alps (Ellenberg 1988; Oberdorfer 1991, 1992; Bogen-
rieder 2001).
The three plant communities include the same tree species (Fagus sylv-
atica, Abies alba, Acer pseudoplatanus, Sorbus aucuparia) and shrubs (Rubus
tereticaulis, Rubus idaeus, Lonicera nigra). Beech (Fagus sylvatica) occupies
a central position in the ecology of the Vosgian forests, outcompeting the
other tree species. Silver fir (Abies alba) is regular but suffers from
browsing (Heuzé 2002). Acer pseudoplatanus is competitive in shallow soils
found in rocky habitats. Spruce (Picea excelsa)is rare, despite the impor-
tance of neighbouring plantations. Ground-level flora is dominated by
Vaccinium myrtillus, Luzula luzuloides, Deschampsia flexuosa and Prenanthes
purpurea.
The GV and FM reserves are 8 km apart (Figure 1). They are close (8–
15 km) to a third natural reserve, the Guebwiller (700–950 m), which also
includes some stands of nearly natural woodlands (Renaud et al. 2000).

Permanent plots

Two plots measuring 3000 m2 (50 · 60 m, i.e. 30 quadrants of 10 · 10 m) each

were selected within the less disturbed mixed-beech woodlands of the two
reserves. Stand coordinates were referenced by means of a global positioning

[5]
1066

Figure 1. Location of FM and GV natural reserves in the Upper Vosges.

system (GPS) (GV: latitude 4827¢N; longitude 665¢E; the plot is situated near
the ‘Grand Ventron’ farm; FM: latitude 4830¢N ; longitude 710¢E: the plot is
situated below the ‘Trois-Fours’ farm). Metallic boundary markers were buried
(forced in the soil) in the ground in order to pursue the study over the very long
term.
The FM plot is located between 900 and 950 m while the GV plot is 50 m
higher (950–1025 m), very close to the multi-stemmed beech forests that form
the timber-line (Carbiener 1966). Both plots face East. Plot GV is characterized
by a succession of steep slopes (50–60%), and flatter (5–30%), moister zones
while the FM plot exhibits more regular slopes, averaging 65%. Plot FM is
adjacent to a large, permanent gap resulting from the accumulation of boulders
of glacial origin.
The fieldwork started in 2000 and ended in 2003.

Soil data

Soil-units were defined according to local topography, regeneration patches

and humus type characteristics. For each soil-unit, one representative
profile was described (horizons, colour, texture, structure, stoniness, root
development). All horizons of the representative profiles were sampled,
with special attention to the first few centimetres (A11 horizon) where seed
germination takes place. In some cases, A11 horizons were also sampled
within one soil-unit, with regard to different slopes, regeneration or humus
characteristics.

[6]
 1067

In 2002, five soil units were defined in the GV plot. In 2003, three soil-units
were identified and sampled in the FM plot.
All samples were air-dried, passed through a 2 mm sieve and then analysed
for chemical characteristics (analysis performed by INRA Laboratory for Soil
Analysis, Arras, France): residual moisture content at 105 C, organic carbon
and total nitrogen (dry combustion method), ‘available’ phosphorus (extracted
by H2SO4 and NaOH; Duchaufour and Bonneau 1959), and exchangeable
cations (based on cobalthexamine method, Orsiny and Remy 1976). Calcula-
tions were performed for C/N ratio, cation exchange capacity (CEC: sum of
exchangeable cations), base saturation (ration of exchangeable ‘basic’ cations –
Ca2+, Mg2+, K+ and Na+ – to CEC), and exchangeable Mg/Al and Ca/Al
ratios.

Stand characteristics

Ground flora
The ground flora coverage was identified and the coverage of each species esti-
mated, (Braun–Blanquet cover coefficient were converted in percentage cover).
Tree seedling densities (height < 130 cm with d.b.h less than 10 cm) were
identified and quantified per 10 · 10 m quadrant (30 per stand).

Size variables and architecture

Tree height, stem diameter and crown area are variables of a tree’s architecture,
and an expression of its growth strategy (Hallé and Oldeman 1970; Hallé et al.
1978; Oldeman 1979, 1990; Oosterhius et al. 1982). Tree height yields infor-
mation about ecological conditions and growth strategies: the harsher the
climate, the shorter the trees. Tree height distribution is also related to
regeneration processes: theoretically, in continuously regenerating stands, the
number of individuals in each height-class is expected to follow an exponential
decline in numbers from shorter to taller trees. Curves are modified under
shady canopies due to the stagnation in growth for much of the tree’s life cycle.
In these cases, stem diameter increases when height progress is nearly at a
standstill (Peters 1992).
Clear allometric relationships exist between these variables. The hf/H ratio
(k) compares the height of the tree trunk (hf) up to the first main fork to the
total height (H). The inversion point k, where the architectural tendencies are
inverted (the axes become smaller and smaller, terminating at the periphery of
the crown), is important in forest stand diagnosis: the higher the inversion
point (i.e. the shallower the crown depth), the higher the competition with
neighbouring trees.
Growth strategy can also be interpreted using tree architecture. Architecture
corresponds to the visible, morphological expression of hereditary growth
development. For tree species, orthotropic versus plagiotropic axes, the ability
to reiterate (i.e. the capacity to partially or totally repeat hereditary architec-

[7]
1068

ture in the same tree, by stimulation of resting meristems, Oldeman 1979) may
explain differences in size variables. Fagus sylvatica grows according to Troll’s
model with plagiotropic differentiation in all axes. The flattened and highly
organized leaf layers (monolayers), as well as the plagiotropy are necessary for
intercepting light over large surfaces. Beech is also very flexible, forming shoots
of different lengths in response to environmental conditions (Hallé and Old-
eman 1970; Nicolini 1997; Roloff 1999). Acer pseudoplatanus follows Rauh’s
model, characterized by orthotropic axes and faster growth than the beech.
However, Acer pseudoplatanus is a monolayer, which increases its ability to
intercept sunlight (Hallé et al. 1978). Abies alba follows Massart’s model,
characterized by a specialized plagiotropic organization of the branches that
confer them high individual survival in the lower forest storeys (Edelin 1977).
Architectural criteria also allow us to define the different phases of tree
development. Oldeman (1990) has defined three main phases (‘potential tree’,
‘tree of the present’ and ‘tree of the past’), considered as the principal social
states in the forest ecosystem. Potential trees and trees of the present can be
seen as two distinct phases in the growth of a tree. In potential trees, the growth
in height is relatively more important than growth in stem diameter and crown
extension. Potential trees may grow faster under the high light levels found in
canopy gaps (‘released-growing trees’) or be suppressed by shading from
canopy trees. After successive intervals of suppression and released growth,
many potential trees reach a minimum height above which they cannot be
suppressed. In this case, they develop their architecture by reiteration and
reach the tree of the present phase. The transition between these two growth
phases is gradual, depending on the site and the forest architecture. To dis-
tinguish the two steps in our study, an empirical threshold was set at the height
of 20 m, which corresponds to the minimum needed to no longer be suppressed
on the steep slopes of the Vosges (suppression is visible when trees present
shallow crowns and reiterations along the trunk, and the h/d.b.h. ratio is
greater than 100). Tall tree species that have gone beyond this threshold are
considered full-grown.
For our purposes, living and dead trees of more than 10 cm d.b.h. and height
>1 m 30 were identified in the plot and measured (diameter at breast height,
total height, height of the main fork. Each (living or dead) tree was carefully
drawn according to its exact proportions and morphology. Crown area was
defined using four perpendicular directions, including one facing the slope.
Tree architecture was represented visually by vertical (six, 10 · 50 m parallel
drawings per stand, oriented towards the slope) and horizontal drawings (one
for living and one for dead trees in each plot).

Relationship between size variables and age

The age distribution of the trees depends on the plot history (past and present
natural disturbances, impact of historical practices). Stem diameter growth
depends on altitude, light intensity, temperature, water and nutrient supply,
neighbouring trees, possible pollution as well as the period in the tree’s life

[8]
 1069

cycle (‘potential’ versus ‘present’). Thus, tree ring analysis is useful for deter-
mining the influence of ecological conditions within the particular context of
the Upper Vosges.
The ring widths of 156 trees (93 in the GV plot and 63 in FM, including
beech, silver fir and Acer pseudoplatanus were noted. Two cores were extracted
at breast height from each tree, one up-slope and one down-slope. Tree-ring
widths were measured using a computer-assisted device (Becker et al. 1995) and
cross-dated against previously existing site chronologies (data from INRA,
Champenoux), in order to correct for missing or duplicate rings.
Stem diameter growth was calculated by dividing core length by the total
number of ring widths. Growth rate comparisons were done at each site for
each state and each species using non-parametric Mann–Whitney U-tests
(Statistica software).

Canopy geometry and light distribution

Since tree growth, architecture and size variables are largely explained by light
dynamics (e.g. Chazdon and Pearcy 1986; Denslow et al. 1990; Vester 1997),
variations in light distribution at points within the canopy were also studied.
Relationships between the tree, the forest and the light level in old-growth
forests have been studied by Koop (1989).
The methodology deals with the architectural parameters of the canopy
(canopy geometry) that are commonly used in the literature (i.e. gap fraction
and foliage area index LAI), and the light variables (i.e. direct and diffuse solar
radiation transmitted throughout the canopy). The simultaneous treatment of
canopy geometry and distribution of incident light (PAR, understood as QPAR,
quantum irradiance, expressed in mol m
2 d
1, Varlet-Granchet et al., 1989)
was studied using hemispherical canopy photography. For each plot, 25 canopy
photographs were taken in each 100 m2 section, 10 m apart from each other, at
50 cm above the ground, in overcast conditions. Because some photographs
were of poor quality, the final raw data included only 21 photographs for GV
and 24 for FM. The camera was equipped with a fish-eye lens with a view angle
of 180, carefully levelled and oriented for true North. The raw data from each
photograph consisted of a matrix with 18 intervals of 5 zenith angles and 24
sectors of 15 azimuth angles which contained the gap fraction. All values were
corrected for latitude, slope, orientation and topographic mask in the GLA
model (Frazer et al. 1999). Calculations were performed at hourly intervals, then
integrated daily, during the vegetative season (June–September). The radiation
intercepted depended upon values calculated for each canopy element involved,
over the whole hemisphere and along the solar tracks. Hemispherical photo-
graphs therefore present spatial auto-correlations with each other.
Values found in FM and GV will be compared with data obtained using
similar methods in the Guebwiller mixed-beech forest. Data concern one plot

[9]
1070

of 1 ha chosen in a nearly natural forest stand growing in deep soil (slope from
20 to 30)(Renaud et al., 2000; Pierrel 2001).

Results

Soil characteristics

In both plots, soils are dark coloured, stony or gravely, with a fluffy to massive
structure and are finely textured (coarser with depth). Soils are shallow. FM
soils show less variability in type than those in GV. Soils belong mostly to the
Rankosol (Ranker) type, except for 2V soil which belong to Alocrisol and 5V
soil which presents hydromorphic features (located in bench slope). Humus
type ranges from oligo-mull to hemi-moder (Brethes et al. 1992).
Chemical analysis (Table 1) shows a humose trend, with relatively high or-
ganic carbon content, particularly for the 5V soil due to waterlogged condi-
tions. Less organic matter accumulation is observed in FM soils which may be
related to better biodegradability of organic materials. In both stands, rather
low C/N ratios, from 14.9 to 17.7 in uppermost horizons (A11), accounted for a
rapid evolution of plant material added to soil and good nitrogen nutrition for
trees. C/N ratios increasing with depth (1Va, 4V and 1F soils) probably indi-
cate a cryptopodzolisation process, morphologically hidden by the humose
character of soils.
Available phosphorus obtained using the Duchaufour and Bonneau (1959)
method is a good indicator of soil fertility. All samples showed available P2O5
values > 0.14 g kg
1, indicating a good phosphorus supply for plants, except
for A1/C horizon (1Va soil) which corresponded to a medium-fertility soil
(Bonneau 1995). No notable differences were observed in regeneration spots, or
between GV and FM plots.
Soils show a medium cation exchange capacity (CEC), mostly related to
organic matter content (organic carbon), decreasing with depth and lower in
FM soils. Low base saturation was always observed in depth, while it exceeded
35% in uppermost horizons in 4V and 3F soils and reached 60% in 2F soil, due
to active nutrient cycling. Among basic exchangeable cations, Ca generally
predominates over Mg and K (low Na). Similarly, among acid exchangeable
cations, Al largely predominates over H (low Mn), except in 1Va, 1Vd, 4V and
2F soils’ uppermost horizons showing a higher H proportion, probably related
to cryptopodzolisation. A plentiful supply of Al on exchange sites is known to
disturb Mg and Ca supply. The lowest Mg/Al and Ca/Al ratios were found in
all deeper horizons (Bw, A1/C and C) except 2F soil, suggesting a possible Mg
and Ca deficiency. On the other hand, all uppermost horizons showed higher
ratios, apart from 3V soil (Mg) and 5Va soil (Ca).
On the whole, no pronounced differences in soil property features were
observed between the two plots.

[10]
 Table 1. Chemical characteristics of soils in GV and FM plots.

Stand Soil Horizon Depth Organic C/N Available Exchangeable cations CEC Base Mg/ Ca/ pH
cm carbon g kg
1 P2O5 g kg
1 saturation% Al Al H 2O
2+ 2+ + + 3+ 2+ +
Ca Mg K Na Al Mn H

cmol+kg
1

GV 1Va A11 0–2 14.84 17.7 0.386 1.95 0.59 0.49 0.08 3.99 0.068 3.56 10.73 29.0 0.148 0.490 3.6
1Va A12 2–7 7.91 17.0 0.195 0.35 0.27 0.24 0.05 4.17 0.013 2.09 7.18 12.6 0.064 0.084 3.6
1Va A1/C 7–30 5.79 21.8 0.125 0.04 0.06 0.07 0.03 5.16 0.006 0.33 5.70 3.6 0.012 0.008 4.1
1Vb A11 0–2 18.48 15.4 0.35 0.63 0.38 0.42 0.07 7.29 0.117 0.55 9.47 15.9 0.053 0.086 4.3
1Vc A11 0–2 15.41 16.4 0.316 1.81 0.57 0.45 0.05 6.54 0.159 1.50 11.09 26.0 0.088 0.276 3.7
1Vd A11 0–2 16.23 17.3 0.289 1.33 0.55 0.44 0.08 5.61 0.036 3.58 11.63 20.7 0.098 0.237 3.6
2V A11 0–2 18.03 17.0 0.404 2.84 0.65 0.49 0.05 7.23 0.298 1.60 13.17 30.7 0.090 0.393 3.9
2V A12 2–10 12.65 15.1 0.3 0.26 0.19 0.20 0.05 7.28 0.042 0.33 8.36 8.4 0.026 0.036 4.0
2V Bw 10–25 6.75 15.4 0.184 0.11 0.10 0.11 0.04 4.83 0.027 0.17 5.39 6.9 0.021 0.024 4.3
2V C 25–50 6.61 15.5 0.16 0.07 0.08 0.07 0.03 4.10 0.014 – 4.37 5.9 0.020 0.018 4.5
[11]

3V A11 0–2 29.24 17.2 0.454 0.78 0.28 0.31 0.03 10.65 0.275 0.35 12.67 11.1 0.027 0.073 4.2
3V A12 2–12 17.92 16.0 0.365 0.27 0.27 0.31 0.05 8.23 0.092 0.30 9.52 9.4 0.032 0.032 4.3
3V C 12–40 6.15 14.6 0.219 0.07 0.09 0.08 0.03 4.06 0.028 0.08 4.45 6.3 0.023 0.018 4.5
4V A11 0–2 30.77 17.7 0.503 6.25 1.31 0.75 0.08 6.06 0.168 3.60 18.21 46.0 0.216 1.032 3.7
4V A12 2–10 19.51 17.2 0.404 1.81 0.59 0.46 0.06 8.16 0.028 2.66 13.77 21.3 0.072 0.222 3.6
4V C 10–33 8.06 21.7 0.197 0.14 0.13 0.11 0.04 6.86 0.006 0.50 7.80 5.6 0.020 0.021 4.0
5Va A11 0–2 34.37 17.6 0.54 0.68 0.47 0.49 0.09 11.41 0.140 0.31 13.58 12.7 0.041 0.059 4.5
5Va A12 2–20 5.58 15.9 0.254 0.12 0.10 0.08 0.02 3.24 0.011 0.16 3.74 8.7 0.032 0.038 4.6
5Vb A11 0–2 18.30 14.9 0.438 0.97 0.40 0.39 0.05 9.12 0.289 0.55 11.77 15.4 0.044 0.106 4.3
FM 1F A11 0–2 6.66 15.9 0.412 0.82 0.23 0.28 0.02 4.05 0.214 0.31 5.91 22.8 0.056 0.203 4.4
1F A12 2–15 4.29 15.4 0.208 0.09 0.08 0.09 0.02 4.18 0.033 – 4.50 6.3 0.020 0.022 4.5
1F A1/C 15–25 3.25 17.4 0.286 0.05 0.03 0.04 0.02 2.57 0.010 – 2.72 5.2 0.012 0.020 4.7
2F A11 0–2 12.45 16.6 0.388 4.13 0.99 0.49 0.06 1.99 0.362 1.39 9.42 60.2 0.495 2.073 4.2
2F A1/C 2–10 8.85 14.9 0.317 0.99 0.33 0.34 0.07 4.40 0.036 1.30 7.46 23.2 0.075 0.225 4.0
3F A11 0–2 7.87 16.5 0.46 1.93 0.45 0.32 0.03 3.37 0.263 0.84 7.20 37.9 0.134 0.574 4.2

1071
3F A12 2–20 5.29 14.8 0.333 0.33 0.15 0.13 0.05 4.41 0.110 0.31 5.49 12.1 0.035 0.074 4.3
Data in g kg
1 of dry matter, except for available P2O5 in g kg
1 of air-dried soil: below detection limits.
1072

Stand characteristics

Density and spatial patterns

Plot GV is more crowded (322 stems ha
1) than FM (275 stems ha
1), but FM
volume values were more important (677.7 m3 ha
1 and 949 m3 ha
1 for GV
and FM respectively)(Table 2). The dominant species is beech, which accounts
for 66% in GV and 70% in FM. Multi-stemmed beech trees were also more
important in GV.
Abies alba and Acer pseudoplatanus are relatively minor components of the
stands, accounting for 18.6 and 13.3% respectively in GV, and 18.1 and 9.4%
in plot FM. Acer pseudoplatanus typically occupies rocky, sunnier sites.
Sapling density is higher in FM than in GV. In plot FM, the dominant
species is Acer pseudoplatanus (42%), followed by Abies alba (32.6%) and
Fagus sylvatica (23.7%). Beech is largely dominant in the GV plot, representing
80% of the total number of saplings. Silver fir development is limited by
ungulate predation and its sensitivity to frost and drought. Picea excelsa and
Sorbus aucuparia are rare. Woody regeneration is strongly clumped in several
ellipsoid patches at gap margins, extending towards the slope (Figure 2a, b).
The biggest gaps in both plots are caused by rocky areas where tree regener-
ation is difficult. These gaps are mainly colonised by herbaceous species: Luzula
luzuloides, ferns in GV; Festuca altissima, Oxalis acetosella, Galeopsis tetrahit,
small woody species (Rubus tereticaulis, Rubus idaeus, Vaccinium myrtillus).

Tree and forest architecture

Because of the steep slopes, nearly all the trees present asymmetrical shapes:
crowns develop marked supplementary axes (reiterations, Oldeman 1974) on
trunks, and secondary axes facing the other end of the valley, while there are no
secondary, supplementary axes on the opposite side (Figure 3a, a1, b, b1).
Asymmetry is particularly developed in Fagus sylvatica and Acer pseudoplat-
anus, very flexible species which reiterate their initial architectural model easily.
Trees of the present reiterate more often than potential trees because they have
sufficient energy for crown expansion. Asymmetrical shapes are explained by
insertion points k which are low (0.32–0.42) on one side of the trunks and high
on the opposite side. There is strong inter-penetration of crowns laterally be-
tween the middle and upper-third of tree height, while foliage is much less
dense above and below.
Potential trees often present triangular crown shapes (suppressed trees
growing below healthy, shading trees of the present). To compensate for the
small foliar volume, some potential trees have developed small reiterations
along the trunk. Firs exhibit many kinds of traumas, including loss of major
branches and double forks resulting from loss of apical buds (frost, wind).
In plot GV, 23.4% of beech trees are multi-stemmed, with 2–14 trunks of
different dimensions. Most trunks have reached the canopy, with some big
trunks reaching more than 30 m high. Within one type of individual, the range
in height varied from 0 to 6 m, with stem diameters varying from 3 to 16 cm. In

[12]
 Table 2. Densities and volumes of tree species in FM and GV.

Number saplings 4 Number, Fagus Number, Number, Acer Number, Number, Sorbus
sylvatica Abies alba pseudoplatanus Picea excelsa aucuparia

GV 8983 7184 592 1194 18 18

FM 52300 12395 17049 22122 680 52
4 stems ha
1; height < 1m30; d.b.h. < 4 cm
Number Number, Number, Number, Total volume Basal area
living trees 1 Fagus sylvatica Abies alba Picea excelsa living trees 2 living trees 3
GV 322 213 43 6 677.7 22.8
[13]

FM 275 193 50 26 949 53.3

Number Number, Number, Total volume
dead trees 1 Fagus sylvatica Abies alba dead trees 2, 4
GV 63 16 47 23.3
FM 16 0 16 24.8
1 Stems ha
1, d.b.h. > 4 cm.
2 m3 ha
1 based on the calculation formula of volume used in Renaud et al. 2000.
3 V = diameter2 · height · 0.35 for coniferous and V = diameter2 · height · 0.5 for broad-leaved tree.
4 m2 ha
1.
5 Dead trunks of multi-stemmed beech trees excluded.

1073
1074

Figure 2. (a) Crown projection map (50 · 60 m) in FM (shaded areas represent regeneration). Fa
for Fagus sylvatica; Ac for Acer pseudoplatanus; Ab for Abies alba. (b) Crown projection map
(50 · 60 m) in VN (shaded areas represent regeneration) Fa for Fagus sylvatica; Ac for Acer
pseudoplatanus; Ab for Abies alba.

[14]
 1075

Figure 3a. Examples of vertical profile (10 m · 50 m) in FM (tree species and age are represented
for potential trees and trees of the present. (Fa: Fagus sylvatica; Ab: Abies alba; Ac: Acer
pseudoplatanus).

general, one or two were big and healthy, one to three smaller and the
remaining trunks (big or small) dead. The proportion of dead trunks increased
as the total number of trunks increased.

Size-distribution, social status and age-distribution

In both plots, tree height distribution presented a simple, global architecture
with foliage concentrated in the canopy:canopy trees (i.e. trees of the present)
accounted for 60% of the registered trees in FM and 68.8% in GV. In spite of a
regular, evenly spaced distribution of trees of the present, canopy stratification

[15]
1076

Figure 3a. (Continued)

was rather complex, with the imbrication of three main layers: 20–25 m, 25–
30 m and some emergents between 35 and 40 m. Plot GV had a nearly equal
number of trees between 20–25 and 25–30 m (Figure 3a, a1), while FM
included more trees in the upper canopy (Figure 3b, b1).
Size-distribution is not related to age-distribution, indicating that shade-
tolerant species cohorts can survive for long periods in a suppressed state.
Correlations are better for older trees in the upper parts of the canopy; i.e.

[16]
 1077

Figure 3b. Examples of vertical profile (10 m · 50 m) in GV (tree species and age are represented
for potential trees and trees of the present. (Fa: Fagus sylvatica; Ab: Abies alba; Ac: Acer
pseudoplatanus).

maximum ages are closer to sizes: 40 m high and 111 cm d.b.h. for a 345-year-
old silver fir, 29 m high and 41 cm d.b.h. for a 214-year-old beech; 23 m high
and 54 cm d.b.h. for a 231-year-old Acer pseudoplatanus (Table 3).
Most trees of the present had a rotten heart, particularly the Acer pseudo-
platanus. However, their foliage and axes were well-developed, without any
sign of bark or leaf loss.
Trees can be considered as reaching the stage ‘of the present’ at various ages.
In general, potential trees in FM reach the canopy earlier than those in GV: a
range of 48–162 years in FM compared to a range of 102 (very rare)–223 years
in GV. Thus, under good light conditions, averages of 2.2 mm growth per year

[17]
1078

Figure 3b. (Continued)

were recorded for three young 45- to 52-year-old beeches in FM. In this plot,
potential trees and trees of the present also had similar growth rates for stem
diameter while in GV, trees of the present grew significantly (p < 0.001) faster
than potential trees (Table 4).
These data indicate that suppression was more marked in GV, which limits
correlations between stem diameter growth and age (Figure 4). For beech trees,
the Spearman correlation coefficient (Rs) is 0.48 while the correlation is above
0.65 in FM. The silver fir presents similar tendencies (Rs = 0.71 in FM; 0.29 in
GV). Average growth rates of suppressed beech trees in GV are only 0.3–
0.6 mm per year and 0.7 mm for silver fir. This explains why some GV saplings
may be rather old: 119 years for a 4.5 m high silver fir; 135 years for a 7.5 m-
high beech. One potential beech was still in a suppressed state at 215 year of age.
Multi-stemmed beech trees present a broad range of ages (from 20 to
78 years) within one individual.

[18]
 1079

Table 3. Ranges of stem diameter and age for beech, silver fir and sycamore in FM and GV.

FM plot GV plot

n Range of Range of n Range of Range of

DBH (cm) ages (years) DBH (cm) ages (years)

Potential trees
< 10 m
Fagus sylvatica 6 8–11 54–102 6 5–19 41–135
Abies alba 3 11–79 30–59 2 6 37–119
Acer pseudoplatanus 2 5–7 21–68 1 10 55
10–20 m
Fagus sylvatica 4 13–32 35–52 5 16–32 119–160
Abies alba 3 21–30 54–80 14 14–54 149–226
Acer pseudoplatanus 1 67 77 1 11 66

Trees of the present

20–30 m
Fagus sylvatica 13 32–76 146–193 36 29–64 102–321
Abies alba 1 45 135 4 48–70 107–193
Acer pseudoplatanus 5 30–73 117–162 14 10–57 78–231
30–40 m
Fagus sylvatica 18 45–83 150–281 8 29–54 147–223
Abies alba 6 57–111 132–345 2 62–64 101–200
Acer pseudoplatanus 1 54 154 14 10–57 78–231

Correlations between stem diameter and age of Acer pseudoplatanus trees

were significant for GV (0.87). For FM, Acer pseudoplatanus trees are too rare
to be analysed.

Dead trees
Dead trees (trees of the past) represented 11.2 and 3.1% of the total volume of
trees in GV and FM, respectively (Table 2). Most of them were silver firs. In
FM no death was recorded among beech trees. Dead trees generated only very
small gaps because they were rarely very large: average stem diameters ranged
from 10–80 cm. The three dead silver fir trees analysed in GV (d.b.h of 13, 13.6
and 25 cm) died at 79, 89 and 191 year of age respectively. Two beech trees
(d.b.h of 25 and 22 cm) died at 145 and 149 years respectively. These trees were
either snapped off at different heights (from 2 to 24 m) or uprooted. Dead trees
presented different degrees of rot. Standing dead trees had woodpecker holes
and were often infected by Fomes fomentarius.

Tree establishment
Tree distribution by species (Fagus sylvatica, Abies alba and Acer pseudoplat-
anus) and age-class (Figure 5) in FM and GV indicates a pattern of estab-
lishment and mortality during the last 350 years. There was a peak in
establishment between 1800 and 1840 in FM (a total of 46 trees in 40 years). In

[19]
1080

Table 4. Stem diameter growth for beech, silver fir and sycamore in FM and GV.

Potential versus present in each plot

Potential Present

n Growth n Growth
rate rate

FM
Fagus sylvatica 10 1.37 31 1.47 NS
Abies alba 6 1.6 7 1.7 NS
Acer pseudoplatanus 3 1.3 6 1.26 NS
GV
Fagus sylvatica 11 0.64 44 0.95 ***
Abies alba 16 0.83 6 2.02 ***
Acer pseudoplatanus 2 0.59 12 0.96 NS
Potential versus potential per plot
FM GV
n Growth n Growth
rate rate
Fagus sylvatica 10 1.37 11 0.64 **
Abies alba 6 1.6 16 0.83 **
Acer pseudoplatanus 3 1.3 2 0.59 NS
Present versus present per plot
Fagus sylvatica 31 1.47 44 0.95 ***
Abies alba 7 1.7 6 2.02 NS
Acer pseudoplatanus 6 1.26 12 0.96 NS
***p < .001
**p < .01
*p < .05
NS: not significant

GV, the peak occurred 20 years earlier, between 1780 and 1820 (a total of 62
trees). Beech was the colonizing tree in more than 50% of the cases. After 1840,
recruitment was continuous but weak (approximately 2–4 trees per 20 year-
period except in 1920–1940 in FM). Before 1780–1800, trees were very sparse
(probably most of them have died since that time): two silver firs born
respectively in 1640 and 1680, two beech trees born in 1680 and 1720 and one
Acer pseudoplatanus in 1760.
Figure 6a, b illustrate age distribution in the two plots. Trees from 100 to
200 years of age were regularly distributed as small groups of similar ages. Trees
in the 200- 300-year-old category, which had survived stress, pathogens or
windstorms, were scattered as relics among these smaller groups. Trees under
100 years of age were clumped around gaps or at the margins of canopy trees.

Canopy geometry and light pattern

The two plots presented differences in canopy geometry and light patterns
(Figure 7a, b; Table 5). Values from FM indicate a more open habitat than

[20]
 1081

Figure 4. Age-stem diameter distribution for Fagus sylvatica, Abies alba, Acer pseudoplatanus in
FM and GV. Rs: Spearmann correlation coeffiecient, ***: p < 0.001, **: p < 0.01, *: p < 0.05,
NS: Not significant

[21]
1082

Figure 5. Tree establishment in GV and FM.

[22]
 1083

Figure 6. (a) Tree age and spatial distribution in FM. (b) Tree age and spatial distribution in GV.

that in GV: canopy openness of 14.6% compared to 11.5% in GV; percentages

of total incident light of 14.5% compared to 6.7% in GV. The differences
can also be visualized when considering the horizontal variations in the gap

[23]
1084

Figure 7a. Variations in canopy geometry and light condition in FM.

fraction and the total incident light (trans total): (i) for gap fraction, the highest
value in FM reaches 22% compared to 16.5% in GV; (ii) for the total incident
light, the highest value is 14.5% compared to 6.7% in GV.

[24]
 1085

Figure 7b. Variations in canopy geometry and light condition in GV.

[25]
 1086
Table 5. Mean and range of values of canopy geometry and of the distribution of light in GV and FM.

n CO% LAI Trans direct Trans Trans diffuse Trans Trans total Trans
m2 m
2 mole m
2 d
1 direct% mole m
2 d
1 diffuse% mole m
2 d
1 total%
[26]

GV 21 Mean values 11.5 2.6 0.65 6.1 0.77 7.3 1.42 6.7
Range of extreme values 7–16.5 1.9–3.8 0.07–1.54 0.67–14.3 0.27–1.65 2.6–15.8 0.36–2.77 1.8–11.8
FM 24 Mean values 14.6 2.5 0.29 14.7 0.89 13.5 1.14 14.5
Range of extreme values 9.8–22 1.8–3 0.07–0.91 5.3–43.3 0.5–1.84 5.9–28 0.71–1.8 7.7–24.3
 1087

Figure 8. Distribution of sun flecks in FM and GV.

The sun fleck distribution points to marked differences between FM and GV

(Figure 8). In GV, the distribution is much more patchy, with a high number of
small 5–10¢ sun flecks (970 compared to only 201 in FM) in relation to the
higher tree density and more complex forest architecture. There were some very
long sun flecks in GV (reaching 155¢) which were not observed in FM (maxi-
mum: 65¢). These long sun flecks, that originated from the horizon, occurred in
only one photo site near the border of the plot.
Canopy geometry and PAR values vary according to the solar zenith angles.
In GV, gap fraction values ranged from 22.5 to 32.5, while in FM the angles
where gap fractions were highest ranged from 2.5 to 22.5. The zenithal angles
where incident light was highest was, in both cases, between 32.5 and
72.5 degrees, corresponding to the slope and the topographic mask.

Discussion

Forest dynamics near the timber-line

The examples given in plots FM and GV indicate that mixed-beech forests are
spatially and temporally heterogeneous. Both plots show wide variations in
stem density, size-class distribution and age distribution that are ecological
traits of woodlands in a nearly natural state.
Despite a higher stem density, the GV site exhibited a lower volume of stem
wood than the FM site.
The importance of deadwood and discontinuities in the distribution of
saplings and seedlings are also ecological features regularly observed in natu-
ral, shady woodlands in Europe and North America (Jones 1945; Lemée 1978;
Mayer and Neumann 1981; Peterken 1996; Schnitzler 2002). These stand
characteristics, combined with the remarkable resistance of mixed-beech for-
ests in the FM and GV reserves (as compared to trees in the surrounding

[27]
1088

managed forest stands) to the historical december 1999 storm (Schwoehrer

personal observation) indicate that forest stands in the two reserves have re-
tained a relatively high degree of ‘naturalness’ (for a detailed discussion of the
word ‘naturalness’, see Peterken 1996). This finding is of interest for the
interpretation of forest dynamics.
The mechanism of gap formation and development is linked to the scale of
disturbance events and biotic processes (pathogens, predation, mast years). In
the two reserves, mixed-beech forests near the timber-line included chronic,
small-scale gap creation associated with the death of single, large trees or small
groups of trees. The causes of death among relatively young trees were un-
known, but no doubt multiple and thought to have resulted from either natural
causes (windbreaks, pathogens, stem exclusion, severe drought) or anthropo-
genic influences (logging, air pollution).
Natural and anthropogenic causes can also have a combined effect. Signifi-
cant impacts were felt from the severe growth declines of silver fir in 1917–1923;
1943–1951 and 1976–1983 due to a combination of reduced rainfall (particu-
larly in 1976) and increased air pollution (Becker 1985, 1989; Ulrich and Williot
1994). The air pollution is worst on the eastern slopes of the Upper Vosges near
the timber line, coming from Eastern Europe (Becker 1985). These events ex-
plain the large numbers of dead silver firs in the two plots. Fir decline may be
due to direct acid deposition on leaves, as well as acidification processes and
nutrient deficiencies in soils with low buffering capacity. In GV and FM stands,
however, soils derive from biotitic granite rock, the weathering of which is
supposed to partially compensate for Ca and Mg losses. In both plots, soils
should thus be less sensitive to acidification than in other parts of the Vosges
mountains, like those with base-poor sandstone and acid granite catchments.
Some small-scale gaps have probably expanded and coalesced in the past,
thus explaining the succession of trees close in age. But there is also evidence of
large-scale disturbance events during the period from 1780 to 1840 which might
be of anthropogenic origin: there is historical evidence of frequent logging in
forests just below summit pastures at the beginning of the 19th century
(Garnier 1994, 1998)
The deep shade cast by the beech and silver fir canopy explains why potential
trees and regeneration are largely confined to gap margins. The tendency for
regeneration and ground flora to form ellipsoid patches below canopy trees is
typical of forest stands growing on very steep slopes into which sun flecks
penetrate obliquely through foliage, thus displaying a multitude of sun flecks
far from the gap (Pierrel 2001).
Steep slopes play a role in the mutual influences seen between the light
regime in understoreys and the canopy architecture. The steeper the slope, the
shallower the soil and the lower the stem density (also related to altitude), the
higher the canopy openness and penetration of incident light. In the Guebwiller
mixed-beech forest canopy, openness averages 9.7% and only 2–15.7% of the
total incident light is transmitted within the plot (Pierrel 2001). Light values are
more important in GV and still more in FM. In the latter, the proximity of a

[28]
 1089

permanent gap further increases the lateral penetration of incident light. This
explains why seedling densities are higher in FM and Acer pseudoplatanus can
regenerate more easily there than in GV. Presence of that specie in FM have a
retroactively impact and directly influences light arrival in the underlayer. A lot
of Acer pseudoplatanus seeds is probably coming from outside, and thus Acer
pseudoplatanus is probably invading the woodlands plots.
Better light conditions also explain why potential trees and trees of the
present have similar growth rates in FM: potential trees reach the canopy in
100 years, and their growth continues thereafter in the canopy at the same
rhythm. In GV many potential trees have grown in shade, and growth rates are
lower: the duration of the potential state lasts lasts 200 years or more. When
potential trees arrive at full light, stem growth increases with the development
of axes and foliage. Beech trees of the present however, grow less rapidly in GV
than in FM which suggests less favorable growth conditions, probably due to
the higher altitude.
The present-day, large range of ages and sizes recorded between trees of the
present in both plots can be interpreted as differences in growth patterns during
tree development: (i.e. alternating suppressed and released-growing periods).
Such growth processes have been recorded in all forests composed of shade
trees (Lemée 1978; Koop and Hilgen 1987; Peters 1992; Korpel 1995; Peterken
1996).
Acer pseudoplatanus presents a different strategy based on its rather low
tolerance for shade. Young Acer pseudoplatanus trees are numerous in open,
rocky areas or at margins, where there is no suppression phase.
Harsher conditions near the summits (for example, only 3 months of age at
900–1000 m) explain why canopy trees are smaller than in forests at lower
altitudes: only 40 m high for 300-year-old silver firs near the crests as compared
to 52–55 m for 180-year-old silver firs in the Guebwiller natural reserve (Re-
naud et al. 2000). In managed stands in the Vosges, some beech trees have been
known to reach 42 m in 120 years (a 2–3 mm annual growth rate between ages
20–70 according to Seynave 1999) compared to only 29 m in height for a 214-
year-old beech in plots of similar density. In the virgin forest of Dobroc (720–
1000 m altitude, granite), Slovakia, there are 45 m beeches that are 230-years-
old (Korpel 1995).
Multi-stemmed beech shape is a particularity of forests growing near the
timber line. Such architecture only occurs under conditions of stress (Carbiener
1966; Peters 1992; Closset 2000). These trees are not lower in stature than
single-stemmed trunks as suggested by Givnish (1984), but they are more
slender than single tree trunk of similar age. Actually, competition between
genetically identical trees has an impact of the lower volume of stem wood.
They form large, very stable individuals in the canopy because a multi-
stemmed growth form ensures better mechanical stability for the tree (Closset-
Kopp and Schnitzler (2000b), thus improving resistance to windthrow.
Multi-stemmed individuals form clusters of genetically similar stems, with the

[29]
1090

potential for separate existence. This explains why stems may have different
sizes and growth rates as related to age and social state.
At the present stage in the evolution of the two plots, old trees are very rare.
The three trees which are more than 300-years-old discovered in the plots are the
only ones recorded in the upper Vosges to date, but clearly more studies could be
done on this subject. The absence of very old trees differentiates forests of the
Upper Vosges from other virgin mixed-beech forests in Europe where there are
silver fir trees more than 400-years-old, and many more beech trees above the age
of 350 (Mayer and Neumann 1981; Korpel 1995; Cenusa 2001; Schnitzler 2002).
Given the present day composition of potential tree, we can predict that
beech will dominate the canopy.

Objectives and priorities of woodland nature conservation in the upper Vosges

Forest stands in the two reserves represent lesser-managed stands in the Upper
Vosges, but human impacts have nonetheless been multiple, and often irre-
versible. Remnants of more natural forest stands are located near the summits
and on steeper slopes, an inaccessibility which limits the data needed for a
comprehensive analysis of forest dynamics. The present-day surface of strictly
protected forests is also too small and too intermixed with managed forests and
open landscapes to serve as reference points for management principles, be-
cause they are not representative enough of a completely pristine landscape.
Given their rarity in Western Europe, these small areas must however be re-
garded as of utmost importance as a class of woodlands for nature conserva-
tion, research and education. A worthy objective of long-term conservation
efforts would be to re-create more substantial examples of missing types of
mixed-beech forests in the upper Vosges, and in the meantime, to leave un-
managed the remaining forests located in natural reserves.

Acknowledgment

We gratefully acknowledge the Parc Naturel Regional des Ballons des Vosges
for financial support through the study project from C. Schwoehrer. We are
also much indebted to J.L. Dupouey for his invaluable assistance in dendrol-
ogy and the logistical support of his laboratory (INRA Champenoux). We also
wish to express their gratitude to Y. Despert, L. Domergue, C. Kieffer and
P. Behr who have contributed core and data sampling.

References

Becker M. 1985. Le dépérissement du sapin dans les Vosges. Quelques facteurs liés à la détério-
ration des cı̂mes. Revue Forestière Française 37: 281–287.

[30]
 1091

Becker M. 1989. The role of climate on present and past vitality of silver fir forests in the Vosges
mountains of northeastern France. Can. J. Forest. Res. 19: 1110–1117.
Bogenrieder A. 2001. Schwarzwald und Vogesen - ein vegetationskundlicher Vergleich. Mitteil-
ungen des Badischen Landesvereins für Naturkunde und Naturschutz 17: 745–792.
Bonneau M. 1995. Fertilisation des forêts dans les pays tempérés. ENGREF, Nancy.
Bonneau M., Faivre P., Gury M., Hétier J.M. and Le Tacon F. 1978. Carte pédologique de la
France à 1/100000. Notice explicative, Saint-Dié. INRA, Service d’Etude des Sols et de la Carte
Pédologique de la France, pp. 11.
Bradshaw R. and Holmqvist B. 1999. Danish forest development during the last 3000 years
reconstructed from regional pollen data. Ecography 22: 53–62.
Brethes A., Brun J.J., Jabiol B., Ponge J.F. and Toutain F. 1992. Typologie des formes d’humus.
In: Baize D. and Girard M.C.R (eds), éférentiel Pédologique, Association Française pour l’Etude
du sol. INRA, Paris, pp. 177–192.
Carbiener R. 1966. La végétation des Hautes Vosges dans ses rapports avec les climats locaux, les
sols et la géomorphologie. Thèse de Doctorat d’Etat, University of Orsay, France.
Cenusa R. 2001. Les montagnes de Calimani. Les forêts vierges de Roumanie, Giurgieu, Donita,
Bandku, Radu, Cenusa, Dissescu, Stoiculescu, Biris. ASBL Forêt wallone, pp.179–184.
Chazdon R.L. and Pearcy R.W. 1986. Photosynthetic light environments in a lowland and tropical
rain forest in Costa Rica. J. Ecol. 72: 533–564.
Closset D. and Schnitzler A. 2000a. Naturalité de la hêtraie d’altitude dans la Réserve naturelle du
Frankenthat Missheimle. University of Metz-Parc Naturel Régional des Ballons des Vosges.
Internal Report 37p.
Closset D. 2000. Sylvigénèse et naturalité de la hêtraie d’altitude dans la Réserve Naturelle du
Frankenthal-Missheimle. DEA Sciences Agronomiques, University of Metz.
COST Action E4 EU 1999. (European corporation in science and technology) Forest Research
Network Final report. Mission, goals, outputs, recommendations, linkage and patterns. Office
for Official Publications of the European Communities, Luxembourg.
Denslow J.S., Schultz J.C., Vitousek P.M. and Strain B.R. 1990. Growth responses of tropical
shrubs to treefall gap environments. Ecology 71: 165–179.
Duchaufour P. and Bonneau M. 1959. Une nouvelle méthode de dosage du phosphore assimilable
dans les sols forestiers. Bulletin de l’Association Française pour l’étude du sol 4: 193–198.
Edelin C. 1977. Images de làrchitecture des conifères. PhD thesis, University of Montpellier.
Ellenberg H. 1988. Vegetation of Middle Europe. Cambridge University Press, Cambridge.
Franks D.A. and McNaughton F. 1991. Stability increases with diversity in plant communities:
empirical evidence from the 1988 Yellowstone drought. Oikos 62: 360–362.
Frazer G.W., Canham C.D. and Lertzman K.P. 1999. Gap Light Analyser (GLA), Version 2.0:
Imaging software to extract canopy structure and gap light transmission indices from true-colour
fisheye photographs, User’s Manual and Program Documentation. Copyright 1999: Simo,
Fraser University, Burnaby, British Columbia, and Institute of Ecosystem Studies, Millbrook,
New York.
Garnier E. 1994. L’homme et son milieu: le massif du Grand Ventron à travers les âges. Parc
Naturel Régional des Ballons des Vosges, Internal Report.
Garnier E. 1998. Jalons pour une histoire de l’environnement: la réserve naturelle du Frankenthal-
Missheimle. Relations des sociétés et du milieu. Parc Naturel Régional des Ballons des Vosges,
Internal report.
Gilg O. 1997. Eléments d’évaluation de la naturalité des écosystèmes forestiers vosgiens. Eléments
conceptuels et méthodologiques, application aux hêtraies-sapinières de la réserve naturelle du
massif du Grand Ventron, DEA, University of Aix Marseille.
Givnish T.J. 1984. Leaf and canopy adaptations in tropical forests. In: Medina E., Mooney H.A.
and Vasques-Yanes C. (eds), Physiological Ecology of Plants of the Wet Tropics. Junk, The
Hague, pp. 51–83.
Hallé F. and Oldeman R.A.A. 1970. Essai sur l’architecture et la dynamique de croissance des
arbres tropicaux. Masson, Paris.

[31]
1092

Hallé F., Oldeman R.A.A and Tomlinson PB. 1978. Tropical Trees and Forests: An Architectural
Analysis. Springer Verlag, Berlin.
Helle T. and Järvinen O. 1986. Population trends in North Finnish land birds in relation to their
habitat selection and changes in forest structures. Oikos 46: 107–115.
Heuzé P. 2002. Impact à moyen terme des grands herbivores sur la hêtraie-sapinière vosgienne.
PhD Thesis, University of Metz.
Jones E. W. 1945. The structure and reproduction of the virgin forest of the north temperate zone.
New Phytologist 44: 130–148.
Koop H. and Hilgen P. 1987. Forest dynamics and regeneration mosaic shifts in unexploited beech
(Fagus sylvatica) stands at Fontainebleau (France). Forest Ecol. Manag. 20: 135–150.
Koop H. 1989. Forest Dynamics. Silvistar: A Comprehensive Monitoring System. Springer Verlag,
Berlin.
Korpel S. 1995. Die Urwälder der Westkarpaten. Gustav Fischer, Stuttgart.
Leibendgut H. 1959. Über Zweck und Methodik der Struktur und Zuwachsanalyse von Urwäldern.
Schweiz Zeitschrift für Forstwesen. 110: 110–124.
Lemée G. 1978. Structure et Fonctionnement des Ecosystèmes Terrestre. Masson, Paris.
Mansuy D. 1992. Les granites et la couverture pédologique dans le bassin du Rouge Rupt. Leurs
participations au contrôle de l’acidification des eaux (Cornimont – Vosges méridionales). PhD
thesis, University of Nancy.
Mayer H and Neumann M. 1981. Struktureller und entwicklungsdynamischer Vergleich der
Fichten-Tannen-Buchen Urwälder Rothwald Niederösterreich und Corkova Urwald. Kroatien
Forstwissenschaftliches Centralblatt 100: 111–132.
Motta R., Nola P. and Piussi P. 2002. Long-term investigations in strict forest reserve in the eastern
Italian Alps: spatio-temporal origin and development in two multi-layered subalpine stands.
J. Ecol. 90: 495–507.
Nicolini E. 1997. Approche morphologique du développement du hêtre (Fagus sylvatica.). PhD
thesis, Université de Montpellier.
Oberdorfer E. 1991. Die hochmontanen Wälder und subalpinen Gebüsche. Der Feldberg im
Schwarzwald. Subalpine Insel im Mittelgebirge. (Landesantstalt für Umweltschutz Baden-
Würtemberg), Karlsruhe.
Oberdorfer E. 1992. Süddeutsche Pflanzen-Gesellschaften. Teil IV: Wälder und Gebüsche. Gustav
Fischer Verlag, Stuttgart.
Office National des Forêts 2000. Evaluation de l’impact des ongulés sur la végétation forestière du
Frankenthal-Missheimle. Parc Naturel Régional des Ballons des Vosges, Internal Report.
Ooisterhius L., Oldeman R.A.A. and Sharik T.L. 1982. Architectural approach to analysis of
North American temperate deciduous forests. Can. J. Forest Res. 12: 835–847.
Oldeman R.A.A. 1974. L’architecture de la forêt guyanaise. Mémoires ORSTOM 73: 73.
Oldeman R.A.A. 1979. Quelques aspects quantifiables de l’arborigenèse et de la sylvigénèse.
Ecologica Plantarum 143: 289–312.
Oldeman R.A.A. 1990. Forests: Elements of Silvology. Springer Verlag, Berlin.
Orsini L. and Remy J.C. 1976. Utilisation du chlorure de cobaltihexamine pour la détermination
simultanée de la capacité d’échange et des bases èchangeables des sols. Science du Sol 4: 268–275.
Peterken G. P. 1996. Natural Woodlands: Ecology and Conservation in Northern Temperate
Regions. Cambridge University Press, Cambridge.
Peters R. 1992. Ecology of beech forests in the northern hemisphere. Ph.D thesis, University of
Wageningen.
Pierrel S. 2001. Détermination du climat lumineux, à l’aide de photographies hémisphériques et de
ses relations avec la répartition de la végétation dans une hêtraie sapinière de la réserve bio-
logique intégrale en forêt domaniale de Guebwiller. DEA. University of Strasbourg, France.
Renaud J.P., Kustner C. and Hauschild R. 2000. La Réserve Biologique Domaniale de Guebwiller
(Haut-Rhin). Présentation générale et résultats d’un premier inventaire réalisé selon le protocole
européen COST. Office National des Forêts, Internal Report.
Roloff A. 1999. Tree vigor and branching pattern. Journal of Forest Science 45: 206–216.

[32]
 1093

Schnitzler A. 2002. Ecologie des forêts naturelles d’Europe. Biodiversité, sylvigénèse, valeur pat-
rimoniale des forêts primaires. Lavoisier, Paris.
Schwoehrer C. and Despert Y. 1999. Réserve naturelle du FM, plan de gestion. Parc Naturel
Régional des Ballons des Vosges, Internal Report.
Schwoehrer C. 1999. Réserve naturelle du massif du Grand Ventron, plan de gestion. Internal
Report PNRBV.
Seynave I. 1999. Analyse de la structure de deux peuplements forestiers mélangés équiennes: la
sapinière-hêtraie et la chênaie-hêtraie dans le nord-est de la France. Ph.D thesis, ENGREF,
Nancy.
Sittler B., Tennhardt T. and Shicarts E. 2000. Die Schützgebiete Russlands vor neuen Herausf-
orderungen. Natur und Landschaft 75: 1–9.
Souchier B. 1971. Evolution des sols sur roches cristallines à l’étage montagnard (Vosges). Mémoire
Service Carte géologique. Alsace Lorraine, pp.33.
Untereiner A., Advocat A. Stoehr B. and Despert Y. 2002. Carte de la végétation de la Réserve
naturelle du Frankenthal Missheimle. Parc Naturel Régional des Ballons des Vosges, Internal
Report.
Ulrich E. and Williot B. 1994. Les dépôts atmosphériques en France de 1850 à 1990. ONF, Internal
Report.
Vallauri D. and Poncet L. 2003. La protection des forêts en France métropolitaine. Livre blanc sur
la protection des forêts naturelles en France. Lavoisier, Paris.
Varlet-Granchet C., Gosse G., Chartier M., Sinoquet H., Bonhomme R. and Allirand J.M. 1989.
Mise au point: rayonnement solaire absorbé ou intercepté par un couvert végétal. Agronomie 9:
419–439.
Vester H. 1997. The Trees and the Forest. The role of tree architecture in canopy development; a
case study in secondary forests (Araracuara, Columbia) PhD thesis, University of Wageningen.
White P.S. 1978. Pattern process and natural disturbance in vegetation. Bot. Rev. 45: 229–299.

[33]
Biodiversity and Conservation (2006) 15:1095–1107
Springer 2006
DOI 10.1007/s10531-004-1868-4

-1

The effects of climate change on the long-term

conservation of Fagus grandifolia var. mexicana,
an important species of the Cloud Forest
in Eastern Mexico

OSWALDO TÉLLEZ-VALDÉS*, PATRICIA DÁVILA-ARANDA and

RAFAEL LIRA-SAADE
Laboratorio de Recursos Naturales, Unidad de Biologı´a, Tecnologı´a y Prototipos, Facultad de Es-
tudios Superiores Iztacala UNAM., Av. de los Barrios 1, Los Reyes Iztacala, Tlalnepantla, C.P.
54090, Estado de México, México; *Author for correspondence (e-mail: [email protected];
phone: +01-55-56-23-11-27; fax: 01-55-56-23-12-25)

Received 23 July 2003; accepted in revised form 22 July 2004

Key words: BIOCLIM, Bioclimatic modeling, Climate change, Cloud forest, Fagus, Sierra Madre
Oriental

Abstract. We examined the effects of climate change on the future conservation and distribution
patterns of the cloud forests in eastern Mexico, by using as a species model to Fagus grandifolia
Ehr. var. mexicana (Martı́nez) Little which is mainly located in this vegetation type, at the
Sierra Madre Oriental. This species was selected because it is restricted to the cloud forest,
where it is a dominant element and has not been considered for protection in any national or
international law. It is probably threatened due to the fact that it plays an important social role
as a source of food and furnishing. We used a floristic database and a bioclimatic modeling
approach including 19 climatic parameters, in order to obtain the current potential distribution
pattern of the species. Currently, its potential distribution pattern shows that it is distributed in
six different Mexican Priority Regions for Conservation. In addition, we also selected a future
climate scenario, on the basis of some climate changes predictions already proposed. The
scenario proposed is characterized by +2 C and
20% rainfall in the region. Under this
predicted climatic condition, we found a drastic distribution contraction of the species, in which
most of the remaining populations will inhabit restricted areas located outside the boundaries of
the surrounding reserves. Consequently, our results highlight the importance of considering the
effects of possible future climate changes on the selection of conservation areas and the urgency
to conserve some remaining patches of existing cloud forests. Accordingly, we believe that our
bioclimatic modeling approach represents a useful tool to undertake decisions concerning the
definition of protected areas, once the current potential distribution pattern of some selected
species is known.

Introduction

The cloud forests represent one of the most interesting biological systems in the
Neotropical region (Luna et al. 1999). They are usually rare, vulnerable and
threatened in the world. Its northern distribution limit is the Sierra Madre
Oriental, in the state of Tamaulipas, Mexico (Briones 1991) and its southern
one reaches Argentina (Webster 1995).

[35]
1096

In Mexico, the cloud forests are characterized by being island-like or archi-

pelagic. In other words, they are arranged in isolated patches that usually bear a
rich flora, with many endemic species (Rzedowski 1996; Luna et al. 2001).
In the last years, the interest for studying the cloud forests, in particular their
species richness and conservation, has been raised (Churchill et al. 1995). The
reason for this interest is based on the high rates of deforestation and loss of
cloud forests due to the introduction of cultivars, especially coffee (Moguel and
Toledo 1999), but also to its irrational use for other agricultural activities, as
well as for forestry and cattle farming purposes. It is recognized that these
forests are threatened all over the world and that the damage that they have
suffered is irreversible, due to their high disturbance vulnerability (Luna et al.
1988; McNeely et al. 1995). Fortunately, many of these forests are restricted to
inaccessible sites in the mountains and consequently, they are still present and
reasonably well conserved. In contrast, those located in places where human
being has access have been drastically transformed to secondary pasture and
cultivated lands.
A few former studies have attempted the identification of priority areas for
the conservation of the Mexican cloud forests, using Parsimony Analysis of
Endemicity and other biogeographic approaches (Morrone and Crisci 1995;
Morrone and Espinosa 1998). Even though these studies have highlighted the
importance and need to protect the cloud forests, they have not considered
either the probable effects that the climatic change might cause in their future
survival, conservation and distribution patterns, nor the proposal of some
general conservation strategies to be undertaken in the coming years. We be-
lieve this information is very relevant, in order to focus our efforts and re-
sources to undertake accurate long-term conservation actions that can assure
the survival of these unique plant communities.
In particular, we decided to use Fagus grandifolia var. mexicana as our study
model, due to its restricted distribution to the cloud forests. Although, this
taxon has been also treated as F. mexicana (López and Cházaro 1995),
F. grandifolia Ehrh. var. mexicana (Martı́nez) Little (Little 1965; Alcántara and
Luna 2001), F. grandifolia Ehrh. (Johnston et al. 1989), or even as the sub-
species Fagus subsp. mexicana (Shen 1992) that has not yet been published, we
recognize the former as the accepted name.
In accordance with the fossil record, Fagus grandifolia was present in eastern
Asia during the late Oligocene and in western North America, including
Alaska, during late Oligocene and early Miocene. However, its current distri-
bution pattern is restricted to eastern North America (Canada and United
States) and small patches of Mexico (Tamaulipas, Hidalgo, Veracruz y Pue-
bla). The latter represent relictual areas of a former extensive cloud forest of
Fagus grandifolia (Pérez 1994).
Fagus grandifolia Ehrh. var. mexicana used to be a dominant and common
tree representative of some of the Mexican cloud forests (Williams et al. 2003).
Some of these cloud forests are restricted to the Sierra Madre Oriental, from
the state of Tamaulipas in northeastern Mexico to the states of San Luis

[36]
 1097

Potosı́, Querétaro, Hidalgo, Puebla and Veracruz in central-eastern Mexico. In

addition, we suspect that the species might be also present in the state of
Oaxaca (Figure 1), but further fieldwork should be done to prove it. Even
though, Fagus grandifolia var. mexicana is restricted to the cloud forests and
plays an important social role, as a source of food and for furnishing activities
(Malda 1990), it has not been considered as either a rare, threatened or
endangered species (Vovides et al. 1997; Oldfield et al. 1998; Williams et al.
2003). However, some authors have already suggested the species rareness
(Malda 1990; López and Cházaro 1995). In particular, Perez (1994, 1999)
considers that the species is endangered at the national level. He estimates that
the total number of individuals of the species existing at the present is below
20,000. He also points out that the largest and most heterogeneous, genetically
speaking, population is located at the state of Hidalgo, where 50% of the total
number of individuals estimated for the country is located in this area. In
addition, all these authors have highlighted the lack of nation and international
laws for protecting and/or conserving the species.

Figure 1. Model of the potential distribution of Fagus grandifolia var. mexicana, on relationship
to the known records. On the right corner the potential distribution of the species in the state of
Oaxaca is shown.

[37]
1098

Some recent data documenting the wild populations status of the species
have been generated, especially in the states of Tamaulipas, Hidalgo and Ve-
racruz (Williams et al. 2003). In some sites the species is considered extinct,
whereas, in other places there are still some small patches of what used to be a
cloud forest of Fagus grandifolia. So far, the species has not been recorded in
the cloud forests of Querétaro, which is a neighbor state of Hidalgo and San
Luis Potosı́ and bears similar environmental conditions for hosting the species.
Probably the absence of Fagus in Querétaro is due to physiographic differences
as suggested by Cartujano et al. (2002). However, it might be also possible that
the species has been misidentified due to its morphological similarity to
Carpinus sp., Ostrya sp. or Ulmus sp., as has been suggested by López and
Cházaro (1995).
Thus, the purpose of this work is to undertake a comprehensive review of the
current situation of the cloud forests in eastern Mexico by using Fagus gran-
difolia var. mexicana as our species model. Consequently, we attempted to
undertake the following actions: (1) to document the current recorded distri-
bution of the species in Mexico; (2) to obtain the potential distribution patterns
of the species; (3) to assess the effects that the potential distribution pattern of
the species might have, under a climatic change scenario; (4) to evaluate the role
that the Protected Natural Areas and the Priority Regions of Mexico will be
playing for the long-term conservation of cloud forests; (5) to propose a general
strategy for attempting the conservation of the oriental Mexican cloud forests.
Accordingly, the approach of this work includes the utilization of biocli-
matic models that enable to explain the current situation of the eastern cloud
forests of Mexico, on the basis of the potential distribution pattern of a rep-
resentative species (Fagus grandifolia var. mexicana) that is used as a model. In
addition, we present an attempt to assess the future distribution of the cloud
forests, using the species data, once a predicted scenario due to climatic change
is included (Téllez and Dávila 2003).

Methods

The plant geographic distribution information that we used in this analysis was
obtained from the database of the World Information Network of Biodiversity
(REMIB) (http://www.conabio.gob.mx/remib/doctos/remibnodosdb.html).
The herbarium data were obtained from the National Herbarium of Mexico
(MEXU), from 29 specimens that beard geo-referenced information
(i.e. complete latitude, longitude, and elevation). The taxonomical identifica-
tion of the specimens was undertaken by Drs. Shen Shung-Fu and Kevin
Nixon who are important specialists of the Fagaceae. On the other hand, the
information concerning the vegetation structure and ecological attributes of
the species that is included in the discussion of this work was obtained from
relevant literature (Malda 1990; López and Cházaro 1995; Luna et al. 2000;
Williams et al. 2003).

[38]
 1099

The bioclimatic modeling approach used in this work was that of the pro-
gram ANUCLIM (Houlder et al. 2000). The program uses mathematically and
statistically interpolated climatic surfaces (digital files in raster format) that
were estimated using the information obtained from a standard network of
meteorological stations. The climatic surfaces or digital files were generated
using thin plate smoothing spline methods in the ANUSPLIN package
(Hutchinson 1991, 1995a, b, 1997; Hutchinson and Gessler 1994). These sur-
faces include long-term monthly mean values of precipitation and temperature
from more than 6200 stations (4000 stations including temperature data and
6000 including precipitation data from the same set of stations). The estimated
mean errors for those surfaces were between 8 and 13% for monthly precipi-
tation values and about 0.4–0.5 C for temperature values. These errors are
similar to those found in the standard meteorological instruments (Nix 1986).
We produced a bioclimatic profile for Fagus grandifolia var. mexicana, using
the program BIOCLIM. The derivation of the bioclimatic profile was based on
selected-simple-matching thresholds. The values for each of the 19 bioclimatic
parameters (Table 1), were assessed by a systematic scanning throughout a grid
of data points. We used the profile to predict potential distribution pattern of
the species. Using the homoclime matching principle, we identified those points
on the climate grid, where the climatic conditions were present within the limits
summarized in the bioclimatic profile of the species (Booth et al. 1987).
We matched the bioclimatic profiles against a grid of data points that con-
tained climatic data from the existing network of stations (bioclimatic

Table 1. Bioclimatic profile of Fagus grandifolia var. mexicana (Fagaceae).

Parameter Minimum–maximum (Mean ± SD)

Annual mean temperature (C) 13.4–22.2 (16.6 ± 2.09)

Mean diurnal range (C) 8.2–15 (11.5 ± 1.88)
Isothermality (2/7) (C) 0.54–0.62 (0.59 ± 0.02)
Temperature seasonality (C of V) (%) 0.61–1.1 (0.78 ± 0.17)
Maximum temperature of warmest period (C) 22.4–33.5 (26.3 ± 3.04)
Minimum temperature of coldest period (C) 5–9.9 (6.8 ± 1.16)
Temperature annual range (5–6) (C) 14.5–24.4 (19.5 ± 2.93)
Mean temperature of wettest quarter (C) 14.3–24.7 (18 ± 2.72)
Mean temperature of driest quarter (C) 12.3–19.5 (14.6 ± 1.59)
Mean temperature of warmest quarter (C) 15.5–25.6 (19.1 ± 2.48)
Mean temperature of coldest quarter (C) 11–17.6 (13.4 ± 1.4)
Annual precipitation (C) 824–2458 (1401 ± 367.19)
Precipitation of wettest period (C) 46–127 (75 ± 17.59)
Precipitation of driest period (C) 0–15 (1 ± 3.67)
Precipitation seasonality (C of V) (%) 66–88 (77 ± 7.28)
Precipitation of wettest quarter (C) 418–1164 (691 ± 168.35)
Precipitation of driest quarter (C) 52–201 (109 ± 42.39)
Precipitation of warmest quarter (C) 243–647 (397 ± 78.18)
Precipitation of coldest quarter (C) 52–239 (126 ± 54.41)

[39]
1100

parameters file). We used a regular grid of 30 arc seconds (0.00083 or

approximately 1 km2) of spatial resolution.
The geocoding errors were detected using the program ArcView 3.2. In
addition, for a more detailed detection of anomalies and potential errors on the
bioclimatic profiles, we used the program BIOCLIM (Houlder et al. 2000).
Whenever possible, we corrected errors by using a 1:50,000 scale topographic
maps. Fortunately, a single anomalous record was detected and removed.
Finally, although the magnitude of climate change is uncertain and many
different future scenarios have been proposed, we generated just one climate
scenario, as proposed by Karl (1998) and some other authors, whom have
predicted similar future climatic conditions (Canziani and Diaz 1998; Giorgi
et al. 1998; Neilson 1998). The program BIOCLIM was used, in order to set up
the proposed future climate change scenario (year 2050), which shows a tem-
perature increment of 2 C and a precipitation decrement of 20%, for any
given present point, at the latitude and longitude where the range and the
localities of the species are located.
For inserting the climate change scenario, we produced a grid of indices in
ARCINFO ASCIIGRID format through the BIOCLIM program and the
Digital Elevation Model (DEM). The predicted distribution patterns of the
selected species were plotted to represent the future potential distribution
patterns found, after climate change conditions were entered. In this paper we
only present the results of an extreme scenario for assessing the role the Pri-
ority Regions for Conservation (PRCs) proposed by CONABIO (Arriaga et al.
2000), will play in the future. The area covered by the potential distribution of
the species was calculated with ArcView 3.2 (ESRI 2000).

Results

The results obtained suggest that the present distribution pattern known for
Fagus grandifolia var. mexicana, is indeed correct and complete, due to the fact
that in all cases, the collecting sites fitted within the limits of the potential
distribution area obtained in the analysis (Figure 1). Thus, the species is re-
stricted to the Sierra Madre Oriental from the state of Tamaulipas to southern
Veracruz, as has been stated by Williams et al. (2003). However, on the basis of
the potential distribution assessment of the species, we believe that probably its
southern limit might extend to the state of Oaxaca.
However, field verifications should be done before we can assure it (Figure 1).
The results also point out that the species is restricted to unique climatic condi-
tions in the Sierra Madre Oriental, as it is shown in its bioclimatic profile
(Table 1). Its climatic uniqueness represents the specific spots or areas along the
Oriental Sierra Madre where it can grow. In other words, although we state that
Fagus grandifolia var. mexicana grows along the Sierra Madre, the fact is that it
only grows in some specific areas that have a unique combination of climatic
attributes and do not grow in others that have other climatic features.

[40]
 1101

On the basis of the species current potential geographic range, it is evident

that it would be distributed in six Priority Regions for Conservation (Arriaga
et al. 2000): (1) El Cielo Biosphere Reserve in the State of Tamaulipas,
(2) Sierra Gorda-Rı́o Moctezuma in the State of Querétaro, (3) Cloud Forest
of the Sierra Madre Oriental in the States of Hidalgo, Veracruz and Puebla,
(4) Cuetzalan in the State of Puebla, (5) Pico de Orizaba-Cofre de Perote in the
State of Veracruz and, (6) Oaxacan northern Sierra.
The current potential distribution model of Fagus (the climatically suitable
environments for the development of this species), covers about 5800 km2.
However, once the climate change scenario was introduced, its potential dis-
tribution pattern contracts drastically in more than 66%. The remaining sites
that will be suitable for the establishment of Fagus populations will be covering
about 1700 km2 or in other words, about 1/3 of the original potential distri-
bution range, including parts of the states of Querétaro, Hidalgo, Puebla, and a
very small portion of the state of Veracruz (Figure 2).
Due to its drastic distribution pattern contraction, the remaining Fagus
patches will probably coincide with only three of the Priority Regions for
Conservation (PRCs 2, 3 and 4) in the states of Querétaro, Hidalgo and Puebla
and none will be present in the state of Veracruz (Figure 2).

Figure 2. Model of the potential distribution of Fagus grandifolia var. mexicana on relationship to
the Priority Regions for Conservation (CONABIO), once the proposed climate change scenario
was entered.

[41]
1102

Discussion

Independently of the taxonomical uncertainty of the studied taxon (whether it

is a species, a variety or a subspecies), evidently, it is seriously threatened due to
its intensive wood extraction, habitat fragmentation and the expansion of the
agricultural land use in areas where it naturally grows. In addition, its
restricted presence in the cloud forests increases its risk.
Currently, the populations of Fagus grandifolia var. mexicana are distributed
within the boundaries of at least five Priority Regions for Conservation
(Arriaga et al. 2000), although the one from Oaxaca, remains to be proved.
From them, the El Cielo Biosphere Reserve represents the only Protected
Natural Area that has been officially declared. Consequently, the future pro-
tection of the cloud forest, Fagus grandifolia var. mexicana and other animal
and plant species of the area is uncertain.
The protection uncertainty of Fagus, has already been pointed out by
Williams et al. (2003) and mentioned the extinction of the species populations
from Teziutlán, Puebla. In the case of the populations located at the Biosphere
Reserve of El Cielo, in the state of Tamaulipas, the agricultural and cattle
farming activities have caused a dramatic reduction of the cloud forest.
Now, when the climatic change scenario is added to the current situation, the
questions to be answered are the following: (1) Is it feasible to have a long-term
conservation strategy to protect the cloud forest of the state of Tamaulipas and
Puebla? and (2) Where do we have higher probabilities of conserving well-
preserved cloud forests in Mexico?
It is clear that the cloud forest of Tamaulipas is already under strong dis-
turbance pressures and consequently its structure and diversity has been al-
ready drastically altered. On the other hand, we believe that these communities
are currently less modified in the states of Querétaro and Hidalgo. Now, if in
addition, the climate changes occur as it is proposed, the results obtained show
that these states also seem to be the adequate cloud forest reservoirs, as has
been partially suggested formerly by Luna et al. (2000).
Alcantara and Luna (1997), mentioned that Hidalgo represents the state
where the cloud forests in Mexico reach their larger extent. They also pointed
out that in the central-eastern part of Hidalgo, this plant community still
remains in the form of wealthy patches that cover around 100 km2 or more. In
these patches, a total of 114 families, 301 genera and 452 species have been
recorded by them. Several species of the region are listed in the Mexican Norm
NOM-059-ECOL-2000 (Anónimo 2000), as vulnerable or in danger of
extinction, such as Cyathea fulva, Deppea hernandezii, Nopalxochia phyllan-
thoides, Magnolia schiedeana, Rhynchostele rosii, Chamaedorea elegans, Psilo-
tum complanatum, Symplocos coccinea and Ceratozamia mexicana (Vovides
et al. 1997; Alcantara and Luna 1997).
Consequently, Luna and Alcántara (2002) emphasize the need to focus the
cloud forest conservation efforts in the state of Hidalgo, where many endemic
plant species for Mexico have been recorded, such as Bouvardia martinezii,

[42]
 1103

Carya palmeri, Ceratozamia mexicana, Cyathea mexicana, Dalbergia palo-esc-

rito, Deppea hernandezii, D. microphylla and Magnolia dealbata, among others.
In addition, these authors pointed out that some other taxa of the cloud forests
that are disjunct between Mexico and the United States show very restricted
distribution ranges in Mexico, as in the case of Illicium floridanum, Nyssa
sylvatica and Schizandra glabra. In summary, this mixture of hardly known,
rare and threatened species is part of a unique natural system that not only
bears taxa from different ancestral biotas, but also has high rates of species
richness and endemicity, as well as, a very fragile habitat that unfortunately do
not have any kind of protection.
In the case of the cloud forests of the states of Querétaro, it is docu-
mented that it bears a very rich flora and plant communities. Cartujano et
al. (2002), recorded 130 families, 465 genera and 774 species of vascular
plants in the cloud forests of the eastern portion of this State. Among this
diverse flora, a number of endemics to Mexico or restricted endemics to the
Sierra Madre Oriental are included (Cinnamomum bractefoliaceum, Clethra
kenoyeri, C. pringlei, Ilex condensata and Inga huastecana, among others), as
well as, some species listed as vulnerable, rare, or threatened (Magnolia
dealbata, Tilia mexicana, Carpinus caroliniana and Litsea glaucescens, among
others) under the Mexican Norm of Endangered Species NOM-059-ECOL-
2000.
Despite the floristic richness and rareness of the cloud forests, timber
extraction, livestock grazing and conversion of forest to farmland, which is
risking its long-term conservation, represent the main recent disturbance
sources of these forests. Unfortunately, precise assessments of the current
destruction rate of these forests have not been done (Pérez 1994, 1999).
Although, in the particular case of cloud forests there is not any former
record documenting their probable shifts due to climate change in Mexico.
A similar exercise assessing future distribution patterns of some cacti species
was done by Téllez and Dávila (2003), in a semiarid region of central
Mexico. They showed the drastic contraction of some of the cacti species
potential distribution patterns, once the climatic changes conditions were
included.
In summary, in this work we attempted to highlight the importance of
including the best biological knowledge available (geographic distribution,
vegetation structure and ecology) and a bioclimatic modeling technique to
assess the possible present and future role of any reserve or protected area. We
also wish to emphasize the need to include Information concerning current and
future environmental conditions and the potential distribution patterns of
plants and animals, should be included in the decisions for selecting and
establishing any reserve or protected area.
Due to the methodology and the available data used, it is important to
consider that the results obtained in this study might be slightly biased by
some unrecorded errors or even by the lack of enough information. The

[43]
1104

inclusion of only 29 records data for the model generation, might seems not
representative of the species distribution pattern. However the records used
cover, in general terms, all the environmental conditions that theoretically the
species might occupy (the geographic, ecological and altitudinal range of the
taxon).
In addition, natural systems complexity represents a challenge for under-
taking a modeling approach. In particular, the evident limitation of the bio-
climatic models is the lack of inclusion of information concerning biotic
interactions, evolutionary changes, as well as relevant biological processes such
as dispersion (Pearson and Dawson 2003). Consequently, the existence of
certain degree of errors is probably unavoidable.
Also, the bioclimatic data, due to its own nature, shows two kinds of errors:
(1) the omission (= the lack of consideration of the space that is occupied by
the niche; (2) commission (= the consideration of a space that is not occupied
by the niche). Consequently, each algorithm used to model a species ecological
niche, has a combination of commission and omission errors (Peterson and
Vieglais 2001).
Even though, the existence of these errors is recognized, we believe that the
bioclimatic modeling represents a useful tool or starting point for under-
standing the current and potential distribution patterns of animals and plants.
Its usefulness has already been proved for some species at certain scales, in
which this approach has generated relevant information (Pearson and Dawson
2003).
In the case of this study, the model clearly reflects that the spatial climatic
resolution used to correlate it to the species records that were included, enabled
a precise and solid bioclimatic profile of Fagus.
Finally, we believe that with the present biological information, it is fea-
sible and recommendable to carry out a similar exercise for other plant
groups. Endemic species and main elements of plant communities should be
especially important to be submitted to a bioclimatic modeling. By this
means, we can increase the probability of proposing adequate conservation
strategies. In the particular case of this study, the results obtained show that
through the bioclimatic approach, we can be able to focus in long-term
management, planning, and development of new, flexible, and dynamic forms
of wildlife and resource conservation (Nix 1986; Lindenmayer et al. 1991;
Téllez and Dávila 2003).

Acknowledgements

We thank to the anonymous reviewers for their valuable comments and

corrections. To PAPCA 2002 program of the FES Iztacala UNAM for the
financial support to carry out part of this study.

[44]
 1105

References

Alcantara O. and Luna V.I. 1997. Florı́stica y análisis biogeográfico del bosque mesófilo de
montaña de Tenango de Doria, Hidalgo, México. Anales del Instituto de Biologı́a, Universidad
Nacional Autónoma de México, Serie Botánica 68: 57–106.
Alcántara O. and Luna V.I. 2001. Análisis florı́stico de dos áreas con bosque mesófilo de montaña
en el estado de Hidalgo, México: Eloxochitlán y Tlahuelompa. Acta Botánica Mexicana 54:
51–87.
Anónimo 2000. Proyecto de Norma Oficial Mexicana PROY-NOM-059-ECOL-2000, protección
ambiental-especies de flora y fauna silvestres de México-Categorı́as de riesgo y especificaciones
para su inclusión, exclusión o cambio. Lista de especies en riesgo. Diario Oficial de la Federa-
ción. de octubre de 2000.
Arriaga L., Espinoza J.M., Aguilar C., Martı́nez E., Gómez L. and Loa E. (coordinadores) 2000.
Regiones Terrestres Prioritarias de México. Comisión Nacional para el Conocimiento y Uso de
la Biodiversidad, México.
Booth T.H., Nix H.A. and Hutchinson M.F. 1987. Grid matching: a new method for homoclime
analysis. Agric. For. Meteorol. 39: 241–255.
Briones O.L. 1991. Sobre la flora, vegetación y fitogeografı́a de la Sierra de San Carlos, Tamau-
lipas. Acta Botánica Mexicana 16: 15–44.
Canziani O.F. and Diaz S. 1998. Latin America. In: Watson R.T., Zinyowera M.C., Moss R.H.
and Dokken D.J. (eds), The Regional Impacts of Climate Change: An Assessment of Vulnera-
bility. Special Report of IPCC Working Group II. Cambridge University Press, Cambridge, UK,
pp. 187–230.
Cartujano S., Zamudio S., Alcantara O. and Luna I. 2002. El bosque mesófilo de montaña en el
municipio de Landa de Matamoros, Querétaro, México. Boletı́n de la Sociedad Botánica de
México 70: 13–44.
Churchill S.P., Griffin III D. and Lewis M. 1995. Moss diversity of the Tropical Andes. In:
Churchill S.P., Balslev H., Forero E. and Luteyn J.L. (eds), Biodiversity and Conservation of
Neotropical Montane Forestes. Proceedings of the Neotropical Montane Forest Biodiversity and
Conservation Symposium. The New York Botanical Garden, 21–26 June 1993, New York,
pp. 335–348.
ESRI (Environmental Scientific Research Institute) 2000. ArcView 3.2. ESRI. Redlands, Califor-
nia, USA.
Giorgi F., Meehl G.A., Kattenberg A., Grassl H., Mitchell J.F.B., Stouffer R.J., Tokiioka T.,
Weaver A.J. and Wigley T.M.L. 1998. Simulated changes in vegetation distribution under global
warning. In: Watson R.T., Zinyowera M.C., Moss R.H. and Dokken D.J. (eds), The Regional
Impacts of Climate Change: An Assessment of Vulnerability. Special report of IPCC working
group II. Cambridge University Press, Cambridge, UK, pp. 427–437.
Houlder D.J., Hutchinson M.F., Nix H.A. and McMahon J.P. 2000. ANUCLIM 5.1 User Guide,
Centre for Resource and Environmental Studies. Australian National University, Australian
Capital Territory, Canberra.
Houghton J.T., Callander B.A. and Varney S.K. 1992. Climate change 1992. The Supplementary
Report to the IPCC Scientific Assessment. Cambridge University Press, Cambridge, UK.
Hutchinson M.F. 1991. The application of thin-plate smoothing splines to continent- wide data
assimilation. In: Jasper J.D. (ed.), BMRC Research Report Series. Bureau of Meteorology,
Melbourne, Australia, pp. 104–113.
Hutchinson M.F. 1995a. Interpolating mean rainfall using thin plate smoothing splines. Int. J.
Geogr. Inform. Syst. 9: 385–403.
Hutchinson M.F. 1995b. Stochastic space-time weather models from ground-based data. Agric.
For. Meteorol. 73: 237–264.
Hutchinson M.F. 1997. ANUSPLIN. Version 4.1. User guide, Centre for Resource and Envi-
ronmental Studies, Australian National University, Australian Capital Territory, Canberra.

[45]
1106

Hutchinson M.F. and Gessler P.E. 1994. Splines – more than just a smooth interpolator. Geoderma
62: 45–67.
Johnston M.C., Nixon K., Nesom G.L., and Martı́nez M. 1989. Listado de plantas vascul-
ares conocidas de la Sierra de Guatemala, Gómez Farı́as, Tamaulipas, México. Biotam 1:
21–33.
Kappelle M., Van Vuuren M.M.I. and Baas P. 1999. Effects of climate change on biodiversity. A
review and identification of key research issues. Biodiv. Conserv. 8: 1383–1397.
Karl T.A. 1998. Regional trends and variation of temperature and precipitation. In: Watson R.T.,
Zinyowera M.C., Moss R.H. and Dokken D.J. (eds), The Regional Impacts of Climate Change:
An Assessment of Vulnerability. Special Report of IPCC Working Group II. Cambridge Uni-
versity Press, Cambridge, UK, pp. 411–425.
Lindenmayer D.B., Nix H.A., McMahon J.P., Hutchinson M.F. and Tanton M.T. 1991. The
conservation of Leadbeater’s possum, Gymnobelideus leadbeateri (McCoy): a case study of the
use of bioclimatic modelling. J. Biogeogr. 18: 371–383.
Little E.L. Jr. 1965. Mexican beech, a variety of Fagus grandifolia. Castanea 30: 167–170.
López M.L. and Cházaro B.M. 1995. Plantas leñosas raras del bosque mesófilo de montaña.
I. Fagus mexicana Martı́nez (Fagaceae). Boletı́n de la Sociedad Botánica de México 57:
113–115.
Luna V.I., Almeida L., Villers L. and Lorenzo L. 1988. Reconocimiento florı́stico y consideraciones
fitogeográficas del bosque mesófilo de montaña de Teocelo, Veracruz. Boletı́n de la Sociedad
Botánica de México 48: 35–63.
Luna V.I., Alcántara A.O., Espinosa O.D.E. and Morrone J.J. 1999. Historical relationships of the
Mexican cloud forests: a preliminary vicariance model applying Parsimony Analysis of Ende-
micity to vascular plant taxa. J. Biogeogr. 26: 1299–1306.
Luna V.I., Alcántara A.O., Morrone J.J. and Espinosa O.D.E. 2000. Track analysis and conser-
vation priorities in the cloud forests of Hidalgo, Mexico. Div. Distribut. 6: 137–143.
Luna V.I., Morrone J.J., Ayala A.O. and Organista D.E. 2001. Biogeographical affinities among
Neotropical cloud forests. Plant Systemat. Evol. 228: 229–239.
Malda G.B. 1990. Plantas vasculares raras, amenazadas y en peligro de extinción en Tamaulipas.
Biotam 2: 55–61.
McNeely J.A., Gadgil M., Leveque C., Padoch C. and Reedford K. 1995. Human influences on
Biodiversity. In: Heywood V.H. and Warton R.T. (eds), Global Diversity Assessment. Cam-
bridge University Press, Cambridge, UK pp. 711–821.
Moguel P. and Toledo M.V.M. 1999. Biodiversity conservation in traditional coffee systems of
Mexico. Conserv. Biol. 13(1): 11–21.
Morrone J.J. and Crisci J.V. 1995. Historical biogeography: introduction to methods. Annu. Rev.
Ecol. Systemat. 26: 373–401.
Morrone J.J. and Espinosa M.D. 1998. La relevancia de los atlas biogeográficos para la conser-
vación de la biodiversidad mexicana. Ciencia (México) 49: 12–16.
Nix H.A. 1986. A Biogeographic analysis of Australian elapid snakes. In: Longmore R. (ed.), Atlas
of the Elapid snakes of Australia. Flora and Fauna. 7: 4–15.
Neilson R.P. 1998. Simulation of regional climate change with global coupled climate models and
regional modelling techniques. In: Watson R.T., Zinyowera M.C., Moss R.H. and Dokken D.J.
(eds), The Regional Impacts of Climate Change: An Assessment of Vulnerability. Special Report
of IPCC Working Group II. Cambridge University Press, Cambridge, UK, pp. 439–456.
Oldfield S.F., Lusty C. and MacKinven A. 1998. The World List of Threatened Trees. World
Conservation Press.
Pearson R.G. and Dawson T.P. 2003. Predicting the impacts of climate change on the distribution
of species: are bioclimate envelope models useful? Global Ecol. Biogeogr. 12: 361–371.
Pérez P.M. 1994. Revisión sobre el conocimiento dendrológico, silvı́cola y un censo de las pob-
laciones actuales del género Fagus en México. Tesis de maestrı́a (Biologı́a). Facultad de Ciencias.
Universidad Nacional Autónoma de México, México, DF, 146 pp.

[46]
 1107

Pérez P.M. 1999. Las hayas de México.Monografı́a de Fagus grandifolia spp. mexicana. Univers-
idad Autónoma Chapingo, Chapingo, México, 51 pp.
Peterson A.T. and Vieglais D.A. 2001. Predicting species invasions using ecological niche modeling:
new approaches from bioinformatics attack a pressing problem. BioScience 51: 363–371.
Rzedowski J. 1996. Análisis preliminar de la flora vascular de los bosques mesófilos de montaña de
México. Acta Botánica mexicana 35: 25–44.
Pérez P.M. 1999. Las hayas de México.Monografı́a de Fagus grandifolia spp. mexicana. Univers-
idad Autónoma Chapingo, Chapingo, México, 51 pp.
Shen C.F. 1992. A monography of the genus Fagus Tourn. ex. L. (Fagaceae). Dissertation, City
University of New York, New York.
Téllez V.O. and Dávila A.P. 2003. Protected areas and climate change: a case study of the cacti in
the Tehuacán-Cuicatlán biosphere reserve, México. Conserv. Biol. 17(3): 846–853.
Vovides A.P., Luna V. and Medina G. 1997. Relación de algunas plantas y hongos mexicanos
raros, amenazados o en peligro de extinción y sugerencias para su conservación. Acta Botánica
Mexicana 39: 1–42.
Webster G.L. 1995. The Panorama of Neotropical Cloud Forests. In: Churchill S.P., Balslev H.,
Forero E. and Luteyn J.L. (eds), Biodiversity and Conservation of Neotropical Montane
Forestes. Proceedings of the Neotropical Montane Forest Biodiversity and Conservation Sym-
posium. The New York Botanical Garden, 21–26 June 1993, New York, pp. 53–77.
Williams L.G., Rowden A. and Newton A.C. 2003. Distribution and stand characteristics of relict
populations of Mexican beech (Fagus grandifolia var mexicana). Biol. Conserv. 109: 27–36.

[47]
Biodiversity and Conservation (2006) 15:1109–1128
Springer 2006
DOI 10.1007/s10531-004-2178-6

Genetic diversity of Dalbergia monticola (Fabaceae)

an endangered tree species in the fragmented oriental
forest of Madagascar

OLIVARIMBOLA ANDRIANOELINA1,
HERY RAKOTONDRAOELINA2, LOLONA RAMAMONJISOA1,
JEAN MALEY3, PASCAL DANTHU4 and JEAN-MARC BOUVET5,*
1
Silo national des Graines Forestières, Ambatobe BP 5091, Antananarivo, Madagascar; 2PCP Forêts
et Biodiversité/Fofifa DRFP BP 904, Antananarivo, Madagascar; 3Institut des Sciences de l’Evolution
Université de Montpellier II, cc 065, Universite´ Montpellier 2, Place Eugène Bataillon, 34095
Montpellier Cedex 05, France; 4PCP Foreˆts et Biodiversité/Cirad, BP 853, Antananarivo,
Madagascar; 5Cirad-Foreˆt, Campus international de Baillarguet TA10/C, BP 5035, 34398 Mont-
pellier Cedex, France; *Author for correspondence (e-mail: [email protected]; phone:
+33-467593728; fax: +33-467593733)

Received 25 February 2004; accepted in revised form 2 August 2004

Key words: Chloroplast microsatellites, Conservation, Gene flow, Genetic structure, Post-glacial
recolonisation, RAPD

Abstract. There is an urgent need to maintain and restore a broad genetic base for the management
of Dalbergia monticola, a very economically important but endangered tree species in Madagascar.
Random amplified polymorphism DNAs (RAPDs) and chloroplast microsatellite markers were
used to quantify the genetic variation and to analyse the geographic distribution of diversity. Ten
locations covering most of the natural range were sampled. Sixty-three RAPD polymorphic and 15
monomorphic loci were obtained from 122 individuals. Genetic diversity was low and very close
among populations and regions. The unrooted neighbour-joining tree exhibited 4 groups, repre-
senting 6% (p = 0.000) of the total variation. The greater part of the variance, 81%, was observed
within populations. A Mantel test suggested that genetic distances between populations were
weakly correlated with geographic distances (R = 0.46, p = 0.12). The three chloroplast micro-
satellite primers assayed on 100 individuals gave 13 chlorotypes. Most of the populations showed 2
or 3 haplotypes. Haplotype diversity for the total population was equal to HeCp = 0.83 and ranged
from 0.00 to 0.80 among the populations. The unrooted neighbour-joining tree exhibited 4 groups
corresponding to the four regions representing 80% (p = 0.0000) of the total variation. Genetic
diversity varies with regions, the north and south being less variable. Chlorotype distribution, the
phylogenetic tree and historical information suggest that putative refugias in the centre-north
region originating from the early Holocene could explain the pattern of variation observed today.
By combining the results obtained at nuclear and organellar loci, a strategy of conservation based
on evolutionarily significant units is proposed.

Introduction

The separation from Gondwana, 158–160 million years ago, has led to high
endemism in Madagascar recognised as one of the most original in the world
(Myers et al. 2000; Briggs 2003). About 80% of the plant species are endemic
and the richness of fauna and flora is great. Present patterns of the Malagasy

[49]
1110

ecosystem have been determined by numerous factors such as climate change

and human practices. Glaciation cycles, and especially the last maximum
glaciation, are known to have had a strong impact on Malagasy species
distributions (Burney 1996; Gasse and Van Campo 2001). Although known
human presence is not very ancient in the island, 2000 years BP, practices
such as fire have also markedly influenced the distribution of species, espe-
cially since the 15th century (Straka 1996). The highlands would have first
undergone the action of fire and today are covered with grass. More recently,
over the two last centuries, the oriental forests have decreased dramatically
due mainly to ‘‘slash and burn’’ practices. Today, primary vegetation prob-
ably still covers about 10% of the original area (Myers et al. 2000), so dense
forest has been reduced to a fragmented landscape. In addition, forest
exploitation has greatly increased over the last 50 years due to rising demand
for wood as energy and saw timber. The combination of fragmentation and
overexploitation threatens some economically and ecologically important tree
species and a conservation strategy for these forest trees is urgently needed.
Much research remains to be done to improve basic biological knowledge
but, as a broad genetic base is required to maintain the evolutionary process
and to preserve the gene pool, assessment of within-species genetic variation
can be a useful tool when starting a conservation strategy (Newton et al.
1999; Cavers et al. 2003).
Among methodologies employed to assess variation, those based on
molecular markers are widely used with forest tree species (Newton et al. 1999).
Random amplified polymorphism DNA (RAPD) is one of the most popular
DNA-based approaches (Bekessy et al. 2002). It is the least technically
demanding and offers a fast method of providing information from a large
number of loci, particularly in species where no study has been undertaken.
Moreover, the diversity assessed by RAPD is comparable to that obtained with
allozymes or RFLP (Wu et al. 1999; Esselman et al. 2000). There are some
limitations, however, owing to their lack of reproducibility, and dominance
prevents the distinction between homozygous and heterozygous individuals
(Gillies et al. 1997). Chloroplast DNA markers are often used to study genetic
diversity and structure, especially in combination with nuclear ones (Viard
et al. 2001) and have provided useful information on colonisation and dispersal
in plant species. Chloroplasts are maternally inherited in most angiosperms
and paternally inherited in gymnosperms, so the level of differentiation is
greater than bi-parental inheritance. Chloroplast microsatellite markers
(cpSSRs) present high polymorphism and are now frequently used in phylog-
eographic forest tree analyses (Marshall et al. 2002; Palmé and Verdramin
2002; Collevatti et al. 2003; Grivet and Petit 2003). Although molecular
techniques have been widely used in tree species, no studies have been
undertaken for Malagasy tree species.
In this study we used both RAPD and cpSSRs to investigate the pattern of
variation at the natural range scale of a species of rosewood: Dalbergia
monticola Baker. This species is one of the major components of the oriental

[50]
 1111

forest of Madagascar but, for the reasons mentioned previously, it is threa-

tened in its natural stands. Little research has been undertaken that provides
information on ecological patterns and no information is available on genetic
variation across the species range.
The aims of this study were then (i) to quantify the genetic variation within
and between populations using these two molecular markers, (ii) to analyse the
geographic distribution of diversity in the natural range, and (iii) to define a
conservation strategy based on molecular diversity.

Material and methods

Species description – plant material

Dalbergia monticola’s natural range extends from the northern part of

Madagascar (region of Antalaha 15 latitude south) to the southern part
(region of Fort Carnot 22 latitude south) forming a fragmented belt
1000 km long and 100 km wide (see the map in Figure 4). Adult trees
frequently reach 20 m in height and 1 m in diameter at chest height.
Recognised as a long-lived tree species (over 200 years), the natural popu-
lations are situated in two main climaxes: the submontane evergreen sea-
sonal forest and the dense rainforest, of altitude ranging from 350 to
1600 m, mean temperature from 18 to 23 C, and mean annual rainfall from
750 to 2500 mm. Dalbergia monticola reproduces mostly sexually and is
mainly insect pollinated, flowering and fruiting from August to November,
with some geographical variations. It fruits between July and September.
The species is mainly barochorous, but its seeds can also be dispersed by
animals such as birds, monkeys, and rodents, although no research has been
conducted on this.
Ten locations were identified, covering most of the natural range from the
north to south of the island (Table 1), to sample the species. Within each
location, between 6 and 25 trees were chosen randomly in fragmented forest.
Due to the lack of trees in some sampling areas, the minimum distance
between two consecutive trees was in some cases only 20 m (see range of
distance in Table 1). The trees were generally small, most of them having a
total height of 5–20 m and a diameter of 15–30 cm. This observation stresses
the lack of adult trees due to the intense exploitation of the species. Five
healthy leaves were collected from each tree and dried rapidly in the field
using silica gel.

DNA extraction

DNA was extracted from dried leaves, following the modified protocol described
by (Bousquet et al. 1990). Leaves (100 mg) were ground to a fine powder with a

[51]
1112

Table 1. Characteristics of the Dalbergia monticola populations sampled in the natural range.

Region Location N Latitude Longitude Rainfall T Elevation Distancea

(mm) min–max C (m) (m)

North Tsaramolotra 12 1657¢S 4844¢E 750 12–28 900–1200 50–500

Ampitsongona 10 1705¢S 4842¢E 750 12–28 900–1200 50–500
Centre- Didy 17 1810¢S 4835¢E 1240 12–29 800–900 20–50
north Antsevabe 11 1755¢S 4832¢E 1240 12–29 800–900 20–100
Ambohijahanary 11 1745¢S 4835¢E 1240 12–29 800–900 50–1500
Centre Bekorakaka 10 1906¢S 4821¢E 1790–2190 10–26 850–900 50–1000
Madiorano 7 1902¢S 4812¢E 1790–2190 10–26 800–1100 100–1500
Ankeniheny 6 1918¢S 4823¢E 1790–2190 10–26 800–1100 100–500
South Ranomafana 25 2116¢S 4726¢E 2900 17–27 900–1000 20–200
Tolongoina 20 2134¢S 4732¢E 2800 10–26 800–1200 20–1000
Total 139
a
range of distance between two consecutive sampled trees.

mortar and pestle in a 1.5 ml Eppendorf tube under liquid nitrogen. DNA
extraction buffer (5 ml) was added (100 mM Tris–HCl (pH 8.0), 20 mM EDTA,
1.4 M NaCl, 1% PEG 6000, 2% MATAB, 0.5% sodium sulphite). The tube was
then incubated at 74 C for 20 min. Samples were washed with wet chloroform
(CIAA, 24:1) to remove cellular debris and protein. After 15 min of centrifu-
gation at 5000 · g, the liquid phase was transferred to 15 ml tubes. Isopropanol
(5 ml) was added and mixed gently to precipitate the DNA. The resulting DNA
pellets were resuspended in 400 ll of sterile water overnight at 37 C and stored
at
20 C until required.

RAPD methods

PCR amplifications were performed in a 20 ll reaction mix with 5 ll of DNA

(3 ng/ll), 2 · buffer, 0.2 lM of each primer (OPB7, OPB11, OPN15, OPR15,
OPW9, OPW12, OPW13, OPW14, OPX3, OPX6, OPX10), 5 U/ll of DNA
Taq polymerase, completed with sterile water. The reaction mixture was
overlaid with 40 ll of sterile mineral oil to prevent fluid evaporation. All
reactions were performed in Techne Cyclogene. Optimal amplification condi-
tions for RAPDs were 1 cycle of 3 min at 94 C (initial denaturation), followed
by 45 cycles of 4 min at 94 C (denaturation), 1 min at 36 C (annealing) and
2 min at 72 C (extension). A final step of 10 min at 72 C ensured full
extension of all amplified products. RAPD bands were separated in 1.5%
agarose gel, staining in ethidium bromide and visualised by UV transillumi-
nation.
To reduce errors in comparison of RAPD profiles between different PCR
runs, the same 10 individuals were included in all PCR runs. Only RAPD
bands that could be unequivocally scored were counted in the analysis. Bands
of weight higher than 1700 bp and molecular weight lower than 300 bp were

[52]
 1113

not used so as to ensure good repeatability of the RAPD process and avoid
misscoring.

Chloroplast microsatellite method

Three universal microsatellite primers (Ccmp4, Ccmp6 and Ccmp7) described

by Weising and Gardner (1999), and 1 tobacco microsatellite (Ntcp9) described
by (Bryan et al. 1999) were tested over a subset of the total population. Among
the 4 primer pairs tested in a sample of 8 individuals, 3 were polymorphic
(Ccmp4, Ccmp6, and Ccmp7). For the primers Ccmp, PCR amplifications
were done in a 8 ll reaction mix with 2 ll of DNA, 2 · buffer, 10 lM of each
primer (R and F), 5 U/ll of DNA Taq polymerase, completed with sterile
water. All reactions were performed in a ‘‘Stratagene’’ Thermocyclor. Optimal
amplification conditions were 1 cycle of 4 min at 94 C (initial denaturation),
following by 30 cycles of 30 s at 94 C (denaturation), 1 min at 56 C
(annealing) and 1 min at 72 C (extension). A final step of 5 min at 72 C
ensured full extension of all amplified products.
PCR amplifications were done for the 9 primers in a 20 ll reaction mix with
5 ll of DNA, 10 · buffer, 2 lM of each primer (R and F), 5 mM of dNTPs,
50 mM of MgCl2, 99% of glycerol, 5 U/ll DNA Taq polymerase, completed
with sterile water. Amplification conditions are similar, except for the annealing
temperature which is 55 C. Bands were separated and visualised in acrylamide
gel.

Data analysis

In the case of RAPD data, amplified DNA marker bands were scored in a
binary manner as either present (1) or absent (0) and entered into a binary
data matrix. Each PCR product was assumed to represent a single locus
because homology is generally high at the intraspecific level. The frequency of
each band and the percentage of polymorphic loci (%P) were calculated in
each population.
Shannon’s diversity index was P used to assess molecular variation. This
parameter, defined as IRAPD =
i¼2 i¼1 pi log2 pi, where pi is the frequency of
the RAPD phenotype (presence (1) or absence (0) of the band), is frequently
used in the absence of assumptions concerning the Hardy–Weinberg equilib-
rium (Gillies et al. 1997; Martin and Hernandez Bermejo 2000). It was calcu-
lated for each locus and averaged over loci to provide the degree of variation
within each population, IRAPDpop. Shannon’s index was also estimated for the
whole sample considered as a single population, IRAPDtot. The expected genetic
heterozygosity PHer was estimated with the fixation index F equal to zero.
HeRAPD ¼ 1
ni¼1 p2i (where pi is the frequency of the allele i in a population),

[53]
1114

and the other diversity parameters, number of haplotypes (na), effective

number of haplotypes (ne ¼ 1 Pn1 2 , where pi is the frequency of the allele i in
p
i¼1 i
a population), percent of polymorphic RAPD loci (%P) and their standard
errors were calculated with Popgene 1.32. (Yeh and Boyle 1997).
For cpSSRs, because of the non-recombining nature of the chloroplast
genome, cpDNA haplotypes were treated as alleles at a single locus. Chloro-
plast haplotype variation within populations was calculated with the same
parameters as for RAPD (ICp HeCp ne) with Popgene software version 1.32.
(Yeh and Boyle 1997).
The genetic structure, for RAPD and cpSSRs, was estimated using analysis
of molecular variance, AMOVA (Excoffier et al. 1992), with Arlequin software
version 2000 (Schneider et al. 2000). The percentage variance within and
among populations were estimated to partition the variation, and the associ-
ated P value was estimated with permutation techniques.
To illustrate the genetic structure obtained with the two markers, a cluster
analysis using the neighbour-joining method was conducted with the software
package DARwin 4.0 (Perrier et al. 2003). The matrix of genetic distances was
calculated using the AMOVA-derived Fst.
The levels of differentiation among populations estimated from AMOVA for
nuclear markers (FstRAPD) and chloroplast markers (FstCp) could be used to
derive the pollen-to-seed migration ratio, using the following formula of
(Ennos 1994):

r ¼ mp =ms ¼ ½ðð1=FstRAPD Þ
1Þ
2ðð1=Fstcp Þ
1Þ=½ð1=Fstcp Þ
1Þ

where mp and ms are pollen and seed migration rates, respectively. In the case
of RAPDs the value is biased, and probably overestimated, due to the
amplification of the cytoplasmic genome.
The association between geographic and genetic distances was estimated as
a Spearman’s rank correlation coefficient (q). The null hypothesis of the
association was tested with the Mantel test using Fstat software (Goudet
2001).
Minimum spanning networks between haplotypes (each network embedding
all minimum spanning trees for a given distance matrix) were computed with the
MINSPNET (Excoffier and Smouse 1994), provided with Arlequin software
version 2000 (Schneider et al. 2000). The distance matrix between haplotypes was
calculated using a distance matrix based on the square of the difference in
microsatellite size with the formula

X
L
Dij ¼ ðail
ajl Þ2
l¼1

where aij and ajl give the allele size in base pairs at the lth locus of individuals i
and j, respectively.

[54]
 1115

Results

Genetic diversity and structure with RAPD markers

The 15 random primers generated a total of 65 RAPD polymorphic and 13

monomorphic loci ranging in size from 152 to 340 bp. This set of loci is ex-
pected to give a good sampling of the total genome and a good assessment of
the genetic diversity. The number of bands per primer varied from 1 to 7.
Table 2 shows that the diversity parameter for the total population was equal
to 0.19 (0.15) and varied from 0.09 (0.16) for the population of Ankeniheni in
the centre to 0.19 (0.21) for the population of Ambohijanahary in the centre
north. Shannon’s diversity index followed the same pattern of variation, with a
value for the total population equal to 0.30 (0.21) and a range between 0.15
(0.24) in Ankeniheni to 0.28 (0.27) in Ambohijanahary. The percentage of
polymorphic loci varied from P = 30% in Ankeniheni to P = 65% in Didy
(Table 2) and the total population value was 83%. Although differences be-
tween populations for na, ne, HeRAPDand IRAPD were marked, they were
smaller than the standard deviation.
No pattern of variation for na, ne, IRAPD, HeRAPD and %P was observed
among the set of populations. For example, the relationship between
diversity parameters and latitude was low, and the coefficient of correlation
between latitude and IRAPD was equal to R = 0.08, (associated p
value = 0.80).
When the four regions were compared, no specific trend was observed for the
diversity parameter (Table 3). The differences between the parameters were
lower than the standard error.
The differentiation assessed among populations was marked. The analysis of
molecular variance showed that 76% (p = 0.000) of the variation was present

Table 2. Population size (N), number of haplotypes (na), effective number of haplotypes (ne),
Shannon’s index (IRAPD), percent of polymorphic RAPD loci (%P), RAPD diversity (HeRAPD) for
each population. The standard error of each parameter is given between brackets.

Country Population N na ne HeRAPD IRAPD %P

North Tsaramalotro 11 1.51(0.50) 1.21(0.30) 0.14(0.17) 0.22(0.24) 51

Ampisotgoina 10 1.49(0.50) 1.23(0.31) 0.14(0.17) 0.22(0.25) 49
Centre-north Ambohijanahary 11 1.60(0.49) 1.30(0.35) 0.18(0.19) 0.28(0.27) 60
Antsevabe 11 1.43(0.50) 1.24(0.34) 0.14(0.19) 0.21(0.27) 44
Didy 17 1.65(0.48) 1.25(0.29) 0.16(0.16) 0.26(0.24) 65
Centre Bekorokaka 10 1.52(0.50) 1.25(0.32) 0.15(0.18) 0.24(0.25) 53
Madiorano 6 1.41(0.50) 1.20(0.30) 0.12(0.17) 0.19(0.25) 41
Ankeniheny 6 1.30(0.46) 1.16(0.29) 0.09(0.16) 0.15(0.24) 30
South Ranomafana 21 1.64(0.48) 1.20(0.24) 0.14(0.15) 0.23(0.22) 64
Tologoina 19 1.52(0.50) 1.25(0.34) 0.19(0.19) 0.23(0.26) 53
Total 122 1.83(0.38) 1.27(0.27) 0.19(0.15) 0.30(0.21) 83

[55]
1116

Table 3. Population size (N), number of haplotypes (na), effective number of haplotypes (ne),
Shannon’s index (IRAPD), percent of polymorphic RAPD loci (%P), RAPD diversity (HeRAPD) for
each region. The standard error of each parameter is given in brackets.

Region N na ne IRAPD HeRAPD %P

North 21 1.59(0.49) 1.24(0.30) 0.15(0.17) 0.24(0.25) 59

Centre-north 39 1.73(0.44) 1.29(0.29) 0.19(0.16) 0.30(0.24) 73
Centre 22 1.61(0.49) 1.23(0.30) 0.15(0.16) 0.24(0.24) 61
South 40 1.74(0.44) 1.25(0.28) 0.16(0.16) 0.27(0.23) 74

within populations, 18% (p = 0.000) among populations within region, and

6% (p = 0.000) among regions. As a result, the unrooted neighbour-joining
tree obtained with RAPD markers (Figure 1) exhibited three main clusters
corresponding to the three regions south, centre and the combination of

Figure 1. Unrooted neighbour-joining tree based on RAPD markers. The four regions are: the
South (S), the Centre (C), the Centre-north (C-N), and the North (N). The tree was drawn with the
Fst matrix given by analysis of molecular variance (Excoffier et al. 1992) with Arlequin software
version 2000 (Schneider et al. 2000).

[56]
 1117

centre-north and north. The link of Antsevabe with the cluster of the south is
difficult to explain. This clustering suggested a good relationship between the
geographical distance and the genetic distance. This was partly confirmed by
the Mantel test showing a coefficient of correlation between genetic and geo-
graphic distances moderately high (R = 0.46), but not significantly different
from zero (p = 0.12), this relationship is illustrated in Figure 2a.

Genetic diversity and structure with chloroplast microsatellite markers

The three chloroplast microsatellite primers assayed on 100 individuals gave 10

different alleles: Ccmp4, 2 alleles, Ccmp6, 4 alleles and Ccmp7, 4 alleles. The
combination of the 3 loci and the 10 alleles gave 13 chlorotypes (Table 4).
Except for Ampitsongoina, each population exhibited several haplotypes and
generally one chlorotype was predominant. In the total population, the C11

Figure 2. Relation between genetic and geographical distances for the RAPD and chsloroplast
microsatellite markers. Matrices of genetic distances were calculated using the AMOVA-derived Fst
(Arlequin software version 2000 (Schneider et al. 2000)).

[57]
1118

Table 4. Allelic characteristics in base pairs for the three loci, allelic combination of each chlor-
otype and frequencies of the chlorotypes present in each population and in the total population of
Dalbergia monticola.

Zone Population N Chlorotype Ccmp4 Ccmp6 Ccmp7 Frequency

North Tsaramalotro 7 C3 121 120 152 0.14

C5 122 120 152 0.86
Ampitsongoina 6 C5 122 120 152 1.00
Centre-north Ambohijanahary 10 C1 122 119 151 0.20
C2 122 120 151 0.10
C4 122 119 152 0.20
C6 122 121 152 0.10
C9 122 120 153 0.30
C10 122 121 153 0.10
Antsevabe 11 C2 122 120 151 0.64
C4 122 119 152 0.36
Didy 15 C1 122 119 151 0.07
C2 122 120 151 0.66
C4 122 119 152 0.07
C5 122 120 152 0.07
C9 122 120 153 0.13
Centre Bekorokaka 6 C3 121 120 152 0.17
C8 121 120 153 0.67
C12 121 120 154 0.16
Madiorano 6 C8 121 120 153 0.83
C12 121 120 154 0.17
Ankeniheny 2 C2 122 120 151 0.50
C9 122 120 153 0.50
South Ranomafana 23 C7 122 122 152 0.04
C11 122 122 153 0.96
Tolongoina 14 C7 122 122 152 0.21
C11 122 122 153 0.58
C13 122 122 154 0.21
Frequencies in total population: C1(0.03), C2 (0.19), C3 (0.02), C4 (0.07), C5 (0.13), C6 (0.01), C7
(0.04), C8 (0.09), C9 (0.06), C10 (0.01), C11 (0.30) C12 (0.02), C13 (0.03)

haplotypes exhibited the highest frequency (30%), followed by C2 (19%) and

C5 (13%) while others were lower than 10%.
Haplotype diversity varied markedly among the populations and ranged
from HeCp = 0.00 (Ampitsongoina) to HeCp = 0.80 (Ambohijanahary)
(Table 5). For the total population it was equal to HeCp = 0.71. The
number of alleles (na), effective number of haplotypes (ne) and Shannon’s
diversity index (ICp) followed the same pattern. No clear pattern of varia-
tion with latitude was observed when considering the within-population
diversity.
However, when the regional level was taken into account the results showed
that the diversity parameters were higher in the centre-north and centre than in
the north and south. For example, HeCp = 0.68 and 0.55 in the centre-north
and centre, while HeCp = 0.14 and 0.32 in the north and south.

[58]
 1119

Table 5. Diversity parameters assessed with chloroplast microsatellite markers for each popula-
tion and the total population of Dalbergia monticola. Population size (N), number of haplotypes
(na), effective number of haplotypes (ne), Shannon’s index (ICp), haplotypic diversity (HCp).

Region Population N na ne HeCp ICp

North Tsaramalotro 7 2 1.32 0.25 0.41

Ampisongoina 6 1 1 0.00 0.00
Centre-north Ambohijanahary 10 6 5 0.80 1.69
Antsevabe 11 2 1.86 0.46 0.66
Didy 15 5 2.10 0.52 1.08
Centre Bekorokaka 6 3 2 0.50 0.87
Madiorano 6 2 1.38 0.28 0.45
Ankeniheny 2 2 2 0.50 0.69
South Ranomafana 23 2 1.09 0.08 0.18
Tologoina 14 3 2.39 0.58 0.98
Total 100 13 6.10 0.83 2.10

With cpSSR, the unrooted neighbour-joining tree exhibited four clusters

corresponding to the four regions more distinctly than with RAPD data
(Figure 3). Ankeniheny from the centre, however, was included in the cluster of
the centre-north region, but the poor sample size of this population (2 indi-
viduals) can explain this unexpected position. This strong differentiation
among populations was confirmed by the analysis of molecular variance. The
variation among groups represented 80% (p = 0.0000) of total variation, the
within-group variation among populations 4% (p = 0.000) and the between-
individual variation within populations 16% (p = 0.000).

Phylogenetic relation and geographic distribution of the haplotypes

The distribution of the 13 haplotypes across the natural range showed a geo-
graphical structure (Figure 4). Some haplotypes were present in contiguous
populations and regions but none was scattered among distant populations
and regions. The network of haplotypes in Figure 4b shows that haplotype C5
has the highest number of connections and C5 is among the most frequent
(13%). Generally, haplotypes that are closely related to each other, differing by
a single mutation at a microsatellite repeat, are geographically close. For
example, the set of haplotypes C7, C11 and C13 is located in the south,
haplotypes C1, C2, C4, C1 and C5 are in the centre-north, haplotypes C8, C9,
C12, and C10 are in the centre and haplotypes C3, C5 in the north. This
relation of isolation by distance is confirmed by the significant correlation
between genetic distance (Fst) and geographic distance (q = 0.57; p = 0.02).
Figure 2b shows that this linear relationship increased strongly up to 300 km
and then reached a plateau, emphasising that Fst is more or less constant and
close to 1 after this distance.

[59]
1120

Figure 3. Unrooted neighbour-joining tree based on chloroplast microsatellites markers. The four
regions are: the South (S), the Centre (C), the Centre-north (C-N), and the North (N). The tree was
drawn with the Fst matrix given by analysis of molecular variance (Excoffier et al. 1992) with the
Arlequin software version 2000 (Schneider et al. 2000).

Discussion

Genetic diversity of Dalbergia monticola

Although the oriental forest is strongly fragmented, the impact of fragmen-

tation due to human agriculture on the dynamics of the genetic diversity within
the remaining fragments of Dalbergia monticola is likely to be still limited
because it is a very recent phenomenon. It takes several generations before the
impact of genetic drift can be observed. However, the natural range of
Dalbergia monticola was likely to be wider in the recent past and the second
wave of settlement from the 15th century had probably reduced the whole
population in Madagascar and consequently the genetic diversity of the
species.
With polymorphic and monomorphic loci used for calculation, the mean
intraspecific genetic diversity of Dalbergia monticola obtained with RAPD
markers is smaller than that estimated for other tree species (Bekessy et al.

[60]
 1121

Figure 4. (a) Geographic distribution of the 13 chlorotypes in the natural range of

Dalbergia monticola (the limit of the natural range is represented by the dark line). The size of
the circle is proportional to the size of the sample. (b) Minimum spanning network among the
13 haplotypes found in the natural range of Dalbergia monticola. The size of the circle is
proportional to the frequency of the haplotype in the total population. The connection length is
equal to one for the dark line and two for the dotted line.

2002; Newton et al. 2002; Bouvet et al. 2004). Shannon’s diversity index for
these species ranged from 0.42 to 0.65, whereas it is 0.30 (0.21) for D. monti-
cola. Values of expected heterozygosity are similar to those estimated for
perennial species growing on smaller islands, but the percentage of polymor-
phic loci in Dalbergia monticola is higher (Kwon and Morden 2002). Although
RAPD markers should be considered with caution as they are not a good
predictor of total genetic diversity (Nybom and Bartish 2000), the low values of
estimated diversity parameters in this species could be explained by the limited
range of the species. An island population contains less genetic diversity than a
mainland population (Barett 1998). Although Madagascar is a single large
island the isolation from other sources gene could favour the effect of genetic
drift. Another factor that can explain the low diversity is a recent and rapid
expansion after a bottleneck (Savolainen and Kuittinen 2000).
There are few published studies concerning the diversity of cpSSrs in
angiosperm tree species. The present study shows, however, that the results
varied according to the species and the number of primers used. With 3
primers, 100 individuals distributed in 10 populations, the 13 haplotypes
identified in Dalbergia monticola are consistent with previous studies on an-
giosperms (Palmé and Verdramin 2002; Grivet and Petit 2003). In addition, the
haplotypic heterozygosity HeCp = 0.88 is high compared with estimates for
other tree species (Palmé and Verdramin 2002) and suggest high diversity
compared to a number of temperate tree species studied.

[61]
1122

Genetic structure and gene flow

The distribution of genetic diversity with RAPD markers indicates that most of
the variation is present within populations (81%). This can be explained by
some biological patterns such as long-lived woody perennials and outcrossed
insect pollination species according to Nybom and Bartish (2000). Nested
analysis of variance with RAPDs shows that only 6% of the total variation is
attributed to among-group variation, 16% to among populations within
groups, and 76% among individuals within populations. This is low but sig-
nificant differentiation among regions and populations is closely related to the
correlation (moderately high but not significant) between the geographic and
genetic distance. This pattern of isolation by distance is often observed in tree
species wind or insect-pollinated when long distances within the natural range
are considered (Bekessy et al. 2002), while no significant correlation is seen
when short distances are taken into account (Schierenbeck et al. 1997).
The distribution of the diversity is very different using chloroplast micro-
satellites. We note a strong differentiation between regions (80%) and a low
percentage of variance within populations (16%). This is a classical result for
angiosperm forest trees (Raspé et al. 2000) because of the maternal inheritance
of the chloroplast DNA, and seeds are dispersed over shorter distances com-
pared with pollen (Echt et al. 1998). In Dalbergia monticola, seed dispersal is
mainly barochorous and contributes to this strong structure. The relationship
between genetic distance and geographic distance for cpSSr (Figure 2b)
confirms this strong structure and shows that above 300 km Fst is close to 1.
The different patterns of genetic structure of RAPD and cpSSr can also be
viewed through the relative rate of pollen flow to seed flow. With FstCp = 0.84
and FstRAPD = 0.23, the calculation of r showed that gene flow by pollen is 15
times greater than by seeds. The difference is marked but is smaller than for
temperate species such as Fagus silvatica (r = 84), Quercus robur (r = 286),
Quercus petraea (r = 500) (King and Ferris 1998). This can be explained by
the fact that they are wind-pollinated whereas D. monticola is insect-pollinated.

Marker distribution and historical factors

One of the most consistent factors that strongly influenced the partition of ge-
netic diversity of tree species over the natural range is the last glaciation period
and the subsequent migrations from the refuges (Aide and Riviera 1998; Willis
and Whittaker 2000). In Europe (Verdramin et al. 1998), South America (Dutech
et al. 2000; Bekessy et al. 2002) and the Sudano-Sahelian region of Africa
(Bouvet et al. 2004), the genetic variability of tree species has been analysed in
connection with the last glaciation period, 15–20,000 years ago, but no studies,
to our knowledge, have attempted to establish this relation in Madagascar.
As many tropical regions, the island underwent climate fluctuation during
the Pleistocene and Holocene periods, but also the impact of human activities.

[62]
 1123

Three critical periods in the Quaternary should be mentioned. The first was the
Last Glaciation Maximum around 20,000–21,000 years ago which led to an
extremely cold and arid climate. According to (Adams and Faure 1997), during
this episode the rainforest in Madagascar was limited to the north-eastern part
of the country (between latitudes 12 and 16), the high plateaus were covered
with ericoid, graminoid and composite-dominated vegetation and the oriental
escarpment and coastal plain presented tropical woodland characterised by
low, open canopy and usually deciduous trees. Other authors have concluded
that tropical montane vegetation belts must have been vertically displaced 900–
1500 m (Burney 1996). Such a cold-driven displacement of vegetation zones
would have confined the island’s humid forest zones to the relatively small
land area along the east coast, with isolated patches elsewhere (low elevation
humid refugia). From the early Holocene (8000 14C years ago) the rainforest
occupied a broader zone than today. Early Holocene warming led to a gradual
replacement of ericoid vegetation in the mid-elevation with forest in wetter
locations of the eastern escarpment and a rise in the level of Lake Alaotra
(Burney 1996). Some analyses, based on pollen records in high plateau loca-
tions (Lake Itasy, Ankatra, Andasibe) suggested that the rain forest was
present up to elevations of 1800–2000 m (Straka 1996). The third period cor-
responds to human settlements 2000 years ago and the effect of burning which
increased the frequency of fire and resulted in disturbance of the too wet or too
dry ecosystems which generally support a natural fire regime (Burney 1996).
Assuming a limitation of the rainforest in the northern part of the range
during LGM, we can hypothesise that this region may be the zone of putative
refuges for sub-montane rainforest tree species. The assumption cannot be
verified by the results of this study because no sample was collected in the
northern part of the range in the latitude 15. However, our study proposes
another putative refuge. Generally, the highest diversity regions are consid-
ered as putative refuges. The higher diversity in the centre-north region and
the decrease in diversity in the southern and northern limits of the natural
range (Table 6) suggest that the zone of Didy, Ambohijanahary and Ant-
sevabe could be a putative refuge for this species. Phylogenetic trees based on
haplotypes (Figures 4b) and their distribution across the natural range
(Figure 4a) also point to a pattern of migration from the centre-north region.

Table 6. Diversity parameters assessed with chloroplast microsatellite markers for the four main
regions of the natural range of Dalbergia monticola. Population size (N), number of haplotypes
(na), effective number of haplotypes (ne), Shannon’s index (ICp), haplotypic diversity (HCp).

Region N na ne HeCp ICp

North 13 2 1.17 0.14 0.27

Centre-north 36 7 3.16 0.68 1.44
Centre 14 5 2.22 0.55 1.13
South 37 3 1.48 0.32 0.61

[63]
1124

Haplotype C5 occupies a central position in the phylogenetic trees with the

highest number of connections (Figure 4b) and is present in the centre-north
region. It is likely to be the ancestor of most of the other chlorotypes which
mutate during the northward and southward expansion. A putative refuge in
the centre-north region, close to Lake Alaotra, is also supported by the study
of Reyes (1993) cited by (Burney 1996) who inferred the Late Quaternary
climatic changes in north-eastern and central Madagascar using diatom
spectra. In this study, Reyes (1993) indicated that early Holocene warming
led to the gradual replacement of ericoid vegetation in the mid-elevation with
forest developing in wetter locations along the eastern escarpments and a rise
in the level of Lake Alaotra.
Human settlement probably started 2000 years ago and the impact on the
forest ecosystem of human practices has increased over the last two millennia.
The probable effect of burning by humans has been to reduce the natural range
of Dalbergia monticola. Without an uncontrolled use of burning, the natural
range of Dalbergia monticola would be wider than today. Our study does not
allow us to measure this effect. More recently, 50 years ago, the impact of
human activity led to a marked increase in fragmentation in the oriental
rainforest. Fragmentation is a factor which can seriously disturb the evolu-
tionary process and which has to be taken into consideration in conservation.
New investigations are needed to measure its impact on the genetic diversity of
Dalbergia monticola.

Consequences for conservation of the species

Forest genetic resources in Madagascar constitute an essential component of

the biological diversity of forest ecosystems. They supply a wide range of goods
and services essential to the life of rural communities and efforts to promote
sustainable management are required. Strategies involving genetic studies are
still rare and exploration of the genetic diversity of Malagasy trees is expected.
Newton et al. (1999) emphasise that intraspecific diversity has become a
parameter fundamental to the management of species with the aim of main-
taining their evolutionary potential.
In this study we used neutral markers to assess diversity. The use of such
markers to predict the diversity of adaptive genes or economic traits within
populations has to be considered with caution (Lynch et al. 1999). RAPD or
cpSSrs are different from genes and may not be a good predictor of gene
function diversity or of complex traits. However, molecular markers can give
indications of the genetic distance between populations and of their different
evolutionary histories (Holsinger 1996), and may indicate a similar genetic
structure for morphological traits (Nesbitt et al. 1995). Different genes could
have been selected, appeared by mutation or have been eliminated by genetic
drift in separated populations which do not exchange genes. Some authors use
intraspecific diversity to elaborate conservation strategies (Rajagodal et al.

[64]
 1125

2000). The concept of evolutionary significant units (ESUs) has received

increasing support as providing a rational basis for identifying suitable units
for conservation (Moritz 1994; Newton et al. 1999) and has been implemented
in forest genetic resources when nuclear and organelle haplotypes are available
(Cavers et al. 2003). This approach can be proposed given the methodology
and the results of the current study. Chloroplast markers can provide a basis
for defining ESUs using the main lineages, their distribution in the natural
range and the result showing the differentiation between populations (AM-
OVA and Mantel test). The set of regions north and centre-north, the centre
region and the south region can be identified as ESUs. Significant divergence
between these three groups with RAPD markers confirms the identification of
the three ESUs. Although pollen gene flow is important, isolation by distance
was observed by means of the Mantel test, and a significant genetic differen-
tiation among populations and regions was shown by AMOVA. RAPD
markers indicate that most of the variability is present within populations,
stressing the importance of gene flow among individuals. This result could help
to define a strategy to elaborate ex and in situ conservation stands. To start
such a programme, the population should consist of many individual trees
selected within a few natural populations within the ESUs to capture a large
proportion of variation.
Much work is needed to elaborate an optimal strategy. As stressed by Cavers
et al. (2003), the best strategy will concord with use of neutral markers and
adaptive information (also ecological information), but a first step in the
identification of ESUs with molecular markers provides a first practical
framework.

Acknowledgements

We are very grateful to colleagues from ‘‘Silo National des Graines Forestières’’
for assistance in collecting samples, to Alexandre Vaillant and Céline Cardi from
Cirad-forêt for their technical assistance in the laboratory, and to the anonymous
reviewers. These results are part of master of science thesis of O. Andrianoelina,
which was financially supported by the Millennium Seed Bank (MSB) project,
managed by the Seed Conservation Department, RBG Kew. The study was
carried out at SNGF, which receives funds from the Malagasy Government, as
well as at Cirad-Forêt, where laboratory work was done. L. Ramamonjisoa, head
of the technique division in SNGF and J.M. Bouvet, head of the Research unit on
forest genetic resources, were responsible for supervision.

References

Adams J.M. and Faure H. 1997. Preliminary vegetation maps of the world since the last glacial
maximum: an aid to archeological understanding. J. Archeol. Sci. 24: 623–647.

[65]
1126

Aide T.M. and Rivera E. 1998. Geographic patterns of genetic diversity in Poulsenia amata
(Moraceae): implications for the theory of Pleistocene refugia and the importance of ripian
forest. J. Biogeogr. 25: 695–705.
Barrett S.C.H. 1998. The reproductive biology and genetics of island plants. In: P.R. Grant (ed),
Evolution on islands. Oxford University Press, Oxford, UK, pp. 19–34.
Bekessy S.A., Allnutt T.R., Premoli A.C., Lara A., Ennos R.A., Burgman M.A., Cortes M. and
Newton A.C. 2002. Genetic variation in the vulnerable and endemic Monkey Puzzle tree,
detected using RAPDs. Heredity 88: 243–249.
Bousquet J., Simon L. and Lalonde M. 1990. DNA amplification from vegetative and sexual tissues
of trees using polymerase chain reaction. Can. J. Forest. Res. 20: 254–257.
Bouvet J.-M., Fontaine C., Sanou H. and Cardi C. 2004. An analysis of the pattern of genetic
variation in Vitellaria paradoxa using RAPD markers. Agroforest. Syst. 60: 61–69.
Briggs J.C. 2003. The biogeographic and tectonic history of India. J. Biogeogr. 30: 381–388.
Bryan G.J., McNicoll J., Ramsay G., Meyer R.C. and De Jong W.S. 1999. Polymorphic simple
sequence repeat markers in chloroplast genomes of solanaceous plants. Theoret. Appl. Genet. 99:
859–867.
Burney D.A. 1996. Climate change and fire ecology as factors in the quaternary biogeography of
Madagascar. In: Lourenço W.R. (ed.), Edition de l’ORSTOM Biogéographie de Madagascar.
Orstrom, Paris, France, pp. 49–58.
Cavers S., Navarro C. and Lowe A.J. 2003. A combination of molecular markers identifies evo-
lutionarily significant units in Cedrela odorata L. (Meliaceae) in Costa Rica. Conserv. Genet. 4:
571–580.
Collevatti R.G., Grattapaglia D. and Hay J.D. 2003. Evidence for multiple lineages of
Caryocar brasiliense populations in the Brazilian Cerrado based on the analysis of chloroplast
DNA sequence and microsatellite haplotype variation. Mol. Ecol. 12: 105–115.
Dutech C., Maggia L. and Joly H.I. 2000. Chloroplast diversity in Vouacapoua americana (Cae-
salpiniaceae), a neotropical forest tree. Mol. Ecol. 9: 1427–1432.
Echt C.S., DeVerno L.L., Anzidei M. and Vendramin G.G. 1998. Chloroplast microsatellites reveal
population genetic diversity in red pine, Pinus resinosa Ait. Mol. Ecol. 7: 307–316.
Ennos R.A. 1994. Estimating the relative rates of pollen and seed migration among plant popu-
lations. Heredity 72: 250–259.
Esselman E.J., Crawford D.J., Brauner S., Stuessy T.F., Anderson G.J. and Silva O.M. 2000.
RAPD marker diversity within and divergence among species of Dendroseris (Asteraceae:
Lactuceae). Am. J. Bot. 87(4): 591–596.
Excoffier L., Smouse P.E. and Quattro J.M. 1992. Analysis of molecular variance inferred from
metric distance among DNA haplotypes: application to human mitochondrial DNA restriction
data. Genetics 131: 479–491.
Gasse F. and Van Campo E. 2001. Late quaternary environmental changes from a pollen and
diatom record in the southern tropics (lake Tritivakely, Madagascar). Paleogeogr. Paleoclimatol.
Paleoecol. 167: 287–308.
Gillies A.C.M., Cornelius J.P., Newton A.C., Navarro C., Hernandez M. and Wilson J. 1997.
Genetic variation in Costa Rican populations of the tropical timber species Cedrela odorata L.,
using RAPDs. Mol. Ecol. 6: 1133–1145.
Grivet D. and Petit R. 2003. Chloroplast DNA phylogeography of the hornbeam in Europe: evidence
for the bottleneck at the outset of the postglacial colonisation. Conserv. Genet. 4: 47–56.
Goudet J. 2001. FSTAT, a program to estimate and test gene diversities and fixation indices
(version 2.9.3). Available from http://www.nil.ch/izea/softwares/fstat.html.
Holsinger K.E. 1996. The scope and the limits of conservation genetics. Evolution 50: 2558–2561.
King R.A. and Ferris C. 1998. Chloroplast DNA phylogeography of Alnus glutinosa (L.) Gaertn.
Mol. Ecol. 7: 1151–1161.
Kwon J.A. and Morden C.W. 2002. Population genetic structure of two rare tree species
(Colubrina oppositifolia and Alphitonia ponderosa, Rhamnaceae) from Hawaiian dry and mesic
forest using random amplified polymorphic DNA markers. Mol. Ecol. 11: 991–1001.

[66]
 1127

Lynch M., Pfrender M., Spitze K., Lehman N., Hicks J., Allen D., Latta L., Ottene M., Bogue F.
and Colbourne J. 1999. The quantitative and molecular genetic architecture of a subdivided
species. Evolution 53: 100–110.
Marshall H.D., Newton C. and Ritland K. 2002. Chloroplast phylogeography and evolution of
highly polymorphic microsatellites in lodgepole pine (Pinus contorta). Theoret. Appl. Genet. 104:
367–378.
Martin J.P. and Hernandez Bermejo J.E. 2000. Genetic variation in the endemic and endangered
Rosmarinus tomentosus Huber-Morath & Maire (Labiatae) using RAPD markers. Heredity 85:
434–443.
Moritz C. 1994. Defining significant units for conservation. Trends Ecol. Evol. 9: 373–375.
Myers N., Mittermeier R.A., Mittermeier C.G., da Fonseca G.A.B. and Kent J. 2000. Biodiversity
hotspots for conservation priorities. Nature 403: 853–858.
Nybom H. and Bartish I.V. 2000. Effects of life history traits and sampling strategies on genetic
diversity estimates obtained with RAPD markers in plants. Perspect. Plant Ecol. Evol. Syst. 3:
93–114.
Nesbitt K.A., Potts B.M., Vaillancourt R.E., West A.K. and Reid J.B. 1995. Partitioning and
distribution of RAPD variation in a forest tree species, Eucalyptus globulus (Myrtaceae).
Heredity 74: 628–637.
Newton A.C., Allnutt T.R., Dvorak W.S., Del Castillo R.F. and Ennos R.A. 2002. Patterns of
genetic variation in Pinus chiapensis, a threatened Mexican pine, detected by RAPD and mito-
chondrial DNA RFLP markers. Heredity 89: 191–198.
Newton A.C., Allnutt T.R., Gillies A.C.M., Lowe A.J. and Ennos R.A. 1999. Molecular
phylogeography, intraspecific variation and the conservation of tree species. TREE 14(4):
140–145.
Palme A.E. and Verdramin G.G. 2002. Chloroplast DNA variation, postglacial recolonization and
hybridation in hazel, Corylus avellana. Mol. Ecol. 11: 1769–1779.
Perrier X., Flori A. and Bonnot F. 2003. Data analysis methods. In: Hamon P., Seguin M., Perrier
X. and Glaszmann J.C. (eds), Genetic diversity of cultivated tropical plants Enfield. Science
Publishers, Montpellier,pp. 43–76.
Rajagopal J., Bashyam L., Bhatia S., Khurana D.K., Srivastava P.S. and Lakshmikumaran M.
2000. Evaluation of genetic diversity in the Himalayan poplar using RAPD markers. Silvae
Genet. 49(2): 49–66.
Raspé O., Saumitou-Laprade P., Cuguen J. and Jacquemart A.-L. 2000. Chloroplast DNA hap-
lotype variation and population differentiation in Sorbus aucuparia L. (Rosaceae: Maloideae).
Mol. Ecol. 9: 1113–1122.
Savolainen O. and Kuittinen H. 2000. Small population processes. In: Youg A., Boshier D. and
Boyle T. (eds), Forest Conservation Genetics. Principles and Practises, CSIRO Publishing
Collingwood Victoria, Australia. pp. 94–100.
Schierenbeck K.A., Skupski M., Lieberman D. and Lieberman M. 1997. Population structure and
genetic diversity in four tropical tree species in Costa Rica. Mol. Ecol. 6: 137–144.
Schneider S., Roessli D. and Excoffier L. 2000. Arlequin version 2000: a software for population
genetics data analysis. Genetics and Biometrics Laboratory, Department of Anthropology,
University of Geneva, Geneva, Switzerland.
Straka H. 1996. Histoire de la végétation de Madagascar oriental dans les derniers 100 millénaires.
In: Lourenço W.R. (ed.), Editions de l’ORSTOM, Biogéographie de Madagascar. Orstrom,
Paris, France, pp. 37–47.
Vendramin G.G., Anzidei M., Madaghiele A., Sperisen C. and Bucci G. 1998. Distribution of
genetic diversity in Pinus pinaster Ait. as revealed by chloroplast microsatellites. Theoret. Appl.
Genet. 97: 456–463.
Viard F., El-Kassaby Y.A. and Ritland K. 2001. Diversity and genetic structure in populations of
Pseudotsuga menziesii (Pinaceae) at chloroplast microsatellite loci. Genome 44: 336–344.

[67]
1128

Weising K. and Gardner R.C. 1999. A set of conserved PVR primers for the analysis of simple
sequence repeat polymorphism in chloroplast genomes of dicotyledonous angiosperms. Genome
42: 9–19.
Willis K.J. and Whittaker R.J. 2000. The refugial debate. Science 287: 1406–1407.
Wu J., Krutovskii K.V. and Strauss S.H. 1999. Nuclear DNA diversity, population differentiation
and phylogenetic relatioships in the California closed-cone pines based on RAPD and allozyme
markers. Genome 42: 893–908.
Yeh F.C. and Boyle T.J.B. 1997. Population genetic analysis of co-dominant and dominant
markers and quantitative traits. Belg. J. Bot. 129: 157.

[68]
Biodiversity and Conservation (2006) 15:1129–1142
Springer 2006
DOI 10.1007/s10531-004-3103-8

-1

Forest management and plant species diversity

in chestnut stands of three Mediterranean areas

HÉLÈNE GONDARD1,*, FRANÇOIS ROMANE1, IGNACIO SANTA

REGINA2 and SALVATORE LEONARDI3
1
CEFE (UMR 5175), 1919 route de Mende, 34293 Montpellier cedex 5, France; 2Instituto de
Recursos Naturales y Agrobiologia de Salamanca, P. box 257 C/Cordel de Merinas 40, E-37071
Salamanca, Spain; 3Dipartimento di Metodologie Fisiche e Chimiche per l’Ingegneria, Facoltà di
Ingegneria, Università di Catania, viale A. Doria 6, 95125 Catania, Italy; *Author for correspondence
(e-mail:[email protected]; phone: +33-4-67613276; fax: +33-4-67412138)

Received 6 February 2004; accepted in revised form 31 August 2004

Key words: Castanea sativa, Coppice stand, Diversity index, Functional trait, Grove

Abstract. Over many centuries, chestnut fruits had an important role as food, while chestnut wood
was used for local purposes. Today sweet chestnut stands are very common around the western
Mediterranean Basin, and it is necessary to analyze the dynamic of plant species diversity in
different chestnut stand types (groves and coppices) to guide management strategies that will allow
the conservation of biodiversity. Our objective was to analyze consequences on plant species
diversity of various management strategies in chestnut stands of three Mediterranean areas, Sal-
amanca (Spain), the Cévennes (France), and Etna volcano (Italy). We found that plant species
diversity is different according to management types; it is higher in groves than in coppice stands.
We also demonstrated that Castanea sativa cultivated groves were characterized by small helio-
phillous therophytes. C. sativa abandoned groves, mixed C. sativa–Quercus pyrenaica coppice
stands, Q. pyrenaica coppice stands, and young C. sativa coppice stands were characterized by
hemicryptophytes with anemochorous dispersal mode and chamaephytes. Medium and old
C. sativa coppice stands (that differ by the shoot age) were characterized by phanerophytes with
zoochorous dispersal mode. Human perturbations maintain a quite high level of species diversity.
In contrast, the abandonment of chestnut stands leads to homogeneous vegetation with decreasing
diversity. One solution could be to maintain a landscape mosaic constituted of diverse chestnut
stands modified by human activities (groves, cultivated or abandoned, and coppice stands). This
could enhance regional plant diversity.

Nomenclature – Flora Europaea (Tutin et al. 1964–1980)

Introduction

Sweet chestnut (Castanea sativa Mill) stands are very common around the
western Mediterranean Basin. Over many centuries, chestnut fruits had an
important role as food for humans and as feed for domestic animals, while
chestnut wood was used for local purposes such as wine barrels, vineyard pegs,
tool handles and carpentry (Arnaud and Bouchet 1995). Today, chestnut

[69]
1130

stands cover large areas particularly in Portugal, Spain, France, Italy and
Greece. Thus, it is necessary to analyze the dynamic of plant species diversity in
different chestnut stand types (groves and coppices) to guide management
strategies that will allow the conservation of biodiversity and at the same time
to optimize productivity and profitability.
The characterization of community response to different management types
in terms of functional traits appears as a promising tool to achieve this goal
(McIntyre et al. 1995; Hadar et al. 1999; Lavorel et al. 1999; Gondard et al.
2003). Indeed, from an ecosystem perspective, species richness (number of
species), which is the conventional metric of biodiversity, is not as important as
functional trait richness. This approach analyzes the functioning of the eco-
system, and its response to abandonment, by focusing on vegetation descrip-
tion defined by functional traits not necessarily linked with taxonomic
attribution (Pillar 1999). Functional traits fall into three biological categories:
morphological traits describing aspect, life history traits indicating plant
behavior in the environment, and regeneration traits (Lavorel et al. 1997). The
use of functional traits for the comprehension and analysis of plant species
dynamics in relation with perturbation is clearly demonstrated by many au-
thors (Dı́az and Cabido 1997; Lavorel and Cramer 1999; McIntyre et al. 1999;
McIntyre and Lavorel 2001; Dı́az et al. 2002; Gondard and Deconchat 2003).
Consequently, our objective was to analyze consequences on plant species
diversity of various management strategies in chestnut stands of three Medi-
terranean areas, Salamanca (Spain), the Cévennes (France), and Etna volcano
(Italy). We hypothesized that, whatever area, species diversity between groves
and coppice stands is different essentially according to dendrometric charac-
teristics and management types. Indeed, groves have, in general, large trees
with regular pruning, understorey cleaning, etc., and coppices have many
shoots without clearing but logging. We assumed that species diversity is
highest in groves. We focused on understorey stratum which is sensitive to
changes of ecosystem conditions (Pregitzer and Barnes 1982; Strong et al. 1991;
Mitchell et al. 1997, 1998) and recognized like a very important component in
ecosystem functioning (Host and Pregitzer 1991; Arsenault and Bradfield 1995;
Brakenhielm and Lui 1998).

Materials and methods

The experiment was carried out in three Mediterranean areas, in the Honfrı́a
forest, located in the southern of Salamanca province in Spain, in the Cévennes
in southern France, and on Etna volcano in Italy (Table 1). The Honfrı́a forest
is representative of traditional chestnut (Castanea sativa) management over
many centuries in Spain, but also a model of possible sustainable management
in the future. In this forest, chestnut is considered as a paraclimax species and
the deciduous oak (Quercus pyrenaica) as a climax species. Thus, we selected
five stands that are representative of this forest: a chestnut cultivated grove, a

[70]
 1131

chestnut abandoned grove, a chestnut coppice stand, a mixed chestnut-oak

stand, and an oak pure stand. In the Cévennes, we identified a succession
following agricultural abandonment from chestnut cultivated grove to chestnut
old coppice stand. Thus, we selected five stages that form the successional
gradient: a cultivated grove, an abandoned grove, a young coppice (<25 years
old), a medium coppice stand (between 26 and 50 years old), and an old
coppice stand (>51 years old). On Etna volcano, the tradition is coppice
management and not grove. Thus, we selected five coppice stands (Fornazzo,
Trisciala, Balilla, Monte Crisimo, Piano Lepre) that are representative of the
study area and differ by their stand characteristics (Table 2).
In each stand, we established five 10 · 10 m plots. The plots were contiguous
because there was only little area available at the site with relatively homo-
geneous topographic conditions, and in order to respect 100 m2 plot size
minimum. In each plot, we recorded all plant species occurring in the under-
storey stratum. The plant cover of each species was estimated by the point
quadrat method (Gounot 1969), using 100 points, i.e. one point each 10 cm,
along a 10 m line traversing each plot. According to previous observations,
realized by one of us, 100 m2 plots appeared to be suitable for monitoring this
kind of vegetation. Data were collected during June month in 2001 in the
Cévennes, 2002 in the Honfrı́a forest, and 2003 in Italy. Moreover, plant
species recorded were characterized by functional traits such as plant height
and life form that refer to morphology, light tolerance that refers to life history
traits, and dispersal mode to regeneration traits (Appendix 1).

Data analyses

The criteria to compare stands were species richness (number of taxa per 100 m2)
and species diversity (Pielou 1975; Magurran 1988). Among the many diversity
indices available, we chose
P the Shannon index (H’), which was recommended by
Pielou (1975): H0 ¼
i¼1;n ðpi log2 ðpi ÞÞ where pi is the abundance ratio of
species (i) in the square, and n is the species number in the square.
Forest stands in the three geographical areas are submitted to different sil-
vicultural management and also to contrasting environment and climate con-

Table 1. Characteristics of the three Mediterranean areas studied.

Honfrı́a Forest Cévennes Etna volcano

Spain France Italy

Altitude (m) 900 650 1000

Mean annual rainfall (mm yr
1) 1500 1400 1100
Mean annual temperature (C) 11 11 12
Parent material Schist Schist volcanic ash, lava
Soil Cambisol Cambisol regosol volcanic

[71]
1132

Table 2. Main characteristics of chestnut stands selected in the Cévennes in France, on Etna
volcano in Italy and in Honfrı́a Forest in Spain. Confidence intervals p = 0.05. For each site, mean
values in the same column followed by different letters are significantly different. p < 0.05, Mann–
Whitney test.

Site Stand Tree age Tree Diameter Shoot density Basal area
(years) height (m) at breast (shoot ha
1) (m 2ha
1)
height (cm)

Honfrı́a C. sativa 90 11.30 ± 1.3b 18.30 ± 2.1c 295 ± 20a 23 ± 5b

Forest cultivated grove
Spain
C. sativa 85 8.90 ± 0.8a 20.40 ± 3.0c 382 ± 30a 19 ± 4a
abandoned grove
C. sativa 70 15.3 ± 1.3c 12.90 ± 1.7b 1892 ± 100b 28 ± 8c
coppice stand
Mixed C. sativa–Q. 60 10.7 ± 0.8b 8.90 ± 1.2a 3208 ± 150c 21 ± 5a
pyrenaica stand
Q. pyrenaica 75 12.2 ± 1.0b 11.60 ± 1.5b 2960 ± 125c 27 ± 7c
pure stand
Cévennes Cultivated grove 70 18.00 ± 1.0a 45.00 ± 7.1a 120 ± 45a 26 ± 18a
France
Abandoned grove 75 17.40 ± 0.5a 44.60 ± 11.5a 440 ± 195a 45 ± 21a
Young coppice 16 11.20 ± 0.8b,c 9.40 ± 1.5b 1040 ± 611b 8 ± 4b
Medium coppice 39 12.40 ± 0.9c 17.80 ± 7.8c,d 1080 ± 396b 17 ± 14c,d
Old coppice 56 16.40 ± 0.5d 24.00 ± 4.8d 840 ± 488b 35 ± 13d
Etna volca Fornazzo coppice 31 17.67 ± 0.7a 9.20 ± 0.5a 4680 ± 1242a,c 38 ± 8a,b
Italy
Trisciala coppice 28 12.17 ± 0.3a 7.20 ± 0.3b 6020 ± 895a 29 ± 3a
Balilla coppice 37 16.67 ± 2.3a 20.9 ± 1.7c 1140 ± 331b 43 ± 8b
Monte Crisimo 26 16.00 ± 0.0a 7.10 ± 1.2d 2900 ± 919c 24 ± 6c
coppice
Piano 27 15.67 ± 3.2a 9.90 ± 0.5a 4180 ± 394d,c 38 ± 1a,b
Lepre coppice

ditions, so differences are expected between them. However, due to the low
number of stands analyzed, we used non-parametric test that allows to work
with low size samples. We chose the Mann–Whitney non-parametric test that
allows to compare means pairwise (Falissard 1998).
In each Mediterranean area, we used Correspondence Analysis (CA) and
Canonical Correspondence Analysis (CCA, ter Braak 1987) to quantify the
effects of management types with species functional traits. We performed a
Correspondence Analysis (CA,Greenacre 1984) of plant species observed on
the entire point quadrat set (67 in Honfrı́a forest in Spain, 41 in the Cévennes in
France, and 40 on Etna volcano in Italy) and management types (coppice
stands and groves in Honfrı́a forest and in the Cévennes, and different coppice
stand types on the Etna volcano). We used CCA to determine the fraction of
variance of the species among management types explained by the species and
functional traits. For each Mediterranean area, we carried out the CCA by
confronting the CA table with another table composed by the same species

[72]
 1133

60 a
55
50
Species richness

45
40
35
30
25
b
bc b
20 c
15
10
5
0
C. sativa C. sativa C. sativa Mixed C. sativa- Q. pyrenaica
cultivated grove abandoned coppice stand Q. pyrenaica stand
grove stand

Figure 1. Mean species richness in the understorey of the C. sativa and Q. pyranaica stands of the
Honfrı́a forest in the southern of Salamanca province in Spain. Error bars at ±95% confidence
limits. Two different letters between the coppice stands indicated significant statistical difference
(Mann–Whitney non-parametric test, p < 0.05).

number and functional traits separated in subclasses. Moreover, hierarchical

ascending classification was used to make easier the identification of groups in
factorial plans (Roux 1985).

Results

Species richness and species diversity (Shannon index)

In the Honfrı́a forest and the Cévennes, species richness was highest in culti-
vated groves, (Figures 1, 2). On Etna volcano, species richness was highest in

45 a
40
Species richness

35
30 b
25 c cd
20 d
15
10
5
0
Cultivated Abandoned Young Medium Old
Grove Grove Coppice Coppice Coppice

Figure 2. Mean species richness along a successional gradient from cultivated chestnut grove to
old C. sativa coppice stand (Le Cros site in the Cévennes). Error bars at ±95% confidence limits.
Two different letters between the coppice stands indicated significant statistical difference (Mann–
Whitney non-parametric test, p < 0.05).

[73]
1134

45
40
Species richness

35 a
30
25
20 b
bc
15 cd
d
10
5
0
Trisciala Monte Fornazzo Balilla Piano Lepre
Crisimo

Figure 3. Mean species richness in the understorey in the five coppice C. sativa stands on the Etna
volcano in Italy. Error bars at ±95% confidence limits. Two different letters between the coppice
stands indicated significant statistical difference (Mann–Whitney non-parametric test, p < 0.05).

the Trisciala coppice stand (Figure 3), and not significantly different from the
abandoned grove in the Cévennes (p > 0.05). Species diversity was also
highest in cultivated groves and Trisciala coppice stand.

Plant functional traits and management types

Consequences of various management types on plant species in term of func-

tional traits were analyzed with CCA, and hierarchical ascending classification
allowed the identification of several groups in each Mediterranean area

1
Axis 2
(inertia 15%)
Honfría forest - Group 2 0,8
C.sativa coppice stands
Geophyte Honfría forest - Group 3
Arpa, Casa, Celo, Prav,
Qupy, Tesc C. sativa abandoned groves & Mixed
0,6
C. sativa-Q. pyrenaica coppice stands
& Q. pyrenaica coppice stands
Anod, Caof, Capa, Daca, Gehi, Gepi,
0,4 Haha, Jamo, Loco, Meme, Feru, Himu

> 50 cm 0,2 Barochorous

Heliophillous
Axis 1 Zoochorous Hydrochorous
Anemochorous Therophyte
(inertia 48%)
0
-1 -0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1 1,2
Phanerophyte Chamaephyte 30-50 cm
Shade tolerant -0,2 1-30 cm

Autochorous Honfría forest - Group 1

-0,4 C. sativa cultivated groves
Hemicryptophyte Anpr, Boma, Brho, Cegl, Cran, Crvi, Cyec,
Gaap, Tran, Trar, Trca, Scan, Vubr
-0,6

Figure 4. Ordination in the plane of the two axes of functional traits after a canonical corre-
spondence analysis from a matrix composed by the 67 plant species observed on the line point
quadrat of the 25 plots in the Honfrı́a forest in Spain and a matrix composed by the same plant
species and their functional traits. Groups were identified by an hierarchical ascending classifica-
tion. Plots, and some plant species associated to each group were indicated on the figure. Codes of
plant species are indicated in Appendix 1. The total variation explained by CCA is 39%.

[74]
 1135

0,8
The Cévennes - Group 2 Axis 2
The Cévennes - Group 3
C. sativa abandoned (inertia 28%)
groves & young coppice
C. sativa medium and & old
Autochorous coppice stands
stands 0,6
Armo, Clvu, Bepe, Csa, Hede, Qupu, Rops
Dagl, Erci,
Hima, Himu, Barochorous
Pone, Prvu, 0,4 Phanerophyte
Trpr, Veof, > 50 cm
Visa
Chamaephyte 0,2
Zoochorous

Shade tolerant
0
-0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 Axis 2 1
30-50 cm Hydrochorous Geophyte
Heliophillous (inertia 37%)

Therophyte -0,2
1-30 cm

Hemicryptophyte -0,4 The Cévennes - Group 1

C. sativa cultivated groves
Anemochorous Ruac, Orpe, Ptaq, Tesc

-0,6

Figure 5. Ordination in the plane of the two axes of functional traits after a canonical corre-
spondence analysis from a matrix composed by the 41 plant species observed on the line point
quadrat of the 25 plots in the Cévennes in France and a matrix composed by the same plant species
and their life traits. Groups were identified by an hierarchical ascending classification. Plots, and
some plant species associated to each group were indicated on the figure. Codes of plant species are
indicated in Appendix 1. The total variation explained by CCA is 46%.

studied. In the Honfrı́a forest, the C. sativa cultivated groves (group 1) were
characterized by small heliophillous therophytes (Figure 4). The C. sativa
coppice stands (group 2) were characterized by shade tolerant phanerophytes
with zoochorous dispersal mode and geophytes. The C. sativa abandoned
groves, mixed C. sativa–Q. pyrenaica coppice stands and Q. pyrenaica coppice
stands (group 3) were composed essentially by hemicryptophytes and
chamaephytes with anemochorous or barochorous dispersal mode.
In the Cévennes, the C. sativa cultivated groves (group 1) were characterized
by therophytes with anemochorous dispersal mode and geophytes (Figure 5).
The C. sativa abandoned groves and the young coppice stands (group 2) were
characterized by heliophillous hemicryptophytes and chamaephytes. The
C. sativa medium and old coppice stands (group 3) were composed more
particularly by phanerophytes with zoochorous dispersal mode.
In the case of C. sativa coppice stands on Etna volcano in Italy, Monte
Crisimo (group 1) were more particularly characterized by therophytes and
chamaephytes (Figure 6), Triciala (group 2) by hemicryptophytes with anem-
ochorous dispersal mode, Piano Lepre (group 3) by geophytes with baroch-
orous dispersal mode, and Balilla and Fornazzo (group 4) by shade tolerant
phanerophytes with zoochorous dispersal mode.

Discussion and concluding remarks

A main trend emerging from our species richness data was higher species
richness in the chestnut cultivated groves than in coppice stands; both in the

[75]
1136

1,5
Etna volcano - Group 2 Axis 2
C. sativa coppice stands (inertia 29%)
Etna volcano - Group 1
Trisciala C. sativa coppice stands
Acli, Brsy, Crle, Himu, Monte Crisimo 1
Heliophillous
Rane, Sivu, Trpu Arth, Avba, Homu, Lasp, 30-50 cm Barochorous
Sete, Stme, Vidi, Vite Geophyte
Anemochorous 0,5
Chamaephyte Etna volcano - Group 3
Hemicryptophyte C. sativa coppice stands
Therophyte Piano Lepre
1-30 cm Hydrochorous Door, Epmi, Lagr, Lave, Leco, Muco, Ptaq
0
-0,8 -0,6 -0,4 -0,2 Autochorous 0 0,2 0,4 0,6 0,8
Axis 1
(inertia 43%)
-0,5

> 50 cm Etna volcano - Group 4

Shade tolerant Phanerophyte C. sativa coppice stands
-1 Zoochorous Balilla & Fornazzo
Casa, Hehe, Prsp, Quda,
Ruid, Ruul

-1,5

Figure 6. Ordination in the plane of the two axes of life traits after a canonical correspondence
analysis from a matrix composed by the 40 plant species observed on the line point quadrat of the
25 plots on Etna volcano in Italy and a matrix composed by the same plant species and their life
traits. Groups were identified by an hierarchical ascending classification. Plots, and some plant
species associated to each group were indicated on the figure. Codes of plant species are indicated in
Appendix 1. The total variation explained by CCA is 49%.

Honfrı́a forest in Spain and in the Cévennes in France. However, species

richness in the cultivated groves of the Honfrı́a forest (53 ± 4 species) was
significantly higher than in the cultivated groves of the Cévennes (38 ± 4
species) (p < 0.01). The strawberry culture in some years under chestnut
groves in the Honfrı́a forest can explain this difference. Indeed, the high bio-
diversity among plants was always related to perturbations (pruning, grazing,
fire, etc.), and often observed in Mediterranean areas (Romane et al. 1992).
The species diversity decrease observed along the successional gradient in the
Cévennes appears to be a general trend in Mediterranean Basin (Tatoni and
Roche 1994; Debussche et al. 1996). On Etna volcano, tradition is coppice
management and not grove, and species diversity in coppice stands was lowest
than in the chestnut cultivated groves of the Honfrı́a forest and the Cévennes.
Nevertheless, species diversity in Trisciala coppice stand was not significantly
different from chestnut abandoned groves in the Cévennes (p>0.05). In Tri-
sciala, stumps have regular and large spacing and thus light, which is recog-
nized as a factor linked positively with species richness (Grime and Jarvis 1975;
Gilliam et al. 1995; Yorks and Dabydeen 1999), is available in the understorey
and favour growth. Like in the study of Rubio et al. (1999) in Extremadure
(Central Spain), or Kitazawa and Ohsawa (2002) in Chiba (Central Japan), the
difference according to management type was well observed.
The species composition differences among management type showed us that
generally small heliophillous therophytes characterized C. sativa cultivated
groves. Low intensity disturbance can explain the persistence of annual species
in cultivated groves (Lavorel 1999). Hemicryptophytes with anemochorous

[76]
 1137

dispersal mode and chamaephytes characterized C. sativa abandoned groves,

mixed C. sativa–Q. pyrenaica coppice stands, Q. pyrenaica coppice stands, and
young C.sativa coppice stands. Phanerophytes with zoochorous dispersal mode
characterized more particularly medium and old coppice stands (coppice
stands that differ by the shoot age). This pattern coincides with the general
trend described in Southern France; annual plants are substituted by perennial
grasses and shrubs with canopy closure (Houssard et al. 1980; Escarré et al.
1983; Tatoni and Roche 1994).
Perturbations were necessary to maintain a quite high level of species
diversity. In contrast, the abandonment of chestnut stands, for decades or even
centuries, will turn into closed and homogeneous vegetation with decreasing
plant diversity. One solution could be to maintain a landscape mosaic con-
sisting of diverse chestnut stands modified by human activities (chestnut
groves, abandoned chestnut groves and chestnut coppice stands) (Gondard
et al. 2001; Rubio and Escudero 2003). This could enhance regional plant
diversity. However, in our study we recorded only common species, thus if rare
species have been observed, the estimation of biodiversity would be review, and
quality aspect take into account. Moreover, due to some of the unsatisfactory
aspects of the experimental design (replication number), our study only indi-
cates, but does not validate, several possible management techniques, of which
remain to be tested further.

Acknowledgements

We thank the European Union (MANCHEST contracts, DG XII). We also

warmly thank Maria Failla, Giuseppe Siracusa, Antonino La Mantia, Mirella
Clausi, Sergio Salazar Iglesias, Alvaro Peix Geldart, Jesús Hernández, Zuheir
Shater, Alain Renaux, Michel Grandjanny, Marie Maistre, Maurice Rapp, and
François Jardon for their participation in collecting data in the chestnut stands.

Appendix 1. Functional traits (life form, dispersal mode, plant height and light tolerance) of plant
species observed along the point quadrat line and used in the Canonical Correspondence Analysis
according to available data: Molinier and Müller 1938; Pignatti 1982; van der Pijl 1982; Bonnier
1990; De Bolos et al. 1993. Th – therophyte, G – geophyte, H – hemicryptophyte, Ch – chamae-
phyte, Ph – phanerophyte.

Code Species Life form Dispersal mode Height Light tolerance

Acli Achillea ligustica H Anemochorous 30–50 cm shade tolerant

Acmi Achillea millefolium H Anemochorous 30–50 cm shade tolerant
Alpe Alliaria petiolata H Anemochorous 30–50 cm shade tolerant
Anod Anthoxantum odoratum H Anemochorous 30–50 cm heliophillous
Anpr Anthemis pratensis Th Anemochorous 1–30 cm heliophillous
Armo Arenaria montana H Anemochorous 1–30 cm heliophillous

[77]
1138

Appendix 1. Continued.

Code Species Life form Dispersal mode Height Light tolerance

Arpa Aristolochia pallida G Barochorous 30–50 cm shade tolerant
Arth Asplenium trichomanes Th Anemochorous 1–30 cm shade tolerant
Asal Asphodelus albus G Barochorous 30–50 cm heliophillous
Astr Asplenium trichomanes H Zoochorous 1–30 cm shade tolerant
Avba Avena barbata Th Anemochorous >50 cm heliophillous
Avsu Avena sativa Th Zoochorous >50 cm heliophillous
Bepe Betula pendula Ph Anemochorous >50 cm shade tolerant
Brsy Brachypodium sylvaticum H Anemochorous 30–50 cm shade tolerant
Brma Bromus maximus Th Anemochorous 30–50 cm heliophillous
Brho Bromus hordeaceus Th Zoochorous 1–30 cm heliophillous
Cadi Carex distachia H Barochorous 1–30 cm shade tolerant
Caof Calamintha officinalis H Zoochorous 30–50 cm heliophillous
Capa Campanula patula H Anemochorous 30–50 cm heliophillous
Casa Castanea sativa Ph Zoochorous >50 cm shade tolerant
Cavu Calluna vulgaris Ch Anemochorous 30–50 cm shade tolerant
Cegl Cerastium glomeratum Th Anemochorous 1–10 cm heliophillous
Celo Cephalanthera longifolia G Anemochorous 1–30 cm shade tolerant
Chju Chondrilla juncea H Anemochorous >50 cm shade tolerant
Coma Conopodium majus G Barochorous >50 cm shade tolerant
Clvu Clinopodium vulgare H Hydrochorous 30–50 cm heliophillous
Coar Convolvulus arvensis H Barochorous 30–50 cm heliophillous
Coav Corylus avellana Ph Zoochorous >50 cm shade tolerant
Coli Corrigiola littoralis H Autochorous 1–30 cm shade tolerant
Cran Crucianella angustifolia Th Anemochorous 1–30 cm heliophillous
Crle Crepis leontodontoides H Anemochorous 30–50 cm heliophillous
Crvi Crepis virens Th Anemochorous 30–50 cm heliophillous
Cyec Cynosurus echinatus Th Anemochorous 1–30 cm heliophillous
Cysc Cytisus scoparius Ph Autochorous >50 cm shade tolerant
Daca Daucus carota H Anemochorous 30–50 cm heliophillous
Dagl Dactylis glomerata H Anemochorous >50 cm heliophillous
Deme Deschampsia media H Zoochorous 30–50 cm shade tolerant
Door Doronicum orientale G Anemochorous 30–50 cm shade tolerant
Epla Epilobium lanceolatum H Anemochorous 30–50 cm shade tolerant
Epmi Epipactis microphylla G Anemochorous 1–30 cm shade tolerant
Erar Erica arborea Ch Barochorous >50 cm shade tolerant
Erci Erica cinerea Ch Barochorous 30–50 cm shade tolerant
Feov Festuca ovina H Anemochorous 30–50 cm heliophillous
Feru Festuca rubra H Anemochorous 30–50 cm shade tolerant
Gaap Galium aparine Th Zoochorous 1–30 cm heliophillous
Gamo Galium mollugo H Barochorous 30–50 cm shade tolerant
Gasa Galium saccharatum Th Zoochorous 1–30 cm shade tolerant
Gefl Genista florida Ch Autochorous 1–30 cm shade tolerant
Gehi Genista hispanica Ch Autochorous 1–30 cm shade tolerant
Gepi Genista pilosa Ch Autochorous 1–10 cm shade tolerant
Gero Geranium robertianum Th Autochorous 1–30 cm shade tolerant
Gnlu Gnaphalium lutescens H Autochorous 1–30 cm heliophillous
Haha Halimium lasianthumsubsp Ch Zoochorous 30–50 cm shade tolerant
.alyssoides
Hehe Hedera helix Ph Zoochorous >50 cm shade tolerant
Hico Hippocrepis comosa H Autochorous 1–30 cm shade tolerant

[78]
 1139

Appendix 1. Continued.

Code Species Life form Dispersal mode Height Light tolerance

Hihi Hispidella hispanica H Anemochorous 1–30 cm heliophillous
Hima Hieracium maculatum H Anemochorous 30–50 cm shade tolerant
Himu Hieracium murorum H Anemochorous 30–50 cm shade tolerant
Hipi Hieracium pilosella H Anemochorous 1–30 cm shade tolerant
Hium Hieracium umbellatum H Anemochorous 30–50 cm heliophillous
Hola Holcus lanatus H Anemochorous >50 cm heliophillous
Homo Holcus mollis H Zoochorous 30–50 cm heliophillous
Homu Hordeum murinum Th Anemochorous 1–30 cm heliophillous
Hyra Hypochaeris radicata H Anemochorous 30–50 cm heliophillous
Ilaq Ilex aquifolium Ph Anemochorous >50 cm shade tolerant
Jamo Jasione montana H Anemochorous 1–30 cm shade tolerant
Laan Lathyrus angulatus Th Barochorous 30–50 cm heliophillous
Lagr Lathyrus grandiflorus G Autochorous >50 cm heliophillous
Lasp Lathyrus sphaericus Th Autochorous 1–30 cm heliophillous
Lave Lathyrus venetus G Barochorous 30–50 cm heliophillous
Leco Leopoldia comosa G Anemochorous 1–30 cm heliophillous
Lihe Linaria heterophylla H Anemochorous 1–30 cm heliophillous
Litr Linaria triornitophora H Autochorous 30–50 cm shade tolerant
Loco Lotus corniculatus H Autochorous 1–30 cm heliophillous
Loet Lonicera etrusca Ph Zoochorous 30–50 cm shade tolerant
Luca Luzula campestris H Zoochorous 1–30 cm heliophillous
Lufo Luzula forsteri H Anemochorous 1–30 cm shade tolerant
Lusi Luzula sieberi H Zoochorous 1–30 cm shade tolerant
Meme Melittis melissophyllum H Zoochorous 30–50 cm shade tolerant
Meun Melica uniflora H Anemochorous 30–50 cm shade tolerant
Muco Muscari commutatum G Anemochorous 1–30 cm heliophillous
Orco Ornithopus compressus Th Zoochorous 30–50 cm heliphillous
Orpe Ornithopus perpusillus Th Zoochorous 1–30 cm shade tolerant
Pehi Petrorhagia hispanica H Autochorous 1–30 cm heliophillous
Pimu Piptaterum multiflorum H Anemochorous >50 cm heliophillous
Plla Plantago lanceolata H Anemochorous 1–30 cm heliophillous
Pobu Poa bulbosa H Barochorous 1–30 cm heliophillous
Pone Poa nemoralis H Anemochorous 30–50 cm shade tolerant
Posy Poa trivialis subsp. sylvicola H Anemochorous 30–50 cm heliophillous
Prav Prunus avium Ph Zoochorous >50 cm shade tolerant
Prgr Prunella grandiflora H Barochorous 1–30 cm shade tolerant
Prsp Prunus spinosa Ph Zoochorous >50 cm shade tolerant
Prvu Prunella vulgaris H Hydrochorous 1–30 cm shade tolerant
Ptaq Pteridium aquilinum G Anemochorous >50 cm shade tolerant
Quda Quercus dalechampii Ph Zoochorous >50 cm shade tolerant
Qupu Quercus pubescens Ph Zoochorous >50 cm shade tolerant
Qupy Quercus pyrenaica Ph Zoochorous >50 cm shade tolerant
Rabu Ranunculus bulbosus H Barochorous 30–50 cm heliophillous
Rane Ranunculus neapolitanus H Anemochorous 1–30 cm shade tolerant
Rops Robinia pseudo-acacia Ph Autochorous >50 cm heliophillous
Ruac Rumex acetosella G Anemochorous >50 cm heliophillous
Ruid Rubus idaeus Ph Zoochorous >50 cm shade tolerant
Rupe Rubia peregrina H zoochorous >50 cm shade tolerant
var . longifolia

[79]
1140

Appendix 1. Continued.

Code Species Life form Dispersal mode Height Light tolerance

Ruul Rubus ulmifolius Ph zoochorous >50 cm shade tolerant
Scan Scleranthus annuus Th zoochorous 1–30 cm heliophillous
Sete Sedum tenuifolium Ch anemochorous 1–30 cm heliophillous
Siin Silene inflata H anemochorous 30–50 cm shade tolerant
Sivu Silene vulgaris subsp. H anemochorous 30–50 cm shade tolerant
angustifolia
Sovi Solidago virgaurea H anemochorous 30–50 cm shade tolerant
Stme Stellaria media Th anemochorous 1–30 cm shade tolerant
Tesc Teucrium scorodonia G barochorous 30–50 cm shade tolerant
Thga Thapsia garganica H anemochorous >50 cm heliophillous
Toar Torilis arvensis Th zoochorous 30–50 cm shade tolerant
Tran Trifolium angustifolium Th anemochorous 1–30 cm heliophillous
Trar Trifolium arvense Th anemochorous 1–30 cm heliophillous
Trca Trifolium campestre Th anemochorous 1–30 cm heliophillous
Trpr Trifolium pratense H anemochorous 1–30 cm heliophillous
Trpu Trifolium pratense subsp. H anemochorous 1–30 cm heliophillous
semi-purpureum
Trre Trifolium repens H anemochorous 1–30 cm shade tolerant
Veof Veronica officinalis H hydrochorous 1–30 cm shade tolerant
Vidi Vicia disperma Th anemochorous 30–50 cm heliophillous
Vilu Vicia lutea Th autochorous 30–50 cm heliophillous
Vips Vicia pseudocracca Th barochorous 30–50 cm heliophillous
Visa Vicia sativa H autochorous 30–50 cm heliophillous
Vite Vicia tenuifolia Th anemochorous 30–50 cm heliophillous
Vubr Vulpia bromoides Th zoochorous 1–30 cm heliophillous

References

Arnaud M.T. and Bouchet M.A. 1995. L’aire écologique du châtaignier (Castanea sativa Mill.) en
Cévennes. Ecologie 26(1): 33–40.
Arsenault A. and Bradfield G.E. 1995. Structural-compositional variation in three age-classes of
temperate rainforests in southern British Columbia. Can. J. Bot. 73: 54–64.
Bonnier G. 1990. La Grande Flore. France, Suisse, Belgique et pays voisins, Belin, Paris.
Brakenhielm S. and Lui Q. 1998. Long-term effects of clear-felling on vegetation dynamics and
species diversity in boreal pine forest. Biol. Conserv. 7: 207–220.
DeBolòs O., Vigo J., Masalles R.M. and Ninot J.M. 1993. Flora Manual dels Paı̈sos Catalans.
Portic S.A. Edition, Barcelone.
Debussche M., Escarré J., Lepart J., Houssard C. and Lavorel S. 1996. Changes in Mediterranean
plant succession: old-fields revisited. J. Veg. Sci. 7: 519–526.
Dı́az S. and Cabido M. 1997. Plant functional types and ecosystem function in relation to global
change. J. Veg. Sci. 8: 463–474.
Dı́az S., McIntyre S., Lavorel S. and Pausas J. 2002. Does hairiness matter in Harare? – Global
comparisons of plant trait responses to disturbance. New Phytol. 154: 7–9.
Escarré J., Houssard C., Debussche M. and Lepart J. 1983. Evolution de la végétation et du sol
après abandon cultural en région méditerranéenne: étude de successions dans les garrigues du
Montpellierais (France). Acta Oecol., Oecol. Plant. 4: 221–239.
Falissard B. 1998. Comprendre et Utiliser les Statistiques dans les Sciences de la vie. Collection
Evaluation et Statistique, Masson, Paris.

[80]
 1141

Gilliam F.S., Turrill N.L. and Bethadams M. 1995. Herbaceous-layer and overstory species in
clear-cut and mature central Appalachian hardwood forests. Ecol. Appl. 5: 947–955.
Gondard H. and Deconchat M. 2003. Effects of soil surface disturbances after logging on plant
species diversity. Ann. Forest Sci. 60: 725–732.
Gondard H., Romane F., Grandjanny M., Junqing L. and Aronson J. 2001. Plant species diversity
changes in abandoned chestnut (Castanea sativa) groves in southern France. Biodivers. Conserv.
10: 189–207.
Gondard H., Jauffret S., Aronson J. and Lavorel S. 2003. Plant functional types: a promising tool
for management and restoration of degraded lands. Appl. Veg. Sci. 6: 223–224.
Gounot M. 1969. Méthodes D’étude Quantitative de la Vègétation. Masson, Paris.
Greenacre M.J. 1984. Theory and Applications of Correspondence Analysis. Academic Press,
London.
Grime J.P. and Jarvis B.C. 1975. Shade avoidance and shade tolerance in flowering plants. II.
Effects of light on the germination of species of contrasted ecology. In: Evans G.C., Bainbridge
R. and Rackham O. (eds), Light as an Ecological Factor II. Blackwell Scientific Publications.
Oxford, pp. 525–532.
Hadar L., Noy-Meir I. and Perevolotsky A. 1999. The effect of shrub clearing and grazing on the
composition of a Mediterranean plant community: functional groups versus species. J. Veg. Sci.
10: 673–682.
Houssard C., Escarré J. and Romane F. 1980. Development of species diversity in some Medi-
terranean plant communities. Vegetatio 43: 59–72.
Host G.E. and Pregitzer K.S. 1991. Ecological species groups for upland forest ecosystems of
northwestern Lower Michigan. Forest Ecol. Manag. 43: 87–102.
Kitazawa T. and Ohsawa M. 2002. Patterns of species diversity in rural herbaceous communities
under different management regimes, Chiba, central Japan. Biodivers. Conserv. 104: 239–249.
Lavorel S. 1999. Ecological diversity and resilience of Mediterranean vegetation to disturbance.
Diversity Distribut. 5: 3–13.
Lavorel S. and Cramer W. 1999. Functional response of vegetation to land use and disturbance.
J. Veg. Sci. 10: 604–732.
Lavorel S., McIntyre S. and Grigulis K. 1999. Plant response to disturbance in a Mediterranean
grassland: How many functional groups? J. Veg. Sci. 10: 661–672.
Lavorel S., McIntyre S., Landsberg J. and Forbes T.D.A. 1997. Plant functional classification:
from general groups to specific groups based on response to disturbance. Tree 12(12): 474–478.
Magurran A.E. 1988. Ecological Diversity and its Measurements. Croom Helm, London.
McIntyre S. and Lavorel S. 2001. Livestock grazing in sub-tropical pastures: steps in the analysis of
attribute response and plant functional types. J. Ecol. 89: 209–226.
McIntyre S., Lavorel S., Landsberg J. and Forbes T.D.A. 1999. Disturbance response in vegeta-
tion-towards a global perspective on functional traits. J. Veg. Sci. 10: 621–630.
McIntyre S., Lavorel S. and Tremont R.M. 1995. Plant life history attributes: their relationships to
disturbance response in herbaceous vegetation. J. Ecol. 83: 31–44.
Mitchell R.J., Marrs R.H. and Auld M.H.D. 1998. A comparison study of the seedbanks of
heathland and succession habitats in Dorset, Southern England. J. Ecol. 86: 588–596.
Mitchell R.J., Marrs R.H., Le Duc M.G. and Auld M.H.D. 1997. A study of succession on lowland
heaths in Dorset, Southern England: changes in vegetation and soil properties. J. Appl. Ecol. 6:
1426–1444.
Molinier R. and Müller P. 1938. La Dissémination des Espèces Végétales. Lesot A. (ed.), Paris.
Pielou R.H. 1975. Ecological Diversity. A Wiley-Interscience Publication, New York.
Pignatti S. 1982. Flora d’Italia. Edagricole (ed.), Bologne, 3 volumes, p. 2302.
Pillar V.D. 1999. On the identification of optimal plant functional types. J. Veg. Sci. 10(5): 631–640.
Pregitzer K.S. and Barnes B.V. 1982. The use of ground flora to indicate edaphic factors in upland
ecosystems of the McCormick experimental forest, Upper Michigan. Can. J. Forest Res. 12: 661–
672.

[81]
1142

Romane F., Bacilieri R., Bran D. and Bouchet M.A. 1992. Natural degenerate Mediterranean
forests: Which future? The examples of the holm oak (Quercus ilex.) and chestnut (Castanea
sativa Mill.) coppice stands. In: Teller A., Mathy P. and Jeffers J.N.R. (eds), Responses of Forest
Ecosystems to Environmental Changes. Elsevier Applied Science, London and New York,
pp. 374–380.
Roux M. 1985. Algorithmes de Classification. Masson, Paris.
Rubio A. and Escudero A. 2003. Clear-cut effects on chestnut forest soils under stressful condi-
tions: lengthening of time-rotation. Forest Ecol. Manag., 183 195–204.
Rubio A., Gavilán R. and Escudero A. 1999. Are soil characteristics and understorey composition
controlled by forest management? Forest Ecol. Manag. 113: 191–200.
Strong W.L., Bluth D.J., LaRoi G.H. and Corns I.G.W. 1991. Forest understorey plants as pre-
dictors of lodgepole pine and white spruce site quality in west-central Alberta. Can. J. Forest
Res. 21: 1675–1683.
Tatoni T. and Roche P. 1994. Comparison of old-field and forest revegetation dynamics in
Provence. J. Veg. Sci. 5: 295–302.
terBraak C.J.F. 1987. The analysis of vegetation-environment relationships by canonical corre-
spondence analysis. Vegetatio 69: 69–77.
Tutin T.G., Heywood V.H., Burges N.A., Moore D.M., Valentine D.H., Walters S.M. and Webb
D.A. 1964–1980. Flora Europea. Cambrige University Press.
van der Pijl L. 1982. Principles of Dispersal in Higher Plants. Springer, Berlin, Heidelberg and New
York.
Yorks T.E. and Dabydeen S. 1999. Seasonal and successional understory vascular plant diversity in
second growth hardwood clearcuts of western Maryland, USA. Forest Ecol. Manag. 119: 217–
230.

[82]
Biodiversity and Conservation (2006) 15:1143–1157
Springer 2006
DOI 10.1007/s10531-004-3105-6

-1

Spatial diversity of dry savanna woodlands

Assessing the spatial diversity of a dry savanna woodland
stand in northern Namibia using neighbourhood-based
measures

FRIEDRICH PATRICK GRAZ

1
Department of Land Management Conservation, Polytechnic of Namibia, P/Bag 13388, Windhoek,
Namibia; (e-mail: [email protected]; phone: +264-061-207-2215; fax: +264-061-207-
2123)

Received 8 March 2004; accepted in revised form 3 August 2004

Key words: Ecology, Mingling index, Namibia, Spatial diversity, Spatial structure, Uniform angle
index, Woodland savanna

Abstract. The dry woodland savannas of Namibia are of significant socio-economic importance.
The paper tests the suitability of a number of diversity indicators developed for species poor
systems in Europe in the woodland context. The indicators that were tested included the species
specific mingling index, MSp, the measure of surround and the uniform angle index. The simple
application of the methods permit relatively unschooled crews to conduct an enumeration in the
field.The results show that the indicators do not only display current diversity status, but also
reflect the ecological context of the individual species.

Introduction and background

The dry woodland savannas of northern Namibia are of significant socio-

economic importance to many rural communities, providing a variety of tim-
ber and non-timber products. The woodland resources that are used range
from building material and wood fuel to food, medicine and grazing (NFSP
1996). The quantities of the different products that are extracted are consid-
erable. Ollikainen (1992) estimated for example, that firewood alone amounted
to a total of 1.5 million cubic metres of wood during 1992. No quantity or
value estimates of non-wood products are available for the Namibian dry
woodland savannas, although these may be considerable.
The various woodland products differ in their importance to the various
communities, and at different times. Although exact quantities were not indi-
cated, Lee (1973) reported, for instance, that the intake of Schinziophyton
rautanenii nuts could comprise up to 90% of the total food intake of some San
communities. While this percentage will have changed in the meantime,
Büschel (1999) indicated that the nuts still represent the staple diet of nomadic
and sedentary San groups. The importance of the nuts increases particularly
when agricultural crops are insufficient to meet normal requirements.

[83]
1144

Other communities in the Kavango region of Namibia depend almost en-

tirely on the production and sale of carvings for the tourism industry, although
no studies seem to have been published in this regard. Some carvings are also
produced in the Caprivi region, but the local population does not appear to
depend as much on this form of income although no estimates are available.
The emphasis on different species for the carving industry is also shifting. In
the early 1990s the industry in the Okavango region made use of Pterocarpus
angolensis almost exclusively. Now, however, species like Guibourtia coleo-
sperma, Baikiaea plurijuga and even S. rautanenii (despite its light weight) are
being utilized extensively. This is primarily due to the overexploitation of
P. angolensis.
Further woodland species, such as Burkea africana or Terminalia sericea
serve mainly for poles or as firewood, although they support a caterpillar that
also represents an important source of food (Leger 1997).

Spatial diversity and woodland structure

The word ‘‘structure’’ generally considers the composition of a population of

trees in terms of specific characteristics. These may include tree age, size,
species or sex (in the case of dioceous trees). Spatial structure, on the other
hand looks at the arrangement of such characteristics in space. Spatial diversity
refers to the arrangement of the characteristics in relation to each-other or in
relation to a particular point on the ground.
The woodland savanna in northern Namibia is supported by coarse Aeolian
sands with poor water holding capacity and nutrient status. The trees that
occur here need to cope with highly variable precipitation and high evapora-
tion rates. Frequent fires and exploitation further affect the environment. Ta-
ken in combination, trees and especially their seedling have to cope with a wide
variety of conditions over a very short period of time and have adapted
accordingly.
A number of the woodland species are frequently, though not exclusively,
found in almost monospecific stands. This may be due to regeneration
requirements, as in the case of P. angolensis (Graz 1996), the ability to com-
pete, especially for water, as in the case of B. plurijuga (Mitlöhner 1997) or
superior fire tolerance as in the case of B. africana (Rutherford 1981).
The monospecificity of stands of S. rautanenii and T. sericea have not been
investigated. T. sericiea, however, is a pioneer that may quickly colonize open
areas where it may actually form thickets (Shackleton 2001).
Büschel (1999) reported on the other hand that stands dominated by
S. rautanenii were comprised of trees of different sizes and species in the
Okavango region of Namibia. Similarly, Mitlöhner (1997) also described
stands of mixed species, comprising of P. angolensis, B. africana and B. plu-
rijuga, while observations near the study site also showed mixed stands
(unpublished data).

[84]
 1145

In addition to being almost monospecific, trees within many stands often

seem to be of similar size although not necessarily of similar age. Childes (1984)
reported, for instance, that B. plurijuga stands were of variable age despite the
equal size of the trees. Plants remain small for a number of years until envi-
ronmental conditions are suitable for further development. This is probably
also the case for B. africana and S. rautanenii, although nothing seems to have
been documented.
The restriction of growth described by Childes for B. plurijuga is similar to
the suffrutex development stage of P. angolensis reported by Vermeulen
(1990). During this period seedlings from a number of years may accumulate
in this developmental stage and develop together to the sapling stage when
environmental conditions permit. In such cases the above ground parts are
not of the same age as the roots. It is unclear if the differences in the ages of
the roots will be reflected in the survival rate of the above ground parts of the
trees.

It is also uncertain whether or not whole stands of any of the above species
will die off and be replaced by others at a different location, or whether the
existing regeneration is sufficient to replace those trees that have died.

The data pertaining to the structure of stands in northern Namibia currently

available is superficial, despite its significant importance for management.
Spatial diversity, or a lack of spatial diversity, has important implications.
Consider for instance the effect of exploitation on an even sized, monospecific
stand; selection based on a minimum diameter may result in a local clear-felling
(Von Breitenbach 1968; Graz 1996). The resulting vegetation structure would
be increasingly prone to fire that may cause further vegetation change, as well
as subsequent erosion and nutrient loss (see Graz 1996).
Causes of mortality are not necessarily only of human origin, however. The
different sizes of a number of species have, for example, their own degree of fire
tolerance. This means that trees up to a particular size class may be removed
from a stand by a sufficiently intense fire. Wilson and Witkowski (2003) found
that the bark-thickness of B. africana increases with tree circumference between
0 and 400 mm. The thickness of the bark is the primary protector against the
effect of fire on the cambium.
Fire tolerance may be overcome if the bark of trees is breached by animals
(Yeaton 1988) or growth stresses (Graz 2003).
Studies relating to spatial aspects have in the past concentrated on the dis-
persion of plants using measures such as the nearest neighbour of Clark and
Evans (1954) or point to plant distances after Pielou (1977). More recently the
uniform angle index (UAI) (Gadow 1999; Staupendahl 2001; Gadow et al.
2003) has been implemented to describe complex forest structures. The
aggregation of tree attributes have only been addressed more recently by other
measures, such as the ‘‘measure of surround’’ (Hui et al. 1998) or the spatial
‘‘mingling’’ (Gadow 1999).

[85]
1146

The mingling measure is used to quantify the degree of interspersion or

mingling of tree characteristics, as illustrated in Figure 1. Trees that are sur-
rounded by others of similar characteristic are aggregated in terms of the
characteristic, implying a lower degree of mingling of this characteristic. On the
other hand, trees surrounded by others of dissimilar characteristic imply a
higher degree of mingling. Mingling should not only be considered in terms of
categorical data, such as species or sex, or whether a tree is alive or dead, but
should be expanded to include any measure with which a tree might be
described, including height or diameter.
Albert and Gadow (1998) reported on the use of these neighbourhood-based
measures to assess the effect of selective thinning on the diversity of a beech stand
in Germany. The authors had found the measures to be sensitive to small-scale
differences and changes of woodland structure, and were able to provide more
intuitively acceptable results than the segregation index of Pielou (1977, p. 227 ff).
This study aims to achieve two main objectives. The first objective is to
assess the applicability of indicators that were developed and assessed in
Europe to the Southern African context where little or no basic stand infor-
mation is available for non-plantation areas. In addition, the study intends to
generate information that will promote the understanding of the ecology of
Namibia’s woodland resources.

Description of the study area

The woodland area that was enumerated covers approximately 70 ha and is

situated between 1930¢ E, 1915¢ S and 1945¢ E, 1930¢ S near the Kanovlei
Forestry Research Station in the western Tsumkwe district of the
Otjozondjupa region, northeastern Namibia.
The area is dominated by linear fossil dunes or sandy plains on calcareous
deposition, similar to those in the adjoining Kavango region described by Graz
(1999). The soils are Kalahari sands, classified as unconsolidated aeolian
material by Coetzee (2001), with very poor water holding capacity and nutrient
status, and subsequently a very low potential for any agricultural development
(Department of Water Affairs 1991).

Figure 1. The mingling of black, grey and white ‘trees’ within two square stands (after Gadow,
1999).

[86]
 1147

The region is traversed by a system of omuramba (vegetated dry riverbed),

with the soils classified as unconsolidated fluvial material (Coetzee 2001). These
soils are shallower and have a heavier texture than the dunes (Department of
Water Affairs 1971).
Precipitation is mostly in the form of thunderstorms amounting to an
average rainfall of between 500 and 600 mm per year (Amakali 1992).
However, the distribution of precipitation is highly variable and prominently
positively skewed. Expected rainfall is therefore significantly lower than the
long-term averages. Rain generally falls in the period September to May,
with most rain occurring between December and March.
Average annual evaporation rates are between 2600 and 2800 mm (Crerar
and Church 1988) resulting in an overall moisture deficit.
de Pauw and Coetzee (1999) have determined an approximate growing
period of between 91 and 120 days, based on the relationship between available
moisture, the amount of evapotranspiration and the average air temperature.
Although the general vegetation is described as tree savanna and woodland
by Giess (1998), there is some significant variation in species and structural
composition. The Directorate of Forestry identifies a number of dissimilar
patches of forest or savanna (Chakanga 1995).
While the sandy planes and dunes are dominated by Burkea africana, various
species of Combretum, Pterocarpus angolensis, Schinziophyton rautanenii and
Terminalia sericea. Scattered patches of Baikiaea plurijuga also occur.
The lower lying omuramba vegetation is comprised primarily of Acacia
erioloba, Dichrostachys cinerea and Philenoptera nelsii.
Nuts from the S. rautanenii trees within the stand are harvested by local
communities to augment their food supply, and by the Directorate of Forestry
to obtain material for the National Tree Seed Centre and for ex situ conser-
vation of genetic material. Additionally the stand shows signs of periodic wood
harvesting of B. plurijuga stems, as well as for firewood. Dry season fires are
frequent (Graz 2003).

Material and methods

The interspersion of tree attributes

The original measure of mingling and its derivatives are based on the pro-
portion of trees with dissimilar characteristics to those of a selected sample
tree. The species mingling index Mi for a given sample tree, i, using n neigh-
bours is, for example, obtained through:
1X n
ð1Þ Mi ¼ mij ;
n j¼1

[87]
1148

where


1; if the tree is of another species;
mij ¼
0; if the tree is of the same species:
When four neighbours are used to determine Mi the index may obtain one of
five possible values:
0/4 none of the neighbours are of a different species,
1/4 one of the neighbours is of a different species,
2/4 two of the neighbours are of a different species,
3/4 three of the neighbours are of a different species, and
4/4 all of the neighbours are of a different species.
The arithmetic mean (MSp) of the Mi values that were obtained for a par-
ticular species sp provides a measure of the degree of interspersion of the
species in the area. MSp provides a value between 0 and 1.
Values close to 0 indicate that trees of the reference species sp occur in
groups therefore implying a low degree of mingling and high degree of
aggregation. High values of MSp, closer to 1, on the other hand, imply a high
degree of mingling, i.e. trees of the reference species do not occur together.
As is the case when examining the distribution of data around a mean value,
additional information may be extracted from the distribution of Mi values of
individual species.
When the proportion that a species contributes to a stand is known, as
assumed in the studies reported on by Lewandowski and Pommerening (1997)
and Hui et al. (1998) a theoretical distribution of Mi values may be calculated
based on the hypergeometric probability distribution. The distribution reflects
the number of expected Mi values that would be obtained if all trees were
interspersed randomly.
The hypergeometric distribution is used to determine the probability, P, that
a number of trees of a particular species may occur in a given sample of n trees
taken from a population of N trees containing k trees of the species of interest.
The probability that x trees in the sample will be of the species of interest is
then determined after Newmark (1997) as:
   
k N
k

x n
x
P¼   for x ¼ 0; 1; 2; . . . ; n;
N
n
which expands to:
ðN
kÞ
k
xðk
xÞ
 ðn
xÞðN
k
ðn
xÞÞ
P¼ N for x ¼ 0; 1; 2; . . . ; n:
nðN
nÞ

The resulting probability multiplied by the total number of samples that

were taken provides the expected number of Mi values for that species. The

[88]
 1149

observed and expected distributions of Mi values may then be compared with

the application of standard statistical methods to test for significance of
deviations from the theoretical (random) distribution.
Although no detailed data is available for any of the woodland areas in
Namibia, and the extent of the woodland areas hampers the collection of such
information, the sample size provided a suitable estimate of the species com-
position of the stand.
The simulation study reported on by Graz (2004) has shown that the min-
gling index is sensitive to the species composition of a stand. In a stand of trees
interspersed randomly, for example the aggregation of a species, 1
MSp,
approximates the proportion that a species Sp contributes to the stand. This
may be more intuitively understood if we consider each sample tree to provide
an estimate of the proportion that its species contributes to the stand. Values of
1
MSp which are greater than the proportion contribution therefore indicate
an overaggregation of the species, while lower values imply overdispersion
within the stand. This relationship provides an important base from which the
index may be interpreted.
This study investigated the interspersion of a number of tree characteristics.
In addition to the mingling of species described above, the interspersion of tree
dominance is quantified on the basis of diameter (TSp) and height (HSp) using
the ‘‘measure of surround’’ described by Hui et al. (1998), and which is applied
in a method analogous to that of the mingling index. More particularly:
1X n
ð2Þ Ti ¼ tij ;
n j¼1

where


1; if the tree; j; is thicker than the sample tree i;
tij ¼
0; otherwise:
The species specific mean interspersion of tree diameter, TSp, is then the
arithmetic mean of the values of Ti for that species.
Similarly, the interspersion of tree height, Hi, is obtained through:
1X n
ð3Þ Hi ¼ hij ;
n j¼1

where


1; if the tree; j; is higher than the sample tree i;
hij ¼
0; otherwise:
The species specific interspersion of tree height, HSp, is then again deter-
mined as the mean of the values of Hi for the species.
An equivalent measure was used to quantify the interspersion of dead trees
(DSp) by counting the number of dead neighbours for each sample tree.

[89]
1150

Uniform angle index

The UAI was initially described by Gadow et al. (1998) and later by Stau-
pendahl (2001) to provide a measure of the overall contagion of trees within a
forest stand.
The index is obtained by identifying the n nearest neighbours of a sample
tree. Starting with the closest neighbour and moving in a clockwise direction
around the sample tree the angle, aj, between two adjacent neighbours is
determined in relation to the sample tree. The number of angles smaller than,
or equal to, a given critical angle, a0, are then counted, i.e.
1X n
ð4Þ Wi ¼ wij ;
n j¼1

where


1; if aj  a0 ;
wij ¼
0; otherwise:
The critical angle (in degrees) is determined as:
360
ð5Þ a0 ¼ :
Number of neighbours
Four neighbours would therefore be evaluated in terms of a 90 critical
angle1. Since all of the indexes used to measure the interspersion of tree
characteristics were based on four trees, the same neighbours could be used for
the UAI.
A practical advantage of choosing a0 = 90 is that two adjoining sides of a
record book or clipboard may be used to determine whether or not an angle is
greater than or less than the critical angle.
Effectively, the index describes the spatial distribution around a particular
reference tree. If the species of the reference tree is noted we may obtain the
mean value for either for the whole population or for a particular species of
interest.
The mean value of the index is strongly correlated with the nearest neigh-
bour index of dispersion of Clark and Evans (1954) that has long been used in
ecological studies. Together with the number of trees in a stand, the UAI may
be used to estimate the distribution of distances between a tree and its neigh-
bours (Gadow et al. 2003).
This information is generally not available and comparison of observed
index values are compared to the simulation results of Gadow et al. (1998) are
used.

1
More recent studies have shown that this statement needs to be modified; a more suitable critical
angle is 72 (see Gadow et al. 2003).

[90]
 1151

Sampling

The extent of the stand was recorded in the field using a Garmin Venture GPS.
The track-log was stored for subsequent mapping. A regular sample grid of one
geographic second was then superimposed on the stand amounting to a sample
point approximately every 30 m at that latitude.
Sampling points were located using a standard GPS receiver. The accuracy
of autonomous GPS readings was considered adequate for the purpose of the
study. While a dense canopy reduces the reliability of a GPS reading within a
stand (Dominy and Duncan 2001), many of the trees in the area had already
shed their leaves and canopy interference was considered negligible after initial
comparison of signal strengths in wooded and in open areas.
Since the enumeration coincided with the war in Iraq it is uncertain whether
GPS readings were affected by selective availability on some days. It was felt,
however, that this was acceptable.
At each sample point the closest tree with a dbh of 5 cm or more was
identified to serve as reference tree. Although trees had, in a few cases, snapped
off below breast height, such trees were nevertheless sampled, since they play a
role in the interspersion of plants.
For each sample tree the four nearest neighbouring trees with a diameter of
greater than 5 cm were determined and compared with the reference tree in
terms of species, mortality, height and diameter, and the UAI was established.
Time was kept short by assigning two persons to each sampling team. While
the enumerator collected the measures, a navigator moved to find the next
sample point.
A total of 1121 sample points were assessed. The data was entered into a
spreadsheet and the indexes were calculated for each species using cross tables.

Results and discussion

The species specific indexes are summarized in Table 1. The table also shows a
surrogate species of ‘Dead’ created to record trees that were still standing but
had been burnt beyond a stage where they might be identified. Also, species of
the genus Combretum and Comiphora were lumped, as individual species could
not readily be identified. The row marked ‘overall’ provides the each index as
calculated over the entire data set.
The overall shows a contagion (Wi) greater than 0.6, here indicating a ten-
dency towards non-random (clumped) dispersion of trees (after Gadow et al.
1998). The dispersion around trees of the individual species does not seem to
diverge very much from the mean value of 0.665, if Philenoptera nelsii and
Securidaka longipedunculata are discounted because of their very low overall
occurrence. This is in line with general observations in the field.
The table shows that most of the species have a tendency to aggregate.

[91]
1152

To compare the proportion of a given species within the stand and the value
1
MSp consider Table 2. The table omits those species with very few
observations (less than 5% of the total). The final column in the table reflects
the parameter M proposed by Graz (2004) to determine the degree of inter-
spersion. The value of M is larger than 0 and less than or equal to 1. Values
close to 0 indicates a very low degree of mingling, and 1 indicates a more
random distribution of the species in the stand.
Table 2 shows that B. plurijuga has the highest degree of aggregation fol-
lowed by Terminalia sericea.
More T. sericea seedlings survive in open areas, i.e. away from conspecific
trees (Smith and Grant 1986) where it has the ability to form thickets
(Shakelton 2001). This was evident in the field. T. sericea would colonize gaps
in the canopy, thus causing the aggregation.
The dispersion of B. plurijuga is also shown in Figure 2. The figure shows
that the species occurs in a very limited area. The accompanying graph shows
the relative distribution of Mi values (bar) and the theoretical hypergeometric
distribution. The graph shows a clear difference between the two, due to the
clumping of the species, reflected by the low value of M (Table 2).
The cause of the aggregation of B. plurijuga is uncertain, since the trees had
few larger neighbours as evidenced by the low value of TSp in Table 1. It is
possible that the patch of B. plurijuga is a remnant of a larger stand that has
been subject to high degrees of mortality. This possibility stems from reports by
Von Breitenbach (1968) who suggested that the almost pure stands in the
Caprivi region developed towards mixed stands as a result of fire.
The possibility is corroborated by the high degree of mortality (DSp) asso-
ciated with the species (see Figure 3). The dead trees within the B. plurijuga

Table 1. Mean of the various indicators for each of the identified species.

Species of sample tree N P(Sp) WSp DSp MSp TSp HSp

Burkea africana 116 0.103 0.688 0.226 0.751 0.323 0.332

Baikiaea plurijuga 194 0.173 0.665 0.116 0.653 0.249 0.256
Combretum species 214 0.191 0.657 0.148 0.697 0.484 0.479
Comiphora species 36 0.032 0.639 0.201 0.875 0.382 0.556
Ochna pulchra 26 0.023 0.635 0.087 0.962 0.567 0.673
Philenoptera nelsii 16 0.014 0.750 0.141 0.820 0.563 0.625
Pterocarpus angolensis 178 0.159 0.647 0.163 0.813 0.198 0.218
Schinziophyton rautanenii 75 0.067 0.653 0.137 0.875 0.243 0.300
Securidaka longipedunculata 2 0.002 0.625 0.125 0.962 0.375 0.625
Strychnos pungens 26 0.023 0.683 0.163 0.893 0.462 0.548
Terminalia sericea 96 0.086 0.641 0.130 0.781 0.565 0.602
Unidentifiable dead tree 142 0.127 0.701 0.285
Overall 1121 1.0000 0.665 0.169 0.794 0.360 0.399
P(Sp) denotes the proportion that a species contributes to the stand as a whole. The species specific
indicators are: WSp = mean UAI, DSp = mean mortality, MSp = mean mingling, TSp = mean
diameter dominance, and HSp = mean height dominance. The overall values for each indicator
was calculated using the entire data set.

[92]
 1153

Table 2. Comparing the proportion P(Sp) that a species contributes to the population with
(1
MSp).
P ðpÞ
Species of sample tree N P(Sp) MSp 1
MSp M ¼ 1
M Sp

Baikiaea plurijuga 116 0.103 0.653 0.347 0.298

Burkea africana 194 0.173 0.751 0.249 0.696
Combretum species 214 0.191 0.697 0.303 0.631
Pterocarpus angolensis 178 0.159 0.813 0.188 0.883
Schinziophyton rautanenii 75 0.067 0.875 0.125 0.627
Terminalia sericea 96 0.086 0.781 0.219 0.391
Unidentifiable dead tree 142 0.127 0.285* 0.447
Total 1121
*Note that the value of DSp is used here (the mean proportion of dead neighbours), rather than the
mingling index.

patch, shown in figure 3, are generally large trees. This is not evident from the
indexes but supports the suggestion by Von Breitenbach cited above.
Actual tree mortality may be caused directly by repeated burning of the
stem, as well as changes in the osmotic potential of the top-soil that is caused
by the accumulation of ash in the upper soil layers (Mitlöhner pers. comm.)
In contrast to B. plurijuga, P. angolensis is interspersed almost randomly
according to Table 2 and in Figure 2. As is evident in the figure, the observed
distribution of Mi values (bars) follow the theoretical distribution much more
closely than those of B. plurijuga. It must be noted, that P. angolensis occurs
comparatively seldom within the B. plurijuga patch. This exclusion from the
patch is more pronounced for B. africana. The reason or cause for this is not
readily apparent. Outside this patch B. africana is more aggregated resulting in
the lower value of M.
The random distribution of P. angolensis is probably a reflection of the
regeneration requirements of the species. Vermeulen (1990) reports that
P. angolensis is especially sensitive to competition in the seedling and estab-
lishment phases. The species therefore often regenerates in areas that have been
cleared by human or other action. Other species would then establish them-
selves later.
The interspersion of trees of different size is reflected in the columns TSp
(diameter specific) and HSp (height specific) in Table 1. Preliminary simulation
results have shown that a random interspersion of tree sizes would result in an
overall average of TSp = 0.5 and HSp = 0.5. The table shows, therefore, that
size classes are not interspersed randomly.
P. angolensis, S. rautanenii and B. plurijuga need to be highlighted. The low
values of TSp and HSp for these species imply that few neighbouring trees are
larger than the reference tree. This is supported by general observations in the
field. The species therefore dominate in the area in which they occur. It also
reflects the regeneration requirements of P. angolensis noted previously, but
highlights the importance of further research into the demography of the other
two species.

[93]
1154

Figure 2. The dispersion of Baikiaea plurijuga, Pterocarpus angolensis and Burkea africana, within
the study area. High values of Mi are shown in large circles and vice versa. The graphs depict the
observed relative distribution of Mi values (bars), and the theoretical hypergeometric distribution
(lines) of the values that would indicate a completely random interspersion of the species.

Table 1 also shows a similarity between the values of TSp and HSp of the
individual species. Unpublished data shows a high degree of correlation be-
tween the dbh and height of B. africana (r2 = 0.8352), as well as for P. an-
golensis (r2 = 0.7317) for nearby stands.

[94]
 1155

Figure 3. The aggregation of dead trees within the stand. The degree of interspersion is reflected
by the size of the points, with a high degree of aggregation shown by larger points.

Differences between the two indexes are due to the number of species
found in the stand, and the differences in their respective diameter
height
relationships. A larger difference occurs for the Comiphora species, however,
reflecting the squat form of the trees; a relatively thick-trunked but short
tree.

Conclusions

In the past the applications of neighbourhood-based spatial measures were

supported by detailed knowledge of the stands that were assessed, as noted
above. This was not the case in this study, where only the extent of the stand
was known. However, despite their simple application the indexes are able to
provide information about the stands they describe, being able to reflect much
of what is currently known about the individual tree species and their eco-
logical circumstances.
The results have also highlighted gaps in our knowledge of the ecology of a
few of the important trees, such as Schinziophyton rautanenii, Baikiaea pluri-
juga, and Burkea africana, as well as the various Combretum species that occur
in the area. These include regeneration requirements and species succession,
and highlights the need for further investigation.
The application of the measures described here has shown that they are
easily applied in the field with relatively little training required, although the
field crews will have to be able to identify the different tree species. This is
particularly useful in view of the trend towards community based natural
resource management in Namibia, where community members will have to
assess their own resources. Since most of the rural community members are
able to identify different plants in their vernaculars, species identification
should not be a problem, despite sometimes limited literacy levels.

[95]
1156

Acknowledgements

I would like to thank my sister Ms. H. Riehmer and Ms. R. Haipinge as well as
the late Mr. H. Roth for their assistance with data collection in the field. My
sincere thanks also to the Directorate of Forestry, Namibia, for allowing me to
use the Kanovlei Forest Station as a base, and the Polytechnic of Namibia who
funded the field work. I would particularly like to thank Prof. K. von Gadow,
Institute of Forest Management, Univeristy of Goettingen for comments.

References

Albert M. and Gadow K.v. 1998. Assessing biodiversity with new neighborhood-based parameters.
In: Proceedings of the International Conference on Data Management and Modelling Using
Remote Sensing and GIS for Tropical Forest Land Inventory. Jakarta, Indonesia, Oct. 26–29,
1998.
Amakali M. 1992. Updated isohyetal rainfall map for Namibia. Unpublished Report. Department
of Water Affairs, Namibia.
Büschel D. 1999. A study of resource utilization: a case from Namibia, Mpungu constituency,
Kavango district, northern Namibia. CRIAA SA-DC Report, Windhoek, Namibia.
Chakanga M. 1995. Forest Cover Reconnaissance Mapping Project. Directorate of Forestry,
Republic of Namibia, Windhoek.
Childes S.L. 1984. The population dynamics of some woody species in the Kalahari sand vegetation
of the Hwange National Park. M.Sc. thesis, University of the Witwatersrand, South Africa.
Clark P.J. and Evans F.C. 1954. Distance to Nearest Neighbor as a Measure of Spatial Rela-
tionships in Populations. Ecology 35(4): 445–453.
Coetzee M.E. 2001. NAMSOTER, A SOTER database for Namibia. Agroecological Zoning
Programme, Ministry of Agriculture, Water and Rural Development, Namibia.
Crerar S.E. and Church J.T. 1988. Evaporation map for south West Africa/Namibia. Hydrological
Report No. 11/1/8/1/H1, Department of Water Affairs, Namibia.
Dominy N.J. and Duncan B. 2001. GPS and GIS methods in an African rain forest: applications to
tropical ecology and conservation. Conserv. Ecol. 5(2): 6. URL: http://www.consecol.org/vol5/
iss2/art6.
Department of Water Affairs. 1971. Consolidated report on reconnaissance surveys of soils of
northern and central south West Africa in terms of their potential for irrigation. Report TS/30/
71, Department of Water Affairs, Windhoek, Namibia.
Department of Water Affairs. 1991. Groundwater investigation in Kavango and Bushmanland,
Namibia. Unpublished Report. Department of Water Affairs, Windhoek, Namibia.
De Pauw E. and Coetzee M.E. 1999. Production of an agro-ecological zones map of Namibia (first
approximation), Part 2: Results. Agricola 10: 8–68.
Gadow K.v. 1999. Waldstruktur und Diversität. Alg. Forst-u. J.-Ztg. 170: 117–122.
Gadow K.v., Hui G.Y. and Albert M. 1998. Das Winkelmass – ein Strukturparameter zur Bes-
chreibung der Individualverteilung in Waldbeständen. Centralblatt das gesamte Forstwesen.
115(1): 1–10.
Gadow K.v., Hui G.Y., Chen B.W. and Albert M. 2003. Beziehungen zwischen Winkelmass und
Baumabständen. Forstw. Cbl. 122: 127–137.
Giess W. 1998. A preliminary vegetation map of Namibia (3rd revised edition). Dinteria 4: 1–112.
Graz F.P. 1996. Management of a Pterocarpus angolensis population under the influence of fire and
land use. M.Sc. thesis, University of Stellenbosch, South Africa.

[96]
 1157

Graz F.P. 1999. A preliminary terrain feature classification of the Okavango region, Namibia. S.
Afr. J. Survey. Geo-Inf. 1: 123–129.
Graz F.P. 2003. Fire damage to Schinziophyton rautanenii (Schinz) trees in North-Eastern Namibia.
Dinteria 28: 39–43.
Graz F.P. 2004. The behavior of the species mingling index MSp in relation to species dominance
and dispersal. Eur. J. Forest Res. 1: 87–92.
Hui G.Y., Albert M. and Gadow K.v. 1998. Das Umgebungsmaß als Parameter zur Nachbildung
von Bestandesstrukturen. Forstw. Cbl. 117: 258–266.
Lee R.B. 1973. Mogongo: the ethnography of a major wild food resource. Ecol. Food Nutr. 2: 307–
321.
Leger S. 1997. The Hidden Gift of Nature. DED, German Development Service, Berlin, Germany.
Lewandowski A. and Pommerening A. 1997. Zur Beschreibung der Waldstruktur – Erwartete und
beobaachtete Arten-Durchmischung. Forstw. Cbl. 116: 129–139.
Mitlöhner R. 1997. Pflanzeninterne Potentiale als Indikatoren für den tropischen Standort. Shaker
Verlag.
Newmark J. 1997. Statistics and Probability in Modern Life, 6th ed. Sauders College Publishing,
Fort Worth.
NFSP. 1996. Namibia Forestry Strategic Plan. Directorate of Forestry, Windhoek, Namibia.
Ollikainen T. 1992. Study on wood consumption in Namibia. Internal Report, Directorate of
Forestry, Windhoek.
Pielou E.C. 1977. Mathematical Ecology. John Wiley & Sons, New York, 384 pp.
Rutherford M.C. 1981. Survival, regeneration and leaf biomass changes in woody plants following
spring burns in Burkea africana – Ochna pulchra savanna. Bothalia 13: 531–552.
Shackleton C.M. 2001. Managing regrowth of an indigenous savanna tree species (Terminalia
sericea) for fuelwood: the influence of stump dimensions and post-harvest coppice pruning.
Biomass Bioenergy 20: 261–270.
Smith T.M. and Grant K. 1986. The role of competition in a Burkea africana – Terminalia sericea
savanna. Biotropica 18: 219–223.
Staupendahl K. 2001. Das flächenbezogene Winkelmass Wf – Ein Index zur quantitativen Bes-
chreibung der horizontalen Baumverteilung. In: Akca A. et al. (HRSG). Waldinventur,
Waldwachstum und Forstplanung – Moderne Technologien, Methoden und Verfahrensweisen.
Festschrift K. von Gadow. Zohab-Verlag, Göttingen. pp. 101–115.
Vermeulen W.J. 1990. A monograph on Pterocarpus angolensis. SARCCUS Standing Committee
for Forestry, Pretoria, South Africa.
Von Breitenbach F. 1968. Long-term plan of forestry development in eastern Caprivi Zipfel.
Unpublished Report. George, South Africa.
Wilson B.G. and Witkowski E.T.F. 2003. Seed banks bark thickness and damage in age and size
structure (1978–1999) of the African savanna tree Burkea africana. Plant Ecol. 167: 151–162.
Yeaton R.I. 1988. Porcupines, fires and the dynamics of the tree layer of the Burkea africana
savanna. J. Ecol. 76: 1017–1029.

[97]
Biodiversity and Conservation (2006) 15:1159–1177
Springer 2006
DOI 10.1007/s10531-004-3509-3

-1

Assessment of threat status and management

effectiveness in Kakamega Forest, Kenya

BÄRBEL BLEHER1,2,*, DANA USTER3 and THOMAS BERGSDORF 4

1
Institut für Zoologie, Abt. V- Ökologie, Johannes Gutenberg-Universität Mainz, Becherweg 13,
55099 Mainz, Germany; 2Department of Ornithology, National Museums of Kenya, Nairobi, Kenya;
3
Universität Bielefeld, Fakultät für Biologie, Abt. Ökologie, Universitätsstrasse 25, 33615 Bielefeld,
Germany; 4Zoologisches Forschungsinstitut & Museum Alexander Koenig, Adenauerallee 160, 53113
Bonn, Germany; *Author for correspondence (e-mail: [email protected]; phone: +49-(0)6131-
3926108; fax: +49-(0)6131-3923731)

Received 6 May 2004; accepted in revised form 6 September 2004

Key words: Conservation, Disturbance indicator, Forest degradation, Logging, Management

Abstract. To counteract an increasing biodiversity decline, parks and protected areas have been
established worldwide. However, many parks lack adequate management to address environmental
degradation. To improve management strategies simple tools are needed for an assessment of
human impact and management effectiveness of protected areas. This study quantifies the current
threats in the heavily fragmented and degraded tropical rainforest of Kakamega, western Kenya.
We recorded seven disturbance parameters at 22 sites in differently managed and protected areas of
Kakamega Forest. Our data indicate a high level of human impact throughout the forest with
illegal logging being most widespread. Furthermore, logging levels appear to reflect management
history and effectiveness. From 1933 to 1986, Kakamega Forest was under management by the
Forest Department and the number of trees logged more than 20 years ago was equally high at all
sites. Since 1986, management of Kakamega Forest has been under two different organizations, i.e.
Forest Department and Kenya Wildlife Service. The number of trees logged illegally in the last
20 years was significantly lower at sites managed by the Kenya Wildlife Service. Finally, logging
was lower within highly protected National and Nature Reserves as compared to high logging
within the less protected Forest Reserves. Reflecting management effectiveness as well as protection
status in Kakamega Forest, logging might therefore provide a valuable quantitative indicator for
human disturbance and thus an important tool for conservation managers. Logging might be a
valuable indicator for other protected areas, too, however, other human impact such as e.g. hunting
might also prove to be a potential indicator.

Introduction

Recent decades have seen a serious biodiversity decline due to habitat loss and
alteration especially of tropical forests leading to a profound species-extinction
crisis (Heywood 1995; Pimm et al. 1995; Whitmore 1997). Thus, much of
tropical biodiversity is unlikely to survive without effective protection (Pimm
et al. 1995; Myers et al. 2000). To counteract the anthropogenic impact and
conserve biodiversity and ecosystem processes parks and protected areas have
been established worldwide. Some studies demonstrate that parks can indeed
provide basic safeguard against land-clearing in the context of high land-use
pressure (Brunner et al. 2001). However, more often parks appear to lack

[99]
1160

adequate management to address a host of threats within their borders

(Brunner et al. 2001; Putz et al. 2001; Ervin 2003a, b). Protected areas face
increasing levels of environmental degradation with more than 70% of 201
parks surveyed across 16 tropical countries being affected by poaching,
encroachment and logging (van Schaik et al. 1997). Consequently, the
improvement of management strategies of protected areas is of top priority for
conservation practitioners.
To improve and optimise management strategies methods to assess the
threat status of protected areas and to measure management effectiveness of
conservation efforts have become a major environmental concern (Margoluis
and Salafsky 1998; Salafsky and Margoluis 1999; Hockings et al. 2000;
Salafsky et al. 2002; Ervin 2003c; Hockings 2003). These assessments are an
essential component of systematic conservation planning (Margules and
Pressey 2000); they can enable conservation managers and policymakers to
identify management strengths and weaknesses, reveal severity and distribution
of levels of human impact, respond to pervasive management problems, refine
their conservation strategies and reallocate budget expenditures (Brunner et al.
2001; Ervin 2003 a, c; Parrish et al. 2003). Therefore, the development of simple
tools to monitor and assess whether conservation succeeds for protected areas
are of great importance and require indicators that are measurable, scientifi-
cally sound, and comparable among protected areas over time, but also
practical and cost-effective (Margoluis and Salafsky 1998).
Traditionally, biological indicators have been used to assess the level of
human impact in protected areas and measure management success. Ideally,
they are supposed to serve as indicators for changes in the overall biodiversity
of a site (Noss 1990; Sparrow et al. 1994). However, relationships between
potential indicator species and total biodiversity as well as critical ecosystem
processes are not that well established (Lindenmayer et al. 2000). Few of these
methods using biologically based indicators are practical and cost-efficient,
especially for use in the developing countries as they require substantial effort
and resources beyond day-to-day project activities (Salafsky and Margoluis
1999). Finally, their results are often difficult to interprete for non-specialists
and generally require the presence of baseline data against which to compare
changes (Salafsky and Margoluis 1999).
Another approach to assess human impact in protected areas and to assess
management effectiveness is to identify and monitor threats directly as a proxy
measurement of conservation success such as e.g. implemented in the Threat
Reduction Assessment (TRA) (Salafsky and Margoluis 1999) and the Rapid
assessment and Prioritization of Protected Area Management (RAPPAM)
program recently established by WWF’s Forest for Life program (Ervin
2003c). This approach of directly identifying threats is sensitive to changes over
short time periods and throughout a site, comparisons among projects and sites
are possible, data can be collected through simple techniques and the method is
practical and cost-effective. Furthermore, the results can be readily interpreted
by conservation staff and can provide detailed, adaptive management guidance

[100]
 1161

to protected area managers. The primary tool for RAPPAM is the rapid
assessment questionnaire which covers management planning, input and pro-
cesses, and the identification of future threats and past pressures (Ervin 2003c).
However, quantitative and objective approaches are still urgently required for
the assessment of threat status and management effectiveness of protected
areas to provide reliable, scientifically sound data.
In this paper we present results from a survey quantifying human impact and
evaluating management effectiveness in Kakamega Forest, western Kenya.
Kakamega Forest is one of the last remaining indigenous forests of Kenya
situated in an agricultural area with a high human density of more than
175 individuals per km2 (Tsingalia 1988). Like many other countries Kenya
harbours an on-going conflict between forest conservation and land use needs
of its increasing population (Tsingalia 1988; Wass 1995). This has put a long-
term pressure on Kakamega Forest leading to its severe reduction and
fragmentation in the last century. Additionally, it has suffered increasing
degradation through both, extensive commercial and local exploitation of
timber (Tsingalia 1988; Fashing et al. 2004; Mitchell, 2004). Large-scale
commercial logging was reduced in the last decades, mostly through official
presidential decree banning all indigenous tree species exploitation in the forest
in the early 1980s (Tsingalia 1988; Mitchell, 2004), the transfer of the northern
part of the forest under the rigorous authority of the Kenya Wildlife Service
(KWS), the establishment of forest stations and ranger patrols, and through
tourism and long-term research (e.g. Zimmerman 1972; Cords 1987; Mutangah
1996; Fashing et al. 2004). However, illegal activities including logging, fuel-
wood collection and extraction of bark for medicinal purposes occur to this
day and appear to be heterogeneously throughout the forest with some sites
providing more protection than others (Kiama and Kiyiapi 2001; Fashing et al.
2004; Fashing, in press). Our study presents a quantitative assessment of the
current threats in the main forest block of Kakamega Forest and its fragments
comprising areas of different management regimes and different protection
priorities. In order to evaluate effectiveness of conservation measures we asked
how differently managed and protected areas differ in their level of human
impact. With this assessment we aim to provide a quantitative, simple site-level
monitoring tool and a first guidance to management planners and decision
makers on problems related to human impact and management in Kakamega
Forest.

Kakamega Forest and its forest history

Study site

We conducted the study at Kakamega Forest (between latitudes 0008¢30.5¢¢ N

(41,236 in UTM 36 N) and 0022¢12.5¢¢ N (15,984) and longitudes 3446¢08.0¢¢
E (696,777) and 3457¢26.5¢¢ E (717,761), G. Schaab, personal communication),

[101]
1162

western Kenya, at an altitude of 1500–1700 m (Figure 1). Kakamega Forest is

a mid-altitudinal tropical rainforest and considered to be the eastern most
remnant of the lowland Congo Basin rainforests of Central Africa (Kokwaro

Figure 1. Satellite image (channel 5 of Landsat 7 ETM+, 05 Feb 2001) of Kakamega main forest
and its five fragments in western Kenya with official forest boundaries as gazetted in 1933 (dashed line)
and official boundaries of National and Nature Reserves (white line). Coordinates in UTM 36 N.

[102]
 1163

1988). Annual rainfall in Kakamega Forest is approximately 2007 mm (as

averaged from FD records at Isecheno Forest Station from 1982 to 2001) and
highly seasonal with a rainy season from April to November and a short dry
season from December to March. The average monthly maximum temperature
ranges from 18 to 29 C while the average monthly minimum ranges from 4 to
21 C (Muriuki and Tsingalia 1990).

Management history

Kakamega Forest was first gazetted as Trust Forest under proclamation

No. 14 in 1933 and has since been managed by the FD; in 1964 it was declared
to be a Central Forest (Blackett 1994). Three small Nature Reserves, Isecheno,
Kisere and Yala, were established and gazetted within the Forest Reserve in
1967 (Blackett 1994). In 1986, the northern part of Kakamega Forest called
Buyangu together with the adjacent Kisere Forest was gazetted as Kakamega
National Reserve and fell under management of the KWS. Today, Kakamega
Forest is part Forest Reserve, part Nature Reserve and part National Reserve,
and management is under the authority of both, FD and KWS, on behalf of
the state.

Fragmentation and disturbance history

Kakamega Forest is a highly fragmented and disturbed forest and has been
continually exploited for many years due to the high surrounding population
pressure (Kokwaro 1988; Wass 1995). The main forest block gazetted in 1933
by the FD to control human activities covered 23,777 ha (Kokwaro 1988, for
original forest boundaries see Figure 1). The FD aimed mostly at provision of
timber for local communities and commercial demand. Clear-felling of indig-
enous forest to make way for fast-growing exotic tree and softwood plantations
was extensive under colonial forest service. Especially the southern parts of
Kakamega were exploited until the late 1980s (Bennun and Njoroge 1999;
Mitchell, 2004). Clearance for settlement and tea plantations slowed over the
1980s as forest protection was better enforced, but more areas were cleared
south of the Yala river (Brooks et al. 1999). Furthermore, selective logging was
intense in the past and the general trend of timber extraction showed a con-
tinued rise from 1933 to 1981 (Tsingalia 1988; Mutangah 1996). Consequently,
the main agents of forest degradation have been mostly logging and extraction
of commercially valuable timber, followed by charcoal burning, cattle grazing,
shamba system farming, hunting for bush-meat, tree debarking and removal of
dead trees for firewood (Oyugi 1996; Mitchell, 2004). In the early 1980s a
presidential decree banned all indigenous tree species exploitation, leading to a
halt of commercial logging, however, tree poaching and other illegal activities
still exist.

[103]
1164

Table 1. List of 22 disturbance survey sites in Kakamega Forest.

Site No. Site name Main forest/ Area Transect Area Management Protection
fragmenta (ha)b length (m) surveyed (ha) regimec statusd

1 Malava f 77 1200 2.4 fd fr

2 Kisere f 420 2600 5.2 kws nr
3 Colobus mf 8245 2400 4.8 kws nr
4 Buyangu mf 8245 1600 3.2 kws nr
5 Shikusa mf 8245 1000 2.0 kws nr
6 Salazar I mf 8245 1000 2.0 kws nr
7 Salazar II mf 8245 2000 4.0 kws nr
8 Shamiloli mf 8245 1000 2.0 fd fr
9 Central II mf 8245 1000 2.0 fd fr
10 Central I mf 8245 1000 2.0 fd fr
11 Chemneko mf 8245 1000 2.0 fd fr
12 Vihiga mf 8245 1000 2.0 fd fr
13 Sawmill mf 8245 1000 2.0 fd fr
14 Isecheno II mf 8245 1000 2.0 fd nr
15 Isecheno I mf 8245 1600 3.2 fd nr
16 Chepsugor f 1370 1000 2.0 fd fr
17 Ikuywa I f 1370 1000 2.0 fd fr
18 Ikuywa II f 1370 1600 3.2 fd fr
19 Yala f 1178 2000 4.0 fd nr
20 Kibiri f 1178 1000 2.0 fd nr
21 Ishiru f 1178 1000 2.0 fd fr
22 Kaimosi f 65 280 0.6 fd fr
a
Abbreviationss: f, fragment; mf, main forest block.
b
Area sizes obtained from satellite image 05 Feb 2001 Landsat 7 ETM+.
c
Abbreviations: fd, Forest Department; kws, Kenya Wildlife Service.
d
Abbreviations: fr, Forest Reserve; nr, National or Nature Reserve.

As a consequence of the long fragmentation and disturbance history

Kakamega Forest was reduced and broken up in several fragments over the
last century and today the main forest block covers only 8245 ha (G. Schaab,
personal communication; for fragment sizes see Table 1) comprising a heter-
ogenous mixture of different succession stages including disturbed primary
forest, secondary forest, clearings and glades, as well as tea and timber plan-
tations (Bennun and Njoroge 1999). For more detailed information on the
fragmentation and disturbance history of Kakamega Forest see Tsingalia
(1988) and Mitchell (2004).

Methods

Disturbance survey

In February and April 2002 and in June and July 2003, disturbance surveys
were carried out at 22 forested sites in Kakamega main forest and its peripheral
fragments (for a complete list of all sites see Table 1). Twelve of the 22 sites

[104]
 1165

Figure 2. Location of 22 disturbance survey sites in Kakamega Forest (left) and the number of
trees logged per hectare in the last 20 years for both, trees 610 cm in diameter and >10 cm in
diameter for each site, respectively (right).

chosen for surveys were close to the 12 sites where Mutangah (1996) carried out
his disturbance surveys in 1992/1994 (see Table 1); an additional 10 new sites
where chosen where Mutangah (1996) had not carried out any surveys in the
past (e.g. in the fragments Kisere, Malava and Kaimosi). This was done in
order to obtain a large sample size of representative sites distributed over the
whole Kakamega Forest. The sites chosen were not necessarily near points of
easy access (Figure 2); in fact, many of the sites are located in the centre of the
forest (e.g. No. 5, 9, 10). With many trails running through the whole of
Kakamega Forest, it is easily accessible for local exploitation. At each site
except for the smallest fragment (Kaimosi), transects were run at least 1000 m
in length (Table 1). Transects sometimes followed existing trails, e.g. at Col-
obus site we chose some of the former overgrown monkey research transects
established by Gathua in 1996 (Fashing and Gathua, in press). In all other
cases where trails did not exist, we made our way through undergrowth along a
line. Surveys included recording any of seven disturbance parameters in a belt

[105]
1166

of 10 m on each side of the transect thereby covering a total area of 56.6 ha

with a median of 2 ha per site (range 0.6–5.2). All disturbances recorded are
thought to present mostly illegal activities.
Disturbance parameters recorded were
1. the number of trees logged: For each tree stump the circumference was
measured to calculate its diameter. Trees with a diameter of less than 10 cm
were assumed to be collected mostly by women and used as firewood,
whereas trees with a diameter of more than 10 cm were assumed to be cut
mostly by men and used as polewood or timber. For each stump the
approximate time since cutting was estimated to be either less than 20 years
or more than 20 years for a distinction between recent and past logging,
respectively. Age was estimated according to the degree of decomposition
and the shape of the remaining tree stump, i.e. stumps with smooth surfaces
or freshly cut stumps with clear cutting profiles were estimated to be logged
in the last 20 years, whereas stumps with wavy or semi-decayed cut surfaces
were estimated to be logged more than 20 years ago (following Mutangah
1996). Tree species were identified when possible; however, for trees having
been logged more than 20 years ago, species identification was not always
possible due to a higher level of decomposition.
2. the number of trees exhibiting any signs of debarking for medicinal use.
3. the number of charcoal kilns, i.e. areas with charcoal remains such as black
half-burned pieces of wood and in some cases still burning charcoal heaps.
4. the number of sawing pits, i.e. pits used for cutting large trees.
5. the number of honey gathering sites, i.e. tree stems with bee-hives from
which honey had been extracted.
6. the number of abandoned and current paths used by locals e.g. for firewood
collection.
7. the number of cattle tracks used to bring cattle to the glades for grazing.

Data analysis

We tested the influence of management regime and protection status on the

disturbance parameters 1–7 for all 22 sites. For disturbance parameter 1, we
tested the influence separately on the number of logged trees of two different
age classes (trees logged in the last 20 years, trees logged more than 20 years
ago) and two different size classes (diameter  10 cm, diameter >10 cm). For
management regime we distinguished between sites being managed by the
KWS (n = 6) and the FD (n = 16) (Table 1). For protection status we dis-
tinguished between sites with high protection priority i.e. National or Nature
Reserves (n = 10) and sites with low protection priority i.e. Forest Reserves
(n = 12) (Table 1). Consequently, 6 National/Nature Reserves are managed
by the KWS and 4 National/Nature Reserves and 12 Forest Reserves by the
FD. Furthermore, we tested the influence of fragmentation on the number of
logged trees for both, the two different age and size classes. We calculated

[106]
 1167

t-tests (t) for normally distributed data with an adjustment in case variances
were unequal and Mann–Whitney U-tests (U) for non-normally distributed
data. We correlated the different disturbance parameters calculating non-
parametric pairwise Spearman correlations.
We compared our data from 12 sites (i.e. site No. 5, 7, 8, 11–13, 15–17, 19–
21) with data from Mutangah (1996) who quantified the same disturbance
parameters in 1992/1994 at the same 12 sites. Although the sites were the same,
transects were not; thus, data are not dependent data. For comparisons of the
two data sets we calculated Pearson and Spearman rank correlations for
normally and non-normally distributed data, respectively.
Data analysis was carried out using JMP (1995).

Results

Evidence for human impact was found at all our 22 sites with a median
number of 21.1 disturbance events per hectare (q1 = 9.8, q3 = 44.6, range
1.8–81.5, n = 22). The sites Salazar II (No. 7) situated in the northern Ka-
kamega National Reserve and managed by the KWS as well as Yala (No. 19)
situated in the Yala Nature Reserve managed by the FD showed lowest
disturbance levels with 2.8 and 4.9 disturbances per hectare, respectively
(Table 2). The site Ishiru (No. 21) had highest disturbance levels with 68.5
disturbances per hectare; is situated at the southern forest edge and is
managed by the FD.

Management, protection status and logging

Of all seven disturbance parameters, logging of trees was the most wide-
spread at all 22 sites (Table 2), thus providing the most useful indicator of
forest disturbance in this study. Over a total survey area of 56.6 ha, we
found 1023 logged trees from 68 species. The most frequent tree species
logged were Funtumia africana (2.1 trees logged per hectare), Prunus africana
(1.3), Celtis new species name: Celtis gomphophylla (1.0), Trilepisium mad-
agascariensis (0.9), Diospyros abyssinica (0.8) and Aningeria altissima (0.7).
The average diameter of tree stumps was 27.0 cm ± 9.6 (if not otherwise
noted mean ± 1 SD).

Past and present logging

We could not find differences in the number of logged trees for past logging
(i.e. more than 20 years ago) between different sites indicating that logging
levels in the past might have been similar throughout Kakamega Forest
(Figures 3a, b; management regime, U: Z = 0.26, p = 0.79; protection status,
t: t = 2.32, df = 1, p = 0.15). In contrast, a significant effect of management
regime and protection status was found for present logging (i.e. in the last

[107]
 1168
Table 2. Summary data of disturbance survey for all 22 sites.

Site No. trees No. trees Total No. trees No. trees No. charcoal No. sawing No. honey No. No. cattle Total no. trees
No. logged/ha logged/ha logged/ha with signs of kilns/ha pits/ha gathering paths/ha tracks/ha logged/ha in
>20 years <20 years debarking/ha sites/ha 1992/1994

1 3.8 5.4 9.2 1.7 0 2.1 0 8.8 0.4 –

2 1.9 8.5 10.4 1.7 0 1.0 0 2.3 0.2 –
3 2.9 2.3 5.2 0 0 0 0 0.2 0 –
4 5.0 4.1 9.1 0 0 0 0 0 0 –
5 1.5 1.5 3.0 0 1.0 0.5 0 1.5 0 14.0
6 3.0 2.0 5.0 0 0 3.0 0 2.0 0 –
7 2.0 0 2.0 0.3 0 0 0 0.5 0 1.0
8 6.0 53.5 59.5 0.5 2.0 1.0 0 1.0 4.5 52.0
9 2.0 8.0 10.0 0 0 2.5 0 7.5 0 –
[108]

10 2.0 10.0 12.0 0 0 0 0 1.0 0 –

11 1.5 40.5 42.0 0 4.5 0 0 6.5 0 74.0
12 6.0 15.0 21.0 3.0 2.5 0 0 7.0 0 29.0
13 1.0 36.0 37.0 0 0.5 0 5.0 0 0.5 8.0
14 0 9.0 9.0 1.0 0 0.5 0 2.0 0 –
15 0 12.5 12.5 2.5 0 0.3 0 0 11.3 16.0
16 8.0 34.5 42.5 0.5 0 0 0 6.5 0 83.0
17 0 3.8 3.8 1.3 1.3 0.3 0 5.9 1.9 23.0
18 3.5 33.5 37.0 0 1.0 0 0 0.5 0 –
19 1.3 1.5 2.8 1.3 0 0.8 0 0 0 11.0
20 3.0 45.4 48.9 0 0 0 0 13.0 0 99.0
21 10.5 65.0 75.5 0.5 0.5 0 0 5.0 0 77.0
22 0 30.0 30.0 6.7 0 0 0 0 0 –
For logging, numbers per hectare are given separately for trees logged >20 years ago, trees logged <20 years ago and for all trees logged. For a comparison of
logging between today and 1992/1994, the total number of trees logged as recorded by Mutangah (1996) are given.
 1169

Figure 3. Number of trees logged more than 20 years ago (a, b) and in the last 20 years (c, d) for
sites under management by the Kenya Wildlife Service (KWS, n = 6) and the Forest Department
(FD, n = 16) (a, c) and for sites having high protection priority (i.e. situated within National/
Nature Reserves, n = 10) and low protection priority (i.e. situated within Forest Reserves, n = 12)
(b, d). Given are medians, quartils, minimum and maximum values and significance levels. n.s., not
significant; * p < 0.05; **p < 0.005.

20 years) with fewer trees logged at sites managed by the KWS and with high
protection priority (Figures 3c, d; management regime, t: t = 19.21, df = 1,
p = 0.0004; protection status, U: Z =
2.54, p = 0.0111).

Firewood and polewood/timber use in the last 20 years

Traces of firewood collection (tree diameter  10 cm) were often found at sites
managed by the FD but rarely at sites managed by the KWS (Figure 2). The
FD-managed sites situated within Nature Reserves (No. 14, 15, 19), in the
centre of the forest (No. 9, 10) and in central Ikuywa (No. 17) had the lowest
levels of firewood collection (Figure 2). In contrast, sites at the forest edge (No.
8, 11, 12, 16, 18, 20, 21) adjacent to local settlements had the highest logging

[109]
1170

levels (Figure 2). Tree cutting for polewood/timber (tree diameter >10 cm)
had low levels at KWS sites, but high levels at FD sites (Figure 2). Both,
firewood collection and logging for polewood/timber was significantly higher
at FD sites as compared to KWS sites (firewood/ha: KWS median = 0,
q1 = 0, q3 = 0.3, range 0–1.2, FD median = 5.5, q1 = 1.6, q3=17.3, range
0–26.0; U: Z =
3.04, p = 0.0024; polewood/ha: KWS 2.9 ± 2.6; FD
16.9 ± 12.7; t: F = 16.54, df = 1, p = 0.0007).
Similarly, significant differences in the number of logged trees for both
firewood and polewood/timber, were found between sites of high and low
protection priority (firewood/ha: high protection median = 0, q1 = 0,
q3 = 2.0, range 0–19.0, low protection median = 10.0, q1 = 1.8, q3 = 17.3,
range 0–26.0; U: Z =
2.55, p = 0.011; polewood/ha: high protection
median = 3.2, q1 = 1.5, q3 = 8.1, range 0–26.5, low protection med-
ian=15.4, q1 = 7.4, q3 = 26.3, range 2.5–47.5; U: Z =
2.61, p = 0.0092).

Fragmentation and logging

No differences were found between the main forest and the fragments for the
number of trees logged more than 20 years ago and less than 20 years ago, as
well as for the number of trees logged for firewood and for polewood/timber
(trees logged >20 years ago: U: Z = 0.44, p = 0.64; trees logged < 20 years
ago: U: Z = 0.83, p = 0.4037; firewood/ha: U: Z = 1.57, p = 0.1164; pole-
wood/ha: U: Z = 0.03, p = 0.97).

Re-assessment of logging: 1992/1994 and today

No differences for the number of logged trees per hectare were found between
our data and those of Mutangah’s (1996) survey in 1992/1994. This suggests a
similar overall logging level (1992/1994: 40.6 ± 34.4, today: 29.2 ± 25.2; t:
t = 0.92, df = 22 p = 0.37). Furthermore, we found a significant positive
correlation between both data sets suggesting that transects at the same sites
still have same logging levels after 10 years (Pearson: r = 0.72, p = 0.0088).

Management, protection status and other human impact

For all other disturbance parameters differences between differently managed

and protected sites were only found for the number of charcoal kilns with sig-
nificantly lower numbers in highly protected sites (high protection: median = 0,
q1 = 0, q3 = 0, range 0–1.0, low protection: median = 0.5, q1 = 0, q3 = 1.8,
range 0–4.5; U: Z =
2.26, p = 0.024). However, in contrast to the number of
logged trees, all other disturbance parameters were mostly rare events and ap-
pear to be indicators only for localized threats (Table 2). Burning of charcoal,
e.g., seems to be a serious threat at the eastern (No. 11, 12) and western edge (No.
8) of the main forest block, whereas cattle tracks appear to be a problem mostly
at Isecheno I (No. 15) and at the eastern edge of the forest (No. 8) (Table 2). In

[110]
 1171

general, no correlation could be found between the disturbance parameters

when calculating non-parametric pairwise Spearman correlations (p > 0.05).

Discussion

Status quo of human impact

According to our survey human impact is found everywhere in Kakamega

Forest with logging being most widespread. This confirms the expressions of
alarm over the misuse and overexploitation of Kakamega’s forest resources
through illegal human activities (Kowkaro 1988; Emerton 1991; Mutangah
et al. 1992; Wass 1995; Oyugi 1996; Fashing et al. 2004). Our data support
Mutangah’s (1996) survey from 1992/1994 indicating the highest logging
levels occur in the most southerly part of the forest as well as along the
western edge. Furthermore, some of the disturbed sites (e.g. No. 11, 21) in
Mutangah’s (1996) survey have been degraded heavily in the meantime and
the canopy cover reduced substantially (N. Saijita and C. Analo, personal
communication). In both, Mutangah’s (1996) and our survey, the lowest
logging levels were found in the northern Kakamega National Reserve,
central Ikuywa and Yala.

Human impact in differently managed areas

Our data do not only show the current status quo of the human impact on
Kakamega Forest, but also reflect its management history in the last
20 years. Before 1986, when all of Kakamega Forest was managed by the
FD, Colobus, Buyangu and Salazar sites (No. 3, 4, 6, 7) in the northern part
were well known for intensive commercial logging through timber companies
(Tsingalia 1988; Mitchell, 2004). Correspondingly, the number of trees logged
more than 20 years ago appears to be equally high at those sites as compared
to others. In 1986, the KWS took over the northern part of Kakamega
Forest as a National Reserve and the changes in management appear to have
resulted in changes in logging numbers in the last 20 years. Illegal tree
poaching was reduced at sites under KWS management probably due to
tightened security, whereas FD sites still experience higher tree poaching rates
today. Furthermore, FD sites show various other local threats such as e.g.
charcoal burning and cattle grazing. Under FD management, sites with high
protection priority such as Yala and Isecheno Nature Reserves (No. 14, 15,
19) still show lower overall threat levels as compared to sites with low pro-
tection priority. For example, Fashing et al.’s (2004) results of a long-term
study of tree populations in Kakamega Forest indicate that their study plots
in Isecheno remained relatively undisturbed over the last 20 years. A decrease
of pioneer species density by 21% in these sites are taken as evidence that the

[111]
1172

forest is maturing towards a climax forest and that at least the conservation
measures applied to Isecheno appear to have succeeded Fashing et al. 2004).
Nevertheless, prospects for other severly disturbed sites are assumed to be
bad, as is the general prognosis for Kakamega Forest if protection efforts are
not increased and illegal exploitation by local people remains high, particu-
larly on its periphery (Cords and Tsingalia 1982; Kokwaro 1988; Tsingalia
1988; Fashing et al. 2004).
How do the two conservation boards KWS and FD differ in their man-
agement aims and strategies? The overall aim of the KWS is ‘to conserve,
protect and sustainably manage the wildlife resources’ and its areas are set
aside for conservation and tourism only (Wass 1995). People are not allowed
to collect any forest products and these policies are strictly enforced through
regular patrols by up to five game rangers (E.W. Kiarie, personal commu-
nication). The overall aim of the FD is to ‘enhance conservation and pro-
tection of indigenous forest, to improve the production of timber and
fuelwood and to establish a framework for the long-term development
forestry’ (Wass 1995). Some sites are also set aside for conservation, how-
ever, some used to be plantations of exotic tree species or mixtures of
indigenous species, while others experienced enrichment planting (A. Oman,
personal communication). Logging, tree debarking and charcoal burning is
prohibited, whereas fuelwood collection was licenced until recently (A.
Oman, personal communication). It appears that the FD has been largely
restricted in its capacity to implement conservation policies effectively due to
the lack of adequate resources in contrast to the better funded KWS,
leading to insufficient levels of staffing, patrols, weaponry etc. These differ-
ences in resources might have led to different disturbance levels as found in
our survey.
Besides overall funding and the number of staff, other potential factors
associated with management regime and effectiveness might be e.g. accessibility
to the forest or proximity of the forest to neighbouring settlements, population
density, community relations and compensation programs to locals. In a recent
assessment of the impact of anthropogenic threats on 93 protected areas of 22
tropical countries park effectiveness was shown to correlate most strongly with
density of guards i.e. the more guards the higher effectiveness (Brunner et al.
2001). Furthermore, effectiveness correlated with the level of deterrence of
illegal activities in the parks and with the degree of border demarcation and
existence of direct compensation programs for local communities (Brunner et
al. 2001). However, it did not correlate with enforcement capacity (i.e. a
composite variable of training, equipment and salary), accessibility, budget,
number of staff working on economic development or education, or the local
involvement of communities in park management (Brunner et al. 2001). To
obtain more information on the factors influencing management effectiveness
in Kakamega Forest more studies are highly recommended following the
RAPPAM guidelines.

[112]
 1173

Management recommendations for Kakamega Forest

The high human impact on Kakamega Forest especially along the western
and eastern edge of the main forest block indicates an imminent danger of
further fragmentation. The main forest block might fall into two separate
forest blocks, i.e. Kakamega National Reserve in the North and Isecheno
Nature Reserve in the South. To prevent this from happening in the near
future, we strongly recommend following the management plan of forest
zoning as outlined by the Kenya Indigenous Forest Conservation Pro-
gramme (KIFCON 1994; Wass 1995): establishing a protection zone to
provide a core for biodiversity conservation extending from the North to
South; setting up a rehabilitation zone with enrichment planting where
degradation has reached high levels; and establishing a subsistence use zone
flanking the protection zones where local people are allowed to extract
forest products. This forest zoning aims both, to maintain as much indig-
enous forest cover as possible and to permit optimal use of forest resources
on a sustainable basis (Wass 1995). We recommend placing the protection
zone under strict KWS management as our survey indicates that areas of
Kakamega Forest managed by the KWS appear to hold surprisingly low
disturbance levels despite high land-use pressure.
The degradation and logging levels in the suggested subsistence zones are
already alarming, so that we suggest enrichment planting there. Finally,
encouragement of on-farm-forestry projects might provide resources in the
long-term and thus might relieve the subsistence use zone. This is supported
by the fact that a tree nursery run by the local grassroot conservation
organization KEEP (Kakamega Environmental Education Program) at Is-
echeno Forest Station has been successfully nursing seedlings of both,
indigenous and exotic tree species, for sale to local farmers. Beyond con-
servation measures for the main forest block, high protection priority must
also be given to the low-disturbance sites central Ikuywa (No. 17), Yala
(No. 19) and the 400 ha fragment of Kisere (No. 2). Kisere Nature Reserve
is of particurlar conservation significance because it has been relatively
undisturbed in the past and still harbours species-rich forest communities
that include the rare DeBrazza monkeys (Cercopithecus neglectus) (Muriuki
and Tsingalia 1990; Chism and Cords 1997). Although managed by KWS, it
appears to have experienced increasing disturbance levels in the last few
years (N. Saijita and C. Analo, personal communication). This might be due
to the lack of ranger outposts in Kisere (KWS headquarters is at Buyangu),
and the fact that the number of rangers (10–20) might not be sufficient to
cover both, Buyangu and Kisere. Therefore, an immediate increase of reg-
ular ranger patrols to control logging more effectively is highly recom-
mended, as suggested by previous authors (Kokwaro 1988; Mutangah 1996;
Chism and Cords 1997).

[113]
1174

Logging as indicator for quantitative threat assessment

In our survey only logging appeared to be an effective indicator for human

impact on the forest and might offer a valuable tool to conservation managers.
First, the recording of the number of logged trees provides a quantitative,
objective measure of the human impact on protected areas. Most other
assessments of threat status and management effectiveness used qualitative
rather than quantitative approaches (see e.g. Salafsky and Margoluis 1999;
Brunner et al. 2001; Ervin 2003c). For example, following the RAPPAM
methodology and using a questionnaire, the question arises whether the pro-
tected area managers themselves answering questions on their own manage-
ment will supply objective answers. Consequently, our method collecting
empirical data on the number of logged trees is more objective.
Second, methods using logging as a disturbance indicator assess distur-
bance directly and not through biological indicators. Often, human impact is
inferred from long-term studies on plant species composition or population
structure as a biological indicator (e.g. Fashing et al. 2004). However, bio-
logical indicators can only assess the present situation resulting from past
human impact. In contrast, quantifying disturbances directly can provide
empirical data on the present human impact. Furthermore, logging as a
disturbance indicator can enable us to differentiate between recent and past
disturbance and might consequently help to evaluate past management
policies.
Third, despite its quantitative approach this method provides a simple, low-
budget method important especially for rapid and repeated assessment of
disturbed forests. Repeated assessments are crucial especially in protected areas
such as Kakamega Forest where heterogeneity in forest condition occurs over
small spatial scales (Fashing et al. 2004). Consequently, surveys using logging
as a disturbance indicator can provide the maximum amount of current up-to-
date and scientifically sound information for management planners in return
for the effort and time involved.
Finally, although the list of potential threats facing protected areas world-
wide is long, logging appears to affect nearly 70% of more than 200 parks
throughout the tropics (van Schaik et al. 1997) and emerges as one of the most
hotly debated issues in tropical forest conservation (Rice et al. 1997; Bowles
et al. 1998; Laurance 2001). Consequences of logging do not only include loss
of habitat, but also changes in the microclimatic environment, erosion of soil
and modification of fire regimes (Barlow et al. 2002; Cochrane and Laurance
2002) with the impact depending on the type of logging, i.e. whether com-
mercial mechanized logging with heavy equipment or local exploitation of
timber through e.g. pit-sawying and firewood collection. Furthermore, sec-
ondary effects of logging might be increased access to remote forested areas
through the creation of roads and paths leading to further logging, forest
colonization and hunting (Wilkie et al. 1992; Rice et al. 1997; Laurance 1998;
Robinson et al. 1999). Consequently, logging appears to be a serious constant

[114]
 1175

threat to tropical forests worldwide making its validity as an useful indicator

even more probable.
Disturbance or impact assessments in combination with long-term studies
on forest structure and composition after logging (e.g. Plumptre 1996;
Chapman and Chapman 1997; Struhsaker 1997; Fashing et al. 2004) can
provide important information on regeneration dynamics after human im-
pact. Studies from Kakamega Forest indicate that regeneration from the
severe human impact of the last century might be possible though not
without rigorous conservation measures (Fashing et al. 2004, this study).
Finally, repeated disturbance assessments are important to keep track of the
human impact in protected areas and can provide feedback to management
planners when evaluating past management decisions and setting up new
conservation goals.

Acknowledgements

The study was funded by the German Federal Ministry of Education and
Research within the framework of BIOTA East Africa (01LC0025 / subpro-
jects E03, E10 & E11). We thank the Kenyan Ministry for Education and
Research for the permission to carry out research in Kakamega Forest, and the
KWS and FD for granting us access to their reserves. We appreciate infor-
mation on management by E.W. Kiarie (Senior Warden, Kakamega National
Reserve) and A. Oman (Assistant District Forest Officer, Kakamega). We
highly acknowledge field assistance by C. Analo and N. Saijita. G. Schaab
kindly provided maps and data on forest patch sizes. We wish to thank the
paperclub at Mainz, N. Mitchell, L. Todt, H.Todt, and two anonymous ref-
erees for comments on earlier drafts of the manuscript and K. Boehning-Gaese,
H. Dalitz and M. Kraemer for overall support.

References

Barlow J., Haugaasen T. and Peres C.A. 2002. Effects of ground fires on understorey bird
assemblages in Amazonian forests. Biol. Conserv. 105: 157–169.
Bennun L. and Njoroge P. 1999. Important Bird Areas in Kenya. The East Africa Natural History
Society, Nairobi, Kenya.
Blackett H.L. 1994. Forest inventory report no. 3: Kakamega. Kenya Indigenous Forest Conser-
vation Programme, Nairobi, Kenya.
Bowles I.A., Rice R.E., Mittermeier R.A. and da Fonseca G.A.B. 1998. Logging and tropical forest
conservation. Science 280: 1899–1900.
Brooks T.M., Pimm S.L. and Oyugi J.O. 1999. Time lag between deforestation and bird extinction
in tropical forest fragments. Conserv. Biol. 13: 1140–1150.
Brunner A.G., Gullison R.E., Rice R.E. and da Fonseca G.A.B. 2001. Effectiveness of parks in
protecting tropical biodiversity. Science 291: 125–128.
Chapman C.A. and Chapman L.J. 1997. Forest regeneration in logged and unlogged forests of
Kibale National Park, Uganda. Biotropica 29: 396–412.

[115]
1176

Chism J. and Cords M. 1997. De Brazza’s monkeys Cercopithecus neglectus in the Kisere National
Reserve, Kenya. Afr. Primates 3: 18–22.
Cochrane M.A. and Laurance W.F. 2002. Fire as a large-scale edge effect in Amazonian forests.
J. Trop. Ecol. 18: 311–325.
Cords M. 1987. Mixed species association of Cercopithecus monkeys in the Kakamega Forest,
Kenya. Univ. Cali. Publ. Zool. 117: 1–109.
Emerton L. 1991. Summary of the Current Value of use of Kakamega Forest. Kenya Indigenous
Forest Conservation Programme, Nairobi, Kenya.
Ervin J. 2003a. Protected area assessment in perspective. BioScience 53: 819–822.
Ervin J. 2003b. Rapid assessment of protected area management effectiveness in four countries.
BioScience 53: 833–841.
Ervin J. 2003c. Rapid assessment and prioritization of protected area management (RAPPAM)
methodology. World Wide Fund for Nature, Gland, Switzerland. Available at http://
www.panda.org/downloads/forests/rappam.pdf.
Fashing P.J. Mortality trends in the African cherry (Prunus africana) and the implications for
colobus monkeys (Colobus guereza) Kakamega Forest, Kenya. Biodiv. Conserv., in press.
Fashing P.J., Forrestel A., Scully C. and Cords M. 2004. Long-term tree population dynamics and
their implications for the conservation of the Kakamega Forest, Kenya. Biodiv. Conserv. 13:
753–771.
Fashing P.J. and Gathua M. Spatial variability in the vegetation structure and composition of an
East African rain forest. Afr. J. Ecol. in press.
Heywood V.H. (ed.) 1995. Global Biodiversity Assessment. Cambridge University Press, Cam-
bridge, UK.
Hockings M. 2003. Systems of assessing the effectiveness of management in protected areas.
BioScience 53: 823–832.
Hockings M., Stolton S. and Dudley N. 2000. Evaluating effectiveness: a framework for assessing
the management of protected areas. IUCN, Gland, Switzerland. Available at http://wcpa.
iucn.org/pubs/pdfs/evaluating_effect.pdf.
JMP 1995. Statistics and Graphics Guide. Version 3.1. SAS Institute Inc., Cary, NC.
Kiama D. and Kiyiapi J. 2001. Shade tolerance and regeneration of some tree species of a tropical
rain forest in western Kenya. Plant Ecol. 156: 183–191.
KIFCON 1994. Kakamega Forest – The offical guide. Kenya Indigenous Forest Conservation
Programme, Nairobi, Kenya.
Kowkaro J.O. 1988. Conservation status of the Kakamega Forest in Kenya: the easternmost relict
of the equatorial rain forests of Africa. Monogr. Syst. Bot. Missouri Bot. Garden 25: 471–489.
Laurance W.F. 1998. A crisis in the making: responses of Amazonian forests to land use and
climate change. Trends Ecol. Evol. 13: 411–415.
Laurance W.F. 2001. Tropical logging and human invasions. Conserv. Biol. 15: 4–5.
Lindenmayer D.B., Margules C.R. and Botkin D.B. 2000. Indicators of biodiversity for ecologi-
cally sustainable forest management. Conserv. Biol. 14: 941–950.
Margules C.R. and Pressey R.L. 2000. Systematic conservation planning. Nature 405: 243–253.
Margoluis R. and Salafsky N. 1998. Measures of Success: Designing, Managing, and Monitoring
Conservation and Development projects. Island Press, Washington, DC.
Mitchell N. 2004. The exploitation and disturbance history of Kakamega Forest, western Kenya.
In: Bleher B. and Dalitz H. (eds), BIOTA Report No. 1. Bielefelder Ökologische Beiträge 20.
Muriuki J.W. and Tsingalia M.H. 1990. A new population of de Brazza’s monkey in Kenya. Oryx
24: 157–162.
Mutangah J.G. 1996. An investigation of vegetation status and process in relation to human
disturbance in Kakamega Forest, western Kenya. Ph.D thesis, University of Wales, UK.
Mutangah J.G., Mwangangi O. and Mwaura P.K. 1992. Kakamega: A Vegetation Survey Report.
Kenya Indigenous Forest Conservation Programme, Nairobi, Kenya.
Myers N., Mittermeier R.A., Mittermeier C.G., da Fonseca G.A.B. and Kent J. 2000. Biodiversity
hotspots for conservation priorities. Nature 403: 853–858.

[116]
 1177

Noss R.F. 1990. Indicators for monitoring biodiversity: a hierarchical approach. Conserv. Biol. 4:
355–364.
Oyugi J.O. 1996. Kakamega Forest is dying. East Afr. Nat. Hist. Soc. Bull. 26: 47–49.
Parrish J.D., Braun D.P. and Unnasch R.S. 2003. Are we conserving what we say we are? Mea-
suring ecological integrity within protected areas. BioScience 53: 851–860.
Pimm S.L., Russell G.J., Gittleman J.L. and Brooks T.M. 1995. The future of biodiversity. Science
269: 347–350.
Plumptre A.J. 1996. Changes following 60 years of selective timber harvesting in the Budongo
Forest Reserve, Uganda. Forest Ecol. Manage. 89: 101–113.
Putz F.E., Blate G.M, Redford K.H, Fimbel R. and Robinson J. 2001. Tropical forest management
and conservation of biodiversity: an overview. Conserv. Biol. 15: 7–20.
Rice R.E., Gullison R.E. and Reid J.W. 1997. Can sustainable management save tropical forests?
Sci. Am. 276: 44–49.
Robinson J.G., Redford K.H. and Bennett E.L. 1999. Wildlife harvest in logged tropical forest.
Science 284: 595–596.
Salafsky N. and Margoluis R. 1999. Thread reduction assessment: a practical and cost-effective
approach to evaluating conservation and development projects. Conserv. Biol. 13: 830–841.
Salafsky N., Margoluis R., Redford K.H. and Robinson J.G. 2002. Improving the practice of
conservation: a conceptual framework and research agenda for conservation science. Conserv.
Biol. 16: 1469–1479.
Sparrow H.R., Sisk T.D., Ehrlich P.R. and Murphy D.D. 1994. Techniques and guidelines for
monitoring neotropical butterflies. Conserv. Biol. 8: 800–809.
Struhsaker T.T. 1997. Ecology of an African Rain Forest: Logging in Kibale and the Conflict
Between Conservation and Exploitation. University of Florida Press, Gainesville, Florida.
Tsingalia M.H. 1988. Animals and the regeneration of an African rainforest tree. Ph.D. Thesis,
University of California, Berkely.
van Schaik C.P., Terborgh J. and Dugelby D. 1997. The silent crisis: the state of the rainforest
nature reserves. In: Kramer R., van Schaik C.P. and Johnson J. (eds), The Last Stand: Protected
Areas and the Defense of Tropical Biodiversity. Oxford University Press, New York, pp. 64–89.
Wass P. 1995. Kenya’s Indigenous Forests – Status, Management and Conservation. IUCN,
Gland, Switzerland.
Whitmore T.C. 1997. Tropical forest disturbance, disappearance and species loss. In: Laurence
W.L. and Bieregaard R.O.Jr. (eds), Tropical Forest Remnants: Ecology, Management, and
Conservation of Fragmented Communities. University of Chicago Press, Chicago, pp. 3–12.
Wilkie D.S., Sidle J.G. and Boundzanga G.C. 1992. Mechanized logging, market hunting, and a
bank loan in Congo. Conserv. Biol. 6: 570–580.
Zimmerman D.A. 1972. The avifauna of the Kakamega Forest, western Kenya, including a bird
population study. Bull. Am. Museum of Nat. Hist. 149: 257–339.

[117]
Biodiversity and Conservation (2006) 15:1179–1191
Springer 2006
DOI 10.1007/s10531-004-4693-x

-1

Influence of forest types and effects of forestry

activities on species richness and composition
of Chrysomelidae in the central mountainous region
of Japan

MASASHI OHSAWA* and TAKUO NAGAIKE

Yamanashi Forest Research Institute 2290-1 Saishoji, Masuho-cho, Minamikoma-gun, Yamanashi
400-0502, Japan; *Author for correspondence (e-mail: [email protected]; phone:
+81-556-22-8001; fax: +81-556-22-8002)

Received 4 May 2004; accepted in revised form 28 October 2004

Key words: Batophila acutangula, Biodiversity, Insect diversity, Larix kaempheri, Leaf beetles,
Species composition, Species richness, Sphaeroderma tarsatum

Abstract. Species richness and composition of the Chrysomelidae (Coleoptera) were studied in
larch (Larix kaempheri [Lamb.] Carrière) plantations, secondary forests, and primary forests. In
addition, the effects of forest management practices, such as thinning and long rotation, were
examined in the larch plantation. The species richness of Chrysomelidae was higher in the larch
plantation than in the secondary forest or in the primary forest. Among the larch plantations, the
species richness in old-aged plantations was higher than that in middle-aged plantations. The
composition of the beetle assemblages in the larch plantation differed from that in the secondary
forest or in the primary forest. Exosoma akkoae (Chujo), Batophila acutangula Heikertinger, and
Calomicrus nobyi Chujo were caught with a bias toward the larch plantation. Longitarsus succineus
(Foudras) and Sphaeroderma tarsatum Baly were caught more in the secondary forest and the
primary forest, respectively. More B. acutangula and S. tarsatum were caught in stands where their
host plants occurred at higher rates. Species richness of understory plants was an important factor
for chrysomlid species richness, and frequency of host occurrence affected the number of indi-
viduals of leaf beetles examined. It seems that forest types and forest management practices affect
host plants as well as Chrysomelidae, and that these effects on the host plants also influence
chrysomelid assemblages.

Introduction

The original vegetation of forestal area in the central mountainous region in

Japan is considered as mixed forest of broad-leaved trees and coniferous trees
dominated by Quercus crispula Blume. Since the 1940s, however, primary forests
have rapidly diminished as a result of forestry activities, and now the area is
known for its larch plantations. To achieve ecologically sustainable forest
management in this region, studies were started to elucidate the status of bio-
diversity in forests and effects of forest management practices in the area, i.e., a
basic information necessary for ecologically susutainable management. The
study area is largely occupied by Japanese larch (Larix kaempferi [Lamb.]
Carrière) plantations and secondary broad-leaved forests, dominated by

[119]
1180

Quercus crispula Blume, (41 and 33% of the total area, respectively). Larch
plantations were established for timber production; thinning has been conducted
twice within a 45-year period in this area, in order to enhance growth of domi-
nant trees. Long rotations have been adopted for some larch plantations to
obtain high quality wood. Secondary forests were formerly used to produce
firewood, charcoal, etc. Now, however, they are mostly left unattended.
Because larch plantations and secondary forests occupy large areas, their
role in conserving biodiversity is important. Primary forests, though frag-
mented today, should also be considered important because of the potential
presence of rare indigenous species inhabiting in them. Three forest types (larch
plantation, secondary forest, and primary forest) and two management prac-
tices frequently used in larch plantations (thinning and long rotations) were
chosen for this study.
Leaf beetles (Chrysomelidae, Coleoptera) are small in size (less than 1.5 cm
in general) and their antennas are usually less than half the length of their
bodies. They distribute worldwide and about 500 species have so far been
reported in Japan (Kimoto and Takizawa 1994). Though they are related with
cerambycid beetles which are saproxylic in their larval stage, leaf beetles feed
on leaves, stems or roots in both larval and adult stages and are regarded as
phytophagous herbivores. The diversity of Cerambycidae, saproxylic beetles,
was investigated in this area, and it was reported that higher species richness of
the beetles was observed in secondary forests than in larch plantations or
primary forests, and that thinning increased cerambycid diversity in larch
plantations (Ohsawa 2004). This time, leaf beetles were chosen for the study of
diversity to clarify the effects of different forest types and forestry activities on
the phytophagous herbivores. There has been one report published on the
subject of the diversity and conservation of Chrysomelidae (Greatorex-Davies
and Sparks 1994). In this study, the diversity of leaf beetles and some other
insects was investigated in the rides of woodlands, and it has been reported that
both species richness and abundance of leaf beetles declined with increasing
levels of shade, and that rides must be actively managed to keep light levels
high if species richness and abundance are to be maintained.
The purpose of this study was to compare the species richness and compo-
sition of Chrysomelidae, phytophagous herbivores, in larch plantations, sec-
ondary forests, and primary forests, and assess the effects of forest practices,
such as thinning and long rotations, on the diversity of this family in order to
obtain information for conservation of leaf beetles in forest area.

Methods

Study site

This study was conducted in forests at altitudes ranging from 1390–1770 m

(mean annual precipitation: ca. 1120 mm; mean annual temperature: ca. 9.9 C)

[120]
 1181

in the central mountainous region of Japan (Figure 1 of Ohsawa 2004). Three

forest types in this area, i.e., larch plantations, secondary broad-leaved forest
(secondary forest), and primary or near-primary broad-leaved forest (primary
forest), were selected for the investigation. A total of 46 stands were chosen: 24
stands of larch plantations, aged 21–79 years; 11 stands of secondary forest,
dominated by Q. crispula, with other broad-leaved trees such as Ilex macropoda
Miq., Fraxinus sieboldiana Blume var. serrata Nakai, Betula platyphylla Su-
katchev var. japonica (Miq.) Hara, Acer spp ., and Prunus spp.; and 11 stands of
primary forest dominated by Q. crispula, but mixed with conifers such as Abies
homolepis Sieb. et Zucc. and Tsuga diversifolia (Maxim.) Masters, with a high
proportion of conifers in two of the stands. Larch plantations consisted of three
types: 13 stands of middle-aged (21–45 years old), six stands of recently thinned
(middle-aged, 1.5 and 2.5 years after thinning), and five stands of old (58–
79 years old) plantations.

Survey of leaf beetles

A Malaise trap was set in each of the 46 stands to capture insects, and leaf
beetles were separated from among those trapped. To minimize forest edge
effects and micro-topographical differences within each stand, efforts were
made to set each trap in a typical spot for the stand on a simple slope (or in a
flat area if there was no slope) in the interior of the stand.
Leaf beetles were captured in three 14-day periods: in middle to late June,
in middle to late July, and in early to mid-August. Because of high eleva-
tion, June to August is an appropriate period for capturing many beetles
including Chrysomelidae, according to the results of an investigation
conducted on insects through spring to autumn in this area (Ohsawa
unpublished data).
Dry specimens were prepared for all species of leaf beetles trapped, and were
kept in the Yamanashi Forest Research Institute.

Environmental factors

Environmental factors examined were altitude, gradient, larch plantation

versus natural broad-leaved forest mostly dominated by Q. crispula, i.e.,
secondary forest and primary forest (larch plantation vs. natural broad-
leaved forest), openness of the canopy, the numbers of vascular plants
(<2 m and 2 m in height), and age of larch plantations. In addition, for
two chrysomelid species, Batophila acutangula Heikertinger and Sphaero-
derma tarsatum Baly, commonly found in this area, the occurrence of their
host plants was investigated as an environmental factor. For the openness of
the canopy, a digital camera with a fish-eye converter (Nikon Cool Pix 950)
was used to take a hemispherical photograph of the canopy from a height
of 1 m above the ground at five locations. The openness of the canopy was

[121]
1182

an average of five values calculated from the hemispherical photographs

using HEMIPHOTO (ter Steege 1993). The number of vascular plant species
was investigated in 11 stands. Vascular plants were divided into two groups
by their height: the one shorter than 2 m and the other 2 m or more.
Counted in the numbers of vascular plants were, for the first group (<2 m),
all species found in 40 plots (1 · 1 m), and for the second group (2 m), all
species found in 40 plots (5 · 5 m) for each stand (Nagaike et al. 2003).
Rubus spp. (Rubus mesogaeus Focke, R. koehneanus Focke, R. palmatus
Thunb. var. coptophyllus A. Gray, and R. pungens Cambessèdes var. oldhamii
[Miq.] Maxim.) and bamboo grass (Sasa nipponica Makino et Shibata and
Sasamorpha borealis [Hack.] Nakai) were host plants of B. acutangula and
S. tarsatum, respectively. In each 46 stand, the occurrences of Rubus spp. and
bamboo grass were investigated in an area of 10 · 30 m including in the center
a spot where a Malaise trap was set in. The occurrence of Rubus spp. in stands
was ranked at four degrees (0: not found; 1: found in less than 10% of the area;
2: found in at least 10% and less than 25% of the area; and 3: found in at least
25%). The occurrence of bamboo grass was also ranked at four degrees (0: not
found; 1: occupied less than 10% of the area; 2: occupied at least 10% and less
than 25%; and 3: occupied at least 25%).

Statistical analyses

One way ANOVA with Bonferroni’s post-hoc test (Bonferroni test) was con-
ducted to compare the number of species, the number of individuals, and
Shannon-Wiener diversity index (H¢, Magurran 1988) among the forest types.
Logarithmic transformation was performed for the number of individuals
before Bonferroni test. Simple regression analysis was conducted to identify the
environmental factors that affected the species richness of leaf beetles. The
regression analysis was also applied to the relationship between the numbers of
two chrysomelid species and occurrence of their host plants.
Detrended correspondence analysis (DCA) was applied to compare the
composition of leaf beetles in all investigated stands. The analysis was conducted
with the software package, PC-ORD (McCune and Mefford 1999). Bonferroni
test was applied to compare DCA scores (the first and second axes) among forest
types. To define the meaning of the first two axes of DCA, simple correlation
coefficients were calculated between the two axes and stand-structure parame-
ters, such as the number of species, the number of individuals, H¢, larch plan-
tation vs. natural broad-leaved forest, altitude, gradient, openness of the canopy,
the number of vascular plants (<2 m and 2 m in height, for 11 stands) and
forest age (for larch plantations),
Indicator species analysis (Dufrêne and Legendre 1997) was carried out to
compare species of Chrysomelidae among the forest types and to identify
species associated with each forest type.

[122]
 1183

Results

Number of species, individuals, and diversity index

In this study, 1303 individuals of 41 species were captured. The species with the
highest number of individuals captured was Aphthona perminuta Baly (635),
followed by S. tarsatum (201), Exosoma akkoae (Chujo) (92), Zipangia lewisi
(Jacoby) (54), Syneta adamsi Baly (51), B. acutangula (47), Cryptocephalus
amiculus Baly (30), Stenoluperus nipponemsis (Laboissiere) (29), Pseudoliprus
nigritus (Baly) (26), and Zeugophora annulata (Baly) (25).
The average number of species, the average number of individuals, and the
average H¢ in each forest type are shown in Table 1. The number of species and
H¢ in the larch plantation was significantly higher than those in the secondary
forest or in the primary forest (Bonferroni test, p < 0.05). The number of
individuals was not significantly different among three forest types.
In the larch plantation, the number of species and that of individuals were
significantly higher in the old plantation than in the middle-aged plantation
(Bonferroni test, p < 0.05, Table 2). H¢ was not significantly different among
three types of larch plantations.

Environmental factors

Among seven environmental factors, openness of the canopy, larch plantation

vs. natural broad-leaved forest, age of larch plantation, and the number of
plant species (<2 m in height) had a significant regression coefficient with the
number of chrysomelid species (Table 3). The species caught in larch planta-
tions were in higher numbers than in natural broad-leaved forests. Among
larch plantations, older plantations had more chrysomelid species. Chryso-
melid species were caught in higher number in stands with wider openness of
the canopy and with more plant species (<2 m in height). Other factors such as
gradient, altitude, and plant species (2 m in height) did not significantly affect
the number of species.

Table 1. Numbers of chrysomelid species, numbers of individuals, and diversity index (H¢) in
three forest types.

n Number of species Number of individuals2) H¢

(Mean1) ± SD) (Mean1) ± SD) (Mean1) ± SD)

Larch plantation 24 5.9a ± 2.1 28.6 ± 26.3 1.8a ± 0.5

Secondary forest 11 3.6b ± 2.1 19.2 ± 30.5 1.2b ± 0.9
Primary forest 11 3.6b ± 2.1 36.8 ± 41.0 0.8b ± 0.8
1)
For each comparison, different alphabetical letters indicate mean with a significant difference
(Bonferroni test, p < 0.05).
2)
Bonferroni test was performed after logarithmic transformation.

[123]
1184

Table 2. Numbers of chrysomelid species, numbers of individuals and diversity index (H¢) in three
types of larch plantations.

n Number of species Number of H¢

(Mean1) ± SD) individuals2) (Mean1) ± SD)
(Mean 1) ± SD)

Middle aged plantation 13 5.4a ± 2.1 19.0a ± 13.8 1.9 ± 0.6

Recently thinned plantation 6 5.2ab ± 1.3 21.8ab ± 10.4 1.9 ± 0.3
Old plantation 5 8.0b ± 1.2 61.8b ± 39.2 1.7 ± 0.7
1)
For each comparison, different alphabetical letters indicate mean with a significant difference
(Bonferroni test, p < 0.05).
2)
Bonferroni test was performed after logarithmic transformation.

Significant correlation coefficients were obtained by calculation between the

number of B. acutangula captured and the occurrence of Rubus spp., and
between the number of S. tarsatum and the occurrence of bamboo grass
(Table 4).

Species composition

The results of detrended correspondence analysis are shown in Figure 1. The

scores of the second axis were higher in the larch plantation than those in the
secondary forest or in the primary forest (Bonferroni test, p < 0.05). There
was no significant difference between the scores in the secondary forest and
those in the primary forest. A stand-structure parameter which possesses a
significant correlation with the first axis was H¢, and that with the second axis
was larch plantation vs. natural broad-leaved forest (Table 5). Correlations
between the other stand-structure parameters and the two axes were not
significant.

Table 3. Relationship between number of species and environmental factors (simple regression
analysis).

Environmental factors Standardized partial

regression coefficient

Openness of the canopy 0.39**

Altitude 0.13
Gradient 0.01
Larch plantation vs. Natural broad-leaved forest
0.49**
Age of foresta 0.50*
Number of vascular plant species (<2 m in height)b 0.74**
Number of vascular plant species (2 m in height)b 0.13
a
The analysis was applied for larch plantations.
b
The analysis was applied for 11 stands.
*p < 0.05; **p < 0.01.

[124]
 1185

Figure 1. Detrended correspondence analysis ordination (first and second axes) performed on leaf
beetles in 46 stands in central Japan.

Indicator species analysis showed that E. akkoae, B. acutangula, and

C. nobyi each had a significant indicator value (Ind Val) for the larch planta-
tion (Table 6). Longitarsus succineus (Foudras) and S. tarsatum showed a
significant Ind Val for the secondary forest and the primary forest, respectively.
Among three types of larch plantation, A. perminuta, S. adamsi, and
C. amiculus each had a significant Ind Val for the old larch plantation, and
E. akkoae had a significant Ind Val for the recently thinned plantation
(Table 7).

Discussion

Species richness

Species richness of leaf beetles in the larch plantation was significantly higher
than that in the secondary forest or in the primary forest. Nagaike (2002)

Table 4. Simple correlation coefficients for two chrysomelid speices between the number of
individuals and occurrence of host plants.

Host plants Correlation coefficients

S. tarsatum B. acutangula

Bamboo grass 0.69** –

Rubus spp. –a 0.33*
a
Not analyzed.
*p < 0.05; **p < 0.01.

[125]
1186

Table 5. Simple correlaton coefficients between scores of detrended corespondence analysis (first
two axes) and stand-structure parameters.

Parameter Axis 1 Axis 2

Number of leaf beetle species 0.22 0.25

Number of leaf beetle individuals
0.14
0.06
H¢ 0.31* 0.28
Larch plantation vs. Natural broad-leaved forest 0.07
0.53**
Openess of the canopy 0.12 0.28
Altitude 0.06
0.10
Gradient 0.06
0.09
Number of vascular plant species (<2 m in height)a 0.50 0.54
Number of vascular plant species (2 m in height)a
0.15 0.22
Age of treeb
0.02
0.21
a
For 11 stands.
b
For larch plantation.
*p < 0.05; **p < 0.01.

studied diversity of vascular plants in this area and reported that species
richness was significantly higher in larch plantations than in secondary forests.
This result in higher species richness of vascular plants coincides with higher
species richness of Chrysomelidae in the larch plantation than in the secondary
forest. In this study, species richness of leaf beetles was higher in stands with
higher species richness of understory plants (<2 m in height), though the
relationship between species richness of leaf beetles and species richness of the
plants (2 m) was not clear. Because leaf beetles feed on leaves, stems, or roots
of various plants, and because host plants of the beetles vary with the beetle
species, it seems that forests with more plant species enable chrysomelid species
to inhabit in them in higher numbers.
Among larch plantations, the old larch plantation had a higher species
richness of leaf beetles than did the middle-aged plantation. This result cor-
responds with the result of regression analysis that the higher number of
chrysomelid species was caught in older larch plantations. Species richness of
vascular plants was higher in the old plantation than in the middle-aged
plantation, though the difference was not significant (Nagaike et al. 2003).
They also found that plant species composition was different between the
middle-aged plantation and the old plantation. This difference may have
resulted in a higher species richness of Chrysomelidae in the old plantation
than in the middle-aged plantation. Further investigations will be necessary to
clarify this point. Thinning did not cause significant difference in species
richness of leaf beetles, though the recently thinned plantation had wider
openness of the canopy than the middle-aged plantation did (Bonferroni test,
p < 0.01).
S. tarsatum and B. acutangula had a significant Ind Val for the primary
forest and the larch plantation, respectively. However, more individuals of
both beetles were found in stands with higher occurrences of host plants. For

[126]
 1187

both species, significant correlations were ascertained by calculation between

the number of the beetles and the occurrence of the host plants. This, therefore,
seems to suggest that the numbers of both chrysomelid species are affected by
the occurrence of their host plants.

Table 6. Indicator value (Ind Val) of chrysomelid species for three forest types.

Species Ind Val Forest type

Exosoma akkoae (Chujo) 53.0* Larch plantation

Batophila acutangula Heikertinger 47.7* Larch plantation
Sphaeroderma tarsatum Baly 47.3* Primary forest
Calomicrus nobyi Chujo 45.8* Larch plantation
Aphthona perminuta Baly 42.5 Primary forest
Zeugophora annulata (Baly) 34.4* Larch plantation
Pseudoliprus nigritus (Baly) 29.2 Larch plantation
Longitarsus succineus (Foudras) 25.7* Secondary forest
Stenoluperus nipponensis (Laboissiere) 23.8 Primary forest
Syneta adamsi Baly 16.8 Primary forest
Zipangia lewisi (Jacoby) 16.0 Larch plantation
Cryptocephalus amiculus Baly 15.5 Larch plantation
Oomorphoides cupreatus (Baly) 14.5 Larch plantation
Stenoluperus cyaneus (Baly) 14.1 Secondary forest
Stenoluperus bicarinatus (Weise) 14.0 Primary forest
Sphaeroderma placidum Harold 12.5 Larch plantation
Basilepta balyi (Harold) 9.1 Secondary forest
Syneta brevitibialis Kimoto 9.1 Secondary forest
Hesperomorpha hirsuta (Jacoby) 9.1 Secondary forest
Cryptocephalus confusus Suffrian 9.1 Secondary forest
Chrysomelidae sp.1 9.1 Primary forest
Aspidomorpha indica Boheman 9.1 Primary forest
Oomorphus japanus Jacoby 8.3 Larch plantation
Lochmaea capreae (Linnaeus) 8.3 Larch plantation
Agelasa nigriceps Motschulsky 8.3 Larch plantation
Calomicrus cyaneus (Jacoby) 6.8 Secondary forest
Cryptocephalus fortunatus Baly 6.2 Primary forest
Asiorestia obscuritarsis (Motschulsky) 6.2 Secondary forest
Zeugophora unifasciata (Jacoby) 6.2 Secondary forest
Lilioceris lewisi (Jacoby) 4.2 Larch plantation
Pyrrhalta annulicornis (Baly) 4.2 Larch plantation
Zeugophora bicolor (Kraatz) 4.2 Larch plantation
Lema cirsicola Chujo 4.2 Larch plantation
Chaetocnema concinna (Marsham) 4.2 Larch plantation
Chaetocnema ingenua (Baly) 4.2 Larch plantation
Lipromela minutissima (Pic) 4.2 Larch plantation
Sphaeroderma nigricolle Jacoby 4.2 Larch plantation
Japonitata nigrita (Jacoby) 4.2 Larch plantation
Sangariola punctatostriata (Motschulsky) 4.2 Larch plantation
Taumacera tibialis (Jacoby) 4.2 Larch plantation
Chrysolina virgata (Motschulsky) 4.2 Larch plantation
*p < 0.05.

[127]
1188

Table 7. Indicator value (Ind Val) of chrysomelid species for larch plantation types.

Species Ind Val Plantation type

A. perminuta 71.5* Old plantation

S. adamsi 69.1* Old plantation
E. akkoae 67.3* Recently thinned plantation
C. amiculus 45.7* Old plantation
S. tarsatum 38.2 Old plantation
S. nipponensis 36.2 Old plantation
Z. annulata 31.6 Recently thinned plantation
B. acutangula 24.3 Recently thinned plantation
P. nigritus 23.3 Middle-aged plantation
C. nobyi 23.1 Recently thinned plantation
P. annulicornis 20.0 Old plantation
Z. bicolor 20.0 Old plantation
C. fortunatus 20.0 Old plantation
L. minutissima 20.0 Old plantation
S. nigricolle 20.0 Old plantation
J. nigrita 20.0 Old plantation
S. punctatostriata 20.0 Old plantation
L. succineus 20.0 Old plantation
O. cupreatus 20.0 Middle-aged plantation
Z. lewisi 18.1 Old plantation
L. cirsicola 16.7 Recently thinned plantation
C. concinna 16.7 Recently thinned plantation
A. obscuritarsis 16.7 Recently thinned plantation
S. bicarinatus 15.4 Middle-aged plantation
C. cyaneus 15.4 Middle-aged plantation
L. capreae 14.4 Old plantation
A. nigriceps 14.4 Old plantation
O. japanus 11.3 Old plantation
S. cyaneus 11.3 Old plantation
S. placidum 9.3 Old plantation
C. ingenua 7.7 Middle-aged plantation
L. lewisi 7.7 Middle-aged plantation
T. tibialis 7.7 Middle-aged plantation
Z. unifasciata 7.7 Middle-aged plantation
C. virgata 7.7 Middle-aged plantation
*p < 0.05.

Species composition

As the results of DCA showed that scores in the second axis was higher in the
larch plantation than in the secondary forest or in the primary forest, the
composition of Chrysomelidae in the larch plantation differed from that in the
secondary forest or in the primary forest.
Six chrysomelid species had significant Ind Vals in relation to the three forest
types in this study. E. akkoae, B. acutangula, and C. nobyi each showed a
significant Ind Val for the larch plantation. B. acutangula and C. nobyi feed on
Rubus spp. and Clematis apiifolia DC., respectively (Kimoto and Takizawa

[128]
 1189

1994). The host plants for E. akkoae are unknown. L. succineus having
a significant Ind Val for the secondary forest feeds on Artemisia spp. S. tarsatum
had a significant Ind Val for the primary forest and feeds on bamboo grass. It is
interesting that host plants of these chrysomelid species having significant Ind
Vals for certain forest types in this study were not dominant tall trees, but
understory herbs.

Comparison with a saproxylic family

Cerambycid beetles, though closely related to Chrysomelidae, are mostly

saproxylic in their larval stage. Cerambycidae had higher species richness in the
secondary forest than in the larch plantation or in the primary forest (Ohsawa
2004). On the other hand, more species of Chrysomelidae inhabited in the larch
plantation than in the secondary forest or in the primary forest. Among forest
management practices for larch plantations, long-rotations brought about high
species richness for Chrysomelidae, whereas thinning increased species richness
for Cerambycidae (Ohsawa 2004). It seems that ecological differences between
these families, i.e., saproxylic (Cerambycidae) and phytophagous herbivores
(Chrysomelidae), can be attributed to the difference in species richness with
these forest types and forest management practices. The amount of dead wood
or decayed wood affects the species richness of saproxylic beetles (Martikainen
et al. 2000), and the secondary forest (unattended) has more amount and
variety of dead or decayed wood than does the larch plantation (well managed)
(Green and Peterken 1997; Kirby et al. 1998; Fridman and Walheim 2000). The
recently thinned larch plantation contains more newly-made stumps and
woody debris as compared with the middle-aged plantation. Consequently,
there are more cerambycid species in the secondary forest and in the recently
thinned larch plantation (Ohsawa 2004). On the other hand, the larch plan-
tation had higher chrysomelid species richness than did the secondary forest or
the primary forest, probably because leaf beetles are phytophagous herbivores
and larch plantations have higher species richness of the plants. In this study,
higher species richness of understory plants positively influenced the species
richness of Chrysomelidae, and some understory plants were important hosts
of leaf beetles characteristic to each forest type. More precise attention should
be paid to understory plants for diversity conservation of leaf beetles, phy-
tophagous herbivores, than for that of saproxylic beetles.

Conclusion

The larch plantation had a higher species richness of Chrysomelidae than did
other two natural forests. However, the composition of chrysomelid assem-
blage in the larch plantation was different from that in the primary forest which
is considered to have indigenous species in the area. Among the three types of

[129]
1190

larch plantation, the species richness of Chrysomelidae was higher in the old
plantation than in the middle-aged plantation. Long rotations of larch plan-
tations seem favorable for chrysomelid diversity.
No difference in chrysomelid species composition between the secondary
forest and the primary forest could be detected by DCA. Since the primary
forest which is thought to contain indigenous species has diminished and
fragmented, the secondary forest is expected to maintain these leaf beetles.
However, considering that the secondary forest contained only 43% (6/14
species) of species caught in the primary forest in this study, the forest type may
not be sufficient to maintain the beetle assemblage in the primary forest.
This study showed that species richness of understory plants is an important
factor for chrysomelid species richness, and that the frequency of host occur-
rence significantly affects the number of individuals. Because forest types affect
plant diversity (Qian et al. 1997; Nagaike 2002; Nagaike et al. 2003), it is likely
that both forest types and forest management practices affect host plants as
well as Chrysomelidae, and that these effects on the host plants also influence
chrysomelid assemblages.

Acknowledgements

We wish to thank Mr. Akihiro Ohashi, Gifu Forest Science Institute, and
Dr. Masahiro Isono, Forestry and Forest Products Research Institute, Japan,
for identifying several leaf beetles and helpful advice. We are also grateful to
Mr. Masao Osawa for correction of English for the manuscript and helpful
advice. We deeply appreciate technical assistance given me by Mrs Miki Saso,
and the helpful advice and encouragement by members of Yamanashi Forest
Research Institute.

References

Dufrêne M. and Legendre P. 1997. Species assemblages and indicator species: the need for a flexible
asymmetrical approach. Ecol. Monogr. 67: 345–366.
Fridman J. and Walheim M. 2000. Amount, structure, and dynamics of dead wood on managed
forestland in Sweden. Forest Ecol. Manage. 131: 23–36.
Greatorex-Davies J.N. and Sparks T.H. 1994. The response of Heteroptera and Coleoptera species
to shade and aspect in rides of coniferised lowland woods in Southern England. Biol. Conserv.
67: 255–273.
Green P. and Peterken G.F. 1997. Variation in the amount of dead wood in the woodlands of the
Lower Wye Valley, UK in relation to the intensity of management. Forest Ecol. Manage. 98:
229–238.
Kimoto S. and Takizawa H. 1994. Leaf Beetles (Chrysomelidae) of Japan. Tokai University Press,
Tokyo.
Kirby K.J., Reid C.M., Thomas R.C. and Goldsmith F.B. 1998. Preliminary estimates of fallen
dead wood and standing dead trees in managed and unmanaged forests in Britain. J. Appl. Ecol.
35: 148–155.

[130]
 1191

Magurran A.E. 1988. Ecological Diversity and its Measurement. Princeton University Press,
Princeton, NJ.
Martikainen P., Siitonen J., Punttila P., Kaila L. and Rauh J. 2000. Species richness of Coleoptera
in mature managed and old-growth boreal forests in southern Finland. Biol. Conserv. 94:
199–209.
McCune B. and Mefford M.J. 1999. PC-ORD. Multivariate Analysis of Ecological data, version 4.
MjM Software Design, Gleneden Beach, Oregon.
Nagaike T. 2002. Differences in plant species diversity between conifer (Larix kaempferi) planta-
tions and broad-leaved (Quercus crispula) secondary forests in central Japan. Forest Ecol.
Manage. 168: 111–123.
Nagaike T., Hayashi A., Abe M. and Arai N. 2003. Differences in plant species diversity in
Larix kaempferi plantations of different ages in central Japan. Forest Ecol. Manage. 183:
177–193.
Ohsawa M. 2004. Species richness of Cerambycidae in larch plantations and natural broad-leaved
forests of the central mountainous region of Japan. Forest Ecol. Manage. 189: 375–385.
Qian H., Klinka K. and Sivak B. 1997. Diversity of the understory vascular vegetation in
40 year-old and old-growth forest stands on Vancouver Island, British Columbia. J. Veg. Sci. 8:
773–780.
ter Steege H. 1993. HEMIPHOTO, A Programme to Analyze Vegetation Indices, Light Quality
from Hemispherical Photographs. The Tropebos Foundation, Wageningen.

[131]
Biodiversity and Conservation (2006) 15:1193–1217
Springer 2006
DOI 10.1007/s10531-004-8230-8

-1

The banana forests of Kilimanjaro: biodiversity and

conservation of the Chagga homegardens

ANDREAS HEMP
Department of Plant Physiology, Universität Bayreuth, Universitätsstr. 30, 95440 Bayreuth (e-mail:
[email protected]; phone: 0921-552630; fax: 0921-552642)

Received 18 May 2004; accepted in revised form 7 December 2004

Key words: Agroforestry, Apophytes, Coffee–banana planations, East Africa, Epiphytes,

Kilimanjaro, Neophytes, Ruderal vegetation, Tropical montane forest

Abstract. Natural flora, vegetation, diversity and structure of 62 traditional coffee–banana plan-
tations on Kilimanjaro were investigated and compared with the other vegetation formations on
this volcano on basis of over 1400 plots following the method of Braun-Blanquet. The vegetation of
the so-called Chagga homegardens belongs floristically to the formation of ruderal vegetation
forming two main communities that are determined by altitude. These coffee–banana plantations
maintain a high biodiversity with about 520 vascular plant species including over 400 non-culti-
vated plants. Most species (194) occurring in the Chagga homegardens are forest species, followed
by 128 ruderal species, including 41 neophytes. Typical of the agroforestry system of the Chagga
homegardens is their multilayered vegetation structure similar to a tropical montane forest with
trees, shrubs, lianas, epiphytes and herbs. Beside relicts of the former forest cover, which lost most
of their former habitats, there are on the other hand (apophytic) forest species, which were directly
or indirectly favoured by the land use of the Chagga people. High demand of wood, the intro-
duction of coffee varieties that are sun-tolerant and low coffee prizes on the world marked endanger
this effective and sustainable system.

Introduction

Humans have continuously inhabitated the slopes of Mt. Kilimanjaro for the
last 2000 years (Schmidt 1989). However, during the last decades the human
population increased dramatically. According to ethnographic studies of
Widenmann (1899) 50,000–60,000 Chagga people were estimated to live on Mt.
Kilimanjaro in 1895. In 2002 the census counted 1,053,204 people (National
Bureau of Statistics 2003). As such, the population has multiplied 20 times
since 1895. Most of the population is concentrated at an altitude between 1000
and 1800 m, with densities varying from 500 to 1000 people per km2 in some
areas (FAO 1986; Timberlake 1986). Here, within the submontane zone, a very
remarkable kind of land use prevails: dense ‘banana forests’ with a scattered
upper tree layer, the so-called Chagga homegardens, in Chagga language
‘vihamba’ (the term ‘homegarden’ refers to the small size and subsistence-level
of the farms, cp. Nair 1993). Due to this sustainable and well developed
agroforestry system (cp. e.g. Fernandes et al. 1984) degradation in this vege-
tation belt is rare, despite the enormous population. In their homegardens the

[133]
1194

Chagga use four vegetation layers. Under a tree layer, which provides fire-
wood, fodder and shadow, banana trees (in about 25 varieties, cp. Simmonds
1966) are grown and under the bananas coffee trees, and under these vegeta-
bles. This multilayer system maximizes the use of limited land. The area is
irrigated by a network of canals fed by main furrows originating from the
montane forest. Rough estimates give over one thousand furrows of varying
lengths and capacities (Ramsay 1965) some dating back to the 17th century
(Hemp and Winter 1999). This farming system evolved over several centuries
and did not change much over the last decades compared with the land uses in
the lower zones.
This densely populated coffee–banana belt stretches on the climatically most
favourable zone of the southern and south-eastern slopes (Figure 1) over an
area of 1000 km2. The Chagga live within their homegardens in single dwell-
ings; villages as such do not exist. But, along the main roads centres with
church, village council, schools and some shops are situated. The average size
of a homegarden varies between 0.5 and 1.7 ha (O’Kting’ati and Kessy 1991;
Mdoe and Wiggins 1997). Livestock – cattle, goats, sheep and pigs – and
sometimes poultry are kept in stalls. Women and children spend a great part of
the day collecting grass along paths, fields and forest edges and on steep
meadow slopes. Pasture farming is rare in submontane zone due to intensive
agriculture. Therefore (based on the nature of their components) the Chagga
homegardens can be classified as an agrisilvicultural system (cp. Nair 1993).
Bee-keeping plays an important role.
The chagga homegardens were the subject of different studies dealing mainly
with socio-economic (Clemm 1963; Brewin 1965; Fernandes et al. 1984;
O’Kting’ati and Kessy 1991; Mdoe and Wiggins 1997) or ethnobotanical and
ethnozoological aspects (O’Kting’ati et al. 1984; Hemp 1999; Hemp 2001). The
array of cultivated species was already described in detail by the first scientists
on Kilimanjaro e.g. Volkens (1897) or Widenmann (1899). The aim of the
present study is to describe natural flora, vegetation and structure of the
Chagga homegardens and to highlight their function for biodiversity and as a
refuge area for natural plant species. For this purpose the species composition
of this man-made habitat is compared with all vegetation formations of Mt.
Kilimanjaro as presented by Hemp (2001a). In a parallel study the function of
the Chagga homegardens as a habitat of endangered and endemic grasshopper
species was investigated on the same plots (Hemp in press).

Study area

Mt. Kilimanjaro, a relic of an ancient volcano, rising from the savanna plains
at 700 m elevation to a snow-clad summit of 5895 m altitude is located 300 km
south of the equator in Tanzania on the border with Kenya. Its climate is
characterized by a bimodal rainfall pattern with the long rains from March to
May forming the main rainy season, and the short rains centred around the

[134]
 1195

Figure 1. Land use and vegetation cover of the study area with location of plots (dots).

month of November of the small rainy season. The foothills of the southern
slopes receive an annual rainfall of 800–900 mm and the lower slopes at
1500 m receive 1500–2000 mm. The forest belt between 2000 and 2300 m re-
ceives partly over 3000 mm (Hemp 2001a), which is more than on other high
mountains of East Africa. In the alpine zone the precipitation decreases to
200 mm.
According to the different climatic conditions several vegetation zones are
apparent on the southern slopes of Mt. Kilimanjaro (Figure 1). Between 700
and 1000 m a.s.l. the dry and hot colline savanna zone stretches around the
mountain base, where most areas are farmed with maize, beans and sunflowers,
in West Kilimanjaro with wheat. Around Lake Chala at the eastern foot of the
mountain, and around Ngare Nairobi of West Kilimanjaro, savanna grass-
lands are still intact. The main cultivation zone with its coffee–banana plan-
tations, the actual study area, is located between 1000 and 1800 m. Natural
forests cover an area of about 1000 km2 on Mt. Kilimanjaro. In the lower parts
of the southern slope the montane forests are characterized by the tree Ocotea
usambarensis and higher up in the cloud forest zone by Podocarpus latifolius,
Hagenia abyssinica and Erica excelsa. On the drier northern slope the vege-
tation zonation starts with Croton-Calodendrum forests, Cassipourea forests at
midaltitudes and Juniperus forests at higher altitudes. At around 3100 m the
forests are replaced by Erica bush. At an altitude of about 3900 m the Erica
heathlands grade into Helichrysum cushion vegetation that reaches up to
4500 m. Higher altitudes are very poor in vegetation while the highest

[135]
1196

elevations of Kibo peak are covered with glaciers. For a more detailed
description of these vegetation types see Hemp (2001a).

Methods

Data have been collected since 1996. Over 1400 relevés (plots) of all vegetation
types were produced on the whole mountain using the method of
Braun-Blanquet (1964), 62 of them representing banana gardens (location see
in Figure 1). Special attention was given to homogeneity and representation of
the stands. The relevé size was chosen with respect to the minimum area, which
was 1000 m2 in the banana gardens.
In the relevés a herb layer (<1 m tall), a shrub layer (1–10 m tall) and a tree
layer (>10 m tall) were differentiated. Climbers (stranglers, scramblers,
tendrillar plants, hook and root climbers and twiners, cp. Lind and Morrison
1974) reaching into the shrub and tree layer were defined as lianas. Epiphytes
(accidental, casual, hemi- and holo-epiphytes, cp. Kress 1986) were treated as
an additional layer. To portray the structure of the banana gardens, a stand-
profile diagram according to Hammen et al. (1989) was produced.
The relevés were clustered according to floristic similarity and the resulting
plant communities were united into nine formations: rocks; ruderal vegetation;
vegetation of trampled grounds; grasslands; salt marshes; freshwater swamps;
forest clearings; forests, and; heathlands. Constancy tables of these formations
(except the vegetation of trampled ground) are presented by Hemp (2001a).
Based on their constancy in the different formations (or vegetation classes)
character species of these formations were determined.
pH was measured in the main root horizon in 5–10 cm depth of selected
plots using a WTW pH-metre (pH 330). Two parallel samples were taken and
measured in distilled water and 0.01 M CaCl2 solution, respectively.
Nomenclature follows FTEA (1952–2003) and (Beentje 1994).

Results

Based on Landsat ETM images from the year 2000 (source: USGS/UNEP-
GRID-Sioux Falls) extend and distribution of the Chagga homegardens on
Kilimanjaro was determined. Excluding other vegetation types of the sub-
montane cultivated zone (e.g. riverine forests, meadows, maize fields etc.)
banana fields cover an area of about 675 km2, which is about two thirds of the
extend of the montane forest belt.

Phytosociological and ecological aspects

The vegetation of the Chagga homegardens belongs floristically to the for-

mation (or vegetation class) of ruderal vegetation (Table 1, cp. Table 1 in

[136]
[137] Table 1. Vegetation of the Chagga homegardens.

1197
 Table 1. Continued.

1198
[138]
 1199
[139]
1200

Hemp 2001a). Ruderal species form in respect of species number and vegeta-
tion cover of the herb layer the most important ecological group. Widespread
ruderal species (character species of the vegetation class) that can be found as
well on fallow arable land, waste places in towns and roadsides from the colline
savanna area up to the montane zone are e.g. Bidens pilosa, Oxalis corniculata,
Commelina benghalensis and Galinsoga parviflora. Phyllanthus odontadenius,
Didymodoxa caffra and Celosia schweinfurthiana occur in different ruderal
vegetation types within the submontane zone. Finally, a characteristic weed
species confined to the banana gardens on Kilimanjaro is Oxalis latifolia.
Based on the floristic composition, two distinct plant communities were
distinguished. Diagnostic species of community 1 are the ruderal species
Euphorbia heterophylla, Amaranthus hybridus, Sida acuta, Blainvillea gayana
and Malvastrum coromandelianum. Impatiens walleriana, Drymaria cordata and
Plectranthus parvus are diagnostic for community 2; the latter species is re-
stricted in the study area to this community. Each of these two main com-
munities can be divided into a vicarious vegetation unit of the southern and the
(north-) eastern slope, respectively, governed by the rainfall regime. This
partition into a wet southern and drier northern slope is also reflected by other
vegetation formations such as forests (cp. Hemp in press a).
Determining factors for the differentiation of the two main communities 1
and 2 are altitude and factors related to altitude: community 1b of the southern
slope occurs between 800 m and about 1300 m, receiving 900–1580 mm rain-
fall, and community 2b between 1300 and 1800 m, receiving 1580–2200 mm
rainfall. Mean annual temperature varies between 23.4 and 18.8 C in com-
munity 1b and between 18.8 and 16.1 C in community 2b (climate data from
Hemp in press a). As a result of the increasing precipitation mean pH-values
decrease from 6.4 (CaCl2) and 6.7 (H2O), respectively in community 1b to 5.5
(CaCl2) and 6.0 (H2O), respectively in community 2b. As in the forests of
Kilimanjaro (Hemp in press a, b) the influence of precipitation is reflected by
the distribution of epiphytes and pteridophytes. Vascular epiphytes are nearly
missing in community 1b and start above 1300 m in community 2b. The same
holds for pteridophytes. Mean cover values of epiphytes (including mosses and
lichens) in both communities are 0 and 8%, respectively.
On the drier eastern slope the lowland community 1a occurs between 1100
and about 1500 m, and community 2b is less distinctly developed than on the
wetter southern slope, occurring only in a narrow strip above 1500 m. Such a
rising and condensation of vegetation zones due to the ‘Mass-
enerhebungseffekt’ was also observed in the forest vegetation on the dryer
leeward slope of Kilimanjaro (Hemp in press a).

Diversity

Five hundred and twenty-three vascular plant species were recorded in the
relevés. These are about three quarter of the species occurring in the ruderal

[140]
 1201

Figure 2. Vascular plant species richness in the main vegetation formations (as presented with
constancy tables by Hemp 2001a) of Mt. Kilimanjaro, based on the evaluation of about 1400
vegetation plots.

vegetation formation on Kilimanjaro (Figure 2). With over 700 species this
formation holds rank three in respect of species richness after the forests and
grasslands. Mean vascular plant species number per plot is 54.
Figure 3 shows the floristic composition of the banana fields in relation to
the different vegetation formations on Kilimanjaro. Most species (194)
occurring in the Chagga homegardens are forest species, followed by 128
ruderal species. Cultivated plants contribute 19% to this floristic spectrum with
99 species.
A characteristic feature of the ruderal vegetation on Kilimanjaro is the high
contribution of neophytes. In the Chagga homegardens nearly a quarter of the
species is introduced, with 99 cultivated species and 41 neophytes (Figure 4,
Table 2).

Vegetation structure

Figure 5 shows a vegetation profile of a Chagga homegarden in the area of

Kidia (Old Moshi). Typical of the agrisilvicultural system of the Chagga ho-
megardens is their multilayered vegetation structure similar to a tropical
montane forest. Therefore the growth form spectrum (Figure 6) displays beside
herbs also trees, shrubs, lianas and epiphytes. Apart from some cultivated fruit
trees, e.g. Persea americana, Mangifera indica and Syzygium cumini or intro-
duced timber trees such as Grewillea robusta and Cupressus lusitanica, most of
the 55 encountered tree species (including young trees of the shrub and herb

[141]
1202

Figure 3. Floristic composition of the banana fields in respect of the different vegetation formations
on Kilimanjaro.

Figure 4. Share of cultivated, neophytic and indigenous plants in the Chagga homegardens.

layer this number adds to 82) are remnants of the former forest cover (Table 3).
Most widespread are Albizia schimperiana, Rauvolfia caffra, Cordia africana,
Commiphora eminii and Margaritaria discoidea. Nearly all banana fields are

[142]
 1203

Table 2. Neophytes of the Chagga home gardens.

Name Frequency

Ageratrum conyzoides 57
Oxalis corniculata 45
Galinsoga parviflora 39
Oxalis latifolia 38
Euphorbia heterophylla 33
Conyza sumatrensis 32
Amaranthus hybridus ssp. hybridus 31
Tagetes minuta 22
Galinsoga quadriradiata 20
Solanum nigrum 20
Richardia scabra 14
Nicandra physalodes 12
Lantana camara 8
Tridax procumbens 8
Adiantum raddianum 7
Euphorbia hirta 7
Bryophyllum pinnatum 5
Lepidium bonariense 5
Senna occidentalis 5
Stachytarpheta jamaicensis 5
Galium aparine 4
Ipomoea hederifolia 4
Lagascea mollis 4
Lantana camara 4
Sonchus oleraceus 4
Acanthospermum hispidum 3
Senna bicapsularis 3
Apium leptophyllum 2
Argemone mexicana 2
Caesalpinia decapetala 2
Chenopodium ambrosioides 2
Tradescantia zebrina 2
Acmella oleracea 1
Brassica juncea 1
Crambe hispanica 1
Datura stramonium 1
Ipomoea nil 1
Sambucus nigra 1
Senna septentrionalis 1
Solanum seaforthianum 1
Spermacoce assurgens 1

covered by at least some trees (Figure 7); mean coverage of the tree layer is
28%, mean species number 4.5.
Twenty-nine species were found growing epiphytically in the plots (Table 4).
Twenty-two were holo-epiphytes, mainly restricted in their occurrence to the
epiphyte layer of the forests on Kilimanjaro (cp. Hemp 2001b). Most wide-

[143]
1204

Figure 5. Profile (27 · 2.5 m) and ground plan (27 · 5 m; bold lines indicate the area used for the
profile) of a typical Chagga homegarden in Kidia (Old Moshi) at 1400 m a.s.l. Exposition: south
west, inclination: 25. An open light upper canopy is formed by Albizia schimperiana var. amani-
ensis, on which epiphytes such as the fern Drynaria volkensii and Telphairia pedata, a liana with oil-
containing seeds, find habitats. Bananas form a dense upper shrub layer of 4–6 m height, coffee
trees a lower shrub layer of 1.5–2 m, intermingled with 1–1.5 m high Coco Yam (Colocasia
esculenta). The lower side of the banana field borders a road; here Dracaena fragans is planted as a
hedge.

spread epiphytes are the petridophytes Pleopeltis macrocarpa (growing quite

often on coffee trees), Asplenium aethiopicum, A. theciferum, Lepisorus excav-
atus and Drynaria volkensii. Vascular epiphytes start to appear above 1300 m
in community 2 (see above).
Forty-one climbers reaching into the shrub and tree layer were found in the
plots (Table 5). Including young climbers found in the herb layer this number
adds to 52 species. Eleven species were cultivated plants with important

[144]
 1205

Figure 6. Growth form spectrum of the Chagga homegardens; (a) species number of the respective
stratum in the vegetation plots; (b) species number of all representatives of a growth form, e.g. of
trees including young trees occurring in the shrub and herb layer or e.g. of herbs excluding young
trees etc.

agricultural crop plants such as three Dioscorea and Passiflora species and the
Cucurbitaceae Telphairia pedata belonging to this growth form.
One hundred twenty-six species were encountered in the shrub layer of the
plots. Excluding young trees, 63 species remain (Table 6). Similar to the trees,
epiphytes and lianas most of the shrubs in the Chagga homegardens were forest
species. However, in the shrub layer the most important cultivated plants oc-
curred: Different varieties of Musa · sapientium (dessert bananas) and
M. · paradisiaca (cooking bananas) and Coffea arabica. Bananas form a dense
(mean cover value 50%) upper shrub layer of about 4–6 m height and coffee
trees a lower layer of 1.5–2 m. Already Volkens (1897) and Widenmann (1899)
reported that the bananas on Kilimanjaro are most luxuriant with heights of 6–
8 m in the area of Kibosho and Kilema. This may be due to the fact that the
bedrocks in these areas of the wet central southern slope consist of rhomb
porphyry instead of porous tuff and ashes as in the adjacent regions of the
southern slope (Downie and Wilkinson 1972) and that the eastern slope
receives less precipitation.
Four hundred nine species were found in the herb layer of the homegardens.
Excluding young trees, shrubs and lianas 304 herbs and grasses occurred in the
plots. In contrast to the other strata the main species group of the herb layer
consisted of representatives of ruderal vegetation. This is shown in Figure 8. In
this figure species of grasslands and wet habitats are lumped together into
‘semi-natural’ species, as such habitats are mostly influenced by cutting and
grazing of cattle. It appears that the tree and epiphyte layer are the most

[145]
1206

Table 3. Trees of the Chagga home gardens.

Name Frequency

Grevillea robusta t 37
Albizia schimperiana 36
Persea americana f 21
Rauvolfia caffra 18
Cordia africana 14
Mangifera indica f 13
Commiphora eminii ssp. zimmermannii 11
Margaritaria discoidea 9
Markhamia lutea 8
Bridelia micrantha 7
(Citrus aurantium) f 7
Croton macrostachyus 7
Albizia petersiana 6
Alangium chinense 5
Ficus sur 5
Casearia battiscombei 4
Ficus thoningii 4
Jacaranga mimosiaefolia t 4
Milicia excelsa 4
Syzygium cumini f 4
Turraea robusta 4
Albizia gummifera 3
(Annona muricata) f 3
Annona reticulata f 3
Artocarpus heterophyllus f 3
(Azaridachta indica) x 3
(Cassia siamea) o 3
Cassia spectabilis o 3
(Citrus limon) f 3
Cupressus lusitanica t 3
Cussonia holstii 3
Eriobotrya japonica f 3
Ficus vallis-choudae 3
Olea capensis ssp. welwitschii 3
Trema orientalis 3
Croton megalocarpus 2
(Deinbollia kilimandscharica) 2
Ficus lutea 2
Prunus africana 2
Trichilia emetica 2
(Allophylus abyssinicus) 1
(Bersama abyssinica) 1
(Bombacopsis glabra) 1
Ceiba pentandra x 1
Celtis africana 1
(Celtis mildbraedii) 1
Cinnamomum verum x 1
(Citrus paradisi) f 1
Citrus sinensis f 1

[146]
 1207

Table 3. Continued.

Name Frequency
(Clausena anisata) 1
Cordia goetzei 1
Diospyros abyssinica 1
(Ehretia cymosa) 1
Ekebergia capensis 1
Erythrina abyssinica 1
Eucalyptus spec. t 1
(Euclea divinorum) 1
Ficus exasperata 1
Kaya anthotheca t 1
Lecaniodiscus fraxinifolius 1
((Lepidotrichilia volkensii)) 1
(Macaranga capensis) 1
(Maesopsis eminii) t 1
Manihot glaziovii x 1
(Maytenus undata) 1
Melia azedarach x 1
(Mystroxylon aethiopicum) 1
(Neoboutonia macrocalyx) 1
((Ochna holstii)) 1
(Olea europaea ssp. africana) 1
Phoenix reclinata 1
(Polyscias fulva) 1
Psidium guajava f 1
Rapanea melanophloeos 1
((Sorindeia madagascariensis)) 1
Synadenium cf. volkensii 1
Syzygium cordatum 1
Syzygium guineense 1
(Tabernaemontana stapfiana) 1
(Tabernaemontana ventricosa) 1
Teclea nobilis 1
(Vepris spec.) 1
f = fruit tree; o = ornamental tree; t = timber tree; x = introduced tree of other uses; no label:
indigenous tree; in brackets = only found in the shrub layer; in two brackets = only found as
young tree in the herb layer.

natural vegetation strata in the Chagga homegardens whereas the herb layer is
dominated by (128) ruderal species followed by 117 forest species and 64
cultivated plants. Plants of the semi-natural grasslands and wet habitats
contributed 51 and 15 species, respectively.

Discussion

The presented distinction of two main communities is confirmed by the com-

position of Saltatoria species of the banana gardens, which show a parallel

[147]
1208

Figure 7. Typical Chagga homegarden in Kidia (Old Moshi) with an open tree canopy and a dense
banana undergrowth.

distinction of two coenoses (Hemp, C. unpub. data). The congruence of veg-

etation communities and Saltatoria coenoses was already proofed on Kili-
manjaro by grassland communities (Hemp and Hemp 2003).
The Chagga homegardens maintain a high biodiversity with over 400 not
cultivated plants. However, their majority serves as forage, for household and
agricultural purposes, for medical applications, as drugs and for magic pur-
poses (Hemp 1999). Most areas of the submontane coffee–banana belt
resemble a woodland with a dense undergrowth of bananas (Figures 5 and 7).
Thus, 193 forest species were found in the studied plots, species that need a
forest-like habitat structure for surviving. These are about 17% of the 1155
forest plants of Kilimanjaro (30% of the forest trees – including young trees
45%—and 17% of the forest epiphytes, cp. Hemp in press a). Compared with
large scale coffee plantations this conserving function becomes evident: Four
surveyed commercial plantations harboured only 6 forest species, and three
quarter of the species were widespread ruderal or cultivated species (Hemp
unpub. data). This holds not only for plants but also for insects such as
Saltatoria. The Chagga homegardens serve as a refuge not only for forest
grasshoppers, which contribute the majority of the species but also for
endemics (Hemp in press).
These findings are in line with the fact that biodiversity in general on Kili-
manjaro culminates at 1000–1300 m with over 900 vascular plant species
(Hemp in press c) inside the coffee–banana belt, the most densely populated
region of the mountain. This is due to the high variety of (moderately) culti-
vated areas (the Chagga homegardens), forest patches, river gorges and
grasslands at this altitude. This (mostly man-made) variety of habitats, the high

[148]
 1209

Table 4. Epiphytes of the Chagga home gardens.

Name Frequency

Pleopeltis macrocarpa 28
Asplenium aethiopicum 21
Asplenium theciferum 12
Lepisorus excavatus 8
Drynaria volkensiii 7
Peperomia tetraphylla 6
Drynaria volkensii 5
Rhipsalis baccifera 5
Impatiens walleriana 4
Phragmanthera usuiensis 4
Loxogramme abyssinica 3
Polystachia simplex 3
Erianthemum dregei 2
Loranthaceae spec. 2
Plicosepalus curviflorus 2
Rangaeris amaniensis 2
(Achyranthes aspera) 1
Aerangis amaniensis 1
Asplenium megalura 1
Asplenium sandersonii 1
Asplenium strangeanum 1
Chamaeangis sarcophylla 1
(Didymodoxa caffra) 1
(Dorstenia zansibarica) 1
(Drymaria cordata) 1
Ficus thoningii 1
(Pilea tetraphylla) 1
(Poa schimperana) 1
(Setaria homonyma) 1
in brackets: accidental epiphytes

beta-diversity, promotes alpha-diversity, allowing species from lower altitudes

to climb up the mountain. A similar phenomenon was observed in the Salta-
toria fauna of Kilimanjaro (Hemp and Hemp 2003).
From the forest species found in the banana gardens it can be assumed that
the forest, which covered the lower slopes of Kilimanjaro before human set-
tlement, resembled in some aspects the Cassipourea forests of the western and
northern slopes (cp. vegetation table in Hemp 2001a). However, due to the
much higher precipitation on the southern slope, major differences to these
forests can be expected; and above 1500 m probably elements of the camphor
forests constituted to their floristic composition. As nearly all of these former
submontane forests disappeared, it is difficult to reconstruct their full floristic
composition. Since humans have continuously inhabitated the lower slopes of
Mt. Kilimanjaro at least for the last 2000 years (Schmidt 1989), it can be
assumed that many submontane forest species were extinguished together with

[149]
1210

Table 5. Lianas of the Chagga home gardens.

Name Frequency

Dioscorea lecardii c 21
(Glycine wightii) 13
Dioscorea bulbifera c 12
Passiflora edulis c 8
Telphairia pedata c 7
Passiflora laurifolia c 6
Zehneria scabra 6
Toddalia asiatica 5
Ficus thoningii 5
Smilax anceps 4
Dioscorea alata c 4
Passiflora quadrangularis c 3
Solanecio angulatus 3
Thunbergia alata 3
Vigna spec. 3
Cyphostemma masukuense ssp-masukuense 3
Cyphomandra betacea 3
Clerodendron johnstonii 3
Basella alba 2
Caesalpinia decapetala 2
Cissus oliveri 2
Cyphostemma kilimandscharicum 2
Diplocyclos schliebenii 2
(Ipomoea obscura) 2
Momordica foetida 2
Pentarrhinum insipidum 2
(Stephania abyssinica) 2
(Teramnus labialis ssp. labialis) 2
Vigna membranacea 2
(Adenia gummifera) 1
Adenia spec. 1
(Begonia meyeri-johannis) 1
Clematis brachiata 1
Dahlbegia lactea 1
(Diplocyclos palmatus) 1
Embelia schimperi 1
Helinus mystacinus 1
Ipomoea hederifolia 1
Ipomoea indica 1
Ipomoea nil 1
Landolphia buchananii 1
(Merremia palmata) 1
(Merremia spec.) 1
Mondia whytei 1
Monstera spec. c 1
Oreosyce africana 1
Paullinia pinnata 1
Piper nigrum c 1
(Rourea thomsonii) 1
Rubus rosifolius 1
Sechium edule c 1
Solanum seaforthianum c 1
c = cultivated; in brackets = only found in the herb layer.

[150]
 1211

Table 6. Shrubs of the Chagga home gardens.

Name Frequency

Musa Sapientium / M. X paradisiaca 62

Coffea arabica 57
Dracaena fragrans i 29
Manihot esculenta 21
Carica papaya 18
Ricinus communis 14
Cyphomandra betacea 13
Dracaena steudneri 11
Morus alba 11
Prunus persica 10
Capsicum fruticosum 9
Adhatoda engleriana i 2
Lantana camara 8
Cajanus cajan 6
Cestrum nocturnum 6
Synadenium compactum i 5
Tetradenia riparia i 5
Colocasia esculenta 4
Euphorbia pulcherrima 4
Ochna insculpta i 4
Aloe volkensii i 3
Datura suaveolens 3
Elettaria cardamomum 3
Keetia gueinzii i 3
Rubus pinnatus i 3
Senna bicapsularis 3
Solanum spec. 3
Bauhinia tomentosa 2
Capsicum cf. baccatum 2
Dracaena steudneri i 2
Ensete edule i 2
Flacourtia indica i 2
Leucas grandis i 2
Phyllanthus ovalifolius i 2
Thevetia peruviana 2
Allophylus rubifolius i 1
Arundinaria alpina i 1
Bougainvillea spec. 1
(Cyathea manniana) 1
Dombeya rotundifolia i 1
Euclea racemosa i 1
Grewia bicolor i 1
Hibiscus calyphyllus i 1
Jatropha curcas 1
Lonchocarpus eriocalyx i 1
Mirabilis spec. 1
Oxyanthus speciosus ssp. globosus i 1
Psychotria lauracea i 1
Punica granatum 1
Rhus vulgaris i 1

[151]
1212

Table 6. Continued.

Name Frequency
Rubus rosifolius i 1
Sambucus nigra 1
Senna septentrionalis 1
Senna spec. 1
(Sesbania sesban) 1
Solanecio cydonifolius i 1
Solanecio mannii i 1
Sterculia cf. stenocarpa i 1
Tephrosia vogelii i 1
Vernonia lasiopus i 1
Vernonia myriantha i 1
Ximenia americana var. caffra i 1
Ziziphus mucronata i 1
i = indigenous species; in brackets = only found as young shrub in the herb layer.

Figure 8. Forest species, semi-natural species (i.e. species of grasslands and wet habitats), cultivated
and ruderal species (including species of trampled ground) with their share in the different strata of
the Chagga homegardens.

the forest cover. Only in deep inaccessible gorges remnants of floristically very
rich forests can be found, resembling the intermediate forests of the Eastern
Arc. Mts. (a chain of ancient crystalline mountains in East Africa), e.g. of the
Pare and Usambara Mts. Here many species, which were believed to be en-
demic to the Eastern Arc. Mts, were encountered on Kilimanjaro (Hemp 2002).
These findings suggest a rich forest flora inhabiting the southern slopes of Mt.
Kilimanjaro in former times. Thus, the lower diversity of the forest flora and

[152]
 1213

Figure 9. Records of Christella dentata on Mt. Kilimanjaro, at the base of the UTM grid. The scale
of the squares is 4 km2. This apophytic fern, which thrives naturally in riverine forests was able to
spread over the whole submontane banana plantation belt due to the network of over one thousand
furrows. Gaps within the distribution area inside the banana plantations (especially in its eastern
part) are mostly due to lack of data.

the lower share of endemic plant species of Kilimanjaro can be explained by the
widely destroyed submontane (intermediate) forest rather than by the higher
age and greater ecoclimatic stability of the Eastern Arc. Mts. as suggested by
e.g. Rodgers and Homewood (1982) or Fjeldså et al. (1997). This is corrobo-
rated by the fact, that forest inhabitants, which become less affected by forest
devastation like Saltatoria, have similar numbers of endemic forest species in
the submontane and montane zone on Mt. Kilimanjaro (including Mt. Meru)
and the East Usambara Mts. (Hemp, C. unpub. data). Endemic grasshopper
species like Ixalidium sjöstedti, Parepistaurus deses and Altiusambilla modicicrus
have coped with the habitat change from forest to plantations (Hemp and
Hemp 2003).
On the other hand there are forest species, which were directly or indirectly
favoured by the Chagga people. An example of an intentionally dispersed
forest plant is Dracaena fragrans. This shrub or small tree has lost almost all
natural habitats on Kilimanjaro, except very few submontane river gorges (cp.
Hemp in press a). However it is one of the most characteristic species in the
banana plantations, where it is used as a hedge plant (cp. Figure 5). Dracaena

[153]
1214

hedges are protected as they serve also as burial ground. Anthropogenic

influence does not only destroy natural habitats but sometimes it enlarges the
distribution of indigenous species by increasing habitat diversity. An example
of such apophytes sensu Rikli (1903) (i.e. indigenous species, which could
extend their natural distribution area due to human influence and occupy
anthropogenic habitats; for examples in the middle European flora cp. e.g.
Sukopp and Kowarik 1987; Sukopp and Langer 1996) is the fern Christella
dentata, a species of riverine forests in the colline and submontane zone of
Kilimanjaro. Its main habitat and main distribution area are nowadays the
coffee–banana plantations with their ramified irrigation system and forest-like
structure on the southern and eastern slopes (Figure 9). The same holds for
Impatiens walleriana and the fern Adiantum poiretii. Another apophyte is e.g.
Pellaea viridis, which naturally thrives on Kilimanjaro in submontane Croton-
Calodendrum forests (Hemp in press b). Some forest plants (e.g. Pilea tetra-
phylla) were only encountered in the banana plantations but in none of the 583
forest plots, highlighting the important conserving function of the Chagga
homegardens.
The agroforestry system of the Chagga homegardens is a unique feature of
Kilimanjaro, distributed over an area of about 1000 km2. If one passes from
north-east to the south-western end of this belt, one could drive for 120 km
through a closed ‘banana forest’ containing about 225 Million banana ‘trees’
(calculated on the base of the vegetation profile) – if there was a continuous
road. The same type of land use, however with a smaller extension, occurs on

Figure 10. Tschibo estate at Mweka, southern slope of Kilimanjaro in November 2003. Hundreds
of old trees were felled to grow new coffee varieties that do not need shade.

[150]
 1215

the Pare Mountains and Mt. Meru, which shows nearly exact the same floristic
and structural composition (Hemp unpub. data).
The Chagga homegardens maintain not only a high biodiversity, they are an
old and very sustainable way of land use that meets several different demands.
Beside crop production, the sparse tree layer provides people with fire wood,
fodder and timber. But the high demand of wood, low coffee prizes on the
world marked and the introduction of coffee varieties that are sun-tolerant
endanger this effective system (Hemp et al. in press). In some areas of the
mountain (e.g. on the eastern slopes) the trees in the banana fields are very
scattered or already missing. A very bad example in this respect was observed
at Mweka on the central southern slope: Here a large foreign coffee company
felled hundreds of old trees in November 2003 to grow coffee (Figure 10). In
order to reduce the pressure on the forest, it is necessary, to support the tree
planting in the Chagga homegardens with their unique agroforestry system.
Similar to environmental programmes for farmers in the European Union (e.g.
for the protection of wetlands or dry meadows), there should be a programme
that rewards farmers to have a certain share of their land covered by trees. It
can be estimated that a homegarden supplies 1/4 to 1/3 of the fuelwood
requirements of a family (Fernandes et al. 1984). As the banana belt is nearly as
extensive as the forest reserve, this will of course have major effects in terms of
forest protection and the water balance. In combination with new marketing
and farming strategies for growing organic coffee through traditional methods
an advertising campaign should be started especially in European countries
where the awareness of environmental problems is high. A concept with this
idea was already proposed to the OECD (Hemp 2003).

Acknowledgements

I gratefully acknowledge grants by the Deutsche Forschungsgemeinschaft and

the Tanzanian Commission for Science and Technology for permitting re-
search. For support in getting permits I owe gratitude to the Chief Park
Wardens of Kilimanjaro National Park, Mr Lomi Ole Moirana and Mr Ny-
amakumbati Mafuru, to the Catchment Forest officers and to my counterpart
Mr Jacob Mushi (Tanzania Association of Foresters), Moshi. I further thank
the keepers of the East African Herbarium, Nairobi (Kenya) Dr Beatrice
Khayota and Kew Herbarium, England, Prof Dr Owens for permission to
study their collections, Quentin Luke and Simon Mathenge (both Nairobi) for
help in identifying difficult species and Prof Dr Inge Lenski, Marburg (Ger-
many) for valuable comments on the manuscript.

References

Beentje H.J. 1994. Kenya Trees, Shrubs and Lianas. National Museums of Kenya, Nairobi.
Braun-Blanquet J. 1964. Pflanzensoziologie. Springer, Wien.

[151]
1216

Brewin D.R. 1965. Kilimanjaro agriculture. Tanganyika Notes Records 64: 115–117.
Clemm V.M. 1963. Agricultural productivity and sentiment on Kilimanjaro. Econ. Bot. 18: 99–121.
Downie C. and Wilkinson P. 1972. The Geology of Kilimanjaro. University of Sheffield.
FAO 1986. Rehabilitation of traditional schemes (Phase I), Tanzania AG: TCP/URT/4522, Mis-
sion Report, Rome.
Fernandes E.C.M., Okting’ati A. and Maghembe J. 1984. The Chagga homegardens: a multistoried
agroforestry cropping system on Mt. Kilimanjaro (Northern Tanzania). Agroforest. Syst. 2: 73–
86.
Fjeldså J., Ehrlich D., Lambin E. and Prins E. 1997. Are biodiversity ‘hotspots’ correlated with
current ecoclimatic stability? A pilot study using the NOAA-AVHRR remote sensing data.
Biodivers. Conserv. 6: 401–422.
FTEA 1952–2003. Flora of Tropical East Africa. Royal Botanic Garden, Kew.
Hammen T.V.D., Mueller-Dombois D. and Little M.A 1989. Manual of Methods for Mountain
Transect Studies. IUBS, Paris.
Hemp A. and Winter J.C. 1999. Ergebnisse der ethnobotanischen Forschung am Kilimanjaro.
Bayreuther Forum Ökologie 64: 117–144.
Hemp A. 1999. An ethnobotanical study on Mt. Kilimanjaro. Ecotropica 5: 147–165.
Hemp A. 2001a. Ecology of the pteridophytes on the southern slopes of Mt. Kilimanjaro. Plant
Biol. 3: 493–523.
Hemp A. 2001b. Life form and strategies of forest ferns on Mt. Kilimanjaro. In: Gottsberger G.
and Liede S. (eds), Life forms and Dynamics in Tropical Forests. Dissertationes Botanicae 346,
pp. 95–130.
Hemp A. 2002. Ecology of the pteridophytes on the southern slopes of Mt. Kilimanjaro. Part I:
Altitudinal distribution. Plant Ecol. 159: 211–239.
Hemp A. 2003. Climate impacts and responses on Mount Kilimanjaro. In: Agrawala S. and
Moehner A. (eds), Development and Climate Change in Tanzania: Focus on Mount Kilimanj-
aro. OECD, Paris.
Hemp A. in press a. Continuum or zonation? Altitudinal gradients in the forest vegetation of Mt.
Kilimanjaro Plant Ecol.
Hemp A. in press b. Ecology and altitudinal zonation of pteridophytes on Mt. Kilimanjaro. In
Proceedings of the XVIIth AETFAT Congress 21–26 September 2003.
Hemp A. in press c. The impact of fire on diversity, structure and composition of Mt. Kilimanjaro’s
vegetation. In Proceedings of the GMBA Symposium 19–24 August 2002 in Moshi, Tanzania.
Hemp A., Lambrechts C. and Hemp C. (in press). Global Trends in Africa: The Case of Mt.
Kilimanjaro. UNEP, Nairobi.
Hemp C. and Hemp A. 2003. Saltatoria coenoses of high-altitude grasslands on Mt. Kilimanjaro,
Tanzania (Orthoptera: Saltatoria). Ecotropica 9: 71–97.
Hemp C. in press. The Chagga home gardens – relict areas for endemic Saltatoria species (Insecta:
Orthoptera) on Mt. Kilimanjaro. Biol. Conserv.
Hemp C. 2001. Ethnozoological research on invertebrates on Mt. Kilimanjaro, Tanzania. Eco-
tropica 7: 139–149.
Kress W.J. 1986. The systematic distribution of vascular epiphytes: an update. Selbyana 9: 2–22.
Lind E.M. and Morrison M.E.S. 1974. East African Vegetation. Longman, London.
Mdoe N. and Wiggins S. 1997. Returns to smallholder dairying in the Kilimanjaro region, Tan-
zania. Agr. Econ. 17: 75–87.
Nair P.K.R. 1993. An introduction to agroforestry. Kluwer, Dordrecht.
National Bureau of Statistics and Central Census Office. 2003. 2002 Population and Housing
census. Genaral report. The United Republic of Tanzania, Dar es Salaam.
O’Kting’ati A. and Kessy J.F. 1991. The farming system on Mount Kilimanjaro. In: Newmark
W.D. (ed), The Conservation of Mount Kilimanjaro. The IUCN Tropical Forest Programme,
pp. 71–80.
O’Kting’ati A., Maghembe J.A., Fernandes E.C.M. and Weaver G.H. 1984. Plant species in the
Kilimanjaro agroforestry system. Agroforest. Syst. 2: 177–186.

[154]
 1217

Ramsay J.C. 1965. Kilimanjaro – sources of water supplies. Tanganyika Notes Records 64: 92–94.
Rikli M. 1903/1904. Die Anthropochoren und der Formenkreis des Nasturtium palustre DC.
Berichte der Zürcherischen Botanischen Gesellschaft 8: 71–82.
Rodgers W.A. and Homewood K.M. 1982. Species richness and endemism in the Usambara
mountain forests, Tanzania. Biol. J. Linn. Soc. 18: 197–242.
Schmidt P.R. 1989. Early exploitation and settlement in the Usambara Mountains. In: Hamilton
A.C. and Bensted-Smith R. (eds), Forest Conservation in the East Usambara Mountains Tan-
zania. The IUCN Tropical Forest Programme, pp.75–78.
Simmonds N.W. 1966. Bananas, 2nd edn. London.
Sukopp H. and Kowarik I. 1987. Der Hopfen (Humulus lupulus L.) als Apophyt der Flora Mit-
teleuropas. Natur und Landschaft 62: 373–377.
Sukopp H. and Langer A. 1996. Campanula rapunculoides – ein Apophyt in der Vegetation
Mitteleuropas. Verhandlungen der Gesellschaft für Ökologie 25: 261–276.
Timberlake L. 1986. Krisenkontinent Afrika. Hammer, Wuppertal.
Volkens G. 1897. Der Kilimandscharo. Darstellung der allgemeineren Ergebnisse eines fünfzehn-
monatigen Aufenthalts im Dschaggalande. Reimer, Berlin.
Widenmann A. 1899. Die Kilimandscharo-Bevölkerung. Anthropologisches und Ethnographisches
aus dem Dschaggalande. Petermanns geographische Mitteilungen, Ergänzungs-Heft 129, 104 pp.

[155]
Biodiversity and Conservation (2006) 15:1219–1252
Springer 2006
DOI 10.1007/s10531-005-0768-6

-1

Biodiversity hotspots and conservation priorities

in the Campo-Ma‘an rain forests, Cameroon

M.G.P. TCHOUTO1,*, M. YEMEFACK2, W.F. DE BOER3, J.J.F.E. DE

WILDE4, L.J.G. VAN DER MAESEN4 and A.M. CLEEF5
1
Limbe Botanic Garden, BP 437, Limbe, Cameroon; 2International Institute for Geo-Information
Science and Earth Observation (ITC), PO Box 6, 7500 AA Enschede, The Netherlands; 3Resource
Ecology Group, Wageningen University, Bornsesteeg 69, 6708 PD, Wageningen, The Netherlands;
4
Biosystematics Group, Wageningen University, Generaal Foulkesweg 37, 6703 BL, Wageningen, The
Netherlands; 5Institute for Biodiversity and Ecosystem Dynamics (IBED) Research Group, Paly-
nology and Paleo/Actuo-ecology, University of Amsterdam, Kruislaan 318, 1090 GB Amsterdam, The
Netherlands; *Author for correspondence (e-mail: [email protected])

Received 3 February 2004; accepted 7 January 2005

Key words: Biodiversity, Cameroon, Campo-Ma’an, Central Africa, Conservation, Endemic

species, Forest refuge, Genetic Heat Index, Pioneer Index, Plant diversity, Tropical rain forest

Abstract. Until recently, patterns of species richness and endemism were based on an intuitive
interpretation of distribution maps with very limited numerical analyses. Such maps based solely on
taxonomic collections tend to concentrate on collecting efforts more than biodiversity hotspots,
since often the highest diversity is found in well-collected areas. During the last decades, there has
been an overwhelming concern about the loss of tropical forest biological diversity, and an
emphasis on the identification of biodiversity hotspots in an attempt to optimise conservation
strategies. Furthermore, the concept of sites of high diversity, or hotspots, has attracted the
attention of conservationists as a tool for conservation priority settings. With the development of
GIS tools, geostatistics, phytosociological and multivariate analysis software packages, more rig-
orous numerical analyses of distributional and inventory data can be used for assessing conser-
vation priorities. In the Campo-Ma’an rain forest, inventory data from 147 plots of 0.1 ha each and
7137 taxonomic collections were used to examine the distribution and convergence patterns of strict
and narrow endemic species. We analysed the trends in endemic and rare species recorded, using
quantitative conservation indices such as Genetic Heat Index (GHI) and Pioneer Index (PI), to-
gether with geostatistic techniques that help to evaluate and identify potential areas of high con-
servation priority. The results showed that the Campo-Ma’an area is characterised by a rich and
diverse flora with 114 endemic plant species, of which 29 are restricted to the area, 29 also occur in
southwestern Cameroon, and 56 others that are also found in other parts of Cameroon. Although
most of the forest types rich in strict and narrow endemic species occur in the National Park, there
are other biodiversity hotspots in the coastal zone and in areas such as Mont d’Eléphant and Massif
des Mamelles that are located outside the National Park. Unfortunately, these areas, supporting 17
strict endemic species that are not found in the park, are under serious threat and do not have any
conservation status for the moment. Taking into consideration that with the growing human
population density, pressure on these hotspots will increase in the near future, it is suggested that
priority be given to the conservation of these areas and that a separate management strategy be
developed to ensure their protection.

[159]
1220

Introduction

Central African rain forests are among the top conservation priority areas
in the world (Davis et al. 1994; Heywood and Watson 1995; Myers et al.
2000). The Campo-Ma’an rain forest, in the southern part of Cameroon,
falls under the Guineo–Congolian Regional Centre of Endemism that is
reported to be species-rich with high levels of endemism (White 1979, 1983;
Davis et al. 1994). It is situated in the middle of the Atlantic Biafran forest
zone that extends from Southeast Nigeria to Gabon and the Mayombe area
in Congo (Letouzey 1968, 1985). The vegetation in the Campo-Ma’an area
is determined by climate, especially rainfall, altitude, soils, proximity to the
sea and human disturbance (Tchouto 2004). The structure and composition
of the forest, as well as the vegetation types change progressively from the
mangrove or coastal forest on sandy shorelines through the endemic low-
land evergreen forest rich in Caesalpinioideae with Calpocalyx heitzii and
Sacoglottis gabonensis, to the submontane forest on hilltops and the mixed
evergreen and semi-deciduous forest in the drier Ma’an area. Other vege-
tation types/sub-types include swamps, seasonally flooded forests, riverine
and secondary forests. The forest in the Ma’an area is described as tran-
sitional between the coastal evergreen forest and the semi-deciduous forest
of the interior.
In view of the rich and diverse flora of the Campo-Ma’an rain forest, as
well as its high level of endemism, it has been identified as one of the key
conservation sites in Cameroon (Gartlan 1989; Foahom and Jonkers 1992).
The Campo-Ma’an area is a Technical Operational Unit (TOU) that com-
prises a National Park, five forest management units, two agro-industrial
plantations, and a multi-uses zone (Tchouto 2004). Despite the low popu-
lation density, there are many stakeholders and different types of land use.
Activities such as logging, industrial and shifting agriculture exert varying
ecological impact on the forest ecosystem. This has led to deforestation,
habitat fragmentation and alteration of the coastal forests. With the
increasing destruction of natural ecosystems, it is important to identify
biodiversity hotspots and conservation priorities in order to enable an
effective management. To achieve this, we need to study the species com-
position and species distribution, so that we can target conservation re-
sources and efforts to rich and diverse areas with a high number of endemic
species. Endemism is commonly regarded as an important criterion for
assessing the conservation value of a given area. In this study, forest
inventory data and taxonomic collections will be used to examine the dis-
tribution and convergence patterns of strict and narrow endemic species. We
will use new quantitative conservation indices such as GHI (Genetic Heat
Index) and Pioneer Index (PI) to analyse trends in endemic and rare species
in the various forest types. Finally, geostatistic analysis and techniques
will help to evaluate and identify potential areas of high conservation
priority.

[160]
 1221

Materials and methods

Study area

The study was conducted in the Campo-Ma’an rain forest in south Cameroon.
The site covers about 7700 km2 and it is located between latitudes 210¢–
252¢ N and longitudes 950¢–1054¢ E. Following the FAO classification sys-
tem, soils in the Campo-Ma’an area are generally classified as Ferrasols and
Acrisols (Franqueville 1973; Muller 1979; van Gemerden and Hazeu 1999).
They are strongly weathered, deep to very deep and clayey in texture (except at
the seashores and in river valleys where they are mainly sandy), acid and low in
nutrients with pH (H2O) values generally around 4. The topography ranges
from undulating to rolling in the lowland area, to steeply dissect in the more
mountainous areas. In the Campo area, altitudes are mostly low, ranging from
sea level to about 500 m. In the eastern part, which is quite mountainous, the
altitude varies between 400 and 1100 m and the rolling and steep terrain brings
about a more variable landscape.
The area has a typical equatorial climate with two distinct dry seasons
(November–March and July to mid-August) and two wet seasons (April–June
and mid-August to October). The average annual rainfall generally decreases
with increasing distance from the coast, ranging from 2950 mm/year in Kribi
and 2800 mm in Campo to 1670 mm in Nyabissan in the Ma’an area. The
average annual temperature is about 25 C and there is little variation between
years. The hydrography of the area shows a dense pattern with many rivers,
small river basins, fast-flowing creeks and rivers in rocky beds containing many
rapids and small waterfalls. Generally, the area has a low population density of
about 10 inhabitants per square kilometre and is sparsely populated (ca. 61,000
inhabitants) with most people living around Kribi, along the coast, and in
agro-industrial and logging camps (ERE Développement 2002; de Kam et al.
2002). Despite the low population density, there are few employment oppor-
tunities. The local people are very poor and so far rely solely on the forest
resources to meet their basic needs. As a result, local pressure on the Campo-
Ma’an rain forest is increasing and there are several activities that are carried
out in the area with varying ecological impacts on the forest ecosystem. These
activities include agriculture, logging, poaching and hunting.

Botanical assessment

Sampling was carried out between 2000 and 2003 in 147 plots of 0.1 ha
(50 m · 20 m) in representative and homogeneous vegetation types (see Table 6
for an overview of the plots per vegetation type). In each 0.1-ha plot, all trees,
shrubs, herbs and lianas with DBH ‡1 cm (diameter at breast height, about 1.3 m
above ground) were measured, recorded and identified. These plots were estab-
lished in undisturbed forests or matured secondary forests within 12 vegetation

[161]
1222

Table 1. Star categories and GHI weight classes as defined for Cameroon.

Star category Weight for GHI Comments

Black (BK) 27 Species which are only found in Campo-Ma’an area (strictly
endemic) or near endemics (species which also occur in some
localities around Campo-Ma’an such as Bipindi, Edea-Kribi,
Lolodorf or southern part of Cameroon). Urgent attention to
conservation of population is needed.
Gold (GD) 9 Cameroon endemics, rare and threatened Lower Guinea
endemics. Cameroon has definitely responsibility for preserving
these species.
Blue (BU) 3 Lower Guinea and Guineo–Congolian endemics which are
widespread internationally but rare in Cameroon, or vice versa.
Scarlet (SC) 1 Common but under serious pressure from heavy exploitation.
Exploitation needs to be curtailed if usage is to be sustainable.
Protection of all scales vital.
Red (RD) 1 Common but under pressure from exploitation.
Pink (PK) 1 Common and moderately exploited.
Green (GN) – Widespread Guineo–Congolian, pantropical and tropical
African species that are not under pressure. No particular
conservation concern.
Adapted from Hawthorne and Abu-Juam (1995), Hawthorne (1996) and Tchouto et al. (1998).

types ranging from coastal forest, mangrove, swamp, lowland evergreen forest,
mixed evergreen and semi-deciduous forest to submontane forest at higher ele-
vations (800–1100 m above sea level). Most of the plots were located in the
National Park and the forest management units which are less affected by human
activities. Furthermore, in each representative vegetation type, a provisional
plant species checklist was made in the field with information on their growth
form, guild and frequency. A guild refers to a group associated with a common
way of life (Table 2). For unknown species, a voucher specimen was collected.
The study also involved the collection of fertile specimens encountered in plots,
vegetation types and specific habitats such as exposed rocks and riverbanks. The
geographic co-ordinates of each plot, sample or specimen were recorded using the
Global Positioning System (GPS, Garmin 12XL model with estimated precision
of ±10 m). These co-ordinates were used for mapping main vegetation types,
species distribution, and biodiversity hotspots. A duplicate of each specimen was
mounted and preserved in the Kribi Herbarium. Others duplicates were sent to
the National Herbarium in Yaounde, Cameroon (YA) and the Nationaal Her-
barium Nederland, Wageningen University Branch (WAG) for further identifi-
cation and preservation.

Criteria for taxa inclusion

A plant species checklist was generated from the inventory data, from the plant
collections made during the study, and from specimens previously collected in
the area by other scientists, stored in the Cameroon and Wageningen herbaria.
Furthermore, a taxonomic search for potential taxa of high conservation

[162]
 1223

Table 2. Guild and weight classes.

Guild Weight for PI Comment

Pioneer (PI) 2 Regenerating only in forest gaps and therefore

indicating disturbed forest (e.g. Ceiba, Musanga,
Harungana, Macaranga).
Non-Pioneer 1 Although some juveniles are also found in the
light demanding (NP) understorey of undisturbed forest, they require
gaps to develop to full maturity. Generally,
non-pioneer light demanding are abundant
in matured disturbed forest (e.g. Albizia,
Entandrophragma, Piptadeniastrum,
Pycnanthus).
Shade-bearers (SB) 0 Understorey herbs, shrubs and trees which
grow, flower and fruit in undisturbed forest
(e.g. Cola, Diospyros, Psychotria, Rinorea).
Adapted from Hawthorne and Abu-Juam (1995), Hawthorne (1996) and Cable and Cheek (1998).

priority such as endemic, rare, new and threatened species was carried out
using existing floras and monographs (Keay and Hepper 1954–1972; Aubré-
ville and Leroy 1961–1992; Aubréville and Leroy 1963–2000; Lebrun and Stork
2003; Satabié and Leroy 1980–1985; Satabié and Morat 1986–2001), the IUCN
(2002) red list categories, and the WCMC (1998) world list of threatened trees.
On the basis of this information, a list of 141 species of high conservation
values was produced with information on their habit, guild, star category
(Table 1) and chorology. In this list, priority was given to taxa that are strictly
endemic to the Campo-Ma’an area. Followed by species that are endemic to
southwestern Cameroon (also occurring in Bipindi and Lolodorf areas) or
Cameroon and Lower Guinea endemics (especially if they reach their northern
or southern limit of distribution in Campo-Ma’an). Furthermore, species that
reach their northern or southern limit of distribution in the Campo-Ma’an area
were also included in the list.

Star rating of species and measurement of forest conservation value

A star rating system, based on the work of Hawthorne and Abu-Juam (1995)
and Hawthorne (1996) in Ghana, Cable and Cheek (1998) and Tchouto et al.
(1998) in Cameroon, was used to define the conservation status of each species
recorded (Table 1). The factors considered when categorising species into star
categories are their distribution, ecology, local abundance, taxonomy, life
history, interaction with ecosystem parameters and economic importance
(Hawthorne 1996). Therefore, species that are endemic, rare, threatened, or
likely to represent a scarce genetic resource, are more valuable than others are.
Hence, forests richer in such species receive a higher score than others.
The GHI concept was developed by Hawthorne (1996) to express the con-
servation value of a given forest, and the PI concept to express the level of

[163]
1224

disturbance in a given forest. GHI is an attempt to provide a scale, on which to

measure the genetic ‘temperature’ or value of the forest. A plot/forest with an
average GHI >150 will be considered warm or hot. In general, for species with
completed monographs, black stars occupy about 1–3 filled degree squares on
a standard distribution map, gold stars 4–14, blue stars 15–30, and green star
more than 30 degree squares.
Hawthorne (1996) defined the guild as a flexible concept used to circumscribe
a group of plant species with a similar ecology and way of life. All the species
were grouped into guild classes as defined in Table 2 and a PI score was
calculated as an expression of the relative contribution of pioneers. Five classes
of human disturbance were used to evaluate the forest quality and condition as
defined in Table 3. These classes were mainly based on the field observation of
the level of human disturbance and the state of forest degradation.
The GHI and PI values of each sample were calculated using the TREMA
database as follows:
Genetic Heat Index (GHI) ¼ ½ð(BK  BK weightÞ þ ðGD  GD weightÞ
þ ðBU  BU weightÞ þ ðRD  RD weight))=
ðBK þ GD þ BU þ GN þ RDÞ  100;
where BK = number of black star species; GD = number of gold star species;
BU = number of blue star species; GN = number of green star species; and
RD = number of red, scarlet and pink star species.
ðPioneer  PI weightÞ þ (NP  NP weightÞ
Pioneer Index (PI) ¼  100;
Total number of species
where PI = number of pioneer species and NP = number of non-pioneer
light demanding species.

Table 3. Forest condition classes showing the degree of human disturbance on the natural forest
cover.

Forest condition Classes Notes

Excellent Virtually Undisturbed forest, with good canopy

undisturbed and few signs of human disturbance
except for hunting and NTFPs collection.
Good Less than 25% Small patches of recent disturbances (<25%) with
disturbed good canopy cover.
Slightly 25–50% disturbed Obviously disturbed with significant patches (25–50%)
degraded of recent degradation but with good predominant
forest and broken upper canopy.
Mostly More than 50% Considerable are (>50%) of recent degradation.
degraded disturbed Patchy with heavily disrupted canopy.
Very poor Farm land No significant forest left (<2% good forest).
Massive land conversion for plantation or farm.

[164]
 1225

Geostatistical analysis

Conservation indices such GHI and PI are likely to vary throughout a region.
Geostatistics (Isaaks and Srivastava 1989; Webster and Oliver 2001) were applied
to quantify the spatial distribution of GHI within the Campo-Ma’an forest.
Geographic analyses were done using ILWIS software (ILWIS 2001) and
GSTAT package (Pebesma and Wesseling 1998) of R software (R Development
Core Team 2002). The semivariance was calculated for GHI data on a minimum
lag distance of 1250 m and each lag distance class contained at least 105 pairs of
points. The semivariogram parameters (nugget, sill and range) were computed
using the GSTAT fit variogram function. During the study of GHI spatial var-
iability, the main objective was to obtain a map from point observations. Since
this also required the estimation of a value at un-visited locations, the technique
commonly used is known as kriging (Isaaks and Srivastava 1989). The semi-
variogram function was then used to extrapolate the GHI values in the Campo-
Ma’an forest at 100 m · 100 m grid, using ordinary kriging. The output map was
reclassified into five classes of conservation value (Hawthorne 1996).

Results

Species richness and endemism

A plant species checklist made of 2297 species of vascular plants, ferns and fern
allies was generated from inventory data and from 2348 herbarium specimens and
4789 ecological specimens collected in the various plots. They belonged to 851
genera and 155 families. More than 67% of the specimens were identified at species
level, 28% at generic level, 4% at family level and 1% remained unidentified. The
20 most important families and genera are shown in Tables 4 and 5. In terms of
growth form, tree species contributed for 26% to the total number of 2297 species
recorded, followed by herbs (24%), shrubs (23%) and climbers (17%), respec-
tively. About 72% of the total number of species recorded was also found in the
Campo-Ma’an National Park and the remaining 28% were only found in the
coastal forest and the semi-deciduous forests located outside the park.
In addition to a list of 92 threatened species (Appendix 2) recorded in IUCN
(2002) and WCMC (1998), a list with 141 plant species of high conservation
priorities was produced, with information on their growth forms, guild, cho-
rology and star categories (Appendix 1). Only species that are endemic to
Cameroon and species that reach their northern or southern limit of distri-
bution are included in this list. The Campo-Ma’an area has about 114 endemic
species, 29 of which are only known from the area, 29 only occur in the
southwestern part of Cameroon, and 56 near endemics that also occur in other
parts of Cameroon (Appendix 1 and Figure 1).
Shrubs contributed for 38% of the 114 endemic species (Appendix 1), herbs
29%, trees 20% and climbers 11%. Moreover, 540 species (23% of the total

[165]
1226

Table 4. Most important families recorded in the Campo-Ma’an area.

No. Family No. of species Predominant growth forms

1 Rubiaceae 279 Trees, shrubs, herbs and climbers

2 Euphorbiaceae 117 Trees, shrubs, herbs and climbers
3 Leguminosae-Caesalpinioideae 96 Trees and shrubs
4 Apocynaceae 80 Trees, shrubs, herbs and climbers
5 Annonaceae 69 Trees, shrubs and climbers
6 Acanthaceae 68 Herbs
7 Leguminosae-Papilionaideae 65 Trees, herbs and climbers
8 Sterculiaceae 62 Shrubs and trees
9 Gramineae 54 Herbs
10 Orchidaceae 54 Terrestrial and epiphytic herbs
11 Melastomataceae 46 Shrubs and herbs
12 Moraceae 40 Trees, shrubs and herbs
13 Celastraceae 39 Climbers
14 Cyperaceae 39 Herbs
15 Dichapetalaceae 39 Climbers
16 Sapindaceae 36 Trees and shrubs
17 Araceae 34 Herbs and hemi-epiphytes
18 Loganiaceae 33 Shrubs and climbers
19 Sapotaceae 30 Trees and shrubs
20 Begoniaceae 29 Terrestrial and epiphytic herbs
Others(135 families) 988 Trees, shrubs, climbers and herbs

Table 5. Most important genera recorded in the Campo-Ma’an area.

No. Genus No. of species Predominant growth forms

1 Dichapetalum (Dichapetalaceae) 37 Climbers

2 Psychotria (Rubiaceae) 35 Shrubs
3 Cola (Sterculiaceae) 32 Trees and shrubs
4 Begonia (Begoniaceae) 29 Terrestrial and epiphytic herbs
5 Diospyros (Ebenaceae) 27 Trees and shrubs
6 Salacia (Celastraceae) 27 Climbers
7 Strychnos (Loganiaceae) 24 Climbers
8 Rinorea (Violaceae) 23 Trees and shrubs
9 Drypetes (Euphorbiaceae) 21 Trees and shrubs
10 Combretum (Combretaceae) 18 Climbers
11 Dorstenia (Moraceae) 18 Herbs
12 Campylospermum (Ochnaceae) 17 Shrubs
13 Bulbophyllum (Orchidaceae) 16 Terrestrial and epiphytic herbs
14 Ficus (Moraceae) 16 Trees and stranglers
15 Garcinia (Guttiferae) 16 Trees and shrubs
16 Asplenium (Aspleniaceae) 15 Epiphytic herbs
17 Culcasia (Araceae) 15 Herbs and hemi-epiphytes
18 Landolphia (Apocynaceae) 15 Climbers
19 Tricalysia (Rubiaceae) 15 Trees an shrubs
20 Bertiera (Rubiaceae) 14 Shrubs
Others (831 genera) 1867 Trees, shrubs, climbers and herbs

[166]
 1227

Figure 1. Distribution of 114 strict and narrow endemic plant species recorded in the Campo-
Ma’an area (gray circle). Black circle represents the distribution of 17 threatened strict endemics
that are not found in the National Park. The size of the circle represents the relative density of
endemics at a given point.

number of species) recorded are endemic to the Lower Guinea Centre of

Endemism, 1123 species (49%) are Guineo–Congolian endemics and 105
species (5%) are Guinea endemics as described by White (1979). Overall, there
was a high concentration of strict and narrow endemic species in the lowland
evergreen forest rich in Caesalpinioideae, coastal and submontane forests lo-
cated in the western and northern parts of Ma’an and a relatively low con-
centration of these species in Ma’an area (Figure 1). Although more than 70%
of the total endemic species recorded were also found in the National Park, 17
of the 29 strict endemic species were not recorded in the park (Appendix 1).
The distribution patterns of these 17 taxa showed a high concentration of
species around Campo, Lobe, Massif des Mamelles, Mont d’Eléphant and
Zingui and a very poor representation in the Ma’an area (Figure 1).

GHI and measurement of forest conservation value

More than 57% of the plots have a high GHI score with the highest score
recorded in the submontane forest (GHI = 294.4) and the lowest score in
mangrove (GHI = 3.1). As shown in Figure 2, the submontane forest had the
highest average GHI score of 214.7, followed by the lowland evergreen forest
rich in Caesalpinioideae with Calpocalyx heitzii and Sacoglottis gabonensis

[167]
1228

Figure 2. The average Genetic Heat Index (GHI = bars) and average Pioneer Index (PI = line)
for the various vegetation types as defined in Table 6.

(GHI = 194.1). The mangrove and the coastal forest on sandy shorelines had
the lowest average GHI score (GHI = 3 and 120.2, respectively). The average
PI was very high in the mangrove forest (PI = 125), coastal forest on sandy
shorelines (PI = 66.9) and in the forest rich in Aucoumea klaineana (PI = 60).
Generally, there was a significant decrease in average GHI with increasing
average PI (Figure 2). As shown in Figure 3, there was a very strong significant
negative correlation between the average GHI scores and the PI scores
recorded in the various vegetation types (F1–10 = 111.71, R2 = 0.918,
p < 0.0001). However, the correlation was rather weak with a low explanatory
factor when the analysis was carried out using all plots as individual data
points (F1–45 = 94.00, R2 = 0.393, p<0.0001). Most of the forest types within
the National Park were virtually undisturbed or less than 25% disturbed
(Figure 4). The coastal forest between Campo and Kribi, as well as the forests
around Massif des Mamelles, Mont d’Eléphant, agro-industrial plantations,
logging concessions and settlements were much more affected by human
activities (Figure 4). These forests were often more than 25% disturbed by
human activities and were characterised by a high PI scores (Figures 2 and 4).

Geostatistical results

The analysis of the spatial structure of the dataset did not show any preferential
spatial trend. Therefore, an omni-directional analysis of the semivariance (best
described by a spherical model) was applied. Figure 5 shows the semivariogram
and its characteristics. The GHI variable showed a strong spatial dependence

[168]
 1229

Figure 3. Correlation between the average GHI scores and the average PI scores for the various
vegetation types.

Figure 4. Impact of human disturbance on the Campo-Ma’an rain forest.

within a range of 10,500 m. The nugget (645) was low compared to the total
variance or sill (3700). This suggests that more than 82% (100*(Sill-Nugget)/Sill)
of the semivariance of GHI could be modelled by the variogram over a range of
10 km. The output map of the ordinary kriging (Figure 6) was reclassified into

[169]
1230

Figure 5. Spherical variogram model for GHI in the Campo-Ma’an rain forest (estimated from
147 points of 0.1 ha each).

Figure 6. Ordinary kriging map showing the distribution of GHI scores and conservation hot-
spots within the Campo-Ma’an rain forest The following GHI values are defined for the various
conservation classes (Hawthorne 1996): Very high conservation value for GHI >200; High con-
servation value (150 ‡ GHI <200); Moderate conservation value (100 ‡ GHI <150); Low con-
servation value (50 < GHI <100) and very low conservation value (GHI <50).

[170]
 1231

five GHI classes, partitioning the conservation value of the Campo-Ma’an forest.
This partition showed that 1% of the area was characterised by a very high
conservation value, 45% by a high conservation value, 30% by an average
conservation value, 15% by a low conservation value and 9% by a very
low conservation value. A considerable portion of the National Park and the
forests around Massif des Mamelles and Mont d’Eléphant was characterised by a
high conservation value, with highest values found in Dipikar Island, Massif des
Mamelles, Mont d’Eléphant and in the submontane forest on hilltops. The for-
ests in the Ma’an area, around Campo and agro-industrial plantations, near
villages and along the roads had a low conservation value. Similar patterns were
observed for the distribution of strict and narrow endemic species (Figure 1).

Discussion

General vegetation patterns

The Campo-Ma’an has a diverse range of vegetation that changes pro-

gressively from sea level to 1100 m at higher altitudes. The wetter Campo
area was dominated by the lowland evergreen forest rich in Caesalpinioideae
and the drier Ma’an area by the mixed evergreen and semi-deciduous forest.
More than 65% of the forest types recorded had at least 250 species
(Table 6). The submontane forest had the highest frequency of species-rich
plots (93% of the plots had above 100 species/0.1 ha). Other rich vegetation
types included the lowland evergreen forest rich in Caesalpinioideae with
Calpocalyx heitzii and Sacoglottis gabonensis (81%), the lowland evergreen
forest rich in Caesalpinioideae with Sacoglottis gabonensis (78%), the low-
land evergreen forest rich in Caesalpinioideae (76%) and the mixed ever-
green and semi-deciduous forest (67%). The mangroves, swamps and the
coastal vegetations on sandy shorelines were species-poor. The explanation
for the diverse range of forest types and habitats might stem partly from the
fact that the Campo-Ma’an vegetation is influenced by several environ-
mental factors such as rainfall, altitude, soil, the proximity to the sea and
the degree of human disturbance (Tchouto 2004). As a result there was a
gradual variation in dominant species, forest type and structure from the
coast to the hilltops and the drier forest in the Ma’an region.

Species richness and endemism

The Campo-Ma’an area is characterised by a rich and diverse flora with more
than 2297 species of vascular plants, ferns and fern allies. The site has about
114 endemic plant species out of which 29 are strictly endemic to the site. The
number of endemic plant species is relatively high considering the size of the
area, and more than 75% of the current vegetation cover was characterised

[171]
1232

Table 6. Number of plots, number of species and number of stem/ha recorded within the various
vegetation types for all plants with DBH ‡1 cm.

Vegetation types No of plots No of species Average No of stems/ha

Caesalp (2.3 ha) 23 557 (75–128) 4380 (2500–5750)

Caesalpcasa (2.5 ha) 25 555 (93–139) 5326 (3120–7020)
Caesalpsa (1.4 ha) 14 474 (86–138) 5935 (4350–7120)
Cosaga (0.9 ha) 9 303 (81–140) 5810 (4940–8010)
Cosaca (0.9 ha) 9 326 (78–108) 5864 (4740–7570)
Cosas (0.4 ha) 4 100 (27–55) 4710 (3630–5700)
Mixevergreen (2.1 ha) 21 523 (63–135) 4983 (3890–6980)
Mixsemideci (1.6 ha) 16 481 (86–147) 5460 (4390–6340)
Submontane (1.4 ha) 14 499 (79–148) 6094 (3680–8449)
Swamps (0.5 ha) 5 246 (18–108) 4276 (2070–5960)
Mangrove (0.2 ha) 2 4 (3–4) 8630 (8150–9100)
Okoumé forests (0.5 ha) 5 234 (18–107) 4802 (3720–5800)
Total for the Campo-Ma’an area 147 1116 (3–148) 5312 (2070–9100)

Minimum and maximum values are given betweem brackets.

Submontane = Submontane forest; Caesalpcasa = Lowland evergreen forest rich in Caesalpi-
nioideae, Calpocalyx heitzii and Sacoglottis gabonensis; Casealpsa= Lowland evergreen forest rich
in Caselpinioideae and Sacoglottis gabonensis; Caesalp= Lowland evergreen forest rich in Cae-
salpinioideae; Cosaga= Coastal forest with Sacoglottis gabonesis; Mixsemideci= Mixed evergreen
and semi-deciduous forest with elements predominant; Mixevergreen= Mixed evergreen and semi-
deciduous forest with semi-deciduous elements of evergreen forest predominant; Okoumé= forest
rich in Aucoumea klaineana; Cosaca= Coastal forest with Sacoglottis gabonensis and Calpocalyx
heitzii; and Cosas= Coastal forest on sandy shorelines.

either by very high, high or average GHI values (Figure 6). Furthermore, the
distributions of strict and narrow endemic species showed a high concentration
of these species in the submontane forest between Ebianemeyong and Akom II,
in Dipikar Island, and in the forests in and around Massif des Mamelles, Lobe,
Mont d’Eléphant and Zingui. Surprisingly, the mixed evergreen and semi-
deciduous forest in the Ma’an area showed a relatively low concentration of
these species (Figure 1). The explanation for the high occurrence of endemics
might stem partly from the fact that the area falls within a series of postulated
rain forest refugia in Central Africa (Hamilton 1982; White 1983; Maley 1987,
1989, 1990, 1993, 1996; Sosef 1994, 1996). In such refugia, the unique combi-
nation of climatic and geological histories, contemporary ecological factors,
and inherent biological properties of taxa and their combinations, may have
contributed to survival and/or speciation (Barbault and Sastrapradja 1995;
Hawksworth and Kalin-Arroyo 1995). Furthermore, the Campo-Ma’an area
forms part of the Guineo–Congolian Regional Centre of Endemism (White
1983). All families endemic to this biogeographic region are also found in the
Campo-Ma’an area (White 1983). They include Hoplestigmataceae, Huaceae,
Lepidobotryaceae, Medusandraceae, Pandaceae, Pentadiplandraceae and
Scytopetalaceae. Moreover, 82% of endemic genera cited by White (1983) also
occur in the area.

[172]
 1233

Forest richness and biodiversity hotspots

Considering the fact that the occurrence of endemic species contributes sig-
nificantly to the conservation value of a forest, it is important to study their
distribution and abundance prior to any conservation initiatives. This is mainly
due to the fact that strict and narrow endemic species are restricted to small
areas, and are therefore highly vulnerable to human disturbance and other
forms of environmental changes (Myers 1988; Williams 1993; Heywood and
Watson 1995). A study carried out in the Campo-Ma’an rain forest has re-
vealed that the submontane forest, the lowland evergreen forest rich in Cae-
salpinioideae with Calpocalyx heitzii and Sacoglottis gabonensis, and the
lowland evergreen forest rich in Caesalpinioideae are richer in strict and nar-
row endemics compared to the other forest types found in secondary forest and
along the coast. This is confirmed by the high average GHI scores recorded in
these forest types (Figure 2). Most of these forest types were located in the
National Park and the lowland evergreen forests around Massif des Mamelles
and Mont d’Eléphant. They were virtually undisturbed or less than 25% dis-
turbed by human activities (Figure 4). This implies that the Massif des
Mamelles and the Mont d’Eléphant areas represent other biodiversity hot-
spots, located outside of the Park (Figure 6).
There was a strong significant negative correlation between the average GHI
scores and the average PI scores recorded in the various vegetation types. Most
plots located near settlements and in secondary forests were characterised by a
low conservation value with low GHI scores and high PI scores. This con-
firmed that disturbed forests are rich in pioneer species but poor in plant
species with high conservation priority. It is worth reiterating that a consid-
erable portion of the Campo-Ma’an area has been selectively logged at least
twice during the past 30 years. Although logging damages were moderate and
had low impact on the total forest biodiversity, it has created forest gaps that
allowed the development of many pioneer species. This might have contributed
to the high average PI scores registered in the coastal forest types. In these
areas with conflicts between human and conservation activities, there is an
urgent need to develop participatory approaches to sustainable natural re-
source management that integrates the objectives of conservation with local
development.

Threatened species

During the selection of species of high conservation priority, taxa were chosen
on a global rather than a Cameroonian or a Campo-Ma’an perspective of
conservation importance. Of the 29 strict endemic species that are only known
from the Campo-Ma’an area, 17 were not recorded in the National Park
illustrating the need for conservation activities outside the park. Although
these 17 strict endemics are not immediately threatened with extinction, the

[173]
1234

most threatened are probably those occurring in the coastal zone and in areas
located at the vicinity of large agro-industrial plantations, since these areas are
heavily exploited. As shown in Figure 4, their habitats are fragmented and
degraded because these areas are surrounded by farms and heavily disturbed
forests.
Considering the fact that extinct species are taxa that are no longer known to
exist in the world after repeated search in their type localities (WCMC 1998;
IUCN 2002), we cannot yet talk about extinction because no attempt has been
made to search for these species. Furthermore, only 67% of the total amount
of specimens collected was identified at species level. However, with the
ongoing speed of forest degradation noticed in the coastal area, eight of these
strict endemics (Beilschmiedia dinklagei, Deinbollia macroura, Ledermanniella
batangensis, Psychotria aemulans, P. batangana, P. dimorphophylla, P. oligo-
carpa, and Strychnos canthioides) that are only known from the coastal zone
can be categorised as endangered species. While the nine others that are located
inland around Efoulan, Fenda, Massif des Mamelles, Mont d’Eléphant and
Zingui can be categorised as vulnerable. They are Afrotrewia kamerunica,
Bulbophyllum alinae, Begonia montis-elephantis, Calvoa stenophylla, Dorstenia
dorstenioides, Guaduella mildbraedii, Hypolytrum sp. nov., Scaphopetalum ac-
uminatum and S. brunneo-purpureum. Some of them so far are only known
from type specimens or from a few collections made in the type locality before
the 60s. Others such as Afrotrewia kamerunica, Begonia montis-elephantis and
Hypolytrum sp. nov. have a restricted range with a small and restricted pop-
ulation. Furthermore, habitat fragmentation may convert a previously more
continuous population structure to a metapopulation structure, with local
populations becoming so small that they may have a substantial risk of
extinction (Hawksworth and Kalin-Arroyo 1995).

Implications for biodiversity conservation

The Campo-Ma’an National Park

The National Park is the core conservation area of the Campo-Ma’an Tech-
nical Operational Unit. It is surrounded by areas under several land uses that
have varying ecological impact on the park and the surrounding forests. The
park is of high conservation priority with about 72% of the 2297 species of
vascular plants, ferns and fern allies recorded so far in the Campo-Ma’an area.
More than 70% of the total endemic species recorded were also found in the
National Park, and most of the forest types with high GHI scores, low PI
scores and high conservation priority species were also found in the park
(Figures 2 and 6). The most important one is the endemic lowland evergreen
forest rich in Caesalpinioideae with Calpocalyx heitzii and Sacoglottis gabon-
ensis, a vegetation type that only occurs in the Campo area (Letouzey 1985;
Gartlan 1989; Thomas and Thomas 1993). Other forest types such as the
submontane forest on hilltops, the lowland evergreen forest rich in

[174]
 1235

Caesalpinioideae and the mixed evergreen and semi-deciduous forests are also
well represented. So far the National Park is the only area with a legal con-
servation status. It is a permanent state forest that is protected by law and
solely used for forest and wildlife conservation. However, its boundaries have
not been marked, the management plan has not yet been produced and pro-
tection is weak. Therefore, it is of urgent need to demarcate the boundary of
the park, to reinforce its protection, and to complete and implement its
management plan as soon as possible.

Massif des Mamelles and Mont d’Ele´phant

This study has demonstrated that other hotspots for biodiversity conservation,
such as Mont d’Eléphant and Massif des Mamelles, are located outside the
National Park (Figure 6). These areas are non-permanent forest estates that
can be allocated for human activities such as logging, agro-industry, agricul-
ture, agro-forestry, community forest, communal forest or private forest.
Moreover, hunting, fishing, mineral exploitation or any other form of eco-
nomic activities is allowed if done in accordance to the 1994 forest law.
Unfortunately, these areas do not have any conservation status and a number
of ongoing human activities have negative impacts on the forest ecosystem
(Tchouto 2004). In addition to the construction of the Tchad-Cameroon oil
pipeline terminal at Grand Batanga and the rock exploitation on Mont
d’Eléphant, there exists a plan to exploit the iron ore deposits of the Massif des
Mamelles. All these activities, if realized, would affect the vegetation and thus
impact the biodiversity. As shown in Figure 4, these fragmented forest patches
with high conservation priorities are more seriously exposed to forest degra-
dation and habitat loss since they are surrounded by disturbed and degraded
forests. Furthermore, they are the type localities for some rare endemic species
such as Afrotrewia kamerunica, Begonia montis-elephantis, Bulbophyllum alinae
and Hypolytrum sp. nov. that are so far only known from the type specimens or
from few collections made in these areas. Pressure on these fragmented hot-
spots will increase in the future with the growing human population density,
the few local employment opportunities and the poverty of the local people, for
whom the forest is a major resource. In order to ensure the protection of these
areas, it is suggested that local community be encouraged to create community
forests with several management zones. Each community forest should have
the identified biodiversity hotspot as the core conservation area, surrounded by
a buffer zone stimulating the sustainable management of non-timber forest
products and hunting practices.

The coastal zone, Ntem basin, Lobe and Memve’ele waterfalls

The coastal zone is a narrow strip (65 km long) along the Atlantic Ocean from
the Lobe waterfalls to the Ntem estuary in the Dipikar Island that extends
about 2–3 km inland. It has suffered and continues to suffer from intense

[175]
1236

human pressure that has led to the destruction of most of its natural vegetation
(Figure 4). However, it is worth mentioning that some rare endemic species
such as Deinbollia macroura, Psychotria batangana, P. dimorphophylla, P. oli-
gocarpa, and Strychnos canthioides are so far only known from this zone.
Furthermore, there is an impressive network of rivers and streams in the
Campo-Ma’an area that presents a number of very specialised riparian habi-
tats. Our study confirmed that the Lobe, Bongola, Memve’ele waterfalls and
Ntem basin (Boucle du Ntem) support a rich riparian flora with many endemic
and rare rheophile species (Cusset 1987; Thomas and Thomas 1993). Most of
the endemic rheophytes are of the genus Ledermanniella in the Podostemaceae
family. These rheophytes which are found on exposed rocks in streambeds, are
seasonally submerged by fast-flowing water, and normally reproduce in drier
periods when the water level recedes. The Ntem basin is also reported to
constitute an important refuge for wildlife and fish fauna because of the
presence of many rare species of freshwater fishes (Vivien 1991; Matthews and
Matthews 2000; Djama 2001). Therefore, it is suggested to develop a separate
management strategy in order to protect these riparian habitats.

Conclusion

The study provides important information on the abundance and distribution

of endemic species, as well as the location of biodiversity hotspots in the
Campo-Ma’an area. This information is essential for any decision-making
process for biodiversity conservation and sustainable natural resource man-
agement. Our study has revealed that the area is characterised by a rich flora
with more than 2297 species of vascular plants and about 114 endemic plant
species, of which 29 are only found in the area. Although most of the forest
types with plant species of high conservation priorities appeared to occur in the
Campo-Ma’an National Park, there are several human activities in the area
with varying negative ecological impacts on the forest ecosystem. Therefore,
the successful management and long-term sustainability of the Park will largely
depend on the ability to reconcile the objectives of conservation and other uses
at its vicinity.
The study also demonstrated that there are other biodiversity hotspots in the
coastal zone and areas such as Mont dEléphant and Massif des Mamelles that
are located outside the National Park. These areas support 17 strict endemic
species that are not found in the park. Unfortunately, these strict endemics are
the most threatened since their habitats are fragmented and degraded as a
result of past and present land conversion to subsistence and industrial plan-
tations. Furthermore, these hotspots are the type localities for some rare en-
demic species that are so far only known from type specimens or from a few
collections made in these areas. Contrary to the National Park, these hotspots
do not yet have any conservation status per se. However, although the park is a

[176]
 1237

permanent state forest which is protected by law and should be solely used for
forest and wildlife conservation, its boundaries have not been marked, the
management plan has not yet been produced and protection is weak. It is,
therefore, of urgent need to demarcate its boundary, reinforce its protection,
and complete and implement its management plan as soon as possible. Fur-
thermore, in view of the fact that pressure on these fragmented hotspots is
likely to increase in the future with the growing human population density, it is
suggested that a separate management strategy be developed to ensure the
protection of these biodiversity hotspots and their endemic species.

Acknowledgements

This study was carried out in the framework of the Campo-Ma’an Biodiversity
Conservation and Management Project, Cameroon, and was financially sup-
ported by Tropenbos International, The Netherlands. We will like to thank G.
Achoundong, J.M. Onana, B. Sonke, L. Zapfack and P. Mezili at the National
Herbarium, Cameroon, and F.J. Breteler and C.C.H. Jongkind at the Na-
tionaal Herbarium Nederland, Wageningen University Branch, who assisted in
plant identification. The staff of Campo-Ma’an Project is also acknowledged
with gratitude for their assistance and support during the fieldwork. Particular
thanks are for my field assistants Elad Maurice and Ossele Mathilde for their
enthusiastic support and cooperation. We will also like to extend our sincere
thanks to all chiefs and village representatives, for their active participation in
the organisation and collection of field data.

[177]
 Appendix 1 List of 141 plant species that are either strictly endemic to the Campo-Ma’an area (only found in Carapo-Ma’an) or near endemic (also occur in

1238
the western parts of south Cameroon or other parts of Cameroon).

No. Family Species Guild Star Habit Chorology Notes

1 Acanthaceae Stenandrium thomense sb GD Hb Cam Akom II, Dipikar Island, Western and South Cameroon
(Milne-Redh.) Vollesen
2 Annonaceae Monanthotaxis elegans sb GD Sh Sw-Cam Akom II, Dipikar Island, Massif des Mamelles, Bipindi and
(Engl. and Diels) Verdc. Lolodorf
3 Annonaceae Monodora zenkeri Engl. sb GD Sh Cam Massif des Mamelles, Bipindi and Lolodorf
and Diels*
4 Apocynaceae Callichilia monopodialis sb GD Sh Cam Ma’an, South, Centre and East Cameroon
(K.Schum.) Stapf*
5 Apocynaceae Landolphia flavidiflora Np GD Lwcl Cam Efoulan, Bipindi, Makak and Mt. Cameroon
(K. Schum.) Persoon*
6 Apocynaceae Petchia africana sb BK Sh Sw-Cam Campo, Bipindi and Lolodorf
Leeuwenb.*
[178]

7 Apocynaceae Tabernaemontana hallei sb GD Sh Lg Northern limit of distribution, from Gabon to Akom II, Ono-
(Boiteau) Leeuwenb. yong and Ma’an
8 Araceae Culcasia bosii Ntepe-Nyame sb BK He Sw-Cam Massif des Mamelles, Dipikar Island, Ma’an and Bipindi
9 Araceae Culcasia panduriformis sb GD Hb Cam Bifa, Zingui, Akom II, Dipikar Island, Bipindi, Mt Cameroon
Engl. and Krause and Eseka
10 Aristolochiaceae Pararistolochia preussii Np GD Swcl Cam Dipikar Island, Ebolowa and Mt. Cameroon
(Engl.) Hutch. & Dalziel
11 Balsaminaceae Impatiens hians Hook.f. var. sb bu Hb Lg Northern limit of distribution, from Gabon to Bipindi, Zingui
bipindensis (Gilg)
Grey- Wilson
12 Balsaminaceae Impatiens gongolana sb bu Hb Lg Northern limit of distribution, from Gabon to Ebianemeyong
N.Hallé
13 Begoniaceae Begonia anisosepala Hook.f. sb bu Hb Lg Northern limit of distribution, from Gabon to Bipindi, Zingui
and Grand Batanga
14 Begoniaceae Begonia clypeifolia Hook.f. sb bu Hb Lg Northern limit of distribution, from Congo, Gabon to Mvini
and Efoulan
 15 Begoniaceae Begonia elaeagnifolia Hook. ep bu Ep Lg Northern limit of distribution, from Gabon to Mvini, Efoulan
f. and around Kom River
16 Begoniaceae Begonia heterochroma Sosef sb bu Hb Lg Northern limit of distribution, from Gabon to Lolabe and
around Kribi
17 Begoniaceae Begonia mbangaensis Sosef sb BK Hb Sw-Cam Akom II, Efoulan, Bipindi and Lolodorf
18 Begoniaceae Begonia microsperma Warb. sb GD Hb Cam Ebianemeyong, Ma’an, South-west and South Cameroon
19 Begoniaceae Begonia montis’-elephantis sb BK Hb Campo-Ma’an Rare species, only known from a small population on Mt
J.J.de Wilde* d’Eléphant
20 Begoniaceae Begonia zenkeriana Smith sb BK Hb Sw-Cam Campo, Massif des Mamelles, Dipikar Island, Bipindi and
and Wassh. Lolodorf
21 Burseraceae Aucoumea klaineana Pierre Pi bu Tr Lg Northern limit of distribution, from Gabon to Ma’an and
Ebianemeyong
22 Burseraceae Dacryodes buettneri (Engl.) Np bu Tr Lg Northern limit of distribution, from Gabon to Ma’an and
Lam. Ebianemeyong
23 Capparaceae Ritchiea simplicifolia Oliv. sb BK Sh Cam Lobe, Campo, Kienke, Dipikar Island, Bipindi, Lolodorf and
var. Caloneura (Gilg) Kers Ebolowa
[179]

24 Celastraceae Pristimera luteoviridis (Ex- Np BK Swcl Campo-Ma’an Rare species, only known from few collections on Mt d’Elé-
ell) N.Hallé var. kribiana phant and Dipikar Island
N.Hallé
25 Chrysobalanaceae Dactyladenia cinera (Engl. sb BK Tr Sw-Cam Rare species, only known from type specimens (Bipindi) and a
ex de Wild) Prance and record from Grand Batanga
F.White**
26 Chrysobalanaceae Dactyladenia icondere (Ba- sb bu Sh Lg Northern limit of distribution, from Congo, Gabon to Grand
ill.) Prance and F.White Batanga, Campo and Dipikar Island
27 Combretaceae Combretum cinnabarinum np bu Lwcl Lg Northern limit of distribution, from Gabon to Bipindi and
Engl. and Diels Dipikar Island
28 Cyperaceae Hypolytrum sp. nov. ined.* sb BK Hb Campo-Ma’an New species only known from Mont d’Eléphant
29 Dichapetalaceae Dichapetalum altescandens np bu Lwcl Lg Northern limit of distribution, from Gabon to Efoulan and
Engl. * Zingui
30 Dichapetalaceae Dichapetalum cymulosum np GD Lwcl Cam Grand Batanga, Campo, Bipindi, Lolodorf and Douala
(Oliv.) Engl.*
31 Dichapetalaceae Dichapetalum librevillense np bu Lwcl Lg Northern limit of distribution, from Gabon to Mt d’Eléphant

1239
Pellegr.* and Campo
 Appendix 1 (Continued)

1240
No. Family Species Guild Star Habit Chorology Notes
32 Dichapetalaceae Dichapetalum oliganthum np BK Lwcl Sw-Cam Grand Batanga, Campo, Mt d’Eléphant, Kribi, Longi and
Breteler* Lolodorf.
33 Dichapetalaceae Tapura tchoutoi Breteler sb BK Sh Campo-Ma’an Rare species, only known from few collections around Bifa and
Dipikar Island
34 Dryopteridaceae Lastreopsis davalliaeformis sb bu He Lg Northern limit of distribution, from Gabon to Bipindi and
(Tardieu) Tardieu* Zingui
35 Ebenaceae Diospyros alboflavescens sb BK Tr Sw-Cam Rare species, only known from few collections from Bifa,
(Gürke) F.White Zingui and Bipindi
36 Ebenaceae Diospyros soyauxii Gürke sb Bu Tr Lg Northern limit of distribution, from Gabon to Campo and
and K. Schum. Zingui
37 Euphorbiaceae Afrotrewia kamerunica Pax sb BK Sh Campo-Ma’an Rare species, only known from Massif des Mamelles
and Hoffm.*
38 Gnetaceae Gnetum buchholzianum Engl. np GD Hcl Cam Dipikar Island, Onoyong, Ma’an, Littoral, South-west and
[180]

South provinces of Cameroon

39 Gramineae Guaduella mildbraedii Pilg.* sb BK Hb Campo-Ma’an Rare species, only known from few collections in the Campo
area
40 Gramineae Hyparrhenia wombaliensis pi bu Hb Lg Northern limit of distribution, from Congo to Campo
(Vanderyst ex Robyns)
Clayton*
41 Guttiferae Garcinia conrauana Engl. Sb GD Tr Cam Akom II, South-west and South Cameroon
42 Guttiferae Garcinia densivenia Engl. ri GD Tr Cam Dipikar Island, Ebianemeyong, Mvini, Littoral and South
Cameroon
43 Icacinaceae Alsodeiopsis zenkeri Engl. rh GD Sh Cam Frequent along the Bongola and Ntem rivers, and other rivers
in Littoral, East and South Cameroon
44 Icacinaceae Iodes kamerunensis Engl. sb GD Swcl Cam Akom II, Dipikar Island, Bipindi, Bertoua and Nanga Eboko
45 Icacinaceae Rhaphiostylis ovalifolia Engl. sb GD Swcl Cam Coastal forest around Kribi, Grand Batanga, Lolabe, Elabi
ex Sleumer* Massif des Mamelles, Littoral and South Cameroon
46 Icacinaceae Rhaphiostylis subsessilifolia sb BK Swcl Campo-Ma’an Rare species, only known from Grand Batanga, Ebianemeyong
Engl. and Mt d’Eléphant
 47 Ixonanthaceae Ochthocosmus calothyrsus np bu Tr Lg Northern limit of distribution, from Gabon to Cameroon
(Mildbr.) Hutch. and Dalziel (frequent in the Campo-Ma’an area)
48 Lauraceae Beilschmiedia cuspida sb BK Tr Campo-Ma’an Rare species, only known from Fenda and Akom II
(K. Krause) Robyns and
Wilczek
49 Lauraceae Beilschmiedia dinklagei sb BK Tr Campo-Ma’an Rare species, only known from few records around Grand
(Engl.) Robyns and Batanga
Wilczek*
50 Lauraceae Beilschmiedia klainei Robyns sb BK Tr Sw-Cam Rare species, only known from few records from Akom II,
and Wilczek Ebianemeyong and Bipindi
51 Lauraceae Beilschmiedia papyracea sb BK Tr Sw-Cam Rare species, only known from Ebianemeyong, Akom II,
(Stapf) Robyns and R.Wil- Fenda and Bipindi
czek
52 Lauraceae Beilschmiedia welczekii sb BK Tr Sw-Cam Akom II, Mvini, Nkoelon, Dipikar Island, Ebianemeyong,
Fouilloy Ma’an, Bipindi and Lolodorf
53 Leguminosae Amphimas ferrugineus Pierre np bu Tr Lg Northern limit of distribution, from Gabon to Dipikar Island,
[181]

-Caesalpinioideae ex Pellegr. Ma’an, Onoyong and Akom II

54 Leguminosae Anthonotha leptorrhachis sb GD Tr Cam Bifa, Campo, Dipikar Island, Lobe, Massif des Mamelles, Mt
-Caesalpinioideae (Harms) J.Léonard d’Eléphant, Bipindi, Lolodorf and Mt Cameroon
55 Leguminosae Aphanocalyx hedinii np GD Tr Cam Akom II, Ebianemeyong, Kom, Ma’an, Bipindi and Eseka
-Caesalpinioideae (A.Chev.) Wieringa
56 Leguminosae Aphanocalyx ledermannii sw bu Tr Lg Northern limit of distribution, occurs along rivers from Gabon,
-Caesalpinioideae (Harms) Wieringa Equatorial Guinea to the Dipikar Island
57 Leguminosae Copaifera religiosa np bu Tr Lg Northern limit of distribution, from Congo to Akom II and
-Caesalpinioideae J.Léonard Efoulan
58 Leguminosae- Daniellia klainei A.Chev. ri bu Tr Lg Northern limit of distribution, from Congo to Akom II, Eoulan
Caesalpinioideae and Ma’an
59 Leguminosae Dialium zenkeri Harms sb BK Tr Sw-Cam Campo, Dipikar Island, Onoyong, Bipindi and Lolodorf
-Caesalpinioideae
60 Leguminosae Gilbertiodendron pachyant- np BK Tr Sw-Carm Ebianemeyong, Kom, Massif des Mamelles, Bipindi and Lol-
-Caesalpinioideae hum (Harms) J.Léonard odorf
61 Leguminosae Plagiosiphon longitubus sb BK Tr Sw-Cam Akom II, Efoulan, Ma’an, Bipindi, and Lolodorf

1241
-Caesalpinioideae (Harms) J.Léonard
 Appendix 1 (Continued)

1242
No. Family Species Guild Star Habit Chorology Notes
62 Leguminosae Plagiosiphon multijugus sb GD Tr Cam Akom II, Dipikar Island, Ma’an, Bipindi and Kribi-Edea areas
-Caesalpinioideae (Harms) J.Léonard
63 Leguminosae Tetraberlinia moreliana sb bu Tr Lg Northern limit of distribution, from Gabon, Bidou and Mt.
-Caesalpinioideae Aubrév.* D’Eléphant
64 Liliaceae Chlorophytum petrophyllum sb GD Hb Cam Bifa, Dipikar Island, Mvini, Littoral and South Cameroon
K.Krause
65 Loganiaceae Mostuea neurocarpa Gilg sb bu Sh Lg Northern limit of distribution, from Gabon to Bifa, Campo and
Dipikar Island
66 Loganiaceae Strychnos canthioides Leeu- np BK Lwcl Campo-Ma’an Rare species, only known from few collections around Grand
wenb. * Batanga and Lolabe
67 Loganiaceae Strychnos elaeocarpa Gilg ex ri GD Tr Cam Akom II, Dipikar Island, Ebianemeyong, Onoyong, Bipindi,
Leeuwenb. Lolodorf, Kribi-Edea and South-west Cameroon
68 Loganiaceae Strychnos mimfiensis Gilg ex np GD Lwcl Cam Dipikar Island, Mvini, Ma’an, Bipindi, Masok, Douala-Edea-
[182]

Leeuwenb. Kribi areas.

69 Loranthaceae Tapinanthus preussii (Engl.) pa GD Pa Cam Grand Batanga, Bongola, Bipindi, Eseka, Barombi and along
Tiegh. the Lokoundje and Nyong rivers.
70 Marantaceae Hypselodelphys zenkeriana pi GD Hb Cam Ma’an, Onoyong and South Cameroon
(K.Schum.) Milne-Redh.
71 Melastomataceae Amphiblemma letouzeyi sb BK Hb Sw-Cam Rare species, only known from few collections recorded on hills
Jacq.-Fél.* around Akom II, Efoulan and Bipindi
72 Melastomataceae Calvoa calliantha Jacq.-Fél. sb BK Hb Sw-Cam Rare species, only known from Ebianemeyong, Akom II and
Bipindi
73 Melastomataceae Calvoa stenophylla sb BK Hb Campo-Ma’an Rare species, only known from type specimens collected in
Jacq.-Fél* Zingui
74 Melastomataceae Guyonia tenella Naud. sb bu Hb Lg Northern limit of distribution, from Equatorial Guinea to Lobe
and Bongola
75 Melastomataceae Memecylon arcuato-margin- sb BK Sh Cam Akom II Dipikar Island, Kom, Mt. D’Eléphant, Kienke, Longi
atum and Kribi-Edea
Gilg ex Engl. var. arcuato-
marginatum
 76 Menispermaceae Albertisia glabra (Diels and sb BK Swcl Sw-Cam Rare species, only known from Dipikar Island and Bipindi
Troupin) Forman
77 Menispermaceae Penianthus camerounensis sb GD Sh Cam Afan, Akom II, Dipikar Island, Ebianemeyong, Mekok, Lit-
A.Dekker toral, South and South-west Cameroon
78 Moraceae Dorstenia dorstenioides sb BK Hb Campo-Ma’an Rare species, only known from few collection around Kienke
(Engl.) and Fenda
Hijman and C.C.Berg*
79 Moraceae Dorstenia involuta sb BK Hb Campo-Ma’an Rare species, only known from Dipikar Island and Ma’an
M.Hijman
80 Myrsinaceae Ardisia dolichocalyx Taton sb GD Hb Cam Bifa, Campo, Dipikar Island, Onoyong, Littoral, South and
South-west Cameroon
81 Myrtaceae Eugenia kameruniana Engl.* sb BK Sh Cam Rare species, only known from Ebianemeyong, Ma’an, Nya-
bissan
82 Ochnaceae Campylospermum letouzeyi sb GD Sh Cam Dipikar Island and South Cameroon
Farron
83 Ochnaceae Campylospermum zenkeri sb GD Sh Cam Campo, Massif des Mamelles, Kribi-Edea and South Camer-
[183]

(Engl. ex Tiegh.) Farron oon

84 Ochnaceae Testulea gabonensis Pellegr. np bu Tr Lg Northern limit of distribution, from Gabon to Dipikar Island,
Ma’an and Onoyong
85 Olacaceae Octoknema dinklagei Engl. sb GD Tr Cam Akok, Grand Batanga, Lolabe, South and South-west Camer-
oon
86 Orchidaceae Bulbophyllum alinae ep BK Ep Campo-Ma’an Rare species, only known from few collections on Mt d’Elé-
Szlachetko* phant
87 Orchidaceae Corymborkis minima sb GD Hb Cam Rare species, only known from few collections around Campo,
P.J.Cribb* Lolabe and Korup National Park
88 Orchidaceae Podandriella batesii sb BK Hb Campo-Ma’an Rare species, only known from Akom II, Efoulan and Ebia-
(la Croix) nemeyong
Szlachetko and Olszewski*
89 Orchidaceae Polystachya letouzeyana ep BK Ep Campo-Ma’an Rare species, only known from Efoulan
Szlachetko and Olszewski*

1243
 Appendix 1 (Continued)

1244
No. Family Species Guild Star Habit Chorology Notes
90 Orchidaceae Vanilla africana Lindley np BK Hcl Sw-Cam Campo, Massif des Mamelles, Mt d’Eléphant and Bipindi
subsp. cucullata
(Kraenzlin and K. Shum.)
Szlachetko and Olszewski *
91 Podostemaceae Ledermanniella annithomae rh BK Hb Campo-Ma’an Rare species, only known from Memve’ele water falls
C. Cusset*
92 Podostemaceae Ledermanniella batangensis rh BK Hb Campo-Ma’an Rare species, only known from Lobe water falls
(Engl.) C. Cusset*
93 Podostemaceae Ledermanniella bosii C.Cus- rh BK Hb Campo-Ma’an Rare species, only known from the Ntem Basin, Bongola, Lobe
set and Memve’ele waterfalls
94 Podostemaceae Ledermanniella boumiensis rh bu Hb Lg Northern limit of distribution, from Gabon to the Bongola and
C. Cusset Memve’ele water falls
95 Podostemaceae Ledermanniella kamerunensis rh BK Hb Campo-Ma’an Rare species, only known from the Bongola water falls in
[184]

(Engl.) C. Cusset Dipikar Island

96 Podostemaceae Ledermanniella linearifolia rh GD Hb Cam Lobe and Bongola falls in the Campo-Ma’an area, and in the
Engl. Nkam river in Yabassi
97 Podostemaceae Ledermanniella variabilis rh GD Hb Cam Bongola and Lobe water falls, and in Mamfe river in South-
(G.Taylor) C.Cusset west Cameroon
98 Rhizophoraceae Cassipourea kamerunensis sb GD Sh Cam Akom II, Littoral and South Cameroon
(Engl.) Alston
99 Rhizophoraceae Cassipourea zenkeri (Engl.) sb GD Sh Cam Akom II Bifa, Ebianemeyong, Eboundja, Lobe, Ma’an, Bip-
Alston indi, Lolodorf and South Cameroon
100 Rubiaceae Chazaliella sciadephora (Hi- sb GD Sh Cam Mvini, Onoyong, Ma’an, Littoral and South Cameroon
ern) Petit
and Verdc, var. condensata
Verdc.
101 Rubiaceae Ecpoma apocynaceum pi BK Sh Sw-Cam Rare species, only known from Bifa, Zingui and Bipindi
K.Schum.
102 Rubiaceae Hymenocoleus glaber Robbr. sb GD Hb Cam Akom II, Dipikar Island, Massif des Mamelles, Mvini, Littoral,
South and South-west Cameroon
 103 Rubiaceae lxora aneimenodesma sb GD Sh Cam Akom II, Dipikar Island, Bipindi and Lolodorf
K.Schum. subsp.
aneimenodesma
104 Rubiaceae Ixora synactica De Block* sb BK Sh Sw-Cam Rare species, only known from Efoulan, Zingui and Bipindi
105 Rubiaceae Oxyanthus oliganthus sb GD Sh Cam Akom II, Ma’an and South Cameroon
K.Schum.
106 Rubiaceae Pavetta camerounensis sb GD Sh Cam Akom II, Bifa, Campo, Dipikar Island, Massif des Mamelles,
S.Manning subsp. Mt d’Eléphant, Littoral and South Cameroon
camerounensis
107 Rubiaceae Pavetta kribiensis J.Manning sb BK Sh Sw-Cam Rare species, only known from Mvini, Bipindi and Lolodorf
108 Rubiaceae Pavetta mpomii S.Manning sb BK Sh Sw-Cam Mt d’Eélephant, Mvini, Nkoelon, Ebianemeyong, Bipindi and
Lolodorf
109 Rubiaceae Pavetta staudtii Hutch. and sb GD Sh Cam Dipikar Island, Mvini, Nkoelon and South Cameroon
Dalziel
110 Rubiaceae Pseudosabicea medusula np GD Hb Cam Ebianemeyong, Nyabissan, Ma’an, Centre and South Camer-
(K.Schum.) N.Hallé oon
[185]

111 Rubiaceae Psychotria aemulans sb BK Sh Campo-Ma’an Rare species, only known from few collections around Grand
K. Schum.** Batanga
112 Rubiaceae Psychotria batangana sb BK Sh Campo-Ma’an Rare species, only known from few collections around Grand
K. Schum.* Batanga
113 Rubiaceae Psychotria camerunensis sb GD Sh Cam Akom II, Bifa, Ma’an, Bipindi, Lolodorf, Centre and South
Petit Cameroon
114 Rubiaceae Psychotria dimorphophylla ri BK Sh Campo-Ma’an Rare species, only known from few collections from Grand
K. Schurm.* Batanga and Lobe
115 Rubiaceae Psychotria lanceifolia sb BK Sh Sw-Cam Rare species, only known from Akom II, Onoyong, Bipindi and
K.Schum. Lolodorf
116 Rubiaceae Psychotria oligocarpa sb BK Sh Campo-Ma’an Rare species, only known from few collections around Grand
K.Schum.* Batanga
117 Rubiaceae Psychotria sadebeckiana sb GD Sh Cam Akok, Bifa, Campo, Dipikar Island, Kom, Mvini and South
K.Schum.var. elongata Petit Cameroon
118 Rubiaceae Psychotria sadebeckiana sb GD Sh Cam Akom II, Dipikar Island, Massif des Mamelles, Mvini, and
K.Scham. var. sadebeckiana South Cameroon

1245
 Appendix 1 (Continued)

1246
No. Family Species Guild Star Habit Chorology Notes
119 Rubiaceae Tricalysia amplexicaulis sb GD Sh Cam Dipikar Island, Massif des Mamelles, Centre and South Cam-
Robbr. eroon
120 Rubiaceae Tricalysia talbotii sb GD Sh Cam Ebianemeyong, Mvini, Centre and South Cameroon
(Wemham) Keay
121 Rubiaceae Vangueriella laxiflora sb GD Swcl Cam Mvini, Nkoelon, Centre and South Cameroon
(K.Schum.) Verdc.
122 Sapindaceae Deinbollia macroura Gilg ex sb BK Sh Campo-Ma’an Rare species, only known from few collections around Campo
Radlkofer*
123 Sapindaceae Deinbollia mezilii D.W.Tho- sb BK Sh Campo-Ma’an Rare species, only known from Bifa, Massif des Mamelles and
mas and D.J.Harris Dipikar Island
124 Sapindaceae Deinbollia pycnophylla Gilg sb bu Sh Lg Northern limit of distribution, from Gabon to Dipikar Island
ex Radlk.
125 Scytopetalaceae Pierrina zenkeri Engl. sb GD Sh Cam Bifa, Campo, Ebianemeyong, Ma’an, Nyabissan, Littoral and
[186]

South Cameroon
126 Scytopetalaceae Rhaptopetalum sessilifoIium sb BK Sh Sw-Cam Rare species, only known from few collections around Efoulan
Engl.* and Bipindi
127 Sterculiaceae Cola fibrillosa Engl. and sb BK Tr Sw-Cam Rare species, only known from few collections around Dipikar
Krause Island and Bipindi
128 Sterculiaceae Cola letouzeyana Nkongm. sb GD Sh Cam Akora II, Dipikar Island, Ebianemeyong, Onoyong, Centre and
South Cameroon
129 Sterculiaceae Cola praeacuta Brenan and sb GD Sh Cam Bifa, Dipikar Island, Massif des Mamelles, South and South-
Keay west Cameroon
130 Sterculiaceae Scaphopetalum acuminatum sb BK Sh Carapo-Ma’an Rare species, only known from few collections from Efoulan
Engl. and K. Krause* and Fenda
131 Sterculiaceae Scaphopetalum brunneo-pur- sb BK Sh Campo-Ma’an Rare species, only known from few collections from Fenda and
pureum Engl. and K. Kra- Zingui
use**
132 Sterculiaceae Scaphopetalum zenkeri sb BK Sh Sw-Cam Akom II, Dipikar Island, Ebianemeyong, Bipindi and Lolodorf
K-Schum.
 133 Thymelaeaceae Dicranolepis glandulosa sb GD Sh Cam Akom II, Dipikar Island, Grand Batanga, Campo, Littoral
H.H.W.Pearson South, and South-west Cameroon
134 Urticaceae Urera gravenreuthi Engl. pi GD Hcl Cam Dipikar Island, Ma’an, Littoral, South and South-west Cam-
eroon
135 Violaceae Allexis zygomorpha sb BK Sh Cam Coastal forest between Edea and Campo, Bidou, Akok, Longi,
Achoundong and Onana* Bipindi and Lolodorf
136 Violaceae Rinorea campoensis sb BK Sh Campo-Ma’an Rare species, only known from Campo, Dipikar Island, Lobe
M. Brandt ex Engl. and Massif des Mamelles
137 Violaceae Rinorea microglossa Engl.* sb BK Sh Sw-Cam Efoulan, Bipindi, Lolodorf, Centre and South Cameroon
138 Violaceae Rinorea sp. nov. 1 ined.* sb GD Sh Cam Coastal forest between Kribi and Campo, Dipikar Island, and
Douala-Edea-Kribi regions
139 Violaceae Rinorea sp. nov. 2 ined.* sb GD Sh Cam Kienke, Massif des Mamelles, Dipikar Island, Kribi, Kribi-
Edea, Douala- Yaounde, and Eseka regions
140 Zingiberaceae Aulotandra kamerunensis Sb BK Hb Sw-Cam Rare species, only known from few collections from Ebia-
Loes. nemeyong, Nyabissan and Bipindi
141 Zingiberaceae Renealmia densispica Sb BK Hb Sw-Cam Rare species, only known from few collections from Dipikar
[187]

Koechlin Island, Ebianemeyong and Ambam

Those species that reach their northern or southern limit of distribution in the Campo-Ma’an area are also included in the list.
*Species strictly endemic to the Campo-Ma’an area that were not recorded in the National Park.
**Species for which the status or range needs more investigation.
Guild: ep, epiphyte; np, non-pioneer light demanding; pi, pioneer; rh, rheophyte; ri, riverine; sb, shade-bearer; and sw, swamp.
Star: as defined in Table 1.
Habit: Ep, epiphyte; Hb, herb; Hcl, herbaceous climber; He, hemi-epiphyte; Lwcl, large woody climber; Swcl, small woody climber; Pa, parasite; Sh, shrub;
and Tr, tree.
Chorology: Campo-Ma’an, strict endemic to Campo-Ma’an; Sw-Cam, endemic to southwestern part of Cameroon; Cam, endemic to Cameroon; Lg, Lower
Guinea endemic (especially those species that reach either their northern or southern limit of distribution in the Campo-Ma’an area).

1247
 Appendix 2 IUCN (1994) threat categories for 92 plant species recorded in the Campo-Ma’an area that are listed in The IUCN (2002) Red List of Threatened

1248
Species and The World List of Threatened Trees (WCMC 1998).

No. Family Species Guild Habit Chorology IUCN/WCMC

1 Acanthaceae Afrofittonia silvestris Lindau sb Hb Lg VU A1c + 2c

2 Acanthaceae Sclerochiton preussii (Lindau) C.B.Clarke sb Hb Lg EN B1 + 2e
3 Anacardiaceae Antrocaryon micraster A. Chev. and Guillaum. pi Tr Lg VU A1cd
4 Anacardiaceae Trichoscypha bijuga Engl. sb Tr Lg CR A1c + 2abc
5 Anacardiaceae Trichoscypha mannii Hook. f. sb Tr Lg VU A1C, B1 + 2c
6 Annonaceae Boutiquea platypetala Le Thomas sb Sh Lg EN A1c + 2c
7 Annonaceae Pachypodanthium barteri (Benth.) Hutch. and Dalziel sw Tr Lg VU A1c
8 Annonaceae Uvariastrum zenkeri Engl. and Diels sb Sh Lg VU A1c, B1 + 2c
9 Annonaceae Uvariodendron connivens (Benth.) R.E.Fr. sb Tr Lg LR/nt
10 Asclepiadaceae Tylophora cameroonica N.E.Br. pi Swcl Lg LR/nt
11 Boraginaceae Cordia platythyrsa Baker pi Tr Gc VU A1d
12 Burseraceae Aucoumea klaineana Pierre pi Tr Lg VU A1cd
[188]

13 Burseraceae Dacryodes igaganga Aubrev. And Pellegr. np Tr Lg VU A1cd + 2cd

14 Celastraceae Salacia lehmbachii Loes. var. pes-ranulae N.Hallé np Swcl Lg VU B1 + 2c
15 Chrysobalanaceae Dactyladenia cinera (Engl. ex de Wild) Prance and F.White sb Tr Sw-Cam CR B1 + 2c
16 Combretaceae Terminalia ivorensis A.Chev. np Tr Gu VU A1cd
17 Connaraceae Hemandradenia mannii Stapf sb Tr Lg LR/nt
18 Ebenaceae Diospyros barteri Hiern sb Tr Gu VU A1c
19 Ebenaceae Diospyros crassiflora Hiern sb Tr Gc EN A1d
20 Euphorbiaceae Amanoa strobilacea Müll.Arg. sb Sh Gu VU A1c, B1 + 2c
21 Euphorbiaceae Crotonogyne manniana Müll.Arg. sb Sh Lg LR/nt
22 Euphorbiaceae Drypetes preussii (Pax) Hutch. sb Tr Lg VU B1 + 2c
23 Euphorbiaceae Drypetes tessmanniana (Pax) Pax and K.Hoffm. sb Sh Lg CR A1c + 2c
24 Euphorbiaceae Neoboutonia mannii Benth. pi Tr Gu LR/nt
25 Euphorbiaceae Pseudagrostistachys africana (Müll.Arg.) Pax and K.Hoffm. sb Tr Lg VU A1c, B1 + 2c
26 Guttiferae Garcinia brevipedicellata (Baker f.) Hutch. and Dalziel sb Tr Lg VU A1c, B1 + 2c
27 Guttiferae Garcinia kola Heckel sb Tr Gc VU A1cd
28 Guttiferae Garcinia staudtii Engl. sb Tr Lg VU A1c, B1 + 2c
29 Hoplestigmataceae Hoplestigma pierreanum Gilg np Tr Lg CRA1c + 2c
30 Huaceae Afrostyrax kamerunensis Perkins and Gilg sb Tr Lg VU A1c, B1 + 2c
 31 Huaceae Afrostyrax lepidophyllus Mildbr. sb Tr Lg VU A1c, B1 + 2c
32 Irvingiaceae Irvingia excelsa Mildbr. np Tr Gc LR/nt
33 Irvingiaceae Irvingia gabonensis (Aubry-Lecomte ex O’Rorke) Baill. np Tr Gc LR/nt
34 Leguminosae-Caesalpinioideae Afzelia bipindensis Harms np Tr Gc VU A1cd
35 Leguminosae-Caesalpinioideae Afzelia pachyloba Harms np Tr Gc VU A1d
36 Leguminosae-Caesalpinioideae Anthonotha leptorrhachis (Harms) J.Léonard sb Tr Cam CR A1c + 2c
37 Leguminosae-Caesalpinioideae Aphanocalyx hedinii (A.Chev.) Wieringa np Tr Cam CR B1 + 2abcd, C1 + 2ab
38 Leguminosae-Caesalpinioideae Daniellia klainei A.Chev. ri Tr Lg LR/nt
39 Leguminosae-Caesalpinioideae Daniellia oblonga Oliv. np Tr Lg VU A1c
40 Leguminosae-Caesalpinioideae Dialium bipindense Harms np Tr Lg LR/nt
41 Leguminosae-Caesalpinioideae Dialium tessmannii Harms sb Tr Lg LR/nt
42 Leguminosae-Caesalpinioideae Didelotia unifoliolata J.Léonard sb Tr Lg LR/nt
43 Leguminosae-Caesalpinioideae Gilbertiodendron pachyanthum (Harms) J.Léonard np Tr Sw-Cam VU D2
44 Leguminosae-Caesalpinioideae Guibourtia ehie (A. Chev.) J. Léonard np Tr Gc VU A1c
45 Leguminosae-Caesalpinioideae Loesenera talbotii Baker f. sb Tr Lg VU A1c, B1 + 2c
46 Leguminosae-Caesalpinioideae Pellegriniodendron diphyllum (Harms) J.Léonard sb Tr Gu LR/nt
47 Leguminosae-Caesalpinioideae Plagiosiphon longitubus (Harms) J.Léonard sb Tr Sw-Cam CR A1 + 2c
[189]

48 Leguminosae-Caesalpinioideae Swartzia fistuloides Harms sb Tr Gc EN A1cd

49 Leguminosae-Mimosoideae Calpocalyx heitzii Pellegr. np Tr Lg VU A1c, B1 + 2c
50 Leguminosae-Mimosoideae Calpocalyx le
testui Pellegr. sb Tr Gc VU D2
51 Leguminosae-Mimosoideae Calpocalyx ngouniensis Pellegr. sb Tr Gc VU A1c
52 Leguminosae-Papilionoideae Craibia atlantica Dunn sb Tr Gc VU A1c
53 Leguminosae-Papilionoideae Millettia laurentii De Wild. np Tr Gc EN A1cd
54 Leguminosae-Papilionoideae Millettia macrophylla Benth. pi Tr Lg VU A1c, B1 + 2c
55 Leguminosae-Papilionoideae Ormocarpum klainei Tisser. sb Sh Lg CR A1c
56 Liliaceae Chlorophytum petrophyllum K.Krause sb Hb Cam CR A1c + 2c
57 Melastomataceae Memecylon candidum Gilg sb Sh Lg VU B1 + 2c
58 Melastomataceae Memecylon dasyanthum Gilg ex Lederman and Engl. sb Tr Lg VU B1 + 2c
59 Melastomataceae Warneckea wildeana Jacq.-Fél. sb Sh Lg VU D2
60 Meliaceae Entandrophragma angolense (Welw.) C.DC. np Tr Tra VU A1cd
61 Meliaceae Entandrophragma candollei Harms np Tr Gc VU A1 cd
62 Meliaceae Entandrophragma cylindricum (Sprague)Sprague np Tr Gc VU A1cd

1249
63 Meliaceae Entandrophragma utile (Dawe and Sprague) Sprague np Tr Gc VU A1cd
64 Meliaceae Guarea cedrata (A.Chev.) Pellegr. np Tr Gc VU A1c
 Appendix 2 (Continued)

1250
No. Family Species Guild Habit Chorology IUCN/WCMC
65 Meliaceae Guarea thompsonii Sprague and Hutch. np Tr Gc VU A1c
66 Meliaceae Khaya anthotheca (Welw.) C. DC. np Tr Gc VU A1cd
67 Meliaceae Khaya ivorensis A.Chev. np Tr Gc VU A1cd
68 Meliaceae Lovoa trichilioides Harms np Tr Gc VU A1cd
69 Meliaceae Turraeanthus africanus (Welw. ex C DC.) Pellegr. sb Tr Gc VU A1cd
70 Moraceae Milicia excelsa (Welw.) C.C.Berg pi Tr Tra LR/nt
71 Myrtaceae Eugenia kameruniana Engl. sb Sh Cam CR A1c
72 Ochnaceae Lophira alata Banks ex Gaertn.f. pi Tr Gc VU A1cd
73 Ochnaceae Testulea gabonensis Pellegr. np Tr Lg EN A1cd
74 Rhizophoraceae Anopyxis klaineana (Pierre) Engl. np Tr Gc VU A1cd
75 Rubiaceae Hallea stipulosa (DC.) Leroy sw Tr Gc VU A1cd
76 Rubiaceae Nauclea diderrichii (De Wild. And T.Durand) Merrill pi Tr Gc VU A1cd
77 Rutaceae Vepris heterophylla Letouzey sb Sh Gc EN A1c, B1 + 2c
[190]

78 Sapotaceae Autranella congolensis (De Wild.) A.Chev. np Tr Gc CR AIcd

79 Sapotaceae Baillonella toxisperma Pierre np Tr Lg VU A1cd
80 Sapotaceae Gluema ivorensis AubréV. and Pellegr. np Tr Gc VU B1 + 2c
81 Sapotaceae Tieghemella africana Pierre np Tr Lg EN A1cd
82 Simaroubaceae Nothospondias staudtii Engl. np Tr Gc VU B1 + 2c
83 Sterculiaceae Cola hypochrysea K.Schum. sw Tr Lg VU A1c
84 Sterculiaceae Cola philipi
jonesii Brenan and Keay sb Sh Lg EN B1 + 2c
85 Sterculiaceae Cola praeacuta Brenan and Keay sb Sh Cam CR A1c + 2c
86 Sterculiaceae Cola semecarpophylla K.Schum. sb Sh Lg LR/cd
87 Sterculiaceae Mansonia altissima (A.Chev.) A.Chev. var. kamerunica Jacq.-Fél. np Tr Gu EN A1cd
88 Sterculiaceae Pterygota bequaertii De Wild. np Tr Gc VU A1cd
89 Sterculiaceae Pterygota macrocarpa K.Schum. np Tr Gc VU A1cd
90 Sterculiaceae Sterculia oblonga Mast. pi Tr Gc VU A1cd
91 Violaceae Allexis cauliflora (Oliv.) Pierre sb Sh Lg VU A1c, B1 + 2c
92 Violaceae Allexis obanensis (Baker f.) Melchior sb Sh Lg VU B1 + 2c

NB: Guild, habit and chorology categories as defined in Appendix 1.

 1251

References

Aubréville A. and Leroy J.-F. (eds) 1961–1992. Flore du Gabon. Muséum National d’Histoire
Naturelle, Paris.
Aubréville A. and Leroy J.-F. (eds) 1963–2000. Flore du Cameroun, vol. 1–20. Muséum National
d’Histoire Naturelle, Paris.
Barbault R. and Sastrapradja S. 1995. Generation, maintenance and loss of biodiversity. In
Heywood V.H. and Watson R.T. (eds), Global Biodiversity Assessment. Cambridge University
Press, UNEP, pp. 192–294.
Cable S. and Cheek M. 1998. The plants of Mount Cameroon. A conservation checklist. Royal
Botanic Gardens, Kew.
Cusset C. 1987. Flore du Cameroun. In: Satabié B. and Morat Ph. (eds), Podostemaceae, vol. 30.
Muséum National d’Histoire Naturelle, Paris.
Davis S.D., Heywood V.H. and Hamilton A.C. 1994. Centres of Plant Diversity. A Guide and
Strategy for their Conservation. WWF, IUCN.
de Kam M., Fines J.-P. and Akogo G. (eds) 2002. Schéma Directeur pour le developpement de
l’Unité Technique Opérationnelle de Campo-Ma’an, Cameroun. Campo-Ma’an Series 1, .
Djama T. 2001. Inventaire quantitatif des poissons dans l’UTO Campo-Ma’an. Projet Campo-
Ma’an, Kribi, Cameroon.
ERE Développement 2002. Etude socio-économique dans l’UTO de Campo-Ma’an. Rapport final
phase 2: résultats d’énquetes auprés des ménages. SNV/Projet Campo-Ma’an.
Foahom B. and Jonkers W.B.J. 1992. A Programme for Tropenbos Research in Cameroon. Final
report. Tropenbos-Cameroon Programme (Phase I). The Tropenbos Foundation, Wageningen.
Franqueville A. 1973. Atlas regional Sud-Ouest 1. République du Cameroun. ORSTOM, Yaoundé,
Cameroun.
Gartlan J.S. 1989. La conservation des ecosystemes forestiers du Cameroun. IUCN.
Hamilton A.C. 1982. Environmental History of Africa: A Study of the Quaternary. Academic
Press, London.
Hawks worth D.L. and Kalin-Arroyo M.T. 1995. Magnitude and distribution of biodiversity. In:
Heywood V.H. and Watson R.T. (eds), Global Biodiversity Assessment. UNEP, Cambridge
University Press, Cambridge, pp. 108–191.
Hawthorne W.D. 1996. Holes and the sums of parts in Ghanian forest: regeneration, scale and
sustainable use. Proc. R. Soc. Edin. 104: 75–176.
Hawthorne W.D. and Abu-Juam M. 1995. Forest protection in Ghana with particular reference
to vegetation and plant species. The IUCN Forest Conservation Programme.
Heywood V.H. and Waston R.T. 1995. Global Biodiversity Assessment. UNEP, Cambridge
University Press, Cambridge.
ILWIS 2001. ILWIS 3.0 Academic: User’s Guide. ITC, Enschede.
Isaaks E.H. and Srivastava R.M. 1989. An Introduction to Applied Geostatistics. Oxford Uni-
versity Press, New York.
IUCN 1994. IUCN Red List Categories. IUCN, Gland.
IUCN 2002. 2002 IUCN Red list of Threatened Species. IUCN Conservation Monitoring Centre,
Cambridge.
Keay R.W.J. and Hepper F.N. (eds) 1954. Flora of West Tropical Africa. 2nd edn., 3 vols. Crown
Agents, London.
Lebrun J.P. and Stork A.L. 2003. Tropical African flowering plants-ecology and distribution,
vol. 1. Annonaceae-Balanitaceae. Conservatoire et Jardin Botanique de la Ville de Genéve.
Letouzey R. 1968. Etude phytogeographique du Cameroun. Le Chevalier, Paris.
Letouzey R. 1985. Notice de la carte phytogéographique du Cameroun, vol. 1–5. Institut de la
Carte Internationale de la Végétation, Toulouse.
Maley J. 1987. Fragmentation de la forêt dense humide Africaine et extention des biotopes mon-
tagnards du Quaternaire recent. Paleoecol. Afr. 18: 307–309.

[191]
1252

Maley J. 1989. Late Quaternary climatic changes in the African rainforest: the question of forest
refuges and the major role of sea surface temperature variations. In Lewen M. and Sarthein M.
(eds), Paleoclimatology and Paleometeolorogy: Modern and Past Patterns of Global Atmo-
spheric Transport. Kluwer, London, pp. 585–616.
Maley J. 1990. Histoire récente de la forêt dense humide africaine: essai sur le dynamisme de
quelques formations forestières. In: Lanfranchi R. and Schwartz D. (eds), Paysage quaternaire de
l’afrique centrale Atlantique pp.367–382.
Maley J. 1993. The climatic and vegetational history of the equatorial regions of Africa during the
upper Quaternary. In: Shaw T., Sinclair P., Andah B. and Okpoko A. (eds), The Archeology of
Africa: Food, Metals and Towns, pp. 43–52.
Maley J. 1996. Le cadre paléoenvironnemental des refuges forestiers africains : quelques données et
hypothéses. In: der Maesen L.J.G., der Burgt X.M. and de Rooy J.M. (eds), The Biodiversity of
African Plants. Kluwer Academic Publishers, Dordrecht, pp. 519–535.
Matthews A. and Matthews A. 2000. Primate Population and Inventory of Large and Medium
Sized Mammals in the Campo-Ma’an Project Area in Southwest Cameroon, including Man-
agement Recommendations. Projet Campo-Ma’an, Kribi, Cameroun.
Muller J.P. 1979. Carte des sols du Cameroun. Atlas de la République Unie du Cameroun.
Myers N. 1988. Threatened biotas ‘hotspots’ in tropical forests. Environmentalist 8: 187–208.
Myers N., Mittermeir R.A., Mittermeir C.G., da Fonseca G.A.B and Kent J. 2000. Biodiversity
hotspots for conservation priorities. Nature 403: 853–858.
Pebesma E.J. and Wesseling C.G. 1998. GSTAT: a program for geostatistical modelling, prediction
and simulation. Comput. Geosci. 24: 17–31.
R Development Core Team 2002. The R Environment for Statistical Computing and Graphics:
Reference Index. The R Foundation for Statistical Computing, Vienna. Version 1.6.2.
Satabié, B. and Leroy J.-F. 1980. Flore du Cameroun, vol. 21–28. Muséum National d’Histoire
Naturelle, Paris.
Satabié B. and Morat Ph. 1986–2001. Flore du Cameroun, vol. 29–37. Muséum National d’Histoire
Naturelle, Paris.
Sosef M.S.M. 1994. Refuge Begonias. Taxonomy, phylogeny and historical biogeography of
Begonia sect. Loasibegonia and sect. Scutobegonia in relation to glacial rain forest refuges in
Africa. PhD Thesis, Wageningen Agricultural University.
Sosef M.S.M. 1996. Begonias and African rain forest refuges : general aspects and recent progress.
In: der Maesen L.J.G., der Burgt X.M. and de Rooy J.M. (eds), The Biodiversity of African
Plants. Kluwer Academic Publishers, Dordrecht, pp. 602–611.
Tchouto P., Pouakouyou D. and Acworth J. 1998. Rapid Botanical Survey methodology. Tech-
nical report to DFID/Mount Cameroon Project. Limbe Botanic Garden and Herbarium.
Tchouto M.G.P. 2004. Plant diversity in a Central African rain forest: implifications for biodi-
versity conservation in Cameroon. Phd Thesis, Wageningen University.
Thomas D.W. and Thomas J.C. 1993. Botanical and ecological survey of the Campo-Ma’an area.
A report to the World Bank, Washington.
Van Gemerden B.S. and Hazeu G.W. 1999. Landscape ecolofical survey (1:100,000) of the Bipindi-
Akom II-Lolodorf region, Southwest Cameroon. Tropenbos-Cameroon Document 1.
Vivien J. 1991. Faune du Cameroun. Guide des mammiféres et poisons. GICAM, Cameroun.
WCMC 1998. The World List of Threatened Trees. World Conservation Press, .
Webster R. and Oliver M.A. 2001. Geostatistics for Environmental Scientists. Wiley & sons,
Chichester.
White F. 1979. The Guineo-Congolian region and its relationships to other phytochoria. Bull. Jard.
Nat. Belg./Bull. Nat. Plant. Belg. 49: 11–55.
White F. 1983. The Vegetation of Africa. UNESCO, Paris.
Williams P.H. 1993. Measuring more of biodiversity for chosing conservation areas, using taxo-
nomic relatedness. In: Moon T.-Y. (ed.), International Symposium on Biodiversity and Con-
servation. Korean Entomological Institute, Seoul, pp. 199–227.

[192]
Biodiversity and Conservation (2006) 15:1253–1270
Springer 2006
DOI 10.1007/s10531-005-0772-x

-1

Tree diversity in western Kenya: using profiles to

characterise richness and evenness

R. KINDT1,*, P. VAN DAMME2 and A.J. SIMONS1

1
ICRAF, PO Box 30677-00100, Nairobi, Kenya; 2Ghent University, Laboratory of Tropical and
Subtropical Agriculture and Ethnobotany, Coupure Links 653, B-9000 Gent, Belgium; *Author for
correspondence (e-mail: [email protected]; phone: 254-2-524000 or 1-650-8336645; fax: 254-2-
524001 or 1-650-8336646)

Received 18 May 2004; accepted in revised form 7 January 2005

Key words: Accumulation, Agroforestry, Diversification, Domestication, Evenness, Rarefaction,

Rényi diversity profile, Richness

Abstract. Species diversity is a function of the number of species and the evenness in the abundance
of the component species. We calculated diversity and evenness profiles, which allowed comparing
the diversity and evenness of communities. We applied the methodology to investigate differences
in diversity among the main functions of trees on western Kenyan farms. Many use-groups (all
trees and species that provide a specific use) could not be ranked in diversity or evenness. No use-
group had perfectly even distributions. Evenness could especially be enhanced for construction
materials, fruit, ornamental, firewood, timber and medicine, which included some of the most
species-rich groups of the investigated landscape. When considering only the evenness in the dis-
tribution of the dominant species, timber, medicine, fruit and beverage ranked lowest (>60% of
trees belonged to the dominant species of these groups). These are also use-groups that are mainly
grown by farmers to provide cash through sales. Since not all communities can be ranked in
diversity, studies that attempt to order communities in diversity should not base the ordering on a
single index, or even a combination of several indices, but use techniques developed for diversity
ordering such as the Rényi diversity profile. The rarefaction of diversity profiles described in this
article could be used in studies that compare results from surveys with different sample sizes.

Introduction

One of the objectives of tree domestication research, in general and more

specific, in western Kenya is the diversification of tree species composition in
agroecosystems (Kindt and Lengkeek 1999; Kindt et al. 2004). In the realm of
agroforestry, underpinning the need for diversification is the desire to enhance
the stability and productivity of agroecosystems (ICRAF 1997; Atta-Krah
et al. 2004).
Diversity means different things to different people. Most often in natural or
agricultural systems, species counts (species richness) are provided as the
measure of diversity. Continuing this logic, diversification means adding more
species. Species diversity, however, is a function of the number of species, and
the evenness in distribution of species’ abundances (Magurran 1988; Purvis and
Hector 2000). Options for diversification can therefore be dissociated into
interventions that target richness and those that target evenness.

[193]
1254

We investigated the diversity of various groups of trees that contributed to

similar service or production functions on farms in western Kenya, as our
target was diversification within these functions, and not of overall diversity.
Since heterogeneity in characteristics of species results in saturating effects of
increased diversity on ecosystem functioning (e.g., Hector et al. 1999; Loreau
et al. 2001; Tilman et al. 2001), we expect larger effects from increasing
diversity within communities of lower diversity. Ranking of use-groups in
diversity, therefore, allows prioritising their scope for diversification measured
as the average expected effect of adding one species. In addition, such
approaches provide the opportunity to model effects of replacement, substi-
tution and expansion at fixed and varying tree densities.
Appropriate techniques for measuring diversity and evenness were used in
this article. Single indices of diversity or evenness can result in wrong inter-
pretation since not all communities can be ranked in diversity or evenness,
whereas they can always be ranked on the basis of a single index (Taillie 1979;
Tóthmérész 1995; Ricotta and Avena 2002; Ricotta 2003). Since diversity
research should not be based on single indices of diversity or evenness, tech-
niques for diversity and evenness ordering were used that produce diversity and
evenness profiles.

Material and methods

Study area

Complete tree inventories were made on 201 stratified-randomly selected farms

(taken to mean all land managed by a single household) in the Vihiga and
Kakamega districts of western Kenya. The study area is inhabited predomi-
nantly by the Luhya (Luyia) ethnic group and belongs to the same agroeco-
logical zone where altitude ranges 1500–1800 m above sea level, annual mean
temperature ranges 18.1–20.4 C, and annual bimodal rainfall ranges 1600–
2000 mm (Jaetzhold 1982). Four villages were selected within the area, each
located in a different stratum that mainly differed in farm sizes and arrange-
ment of woody biomass in the landscape (Bradley et al. 1985; Bradley 1991).
The selection of villages coincided with a gradient towards the species-rich
Kakamega Forest National Reserve.

Information recorded on tree species

All trees (woody perennials) were censused using Beentje (1994) as the key
reference. For each tree species encountered on a farm, its abundance (the total
number of trees) and uses (see below) were recorded by participatory interviews
with household informants involving farm walks, tree counting by the inter-
viewer and data recording on a species-by-species basis.

[194]
 1255

Households listed all the products or services (uses) that are provided by
the different species encountered on their farm. Free responses were obtained
on tree uses (primary and all secondary uses) that were postcoded during
data entry and checking. In total, 60 use categories were recorded, but
analyses were only conducted for the 12 use categories that occurred on more
than 20% of farms. Because it is possible that some informants could have
forgotten some uses of particular species, information was adjusted
(increasing species–farm-use combinations from 6859 to 7526) by always
including all trees of a species in a use-group if more than 50% of farmers
and minimum five farmers with the species mentioned the use.
Use-groups were defined as all the trees encountered in the survey that
provided one particular type of use. These use-groups were analysed by
matrices with as rows the information collected on a particular farm and as
columns the various species encountered. Each cell of a use-group matrix
provides the number of trees of each species that was used for a particular
function on a particular farm.

Diversity and evenness profiles

The Rényi diversity profile is one of the techniques for diversity ordering that
were specifically designed to rank communities from low to high diversity.
Rényi diversity profile values (Ha) are calculated from the frequencies of each
component species (proportional abundances pi = abundance of species i/
total abundance) and a scale parameter (a) ranging from zero to infinity
(Tóthmérész 1995; Legendre and Legendre 1998) as:
P a 
ln pi
Ha ¼
1
a
It can be demonstrated that values of the Rényi profile at the respective
scales of 0, 1, 2 and ¥ are related to species richness S, the Shannon diversity
index H, the Simpson diversity index D
1 and the Berger–Parker diversity
index d
1 (Magurran 1988; Legendre and Legendre 1998; Shaw 2003):
H0 ¼ lnðSÞ

X
H1 ¼ H ¼
pi log pi

X 

1
H2 ¼ lnðD
1 Þ ¼ ln ðp2i Þ

H1 ¼ lnðd
1 Þ ¼ lnðp
1
max Þ

[195]
1256

Community A is more diverse than a community B if the diversity profile for

community A is everywhere above the diversity profile for community B.
Communities that have intersecting profiles cannot be ordered in diversity. The
fact that intersecting profiles (partial diversity ordering) could occur explains
why ordering techniques such as the Rényi series are needed, since a single
diversity index will not provide sufficient information. The values of the series
for the various use-groups were calculated for scales a 2{0,0.25, 0.5, 1, 2, 4, 8,
¥}.
From the diversity profile we derived an evenness profile (ln Ea,0) that orders
communities in evenness in a similar way that diversity profiles order com-
munities in diversity (Kindt et al. 2001; Ricotta and Avena 2002):
ln Ea;0 ¼ Ha
H0
Partial ordering may occur where evenness profiles intersect, indicating that
intrinsic evenness ordering is not possible. For example, community A with
species abundances of (70, 20, 10) can only be partially ordered with com-
munity B with species abundances of (60, 35, 5) and it is therefore not possible
to identify the community of largest evenness. The example above is a rare case
where diversity ordering equals evenness ordering since species richness is equal
in community A and B. Since species richness differs between most commu-
nities, diversity and evenness ordering can only be inferred through separate
ordering techniques that conform to diversity ordering (such as the Rényi
diversity series) or to evenness ordering (such as Hill’s Ea,0 subfamily, see
below).
Since H0 = ln S (as shown above), rearranging the above formula shows
that the Rényi diversity profile can be decomposed as:
Ha ¼ ln S þ ln Ea;0

This decomposition has the attractive feature in showing mathematically

that diversity combines information on species richness and evenness, which
conforms to the definition of diversity (Magurran 1988; Purvis and Hector
2000; Shaw 2003). The decomposition of the Rényi profile is a generalisation of
the decomposition of the Shannon index H1 as ln S + ln E (Hayek and Buzas
1997; Kindt et al. 2001).
Ea,0 is a subfamily of Hill’s parametric evenness (Hill 1973; Ricotta and
Avena 2002; Ricotta 2003) defined as:
Na exp Ha
Ea;0 ¼ ¼
N0 exp H0

Ea,0 has several desirable properties as an evenness function: it is consistent

with the Lorenz ordering and therefore meets the general requirements for an
evenness index (Taillie 1979; Rousseau et al. 1999; Ricotta 2003), it is formally
related to diversity (Ricotta and Avena 2002; Ricotta 2003), and it is norma-
lised between zero and one.

[196]
 1257

Since ln Ea,0 is a monotone transformation of Ea,0, both these measures

provide the same rank order when communities are compared for evenness
despite the fact that ln Ea,0 is not normalised between zero and one. In analogy
with diversity profiles (Tóthmérész 1995), the various techniques provide the
same ordering, but differ in the resolution of the graphs that they produce.

Accumulation patterns for the Re´nyi series

We calculated the Rényi diversity profile for each use-group by calculating

the frequencies of each species in the entire survey. These profiles allow
comparing the total diversity of the various use-groups at the scale of the
total survey. Since some use-groups did not occur on each farm, another
comparison was made by only including farms where a specific use-group
occurred. Because the total number of farms per use-group differed, profiles
were rarefied to the same number of farms (the number of farms of the
smallest use-group) so that effects from differences in sample size could be
removed from the analysis.
The rarefaction of the diversity profile was achieved through a Monte-Carlo
approach of 1000 random additions of farms (selecting the first farm at ran-
dom, adding the second farm at random,…) using sampling without replace-
ment. We obtained accumulation surfaces for diversity profiles by calculating
the average Ha for each subset of 1, 2, …, all farms combined, and for each
a2{0,0.25, 0.5, 1, 2, 4, 8, ¥} (Kindt et al. 2001). A similar approach of rare-
faction is used to calculate species accumulation curves for random pooling of
sample units (e.g., Colwell 1997; Gotelli and Colwell 2001). FORTRAN and R
statistical programmes developed to carry out the computations can be
obtained from the authors (Kindt 2001, 2004).
H¥ was the only value for which 95% confidence interval (CI) limits for the
expected values for the entire survey area could be calculated as it is obtained
solely from the proportion of the dominant species, while other values in the
diversity series include an effect of species richness. Hayek and Buzas (1997)
propose to calculate the variance of species frequency (p) in the case of cluster
sampling as
P
n
m2i ðpi
pÞ2
_2
r pclus ¼ i¼1
ðn
1Þm2 n
1
(mi = total abundance of farm i, pi = species frequency in farm
i, p = species frequency of the total survey, n = number of farms, and
m = total abundance). 95% CI limits for the proportion of the dominant
species can then be calculated as:
_
p  t  r pclus

[197]
1258

The limits of the CI for the proportion were y=ln(x
1) transformed to

calculate a CI for H¥.

Results

Diversity ordering and examination of richness and evenness contributions

Figure 1 shows the Rényi diversity ordering of all trees and the 12 most fre-
quent use-groups. By examining the values at scales 0, 1, 2, and ¥, species
richness and values of Shannon, Simpson, and Berger–Parker diversity indices
can be inferred. The many intersections in the figure show the difficulties to
order most groups in diversity. Figure 2 shows many groups with intersecting
evenness profiles, which is an indication that many groups cannot be ranked in
evenness. Table 1 provides some more precise statistics for certain profile
values than can be inferred from the figures, and also gives some parameters
that describe the pattern of the various diversity profiles (see below).
We differentiated between five pools of use-groups based on their total
species richness (H0) (which is an ordering of use-group only based on rich-
ness): (i) pool A – very high richness, including all trees and firewood; (ii) pool
B – high richness, including shade; (iii) pool C – medium richness, including

Figure 1. Diversity profiles based on the Rényi series Ha for all trees and trees belonging to
particular use-groups.

[198]
 1259

Figure 2. Evenness profiles for all trees and trees belonging to particular use-groups.

Table 1. Values for specific profile values of the diversity and evenness profiles depicted in
Figures 1 and 2 with for all trees and for the most frequent use-groups in western Kenya.

Use-group H0 H1 H2 H¥ ln E1,0 E1,0 H¥,I H¥,u

All trees 5.16 2.79 2.38 1.78
2.37 0.09 1.61 1.98
Firewood 5.04 2.69 2.29 1.71
2.35 0.10 1.54 1.92
Shade 4.43 3.00 2.60 2.04
1.43 0.24 1.43 3.84
Medicine 4.06 1.59 0.79 0.41
2.47 0.08 0.15 0.77
Ornamental 3.97 1.34 0.94 0.68
2.63 0.07 0.37 1.12
Timber 3.89 1.11 0.77 0.48
2.78 0.06 0.36 0.61
Boundary 3.53 1.95 1.69 1.17
1.58 0.21 0.97 1.44
Soil fertility 3.30 1.44 1.08 0.64
1.86 0.16 0.17 1.57
Charcoal 3.26 1.29 0.95 0.70
1.97 0.14 0.29 1.41
Fruit 3.22 1.08 0.61 0.32
2.13 0.12 0.24 0.41
Construction 3.00 0.66 0.43 0.24
2.33 0.10 0.17 0.32
Fodder 1.95 1.61 1.40 0.87
0.33 0.72 0.44 1.63
Beverage 1.39 0.50 0.38 0.22
0.89 0.41 0.04 0.44
H¥;l and H¥;u are 95% CI limits for H¥. See methods for the formulas.

medicine, ornamental and timber; (iv) pool D – moderate richness, including

boundary demarcation, soil fertility enhancement, charcoal, fruit and con-
struction wood; and (v) pool E – low richness, including fodder and beverage.
Some differences and similarities in diversity and evenness can subsequently be
detected between and within pools (Figures 1 and 2).

[199]
1260

When comparing diversity and evenness patterns for pools, it can be

observed that use-groups of pool A and B were more diverse than the other
use-groups. The evenness profile (Figure 2) shows that shade (pool B) has a
more even distribution than use-groups of pool A. Diversity profiles of
use-groups of pool C intersected all use-groups of pool D and E, with the
exception of fruit and construction (pool D) and beverage (pool E). The
evenness profiles, however, show that ornamental and timber (pool C) had
lower evenness than all use-groups of pool D, with exception of construction
for which intersection can be observed. Fodder and beverage (pool E) had the
most even distribution of all use-groups, with the exception of an intersection
for beverage and shade.
When comparing diversity for use-groups that belong to the same pool, it
can be seen that all trees are more diverse but that firewood is more even within
pool A although the profiles were almost parallel to each other. The evenness
profile of medicine intersected the other use-groups of pool C. All use-groups
of pool C had intersecting evenness profiles. These groups can therefore not be
ranked in evenness within pool C. Most use-groups within pool D could be
ranked in diversity, since their diversity profiles did not intersect. Boundary
demarcation is the most diverse and evenly distributed use-group within the
pool.
Construction is the least diverse use-group and fruit the second least diverse
of pool D. Their evenness profiles intersect, however. Charcoal and soil fertility
had intersecting diversity and evenness profiles. Within pool E, fodder is more
diverse and more evenly distributed.
Values of H¥ (Figure 1 and Table 1) lower than 0.5 indicate that the dom-
inant species of timber, medicine, fruit, construction and beverage contains
more than 60% of all trees of these use-groups. Values of H¥ between 0.5 and
1.5 indicate that the dominant species contains between 22 and 60% of all trees
for soil fertility, charcoal, ornamental, fodder and boundary demarcation. The
dominant species contains a smaller percentage of trees for the other use-
groups. However, values of ln E¥,0 (Figure 2) show that the dominant species
for all trees and firewood are less evenly distributed when comparing with
many other use-groups (the formula ln E¥,0 = H¥
H0 implies that when
species richness increases, the frequency of the dominant species should de-
crease to maintain the same ln E¥,0). Medicine had the least evenly distributed
dominant species.

Effects of sample size on diversity

Figure 3 shows the average profile values and associated 95% CI for subs-
amples of 47 farms (the number of farms with fodder, the least frequent use-
group). Figure 3 indicates that, on average, similar profiles are obtained as for
the full sample. Diversity profiles, especially medicine and shade, had large
95% CI, however. The 95% CI still allow classifying use-groups in the richness

[200]
 1261

pools that we differentiated above. Construction of the original pool D ob-

tained richness values closer to the range of pool E. Within pools, most 95% CI
of diversity profile values overlap, although the averages showed a similar
pattern in Figures 1 and 3.
Figures 4–6 show that several use-groups (especially all trees, firewood,
boundary, timber, fruit and construction) obtain stable (asymptotic) values for
H1, H2 and H¥ after 40 farms were sampled. Some other use-groups showed
some strong non-linear patterns throughout the range of pooled farms. These
use-groups were especially shade, fodder, medicine, and soil fertility. Whereas
the non-linear patterns were for the larger part increments of the index with
increasing sample size, a declining pattern could be observed for medicine for
H2 and H¥. The observed patterns suggest that to extrapolate profile values, in
some cases (asymptotic patterns) similar values can be expected, whereas in
other cases different values would be expected (non-linear patterns). However,
extrapolation should be used carefully since we had no actual data for larger
sample sizes.
Table 1 (H¥,l and H¥,u) indicates the CI that is expected for the average
value of H¥ when all farms would be sampled in the survey area (assuming
random sampling of the farms). Relatively large CI were even obtained for

Figure 3. Averages of rarefied diversity profiles based on the Rényi series Ha for all trees and trees
belonging to particular use-groups calculated from 1000 random subsamples of 47 farms, with 95%
CI. All profile values were calculated for the same scales, but groups were presented at different
scales for better discrimination.

[201]
1262

Figure 4. Accumulation curve for the Shannon diversity index for all trees and trees belonging to
particular use-groups. The vertical reference indicates the sample size of 47 farms depicted in
Figure 3.

groups where stable values were obtained at the respective scale in Figure 6
(for example timber, fruit, construction, and beverage). However, these CI
were among the smallest. Large CI corresponded to non-linear patterns for the
Berger–Parker index (Figure 6, see discussion above). Medicine which had a
non-linear negative pattern for the Berger–Parker index had a relatively small
CI, however.

Discussion

Using diversity and evenness profiles

We compared the diversity of various use-groups by using the Rényi diversity

profile. The numerous intersections of diversity profiles indicate situations of
partial ordering in our data set. If we would have used a single diversity index
such as the Shannon, Simpson or Berger–Parker index, then we have obtained
erroneous results. Studies that attempt to order communities in diversity (e.g.,
Dougall and Dodd 1997; Slik et al. 2002; Mishra et al. 2004; Zilihona et al.
2004) should therefore not base the ordering on a single index, or even a

[202]
 1263

Figure 5. Accumulation curve for H2 for all trees and trees belonging to particular use-groups.
The vertical reference indicates the sample size of 47 farms depicted in Figure 3.

combination of several indices, but use techniques developed for diversity

ordering such as the Rényi diversity profile. Although the aim of our study was
not to compare the performance of several diversity ordering techniques, we
agree with Tóthmérész (1995) that the Rényi profile is one of the most useful
methods for diversity ordering.
Various researchers have used models of rank-abundance curves to study
biodiversity (e.g., Magurran 1988; Hayek and Buzas 1997; Hubbell 2001;
Belaoussoff et al. 2003; Magurran and Henderson 2003; McGill 2003). Because
rank-abundance curves do not provide a direct graphical method for diversity
ordering (some diversity ordering techniques are based on cumulative fre-
quencies, but none use the frequencies of the individual species; Tóthmérész
1995), and since information is lost when rank-abundance curves are modelled,
we do not recommend using (models of) rank-abundance distribution for
diversity ordering.
As for the diversity profile, we observed various intersections in the evenness
profiles for our dataset. For this reason, a single index of evenness (e.g.,
Zilihona et al. 2004) is not sufficient information to order communities in
evenness in the same way that a single index of diversity is not sufficient for
diversity ordering.

[203]
1264

Figure 6. Accumulation curve for H¥ for all trees and trees belonging to particular use-groups.
The vertical reference indicates the sample size of 47 farms depicted in Figure 3.

The influence of sample size on the diversity profile

It has been observed for a long time in ecological research that sample size has
an influence on species richness (e.g., Arrhenius 1921). Since diversity is
influenced by richness, sample size also has an effect on diversity. We were able
to investigate the influence of sample size on diversity by studying the range of
values observed in subsets of the data by using a randomisation approach
similar to the randomisation approach to calculate species accumulation curves
(Kindt et al. 2001). This approach allowed rarefaction to the same number of
sample units (farms in our example), thus removing the effects from differences
in sample size. This approach could be useful in other studies that compare
results obtained from surveys with different sample sizes, especially since the
approach includes a diversity ordering technique which is necessary for an
accurate description of the diversity of a community.
Magurran (1988) indicated that a Rényi series value with larger scale
parameter value has reduced sensitivity to sample size. Gimaret-Carpentier
et al. (1998) observed that the Simpson index reached stable values at lower
sample sizes compared to the Shannon index.
We observed such asymptotic patterns both for H2 and H¥ for those groups
where stable values were obtained for the Shannon index (timber, fruit,

[204]
 1265

construction, and beverage). Shade and fodder, however, showed the other
extreme with increasing profile values with increasing sample size, and not
asymptotic values.
Hayek and Buzas (1997) proposed to investigate accumulation patterns of
H, ln(E) and ln(E)/ln(S) to choose the best abundance distribution model
(‘SHE analysis’). Reaching asymptotic values for these statistics with
increasing sample size would indicate that the species abundance distribution
corresponds to respectively a log-series, a broken-stick or log-normal
distribution. Based on our results, we could therefore conclude that most use-
groups (and especially timber, fruit, construction and beverage) have abun-
dance distributions more typical of the log series distribution (asymptotic H),
while the distributions of shade and fodder corresponded more to a log-
normal distribution (asymptotic ln(E)/ln(S), figure not included). (A pre-
liminary analysis found that shade, medicine and ornamental conformed to
the log-normal distribution, whereas all trees, firewood, timber, and bound-
ary did not – for other use-groups, unreliable results were obtained for chi-
square goodness-of-fit tests.) We suggest expanding the SHE analysis to the
complete diversity profile, as in Figure 3. For example, for a pure log-series
(and geometric series), the Simpson and Berger–Parker indices are also
expected to reach constant values (May 1975), which was a phenomenon that
we could observe most clearly for the groups with the clearest asymptotic H1
pattern. Since asymptotic H1 and H2 are only expected for the log-series
distribution, the findings of Gimaret-Carpentier et al. (1998) described above
most likely correspond to studies of systems more conform to this distribu-
tion. For the same reason, the observation of reduced sensitivity to sample
size with increasing scale provided by Magurran (1988) could only apply to
the log-series.

Planning diversification based on diversity profiles

The basic aim of our study was to plan for diversification. The diversity and
evenness profiles were calculated on several use-groups. By differentiating
between several use-groups, use-groups of lower diversity or evenness were
identified and a benchmark dataset was created to measure the impact of future
diversification. By identifying use-groups of lower relative diversity (such as
beverage and construction), priority can be given to these use-groups for
diversification. Such approach can be described as a coldspot approach that
focuses on subregions of lowest diversity in a study area. For example, it would
be less efficient for diversification to add new species to firewood than adding
new species to construction. With the rarefaction procedure, comparisons are
made only for farmers that are already growing trees for a use-group. This
approach should be preferred when there is no scope for increasing the number
of farmers for each use-group: this is a better benchmark to plan diversification
efforts (see below).

[205]
1266

The value of using diversity profiles lays not only in determining which use-
groups are more diverse. Diversity profiles also allow discrimination between
richness and evenness contributions to diversity. If the intervention would
attempt to increase diversity for two groups with intersecting diversity profiles,
then for one group improvement of richness could be attempted, while
improvement of evenness would be the target for the other group.
The lack of evenness in the distribution of the dominant species and the
many steep decreases in profiles with increasing scale parameter value showed
that diversity could be increased substantially in many use-groups by targeting
evenness, rather than targeting richness. No group had perfectly evenly dis-
tributed species. Evenness increment could be achieved by encouraging farmers
to establish trees in more even numbers (influencing the demand for tree
germplasm) or by more species-even germplasm distribution (influencing the
supply of tree germplasm). The analysis shows use-groups with steep diversity
and evenness profiles where such diversity improvements would be most useful:
i.e. the construction, fruit, ornamental (although this group will probably not
constitute a priority to farmers), firewood, timber and medicine groups. When
considering the frequency of the dominant species only (not how evenly this
species is distributed), timber, medicine, fruit and beverage are the groups with
frequencies larger than 60%. Interestingly, these are also the use-groups that
could be categorised as providing more cash income to farmers (‘high value
trees’ that could be selected to be of higher priority for domestication for this
reason).
The analysis of diversity did not include all aspects that could influence
decisions on alterations in tree species composition on farms. Such factors
include potential differences in importance attributed by farmers to different
use-groups so that diversification of a more important but more diverse use-
group could be given priority over that of a less diverse but also less important
use-group. Another aspect is potential differentiation among farms in alpha
diversity, so that farms with low diversity for a specific use-group could be
targeted, rather than only targeting use-groups with lower diversity at the
survey level (Kindt et al. 2004).
Effective diversity planning will require that the relative abundances of the
composing species are analysed within use-groups that were prioritised for
diversification. When planning for increments in evenness, the potential for
increasing the abundances of rare species should be investigated. Where it is
not possible to add new trees to a particular farm or landscape, the potential of
substituting some trees of the dominant species with trees of the rare species
should be explored. Participatory research should investigate why some species
occur in higher numbers than other species: substitution of common species by
rare species may be easier when differences in abundance are not related to
differences in farmer preference, but caused by factors such as differences
between species in natural regeneration, historic promotion by development
agencies, or erosion in local knowledge. Since all species within a particular
use-group are used for that particular purpose, there is a definite potential for

[206]
 1267

increasing evenness, but such increments may require balancing diversity with
differences in preference – some species may need to be promoted for the
insurance that they bring at a cost for short-term productivity.
McNeely and Scherr (2002) describe that a new type of agriculture is needed
that leads to increased food security and conservation gains since human
population density and biodiversity are positively correlated in many areas.
Their book provides examples of innovative landscape management strategies
that successfully combined both objectives by applying ecoagriculture strate-
gies. Our study documented that many tree species have been integrated in
farming systems already (conform Ecoagriculture Strategy 4 of mimicking
natural habitats by integrating productive perennial plants). Since careful
scrutiny of species identities of the various use-groups (such detailed study was
beyond the scope for this article, but is a logical next step as described above)
showed that the dominant species in most use-groups were exotic species
(exceptions were boundary demarcation, fodder and medicine), whereas the
majority of rare species were indigenous species, increments of evenness could
result in increments of indigenous tree species in the farming landscape (since
not only exotic species are planted, increasing abundances for indigenous
species entails small relative changes in planting practices by local communi-
ties). Diversification could therefore result in improved conservation, although
the links between development and conservation goals need to be explored
carefully (Adams et al. 2004).
Landscape diversification could also consider population structure and
geneflow of particular species, especially for indigenous species. Where less
frequent species are promoted, interventions should attempt to ensure that
population sizes are large enough to avoid substantial genetic erosion. Diver-
sification planning could for example consider corridors in farmland between
natural populations to reduce genetic erosion (O’Neill et al. 2001). Such con-
siderations of genetic diversity may indicate limits to diversification (not all
species could potentially be maintained at large enough population sizes for a
more even distribution of species) that could be incorporated in planning of
diversification (finding a more even species abundance distribution that avoids
too small population sizes for each species, possibly with a smaller number of
species). Alternatively, species can be maintained at very small population sizes
given that new genetic diversity is regularly introduced from genetically diverse
seed sources.
Ecological reasons for diversification within a use-group could include
minimizing the chances of pest and disease outbreaks. Promotion of single
species for a particular use-group should especially be avoided since several
pest outbreaks on agroforestry species have been experienced after large-
scale promotions of monoculture agroforestry technologies (Atta-Krah et al.
2004). Ecological research has indicated that biodiversity can affect ecosys-
tem function, but that differences in species function are conditions for
positive effects of biodiversity on ecosystem stability and productivity (e.g.,
Hector et al. 1999; Loreau et al. 2001, 2002; Tilman et al. 2001). Natural

[207]
1268

communities with specific species extinction and abundance patterns func-

tion differently from experimental communities of similar richness but dif-
ferent composition and equalized abundances, however (Zavaleta and
Hulvey 2004). The ecological consequences of increasing the diversity or
evenness of trees on farms can therefore not be predicted and thus needs to
be evaluated on a case-by-case basis, although natural communities can
provide some clues on potential richness and composition (e.g., Van Noo-
rdwijk and Ong 1999).
Although the methods shown in this article are unlikely to provide all the
answers for diversification planning, they did provide meaningful insights in
relationships between richness, evenness, and sample size of use-groups. They,
thus, provide accurate guidance for attempts in alterations of these charac-
teristics. Obviously, they allow also for detailed monitoring of the impact of
interventions of these characteristics by providing a baseline to compare
diversity before and after interventions and by using a technique that is
specifically tailored to study differences in diversity.

Acknowledgements

We are very grateful for information provided by farmers. Roeland Kindt

wants to thank Wim Buysse, Richard Coe, Ian Dawson, Ard Lengkeek, Meine
Van Noordwijk and two anonymous reviewers for inputs in the article, assis-
tance provided by the ICRAF East and Central African regional programme
especially through Stephen Ruigu and Amadou Niang, assistance during data
collection by Joseph Njeri, and funding provided by DFID and VVOB.

References

Adams W.M., Aveling R., Brockington D., Dickson B., Elliott J., Hutton J., Roe D., Vira B. and
Wolmer W. 2004. Biodiversity eradication and the eradication of poverty. Science 306: 1146–
1149.
Arrhenius O. 1921. Species and area. J. Ecol. 9: 95–99.
Atta-Krah K., Kindt R., Skilton J.N. and Amaral W. 2004. Managing biological and genetic
diversity in tropical agroforestry. Agroforest. Syst. 61: 183–194.
Beentje H.J. 1994. Kenya Trees, Shrubs and Lianas. National Museums of Kenya, Nairobi 722 pp.
Belaoussoff S., Kevan P.G., Murphy S. and Swanton C. 2003. Assessing tillage disturbance on
assemblages of ground beetles (Coleoptera: Carabidae) by using a range of ecological indices.
Biodivers. Conserv. 12: 851–882.
Bradley P.N. 1991. Woodfuel, Women and Woodlots, vol. 1. Macmillan Education Ltd, London
and Basingstoke 338 pp.
Bradley P.N., Chavangi N. and Van Gelder A. 1985. Development research and energy planning in
Kenya. Ambio 14: 228–236.
Colwell R.K. 1997. Estimates: Statistical Estimation of Species Richness and Shared Species from
Samples. University of Connecticut, Storrs.
Dougall T.A.G. and Dodd J.C. 1997. A study of species richness and diversity in seed banks and its
use for the environmental mitigation of a proposed holiday village development in a coniferized
woodland in south east England. Biodivers. Conserv. 6: 1413–1428.

[208]
 1269

Gimaret-Carpentier C., Pelissier R., Pascal J.P. and Houllier F. 1998. Sampling strategies for the
assessment of tree species diversity. J. Veg. Sci. 9: 161–172.
Gotelli N.J. and Colwell R.K. 2001. Quantifying biodiversity: procedures and pitfalls in the
measurement and comparison of species richness. Ecol. Lett. 4: 379–391.
Hayek L.-A.C. and Buzas M.A. 1997. Surveying Natural Populations. Columbia University Press,
New York xvi, 563 pp.
Hector A., Schmid B., Beierkuhnlein C., Caldeira C., Diemer M., Dimitrakopoulos P.G.,
Finn J.A., Freitas H., Giller P.S., Good J., Harris R., Högberg P., Huss-Danell K., Joshi J.,
Jumpponen A., Körner C., Leadley P.W., Loreau M., Minns A., Mulder C.P., O’Donovan
G., Otway S.J., Pereira J.S., Prinz A., Read D.J., Scherer-Lorenzen M., Schulze E.D.,
Siamantziouras A.D., Spehn E.M., Terry A.C., Troumbis A.Y., Woodward F.I., Yachi S. and
Lawton J. 1999. Plant diversity and productivity in European grasslands. Science 286: 1123–
1127.
Hill M.O. 1973. Diversity and evenness: a unifying notation and its consequences. Ecology 54: 427–
431.
Hubbell S.P. 2001. The Unified Neutral Theory of Biodiversity and Biogeography. Princeton
University Press, Princeton xiv + 375 pp.
ICRAF 1997. ICRAF Medium-Term Plan 1998–2000. International Centre for Research in
Agroforestry, Nairobi 77 pp.
Jaetzhold R. 1982. Farm Management Handbook of Kenya: Natural Resources and Farm Man-
agement Information vol. iia – West Kenya (Nyanza and Western Provinces). Kenya Ministry of
Agriculture, 397 pp.
Kindt R. 2001. RenyiAccum. Program to Calculate Average Values and Ranges for the Rényi
Diversity Series Associated with Random Site Sequences. International Centre for Research in
Agroforestry (ICRAF), Nairobi.
Kindt R. 2004. Biodiversity Analysis Functions for R (Biodiversity.R). ICRAF, Nairobi.
Kindt R., Degrande A., Turyomurugyendo L., Mbosso C., Van Damme P. and Simons A.J. 2001.
Comparing species richness and evenness contributions to on-farm tree diversity for data sets
with varying sample sizes from Kenya, Uganda, Cameroon, and Nigeria with randomized
diversity profiles. In: IUFRO Conference on Forest Biometry, Modelling and Information
Science, 26–29 June 2001. University of Greenwich, UK. URL: http://cms1.gre.ac.uk/confer-
ences/iufro/proceedings/ (last accessed 4–2004).
Kindt R. and Lengkeek A.G. 1999. Tree diversity on farm – use it or lose it. In: National Workshop
on Agricultural Biodiversity Conservation, 27–29 January 1999. Intermediate Technology
Development Group (ITDG, Nairobi, pp 75–85.
Kindt R., Simons A.J. and Van Damme P. 2004. Do farm characteristics explain differences in tree
species diversity among western Kenyan farms? Agroforest. Syst. 63: 63–74.
Legendre P. and Legendre L. 1998. Numerical Ecology. Elsevier Science BV, Amsterdam 853 pp.
Loreau M., Naeem S., Inchausti P., Bengtsson J., Grime J.P., Hector A., Hooper D.U., Huston
M.A., Raffaelli D., Schmid B., Tilman D. and Wardle D.A. 2001. Biodiversity and ecosystem
functioning: current knowledge and future challenges. Science 294: 804–807.
Loreau M., Naeem S. and Inchausti P. 2002. Biodiversity and Ecosystem Functioning: Synthesis
and Perspectives. Oxford University Press, Oxford 304 pp.
Magurran A.E. 1988. Ecological Diversity and Its Measurement. Princeton University Press,
Princeton, N.Jx, 179 pp.
Magurran A.E. and Henderson P.A. 2003. Explaining the excess of rare species in natural species
abundance distributions. Nature 422: 714–716.
May R.M. 1975. Patterns of species abundance and diversity. In: Cody M.L. and Diamond J.M.
(eds), Ecology and the Evolution of Communities. The Belknap Press of Harvard University
Press, Cambridge and London, pp. 81–120.
McGill B.J. 2003. A test of the unified neutral theory of biodiversity. Nature 422: 881–885.
McNeely J.A. and Scherr S.J. 2002. Ecoagriculture: Strategies to Feed the World and Save Wild
Biodiversity. Island Press, Washington 323 pp.

[209]
1270

Mishra B.P., Tripathi O.P., Tripathi R.S. and Pandey H.N. 2004. Effects of anthropogenic dis-
turbance on plant diversity and community structure of a sacred grove in Meghalaya, northeast
India. Biodivers. Conserv. 13: 421–436.
O’Neill G.A., Dawson I.K., Sotelo-Montes C., Guarino L., Current D., Guariguata M. and Weber
J.C. 2001. Strategies for genetic conservation of trees in the Peruvian Amazon basin. Biodivers.
Conserv. 10: 837–850.
Purvis A. and Hector A. 2000. Getting the measure of biodiversity. Nature 405: 212–218.
Ricotta C. 2003. On parametric evenness measures. J. Theor. Biol. 222: 189–197.
Ricotta C. and Avena G.C. 2002. On the information-theoretical meaning of Hill’s parametric
evenness. Acta Biotheor. 50: 63–71.
Rousseau D., Van Hecke P., Nijssen D. and Bogaert J. 1999. The relationship between diversity
profiles, evenness and species richness based on partial ordering. Environ. Ecol. Stat. 6: 211–223.
Shaw P.J.A. 2003. Multivariate Statistics for the Environmental Sciences. Hodder Arnold, London
ix + 233 pp.
Slik J.W.F., Verburg R.W. and Keßler P.J.A. 2002. Effects of fire and selective logging on tree
species composition of lowland dipterocarp forest in East Kalimantan, Indonesia. Biodivers.
Conserv. 11: 85–98.
Taillie C. 1979. Species equitability: a comparative approach. In: Grassle J.F., Patil G.P., Smith
G.K. and Taillie C. (eds), Ecological Diversity in Theory and Practice. International Cooperative
Publishing House, Fairland, pp. 51–62.
Tilman D., Reich P.B., Knops J., Wedin D., Mielke T. and Lehman C. 2001. Diversity and
productivity in a long-term grassland experiment. Science 294: 843–845.
Tóthmérész B. 1995. Comparison of different methods for diversity ordering. J. Veg. Sci. 6: 283–
290.
Van Noordwijk M. and Ong C.K. 1999. Can the ecosystem mimic hypotheses be applied to farms
in African savannahs? Agroforest. Syst. 45: 131–158.
Zavaleta E.S. and Hulvey K.B. 2004. Realistic species losses disproportionately reduce grassland
resistance to biological invaders. Science 306: 1175–1177.
Zilihona I.J.E., Niemela J. and Nummelin M. 2004. Effects of a hydropower plant on Coleopteran
diversity and abundance in the Udzungwa Mountains, Tanzania. Biodivers. Conserv. 13: 1453–
1464.

[210]
Biodiversity and Conservation (2006) 15:1271–1301
Springer 2006
DOI 10.1007/s10531-005-2576-4

-1

Vascular plant species inventory of a Philippine

lowland rain forest and its conservation value

GERHARD LANGENBERGER, KONRAD MARTIN* and

JOACHIM SAUERBORN
Institute of Plant Production and Agroecology in the Tropics and Subtropics (380), Agroecology
Section, University of Hohenheim, 70593 Stuttgart, Germany; *Author for correspondence (e-mail:
[email protected]; phone: +49-711-459-3605; fax: +49-711-459-3843)

Received 22 June 2004; accepted in revised form 8 February 2005

Key words: Conservation value, Dipterocarp forests, Gene bank, Molave forest, Native species,
Species richness, Tropical rain forest, Vascular plant species

Abstract. The Philippines are one of the most important biodiveristy hotspots on earth. Due to the
extraordinary rate of environmental destruction, leaving only 3% of the land with primary forest,
this biodiversity is at high risk. Despite that situation information on Philippine forest vegetation is
fragmentary and focused on trees. This study aimed at analysing forest remnants in the Leyte
Cordillera on the Island of Leyte, and at evaluating their role as refuge to the largely destroyed
lowland forest vegetation. A total of 49 plots (100 m2 each) between 55 and 520 m a.s.l. were
studied. All vascular plant species except epiphytes were included. Records include 685 taxa from
289 genera and 111 families, representing nearly 8% of the known Philippine vascular plant species.
More than half (52%) of the species are Philippine endemics. A number of 41 tree species, or 6% of
all taxa recorded, are included in the IUCN red list, either as vulnerable, endangered, or critically
endangered. Life form composition was dominated by phanerophytes (65.3%), followed by lianas
and chamaephytes (17.1 and 16.9%, respectively). The most common families were the Rubiaceae
with 35 and the Euphorbiaceae with 32 species. All five Philippine dipterocarp forest types as well
as the molave forest type were represented by typical tree species. The area provides an important
gene bank of the highly threatened Philippine lowland forest vegetation and is of high value for
biodiversity conservation. Additionally, it can play an important role as seed source of valuable tree
species for the increasing initiatives to rehabilitate and reforest degraded land with native species.

Introduction

The destruction of tropical rain forests is still continuing at high rates (FAO
2003). This process, especially threatens the earth’s biodiversity hotspots such as
the Philippines (Myers et al. 2000; Brooks et al. 2002). Despite this, there are only
very few studies worldwide which aimed at the documentation of the total plant
species richness of such sites. Most inventories were restriced to selected life forms
such as ground herbs (e.g. Kiew 1987; Poulsen and Balslev 1991; Poulsen 1996) or
trees of a defined minimum diameter (e.g. Valencia et al. 1994; Lieberman et al.
1996; Newbery et al. 1996; Rennolls and Laumonier 2000; Slik et al. 2003).
Vascular plant species composition of tropical lowland forests was studied in
Ghana on 0.5- and 1-ha plots by Hall and Swaine (1981), in Amazonia on 0.02-ha plots
by Takeuchi (1960) and on 10 non-contiguous plots of 0.1 ha by Duivenvoorden

[211]
1272

(1994), in Ecuador on 0.1-ha plots of three lowland forest types by Gentry and
Dodson (1987), and in stratified plots with a total area of about 2 ha in Puerto
Rico by Smith (1970). In Southeast Asia, Kochummen et al. (1992) studied the
trees and shrubs (>1 cm diameter at breast height (dbh)) in a 50-ha plot in the
Pasoh Forest Reserve in Malaysia. The most comprehensive study including all
vascular plants as well as mosses was conducted by Whitmore et al. (1985) on a
single 100-m2 plot in the lowland rain forest of Costa Rica. However, no study
representing a complete inventory of vascular plant species richness of any site of
lowland rain forest in Southeast Asia was found.
The Philippines are among the most seriously depleted tropical countries
with only 3% of the land area still covered by primary forest (Myers et al.
2000). From 1990 to 2000, the Philippines lost 1.4%, or 89,000 ha, of the forest
area annually (FAO 2003). At the same time, the Philippine archipelago is one
of the most important biodiversity hotspots on earth (Myers et al. 2000) with
high proportions of endemic plant and animal species (Heaney and Regalado
1998). The endemism rate of plants was estimated to be 39% (Davis et al.
1995), but for certain taxa, it can be much higher. For example, 11 of 12 species
of pitcher plants (Nepenthes spp.) known from the Philippines are endemic
(Cheek and Jebb 2001). Similarly, there are high rates of endemism among the
fauna. Referring to terrestrial vertebrates, 64% of the archipelago’s land
mammals are endemic, as well as 44% of the breeding land birds, 68% of the
reptiles, and ca. 78% of the amphibians (Heaney and Regalado 1998). Most of
them depend on forest ecosystems.
Despite the ecological uniqueness on the one hand and the extensive
destruction on the other, the study of Philippine forest vegetation has been
neglected (Tan and Rojo 1989; Kartawinata 1990; Soerianegara and Lemmens
1994). Much of the current knowledge is still based on studies conducted in the
early 20th century (Whitford 1906; Whitford 1911; Brown and Mathews 1914;
Brown 1919), which were mainly dealing with timber trees under economical
aspects. Recent studies focused on the vegetation of montane and submontane
forest types on different islands. However, in most cases (Aragones 1991;
Pipoly and Madulid 1998; Proctor et al. 1998; Hamann et al. 1999) these were
largely restricted to trees of a defined size, which usually is ‡10 cm dbh. Buot
and Okitsu (1997) only considered woody plants higher than 1.3 m, and Ingle
(2003) those of at least 5 cm dbh. The only data without size limitations are
provided by Gonzales-Salcedo (2001) from Mt. Amuyao, Luzon, at elevations
of 1600–1800 a.s.l. and by Gruezo (1998) from the highly degraded vegetation
of Pagbilao Grande Island. No study dealing with lowland forest vegetation
was found in the literature.
In order to provide more substantial information on species richness and
composition of Philippine lowland forests, we analysed forest remnants in the
rugged foothills of the Leyte Cordillera. The island of Leyte is located in the
central part of the Philippine archipelago and represents a typical example of
the environmental situation in the Philippines. In 1987, the remaining forest
cover of Leyte was 12%, and in 1994 only 2% of the island’s area have been

[212]
 1273

estimated to be primary forest (Dargantes and Koch 1994). More recent data
(DENR 1998) show that about 40% of the land area of Leyte is covered by
grassland and barren land, resulting from abandoned cultivation and grazing
land that marginalised in productivity through erosion and leaching. Another
40% of the island’s area is under coconut plantations. The remaining area is
composed of settlements, agricultural land and forest. In the view of this sit-
uation, the objectives of this study were (a) to analyse the vascular plant species
composition and diversity of selected plots of mature primary forest and (b) to
evaluate the role of the study area as refuge to lowland forest vegetation and its
significance for conservation and as a gene bank.

Material and methods

Study area

The Island of Leyte (Figure 1) belongs to the biogeographic region of the

Eastern Visayas (DENR and UNEP 1997). It is located between 955¢ N–
1148¢ N and 12417¢ E–12518¢ E, with an extension of 214 km from north to
south. Located offshore the northeastern part of Leyte is the island of Samar.
The southern part of Leyte is exposed to the Pacific Ocean (Leyte Gulf).
Leyte is characterised by the north–south running Leyte Cordillera which is
part of the Philippine Fault Line. The Cordillera reaches a maximum elevation of
ca. 1350 m (Mt. Lobi) in the northern part of the island. As geologically young
volcanic mountain range, it shows a typical rugged topography of narrow ridges
and steep slopes, where landslides are common (Bremer 1995, 1999; Walsh 1996).
In its foothills, patches of primary forest without discernible human interference
can still be found although the island is densely populated (ca. 262 inhabitants/
km2 as calculated after: NSCB RU-8 2001). The coastal plains have already been
deforested in the first half of the last century (Barrera et al. 1954).
The study area is located in the western part of the Leyte Cordillera,
ca. 8 km north of the provincial capital of Baybay, in the foothills of Mt.
Pangasugan (1150 m, 1046¢ N, 12450¢ E). In this part of the island, the
Cordillera reaches close to the coast. Large parts of the mountain’s western
range are extremely steep and are free from trees of this reason. In the eastern
part of the Cordillera, the slope has a lower gradient.
Primary forest can be found from about 250 m a.s.l. up to the mountain’s
summit at 1150 m a.s.l. In hillsides below this elevation, the forests has largely
been replaced by coconut plantations and slash and burn agriculture. Only
along the small creeks, near natural vegetation can still be found at these lower
elevations. Above ca. 600 m a.s.l., the lowland forest formation changes into
mossy forest, with its stunted trees and a rich epiphyte community.
Within the study area, no recent logging was observed, although forest
clearing continues at other localities of the mountain. This can be explained by
the area’s status as Forest Reserve of the Leyte State University. Despite this,
rattan collection and hunting was observed within the Forest Reserve.

[213]
1274

Figure 1. The Philippine archipelago, the Island of Leyte, and the location of the study area.

Geology and soils

The soil type in the primary forest between 370 and 520 m a.s.l. is a haplic
Andosol with rudic phase (FAO/UNESCO 1988) overlying basaltic and
andesitic breccia (Zikeli 1998). The soil at lower elevations (100 m a.s.l.) has
been classified as haplic Alisol (FAO/UNESCO 1988) over basalt (Asio 1996).

Climate

According to the climatic classification of the ‘Modified Corona’s System’,

Leyte is climatically divided (Kintanar 1984). Southern Leyte belongs to the
climatic type II (i.e. no pronounced dry season), but exhibits a distinct rainfall
peak in December and January as a result of the northeast monsoon. The
northern part of Leyte which includes the study area has been assigned to
climatic type IV, showing a more or less even rainfall distribution throughout

[214]
 1275

the year. The standardization of the rainfall pattern in northern Leyte com-
pared to that of southern Leyte might be explained by the protective effect of
the Island of Samar off the northeast-coast of Leyte, although Samar’s
mountains are lower than those of Leyte (ca. 850 m a.s.l.).
Local climatic conditions have been analysed from a 23-years period (1976–
1998) of record by using data from the PAGASA (Philippine Atmospheric,
Geophysical and Astronomical Services Administration) weather station on
the campus of the Leyte State University (7 m a.s.l.), ca. 1–3 km west of the
study plots. The annual average temperature is 27.4 C and the average annual
precipitation is 2586 mm. Highest precipitation occurs during November to
January. Lowest rainfall is observed between March and May. On average, all
months receive at least 100 mm precipitation, i.e. there is no dry month
according to the definition of Walsh (1996). However, drought periods (i.e. less
than 50 mm of monthly precipitation according to Walsh 1996) of up to
4 months have been recorded during El Niño Southern Oscillation events. The
general rainfall patterns and the climatic conditions measured at the PAGASA
station are more similar to climatic type II with its clear impact of the northeast
monsoon than to climatic type IV, implying that neither the mountains
of Samar nor the Leyte Cordillera itself causes a distinct rain shadow west
of Mt. Pangasugan.
Orographic rainfalls are an important factor in the Leyte Cordillera, espe-
cially in the vicinity of the mountain summits. The summit of Mt. Pangasugan
is often observed being cloud covered, and during field work heavy rainfalls
have been experienced, while the coastal plain did not receive any precipitation.
An important climatic feature of the area are typhoons. Leyte lies at the
southern margin of the typhoon tracks entering the Philippines, and is hit at a
rate of five typhoons in three years, mainly during the summer months (Parong
1984; cited in Kintanar 1984).

Vegetation analysis

Field studies were conducted in 1997 and 1998. The attempt to identify a
minimum area in mature primary forest failed due to the heterogeneity of the
vegetation. A plot size allowing a reasonable number of replications proved to
be 100 m2. Where relief conditions and homogeneity of the vegetation allowed,
plots were arranged along a catena from ridge to river bank. The 100-m2 plots
were generally designed as quadrats, but on narrow ridges and river banks,
other rectangular design was used due to relief constraints. A total of 49 plots
was established, with 15 on the ridge, 21 on slopes, and 11 on riverbanks. Two
plots were established in ca. 6-year-old land slide successions, with one of them
located ca. 2 km south of the main study area.
The vegetation analysis procedure was based on a ‘nested quadrat
design’ (Kent and Coker 1992). All plants >2.5 m were identified from
the total plots (100 m2). On central subplots of 25 m2, all plants £ 2.5 m

[215]
1276

as well as the lianas were considered. Records included epiphytic and climbing
plants on the stem bases of trees up to a height of 2.5 m. Crown epiphytes were
not included in the analysis, but epiphytes found on the ground were identified
and added to the species lists. From species which could not be identified in the
field, voucher specimens were collected. Tall trees were sampled with the help
of a tree climber. However, no samples were taken from erect and climbing
palms (rattans), because this would have been destructive and the chance of
identification was very low due to the lack of fertile specimens. Therefore, most
palms had to be distinguished as morphospecies.
Taxa were assigned to life forms on the base of field observations, or with the
help of literature information in the case of juveniles. Life form classification
followed Ellenberg and Mueller-Dombois (1967). Plant samples collected in
this study were deposited at the Department of Plant Breeding, Herbarium,
Leyte State University, ViSCA, Baybay, Leyte, 6521A, Philippines.

Species identification and nomenclature

Identification of specimens was conducted with the help of literature and

specialists. Publications referring to the Philippine flora included de Guzman
et al. (1986), Merrill (1912), Pancho (1983), Santos et al. (1986), van Steenis
(1950-ongoing), and Zamora and Co 1986). Sources referring to neighbouring
countries, which also include many Philippine taxa, were used in addition
(Henderson 1974a, b; Ng 1978, 1989; Keng 1983, 1990; Whitmore 1983a, b;
Corner 1988; Anonymous 1993, 1994, 1996; Soepadmo and Wong 1995;
Soepadmo et al. 1996). Identification of seedlings, infertile and juvenile plants
was not possible in all cases. For pre-identification of taxa and delimitation of
morphospecies, field characters were very important. Besides the description of
such characters in the above mentioned literature, the specific publications by
van Balgooy (1997a, b) and Keller (1996) were used. Tree seedlings were
identified with help of Ng (1991) and Burger (1972).
In addition, plants were identified by taxonomists from the Philippine
National Herbarium, Manila, the National Herbarium of the Netherlands,
Leiden Branch, and collaboratively during meetings of the Philippine Native
Plant Group. Nomenclature followed various sources as cited above. However,
priority was given to the Flora Malesiana (van Steenis 1950-ongoing), when-
ever possible. The legumes were assigned to the traditional family of
Leguminosae. Scientific names were not derived from the translation of local
names in any case.

Data analysis

Species richness, diversity and evenness were determined for each of the 49
plots. Only those plants rooting within the plots were considered in the

[216]
 1277

analysis. The Shannon-Index (H¢) was used as a robust and simple diversity
measure (Magurran 1988). For the analysis of species dominance patterns,
Evenness (E) based on the Shannon-Index was calculated for each of the plots.
To assess the area’s value as a refuge to Philippine tree species, the char-
acterization of forest types by Whitford (1911) was used. His classification and
characterization is based on the occurrence of typical tree species and tree
species combinations. He often used vernacular names or typical families or
genera as e.g. ‘Apitong’ for Dipterocarpus spp. to characterise his forest types.
For many of these vernacular names a scientific species could not be assigned
with certainty, and therefore, were not used for comparisons. Whitford (1911)
pointed out that the description of his forest types was based on a still frag-
mentary knowledge of Philippine forests. Most of his ‘typical’ tree species –
with few exceptions such as mangroves – occur in the other forest types as well.
For example, many species of the dipterocarp forest types occur at wet local-
ities in the Molave forest (limestone forest). On the other hand, the typical
Molave forest species also exist in the dipterocarp forest types, especially on
dry sites. Of such reasons, Whitford’s (1911) forest types are primarily related
to the major habitat conditions in the Philippines and do not represent real
plant associations. The comparison of the species recorded in this study with
Whitford’s (1911) forest types merely demonstrates the diverse habitat condi-
tions in the present study area. Unfortunately, not much work has been con-
ducted so far to improve Whitford’s system, and information on species
composition of the undergrowth vegetation, which might be especially valuable
to characterise habitat conditions (Schulze and Whitacre 1999), is still missing.

Results

From the 49 plots, a total of 685 taxa was recorded. Of these, 58.3% were
identified to species level, 86.2% to genus level, and 96.7% to family level. The
remaining 3.3% of the taxa could only be assigned to higher taxonomic levels.
All taxa identified to species level are listed in the Appendix. Species inventory
was clearly dominated by angiosperms, accounting for 92.1% of all species.
The pteridophytes represented 7.5% of the species. Only three species of
gymnosperms (Podocarpus rumphii, Gnetum gnemon, G. latifolium) were found
(Table 1).
More than half (52%) of all species identified are endemic to the Philippines,
including one endemic genus (Greeniopsis, Rubiaceae). The most common
families were the Rubiaceae (35 species) and the Euphorbiaceae (32 species),
followed by the herbaceous family of Araceae and the erect and climbing palms
(Arecaceae) with 28 species each. The Meliaceae and Moraceae included 27
species each (Figure 2). The ratio between the number of genera and the
number of species ranged between 1:1.5 (Anacardiaceae) and 1:6.7 (Moraceae).
The frequency of taxa was low. Nearly half (48.5%) of all taxa were recorded
from only one of the 49 plots, and nearly one third (30.5%) of the taxa were

[217]
1278

Table 1. Taxonomic composition of 49 non-contiguous plots (100 m2 each) in lowland forest

remnants of the study area in the foothills of Leyte Cordilliera.

Spermatophytes (%) Pteridophytes (%) Total (%)

Gymnosperms Angiosperms

Families 2 (1.8) 94 (84.7) 15 (13.5) 111 (100)

Genera 2 (0.7) 261 (90.3) 26 (9) 289 (100)
Species 3 (0.4) 631 (92.1) 51 (7.5) 685 (100)

Figure 2. The 20 most common plant families recorded from 49 plots (100 m2 each) in the study
area. Figures in brackets indicate the ratio between the number of genera and the number of
species.

[218]
 1279

represented by only one single individual. Very few species showed high fre-
quencies as e.g. the two tree species, Calophyllum blancoi (present in 32 plots)
and Dacryodes rostrata (present in 31 plots), which was due to a high rate of
juveniles.
The average number of species per plot was 47 and ranged between 17 and
80. Shannon diversity (H¢) reached values between 2.2 and 3.9, and evenness
(E) ranged between 0.64 and 0.98. The species–area curve for all plots shows a
steady increase of species numbers with only a weak tendency to level off
(Figure 3). The flattening of the curve at its beginning is the result of the river
bank vegetation which was comparatively species poor and homogenous. The
species–area curve starts to rise again with the addition of the slope plots.
Life form composition is clearly dominated by phanerophytes (65.3% of all
taxa), followed by lianas (17.1%) and chamaephytes (16.9%). Geophytes were
rare (0.7%) and largely represented by few ground orchids. Hemicryptophytes
and therophytes were absent (Figure 4). Epiphytes were not the focus of this
study and are therefore not included in the calculation of life form composi-
tion. A rough estimate of epiphyte contribution to the area’s species inventory
is ca. 10%. The most conspicuous epiphytic plant group observed were orchids.
Many of the vegetation clusters observed in the tree crowns were composed
of the accumulation of orchid bulbs belonging to a single species (e.g.
Grammatophyllum multiflorum).
The following taxa occurring in the study plots have been classified by
Soepadmo (1995) as endangered and economically important lowland forest
genera in SE Asia: Anisoptera, Dipterocarpus, Parashorea, Shorea, Vatica
(Dipterocarpaceae), Artocarpus (Moraceae), Mangifera (Anacardiaceae), and
Calamus (Arecaceae). Additionally, 41 of the tree species recorded are listed as
endangered for the Philippines by IUCN (2000). Of these, 23 are classified as
vulnerable, one as endangered, and 17 as critically endangered (see Appendix).

Figure 3. Species–area curve for 49 plots (100 m2 each) in the study area.

[219]
1280

Figure 4. Life form spectrum (after Ellenberg and Müller-Dombois 1967) of species recorded
from 49 plots (100 m2 each) in the study area.

Discussion

The 685 taxa recorded from the 49 plots account for nearly 8% of the ca. 8900
vascular plant species so far described for the Philippines (Davis et al. 1995).
Although the plots were not contiguous and species numbers can therefore be
expected to be higher than in contiguous plots (Whitmore 1985) this figure is
high, considering the small overall study area (4900 m2 in total). Only very few
datasets cover tropical lowland forest vegetation comprehensively and are
therefore suitable for comparison. The only study using the same plot size was
conducted by Whitmore et al. (1985) in the tropical lowland rain forest of
Costa Rica, who analysed a plot of 100 m2, considering all vascular plants.
They recorded a total of 233 species, including 59 (25%) epiphyte species. In
the present study, the highest number of species recorded from a single 100-m2
plot was 80 and thus much lower than the number found by Whitmore et al.
(1985). However, in the present study the vegetation up to 2.5 m tall as well as
the lianas were collected from subplots of 25 m2, and crown epiphytes were
excluded. Despite this, the maximum number of vascular plant species on 100-
m2 plots in the study area can expected to be clearly lower than the number of
233 species recorded by Whitmore et al. (1985).
An estimate of the overall vascular plant species richness of Mt. Pangasugan
area, including mossy forest as well as the different stages of succession, results
in 1500–2000 species. This estimate is based on the very conservative
assumption that 50% of the lowland forest species was recorded in this study,
and that the mossy forest has a similar species richness as a 1-ha plot studied by

[220]
 1281

Meijer (1959) in a montane rainforest in Indonesia (333 vascular plant species).

The numbers of tree species given by Ingle (2001) (100 species ‡5 cm dbh) and
Hamann et al. (1999) (92 species ‡10 cm dbh) for 0.75- and 1-ha plots,
respectively, in Philippine mountain environments show that the overall species
richness including all life forms can be expected to be roughly similar to that of
Meijer (1959) in Indonesia.

Representation of taxa

The composition of taxa observed in this study is similar to other areas in

Southeast Asia. Differences to such sites are related to the proportions of
families and result mainly from different inventory approaches. For trees
alone, the dominance of the Dipterocarpaceae and the Euphorbiaceae con-
cerning number of species is well documented (Manokaran and Kochummen
1990; Sist and Saridan 1998; Slik et al. 2003; Wilkie et al. 2004). Sist and
Saridan (1998) report that the Dipterocarpaceae represent 70% of all
trees ‡ 50 cm dbh in a primary forest in East Kalimantan. In our study, the
Dipterocarpaceae were the most common family in the canopy layer (12 of 44
species).
Turner (1994) analysed the vascular flora of Singapore and its main habitat
types from herbarium collections.The Orchidaceae are the most speciose family
in his taxonomic spectrum. This reflects the large number of orchid species in
Malesia (6500 species according to Soepadmo 1995). In our study, however,
Orchidaceae are poorly represented because we did not include crown epi-
phytes. Without considering the orchids in both studies, the pteridophytes are
the most speciose group, followed by the Rubiaceae and the Euphorbiaceae in
both studies. The other predominant families in terms of species richness listed
by Turner (1994) are Annonaceae, Moraceae, Arecaceae/Palmae, Myrtaceae,
Melastomataceae and Lauraceae. With exception of the Melastomataceae,
these are also the most speciose families in our study (Figure 2).

Representation of forest types

The rugged relief of the study area represents a broad spectrum of Philippine
habitats. The comparison of the tree species recorded from our study with the
typical tree species composition of the forest types described by Whitford
(1911) showed a high degree of correspondence. Many tree species typical of
the five dipterocarp forest types as well as the Molave type (Figure 5) were
present. From the 18 tree species listed by Whitford (1911) as typical for the
Lauán-hagághak, 15 were also present in our study area. Originally, this forest
type is established on lowland plains on wet soils (Whitford 1911), but was
transformed into rice fields in the study area. However, tree species repre-
senting this type of forest still occur on the banks of the small creeks at low

[221]
1282

Figure 5. Comparison of the number of characteristic species of the different lowland forest types
in the Philippines (after Whitford 1911) with the number of respective species recorded in this
study.

elevations. The typical tree species of other forest types were also well repre-
sented (Figure 5). The high number of Molave type species (50% of the typical
species as mentioned by Whitford 1911) in the study area is remarkable, as this
forest represents dry limestone areas (Whitford 1911). This is another indica-
tion that the area’s vegetation might be strongly influenced by drought periods.

Life form composition

The dominance of trees and phanerophytes is a typical feature of tropical rain

forests (Richards 1996). In our study, Phanerophytes account for 65.3% of the
species. Richards (1996) provides figures from a rain forest in Guyana, which
are based on the Raunkiaer System (Raunkiaer 1934) and cannot be directly
compared with our data. We therefore recalculated his data by excluding the
lianas from the phanerophytes and excluded the epiphytes in addition. This
resulted in a life form composition of 60% phanerophytes, 16% chamaephytes,
and 24% lianas. A similar recalculation of figures provided by Cromer and
Pryor (1942) for a rain forest in Queensland results in 77.1% phanerophytes,
12.5% chamaephytes, and 10.4% lianas. Figures for a terra firme rain forest in

[222]
 1283

Brasilia (Cain et al. 1956) are: phanerophytes 74.3%, chamaephytes 0.9%,

hemicryptophytes 2.8%, geophytes 0.9%, lianas 12.8%, and epiphytes 8.3%.
Therophytes and hemicryptophytes are usually absent from undisturbed
tropical rain forests (Richards 1996). Geophytes are also often absent as in
Richards’ Guyana study or represented by few species as in the present study
(0.7%), where they were mainly made up of ground orchids.
The estimated proportion of epiphyte species of the total number of species
in this study (ca. 10%) is clearly lower than the numbers given by Whitmore
et al. (1985) (25%) for Costa Rica or by Gentry and Dodson (1987) (35%) for
Ecuador.

Conservation value

Kochummen et al. (1992) stated that comparatively small areas might represent
high numbers of a regional flora. They found that their 50-ha plot in the Pasoh
Forest Reserve (Malaysia) included 25% of all trees and shrubs (‡ 1 cm dbh)
of the Malay Peninsula. In our study, an overall sample area of approximately
half hectare included ca. 8% of all Philippine vascular plant species. Given the
small area considered as well as the fact that neither the successional vegetation
nor the mossy forest is included, the representation of Philippine flora in the
Mt. Pangasugan area is clearly higher than 8%.
The proportion of 52% endemic taxa recorded in this study is clearly higher
than the proportion of 39% stated as an average for the Philippines (Davis
et al. 1995). This result agrees with Ashton (1993) who stated that the south-
eastern part of the Philippines is especially rich in endemic plants. The area’s
endemism might be even higher than 52%, as a number of taxa could not be
identified. For example, only 3 of the 16 rattan species (Arecaceae) recorded,
which generally show a high degree of endemism (Dransfield 1990), could be
assigned to a scientific name. Two of them were Philippine endemics.
Another aspect referring to the conservation value of the area is the
occurrence of 41 tree species in the red list of IUCN (2000). However, from the
species recorded from this study, other than trees are not represented in the red
list. Despite this, it can be expected that many of the non-tree taxa recorded are
threatened by habitat destruction as well. For example, no rattans are listed by
IUCN although this plant group is still heavily exploited and shows high rates
of endemism. The IUCN red list seems to have a strong focus on well known
and economically important tree species. This is supported by the fact that only
dipterocarps are classified as critically endangered, although many other tree
species are more rare in the study area. This was e.g. true for the valuable tree
species Heritiera sylvatica (Sterculiaceae) and Xanthostemon verdugonianus
(Myrtaceae) which were known by local farmers from only one mature tree
each in the entire western foothills of Mt. Pangasugan.
Taken together, the Mt. Pangasugan region on Leyte represents a unique
refuge for a high number of species, which are characteristic of all Philippine

[223]
1284

dipterocarp forest types and the molave type. In view of the large areas of
degraded land in the Philippines, the conservation value of the Mt. Pangasugan
region is very high and represents an important gene bank of the Philippine
forest vegetation.

Acknowledgements

This study was part of the ViSCA-gtz Applied Tropical Ecology Program (PN
95.2290.5–001.00) and partly funded by the Tropical Ecology Support Pro-
gram (TÖB) of GTZ (PN 90.2136.1–03.107). We are grateful to the President
of the Leyte State University, Dr. P.P. Milan and her staff for their support and
help. We are also grateful to the Cienda San Vicente Farmer Association
(CSVFA) and their community organiser Marlito Bande, who made the
extensive field trips possible. Special thanks to the curator of the Philippine
National Herbarium, Dr. Madulid, and his staff, as well as to the director of
the National Herbarium of the Netherlands, Leiden branch, Prof. Baas and his
staff who helped to put species identification on a firm ground. We are also
very grateful to Leonardo Co, Nina Ingle, David and Luze Bicknell, Franz
Seidenschwarz, and B.C. Tan for various assistance. We also like to thank the
two anonymous referees for their comments and constructive criticism.

Appendix

Species list of the vascular plant species found in 49 plots (100 m2 each) in the
foothills of the Leyte Cordillera at Mt. Pangasugan, Leyte, Philippines.
The list includes only those species which could be identified to species level.
Some species recorded outside the plots are provided in addition.
Numbers in brackets following the species name indicate the first voucher
specimen collected of this species.
Life form classification of species is based on observations of mature indi-
viduals in the study area, or from species descriptions in literature. Life form
definitions follow Ellenberg and Mueller-Dombois (1967) with a minor revision
by Richter (1997). MacP, Macrophanerophyte (>20–50 m); MesP, Meso-
phanerophyte (>5–20 m); MiP, Microphanerophyte (>2–5 m); NP, Nano-
phanerophyte (>1–2 m); NP herb, herbaceous Nanophanerophyte; Ch,
Chamaephyte ( £ 1 m); Ch frut, fruticose Chamaephyte; Ch suff, suffruticose
Chamaephyte; Ch herb, herbaceous Chamaephyte; G rhiz, rhizome Geophyte;
PL, Phanerophytic Liana; r PL, root PL; st PL, winding PL; el PL, tendril PL; d
PL, spread climber; E, Epiphyte.

[224]
 1285

Species classified by IUCN (2000) as endangered are listed along with their
status in bold letters. Short definitions of the status are:
CR, critically endangered (‘… facing an extremely high risk of extinction in
the wild in the immediate future …’); EN, endangered (‘… not critically
endangered but facing a very high risk of extinction in the wild in the near
future …’); VU, vulnerable (‘… not critically endangered or endangered but
facing high risk of extinction in the wild in the medium-term future …’).
For comprehensive definitions and criteria of classification see www.
iucnredlist.org/search-basic.html
I.

Spermatophyta Life form

Aceraceae
Acer laurinum Hassk. (1221) MacP
Actinidiaceae
Saurauia cf. denticulata C.B. Rob. (1078) MiP
Saurauia samarensis Merr. (235) MiP
Alangiaceae
Alangium longiflorum Merr. (1331) VU MesP
Amaranthaceae
Deeringia polysperma (Roxb.) Moq. (2214) Ch herb
Anacardiaceae
Dracontomelon dao (Blco.) Merr. & Rolfe (660) MacP
Dracontomelon edule (Blco.) Skeels Mes
Koordersiodendron pinnatum (Blco) Merr. (162) MacP
Mangifera altissima Blco. (971) VU MacP
Rhus taitensis Guill. (818) MesP
Semecarpus cuneiformis Blco. (538) MiP
Annonaceae
Alphonsea arborea (Blco.) Merr. (1009) MesP
Anaxagorea javanica Bl. (1509) MiP
Artabotrys cf. rolfei Vid. (2159) el PL
Cananga odorata (Lamk.) Hook. f. & Thoms. MesP
Goniothalamus elmeri Merr. (327) MiP
Meiogyne virgata (Bl.) Miq. MesP
Papualthia cf. lanceolata (Vid.) Merr. (206) MesP
Popowia pisocarpa (Bl.) Endl. (1054) MesP
Apocynaceae
Alstonia macrophylla Wall. ex. G. Don (1774) MesP
Alstonia scholaris (L.) R. Br. MesP
Kibatalia blancoi (Rolfe) Merr. (467) MesP
Lepiniopsis ternatensis Val. (2220) MesP
Tabernaemontana pandacaqui Poir. (140) MicP
Voacanga globosa (Blco.) Merr. (218) MiP
Araceae
Alocasia cf. zebrina Schott ex van Houtte (1677) Ch herb
Amorphophallus paeoniifolius (Dennst.) Nicolson (1924) NP herb
Costus speciosus (J. Konig) Sm NP herb

[225]
1286

Spermatophyta Life form

Pothos cylindricus Presl (1226) r PL
Raphidophora korthalsii Schott (881) d PL
Araliaceae
Arthrophyllum ahernianum Merr. (1911) MesP
Osmoxylon trilobatum (Merr.) Philipson (220) NP
Polyscias nodosa (Bl.) Seem. MesP
Arecaceae
Calamus cf. merrillii Becc. d PL
Caryota cf. cumingii Lodd. ex Mart MiP
Caryota cf. mitis Lour. (560) MesP
Daemonorops cf. mollis (Blco.) Merr. (593) PL
Korthalsia laciniosa Mart. (1120) d PL
Pinanga maculata Porte MiP
Caryota rumphiana Mart. var. philippinensis Becc. MacP
Aristolochiaceae
Aristolochia philippinensis Warb. (702) Ch suff
Asclepiadaceae
Hoya multiflora Bl. (689) PL
Asteraceae
Vernonia arborea Buch.-Ham. (826) MesP
Bignoniaceae
Oroxylum indicum (L.) Vent. MesP
Radermachera pinnata (Blco.) Seem. (1129) MesP
Burseraceae
Canarium asperum Benth. (265) MacP
Canarium denticulatum Bl. (428) MesP
Canarium euryphyllum Perk. (1265) MacP
Canarium gracile Engl. (611) MesP
Canarium hirsutum Willd. (1714) MesP
Dacryodes rostrata (Bl.) H. J. Lam (247) MacP
Caprifoliaceae
Sambucus javanica Reinw. ex Bl. MiP
Casuarinaceae
Gymnostoma rumphianum (Miq.) L.A.S. Johnson (1915) MesP
Cecropiaceae
Poikilospermum erectum (Blco) Merr. (321) d PL
Poikilospermum suaveolens (Bl.) Merr. (328) d PL
Celastraceae
Bhesa paniculata Arn. (812) MesP
Euonymus cochinchinensis Pierre (495) MesP
Euonymus javanicus Bl. (232) MesP
Lophopetalum javanicum (Zoll.) Turcz. (1127) MacP
Chloranthaceae
Chloranthus erectus (Buch.-Ham.) Verdc. (1522) Ch suff
Sarcandra glabra (Thunb.) Nakai (562) Ch frut

[226]
 1287

Spermatophyta Life form

Chrysobalanaceae
Maranthes corymbosa Bl. (790) MacP
Clusiaceae
Calophyllum blancoi Pl. & Tr. (278) MesP
Calophyllum soulattri Burm. f. (1250) MesP
Cratoxylum formosum Benth. & Hook. f. ex Dyer (452) MesP
Combretaceae
Terminalia microcarpa Decne. (672) MacP
Terminalia nitens Presl (481) VU MesP
Commelinaceae
Floscope scandens Lour. Ch herb
Forrestia hispida Less. & A. Rich. (423) Ch herb
Pollia sorzogoniensis (E. Meyer) Steud. Ch herb
Pollia thyrsiflora (Bl.) Steud. Ch herb
Rhopalephora cf. vitiensis (Seem.) Fader (2102) Ch herb
Connaraceae
Agelaea borneensis (Hook. f.) Merr. (491) st PL
Connarus culionensis Merr. (686) PL
Connarus semidecandrus Jack (623) PL
Ellipanthus tomentosus Kurz (396) MesP
Crypteroniaceae
Crypteronia cumingii (Planch.) Planch. ex Endl. (1693) MesP
Cunoniaceae
Weinmannia cf. hutchinsonii Merr. (130) MesP
Datiscaceae
Octomeles sumatrana Miq. MacP
Dilleniaceae
Dillenia megalantha Merr. (2007) VU MesP
Dillenia philippinensis Rolfe VU MesP
Tetracera fagifolia Bl. (674) st PL
Dioscoreaceae
Dioscorea hispida Dennst. PL
Dipterocarpaceae
Anisoptera thurifera Foxw. ssp. thurifera (353) MacP
Dipterocarpus gracilis Bl. (486) CR MacP
Dipterocarpus validus Bl. CR MacP
Hopea acuminata Merr. (292) CR MacP
Hopea malibato Foxw. ex Elm. (20) CR MacP
Hopea philippinensis Dyer (925) CR MacP
Hopea plagata (Blco.) Vid. (305) CR MacP
Parashorea malaanonan (Blco.) Merr. (267) CR MacP
Shorea almon Foxw. (430) CR MacP
Shorea assamica Dyer forma philippinensis (Brandis) Sym. (269) CR MacP
Shorea astylosa Foxw. (1796) CR MacP
Shorea cf. hopeifolia (Heim) Sym. (2110) CR MakP
Shorea contorta Vid. (1001) CR MacP

[227]
1288

Spermatophyta Life form

Shorea falciferoides Foxw. ssp. falciferoides (290) CR MacP
Shorea guiso (Blco) Bl. (384) CR MacP
Shorea palosapis (Blco) Merr. (263) CR MacP
Shorea polysperma (Blco) Merr. (297) CR MacP
Vatica mangachapui Blco. (528) EN MacP
Ebenaceae
Diospyros blancoi A. DC. (163) VU MacP
Diospyros cf. nitida Merr. (1901) MiP
Diospyros curranii Merr. (1631) MesP
Diospyros multibracteata Merr. (598) MiP
Diospyros pilosanthera Blco. MesP
Diospyros pyrrhocarpa Miq. (385) MesP
Elaeagnaceae
Elaeagnus triflora Roxb. var. triflora (412) dPL frut
Elaeocarpaceae
Elaeocarpus cumingii Turcz. (1123) MesP
Euphorbiaceae
Acalypha amentacea Roxb. (254) NP
Antidesma digitaliforme Tul. (371) NP
Antidesma nitidum Tul. (268) MiP
Antidesma tomentosum Bl. (919) MicP
Aporosa benthamiana Hook. f. (573) MiP
Baccaurea tetrandra (Baill.) Müll. Arg. (360) MesP
Bridelia glauca Bl. (233) MesP
Claoxylon brachyandrum Pax & K. Hoffm. (379) MesP
Cleistanthus cf. glaber Airy Shaw (628) MesP
Cleistanthus sumatranus (Miq.) Müll. Arg. (396) MesP
Codiaeum luzonicum Merr. MiP
Croton cascarilloides Raeusch. (205) NP
Drypetes cf. megacarpa (Bl.) Pax & Hoffm. (374) MiP
Drypetes longifolia (Merr.) Pax et Hoffm. (372) MiP
Glochidion rubrum Bl. (715) MiP
Macaranga caudatifolia Elm. (735) VU MiP
Macaranga grandifolia (Blcol.) Merr. VU MesP
Macaranga hispida (Bl.) Muell.-Arg. MiP
Macaranga tanarius (L.) Muell.-Arg. MesP
Mallotus cf. paniculatus (Lam.) Muell.-Arg. (330) MesP
Mallotus floribundus (Bl.) Muell.-Arg. (228) MesP
Mallotus lackeyi Elm. (1800) MesP
Mallotus philippensis (Lam.) Muell.-Arg. (302) MesP
Neotrewia cumingii (Muell.-Arg.) Pax & Hoffm. (343) MesP
Omalanthus populneus (Geisel.) Pax MiP
Phyllanthus leytensis Elm. (250) Ch frut
Suregada glomerulata (Hassk.) Jones (287) NP
Fagaceae
Lithocarpus buddii (Merr.) A. Camus (15) MacP
Lithocarpus caudatifolia (Merr.) Rehd. (555) MesP
Lithocarpus coopertus (Blco) Rehd. (387) MesP

[228]
 1289

Spermatophyta Life form

Flacourtiaceae
Casearia cf. mindanaensis Merr. (1675) P
Casearia grewiaefolia Vent. var. gelonioides (Bl.) Sleum. (794) MesP
Flacourtia cf. indica (Burm. f.) Merr. (378) MesP
Osmelia philippina (Turcz.) Benth. (352) MesP
Flagellariaceae
Flagellaria indica L. (1128) el PL
Gesneriaceae
Cyrtandra angularis Elm. (2212) Ch herb
Cyrtandra glaucescens Kranzl. (960) Ch herb
Monophyllaea merrilliana Kranzl. (2027) Ch herb
Rhynchoglossum obliquum Bl. Ch herb
Gnetaceae
Gnetum gnemon L. var. gnemon (375) MesP
Gnetum latifolium Bl. PL
Hamamelidaceae
Sycopsis dunnii Hemsl. (739) MesP
Hernandiaceae
Illigera megaptera Merr. (721) PL
Icacinaceae
Gomphandra cumingiana (Miers) F.-Vill. (1118) MesP
Gonocaryum calleryanum (Baill.) Becc. (700) MesP
Miquelia celebica Bl. PL
Phytocrine macrophylla (Bl.) Bl. var. macrophylla PL
Platea excelsa Bl. var. borneensis (Heine) Sleum. (1217) MesP
Ixonanthaceae
Ixonanthes petiolaris Bl. MesP
Juglandaceae
Engelhardtia serrata Bl. (411) MacP
Lamiaceae
Gomphostemma javanicum (Bl.) Bth. (285) Ch herb
Lauraceae
Actinodaphne apoensis Merr. (1083) MesP
Actinodaphne bicolor (Elm.) Merr. MesP
Actinodaphne cf. multiflora Benth. (833) MesP
Caryodaphnopsis tonkinensis (Lec.) Shaw (441) MesP
Cinnamomum mercadoi Vid. (468) VU MacP
Endiandra coriacea Merr. (1883) MesP
Litsea garciae Vid. (478) MesP
Litsea leytensis Merr. (805) VU MesP
Neolitsea cf. vidallii Merr. (272) VU MiP

[229]
1290

Spermatophyta Life form

Leeaceae
Leea aculeata Bl. ex Spreng. Mip
Leea guineensis G. Don (255) MiP
Leea quadrifida Merr. (546) MiP
Leguminosae
Afzelia rhomboidea (Blco.) Vid. VU MesP
Albizia procera (Roxb.) Benth. MacP
Albizia saponaria (Lour.) Bl. ex Miq. MesP
Archidendron clypearia var. casai (Blco.) I.C. Nielsen (1082) MesP
Archidendron pauciflorum (Benth.) Nielsen (1852) MiP
Archidendron scutiferum (Blco.) I.C. Nielsen (323) MesP
Bauhinia integrifolia Roxb. ssp. cumingiana (Benth.) K. & S.S. Larsen (364) PL
Dalbergia cf. mimosella (Blco) Prain (1435) MesP
Desmodium laxum DC. (820) Ch herb
Entada phaseoloides (L.) Merr. el PL
Erythrina subumbrans (Hassk.) Merr. MacP
Euchresta horsfieldii (Lesch.) Benn. (847) Ch herb
Kingiodendron alternifolium (Elm.) Merr. & Rolfe (357) MacP
Ormosia calavensis Azaola MesP
Pterocarpus indicus Willd. VU MesP
Wallaceodendron celebicum Koord. (395) MacP
Liliaceae
Dianella ensifolia (L.) DC. Ch herb
Loganiaceae
Fagraea auriculata Jack ssp. auriculata (851) st PL
Fagraea racemosa Jack ex Wall. MiP
Strychnos luzoniensis Elm. (748) el PL
Magnoliaceae
Magnolia liliifera (L.) Baill. var. angatensis (719) MesP
Marantaceae
Donax cannaeformis (Forst. f.) K. Schum. (1006) MiP
Marattiaceae
Angiopteris evecta (Forst.) Hoffm. (1188) NP herb
Marattia pellucida Presl (1444) Ch herb
Melastomataceae
Memecylon paniculatum Jack (311) MiP
Meliaceae
Aglaia argentea Bl. (642) MesP
Aglaia costata Merr. (275) VU MesP
Aglaia elliptica Bl. (1295) MesP
Aglaia luzoniensis (Vid.) Merr. & Rolfe (511) MiP
Aphanamixis polystachia (Wall.) R.N. Parker (941) MesP
Chisocheton ceramicus (Miq.) C. DC.) (1184) MesP
Chisocheton cumingianus (C. DC.) Harms (211) MesP
Chisocheton pentandrus (Blco.) Merr. (753) MesP

[230]
 1291

Spermatophyta Life form

Dysoxylum arborescens (Bl.) Miq. (664) MesP
Dysoxylum cumingianum C. DC. (316) MesP
Reinwardtiodendron humile (Hassk.) Mabb. (965) MesP
Toona calantas Merr. & Rolfe (918) MacP
Vavaea amicorum Benth. (273) NP
Walsura cf. pinnata Hassk. (2082) MesP
Menispermaceae
Arcangelisia flava (L.) Merr. (1798) el PL
Monimiaceae
Matthaea pubescens Merr. (139) MiP
Moraceae
Ficus aurita Bl. (210) MiP
Artocarpus blancoi (Elm.) Merr. (1701) VU MacP
Artocarpus elastica Reinw. ex Bl. (697) MacP
Ficus balete Merr. (v) MacP
Ficus benjamina L. (1075) MacP
Ficus cumingii Miq. var. angustissima (Merr.) Corner (778) MesP
Ficus fistulosa Reinw. ex Bl. (1307) MiP
Ficus heteropoda Miq. (425) MiP
Ficus odorata (Blco.) Merr. MesP
Ficus pedunculosa Miq. (1780) MiP
Ficus pseudopalma Blco. NP
Ficus punctata Thunb. (406) r PL
Ficus ribes Reinw. ex Bl. (405) MiP
Ficus ruficaulis Merr. MesP
Ficus subulata Bl. (646, 1966) PL
Ficus ulmifolia Lam. (451) VU MesP
Maclura cochinchinensis (Lour.) Corner (1417) d PL
Streblus ilicifolia (Vid.) Corner (1700) MesP
Streblus macrophyllus Bl. (216, 335, 613) MesP
Myristicaceae
Endocomia macrocoma (Miq.) W.J.J. de Wilde subsp. prainii MesP
(King) W.J.J.de Wilde (479)
Gymnacranthera farquhariana (Hook. f. & Th) Warb. var. MesP
paniculata (A. DC.) R. Schouten (541)
Horsfieldia cf. costulata (Miq.) Warb. (2002) MesP
Knema glomerata (Blco.) Merr. (641) MesP
Knema stellata Merr. (1481) MesP
Myristica cf. frugifera W. J. J. de Wilde (743) VU MesP
Myristica cf. philippensis Lam. VU MesP
Myristica simiarum A. DC cf ssp. simiarum (417) MesP
Myrsinaceae
Ardisia pardalina Mez. (815) Ch frut
Ardisia squamulosa Presl (204) VU Ch frut
Maesa denticulata Mez (241) MiP
Myrtaceae
Acmena acuminatissima (Bl.) Merr. & Perry (503) MacP

[231]
1292

Spermatophyta Life form

Syzygium cf. densinervium (Merr.) Merr. (749) MesP
Syzygium cf. xanthophyllum (C.B. Rob.) Merr. MesP
Syzygium cumini (L.) Skeel MesP
Tristaniopsis decorticata (Merr.) P.G. Wilson MesP
& J.T. Waterh. (142) VU
Tristaniopsis micrantha (Merr.) P.G. Wilson & J.T. Waterh. (301) MesP
Xanthostemon verdugonianus Naves (2209) VU MacP
Ochnaceae
Gomphia serrata (Gaertn.) Kanis MesP
Olacaceae
Erythropalum scandens Bl. (780) el PL
Strombosia philippinensis (Baill.) Rolfe (380) MesP
Oleaceae
Olea borneensis Boerl. (306) MesP
Opiliaceae
Champereia manillana (Bl.) Merr. (100) MesP
Melientha suavis Pierre ssp. suavis (366) MesP
Orchidaceae
Calanthe triplicata (Willem.) Ames G rhiz
Ceratostylis retisquama Rchb. f.B143 E
Cymbidium aliciae Quis. (880) E
Eulophia zollingeri (Reichb.f.) J.J.Smith G rhiz
Grammatophyllum multiflorum var. tigrinum Lindley. E
Lepidogyne longifolia (Bl.) Bl. G rhiz
Liparis wenzelii Ames G rhiz
Phalaenopsis hieroglyphica (Rchb. f.) Sweet E
Robiquetia cf. compressa Schltr. E
Trichoglottis latisepala Ames E
Trichoglottis rosea (Lindl.) Ames (1055) E
Pandanaceae
Freycinetia cf. philippinensis Hemsl. (1353) r PL
Freycinetia cumingiana Gaudich. (388, 1234) r PL
Freycinetia multiflora Merr. (1130) r PL
Freycinetia vidalii Hemsl. (1352) r PL
Freycinetia membranifolia Elm. (955) r PL
Pentaphragmataceae
Pentaphragma grandiflorum Kurz (457, 458, 1407) Ch herb
Piperaceae
Piper abbreviatum Opiz (638) st PL
Piper halconense C. CD. st PL
Piper toppingii C. CD. (654, 1143) st PL
Piper viminale Opiz (1205) st PL
Pittosporaceae
Pittosporum resiniferum Hemsl. (448) Mesp

[232]
 1293

Spermatophyta Life form

Poaceae
Bambusa spinosa Roxb. MesP
Dinochloa cf. pubiramea Gamble PL
Dinochloa cf. scandens (Bl.) O. Ktze. PL
Podocarpaceae
Podocarpus rumphii Bl. (1520) MacP
Polygalaceae
Polygala venenosa Juss. ex Poir. (284, 2011) Ch herb
Xanthophyllum vitellinum (Bl.) Dietr. (992) MesP
Proteaceae
Helicia graciliflora Merr. (1154) MiP
Helicia loranthoides Presl. (1079) MesP
Helicia robusta (Roxb.) R. Br. ex Wall. (588) MiP
Ranunculaceae
Clematis javana DC. (159, 1997) PL
Rhamnaceae
Ventilago dichotoma (Blco.) Merr. (723) PL
Ziziphus angustifolius (Miq.) Hatusima ex Steenis (488) MesP
Ziziphus crebrivenosa C.B. Rob. (492, 661) d PL
Rhizophoraceae
Gynotroches axillaris Bl. (1538) MacP
Rosaceae
Prunus arborea (Bl.) Kalkm. var. arborea (1624) MesP
Prunus cf. fragrans (Elm.) Kalkm. (795) MesP
Prunus grisea (Bl.) Kalkm. var. grisea (71, 490) MesP
Rubus fraxinifolius Poiret (2017) d PL suff
Rubiaceae
Boholia nematostylis Merr. (1919) Ch herb
Canthium gynochthodes Baill. (563) MesP
Diodia ocynifolia (Willd.) Brem. (1424) PL
Diplospora cf. fasciculiflora Elm. (663) MiP
Dolicholobium philippinense Trenteuse (260) MiP
Greeniopsis multiflora (Elm.) Merr. (279) MesP
Hedyotis baruensis (Miq.) Val. ex Merr. (329) Ch herb
Hypobathrum purpureum (Elm.) Merr. (1507) MesP
Ixora bartlingii Elm. (1060) Mip
Ixora cf. cumingiana Vidal (509) MiP
Ixora cf. macrophylla Bartl. (207) MiP
Ixora longistipula Merr. (1122) MiP
Ixora macrophylla Barth. MiP
Ixora salicifolia (Bl.) DC. (288) NP
Lasianthus cf. obliquinerva Merr. (701) MiP
Morinda bracteata Roxb. (326) MiP
Mussaenda philippica A. Rich. MiP
Mussaenda vidallii Elm. (129) MiP
Mycetia javanica (Bl.) Korth. (258) Ch suff
Nauclea subdita (Korth.) Stend. (1958) MiP

[233]
1294

Spermatophyta Life form

Neonauclea formicaria (Elm.) Merr. (793) MiP
Neonauclea lanceolata (Bl.) Merr. subsp. gracilis (Vidal) Ridsdale (402) MesP
Praravinia cf. mindanensis (Elm.) Brem. (289) NP
Psychotria cf. ixoroides Bartl. ex DC. (515) st PL
Psychotria membranifolia Bartl. ex DC. (257) NP
Tarenna cumingiana (Vid.) Elm. (464) MesP
Tarrenoidea wallichii (Hook. f.) D.D.Tirvengadum & C. Sastre (307) MesP
Timonius arboreus Merr. (248) MiP
Uncaria cf. perrottetii (A. Rich.) Merr. (325) el PL
Uncaria lanosa Wall. f. philippinensis (Elm.) Ridsd. (900) el PL
Uncaria longiflora (Poir.) Merr. (1300) el PL
Wendlandia luzoniensis DC. (444) MesP
Xanthophytum fruticulosum Reinw. ex Bl. (1005) NP
Rutaceae
Clausena anisum
olens (Blco.) Merr. (605) NP
Lunasia amara Blco. (158) NP
Micromelum compressum (Blco.) Merr. (771) NP
Severinia disticha (Blco) Merr. (398) NP
Sapindaceae
Allophyllus cobbe (L.) Raeuschel (823) MesP
Cubilia cubili (Blco.) Adelh. (586) MacP
Dictyoneura acuminata Bl. ssp. acuminata (246) MesP
Dimocarpus fumatus (Bl.) Leenhouts ssp. philippinensis Leenhouts (72) MesP
Euphorianthus obtusatus Radlk. ex Koord. (1641) MesP
Ganophyllum falcatum Bl. (1212) MesP
Guioa cf. diplopetala (Hassk.) Radlk. (1104) MesP
Harpullia cupanioides Roxb. (212) MesP
Lepisanthes fruticosa (Roxb.) Leenh. (933) MesP
Nephelium cf. ramboutan
ake (Labill.) Leenh. (442) MesP
Paranephelium cf. xestophyllum Miq. (727) MesP
Pometia pinnata Forst. (578, 1546) MesP
Sapotaceae
Palaquium philippense (Perr.) C. B. Rob. (443) VU MacP
Planchonella mindanaensis Clemens (1126) MacP
Pouteria firma (Miq.) Baehni (1237) MacP
Saxifragaceae
Dichroa philippinensis Schltr. Ch frut
Polyosma integrifolia Bl. (1219) NP
Dichroa fibrifuga (807)
Simaroubaceae
Picrasma javanica Bl. (218) MesP
Solanaceae
Solanum anisophyllum Elm. (225) Ch herb
Solanum ferox L. (282) Ch herb
Sonneratiaceae
Duabanga moluccana Bl. MacP

[234]
 1295

Spermatophyta Life form

Staphyleaceae
Turpinia borneensis (Merr. & Perry) B.L. Linden (1802) MesP
Sterculiaceae
Heritiera sylvatica Vidal (1768) MesP
Pterocymbium tinctorium (Blco.) Merr. (345) MacP
Pterospermum diversifolium Bl. (270) MesP
Pterospermum elongatum Korth. (434) MesP
Pterospermum obliquum Blco. (120) MesP
Sterculia multistipularis Elm. (251) MiP
Sterculia oblongata R. Br. (678) MesP
Sterculia philippinensis Merr. (1898) MesP
Sterculia stipulata Korth. var. jagorii (Warb.) Tantra MesP
Symplocaceae
Symplocos cochinchinensis(Lour.) Moore var. cochinchinensis (1954) MicP
Taccaceae
Tacca palmata Bl. (303) G rhiz
Theaceae
Eurya acuminata DC. (1314) NP
Ternstroemia philippinensis Merr. var. philippinensis (1491) MesP
Thymelaeaceae
Aquilaria cumingiana (Decn) Ridl. (300) NP
Phaleria perrottetiana (Dcne) F.-Vill. (160) Ch suff
Tiliaceae
Colona serratifolia Cav. (626) MesP
Diplodiscus paniculatus Turcz. (271) VU MesP
Ulmaceae
Celtis cf. philippinensis Blanco MesP
Gironniera celtidifolia Gaudich. (238) MiP
Trema orientalis (L.) Bl. (2202) MesP
Urticaceae
Cypholophus moluccanus (Bl.) Miq. Ch frut
Leucosyke capitellata (Poir.) Wedd. (242) MiP
Maoutia setosa Wedd. NP
Villebrunea rubescens (Bl.) Bl. (324) MesP
Villebrunea trinervis Wedd. (733) MesP
Verbenaceae
Clerodendrum villosum Bl. NP
Teijsmanniodendron pteropodum (Miq.) Bakh. (157) MesP
Vitex parviflora Juss. (1837) MacP
Vitex turczaninowii Merr. (705) MacP
Premna odorata Blco. (633) MesP

[235]
1296

II.

Ptaridophyta Life form

Aspidiaceae
Didymochlaena cf. truncatula (Sw.) J. Sm. (2056) Ch herb
Aspleniaceae
Asplenium nidus L. (1902) Ch herb
Asplenium tenerum Forst. (2096) Ch herb
Athyriaceae
Diplazium asperum (Bl.) Milde (1809) Ch herb
Diplazium esculentum (Retz.) Sw. (1846) Ch herb
Cyatheaceae
Cyathea cf. contaminans (Hook.) Copel. MesP
Davalliaceae
Davallia solida (G. Forst.) Sw. (1462) Ch herb
Davallia trichomanoides Bl. var. lorrainii (Hance) Holttum (222) Ch herb
Hymenophyllaceae
Trichomanes javanicum Bl. (1042) Ch herb
Lindsaeaceae
Lindsaea lucida Bl. ssp. lucida (533) Ch herb
Sphenomeris chinensis (L.) Maxon Ch herb
Tapeinidium pinnatum (Cav.) C.Chr. (1267) Ch herb
Lomariopsidaceae
Bolbitis cf. guoyana (Gaudich.) Ching (2016) Ch herb
Bolbitis guoyana (Gaudich.) Ching Ch herb
Bolbitis heteroclita (Presl) Ching (1049) r PL
Lomogramma cf. copelandii Holttum (1851) r PL
Lomogramma copelandii Holttum r PL
Teratophyllum arthropteroides (Christ) Holttum (2084) Ch herb
Teratophyllum cf. articulatum (J. Sm. ex Fèe) Mett. (516) Ch herb
Osmundaceae
Osmunda banksiaefolia (Pr.) Kuhn (1261, 1392) Ch herb
Polypodiaceae
Drynaria quercifolia (L.) J. Sm E
Leptochilus cf. decurrens Bl. Ch herb
Microsorum cf. longissimum J. Sm. ex Fée (964) Ch herb
Microsorum membranifolium (R. Br.) Ching Ch herb
Microsorum punctatum (L.) Copel. (1821) Ch herb
Microsorum scolopendria (Burm. f.) Copel. (1445) Ch herb
Pyrrosia cf. lanceolata (L.) Farwell PL
Microsorum plukenetii (Presl) M.G. Price (1860)
Pteridaceae
Pteris cf. pellucida Presl Ch herb
Pteris ensiformis Burm. f. (1806) Ch herb
Pteris longipinnula Wall. (334) Ch herb
Schizaeaceae
Lygodium auriculatum (Willd.) Alst. et Holtt. (1974) st PL
Lygodium circinnatum (Burm. f.) Sw. (1603) st PL

[236]
 1297

Ptaridophyta Life form

Selaginellaceae
Selaginella cf. involvens (Sw.) Spring (856) Ch frut
Selaginella cf. springiana Alderw. (1526) Ch frut
Selaginella cupressina (Willd.) Spring (745) Ch frut
Selaginella engleri Hieron. (1011) Ch frut
Taenitidaceae
Taenitis blechnoides (Willd.) Sw. (1091) Ch herb
Tectaria group
Ctenitis cf. silvatica Holttum (939) Ch herb
Cyclopeltis crenata (Fée) C. Chr. (1807) Ch herb
Pleocnemia cf. presliana Holttum (1849) Ch herb
Pleocnemia irregularis (Presl) Holttum (1007) Ch herb
Tectaria crenata Cav. (1301) Ch herb
Thelypteridaceae
Cyclosorus sumatranus (v. Ald. v. Ros.) Ching Ch herb
Pneumatopteris laevis (Mett.) Holttum (1812) Ch herb
Pronephrium · xiphioides (Christ) Holttum (498) Ch herb
Pseudophegopteris paludosa (Bl.) Ching (2093) Ch herb
Pronephrium granulosum (Presl) Holtt. (997) Ch herb

References

Anonymous 1993. Flora of Taiwan Vol. 3, Angiosperms, Dicotyledons, , 2nd edn. Editorial
Committee of the Flora of Taiwan, Department of Botany, National Taiwan University, Taipei,
Taiwan, ROC.
Anonymous 1994. Flora of Taiwan Vol. 1, Pteridophyta, Gymnospermae, 2nd edn. Editorial
Committee of the Flora of Taiwan, Department of Botany, National Taiwan University, Taipei,
Taiwan, ROC.
Anonymous 1996. Flora of Taiwan Vol. 2, Angiospermae, 2nd edn. Editorial Committee of the
Flora of Taiwan, Department of Botany, National Taiwan University, Taipei, Taiwan, ROC.
Aragones E.G. Jr. 1991. Vegetation-soil pattern along altitudinal gradient in the western slopes of
Mt. Banahaw, Luzon, Philippines: I. The forest communities and changes in forest composition
with altitude. Sylvatrop 1(1): 15–45.
Ashton P.S. 1993. Philippine phytogeography. Asia Life Sciences 2(1): 1–8.
Asio V. 1996. Characteristics, weathering, formation and degradation of soils from volcanic rocks
in Leyte, Philippines. Hohenheimer Bodenkundliche Hefte 33 Universität Hohenheim Institut
für Bodenkunde und Standortslehre, Stuttgart, 209 pp.
Barrera A., Aristorenas I. and Tingzon J.A. 1954. Soil Survey of Leyte Province, Philippines. Soil
Survey Report No. 18, Bureau of Printing, Manila, 103 pp.
Bremer H. 1995. Boden und Relief in den Tropen: Grundvorstellungen und Datenbank. Gebrüder
Bornträger, Berlin, Stuttgart, 324 pp.
Bremer H. 1999. Die Tropen – Geographische Synthese einer fremden Welt im Umbruch. Gebrüder
Bornträger, Berlin, Stuttgart, 428 pp.
Brooks T.M., Mittermeier R.A., Mittermeier C.G., da Fonseca G.A.B., Rylands A.B., Konstant
W.R., Flick P., Pilgrim J., Oldfield S., Magin G. and Hilton-Taylor C. 2002. Habitat loss and
extinction in the hotspots of biodiversity. Conservation Biology 16(4): 909–923.

[237]
1298

Brown W.H. 1919. Vegetation of Philippine Mountains: The Relation between the Environment
and Physical Types at Different Altitudes. Bureau of Science, Manila, 434 pp.
Brown W.H. and Mathews D.M. 1914. Philippine dipterocarps forests. Philippine Journal of
Science A 9(5): 413–561.
Buot I.E. Jr. and Okitsu S. 1997. Woody species composition in the altitudinal zones of the mossy
forest of Mt. Pulog, Luzon, Philippines. Flora Malesiana Bulletin 12(1): 6–11.
Burger D. 1972. Seedlings of Some Tropical Trees and Shrubs Mainly of South East Asia. Pudoc,
Wageningen, 399 pp.
Cain S.A., de Oliveira Castro G.M., Pires J.M. and da Silva N.T. 1956. Application of some
phytosociological techniques to Brazilian rain forest. American Journal of Botany 43: 911–941.
Cheek M. and Jebb M. 2001. Nepenthaceae. Flora Malesiana Vol. 15, Nationaal Herbarium
Nederland, Universiteit Leiden Branch, the Netherlands, 164 pp.
Corner E.J.H. 1988. Wayside Trees of Malaya. Vol. 1 and 2. Malayan Nature Society, Kuala
Lumpur, 774 pp.
Cromer D.A.N. and Pryor L.D. 1942. A contribution to rain-forest ecology. Proceedings of the
Linnean Society 67: 249–268.
Dargantes B.B. and Koch W. 1994. Case studies on the occupation and cultivation of the forest
lands of Leyte, Philippines. Annals of Tropical Research 16: 13–29.
Davis S.D., Heywood V.H. and Hamilton A.C. (eds) 1995. Centres of Plant Diversity – A Guide
and Strategy for their Conservation, Vol. 2: Asia, Australasia and the Pacific. IUCN Publications
Unit, Cambridge, UK.
de Guzman E.D., Umali R.M. and Sotalbo E.D. 1986. Guide to Philippine Flora and Fauna, Vol.
3: Dipterocarps, Non-Dipterocarps. Natural Resources Management Center, Ministry of Nat-
ural Resources and University of the Philippines, 414 pp.
DENR 1998. Philippine atlas. Department of Environment and Natural Resources (DENR),
Quezon City, Philippines.
DENR and UNEP 1997. Philippine Biodiversity: An Assessment and Action Plan. Bookmark Inc.,
Makati City, The Philippines, 298 pp.
Dransfield J. 1990. Outstanding problems in Malesian palms. In: Baas P., Kalkmann K. and
Geesink R. (eds), The Plant Diversity of Malesia, Proceedings of the Flora Malesiana Sympo-
sium, Leiden, the Netherlands, August 1989. Kluwer Academic Publishers, Dordrecht-Boston-
London, pp. 17–25.
Duivenvoorden J.F. 1994. Vascular plant species counts in the rain forests of the middle Caquetá
area. Colombian Amazonia. Biodiversity and Conservation 3: 685–715.
Ellenberg H. and Mueller-Dombois D. 1967. A key to Raunkiaer plant life forms with revised
subdivisions. Berichte des geobotanischen Instituts Rübel 37: 56–73.
FAO 2003. State of the World’s Forests 2003. FAO, Rome, Italy. www.fao.org/DOCREP/005/
Y7581E/Y7581E00.HTM, accessed 25 February 2004.
FAO/UNESCO 1988. Soil Map of the World, Revised Legend. World Soil Resources Report 60.
FAO, Rome.
Gentry A.H. and Dodson C. 1987. Contribution of nontrees to species richness of a tropical rain
forest. Biotropica 19(2): 149–56.
Gonzales-Salcedo P.V. 2001. Floral diversity and vegetation zones of the northern slope of
Province, Luzon, Philippines. Asia Life Sciences 10(2): 119–157.
Gruezo W.S. 1998. Species alpha diversity of Pagbilao Grande Island vegetations, Quezon Prov-
ince, Philippines. Asia Life Sciences 7(1): 39–92.
Hall J.B. and Swaine M.D. 1981. Distribution and Ecology of Vascular Plants in a Tropical Rain
Forest: Forest Vegetation in Ghana. Dr. W. Junk Publishers, The Hague.
Hamann A., Barbon E.B., Curio E. and Madulid D.A. 1999. A botanical inventory of a submontane
tropical rainforest on Negros Island, Philippines. Biodiversity and Conservation 8: 1017–1031.
Heaney L. and Regalado J.C. Jr. 1998. Vanishing Treasures of the Philippine Rain Forest: A
Comprehensive Introduction to the Biodiversity of the Philippines. The Field Museum Uni-
versity of Chicago Press, Chicago, 88 pp.

[238]
 1299

Henderson M.R. 1974a. Malayan Wild Flowers: Monocotyledons, reprint of the 1954 edition. The
Malayan Nature Society, Kuala Lumpur, Malaysia, 357+27 pp.
Henderson M.R. 1974b. Malayan Wild Flowers: Dicotyledons, Reprint. The Malayan Nature
Society, Kuala Lumpur, Malaysia, 478 pp.
Ingle N. 2003. Seed dispersal by wind, birds, and bats between Philippine montane rainforest and
successional vegetation. Oecologia 134: 251–261.
IUCN 2000. Red List – Philippines. www.iucnredlist.org/search-basic.html. Accessed 14.01.2004.
Kartawinata K. 1990. A review of natural vegetation studies in Malesia, with special reference to
Indonesia. In: Baas P., Kalkmann K. and Geesink R. (eds), The Plant Diversity of Malesia,
Proceedings of the Flora Malesiana Symposium, Leiden, the Netherlands, August 1989. Kluwer
Academic Publishers, Dordrecht-Boston-London, pp. 120–132.
Keller R. 1996. Identification of Tropical Woody Plants in the Absence of Flowers and Fruits.
Birkhäuser Verlag, Basel, Boston, Berlin, 229 pp.
Keng H. 1983. Orders and Families of Malayan Seed Plants, 3rd edn. Singapore University Press,
National University of Singapore, 441 pp.
Keng H. 1990. The Concise Flora of Singapore: Gymnosperms and Dicotyledons. Singapore
University Press, National University of Singapore, Singapore, 222 pp.
Kent M. and Coker P. 1992. Vegetation Description and Analysis. Belhaven Press, London, 363
pp.
Kiew R. 1987. The herbaceous flora of Ulu Endau, Johore-Pahang, Malaysia, including taxonomic
notes and descriptions of new species. Malayan Nature Journal 41(2 and 3): 201–234.
Kintanar R.L. 1984. Climate of the Philippines. PAGASA, Quezon City, Philippines, September
1984, 38 pp.
Kochummen K.M., LaFrankie J.V. and Manokaran N. 1992. Floristic composition of Pasoh
Forest Reserve, a lowland rainforest in peninsular Malaysia. In: Kheong Y.S. and Win L.S.
(eds), In Harmony with Nature, Proceedings of the International Conference on Conservation of
Tropical Biodiversity, Kuala Lumpur, Malaysia, 12–16 June 1990. Malayan Nature Society,
Kuala Lumpur, Malaysia, pp. 545–554.
Lieberman D., Lieberman M., Peralta R. and Hartshorn G.S. 1996. Tropical forest structure and
composition on a large-scale altitudinal gradient in Costa Rica. Journal of Ecology 84: 137–152.
Magurran A.E. 1988. Ecological Diversity and Its Measurement. Chapman and Hall, London, 179
pp.
Manokaran N. and Kochummen K.M. 1990. A re-examination of data on structure and floristic
composition of hill and lowland dipterocarp forest in peninsular Malaysia. Malayan Nature
Journal 44: 61–75.
Meijer W. 1959. Plantsociological analysis of montane rainforest near Tjibodas, West Java. Acta
Botanica Neerlandica 8: 277–291.
Merrill E.D. 1912. A Flora of Manila. Department of the Interior, Bureau of Science, Manila, 491
pp.
Myers N., Mittermeier R.A., Mittermeier C.G., da Fonseca G.A.B. and Kent J. 2000. Biodiversity
hotspots for conservation priorities. Nature 403: 853–858.
Newbery D.McC., Campbell E.J.F., Proctor J. and Still M.J. 1996. Primary lowland dipterocarp
forest at Danum Valley, Sabah, Malaysia. Species composition and patterns in the understorey.
Vegetatio 122: 193–220.
Ng F.S.P. (ed.) 1978. Tree Flora of Malaya Vol. 3. Forest Department, Ministry of Agriculture and
Lands, Malaysia, Longman Singapore, 339 pp.
Ng F.S.P. (ed.) 1989. Tree Flora of Malaya Vol. 4. Forest Department, Ministry of Agriculture and
Lands, Malaysia, Longman Malaysia, 548 pp.
Ng F.S.P. 1991. Manual of Forest Fruits, Seeds and Seedlings. Vols. I and II. Malayan Forest
Record No. 34. Forest Research Institute Malaysia, Kepong, 52109 Kuala Lumpur, Malaysia,
997 pp.
NSCB RU-8 2001. Provincial/City statwatch: Regional Unit VIII – Eastern Visayas. National
Statistical Coordination Board, www.nscb.gov.ph. Accessed 20.11.2003.

[239]
1300

Pancho J.V. 1983. Vascular Flora of Mount Makiling and Vicinity, Part 1. Kalikasan, The Phil-
ippine Journal of Biology, Supplement No. 1. New Mercury Printing Press, Quezon City,
Philippines, 476 pp.
Pipoly J.J.III and Madulid D.A. 1998. Composition, structure and species richness of a submon-
tane moist forest on Mt. Kinasalapi, Mindanao, Philippines. In: Dallmeier F. and Comiskey J.A.
(eds), Forest Biodiversity Research, Monitoring and Modeling: Conceptual Background and Old
World Case Studies. UNESCO and The Parthenon Publishing Group Limited, Carnforth, UK,
UNESCO, France, pp. 591–600.
Poulsen A.D. 1996. Species richness and density of ground herbs within a plot of lowland rainforest
in north-west Borneo. Journal of Tropical Ecology 12: 177–190.
Poulsen A.D. and Balslev H. 1991. Abundance and cover of ground herbs in an Amazonian rain
forest. Journal of Vegetation Science 2: 315–322.
Proctor J., Argent G.C. and Madulid D.A. 1998. Forests of the ultramufic Mt. Giting-Giting,
Sibuyan Island, the Philippines. Edinburgh Journal of Botany 55(2): 295–316.
Raunkiaer C. 1934. The Life-forms of Plants and Statistical Plant Geography. Oxford University
Press, Oxford.
Rennolls K. and Laumonier Y. 2000. Species diversity structure analysis at two sites in the tropical
rain forest of Sumatra. Journal of Tropical Ecology 16: 253–270.
Richards P.W. 1996. The Tropical Rain Forest: An Ecological Study. Cambridge University Press,
Great Britain, 575 pp.
Richter M. 1997. Allgemeine Pflanzengeographie. B.G. Teubner, Stuttgart, 256 pp.
Santos J.V., de Guzman E.D. and Fernando E.S. 1986. Guide to Philippine Flora and Fauna. Vol. IV:
Bamboos, Grasses, Palms. Natural Resources Management Center, Ministry of Natural Resources
and University of the Philippines, JMC Press, Incorporated, Quezon City, Philippines, 256 pp.
Schulze M.D. and Whitacre D.F. 1999. A classification and ordination of the tree community of
Tikal National Park, Peten, Guatemala. Bulletin of the Florida Museum of Natural History
41(3): 169–297.
Sist P. and Saridan A. 1998. Description of the primary lowland forest of Berau. In: Bertault J.-G.
and Kadir K. (eds), Silviculture research in a lowland mixed dipterocarp forest of East Kali-
mantan, the contribution of STREK project. CIRAD-Forêt Publication, pp. 51–93.
Slik J.W.F., Poulsen A.D., Ashton P.S., Cannon C.H., Eichhorn K.A.O., Kartawinata K.,
Lanniari I., Nagamasu H., Nakagawa M., van Nieuwstadt M.G.L., Payne J., Purwaningsih,
Saridan A., Sidiyasa K., Verburg R.W., Webb C.O. and Wilkie P. 2003. A floristic analysis of the
lowland dipterocarp forests of Borneo. Journal of Biogeography 30: 1517–1531.
Smith R.F. 1970. The vegetation structure of a Puerto Rican rain forest before and after short-term
irradiation. In: Odum H. (ed.), A Tropical Rain Forest. Oak Ridge, Atomic Energy Commission,
USA, pp. D103–140.
Soepdadmo E. 1995. Plant diversity of the Malesian tropical rainforest and its phytogeographical
and economic significance. In: Primack R.B. and Lovejoy T.E. (eds), Ecology, Conservation, and
Management of Southeast Asian Rainforests. Yale University, pp. 19–40.
Soepadmo E. and Wong K.M. (eds) 1995. Tree Flora of Sabah and Sarawak. Vol. 1. Forest
Research Institute Malaysia, Sabah Forestry Department, and Sarawak Forestry Department,
Ampang Press Sdn. Bhd., Kuala Lumpur, Malaysia, 513 pp.
Soepadmo E., Wong K.M. and Saw L.G. (eds) 1996. Tree Flora of Sabah and Sarawak. Vol. 2.
Forest Research Institute Malaysia, Sabah Forestry Department, and Sarawak Forestry
Department. Ampang Press Sdn. Bhd., Kuala Lumpur, Malaysia, 443 pp.
Soerianegara I. and Lemmens R.H.M.J. (eds) 1994. Plant Resources of South-East Asia No. 5(1):
Timber Trees: Major Commercial Timbers. Prosea Foundation, Bogor, Indonesia and Pudoc-
DLO, Wageningen, the Netherlands, 610 pp.
Takeuchi M. 1960. A estrutura da vegetaçao na Amazonia III: A mata da campina na regiâo do
Rio Negro. Boletim do Museo Paraense Emilio Goeldi 8: 1–13.
Tan B.C. and Rojo J.P. 1989. The Philippines. In: Campbell D.G. and Hammond H.D. (eds),
Floristic Inventory of Tropical Countries – The Status of Plant Systematics, Collections, and

[240]
 1301

Vegetation, Plus Recommendations for the Future. New York Botanical Garden, New York, pp.
44–62.
Turner I.M. 1994. The taxonomy and ecology of the vascular plant flora of Singapore: a statistical
analysis. Bot. J. Linn. Soc. 114: 215–227.
Valencia R., Balslev H. and Paz Y. Miño C.G. 1994. High tree alpha-diversity in Amazonian
Ecuador. Biodiversity and Conservation 3: 21–28.
van Balgooy M.M.J. 1997a. Malesian Seed Plants. Vol. 1: Spot-characters. Rijksherbarium /Hortus
Botanicus, Leiden, 154 pp.
van Balgooy M.M.J. 1997b. Malesian Seed Plants. Vol. 2: Portraits of Tree Families. Rijksher-
barium /Hortus Botanicus, Leiden, 307 pp.
van Steenis C.G.G.J. (ed.) 1950-ongoing. Flora Malesiana. Martinus Nijhoff Publishers, Dordr-
echt, the Netherlands.
Walsh R.P.D. 1996. Climate. In: Richards P.W. (ed.), The Tropical Rainforest. Cambridge Uni-
versity Press, pp.159–205.
Whitford H.N. 1906. Vegetation of the Lamao Forest Reserve. Philippine Journal of Science I:
373–431 and 637–682.
Whitford H.N. 1911. The Forests of the Philippines. Part I: Forest Types and Products. Depart-
ment of the Interior Bureau of Forestry, Bulletin No. 10, Manila, 94 pp.
Whitmore T.C. (ed.) 1983a. Tree Flora of Malaya. Vol. 1. Corrected Reprint of the 1972 Edition.
Forest Department, Ministry of Agriculture and lands, Malaysia Longman Malaysia, 473 pp.
Whitmore T.C. (ed.) 1983b. Tree Flora of Malaya. Vol. 2. Corrected Reprint of the 1972 Edition.
Forest Department, Ministry of Primary Industries, Malaysia, Longman Malaysia, 442 pp.
Whitmore T.C. 1985. Tropical Rain Forests of the Far East. Corrected Reprint of the 1984 Edition.
Clarendon Press, Oxford, 352 pp.
Whitmore T.C., Peralta R. and Brown K. 1985. Total species count in a Costa Rican tropical rain
forest. Journal of Tropical Ecology 1: 375–378.
Wilkie P., Argent G., Cambell E. and Saridan A. 2004. The diversity of 15 ha of lowland mixed
dipterocarp forest, Central Kalimantan. Biodiversity and Conservation 13(4): 695–708.
Zamora P.M. and Co L. 1986. Guide to Philippine Flora and Fauna. Vol. II: Economic Ferns,
Endemic Ferns, Gymnosperms. Natural Resources Management, Center Ministry of Natural
Resources, and University of the Philippines, JMC Press, Incorporated, Quezon City, Philip-
pines, 273 pp.
Zikeli S. 1998. Nutrient status and nutrient cycles of the tropical rainforest, Mt. Pangasugan, Leyte,
Philippines. Diploma thesis, Martin - Luther - University Halle - Wittenberg, Landwirtschaf-
tliche Fakultaet, Institut für Bodenkunde und Pflanzenernaehrung, 100 pp.

[241]
Biodiversity and Conservation (2006) 15:1303–1318
Springer 2006
DOI 10.1007/s10531-005-3873-7

-1

Status and conservation of Trigonobalanus

doichangensis (Fagaceae)#

WEIBANG SUN1,2,*,YUAN ZHOU1, CHUNYUAN HAN1, CHUNXIA

ZENG3, XIAODONG SHI3, QIBAI XIANG2 and ALLEN COOMBES4
1
Kunming Institute of Botany, the Chinese Academy of Sciences, Kunming 650204, the People’s
Republic of China (post address); 2Nanjing Forestry University, Nanjing 210037, the People’s
Republic of China; 3Qujian Normal University, Yunnan, the People’s Republic of China; 4Sir Harold
Hillier Gardens, Jermyns Lane, Ampfield, Romsey, Hampshire SO51 0QA, UK; *Author for corre-
spondence (e-mail: [email protected]; phone: +86-871-5223622; fax: +86-871-5216350)

Received 20 January 2004; accepted in revised form 8 March 2005

Key words: Conservation, Endangered, Genetic diversity, Habitat degradation, Reproductive

barriers, Trigonobalanus doichangensis (A.Camus) Forman

Abstract. Trigonobalanus doichangensis is a national rare and endangered plant of China. It is

restricted to 4 sites in southwest Yunnan, China and 1 site in Chiang-Rai, northern Thailand.
Investigations revealed that 4 community types are currently extant in Yunnan: isolated individuals,
sprouting woods, mono-dominant forest and co-dominant forest. The habitats have been severely
damaged and the populations there are facing a high risk of extinction. The adult phase of
T. doichangensis is reached when a tree attains a height of about 4 m. The flowering and fruiting
time varies slightly among populations and/or across the micro-habitats. In comparison with other
fagaceous plants in the community, T. doichangensis has an inverse flowering and fruiting period
from October through to May. Although microspore genesis and the development of male gametes
are normal, the pollen does not germinate until at least 8 h culture. The highest germination rate
was 37.8%. In addition, only 9.8% of the nuts contain well-developed seeds. Genetic variation
analyzed with random amplified polymorphic DNA indicated that its total genetic diversity was
0.1600 and genetic diversity within populations was 0.0749, the coefficient of gene differentiation
was 0.5320. Therefore, T. doichangensis has high genetic differentiation, a low level of genetic
diversity and a poor gene flow compared to the other fagaceous species. It seems that habitat
degradation, over exploitation and reproductive barriers, are most likely to be the factors threat-
ening the species. However, it may be a combination of geographical catastrophic events, habitat
deterioration, declining genetic diversity and physiological stress. Therefore, a practical conser-
vation strategy for T. doichangensis is urgently needed.

Introduction

The fagaceous genus Trigonobalanus Forman shows many ancestral features of

the family (Forman 1964; Crepet and Nixon 1989; Nixon and Crepet 1989) and
was probably widely distributed (perhaps over the whole northern hemisphere)

#
Supported by the important directional item of the Chinese Academy of Sciences (KSCX2-SW-
104)

[243]
1304

during the tertiary period or even before (Zhou 1992). The genus has only three
extant species and Trigonobalanus doichangensis is the only one distributed in
China (Hsu et al. 1981). T. doichangensis is most closely related to the genus
Quercus L. (Zhou 1992) and has a high scientific value for studies on Fagaceae
phylogeny, continental drift theory and global environmental changes.
Due to heavy exploitation for fuel wood, vegetation destruction in China
and its restricted distribution, T. doichangensis is seriously threatened and is
being pushed to the verge of extinction (Fu 1992). Hence, it was listed as a
national rare and endangered plant in 1984 and has also been proposed as a
second-ranked plant for national protection in China (Anon 1999). So far, its
phylogeny (Wang and Zhang 1988; Nixon and Crepet 1989; Wu and Xiao
1989; Liao et al. 1998; Wang et al. 1998) and the community floristic elements
(Li 1994) have been comprehensively studied. However, there are only a few
recent reports on its conservation biology (Zhou et al. 2003; Zhou, 2003; Sun
et al. 2004).
It is well known that a practical conservation strategy for an endangered
plant is based on an understanding of the threats it faces. In this study we
describe the status of the current distribution, population ecology, reproduc-
tive biology and genetic variation of T. doichangensis. Our goal was to provide
comprehensive information relevant to the conservation and restoration of this
species.

Methods

Trigonobalanus doichangensis is restricted to Thailand and south Yunnan in

southwest China (Fu 1992). In this study all localities from recorded herbarium
specimens, both from Thailand and China were used to determine its current
distribution (including collection of vouchers and research materials). Based on
this, the 4 known localities of T. doichangensis in Yunnan, China, Menglian,
Lancang, Ximeng and Cangyuan, were regarded as 4 populations on which to
carry out the comprehensive study. These 4 populations are situated in areas
bordered by southeast Myanmar, within the southern part of the Tropic of
Cancer. The climate belongs to the south subtropics type (EBYV 1987) and the
soil is mainly red acid (Xiong 1987). Dai, Wa and Laku are the three indige-
nous ethnic groups in the region.

Threat category and the characteristics of population ecology

Data analysis from the general surveys and ecological plotting studies, deter-
mined the threat category by assessment of current distribution and population
size following to the IUCN Red List Categories and Criteria (IUCN 2001).
Investigations were performed in the flowering and fruiting season. Plot sizes of
10 · 20 m, 20 · 20 m, 30 · 30 m and 50 · 60 m were applied in accordance with

[244]
 1305

the population size. The characteristics recorded were: (1) Site conditions, (2)
Plant height and crown size (m·m), trunk base diameter and diameter at breast
height of all T. doichangensis in the plot, (3) Habitat quality, (4) The main
accompanying higher plants and their abundance, and (5) Regeneration
capacity of T. doichangensis from seeds.

Analysis of genetic diversity

The random amplified polymorphic DNA (RAPD) technique was applied in

this study. Each locality was taken as a population, which, including that in
Thailand, made a total of 5 populations. Young but fully expanded leaves of 19
or 20 individuals within each population were selected and harvested, dried
quickly with silica gel and stored in a refrigerator at 4 C.

DNA extraction
Total DNA was extracted from dried leaves using a modified CTAB method
(Doyle and Doyle 1987). Modifications included incubation in CTAB at 65 C
for 1.5 h; a second extraction in chloroform–isoamyl alcohol; repeated wash-
ings in 70% ethanol without ammonium acetate; a final washing in 100%
ethanol and the addition of RNase to the TE buffer. DNA was visualized on
1.0% agarose gels.

RAPD PCR amplification

A mixed DNA from 5 randomly selected individuals was used to screen 137
primers (Operon Technologies, Alameda, CA, USA). The primers that gen-
erated clear bands with high polymorphism were determined for PCR ampli-
fication on a T3 thermocycler (Biomotra, Goettingen, Germany) using a final
volume of 20 ll. The reaction mix consisted of 2 ng of template DNA,
2.0 mmol/L MgCl2, 0.5 lmol/l dNTP, 10·buffer, 2.5 lmol/L of primer and 2
units of Taq DNA polymerase. The reaction included 4 min of initial dena-
turation at 94 C, followed by 45 cycles of 15 s of denaturation at 94 C, 45 s
of annealing at 36 C, 90 s of extension at 72 C, and a final extension step of
4 min at 72 C. Subsequently 10 ll amplification products were visualized on
1.5% agarose gels after electrophoresis in TAE at pH 8.0 for 3 h. The gels were
photographed with a GekDoc2000 (BIORAD) image analyzer (BIORAD,
Hercules, CA, USA).

Data analysis
The individuals were scored for the presence (1) or absence (0) of amplified
bands. The data matrix of RAPD bands was statistically analyzed using
POPGENE (Yeh et al. 1997). The parameters of the genetic diversity and
genetic structure are the percentage of polymorphic loci P(%), Shannon’s in-
dex of diversity (I), Nei’s gene diversity (h), the coefficient of gene differen-
tiation (Gst) and an estimate of gene flow (Nm).

[245]
1306

Reproduction biology

Flowering and fruiting time of the 4 T. doichangensis populations in China were

observed and compared. The floral organs at different developmental phases
were fixed with FAA (50% ethanol/acetic acid/formaldehyde = 90/6/4 by
volume) and preserved. The conventional paraffin slice thickness was 5–7 lm.
These were stained with iron alum–haematoxylin, dyed with chrysoidine G,
then observed and photographed using an Olympus BX-51 optical microscope.
Various levels of agar, sucrose, calcium nitrate and boric acid were selected
and combined for pollen germination tests. Fresh pollen collected from each
population was germinated at 25±1 C under darkness. The germination
process was observed every 2 h and the final germination rate (%) was mea-
sured at hour 24.
Seeds (nuts) were harvested between April and June from each population
and stored in a refrigerator at 4 C. Seed structure and embryo development
were observed and recorded under the ZS-PT Olympus anatomical lens. Seed
germination was tested in the LRH-250-GS climatic box made in China. A
conventional seed sowing was also conducted under Kunming climatic con-
ditions in Spring.

Results

Current distribution and threat status

When it was originally described, T. doichangensis was only known from

northern Thailand (Forman 1964). Later it was also found to occur in China
(Hsu et al. 1981). Since it was recorded in China some field surveys and
specimen identifications showed that it occurs in Menglian, Lancang, Ximeng
and Mengla,Yunnan, China (Fu 1992). Recent investigations revealed that
T. doichangensis still exists in fragmented populations of various sizes in
Menglian (Vouchers: SWB 02T001 020), Lancang (Vouchers:
SWB02T021 040), Ximeng (Vouchers: SWB02T041 060) and Chiang-Rai
Province in northern Thailand (Vouchers: SWB02T081 100). In addition, the
new localities in Cangyuan (Vouchers: SWB02T061 080) were recorded, and it
was shown that the species does not occur in the previously recorded locality in
Mengla. Therefore, it can be concluded that the natural distribution of
T. doichangensis is in the south and southwest Yunnan, China and in northern
Thailand, and currently only 5 populations are extant.
Trigonobalanus doichangensis has been listed as one of the national second-
ranked protected plants in the No.4 Order of the Forest and Agricultural
Ministry of China (Anon 1999) because of its threatened status in the wild. It
has also been recognized as a seriously threatened species because of the
amount of vegetation destruction and cutting for fuel-wood (Fu 1992). A case
study from the township market in Lancang, showed that the annual trading of

[246]
 1307

fuel-wood of T. doichangensis was about 1000–1400 m3 . Also a natural stand

close to the township is still cut continuously (Sun et al. 2004). At present the
indigenous ethnic people of Dai, Wa and Laku are still causing great
destruction to the habitats and using the wood resources, and as a result the
populations are fragile and fragmented. T. doichangensis in China is facing a
high risk of extinction.

Attributes of the communities

The populations in Yunnan, China showed different characteristics in their

tree-age structure and floristic composition (Table 1). Four community types
were recognized during the recent investigations.
(1) Type IsI (Isolated individuals): Individuals of T. doichangensis were often
found in secondary woods, by roadsides, in mixed woods or farmland, and
occasionally within the evergreen broadleaf forests. The forests surrounding

Table 1. Population characteristics of Trigonobalanus doichangensis in different communities.

Communities SW MDF CDF

Localities Menglian Lancang Cangyuan

BTS RBS MDS

Plots
Altitude (m) 1020 1450 1550 1590 1730
Slope N, 30 degrees W, 40 degrees SE, 40 degrees S, 30 degrees S, 30 degrees
Coverage (%) >90 >90 >95 >95 >95
Size (m2) 800 400 3000 600 400
TD no. in the plot 195 140 41 101 118
TD no. per 100 m2 24 35 2 17 30
TD Avg. height (m) 5.1 5.6 17.9 6.8 14.4
Hst/Shst of TD (m) 13.5/0.2 9.5/0.3 35/0.4 32/0.28 20/0.2
% of TD ‡20 m / / 51 6 0.8
% of TD ‡4–20 m 59 73 44 55 92
% of TD 2–3 m 15 21 / 23 5
% of TD £ 1.0 m 26 6 5 16 3
% of dead TD / / / 17 15
Avg. height of DP (m) / / / 1.5 10 3.5 18
Accompanier no. 45 30 25 27 25
Main companions (TP) 1 2 3 2 3 4 5 4 5 4 5

Evaluation HAD PR RSS USD MDSD

Notes: SW = Sprouting wood; MDF = Mono-dominant forest; CDF = Co-dominant forest;
HTS = Huge-tree structure; RBS = Relatively balance structure; MDS = Mono-dominant
structure; No. = numbers; TD = Trigonobalanus doichangensis; Hst = Highest; Shst =
Shortest; Avg. = Average; DP = dead plants; TP = top layer; 1 = Vaccinium bracteatum;
2 = Castanopsis echinocarpa; 3 = Castanopsis hytrix; 4 = Castanopsis calathiformis; 5 = Schi-

ma wallichii; HAD = heavily disturbed by human activities; PR = Population in recovery;

RSS = relatively stable structure; USD = unstable structure in developing; MDSD = mono-dom-
inant structure in developing.

[247]
1308

these individuals were mostly replaced by other native plant species, alien plant
invaders, such as Eupatorium adenophorum (Ageratina adenophora) and Eup-
atorium odoratum (Chromolaena odorata), and agricultural crops. Occasionally,
some seedlings or young trees appeared around scattered individuals inside the
evergreen broadleaf forests. However, other plants such as Castanopsis ca-
lathiformis, Castanopsis echinocarpa, Lithocarpus fenestratus, Schima wallichii
and Anneslea fragrans had already dominated in the vegetation. The formation
of Type IsI is due to heavy cutting and vegetation destruction. The isolated
individuals are found in the most endangered habitat and their genetic diversity
is easily reduced by further human activities, grazing animals, farming and
further biotic invasion.
(2) Type SW (Sprouting woods): Type SW is the result of fuel wood cutting by
indigenous people. Investigations show that the indigenous ethnic groups of
Dai, Wa and Laku are familiar with T. doichangensis and have realized that the
tree can sprout easily after top-cutting and thinning and thus they have adopted
the methods of ‘alternate cutting or thinning cutting’ for the primitive sustain-
able use of the tree as fuel wood. As a result of these practices, plants of
T. doichangensis in these woods showed some unique characteristics in tree
shape, tree height structure and associated floristic composition (Table 1). In
this community the tallest T. doichangensis was about 13 m and the average
height was around 5 m. Some 60% of T. doichangensis reached the reproductive
phase and about 25.6% of the individuals were below 1 m in height. T. doi-
changensis was the dominant species in woods, and 50 species of accompanying
higher plants were present. The most important trees were Vaccinum bractea-
tum, Castanopsis hystrix, Lithocarpus fenestratus, Craibiodendron stellatum,
Anneslea fragrans, Ternstroemia gymnanthera and Schima wallichii. Vernonia
parishii, Arthraxon lanceolatus, Carex baccans and Zingiber striolatum were the
typical herbaceous plants. Among epiphytes, Phymatodes lucida, Vanda coerulea
and Eria pannea were also commonly found on the trunk. As ontogenesis of
T. doichangensis was prevented by cutting, most trees could not complete their
natural growth and their reproductive capacity was relatively restrained.
(3) Type MDF (Mono-dominant forest): Plants of T. doichangensis in type
MDF are found as small mono-dominated patches scattered in the secondary
evergreen broadleaf forest. The tallest plant of T. doichangensis was about
10 m and the average height was around 5–6 m. Approximately 70% of the
trees in the plots were mature with a height of ‡4 m. Around 20% of the plants
were 2–3 m in height, while seedlings and young trees represented only 6% of
the total population. Accompanying higher plants were represented by some 30
species and most of the woody species were the same as in Type SW, but there
were far fewer herbaceous plants and epiphytes. Osyris wightiana, Viburnum
cylindricum, Broussonetia papyrifera and Phyllanthus emblica were present in
the community. T. doichangensis in Type MDF showed vigorous growth and a
strong regenerative ability.
(4) Type CDF (Co-dominant forest): Plants of T. doichangensis in Type CDF
often formed a mosaic of mono-dominant patches in the primitive evergreen

[248]
 1309

broadleaf forest. In this type of community, T. doichangensis grows naturally

without destructive disturbance from human activities. Based on the tree
height-grade Type CDF can be divided into 3 ranks. These are Big Tree
Structure (BTS), Relatively balanced structure (RBS) and Mono-dominant
structure of mature trees (MDS) (Table 1). In rank BTS over 50% of T. doi-
changensis were trees with a height of ‡20 m and approximately 90% of the
plants were in the reproductive phase. Compared with the other 2 ranks BTS
had a lower percentage of young trees and seedlings. In rank RBS only 6% of
T. doichangensis were trees with a height of ‡20 m, while some 17% of the total
were dead, and almost 60% of the plants were in the flowering and fruiting
stage. Rank RBS had almost equal number of flowering and non-flowering
individuals of T. doichangensis. Compared with the ranks of BTS and RBS,
rank MDS had the highest number of flowering individuals and nearly 90% of
them were in the reproductive phase. The percentages of trees over 20 m tall or
less than 1 m in the MDS were the lowest, and around 15% of all trees with a
height of 3–18 m were dead. Both ranks of RBS and MDS had more indi-
viduals per 100 m2, however, both of them also had a certain percentage of
plants that had died naturally. It can be inferred that T. doichangensis in the co-
dominant forest is dynamic, and both types of RBS and MDS would develop
into BTS with some individuals dying gradually because of strong competition
for resources. The interrelationship of MDS M RBS M MDS may be the main
characteristic for T. doichangensis in the primitive evergreen broadleaf forest.
About 25 species of accompanying higher plant were found in CDF. Of these,
Castanopsis calathiformis, Lithocarpus echinotholus and Lithocarpus fenestra-
tus, were the most common in the upper canopy and Castanopsis calathiformis
was the most competitive species with T. doichangensis. Other woody plants,
such as Olea rosea, Eriobotrya cavaleriei, Cinnamomum bejolghota, Phoebe
macrophylla, Helicia nilagirica, Ternstroemia gymnanthera, Pyrenaria dios-
pyricarpa, Pithecellobium clypearia, and Eurya groffii, were also commonly
found.

Genetic diversity and population genetic structure

One hundred and fifty seven bands from 100 to 2900 bp were generated by the
16 selected primers, with each primer producing 6–13 bands with an average of
9.81 bands. Eighty three of the total bands were polymorphic and accounted
for 52.87% (Table 2). These findings show that genetic diversity at the species
level was abundant. Table 2 also indicates that the percentages of polymorphic
bands in both populations of Lancang and Chiang-Rrai were far lower than
those in the 3 other populations of Cangyuan, Menglian and Ximeng.
Therefore, the genetic variation within both populations of Lancang and
Chiangrai was lower that that in others.
Table 3 shows that the effective number of alleles (ne), Shannon’ s index of
diversity (I) and Nei’ s gene diversity (h) were 1.2646, 0.2431 and 0.1595,

[249]
1310

Table 2. The sequences of random oligonucleotide primers and the number of polymorphic bands.

Primers Sequences Total bands Polymorphic bands

OPC05 TCGTCTGCCC 9 4
OPC14 TGCGTGCTTG 10 5
OPC15 GACGGATCAG 11 6
OPJ 09 TGAGCCTCAC 11 6
OPM02 ACAACGCCTC 6 0
OPM06 CTGGGCAACT 8 5
OPM13 GGTGGTCAAG 10 4
OPM15 GACCTACCAC 12 5
OPM16 GTAACCAGCC 13 7
OPM20 AGGTCTTGGG 10 4
OPN14 TCGTGCGGGT 12 9
OPN20 GGTGCTCCGT 7 3
OPS03 CAGAGGTCCC 11 6
OPV06 ACGCCCAGGT 10 9
OPV10 GGACCTGCTG 9 4
OPV15 CAGTGCCGGT 8 6

Table 3. The genetic diversity and genetic structure of T. doichangensis.

Population na ne H I P/% Ht Hs Gst Nm

LC 1.1019 1.0538 0.0311 0.0472 10.19

CHR 1.1911 1.0804 0.0513 0.0812 19.11
CY 1.3185 1.1675 0.0985 0.1502 31.85
ML 1.2930 1.1469 0.0911 0.1403 29.30
XM 1.3312 1.1696 0.1024 0.1568 33.12
Mean 1.5287 1.2646 0.1595 0.2431 52.87 0.1600 0.0749 0.5320 0.4398
Notes: na, observed number of alleles; ne, effective number of alleles Kimura and Crow (1964); h,
Nei’s (1973) gene diversity; I, Shannon’s Information index; P, the percentage of polymorphic loci;
Ht, gene diversity of species; Hs, gene diversity within populations; Nm, gene flow; Gst, coefficient
of gene differentiation.

respectively, and the genetic variation of T. doichangensis is rather low (the

effective number of alleles<1.5000). As the species gene diversity (Ht) was
0.1600 and the gene diversity within populations (Hs) was 0.0749, the genetic
variation within populations was lower than that between populations. Table 3
also indicates that T. doichangensis shows a very strong genetic differentiation
between its populations because the coefficient of gene differentiation (Gst)
reached 0.5320, more than 10 times that of other fagaceous plants (0.020–
0.065) (Chen et al. 1997).
The analysis also indicated that T. doichangensis had a lower gene flow
(Nm = 0.4398) (Table 3) compared with other fagaceous plants, such as
Castanopsis fargesii (Nm = 11.2645) (Zhu et al. 2002), Cyclobalanopsis glauca
(Nm = 20.49) (Zhu et al. 2002) and Quercus aquifolioides (Nm = 2.66) (Li
et al. 1997). Thus, there is less gene movement between T. doichangensis
populations compared to other plants in Fagaceae.

[250]
 1311

Microspore genesis and development of male gametes

Some 400 floral organs of T. doichangensis were sampled at different devel-

opment stages from the 4 populations. The anther is 4-sporangiate and the
anther wall formation conforms to the dicotyledonous type. The tapetum is of
the glandular type and most cells are 2-nucleate. Cytokinesis at meiosis of
microspore mother cells is simultaneous and tetrads are tetrahedral, occa-
sionally decussate. Mature pollen grains are 2-celled. These characteristics are
consistent with the result of Stairs’s results on Quercus (Johri et al. 1992). The
meiosis process of microspore mother cell in T. doichangensis is the same as the
common angiosperms (Hu 1982).
These observations also showed that different anthers or different anther sacs
of the same anther in the meiosis phase of the microspore-mother-cell showed
asynchronism in T. doichangensis. Various authors (Whelan 1974; Li and Cao
1986; Huang et al. 2001) have given different opinions of this asynchronism.
However, it seems certain that the asynchronism of the meiosis phase in the
microspore-mother-cell is a universal phenomenon in angiosperms, and it does
not relate to the normal development of anthers and the formation of fertile
pollen. Therefore, the observations did not find any abortion or other
abnormal phenomena in anther wall development, microspore genesis and
male gamete formation in T. doichangensis.

Pollen germination in vitro and its abortion

Pollen germination started after 8 h of culture, and was completed at 24 h.

Different combinations of agar, sucrose, boric acid and calcium nitrate could
affect pollen germination, but the highest germination rate was only 37.8%
(Table 4) and more than 63% of the pollen of T. doichangensis was aborted.
Pollen germination has rarely been reported in Fagaceae, but some major
chestnut cultivars were tested (Xia et al. 1989). The pollen germination rate of
Castanea mollissima was lower (31%) than that of its cultivars (60–78%), and
the lower pollen germination of C. mollissima was one of the causes of its poor
fruiting (Xia et al. 1989). The cause of pollen abortion in T. doichangensis may
be complicated, and yet its lower pollen germination rate and the delayed
germination (8 h after culture) may contribute to the lower fruiting percentage
(9.8%) (Sun et al. 2004).

Ovule abortion and embryo development

In T. doichangensis the ovary has 3 locules with 2 ovules per locule. Obser-
vation on the ovary in transverse section showed that 6 ovules in each ovary
were generated at an early stage, and at the developmental anaphase 1 of 2
ovules per chamber were well developed, while another one was aborted.
Meanwhile, both axile and parietal placentas were observed, and thus

[251]
1312

Table 4. Pollen germination of T. doichangensis on different media.

Medium code Germination rate (%)

1 1.9
2 8.1
3 18.3
4 22.9
5 22.1
6 22.4
7 22.2
8 20.2
9 20.9
10 21.8
11 37.8
12 37.2
13 22.7
14 21.4
15 18.3
Notes: 1 agar 7; 2 agar 7 + sucrose 100; 3 agar 7 + sucrose 150; 4 agar 7 + sucrose 200;
5 agar 7 + sucrose 250; 6 agar 7 + sucrose 200 + boric acid 100; 7 agar 7 + sucrose

200 + boric acid 150; 8 agar 7 + sucrose 200 + boric acid 200; 9 agar 7 + sucrose 200 +
boric acid 300; 10 agar 7 + sucrose 200 + boric acid 400; 11 agar + sucrose 200 + calcium
nitrate 100; 12 agar 7 + sucrose 200 + calcium nitrate 200; 13 agar 7 + sucrose 200 +
calcium nitrate 300; 14 agar 7 + sucrose 200 + calcium nitrate 400; 15 agar 7 + sucrose
200 + calcium nitrate 500 (All units are g/L).

T. doichangensis may show a transition trend from axile to parietal placenta-

tion. The ovule is anatropous and is enclosed by outer and inner integuments.
The following 3 phenomena were also observed in the young fruits: (1) Only
1 of the 6 ovules per ovary developed into an embryo and the others were
aborted. The percentage of the one-ovule developed nuts was less than 10%;
(2) In some fruits all the 6 ovules in the ovary were aborted and the ovary was
lignified; (3) The ovules in some fruits seem to be developed but the embryos
were membranous. It can be concluded that considerable embryo abortion
occurs in T. doichangensis.

Flowering and fruiting

Investigations in the wild indicated that T. doichangensis becomes of repro-

ductive age once its height reaches 4 m. The flowers are unisexual and plants
are monoecious. Staminate flowers are often clustered on the rachis forming
axillary catkins or occur terminally on leafless branchlets. Pistillate inflores-
cences are unbranched in the axils of the upper (younger) leaves and bear 1–3
flowers on distal parts of rachis.
The pollination mechanism in Fagaceae is complex. Flower characters in
Quercus Subgen. Cyclobalanopsis show a transition from entomophily to
anemophily (Kaul and Abbe 1989; Zhou 1992), and the upright or curved
staminate inflorescence in genus Trigonobalanus may also represent such a

[252]
 1313

transition (Hsu et al. 1981). The staminate inflorescences of T. verticillata are

erect and often branched, and are reminiscent of inflorescences found in sub-
family Castaneoideae. And this, along with reports of odor in flowers (Soep-
admo 1972) suggests the possibility that T. verticillata is an entomophilous
species, as are the species of Castaneoideae such as Lithocarpus densiflorus,
Castanea sativa, C. mollissima etc (Nixon and Crepet 1989). T. doichangensis
and T. excelsa have lax staminate inflorescences like those of Quercus, which is
generally accepted as anemophilous (Nixon and Crepet 1989). However, our
observation on fresh male flowers of T. doichangensis under the stereo-ana-
tomical lens, found that tiny beetles frequently visited the anthers to collect
pollen. Therefore, we infer that these tiny beetles may also participate in the
pollination of T. doichangensis.
The flowering time of T. doichangensis varies slightly between different
populations and micro-habitats. However, the period of flowering and fruiting
is generally from October to the following May, a period in which the weather
is cool and wet (YWB 1983). This phenology is unique compared to most other
plants in the community, which flower in Spring and fruit in Autumn; a factor
that may also contribute to its low fruiting percentage.

Seeds and seed germination

The nut of T. doichangensis is normally called a seed and has a kilo-grain

weight of 12.7 g. In fact, the seed is exoendosperm enclosed by a semi-trans-
parent episperm and endocarp. T. doichangensis fruits well, but only 9–11% of
the nuts contain well-developed seeds (with fertile embryos).
Nuts containing well-developed seeds germinated 4 d after culturing at
various temperatures under darkness, and after 11 d all germination was
completed. Lighting slightly stimulated the germination and the highest ger-
mination rate was around 63–73%. Epigeal germination has been reported in
T. doichangensis (Nixon and Crepet 1989; Liao et al. 1998) and this was con-
firmed by sowing nuts in Spring at Kunming,Yunnan, China. Observations
indicated that hypocotyls grew rapidly at 8 d and at 11 d, 2 or occasionally 3
cotyledons emerged from the compost. Germination in compost lasted around
20 d and the germination rate was close to that found in tests in the LRH-250-
GS climatic box. Transplanted young plants have been growing well at Kun-
ming (where the recorded minimum absolute temperature is
5.4 C) for
3 years. And thus, T. doichangensis may have a broad climatic adaptability.

Discussion

Global threat category of T. doichangensis

Trigonobalanus doichangensis in China has been considered as an endangered

plant for more than 10 years (Fu 1992). However, it was ranked as DD (Data

[253]
1314

Deficient) in The World List of Threatened Trees (Oldfield et al. 1998) and its
threat category has not yet been evaluated by using any version of IUCN Red
List Categories and Criteria.
Based on our observations and data analyses, the threat category of
T. doichangensis in China is proposed as EN (Endangered) [EN, B1a,
B1b(i,ii,iii)] (IUCN 2001). It is therefore considered to be facing a very high
risk of extinction in the wild, because: (1) Its extent of occurrence is estimated
to be less than 5000 km2, (2) Its populations are severely fragmented and (3)
There is a continuing decline in the extent of occurrence, area of occupancy
and quality of habitat. Judging by the limited distribution of the species in
northern Thailand, and its only postulated occurrence in Myanmar, it is likely
that EN will also be the Global threat category.

Identifying threats

Rarity and endangerment of a plant species may be due to intrinsic (related to

the biology of the species) or extrinsic (environmental) factors (Rabinowitz
1981; Fiedler and Ahouse 1992) and anthropogenic habitat fragmentation has
been widely cited as a major threat to biodiversity (Simberloff 1988).
Destruction and disappearance of habitat are the major extrinsic factors
threatening biodiversity (Pang et al. 2003; Jiang et al. 1997). Some remnant
plant species, such as Metasequoia glyptostroboides, Cathaya argyrophylla and
Liriodendron chinense, may maintain a present population in the ‘endangered
habitats’ because large parts of their populations have disappeared due to
historical events such as climate change (He et al. 1996). As a remnant species
T. doichangensis may also be in an ‘endangered habitat’. Its present habitat
destruction and disappearance, caused by expansion of agricultural land and
over cutting, are threatening its survival. Therefore, a very low gene flow
among populations in T. doichangensis may certainly be due to its habitat
degeneration and population isolation.
Reproductive barriers are among the main factors to threaten a plant species
(Pan et al. 2003). T. doichangensis does not show any abnormal phenomena in
anther wall development, microspore genesis and male gamete formation. It did,
however, show serious ovule development abortion, the causes of which need to
be studied. The low pollen germination, as well as the length of time for pollen to
germinate and the inverse flowering and fruiting behaviour compared to other
fagaceous plants in the population, show that T. doichangensis exhibits a
reproductive barrier in seed development. In addition, its seed (nut) germination
and seedling establishment need a moist environment and the establishment of
young trees from seedlings needs comparatively sunny conditions (Sun et al.
2004). Evidently, the environment in the severely fragmented habitat is not ideal
for seed germination and plant development into the reproductive phase.
Some rare species that existed in glacial refuges, have long survived in certain
restricted habitats even though their populations are isolated and the habitats

[254]
 1315

are already fragmented (Li et al. 2003). However, long-term species survival
depends on the maintenance of sufficient genetic variability within and among
populations to accommodate new selection pressures brought about by envi-
ronmental changes (Elena et al. 2003), and populations with low genetic var-
iation have a high risk of extinction (Li et al. 2003; Ledig et al. 2002).The
decline of seed viability of Metasequoia glyptostroboides, is caused by its poor
genetic base and inbreeding, and its effective seed germination relies on suitable
conditions (Liao and Zhou 1989). Accordingly, lower levels of genetic diversity
and intense genetic differentiation in T. doichangensis will not be able to re-
spond as well to changes in abiotic or biotic environmental conditions as can
other fagaceous species. The low seed germination and plant establishment
may also partly contribute to its present genetic status caused by habitat
fragmentation and population isolation.
We may conclude that endangerment of T. doichangensis is not caused by a
single factor. It seems that habitat destruction, over-exploitation and repro-
duction barriers, are the most likely factors. However, it may be a combination
of historic factors such as climate changes, habitat degeneration, a poor genetic
base and physiological stress. Therefore, a practical conservation strategy for
T. doichangensis is urgently needed.

Conservation considerations

Protection and restoration of natural habitats is the best and cheapest method
of preserving the biological diversity and stability of global ecosystem (Lande
1988). Undoubtedly, the long-term survival of T. doichangensis is dependent on
habitat conservation. However, only 1 of the 4 extant populations of T. doi-
changensis in China has been legally preserved by government ownership and
others are facing a high risk of habitat disappearance. Perhaps more than most
endangered plant species in China, the unique characteristics of T. doichang-
ensis, population isolation and the complex community interaction, exemplify
the importance of habitat preservation and in-situ conservation. Nevertheless,
both in-situ and ex-situ measures are needed for preserving T. doichangensis
and its genetic diversity, and following aspects should be particularly consid-
ered:
(1) Preservation of the habitat and population of T. doichangensis in
Canyuan must be reinforced. The Canyuan population is the only 1 of the 4
populations in China which has been well protected inside a national natural
reserve. Therefore, the population is a vital resource for further research into
the origin and evolution, eco-biological characteristics and genetic realities of
T. doichangensis. Furthermore, the population also has a great value in pop-
ulation restoration research.
(2) As T. doichangensis populations in Menglian, Lancang and Ximeng are
still exposed to high habitat destruction and cutting for fuel wood, it is essential
that new in-situ conservation sites in these area should be urgently planned.

[255]
1316

These new sites will play an important role in habitat recovery and population
restoration.
(3) Although cross-planting is often controversial in plant conservation
planning, it is proposed that seedling cross-planting between different popu-
lations of T. doichangensis should be considered. At least, the potential impacts
of this measure should be studied to determine if low gene flow would be
enhanced and if overall diversity would be increased.
(4) As young T. doichangensis plants propagated from seeds can tolerate
temperatures below
2 C at Kunming (Sun et al. 2004), it may be cultivated
for fuel wood and as a landscaping plant in northern parts of the Tropic of
Cancer. However, the mixed planting of trees propagated from various pop-
ulations is essential.
(5) Ex-situ efforts need to be undertaken to preserve genetic diversity and
multiply specimens. This applies in particular to those populations outside of
the well-protected national nature reserve. Ideally, ex-situ sites will be close to
nature reserves or to botanical gardens. Propagation from seeds is preferred,
since it would be the least detrimental to the extant populations and would
include the widest range of genetic diversity. Optimally, seeds should be taken
from many individuals from each of the populations.

References

Anon 1999. List of National Key Protected Wild Plants (First Group). The Order of National
Forestry Bureau and Agriculture Ministry of China 4: 2–13.
Chen X.Y., Wang X.H. and Song Y.C. 1997. Genetic diversity and differentiation of Cycloba-
lanopsis glauca populations in east China. Acta Botanica Sinica 39: 149–155.
Crepet W.L. and Nixon K.C. 1989. Earliest megafossil evidence of Fagaceae: phylogenetic and
biogeographic implications. American Journal of Botany 76: 842–855.
Doyle J.J. and Doyle J.L. 1988. Isolation of plant DNA from fresh tissue. Focus 12: 13–15.
EBYV (Editorial Board of Yunnan Vegetation). 1987. Vegetation of Yunnan. Science Press,
Beijing, China.
Elena T., Jose M.I. and Cesar P. 2003. Genetic structure of an endangered plant, Antirrhinum
microphyllum (Scrophulariaceae): Allozyme and RAPD analysis. American Journal of Botany
90: 85–92.
Fiedler P.L. and Ahouse J.J. 1992. Hierarchies and cause: toward an understanding of rarity in
vascular plant species. In Fielder P.L. and Jain S.K. (eds), Conservation biology: the theory and
practice of nature conservation, preservation and management. Chapman and Hall, New York,
USA, pp. 23–47.
Forman L. 1964. Trigonobalanus: a new genus of Fagaceae. Kew Bulletin 17: 381–396.
Fu L.G. 1992. China Plant Red Data Book, Vol.1. Science Press, Beijing, China.
He S.A., Hao R.M. and Tang S.J. 1996. A study on the ecological factors of endangering mechanism
of Liriodendron chinensis (Hemsl.) Sarg. Journal of Plant Research and Environment 5: 1–8.
Hsu Y.C., Wang C.J, Wu C.Y. and Li H.W. 1981. Trigonobalanus Forman-A new recorded genus
of Fagaceae in China. Acta Botanica Yunnanica 3: 213–215.
Hu S.Y. 1982. Embryology of Angiosperm. Higher Education Press, Beijing, China.
Huang H.Y., Ma S.B. and Li L. 2001. The genesis of microspore and the formation of male
gametophyte in Dysosma versipellis. Bulletin of Botanical Research 21: 561–566.
IUCN 2001. IUCN Red List Categories and Criteria: Version 3.1. Information Press, Oxford, UK.

[256]
 1317

Jiang Z.G., Ma K.P. and Han X.G. 1997. Conservation Biology. Sciences and Technology Press,
Hangzhou, Zhejiang, China.
Johri B.M., Ambegaokar K.B. and Srivastava P.S. 1992. Comparative Embryology of Angio-
sperms, Vol. I. Springer-Verlag, Berlin, Heidelberg.
Kaul R.B. and Abbe E.C. 1989. Inflorescense architecture and evolution in the Fagaceae. Journal
of Arnold Arboretum 65: 375–401.
Kimura M. and Crow J.F. 1964. The number of alleles that can be maintained in a finite popu-
lation. Genetics 49: 725–738.
Lande R. 1988. Genetics and demography in biological conservation. Science 241: 1455–1460.
Ledig F.T., Hodgeskiss P.D. and Jocob-Cervantes V. 2002. Genetic diversity, mating system, and
conservation of a Mexican subalpine relict, Picea Mexicana Martinez. Conservation Genetics 3:
113–122.
Li T.Q. and Cao H.J. 1986. Microsporogenesis and development of male gametophyte of Camellia
chrysantha (Hu) Tuyama. Journal of Beijing Forestry University 8: 30–34.
Li J. 1994. A preliminary study on the floristic elements of the community of Formanodendron
doichangensis. Acta Botanica Yunnanica 16: 17–24.
Li J., Chen K.Y. and Li B.S. 1997. Preliminary study on genetic structure of alpine oaks in Tibet.
Journal of Beijing Forestry University 19: 93–98.
Li Z.Z., Gong J.J., Wang Y. and Huang H.W. 2003. Spatial structure of AFLP genetic diversity of
remnant populations of Metasequoia glyptostroboides (Taxodiaceae). Biodiversity Science 11: 265–275.
Liao H.M., Gou G.Q. and Ye N.G. 1998. A study on seedling morphology and anatomy and
systematic position of Trigonobalanus doichangensis Form. Journal of Wuhan Botanical Re-
search 16: 223–226.
Liao S.Z. and Zhou W.L. 1989. Seed germination experiment of Metasequoia glyptostroboides.
Sichuan Forestry Science and Technology 10: 71–74.
Nixon K.C. and Crepet W.L. 1989. Trigonobalanus (Fagaceae): Taxonomic status and phylloge-
netic relationships. American Journal of Botany 76: 828–841.
Oldfield S., Lusty C. and MacKinven A. 1998. The World List of Threatened Trees. World
Conservation Press, Cambridge, UK.
Pan Y.Z., Gong X. and Liang H.X. 2003. A study on the embryology of endangered plant
Manglietia aromatica. Journal of Wuhan Botanical Research 21: 1–8.
Pang X.A., Liu X., Liu H., Wu C., Wang J.Y., Yang S.X. and Wang Q.F. 2003. The geographic
distribution and habitat of the Isoetes plants in China. Biodiversity Science 11: 288–294.
Rabinowitz D. 1981. Seven forms of rarity. In: Synge H. (ed.), The Biological Aspects of Rare Plant
Conservation. Wiley, New York, USA, pp. 205–217.
Simberloff D. 1988. The contribution of population and community biology to conservation
sciences. Annual Review of Ecology and Systematics 19: 473–511.
Soepadmo E. 1972. Fagaceae. In: Van Steenis C.G.G.J. (ed.), Flora Malesiana, Ser.1, Vol.7, Pt.
2:265–403.
Sun W.B., Zhou Y., Zhao J.C. and Chen G. 2004. Current distribution,population attributes and
biological characters of Trigonobalanus doichangensis in relation to its conservation. Acta Eco-
logia Sinica 24: 352–348.
Wang P.L., Pu F.T. and Zheng Z.H. 1998. Palynological evidence for taxonomy of Trigonobalanus
(Fagaceae). Acta Phytotaxonomica Sinica 36: 238–241.
Wang P.L. and Zhang J.T. 1988. On the pollen morphology and systematic position of Trigono-
balanus doichangensis. Acta Phytotaxonomica Sinica 26: 44–46.
Whelan E.D.P. 1974. Discontinuities in the callose wall, inter-meiocyte connections, and cytomixis
in angiosperm meiocytes. Canadian Journal of Botany 53: 1219–1224.
Wu S.M. and Xiao S.Q. 1989. Comparative anatomical studies on the woods of Castanea Mill. and
Trigonobalanus Forman in China. Guihaia 22: 567–571.
Xia R.X., Ma M.T. and Mao Q.S. 1989. Observation of some features of staminate inflorescence
and pollen of the major chestnut cultivars in Luotian county. Journal of Huazhong Agricultural
University 8: 189–191.

[257]
1318

Xiong Y. 1987. China Soils. Science Press, Beijing, China.

YWB (Yunnan Weather Bureau). 1983. Yunnan Agricutural Weather Data Collection. Yunnan
People’s Press, Yunnan, China.
Zhou Y., Sun W.B. and Li C.R. 2003. Preliminary study on seed germination of Trigonobalanus
doichangensis. Journal of Wuhan Botanical Research 21: 73–76.
Zhou Y. 2003. Effect of different seed provenances on seed germination and seedling growth in
Trigonobalanus doichangensis. Seeds 130: 3–4.
Zhou Z.K. 1992. Origin, phylogeny and dispersal of Quercus from China. Acta Botanica
Yunnanica 14: 227–236.
Zhu Q.H., Pan H.X., Zhu G.Q.; Yin T.M., Zou H.Y. and Huang M.R. 2002. Analysis of genetic
structure of natural populations of Castanopsis fargesii by RAPDs. Acta Botanica Sinica 44:
1321–1326.

[258]
Biodiversity and Conservation (2006) 15:1319–1338
Springer 2006
DOI 10.1007/s10531-005-3875-5

-1

Ghyll woodlands of the Weald: characterisation

and conservation

NIALL G. BURNSIDE1,*, DAN J. METCALFE2, ROGER F. SMITH1

and STEVE WAITE3
1
Biogeography & Ecology Research Group (BERG), School of the Environment, University of
Brighton, Cockcroft Building, Moulsecoomb, Brighton BN2 4GJ, United Kingdom; 2CSIRO Tropical
Forest Research Centre, PO Box 780, Atherton, Qld 4883 Australia; 3Biogeography & Ecology
Research Group (BERG), Biology Division, University of Brighton, Cockcroft Building, Moulse-
coomb, Brighton BN2 4GJ, United Kingdom; *Author for correspondence (e-mail: N.G.Burnside@
brighton.ac.uk; fax: +44-0-1273-642285)

Received 24 May 2004; accepted in revised form 8 March 2005

Key words: Biodiversity conservation, Geographical information systems, Ghyll woodlands, The
Weald, Woodland characterisation

Abstract. Ghylls are linear valley features cut into the sandy beds of the Weald of south-eastern
England. The ghyll’s indigenous woodlands are highly species rich at the small scale, support
distinctive assemblages of cryptogamic plants, and are unique to south-east England. Field surveys
were carried out for 48 ghyll woodlands in the Weald with a GIS used to examine the ecology,
landform and conservation status of the ghyll woodlands. The data were analysed using spatial and
multi-variate techniques in order to identify sub-groups or ghyll woodland communities based
upon species composition, topography and geology. The ghylls are shown to be reasonably uniform
for canopy vegetation type and structure and for their geological and soil characteristics. However,
analysis shows that geomorphology, understorey and field layer variability may act as stronger
indicators of site conditions and character. Further analysis focused on the level and extent of
nature conservation protection that these unique and ancient systems receive. The study concludes
that despite their ecological importance and potentially international significance, ghyll woodlands
are poorly understood and protected.

Introduction

The Weald of southeast England supports woodland valley systems interna-

tionally distinct in both their ecology and geomorphology. The ghyll wood-
lands, around 1000 in total, are typically linear features occupying deep and
narrow valleys cut into the sandy and silty Hastings Beds and the Weald and
Wadhurst clays (Woolridge and Goldring 1962). The sheltered valleys occupied
by ghyll woodlands buffer temperature fluctuations and maintain high
humidity levels resulting in unusually oceanic micro-climatic conditions (Rose
1995). Elsewhere in northwestern Europe, ghyll-like valleys are rare, and thus
the concentration of these features in lowland England has international
geomorphological and ecological significance (Rose and Patmore 1997).
The Weald has retained extensive tracts of a variety of ecologically impor-
tant woodland types, and notably the central High Weald (HW) is the most

[259]
1320

wooded natural area in England (Countryside Commission 1994; Rose 1995;

Reid et al. 1996; Forestry Commission 2001). Spatial and structural analyses
suggest that the ghyll woodland systems are an important and significant
landscape feature in the Weald and are of high conservation value (Rose 1995;
Rose and Patmore 1997; Burnside et al. 2002). Yet, ghyll woodlands have
received only scant attention in the past.
The international and national ecological significance of these ghyll wood-
lands is that they support a unique assemblage of cryptogamic plants with both
oceanic and sub-oceanic affiliations (Ratcliffe 1968; Hodgetts 1997). The
presence of rich liverwort and moss communities suggests that the ghyll
woodlands are of considerable age, and are therefore also likely to support a
high biodiversity of other species of conservation concern, particularly ter-
restrial invertebrates, which share similar micro-environmental requirements
(Peterken 1993; Rose and Patmore 1997; Woodland Trust 2000). Most can be
regarded as ‘ancient woodland’ (sensu Rackham 1980; Peterken 1981; Forestry
Commission 2001), and Rose and Patmore (1997) suggest that field investi-
gations may indicate that some fragments could represent actual remnants of
prehistoric woodland.
Much of south-eastern and southern England underwent preferential
exploitation of woodland following a phase of human expansion in the late
Neolithic/early Bronze Age (Waller and Marlow 1994). From a land-use per-
spective, however, the steep slopes of the ghylls have meant that, for the most
part, they have remaining relatively uncultivated and thus wooded. Histori-
cally, the Weald area, though difficult to cultivate, benefited from the iron
industry with many woods managed (coppiced1) for charcoal production and
subsequent iron smelting (Brandon 1977; Cleere and Crossley 1985; Sussex
Biodiversity Partnership 2000). This provided early conservation of the
woodlands, as the resource was managed on a long-term sustainable basis, and
may have resulted in many ghylls maintaining ‘old forest type epiphytic lichens
and bryophytes’ (Rose and Patmore 1997).
The geology and geomorphology of the ghylls is also distinctive. In the
sandstone beds of the Weald, ghylls can take the form of narrow rock-walled
gorges. Conversely, where rocks are softer, slopes tend to be shallower and the
valleys less confined (Rose 1995). The Wealden ghylls differ from similar fea-
tures in southwest England, as a result of greater continental climatic influ-
ences, and the presence of relatively clay-rich soils. Rose and Patmore (1997)
report that similar ghyll-like features do occur in other parts of lowland Britain
that were not blanketed by glacial drift. Most, however, are shallow with
gentler relief and do not exhibit the steep morphology present in the Weald.
Many ghyll valleys extend over 1 km in length and, as a result of the com-
plexity of Wealden geology, traverse a number of different geological strata

1
Coppicing is the productive management of woodland. Trees are cut down and encouraged to
grow again from the stump. Coppicing produces a large number of thin stems, which are harvested
on a regular cycle of about 5–15 years.

[260]
 1321

(Gallois 1965). This variability in geology and unusual micro-climate has clear
implications for ecological diversity, and is considered to influence the high
levels of bryophyte diversity observed in ghyll systems (Rose and Patmore
1997).
The study presented within this paper constitutes the first attempt to use
geographical information system (GIS) approaches and contemporary surveys
to analyse and characterise these unique systems. The analysis focuses on the
geomorphological and ecological characteristics of the ghylls, but also con-
siders how the ghyll systems relate to existing conservation provision. GIS-
based cadastral approaches and multi-variate statistical techniques are used to
compare and analyse environmental, management and species information
derived from field surveys undertaken for a sample of ghyll woodlands within
southeast England.

Methodology

GIS techniques coupled with recent field survey data and secondary data
derived from digital maps have been used in an inductive study to analyse a
range of environmental and ecological characteristics for ghyll woodlands
within the central Weald.

Development of a GIS database

The study used boundary data for 1130 ghylls in the Weald. The development
of the GIS required the collation of large amounts of secondary data held in
digital and non-digital form. Data, such as ghyll woodland boundaries, nature
conservation designations, topography and soil associations, were obtained in
digital form. The ghyll woodland boundary data were compiled from field
reports held at the regional Biological Records Centre. The statutory nature
conservation designations examined included the boundaries of Sites of
Special Scientific Interest (SSSI), Special Areas of Conservation (SAC),
National Nature Reserves (NNR), and also ancient woodland (AW). The
resultant GIS was structured around geographic and ecological landscape
features.

Field survey

This study has used data from field surveys of 48 ghyll woodland systems
throughout East and West Sussex (Figure 1). Using a uniform walkover survey
approach, field survey data recorded both biotic and abiotic characteristics of
the ghylls (Table 1). The data were used to construct a GIS database con-
taining site-specific information on the ecology, landform and management of

[261]
1322

48 ghyll woodlands randomly selected from 1130 ghyll woodlands. Where there
were substantial differences in the vegetational or geomorphological structure
in individual ghylls, multiple surveys were undertaken so as to represent this
internal variation. The variables included in the survey reflected those con-
sidered important in characterising ghylls (Table 1), and were integrated in a
GIS and analysed using spatial and multi-variate techniques to identify sub-
groups or types of ghyll systems.

Table 1. Field survey information for the ghyll woodland sample in the Weald.

Characteristics Descriptors

Geology Predominant geological bed, geological beds in choronological

sequence
Geomorphology Valley form profiles, channel dimensions, height of ghyll,degree of fall,
length of ghyll
Present land use Coppicing, thinning/selective felling, clear felling, adjacent land use
Woodland type Broadleaved, coniferous, mixed-woodland NVC – e.g. W6,W8,W10
Vegetation structure Canopy, understorey, field layer vegetation composition using
DAFOR scale, presence of alien species
Presence of important Geomorphology – waterfalls, sandstone outcrops
features Ecology – bryophytes, invertebrates

Figure 1. The location of the ghyll woodland field survey sites.

[262]
 1323

Spatial and statistical multi-variate analysis of the data

FragStats (McGarigal and Marks 1994) was used to measure area,

perimeter length, and inter-patch distance for the ghyll woodlands.
Additional spatial statistics were calculated for the digital boundary data
of ghyll woodlands, to permit an assessment of the woodland pattern and
physiognomy, relevant to the habitat and species of concern (Haines-
Young and Choppin 1996; Gkaraveli et al. 2001). Using a cadastral
approach, each ghyll woodland was examined to establish if it was either
in [within] or abutted [intersected] one of the nature conservation desig-
nations (SSSI, SAC, NNR, AW) to facilitate examination of existing
conservation designation for these sites. This analysis of ghyll woodland
location was also undertaken for relevant English Nature landscape
character areas, LCA (English Nature 2002). This analysis facilitated the
classification of ghylls within broad-scale landscape management descrip-
tions, for example, High Weald or Low Weald character areas (English
Nature 2002).
The GIS was further utilised to examine the relationship between the ghyll
woodland systems and soil associations using the woodland digital boundaries
and digital soil information (SSLRC 2001). Soil associations known to be
allied with ghyll woodlands were examined and proportional areas of ghyll
woodlands situated in these soil types calculated using standard cadastral
approaches (ESRI 1998).
Statistical analysis involved the application of multi-variate techniques to the
field survey data. Cluster analysis and principal component analysis (PCA)
were used to examine the differences in the ghyll woodlands for all major
attributes recorded in the survey (the data format did not support detrended
correspondence analysis) (Kent and Coker 1992; Everitt et al. 2001). The
cluster analysis technique used average linkage and the Manhattan similarity
index to describe the group of classification procedures (Waite 2000). In this
analysis all ghyll woodland sites were initially considered to represent unique
ghyll woodland systems or sites. A hierarchical classification was derived from
a matrix, and based upon the values within that matrix samples were joined to
form similar groups (Waite 2000). At each level, groups were joined based on
their similarity. The process continued until all groups were joined to a com-
mon root.
PCA, was also used to assess the similarity or variation between these
woodland systems based upon their vegetation and environmental features
(Fowler et al. 1999; Waite 2000). A PCA ordination was undertaken to
examine the gradients and variation in vegetation present within the ghyll
systems. Analysis was undertaken collectively for the main canopy, understo-
rey, and field layer vegetation data along with geological and environmental
features (e.g. geological strata and valley form). This provided an initial
analysis of a range of biotic and abiotic features associated with these distinct
systems.

[263]
1324

Results

Spatial patterns

Analysis indicates that ghyll woodlands cover 9332 ha of East and West Sus-
sex, and are divided into 1130 patches (Table 2). The mean patch size (or
habitat area) of ghyll woodland fragments is 8 ha. The ghyll woodlands have a
high density of patches with 12 patches per 100 ha in the area of the Weald
examined (Table 2). The mean patch shape calculations show that the ghyll
woodland habitats (Shape Index = 17) are not uniform in shape (i.e. patch
shapes move away from being regular shapes such as squares or circles). The
calculation of mean nearest neighbour distance (NND) between ghyll wood-
land habitats gives values of 217 m (±250.5). A number of ghylls are however
joined by contiguous areas of other woodland (managed and unmanaged).
Thus, care must be taken to ensure that conclusions are not drawn regarding
potential fragmentation effects (Wiens 1989; Bailey et al. 2002).

Conservation designations

Cadastral analysis of the ghyll woodland systems and national conservation

designations indicate that the ghyll systems have not been a focus of conser-
vation attention and on a site basis only 9% are included within SSSI. This
proportion is also reflected in the sample of surveyed ghylls, where only 12%
were identified as being associated with SSSIs. Furthermore, only 4% of ghyll
woodland sites intersect the boundaries of SAC, and no ghyll woodlands are
designated as NNR
The GIS analysis shows, however, that on a site basis 85% of the ghyll
woodland systems are associated with areas of AW(both semi-natural and
replanted) in East and West Sussex. Additionally, a substantial number of
ghyll woodlands are included in the locally designated sites of nature
conservation interest (SNCI), and a number fall within important landscape

Table 2. Spatial statistics for the Ghyll woodlands within the Central Weald of East and West
Sussex (raster format).

Raster indices Ghyll woodland

Total area (ha) 9332

Mean area (ha) 8
Mean area std dev 13.3
Patch number 1130
Patch density (#/100 ha) 12
Mean Shape Index 2.17
Mean NND (m) 217
Mean NND std dev 250.5

[264]
 1325

Figure 2. The relationship between ghyll woodland systems and English Nature landscape
character areas.

designations such as area of outstanding natural beauty (AONB) and the HW.
Comparison of the distribution of ghyll woodlands and LCA classification
(English Nature 2002) reveals that 92% of ghyll systems are associated with the
HW LCA. A further 6% are found in the Low Weald LCA and, as might be
anticipated, only 2% are found in the Romney Marsh LCA (Figure 2).

Plant community characteristics

GIS data and field surveys revealed that Quercus robur – Pteridium aquilinum
– Rubus fruticosus woodland (NVC – W10) and Fraxinus excelsior – Acer
campestre – Mercurialis perennis woodland (NVC – W8) were the most
common woodland community types found. Of the woodland surveyed, 47%
was found to be predominantly W10, and a further 43% dominated by W8
woodland. The remaining areas were comprised of W7, W12, W15 and some
pockets of W4, W6 and W16 woodland (Rodwell 1998). Analysis using multi-
variate statistical approaches (Cluster analysis and PCA) was performed on
the digital survey records. This provided a descriptive means of assessing
similarity and differentiation both within individual ghylls and between dif-
ferent ghyll woodland sites. Cluster analysis was initially undertaken for the

[265]
1326

three vegetation components surveyed; the canopy, the understorey and the
field layer.

The canopy

The canopy data shows that overall there is a moderately good level of
similarity in the composition of the canopy within the sampled ghyll wood-
land systems (Figure 3a). Values of between 38 and 55% similarity were
obtained when comparing the overall data set and the various sub-groupings.
When clusters were grouped, analysis showed an increase in the similarity
values with groups 1–3 and 8 between 50 and 60% similar whilst groups 4–7
showed similarity values greater than 60% (see appendices). Within these
groups, one major anomaly was seen at Fairlight Glen, Hastings which was
clearly identified as an outlier. This is because the canopy of this southern,
more coastal, section of Fairlight Glen ghyll is solely dominated by Salix
species.
Further analysis identified some clear ghyll woodland associations, all
showing comparatively high similarity values (>64%). The data may suggest
that there are some ‘canopy’ based associations within these ghyll woodlands.
For example, Group 1 is shown to be 42% similar to the other ghyll woodlands
surveyed yet, there is a moderately high level of canopy similarity within the
group itself (56%) (see appendices).
More specifically, two classes are identified and, of these, the two Brick Kiln
ghyll woodland areas are identical in canopy structure. Furthermore, the
analysis picks out two, somewhat individual, ghyll woodlands based upon
canopy composition, Kiln Wood (Lower ghyll) (no. 11) and Sandyden Wood
(no. 49). These woodlands are shown as only 38% similar to all other ghyll
woodlands surveyed and have no comparative ghylls at the group or class level
(Figure 3a).

The understorey

The data shows that, generally, there is similarity in the composition of the
understorey within the ghyll woodland systems (Figure 3b). Values of between
31 and 61% similarity were obtained when comparing the overall data set and
the various sub-groupings provided through cluster analysis. The analysis at
the group level showed a substantial increase in the similarity values with all
groups, apart from group 1, greater than 60% similar. These findings may
indicate some potential categories or communities.
Analysis at the class level shows strong association within the ghyll
woodland systems, with similarity values around 75–100%. The analysis
shows that when the understorey vegetation is examined many of the ghyll
sites display little difference (e.g. Marline Valley, Batemans and Northlands

[266]
 1327

Figure 3. Cluster analysis dendrogram of the ghyll woodland canopy (a), under-storey (b), and
field (c) survey data. A complete table of percentage similarity values derived from cluster analysis
is provided in the appendix.

[267]
1328

Wood). It is, however, interesting to note that the analysis shows that
Marline valley does have some zonation within the ghyll – reflected in the
fact that two surveys were conducted at the site. Marline valley (away from
stream) and Marline Valley (near stream) are identical, as are Marline Valley
(by watercourse) and Marline Valley (away from watercourse). When com-
paring these two groups the analysis only shows a similarity value of 47%
between them (Figure 3b).
The analysis isolates Waterfall Wood as relatively distinctive. When the GIS
database and survey sheets are examined, this separation appears to be an
artefact of the understorey composition. Waterfall wood is the only ghyll with
the understorey solely dominated by Alnus species. Highams Ghyll and
Sandyden Wood also have Alnus species present within the understorey;
however, closer inspection shows that Highams also has Crataegus monogyna
dominant and Ilex aquifolium occasional within the understorey, whilst
Sandyden Ghyll has Sambucus nigra frequent within the understorey and
Corylus avellana occasional.

The field layer

The analysis of the field layer data using cluster analysis shows that some data
sets appear very dissimilar with values ranging from 21 to 39% similarity
(derived from 11 site surveys). In contrast, 48 of the remaining sites show high
levels of similarity (Figure 3c). When considering the cluster analysis for field
data, the dendrogram clearly shows that those sites to the right of the figure are
on the whole largely dissimilar whilst those sites to the left of the figure reveal
moderately good levels of similarity, >50% (Figure 3c).
When grouped, over half of the sites show similarity levels of 62%. Fur-
thermore, within group 1, for example, the associations show 13 sites with
80% similarity and a further 15 sites with 81% similarity (see appendices).
Despite these high levels of field layer similarity within the majority of ghyll
systems, there are additional ghyll systems which are shown to be relatively
distinctive in character and field layer composition. Sandyden Wood and
Waterfall Wood are shown to be 50% similar to each other in presence/
absence, but are only 39% similar to all other ghyll woodlands surveyed.
Wicks Copse and Tilsmore Wood are also outliers with similarity values of
25 and 21% respectively (Figure 3c). Wren’s Warren is also distinct, sepa-
rated from other ghylls by the presence of Vaccinium spp. within the field
layer (see appendices).

Ghyll woodlands and soils

Cadastral analysis via the GIS demonstrates that the ghyll woodlands occupy a
range of soil types. At a general level, GIS analysis indicates that 85% of the

[268]
 1329

ghyll woodlands are associated, to some extent, with brown soils and partic-
ularly with argillic brown earths of the Curtisden association (Association 5.72i
of Jarvis et al. 1984). This association is widespread throughout the HW,
occurring on siltstones and sandstones. In this association, slope relief is typ-
ically strong with moderately to steeply sloping valleys [ghylls] separated by
gently sloping interfluves. Soils of the Curtisden association are usually mod-
erately deep on gentle or moderate slopes. Characteristically, the soils comprise
slowly permeable compact subsoil, which is subject to seasonal water logging.
Springs and flushes are often common on the sloping ground of permeable and
impermeable strata (Jarvis et al. 1984).
Ghyll woodlands are also commonly found on surface-water gley soils
(Stagnogley Associations 7.11e, 7.11i and 7.12b of Jarvis et al. 1984), which
are linked, again to varying degrees, with 55% of ghyll systems. The
Wickhams 1 and 5 Associations are extensive on the Low Weald. The
Wickham Associations have slowly permeable subsoils and are often water-
logged for prolonged periods in winter. Often heavily wooded, these soils are
naturally acidic and can have good reserves of available water (Jarvis et al.
1984). Other associations are linked with the ghyll systems but their
importance is low. Podzolic soils, in particular gley podzolic soils, are
associated with 4% of the ghyll systems. Only 1% are associated with ground
water gley soils.

Cluster analysis of geological and topographical data

Cluster analysis was undertaken for the geological components of the surveys,
including surrounding geology, 1st, 2nd, 3rd Geological Beds (in chronological
sequence), height at top of ghyll, and height difference within the ghyll valley.
The geological and topographical data shows substantial level of similarity
within the ghyll woodland systems with 89% of the sites being >68% similar
(Figure 4). All reside on the clays and sands of the Hastings Beds. Beyond this
level of similarity there are some additional sub-groupings with particularly
strong associations yielding similarity values >80% (Figure 4), which reflect
the presence of ghylls on Ashdown sands, Tunbridge sands, Wadhurst clays
and Fairlight clays.
Figure 4a also shows that some sites appear relatively distinctive in com-
parison to the majority of ghylls surveyed. Site 41 (Guestling Wood) shows
low similarity values of 28% to all other ghyll systems examined. Two of the
Fairlight Glen surveys (Sites 46 and 27) show good similarity when compared
to each other (87%), but are only 41% similar to all other systems. Equally,
Courtlands Wood (Site 5) and Little Iwood (Site 10) are similar to one
another (90%) but are only 54% similar to all other survey sites. Examina-
tion of the GIS database and survey sheets suggests that these differences are
related to surrounding geology, strata and average levels of fall along the
ghyll. Further cluster analysis was undertaken for the geological components

[269]
1330

Figure 4. Cluster analysis dendrogram of the ghyll woodland geological (a) and geological and
vegetation survey data combined (b).

and the vegetational data from the ghyll woodland surveys combined (Figure
4b). The dendrogram repeats patterns observed within the previous cluster
analysis showing a strong association (>62%) between the majority of ghyll
woodland sites surveyed (over 80% of sites). In addition, the dendrogram
isolated the same sites as being distinctive including Fairlight Glen, Guestling
Wood, Courtlands Wood, Little Iwood and Marline Valley (Figure 4b). This
similarity of outcomes may be associated with the dominance of geological
data within the analysis of both data matrices and the relatively weak

[270]
 1331

contribution of the vegetational components to the geology and vegetational

matrix.

PCA of geological and vegetational data

PCA analysis of the geological and vegetational survey data reinforces the
conclusions drawn from the cluster analysis. Analysis shows high levels of
association between the majority of ghyll systems based upon the vari-
ables (shown by the main grouping on the ordination, Figure 5). This
suggests that the majority of ghyll systems have similar geological and
vegetational characteristics and little or no separation is present along
either axis.
It is however, interesting to note that there are two ghylls separated from
the main grouping along the first component showing 24% of the variation
(eigenvalue = 2.419) (Figure 5). Both Fairlight Glen and Marline Valley
(near watercourse) are separated along this axis and examination of the GIS
database suggests that this relates to strong differences in their geological

Figure 5. PCA analysis ordination of the ghyll woodland geological and vegetational survey data.

[271]
1332

beds, valley form and average degree of fall along the ghyll. Equally, a
further three ghylls are separated from the main grouping along the second
component showing 20% of the variation (eigenvalue=2.002). Guestling
Wood, Courtlands Wood and Little Iwood are separated here and the survey
data again points to geological conditions although some differences are
evident within their National Vegetational Classifications (Rodwell 1998)
(Figure 5).

Discussion

Using a GIS-based spatial approach and multi-variate statistical techniques,

this inductive study has provided an initial analysis of a sample of ghyll
woodlands within the Weald of southeast England which suggests the sig-
nificance of these important ecosystems within the broader landscape con-
text. Biological Records Centre digital data sets identify 1130 ghyll
woodlands in the High and Low Weald LCA. Spatial analysis suggests that,
on an area basis, these ghylls comprise 23% of the woodland within East
Sussex alone. This, however, is likely to be an over-estimate due to
boundary issues, and the previous classification of substantial woodland
areas as ghylls beyond the confines of the characteristic valley systems
(Burnside et al. 2002). This latter point is of particular relevance to area
calculations, as ghyll systems are normally regarded as restricted to small
linear valley features.
The analysis of the composite GIS survey database and field surveys using
multi-variate techniques has shown that the vegetation within the ghyll
woodlands is relatively similar in the canopy, understorey and field layers with
similarity values of around 50–60%. The similarity reflects the common
occurrence of NVC type W10 and W8 woodlands and a dominance of Quercus
sp. and Fraxinus sp. respectively in the canopy. Equally, within the understorey
and field layer the often widespread incidence of Ilex aquifolium and Corylus
avellana, and in other cases of Rubus fruticosus and Pteridium aquilinum,
provided good similarity levels.
In some cases, however, the vegetation within particular ghylls does appear
distinctive with percentage similarity values dropping to 20–30%. In relation to
the canopy this reflected cases whereby the vegetation was solely dominated by
Salix species. In relation to the understorey layer, the distinctive nature of
some ghylls reflected the strong presence of Castanea sativa and the manage-
ment practice of rotation coppicing.
Analysis of the geological data also illustrates that the ghylls are undiffer-
entiated at the general level. Cluster analysis identified that around 90% of
sites were 70% similar for geological beds (e.g. Ashdown beds or Tunbridge
Wells sands) and topographic characteristics (e.g. depth of ghyll and degree of
fall). However, as shown by the vegetational analysis, some of the ghylls do
remain relatively distinctive in topography (levels including 30–40%

[272]
 1333

similarity), which reflects the importance of landscape influence. The PCA

shows, from a geological, geomorphological and ecological perspective that
valley form, underlying geology and woodland type (sensu NVC) can be used
to differentiate some ghyll woodlands (Figure 5). For example, Fairlight Glen
and Marline valley are shown to be representative of steep-sided, confined
valleys, on harder Wadhurst clays with a dominance of traditional Quercus
robur – Pteridium aquilinum – Rubus fruticosus woodland (NVC – W10). Whilst
Guestling wood, Little Iwood and Courtlands wood, are representative of
more open valleys on unconsolidated and unlithified head-material. The latter
ghylls exhibit a strong presence of Fraxinus excelsior – Acer campestre –
Mercurialis perennis woodland (NVC – W8), vegetation more characteristic of
base-rich soils (Rodwell 1998). In comparative terms, many of the ‘main
groupings’ display less distinctive traits and exhibit the more general charac-
teristics, which include Ashdown or Tunbridge geology, a moderate depth of
ghyll (circa 20 m) and more than one woodland community along the entire
valley profile.
Comparison of the ghyll woodland data and nature conservation des-
ignation confirms that in many cases the ghyll systems have received little
protection. It is apparent that a substantial number of the ghyll woodlands
fall within the broad remit of landscape designations, such as AONB, but
on a site-by-site basis only about 10% have benefited from a nationally
recognised conservation designation. This may result from the small size
and linear character of individual ghyll woodlands when placed in the
context of the broader wooded landscape. Effectively, ghyll woodlands
represent small pockets of exceptionally high diversity, but often occur in
broader woodlands that do not warrant designation. The analysis pre-
sented here may well indicate an under-representation of these intrinsically
small and fragmented sites within existing statutory protection (Kirby
2003).
These linear and fragmented sites may also be susceptable to the negative
aspects of edge effect and incursion by more robust and competitive species
(Harrison and Bruna 1999). Alterations to the understorey and canopy layers
as a result of species change could have severely detrimental effects to the
structure of the woodland systems (Burke 1998), and may result in a reduction
in biodiversity and the loss of more specialised species indicative of the ghyll
woodland habitats.
The study emphasises the importance of developing appropriate man-
agement plans for these sites, and the need to set appropriate nature con-
servation designation. As stated earlier, many of the ghyll woodlands are
AW and have for the most part remained relatively undisturbed (Hodgetts
1997; Rose and Patmore 1997). This is an important consideration in
relation to both biodiversity management and biological conservation. In
many cases, whilst maintaining rich (and often locally rare) cryptogamic
communities, ghyll valleys have no prescribed management plans and are
under private ownership. Sustaining and promoting biodiversity at the

[273]
1334

landscape scale requires the characterisation and maintenance of pockets of

high biodiversity (Ernoult et al. 2003). In the case of some ghyll woodlands,
this may only involve non-interventionist management but, nevertheless,
designation must be secured. Some ghylls have been given regional status
via SNCI designation, but this is a local designation and provides only
limited protection.

Conclusion

GIS database and multi-variate statistical approaches have been used to

investigate ghyll woodland systems of the Sussex Weald. The ghylls have
been shown to be a reasonably uniform vegetation type in terms of the
canopy and their geological and soil characteristics. Yet, the variation in
geology and climate, along with historical factors creates a system which
is very species rich at the small scale (Waller and Marlow 1994; Rose
and Patmore 1997; Sussex Biodiversity Partnership 2000). Analysis shows
that the ghyll woodlands offer substantial opportunity for analysis of
woodland characterisation and diversity. Some workers consider that
bryophyte species occurrence and frequency may be the key to the sys-
tematic differentiation of ghyll woodland communities (Rose 1995; Rose
and Patmore 1997), whilst others propose a more balanced approach
looking at both ecological and landscape factors together (Burnside et al.
2002). However, this study shows that the geomorphology, understorey
and field layer may act as an initial indicator of site conditions and
character.
The research has also demonstrated that ghylls have relatively weak levels of
habitat protection in respect of their known conservation value for lower
plants. Addressing this protection is of critical importance given the national,
and potentially international, significance of the ghyll woodlands of the Weald.
Further GIS investigations linking surveys of mosses and lichens with climatic
and aspect data could provide a means to identify noteworthy ghyll woodlands
for conservation targeting.

Acknowledgements

Special thanks go to Neil Carrett for his assistance and input in the project, and
Colin Reader (Habitat Management Services) and John Patmore (University
of Brighton) for conducting the surveys. We would also like to thank David
Saunders and Laetitia Tual (East Sussex County Council) for their support and
provision of additional data for this project.
GIS analysis was performed on ArcView 3.x software, Environmental Sys-
tems Research Institute and FragStats (ver 2), McGarigal and Marks (1992).

[274]
 1335

Statistical analysis performed using WinSTAT 3.1, Kalmia Co. Inc., 1995.
Terrain data supplied by EDINA Edinburgh University, 2002; soil data sup-
plied by SILSOE, Cranfield University, 2002; forestry and woodland data
supplied by Forestry Commission, 2002; English Nature GUI, 2002, and
Sussex Biological Records Centre, 2001.

Appendices

Table A1. Percentage Similarity of cluster analysis for Canopy data.

[275]
1336

Table A2. Percentage Similarity of cluster analysis for Understory data.

[276]
 1337

Table A3. Percentage similarity of cluster analysis for Field data.

References

Bailey S.A., Haines-Young R.H. and Watkins C. 2002. Species presence in fragmented landscapes:
modelling of species requirements at the national level. Biological Conservation 108: 307–316.
Brandon P. 1977. The Sussex Landscape. Hodder and Stoughton, London.
Burke D.M. 1998. Edge and fragment size effects on the vegetation of deciduous forests in Ontario,
Canada. Natural Areas Journal 18: 45–53.
Burnside N.G., Carrett N.J., Metcalfe D. and Waite S. 2002. GIS analysis of Ancient and Ghyll
Woodlands within East Sussex. Biogeography & Ecology Research Group, University of
Brighton, Brighton.
Cleere H. and Crossley D. 1985. The Iron-industry of the Weald. Leicester University Press,
Leicester.
Countryside Commission. 1994. The High Weald. Countryside Commission, Northhampton.
English Nature. 2002. Landscape Character Areas, Data prepared by English Nature GUI. URL
http://www.english-nature.org.uk/.
Ernoult A., Bureau F. and Poudevigne I. 2003. Patterns of organisation in changing landscapes:
implications for the management of biodiversity. Landscape Ecology 18(3): 239–251.
ESRI. 1998. ArcView 3.2, Software System. Environmental Systems Research Institute Inc.
Everitt B.S., Landau S. and Leese M. 2001. Cluster Analysis, 4th ed. Arnold, London.

[277]
1338

Forestry Commission. 2001. National Inventory of Woodland and Trees: England. Forestry
Commission, Edinburgh.
Fowler J., Cohen L. and Jarvis P. 1999. Practical Statistics for Field Biology. Wiley, Chichester,
West Sussex.
Gallois R.W. 1965. British Regional Geology, 4th ed. Department of Scientific and Industrial
Research, London, UK.
Gkaraveli A., Williams J.H. and Good J.E.G. 2001. Fragmented native woodlands in Snowdonia
(UK): assessment and amelioration. Forestry 74(2): 89–103.
Haines-Young R. and Chopping M. 1996. Quantifying landscape structure: a review of landscape
indices and their application to forested landscapes. Progress in Physical Geography 20: 418–445.
Harrison S. and Bruna E. 1999. Habitat fragmentation and large-scale conservation: what do we
know for sure? Ecography 22: 225–232.
Hodgetts N.G. 1997. A National and International Context for the Cryptograms in the Weald with
Reference to Current and Future Conservation Initiatives for UK Cryptograms. In: Jackson A.
and Flanagan M. (eds), Conservation of Cryptogams in the Weald. Proceedings of the Work-
shop held at Wakehurst Place 2nd May 1996. Royal Botanic Gardens, Kew pp.1–18.
Jarvis M.G., Allen S.J., Fordham S., Hazelden J., Moffat A.J. and Sturdy R.G. 1984. Soils and
their Use in South East England. Soil Survey, Harpenden.
Kent M. and Coker P. 1992. Vegetation Description and Analysis a Practical Approach. John
Wiley and Sons, Chichester.
Kirby K.J. 2003. Woodland conservation in privately-owned cultural landscapes: the English
experience. Environmental Science and Policy 6(3): 253–259.
McGarigal, K. and Marks B.J. 1994. FRAGSTATS Spatial Pattern Analysis Program for Quan-
tifying Landscape Structure, Version 2. Forest Science Department, Oregon State University,
Corvallis.
Peterken G.F. 1981. Wood anemone in central Lincolnshire: an ancient woodland indicator?
Transactions Lincolnshire Natural Union 20: 78–82.
Peterken G.F. 1993. Woodland Conservation and Management, 2nd ed. Chapman and Hall,
London.
Rackham O. 1980. Ancient Woodland. Arnold, London.
Ratcliffe D.A. 1968. An ecological account of the Atlantic bryophytes in the British Isles. New
Phytologist 67: 365–430.
Reid C.M., Kirby K.J. and Cooke R. 1996. A Preliminary Assessment of Woodland Conservation
in England by Natural Area. English Nature, Peterborough, UK.
Rodwell J.S. 1998. British Plant Communities Volume 1: Woodlands and Scrub. Cambridge
University Press, Cambridge.
Rose F. 1995. The Habitats and Vegetation of Sussex. The Booth Museum of Natural History,
Brighton Borough Council.
Rose F. and Patmore J.M. 1997. Gill Woodlands in the Weald. English Nature, Peterborough.
SSLRC. 2001. Sussex Soil Associations. Data prepared by National Soil Resources Institute
(SSLRC), Digital Data Copyright (c) Cranfield University, 2001 (SSLRC Project Code: JP7017v/
9).
Sussex Biodiversity Partnership. 2000. Woodland Habitat Action Plan. Sussex Biodiversity Part-
nership, Sussex.
Waller M.P. and Marlow A.D. 1994. Flandrian vegetational history of southeastern England –
stratigraphy of the Brede Valley and pollen data from Brede Bridge. New Phytologist 126(2):
369–392.
Waite S. 2000. Statistical Ecology: A Practical Guide. Prentice Hall, Harlow, UK.
Wiens J.A. 1989. The Ecology of Bird Communities, Vol. 2. Cambridge University Press, New
York.
Woodland Trust. 2000. Woodland Biodiversity: Expanding Our Horizons. Woodland Trust,
Grantham, Lincs.
Woolridge S.W. and Goldring F. 1962. The Weald, 3rd ed. Collins, London. UK.

[278]
Biodiversity and Conservation (2006) 15:1339–1351
Springer 2006
DOI 10.1007/s10531-005-4875-1

-1

Effects of fragmentation of evergreen broad-leaved

forests on genetic diversity of Ardisia crenata var.
bicolor (Myrsinaceae)

AI-LIAN ZHAO1, XIAO-YONG CHEN1,2,*, XIN ZHANG1 and

DONG ZHANG1
1
Department of Environmental Sciences, East China Normal University, Shanghai 200062, P. R.
China; 2Shanghai Key Laboratory for Ecological Processes and Restoration in Urban Areas,
Zhongshan R. (N.) 3663, Shanghai 200062, P. R. China; *Author for correspondence (e-mails:
[email protected], [email protected]; phone: +86-21-62232697; fax: +86-21-62233669)

Received 11 May 2004; accepted in revised form 20 March 2005

Key words: Ardisia crenata var. bicolor, Differentiation, Forest fragmentation, Genetic diversity,
Population size, RAPD markers

Abstract. Due to the long generation times and high densities, dominant tree species usually did
not respond consistently with theoretical predictions to the recent fragmentation. Genetic struc-
tures of shrubs and herbs, especially those with low densities, may be more sensitive to forest
fragmentation. We studied the genetic structure of a self-compatible subshrub, Ardisia crenata var.
bicolor (Myrsinaceae) in a recently fragmented landscape. Ten RAPD primers used for analysis
generated a total of 76 bands. We found that A. c. var. bicolor had relatively low species-level
(P95 = 63.2%; H = 0.106; Shannon diversity index (SI) = 0.246) and within-population diversity
(P95 = 5.3
46.1%; H = 0.026
0.175; SI = 0.032
0.253), and significant population differenti-
ation (GST = 0.445). Significantly positive relationships were found between measures of diversity
(P95, H and SI) and the log of estimated population size. No significant relationship was observed
between Nei’s genetic distance and spatial distance of pairwise populations, indicating no isolation-
by-distance. Given most species of forests are shrubs and herbs with short generation times, our
observation indicated that distinct genetic consequences of recent fragmentation may be expected
for quite a number of plant species.

Introduction

Due to increased urbanization, intensive agricultural practices and habitat

destruction, many plant species occur in highly fragmented habitats (Van
Rossum et al. 2004). Usually, consequences of habitat fragmentation consist of
reduced population size and increased isolation (Saunders et al. 1991; Van
Rossum et al. 2004), leading to genetic erosion and increased genetic differ-
entiation among populations, through random drift, increased levels of
inbreeding and reduced gene flow (Young et al. 1996; Chen 2000; Van Rossum
et al. 2004). Ultimately these genetic processes may result in fitness declines and
extinction (Keller and Waller 2002; Bacles et al. 2004).
There have been increasing studies concerning genetic effects in plants of
habitat fragmentation, and loss of genetic diversity and increased differen-

[279]
1340

tiation have been found in some systems (e.g., Raijmann et al. 1994; Hall
et al. 1996; Morden and Loeffler 1999; Frankham et al. 2002). However,
these responses to increased fragmentation are unlikely to be common, and
other factors may influence the genetic consequences of fragmentation. First,
most studies were conducted on long-lived species in a recently fragmented
landscape and there were not sufficient time for bottleneck and inbreeding to
take action (Young et al. 1993; Cardoso et al. 1998). Long-lasting, dormant
seed banks also can buffer against genetic effects for decades or centuries
(Morris et al. 2002). Second, many plant species studied were dominant
species with high density. Thus, populations in fragmented habitats were
large enough to maintain relatively high genetic diversity. Thirdly, some
species are naturally rare species, and have evolved mechanisms to overcome
the disadvantages of small population size. Thus, genetic consequences of
recent fragmentation may not be detectable for a long time (England et al.
2002).
Shrubs and herbs constitute the main part of species composition of
forests. Therefore, genetic effects observed in dominant species of forests
might not be general for most forest species. Shrubs and herbs – especially
those with low densities – may be more genetically sensitive to forest frag-
mentation because of their much shorter life span. A shorter life span means
they pass many generations even in recently fragmented habitats and, given
their low densities, they therefore experience large declines in population size
even if the fragmentation is not serious. However, much fewer studies have
been conducted on herbs and shrubs than on tree species. For example,
populations of herbaceous Swertia perennis in small, isolated habitats had
reduced genetic variability and the highest within-population inbreeding
coefficients (Lienert et al. 2002). In the herb, Scutellaria montana, popula-
tions that were less than 100 individuals tended to have lower proportions of
polymorphic loci than that of populations more than 100 individuals (Cru-
zan 2001).
Ardisia (Myrsinaceae) is a tropical and subtropical genus and includes
about 200 species. Coral ardisia, A. crenata, native to Japan to north India, is
an insect-pollinated and self-compatible evergreen subshrub (Cheon et al.
2000). In China, a variant, A. crenata var. bicolor, was identified according to
the purple color of the lower side of its leaves, whereas some researchers
thought it as a distinct species, i.e. A. bicolor. A. crenata var. bicolor is a small
upright-growth shrub. Although outcrossing rate of A. c. var. bicolor was
estimated to be about 1 based on allozyme using Ritland’s (1990) MLT
program (Chen et al. 2001), bag-pollination treatments indicated that it is self-
compatible (unpublished data). This species can reproduce vegetatively via
rhizome, but spreading to a short distance, usually less than 1 m (personal
observations).
In the present study, populations of A. c. var. bicolor in a fragmented
landscape were selected to determine whether there is a relationship between
population size and the level of variation and to evaluate the degree of

[280]
 1341

population subdivision and differentiation, using RAPD markers. Although

RAPDs have some limitations – such as dominant allelic expression and
occasionally low reproducibility – they have advantages in investigating genetic
variation, such as random sampling in the whole genome, high levels of
polymorphism, and fast and easy to perform, and have been widely used in
estimating genetic variation of plant populations (Nybom and Bartish 2000;
Nybom 2004).

Methods

Population sampling

The study sites were located in Tiantong Forest Park (TFP) and adjacent
areas (Figure 1). TFP was distributed by evergreen broad-leaved forests
(EBLFs) dominated by Fagaceae species, such as Castanopsis fargesii, Ca.
carlesii, Ca. sclerophylla, Lithocarpus glaber, L. henryi, Cyclobalanopsis nubi-
um, and species of Theaceae (Schima superba) and Lauraceae (Machilus
thunbergii) (Song and Wang 1995). Around TFP, there were EBLFs fragments
of previous continuous forests or recovered from abandoned or unmanaged
plantations. These fragmented EBLFs were usually dominated by Cycloba-
lanopsis glauca, Cy. gilva, Ca. sclerophylla, L. glaber, Ca. carlesii, M. thun-
bergi. Surrounding these EBLFs, there were Cunninghamia lanceolata
plantations, Phyllostachys pubescens forests, and shrubs dominated by Quer-
cus fabra and bamboos.
In TFP and adjacent areas, A. c. var. bicolor usually appears in forests of
lower than 300 m above sea-level. Based on detailed surveys, 10 populations of
A. c. var. bicolor were sampled (Figure 1). The estimated sizes of each popu-
lation ranged from 5 to about 1000 individuals (Table 1). Leaves were col-
lected randomly from individuals with a distance of at least 2 m between each
other in medium and large populations, avoiding collecting the same clones. In
small populations, as many as possible individuals were collected with a dis-
tance of at least 2 m between sampled individuals.

DNA extraction and PCR condition

We isolated DNA with modified Doyle and Doyle’s (1987) procedure (Fan
et al. 2004). A set of random 10-mer primers was purchased from Sagon Inc.,
Shanghai. After screening more than 100 arbitrary primers, 10 primers that
consistently amplified clear banding patterns were chosen for further studies
(Table 1). RAPD assays were performed using the conditions described by Fan
et al. (2004). Samples were amplified at least two replicates and same pattern
was obtained by the primers used in this study. Five ll amplification product
was separated on 1.6% agarose gel in 0.5· TBE buffer and visualized by

[281]
1342

Figure 1. Locations of sampling sites of Ardisia crenata var. bicolor in Tiantong Forest Park and
adjacent areas.

[282]
 1343

Table 1. RAPD primers used in the survey of Ardisia bicolor and number of scored bands.

Primer Sequence 5¢-3¢ Number of Primer Sequence 5¢-3¢ Number of

scored bands scored bands

S59 CTGGGGACTT 8 S1361 TCGGATCCGT 10

S1200 GTGAACGCTC 9 S2068 CATACGGGCT 8
S1221 CACACCGTGT 3 S2084 CCCAAGCGAA 4
S1238 GTTGCGCAGT 11 S2100 CAAAGGCGTG 9
S1341 GTCCACCTCT 4 S2160 CACCGACATC 10

staining with ethidium bromide and photographed under UV light with Bio-
RAD Gel Doc2000TM.

Data analysis

Each PCR product was assumed to represent a single locus and was scored for
presence and absence. The resulting data matrix was analyzed using Popgene
1.31 (Yeh et al. 1999). Gene diversities (H) at population and at the species
level were calculated based on Lynch and Milligan’s (1994) Taylor expansion
estimate using TFPGA (Tools For Population Genetic Analyses) v1.3 (Miller
1997).
Nei’s unbiased genetic identity (I) and genetic distance (D) between popu-
lations were also analysed using Popgene 1.31. Because the data were larger
than the up-limit of AMOVA, coefficient of gene differentiation (GST) was
calculated to estimate population
P differentiation. Shannon diversity index
(Lewontin 1972), SI =
pi log2 pi, was calculated to provide a relative
estimate of the degree of variation at population and species levels using
Popgene 1.31 (Yeh et al. 1999). The proportion of diversity among populations
was estimated as (SIsp
SIpop)/SIsp, whereas SIsp and SIpop were SI at species
and population level, respectively. A Mantel type matrix randomization test
(Mantel 1967) was performed to evaluate the relationship between the matrix
of genetic distances and the matrix of geographic distances using TFPGA
(Miller 1997).
Relationships between measures of within-population genetic variation and
population size were analyzed using Regression methods in Microsoft Excel
program.

Results

The RAPD profile

The 10 primers used for analysis generated a total of 76 bands, among which
polymorphic bands were 48 (or 63.2%) and 51 (or 67.1%) based on 95 and

[283]
1344

Table 2. Patterns of genetic diversity for Ardisia bicolor populations.

Population Estimated population size Sample size P95 P H SI

A 1000 34 35.5% 40.8% 0.144 0.208

B 1000 33 32.9% 35.5% 0.122 0.178
C 100 15 29.0% 30.3% 0.120 0.168
D 100 14 30.3% 32.9% 0.128 0.179
E 1000 37 46.1% 47.4% 0.175 0.253
F 500 34 32.9% 38.2% 0.127 0.186
G 5 2 7.89% 7.9% 0.039 0.048
H 200 21 30.3% 31.6% 0.122 0.174
I 5 2 5.3% 5.3% 0.026 0.032
J 100 32 22.4% 27.6% 0.073 0.112
K 50 22 27.6% 32.9% 0.113 0.163
L 10 5 15.8% 15.8% 0.065 0.089
M 100 25 21.1% 23.7% 0.077 0.112
N 200 23 34.2% 36.8% 0.129 0.186
O 300 13 39.5% 40.8% 0.171 0.236
P 30 8 19.7% 19.7% 0.072 0.103
Mean 294 20 26.9% 29.2% 0.106 0.152
Total 4700 320 63.2% 67.1% 0.192 0.246
P95 and P are percentages of polymorphic bands based on 95 and 100% criteria, respectively; H
denotes mean Nei’s gene diversity based on Nei’s (1972) unbiased estimates; SI, Shannon’s diversity
index.

100% criteria, respectively (Table 2). The number of scored bands ranged from
3 for primer S1221 to 11 for primer S1238. 308 of the 320 individuals from the
16 populations were found to have a unique multilocus genotype, and six
genotypes have two individuals. The individuals having the same multilocus
genotypes belonged to same populations. No population-specific band was
observed in the data set.

Genetic diversity

Percent polymorphic RAPD loci varied from 5.3 (population I) to 46.1 %

(E), based on 95% criterion, with a mean of 26.9% (Table 2). The Nei’s gene
diversity ranged from 0.026 to 0.175 with a mean of 0.106 and the pooled
species-level value was 0.192. The relative degree of diversity in each popu-
lation as measured by Shannon’s index varied from 0.032 to 0.253 (Table 2).
The mean Shannon diversity for all populations was 0.152 and the pooled
species-level value was 0.246. Among the 16 populations, population E
exhibited the highest level of genetic variability (P, H and SI) and population
O was the next, whereas population I was the lowest (Table 2). Regression
analysis indicated significantly positive relationships between measures of
within-population genetic variation (P95, H and SI) and the log of population
sizes (Figure 2).

[284]
 1345

Figure 2. Relationships between measures of within-population genetic variation and population

size in Ardisia crenata var. bicolor. P95, H and SI were the percentage of polymorphic loci at 95%
criterion, expected heterozygosity and Shannon diversity index, respectively.

Genetic differentiation

The coefficient of genetic differentiation between populations (GST) was 0.445,

indicating a high differentiation among populations. The Shannon’s index
analysis partitioned 38.4% of the total variation among populations. Genetic
distances (D) between populations varied from 0.018 to 0.149 (Table 3) with a
mean of 0.055±0.023. The level of gene flow (Nm) was estimated to be 0.312.
Mantel test indicated no significant relationship between genetic distance and
spatial distance (r = 0.021, P = 0.422).

[285]
 1346
Table 3. Nei’s unbiased genetic identity (below diagonal) and genetic distance (above diagonal) between populations of Ardisia crenata var. bicolor.

Population A B C D E F G H I J K L M N O P

A – 0.0372 0.0361 0.0212 0.0441 0.0254 0.1092 0.0429 0.0558 0.0603 0.0345 0.0383 0.0354 0.0382 0.0581 0.0445
B 0.9635 – 0.0397 0.0380 0.0442 0.0481 0.0893 0.0300 0.0552 0.0538 0.0300 0.0490 0.0296 0.0247 0.0676 0.0606
C 0.9646 0.9611 – 0.0324 0.0505 0.0452 0.0959 0.0372 0.0479 0.0655 0.0418 0.0626 0.0464 0.0296 0.0716 0.0631
D 0.9790 0.9627 0.9682 – 0.0426 0.0370 0.0922 0.0395 0.0532 0.0589 0.0398 0.0393 0.0356 0.0331 0.0462 0.0530
E 0.9569 0.9567 0.9507 0.9583 – 0.0389 0.0895 0.0338 0.0707 0.0574 0.0471 0.0725 0.0519 0.0281 0.0287 0.0457
F 0.9749 0.9530 0.9558 0.9636 0.9618 – 0.0953 0.0294 0.0718 0.0548 0.0499 0.0624 0.0414 0.0456 0.0660 0.0584
[286]

G 0.8965 0.9145 0.9085 0.9119 0.9144 0.9091 – 0.0892 0.1155 0.0858 0.0894 0.1216 0.0827 0.0608 0.0722 0.1487
H 0.9580 0.9705 0.9635 0.9613 0.9668 0.9711 0.9147 – 0.0411 0.0555 0.0397 0.0676 0.0385 0.0339 0.0526 0.0341
I 0.9458 0.9463 0.9532 0.9482 0.9317 0.9307 0.8909 0.9597 – 0.0732 0.0570 0.0771 0.0562 0.0526 0.0754 0.0447
J 0.9415 0.9477 0.9366 0.9428 0.9442 0.9467 0.9178 0.9460 0.9294 – 0.0184 0.0397 0.0387 0.0398 0.0713 0.1064
K 0.9661 0.9704 0.9590 0.9610 0.9540 0.9513 0.9145 0.9611 0.9446 0.9818 – 0.0304 0.0290 0.0296 0.0609 0.0789
L 0.9625 0.9522 0.9393 0.9614 0.9301 0.9395 0.8855 0.9346 0.9258 0.9610 0.9701 – 0.0257 0.0608 0.0926 0.1034
M 0.9652 0.9709 0.9547 0.9650 0.9494 0.9595 0.9206 0.9623 0.9454 0.9620 0.9714 0.9746 – 0.0413 0.0755 0.0876
N 0.9625 0.9756 0.9709 0.9675 0.9723 0.9555 0.9411 0.9667 0.9488 0.9609 0.9708 0.9410 0.9595 – 0.0518 0.0565
O 0.9435 0.9347 0.9309 0.9548 0.9717 0.9361 0.9303 0.9488 0.9274 0.9312 0.9409 0.9115 0.9273 0.9495 – 0.0536
P 0.9565 0.9412 0.9389 0.9484 0.9553 0.9433 0.8618 0.9665 0.9563 0.8991 0.9242 0.9018 0.9161 0.9451 0.9478 –
 1347

Discussion

The present study reveals relative low genetic diversity in A. c. var. bicolor
compared to other species based on RAPD markers. Percentage of poly-
morphic bands, Nei’s gene diversity and Shannon index of the pooled data
were 67.1%, 0.192 and 0.246, respectively (Table 2). These values were even
lower than many endangered species, Metasequoia glyptostroboides
(P: 87.9%, H=0.318, SI = 0.476) (Li et al. 2005), Caesalpinia echinata
(P = 95.7%) (Cardoso et al. 1998), Leucadendron elimense (P = 98.8%)
(Tansley and Brown 2000), Boloria aquilonaris (H = 0.402) (Vandewoestijne
and Baguette 2002), but higher than Haplostachys haplostachya (H = 0.166)
(Morden and Loeffler 1999), Dryopteris cristata (P = 2.5%) (Landergott
et al. 2001).
Our results were not in accordance with the predictions based on the asso-
ciation of life history traits and genetic variation. According to the data of
RAPDs, there were strong associations between genetic diversity and breeding
system or successional status (Nybom and Bartish 2000; Nybom 2004). Out-
crossing species had significantly high genetic diversity than selfers. Higher
genetic diversity was found in late- than early - successional species. Species of
ingested seed dispersal also possessed relatively high genetic diversity (Nybom
2004). Given such considerations, high genetic diversity was expected based on
its high outcrossing rate (Chen et al. 2001), mid to late-successional status and
bird dispersal manner.
Relatively low genetic diversity in the studied A. c. var. bicolor populations
might be explained by their geographical positions. The studied populations
are located on the eastern margin of its distribution in mainland China (Figure
1). Due to effects of founder events, genetic drift and inbreeding, marginal
populations usually possess relative low genetic variation, which had been
confirmed in diverse species (Chen et al. 1997; Tyler 2002; Cassel and Tam-
maru 2003). Restricted geographical range in the present study might be
another explanation of the low genetic diversity. Though 10 populations were
sampled, their spatial distances were small. The largest distance of pairwise
populations is 4.6 km. If populations were sampled in a large range, more
genetic variation might be expected.
Though some studies failed to observe the distinct genetic consequences of
forest fragmentation on plant populations (Ellstrand and Elam 1993; Young
et al. 1996), our results indicated that habitat fragmentation had played a vital
role in genetic structure of A. c. var. bicolor populations. Fragmentation led to
the loss of genetic diversity. In small populations, significantly lower diversity
was observed than large and medium ones. Significant relationship was found
between estimated population size and within-population genetic variation as
measured by P, H and Shannon index.
Decreased genetic diversity in small populations was due to various reasons.
Firstly, the instantaneous effects of fragmentation (i.e., sampling effects) lead
to stochastic loss of rare alleles because only a small portion of the original

[287]
1348

gene pool remains after the decrease. Buchert et al. (1997) had compared the
genetic diversity in pre-harvest and post-harvest gene pools of two virgin
stands of eastern white pine (Pinus strobus). They found total and mean
number of alleles was reduced by 25% after tree density reductions of 75%.
About 40% of the low frequency alleles and 80% of the rare alleles were lost
because of harvesting (Buchert et al. 1997). Secondly, inbreeding and genetic
drift further decreased genetic variation in small populations (Young et al.
1996). In the present study, most small populations experienced bottleneck for
more than 10 generations, given a generation of 3 years and deforestation of at
least 50 years. It is enough for inbreeding and drift to virtually change the
genetic composition of small populations of less than 50 individuals. Thirdly,
founder effect might also contribute to low genetic diversity in some small
populations (Frankham et al. 2002). Population G, for instance, located in
dense high shrubs, had only two small individuals, indicating a recent founding
event.
High genetic differentiation was observed among populations of A. c. var.
bicolor. GST indicated that about 44.5% of the genetic variation occurred
among populations with short spatial distances. This value is higher than the
RAPD-based estimates of other widespread, or animal-dispersal species
(Nybom and Bartish 2000). High genetic differentiation among populations
was also in accordance with theoretical prediction of fragmentation, indicating
the effects of bottleneck and inbreeding.
No significant relationship between genetic distance and spatial distance was
found in the present study. This is usually interpreted that selection or drift
plays a more significant role than gene flow. In this study, no distinct difference
in habitats was found among populations, though the dominant species were
different. At local scale, populations from different communities usually
showed similar genetic composition in studied species, such as, Cyclobalanopsis
glauca (Chen and Song 1998). Therefore, selection plays a minor role in the
differentiation of A. c. var. bicolor populations, and drift led by fragmentation
contributed to the high differentiation.
Our findings in A. c. var. bicolor give a gloomy implication for forest
species because most species of forests are shrubs and herbs and among them
most are moderate- or low-density species. For example, there were about a
dozen of species in tree layer of EBLFs; among them, usually less than 3
species dominated the community. However, the number of species in shrub
and herb layers was about three to four folds of that in tree layer (Song and
Wang 1995). Among them, most are moderate or low density, like A. c. var.
bicolor, and are vulnerable to fragmentation. This situation is also common
in tropical, temperate or boreal forests. Thus, although some studies showed
no distinct effects on long-lived tree species which have survived hundreds of
years of fragmentation, our study indicated that distinct genetic consequences
of recent fragmentation may be expected for quite a number of plant species.
More attention should be paid to these species and conservation efforts are
needed.

[288]
 1349

Acknowledgements

We thank Mr Ling-jian Li and Ms Xiao-xia Fan for the assistance in sample

collection and preparation. We thank J. P. Sniadecki of Grand Valley State
University for English improvement and helpful comments and reviewers for
their critical comments and suggestions. This study was supported by Natural
Science Foundation of China (39870128, 30170060), The State’s Tenth Five-
year ‘‘211 Project’’ and Shanghai Priority Academic Discipline.

References

Bacles C.F.E., Lowe A.J. and Ennos R.A. 2004. Genetic effects of chronic habitat fragmentation on
tree species: the case of Sorbus aucuparia in a deforested Scottish landscape. Molecular Ecology
13: 573–584.
Buchert G.P., Rajora O.P., Hood J.V. and Dancik B.P. 1997. Effects of harvesting on genetic
diversity in old-growth eastern white pine in Ontario, Canada. Conservation Biology 11: 747–758.
Cardoso M.A., Provan J., Powell W., Ferreira P.C.G. and de Oliveira D.E. 1998. High genetic
differentiation among remnant populations of the endangered Caesalpinia echinata Lam.
(Leguminosae-Caesalpinioideae). Molecular Ecology 7: 601–608.
Cassel A. and Tammaru T. 2003. Allozyme variability in central, peripheral and isolated popu-
lations of the scarce heath (Coenonympha hero: Lepidoptera, Nymphalidae); implications for
conservation. Conservation Genetics 4: 83–93.
Chen X.Y. 2000. Effects of habitat fragmentation on genetic structure of plant populations and
implications for the biodiversity conservation. Acta Ecologica Sinica 20: 884–892.
Chen X.Y., Li N. and Shen L. 2001. The mating system of Ardisia crenata var. bicolor (Myrsin-
aceae), a subtropical understory shrub, in Tiantong National Forest Park, Zhejiang Province.
Acta Phytoecologica Sinica 25: 161–165.
Chen X.Y. and Song Y.C. 1998. Microgeographic differentiation in a Cyclobalanopsis glauca
poplation in western Huangshan, Anhui Province. Journal of Plant Resources Environment
7: 10–14.
Chen X.Y., Wang X.H. and Song Y.C. 1997. Genetic diversity and differentiation of Cycloba-
lanopsis glauca populations in East China. Acta Botanica Sinica 39: 149–155.
Cheon C.P., Chung M.Y. and Chung M.G. 2000. Allozyme and clonal diversity in Korean pop-
ulations of Ardisia japonica and Ardisia crenata (Myrsinaceae). Israel Journal of Plant Science
48: 239–245.
Cruzan M. 2001. Population size and fragmentation thresholds for the maintenance of genetic
diversity in the herbaceous endemic Scutellaria montana (Lamiaceae). Evolution 55: 1569–1580.
Doyle J.J. and Doyle J.L. 1987. A rapid DNA isolation procedure for small quantities of fresh leaf
tissue. Phytochemical Bulletin 19: 11–15.
Ellstrand N.C. and Elam D.R. 1993. Population genetic consequences of small population
size: implications for plant conservation. Annual Review of Ecology and Systematics 24:
217–242.
England P.R., Usher A.V., Whelan R.J. and Ayre D.J. 2002. Microsatellite diversity and genetic
structure of fragmented populations of the rare, fire-dependent shrub Grevillea macleayana.
Molecular Ecology 11: 967–977.
Fan X.X., Shen L., Zhang X., Chen X.Y. and Fu C.X. 2004. Assessing genetic diversity of Ginkgo
biloba L. (Ginkgoaceae) populations from China by RAPD markers. Biochemical Genetics 42:
269–278.
Frankham R., Ballou J.D. and Briscoe D.A. 2002. Introduction to Conservation Genetics.
Cambridge University Press, Cambridge.

[289]
1350

Hall P., Walker S. and Bawa K. 1996. Effect of forest fragmentation on genetic diversity and
mating system in a tropical tree, Pithecellobium elegans. Conservation Biology 10: 757–768.
Keller L.F. and Waller D.M. 2002. Inbreeding effects in wild populations. Trends in Ecology and
Evolution 17: 230–241.
Landergott U., Holderegger R., Kozlowski G. and Schneller J. 2001. Historical bottlenecks
decrease genetic diversity in natural populations of Dryopteris cristata. Heredity 87: 344–355.
Lewontin R.C. 1972. The apportionment of human diversity. Evolutionary Biology 6: 381–398.
Li Y.-Y., Chen X.-Y., Zhang X., Wu T.-Y., Lu H.-P. and Cai Y.-W. 2005. Genetic differences
between wild and artificial populations of Metasequoia glyptostroboides Hu et Cheng (Taxodi-
aceae): Implications for species recovery. Conservation Biology 19: 224–231.
Lienert J., Fischer M., Schneller J. and Diemer M. 2002. Isozyme variability of the wetland spe-
cialist Swertia perennis (Gentianaceae) in relation to habitat size, isolation, and plant fitness.
American Journal of Botany 89: 801–811.
Lynch M. and Milligan B.G. 1994. Analysis of population genetic structure with RAPD markers.
Molecular Ecology 3: 91–99.
Mantel N. 1967. The detection of disease clustering and a generalized regression approach. Cancer
Research 27: 209–220.
Miller M.P. 1997. Tools for population genetic analyses (TFPGA) v1.3: A windows program for
the analysis of allozyme and molecular genetic data. Department of Biological Sciences Northern
Arizona University, Flagstaff.
Morden C.W. and Loeffler W. 1999. Fragmentation and genetic differentiation among subpopu-
lations of the endangered Hawaiian mint Haplostachys haplostachya (Lamiaceae). Molecular
Ecology 8: 617–625.
Morris A.B., Baucom R.S. and Cruzan M.B. 2002. Stratified analysis of the soil seed bank in the
cedar glade endemic Astragalus bibullatus: Evidence for historical changes in genetic structure.
American Journal of Botany 89: 29–36.
Nybom H. 2004. Comparison of different nuclear DNA markers for estimating intraspecific genetic
diversity in plants. Molecular Ecology 13: 1143–1155.
Nybom H. and Bartish I.V. 2000. Effects of life history traits and sampling strategies on genetic
diversity estimates obtained with RAPD markers in plants. Perspectives in Plant Ecology,
Evolution and Systematics 3/2: 93–114.
Raijmann L.E.L., Van Leeuwen N.C., Kersten R., Oostermeijer J.R.B., Den Nijs H.C.M. and
Menken S.B.J. 1994. Alleleic variation and outcrossing rate in relation to population size in
Gentiana pneumonanthe L. Conservation Biology 8: 1014–1026.
Ritland K. 1990. A series of FORTRAN computer programs for estimating plant mating systems.
Journal of Heredity 81: 235–237.
Saunders D.A., Hobbs R.J. and Margules C.R. 1991. Biological consequences of ecosystem frag-
mentation: A review. Conservation Biology 5: 18–32.
Song Y.C. and Wang X.R. 1995. Vegetation and Flora of Tiantong National Forest Park, Zhejiang
Province. Shanghai Scientific Documentary Press, Shanghai.
Tansley S.A. and Brown C.R. 2000. RAPD variation in the rare and endangered Leucadendron
elimense (Proteaceae): implications for their conservation. Biological Conservation 95: 39–48.
Tyler T. 2002. Large-scale geographic patterns of genetic variation in Melica nutans, a widespread
Eurasian woodland grass. Plant Syst. Evol. 236: 73–87.
Van Rossum F., De Sousa S.C. and Triest L. 2004. Genetic consequences of habitat fragmentation
in an agricultural landscape on the common Primula veris, and comparison with its rare con-
gener, P. vulgaris. Conservation Genetics 5: 231–245.
Vandewoestijne S. and Baguette M. 2002. The genetic structure of endangered populations in the
Cranberry Fritillary, Boloria aquilonaris (Lepidoptera, Nymphalidae): RAPDs vs allozymes.
Heredity 89: 439–445.
Yeh F.C., Yang R.C. and Boyle T. 1999. Popgene Version 1.31: Microsoft Window-Based Free-
ware for Population Genetic Analysis. Department of Renewable Resources, University of
Alberta, Edmonton, AB Canada.

[290]
 1351

Young A., Boyle T. and Brown T. 1996. The population genetic consequences of habitat frag-
mentation for plants. Trends in Ecology and Evolution 11: 413–418.
Young A., Merriam H.G. and Warwick S.I. 1993. The effects of forest fragmentation on genetic
variation in Acer saccharum Marsh. (sugar maple) populations. Heredity 71: 277–289.

[291]
Biodiversity and Conservation (2006) 15:1353–1374
Springer 2006
DOI 10.1007/s10531-005-5394-9

-1

Diversity patterns in the flora of the Campo-Ma’an

rain forest, Cameroon: do tree species tell it all?

M.G.P. TCHOUTO1,*, W.F. DE BOER2, J.J.F.E. DE WILDE3 and

L.J.G. VAN DER MAESEN3
1
Limbe Botanic Garden, BP 437 Limbe, Cameroon; 2Resource Ecology Group, Wageningen Uni-
versity, Bornsesteeg 69, 6708 PD Wageningen, The Netherlands; 3Biosystematics Group, Wageningen
University, Generaal Foulkesweg 37, 6703 BL Wageningen, The Netherlands; *Author for
correspondence (e-mail: [email protected])

Received 9 February 2004; accepted in revised form 31 March 2005

Key words: Biodiversity, Cameroon, Campo-Ma’an, Central Africa, Conservation, Endemic spe-
cies, Forest refuge, Plant diversity, Tropical rain forest

Abstract. This study describes diversity patterns in the flora of the Campo-Ma’an rain forest, in
south Cameroon. In this area, the structure and composition of the forests change progressively
from the coastal forest on sandy shorelines through the lowland evergreen forest rich in Cae-
salpinioideae with Calpocalyx heitzii and Sacoglottis gabonensis, to the submontane forest at higher
elevations and the mixed evergreen and semi-deciduous forest in the drier Ma’an area. We tested
whether there is a correlation between tree species diversity and diversity of other growth forms
such as shrubs, herbs, and lianas in order to understand if, in the context of African tropical rain
forest, tree species diversity mirrors the diversity of other life forms or strata. Are forests that are
rich in tree species also rich in other life forms? To answer this question, we analysed the family and
species level floristic richness and diversity of the various growth forms and forest strata within 145
plots recorded in 6 main vegetation types. A comparison of the diversity within forest layers and
within growth forms was done using General Linear Models. The results showed that tree species
accounted for 46% of the total number of vascular plant species with DBH ‡1 cm, shrubs/small
trees 39%, climbers 14% and herbs less than 1%. Only 22% of the diversity of shrubs and lianas
could be explained by the diversity of large and medium sized trees, and less than 1% of herb
diversity was explained by tree diversity. The shrub layer was by far the most species rich, with both
a higher number of species per plot, and a higher Shannon diversity index, than the tree and the
herb layer. More than 82% of tree species, 90% of shrubs, 78% of lianas and 70% of herbaceous
species were recorded in the shrub layer. Moreover, shrubs contributed for 38% of the 114 strict
and narrow endemic plant species recorded in the area, herbs 29%, trees only 20% and climbers
11%. These results indicate that the diversity of trees might not always reflect the overall diversity
of the forest in the Campo-Ma’an area, and therefore it may not be a good indicator for the
diversity of shrubs and herbaceous species. Furthermore, this suggests that biodiversity surveys
based solely on large and medium sized tree species (DBH ‡10 cm) are not an adequate method for
the assessment of plant diversity because other growth form such as shrubs, climbers and herbs are
under-represented. Therefore, inventory design based on small plots of 0.1 ha, in which all vascular
plants with DBH ‡1 cm are recorded, is a more appropriate sampling method for biodiversity
assessments than surveys based solely on large and medium sized tree species.

Introduction

In a large, heterogeneous and structurally complex forest ecosystem such as the

Campo-Ma’an tropical rain forest, selection of the most appropriate methods

[293]
1354

for the assessment of plant biodiversity is a difficult matter. So far, many

botanical biodiversity studies in tropical rain forest are often limited to tree
species (mainly medium and large trees, or for some cases trees with DBH
‡10 cm) which are assumed to reflect the forest floristic composition and
physical structure (Letouzey 1968; Reitsma 1988; Hart et al. 1989; Mosango
1990; Koubouana 1993; Wolter 1993; Lejoly 1995a, b; Newbery and Gartlan
1996; White 1996; Sonké 1998; Sonké and Lejoly 1998; van Valkenburg et al.
1998). Moreover, for most of these studies tree species accounted for more than
50% of the overall species composition. This traditional approach of forest
inventory might not be sufficient for biodiversity assessment because other taxa
belonging to other life forms such as shrubs, small trees, woody lianas, her-
baceous climbers, herbs and epiphytic flora are not or under-represented.
Furthermore, it has been shown in Central and West Africa that many plant
species of high conservation value such as endemic and rare species are shrub
and herbaceous species (Letouzey 1968, 1985; Robbrecht 1996; Sosef 1996;
Achoundong 2000; Cable and Cheek 1998).
However, during the last two decades shrubs, herbs and climbers are pro-
gressively being taking into consideration during biodiversity assessment,
forest dynamic and ecological studies (Gentry and Dodson 1987; Poulsen and
Balslev 1991; Valencia et al. 1994; Balslev et al. 1998; Condit et al. 2000). But,
there is still a gap in knowledge regarding their contributions in the overall
vascular plant species diversity in tropical rain forests. Some work has been
done in this respect in Iquitos, Colombia and Guyana (Gentry 1988a, b; Du-
ivenvoorden and Lips 1995; ter Steege 2000). This study is the first attempt to
study the diversity patterns in the flora of a Central African tropical rain forest,
and the contribution of the different plant layers to the total species diversity.
We will analyse the diversity of the flora in the Campo-Ma’an rain forest, and
test whether there is a correlation between tree species diversity and diversity of
other growth forms such as shrubs, herbs and lianas. This will help us to
understand if, in the context of African tropical rain forest, tree species
diversity tells it all.

Methods

Study area

The study was conducted in the Campo-Ma’an rain forest in south Camer-
oon. The site covers about 7700 km2 and is located between latitudes 210¢–
252¢ N and longitudes 950¢–1054¢ E. The Campo-Ma’an area is a Tech-
nical Operational Unit (TOU) that comprises a National Park, five forest
management units, two agro-industrial plantations, and a multi-uses zone.
Following the FAO classification system, soils in the Campo-Ma’an area are
generally classified as Ferrasols and Acrisols (Franqueville 1973; Muller 1979;
van Gemerden and Hazeu 1999). They are strongly weathered, deep to very

[294]
 1355

deep and clayey in texture (except at the seashores and in river valleys where
they are mainly sandy), acid and low in nutrients with pH (H2O) values
generally around 4. The topography ranges from undulating to rolling in the
lowland area, to steeply dissect in the more mountainous areas. In the
Campo area, altitudes are mostly low, ranging from sea level to about 500 m.
In the eastern part, which is quite mountainous, the altitude varies between
400 and 1100 m and the rolling and steep terrain brings about a more var-
iable landscape.
The area has a typical equatorial climate with two distinct dry seasons
(November–March and July–mid-August) and two wet seasons (April–June
and mid-August–October). The average annual rainfall generally decreases
with an increasing distance from the coast, ranging from 2950 mm/year in
Kribi and 2800 mm in Campo to 1670 mm in Nyabissan in the Ma’an area.
The Ma’an region has significantly less rainfall than other areas. The average
annual temperature is about 25 C and there is little variation between years.
The hydrography of the area shows a dense pattern with many rivers, small
river basins, fast-flowing creeks and rivers in rocky beds containing many
rapids and small waterfalls. Generally, the area has a low population density of
about 10 inhabitants per km2 and is sparsely populated (ca. 61,000 inhabitants)
with most people living around Kribi, along the coast, and in agro-industrial
and logging camps (ERE Développement 2002; de Kam et al. 2002). Despite
the low population density, there are few employment opportunities. The local
people are very poor and so far rely solely on the forest resources to meet their
basic needs. As a result, local pressure on the Campo-Ma’an rain forest is
increasing and there are several activities that are carried out in the area with
varying ecological impacts on the forest ecosystem. These activities include
agriculture, logging, poaching and hunting.

Field sampling

After a study of satellite images, topographic and vegetation maps, a recon-

naissance trip was carried out in the study area to identify representative and
homogeneous vegetation types to be sampled. Sampling sites were selected on
the basis of physical and human factors such altitude, slope, rainfall, soils, the
proximity to the sea, and the degree of forest use. Sampling was carried out in
small plots of 0.1 ha (50 m · 20 m) at irregular intervals along a line transect
from a random starting point, In total 145 plots covering 14.5 ha were
established in undisturbed forests or matured secondary forests within 6 main
vegetation types ranging from coastal forest, swamp, lowland evergreen forest,
to submontane forest at higher elevations (800–1100 m above sea level).
Twenty two (22) plots were established in coastal forest, 26 in the lowland
forest rich in Caesalpinioideae, 39 in the lowland forest rich in Calpocalyx
heitzii and Sacoglottis gabonensis, 39 in mixed evergreen and semi-deciduous
forest, 14 in the submontane forest and 5 in swamps proportion to their area

[295]
1356

coverage. Most of the plots were located in the National Park and the forest
management units, which are less affected by human activities.
In each 0.1 ha plot, all trees, shrubs, herbs and lianas with DBH ‡1 cm were
measured, recorded and identified as far as possible. For unknown species, a
voucher specimen was collected. Herbaceous species and seedlings of trees,
shrubs and climbers were sampled in subplots of 5 m · 5 m each that were
established in the 0.1 ha plots. These subplots were not used for the analyses,
the output was only used to illustrate the contribution of the ground layer and
herbaceous species when all vascular plant species are included in the floristic
assessment of the forest.

Data analysis

The analysis focused on family and species level floristic richness within the
various life forms and forest strata recorded in the 145 plots. In this study tree
layer comprised all vascular plant species with DBH ‡10 cm, shrub layer
(1.5 cm £ DBH < 10 cm) and herbaceous layer (1 cm £ DBH < 1.5 cm).
Diversity was measured by recording the number of species and their relative
abundance in the different plots and vegetation types. This study focused on
the a diversity (species richness), which is defined as the number of species
within a chosen area, given equal weight to each species, and the b diversity,
which is the difference in species diversity between areas or communities
(Magurran 1988; Kent and Coker 1992; Bisby 1995). b diversity was quantified
with the Shannon diversity index (H¢) using all individuals above 1 cm DBH
and all species per plot. Phytosociological parameters (relative density and
relative frequency) and Shannon diversity index were calculated following
Whittaker (1975), Kent and Coker (1992) and Magurran (1988). The SPSS
package version 10.0 for Windows was used for statistical analyses. The
Spearman’s correlation test was used to correlate the species richness and
diversity between the various growth forms and forest layers. We compared
species diversity within forest layers and within growth forms using a General
Linear Model (GLM) followed by a Tukey Multiple Comparison test
(p<0.05).

Results

General patterns of species richness within forest types

A total of 76360 trees, shrubs, climbers and other vascular plants with DBH
‡1 cm was recorded in 145 plots of 0.1 ha each in the various vegetation types.
They belonged to 1112 species, 420 genera and 97 families. In addition, 759
species of vascular plants (herbs, hemi-epiphytes, shrubs and seedlings of tree
species) belonging to 101 families and 327 genera were recorded in the subplots

[296]
 1357

Table 1. Summary of the number of species, number of families, number of stems/ha and Shannon
diversity (H¢) recorded in each vegetation type for all vascular plants with DBH ‡ 1 cm.

Vegetation types No of No of No of No of Shannon diversity

plots species families stems/ha index (H¢)

Coastal forest (mangroves excluded) 22 381(78–140) 69 6208 4.73

Swamps 5 293 (18–108) 58 5798 4.58
Forest rich in Caesalpinioideae 26 468 (93–147) 68 6033 5.16
Forest rich in Calpocalyx heitzii 39 416 (81–143) 65 5990 4.93
and Sacoglottis gabonensis
Mixed evergreen and 39 441 (63–145) 66 5867 4.77
semi-deciduous forest
Submontane 14 412 (79–148) 64 6912 5.14
Minimum and maximum values are given between brackets.

of 5 m · 5 m each located within the 0.1 ha plots. Overall, 1471 species of

vascular plants, including ferns and fern allies belonging to 542 genera and 126
families were recorded. More than 73% of all specimens collected were iden-
tified at species level, 23% at generic level, 3% at family level and 1%
unidentified. The number of stems/ha for all vascular plants ‡1 cm DBH varied
from 5798 in swamps to 6912 in the submontane forest (Table 1). The number
of species/ha for all vascular plants‡1 cm varied from 293 in swamps to 468 in
the lowland evergreen forest rich in Caesalpinioideae. The Shannon diversity
(H¢) varied from 4.58 in coastal forests to 5.16 in forests rich in Caesalpi-
nioideae. More than 57% of the plots have above 100 species/0.1 ha and a
Shannon diversity (H¢)>4 with the most diverse, and species rich plots located
in the submontane forests, forests rich in Caesalpinioideae, and forests rich in
Calpocalyx heitzii and Sacoglottis gabonensis.

Floristic composition and diversity within forest strata

The number of stems/ha and the number of vascular plant species per plot were
generally higher in the shrub layer compared to the herbaceous layer and the
tree layer. The number of stems/ha (Table 2) in the shrub layer varied from
3914 (mixed evergreen and semi-deciduous forest) to 4572 (coastal forest), in
the herbaceous layer from 905 (swamps) to 1963 (submontane forest), and in
the tree layer from 489 (coastal forest) to 785 stems/ha (submontane forest).
The number of species in the shrub layer varied from 231 species (swamps) to
413 (mixed evergreen and semi-deciduous forest), in the herbaceous layer from
99 (swamps) to 229 (Calpocalyx heitzii and Sacoglottis gabonensis forest) and in
the tree layer from 100 (swamps) to 183 species (submontane forest).
In terms of Shannon diversity (H¢), the shrub layer was the most diverse
followed by the tree and herbaceous layers (Table 2). The Shannon diversity
varied in the shrub layer from 4.39 (swamps) to 5.13 (forest rich in Cae-
salpinioideae), in the tree layer from 3.82 (swamps) to 4.83 (forest rich in

[297]
 1358
Table 2. Summary of the number of species, number of families, number of stem/ha and Shannon diversity (H¢) recorded in the tree, shrub and herbaceous
layers for each vegetation types for all vascular plants with DBH ‡ 1 cm.

Floristic composition Forest types

Coastal Swamps Forest rich in Forest rich in Mixed evergreen Submontane

forest Caesalpinioideae Calpocalyx and Sacoglottis and semi-deciduous forest
forest

Tree layer: DBH‡10 cm

No. of stems/ha 489 741 586 603 562 785
No. of species 147 100 181 143 181 183
No. of families 43 37 48 45 48 44
Shannon diversity index (H¢) 4.62 3.82 4.83 4.26 4.70 4.75
Shrub layer: 1.5 cm £ DBH<10 cm
[298]

No. of stems/ha 4572 4152 4219 4316 3914 4164

No. of species 332 231 399 349 413 344
No. of families 69 53 66 57 65 53
Shannon diversity index (H¢) 4.63 4.39 5.13 4.82 4.93 5.05
Herb layer: 1 £ DBH<1.5 cm
No. of stems/ha 1147 905 1228 1071 1391 1963
No. of species 190 99 225 229 217 225
No. of families 53 34 54 47 52 51
Shannon diversity index (H¢) 4.20 3.76 4.58 4.65 3.97 4.66
Total no. of stems/ha 6208 5798 6033 5990 5867 6912
Total no of species 381 293 468 416 441 412
Total no of families 69 58 68 65 66 64
Shannon diversity index (H¢) 4.73 4.58 5.16 4.93 4.77 5.14
Note that species may overlap within forest strata.
 1359

Table 3. Contribution to the total species richness from the various forest strata in 6 main veg-
etation types.

Forest types Total no. Forest strata

of species
Tree layer Shrub layer Herbaceous
DBH ‡10 cm 1.5 cm £ DBH layer 1 £ DBH
<10 cm <1.5 cm

Species % Species % Species %

Coastal forest 381 147 39 332 87 190 50

Swamps 293 100 34 231 79 99 34
Forest rich in Caesalpinioideae 468 181 41 399 90 225 51
Forest rich in Calpocalyx heitzii 416 143 34 349 84 229 55
and Sacoglottis gabonensis
Mixed evergreen and 441 181 39 413 88 217 46
semi-deciduous forest
Submontane forest 412 183 44 344 83 225 55
Note that some species may overlap within forest strata.

Caesalpinioideae) and in the herbaceous layer from 3.76 (swamps) to 4.66

(submontane forest). The Shannon diversity (H¢) was significantly different
among forest layers (F2, 26.7 = 38.905, p<0.001) when correcting for dif-
ferences in vegetation types, by including vegetation type as a random factor
in a generalized linear model (F10, 20 = 6.605, p<0.001). A Tukey multiple
comparison test showed that all layers were significantly different from each
other (p<0.05) with the shrub layer having the highest mean value
(H¢ = 3.57) and the herbaceous layer the lowest (H¢ = 2.73). Species richness
was also significantly different among forest layers (F2, 24.2 = 151.28,
p<0.001) when correcting for differences in vegetation types (GLM, with
vegetation type as a random factor, F10, 20 = 4.412, p<0.01), with highest
values also recorded in the shrub layer.
The shrub layer generally contributed more than 80% to the total number of
species recorded in each vegetation type (Tables 2 and 3), followed by the
herbaceous layer (40%) and tree layer (35%). There was a significant positive
correlation between the total number of vascular plant species per plot in the
tree layer and that of the shrub (F1, 143 = 24.059, R2 = 0.144, p< 0.001) and
herbaceous (F1, 143 = 15.702, R2 = 0.099, p<0.001) layers. In terms of the
Shannon Diversity Index, there was also a significant positive correlation be-
tween the diversity of the tree layer and that of the shrub and to a lesser extent
with that of the herbaceous layers (Figures 1 and 2).

Floristic composition and diversity by life forms

Tree species richness was relatively higher than that of shrubs, lianas and
herbs (Table 4). The total number of species varied from 172 species

[299]
1360

Figure 1. Correlation between the Shannon diversity (H¢) of all vascular plant species recorded in
the tree layer and that of the shrub/small tree layer within 145 plots of 0.1 ha each.

(swamps) to 237 (forest rich in Caesalpinioideae) for trees, 71 (swamps) to

164 (forest rich in Caesalpinioideae) for shrubs, 42 (swamps) to 63 (forest
rich in Caesalpinioideae) for lianas and 4 (submontane) to 9 (mixed evergreen
and semi-deciduous forest) for herbs. About 63% of the total number of tree
species was recorded in the tree layer, 82% in the shrub layer and 41% in
the herbaceous layer. Less than 10% of the total number of shrub/small tree
species was found in the tree layer, 90% in the shrub layer and 62% in the
herbaceous layer. No herbaceous species was found in the tree layer, 70%
was recorded in the shrub layer and 97% in the herb layer. The contribution
of herbaceous species was very low in the 0.1 ha plots since many herbaceous
species do not have a DBH ‡1 cm. However, their contribution was well
illustrated in the 5 m·5 m subplots where there was a considerable increase in
the number of herbaceous species (Table 5), moving from 25 species (in 145
plots of 0.1 ha each) to 257 species (in 136 subplots of 25 m2 each covering
0.34 ha).
Trees and shrubs were the most diverse growth forms followed by lianas and
herbaceous species (Table 4). The Shannon diversity (H¢) of trees varied from
3.96 (swamps) to 4.67 (submontane forest), for shrubs from 3.32 (coastal
forests) to 4.66 (submontane forest) and for herbaceous species from 0.53
(coastal forests) to 1.03 (forest rich in Calpocalyx and Sacoglottis).

[300]
 1361

Figure 2. Correlation between the Shannon diversity (H¢) of all vascular plant species recorded in
the tree layer and that of the herbaceous layer within 145 plots of 0.1 ha each.

There was a significant positive correlation between the number of large and
medium sized tree species and that of shrubs/small trees (F1, 143 = 112.033,
R2 = 0.439, p<0.001) and woody climbers (F1, 143 = 26.986, R2 = 0.159,
p<0.001). There was also a significant positive correlation between the
diversity of large and medium sized trees and that of shrubs/small trees and
woody climbers (Table 6; Figures 3 and 4). The correlations between the
diversity/species richness of large and medium sized trees and that of the
herbaceous species were not significant (F1, 143 = 0.001, R2 = 0.00002,
p = 0.975 for Shannon diversity and F1, 143 = 0.0387, R2 = 0.0003, p = 0.844
for species richness). The Shannon diversity (H¢) was significantly different
among the various growth forms (F3, 34.6 = 151.290, p<0.001) when correcting
for differences in vegetation types (GLM: F10, 30 = 2.727, p<0.01 for vege-
tation type included as a random factor). A Tukey multiple comparison test
showed that all growth forms were significantly different from each other
(p<0.05) with trees having the highest mean value (H¢ = 3.37) and the lowest
(H¢ = 0.24) for the herbaceous species. The species richness was also signifi-
cantly different between growth forms (GLM: F3, 33.2 = 221.889, p<0.001 for
layer and F10, 30 = 2.973, p<0.01 for vegetation type included as a random
factor), with similar relative difference.

[301]
 1362
Table 4. Summary of the total number of species for the various growth forms recorded in 6 main vegetation types for all vascular plants with DBH ‡1 cm.

Growth form Forest types

Coastal Swamps Forest rich in Forest rich in Mixed evergreen Submontane

Forest Caesalpinioideae Calpocalyx and and semi-deciduous forest
Sacoglottis forest

Total no. of tree species 207 172 237 211 222 216
Tree layer: DBH ‡ 10 cm 128 82 158 126 156 162
Shrub layer: 1.5 cm £ DBH <10 cm 177 132 198 169 192 173
Herb layer: 1 £ DBH <1.5 cm 94 44 97 100 86 102
Shannon diversity (H¢) 4.26 3.96 4.47 4.13 4.44 4.67
Total no. of shrubs/small tree species 127 71 164 155 161 152
Tree layer: DBH ‡10 cm 11 10 20 10 13 12
[302]

Shrub layer: 1.5 cm £ DBH <10 cm 113 58 153 143 152 137
Herb layer: 1 £ DBH <1.5 cm 76 37 105 104 100 101
Shannon diversity (H¢) 3.32 4.39 4.26 4.08 4.32 4.66
Total no. of herbaceous species 8 7 5 4 9 4
Tree layer: DBH ‡10 cm 0 0 0 0 0 0
Shrub layer: 1.5 cm £ DBH <10 cm 7 4 3 3 6 3
Herb layer: 1 £ DBH <1.5 cm 8 6 5 4 8 4
Shannon diversity (H¢) 0.53 0.54 0.78 1.03 0.85 0.90
Total no. of liana species 46 42 63 50 59 43
Tree layer: DBH ‡10 cm 8 7 7 8 12 6
Shrub layer: 1.5 cm £ DBH <10 cm 40 39 60 8 54 36
Herb layer: 1 £ DBH <1.5 cm 16 11 25 26 23 21
Shannon diversity (H¢) 3.13 3.29 3.34 3.27 3.65 3.19
Herbaceous species include herbs, herbaceous climbers and hemi-epiphytes. Lianas include small and large woody climbers. Note that species may overlap
within forest strata.
 Table 5. Summary of the number of species of the various growth forms recorded in 136 vegetative subplots of 5 m · 5 m each covering 0.34 ha.

Floristic composition Forest types

Coastal Swamps Forest rich in Forest rich in Calpocalyx Mixed evergreen and Submontane
forest Caesalpinioideae and Sacoglottis semi-deciduous forest forest
[303]

Trees 27 6 35 38 37 26
Shrubs/small trees 87 14 148 102 134 74
Herbs 88 14 142 98 105 77
Lianas 30 3 36 33 32 11
Total No of species 232 37 368 271 308 188
Total No of families 67 27 76 63 68 47

1363
1364

Table 6. Spearman correlation coefficients between the Shannon diversity (H¢) of the various
growth forms recorded for all vascular plants with DBH ‡1 cm in 145 plots of 0.1 ha each.

Life forms Large trees Shrubs/small trees Herbs Lianas

Large trees 1
Shrubs/small trees 0.29** 1
Herbs 0.13 0.20* 1
Lianas 0.24** 0.20* 0.18* 1
Spearman correlation is significant at the 0.05 level for * and at the 0.01 level for **.

Figure 3. Correlation between the Shannon diversity (H¢) of large and medium sized tree species
and that of the shrub/small tree species within 145 plots of 0.1 ha each.

Species richness within families

Overall, shrubs/small trees have the highest number of families (75) followed
by medium trees (61), herbs and hemi-epiphytes (58), large trees (54) and
woody climbers (37). Rubiaceae was by far the most species rich family (204
species) followed by Euphorbiaceae (88), Leguminosae-Caesalpinioideae (85),
Annonaceae (63), Sterculiaceae (50), Apocynaceae (47) and Sapindaceae (40).
Leguminosae (especially the subfamily Caesalpinioideae) was the dominant
family for the large trees species (DBH ‡30 cm) in terms of relative density
and frequency in the Campo-Ma’an area (Table 7). Dominant large tree
species included Calpocalyx heitzii, Desbordesia glaucescens, Erythrophleum
ivorensis, Lophira alata, Lovoa trichilioides, Pycnanthus angolensis, Sacoglottis

[304]
 1365

Figure 4. Correlation between the Shannon diversity (H¢) of large and medium sized tree species
and that of the climbers within 145 plots of 0.1 ha each.

gabonensis and Terminalia superba. Common shrubs and small tree species
included many species of the genera Cola, Diospyros, Drypetes, Psychotria,
and Rinorea. Common large woody climber species were from the genera
Agelaea, Dichapetalum, Combretum, Millettia, and Strychnos. The most
important herb species were of the genera Begonia, Culcasia, Dorstenia,
Geophila, Haumania, Hymenocoleus, Mapania, Marantochloa, Microcalamus
and Palisota.

Discussion

General patterns of species composition and diversity within forest types

The Campo-Ma’an area is dominated by the lowland evergreen rain forest rich
in Caesalpinioideae and is characterized by a rich and diverse flora Letouzey
1968 and 1985; Thomas and Thomas 1993; Tchouto 2004). The number of
species/ha for all vascular plants ‡1 cm recorded varied from 293 in swamp
forest to 468 in the lowland evergreen forest rich in Caesalpinioideae. The
Shannon diversity (H¢) varied from 4.73 in coastal forests to 5.16 in forests rich
in Caesalpinioideae. The explanation for the high level of species richness and
diversity might stem partly from the fact that the area is part of a series of
postulated rain forest refugia in Central and West Africa (Hamilton 1982;

[305]
1366

Table 7. Five most important families/subfamilies recorded for the various growth forms in 145
forest plots (0.1 ha each) and the vegetative subplots. Note that species may overlap between size
classes.

Family/subfamily Relative density Relative frequency

Canopy tree: DBH ‡30 cm (286 species and 54 families)

Leguminosae-Caesalpinioideae 15.00 9.03
Olacaceae 9.72 8.49
Burseraceae 8.23 6.00
Euphorbiaceae 7.94 6.93
Myristicaceae 6.44 5.22
Medium sized trees: 10 cm £ DBH<30 cm (316 species and 61 families)
Olacaceae 14.47 3.23
Leguminosae-Caesalpinioideae 13.25 2.96
Euphorbiaceae 11.19 3.01
Annonaceae 8.48 2.45
Burseraceae 5.70 1.94
Shrub/small trees: 1.5 cm £ DBH<10 cm (389 species and 75 families)
Euphorbiaceae 13.60 3.59
Rubiaceae 8.50 3.64
Leguminosae-Caesalpinioideae 7.01 3.59
Olacaceae 6.33 3.51
Ebenaceae 6.07 3.41
Small and large climbers (158 species and 37 families)
Dichapetalaceae 26.65 12.29
Palmae 13.48 8.03
Connaraceae 12.24 11.06
Loganiaceae 8.99 8.79
Celastraceae 8.30 8.51
Herbs and hemi-epiphytes (257species and 58 families)
Araceae 20.43 11.32
Marantaceae 12.69 10.78
Gramineae 8.94 8.20
Commelinaceae 8.42 8.91
Cyperaceae 7.34 7.13

White 1983; Maley 1987 and 1989; Sosef 1994). Furthermore, the Campo-
Ma’an forest falls within the Guineo-Congolian Centre of Endemism (White
1983; Gartlan 1989; Thomas and Thomas 1993; Davis et al. 1994). It is situated
in the middle of the rich Biafran forest type that extends from southeast
Nigeria to Gabon and the Mayombe area in Congo (White 1983; Letouzey
1968 and 1985) and shares with these sites the overall characteristic of lowland
evergreen rain forest with some semi-deciduous species.
In agreement with Hubbell and Foster (1983, 1986c), many species
showed no apparent distributional biases with respect to habitat boundaries.
Nevertheless, some species were strongly positively or negatively associated
with specific habitats such as swamps, hilltops, riverbank and disturbed
forests. Species composition was influenced by rainfall, altitude, soils, the
proximity of the sea, and the level of human disturbance that contributed to

[306]
 1367

several environmental hostilities that may have limited the influx of plant
species and weaken competitive abilities of poorly adapted species (Tchouto
2004). The total number of vascular plant species was relatively influenced
by the proximity of the sea and rainfall, thus resulting in a gradual vari-
ation in dominant species and an increase in species richness with increasing
annual rainfall and distance from the sea. There was also an increase in
deciduous and semi-deciduous elements with decreasing annual rainfall.
Forest at higher altitudes between 800 and 1100 m above sea level appeared
to be relatively more species-rich than the disturbed lowland and the coastal
forests on sandy shorelines. In the undisturbed lowland evergreen forest rich
in Caesalpinioideae and the submontane forest, more than 93% of the
0.1 ha plots had above 100 species. Swamps and seasonal flooded forests on
hydric soils with poor drainage conditions and low nutrient concentrations
were species-poor. This result is in full agreement with several studies that
have shown that permanent water logging in soils is a main factor limiting
vascular plant species alpha diversity in Neotropical swamps (Duivenvoor-
den and Lips 1995; Gartlan et al. 1986; Newbery et al. 1996; Sheil et al.
2000).

Floristic composition and diversity within forest strata

In general, the difference in species composition and diversity within forest

strata or life forms followed the same trend within the various plots and
vegetation types, with the shrub layer being the most diverse and species-rich
layer, followed by the herbaceous and tree layers respectively. The number
of stems/ha, species diversity (H¢), and number of vascular plant species/ha
were generally higher in the shrub layer than in the herbaceous and tree
layers. This must stem partly from the fact that the shrub layer is made up
of many life forms such as shrubs, small trees, young large trees, woody
climbers, small herbaceous and woody climbers, tall herbs, and hemi-epi-
phytes, which are not found in the upper tree layer. In terms of species
richness, the shrub layer generally contributes to more than 80% of the total
number of species recorded in each forest type, followed by the herbaceous
layer (40%), and the tree layer (35%). Except for some species-poor plots
recorded in swamps and disturbed forests, this trend was noticed within
most of the plots. It is worth mentioning that there was some floristic
overlap between the different forest layers, since more than 50% of the
species occurred in more than one stratum. Many large tree and woody liana
species were found in all the strata depending on their development stage
(seedlings, juveniles and immature or mature individuals). There was a po-
sitive correlation between the diversity of the tree layer and that of the shrub
layer. This is partly attributed to the fact that a high proportion of the
immature large tree and liana species of the upper tree layer was also
recorded in the shrub layer.

[307]
1368

Floristic composition and diversity by life forms

A total of 533 tree species with DBH ‡10 cm was recorded in 145 plots of
0.1 ha each, with an average of 12–50 species/0.1 ha. Most plots in the lowland
evergreen forest and the submontane forest had above 35 tree species/0.1 ha.
These results are comparable to that of 15–60 species/0.1 ha found by Gentry
(1988b) in the tree species rich Mishana and Yanamono plots near Iquitos.
Balslev et al. (1998) found 307 tree species with DBH ‡10 cm in 1 ha plots of
forest in Ecuador and Condit et al. (2000) between 673 and 996 tree species in
two 50-ha plots in Malaysia. In this study, shrubs and small trees were by far
the most species rich growth forms, more than 756 species were recorded
among treelets and shrubs with DBH £ 2.5 cm, and 793 species for treelets
between 2.5<DBH <10 cm in 145 0.1-ha plots, with an average of 13–101
species and 10–90 species/0.1 ha, respectively. Gentry and Dodson (1987)
found between 37 and 137 species/0.1 ha among treelets with DBH < 2.5 cm,
and 38–87 species among treelets between 2.5<DBH <10 cm in 3 plots in
western Ecuador. These results correspond to some extent to the general trend
of tree and shrub species densities recorded in other species-rich Neotropical
rain forests (Gentry 1988a, b; Valencia et al. 1994; Duivenvoorden and Lips
1995; Condit et al. 2000; ter Steege 2000). This suggests that the world’s highest
level of tree and shrub species richness are not only confined to the Amazonian
and south east Asian forests, but may be also found in the Atlantic Biafran
lowland evergreen rain forest in Central Africa.
The species richness of trees, treelets and shrubs was higher compared to that
of the climbers and herbs. Species richness among terrestrial herbs and climbers
were relatively low. A total of 161 species of vascular climbers, and 29 species
of herbs were recorded with an average of 2–23 species of climbers/0.1 ha and
0–5 herb species/0.1 ha with DBH ‡ 1 cm. Poulsen and Balslev (1991) found
25–41 obligate herb species in 0.1 ha in an upland forest in Ecuador, and
Duivenvoorden and Lips (1995), 21–25 (excluding palm) in Colombia. The
average of 9.7 individuals/ha (3–18 individuals/ha) recorded for large lianas
with DBH ‡10 cm in the Campo-Ma’an area, falls within the pan-tropical
range presented by Rollet (1974) for large lianas. Climbers represented 14% of
the total number of vascular plant species, indicating that climbers contribute
considerably to forest plant species richness in the area. Amazonian and
Australian moist forests are reported to be richer in climber species than
African rain forests since they contribute up to 25% of the total vascular plant
species richness (Putz and Chai 1987; Nabe-Nielsen 2001; Schnitzer and
Bongers 2002), however, the Campo-Ma’an area appeared to be one of the
richest sites for climbers in tropical Africa. The explanation for the high
number of climbers species might stem partly from the fact that the area falls
within the Congo Basin, which is known to be the richest in climbers in Central
and West African (Parren 2003).
The striking low number of herbs in the present study is partly due to the
fact that many herbaceous species were below 1 cm DBH and could not be

[308]
 1369

sampled within the 0.1 ha plots. As a result, herbaceous species contributed less
than 1% of the total vascular plant species count. However, in the 136 subplots
of 5 · 5 m, there was a considerable increase in herb species, from 25 species in
145 plots of 0.1 ha to 257 species in the 5 · 5 m plots (Table 5). A stratified
sampling approach, with larger plots for larger individuals (e.g. 1 ha plot for
DBH ‡ 10 cm) divided into small subplots for smaller individuals (e.g. 0.1 ha
for DBH ‡ 1 cm and 0.01 ha for DBH <1 cm) seems therefore the best
approach to measure species richness in tropical rainforests.
Tree species appeared to be more diverse than shrubs, climbers and herbs.
Although there was a significant positive correlation between the diversity of
trees and that of the shrubs/small trees and woody climbers, the correlations
between the diversity of trees and that of the herbaceous species were not
significant. Moreover, there was a significant positive correlation between the
diversity of shrubs/small trees and that of herbaceous species. This is partly due
to the fact that most of the shrubs, small trees and herbaceous plants are
understorey species that live under the same physiological and biological
conditions. They are either shade bearers or non-pioneer light demanding
species that require little sunlight for survival. More than 40% of the herba-
ceous layer species contribution came from shade hemi-epiphytes that are often
restricted to the lower trunk of shrubs and small trees. Hemi-epiphytic species
of the genera Culcasia and Cercestis (Araceae) and several fern genera such as
Lomariopsis, Hymenophyllum and Trichomanes were very common (Table 7).
It is worth mentioning that species richness and diversity was higher in
undisturbed forest types than in the coastal forests where there was a pro-
nounced human disturbance. With the exception of swamps and mangroves,
the coastal forest types were floristically poorer with a lower diversity index
(H¢) than other forest types (Tables 1 and 2). The degree of human activities
seems to have an effect on species composition and decrease species diversity
within the various forest types. Plots located in secondary forests, past logging
concessions and plots near forest edges were more disturbed, and characterized
by a high number of herbaceous, climber and pioneer species. Furthermore,
undisturbed forests with open canopy, such as the mixed evergreen and semi-
deciduous forest in the drier Ma’an area, were also characterized by an in-
creased number of semi-deciduous, herbaceous and climber species.

Does tree diversity tell it all?

Many botanical studies in tropical rain forest emphasise the structural aspect
of the forest, assuming that the diversity of large and medium sized trees (DBH
‡ 10 cm) reflect the overall diversity of the forest. When comparing the tree
diversity and floristic composition in 6 different forest types in the Campo-
Ma’an area, we noticed that tree species accounted for 46% of the total vas-
cular plant species with DBH ‡ 1 cm, shrubs/small trees 39%, climbers 14%
and herbs less than 1%. Only 22% of the diversity of shrubs and lianas could

[309]
1370

be explained by the diversity of large and medium sized trees, and less than 1%
of herb diversity was explained by the tree diversity (Figures 3 and 4). A higher
percentage of tree, shrub and climber species occurred in the shrub layer than
in the tree and herbaceous layers. Moreover, only 63% of the tree species were
recorded in the tree layer against 82% in the shrub layer. Less than 10% of the
total number of shrub/small tree species was found in the tree layer compared
to 90% in the shrub layer. Furthermore, shrubs contributed for 38% to the 114
strict and narrow endemic plant species recorded in the area, herbs 29%, trees
only 20% and climbers 11% (Tchouto 2004). It is worth mentioning that,
although there was a significant positive correlation between the diversity of
trees and that of shrubs and woody climbers, the correlation between tree and
herb diversity was not significant.
This study also demonstrated that the shrub layer was by far the most
species-rich in the different plots and vegetation types. It was significantly more
diverse and species-rich than the tree and herbaceous layers. More than 82% of
tree species, 90% of shrubs, 78% of lianas and 70% of herbaceous species were
recorded in this layer. The high number of species found in this layer can be
attributed to the fact that, in addition to immature large trees and woody
climbers, the shrub layer comprises shrubs, small trees, tall herbs, small
climbers and hemi-epiphytes which are not found in the upper tree layer. This
leads to the conclusion that tree diversity does not always reflect the overall
diversity of the forest. In addition, more than 75% of plant species of high
conservation value such as endemic species are shrub and herbaceous species
(Tchouto 2004). Similar studies carried out by Duivenvoorden and Lips (1995)
in Colombia, and by ter Steege (2000) and van Andel (2001) in Guyana, have
also shown that tree diversity is not a good indicator for the diversity of shrubs
and herbs. This suggests that sampling design, based on small plots of 0.1 ha, in
which all vascular plants with DBH ‡ 1 cm are recorded, is a more appropriate
sampling method for biodiversity conservation purposes, than assessments
based solely on large and medium sized trees. Although it requires additional
effort, time and financial involvement, it provides more information on other
growth forms such as shrubs, climbers and herbs that are under-represented
when the sampling design only includes large and medium sized trees (DBH ‡
10 cm).
In our study, large and medium sized tree species richness showed a strong
positive correlation with that of lianas, indicating that in undisturbed Central
African lowland evergreen rain forests, tree species richness may predict woody
climber species richness relatively well. Similar results were obtained by ter
Steege (2000) in Guyana. This is partly explained by the fact that woody
climbers are dependent on the presence of trees for their support. Although,
large trees are reported to have a negative impact on liana density through
direct competition with the lianas for light and nutrients, their composition and
physiognomy, as well as the forest structure, are important factors influencing
the species composition and diversity of liana (Schnitzer and Bongers 2002;
Parren 2003). Considering the fact that most of the Campo-Ma’an area is

[310]
 1371

dominated by a lowland forest rich in large canopy and emergent tree species
(up to 60 m tall) that forms a fairly continuous canopy, only large non-pioneer
light-demanding climber species and small woody shade bearer climbers that
are adapted to low light understorey conditions can survive in such an envi-
ronment. The density and species richness of small light-demanding herbaceous
and woody climbers is therefore relatively low in undisturbed forests and high
at forest edges, in natural tree fall gaps, in secondary forests and in opened
forest with a discontinuous canopy.

Conclusion

There is a general perception among scientists that, in the tropical rain forest,
the diversity of large and medium sized trees (DBH ‡ 10 cm) can be used to
predict the diversity of other life forms, since, in most of these studies tree
species account for more than 50% of the overall species composition. This
study has demonstrated that the diversity of trees does not always reflect the
overall diversity of forest in the Campo-Ma’an area and, therefore, it may not
be a good indicator for the diversity of shrubs and herbaceous species. How-
ever it is a relatively good indicator for the diversity of lianas. In terms of
floristic composition, the number of tree species can be used to some extent to
predict the number of species of other growth forms. Furthermore, the shrub
layer (1.5 cm £ DBH <10 cm) was by far the most species rich in the different
vegetation types sampled and appeared to be more diverse and species-rich
than the tree and herbaceous layers. This suggests that sampling design, based
on small plots of 0.1 ha, in which all vascular plants with DBH ‡1 cm are
recorded, is a more appropriate sampling method for biodiversity conservation
purposes than assessments based solely on large and medium sized trees (DBH
‡ 10 cm). Moreover, if there are enough means, time and staff, a stratified
sampling approach, with larger plots for larger individuals (e.g. 1 ha plot for
DBH ‡ 10 cm) divided into small subplots for smaller individuals (e.g. 0.1 ha
for DBH ‡ 1 cm and 0.01 ha for DBH < 1 cm) is the best.

Acknowledgements

This study was carried out within the framework of the Campo-Ma’an Bio-
diversity Conservation and Management Project, Cameroon, and was finan-
cially supported by Tropenbos International, The Netherlands. We thank
G. Achoundong, J.M. Onana, B. Sonke, L. Zapfack and P. Mezili at the
National Herbarium, Cameroon, and F.J. Breteler, and C.C.H. Jongkind at
the Nationaal Herbarium Nederland, Wageningen University Branch, who
assisted in plant identification. The staff of Campo-Ma’an Project is also
acknowledged with gratitude for their assistance and support during the
fieldwork. Particular thanks are for my field assistants Elad Maurice and

[311]
1372

Ossele Mathilde for their enthusiastic support and cooperation. We also extend
our sincere thanks to all chiefs and village representatives, for their active
participation in the organization and collection of field data.

References

Achoundong G. 2000. Les Rinorea et l’étude des refuges forestiers en Afrique. In: Servant M. and
Servant-Vildary S. (eds), Dynamique à long terme des écosystèmes forestiers intertropicaux,
pp. 19–29.
Balslev H., Valencia R., Paz y Mino G., Christensen L.B. and Nielsen I. 1998. Species count of
vascular plants in one hectare of humid lowland forest in Amazonian Ecuador. In: Dallmeier F.
and Comiskey J.A. (eds), in Forest Biodiversity in North, Central and South America, and the
Caribbean. MAB series, 21: 585–594.
Bisby F.A. 1995. Characterization of biodiversity. In: Heywood V.H. and Watson R.T. (eds),
Global Biodiversity Assessment. UNEP, Cambridge University Press, pp. 192–294.
Cable S. and Cheek M. 1998. The plants of Mount Cameroon. A Conservation Checklist. Royal
Botanic Gardens, Kew.
Condit R., Ashton P.S., Baker P., Bunyavejchewin S., Gunatilleke S., Gunatilleke N., Hubbell S.P.,
Foster R.B., Itoh A., LaFrankie J.V., Lee S.H., Losos E., Manokaran N., Sukumar R. and
Yamakura T. 2000. Spatial patterns in the distribution of tropical tree species. Science 228: 1414–
1417.
Davis S.D., Heywood V.H. and Hamilton A.C. 1994. Centres of Plant Diversity. A Guide and
Strategy for Their Conservation 1.
de Kam M., Fines J.-P. and Akogo G. (eds) 2002. Schéma Directeur pour le developpement de
l’Unité Technique Opérationnelle de Campo-Ma’an. Cameroun. Campo-Ma’an Series 1.
Duivenvoorden J.F. and Lips J.M. 1995. A Land Ecological Study of Soils, Vegetation and Plant
Diversity in Colombia Amazonia. Tropenbos Series 12The Tropenbos Foundation, Wageningen.
ERE Développement 2002. Etude socio-économique dans l’UTO de Campo-Ma’an. Rapport final
phase 2: résultats d’enquètes auprès des ménages. SNV/Projet Campo-Ma’an.
Franqueville A. 1973. Atlas Regional Sud-Ouest 1. République du Cameroun. ORSTOM, Yao-
undé, Cameroun.
Gartlan J.S. 1989. La conservation des ecosystemes forestiers du Cameroun. IUCN.
Gentry A.H. 1988a. Change in plant community diversity and floristic composition on environ-
mental and geographical gradients. Annals of the Missouri Botanical Garden 75: 1–34.
Gentry A.H. 1988b. Trees species richness of upper Amazonian forests. Proceedings of the
National Academy of Science USA 85: 156–159.
Gentry A.H. and Dodson C. 1987. Contribution of non trees to species richness of a tropical rain
forest. Biotropica 19: 149–156.
Hamilton A.C. 1982. Environmental History of Africa: A Study of the Quaternary. Academic
Press, London.
Hart T.B., Hart J.A. and Murphy P.G. 1989. Monodominant and species-rich forests of the humid
tropics: causes for their co-occurrence. The American Naturalist 133: 613–633.
Hubbell S.P. and Foster R.B. 1983. Diversity of canopy trees in a neotropical forest and impli-
cations for conservation. In Sutton S.L., Whitemore T.C. and Chadwick A.C. (eds), in Forest:
Ecology and Management. Blackwell Science, Oxford, pp. 24–41.
Hubbell S.P. and Foster R.B. 1986c. Commonness and rarity in a neotropical forest: implications
for tropical tree conservation. In: Soule M.E. (ed.), Conservation Biology: The Science of
Scarcity and Diversity. Sinauer Associates, Massachusetts, pp. 205–231.
Kent M. and Coker P. 1992. Vegetation Description and Analysis. Belhaven Press, London.
Koubouana F. 1993. Les forêts de la vallée du Niari, Congo: Etudes floristiques et structurales.
Thèse de doctorat, Université de Paris 6.

[312]
 1373

Lejoly J. 1995a. Biodiversité végétale dans le Parc National d’Odzala, Congo. Rapport technique.
Groupement Agreco-CTFT.
Lejoly J. 1995b. Utilisation de la méthode de transects en vue de l’étude de la biodiversité dans la
zone de conservation de la forêt de Ngotto, République Centrafricaine. Rapport technique.
Projet ECOFAC, Agreco-CTFT.
Letouzey R. 1968. Etude phytogéographique du Cameroun. Le Chevalier, Paris.
Letouzey R. 1985. Notice de la carte phytogéographique du Cameroun, vol. 1–5. Institut de la
Carte Internationale de la Végétation, Toulouse.
Magurran A.E. 1988. Ecological Diversity and its Measurement. Croom Helm, London.
Maley J. 1987. Fragmentation de la forêt dense humide Africaine et extention des biotopes mon-
tagnards du Quaternaire recent. Paleoecology of Africa 18: 307–309.
Maley J. 1989. Late Quaternary climatic changes in the African rainforest: the question of forest
refuges and the major role of sea surface temperature variations. In: Lewen M. and Sarthein M.
(eds), Paleoclimatology and Paleometeology: Modern and Past Patterns of Global Atmospheric
Transport. Kluwer, Londonmeteo, pp. 585–616.
Mosango M. 1990. Contribution à l’étude botanique et biogéochimique de l’écosystème des forêts
en région équatoriale (Ile Kongolo, Zaire). Thèse de doctorat, Université Libre de Bruxelles.
Muller J.P. 1979. Carte des sols du Cameroun. Atlas de la République Unie du Cameroun.
Nabe-Nielson J. 2001. Diversity and distribution of lianas in a neotropical rain forest, Yasuni
National Park, Ecuador. Journal of Tropical Ecology 17: 1–19.
Newbery D. Mc and Gartlan J.S. 1996. A Structural Analysis of Rainforest at Korup and Douala-
Edea, Cameroon. Proceedings of the Royal Society of Edinburg 104B: 177–224.
Parren M.P.E 2003. Lianas and logging in West Africa. Tropenbos-Cameroon Series 6. PhD
Thesis, Wageningen University, The Netherlands.
Poulsen A.D. and Balslev H. 1991. Abundance and cover of ground herbs in an Amazonian rain
forest. Journal of Vegetation Science 2: 315–322.
Putz F.E and Chai P. 1987. Ecological studies of lianas in Lambir National Park, Sarawak,
Malaysia. Journal of Ecology 75: 523–531.
Reitsma J.M. 1988. Forest vegetation in Gabon. Tropenbos Technical Series 1.
Robbrecht E. 1996. Geography of African Rubiaceae with reference to glacial rain forest refuges.
In: van der Maesen L.J.G., vander Burgt X.M. and van Medenbach de Rooy J.M. (eds), The
Biodiversity of African Plants. Kluwer Academic Publishers, Wageningen, pp. 564–581.
Rollet B. 1974. La régénération naturelle dans les trouées Un processus général de la dynamique
des forêts tropicales humides. Revue Bois et Forêts des Tropiques 201: 3–34.
Schnitzer S.A. and Bongers F. 2002. The ecology of lianas and their role in forests. Trend in
Ecology and Evolution 17: 223–230.
Sheil D., Jennings S. and Savill P. 2000. Long-term permanent plot observations of vegetation
dynamic in Budongo, a Ugandan rain forest. Journal of Tropical Ecology 16: 765–800.
Sonké B. 1998. Etudes floristiques et structurales des forêts de la Reserve de Faune de Dja,
Cameroun. Thèse présentée en vue de l’obtention du grade de Docteur en sciences, option
écologie végétale. Université Libre de Bruxelles.
Sonké B. and Lejoly J. 1998. Biodiversity study in the Dja Fauna Reserve (Cameroon): using
the transect method. In: Huxley C.R., Lock J.M. and Cutler D.F. (eds), Chorology, Taxonomy
and Ecology of the Floras of Africa and Madagascar. Royal Botanical Gardens, Kew,
pp. 171–179.
Sosef M.S.M. 1994. Refuge Begonias. Taxonomy, phylogeny and historical biogeography of
Begonia sect. Loasibegonia and sect. Scutobegonia in relation to glacial rain forest refuges in
Africa, PhD Thesis, Wageningen Agricultural University.
Sosef M.S.M. 1996. Begonias and African rain forest refuges: general aspects and recent progress.
In: van der Maesen, L.J.G., van der Burgt X.M. and van Medenbach de Rooy J.M. (eds), The
Biodiversity of African Plants. Kluwer Academic Publishers, Wageningen, pp. 602–611.
Tchouto M.G.P. 2004. Plant diversity in a Central African rain forest: implications for biodiversity
conservation in Cameroon. PhD Thesis, Wageningen University.

[313]
1374

ter Steege H. 2000. Plant diversity in Guyana, with recommendations for a National Protected area
strategy. Tropenbos Series 18.
Valencia R., Balslev H. and Pazy Mino G. 1994. High tree alpha-diversity in Amazonian Ecuador.
Biodiversity and Conservation 3: 21–28.
van Andel T. 2001. Floristic composition and diversity of mixed primary and secondary forests in
northwest Guyana. Biodiversity and Conservation 10: 1645–1682.
van Gemerden B.S. and Hazeu G.W. 1999. Landscape ecological survey (1:100,000) of the Bipindi-
Akom II-Lolodorf region, Southwest Cameroon. Tropenbos-Cameroon Document 1.
van Valkenburg J.L.C.H., Ketner P. and Wilks C.M. 1998. A floristic inventory and preliminary
vegetation classification of the mixed semi-evergreen rain forest in the Minkébé region, North
East Gabon. Adansonia 20: 139–162.
White F. 1983. The Vegetation of Africa. UNESCO, Paris.
White L.J.T. 1996. Determinants of vegetation composition in the Lopé Reserve. Report to
AGRECO/CTFT, Gabon.
Whittaker R.J. 1975. Communities and Ecosystems, 2nd ed.. Macmillan, London.
Wolter F. 1993. Etude des possibilités techniques, économiques et financières d’un aménagement
des forêts tropicales denses humides de la cuvette centrale du Zaı̈re, basée sur ses capacités
naturelles. Thèse de doctorat. Université de Louvain.

[314]
Biodiversity and Conservation (2006) 15:1375–1397
Springer 2006
DOI 10.1007/s10531-005-5397-6

-1

Impacts of selective logging and agricultural clearing

on forest structure, floristic composition and diversity,
and timber tree regeneration in the Ituri Forest,
Democratic Republic of Congo

JEAN-REMY MAKANA1,2 and SEAN C. THOMAS1,*

1
University of Toronto, Faculty of Forestry, 33 Willcocks Street, Toronto, ON M5S 3B3, Canada;
2
Centre de Formation et de Recherche en Conservation Forestie`re (CEFRECOF), Epulu, Eastern
Province, Democratic Republic of Congo; *Author for correspondence (e-mail: sc.thomas@utoronto.
ca; phone: 416-978-1044; fax: 416-978-3834)

Received 16 June 2004; accepted in revised form 5 April 2005

Key words: Forest composition, Ituri Forest, Secondary forest, Selective logging, Shifting culti-
vation, Timber tree regeneration, Tree diversity

Abstract. Mature tropical forests at agricultural frontiers are of global conservation concern as the
leading edge of global deforestation. In the Ituri Forest of DRC, as in other tropical forest areas,
road creation associated with selective logging results in spontaneous human colonization, leading
to the clearing of mature forest for agricultural purposes. Following 1–3 years of cultivation,
farmlands are left fallow for periods that may exceed 20 years, resulting in extensive secondary
forest areas impacted by both selective logging and swidden agriculture. In this study, we assessed
forest structure, tree species composition and diversity and the regeneration of timber trees in
secondary forest stands (5–10 and 40 years old), selectively logged forest stands, and undisturbed
forests at two sites in the Ituri region. Stem density was lower in old secondary forests ( 40 years
old) than in either young secondary or mature forests. Overall tree diversity did not significantly
differ between forest types, but the diversity of trees ‡10 cm dbh was substantially lower in young
secondary forest stands than in old secondary or mature forests. The species composition of
secondary forests differed from that of mature forests, with the dominant Caesalpinoid legume
species of mature forests poorly represented in secondary forests. However, in spite of prior log-
ging, the regeneration of high value timber trees such as African mahoganies (Khaya anthotheca
and Entandrophragma spp.) was at least 10 times greater in young secondary forests than in mature
forests. We argue that, if properly managed and protected, secondary forests, even those impacted
by both selective logging and small-scale shifting agriculture, may have high potential conservation
and economic value.

Abbreviations: CEFRECOF – Centre de Formation et de Recherche en Conservation Forestière;

CTFS – Center for Tropical Forest Science; DRC – Democratic Republic of Congo; ENRA –
Enzyme Refiners Association; ITTO – International Tropical Timber Organization; NSERC –
Natural Sciences and Engineering Research Council of Canada

Introduction

The intrusion of human populations into primary forest areas that were
previously free of anthropogenic disturbance is becoming increasingly

[315]
1376

common in all major blocks of remaining tropical forest (Witte 1992;

Verissimo et al. 1995; Coomes et al. 2000; Kammesheidt 2002). Selective
logging of high value timber species is generally the first stage of this process.
Shifting cultivators in search of available land use roads constructed for
logging operations to penetrate deep into the forest interior and clear the
forest for agricultural purposes (Witte 1992; Laurance 2001). Not all farmers
are traditional shifting cultivators who are familiar with the fallow systems
used to restore soil fertility and protect some of the original biodiversity. The
combined effects of selective logging and unsustainable farming practices in
agricultural frontiers are expected to result in considerable loss of biodiversity
and degradation of timber resources.
The Congo basin, in Central Africa, contains the second largest block of
undisturbed, continuous tropical rain forests after the Amazon. Until re-
cently, commercial logging was not causing excessive degradation in the
region, but is considered a major threat as an increasing number of trans-
national logging companies seek to operate in the Congo basin (Wolfire
et al. 1998). Valuable timber trees are generally present at low density in the
natural forests of Central Africa and rates of extraction rarely exceed 2 trees
ha
1 (White 1994; Malcolm and Ray 2000; Hall et al. 2003). Although such
selective logging may be relatively ecologically benign (Wilkie et al. 1992;
White 1994; Hall et al. 2003), the ecological and biological complexity of the
forest may be profoundly disrupted if the forest is logged repeatedly
(Panayotou and Ashton 1992), or if offtake rates are high enough and
logged areas serve as foci for elephant disturbance (Struhsaker 1997;
Struhsaker et al. 1996). In addition, the highly selective character of such
timber harvesting practices can lead to severe depletion and may eventually
result in the local extinction of some high value tree species.
A much greater potential impact than the direct biological and environ-
mental damage of selective logging, however, is the opening of mature forest
areas for colonization (Verissimo et al. 1995; Johns 1997; Whitmore 1999;
Laurance 1999, 2001). Logging companies construct new roads in forest
areas that were previously inaccessible, thereby facilitating spontaneous
colonization of logged forests by agricultural colonists (Wilkie et al. 2000;
Mittelman 2001). The existence of a road network facilitates entry into the
forest and increases potential agricultural economic returns because of
increased opportunities to transport agricultural product surpluses to local
and regional markets (Southgate et al. 1991). The consequences of increases
in these returns may include a substantial increase in farm sizes and
shortening of the fallow period, eventually leading to large scale and severe
destruction of natural forest areas. Although this pattern is widely recog-
nized as a central conservation concern in tropical rain forests, very little
data are available on the actual impacts of selective logging and subsequent
swidden agriculture on forest structure, species composition and diversity,
and timber tree regeneration in Central Africa. Studies in the Neotropics
suggest that land use type and intensity are important factors determining

[316]
 1377

the regrowth of woody vegetation on cleared areas (Aide et al. 1995;

Guariguata and Ostertag 2001). In lightly used areas, secondary forests
regrown from abandoned farms rapidly recover structural characteristics
and tree diversity similar to old-growth stands. Where land use is intense,
such as in grazed pastures, forest recovery is seriously impeded by the
presence of grasses and other herbaceous species that inhibit the establish-
ment of woody vegetation (Guariguata et al. 1997; Smith et al. 1999;
Zahawi and Augspurger 1999; Kennard 2002).
This study reports on the impacts of logging and subsequent forest clearing
on forest regeneration, structure and composition in the Ituri region, north-
eastern Congo basin. The history of mechanized timber harvesting in the
region dates back to the early 1980s, when a 52,000 ha logging concession was
awarded to ENRA in 1982. Due to the collapse of local markets in the early
1990s, logging is very selective with only very large trees of highly valuable
species being extracted for export. Milicia excelsa (Welw.) C. Berg. (Moraceae)
and five species of African mahogany (Khaya anthotheca (Welw.) C. DC,
Entandrophragma angolense (Welw.) C. DC, E. cylindricum (Sprague) Sprague,
E. candollei Harms, and E. utile (Dawe and Sprague) Sprague) accounted for
87% of total volume harvested in 1999 (R. Ducarme, personal communica-
tion). High human population density in the neighboring eastern savanna
regions leads to immigration of landless farmers, who take advantage of log-
ging roads to enter the primary forest areas where land is plentiful and cheap
(Witte 1992). The activities of colonists have resulted in extensive degradation
of the natural forest and its conversion to farmlands. The landscape in logged
forests in the Ituri region is thus dominated by regenerating secondary forest of
varying ages, intermingled with patches of mature forests, active farmlands and
isolated human settlements.
Disturbed and fragmented tropical forests have been increasingly recognized
as being important economically, socially, and for biodiversity conservation,
especially in light of the destruction of the original primary forests (Lugo 1995;
Cannon et al. 1998; Fredericksen 1998; Kammesheidt 2002). The aim of this
study is to compare forest structure and composition and the regeneration of
major timber trees, particularly African mahoganies, between secondary for-
ests regrown after slash-and-burn agriculture in selectively logged forests and
undisturbed mature forest stands in the Ituri Forest. The combined effects of
selective logging and forest clearing for agriculture in one of our study sites can
be expected to result in secondary forests having low species diversity and
reduced major timber tree regeneration. The specific hypotheses tested here are
(1) secondary forests impacted by both selective logging and agriculture will
have lower tree diversity than mature forests, (2) species composition of
secondary forests will favor fast growing early successional species; however,
(3) those species that are economically important, such as African mahoganies,
will show reduced abundance in secondary forests as a result of prior selective
logging.

[317]
1378

Study sites

The study was conducted at two sites in the Ituri Forest, in the northeastern part
of the Congo basin forest block (Democratic Republic of Congo, DRC). The
first site (Mandumbi) was a 17 years old logging concession located 25 km
northwest of the town of Beni (045¢ N latitude, 2915¢ E longitude), whereas a
second site was located at Epulu, in the 1,350,000 ha Okapi Wildlife Reserve.
Field investigation at this site was carried out within the 5 km2 Lenda Study
Area (LSA, 119¢ N and 2838¢ E) established by CEFRECOF. The elevation in
the region from varies 750 m to 950 m above sea level. Mean annual rainfall in
Beni is 1639 mm and 1725 mm in Epulu. A dry season occurs from December to
February, during which monthly average rainfall is less than 100 mm. May and
October are the wettest months of the year, with average precipitations of
186 mm and 200 mm, respectively. Annual average daily temperature at both
sites is 23–25.5 C and varies little through the year (Figure 1).
The vegetation in the region is a mixture of evergreen forest, including
extensive areas of ‘‘mbau forest’’ dominated by Gilbertiodendron dewevrei (De
Wild.) J. Léonard, and ‘‘mixed forests’’ in which no species is predominant, but
other Caesalpinoid legumes, such asJulbernardia seretii (De Wild.) Troupin
and Cynometra alexandri C.H. Wright, are abundant (Makana et al. 2004). At
the eastern edge of the region, evergreen forests grade into a semi-deciduous
forest whose canopy is dominated by light-demanding tree species that include
Entandrophragma spp. (Meliaceae), K. anthotheca (Meliaceae), Albizia spp.
(Mimosaceae) andCanarium schweinfurthii Engl. (Burseraceae). Evergreen
mixed forest is the main vegetation type in Epulu, while semi-deciduous forest
prevails at the Mandumbi site. Large-scale human activities at the Mandumbi
site have resulted in extensive areas of active crop fields and secondary vege-
tation of various ages. Secondary forests were generally young, less than
10 years old, and were dominated by the early pioneer tree Musanga cecro-
pioides R. Br. There was also an old secondary forest created by shifting
agriculture fields abandoned in the 1960s at the Epulu site. However, no
selective logging took place prior to forest clearing at this site.
Soils at both sites are derived from granitic or alluvial rocks and fall under
the order Oxisols, which dominates most of the Congo basin rain forest block
in central Africa. Their texture ranges from loamy sand to sandy clay. The soils
are very acidic, with mean pH values at 20 cm averaging 4 in Epulu, and low
in available nitrogen and phosphorus. Mean soil sand content at LSA is 70%
(Hart 1985).
Logging activities in the Ituri region are concentrated in the relatively drier
semi-deciduous forests near the transition between closed canopy forest and
eastern savanna woodlands, likely due to the proximity of export routes to the
Indian Ocean through Uganda and Rwanda. The forests at the savanna
margin are also richer in high-value timber trees such as African mahoganies
and Milicia excelsa than moist evergreen forests found in central and western
Ituri (Makana 2004).

[318]
 1379

Figure 1. Climatic data of the two study sites, Epulu and Mandumbi. Data for Mandumbi came
from a weather station in the town of Beni. Mean annual rainfall and mean average daily
temperature are given at the top corners of each graph.

Methods

Vegetation sampling

Censuses were conducted in regenerating secondary forests and in mature for-

ests at both Mandumbi and Epulu in nested plots. In Epulu, plots were located
every 50 m along a 500-m long transect in each forest type. Transects from the

[319]
1380

two forest types were parallel and 150 m apart. In Mandumbi, where small
farms are intermingled with patches of mature forest fragments, plot location
was dependent on the availability of appropriate secondary forest stands (e.g.
secondary forests of 5–10 years old). Plots were spatially interspersed across
forest patches, with a minimum distance of 50 m between any two adjacent
plots. At both sites, secondary forest plots were located at least 20 m away from
the adjacent mature forest edge. All free-standing trees ‡1 cm dbh (diameter at
breast height) were identified and measured for dbh in 5 m · 5 m plots. Trees
‡10 cm dbh were identified and measured in 10 m · 10 m plots extending from
each 5 m·5 m plot. For the most common tree species identifications were made
directly in the field. When definitive field identification was not possible leaf
samples were collected and compared to voucher specimens at CEFRECOF’s
herbarium in Epulu. Species names follow Lebrun and Stork (1997). In addition
to botanical data, environmental information was collected at each plot. This
information included soil texture, herbaceous cover, exposed mineral soil and
litter depth. Soil texture was assessed according to the finger assessment of soil
texture of the Ontario Institute of Pedology (1985). For the purposes of this
study, soil texture was classified only in three major categories: sandy, loamy
and clay soils. Herbaceous cover and the proportion of surface area made of
exposed mineral soil were visually estimated and recorded as a percent of total
area. Litter depth was measured by inserting a knife through the litter until it
reached mineral soil, then the thickness of the litter and organic matter layer
was determined to the nearest half centimeter by measuring the length of the
portion of the knife that was inserted into the litter. A total of 54 plots were
inventoried; 32 were at the Mandumbi site and 22 at the Epulu site.

Seedling demography

The regeneration of timber trees was assessed in each of the 54 vegetation plots
described above. Seedlings (30 cm height to 0.9 cm dbh) were identified, tagged
and measured for total aboveground height and collar diameter in 5 m · 5 m
plots. Saplings (1–9.9 cm dbh) were identified and measured for dbh in
10 m · 10 m. Diameter and height were taken at two different times to assess
growth. The first measurements were done in 2000–2001 and the second ones
took place in August–October 2002. Diameter was measured to the nearest
0.01 mm using an electronic caliper for seedlings and trees <4 cm dbh or to
the nearest millimeter for individuals ‡4 cm dbh, whereas height was measured
to the nearest cm using a graduated stick. Stem diameter of seedlings was
measured at 10 cm from the ground.

Data analysis

For each plot, which was the experimental unit of this study, the total number
of individuals, the number of species and Shannon-Wiener diversity index were

[320]
 1381

calculated for seedlings, saplings, trees ‡1 cm dbh, and trees ‡10 cm dbh.
Seedling abundance was log-transformed before the analysis to homogenize
variance and produce approximately normal residuals (Sokal and Rolhf 1981).
ANOVA was used to assess the effects of site, forest type and their interaction
on tree and seedling abundance and diversity. If a significant interaction was
found, the means of site by forest type combinations were compared by a
Tukey-Kramer test. Analysis of covariance served to assess the effects of forest
type on the relationship between light availability and seedling relative growth
rates. Multivariate analysis was used to explore the variation of floristic
composition across sites and forest types. Detrended correspondence analysis
(DCA) was utilized because an arch effect was observed for correspondence
analysis (ter Braak 1995). Logistic regression was utilized to compare the
abundance of seedlings and saplings of individual timber tree species in sec-
ondary and primary forest plots. The number of seedlings/saplings per plot was
transformed into three categories: 0 for no seedling recorded, 1 for 1–4 seed-
lings, and 2 for >5 seedlings.

Results

Vegetation structure

The density of trees ‡1 cm dbh and ‡10 cm dbh differed significantly between
the two study sites, Mandumbi and Epulu (Table 1). On average, Mandumbi
had higher tree density than Epulu for both dbh cut-offs. There was a mar-
ginally significant effect of forest type on the density of trees ‡1 cm dbh
(F1,50 = 2.5, p = 0.094). In Epulu, primary forest had significantly more
stems than secondary forest, whereas the density of trees ‡1 cm dbh was only
slightly higher in primary forest compared to secondary forest in Mandumbi.
Basal area was significantly higher in secondary than in primary forests for
both dbh cut-offs (Table 1). Higher basal area in secondary forests was in the
most part due to the presence of fast growing pioneer species. The early suc-
cessional and short-lived Musanga cecropioides was very common in the sec-
ondary forest of Mandumbi and it represented 55% of total basal area. Some
of the basal area in secondary forest stands was also accounted for by remnant
trees. In Epulu, long-lived early colonizers such as Albizia sp., Petersianthus
macrocarpus (P. Beauv.) Liben, and Alstonia boonei De Wild. made up most of
basal area in secondary forest plots.
The number of stems decreased rapidly with increasing tree size (Figure 2).
Trees <10 cm dbh represented >80% of the total number of stems in each
forest type at both study sites. Although primary forest sites in Epulu had more
trees ‡10 cm dbh than secondary forest, the overall size distribution was similar
between the two forest types (v2 = 4.5, d.f. = 2, p = 0.210). In Mandumbi,
the size distribution was significantly different between the two forest types
(v2 = 16.3, d.f. = 2, p<0.001); small trees (<20 cm dbh) were more abun-

[321]
 1382
Table 1. Structural characteristics and diversity of mature logged and undisturbed, and secondary forest stands in northeastern Congo basin. Values of stem
density are averages per hectare, and were calculated on the basis of 5 m · 5 m subplots. Species richness and Shannon’s diversity index were calculated for
5 m · 5 m quadrats for trees ‡1 cm dbh and for 10 m · 10 m quadrats for trees ‡10 cm dbh. Figures are least squares means and standard errors1.

Epulu Mandumbi Significance2

PF3 (n = 11) SF (n = 11) PF (n = 16) SF (n = 16) Site FT Site*FT

Trees ‡1 cm dbh
Stem density (ha
1) 6473(678)a 4509 (678)b 7571 (601)a 7343 (601)a 0.004 0.094 0.182
Basal area (m2 ha
1) 26.82 (5.75)a 44.92 (5.75)b 28.17 (4.93)ab 39.16 (4.93)b 0.683 0.009 0.510
Richness 10.9 (0.89) 11.5 (0.96) 12.9 (0.86) 12.3 (0.76) 0.072 0.683 0.796
Shannon’s index 2.0 (0.14) ab 1.9 (0.14)a 2.3 (0.13)ab 2.3 (0.11)b 0.007 0.955 0.440
[322]

Trees ‡10 cm dbh

Stem density (ha
1) 427 (62)a 356 (62)a 636 (55)b 721 (55)b <0.001 0.912 0.181
Basal area (m2 ha
1) 22.63 (6.32)a 41.94 (6.32)b 26.49 (5.60)ac 35.40 (5.41)ab 0.955 0.014 0.482
Richness 3.3 (0.41)a 3.2 (0.48)a 4.67 (0.77)a 1.9 (0.41)b 0.010 0.001 <0.001
Shannon’s index 1.1 (0.12)ab 1.0 (0.16)ab 1.6 (0.14)a 0.8 (0.18)b 0.450 0.010 0.027
Environmental characteristics
Herb cover (%) 11.1 (3.97)a 35.5 (4.64)b 15.4 (5.81)a 16.5 (6.0)ab 0.164 0.014 0.047
Litter depth (cm) 1.2 (0.14)a 0.9 (0.12)a 2.7 (0.37)b 3.8 (0.86)b <0.001 0.612 0.167
1
Means with common subscripts are not significantly different according to Tukey’s multiple comparison test (a = 0.05).
2
Abundance was included as explanatory variable for ‘‘richness’’ to remove the effects of tree density on species richness.
3
Forest type codes: PF for primary forest and SF for secondary forest. The primary forest at Mandumbi was disturbed by logging and tree harvesting for
construction; it is undisturbed at Epulu. The secondary forest stands at Mandumbi were young (<10 years old), but those at Epulu were much older
( 40 years old).
 1383

Figure 2. Whole-forest diameter distribution in 10 cm dbh intervals for (a) Epulu primary forest,
(b) Mandumbi primary forest, (c) Epulu secondary forest, and (d) Mandumbi secondary forest.

dant in primary forest, whereas secondary forest had more intermediate-sized

trees (20–39 cm dbh) than primary forest stands. At both sites, understory and
shade tolerant trees constituted the majority of trees in primary forest stands
(Figure 2a and b), whereas light demanders and pioneers were dominant in
secondary forests (Figure 2c and d).
Herbaceous cover varied significantly with respect to forest type
(F1,50 = 6.6, p = 0.014). At both sites, herbaceous vegetation was more
abundant in secondary forest than in primary forest, and it was dominated by
species of the family Maranthaceae and Zingiberaceae. The litter layer was at
least twice as thick in Mandumbi as in Epulu (F1, 50 = 33.3, p<0.001) but it
did not vary significantly with respect to forest disturbance. No significant
differences were observed in soil texture or percent of ground area in exposed
mineral soil. At both sites, soil texture ranged from sandy to clay. Small areas
of exposed mineral soil were recorded in seven plots, representing 0.17% of
the total area sampled.

Species composition and diversity

A total of 159 species were recorded in the census plots, of which 121 taxa were
identified to the species level, 17 to genus, 17 to family, and the remaining four

[323]
1384

were unidentified. Overall, Mandumbi plots had 122 species, while 82 species
were represented in Epulu plots. Species richness and Shannon’s diversity index
showed different patterns relative to tree sizes (Table 1). For trees ‡1 cm, mean
number of species per plot was similar in the two sites and forest types;
however, the value of Shannon’s index was significantly higher in Mandumbi
than in Epulu (F1,50 = 8.1, p = 0.007). In contrast, species richness of trees
‡10 cm dbh varied significantly according to both site and forest disturbance,
while Shannon’s index was only affected by forest disturbance. There were
significant interactions in both diversity measures between site and forest type.
In Epulu, both the mean number of species per plot and the mean value of
Shannon’s index were similar for primary and secondary forests. In contrast,
there was a significant difference for both parameters between the two forest
types in Mandumbi. The young secondary forest at the latter site was much less
diverse than primary forest for trees ‡10 cm dbh (Table 1). Among the four
combinations of site and forest type, richness and Shannon’s diversity index of
trees ‡10 cm dbh were highest for the primary forest of Mandumbi and lowest
for the young secondary forest of the same site.
The species accumulation curves in Figure 3 are far from asymptotic,
indicating that the area sampled was too small to estimate the total number

Figure 3. Species-area relationships for (a) all trees ‡1 cm dbh and (c) trees ‡10 cm dbh. Species–
individual curves for (b) all trees above 1 cm dbh and (d) trees ‡10 cm dbh. EP = Epulu and
MN = Mandumbi, PF = primary forest and SF = secondary forest.

[324]
 1385

Figure 4. Axes 1 and 2 from a detrended correspondence analysis on tree species represented by at
least 8 individuals in the plots: (a) plot scores (triangles represent Mandumbi site and circles Epulu
site, filled symbols are for primary forest plots and open symbols for secondary forest sites) and (b)
species scores (symbols with circles represent commercial timber species). Species codes as in
Table 2.

of species in the studied communities. At each site, species–area curves were

similar between primary and secondary forests for trees ‡1 cm dbh. The rate
of species accumulation was different between primary and secondary forests
in Mandumbi for trees ‡10 cm dbh, while it was quite similar between the
two forest types in Epulu (Figure 3c and d). In Mandumbi, the rate of

[325]
1386

species accumulation for trees ‡10 cm dbh was much higher in primary
forest than in secondary forest.

Variation in tree species composition

The two sites and forest types shared most of their common species (Ta-
ble 2). Overall, Epulu and Mandumbi shared 14 of their 20 most abundant
species, while primary and secondary forests had 18 of their top 20 species
in common at each site. In spite of sharing the majority of the most
common species, there was no significant positive correlation of ranked
abundances of these species between sites or forest types. Epulu and
Mandumbi showed a significant negative correlation among abundance
ranks (Spearman’s rS =
0.549, p = 0.012), as did primary and secondary
forests (rS =
0.739, p<0.001). The most abundant species in Epulu,
Scaphopetalum dewevrei Wildem. and Th. Dur., was totally absent from
Mandumbi plots. Moreover, two of the most common species in the latter
site (Musanga cecropioides and Rinorea oblongifolia Marquand) were not
represented in Epulu plots. In addition, two species of African mahogany
(E. utile and K. anthotheca) were much more abundant in Mandumbi than
in Epulu (Table 2).
Pair-wise comparisons of species abundances for site-forest type combina-
tions showed that the young secondary forest of Mandumbi had significant
negative correlation with primary forests of both sites (rs =
0.743, p<0.001
and rs =
0.797, p<0.001 for Mandumbi and Epulu respectively). No sig-
nificant correlation was found between the old secondary forest of Epulu and
the primary forest of either site. Thus, the composition of the young secondary
forest in Mandumbi was significantly different from that of primary forests at
both sites, while that of the old secondary forest at Epulu was not.
The patterns of variation in floristic composition observed from correla-
tion analysis were corroborated by the results of multivariate analysis. The
first axis of a detrended correspondence analysis (DCA), which explained
12.9% of the total variation in the species data, appeared to be related to
the composition of canopy flora and it separated primary and secondary
forests at each site. For each site, secondary forests had higher scores of
axis 1 than primary forests. The separation between primary and secondary
forests on axis 1 was more distinct for Mandumbi than Epulu forests
(Figure 4a). The second axis (7.6% of the total variation) also distinguished
between primary and secondary forests of each site, but seemed to be more
related to the composition of understory vegetation. Primary forests had
higher scores than secondary forests on the second DCA axis. Primary
forests of both sites and the old secondary forest in Epulu overlapped on
both DCA axes. Mandumbi primary forest and Epulu secondary forest were
indistinguishable on the first axis, but the former had higher scores of axis 2
than the latter.

[326]
 Table 2. Abundance of the 20 most common tree species for stems ‡1 cm dbh in primary forests (PF) and secondary forests (SF) at two study sites (Epulu
and Mandumbi) in northeastern Congo Basin. Figures are number of trees per hectare.

Species name Code Status EPULU MANDUMBI

PF SF PF SF

Scaphopetalum dewevrei De Wild. and T. Durand SCAPDE Understory 1782 109 0 0

Gilbertiodendron dewevrei (De Wild.) J. Léonard GILBDE Canopy 764 0 833 0
Julbernardia seretii (De Wild.) Troupin JULBSE Canopy 800 109 401 48
Alchornea floribunda Müll. Arg. ALCHFL Understory 145 182 668 175
Rinorea oblongifolia (C.H. Wright) Marquand ex Chipp RINOOB Understory 0 0 1012 110
Diospyros bipendensis Gürke DIOSBI Understory 327 109 467 150
Celtis mildbraedii Engl. CELTMI Canopy 255 618 67 99
[327]

Khaya anthotheca (Welw.) C. DC. KHAYAN Canopy 36 73 167 724

Trichilia rubescens Oliv. TRICRU Understory 109 545 166 174
Pancovia harmsiana Gilg PANCHA Understory 327 0 401 98
Myrianthus preussii Engl. MYRPR Understory 100 440 33 101
Cynometra alexandri C.H. Wright CYNOAL Canopy 218 400 0 0
Musanga cecropioides R. Br. MUSACE Canopy 0 0 0 653
Albizia gummifera (J.F. Gmel.) C.A. Sm. ALBIGU Canopy 0 109 68 399
Greenwayodendron suaveolens (Engl. and Diels) Verdc. GREESU Understory 327 72 66 0
Pycnanthus angolensis (Welw.) Warb. PYCNAN Canopy 0 0 264 224
Antiaris toxicaria Lesch. ANTITO Canopy 73 36 67 275
Cola lateritia K. Schum COLALA Canopy 110 37 233 48
Entandrophragma utile (Dawe and Sprague) Sprague ENTAUT Canopy 0 40 65 301
Microdesmis puberula Hook. F. ex Planch. MICRPU Understory 182 218 0 0

1387
 1388
Table 3. Abundance of seedlings and saplings of commercial timber tree species in primary forest (PF) and secondary forest (SF) in northeastern Congo
basin. Abundances are averages of the number of stems per hectare, and were calculated on the basis of 5 m · 5 m quadrats for seedlings and of 10 m · 10 m
quadrats for saplings. Significance levels for the difference in regeneration abundance between primary and secondary forests as follows: ns p>0.05, * p<0.05,
** p<0.01.

Species name Shade Seedlings ha
1 Saplings ha
1

tolerance
guild1 PF SF Difference PF SF Difference

Khaya anthotheca (Welw.) C.DC. NPLD 237 4978 ** 28 120 **

Entandrophragma cylindricum (Sprague) Sprague NPLD 356 370 ns 31 4 ns
Entandrophragma angolense (Welw.) C. DC. NPLD 30 978 ** 14 52 *
Entandrophragma utile (Dawe and Sprague) Sprague NPLD 0 89 – 4 40 ns
[328]

Lovoa trichilioides Harms NPLD 30 0 – 10 24 ns

Guarea cedrata (A. Chev.) Pellegr. Tolerant 89 59 ns 7 0 –
Albizia gummifera (J.F. Gmel.) C.A. Sm. NPLD 607 119 ns 31 60 ns
Tieghemella africana Pierre NPLD 59 59 ns 21 20 ns
Gilbertiodendron dewevrei (De Wild.) J. Léonard Tolerant 5956 30 ** 162 0 –
Zanthoxylum gillettii (De Wild.) P.G. Waterman NPLD 0 0 – 7 20 ns
Nauclea diderrichii (De Wild.) E.M.A. Petit Pioneer 0 30 – 0 24 –
Canarium schweinfurthii Engl. Pioneer 0 0 – 4 32 ns
Alstonia boonei De Wild. Pioneer 0 0 – 0 32 –
Klainedoxa gabonensis Pierre ex Engl. NPLD 0 0 – 7 8 ns
TOTAL 7364 6712 326 436
1
NLPD: non-pioneer light demander (Hawthorne 1993)
 1389

A greater proportion of common species were associated with Mandumbi

forests than Epulu forests (Figure 4b). Most of Epulu primary forest plots
and Mandumbi secondary forest plots were more than 3 standard deviations
apart on the first axis. This suggests that the two communities share few of
their most common canopy species. Examination of Figure 4 reveals that
the floristic composition of the primary forest in Mandumbi was more
similar to those of the primary and old secondary forests in Epulu than to
that of the young secondary forest at the same site.

Regeneration of timber species

The abundance of seedlings of commercial timber trees differed significantly

between primary and secondary forest forests (F1,44 = 8.0, p<0.001), and
there was also a significant site by forest type interaction (F1,44 = 7.9,
p = 0.008). The difference in seedling abundance with respect to forest type
was entirely due to the extremely low abundance of timber tree seedlings in
the secondary forest of Epulu, which averaged 15 times less seedlings than
any other combination of site and forest type. The number of seedlings in
the mature and undisturbed forest of Epulu was similar to that of the
logged or secondary forests of Mandumbi. The abundance of saplings of
timber species did not vary significantly according to forest type. Although
Mandumbi forests had consistently higher densities of saplings than Epulu
forests, the difference was significant only at p<0.1.
Two species were particularly abundant in the seedling and sapling
populations. Khaya anthotheca was the most abundant timber species in the
secondary forest of Mandumbi, averaging densities of 8350 seedlings and
190 saplings ha
1. In Epulu, K. anthoteca was totally absent from the
seedling population and occurred only at a density of 73 saplings ha
1. G.
dewevrei dominated the seedling and sapling populations in primary forests
of both sites. Mean densities of G. dewevrei seedlings and saplings in pri-
mary forests were 6000 and 162 individuals ha
1, respectively. While K.
anthoteca was also well represented among the seedling and sapling popu-
lations of primary forests, G. dewevrei seedlings were very rare in secondary
forests ( 30 individuals ha
1) and no saplings of this species were recorded
in secondary forest (Table 3).

Mahogany regeneration

Seedlings of four species of African mahogany (K. anthoteca, Entandrophragma

angolense, E. cylindricum and E. utile) were much more abundant in secondary
forests (10,775±3776 ha
1) than in primary forests (1050±336 ha
1) at

[329]
1390

Mandumbi. No seedlings of these species were found in any of the forest plots in
Epulu. The abundance of mahogany saplings (1–9.9 cm dbh) was also higher in
secondary forests than in primary forests at both sites, 331 (125) vs. 113 (39) stems
ha
1 in Mandumbi and 27 (14) vs. 9 (9) stems ha
1 in Epulu. To determine
environmental factors that affect the regeneration of African mahogany tree
species in the region, the abundance of mahogany regeneration was compared
between primary and secondary forests and the relationships between the
abundance of the regeneration and recorded environmental variables was tested
through regression analysis and analysis of variance. This analysis was limited to
data from Mandumbi because mahogany regeneration in Epulu was negligible.
The abundance of the regeneration of African mahogany species varied
significantly according to forest type (F1,44 = 17, p<0.001) and soil texture
(F1,44 = 2.43, p = 0.038). There were at least ten times more seedlings of these
species in secondary forests than in primary forests at Mandumbi (Table 3).
Loamy and sandy soils supported a higher density of seedlings and saplings of
mahogany than did clay-textured soils. Simple regression analysis showed that
two factors were significantly associated with mahogany regeneration. The
abundance of M. cecropioides, an early pioneer tree species, and litter depth
were positively associated with mahogany regeneration (r2 = 0.431, p<0.001
and r2 = 0.158, p = 0.022 respectively). Herb cover was marginally negatively
associated with the abundance of mahogany regeneration (r2 = 0.099,
p = 0.075).
A simultaneous test of all factors and their interactions through multiple
regression (stepwise selection) revealed that only the abundance of M. cecro-
pioides and the interaction between the latter factor and herb cover were sig-
nificant (p<0.001 for both variables). This model explained 61.4% of the
variation in the abundance of mahogany regeneration in Mandumbi. Litter
depth was not maintained in the model likely due to its strong positive cor-
relation with the abundance of M. cecropioides (r = 0.525, p = 0.002).

Discussion

Secondary forests and the conservation of tree diversity

It was expected that the combination of selective logging and agricultural

clearing would result in the degradation and impoverishment of natural forests.
Secondary forests are generally seen as having much lower conservation value
than mature forests. They generally have fewer tree species, are dominated by
widespread pioneer trees, and have a simpler structure (Garcia-Montiel and
Scatena 1994; Guariguata et al. 1997; Whitmore 1999; Aide et al. 2000). The
return of secondary forests to the complex and species-rich primary forest
conditions can be very slow, partly due to the limited availability of seeds of
primary forest tree species (Holl et al. 2000; Wijdeven and Kuzee 2000).
However, human-induced disturbances in tropical forests span a wide gradient,

[330]
 1391

depending on land-use type and intensity. Forest clearing for traditional slash-
and-burn agriculture, such as practiced in our study sites, occupies the lower
end of severe forest disturbance as compared to clearing for large commercial
pastures or oil palm plantations (Lawrence et al. 1998; Mesquita et al. 2001).
This study shows that secondary forests growing after the initial clearing
of primary forests for shifting cultivation in the Ituri harbor surprisingly
high levels of tree species diversity for small stem sizes. Overall diversity
measures of trees ‡1 cm dbh were similar between secondary and primary
forest stands. Two factors that may account for the observed diversity
patterns in the secondary and primary forest stands include edge effects and
the dominance of primary forest stands by G. dewevrei. The proximity of
many secondary forest plots to primary forest stands will likely result in
increased diversity in these plots due to potentially high seed input from
mature forests (Mesquita et al. 2001; Kennard 2002). On the other hand,
most mature forest plots especially at Epulu were located in forest stands
dominated by G. dewevrei, which are known to have very low diversity at
small spatial scales (Hart et al. 1989; Makana et al. 2004). The diversity of
larger trees (i.e. ‡10 cm dbh) was, however, significantly lower in young
secondary forest stands (<10 years old). However, notwithstanding the
similarity in overall tree diversity, the floristic compositions of the two forest
types were very different. The flora of secondary forests was particularly
depauperate in common species characteristic of old-growth forests in the
region, particularly G. dewevrei and J. seretii, and understory specialists
such as S. dewevrei, Drypetes spp., Rinorea spp., and Pancovia harmsiana
Gilg. In this respect, our results corroborate those of other studies on
tropical forest succession (Brown and Lugo 1990; Guariguata et al. 1997;
Aide et al. 2000). For example, Aide et al. (2000) noted that a 40-year old
secondary forest derived from abandoned pasture had comparable tree
diversity to adjacent mature forest, whereas the floristic composition was
substantially different.
Stem density and basal area increased very rapidly after farm abandonment
with higher basal areas observed in 10-year old secondary forests than in
adjacent primary forest. However, most of the basal area in the young sec-
ondary forest of Mandumbi was accounted for by the presence of the early
pioneer tree M. cecropioides, which does not persist beyond the senescence of
the initial cohort, which may result in a reduction in the basal area of those
stands when the initial cohort of that species dies out. Similar trends have been
observed elsewhere in tropical forest succession (Aide et al. 2000). Our result
are consistent with other findings suggesting that many structural character-
istics of secondary forest stands in the tropics can reach levels encountered in
mature forest stands quite early during succession (Brown and Lugo 1990;
DeWalt et al. 2003). The rapid forest recovery observed in this study may be
the result of light land-use intensity and high seed fall into abandoned fields
due to the relatively small sizes of farms and the presence of remnant trees that
attract seed dispersers.

[331]
1392

Forest disturbance and the regeneration of major timber trees

A pattern of reduced abundance of timber tree regeneration in selectively

logged forests has been frequently observed in the tropical forests of Africa and
Latin America (Verissimo et al. 1995; Struhsaker 1997; Mwima et al. 2001;
Hall et al. 2003). Low seed availability due to the removal of most large
reproductive trees (Plumptre 1995; Makana and Thomas 2004), small sizes of
canopy openings created by single-tree removal, high seed and seedling pre-
dation, and the rapid invasion of logging gaps by lianas and herbaceous veg-
etation (Struhsaker 1997; Fredericksen 1998) are commonly blamed for this
lack of timber tree regeneration after selective logging. On the other hand,
several studies have reported good regeneration of Neotropical mahogany
(Swietenia macrophylla King) following severe disturbances such as hurricane,
fire, flooding or agricultural clearing (Lamb 1966; Snook 1996; Gullison et al.
1996; Mostacedo and Fredericksen 1999; Negreros-Castillo et al. 2003). Our
results support the hypothesis that African mahoganies, like their Neotropical
counterpart, regenerate well after severe disturbances (i.e. forest clearing for
agriculture) that destroy most of the existing vegetation. Secondary forests
regrowing on areas that were previously selectively logged and cleared for
agricultural purposes had at least 10 times more seedlings of African mahog-
anies than unlogged mature forests. The positive effects of forest clearing on
tree regeneration were not limited to mahoganies; other light demanding
timber trees such as Albizia gummifera C. A. Smith, Canarium schweinfurthii
Engl., and Zanthoxylum gillettii (De Wild.) Waterman also showed better
seedling recruitment in secondary vegetation than in mature forests.
Good regeneration of timber trees in the secondary forest stands of this
study may be the result of traditional shifting cultivation practices in the Ituri
region. Mature individuals of valuable tree species (fruit trees, timber or
medicinal species) are commonly left alive during forest clearing for agricul-
tural purposes. These ‘‘remnant’’ trees likely dispersed seeds in abandoned
farms while competition from existing vegetation was low, leading to abundant
regeneration of major timber trees such as African mahoganies (Carrière et al.
2000). Slash-and-burn agriculture, which mimics hurricanes followed by
wildfires, is now considered as one of the silvicultural treatments that favor the
establishment of mahogany in the Neotropics (Negreros-Castillo et al. 2003).
In the Peruvian Amazon, Smith et al. (1999) reported a high representation of
valuable timber trees in secondary forests regrown after farm abandonment.
Twenty of the 22 most important timber trees in the region were present as
poles (>10 m height) in old secondary forests.
The regeneration of African mahogany species was much more abundant in
Mandumbi than in Epulu. Mandumbi is located at the margin of closed can-
opy forest and eastern savanna woodlands, and it has semi-deciduous vege-
tation, whereas the vegetation at Epulu (located in central Ituri basin) is
primarily evergreen. Although African mahogany species are widely distrib-
uted across tropical Africa, their local abundance can be influenced by climate

[332]
 1393

(rainfall), soil conditions (moisture and nutrients), historic events, and dis-
turbances (Bongers et al. 1999). Most African mahoganies have been described
as belonging to the deciduous or semi-deciduous forests (Hall and Swaine
1981). The relatively drier and semi-deciduous forests of the eastern fringe of
the Ituri basin probably offer more favorable conditions for the regeneration
and growth of African mahoganies than the evergreen forests of Epulu.
Densities of commercial size trees (‡ 80 cm dbh) were 5 times higher in
Mandumbi than in Epulu (J.-R. Makana, unpublished data). Low density of
adult trees may therefore partly explain the poor regeneration of K. anthotheca
and Entandrophragma spp in Epulu, due to limited seed availability (Plumptre
1995; Makana and Thomas 2004).

Implications for conservation and management

Natural forests in the Ituri region host an impressive diversity of tree and
animal species which has led to the designation of the Ituri forest as a
refugium during late Quaternary climate fluctuations (Grubb 1982; Hart
et al. 1996). The Ituri forest contains large mammal species including the
endemic Okapi, forest elephant, leopard, buffalo, as well as 13 species of
anthropoid primates, seven species of forest antelopes (duikers), and many
other species of mammals, birds and reptiles (Hart 1986; Thomas 1991;
Hart et al. 1986; Plumptre 1996). This abundant and diverse wildlife
assemblage has coexisted with small communities of horticulturalists in the
Ituri for centuries (Wilkie and Finn 1990). While the presence of small areas
of secondary vegetation derived from swidden agriculture may benefit some
mammal populations (Short 1983; Thomas 1991), the creation of large areas
of secondary forests poses a threat to most mature forest dwelling species
such as forest ungulates, okapi, leopard, and others. These species, espe-
cially forest ungulates, are the main source of dietary protein for local
populations (Hart and Hart 1986). Therefore maintaining large tracks of
undisturbed primary forest is essential to the conservation of animal
diversity and to the well-being of local hunter–gatherer communities.
Forest recovery on abandoned agricultural lands is possible through natural
regeneration. It is shown here that some aspects of forest structure and the
diversity of small trees can be rapidly restored to levels comparable to those of
mature forests. However, forest recovery is possible only if secondary forests
are protected from repeated clearing because the return of the species com-
position of secondary vegetation to assemblage similar to that of old-growth
forests may require over 100 years due to limited seed availability and dis-
persal, and slow growth of mature forest tree species. Small-sized clearings,
moderate land use intensity (long fallow periods), and a fine scaled landscape
mosaic may speed up the return of secondary forests to mature forest species

[333]
1394

composition. Human intervention in the recovery process may be desirable.

Makana and Thomas (2004) showed that the addition of seeds in light-gaps
greatly increased seedling recruitment in the study site. Thus, seed supple-
mentation and/or enrichment planting can be valuable interventions for the
rapid return of secondary forests to the complex and species-rich mature forest
conditions.
Our results are consistent with the generalization that tropical secondary
forest resulting from farm abandonment may provide a favorable environment
for the regeneration of high value timber trees (Negreros-Castillo et al. 2003).
The high abundance of African mahoganies’ regeneration and the enhanced
performance of their seedlings in secondary forest stands suggest that these
species require sufficient light availability to regenerate well. Therefore, suc-
cessful regeneration of African mahoganies may depend on the adoption of
forestry practices that provide for larger clearings, followed by post-harvest
silvicultural treatments to control competing vegetation (see Fredericksen and
Putz 2003; Negreros-Castillo et al. 2003). It has been shown here that young
secondary forests, growing on areas that were both selectively logged and
cleared for agricultural purposes, have both high tree species diversity and
abundant regeneration of valuable timber trees. This potentially favorable
situation offers a window of opportunity to reconcile social and ecological
demands on tropical forests. Planned and controlled clearing of selectively
logged forests for shifting cultivation can have the potential create favorable
conditions for African mahogany regeneration as well as to provide for the
needs of local communities that depend on the forest, while still conserving a
high portion of the original biodiversity through appropriate fallow systems
and protection of primary forest stands as a source of propagules.

Acknowledgements

Financial support for the research came through grants from ITTO, CTFS and
CEFRECOF, a University of Toronto fellowship, and an NSERC grant to
Sean Thomas. Logistical support at the field sites was provided by CEFRE-
COF and ENRA. We thank Sabuni Paluku, Simende, Nzambe, and Amisi for
assisting with fieldwork. Jay Malcolm, Justina Ray, Terry Carleton and two
anonymous reviewers provided constructive comments on an earlier version of
this manuscript.

References

Aide T.M., Zimmerman J.K., Herrera L., Rosario M. and Serrano M. 1995. Forest recovery in
abandoned tropical pastures in Puerto Rico. Forest Ecology and Management 77: 77–86.
Aide T.M., Zimmerman J.K., Pascarella J.B., Rivera L. and Marcano-Vega H. 2000. Forest
regeneration in a chronosequence of tropical abandoned pastures: implications for restoration
ecology. Restoration Ecology 8: 328–338.

[334]
 1395

Bongers F., Poorter L., van Rompaey R.S.A.R. and Parren M.P.E. 1999. Distribution of twelve
moist forest canopy species in Liberia and Côte d’Ivoire: response curves to a climatic gradient.
Journal of Vegetation Science 10: 371–382.
Brown S. and Lugo A.E. 1990. Tropical secondary forests. Journal of Tropical Ecology 6: 1–32.
Cannon C.H., Peart D.R. and Leighton L. 1998. Tree diversity in commercially logged Bornean
rainforest. Science 281: 1366–1368.
Carrière S.M., Letourmy P. and McKey D.B. 2002. Effects of remnant trees in fallows on diversity
and structure of forest regrowth in a slash-and-burn agricultural system in southern Cameroon.
Journal of Tropical Ecology 18: 375–396.
Coomes O.T., Grimard F. and Burt G.J. 2000. Tropical forests and shifting cultivation: secondary
forest dynamics among traditional farmers of the Peruvian Amazon. Ecological Economics 32:
109–124.
DeWalt S.J., Malakal S.K. and Denslow J.S. 2003. Changes in vegetation structure and compo-
sition along a tropical forest chronosequence: implications for wildlife. Forest Ecology and
Management 182: 139–151.
Fredericksen T.S. 1998. Limitations of low-intensity selection and selective logging for sustainable
tropical forestry. Commonwealth Forestry Review 77: 262–266.
Fredericksen T.S. and Putz F.E. 2003. Silvicultural intensification for tropical forest conservation.
Biodiversity and Conservation 12: 1445–1453.
Garcia-Montiel D.C. and Scatena F.N. 1994. The effect of human activity on the structure and
composition of a tropical forest in Puerto Rico. Forest Ecology and Management 63: 57–78.
Grubb P. 1982. Refuges and dispersal in the speciation of African forest mammals. In: Prance G.T.
(ed.), Biological Diversification in the Tropics. Columbia University Press, NY, pp. 537–553.
Guariguata M.R. and Ostertag R. 2001. Neotropical secondary forest succession: changes in
structural and functional characteristics. Forest Ecology and Management 148: 185–206.
Guariguata M.R., Chazdon R.L., Denslow J.S., Dupuy J.M. and Anderson L. 1997. Structure and
floristics of secondary and old-growth forest stand in lowland Costa Rica. Plant Ecology 132:
107–120.
Gullison R.E., Panfil S.N., Strouse J.J. and Hubbell S.P. 1996. Ecology and management of
mahogany (Swietenia macrophylla King) in the Chimanes Forest, Beni, Bolovia. Botanical
Journal of Linnean Society 122: 9–34.
Hall J.B. and Swaine M.D. 1981. Distribution and ecology of vascular plants in a tropical rain
forest: forest vegetation in Ghana. Dr W. Junk, The Hague.
Hall J.S., Harris D.J., Medjibe V. and Ashton P.M.S. 2003. The effects of selective logging on forest
structure and tree species composition in a Central African forest: implications for management
of conservation areas. Forest Ecology and Management 183: 249–264.
Hart J.A. and Carrick P. 1996. Climate of the Réserve de Faune à Okapis: Rainfall and temper-
ature in the Epulu Sector, 1986–1995. CEFRECOF Working Paper n 2.
Hart J.A., Hart T.B. and Thomas S.C. 1986. The Ituri Forest of Zaire: primate diversity and
prospects for conservation. Primate Conservation 7: 42–44.
Hart T.B. 1985. The Ecology of a Single-Species Dominant Forest and a Mixed Forest in Zaire,
Equatorial Africa. PhD thesis, Michigan State University, East Lansing, MI.
Hart T.B., and Hart J.A. 1986. The ecological basis of hunter-gatherer subsistence in African rain
forests-the Mbuti of Eastern Zaı̈re. Human Ecology 14: 29–55.
Hart T.B., Hart J.A. and Murphy P.G. 1989. Monodominant and species-rich forests of the humid
tropics: causes for their co-occurrence. American Naturalist 133: 613–623.
Hart T.B., Hart J.A., Dechamps R., Fournier M. and Ataholo M. 1996. Changes in forest com-
position over the last 4000 years in the Ituri basin, Zaire. In: van der Maesen L.J.G. (ed.), The
Diversity of African Plants. Kluwer Academic Publishers, the Netherlands, pp. 545–563.
Holl K.D., Loik M.E., Lin E.H.V. and Samuels I.A. 2000. Tropical montane forest restoration in
Costa Rica: overcoming barriers to dispersal and establishment. Restoration Ecology 8: 339–349.
Johns A.G. 1997. Timber production and biodiversity conservation in tropical rain forests.
Cambridge University Press, Cambridge, UK.

[335]
1396

Kammesheidt L. 2002. Perspectives on secondary forest management in tropical humid lowland

America. Ambio 31: 243–250.
Kennard D.K. 2002. Secondary succession in tropical fry forest: patterns of development across a
50-year chronosequence in lowland Bolivia. Journal of Tropical Ecology 18: 53–66.
Lamb F.B. 1966. Mahogany of Tropical America: Its Ecology and Management. University of
Michigan Press, Ann Arbor.
Laurance W.F. 1999. Reflections on the tropical deforestation crisis. Biological Conservation 91:
109–117.
Laurance W.F. 2001. Tropical logging and human invasions. Conservation Biology 15: 4–5.
Lawrence D., Peart D.R. and Leighton M. 1998. The impact of shifting cultivation on rainforest
landscape in West Kalimantan: spatial and temporal dynamics. Landscape Ecology 13: 135–148.
Lebrun J. and Stork A.L. 1997. Enumération des plantes à fleurs d’Afrique tropicale. Vol. 4.
Publication hors-série n 7C, Conservatoire et Jardin Botaniques de ville de Genève, Genève.
Lugo A.E. 1995. Management of tropical biodiversity. Ecological Applications 5: 956–961.
Makana J.-R. 2004. Ecology and sustainable management of African mahoganies and other se-
lected timber species in northeastern Congo Basin, Democratic Republic of Congo. PhD thesis,
University of Toronto, Toronto, Canada.
Makana J.-R. and Thomas S.C. 2004. Dispersal limits the natural recruitment of African
mahoganies. Oikos 106: 67–72.
Makana J.-R., Hart T.B., Hibbs D.E. and Condit R. 2004. Forest structure and diversity in the
Ituri Forest dynamics plots. In: Losos E. and Leigh E.Jr. (eds), Tropical Forest Diversity and
Dynamism: Findings from a Large-Scale Network. University of Chicago Press, Chicago, pp.
149–174.
Malcolm J.R. and Ray J.C. 2000. Influence of timber extraction routes on Central African small-
mammal communities, forest structure, and tree diversity. Conservation Biology 14: 1623–1638.
Mesquita R.C.G., Ickes K., Ganade G. and Williamson G.B. 2001. Successional pathways in the
Amazon Basin. Journal of Ecology 89: 528–537.
Mittelman A. 2001. Secondary forests in the lower Mekong subregion: an overview of their extent,
roles and importance. Journal of Tropical Forest Science 13: 671–690.
Mostacedo B. and Fredericksen T.S. 1999. Regeneration status of important tropical forest tree
species in Bolivia: assessment and recommendations. Forest Ecology and Management 124: 263–
273.
Mwima P.M., Obua J. and Oryem-Origa H. 2001. Effect of logging on the natural regeneration of
Khaya anthotheca in Budongo Forest Reserve, Uganda. International Forestry Review 3: 131–
135.
Negreros-Castillo P., Snook L.K. and Mize C.W. 2003. Regenerating mahogany (Swietenia mac-
rophylla) from seed in Quintana Roo, Mexico: the effects of sowing method and clearing
treatment. Forest Ecology and Management 183: 351–362.
Okali D.U.U. and Ola-Adams B.A. 1987. Tree population changes in treated rain forest at Omo
Forest Reserve, south-western Nigeria. Journal of Tropical Ecology 3: 291–313.
Ontario Institute of Pedology 1985. Field Manual for Describing Soils. 3rd ed. Ontario Institute of
Pedology, Guelph.
Panayotou T. and Ashton P.S. 1992. Not by Timber Alone: Economics and Ecology for Sustaining
Tropical Forests. Island Press, Washington, D.C.
Plumptre A.J. 1995. The importance of ‘seed trees’ for the natural regeneration of selectively logged
tropical forest. Commonwealth Forestry Review 74: 253–258.
Plumptre A.J. 1996. Shifting Cultivation Along the Trans-African Highway and Its Impact on the
Understory Bird Community in the Ituri Forest, Zaire. University of Oxford, Unpublished
report.
Short J.C. 1983. Density and seasonal movement of forest elephant (Loxondonta africana cyclotis
Matschie) in Bia National Park, Ghana. African Journal of Ecology 21: 175–184.

[336]
 1397

Smith J., van de Kop P., Reategui K., Lombardi I., Sabogal C. and Diaz A. 1999. Dynamics of
secondary forests in slash-and-burn farming: interactions among land use types in the Peruvian
Amazon. Agriculture, Ecosystems and Environment 76: 85–98.
Snook L.K. 1996. Catastrophic disturbance, logging, and the ecology of mahogany: ground for
listing a major tropical timber species in CITES. Botanical Journal of the Linnean Society 122:
35–46.
Sokal R.R. and Rohlf F.J. 1981. Biometry: the Principles and Practice of Statistics in Biological
Research. W. H. Freeman and Company, NY.
Southgate D., Sierra R. and Brown L. 1991. The causes of tropical deforestation in Ecuador: a
statistical analysis. World Development 19: 1145–1151.
Struhsaker T.T. 1997. Ecology of an African Rain Forest: Logging in Kibale and the Conflict
Between Conservation and Exploitation. University Press of Florida, Gainesville, Florida.
Struhsaker T.T., Lwanga J.S. and Kasenene J.M. 1996. Elephants, selective logging, and forest
regeneration in the Kibale Forest, Uganda. Journal of Tropical Ecology 12: 45–64.
ter Braak C.J.F. 1987. Ordination. In: Jongman R.H.G., ter Braak C.J.F. and van Tongeren
O.F.R. (eds), Data analysis in community and landscape ecology. Cambridge University Press,
Cambgridge, UK.
Thomas S.C. 1991. Population densities and patterns of habitat use among anthropoid primates of
the Ituri Forest, Zaire. Biotropica 23: 68–83.
Verissimo A., Bareto P., Tarifa R. and Uhl C. 1995. Extraction of a high-value resource in
Amazonia: the case of mahogany. Forest Ecology and Management 72: 39–60.
White L.J.T. 1994. The effects of commercial mechanized selective logging on a transect in lowland
rainforest in Lopé Reserve, Gabon. Journal of Tropical Ecology 10: 313–322.
Whitmore T.C. 1999. Arguments on the forest frontier. Biodiversity and Conservation 8: 865–868.
Wijdeven S.M.J. and Kuzee M.E. 2000. Seed availability as a limiting factor in forest recovery
processes in Costa Rica. Restoration Ecology 8: 414–424.
Wilkie D.S. and Finn J.T. 1990. Slash-burn cultivation and mammal abundance in the Ituri forest,
Zaire. Biotropica 22: 90–99.
Wilkie D.S., Sidle J.G. and Boundzanga G.C. 1992. Mechanized logging, market hunting, and a
bank loan in Congo. Conservation Biology 6: 570–580.
Wilkie D.S., Shaw E., Rotberg F., Morelli G. and Auzel P. 2000. Roads, development, and con-
servation in the Congo basin. Conservation Biology 14: 1614–1622.
Witte J. 1992. Deforestation in Zaire: logging and landlessness. The Ecologist 22: 58–65.
Wolfire D.M., Bruner J. and Sizer N. 1998. Forests and the Democratic Republic of Congo:
opportunity in a time of crisis. World Resources Institute, Washington, D.C.
Zahawi R.A. and Augspurger C.K. 1999. Early plant succession in abandoned pastures in Ecuador.
Biotropica 31: 540–552.

[337]
Biodiversity and Conservation (2006) 15:1399–1415
Springer 2006
DOI 10.1007/s10531-005-5410-0

Forest management considerations for conservation

of Black Woodpecker Dryocopus martius and
White-backed Woodpecker Dendrocopos leucotos
populations in Quinto Real (Spanish Western
Pyrenees)

ALFONSO GARMENDIA1,*, SUSANA CÁRCAMO2 and

OSCAR SCHWENDTNER2
1
Department of Agroforest Ecosystems, Polytechnic University of Valencia, E.T.S.M.R.E.
(‘‘Agrı´colas’’), Blasco Ibáñez 21, 46010 Valencia, Spain; 2BASOA, C/Tafalla 20, Pamplona, Spain;
*Author for correspondence (e-mail: [email protected])

Received 9 November 2004; accepted in revised form 5 April 2005

Key words: Black woodpecker, Dead wood, Dendrocopos leucotos ssp. lilfordi, Dryocopus martius,
Forest structure, Population dynamics, Spain, Sustainable forest management, White-backed
woodpecker

Abstract. The woodlands of Quinto Real (Quinto Real, Erreguerena and Legua Acotada) are a
3,000 hectare beech (Fagus sylvatica) forest managed by the shelterwood system applied to even-
aged (regular) stands. This study analyses how forest management determines the local distribution
of the white-backed woodpecker (Dendrocopos leucotos) and black woodpecker (Dryocopus mar-
tius) and its relationship with the type, structure and size of the stands used for nesting by both
species, as well as their dead wood requirements. The most suitable nesting habitat of both species
is the mature forest (stands of regular large final crop trees), but the size of the mature fragments
and a minimum quantity of dead wood is also important.

Introduction

Woodpeckers (Picidae), particularly the white-backed (Dendrocopos leucotos)

and black woodpeckers (Dryocopus martius), are associated with the existence
of mature forests (Voous 1947). The black woodpecker and the white-backed
woodpecker are two of the most threatened species in the Pyrenean Region of
Spain. They need large areas of mature deciduous forest, particularly wood-
lands with beech (Fagus sylvatica), or beech-fir (Abies alba) or beech-pine (Pinus
spp.) mixed forests (Cuisin 1967; Purroy et al. 1990; Fernández et al. 1994). The
surfaces of these forests are decreasing and those that still remain are variously
degraded in most of continental Europe (Cramp 1985; Avery and Leslie 1990),
especially in Mediterranean countries (Purroy et al. 1990; Tellerı́a 1992).
Two legal instruments that were recently approved by the Regional Gov-
ernment of Navarre, Spain – The forest plan (Gobierno de Navarra 1998), and
The Biological Diversity Conservation Strategy (Gobierno de Navarra 1999) –
are based on the sustainable management of natural resources. In Quinto Real

[339]
1400

woodland, this environmental standard involves forestry management and

offers the chance to provide, if not an optimum, at least a suitable habitat for
these species.
Woodpecker species are an interesting group as ‘umbrella species for bio-
diversity conservation (Simberloff 1999; Fleishman et al. 2000; Fleishman et al.
2001; Roberge and Angelstam 2004) because they need large areas of well
conserved forest with little alteration of their structure and with a sufficient
amount of old trees and recently dead large trees (Hogstad 1970; Angelstam
1990; Angelstam and Mikusinsky 1994; Mikusinsky and Angelstam 1997;
Murphy and Lehnhausen 1998; Imbeau and Desrochers 2002; Butler et al.
2004). Therefore, the modern forestry practices of removing old and dead trees
can compromise their conservation (Nilsson 1992, Tucker and Heath 1994).
The Pyrenees white-backed woodpecker (Dendrocopos leucotos ssp. lilfordi)
population is particularly interesting from a bio-geographical perspective. The
Pyrenees represent the south – western fringe of its world distribution (Fern-
ández et al. 1994), and it has been catalogued as endangered in Spain (Blanco
and González 1992). This subspecies has a patchy and discontinuous distri-
bution in mountains of southern Europe, such as the Pyrenees, the Apennines
and the Carpathian ranges (Voous 1947). The Pyrenees population is the only
one in the Iberian Peninsula and its conservation is therefore very important.
This population has been estimated to be composed of 60–70 pairs, of which
approximately 20% can be found in the woodlands of Quinto Real (Fernández
et al. 1994).
The white-backed woodpecker feeds mainly on wood-boring beetle larvae,
mostly Cerambycidae. Therefore, the availability of dead wood in this area is an
important factor for its survival (Aulén 1988, Aulén and Lundberg 1991). Of all
Picidae, this species is the most specialised in feeding habits. Thus, it is probably
the most vulnerable to changes in forestry management (Conner 1979).
Another limiting factor for the distribution of this species is the presence of
optimum nesting sites. During the breeding season, foraging is confined to
relatively small areas around the nest site. This forces this species to choose
nesting sites with abundant insects, such as groups of dead trees and borders
between different forest areas (Fernández et al. 1994).
The black woodpecker (Dryocopus martius) is widely distributed through-
out northern and temperate forests of Europe and Asia. It inhabits mature
forests, where there are usually beech trees. However, in northern Europe and
Siberia it is also found in coniferous forests. In the Iberian Peninsula, this
species occupies the beech forests of the Pyrenees and the Cantabrian Moun-
tains (Martinez-Vidal 1999). It has also been spotted in the relict beech forests
situated in the central portion of the Iberian Peninsula, but it seems not to be
an established population, but rather composed of young dispersing individ-
uals (Brooks 1985). It has a broad feeding spectrum: most of its food consists
of ants, although it also eats many kinds of wood-boring beetle larvae. It
frequently feeds on the ground, digging out ant nests or forages in the stumps
of felled trees and other types of dead wood (pers. obs.).

[340]
 1401

The black woodpecker is an indicator species of mature forests. It is espe-

cially reliant upon dense, tall stands, and tends to disappear when the forest is
degraded (Brooks 1985, Fernández and Azkona 1996). Nevertheless, a limited
use of forest resources can be beneficial for this species if small clearings are
created and a considerable amount of standing dead wood are left over in the
process. Indeed, this can increase the availability of its prey (Brooks 1985).
The conservation of the black woodpecker and the white-backed wood-
pecker depends, to a large extent, on how the beech forests they inhabit are
managed. Inappropriate management may lead to the disappearance of one or
both of these species. Consequently, it is very important to implement suitable
management programmes in these woodlands.
The most widely-used cutting method in the beech forests of Navarre is the
shelterwood system. Current management has resulted in the development of
patches of different ages. Protected patches have also been designated where no
exploitation can be carried out for logistical reasons. On the other hand, there
is a legal obligation to designate non-exploitable areas on at least 5% of the
area (Gobierno de Navarra 1990,1992) which involves the creation of several
core ‘biodiversity reserves’ areas, that complement other extensive measures for
protecting the habitat.
During spring 1993, Fernández and Azkona (1996) conducted a census of
both species in the Quinto Real group of woodlands. The study revealed that
the population density of the two species was relatively high in all the study
area. They linked the presence of territories for both species with forest stands
with high basal areas (>20 m2/ha). They also noticed a clear overlap between
the locations of territories of both species. Dead wood amount in these
woodpeckers habitats have not been quantified for this area, and the existence
of dead wood threshold value has never been tested.
The objective of this study was to establish how forest management affects
woodpeckers’ density. In particular, the addressed question is how the type and
size of the forest patches, and the amount of dead wood, affects the nesting sites
selection for both species. Do these species utilize all the forest stands for their
territories or do they prefer one? Which is the threshold stand size of the pre-
ferred type? What is more important in this area, the stand type and its size or
the dead wood amount? Do all kinds of dead wood serve the same or there are
some preferred diameters? Is it really necessary that the dead wood is standing,
or could it be lied down? These are the questions addressed in this article.

Methods

Study area

The study was carried out in the Quinto Real group of woodlands, located in
northern Navarre in the Baztán and Erro valleys on the Spanish side of the
Pyrenees. The Quinto Real area consists of three woodlands: Quinto Real

[341]
1402

(1666 ha), Erreguerena (941 ha) and Legua Acotada (907 ha), listed under
numbers 2, 3 and 4 in the Navarre’s Public Utility Woodlands Catalogue
(Gobierno de Navarra 1998). These woodlands are some of the best-preserved
beech forests in the Pyrenees. As a result, the Regional Government of Navarre
recently proposed them as a Place of Community Interest within the frame-
work of the European Union’s Natura 2000 Network. Forest management in
this area is governed by a Management Project and its subsequent reviews
(Schwendtner and Larrañaga 2001).
The Quinto Real group of woodlands has been managed for timber pro-
duction since 1904. Extensive areas of regeneration, resulting from shelterwood
system harvesting, are present. This kind of exploitation was particularly
intensive in the period 1950–1970. There are also mature areas where thinning
has been carried out with varying degrees of intensity, while others have not
been harvested in the last 70 years. Other areas are characterised by their
heterogeneity and unevenness as a result of high-grading of the valuable
timber.

Forest characterization

The Quinto Real Natural Resources Management Plan carried out a detailed
inventory for the different forest stands and their classification according to
structural criteria (Schwendtner and Larrañaga 2001). Table 1 provides
information on the different stand types on which the three woodlands of
Quinto Real were divided. A total of 397 different homogenous forest patches
(stands) were distinguished in the three woodlands. The following variables
were measured to characterize the forest structure: dominant height (DH),
measured as the mean height of those trees with an average diameter, excluding
the stems under 20 cm of diameter, except for the young stands (RY), in which
only stems under 10 cm of diameter are excluded; basal area (BA) of all stems
over 10 cm of diameter; average diameter (AD) of the stems over 10 cm,
measured at breast height; and average age (AA) of the stand, based on several
individuals (approximately 1%) of each stand, by counting the growth rings on
wood samples. All these measures were made on all stands over all the stand
area for management purposes.
Site quality is a variable that is calculated from the relationship between the
mean growth rate and tree age. Site quality can be categorized on a scale
ranging from I to V: very good and good quality sites are I and II, while III and
IV would represent intermediate qualities, and V poor quality sites
(Schwendtner and Larrañaga 2001). In better quality sites, trees grow more
rapidly that on poorer quality sites, where the harvesting cycle is not as rapid.
Schwendtner and Larrañaga (2001) proposed to exploit site qualities I and II,
but advised against timber harvest on the lower quality sites.
Basal area of standing dead trees (DBA) and the amount of felled trunks
were also determined for each stand.

[342]
 1403

Table 1. Stand classification by management objectives on the Quinto Real Natural Resources
Management Plan (Schwendtner and Larrañaga 2001).

Stand type Abreviation Principal characteristics

1. Stand of regular large RLF Mature forest. Average tree diameter >45 cm.
final crop trees Suitable for final cutting.
2. Stand of regular medium RM Medium sized and aged stand. Average tree diameter
sized crop trees from 20 to 45 cm. For intermediate cuttings
with economic value.
3. Stand of regular RY Young stand. Average diameter from 10 to 20 cm.
young trees For thinning without economic value.
4. Heterogeneous and HI Mixed stand, It is heterogeneous when it has
irregular stand different species and irregular when it has
different age classes and structures.
Various diameters.
5. Low forest stand LF Low forest stand. Generally on sites
of poor quality.
6. Open large final crop trees OF Open zones in regeneration process.
Some residual large trees.
Basal Area <15 m2/ha.

Woodpeckers census

The distribution and density of the black woodpecker and the white-backed
woodpecker were established by determining their breeding territories during
spring 2001. The low density of both species makes sampling difficult.
However, since they are highly territorial animals, the location of breeding
territories was used to census these species (Svenssons 1979; Tellerı́a 1986;
Bibby et al. 1992). The method was the same as the one used in the previous
censuses by Fernández and Azcona (1996), so that the densities and distri-
bution of territories could be compared. Recordings of the birdcalls and
tapping patterns of both species were used as decoys for locating the breeding
territories. For the density estimates, the ‘open land’ or patches with no trees
were excluded of the total study area (3200 ha) and not sampled, though all
the other stands of the study area were sampled. To attract territorial birds
or to provoke their response during the search, the tapped calls were played
every one or two hundred meters, alternating with periods of 30 s of silence,
thus permitting to detect the bird response and to locate the individual. Once
an individual was located, it was followed to locate the nest and the partner,
registering the stands they defended and used for foraging. The wood boring
signs were only taken into account when they were extremely recent and very
abundant and they were only used to determine areas where investigation
should be intensified. Alone they were not considered as sufficient proof of
the existence of a nesting area. Sightings of non-territorial individuals were
excluded.

[343]
1404

Habitat selection

The effect of different forest variables on the breeding area selection of the two
bird species (presence or absence of bird territories) was investigated. There-
fore, the conclusions that may be drawn from this study concern the territories
used for breeding and not the habitats used during other seasons. For the study
on dead wood, the three woodlands of Quinto Real were subdivided into
‘quarters’ that represent smaller management units. Quinto Real was divided
into three quarters, and Erreguerena and Legua Acotada into two quarters
each (Schwendtner and Larrañaga 2001).

Statistical analysis

The v2 test was used to compare the presence/absence frequencies of wood-

peckers in the different stand types and site qualities. Only the stands included
in the territories were considered as presence, wandering individuals were not
taken into account. As the number of low forest (LF) stands was small, these
data were eliminated from the statistical analysis.
To estimate the threshold stand size of the type preferred by woodpeckers,
different size classes have been separated to compare the percentages of stands
of different size classes included in territories. Also an analysis of variance
(ANOVA) has been made to compare the sizes of these stands used and not
used by each species.
To compare quantitative variables for the forest stands used by each
woodpecker species in their breeding territories, analysis of variance (ANOVA)
was combined with the comparison of means of each group (LSD, student’s t).
Also the differences between woodlands and stand types have been tested.
To reduce the complexity of the data set and to detect the interactions
between species occurrence and environmental variables, a principal compo-
nent analysis (PCA) based on the correlation matrix was carried out. In this
analysis only beech forest stands with data available for the four variables were
used (n = 243).
To determine the importance of the amount of dead wood in the territories,
different ANOVA analyses have been done at different scales. A first analysis
was made comparing the amount of dead wood in the stands included and not
included in each woodpecker territories. As no results were found with this
analysis, a more detailed analysis was made, repeating it for each quarter and
for each stand type. Also, the amount of dead wood in large RLF stands is
analysed.
The comparisons between woodlands were made on means of each quanti-
tative variable for each stand, weighted by the area of the stand. For all the
ANOVA analyses, a test for normality has been carried out to fulfil the
assumptions. All of the statistical analyses were conducted using the Statistica
4.5
for windows from Statsoft, Inc.

[344]
 1405

Results

Forest stand classification

From the analysis of forest stands, it appears that approximately 40% of the
area consist of high-quality sites (I and II), 30% of intermediate quality sites
(III and IV), and 30% of low quality sites. In the low quality sites are included
the non-exploitable sites due to environmental constraints (known as protec-
tion patches). Extraction priority was given to areas that are more productive
or easily accessible (Schwendtner and Larrañaga, 2001), so there is a certain
imbalance in the age histogram according to site qualities (results not shown).
Stand type distribution in the studied area is explained in Table 2.

Census

As can be seen in Table 3, there were 11 reproductive pairs of white-backed

woodpecker (7 pairs in Quinto Real, 1 pair in Erreguerena, 1 pair between
Quinto Real and Erreguerena and 2 pairs in Legua Acotada) and 14 pairs of
black woodpecker (7 pairs in Quinto Real, 4 pairs in Erreguerena, 1 pair
between Quinto Real and Erreguerena and 2 pairs in Legua Acotada).
Comparing these data with the densities found by Fernandez and Azkona
(1996) on spring 1993 (Table 3), can be seen that the population remains stable
with a downward trend in the case of the white-backed woodpecker and an
upward trend in the case of the black woodpecker in the whole study area, but
a clear decline is detected for the white-backed woodpecker in Erreguerena and
for the black woodpecker in Legua Acotada. This decline is compensated in
both cases with increases in Quinto Real and for the black woodpecker in
Erreguerena.

Table 2. Area (hectares), number of stands and average stand size (hectares) of the tree woodlands
of Quinto Real: Quinto = Quinto Real; Erreg. = Erreguerena; Legua = Legua Acotada.

Stand Area (ha) Number of stands Average stand size (ha)

type
Quinto Erreg. Legua Total Quinto Erreg. Legua Total Quinto Erreg. Legua Total

1. RLF 355.1 300.7 260.4 916.2 33 32 17 82 10.8 9.4 15.3 11.2

2. RM 382.6 140.3 265.3 788.2 54 16 8 88 7.1 8.8 14.7 9.0
3. RY 236.7 233.1 148.6 618.3 32 22 17 71 7.4 10.6 8.7 8.7
4. HI 346.1 152.0 135.4 633.5 58 38 19 115 6.0 4.0 7.1 5.5
5. LF 200.5 0.0 6.3 206.8 31 0 1 32 6.5 6.3 6.5
6. OF 18.3 3.6 16.1 37.9 4 1 4 9 4.6 3.6 4.0 4.2
Total 1539.3 829.7 832.0 3200.9 212 109 76 397 7.3 7.6 10.9 8.1
The abbreviations of the stand types come from Table 1: RLF, stands of regular large final crop
trees; RM, stands of regular medium sized crop trees; RY, stands of regular young trees; HI,
heterogeneous and irregular stands; LF, low forest stands.

[345]
1406

Table 3. Number of territories and density (pairs/Km2) of each woodpecker species

(WW = White backed woodpecker, BW = Black woodpecker) in all the forested area (Total) of
the three Woodlands, and in each of them (Quinto = Quinto Real, Legua = Legua Acotada).
The 2001 are the census carrried by the authors of this article and the 1993 are the census carried by
Fernández and Azcona in 1993 (see Fernández and Azcona 1996).

2001 1993

WW BW WW BW

Number of territories
Total 11 14 12 13
Quinto 7.5 7.5 6 6
Erreguerena 1.5 4.5 4 4
Legua 2 2 2 3
Density (pairs/Km2)
Total 0.34 0.44 0.38 0.41
Quinto 0.49 0.49 0.48 0.48
Erreguerena 0.18 0.54 0.38 0.38
Legua 0.33 0.33 0.33 0.49

The territories of both species reveal certain mobility compared to the 1993
census carried out by Fernández and Azkona (1996). Many of the territories
are still located in exactly the same forest stands. Others clearly occupy the sites
situated between former territories, presumably using the areas that were less
used on the 1993 territory distribution. In territories where felling has been
carried out, the pairs affected have moved out, probably to other unoccupied
patches.

Preferences according to stand classification and site qualities

Frequencies of sightings of the two species in the different forest types show
significant differences (v2, p lt 0.005 for the black woodpecker and p < 0.00001
for the white-backed woodpecker). Both species show a clear preference for
regular large final crop stands (RLF). For both species there is also a distinct
negative selection against heterogeneous (mixed with conifers) and uneven
(mixed ages) stands (HI); this is less marked in the case of the black wood-
pecker. Sightings in regular medium sized crop stands (RM) and regular young
stands (RY) do not reveal any significant differences.
Although most of the territories cover various forest stand types, nearly all
cases – except in one black woodpecker and one white-backed woodpecker
territories – include an RLF stand. When nests were found, they were usually
located in this stand type, while the others – mainly RM and RY – are also
defended and used for feeding. In the two territories identified in a place
without RLF, there were RM stands of a considerable age (on the boundary
of stands regarded as RLF). In one of these cases, the territory may have
been moved from a recently exploited mature stand (RLF). Nevertheless, in

[346]
 1407

Table 4. Means comparison (LSD) of dominant height (DH), average age, average diameter (AD)
and basal area (BA) from the Quinto Real Natural Resources Management Plan (Schwendtner and
Larrañaga 2001).

Stand use DH (m) Age (years) AD (cm) BA (m2/ha)

Without territory 20.1 D 97.8 C 26.4 C 22.7 B

WT without BT 21.8 C 109.9 B 31.5 B 23.1 B
BT without WT 22.8 B 104.9 B 27.4 C 22.1 B
BT and WT 25.6 A 148.7 A 33.8 A 26.5 A
Woodlands
Quinto Real 20.4 B 108.3 A 28.0 AB 21.5 C
Erreguerena 22.1 A 106.5 A 28.2 A 25.8 A
Legua Acotada 21.9 A 96.5 B 27.1 B 23.2 B
Stand types
RLF 26.0 A 155.3 A 36.7 A 26.9 B
RM 21.9 B 91.2 D 28.8 B 28.5 A
RY 15.8 D 45.3 E 15.0 D 13.0 E
HI 20.0 C 119.6 B 28.4 B 21.9 C
LF 15.9 D 113.3 C 25.6 C 16.9 D
Stands with black woodpecker territories (BT), with white-backed woodpecker territories (WT),
both or none, the different woodlands, and stand types (see Table 1: RLF, stands of regular large
final crop trees; RM, stands of regular medium sized crop trees; RY, stands of regular young trees;
HI, heterogeneous and irregular stands; LF, low forest stands) are compared. Within each com-
parison, different letters represent significative mean differences (p < 0.05).

all cases in which a territory is included in only one big stand (3 white-backed
woodpecker territories and 4 black woodpecker territories), this is a RLF
stand.
No significant differences were found between the site quality of the stands
used by each species with those not used. Therefore, woodpeckers do not
appear to choose stands for their site quality, but rather for the physiognomic
characteristics of the forest which is best reflected in the stand classification (see
Table 1 and 4). This supports the option for only harvesting stands with high
site quality, and conserving the poorer quality sites with well-preserved mature
forest (RLF).

Stand size

One of the typical questions that arises from the management of these forests
concerns the minimum stand size that must be left as mature forest for these
species to establish their territory. To answer this question, a study was con-
ducted on the size differences between RLF stands where territories were
present and those where they were not. Figure 1 illustrates that the stands
where territories for both species were present were considerably larger than
those where there were not.
The RLF stands in which the black woodpecker appears have an average
size of 24 ha, while those of the white-backed woodpecker average 19 ha. The

[347]
1408

Figure 1. Analysis of variance on the regular large final crop trees (RLF) forest stands size
according to their occupation for each woodpecker species: BN, stands where the black wood-
pecker territories are studied; WN, stands where the white-backed woodpecker does not appear;
WT, stands where the white-backed woodpecker territories are situated. The rectangles represent
the standard error and the lines the standard.

Figure 2. Percentage of RLF stands of different sizes included in territories of black woodpecker
(triangles) and white backed woodpecker (circles).

actual surfaces required are probably somewhat greater, since some territories
occupy more than one mature stand. In fact, most RLF stands with an area
exceeding 30 ha (8 stands) are included in a territory, except three stands in
which recent cutting was carried out. In Figure 2 can be seen the percentages of
the different sizes of RLF stands that are included in the territories of each
species. The total percentage for each species is higher that 100 because one
territory usually extend over several stands. The occupation percentage in large
stands is much higher than that for small stands.

[348]
 1409

Forest physiognomy

Although the stand types reveal considerable differences with regard to

habitat selection by Picidae, it is still basically a subjective classification that
depends on the criterion of the forest engineer responsible for the manage-
ment plan. For this reason, and in order to assess the classification more
objectively, an analysis of the quantitative variables was performed using
variance analysis (ANOVA), LSD means comparison and principal compo-
nent analysis (PCA).
There was a considerable difference between stands where the two species
shared territories and the other stands (see Table 4). These stands had the
tallest, oldest and largest trees, and also the largest basal area. Stands with a
territory of only one of both species have intermediate values, and those with
no territories have the lowest values. The white-backed woodpecker clearly
prefers forests with a larger average tree diameter. However, when both
woodpeckers do not occur in the same area, it selects stands of a lower height
than those chosen by the black woodpecker.
The first PCA axis represents 72.6% of the total variance, while the second
axis represents 12.8% (Figure 3a). On the same axes, Figure 3b shows the
larger RLF stand of each territory for the black and white-backed woodpecker.
It can be noted that the first axis is a good predictor for the presence or absence
of each species. Their territories were linked to high values of the first axis,
which corresponds to high values of the four morphometric variables used. The
second axis separates the four variables, but there is not a clear relationship
with territories and it is much less reliable.

Dead wood

When stands of the three woodlands of Quinto Real were used, no significant
differences were found for total dead wood (number of trunks per ha) between
stands used by each species or those that were not used, probably because
many stands with considerable amounts of dead wood were not used by neither
of the two species. Nevertheless, there were notable differences in the amount
of thin dead wood (from 10 to 20 cm), i.e., the type of dead wood most
abundant and most representative (Table 5).
Comparing only stands from quarters that had comparatively little dead
wood, significant differences were found among stands where the white-backed
woodpecker appeared, particularly for the 20–30 cm range (p < 0.05). This
species did not choose stands with more dead wood in quarters where it was
abundant, but in those quarters with little dead wood, it did a positive selection
for stands with more dead wood.
For the dead wood in RLF, considerable differences might be identified
between stands where the white-backed woodpecker was found and stands
where it was absent for the two largest diameter classes of dead wood

[349]
1410

Figure 3. Principal components analysis of the stands described by the four variables that describe
the forest structure. The first two axis do account for the 85% of the variance. (a) scatter diagram
of all the stands; symbols indicate the stand classification according to Table 1. RLF, stands of
regular large final crop trees; RM, stands of regular medium sized crop trees; RY, stands of regular
young trees; HI, heterogeneous and irregular stands; LF, low forest stands. (b) scatter diagram of
the larger RLF stand of each woodpecker territory on the same axis that (a). Circles represent the
black woodpecker territories and triangles the white backed woodpecker territories. Also the de-
scriptors of the four variables that describe the forest structure are represented. BA, basal area;
DH, dominant height; AD, average diameter; AA, average age.

(20–30 cm and >30 cm; p < 0.001). This implies that the species clearly
chooses those RLF that contain a larger amount of dead wood. This
relationship remains significant comparing the differences between large-size
RLF stands (>15 ha) whether the white-backed woodpecker is present or not.
For the black woodpecker no differences were found for all the comparisons.

[350]
 1411

Table 5. Standing dead wood on the seven quarters of Quinto Real Woodlands: three in Quinto
Real (Q1,Q2, Q3), two in Erreguerena (E1, E2) and two in Legua Acotada (L1, L2).

Quarters Standing dead trees

Trunk diameter

10–20 cm 20–30 cm >30 cm DBA

Q1 22.2 A 5.4 AB 2.0 A 0.915 A

Q2 17.4 A 5.7 AB 1.1 A 0.728 AB
Q3 12.0 B 4.6 AB 1.6 A 0.645 AB
E1 7.0 B 4.5 AB 0.4 A 0.394 B
E2 6.4 B 6.9 A 1.7 A 0.666 AB
L1 22.7 A 3.5 AB 0.7 A 0.662 AB
L2 4.0 B 1.2 B 0.6 A 0.208 B
Trunk number per ha is given for each of the three diameter classes and dead trees basal area
(DBA) is given in m2/ha. Within each column, different letters indicate significant differences
(p < 0.05).

Discussion

Although the overall densities of the two woodpecker species remain rather
stable for each of the woodlands of Quinto Real, according to the 1993 census
(Fernández and Azkona 1996) and the one obtained in 2001, it appears that the
white-backed woodpecker is declining in Erreguerena, and the black wood-
pecker is declining in Legua Acotada and increasing in Erreguerena. These
differences are probably due to forestry management. Nevertheless, the Quinto
Real populations have increased, which seems to indicate that management has
been more appropriate than in Erreguerena and Legua Acotada. It does not
seem that this difference can be explained by other ecological variables, because
topography, climate, and other non-anthropic factors are quite similar.
Over the years both species remained faithful to their breeding territories (as
also observed by McClelland and McClelland 1999). By comparing both
censuses (1993–2001), it appears that there have been some shifts of territories,
possibly due to forestry activities. A possible explanation could be that changes
on the forest structure due to felling in a woodpecker territory, (a RLF stand is
converted on a RY stand) may cause a territorial movement towards another
place with more mature forest, thus ‘pushing’ adjacent territories.
One of the most obvious conclusions is that the most suitable habitat for
both species is the mature forest stands (RLF). These are also referred to in the
Natural Resources Management Plan (Schwendtner and Larrañaga 2001) as
stands where final cutting is most likely to be carried out because these sites
have the largest amount of timber trees for felling. If the exploitation of these
woodlands by town councils is intensified, these stands will soon become
extremely scarce.
A negative selection is observed, in both species, against heterogeneous and
irregular stands. The same occurs within the stands in which other species

[351]
1412

rather than beech are dominant (Larix, Pinus, Quercus). Although the black
woodpecker occasionally feeds in these forests, the white-backed woodpecker
has been found exclusively in monospecific beech forests.
The dominant height (26 m), age (149 years), mean diameter (34 cm) and
basal area (27 m2/ha) of stands where both the black and white-backed
woodpecker territories coincide may help to determine the characteristics that
the stands left for conservation should have (5% of the total). A sufficient
amount of this type of stand should be left in the rest of the woodlands to
ensure that the population of these species do not decrease.
It has been confirmed that the size of the RLF stand is another decisive
factor for both species in establishing breeding territories, with minimum sizes
close to 20–30 ha. As these are territorial birds, it does not appear to be a good
idea to leave all the RLF stands grouped in one area. It seems much more
appropriate to keep sufficiently large (>30 ha) RLF patches separated from
one other (the number of patches depends on the desired size of the popula-
tion). Moreover, given that there is no correlation between the site quality and
the distribution of territories, it is advisable to concentrate exploitation in the
best sites with short felling cycles. The worst sites should be left unexploited in
order to fulfill the above objectives.
The amount of dead wood does not appear to be the main factor for
choosing breeding territories for these birds at the scale of the whole study
area, probably because it is very abundant in most of the area and therefore it
is not a limiting factor. In fact, when the analysis is concentrated in the areas
where dead wood is scarcer, this variables becomes an important factor for the
distribution of the white-backed woodpecker territories, but the black wood-
pecker territories distribution do not seem to be affected by this variable. This
can be explained by the diet of the black woodpecker, which feeds mainly on
ants, and is not so dependent on dead wood.
The white-backed woodpecker territories distribution shows a relationship
with the amount of thicker standing dead wood in the quarters where total
dead wood is less abundant. Also the amount of dead wood seems important
when comparing the territories occupancy frequencies between all RLF stands
and also only with large RLF stands. But not all dead wood classes are of the
same importance: thick standing dead wood seems to be more important than
other classes of dead wood. This reinforce the importance of well conserved,
large enough RLF stands, also with sufficient amount of standing thick dead
trees, that could be increased by ringing some trees if necessary. It is more
important in the places where the surroundings have less dead wood. There are
other authors that have also found important the amount of dead wood for the
woodpeckers, in particular with the specialist species (Angelstam et al. 2003;
Butler et al. 2004).
Although the number of felling activities is insufficient to analyse their effect
from a statistical point of view, their effects on the territories of both species
seem to be very clear. When a RLF stand included in a 1993 territory disap-
peared, the territory has ‘moved’ to include another RLF stand in it. In this

[352]
 1413

study, the movement of territories have not been analysed, and we do not know
if the new territories (or even the old ones) are done by the same individuals or
different ones. As example, the black woodpecker territory that has disap-
peared in Legua Acotada, correspond to a place where the RLF stand included
in it has been fallen down, and no mature stand can be found nearby.
At the moment in Legua Acotada it seems unlikely that the population of
both species can be increased to levels similar to those in Quinto Real unless
management is changed, for example by using smaller stand sizes and leaving
the small number of mature RLF stands (there were 3 in 1993, but one has
already been harvested). There is a territory of each species in each of these
stands, but the nearby forests are regular young stands (RY) or final cuttings
that are extremely homogenous and have been exhaustively ‘cleaned out’. As a
result, they are of no use to the picidae. If the RLF stands that remain are cut
down, the territories in them will surely disappear.
Nevertheless, the situation in Erreguerena is slightly more encouraging.
There are sufficiently large RLF stands in this area. Although it appears that
several white-backed woodpecker territories have disappeared due to recent
cuttings, they may have established in other RLF stands that were unoccupied.
Black woodpecker seems to be attracted by these felling activities, possibly due
to an increase of felled dead wood and therefore an increase of the amount of
ants. Appropriate management would mean cutting the unoccupied stands in
order not to disturb existing pairs. As the territories may vary in location, it is
necessary to conduct yearly censuses in order to determine the situation before
planning felling activities.
The patchwork situation in Quinto Real – stands that are relatively small,
with a relatively high abundance of large enough RLF stands sufficiently
separated ones from the others – has allowed the creation of a large number of
breeding territories. Nevertheless, it is advisable to exploit only the stands that
remain between territories and leave those that currently contain breeding
territories.
Another general recommendation from the results is to leave enough dead
wood in all stands and to leave dead trees standing because this is where the
white-backed woodpecker mainly feeds. Girdling can be carried out instead of
harvesting in some cases, since this technique leaves standing dead wood. Some
management plans which aimed at protecting yew trees (Taxus baccata) by
girdling the beech trees that overshadow them, may be also beneficial to spe-
cialist species (see Carlson 2000). White-backed woodpeckers are also benefi-
cial for the health of the beech forest as it eats a lot of the forest plagues, and
keeps them under control (Butler and Schlaepfer 2003).

Acknowledgements

We gratefully acknowledge Carmelo Fernández for his support and advice

from the outset of this study, and Miguel Ángel Salas for the information he

[353]
1414

provided about Picidae. Our thanks also to the ‘‘Sección de Montes del Servicio
de Conservación de la Biodiversidad del Gobierno de Navarra’’ for their support
and to the Foreign Language Co-ordination Office at the Polytechnic Uni-
versity of Valencia for their help in translating this paper.

References

Angelstam P. 1990. Factors determining the composition and persistence of local woodpecker
assemblages in taiga forests in Sweden – a case for landscape ecological studies. In: Carlson A.
and Aulen G. (eds), Conservation and Management of Wo‘odpecker Populations. Swedish
University of Agricultural Sciences, Department of Wildlife Ecology, Uppsala, pp. 147 – 191.
Angelstam P. and Mikusinski G. 1994. Woodpecker assemblages in natural and managed boreal
and hemiboreal forest – a review. Annales Zoologici Fennici 31: 157 – 172.
Angelstam P., Butler R., Lazdinis M., Mikusinski G. and Roberge J.M. 2003. Habitat thresholds
for focal species at multiple scales and forest biodiversity conservation – dead wood as an
example. Annales Zoologici Fennici 40(6): 473 – 482.
Aulén G. 1988. Ecology and distribution history of the White-backed Woodpecker Dendrocopos
leucotos in Sweden. Ph.D. Thesis, Report 14, Swedish University of Agricultural Sciences
Department of Wildlife Ecology, Uppsala.
Aulén G. and Lundberg A. 1991. Sexual dimorphism and patterns of territory use by the White-
backed Woodpecker Dendrocopos leucotos. Ornis Scandinavica 22: 60 – 64.
Avery M. and Leslie R. 1990. Birds and Forestry. T. & A.D. Poyser, London.
Bibby C.J., Burgess N.D. and Hill D.A. 1992. Bird Census Techniques. Academic Press, London.
Blanco J.C. and González J.L. 1992. Libro Rojo de los Vertebrados de España. ICONA, ed.
Ministerio de Agricultura Pesca y Alimentación, Madrid.
Brooks J. 1985. Handbook of the Birds of Europe, the Middle East and North Africa. Vol IV.
Oxford University Press.
Butler R., Algelstam P., Ekelund P. and Schlaepfer R. 2004. Dead wood threshold values for the
three-toed woodpecker presence in boreal and subalpine forest. Biological Conservation 119:
305 – 318.
Butler R. and Schlaepfer R. 2003. Three-toed woodpecker as an alternative to bark beetle control
by traps?. In: Pechacek P and d’Oleire-Oltmanns W. (eds), International Woodpeckers Sym-
posium. Berchtesgaden, pp. 13 – 26.
Carlson A. 2000. The effect of habitat loss on a deciduous forest specialist species: the White-
backed Woodpecker (Dendrocopos leucotos). Forest Ecology and Management 131: 215 – 221.
Conner R. 1979. Seasonal changes in woodpecker foraging methods: Strategies for winter survival
Pages 95 – 105. In: Dickson J.G., Conner R.N., Fleet R.R., Kroll J.C. and Jackson J.A. (eds),
The role of insectivorous birds in forest ecosystems. Academic Press, NY, pp. 381.
Cramp S. (ed.) 1985. Handbook of the Birds of Europe, the Middle East and North Africa. Vol. 4.
Oxford, Oxford University Press.
Cuisin M. 1967. Essai d’une monographie du Pic noire (Dryocopus martius). Oiseau Rev Fr
Ornithol. 37:163 – 192, 37:285 – 315, 38:20 – 52, 38:103 – 126, 38:209 – 224.
Fernández C., Azkona P. and Lorente L. 1994. Corologı́a y caracterización del hábitat del pico
dorsiblanco (Dendrocopos leucotos) en el Pirineo occidental español. Ardeola 41(2): 135 –
140.SEO.
Fernández C. and Azkona P. 1996. Influence of forest structure on the density and distribution of
the White-backed Woodpecker Dendrocopos leucotos and black Woodpecker Dryocopus martius
in Quinto Real (Spanish western Pyrenees). Bird Study 43: 305 – 313. British Trust for
Ornithology.
Fleishman E., Blair R.B. and Murphy D.D. 2001. Empirical validation of a method for umbrella
species selection. Ecological Applications 11: 1489 – 1501.

[354]
 1415

Fleishman E., Jonsson B.G. and Sjagren-Gulve P. 2000. Focal species modeling for biodiversity
conservation. Ecological Bulletins 48: 85 – 99.
Gobierno De Navarra 1990. Ley Foral 13/1990, de 31 de diciembre. Protección Y Desarrollo del
Patrimonio Forestal de Navarra. Boletı́n Oficial de Navarra núm. 6, de 14 de enero de 1991.
Gobierno De Navarra 1992. Decreto Foral 59/1992, de 17 de febrero. Reglamento de Montes en
Desarrollo de la Ley Foral 13/1990, de 31 de Diciembre, de Protección y Desarrollo del Patri-
monio Forestal de Navarra. Boletı́n Oficial de Navarra núm. 48, de 20 de abril de 1992.
Gobierno De Navarra 1998. Plan Forestal de Navarra. www.cfnavarra.es, .
Gobierno De Navarra 1999. Estrategia Navarra para la Conservación de la Biodiversidad.
www.cfnavarra.es.
Hogstad O. 1970. On the ecology of the three-toed woodpecker Picoides tridactylus (L.) outside the
breeding season. Nytt Magasin for Zoologi 18: 221 – 227.
Imbeau L. and Desrochers A. 2002. Foraging ecology and use of drumming trees by three-toed
woodpeckers. Journal of Wildlife Management 66: 222 – 231.
Martinez-Vidal R. 1999. Hábitat de crı́a del pito negro (Dryocopus martius) en las Sierras de Cadı́ y
Moixeró: caracterización, tipologı́a y pérdidas de árboles nido. Gestión y Conservación de la
biodiversidad en ecosistemas forestales. C.T.F.C, Solsona.
Mcclelland B. and Mcclelland P. 1999. Pileated woodpecker nest and roost trees in Montana: links
with old-growth and forest ‘‘health’’. Wildlife Society Bulletin 27(3): 846 – 857.
Mikusinski G. and Angelstam P. 1997. European woodpeckers and anthropogenic habitat change:
a review. Vogelwelt 118: 277 – 283.
Murphy E.C. and Lehnhausen W.A. 1998. Density and foraging ecology of woodpeckers following
a stand-replacement fire. Journal of Wildlife Management 62: 1359 – 1372.
Nilsson S.G. 1992. Population trends and fluctuations in Swedish woodpeckers. Ornis Svecica 2:
13 – 21.
Purroy F.J. 1972. El pico dorsiblanco Dendrocopos leucotos del Pirineo. Ardeola 20: 145 – 158.
Purroy F.J., Alvarez A. and Pettersson B. 1990. Bosque y fauna de vertebrados terrestres en
España. Ecologı́a 1: 349 – 363.
Roberge J.M. and Angelstam P. 2004. Usefulness of the umbrella species concept as a conservation
tool. Conservation Biology 18(1): 1 – 10.
Schwendtner O. and Larrañaga A. (coords.) 2001. Sexta Revisión de la Ordenación del grupo de
Montes de Quinto Real. BASOA/Gobierno de Navarra.
Simberloff D. 1999. The role of science in the preservation of forest biodiversity. Forest Ecology
and Management 115: 101 – 111.
Svenssons S.E. 1979. Census Efficiency and number of visits to a study plot when estimating bird
densities by the territory mapping method. Journal of Applied Ecology 16: 61 – 68.
Tellerı́a J.L. 1986. Manual para el Censo de los Vertebrados Terrestres. Ed. Raı́ces, Madrid.
Tellerı́a J.L. 1992. Gestión forestal y conservación de las aves en España peninsular. Ardeola 39:
99 – 114.
Tucker G.M. and Heath M.F. (eds) 1994. Birds in Europe – their conservation status. BirdLife
International, Conservation Series No. 3, Cambridge.
Voous K.H. 1947. On the history of the distribution of the genus Dendrocopos. Limosa 20: 1 – 142.

[355]
Biodiversity and Conservation (2006) 15:1417–1424
Springer 2006
DOI 10.1007/s10531-005-0308-4

A reconsideration of the reproductive biology of the

Atlantic forest in the Volta Velha Reserve

STEVEN M. VAMOSI
Department of Biological Sciences, University of Calgary, 2500 University Drive NW, Calgary,
Alberta, Canada, T2N 1N4; E-mail: [email protected]

Received 6 October 2004; accepted in revised form 1 June 2005

Key words: Brazil, Dioecy, Floristic survey, Reproductive biology, Seed shadow effect

Abstract. Published species lists that include breeding system designations of vascular plants are
rare in the primary literature and, thus, can be potentially valuable sources of information for
comparative studies. The published list for vascular plants in the Volta Velha Reserve suffered from
a number of errors, notably applying the designation of monoecious to all species with imperfect
flowers. Here, I reconsider the breeding systems for 97 woody vascular plant species. The majority
of species initially categorized as monoecious are found to be hermaphroditic. I then examine the
relationship between breeding system and numbers of individuals in a 1 ha plot. The mean number
of individuals was marginally higher in dioecious than hermaphroditic and monoecious species
combined. Furthermore, although only 28% of the species were characterized as possessing a
dioecious breeding system, 42% of the individuals encountered belonged to a dioecious species.
These results suggest that dioecious species can, at least under certain circumstances, overcome the
reductions in the number of seed-bearing individuals and mate assurance that accompany pos-
sessing spatially segregated sexes.

Introduction

Understanding the mechanisms responsible for the origination and mainte-

nance of breeding system variation in flowering plants has long been a fun-
damental issue in evolutionary biology (e.g., Darwin 1877, 1878; Geber et al.
1999; Heilbuth 2001). Although the vast majority of angiosperm species are
hermaphroditic (i.e., perfect flowers), a number of other breeding systems also
occur. Dioecy (imperfect flowers, with separate male and female individuals),
for example, is present in approximately 6% of species globally (Renner and
Ricklefs 1995), although its incidence in local floras can be as high as 28%
(Arroyo and Squeo 1990). Although relatively rare and possessing a wide
taxonomic distribution (Yampolsky and Yampolsky 1922; Renner and Rick-
lefs 1995; Heilbuth 2001; Vamosi and Vamosi 2004), many authors have
commented on the associations between these breeding systems and various
ecological and life history traits. Dioecious species, for example, frequently
have a tropical distribution, woody growth form, fleshy fruits, and plain white
flowers (reviewed by Vamosi and Vamosi 2004).

[357]
1418

In an effort to explore additional correlations, I have been searching

monographs, books and papers for information on breeding systems. A recent
article in this journal on the reproductive biology of plants from a rainforest in
southern Brazil provided the breeding systems for 97 woody species in 38
families (Negrelle 2002). Many floristic surveys do not contain species-specific
information (e.g., Flores and Schemske 1984); hence, such lists can be quite
useful. Unfortunately, the list is currently misleading in its present form pri-
marily because of the way in which taxa were scored for breeding system.
Negrelle (2002, p. 891) stated that: ‘definitions of dioecy and monoecy followed
standard definitions, except in those few cases where previous literature iden-
tified monoecious trees as functionally dioecious.’ However, the designation of
monoecious (imperfect flowers, with staminate and carpellate flowers on the
same plant) was applied to all species thought to possess imperfect flowers,
including hermaphroditic and polygamous species. For example, four species
in the family Annonaceae were scored as monoecious, although the vast
majority of the ca. 2300 species in this family are hermaphroditic (Kessler 1993)
and monoecy is known only from a single genus (Uvariopsis Engl.) not
encountered in the present survey. It seems likely that the source of the coding
error stems from an overly liberal interpretation of the etymology of monoecy
(i.e., ‘one home’). Furthermore, I encountered a smaller number of errors in
breeding system designations in the list. For example, all of the Ocotea Aubl.
(Lauraceae) species encountered in the survey were scored as being dioecious.
Two of the seven species are, in fact, hermaphroditic (Rohwer 1986). Although
such errors are nearly inevitable when dealing with breeding systems, their
presence, in addition to the unorthodox scoring method, warranted a recon-
sideration of the entire data set.
In addition to breeding system designations, Negrelle (2002) also provided
potentially valuable data on the number of individuals, basal area occupied,
and importance value (which accounts for density, frequency and dominance)
for each species in a 1 ha plot. Such data can be used to explore the ecological
consequences of spatial segregation, such as reduced mate assurance and fewer
seed-bearing individuals (hereafter, seed shadow effect), that are predicted to
affect plant species with spatially separate sexes (e.g., Pannell and Barrett 1998;
Heilbuth et al. 2001). Reduced mate assurance may be alleviated if the density
of individuals is high, as pollen is more likely to be transferred to females.
Similarly, the disadvantage of a reduction in the number of seed-dispersing
individuals by half that accompanies the evolution of dioecy will be alleviated if
more females are present in a dioecious population.

Methods

The Volta Velha Reserve is located in the Municipality of Itapoá, Santa

Catarina State, Brazil (2604¢ S, 4838¢ W). Although south of the Tropic of
Capricorn, the flora in this Reserve was found to be similar to that of a number

[358]
 1419

of Neotropical sites (Negrelle 2002). A total of 398 species and 99 families

of vascular plants have been collected in the Volta Velha Reserve, of which
the vast majority are angiosperms. In a single 1 ha plot, 734 individuals with
a diameter at breast height (DBH) ‡10 cm from 97 species were sampled
(Negrelle 2002; pp. 910–911).
Information on breeding systems for the 97 species in the 1 ha plot was
obtained from a variety of sources, especially Bush (1995), Charlesworth
(1985), Hutchinson (1959), Kubitzki et al. (1993), Kubitzki (2004) and Watson
and Dallwitz (1992 onwards). I consulted authorities (J. Rohwer, M. Freitas
and J. Pipoly III, and L. Bohs, respectively) for Ocotea, Rapanea (= Myrsine)
and Solanum species. For all species, I first consulted the Families and Genera
of Vascular Plants volumes, and subsequently used other sources for families
that have not yet been catalogued in this series and for genera that have more
than one breeding system represented by their constituent species. A complete
list of sources is available from the author on request.
Because only dioecious species are expected to suffer from a seed shadow
effect (Heilbuth et al. 2001), I compared number of individuals, basal area and
importance value (Negrelle 2002, pp. 910–911) of dioecious species to her-
maphroditic and monoecious species combined. Indeed, hermaphroditic and
monoecious species were not significantly different for any of the measures
(p > 0.55 in all cases). The three measures are related and produced qualita-
tively similar results, thus I present only the results for number of individuals.
I applied a square root transformation on the number of individuals per species
prior to conducting a one-way ANOVA. Means are presented as the square of
the mean of square-root transformed values.

Results and discussion

The main effect of re-scoring breeding systems was, perhaps not surprisingly, to
drastically reduce the apparent prevalence of monoecious species (Table 1). In
agreement with other studies, hermaphroditic species were most common
(68%), followed by dioecious and androdioecious (28%) species combined,
and monoecious and polygamous (4%) species combined. These values are
remarkably similar to those observed for a sample of 139 tree species from a
tropical rainforest in Mexico (Ibarra-Manrı́quez and Oyama 1992). In Los
Tuxtlas, the frequency of species with the different sexual systems was reported
as 63, 27, and 9%, respectively. The apparently high incidence of dioecy in the
current survey likely reflects the fact that only individuals with DBH ‡10 cm
were sampled. Because of the correlation between dioecy and a woody growth
form, especially in the tropics (Vamosi and Vamosi 2004), the incidence of
dioecy in the Volta Velha Reserve will almost certainly be lower with the
inclusion of herbaceous species (see also Ibarra-Manrı́quez and Oyama 1992).
There was a marginally higher number of individuals in dioecious
(mean = 7.92) than hermaphroditic + monoecious (mean = 4.81) species

[359]
1420

Table 1. Breeding system of 97 plant species surveyed in a 1 ha plot of the Volta Velha Reserve.
Podocarpus sellowii (Podocarpaceae) is a gymnosperm, whereas the remaining species are angio-
sperms.

Family Species Breeding

System

Anacardiaceae Tapirira guianensis Aubl. D

Annonaceae Annona cacans Warm. H
Annonaceae Duguetia lanceolata A. St.-Hil. H
Annonaceae Guatteria australis A. St.-Hil. H
Annonaceae Xylopia brasiliensis Spreng. H
Apocynaceae Aspidosperma parvifolium A. DC. H
Aquifoliaceae Ilex dumosa Reissek D
Aquifoliaceae I. integerrima Reissek D
Aquifoliaceae I. pseudobuxus Reissek D
Aquifoliaceae I. theezans Mart. ex Reissek D
Arecaceae Attalea dubia (Mart.) Burret M
Bignoniaceae Tabebuia alba (Cham.) Sandwith H
Bombacaceae Spirotheca passifloroides Cuatrec. H
[= Malvaceae]
Burseraceae Protium kleinii Cuatrec. D
Caesalpinaceae Copaifera trapezifolia Hayne H
Celastraceae Maytenus robusta Reissek H
Chrysobalanaceae Hirtella hebeclada Moric. ex DC. H
Chrysobalanaceae Parinari sp. H
Clethraceae Clethra scabra Pers. H
Clusiaceae Calophyllum brasiliense Cambess. AD
Clusiaceae Clusia parviflora Humb. & Bonpl. ex Willd. D
Clusiaceae Garcinia gardneriana (Planch. & Triana) Zappi D
Combretaceae Buchenavia kleinii Exell H
Cunoniaceae Weinmannia paulliniifolia Pohl H
Elaeocarpaceae Sloanea guianensis (Aubl.) Benth. H
Euphorbiaceae Alchornea triplinervia (Spreng.) Müll. Arg. D
Euphorbiaceae Aparisthmium cordatum (Juss.) Baill. D
Euphorbiaceae Hyeronima alchorneoides Allemão D
Euphorbiaceae Maprounea guianensis Aubl. M
Euphorbiaceae Pera glabrata (Schott) Poepp. ex Baill. D
Fabaceae Andira anthelminthica Benth. H
Lauraceae Aiouea saligna Meisn. H
Lauraceae Aniba firmula (Nees & C. Mart.) Mez H
Lauraceae Nectandra grandiflora Nees & C. Mart. ex Nees H
Lauraceae Nec. megapotamica (Spreng.) Mez H
Lauraceae Nec. oppositifolia Nees & Mart. H
Lauraceae Ocotea aciphylla (Nees) Mez H
Lauraceae O. dispersa (Nees) Mez D
Lauraceae O. elegans Mez H
Lauraceae O. glaziovii Mez D
Lauraceae O. odorifera (Vellozo) Rohwer D
Lauraceae O. pulchella Mart. D
Lauraceae O. pulchra Vattimo D
Lauraceae Persea venosa Nees & Mart. ex Nees H
Malphigiaceae Byrsonima ligustrifolia St. Hilaire H
Melastomataceae Miconia cabuçu Hoehne H

[360]
 1421

Table 1. (Continued).

Family Species Breeding

System
Melastomataceae M. cubatanensis Hoehne H
Melastomataceae Mouriri chamissoana Cogn. H
Mimosaceae Inga heterophylla Willd. H
Mimosaceae Pithecellobium langsdorffii Benth. H
Monimiaceae Mollinedia triflora (Spreng.) Tul. D
Monimiaceae Mol. uleana Perkins D
Myristicaceae Virola oleifera (Schott) A.C. Sm. D
Myrsinaceae Conomorpha peruviana A. DC. D
[= Cybianthus peruvianus (A. DC.) Miq.]
Myrsinaceae Rapanea ferruginea (Ruiz & Pav.) Mez D
[= Myrsine coriacea (Sw.) R. Br. ex Roem. & Schult.]
Myrsinaceae R. venosa (A. DC.) Mez [= Myrsine venosa A. DC.] D
Myrtaceae Aulomyrcia obscura O. Berg H
Myrtaceae Blepharocalyx salicifolius (Kunth) O. Berg H
Myrtaceae Calyptranthes concinna DC. H
Myrtaceae Ca. lucida Mart. ex DC. H
Myrtaceae Campomanesia guaviroba (DC.) Kiaersk. H
Myrtaceae Eugenia cerasiflora Miq. H
Myrtaceae E. obovata Poir. H
Myrtaceae E. subavenia O. Berg H
Myrtaceae E. tristis D. Legrand H
Myrtaceae E. umbelliflora O. Berg H
Myrtaceae Gomidesia affinis (Cambess.) D. Legrand H
Myrtaceae Go. schaueriana O. Berg H
Myrtaceae Marlierea eugeniopsoides H
(D. Legrand & Kausel) D. Legrand
Myrtaceae Ma. reitzii D. Legrand H
Myrtaceae Myrceugenia campestris (DC.) D. Legrand & Kausel H
Myrtaceae My. reitzii D. Legrand & Kausel H
Myrtaceae Myrcia acuminatissima O. Berg H
Myrtaceae My. fallax (Rich.) DC. H
Myrtaceae My. pubipetala Miq. H
Myrtaceae Neomitranthes cordifolia (D. Legrand) Legr. H
Myrtaceae Neo. glomerata (D. Legrand) D. Legrand H
Myrtaceae Psidium cattleyanum Sabine H
Ochnaceae Ouratea parviflora (DC.) Baillon H
Olacaceae Heisteria silvianii Schwacke H
Olacaceae Tetrastylidium grandifolium (Baill.) Sleumer H
Oleaceae Chionanthus filiformis (Vell.) P.S. Green H
Podocarpaceae Podocarpus sellowii Klotzsch ex Endl. D
Polygonaceae Coccoloba ovata Benth. D
Rosaceae Prunus sellowii Koehne H
Rubiaceae Amaioua guianensis Aubl. D
Rubiaceae Faramea marginata Cham. H
Rubiaceae Posoqueria latifolia (Rudge) Roem. & Schult. H
Rubiaceae Psychotria carthagenensis Jacq. H
Rutaceae Esenbeckia grandiflora Mart. H
Sapindaceae Cupania oblongifolia Mart. M
Sapindaceae Matayba guianensis Aubl. PG

[361]
1422

Table 1. (Continued).

Family Species Breeding

System
Sapotaceae Manilkara subsericea (Mart.) Dubard H
Sapotaceae Pouteria beaurepairei (Glaz. & Raunk.) Baehni H
Sapotaceae Po. venosa (Mart.) Baehni H
Solanaceae Solanum inaequale Vell. [= S. pseudoquina A. St.-Hil.] H
Styracaceae Styrax glabratus Schott H

AD = androdioecious, D = dioecious, M = monoecious, H = hermaphroditic, PG = polyga-

mous. To facilitate comparison with the original list (Negrelle 2002), all original species and family
names are given, with entries in square parentheses indicating the currently accepted synonym.

combined (p = 0.075) (Figure 1). The dioecious species Tapirira guianensis

(Anacardiaceae) and Aparisthmium cordatum (Euphorbiaceae) had the highest
and second highest number of individuals, respectively, and seven of 18 (39%)
species with ‡10 individuals were dioecious. Overall, 309 of the 734 (42%)
individuals belonged to a dioecious species, which is significantly greater than
the proportion of species that were dioecious (contingency test, p = 0.007).
One possible explanation for this result is that the strong correlation between a
dioecious breeding system and zoochory in this flora helps these species
overcome the seed shadow effect (Heilbuth et al. 2001). Birds are thought to
disperse seeds from fleshy fruits to such an extent as to obscure the spatial
segregation that accompanies dioecy (e.g., Nanami et al. 1999). Although there
are few comparable studies (Ibarra-Manrı́quez and Oyama 1992; Lieberman and
Lieberman 1994; Nanami et al. 1999; Pitman et al. 2001; Chazdon et al. 2003),

10
9
8
Square root number

7
of individuals

6
5
4
3
2
1
0
Dioecious Nondioecious
Breeding system
Figure 1. Square root transformed number of individuals with a dioecious (diamonds) or non-
dioecious (hermaphroditic and monoecious; triangles) breeding system in a 1 ha plot of the Volta
Velha Reserve. N = 27 dioecious species, 70 nondioecious species.

[362]
 1423

an overrepresentation of individuals from dioecious species has been docu-

mented into two other forests. Ibarra-Manrı́quez and Oyama (1992; p. 387)
remarked that: ‘some dioecious species are common in open large patches
of the forest … and probably the density of these species is higher than that of
plants with hermaphroditic or monoecious flowers.’ On Mt. Mikasa, Japan,
the dioecious Podocarpus nagi (Thunb.) Makino (Podocarpaceae) and Neolit-
sea aciculata (Blume) Koidz. (Lauraceae) were numerically the most dominant
species, accounting for 72 and 20%, respectively, of 368 plants observed >5 cm
diameter at breast height (Nanami et al. 1999). Although the other community
members were not discussed in detail, the remaining 8% of plants were rep-
resented by 12 other species. Given the relative rarity of dioecy among all
angiosperms, such dominance of dioecious species in local floras will likely be
shown to occur only under particular circumstances (e.g., forests in sub- and
tropical-regions). Further explorations of the relative representation of indi-
viduals with a dioecious breeding system, especially in tropical vs. temperate
floras, are likely to reveal interesting insights into the costs and benefits of the
evolution and maintenance of spatially separated sexes (Heilbuth et al. 2001;
Vamosi and Vamosi 2004).

Acknowledgements

I wish to express my sincere thanks to L. Bohs, M. Freitas, J. Pipoly III,

J. Rohwer, and T. Taggart for their assistance with breeding system designations,
J. Ricketson for kindly putting me in touch with J. Pipoly III, J. Vamosi and an
anonymous reviewer for insightful comments, and the Natural Sciences and
Engineering Research Council of Canada and University of Calgary URGC
Starter Grant for financial support.

References

Arroyo M.T.K. and Squeo F. 1990. Relationship between plant breeding systems and pollination.
In: Kawano S. (ed.), in Biological Approaches and Evolutionary Trends in Plants. Academic
Press, London, pp. 205–227.
Bush M.B. 1995. Neotropical plant reproductive strategies and fossil pollen representation. Am.
Nat. 145: 594–609.
Charlesworth D. 1985. Distribution of dioecy and self-incompatibility in angiosperms. In:
Greenwood J., Harvey P.H. and Slatkin M. (eds), Evolution: Essays in Honour of John May-
nard Smith. Cambridge University Press, Cambridge, pp. 237–268.
Chazdon R.L., Careaga S., Webb C. and Vargas O. 2003. Community and phylogenetic structure
of reproductive traits of woody species in wet tropical forests. Ecol. Monogr. 73: 331–348.
Darwin C. 1877. The Different Forms of Flowers on Plants of the Same Species. John Murray,
London.
Darwin C. 1878. The Effects of Cross and Self Fertilisation in the Vegetable Kingdom. John
Murray, London.
Flores S. and Schemske D.W. 1984. Dioecy and monecy in the flora of Puerto Rico and the Virgin
Islands: ecological correlates. Biotropica 16: 132–139.

[363]
1424

Geber M.A., Dawson T.E. and Delph L.F. (eds) 1999. Gender and Sexual Dimorphism in
Flowering Plants. Springer-Verlag, New York.
Heilbuth J.C. 2000. Lower species richness in dioecious clades. Am. Nat. 156: 221–241.
Heilbuth J.C., Ilves K. and Otto S.P. 2001. The consequences of dioecy on seed dispersal: modeling
the seed-shadow handicap. Evolution 55: 880–888.
Hutchinson J. 1959. The Families of Flowering Plants, 2nd edn. Clarendon Press, Oxford.
Ibarra-Manrı́quez G. and Oyama K. 1992. Ecological correlates of reproductive traits of Mexican
rain forest trees. Am. J. Bot. 79: 383–394.
Kessler P.J.A. 1993. Annonaceae. In: Kubitzki K., Rohwer J.G. and Bittrich V. (eds), The Families
and Genera of Vascular Plants. Volume II. Flowering Plants. Dicotyledons: Magnoliid,
Hamamelid and Caryophyllid Families. Springer-Verlag, Berlin, pp. 93–129.
Kubitzki K., Rohwer J.G. and Bittrich V. (eds) 1993. The Families and Genera of Vascular Plants.
Volume II. Flowering Plants. Dicotyledons: Magnoliid, Hamamelid and Caryophyllid Families.
Springer-Verlag, Berlin.
Kubitzki K. (ed.) 2004. The Families and Genera of Vascular Plants. Volume VI. Flowering Plants
Dicotyledons: Celastrales, Oxalidales, Rosales, Cornales, Ericales. Springer-Verlag, Berlin.
Lieberman M. and Lieberman D. 1994. Patterns of density and dispersion of forest trees. In:
McDade L.A., Bawa K.S., Hespenheide H.A. and Hartshorn G.S. (eds), in La Selva: Ecology
and Natural History of a Neotropical Rain Forest. University of Chicago Press, Chicago,
pp. 106–119.
Nanami S., Kawaguchi H. and Yamakura T. 1999. Dioecy-induced spatial patterns of two
codominant tree species, Podocarpus nagi and Neolitsea aciculata. J. Ecol. 87: 678–687.
Negrelle R.R.B. 2002. The Atlantic forest in the Volta Velha Reserve: a tropical rain forest site
outside the tropics. Biodiv. Conserv. 11: 887–919.
Pannell J.R. and Barrett S.C.H. 1998. Baker’s Law revisited: reproductive assurance in a meta-
population. Evolution 52: 657–668.
Pitman N.C.A., Terborgh J.W., Silman M.R., Núñez V.P., Neill D.A., Cerón C.E., Palacios W.A.
and Aulestia M. 2001. Dominance and distribution of tree species in upper Amazonian terra
firme forests. Ecology 82: 2101–2117.
Renner S.S. and Ricklefs R.E. 1995. Dioecy and its correlates in the flowering plants. Am. J. Bot.
82: 596–606.
Rohwer J.G. 1986. Produmus einer monographie der gattung Ocotea Aubl. (Lauraceae), sensu lato.
Mitteilungen aus dem Institut für Allgemeine Botanik Hamburg 20: 3–278.
Vamosi J.C. and Vamosi S.M. 2004. The role of diversification in causing the correlates of dioecy.
Evolution 58: 723–731.
Watson L. and Dallwitz M.J. 1992 onwards. The Families of Flowering Plants: Descriptions,
Illustrations, Identification, and Information Retrieval. Version: 14th December 2000. http://
biodiversity.uno.edu/delta/.
Yampolsky C. and Yampolsky H. 1922. Distribution of sex forms in the phanerogamic flora. Bib.
Genet. 3: 1–62.

[364]
Biodiversity and Conservation (2006) 15:1425–1440
Springer 2006
DOI 10.1007/s10531-005-0310-x

-1

Patterns of rodent species diversity and abundance

in a Kenyan relict tropical rainforest

ALESSIO MORTELLITI* and LUIGI BOITANI

Dipartimento di Biologia Animale, Università ‘‘La Sapienza’’, Viale dell’Università 32, 00185, Rome,
Italy; *Author for correspondence (e-mail: [email protected])

Received 26 July 2004; accepted in revised form 1 June 2005

Key words: African tropical rainforests, Anthropogenic disturbance, Diversity, Equitability,

Population abundance, Rodents, Species richness

Abstract. Patterns of rodent species abundance and diversity were examined over a 5 months
period in two areas of a Kenyan relict tropical rainforest. The two areas are subjected to different
administrations which lead to various levels of anthropogenic disturbance: one can be considered
relatively disturbed and one relatively undisturbed. Anthropogenic disturbance causes a reduction
in woody stem density between 0 and 1.5 m and reduced understory tree canopy cover. Rodent
abundance was estimated using the program CAPTURE and compared with the number of
individuals actually captured. Density was estimated with three different methods, two of these
utilised a boundary strip to estimate effective size of the area trapped. Density resulted in being
relatively high in both areas, so population might have been at a peak. Species richness was higher
in the disturbed forest, while species diversity and evenness was higher in the undisturbed forest.
We suggest that in the disturbed forest the increase in number of species might be due to sporadical
entrance in the forest by non-forest species, while the decrease in diversity might be due to the
decrease of lower strata vegetation that occurs in the disturbed forest, hence this factor might affect
species equitability. Bibliographic data supports this hypothesis as rodent species diversity and
ground vegetation cover have been found to be correlated.

Introduction

Most studies on the ecology of African rodents have focused on communities

in savanna, secondary bush, or formerly cultivated land, hence tropical rain-
forests are the most understudied of major habitat types (Delany 1986;
Isabyrie-Basuta and Kasenene 1987).
Notwithstanding this lack of attention, rodents must be seen to play an
important role in tropical rainforests as seed predators and seed dispersers
(Fleming 1975; Struhsaker 1997; Chapman and Chapman 1999). From a forest
management perspective, changes in the ecological parameters of rodent
communities can have repercussions on tree regeneration and floristic com-
position of the forest (Genest-Villard 1980; Isabirye-Basuta and Kasenene
1987).
Modification of natural habitat types can have quite marked effects on the
rodent fauna (Delany 1986). According to Jefferey (1977), Delany (1986),
Isabyrie-Basuta and Kasenene (1987) and Sthruhsaker (1997), rodent

[365]
1426

abundance, as well as species richness (sensu number of species in the com-

munity – Krebs (1999)) and diversity indices (Shannon–Wiener and Simpson
indices), increase with logging. Isabirye-Basuta and Kasenene (1987) in the
Kibale Forest found a positive correlation between rodent species richness and
diversity with ground vegetation cover (0–1.2 m). All the studies in the Kibale
Forest (Uganda) found that ground vegetation cover was significantly greater in
the logged rather than the unlogged forest (Struhsaker 1997). Similar results
were found by Malcolm (1995) in the Amazon Forest of Brazil: as understory
vegetation increased and overstory density decreased due to logging, the
abundance, species richness, and diversity of the terrestrial small mammals
increased, Struhsaker (1997) concludes that ground vegetation cover is an
important ecological variable that probably affects rodent populations. These
studies focus on a particular dynamic: logging affects overstory canopy, the
subsequent increase in light and water reaching ground leads to an increase of
ground vegetation that seems to affect the composition and structure of the
rodent community (abundance of some species and diversity). However, logging
is only one of the many types of anthropogenic disturbance in tropical rain-
forests (Jordan 1986; Whitmore 1998; Primack 2000). Factors such as forest
fragmentation (and consequent edge effect) and degradation of understory
vegetation might lead to different reactions of the rodent community. With this
study we wanted to focus on the effects of these types of anthropogenic dis-
turbance on various parametres of the rodent community: species richness,
species diversity, community equitability and population abundance.
We chose the Kakamega Forest (western Kenya) because of the presence
of two areas subjected to different levels of these types of anthropogenic
disturbance: more specifically the degradation of ground vegetation structure
and the edge effect due to high interspersion between forest and pastures –
cultivated fields. Details on the vegetation structure are discussed elsewhere
(Mortelliti e Boitani submitted). The forest area is managed by two different
authorities: the northern part (the Buyango-Area) is managed by the Kenya
Wildlife Service and locals are not allowed into the forest; the southern part
(the Isecheno-Area) is managed by the Forestry Department which permits
locals to enter the forest for the collection of wood and other forest re-
sources such as fruit and medicinal plants (Rogo et al. 1999). They are also
allowed to pass with their cattle (personal observation). Both these activities
create foothpaths and a generalised tread and thus degradate the lower
strata of vegetation (ground vegetation). This alteration of forest structure,
that more specifically coincides with lower woody stem density between 0
and 1.5 m and reduced understory tree canopy cover, is thus opposite to
that indirectly caused by logging, which results in an increase in the vege-
tation of the lower strata. Hence the Kakamega Forest appears as a suitable
context to further investigate and clarify the relationships between com-
munity diversity, richness, density and ground vegetation cover plus increase
in the forest open areas boundaries.

[366]
 1427

Materials and methods

The study area was located in the Kakamega Forest, Kenya (latitude: 0010¢ N–
0021¢ N, longitude 3447¢ E–3458¢ E; 1500–1700 m a.s.l). According to Lucas
(1968) the forest is the only reasonably large patch of Central African type
lowland rainforest in Kenya. Mean annual precipitation is 2000 mm (Cords
1990). Some of commonest trees are Celtis africana, Prunus africana, Albizia
gummifera, Antiaris toxicaria (Cords 1990; KIFCON 1994).
Research was carried out in the field from November 2002 to April 2003,
during the dry season.
Rodents were live-trapped using large aluminium Sherman traps. Diced fried
coconut mixed with peanut butter was used as bait. Traps were inspected once
a day. Animals were marked by toe clipping and standard data was taken from
each animal before releasing it: specimen number, trap location, sex, repro-
ductive condition, body weight.
The trapping pattern was determined by the number of traps temporarily
available and by logistical factors. The main objective was to obtain two
specular sets of grids in each type of forest in order to allow comparisons. All
grids covered an area of 0.81-ha. Each forest area had:
(1) one 7 · 7 grid with 15 m of trap spacing, two traps at each station, one at
ground level and one arboreal trap (1–3 m), a trapping period of 4 days;
(2) one 7 · 7 grid with 15 m of trap spacing, one ground level trap at each
station, a trapping period of 4 days;
(3) one 10 · 10 grid with 10 m of trap spacing, two traps at each station, one at
ground level and one arboreal trap (1–3 m), a trapping period of 3 days
(Table 1).
Due to the higher interspersion between cultivated fields and forest in the
disturbed forest area, the grids resulted in being closer to the forest edge
(nearest trap of each grid approximately 50, 80 and 150 m) while the grids of
the undisturbed forest were relatively distant (nearest trap approximately 300,
600 and 700 m).
Three trapping sessions (December, February, March) were performed in
the undisturbed forest whereas, due to unpredictable logistic problems, two
trapping sessions (February, March) in the disturbed Forest. The 10 · 10 grid

Table 1. Grid label and grid collocation, trap numbers and trap collocation.

Grid number Collocation Number of Number of

terrestrial traps arboreal traps

1 Disturbed forest (Isecheno forest) 49 0

2 Disturbed forest (Isecheno forest) 100 100
3 Disturbed forest (Isecheno forest) 49 49
4 Undisturbed forest (Buyango forest) 49 0
5 Undisturbed forest (Buyango forest) 100 100
6 Undisturbed forest (Buyango forest) 49 49

[367]
1428

was used only in February and March sessions of each forest. A removal grid
with variable number of traps was occasionally added in each forest in order to
gather 10 skulls of each sex of each species for accurate species determination.
We followed the classification and diagnostic characters of Lecompte et al.
(2001, 2002) and Delany (1975).
Rodent abundance in each grid was estimated using the program
CAPTURE which selects the most appropriate model (Otis et a1. 1978).
Density was estimated in three different ways: the first method consisted in
dividing the number of individuals/grid area; the other two methods applied a
correction to the grid area by adding a boundary strip in order to estimate the
effective size of the area trapped (Krebs 1999). The simplest procedure is to add
a strip one-half the movement radius of the animals under study (Krebs 1999).
In the first case, we used movement radius of animals of the same species in the
same grid. As this first method may be biased by a low number of recaptures, in
the second case we used movement radius of all individuals of the same species
captured in the whole trapping period. Abundance comparisons were per-
formed for each species using a capture index (number of unique individuals/
trap nights for each trapping session, Nichols and Dickman 1996) with a t-test.
Species richness was estimated by the number of species captured. Rodent
species diversity was estimated using the Shannon–Wiener and Simpson indices
(Krebs 1999); the importance of species was measured by the number of
individuals actually captured (Krebs 1999). Values for Shannon–Wiener Index
are espressed as N1 that is number of equally common species that would
produce the same diversity as H¢, while values for Simpson Index are espressed
as 1/D (Krebs 1999). Community evenness was measured using the Shannon–
Wiener measurement (H/Hmax) and Simpson measurement. (1/D/number of
species) (Hair 1980; Krebs 1999).

Results

The study covered a total of 5340 trap nights: 2376 in the undisturbed forest
and 2964 in the disturbed forest. In the whole forest a total of eight species was
captured. Species and number of individuals caught were: 274 Praomys jack-
soni, 82 Hylomyscus stella, 13 Lophuromys flavopunctatus, 4 Mus (Nannomys)
minutoides, 1 Graphiurus sp., 1 Otomys sp., 1 Lemniscomis sp., 1 Mastomys sp.).
Less abundant species (one single capture) were identified only to the genus
level as no skull was available for accurate species determination. All species
except for Graphiurus sp. were caught in the disturbed forest, instead only the
first five species were caught in the undisturbed forest. The number of indi-
viduals of both Praomys and Hylomyscus varied, changing both in grids and in
time; no significant correlation was found between rodent abundance and trap
numbers (Spearman correlation: r = 0.153, p = 0.43, n = 28). Comparisons
of the abundance of each species between macrohabitats was performed with a
t-test on the capture index (unique individuals captured/trap nights, Table 2)

[368]
 1429

Table 2. Capture index (unique individuals captured/trap nights) for Praomys jacksoni and
Hylomyscus stella.

g6 g6 g6 g3 g3 g1 g1 g4 g4 g4 g5 g5 g2 g2
s1 s2 s3 s1 s2 s1 s2 s1 s2 s3 s1 s2 s1 s2

Praomys 0.06 0.09 0.07 0.08 0.11 0.14 0.1 0.07 0.11 0.11 0.03 0.02 0.05 0.06
jacksoni
Hylomyscus 0.04 0.02 0.01 0.03 0.04 0 0.01 0.04 0.04 0.02 0.01 0.01 0.02 0.01
stella
Results are shown for each session of each trapping grid; g – grid; s –session.

no significant difference was found (Praomys: t = 1.1 p = 0.28 n = 14;

Hylomyscus: t = 0.699, p = 0.49, n = 14).
Capture estimate was performed only for the two most abundant species:
Praomys jacksoni and Hylomyscus stella. The model most used for Praomys
jacksoni was model 0 which assumes constant capture probabilities; the model
most used for Hylomyscus stella, instead, was model th: capture probabilities
vary according to time and the individual animal (Figure 1) (see Otis et al. 1978
for models description).
The estimate of CAPTURE and the actual number of individuals captured
were often very close, or even coincident (Figure 2). In some cases, though,
differences are relevant, for example Praomys jacksoni: second session of grid 6
and second session of grid 3 (Figure 2). The precision of the CAPTURE
estimate is given by the confidence interval (Figures 3 and 4).
The density estimate varies with the methodology of correction used, values
range from 7.17 to 69.09/ha (Praomys jacksoni) and 1 to 23.45/ha (Hylomyscus
stella) (Tables 3 and 4). No significant difference was found when we compared
results from Corrected Density (1) and Corrected Density (2) for both species
(P. jacksoni t = 0.229, df=26, p = 0.82; H. stella t = 0.73, df=26, p = 0.94).
Praomys jacksoni, Hylomyscus stella and Lophuromys flavopunctatus were
caught in all the grids; Mus (Nannomys) minutoides was caught in grid 3

Figure 1. Models used by program CAPTURE. In black Praomys jacksoni, in white Hylomyscus
stella. X axis – model name; Y axis – absolute frequencies. See Otis et al. (1978) for models
description.

[369]
1430

Figure 2. Comparison between CAPTURE estimate and actual number of individuals captured.
Praomys jacksoni (above) and Hylomyscus stella (below): in white CAPTURE estimate, in black
actual number individuals captured. X axis – number of rodents; Y axis – grid number and session
number.

Figure 3. Praomys jacksoni CAPTURE estimate (with selected model) and confidence interval.
Leg: p – Praomys jacksoni; g6-1 – grid 6 session 1.

[370]
 1431

Figure 4. Hylomyscus stella. CAPTURE estimate (with selected model) and confidence interval.
Leg: h – Hylomyscus stella; g6-1 – grid 6 session 1.

(undisturbed forest) and grid 1 and 2 (disturbed forest); Graphiurus sp. was
caught only in grid 6 (undisturbed forest), Lemniscomys was caught only in
grid 3 (disturbed forest); Otomys sp. and Mastomys sp. were caught only in grid
2 (disturbed forest). Grid 1 and 2 of the disturbed forest supported a higher
number of species than their corresponding grids in the undisturbed forest, but
in grid 6 of the undisturbed forest a higher number of species was caught in
comparison to its analogous in the disturbed forest (Table 5).
In all the comparisons with the Shannon–Wiener Index between analogous
grids, species diversity resulted higher in the undisturbed forest, while with the
Simpson Index in grid 6 (undisturbed forest) we found a lower diversity than its
equivalent grid in the disturbed forest. As one more trapping session was
performed in grid 4 and 6 of the undisturbed forest, the Diversity Indices were
also calculated without considering individuals caught in the last session, re-
sults partially confirm the higher values for the undisturbed forest: in this case
in grid 6 (undisturbed forest) we found a higher value of the Simpson Index in
comparison to the analogous grid (grid 3) in the disturbed forest thus, all the
undisturbed forest grids were characterised by a higher diversity (Table 5).
In grid 4 and 5 of the undisturbed forest there was a higher Evenness than
their correspondant grids in the disturbed forest, but this pattern was inverted
in grid 6 of the undisturbed forest.
If we pool data from all the grids of the same forest type we obtain a higher
community diversity and higher evenness in the undisturbed forest (Table 5).
In the double-trap grids Hylomyscus stella was captured above ground (1–3
m) 55.7% of times (n = 159), Praomys jacksoni 25.3 % of times (n = 247).
The number of above ground captures for Hylomyscus stella resulted sig-
nificantly higher than the number of above ground captures for Praomys

[371]
 1432
Table 3. Praomys jacksini: density estimates.

Praomys jacksoni Grid 4 Grid 1 Grid 6 Grid 3 Grid 5 Grid 2

Buyango/ha Isecheno/ha Buyango/ha Isecheno/ha Buyango/ha Isecheno/ha

Session 1 Density 18.51 / 41.97 / / /

Corrected density (1) 7.17 / 25 / / /
Corrected density (2) 11.02 / 25 / / /
Session 2 Density 32.09 34.56 93.82 66.66 23.45 46.91
[372]

Corrected density (1) 18.7 17.5 69.09 43.90 16.1 33.62

Corrected density (2) 19.11 20.5 55.8 39.7 13.9 27.9
Session 3 Density 27.16 28.39 37.03 55.55 17.28 56.79
Corrected density (1) 14.37 15.75 19.6 30.4 10 41.81
Corrected density (2) 16.17 16.9 22.05 33 10.29 33.8
Density – CAPTURE estimate of number of individuals/grid area. Corrected density 1 – CAPTURE estimate of number of individuals/grid area plus
boundary strip (mean movement radius of Praomys of the grid). Corrected density 2 – CAPTURE estimate of number of individuals/grid area plus boundary
strip (mean movement radius of Praomys of all grids). Buyango – undisturbed forest; Isecheno – disturbed forest.
 Table 4. Hylomyscus stella: density estimates.

Hylomyscus Stella Grid 4 Grid 1 Grid 6 Grid 3 Grid 5 Grid 2

Buyango/ha Isecheno/ha Buyango/ha Isceheno/ha Buyango/ha Isecheno/ha

Session 1 Density 22.2 / 30.86 / / /

Corrected density (1) 7.5 / 18.11 / / /
Corrected density (2) 11 / 15.7 / / /
Session 2 Density 18.51 1.23 9.87 16.04 8.64 32
[373]

Corrected density (1) 9.67 * 4.21 8.96 3.91 17.8

Corrected density (1) 9.43 0.62 5.03 8.17 4.4 16.35
Session 3 Density 6.17 2.46 6.17 23.45 7.4 12.3
Corrected density (1) 3.16 ** 3.67 9.94 4.19 6.09
Corrected density (1) 3.14 1.25 1.88 11.94 3.77 6.2
Density – CAPTURE estimate of number of individuals/grid area. Corrected density 1 – CAPTURE estimate of number of individuals/grid area plus
boundary strip (mean movement radius of Praomys of the grid). Corrected density 2 – CAPTURE estimate of number of individuals/grid area plus boundary
strip (mean movement radius of Praomys of all grids).
*1 capture, **2 captures Buyango – undisturbed forest; Isecheno – disturbed forest.

1433
 1434
Table 5.

Grid Species richness Shannon-Wiener Simpson Shannon-Wiener Simpson’s

indexa (ln) Indexb measure measure
of evenness of evenness

Grid 6 undisturbed 5 (P. jacksoni, H. stella, 2.03 (2.11) 1.66 (2.08) 0.44 (0.46) 0.33 (0.41)
L. flavopunctatus, M. minutoides,
Graphiurus sp.)
Grid 3 (disturbed) 4 (P. jacksoni, H. stella, 1.91 1.75 0.59 0.58
L. flavopunctatus, Lemniscomys sp.)
Grid 4 undisturbed 3 (P. jacksoni, H. stella, 1.93 (2.05) 1.74 (1.83) 0.60 (0.37) 0.58 (0.61)
L. flavopunctatus)
Grid 1 (Disturbed) 4 (P. jacksoni, H. stella, 1.5 1.29 0.29 0.32
L. flavopunctatus,
M. minutoides)
Grid 5 undisturbed 3 (P. jacksoni, H. stella, 2.24 2.14 0.74 0.71
[374]

L. flavopunctatus)
Grid 2 (disturbed) 6 (P. jacksoni, 2.1 1.78 0.43 0.29
H. stella, L. flavopunctatus,
M. minutoides, Otomys sp.,
Mastomys sp.)
Undisturbed forest 5 (P. jacksoni, H. stella, 2.24 2.03 0.5 0.4
(all data pooled) L. flavopunctatus, M. minutoides,
Graphiurus sp.)
Disturbed forest 7 (P. jacksoni, H. stella, 2.03 1.69 0.36 0.24
(all data pooled) L. flavopunctatus, M. minutoides,
Otomys sp.,
Mastomys sp., Lemniscomys sp.)
a
The value is the number of equally common species that would produce the same diversity as H¢ (Krebs 1999).
b
The value is 1/D (Krebs 1999); in brackets, value calculated excluding last session.
 1435

jacksoni (paired t-test: t =
8.45 df=8 p < 0.001). No significant difference

was found in the number of above ground captures for each species between
the two forest types (P. jacksoni t =
1.11 df=7 p = 0.3; H. stella t = 1.5
df=7 p = 0.15). The individual Graphiurus was captured above ground.

Discussion

Our results show that anthropogenic disturbance does not necessarily lead to
an increase in rodent species diversity. We found that damage to the vegetation
of the lower strata and increase in the forest – open area boundaries lead to a
variation in the diversity, evenness and richness of the rodent community.
More specifically it appears that the proximity to forest edge increases the
species richness while the decrease in ground vegetation cover leads to a
decrease in the community diversity.
The population study results show that there is a high similarity between the
CAPTURE estimate and the actual numbers of individuals trapped. This might
suggest that the estimate was accurate and that most of the animals present in
the trapping area were caught. At the same time from Figures 3 and 4 we can see
that the estimates were often precise as they have a very small confidence
interval. The comparison of the capture index of each species between macro-
habitats did not give significant results, in fact most variations occur between
grids and between trapping sessions and not between macrohabitats: this might
depend on actual differences in population sizes, or sampling error as well as
differences in trap numbers (but no significant correlation was found between
rodent abundance and trap numbers). Struhsaker (1997), reports that in some
cases these two species showed higher abundance in logged forest, but in other
cases differences were not significant. Chapman and Chapman (1999) reported
that their capture success doubled in the disturbed forest, while Isabirye-Basuta
and Kasenene (1987) found significant differences for Hylomyscus stella but not
for Praomys jacksoni. Waweru and Odanga (2004) in the Kakamega Forest
found higher abundance in a portion of mature forest when compared to a
fragment of regenerating forest (clear felled 15 years before). In all these cases
anthropogenic disturbance coincided with logging; as previously stated this is
not the case for our study areas, so the fact that we did not observe significant
differences could be due the fact that this kind of anthropogenic disturbance
does not affect this demographic parametre. However, since populations of
African rainforest rodents fluctuate to some extent (Struhsaker 1997) long-term
studies are needed to confirm this hypothesis.
The density estimate varies with the method used. For example in grid 5
(disturbed forest), with the first method, that might overestimate density, we
obtain a value of 46.91 individuals/ha, with one of the corrections the estimate
drops down to 27.9 individuals/ha. Even if we consider the lowest values still we
obtain very high values of density. The boundary strip correction is a very useful
tool to estimate the effective size of Area trapped and is often used, also in

[375]
1436

tropical rainforests (Isabyrie-Basuta and Kasenene 1987). No significant dif-

ference was found between these two corrections for both species, however,
since the first method may be biased by a low number of recaptures, the second
method, that considers movement radius of all individuals of the same species
captured in the whole trapping period, might be relatively more reliable al-
though it is insensitive to variations in movement radius occuring between grids.
Struhsaker (1997) reviewed data of 10 years of study in the Kibale Forest,
for Praomys jacksoni density estimates vary from 1.98 to 4.98 individuals/ha
while for Hylomyscus stella they vary from 1.78 to 6.76 individuals/ha. How-
ever, in another forest in Uganda, Delany (1986) reports, for all rodent species,
density peaks up to 58.7 rodents/ha. If we pool data for Praomys and Hylo-
myscus, data is consistent with Delany’s results. Thus it appears that, during
the period of study, the population of Praomys jacksoni and Hylomyscus stella
might have been at a peak. If we extend our comparisons, density in the
Kakamega Forest is still very high: Mares and Ernest (1995) in a gallery forest
of central Brazil found a density of rodents below 10/ha, while in his review of
many studies in tropical rainforest Fleming (1975) reports values of rodent
density below 15/ha.
There is a need for long-term studies in order to investigate whether this is a
temporary situation; in any case this research highlights the presence (tempo-
rary or not) of extreme abundances of rodents, which could have drastic
consequences on seed ecology, thus it underlines the necessity and importance
of studies focusing on the effects of rodent communities on seed predation and
seed dispersal.
The comparisons relative to community richness, diversity and equitability
regard corresponding grids of the two forest types: grids with the same number
of traps and trapping pattern, that is grids with the same probability of cap-
turing rare species. Some factors that may bias results are: (a) traps and baits
are species-specific so it is possible that many species are not trapped (Gurnell
and Flowerdew 1994; Barnett and Dutton 1995); (b) rare species were repre-
sented by one single capture, so probability of capture for these species was
very low, hence prone to stochastic variability; in this case the Simpson Index,
as it is less sensitive to rare species, might be more indicative (Hair 1980); (c)
although we used arboreal traps, our efforts were limited to the first 3 m: to
obtain a complete inventary of species one should sample up to the canopy
(Delany 1986; Barnett and Dutton 1995). Our list of species is similar to the
one obtained by Waweru and Odanga (2004) in the Kakamega Forest except
for Colomys goslongi and Rattus rattus which they caught in a fragment of
regenerating forest (clear-felled 15 years before) therefore in a different habitat.
Our list of species and community structure is also similar to the ones of similar
studies in forests of East Africa: Isabyrie-Basuta and Kasenene (1987) and
Chapman and Chapman (1999), in the Kibale Forest found that the two forest
specialist Praomys jacksoni and Hylomyscus stella, were the most abundant
species, followed by Lophuromys flavopunctatus and Mus (Nannomys) min-
utoides. At the grid level, species richness, in two out of three comparisons, is

[376]
 1437

higher in the disturbed forest. If we pool data from all the grids of the same
forest type, species richness is higher in the disturbed forest (7 species) than in
the undisturbed forest (5 species). With the Shannon–Wiener Function we
found a higher community diversity in all the grids of the undisturbed forest,
while with the Simpson’s Index we obtained the same result in two of the three
comparisons. The decrease of rodent species diversity together with the de-
crease in lower strata vegetation seems to be coherent with the considerations
of Isabyrie-Basuta and Kasenene (1987) on the positive correlation between
rodent species diversity and ground vegetation cover. The only difference is
that in their case anthropogenic disturbance (logging) leads to an increase of
ground vegetation cover, which is the opposite of our case. On the other hand,
our data reveals an increase in species richness in the disturbed forest which is
characterised by lower ground vegetation density. Isabyrie-Basuta and
Kasenene (1987) instead, found a positive correlation between ground vege-
tation cover and rodent species richness. This particular pattern of inversion
between rodent species richness and diversity in the two areas of the Kakamega
Forest may be due to two different aspects of anthropogenic disturbance: (1)
proximity to the forest edge might favour the occasional entrance of non-forest
species, thus increasing rodent species richness; (2) alteration of lower strata
vegetation might increase the other component of diversity: equitability, that is
the relationships of dominance between species.
First, as was anticipated in the Introduction, the disturbed forest grids
were near to the forest edge, so very near to the cultivated fields and pastures,
that sustain a different rodent fauna (Delany 1975; Delany 1986). Although
identification of single capture species stops at the genus level, Mastomys,
Lemniscomys and Otomys are three genus with only non-forest species and
they were captured only in the disturbed forest, Lophuromys flavopunctatus
and Mus (Nannomys) minutoides are habitat generalists and are often cap-
tured in forests (Kingdon 1974; Delany 1975; Delany 1986), these species
were captured in both forest types. Graphiurus, Praomys jacksoni and Hylo-
myscus stella are the only forest-specialist species (Kingdon 1974; Delany
1975; Delany 1986; Lecompte et al. 2002). Graphiurus sp. was captured only
in grid 6 of the undisturbed forest, which is the only grid of the undisturbed
forest characterised by higher species richness than its corresponding in the
disturbed forest. Thus this data supports the hypothesis that the proximity to
forest edge is the main factor responsible for increase in species richness. A
short-term study such as this is not able to assess whether the 3 non-forest
species captured in the disturbed forest are permanent components of the
rodent forest community, however, these same open area species are known
to invade forest, particularly secondary forest (Delany 1975; Isabyrie-Basuta
and Kasenene 1987; Struhsaker 1997).
As regards point two, in two out of three comparisons evenness is higher in
the undisturbed forest, if we pool data we confirm that evenness is higher in
the undisturbed forest. As structural complexity or heterogeneity of a habitat
increases, the number of microhabitats potentially available increases (Hair

[377]
1438

1980). In the other part of our research (Mortelliti and Boitani, submitted)
we found that in the Kakamega Forest microhabitat heterogeneity decreases
in the disturbed forest as a result of the reduction of low strata vegetation.
The higher variability of microhabitats in the undisturbed forest might hence
lead to a more even distribution of species increasing and/or decreasing the
proportion of individuals of a certain species in the community. The pattern
observed in the corresponding grids 6 and 3 does not fit in this hypothesis, so
one can only postulate that some other factor might be affecting community
diversity and evenness. In synthesis our results support the hypothesis that
proximity to forest edge increases species richness while degradation of
ground vegetation cover reduces the evenness parametre, thus decreasing
community diversity.
A comprehensive formulation of this hypothesis will need further long-term
studies with the utilisation of various types of traps at various heights in order
to reduce bias of the species checklist. Furthermore it is important to highlight
that woody stem density or low-strata vegetation in general might not be the
actual feature responsible for this change, it might be correlated to the actual
factors effectively influencing rodent species diversity.
Data on vertical stratification appears to be consistent with bibliographic data
(Kingdon 1974; Delany 1975; Delany 1986; Struhsaker 1997): Praomys jacksoni
and Hylomyscus stella are two scansorial species, the latter being relatively more
arboreal. Our results show that in the Kakamega forest anthropogenic distur-
bance does not seem to influence vertical stratification since no significant
difference in above-ground captures was found between macrohabitats.
Struhsaker (1997) suggested that anthropogenic disturbance leads to an in-
crease in the degree of segregation between arboreal and terrestrial niches
(more specifically Hylomyscus stella appeared to be strictly arboreal rather
than scansorial). However, in his case anthropogenic disturbance coincided
with logging and thus with an increase of ground vegetation cover, which is the
opposite of our case. Our data refers only to the first 3 m of height, a detailed
analysis of vertical stratification will require traps located right up to the
overstory canopy (Delany 1986).
Our short-term study shows that during the period of study the Kakamega
forest supported an extremely high abundance of Praomys jacksoni, and in
some cases of Hylomyscus stella, this highlights the importance of studies
focusing on possible effects of this high density of rodents on seed ecology. This
study also supports the hyphothesis that ground vegetation might be (directly
or indirectly) the factor responsible for variation in community diversity
(through a variation in evenness), while proximity to the forest edge might be
responsible for variation in species richness. This study further demonstrates
the ability of man to modify the structure of animal communities, even in the
form of a ‘light’ impact such as wood collection and cattle passage by local
communities inhabiting forest surroundings and thus pones evidence on the
importance of these studies to forest management.

[378]
 1439

Acknowledgements

Thanks to the following Institutions that provided us with the necessary

permits to conduct research: National Museums of Kenya, Kenya Wildlife
Service, Isecheno Forestry Department. This study was financed with a grant
from ‘‘La Sapienza’’ University of Rome. Two anonymous reviewers
significantly improved the manuscript.

References

Barnett A. and Dutton J. 1995. Expedition field techniques: small mammals (excluding bats).
Expedition Advisory Centre, London, England.
Chapman C.A. and Chapman L.A. 1999. Forest restoration in abandoned agricultural land: a case
study from east Africa. Conserv. Biol. 13(6): 1301–1311.
Cords M. 1990. Mixed-species association of East-African Guenons: general pattern or specific
examples? Am. J. Primat 21: 101–114.
Delany M.J. 1975. The Rodents of Uganda. British Museum of Natural History.
Delany M.J. 1986. Ecology of Small Rodents in Africa. Mammal review Vol. 16.
Fleming T.H. 1975. The role of small mammals in tropical ecosystems. In: Golley F.B., Petrusewicz
K. and Ryszkowski L. (eds), Small Mammals: Their Productivity and Population Dynamics.
Cambridge University Press.
Genest-Villard H. 1980. Regime alimentaire des rongeurs Myomorphes de foret equatoriale (region
de M’Baiki Republique Centrafricaine). Mammalia 44: 432–484.
Gurnell J. and Flowerdew J.R. 1994. Live Trapping Small Mammals –A Practical Guide –
Occasional Publication No3. Mammal Society, London.
Hair J.D. 1980. Measurements of ecological diversity. In Schennitz (ed.), Wildlife Management
Techniques Manual.
Isabirye-Basuta G. and Kasenene J.M. 1987. Small rodent population in selectively felled and
mature tracts of Kibale forest, Uganda. Biotropica 19(3): 260–266.
Jefferey S.M. 1977. Rodent ecology and land use in western Ghana. J. Appl. Ecol. 14: 741–755.
Jordan C.F. 1986 Local effects of tropical deforestation. In: Soulè E. (ed.)1986 Conservation
Biology: The Science of Scarcity and Diversity. Sinauer Associates.
KIFCON (Kenya Indigenous Forest Conservation) 1994. Kakamega Forest: The Official Guide.
KIFCON, Nairobi.
Kingdon and J. 1974. East African Mammals (Rodentia, Insectivora, Macroscelida). Academic Press.
Krebs Charles J. 1999. Ecological Methodology 2nd edn. Addison Wesley Longman Inc.
Lecompte E., Denys C. and Granjon L. 2001. An identification key of the Praomys species (Rodentia:
Muridae). In: 1’IRD (ed.), African Small Mammals. colloques et séminaires, paris, pp. 127–139.
Lecompte E., Granjon L. and Denys C. 2002. The phylogeny of the Praomys complex (Rodentia:
Muridae) and its phylogeographic implications. J. Zool. Syst. Evol. Res. 40(2002): 8–25.
Lucas G.L. 1968. Kenya. In: Hedberg I. and Hedberg O. (eds), Conservation of Vegetation in
Africa South of the Sahara. Acta phytogeographica suecica 54, pp.152–166.
Mares M.A. and Enest K.A. 1995. Population community ecology of small mammals in a gallery
forest of central Brazil. J. Mammal. 76(3): 750–768.
Malcolm J.R. 1995. Forest structure and the abundance and diversity of neotropical small mammals.
In: Lowman M.D. and Nadkarni N.M. (eds), Forest Canopies. Academic Press, New York, 624 pp.
Nichols and Dickmanin Wilson D.E., Cole R.T., Nichols J.D., Rudran R. and Foster M.S. 1996.
Measuring and Monitoring Biological Diversity: Standard Methods for Mammals. Smithsonian
Institution Press.
Primack R.B. 2000. A Primer in Conservation Biology. Sinauer Associates.

[379]
1440

Otis D.L., Burnham K.P., White G.C. and Anderson D.R. 1978. Statistical Inference from Capture
Data on Closed Animal Populations. Wildlife Monographs – A publication of the Wildlife
Society No62.
Rogo L., Lwande W., Miller S., Herren H. and Chapya A. 1999. Kakamega Forest: an integrated
conservation project. Bull. East African Nat. History Soc. 29(3): 9–13.
Struhsaker T.T. 1997. Ecology of an African Rainforest: Logging in Kibale and the Conflict
between Conservation and Exploitation. University Presses of Florida, Gainesville.
Waweru C. and Odanga J.C. 2004. Demographic aspects of sympatric Praomysjacksoni and
P. Stella in a tropical lowland forest in Kekamega, Kenya. Afr. J. Ecol. 42: 93–99.
Whitmore T.C. 1998. An Introduction to Tropical Rainforests. Oxford University press.

[380]
Biodiversity and Conservation (2006) 15:1441–1457
Springer 2006
DOI 10.1007/s10531-005-0598-6

-1

The role of landscape patterns of habitat types

on plant species diversity of a tropical forest in Mexico

J. LUIS HERNANDEZ-STEFANONI1,2,*
1
Watershed Ecosystems Graduate Program, Trent University, Peterborough, Ontario, Canada;
2
Present address: Servicio de Información y Estadı´stica Agroalimentaria y Pesquera, Av. Benjamı´n
Franklin 146, Col. Escandon, C.P. 11800, Me´xico, USA; *Author for correspondence (e-mail:
[email protected]; phone: +1-52-55-5271-7111, ext. 134)

Received 5 April 2004; accepted in revised form 24 June 2005

Key words: Landscape fragmentation, Landscape patterns, Plant diversity, Shannon diversity
index, Simpson diversity index, Tropical forest

Abstract. The relationships among landscape characteristics and plant diversity in tropical forests
may be used to predict biodiversity. To identify and characterize them, the number of species, as
well as Shannon and Simpson diversity indices were calculated from 157 sampling quadrats (17,941
individuals sampled) while the vegetation classes were obtained from multi-spectral satellite image
classification in four landscapes located in the southeast of Quintana Roo, Mexico. The mean
number of species of trees, shrubs and vines as well as the mean value of the total number of species
and the other two diversity indices were calculated for four vegetation classes in every one of the
four landscapes. In addition, the relationships between landscape patterns metrics of patch types
and diversity indices were explored. The multiple statistical analyses revealed significant predictor
variables for the three diversity indices. Moreover, the shape, similarity and edge contrast metrics
of patch types might serve as useful indicators for the number of species and the other two diversity
variables at the landscape scale. Although the association between the three diversity indices and
patch types metrics showed similar behavior, some differences were appreciated. The Shannon
diversity index, with its greater sensitivity to rare species, should be considered as having a greater
importance in interpretation analysis than Simpson index.

Introduction

Tropical forests of the world are being destroyed by degradation and con-
version to other forms of land use, induced by increasing human needs or
simply by economic gain. The loss of biodiversity is considered to be one of the
most important of all negative effects on these forests. High diversity implies
that there is a source of new species executing functions or ecosystem services
for human needs (Bengtsson 1998). Therefore, a reduction of biological
diversity means less environmental functions and ecological processes that
generate and maintain soils, convert solar energy into plant tissue, absorb
pollutants, supply clean air and water, store essential nutrients, regulate
weather, and climate and so on (Myers 1995).
The Yucatan peninsula has been recognized as one of the world’s biodi-
versity ‘‘hotspots’’ areas with high levels of biological diversity (Myers et al.

[381]
1442

2000). Nevertheless, forests of the peninsula are subject to disturbance by

hurricanes and forest fires (Whigham et al. 1991) but also by land use changes.
These disturbances produce a mosaic of forest in different stages of succession,
as well as small forest remnants embedded in a matrix of agriculture and
grassland areas. Conversely, the practice of ‘‘slash and burn’’ agriculture, by
Mayan farmers, allows the re-growth of native vegetation over a period of time
or fallow in the areas opened to cultivation (Hernandez-Xolocotzi et al. 1995).
These changes have profound consequences in the forest area and on its bio-
logical diversity. Although these changes are occurring in the Yucatan penin-
sula, the exact magnitude on which these changes affect the diversity of the area
is not well understood.
Disturbance is one of the main functional elements in landscapes. It is also a
key factor to maintain biological diversity (Roberts and Gilliam 1995). There
are different phases of forest succession after a disturbance event takes place
(Whitmore 1989); each of them poses particular problems and advantages for
arrival, regeneration and establishment of different species (Whitmore 1989;
Alvarez-Buylla and Garcia-Barros 1991). For example, pioneer or shade-
intolerant species generally reach maturity if they establish themselves in large,
newly opened gaps (Brokaw 1987). However, the greatest likelihood of
regeneration of pioneer species occurs in the neighborhood of maturing gaps
(Schupp et al. 1989) because of seeds and seedlings being released from adja-
cent forested areas. On the other hand, non-pioneer will germinate and get
established almost entirely beneath the forest canopy, due to few seeds reach a
gap, and most of the seeds that find a gap are eaten by mammals (Schupp et al.
1989; Alvarez-Buylla and Martinez-Ramos 1992). Nevertheless, some of the
species of this group requires an open canopy for growth and reproduction
(Denslow 1995). In other words, non-pioneer species become gap dependant or
need a condition with certain amount of canopy opening, such as that created
at the edge of the patches (Howe 1990). Therefore, the presence or absence and
spatial distribution of plant species may be influenced by both physical and
biotic conditions created in the different phases of forest succession and then
the area, distance and similarity or contrast of adjacent forest areas as well as
the perimeter or shape of the habitat types may be factors that can describe the
species composition (Mazerolle and Villard, 1999; Debinski and Holt, 2000).
The quantification of spatial heterogeneity of landscapes can be studied
in term of patches and their characteristics, through landscape metrics
(McGarigal et al. 2002), which are used to create quantitative measures of
spatial patterns. Such patterns are found on categorical maps, classified air
photographs and remote sensing imagery and their metrics have been linked to
ecological functions. Thus, the most important reason to use landscape metrics
is that landscape patterns can be linked in a quantitative way with ecological
and environmental processes (Krummel et al. 1987; McGarigal et al. 2002).
Consequently, it is possible to study changes in habitat of particular species or
communities of organisms and determine whether or not the habitats are too
fragmented for the species or communities to survive.

[382]
 1443

The purpose of this paper is to examine the relationships between landscape

patterns of habitat types and plant diversity estimates. This was based on the
understanding that the environmental conditions in the study area have small
variation and in particular based on the fact that several studies have sug-
gested that spatial patterns may be important determinant of species distri-
bution at landscape scale (Mazerolle and Villard, 1999; Debinski and Holt
2000; McGarigal and Kushman 2002). The main goal of this study was to
investigate the relationships between the estimates of plant diversity and
landscape metrics of patch types in order to find out whether spatial patterns
of habitat types are reliable mechanisms for predicting biodiversity. This was
done with a view of predicting plant species diversity from landscape features
easily observable/measurable from satellite images or maps. The assessment
and modeling of the spatial distribution of plant diversity is of considerable
importance for conservation and forest management purposes. To preserve
biodiversity, knowledge of where species richness is the highest and how
species assemblages change over the space is required. For example the pres-
ence of species and the habitat occupied by species are significant criteria for
prioritizing and selecting sites for preservation in land use planning (Rossi and
Kuitunen 1996) or for locating areas where plant diversity is critical (Carroll
1998).

Methods

Study area and plant diversity data

The study was conducted in a tropical forest over four contiguous landscapes
of 4 km · 4 km, located in the southeastern portion of the Yucatan peninsula,
Mexico. Tropical sub-deciduous forests in different stages of succession, which
are characterized by the age, as well as secondary associations that prosper
mainly in flood areas, cover the majority of the four landscapes (Cabrera et al.
1982). The forest consisting of 2 or 3 canopy layers with trees, shrubs and vines
between 3 and 25 m of height. Mayan farmers identified the stages of succes-
sion with indigenous local names. ‘‘kanah kax’’ refers to a forest from 20 to
60 years old; ‘‘kelenche’’ is used for vegetation between 11 and 19 years of age;
‘‘juche’’ used for plant species between 4 and 10 years of age and ‘‘saakab’’
with plants species of 3 or less years of age. The secondary plant associations in
the area are ‘‘savanna’’, which have few sparse tree species between 3 and 10 m of
height and ‘‘akalche’’ (in local Mayan language) consisting of a shrub stratum.
A plant survey based on a stratified random sampling design was performed
in the study area during the summer of 2000, 2001 and 2003. This survey had a
total of 157 sampling quadrats, which were located on the ground using a GPS
unit in the six vegetation types. Of the total number of quadrats 42 fell within
the class ‘‘kanah kak’’, 27 in ‘‘kelenche’’, 22 in ‘‘juche’’, 25 in ‘‘saakab’’, 22 in
‘‘akalche’’ and 19 in the ‘‘savanna’’ vegetation class. The sampling quadrat

[383]
1444

consisted of two nested sites, one of them of 10 · 10 m used to sample trees and
vines that have 3 or more meters of height, the other, a nested sub-site of
5 · 5 m used for sampling all the shrubs taller than 1.0 m. In every one of the
quadrats three diversity indices were computed, those are species richness (i.e.,
the number of species present in an area) and two measures based on species
frequencies or abundance, including exponent Shannon and reciprocal Simp-
son indices (Magurran 1988; Krebs 1989). A total of 17,941 sampled individ-
uals were identified to species and enumerated.

Land cover mapping

A land cover map for the entire area was obtained from Landsat 7 Thematic
Mapper (TM) imagery acquired on April 2000, after applying a supervised
classification on bands 5 (short-wave infrared: 1.55–1.75 gm), 4 (near infrared:
0.76–0.90 gm) and 3 (red: 0.63–0.69 gm). Each band was geo-referenced and
radiometrically corrected. The ‘‘Maximum Likelihood Algorithm’’ imple-
mented by the image analysis software ER MapperTM 6.1 (Earth Resource
Mapping Ltd. 1998) was used as the classification method. The sampling
quadrats were used for assessing the accuracy of the classified land cover maps,
which resulted in an overall accuracy of 82.3%. The final land cover map of the
four landscapes is shown in Figure 1. Details of the classification and the
accuracy assessment procedures of the resulting land cover maps are found in
Hernandez-Stefanoni (2004) and Hernandez-Stefanoni and Ponce-Hernandez
(2004).

Calculation of landscape-pattern metrics

The ER MapperTM raster files of the four landscape mosaics were exported to
the GIS program IDRISI (Eastman 1999), in order to calculate the landscape-
pattern metrics using the program software FRAGSTATS 3.0 (McGarigal
et al. 2002). The six vegetation types identified during the classification and the
remaining of the land cover classes grouped as ‘‘background’’ were considered
for the calculations. The individual patches were classified as clusters of ver-
tical, horizontal or diagonal pixels as in other studies (Gustafson et al. 1994),
while a patch type considers all the individual patches of the same vegetation
class. Most of the metrics applied to vegetation classes can be interpreted as
fragmentation indices, because they measure the configuration of a particular
patch type (McGarigal et al. 2002). The four landscapes in the study area were
considered for the computation of the indices per class (patch type).
The division of the study area in four landscapes was done to obtain different
replicas of landscape configurations. The four landscapes, defined for esti-
mating landscape metrics of patch types, still retain a sufficient size as to allow
for the occurrence of several patches of plant diversity. Gustafson (1998) used

[384]
 1445

Figure 1. Land cover maps obtained from supervised classification of the four landscapes.

the range of influence, as depicted from the spatial auto-correlation of sample

units, to define the maximum patch size of the variable of interest. Here, the
range of influence of diversity indices found from the semi-variance analysis is
ranged between 1.2 and 2.1 km (Hernandez-Stefanoni 2004), this indicates that
the four landscapes are sufficiently large to include patches of various sizes
while affording for replication. Moreover, the size of the landscape maybe
arbitrary but needs to be relevant to the process or organisms studied
(McGarigal et al. 2002); in this case the four landscapes continue to be relevant
to define patches of plant diversity.
Several indices at this level are calculated by FRAGSTATS all of them were
considered for reviewing and further examination. Many of patch-type indices
are redundant or represent an alternative formulation of the same information
(Riitters et al. 1995; Hargis et al. 1998). Thus, only selected measurements were
considered in this study. To choose the group of indices, their correlations were
analyzed. Pearson correlation coefficients between each pair of landscape
metrics, as well as the correlation of these metrics with plant diversity values
were computed. The selection of the indices also considered criteria including
variables that quantify different aspects of the patch type configuration. Six
indices were finally included in this study. The selected indices represent factors
(landscape metrics) that might explain the plant diversity in the tropical forests
of the study area. The final selection of the landscape metrics was made

[385]
1446

considering both, how commonly such measurements of landscape metrics are

in landscape studies literature (Mazerolle and Villard 1999), as well as the
explanatory power that such metrics may have to describe plant species com-
position. On these bases, four groups of metrics were selected to relate plant
diversity indices and landscape-patterns of habitat types. Those groups and
metrics are: area/density/edge (percentage of landscape, patch density and edge
density), shape (mean area weighted shape index), isolation/proximity (mean
area weighted similarity index) and contrast (total edge contrast index). A
description of each metric is given next.
Percentage of landscape (PLAND) is calculated as the sum of the areas (m2)
of all patches of the corresponding patch type, divided by the total landscape
area (m2), multiplied by 100 (to convert a percentage). This metric is a measure
of landscape composition, specifically, how much of the landscape is comprised
of a particular patch type (vegetation class).
Patch Density (PD) is calculated as the number of patches of a given patch
type divided by the total landscape area (m2), multiplied by 10,000 and 100 (to
convert to 100 ha). This metric represents the number of patches on a per unit
area basis, which facilitates comparisons among landscapes of varying size.
Edge Density (ED) is calculated as the sum of the length (m) of all edge
segments involving the corresponding patch type, divided by the total land-
scape area (m2), multiplied by 10,000 (to convert to hectares). ED is a measure
of total edge length of all patch types on a per unit area bases that facilitates
comparisons among landscapes of varying size.
Mean area weighted shape Index (SHAPE_AM) is the sum, across all patches
of the corresponding patch type, of the shape index value multiplied by the
proportional abundance of the patch [i.e., patch area (m2) divided by the sum
of patch areas]. The shape index is calculated as the perimeter of a patch (m)
divided by the square root of a patch area (m2). Therefore, this metric is a
measure of shape complexity of a patch compared to a standard shape (square)
of the same size.
Mean area weighted similarity index (SIMI_AM) is the sum, across all pat-
ches of the corresponding patch type, of the similarity index value multiplied
by the proportional abundance of the patch [i.e., patch area (m2) divided by the
sum of patch areas]. The similarity index is calculated as the sum, over all
neighboring patches with edges within a specified distance (m) of the focal
patch, of neighboring patch area (m2) times a similarity coefficient between the
focal patch type and the class of the neighboring patch, divided by the nearest
edge-to-edge distance squared (m2) between the focal patch and the neigh-
boring patch. The similarity index considers the size and proximity of all
patches, regardless of class, whose edges are within a specified search radius of
the focal patch. In resume, the similarity index quantifies the spatial context of
a (habitat) patch in relation to its neighbors of the same or similar class;
specifically, the index distinguishes sparse distributions of small and insular
habitat patches from configurations where the habitat forms a complex cluster
of larger, hospitable (i.e., similar) patches (McGarigal et al. 2002).

[386]
 1447

Table 1. Values used to give a similarity weight between the different patch types.

Kanah Kax Kelenche Juche Saakab Akalche Savanna

Kanah Kax 1.00

Kelenche 0.73 1.00
Juche 0.60 0.77 1.00
Saakab 0.39 0.47 0.52 1.00
Akalche 0.08 0.09 0.10 0.11 1.00
Savanna 0.01 0.02 0.02 0.02 0.08 1.00

In order to calculate the SIMI_AM, a search radius of 10 pixels (300 m) was

considered. This radius is arbitrary but coincides with empirical evidence
gathered in the field about an expected average size of a patch type. In addition
to the radius, this index requires for its calculations some similarity weights for
each pairwise of patch types. In this case the mean values of 4 estimates of beta
diversity between each pair of vegetation types were used as those weights
(Table 1). These beta diversity estimates are similarity measures and are de-
scribed by Magurran (1988). They were calculated using cumulative values of
the sampling quadrats for each vegetation class. Two of these measures use
presence and absence of species (i.e., Jaccard and Sorenson) while the other
two require abundance data for their calculations (i.e., Sorenson-abundance
and Morisita-Horn).
Total edge contrast index (TECI) is the sum of the lengths (m) of each edge
segment involving the corresponding patch type multiplied by the corre-
sponding contrast weight, divided by the sum of the lengths (m) of all edge
segments involving the same type, multiplied by 100 (to convert to a percent-
age). This metric is a relative measure. That is to say, given any amount or
density of edge, they measure the degree of contrast in that edge. High values
of these indices mean that the edge present is of high contrast, and vice versa
(McGarigal et al. 2002). The weighted edge contrast between vegetation classes
demanded to compute this metrics was calculated as the inverse values of the
similarity weights (Table 1).

Statistical analysis

The statistical analysis to evaluate the relationship between landscape-patterns

metrics and mean diversity values of trees, shrubs and vines per unit area,
includes simple correlations and regression analysis. The mean diversity value
of each patch type was calculated as the average value of all plots inside of a
vegetation class. Since the dimensionality of the problem was large due to the
multiple number of landscape metrics in the data, a principal components
analysis was used to summarize the configuration for the 6 metrics of classes to
obtain new variables that were independent of each other (i.e., orthogonal), and
yet represent the ‘‘aggregation’’ of the data reducing with it the dimensionality

[387]
1448

of the problem. This was done while trying to avoid the well-known multico-
linearity problem that often emerges among explanatory variables when the
correlations between them are high. The variables need to be transformed with
1/x, log10(x), log10(x + 1) and sqrt(x) as necessary to meet the assumptions of
normality and linearity (Tabachnick and Fidell 1996). The final components at
each level were found by a rotation procedure using the varimax method,
yielding the final principal components. Then, these were interpreted using
component loading (correlation between the principal component and the
original variable). In order to create a predictive model of plant diversity as a
function of the computed ‘‘compound’’ variables represented by the principal
components, the factor scores and the plant diversity indices were related using
multiple regression analysis, producing models to predict plant diversity indices
based on components created from landscape metrics of patch types. Finally, to
avoid that the correlations between landscape pattern metrics and species
diversity were confounded by habitat type, the analysis was conducted just for
the four stages of succession in the tropical sub-deciduous forest class.

Results

Estimations of plant diversity indices

The mean values of the tree plant diversity indices in each patch type for the
four landscapes are presented in Table 2. The mean values of the diversity
indices for the six vegetation types in every one of the four landscapes showed a
similar pattern. Thus, kanah kax class (i.e., the oldest stage of succession in the
forest) has more species and less dominance than kelenche, juche and saakab,
which are early successional stages of the forest.

Relating landscape-pattern metrics and plant diversity measures

To evaluate the degree of association between metrics of patch types and plant
diversity indices, correlation coefficients between them were computed (see
Table 3). The total edge contrast index (TECI), percentage of land (PLAND),
edge density (ED) and mean area-weighted shape index (SHAPE_AM) showed
the highest correlation coefficients with most of the plant diversity variables.
These coefficients varied from 0.351 to 0.869 in absolute values. PLAND, ED
and SHAPE_AM are positively correlated while TECI is negatively corre-
lated with the diversity indices. This means that the diversity of a patch type
increases with the augment in its area, irregular shape and perimeter, and when
the contrast with other patch types decreases. So, classes that occupy larger
proportion of the area of the landscape and show less contrast with neigh-
boring patch types favor diversity of plants. Moreover, another variable of
patch type metrics (SIMI_AM) was found moderately correlated with plant
diversity indices varying from 0.351 and 0.395 in absolute values.

[388]
 1449

Table 2. Mean plant diversity values for the four landscapes of the study area.

Vegetation type Number of species Exponent Shannon Reciprocal Simpson

Total Trees Shrubs Vines

Landscape 1
Kanah Kax 34.45* 26.70* 5.09* 2.63* 23.28* 16.95*
Kelenche 33.00* 27.57* 6.00* 1.92* 20.13* 14.12*
Juche 27.67* 19.00* 6.00* 2.66* 12.52* 7.31*
Saakab 19.67 12.00 4.67* 2.00* 13.22* 9.32*
Landscape 2
Kanah Kax 35.36* 27.57* 5.78* 1.92* 22.74* 16.10*
Kelenche 34.67* 25.66* 7.00* 3.00* 20.06* 13.38*
Juche 29.75* 22.50* 5.37* 1.87* 16.78* 10.86*
Saakab 16.20 10.90 3.90 1.40* 9.49 6.67
Landscape 3
Kanah Kax 34.08* 28.08* 4.33* 1.66* 20.63* 13.65*
Kelenche 30.27* 23.45* 5.00* 1.88* 18.50* 13.18*
Juche 28.71* 21.57* 5.28* 2.00* 15.90* 10.96*
Saakab 15.11 9.77 3.33* 1.88* 8.57 6.24
Landscape 4
Kanah Kax 36.60* 28.00* 4.80* 4.00* 19.66* 12.51*
Kelenche 30.71* 24.85* 4.42* 1.42* 18.04* 12.06*
Juche 27.00* 20.25* 3.75* 3.00* 15.32* 10.45*
Saakab 17.66 10.67 4.33* 1.66* 9.85* 6.95*
*A Turkey HSD test was performed to compare the mean diversity values among vegetation classes
by landscape, no significant differences between these groups.

Considering the correlation of number of species of the different group of

plants (trees, shrubs and vines) with the landscape metrics of classes, the results
vary according to the group of species considered. In the case of number of

Table 3. Pearson correlation coefficients between landscape spatial patterns of patch and between
plant diversity indices and landscape patterns of patch types.

VARIABLE PLAND PD ED SHAPE_AM SIMI_AM TECI

PLAND
PD 0.340*
ED 0.513 0.411
SHAPE_AM 0.651 0.178* 0.590
SIMI_AM 0.470
0.305* 0.151* 0.736
TECI
0.740 0.514 0.091* 0.333*
0.283*
Number of species
Total 0.791 0.317* 0.596 0.561 0.353
0.490
Trees 0.829 0.289* 0.616 0.652 0.357
0.520
Shrubs 0.351 0.093* 0.593 0.464 0.395
0.125*
Vines 0.186* 0.354* 0.051* 0.016* 0.074*
0.046*
Exp Shannon 0.865 0.323* 0.591 0.583 0.351
0.603
Rec Simpson 0.869 0.314* 0.575 0.575 0.369
0.619
*Correlations are not significant at p < 0.05.
Values in bold are for pairs highly correlated (r > |0.5|).

[389]
1450

tree-species, the correlation with PLAND is very high (r = 0.829, Table 3),
alternatively in the case of shrubs, there is a moderate correlation with the
percentage of land (r = 0.351, Table 3) and there was not significant corre-
lation between PLAND and number of species of vines. This gives some idea of
the importance of the percentage of land in the distribution of number of tree-
species. In the case of richness of trees and shrubs it can be observed that edge
density and shape have an important weight as explanatory factors, as opposed
to the richness of vines, which is not associated with any of landscape metrics.
On the other hand, no significant correlation was found between the number
of patches by hectare or patch density (PD) and plant diversity variables
(p > 0.05, Table 3). The correlations between the 6 metrics of landscape pat-
terns for patch types are also shown in Table 3. Most of the paired combi-
nations of the metrics of patch types were significantly correlated between them
and 6 of the 15 possible paired combinations are highly correlated (r > |0.5|,
Table 3), which may indicate a degree of redundancy in terms of the infor-
mation that they provide about the structure of the landscape.
Principal component analysis was performed on 6 landscape metrics of patch
types for 16 observations. Two components in the studied area were selected as
meaningful factors with eigenvalues greater than one, which explained 77.24%
of the variation. After applying a varimax rotation, the components were
interpreted as a gradient in percentage of land corresponding to a class
(PLAND), the shape (SHAPE_AM), similarity (SIMI_AM) and the total edge
contrast of the patch type (TECI). This first principal component (PC1) ex-
plained 50.50% of the total variation in the original data. The second principal
component (PC2) explained the additional 26.74% of the variation, and it is
essentially a measure of number of patches found in a hectare (PD) and the
edge density (ED). The percentage of variance explained by the two compo-
nents and the correlation between the principal component and the original
metrics of patch types are shown in Table 4.
To find a model for predicting plant diversity indices from uncorrelated
variables derived from landscape metrics of patch types, a regression analysis
was performed. The regression uses number of species (total, trees, shrubs and
vines), exponent Shannon and reciprocal Simpson indices as dependent vari-
ables and the two components retained from the Principal Component Anal-
ysis (PCA) as independent variables. The results are shown in Table 5. The
variability in total number of species and in the other two indices is explained
by a moderately correlated variable related to percentage of land, similarity,
shape and total edge contrast index (PC1) with Sr2 ranged from 0.453 to 0.479.
Also the number of patches and edge density (PC2) explained the variability of
the diversity indices, with Sr2 ranged from 0.113 to 0.154. The positive cor-
relation with axis 1 of the PCA indicates that a patch types is more diverse
when it is larger in area, more similar to its neighbors, has irregular shape and
it is less contrasted with other patch types.
The results of the regression considering the number of species of the three
groups of plants and the two factors of landscape structure of patch types show

[390]
 1451

Table 4. Variance explained by two principal components derived from metrics of patch types and
the weights of the variables in each component after rotation.

PC1 PC2

Explained variance
Observed Eigenvalue 3.03 1.60
% Variance 50.50 26.74
Cum. % Variance 50.50 77.24
Variable
PLAND 0.90 0.11
PD 0.50 0.79
ED 0.42 0.85
SHAPE_AM 0.83 0.36
SIMI_AM 0.75 0.05
TECI
0.73
0.33
*Marked loadings are >0.70.

different responses and reveal that the factor associated with area, shape,
similarity and contrast is more related to the richness of trees, and shrubs
(Sr2 = 0.528 and 0.313, Table 5). On the other hand the factor (PC2) associated
with number of patches and edge density is of variable importance depending on
the group of species considered. Axis 2 of PCA appears to be only correlated
with the richness of trees (Sr2 = 0.115, Table 5).

Discussion

The analysis of the relationships between plant diversity variables and metrics
of habitat types yield the following main results. The percentage of land
(PLAND), a measure of landscape composition, showed a strong correlation

Table 5. Summary of the regression procedures for predicting plant diversity indices from prin-
cipal components derived from metrics of patch types.

Dependent variable Intercept* Model parameters B Sr2 Adj R2 R2

Number of species
Total 28.18 PC1* 4.84 0.453 0.500 0.566
PC2** 2.41 0.113
Trees 21.15 PC1* 4.90 0.528 0.588 0.643
PC2** 2.28 0.115
Shrubs 4.94 PC1* 0.54 0.313 0.264 0.313
Vines – – – – – –
Exponent Shannon 16.54 PC1* 3.26 0.485 0.584 0.640
PC2* 1.84 0.154
Reciprocal Simpson 11.29 PC1* 2.29 0.479 0.577 0.632
PC2* 1.34 0.153
*Variables included in the model with p < 0.05.
**Variables included in the model with p < 0.09.

[391]
1452

with species richness (total, trees, and shrubs) and the other two additional
diversity indices. This result together with the weak relationship between plant
diversity and area of a fragment (Hernandez-Stefanoni 2005) may indicate the
importance that fragmentation and diversity of habitats have on plant species
diversity. Several small spread habitat patches usually contain more plant
species than a few large habitat patches (Margules et al. 1994; Honnay et al.
1999), indicating that the availability of different resources may be important
for the establishment of plant species. Canopy openings events such as treefall,
and ‘‘slash and burn’’ agriculture promote spatial variation of patch types that
create several physical environments for plants that offer an increasing in the
availability of resources (Martinez-Ramos et al. 1988; Denslow 1995). This can
improve species diversity for a specific patch type. However, forests with both
very high or low frequency of disturbances, may both lead to low diversity, due
to the fact that either pioneer species (fast-growing) or non-pioneer species
(highly competitive and slow-growing) are respectively selected in each situa-
tion (Martinez-Ramos et al. 1988). Thus, this allows for the regeneration of
similar group of species.
The different response to percentage of land by various groups of species
(trees, shrub and vines) shows the importance of the frequency of disturbances
and the different life history of the species within each group. Given the fact
that newly formed patches cover a lower proportion of the forest in the studied
area (Hernandez-Stefanoni 2004), and that the community of oldest patches
contains more individuals and includes the shade-intolerant species, the pre-
dominant class life history in the tropical forest (Whitmore 1989), it would be
expected that a higher association between tree-species and percentage of land
existed. The reason for this expectation is that the development of a similar
group of species established under the canopy (shade-intolerant species) is
favored. In contrast, in the shrub community the pioneer, shade-intolerant
species are the predominant group (Denslow et al. 1990), making this group
more dependant of gaps.
In the case of vines, they depend on large plants for support and living,
where vines maturation takes place. The availability of light however, has been
proposed as other factor that promotes liana-species distribution, particularly
near the edges where lianas can grow faster (Ibarra-Martinez and Martinez-
Ramos 2002). Not all liana species however, are light demanding (Putz and
Chai 1987). Therefore, vines-species presence could be favored by two main
factors. First, vines can be established in oldest patches, where trees can pro-
vide them support. Second, several liana species are light demanding and grow
well in natural or man-made disturbances (Putz 1984). Consequently, it is
difficult to establish an association between richness of vines and the different
metrics if there is not a division of these two main groups of vines.
The degree of contrast between a patch type and its neighbors classes,
measured as total edge contrast index, was a metric highly related to species
richness (total, tree and shrub) and the other two plant diversity indices. These
results may be explained by the fact that resource availability of a class is given

[392]
 1453

by the quality of the surrounding areas (Alvarez-Buylla and Garcia-Barrios

1991). However, Multivariate analysis of plant species diversity and metrics of
patch types revealed that the three diversity indices are best predicted by the
degree of contrast and similarity between classes as well as the percentage of
land and the shape of a habitat type. These results may indicate that plant
diversity of the tropical forest in the studied area is explained by a combination
of factors. The relationship among plant diversity and various landscape-
metrics shows that the number of species present in a given patch type is
determined by several conditions, such as number of seed dispersed to the site,
number of dormant seeds on the soil, the probabilities of germination and
survival until they can reach maturity, and the resources provided by the
environmental conditions of the patches or classes during their development
(Schupp et al. 1989).
Several studies have found that the configuration or spatial arrangement of
habitat types such as the degree of isolation, connectivity and fragmentation
can be significant predictors of species richness and other diversity variables.
For instance Mazerolle and Villard (1999) found that landscape variables were
significant predictors of animal species richness in 36 studies reviewed, the
groups of animal analyzed include amphibians, reptiles, mammals, arachnids,
birds and several insects. There have been reports, also of similar results be-
tween plant species and landscape configuration, for example Kollmand and
Schneider (1999), investigating the correlation among mean a-diversity of
woody species and attributes of habitat patches, found that landscape attri-
butes such as patch size, total edge and patch fractal dimension are important
in explaining species richness at forest edges. Similarly, Grasholf-Bokdam
(1997) found that habitat heterogeneity and isolation of the habitat types
might determine plant species composition and species richness within a
habitat type.
The statistical analysis applied to relate diversity variables with landscape
metrics of patch types, shows the three diversity indices as having similar
behavior. This behavior can be explained by the fact that these variables are
highly correlated amongst themselves. This can be appreciated by examining
the Spearman correlation coefficient between species richness and the other two
diversity indices of the ranked diversity values for the 16 patch types (0.96 and
0.91, p > 0.001 for Shannon and Simpson respectively).
Although the strength and nature of the association between the diversity
variables (number of species, exponent Shannon and reciprocal Simpson) and
the metrics of patch types show a similar behavior, some differences can be
appreciated. First, the relationships between percentage of land and species
richness, exponent Shannon, as well as reciprocal Simpson indices (Table 3)
showed a strong correlation. As it was discussed before, this result may indicate
that events resulting in canopy openings, such as treefall, and ‘‘slash and burn’’
agriculture promote different environmental conditions in the boundaries of
the patches, compared with those at the center. This offers an increasing in the
availability of resources for plant species (Martinez-Ramos et al. 1988), and

[393]
1454

creates an augmentation of habitat diversity (Honnay et al. 2003), which can

improve plant species diversity for a specific patch type. However, the asso-
ciation between percentage of land and plant diversity indices is higher in
Shannon and Simpson indices, compared with that of number of species.
Second, in a similar way the shape index of patch types is more associated with
Shannon and Simpson indices than with the number of species (Table 3 and 5).
One of the reasons that may explain these relationships is that both plant
species diversity and the number of individuals in a population increase as a
consequence of a larger amount of resource availability (Brokaw 1985;
Denslow 1995). That is, after a disturbance process takes place in a tropical
forest, such as treefall, or a ‘‘slash and burn’’ event, the creation of canopy
opening allows for the regeneration, establishment and growth of some par-
ticular species that require high light levels. These species can be pioneer species
(Brokaw 1987) or those non-pioneer species that are suppressed under the
canopy (Hubbell and Foster 1986). The canopy opening affects species diver-
sity in part because disturbances generally produce a local increase in tree
density as a result of few large trees being replaced with numerous small ones
(Uhl et al. 1988; Denslow 1995). Therefore, Shannon and Simpson indices may
be more correlated to percentage of land and shape indices because they
consider not only the number of species, but also their abundance.
Additionally, it was also observed that the Simpson index was more strongly
associated to the percentage of land and to the shape of a given patch type than
the Shannon diversity index (Table 5). A possible explanation for this result
lies in the differential sensitivity to rare and dominant species of both diversity
indices (Peet 1974). The stronger association found between the Simpson
diversity index and indices of patch types possibly results from the fact that
Simpson index is more sensitive to the presence of the dominant species, such
as those species that allows for an increase on plant density within the patch
type.
The results of this paper indicated that plant diversity estimates of patch
types at landscape scale, such as number of species (total, tree and shrubs),
Shannon and Simpson indices can be predicted from landscape metrics of
habitat types. The appearance of shape, similarity and edge contrast indices
in most of the models might imply that some generalization can be made
about the effects of patch type metrics over plant species, which may have a
conservation use. Although it is difficult to find strategies that can help in the
design of conservation areas or maintenance of particular groups of species,
some generalizations can be made based on the results of this investigation. For
instance, the maintenance of the ‘‘slash and burn’’ agricultural activities
(Hernandez-Xolocotzi et al. 1995) or the promotion of small clear-cutting
areas distributed in the forest (Hartshorn 1989), may lead to a high level of
diversity not only allowing the establishment of pioneer species but also
increasing the availability of resources at the limits of the patch types, thus
providing conditions for the regeneration and establishment of other species.

[394]
 1455

On the other hand the use of Shannon and Simpson diversity indices to
assess plant diversity and its relationships to landscape patterns should be
considered with caution. However, the question remains as to which diversity
index should de used, and under what situations. Occurrences of rare species
are among the most frequently used criteria for selecting and prioritizing
habitat sites for preservation (Prendergast et al. 1993; Rossi and Kuitunen
1996). Moreover, the relative species abundance distribution of a tropical
forest often follows a J-inverted shape (Magnussen and Boyle 1995), in par-
ticular in the studied area (Hernandez-Stefanoni 2004), which implies the
probability of finding many species with low abundances. Thus, the Shannon
diversity index, with its greater sensitivity to rare species, should be considered
as having a greater importance in interpretation of the analysis. In fact, several
published studies of plant diversity in tropical forests have elected to use this
index (Fanliang et al. 1996; Nangendo et al. 2002). However, in particular
cases where a single dominant species is of interest for management or con-
servation proposes, the Simpson diversity index should be preferred.

References

Alvarez-Buylla E. and Garcia-Barros R. 1991. Seed and forest dynamics: a theoretical framework
and an example from the neotropics. Am. Nat. 137(2): 133–154.
Alvarez-Buylla E. and Martinez-Ramos M. 1992. Demography and allometry of Cecropia
obtusifolia, a neotropical pioneer tree and evaluation of the climax-pioneer paradigm for tropical
rain forest. J. Ecol. 80: 275–290.
Bengtsson J. 1998. Which species? What kind of diversity? Which ecosystem function? Some
problems in studies of relations between biodiversity and ecosystem function. Appl. Soil. Ecol.
10: 191–199.
Brokaw N.V.L. 1985. Treefalls, regrowth and community structure in tropical forest. In: Pickett
S.T.A. and White P.S. (eds), The Ecology of Natural Disturbances and Patch Dynamics. Aca-
demic Press, Orlando, Florida, USA, pp. 53–69.
Brokaw N.V.L. 1987. Gap-phase regeneration of three tree pioneer species in a tropical forest.
J. Ecol. 75: 9–19.
Cabrera C.E., Souza S.M. and Tellez V.O. 1982. Imagenes de la flora Quintanaroense. Centro de
Investigaciones de Quintana Roo, Mexico.
Carroll S.S. 1998. Modelling abiotic indicators when obtaining spatial predictions of species
richness. Environ. Ecol. Stat. 5: 257–276.
Debinski D. and Holt R.D. 2000. A survey and overview of habitat fragmentation experiments.
Conserv. Biol. 14: 63–76.
Denslow J.S., Schultz J.C., Vitousek P.M. and Strain B.R. 1990. Growth responses of tropical
shrubs to treefall gap environments. Ecology 71(1): 165–179.
Denslow J.S. 1995. Disturbance and diversity in tropical rain forest: the density effect. Ecol. Appl.
5(4): 962–968.
Eastman J.R. 1999. IDRISI 32: User’s Guide. Clark Labs for Cartography, Technology and
Geographic Analysis. Clark University, Worcester, Massachusetts, USA.
Earth Resource Mapping Ltd. 1998. ER Mapper 6.1. User Guide. San Diego CA.
Fanliang H., Legendre P. and LaFrankie J.V. 1996. Spatial patterns of diversity in a tropical rain
forest in Malaysia. J. Biogeogr. 23: 57–74.
Grashof-Bokdam C.J. 1997. Forest plants in an agricultural landscape in the Netherlands: effect of
habitat fragmentation. J. Veg. Sci. 8: 21–28.

[395]
1456

Gustafson E.J. 1998. Quantifying landscape spatial patterns: what is the state of the art? Ecosys-
tems 1: 143–156.
Gustafson E.J., Parker G.R. and Backs S.E. 1994. Evaluating spatial patterns of wildlife habitat: a
case study of the wild turkey (Meleagris gallopavo). Am. Midl. Nat. 131: 24–33.
Hartshorn G.S. 1989. Application of gap theory to tropical forest management: natural re-
generation on strip clear-cuts in the Peruvian Amazon. Ecology. 70(3): 567–569.
Hargis C.D., Bossonette J.A. and David J.L. 1998. The behavior of landscape metrics commonly
used in the study of habitat fragmentation. Landscape Ecol. 13: 167–186.
Hernandez-Stefanoni J.L. 2004. A Framework for Evaluating the Diversity Status of Trees, Shrubs
and Vines in a Tropical Forest in Mexico based on Landscape Patterns Metrics and Spatial
Modelling with GIS. Ph. D. dissertation, Trent University, Peterborough, Ontario, Canada.
Hernandez-Stefanoni J.L. 2005. Relationships between landscape patterns and species richness of
trees, shrubs and vines in a tropical forest. Plant Ecol. 179: 53–65.
Hernandez-Stefanoni J.L. and Ponce-Hernandez R. 2004. Mapping the spatial distribution of plant
diversity indices in a tropical forest using multi-spectral satellite image classification and field
measurements. Biodiv. Conserv. 13: 2599–2621.
Hernandez-Xolocotzi E., Baltaza E. and Tache S. 1995. La milpa en Yucatan, Un sistema agricola
tradicional. Colegiode Postgraduados, Mexico.
Honnay O., Hermy M. and Coppin P. 1999. Effects of area, age and diversity of forest patches in
Belgium on plant species richness, and implications for conservation and reforestation. Biol.
Conserv. 87: 73–84.
Honnay O., Piessens K., Van Landuyt W., Hermy M. and Guilinck H. 2003. Satellite based land
use and landscape complexity indices as predictors for regional plant species diversity. Landscape
Urban Planning 63(4): 241–250.
Howe H.F. 1990. Habitat implications of gap geometry in tropical forest. Oikos. 59: 141–143.
Hubbell S.P. and Foster R.A. 1986. Biology, change and history and the structure of tropical rain
forest communities. In: Diamond J. and Case T.J. (eds), Community Ecology. Harper and Row,
New York, pp. 314–329.
Ibarra-Martinez G. and Martinez-Ramos M. 2002. Landscape variation of liana communities in a
neotropical rain forest. Plant ecology. 160: 91–112.
Kollman J. and Schneider B. 1999. Landscape structure and diversity of flashy-fruited species at
forest edges. Plant Ecol. 144: 37–48.
Krebs C.J. 1989. Ecological Methodology. Harper Collins, New York.
Krummel J.R., Gardner R.H., Sugihara G., O’Neill R.V. and Coleman P.R. 1987. Landscape
patterns in a disturbed environment. Oikos. 48: 421–424.
Magnussen S. and Boyle T.J.B. 1995. Estimating sample size for inference about Shannon-Weaver
and the Simpson indices of species diversity. For. Ecol. Manage. 78: 71–84.
Magurran A.E. 1988. Ecological Diversity and its Measurement. Princeton University Press,
Princeton, NJ.
Margules C.R., Nicholls A.O. and Usher M.B. 1994. Apparent species turnover, probability of
extinction and the selection of natural reserves: a case of study of the Ingleborough limestone
pavements. Conserv. Biol. 8: 398–409.
Martinez-Ramos M., Alvarez-Buylla E., Sarukhan J. and Piñero D. 1988. Treefall age determi-
nation and gap dynamics in a tropical forest. J. Ecol. 76: 700–716.
Mazerolle M.J. and Villard M.A. 1999. Patch characteristics and landscape context as predictor of
species presence and abundance: a review. Ecoscience 6(1): 177–124.
McGarigal K. and Cushman S.A. 2002. Comparative evaluation of experimental approaches to the
study of habitat fragmentation studies. Ecol. Appl. 12(2): 335–345.
McGarigal K., Cushman S.A., Neel M.C. and Ene E. 2002. FRAGSTATS: Spatial Pattern
Analysis for Categorical Maps. University of Massachusetts.
Myers N. 1995. Environmental services of biodiversity. Proc. Natl. Acad. Sci. USA. 3: 2764–2763.
Myers N., Mittermeier R.A, Mittermeier C.G., da Fonseca G.A.B. and Kent J. 2000. Biodiversity
hotspots for conservation priorities. Nature 43(24): 853–858.

[396]
 1457

Nangendo G., Stein A., Gelens M., Gier A. and Albricht R. 2002. Quantifying differences in
biodiversity between a tropical forest area and a grassland area subject to traditional burning.
For. Ecol. Manage. 164: 109–120.
Peet R.K. 1974. The measurement of species diversity. Annu. Rev. Ecol. Syst. 5: 285–307.
Prendergast J.R., Quinn R.M., Lawton J.H., Eversham B.C. and Gibbons D.W. 1993. Rare spe-
cies, the coincidennce of diversity hotspots and conservation strategies. Nature 365: 335–337.
Putz F.E. 1984. The natural history of lianas on Barro Coloraco Island, Panama. Ecology 65(6):
1713–1724.
Putz F.E. and Chai P. 1987. Ecological studies in lianas in Lambir National Park, Sarawak. J. Ecol.
75: 523–531.
Riitters K.H., O’Neill, R.V., Hunsaker C.T., Wickham J.D., Yankee D.H., Timmins S.P., Jones
K.B. and Jackson B.L. 1995. A factor analysis of landscape pattern and structure metrics.
Landscape Ecol. 10: 23–39.
Roberts M.R. and Gilliam F.S. 1995. Patterns and mechanisms of plant diversity in forested
ecosystems: implications for forest management. Ecol. Appl. 5(4): 969–977.
Rossi E. and Kuitunen M. 1996. Ranking of habitats for the assessment of ecological impact in
land use planning. Biol. Conserv. 77: 227–234.
Schupp E.W., Howe H.F., Augspurger C.K. and Levey D.J. 1989. Arrival and survival in tropical
treefall gaps. Ecology 70(3): 562–564.
Tabachnick B.G. and Fidell L.S. 1996. Using Multivariate Statistics. Harper Collins College
Publishers, New York.
Uhl C., Clark K., Dezzeo N. and Maquirino P. 1988. Vegetation dynamics in Amazonian treefall
gaps. Ecology 69: 751–763.
Whigham D.F., Olmsted I., Cabrera Cano E. and Harmon M.E. 1991. The impact of hurricane
Gilbert on trees, literfall and woody debris in a dry tropical forest in the northeastern Yucatan
peninsula. Biotropica 23: 434–441.
Whitmore T.C. 1989. Canopy gaps and the mayor groups of forest trees. Ecology 70(3): 536–538.

[397]
Biodiversity and Conservation (2006) 15:1459–1466
Springer 2006
DOI 10.1007/s10531-005-0599-5

-1

Distribution, habitat and Red List status of the New

Caledonian endemic tree Canacomyrica monticola
(Myricaceae)

JANE HERBERT
School of Biology, Sir Harold Mitchell Building, University of St Andrews, KY16 9TH, UK; Present
address: School of Integrative Biology, University of Queensland, Brisbane QLD 4072, Australia;
(e-mail: [email protected]; phone: +01334-463372; fax: +01334-463366)

Received 24 May 2004; accepted in revised form 24 June 2005

Key words: Bush fire, Endangered, In situ conservation, Mining, Primary forest, Serpentine,
Ultramafic

Abstract. The monotypic genus Canacomyrica Guillaumin is a small tree endemic to the rare
remaining fragments of primary forest growing on ultramafic geology in New Caledonia. In the
rich flora of this island it is one of many endemics to be threatened by habitat loss due to a variety
of factors, most significantly open-cast mining for nickel. Using field observations and data from
herbarium specimens the extent of occurrence of Canacomyrica monticola is established to be
approximately 1420 km2. Within this area the distribution of C. monticola is very fragmented and
limited to just 11 known localities. Six localities are outside protected areas; two of these may be
imminently threatened by mining activity and another may be threatened by bush fires. It is
recommended that the IUCN Red List status of Endangered (EN B1ab (i,ii,iv,v)) is assigned to this
species.

Introduction

Canacomyrica is a monotypic genus endemic to the pacific island group of New

Caledonia, a territory renowned for its rich flora (Jaffré et al. 2001b). Grande
Terre, the main island has an area of just 19,000 km2 yet it is home to five
endemic families, 106 endemic genera and more than 3200 species of vascular
plants, a remarkable 74–75% of which are endemic (Jaffré et al. 2001b).
Among these endemics are representatives of ancient lineages such as the taxon
considered to be sister to all other extant flowering plants, Amborella tricho-
poda (Zanis et al. 2002). The combination of high endemism and the presence
of ‘relict’ taxa has lead to the recognition of New Caledonia as a distinct
phytogeographic region (Takhtajan 1986).
The unique flora of New Caledonia is under threat from a number of factors
including deforestation, introduced species, fire, agriculture and livestock
grazing (Olson et al. 2000). However, the greatest single threat posed to the
island’s plants is the practice of open-cast mining for metals. New Caledonia
has one third of the world’s reserves of nickel ore, found in the ultramafic
geology of the southern region and in the isolated massifs of the west coast of

[399]
1460

Grande Terre (Brooks 1987). It is to these ultramafic outcrops with their

economically valuable, nutrient-deficient soils, that many endemic plant taxa
are restricted (eg. Pintaud et al. 1999; Jaffré et al. 2001a; Herbert et al. 2002;
Whitlock et al. 2003).
According to IUCN criteria (IUCN 1997) 14.4% of plant species in New
Caledonia are Red Listed. However, this figure is likely to be much higher for
plants endemic to, or growing predominantly on, ultramafic substrate. The
lack of sufficient data on many species is likely to be a further factor con-
tributing to a misrepresentation of the true number of plants threatened with
extinction in New Caledonia. Where specific groups have been examined in
detail, the number of threatened species is found to be high. For example, the
43 species of conifer in the Territory are all endemic and of these, 67% are Red
Listed (Farjon and Page 1999); similarly, of the 37 endemic palms, 35% are
considered to be threatened (Pintaud et al. 1999).
Regrettably only 5000 km2 of primary vegetation (28% of the original ex-
tent) remains in New Caledonia (Myers et al. 2000) and less than 10% of the
land area has protected status (WWF and IUCN 1995; Jaffré et al. 1998;
Pintaud et al. 1999). It has become clear in recent years that there is an urgent
need for understanding and protection of the flora of this island, along with its
unique terrestrial fauna and marine biota (Bouchet et al. 1995; Jaffré et al.
1998; Mittermeier et al. 1996; Proctor 2003). New Caledonia has been
acknowledged as one of the world’s 25 biodiversity ‘hotspots’ (Myers et al.
2000) and has been identified as one of the WWF’s ‘Global 200’ ecoregions,
singled out as priority targets for conservation action (Olson et al. 2000).
Canacomyrica monticola is an evergreen shrub or small tree (up to 7 m) with
coriaceous leaves, flowers borne in spikes and black drupaceous fruits. It is
entirely restricted in its distribution to primary forest on the ultramafic soils in
the south of Grande Terre. Since its description (Guillaumin 1940) Canac-
omyrica has been largely neglected, almost nothing is known of its ecology, the
plant is not in cultivation anywhere in the world, and its conservation status
remains unknown.
As the sole member of a geographically isolated genus, Canacomyrica has an
important bearing on our understanding of evolutionary processes in Myric-
aceae (Herbert 2005). Basic data about its ecology and distribution is much
needed to further knowledge about the entire family. Such information will
also contribute to a better appreciation of the ecology of the rare remaining
fragments of primary vegetation in New Caledonia. The aim of this study is to
determine the distribution and habitat of Canacomyrica in New Caledonia and
to assess its conservation status.

Materials and methods

A list of all known localities for Canacomyrica was compiled from herbarium
specimens held in three collections: Royal Botanic Garden, Edinburgh (E) (16

[400]
 1461

specimens); Institut de recherche pour le développement, Nouméa (NOU) (35

specimens); the Muséum National D’Histoire Naturelle, Paris (P) (47 speci-
mens). The latter two institutions hold large numbers of collections of the New
Caledonian flora. Herbarium specimens have been acknowledged as suitable
data sources for assessing plant distributions in the absence of other infor-
mation (Willis et al. 2003).
During a 3 week expedition to New Caledonia (RBGE expedition, May
2001) populations of Canacomyrica were sought from throughout the ultra-
mafic region of Grande Terre. Field observations, collections of herbarium
specimens, seed and seedling collections were made by the author. Two pop-
ulations were located and at each field site, data were recorded on substrate,
elevation, habitat features and threats to habitat. An attempt was made in each
case to assess the extent and demography of the population.
The conservation status of Canacomyrica was determined using the IUCN
Red List criteria (IUCN 2001). Measurement of extent of occurrence is the first
step in estimating the geographic distribution of a species. The known localities
of Canacomyrica were plotted on a map and a boundary drawn between them,
the extent of occurrence was estimated by manual measurement of the area
within this boundary.

Results

A survey of the herbarium specimens held in Edinburgh (E), Paris (P) and
Noumea (NOU) revealed 10 localities for Canacomyrica, and another was
reported by T. Jaffré (personal communication). The exact positions of col-
lection sites given on herbarium specimens were determined by consulting the
H.S. MacKee gazetteer (Muséum National D’Histoire Naturelle, Paris; http://
phanero.novcal.free.fr/site.html). The eleven localities, all of which were in the
south of the island on ultramafic substrate, are shown in Figure 1 and details
are given in Table 1. Manual measurement of the area between the 11 known
localities gave an extent of occurrence of 1420 km2.
Canacomyrica was examined in the field at Mont Bouo and Mont Mamié
(Table 1, Figure 1). Plants at both sites were highly localised, occurring in
almost monospecific stands with few or no outlying individuals. The local
distribution of the populations appeared to be limited by water availability.
At Mont Bouo plants of Canacomyrica were found only between altitudes of
1050 m and approximately 1150 m, growing in a rainforest community on
ultramafic substrate. Mature individuals were 3–4 m in height. The habitat at
Mont Bouo appeared to be undisturbed and access was difficult.
At Mont Mamié plants of Canacomyrica were found at 500 m, the lowest
recorded altitude for the plant (most collections have been made above
800 m). Canacomyrica was growing in a low scrub community co-dominated
by Cyperaceae species on ultramafic substrate, inundated with water from
abundant natural springs. Mature individuals were up to 1 m in height; many

[401]
1462

Figure 1. Map of New Caledonia showing the extent of ultramafic geology and the position of the
11 known localities for Canacomyrica (extent of ultramafic follows Jaffré et al. 1987).

seedlings were observed in this population. The habitat at Mont Mamié was
disturbed and access was relatively easy due to the presence of mining pros-
pecting tracks.
At both sites there were more than 30 mature individuals but, it was not
possible to estimate the total number of mature plants at either site due to the
time constraints of the expedition. The total area occupied by the population at
Mont Bouo was estimated to be approximately 100 m2. Time constraints also
prevented estimation of the total area occupied by the population at Mont
Mamié.
It was expected that further populations would be found at Mont Mou and
Rivière Bleue but plants of Canacomyrica were not found at these localities
during the expedition. All suitable habitat for Canacomyrica was searched at
Mont Mou. At Rivière Bleue it was not possible to search all suitable habitat
due to the difficulty of the terrain. On the basis of these observations, it is
considered that a single collection from Mont Mou (Baumann-Bodenheim
15679, P) is a doubtful locality for Canacomyrica. It is thought likely that the
specimen was collected elsewhere and incorrectly labelled, alternatively (but
less likely) this collection may represent an extinct population. It was not
possible to gain access to populations within the strictly protected Nature
Reserve of Montagne des Sources. Other locations were not visited due to the
time constraints of the field study.

[402]
 Table 1. Known localities for Canacomyrica, with coordinates and protected status details.

Locality Coordinates Protected status and notesa

Mont Bouo (Koghis range)b 2210¢ S 16630¢ E Protected – amenity protected area, adjacent to the Strict Nature Reserve of Montagne des Sources
Mont Mamiéb 2206¢ S 16653¢ E Unprotected – mining activity observed
N’Goic 2149¢ S 16630¢ E Unprotectede – adjacent to existing Special Botanical Reserve of Mt Humboldt (5 km)
Montagnes des Sourcesc 2207¢ S 16633¢ E Protected – Strict Nature Reserve of Montagne des Sources
Pourinac 2201¢ S 16644¢ E Unprotected – adjacent to Special Botanical Reserve of Haute Pourina (3 km) and Natural park
of Riviére Bleue (5 km)
Ouinnéc 2157¢ S 16642¢ E Unprotectede
[403]

Kouakouéc 2157¢ S 16632¢ E Protectedf – Special Fauna and Flora Reserve of Mt Kouakoué
Humboldtc 2153¢ S 16625¢ E Protectedf – Special Botanical Reserve of Mt Humboldt
Nembrouc 2145¢ S 16613¢ E Unprotected – adjacent to Special Botanical Reserve of Forêt de Saille (5 km)
Nékandoc 2150¢ S 16620¢ E Unprotected – adjacent to Special Botanical Reserve of Mt Humboldt
Forêt de Sailled 2140¢ S 16613¢ E Protected – Special Fauna and Flora Reserve of Forêt de Saille
a
Sources of information on protected status: Pintaud et al. (1999), J. Manaute (personal communication).
b
Locality visited by the author.
c
Locality determined from herbarium specimens.
d
Locality according to T. Jaffré (personal communication).
e
Site adjacent to proposed ‘Ni-Kouakoué-Ouinné’ reserve.
f
Site expected to be included in the proposed ‘Ni-Kouakoué-Ouinné’ reserve. Map coordinates are intended as a guide only.

1463
1464

Discussion

Canacomyrica is known to occur in 11 localities in the south of Grande Terre,

New Caledonia where it grows exclusively on ultramafic substrates. Of these
localities, six are afforded no official protection and disturbance has been ob-
served at the Mont Mamié site. There is evidence of disturbance caused by
mining prospecting at the Ouinné site and the Nembrou site is potentially
threatened by bush fire (J. Manaute and T. Jaffré, personal communication).
Canacomyrica is estimated to have an extent of occurrence of 1420 km2 and
within this area its distribution is patchy. If the population at Mt Bouo is
typical, where Canacomyrica occurs in an area of approximately 100 m2, then
it is tempting to speculate that the total area of occurrence for the species is
significantly smaller than its area of occupancy. Continuing decline is projected
in both the area and extent of occurrence, and number of locations. Also,
continuing decline in number of mature individuals is inferred at the Mont
Mamié, Ouinné and Nembrou sites. It is therefore recommended that Canac-
omyrica monticola is given the IUCN (2001) Red List status of Endangered
(EN B1ab (i,ii,iv,v)).
The principal threats to Canacomyrica are likely to be open-cast mining and
bush fires in areas of its occurrence lacking protected status. The most immi-
nently threatened populations are at Mont Mamié, Ouinné and Nembrou.
Habitat destruction in an area where Canacomyrica occurs may entirely de-
stroy localised and fragmentary populations that are characteristic of this
species. It is encouraging that age structure, indicative of recent regeneration,
was observed at Mont Mamié although it is stressed that this was only ob-
served on undisturbed substrate. In experimental work, approximately 90% of
fruits examined were sterile and ex situ germination of seeds and cultivation of
seedlings was unsuccessful (Herbert 2005). This suggests that in situ measures
are likely to be the best if not the only approach for the conservation of this
species.
Conservation achieved through a network of protected areas is considered to
be the most effective way to preserve biodiversity (Primack 2000). This is never
more appropriate than in the case of species with highly specialised ecological
requirements that are unlikely to thrive anywhere but in their natural habitat,
such as Canacomyrica. Whilst it is desirable for additional data to be collected
on numbers of individuals, area of occurrence and threats to localities other
than those detailed here, the data presented are sufficient to permit the fol-
lowing recommendations to be made: (1) expansion of the Special Botanical
Reserves of Haute Pourina and the Forêt de Saille should be undertaken
to protect the populations at Pourina and Nembrou; (2) the proposed ‘Ni-
Kouakoué-Ouinné’ reserve (project under consideration; J. Manaute, personal
communication) should include the sites at N’Goı̈ and Ouinné; (3) special
attention should be given to Canacomyrica when botanical surveys or inven-
tories are carried out at sites on ultramafic to enhance knowledge of the dis-
tribution and demography of this species; (4) an investigation to assess the level

[404]
 1465

of genetic diversity, both within and among populations, should be carried out
to act as a guide for the prioritisation of populations in future conservation
management.
Protection of the above mentioned populations would raise the number of
protected localities for Canacomyrica from five to nine, or 82% of all known
sites. These measures represent the first steps towards ensuring the continued
survival of Canacomyrica. Furthermore, conservation measures targeted at
Canacomyrica will help to raise the profile and survival prospects of some of
the last remaining fragments of New Caledonia’s primary forest.

Acknowledgements

I would like to thank two anonymous reviewers for comments on an earlier

draft of this manuscript. T. Jaffré (IRD, Nouméa, New Caledonia) and
J. Manaute (Province Sud) are acknowledged for valuable additional infor-
mation and comments on earlier drafts of the manuscript. M. Hughes (RBGE)
is thanked for comments on the manuscript. P.M. Hollingsworth and M.
Gardner (RBGE) are thanked for organising the expedition to New Caledonia
and, along with A. Ponge, for their help with collecting Canacomyrica. The
administration of Province Sud, New Caledonia is acknowledged for permis-
sion to collect. H. Hodge (University of St Andrews) and A. Ensoll (RBGE)
are thanked for cultivation of plants. Financial support for field work was
provided by the Russell Trust (University of St Andrews) and the Davis
Expedition Fund (University of Edinburgh). This study was supported by a
NERC studentship (NER/S/A/2000/03638).

References

Bouchet P., Jaffré T. and Veillon J.M. 1995. Plant extinction in New-Caledonia – Protection of
sclerophyll forests urgently needed. Biodivers. Conserv. 4: 415–428.
Brooks R.R. 1987. Serpentine and its Vegetation. A Multidisciplinary Approach. Croom Helm,
London, UK, pp. 330–353.
Farjon A. and Page C.N. (compilers) 1999. Status Survey and Conservation Action Plan. IUCN/
SSC Conifer Specialist Group, IUCN, Gland, Switzerland, pp. 41–50.
Guillaumin A. 1940. Matériaux pour la flore de la Nouvelle-Calédonie. LVII. La présence d’une
Myricacée. Bulletin de la Société Botanique de France 87: 299–300.
Herbert J. 2005 Systematics and biogeography of Myricaceae. Unpublished Ph.D. Thesis, Uni-
versity of St Andrews, St Andrews.
Herbert J., Hollingsworth P.M., Gardner M.F., Mill R.R., Thomas P.J. and Jaffré T. 2002.
Conservation genetics and phylogenetics of New Caledonian Retrophyllum (Podocarpaceae)
species. New Zeal. J. Bot. 40: 175–188.
IUCN 1997. Red List of Threatened Plants. IUCN Species Survival Commission, http://iucn.org/
themes/ssc/97plrl/table5.htm [accessed 26 April 2004].
IUCN 2001. IUCN Red List Categories and Criteria: Version 3.1. IUCN Species Survival Com-
mission, Gland, Switzerland.

[405]
1466

Jaffré T., Bouchet P. and Veillon J.M. 1998. Threatened plants of New Caledonia: Is the system of
protected areas adequate? Biodivers. Conserv. 7: 109–135.
Jaffré T., McCoy S., Rigault F. and Navarro E. 2001a. A comparative study of flora and symbiotic
microflora diversity in two Gymnostoma formations on ultramafic rocks in New Caledonia.
S. Afr. J. Sci. 97: 599–603.
Jaffré T., Morat Ph., Veillon J.M. and MacKee H.S. 1987. Changements dans la végétation de la
Nouvelle-Calédonie au cours du Tertiaire: la végétation et la flore des roches ultrabasiques.
Adansonia 4: 356–391.
Jaffré T., Morat Ph., Veillon J.M., Rigault F. and Dagostini G. 2001b. Composition and Char-
acteristics of the Native Flora of New Caledonia. IRD, Nouméa, pp. 28–29.
Mittermeier R.A., Werner T.B. and Lees A. 1996. New Caledonia – A conservation imperative for
an ancient land. Oryx 30: 104–112.
Myers N., Mittermeier R.A., Mittermeier C.G., da Fonseca G.A.B. and Kent J. 2000. Biodiversity
hotspots for conservation priorities. Nature 403: 853–858.
Olson D.M., Dinerstein E., Abell R., Allnutt T., Carpenter C., McClenachan L., D’Amico J.,
Hurley P., Kassem K., Strand H., Taye M. and Thieme M. 2000. The Global 200: A Repre-
sentation Approach to Conserving the Earth’s Distinctive Ecoregions. World Wildlife Fund-US,
Conservation Science Program, Washington, DC. http://www.panda.org/about_wwf/where_
we_work/ecoregions/global200/downloads [accessed 26 April 2004].
Pintaud J.C., Jaffré T. and Veillon J.M. 1999. Conservation status of New Caledonia palms. Pac.
Conserv. Biol. 5: 9–15.
Primack R.B. 2000. A Primer of Conservation Biology. Sinauer Associates Inc., Sunderland,
Massachusetts, pp. 183.
Proctor J. 2003. Vegetation and soil and plant chemistry on ultramafic rocks in the tropical Far
East. Perspect. Plant Ecol. Evol. Syst. 6: 105–124.
Takhtajan A. 1986. Floristic Regions of the World. University of California Press, Berkeley,
pp. 248–253.
Whitlock B.A., Karol K.G. and Alverson W.S. 2003. Chloroplast DNA sequences confirm the
placement of the enigmatic Oceanopapaver within Corchorus (Grewioideae : Malvaceae S. L.,
formerly Tiliaceae). Int. J. Plant Sci. 164: 35–41.
Willis F., Moat J. and Paton A. 2003. Defining a role for herbarium data in Red List assessments: a
case study of Plectranthus from eastern and southern tropical Africa. Biodivers. Conserv.
12: 1537–1552.
WWF and IUCN 1995. Centres of Plant Diversity. A Guide and Strategy for their Conservation
Vol. 2. IUCN Publications Unit, Cambridge, UK, pp. 529–541.
Zanis M.J., Soltis D.E., Soltis P.S., Mathews S. and Donoghue M.J. 2002. The root of the
angiosperms revisited. Proc. Natl. Acad. Sci. USA 99: 6848–6853.

[406]
Biodiversity and Conservation (2006) 15:1467–1495
Springer 2006
DOI 10.1007/s10531-005-1876-z

-1

Composition of woody species in a dynamic

forest–woodland–savannah mosaic in Uganda:
implications for conservation and management

GRACE NANGENDO1,*, HANS TER STEEGE2 and

FRANS BONGERS3
1
International Institute for Geo-information Science and Earth Observation, P.O Box 6, 7500 AA
Enschede, The Netherlands; 2National Herbarium Netherlands, Utrecht University Branch, Heidel-
berglaan 2, 3584 CS Utrecht, The Netherlands; 3Forest Ecology and Forest Management Group,
Centre for Ecosystem Studies, Wageningen University, P.O. Box 47, 6700 AA Wageningen,
The Netherlands; *Author for correspondence (phone: +31-53-4874444; fax: +31-53-4874388)

Received 21 December 2004; accepted in revised form 19 July 2005

Key words: Budongo Forest Reserve, Fire disturbance, Forest–woodland–savannah mosaics,

Species gradient, Woody species composition

Abstract. Forest–woodland–savannah mosaics are a common feature in the East African

landscape. For the conservation of the woody species that occur in such landscapes, the species
patterns and the factors that maintain it need to be understood. We studied the woody species
distribution in a forest–woodland–savannah mosaic in Budongo Forest Reserve, Uganda. The
existing vegetation gradients were analyzed using data from a total of 591 plots of 400 or 500 m2
each. Remotely sensed data was used to explore current vegetation cover and the gradients there in
for the whole area. A clear species gradient exists in the study area ranging from forest, where there
is least disturbance, to wooded grassland, where frequent fire disturbance occurs. Most species are
not limited to a specific part of the gradient although many show a maximum abundance at some
point along the gradient. Fire and accessibility to the protected area were closely related to vari-
ation in species composition along the ordination axis with species like Cynometra alexandri and
Uvariopsis congensis occurring at one end of the gradient and Combretum guenzi and Lonchocarpus
laxiflorus at the other. The vegetation cover classes identified in the area differed in diversity,
density and, especially, basal area. All vegetation cover classes, except open woodland, had indi-
cator species. Diospyros abyssinica, Uvariopsis congensis, Holoptelea grandis and all Celtis species
were the indicator species for the forest class, Terminalia velutina and Albizia grandbracteata for
closed woodland, Grewia mollis and Combretum mole for very open woodland and Lonchocarpus
laxiflorus, Grewia bicolor and Combretum guenzi for the wooded grassland class. Eleven of the
species occurred in all cover classes and most of the species that occurred in more than one
vegetation cover class showed peak abundance in a specific cover class. Species composition in the
study area changes gradually from forest to savannah. Along the gradient, the cover classes are
distinguishable in terms of species composition and vegetation structure. These classes are, how-
ever, interrelated in species composition. For conservation of the full range of the species within
this East African landscape, the mosaic has to be managed as an integrated whole. Burning should
be varied over the area with the forest not being burnt at all and the wooded grassland burnt
regularly. The different vegetation types that occur between these two extremes should be main-
tained using a varied fire regime.

[407]
1468

Introduction

Large areas of East Africa are covered with forest–woodland–savannah (FWS)

mosaics. Fire, both of natural and anthropogenic origin, is typical for the
woodland–savannah part of the mosaic (Walter 1985). The fresh grass that
regrows after fire is advantageous for the many ungulates of East Africa and
many of these ecosystems are, therefore, managed mainly for wildlife conser-
vation. Forests, on the other hand, are managed for timber production, for
woody species conservation, as water catchment areas, and for their aesthetic
values. Consequently the forest and the woodland–savannah areas, even where
they occur together as a mosaic, are often treated as independent conservation
entities.
Uganda is has large tracts of FWS mosaics. Several of these FWS mosaics
have been enlisted for demarcation as conservation areas because of their high
biodiversity value (Forest Department Uganda 1999), yet surveys in such areas
have focused on the forest and have not taken the woodland areas into
account. Although forests are arguably richer in species (Sheil and Burslem
2003), this does not do justice to the specific flora and fauna found in wood-
lands that are adjacent to or within the forests.
In our study area, Budongo Forest Reserve, fire has been used as a man-
agement tool in the woodland areas for hundreds of years (Paterson 1991). The
local people, resident on the outskirts of the forest reserve, set most of the fires.
Changes in management of the area, which resulted in more active policing of
the protected area and the establishment of a gate entrance to the only access
route, have led to less use of fire in the woodland areas. As a consequence,
forest vegetation is now colonizing the woodland areas (van Straaten 2003). It
is unknown to what extent accessibility to the protected area acts as a con-
trolling factor. As specific floristic information for the area is lacking, it is as
yet unknown what the effects of continuing reforestation will be on the tree
diversity of Budongo Forest Reserve. In this paper, we explore the current
status of the forest in terms of species composition and diversity, and how it is
distributed in space.
Studies that have explored the species interrelationship within FWS mosaics
are rare (Hovestadt et al. 1999). Most studies have concentrated on species
distribution within the forest (Eggeling 1947; Sheil et al. 2000; Mwami and
McNeilage 2003; Eilu et al. 2004) or the woodland–savannah (Swaine et al.
1992; Schwilk et al. 1997; Schwartz and Caro 2003; Li et al. 2004). Previous
data of Budongo forest (Eggeling 1947) stimulated the emergence of the
important Intermediate Disturbance Hypothesis (Connell 1978), which states
that in a landscape, species diversity is highest in areas with an intermediate
level of disturbance. In areas of high disturbance and areas of low or no
disturbance, species diversity is low. At the time of Eggeling’s (Eggeling 1947)
study, frequent burning had arrested the forest succession into the woodland.
Eggeling’s gradient mainly reflects succession within the forested area. With the
increased control of burning over the years, resulting in variation of burning

[408]
 1469

with some areas burnt more often than others, forest succession into the
woodland became possible. Hence, an added component of our study is the
extension of a historically important succession gradient, exploring its range
into the woodland areas.
In our study, special emphasis was placed on how the existing vegetation
types can be characterized in terms of the woody plants and the implication of
the observed species patterns to conservation of woody plants in such land-
scapes. The hypothesis made is that all the vegetation types that exist within the
study area, and the species they support, are an integral part of a composi-
tional/successional gradient that stretches across the FWS mosaic. We asked
the following questions: Is it possible to quantify the gradient? What species are
specific for certain areas? How does the species composition vary along the
succession gradient? Can the gradient be explained in relation to environmental
variables?
A further question we address is whether a satellite image classification of the
area can be used to adequately map the vegetation and its composition in the
area. For this we made use of discrete vegetation cover classes, obtained from a
classification carried out using a combination of spectral information and
environmental variables’ information (Nangendo et al., submitted). The veg-
etation classes are considered a proxy of the vegetation types found in the area.
Standard vegetation indices (NDVI and Tasseled Cap vegetation index) based
on the same image were also compared in their ability to explain the observed
gradient. Finally, we discuss the conservation and management implications of
our results.

Materials and methods

Study area

The work was carried out in the northern part of Budongo Forest Reserve in
north-western Uganda. The area is located between 135¢ and 155¢ N and
3118¢ and 3142¢ E. It receives between 1397 and 1500 mm of rain annually on
100 to 150 days. There are two main forest blocks: the main Budongo Forest
block and the Kaniyo-Pabidi Forest block (Figure 1). A woodland area,
interspersed with forest patches, commonly referred to as Kaniyo-Pabidi
woodland, separates these two blocks.
The underlying geology of the Budongo Forest is Precambrian origin con-
sisting of high-grade metamorphic rocks of the 2.9 billion-year-old granulite
group (van Straaten 1976). The soils over 90% of the study area are orthic
Ferralsols: highly weathered, deep, well drained soils with low pH. The
remaining 10% of the area has typically shallow soils, called Lithosols. These
soils are mainly found on hilltop regions and are predominantly underlain by
rocks. In river valleys, eutric Fluvisols are present.

[409]
1470

Figure 1. Map of study area location.

In the woodlands, fire has been prevalent for hundreds of years (Paterson
1991). The woodland burning was initially carried out by the local people for
purposes of hunting and refreshing grass for both domestic and wild ungulates
(Buechner and Dawkins 1961). With the transfer of the control of the wood-
lands from the local people (Bunyoro Kingdom) to the central government
(Forest Department) in 1968, measures to control burning were put in place
(Forest Department Uganda 1997). These were not very effective, however,
until the establishment of the joint management between Forest Department
and Uganda Wildlife Authority in the mid 1980s. Fewer, and smaller, areas are
now burnt and the burning is also less frequent. The woodland is therefore
heterogeneous and made up of vegetation patches at varying stages of recovery
since they were last burnt.

Data collection

Data was collected from 591 plots, 266 of which had an area of 400 m2 and
326 with an area of 500 m2. All data were collected during the same period
(August–October 2002). Along a transect, perpendicular lines were laid every
300 m. Along each perpendicular line, data were collected at every 75 m. For
sites 1–5, a plot size of 400 m2 was used (Figure 2), while for sites a–e, it was
500 m2. Based on a 2002 satellite image of the study area, sites 1–5 were
located in areas that showed a similar spectral reflectance, whereas sites a–e
were located in areas that showed varying spectral reflectance. The variation

[410]
 1471

Figure 2. The location of the data collection points. 1, 2, 3, 4 and 5 are locations where the plots
were 400 m2 and a, b, c, d, e and f are locations where the plot size was 500 m2.

[411]
1472

of the site locations was to ensure that we capture as much as possible of the
species variation within the area. In each plot, the following data were col-
lected:
• Plot coordinates
• Species names, diameter at breast height (DBH) for all woody plants ‡10 cm
DBH, measured at 130 cm. If the tree was buttressed and abnormal at
130 cm, the diameter was measured just above the buttress where the stem
assumes a near cylindrical shape.
• Canopy cover percentage, using a canopy densiometer (Robert E: Lemmon,
Forest Densiometers, Oklahoma, USA), following the provided guidelines.
Four measurements were taken in each plot and an average of these mea-
surements was calculated to determine the final canopy cover of the plot.
• A fire indicator value. The fire indicator value was based on several factors
(1) the degree of scorching on the woody stems i.e. if it was fresh or old, (2) if
there existed remains of burnt grass in the undergrowth and (3) whether fresh
ash was found in the area. The last two factors were used to confirm areas
with recent fire. Plots with fresh fire scorching on the woody stems, remains
of burnt grass or ash were recorded as ‘recent burns’ and labelled class 2.
Plots with old signs of fire were labelled class 1 (old fires) and plots with no
sign of fire were labelled class 0 (no fire).
Species identification was based on Eggeling and Dale (1952) and Hamil-
ton (1991). Samples of the species that could not be clearly identified in the
field by the botanists on the team (Israel Tinka and Hezekias Ddumba) were
sent to the Uganda National Herbarium, Makerere University, where they
were identified.

Data preparation

A Detrended Correspondence Analysis (DCA) (Multi-Variate Statistical

Package MVSP 3.11, Kovach Computing Services, UK) was run using the two
data sets i.e. for the 400 m2 plots and the 500 m2 plots. When the plot scores of
DCA axis one and two were plotted together, using a separate symbol for each
plot size, the data for the two sets fell within the same range i.e. they showed
near to identical results and complemented each other. As the plots also
overlap spatially, it was therefore decided to pool the two data sets. Expressing
density as the number of trees per 1000 m2, the abundance values were cal-
culated for each plot. Plots with less than 10 individuals were removed from the
data. As the larger plots have more individuals and thus capture more species
on average than the smaller plots, rare species, defined as those having a total
of less than 25 individuals, were also removed from the data set. The final
dataset consisted of 491 plots with 45 species.
From the DBH values measured in the field, basal area (BA, m2 ha
1) was
calculated for each plot, including all trees of the actual plot data.

[412]
 1473

Remote sensing

Values of the Normalized Difference Vegetation Index (NDVI), which is well

correlated with vegetation biomass (Tucker 1979) and Tasseled Cap vegetation
index (TC), which has a good correlation with forest stand density (Crist et al.
1986), were extracted for each plot from the respective vegetation index maps
calculated using a 2002 Landsat ETM+ satellite image. These values were used
for further analyses. Vegetation indices provide values that are indicative of the
spectral reflectance of the vegetation at a given place. Depending on the sa-
tellite image bands selected and the ratios used, each vegetation index measure
will result in a different value for a specific plot. Because there is a high vari-
ation in reflectance over a forested area, the resultant pixel values for a given
index vary from point to point resulting in continuous values over the forested
area.
NDVI used two bands, red and near infrared. Tasseled Cap incorporates
more information by using six different light bands (blue, green, red, near
infrared and far infrared). Depending on the ratios of combination of the six
bands, different multispectral features are obtained (Crist and Cicone 1984).
The first three features usually account for most of the variation in a single date
image (Collins and Woodcock 1996). These three have been labelled bright-
ness, greenness and wetness, respectively. All three were used in this study.
Vegetation cover class values for each plot were extracted from a vegetation
cover map of the area obtained from an earlier classification (Nangendo et al.,
submitted) of a Landsat ETM+ image using both spectral and environmental
information. All 592 plots were separated into the discrete cover classes (forest,
closed woodland, open woodland, very open woodland and wooded grass-
land). Having used a Landsat satellite image, with a pixel size of 30 m, in the
classification, the minimum area belonging to a specific cover class is 900 m2.
Species composition, diversity and forest structure were analyzed in consid-
eration of the cover class in which each plot fell with the assumption that these
vegetation cover classes were representative of the major vegetation variation
within the area.

Accessibility

We used distance from the southern forest boundary to each plot as a surrogate
for accessibility, by the local people, to the sampled areas. The conservation
area gate marks the southern boundary between the conservation area and the
local people’s settlements. From here on, distance will be referred to as ‘dis-
tance from gate.’ During fieldwork, it was observed that because of the gate
control, the local people entered the protected area at other points along the
boundary of the protected area, instead of using the road. Having recorded the
coordinate of the gate location, an east–west line was established at this point

[413]
1474

and distance for each plot was calculated based on this line. This provided the
plot distance relative to the conservation area gate.

Analysing the gradient

We used Detrended Correspondence Analysis to explore the species distribu-

tion within the study area. To determine which variables best explained the
gradient in species composition, the plot scores on the DCA axes were related
to the site variables using stepwise regression.

Linking remote sensing with the gradient

Two approaches were used in analysing remote sensing outputs. First, plot
values obtained from vegetation indices (such as TC), which are continuous
classifiers, were compared to DCA plot scores. To identify the vegetation index
that best explained the gradient, a non-linear regression method was used since
the scatter plot of the DCA vs. the index values showed a non-linear rela-
tionship. Second, discrete classes obtained from an earlier classification
(Nangendo et al., submitted) were analyzed for differences in terms of species
composition and diversity and, in basal area. Although the same satellite image
was used for the classification and for the creation of the index maps, the plots
used for the classification are not the same as those used in the analysis.

Differences in composition

We used the Multiple-Response Permutation Procedure (MRPP) and Indicator

Species Analysis of PC-ORD (McCune and Mefford 1999; McCune et al. 2002)
to test for differences in composition between the different vegetation units.
MRPP, a non-parametric procedure was used for testing the hypothesis that no
difference existed in composition between two or more groups of plots. For
distance in composition between the plots, Relative Sørensen (Bray–Curtis)
was used because it takes into account both compositionP (presence–absence of
species) and abundance. For weighting option: CI = nI/ nI was used, which
is the most widely used and recommended measure. CI is the weight and is
dependent on the number of items in a group, say I, and nI is the number of
items in group I. The software uses 9999 permutations in the test. Two tests
were carried out based on a priori selection: cover classes and fire classes. An
Indicator Species Analysis was also carried out on the basis of these two
classifications.
Indicator species Analysis combines information on the concentration of
species abundance in a particular group (transect) and the faithfulness
of occurrence of the species in that group. A perfect indicator species of a

[414]
 1475

particular group should always be present and should also be exclusive to that
group (not occurring in other groups). From the analysis, an indicator value is
obtained for each species in each group (Dufrêne and Legendre 1997; McCune
and Mefford 1999; McCune et al. 2002). The indicator values are tested for
statistical significance using a Monte Carlo randomization. Species diversity
was expressed as species dominance, which was calculated using the Simpson
Index (SI) (Magurran 1988), and Fisher’s a (Fa), (Fisher et al. 1943). These
indices have low sensitivity to plot size differences (Magurran 1988). Differ-
ences between plots in different fire and cover classes with respect to SI, Fa and
BA were tested with ANOVA using SPSS (SPSS 10, SPSS Inc. USA).
To check for variation in species abundance and diversity in relation to
disturbance, graphs of number of species per 100 m2 and Fisher’s a per plot
were made. Having the assumption, which was also backed by field observa-
tion, that disturbance was lowest in the forest class and highest in the wooded
grassland class, plots were arranged according to vegetation cover classes. The
order of plot arrangement was; forest (1–147), closed woodland (148–310),
open woodland (311–459), very open woodland (460–555) and wooded
grassland (556–592). Within each vegetation cover class the plots are randomly
ordered.

Results

Species distribution

A total of 26,076 individuals from 121 species, 89 genera, and 38 families were
recorded on the 591 plots. The most species-rich family was Moraceae with
11% of all species found (13), followed by Euphorbiaceae and Mimosaceae
with 8% each (10). The most species-rich genus was Ficus with 5% of all
species (6), followed by Acacia, Albizia, Celtis and Combretum with 3% each
(4). Nine species or 7% of all species could not be identified to genus level. A
full species list with abundances is given in Appendix 3. The most abundant
genus, in terms of total individuals encountered, was Combretum, with close to
16% of all individuals, followed by Terminalia (14%), Grewia (13), Stereo-
spermum (6%), and Uvariopsis (6%).
The DCA analysis on combined and trimmed data (491 plots and 45 species)
ordered the plots mainly along 1 axis (Figure 3a). This axis had a relatively
high eigenvalue (0.465) suggesting significant woody species variation along
this axis. The eigenvalue for the second axis was 0.172. With 491 plots in-
cluded, axis 1 explained 11.8% of the variation. There was, however, one
outlier plot strongly influencing the second axis. This outlier plot was domi-
nated by Sapium elipticum, a species that rarely occurred in the study area.
After removing this plot, axis 1 explained 12.5% of the variation and axis 2 an
additional 4.6%. Plots with a low axis score (close to 0) are found in the forest
area, plots with a high score (>7) are found in the most open areas. As most of

[415]
1476

[416]
 1477

the discussion here on will pertain to axis 1, the main gradient, we will
abbreviate ‘DCA axis 1 plot scores’ to ‘DCA scores.’
The species plot (Figure 3b) also shows most of the variation along the
first axis. The effect of the second axis is only evident close to zero along
axis 1, the forest side, where there appear to be two groups (the same can be
said for the plot scores). Based on this interpretation the species can be
divided into three groups; A, B and C (Figure 3b). Groups A and B occur
within the forest area and group C, probably starting at the forest edge,
stretches through to the woodland area. Species found in group A include
Cynometra alexandri, Diospyros abyssinica and Khaya anthotheca. Group B
species include Uvariopsis congensis, Celtis wightii, Holoptelea grandis
and Funtumia elastica. And species found in group C include Albizia
grandibracteata, Terminalia velutina, Grewia mollis, Combretum molle and
Lonchocarpus laxiflorus.
Fire indicator best explained the gradient in species composition followed by
slope and then distance from gate. Using the stepwise regression analysis, fire
alone had r2 of 0.324 with a standard error of 1.395. Including slope in the
model the r2 was raised to 0.354 and the standard error reduced to 0.365. When
distance from gate was included, the r2 increased to 0.359 and the standard
error was reduced to 1.361. Vegetation cover type was not significant and so it
does not appear in the results table. Relating the site variables individually to
DCA (results not shown) showed that while all the other variables had a
positive correlation with the DCA, distance from gate had a negative corre-
lation.

Species composition and vegetation indices

All the vegetation-indices explained well the DCA variation. TC-wetness and
TC-greenness showed the best relationship with DCA scores with r2 of 0.73
and 0.70, respectively. TC-brightness had the lowest value (r2 = 0.46). NDVI
had an r2 of 0.64.

Species distribution in discrete vegetation cover classes

The classes derived from the analyses of the satellite image differed consider-
ably in their DCA scores (Figure 4a and b). Plots of the ‘No-fire’ class had
consistently low DCA scores, whereas the plots from the class ‘Recent-fire’
have high DCA scores. Plots from the class ‘Old-fire’ were intermediate. The
Fire classes also differed considerably in their TC-greenness values. Conse-
quently a combination of DCA scores and TC-greenness value segregated the
fire classes well.
A similar result was found for the cover classes. These classes are segregated
both by their DCA scores and TC-greenness values (Figure 4b).

[417]
1478

Figure 4. DCA axis 1-Tasseled Cap relationship as subdivided by (a) fire regimes and (b)
vegetation cover classes.

Plots of different fire classes also differed significantly in their species

composition (MRPP, A = 0.061, p > 0.0001). Uvariopsis congensis, Celtis
wightii, Diospyros abyssinica, Phyllanthus discoideus, Celtis zenkeri, Alstonia
boonei, Cynometra alexandri and Trichilia prieuriana, all exclusively occur in
the No-fire class i.e. relative abundance (RA) equals 100% for each of the
species. The indicator species analysis also identified the above named species
as indicators for the No-fire class i.e. significant p values (Appendix 1). Al-
though no species had 100% relative frequency in any class, Terminalia velutina
and Grewia mollis had very high relative frequency, 90 and 87%, in Old-fire
and Recent-fire classes, respectively. For the Old-fire class, species that had
significant species indicator values include Terminalia velutina, Stereospermum
kunthianum and Piliostigma thoningii. And for the Recent-fire, species that had
significant species indicator values include Grewia mollis, Annona senegalensis,
Combretum molle, Loncocarpus laxiflorus and Grewia bicolor.
Plots of different cover classes also differed significantly in their species
composition (MRPP, A = 0.148, p > 0.0001). Of the species exclusively
found in the no fire area, Uvariopsis congensis, Celtis wightii, Celtis zenkeri,
Cynometra alexandri and Trichilia prieuriana were also exclusively found in the

[418]
 1479

forest area. In addition, Pterygota mildbreadii was also exclusively found in the
forest (Appendix 2). Funtumia elastica, Uvariopsis congensis and Celtis wightii
had the highest relative frequency in the forest class; 55, 54 and 50%,
respectively. Species with the highest relative frequency in the closed woodland
are Terminalia velutina and Grewia mollis with 97 and 70%, respectively. In the
open woodland plots, Terminalia velutina and Grewia mollis still had the
highest relative frequency of 85 and 88%, respectively. In the very open
woodland, Grewia mollis occurred in 99% of the plots while in the wooded
grassland, Stereospermum kunthianum had the highest relative frequency of
67%. Overall, Grewia mollis in the very open woodland had the highest relative
frequency i.e. it occurred in 99% of the closed woodland plots.
Whereas in the forest some of the species that had the highest relative fre-
quency are part of those that had the highest relative abundance, it is different
for the other cover classes. In the closed woodland, the species with the highest
relative abundance were Bridelia michrantha (70%), Albizia grandibracteata
(55%) and Maesopsis eminii (55%). In the open woodland there were no
species with relative abundance above 50%. The highest was Ficus exasperata
with 48%. In the very open woodland, Combretum molle, Securinega virosa and
Dombeya rotundifolia had the highest relative abundance with 74, 71 and 75%,
respectively. Combretum guenzi exclusively occurred in the wooded grassland.
Other species with high relative abundance in the wooded grassland
were Combretum binderanun, Grewia bicolor, Lonchocarpus laxiflorus and
Hymenocardia acida with 78, 66, 58 and 50%, respectively.
Most of the species identified as belonging to groups A and B e.g. Cynometra
alexandri, Khaya anthotheca, Diospyros abyssinica, Uvariopsis congensis and
Holoptelea grandis (Figure 3b) were also identified through indicator species
analysis as good indicators for the No-fire class. Of these, Diospyros abyssinica,
Uvariopsis congensis, Holoptelea grandis and all Celtis species were also good
indicators of the forest class (Appendix 2). The species in group C belonged
both to Old-fire and Recent-fire classes. Considering the cover classes,
Terminalia velutina and Albizia grandbracteata were good indicators for closed
woodland, Grewia mollis and Combretum mole for very open woodland and
Lonchocarpus laxiflorus, Grewia bicolor and Combretum guenzi were good
indicators for the wooded grassland class. Several of these species e.g.
Uvariopsis congensis, Terminalia velutina and Grewia mollis have distinctively
high abundance in specific areas along the gradient (Figure 5).
Although the closed woodland had the largest area sampled followed by the
open woodland, the forest had the highest number of species and genera
identified (Appendix 3). The lowest number of species and genera was found in
the wooded grassland. The highest ratio of species to genera was in very open
woodland (1.4) and the lowest in wooded grassland (1.2). Eleven species occur
in all classes and most species occur in more than one cover class but their
abundance varies greatly between classes. Forest and closed woodland classes
had an equal number of families and wooded grassland class had the lowest
number of families.

[419]
1480

Figure 5. Relationship between DCA axis 1 and some of the most abundant species whose
maximum abundance occur in different areas along the gradient. The selected species also display a
variation in their distribution range.

The Simpson index of all vegetation classes differed only slightly except that
of wooded grassland (Figure 6a). The wooded grassland had the highest value
and the highest standard error. The forest class had the highest mean Fisher’s a
(Figure 6b) followed by the closed woodland class. These two classes were
significantly different from all other classes but not from each other. The open
woodland was also significantly different from the wooded grassland. The
wooded grassland had the lowest Fisher’s a.
The basal area (Figure 6c) decreased from the forest, which had the highest
value, to the wooded grassland, which had the lowest. The forest also showed
the highest variation. All cover types were significantly different from each
other. The mean stem density values for the forest, closed woodland and open
woodland were very close (Figure 6d) and there was no significant difference
between them. The very open woodland also had a high mean value although
slightly lower than the other 3. The wooded grassland is much lower than all
others. The very open woodland and the wooded grassland are each signifi-
cantly different from all others. So while many individual trees may be found in
each cover type, they vary in size with the forest having larger trees than any of
the other cover types. Details of the species occurring in each cover type and
their abundance are indicated in Appendix 3.

Discussion

Variation in species composition along the gradient

The species composition along the gradient gradually changes from species
that attain maximum abundance in areas of minimum disturbance e.g.

[420]
 1481

Figure 6. comparison of cover class mean and standard deviation for (a) Simpson index, (b)
Fisher’s a, (c) basal area and (d) stem density. The class numbers consistently represent 1, forest; 2,
closed woodland; 3, open woodland; 4, very open woodland; and 5, wooded grassland. The letters
beside each bar indicate significance differences. Bars, for a specific variable, which have the same
letter mean that they are not significantly different (ANOVA: p = 0.05).

Cynometra alexandri and Uvariopsis congensis to species that attain maximum

abundance in areas with frequent disturbance e.g. Grewia mollis. On the other
hand, species like Terminalia velutina attain maximum abundance in the
moderately disturbed areas (Smart et al. 1985). Many species, as evidenced by
the species abundance plot (Figure 5), are wide ranging although they have a
clear optimum, which occurs at species specific locations along the gradient.
Identification of a vegetation type should, therefore, be based on species
abundance proportions rather than species incidence alone. This variation in
species tolerance range has also been observed in a Mexican dry forest
(Balvanera et al. 2002). In another study (Nangendo et al., submitted), it was
observed that the wide-ranging species often have their different development
sizes (seedlings, saplings and trees) in species specific locations along the
gradient.

[421]
1482

Relationship between site variables and the observed gradient

Of the environmental variables recorded, fire best explained the gradient. This
is evidenced by the high correlation between DCA and fire (Table 1) and the
fact that the compositional gradient could be divided using the fire regime
(Figure 4a). Areas that had recent fires, and are probably most frequently
burnt, had species that characteristically display fire resistant traits e.g. a thick
bark, pealing off of the old bark and good sprouting ability after a fire
(Gashaw et al. 2002; Saha and Howe 2003; Vesk and Westoby 2004). The
occurrence of some species is thus influenced by their fire-tolerance level
(Cauldwell and Zieger 2000) with increasingly more of the less fire resistant
species in the Old-fire class. Here, seed dispersal (a factor not explored in this
study) may have an important role. A number of the species that occurred in
the Old-fire class were most abundant in the No-fire class. Their seeds were
probably dispersed into the Old-fire class areas e.g. by wind and, when con-
ditions became favorable, they got established. Hence we suggest that the
existent fire regime influences their low occurrence (Huston 1994).
Although water is often a limiting factor for plant survival, in humid FWS
mosaics, water distribution is not a critical controlling factor (Favier et al.
2004). Despite the variation in rainfall over Budongo Forest Reserve, with the
northern part receiving less rain than the south (Plumptre 1996), the north still
receives over 1200 mm a year (Forest Department Uganda 1997) which is
sufficient for forest maintenance. Also elephants that previously restricted
forest expansion (Laws et al. 1975) are no longer present. The species turnover
could possibly be explained by an additive effect of the environmental variables
considered in this study, the historical impact by elephants and probably other
factors that were not considered in this study e.g. seed dispersal mechanisms,
which have been shown to favor establishment of species with higher dispersal
ability in the post disturbance period (Hovestadt et al. 1999; Ohsawa et al.
2002). However, just like in other studies where FWS occur (Elliott et al. 1999;
Hovestadt et al. 1999), fire plays a major role in controlling species distribution
pattern but it does not explain all the variation (Weiher 2003). Accessibility to
the protected areas, where local people mainly utilize areas closest to them
(Acharya 1999; Obiri et al. 2002), also showed a significant relationship with
the species composition gradient.

Vegetation variation and composition as mapped using satellite image

classification

The image classification provided a good representation of the vegetation

types. Each cover class had significant indicator species and differences in
structural and species diversity existed among the cover classes (Figure 6).
Although the best differentiating factor was basal area, where each cover class
was significantly different from the others, indicator species have also been

[422]
 1483

shown (Cousins and Lindborg 2004) to correspond well with the succession
gradient. Classification of mosaic areas using remotely sensed data could
therefore be a good start for identification of the vegetation types that exist
within them. This would require less time (Schmidt et al. 2004) as compared to
when only field surveys would have been used.
Our study has shown that although the forest significantly differed in species
diversity and vegetation structure, especially basal area, there was a systematic
decrease in variation from forest to wooded grassland (Figure 5). A major
gradient stretching from the forest to the wooded grassland is evident (Dezzeo
et al. 2004) and species composition and forest structure vary along this gra-
dient. Most of the areas sampled by Eggeling (1947) and followed up in Sheil
et al. (2000) had not had disturbance for a long time. Areas sampled in this
study, however, cover both areas with ranging times since last disturbance and
areas that are still experiencing frequent disturbance. Thus, in this study we
observe a wider range of vegetation variation.
Although subtle variations in vegetation structure may be evident in some
landscapes, the species composition variation is often more complex (Muh-
lenberg et al. 1990). In our study, the observed gradual change in species
composition along the gradient and the compositional interrelationship
between the vegetation cover classes indicate that the FWS mosaic is a single,
interacting, integrated unit.

The effect of continuing reforestation on the biodiversity of Budongo

Eggeling (1947), also followed up in Sheil (1999), identified successional stages

within the forest, with ironwood (Cynometra alexandri) at the climax end of the
spectrum and the colonizing (woodland) forest as the starting point. In their
study, the lowest woody plant diversity occurred in the ironwood forest. In our
study, although diversity is low at the ironwood end of the gradient, it is even
lower on the wooded grassland side of the gradient (plot results not shown)
indicating a drop on either end of the gradient. The highest diversity is within
the forest area and it gradually reduces until the lowest level, which occurs in
the wooded grassland. The colonizing forest, identified by Eggeling (1947) as
the starting point of the succession, occurs somewhere towards the middle of
the current gradient. The current study has, therefore, extended the succession
gradient to further into the wooded grassland and yet still conforms with the
Intermediate Disturbance Hypothesis (IDH) (Connell 1978).
Another DCA run, after combining a resampled set of Eggeling’s data with
data used in this study, revealed more of the similarities between the two
gradients. It, additionally, emphasized the existence of more than one succes-
sion path in the forest (Eggeling 1947; Sheil et al. 2000) and the variation
within the forest (Plumptre 1996). To incorporate Eggeling’s data, resampling
from the original data set was carried out. Having known the plot size and the
number of individuals collected from each of his plots, the number of

[423]
1484

individuals expected to occur in a 500 m2 plot was calculated. The calculated

number of individuals was then randomly sampled from the original individ-
uals of the respective plot. The abundance of each species in each plot was then
raised to that which would occur in an 1000 m2 plot. After crosschecking the
species names for possible changes in naming, the data were combined with
the rest of the tree data used in this study and a DCA was carried out. All
Eggeling’s plots, considering the first axis, occurred at one end of the gradient
but in line with the rest of the plots (Figure 7). Eggeling’s observation of
compositional convergence (Eggeling 1947) is still evident in his plots (see
dotted lines in Figure 7).
Plumptre (1996) identified a north–south compositional gradient. In our
study, the variation along the second axis of Figure 7 is an indicator of this
gradient. Eggelling’s plots collected from the southern part of the forest occur
separate from most of our plots, which were collected from the northern part
of the forest.
Succession always starts with very few species, then progresses awhile along
one line with more species coming in as conditions become more favorable
(Huston 1994). In our study, few species were observed in the wooded
grassland end of the gradient and species numbers increased as one moved
towards the forest (Figure 3b). Although the forest side of the gradient had
more species, other species occur away from the forest environment. The
diversity of an area is influenced by the type, frequency and intensity of the
disturbance (Trapnell 1959; Petraitis et al. 1989). Hence, if the whole suc-
cession gradient occurs in an area, there would be more species (Connell
1978; Huston 1994) than if one or a few stages of the succession gradient
were conserved. So while the areas that have high species numbers e.g. forest

Figure 7. DCA graph obtained after combining a resampled set of Eggeling’s data to the data
used in this paper. Axis 1 had an eigenvalue of 0.38 and explained 9.8% of the variation. The
second axis had an eigenvalue of 0.19 and explained 4.9%.

[424]
 1485

ought to be preserved (Sheil and Burslem 2003), the woodland areas should
not all be allowed to become forest since that would mean loosing the
woodland dependant species. And the highest number of species can only be
conserved when complementary areas are included in the conservation plan
(Howard et al. 1998). The maintenance of the high diversity of Budongo,
being an isolated forest with no immediate source of additional forest species,
may be more attributed to the existence of all stages of the succession gra-
dient (Richardson-Kageler 2004; Shea et al. 2004) than acquisition of more
forest species from elsewhere, which, additionally, often takes a long time
(Chapman et al. 1997). Hence, if reforestation of Budongo Forest Reserve
would continue to the extent that the woodland areas would be lost, the
biodiversity of the reserve would probably decrease. For purposes of con-
serving woody plants in a dynamic landscape, it is thus important that each
vegetation type represented is included and maintained within the conserva-
tion area (Bengtsson et al. 2003). In the area under study, fire disturbance is a
requirement for species coexistence (Shea et al. 2004).
In areas where fire may be applied, the vegetation type and its develop-
ment stage may affect the potential for ignition and spread of the fire
(Everett et al. 2000). Although no evidence exists of fires having destroyed
tropical rain forests in Uganda, it has been observed elsewhere that tropical
forests can burn (Cochrane and Schuize 1999; Cochrane and Laurance 2002;
Laurence 2003). This, however, mainly occurs in the presence of very dry
conditions, in fragmented forest landscapes and when fire is carelessly
applied in or adjacent to logged over areas. Fire also remains a highly de-
bated conservation management tool (Mentis and Bailey 1990; Trollope et al.
1995; van Wilgen et al. 1998). It is therefore important that fire be used
cautiously and, probably learning and using burning methods that have been
used in the past (Goma et al. 2001) will be a prerequisite. In this respect,
conservationists need to focus more attention on the current vegetation
management practices of local people surrounding conservation areas (Leone
and Lovreglio 2004) since they have been noted to use fire destructively
(Condit et al. 1998; Wheater 1971).
In Africa FWS mosaics are prevalent in areas surrounding the Congo basin
forests, including Uganda. These areas have been defined as transitional zones
between the moist tropical forest and the drier savanna landscape typical of
much of Africa. On the northern side, the transition occurs at about 8 N with
the exception of Togo and Benin and part of Ivory Coast (Gautier and Spi-
chiger 2004). Many FWS mosaics occur in Uganda because of its location in a
zone of overlap between the ecological communities characteristic of the dry
East African savannas and the West African rainforests (Howard 1991). The
observations made in this study and their management implications are,
therefore, relevant to many areas in Africa and in much of the tropical world
where such landscapes occur.

[425]
1486

Conclusions and recommendations

• Species composition in the area gradually changes from the forest to

savannah.
• Although many of the species occurred in more than one vegetation cover
class, each class had species that can be used to identify it. These are the
classes where such species had a significantly higher relative abundance as
compared to other classes.
• The gradient could be divided into sections using vegetation cover classes
and the fire indicator. These cover classes were compositionally separable
and vegetation structure significantly differed between the classes.
• Among the environmental variables, fire best explained the compositional
variation along the gradient.
Areas with such a dynamic FWS mosaic need a purposeful management that
takes into account the relationship between the observed vegetation pattern
and how this has been generated over time (Alados et al. 2004). Since each
vegetation cover class was compositionally separable from the others, a portion
of each of these classes needs to be conserved. An area where all cover classes
occur would be preferable since many species tend to occur in more than one
cover class and another study (Nangendo 2005) showed that the juveniles and
adults of some species do not occur in the same vegetation patches.
A well balanced management, including a controlled fire management sys-
tem that will prevent forest from colonizing the whole area yet allowing the
existence of varying disturbance regimes is a prerequisite for maintaining
species diversity (Crow and Perera 2004).

Acknowledgements

We thank Mr Hezekias Ddumba for being there for us to sort out our species
identification problems and for all the logistical support he provided while we
stayed at Kaniyo-Pabidi ecotourism camp. We express our gratitude to Mr Oli-
ver van Straaten who contributed part of the data used in this study. We also
express our gratitude to Professor Dr Alfred De Gier for his continued support
and for his valuable comments during the preparation of this manuscript.

Appendix 1. The Indicator Species Analysis output based on fire indicator classes.

Species names RA-0 RF-0 RA-1 RF-1 RA-2 RF-2 IV p Fire class

Terminalia velutina 39 66 52 90 9 47 46.7 0.001 1

Grewia mollis 14 47 37 80 49 87 42.4 0.001 2
Combretum collinum 23 37 39 56 38 49 21.7 0.046
Uvariopsis congensis 100 26 0 1 0 0 25.7 0.001 0
Annona senegalensis 18 33 35 59 47 55 25.7 0.005 2

[426]
 1487

Appendix 1. Continued.

Species names RA-0 RF-0 RA-1 RF-1 RA-2 RF-2 IV p Fire class
Albizia grandibracteata 64 41 31 22 5 6 26.2 0.001 0
Stereospermum kunthianum 22 28 44 50 34 47 21.9 0.003 1
Combretum molle 1 5 31 24 68 49 33.7 0.001 2
Lonchocarpus laxiflorus 10 12 19 22 71 45 31.7 0.001 2
Vitex doniana 38 33 36 36 26 25 12.8 0.585
Funtumia elastica 94 31 6 3 0 0 28.9 0.001 0
Lanea barteri 27 25 41 32 32 29 13.1 0.174
Celtis wightii 100 24 0 0 0 0 24.2 0.001 0
Acacia hockii 22 19 46 29 32 23 13.4 0.033
Piliostgma thonningii 25 18 53 31 23 15 16.2 0.004 1
Caloncoba schweinfurthii 86 23 14 6 0 0 20.2 0.001 0
Holoptelea grandis 99 22 1 1 0 0 21.6 0.001 0
Maesopsis eminii 62 21 38 12 0 0 12.8 0.003 0
Diospyros abyssinica 100 10 0 0 0 0 9.6 0.001 0
Ficus sur 43 15 48 15 9 3 7.1 0.245
Grewia bicolor 8 5 23 9 69 18 12.8 0.001 2
Khaya anthotheca 86 15 14 2 0 0 12.6 0.001 0
Dombeya mukole 79 8 17 3 5 1 6.4 0.013 0
Bridelia micrantha 87 12 9 1 4 1 10.7 0.001 0
Celtis durandii 86 12 14 2 0 0 9.9 0.003 0
Combretum binderanum 2 0 37 6 61 9 5.6 0.012 2
Margaritaria discoidea 71 11 29 2 0 0 7.6 0.004 0
Phyllanthus discoideus 100 10 0 0 0 0 9.6 0.001 0
Albizia zygia 56 2 20 2 24 3 1.3 0.829
Celtis zenkeri 100 10 0 0 0 0 10.4 0.001 0
Pterygota mildbreadii 70 5 30 1 0 0 3.8 0.059
Hymenocardia acida 11 3 27 5 62 9 5.7 0.008 2
Olea welwitschii 95 9 5 1 0 0 8.8 0.001 0
Oncoba spinosa 37 5 14 1 49 6 2.8 0.326
Tapura fisheri 80 9 20 2 0 0 7.1 0.009 0
Securinega virosa 0 0 38 5 62 10 6.4 0.004 2
Dichrostachys cinerea 73 5 27 1 0 0 3.9 0.064
Alstonia boonei 100 9 0 0 0 0 8.8 0.001 0
Cynometra alexandri 100 5 0 0 0 0 5 0.007 0
Ficus exasperata 37 3 63 3 0 0 2.2 0.272
Combretum gueinzii 0 0 73 1 27 1 0.5 0.749
Sapium ellipticum 81 2 19 1 0 0 1.2 0.403
Carpololobia alba 0 0 21 3 79 6 4.5 0.006 2
Dombeya rotundifolia 8 2 25 5 67 8 5.4 0.012 2
Trichilia prieuriana 100 5 0 0 0 0 5.4 0.007 0

It indicates the concentration of each species in each class (Relative abundance, RA), the faith-
fulness of occurrence of the species in that class (Relative frequency, RF), the highest species
indicator value across the classes (IV) the statistical significance of the indicator value (p) and the
class in which a particular species had the highest indicator value (Fire class). For species that were
not significant indicators for any class, fire class was left blank. RA is expressed as a proportion of a
particular species in a particular class relative to its abundance in other classes. RF is expressed as
the percentage of sample units in a class that contain that species.
p is significant at 0.01.
0, No fire; 1, old fire; and 2, recent fire.

[427]
1488

Appendix 2. The Indicator Species Analysis output based on vegetation cover classes.

Species RA-1 RF-1 RA-2 RF-2 RA-3 RF-3 RA-4 RF-4 RA-5 RF-5 IV P Cover
cord

Terminalia 12 30 44 97 31 85 8 57 5 44 42.8 0.001 2

velutina
Grewia mollis 2 16 16 70 32 88 48 99 1 11 47 0.001 4
Combretum 11 18 20 52 38 64 22 43 10 22 24.3 0.042
collinum
Uvariopsis 100 54 0 0 0 0 0 0 0 0 54.4 0.001 1
congensis
Annona 6 14 23 52 30 58 42 64 0 0 26.9 0.02
senegalensis
Albizia 28 26 55 60 14 15 3 3 0 0 32.9 0.002 2
grandibracteata
Stereospermum 7 11 26 51 22 45 17 39 29 67 19.2 0.063
kunthianum
Combretum 0 2 2 6 15 23 74 66 8 22 48.4 0.001 4
molle
Lonchocarpus 0 0 2 10 14 36 26 47 58 56 32 0.001 5
laxiflorus
Vitex doniana 5 11 33 51 23 36 16 24 23 22 16.8 0.095
Funtumia 89 55 11 11 0 0 0 0 0 0 49.4 0.001 1
elastica
Lanea barteri 2 3 23 30 35 40 34 43 6 11 14.7 0.095
Celtis wightii 100 50 0 1 0 0 0 0 0 0 49.4 0.001 1
Acacia hockii 7 10 16 22 31 37 17 21 29 11 11.6 0.15
Piliostgma 5 8 18 24 22 31 10 16 45 44 19.8 0.024
thonningii
Caloncoba 74 38 24 13 2 2 0 0 0 0 27.9 0.005 1
schweinfurthii
Holoptelea 94 42 6 3 0 0 0 0 0 0 39.8 0.001 1
grandis
Maesopsis 45 23 55 28 0 0 0 0 0 0 15.7 0.033
eminii
Diospyros 96 18 3 1 1 1 0 0 0 0 17.7 0.005 1
abyssinica
Ficus sur 18 10 43 20 31 14 8 4 0 0 8.8 0.133
Grewia bicolor 0 0 3 3 11 13 20 20 66 56 36.9 0.001 5
Khaya 71 20 23 10 6 1 0 0 0 0 14.2 0.03
anthotheca
Dombeya 82 14 5 2 11 4 3 1 0 0 11.1 0.04
mukole
Bridelia 26 7 70 16 2 1 3 1 0 0 11.3 0.046
micrantha
Celtis durandii 93 23 7 3 0 0 0 0 0 0 21.5 0.005 1
Combretum 0 0 1 1 9 5 11 10 78 22 17.4 0.005 5
binderanum
Margaritaria 61 14 39 9 0 0 0 0 0 0 8.3 0.078
discoidea
Phyllanthus 97 18 3 1 0 0 0 0 0 0 17.8 0.011 1
discoideus
Albizia zygia 67 3 12 2 7 1 15 4 0 0 2.1 0.469

[428]
 1489

Appendix 2. Continued.

Species RA-1 RF-1 RA-2 RF-2 RA-3 RF-3 RA-4 RF-4 RA-5 RF-5 IV P Cover
cord
Celtis zenkeri 100 22 0 0 0 0 0 0 0 0 21.6 0.006 1
Pterygota 100 13 0 0 0 0 0 0 0 0 12.8 0.025
mildbreadii
Hymenocardia 0 0 3 2 20 9 27 7 50 22 11.2 0.045
acida
Olea 92 17 8 3 0 0 0 0 0 0 15.4 0.016
welwitschii
Oncoba 48 9 6 1 12 3 33 6 0 0 4.2 0.166
spinosa
Tapura fisheri 90 18 10 2 0 0 0 0 0 0 16.6 0.014 1
Securinega 0 0 8 1 21 3 71 14 0 0 10.1 0.03
virosa
Dichrostachys 63 6 18 4 12 1 8 1 0 0 4 0.138
cinerea
Alstonia 86 14 14 3 0 0 0 0 0 0 12.4 0.033
boonei
Cynometra 100 10 0 0 0 0 0 0 0 0 10.4 0.022
alexandri
Ficus 6 1 46 5 48 4 0 0 0 0 2.2 0.454
exasperata
Combretum 0 0 0 0 0 0 0 0 100 22 22.2 0.001 5
gueinzii
Sapium 82 2 18 1 0 0 0 0 0 0 2 0.174
ellipticum
Carpololobia 0 0 32 1 38 4 30 4 0 0 1.4 0.607
alba
Dombeya 0 0 8 2 17 4 75 13 0 0 9.7 0.04
rotundifolia
Trichilia 100 11 0 0 0 0 0 0 0 0 11.2 0.03
prieuriana

It indicates the concentration of each species in each class (Relative abundance, RA), the faithfulness of
occurrence of the species in that class (Relative frequency, RF), the highest species indicator value across the
classes (IV) the statistical significance of the indicator value (p) and the class in which a particular species had the
highest indicator value (Cover cord). For species that were not significant indicators for any class, cover cord was
left blank. RA is expressed as a proportion of a particular species in a particular class relative to its abundance in
other classes. RF is expressed as the percentage of sample units in a class that contain that species.
p is significant at 0.01.
1, Forest; 2, closed woodland; 3, open woodland; 4, very open woodland; and 5, wooded grassland.

Appendix 3. The 121woody species identified in the field, their abundance per class.

Number of individuals

Family Species Fo cw Ow Vow Wg

Mimosaceae Acacia hockii De Wild. 20 55 97 25 7
Mimosaceae Acacia seyal Delile . . . 1 .

[429]
1490

Appendix 3. Continued.

Number of individuals
Mimosaceae Acacia sieberiana Dc. Var. 1 6 . 3 .
woodii (Burtt Davy)
Keay & Brenan
Mimosaceae Acacia spp. 2 . . . .
Euphorbiaceae Acalypha neptunica Müll. 9 . . . .
Arg. Var.
Mimosaceae Albizia coriaria Oliver 9 8 2 . .
Mimosaceae Albizia grandibracteata Taub. 120 257 57 7 .
Mimosaceae Albizia spp. . . 1 . .
Mimosaceae Albizia zygia (DC.) Macbr. 33 5 3 3 .
Apocynaceae Alstonia boonei de Wild 24 5 . . .
Sapotaceae Aningeria altissima (A. Chev.) 15 2 1 2 .
Aubr. & Pellegr.
Annonaceae Annona senegalensis Pers. 35 157 194 129 1
Balemetea gramofolia 1 . . . .
Rubiaceae Belonophora glomerata 2 1 . . .
Sapindaceae Blighia unijugata Baker 2 6 . . .
Euphorbiaceae Bridelia micrantha (Hochst.) Baill. 15 49 1 1 .
Euphorbiaceae Bridelia scleroneuroides Pax. . 2 5 2 .
Elacourtiacea Caloncoba schweinfurthii Glig. 107 42 3 . .
Polygalaceae Carpololobia alba G. Don . 9 12 4 .
Caesalpinioideae Cassia siamea Lam. 2 13 . . .
Caesalpinioideae Cassia spp. . 1 . . .
Ulmaceae Celtis durandii Engl. 59 6 . . .
Ulmaceae Celtis mildbraedii Engl. 7 . . . .
Ulmaceae Celtis wightii Planch. 204 1 . . .
Ulmaceae Celtis zenkeri Engl. 43 . . . .
Moraceae Chlorophora excelsa (Welw.) Benth . 3 . . .
Sapotaceae Chrysophyllum albidum G. Don 6 . . . .
Rutaceae Citropsis articulata (Wild. Ex Spreng) 1 . . . .
Swingle & M. Kellerm
Annonaceae Cleistopholis patens (Beth.) Engl. & Diels 1 . . . .
Closophila magida 1 . . . .
Rubiaceae Coffea canephora Pierre ex Froechner. 5 . . . .
Rubiaceae Coffea euginiodes 5 . . . .
Sterculiaceae Cola gigantea A. Chev. 13 . . . .
Combretaceae Combretum binderanum Kotschy . 3 16 14 17
Combretaceae Combretum collinum Fresen. 107 240 431 137 12
Combretaceae Combretum gueinzii Sond. . . 2 . 21
Combretaceae Combretum molle R. Br. Ex G. Don 4 18 92 243 6
Boraginaceae Cordia millenii Baker 7 7 . . .
Aralliaceae Cussonia arborea Hochst. Ex A. Rich. . 1 4 12 .
Caesalpiniaceae Cynometra alexandri CH Wright 29 . . . .
Mimosaceae Dichrostachys cinerea (L.) Wright & Arn 19 8 3 1 .
Ebenaceae Diospyros abyssinica (Hiern) F. White 90 3 1 . .
Sterculiaceae Dombeya mukole Sprague 56 3 6 1 .
Sterculiaceae Dombeya rotundifolia (Hochst.) Planch. 2 3 6 12 .
Mimosaceae Entada abyssinica Steud. Ex A. Rich . . 1 . .
Meliaceae Entandrophragma angolense (Welw.) C. DC. 3 . . . .

[430]
 1491

Appendix 3. Continued.

Number of individuals
Meliaceae Entandrophragma cylindricum 1 . . . .
(Sprague) Sprague
Papilionaceae Erythrina abyssinica Lam. Ex DC 1 4 1 3 .
Leguminosae Erythrophleum suaveolens 8 . . . .
(Guill. & Perr.)Brenan
Rutaceae Fagaropsis angolensis (Engl.) 5 1 . . .
HM. Gardner
Moraceae Ficus capensis Thunb 1 1 . . .
Moraceae Ficus casuarina . 1 . . .
Moraceae Ficus exasperata Vahl 1 10 8 . 5
Moraceae Ficus mucuso Welw ex Ficalho 1 5 1 . .
Moraceae Ficus polita Vahl 9 5 1 . .
Moraceae Ficus saussureana DC. . 1 . . .
Moraceae Ficus spp. 1 . . . .
Moraceae Ficus sansibarica Warb. 1 . . . .
Moraceae Ficus sur Forssk 17 41 26 3 .
Apocynaceae Funtumia elastica (Preuss) Stapf 227 32 . . .
Rubiaceae Gardenia Jovis-tonantis (Welw.) Hiern. . 1 1 1 .
Tiliaceae Grewia bicolor Juss. . 10 27 26 17
Tiliaceae Grewia mollis Juss. 58 424 761 531 6
Simaroubaceae Harrisonia abyssinica Oliv. . 1 . . .
Ulmaceae Holoptelea grandis (Hutch.) Mildbr. 99 10 . . .
Euphorbiaceae Hymenocardia acida Tul. 3 3 22 15 3
Meliaceae Khaya anthotheca (Welw.) C. DC. 50 16 4 . .
Meliaceae Khaya grandifolia C. DC. 1 . . . .
Bignoniaceae Kigeria africana (Lam.) Benth . . 5 1 .
Anacardiaceae Lanea barteri (Oliv.) Engl. 5 78 107 48 2
Anacardiaceae Lannea welwitschii (Hiern.) Engl. . 3 . 1 .
Rhamnaceae Lasiodiscus mildbraedii Engl. 1 . . . .
Sapindaceae Lepisanthes senegalensis (Juss. Ex Poir.) 6 5 . . .
Papilionaceae Lonchocarpus laxiflorus Guill. & Perr. . 20 116 108 34
Capparidaceae Maerua duchensii 12 . . . .
Rhamnaceae Maesopsis eminii Engl. 47 61 . . .
Meliaceae Mahogany spp. 8 . . . .
Euphorbiaceae Margaritaria discoidea (Baill.) Webster 29 21 . . .
Rignoniaceae Markhamia platycalyx (Baker) Sprague 2 1 . . .
Celastraceae Maytenus undata (Thunb.) Blakelock . 1 4 7 4
Papilionaceae Mildbraediodendron excelsum (Harms) 3 . . . .
Moraceae Milicia excelsa (Welw.) CC Berg 2 2 . . .
Rubiaceae Mitragyna stipulosa (DC.) O. Ktze 1 . . . .
Moraceae Morus lactea (Sim) Mildbr. . 1 . . .
Moraceae Myrianthus holstii Engl. 4 . . . .
Oleaceae Olea welwitschii (Knobl.) Gilg & Schellenb. 36 4 . . .
Flacourtiaceae Oncoba spinosa Forsk. 24 3 5 8 .
Palmae Phoenix reclinata Jacq. 12 . . . .
Euphorbiaceae Phyllanthus discoideus Muell. 46 2 . . .
Caesalpiniaceae Piliostgma thonningii (Schum.) 21 66 75 18 12
Verbenaceae Premna angolensis Guerke 18 2 . . .
Proteaceae Protea madiensis Oliv. . . . 9 .
Anacardiaceae Pseudospondias microcarpa (A. Rich.) Engl. 4 2 . . .

[431]
1492

Appendix 3. Continued.

Number of individuals
Sterculiaceae Pterygota mildbraedii Engl. 40 2 . . .
Euphorbiaceae Ricinodendron excelsum 8 2 . . .
Violaceae Rinorea dentata (P. Beauv.) Kuntze 1 . . . .
Violaceae Rinorea ilicifolia (Welw. Ex Oliv.) 13 . . . .
Capparidaceae Ritchiea albersii Gilg 1 . . . .
Rubiaceae Rothmannia urcelliformis (Hiern) . . 4 5 6
Bullock exRobyns
Celestraceae Salacia elegans Welw. Ex Oliv. . . 3 2 5
Euphorbiaceae Sapium ellipticum Pax. 20 1 . . .
Oleaceae Schrebera arborea A. Chev. 9 . . . .
Polygalaceae Securidaca spp. . 4 10 19 .
Euphorbiaceae Securinega virosa (Roxb. Ex Willd.) Baill 2 3 5 . .
Bignoniaceae Spathodea campanulata P. Beauv. 4 12 . . .
Umbelliferae Steganotaenia araliacea Hochst. 2 2 2 9 .
Bignoniaceae Stereospermum kunthianum Cham. 41 162 126 55 18
Apocynaceae Tabernaemontana holstii K. Schum 13 1 . . .
Chailletiaceae Tapura fisheri 32 6 . . .
Rutaceae Teclea nobilis Del. 11 10 2 . .
Combretaceae Terminalia velutina Rolfe 308 1288 824 103 14
Euphorbiaceae Thecacoris lucida . . . 4 .
Ulmaceae Trema orientalis (L.) Blume 1 2 . . .
Meliaceae Trichilia prieuriana A. Juss 23 . . . .
Meliaceae Trichilia spp. 3 . . . .
Meliaceae Turrae floribunda 1 . 1 . .
Annonaceae Uvariopsis congensis Robyns & Ghesq. 663 . . . .
Rubiaceae Vangueria apiculata K. Schum . 2 . . .
Compositae Vernonia amygdalina Delile 1 2 . . .
Verbenaceae Vitex doniana Sweet. 20 143 92 29 6
Rhamnaceae Zizyphus abyssinica Hochst. Ex A. Rich 1 . 1 . .
Total individuals 3042 3394 3172 1602 196
Total species 95 77 48 39 18
Total genera 73 60 35 28 15
Total families 33 33 23 19 13
Total area (sq. m) 65300 71000 65500 42100 16500

References

Acharya B. 1999. Forest biodiversity assessment: a spatial analysis of tree species diversity in
Nepal. Ph.D. Thesis, Leiden University, Enschede, 199 pp.
Alados C.L. et al. 2004. Variations in landscape patterns and vegetation cover between 1957 and
1994 in a semiarid mediterranean ecosystem. Landscape Ecol. 19: 543–559.
Balvanera P., Lott E., Segura G., Siebe C. and Islas A. 2002. Patterns of B-diversity in a Mexican
tropical dry forest. J. Vege. Sci. 13(2): 145–158.
Bengtsson J. et al. 2003. Reserves, resilience and dynamic landscapes. Ambio 32(6): 389–396.
Buechner H.K. and Dawkins H.C. 1961. Vegetation change induced by elephants and fire in the
Murchison falls national park, Uganda. J. Ecol. 42(4): 752–766.

[432]
 1493

Cauldwell A.E. and Zieger U. 2000. A reassessment of the fire-tolerance of some miombo woody
species in the Central Province, Zambia. Afr. J. Ecol. 38: 138–146.
Chapman C.A., Chapman L.J., Wrangham R., Isabirye-Basuta G. and Ben-David K. 1997. Spatial
and temporal variability in the structure of a tropical forest. Afr. J. Ecol. 35: 287–302.
Cochrane M.A. and Laurance W.F. 2002. Fire as a large-scale edge effect in Amazonian forests.
J. Trop. Ecol. 18: 311–325.
Cochrane M.A. and Schuize M.D. 1999. Fire as a recurrent event in tropical forests of eastern
Amszon: effects on forest structure, biomass, and species composition. Biotropica 31: 2–16.
Collins B.J. and Woodcock E.C. 1996. An assessment of several linear change detection techniques
for mapping forest mortality using multitemporal Landsat TM data. Remote Sens. Environ.
56: 66–77.
Condit R., Sukumar R., Hubbell S.P. and Foster R.B. 1998. Predicting population trends from size
distributions: a direst test in a tropical tree community. Am. Natural. 152(4): 495–509.
Connell J.H. 1978. Diversity in tropical rain forests and coral reefs. Science 199: 1302–1310.
Cousins S.A.O. and Lindborg R. 2004. Assessing changes inplant distribution patterns-indicator
species versus plant functional types. Ecol. Indicators 4: 17–27.
Crist E.P. and Cicone R.C. 1984. Application of the Tasseled Cap concept to simulated Thematic
Mapper data. Photogramm. Eng. Remote Sens. 50: 343–352.
Crist E.P., Laurin R. and Cicone R.C. 1986. Vegetation and soils information contained in
transformed Thematic Mapper data. Proceedings of the IGARSS ’86 Symposium, Zurich,
Switzerland, ESA, Paris, pp. 1465–1470.
Crow T.R. and Perera A.H. 2004. Emulating natural landscape disturbance in forest management –
an introduction. Landscape Ecol. 19: 231–233.
Dezzeo N., Chacón N., Sanoja E. and Picón G. 2004. Changes in soil properties and vegetation
characteristics along a forest–savanna gradient in southern Venezuela. Forest Ecol. Manag.
200: 183–193.
Dufrêne M. and Legendre P. 1997. Species assemblages and indicator species: the need for a flexible
asymmetrical approach. Ecol. Monogr. 67(3): 345–366.
Eggeling W.J. 1947. Observations on the ecology of the Budongo rain forest, Uganda. J. Ecol.
34(1): 20–87.
Eggeling W.J. and Dale R.I. (eds) 1952. The Indigenous Trees of the Uganda Protectorate. Uganda
Government printer, Entebbe.
Eilu G., Hafashimana D.L.N. and Kasenene J.M. 2004. Tree species distribution in forests of the
Albertine Rift, Western Uganda. Afr. J. Ecol. 42: 100–110.
Elliott K.J., Hendrick R.L., Major A.E., Vose J.M. and Swank W.T. 1999. Vegetation dynamics
after a prescribed fire in the southern Appalachians. Forest Ecol. Manag. 114: 199–213.
Everett R.L., Schellhaas R., Keenum D., Spurbeck D. and Ohlson P. 2000. Fire history in the
ponderosa pine/Douglas-fir forests on the east slope of the Washington Cascades. Forest Ecol.
Manag. 2000: 207–225.
Favier C., Chave J., AFabing A., Schwartz D. and Dubois M.A. 2004. Modelling forest–savanna
mosaic dynamics in man-influenced environments: effects of fire, climate and soil heterogeneity.
Ecol. Model. 171: 85–102.
Fisher R.A., Corbet A.S. and Williams C.B. 1943. The relation between the number of species and
the number of individuals in a random sample of an animal population. J. Anim. Ecol. 12: 42–58.
Forest Department Uganda 1997. Forest management plan for Budongo Forest Reserve for 1997
to 2007. Forest Department, Kampala, 107 pp.
Forest Department Uganda 1999. Forestry Nature Conservation Master Plan, Ministry of Water.
Lands and Environment, Kampala.
Gashaw M. et al. 2002. Post fire regeneration strategies and tree bark resistance to heating in
frequently burning tropical savanna woodlands in Ethiopia. Nord. J. Bot. 22(1): 19–33.
Gautier L. and Spichiger R. 2004. The forest–savanna transition in West Africa. In: Poorter L.,
Bongers F., Kouamé F.N. and Hawthorne W.D. (eds), Biodiversity of West African Forests: An
Ecological Atlas of Woody Plant Species. CAB International, Wallingford, pp. 33–40.

[433]
1494

Goma H.C., Rahim K., Nangendo G., Riley J. and Stein A. 2001. Participatory studies for agro-
ecosystem evaluation. Agric. Ecosyst. Environ. 87: 179–190.
Hamilton A. 1991. A Field Guide to Ugandan Forest Trees. Makerere University, Kampala.
Hovestadt T., Yao P. and Linsenmair E.K. 1999. Seed dispersal mechanisms and the vegetation of
forest islands in a West African forest savanna mosaic (Comoé National Park, Ivory Coast).
Plant Ecol. 144: 1–25.
Howard P.C. 1991. Nature Conservation in Uganda Tropical Forest Reserves. IUCN, Gland.
Howard P.C. et al. 1998. Complementarity and the use of indicator groups for reserve selection in
Uganda. Nature 394: 472–475.
Huston M.A. 1994. Biological Diversity. The Coexistence of Species on Changing Landscapes.
Cambridge University Press, London.
Laurence W.F. 2003. Slow burn: the insidious effects of surface fires on tropical forests. Trends
Ecol. Evol. 18(5): 209–212.
Laws R.M., Parker I.S.C. and Johnstone R.C.B. 1975. Elephants and Their Habitats. The Ecology
of Elephants in North Bunyoro, Uganda. Oxford University Press, London.
Leone V. and Lovreglio R. 2004. Conservation of Mediterranean pine woodlands: scenarios and
legislative tools. Plant Ecol. 171: 221–235.
Li J., Loneragan W.A., Duggin J.A. and Grant C.D. 2004. Issues affecting the measurement of
disturbance response patterns in herbaceous vegetation – a test of the intermediate disturbance
hypothesis. Plant Ecol. 172: 11–26.
Magurran A.E. 1988. Ecological Diversity and its Measurement. Princeton University Press,
Princeton, New Jersey, 179 pp.
McCune B., Grace J.B. and Urban D.L. 2002. Analysis of the Ecological Communities. MjM
Software, Gleneden Beach, Oregon, US.
McCune B. and Mefford M.J. 1999. Multivariate Analysis of Ecological Data. Version 4.25. MjM
Software, Gleneden Beach, Oregon, US.
Mentis M.T. and Bailey A.W. 1990. Changing perceptions of fire management in savanna parks.
J. Grassland Soc., Southern Africa 7: 81–85.
Muhlenberg M., Galat-Luong A., Poilecot P., Steinhauer-Burkart B. and Kuhn I. 1990. The
importance of forest islands within wet savannas for the conversation of rainforest animals in the
Ivory Coast. Revue d’Ecologie (Terre et la Vie) 45(3): 197–214.
Mwami P.M. and McNeilage A. 2003. Natural regeneration and ecological recovery in Bwindi
Impenetrable National Park, Uganda. Afr. J. Ecol. 41: 93–98.
Nangendo G. 2005. Changing forest–woodland–savanna mosaics in Uganda: with implications for
conservation. Ph.D. Thesis, Wageningen University, Wageningen, 139 pp.
Nangendo G., Skidmore A.K. and van Oosten H. submitted. Mapping East African tropical forests
and woodlands: a comparison of classifiers. ISPRS J. Photogramm. Remote Sensing.
Nangendo G., van Straaten O.ter Steege H., de Gier A. and Bongers F. submitted. Vegetation
cover change and its relationship with species distribution dynamics. J. Biogeogr..
Obiri J., Lawes J. and Mukolwe M. 2002. The dynamics and sustainable use of high-value tree
species of the coastal Pondoland forests of the Eastern Cape Province, South Africa. Forest Ecol.
Manag. 166: 131–148.
Ohsawa K., Kawasaki K., Takasu F. and Shigesada N. 2002. Recurrent habitat distribution and
species diversity in a multiple-competitive species system. J. Theor. Biol. 216: 123–138.
Paterson D.J. 1991. The ecology and history of Uganda’s Budongo forest. Forest Conserv. Hist.
35: 179–186.
Petraitis P.S., Latham R.E. and Niesenbaum R.A. 1989. The maintenance of species diversity by
disturbance. Quarter. Rev. Biol. 64: 393–418.
Plumptre A.J. 1996. Changes following 60 years of selective timber harvesting in the Budongo
Forest Reserve, Uganda. Forest Ecol. Manag. 89: 101–113.
Richardson-Kageler S.J. 2004. Effects of large herbivore browsing on the functional groups of
woody plants in a southern African savanna. Biodivers. Conserv. 13: 2145–2163.

[434]
 1495

Saha S. and Howe H.F. 2003. Species composition and fire in a dry deciduous forest. Ecology
84(12): 3118–3123.
Schmidt S.K. et al. 2004. Mapping costal vegetation using an Expert System and Hyperspectral
imagery. Photogramm. Eng. Remote Sens. (in press).
Schwartz M.W. and Caro T.M. 2003. Effect of selective logging on tree and understory regener-
ation in Miombo woodland in western Tanzania. Afr. J. Ecol. 41: 75–82.
Schwilk D.W., Keeley J.E. and Bond W.J. 1997. The intermediate disturbance hypothesis does not
explain fire and diversity pattern in fynbos. Plant Ecol. 132: 77–84.
Shea K., Roxburgh S.H. and Rauschert E.S.J. 2004. Moving form pattern to process: coexistence
mechanisms under intermediate disturbance regimes. Ecol. Lett. 7: 491–508.
Sheil D. 1999. Developing tests of successional hypotheses with size-structured populations, and an
assessment using long-term data from a Ugandan rain forest. Plant Ecol. 140: 117–127.
Sheil D. and Burslem F.R.P.D. 2003. Disturbing hypothesis in tropical forests. Trends Ecol. Evol.
18(1): 18–26.
Sheil D., Jennings S. and Savill P. 2000. Long-term permanent plot observations of vegetation
dynamics in Budongo, a Ugandan rain forest. J. Trop. Ecol. 16: 765–800.
Smart N.O.E., Hatton J.C. and Spence D.H.N. 1985. The effect of long-term exclusion of large
herbivores on vegetation in Murchison Falls National Park, Uganda. Biol. Conserv. 33: 229–245.
Swaine M.D., Hawthorne W.D. and Orgle T.K. 1992. The effects of fire exclusion on savana
vegetation at Kpong, Ghana. Biotropica 24: 166–172.
Trapnell C.G. 1959. Ecological results of woodland burning experiments in Northern Rhodesia.
J. Ecol. 47: 129–168.
Trollope W.S.W., Biggs H.C., Potgieter A.L.F. and Zambatis N. 1995. A structured versus a
wilderness approach to burning in the Kruger National Park in South Africa. In: West N.E.
(ed.), Rangelands in a Sustainable Biosphere. Society for Rangeland Management, Denver.
Tucker J.C. 1979. Red and Photographic Infrared linear combinations for monitoring vegetation.
Remote Sens. Environ. 8: 127–150.
van Straaten H.P. 1976. Präkambrium und junges Western Rift im Bunyoro District NW-Uganda.
Unpublished Ph.D., Universität Göttingen, Hanover.
van Straaten O. 2003. Changing woodland systems: post-disturbance woody species succession
dynamics and spatial trends. M.Sc. Thesis, International Institute for Geoinformation Science
and Earth Observation, Enschede, 70 pp.
van Wilgen B.W., Biggs H.C. and Potgieter A.L.F. 1998. Fire management and research in the
Kruger National Park, with suggestion on the detection of thresholds of potential concern.
Koedoe 41: 69–87.
Vesk P.A. and Westoby M. 2004. Sprouting ability across diverse disturbances and vegetation types
worldwide. J. Ecol. 92(2): 310–320.
Walter H. 1985. Vegetation of the Earth and Ecological Systems of the Geo-Biospere. Springer-
Verlag, 318 pp.
Weiher E. 2003. Species richness along multiple gradients: testing a general multivariate model in
oak savanna. OIKOS 101: 311–316.
Wheater R.J. 1971. Problems of controlling fires in Uganda National parks, Annual Tall Timbers
Fire Ecology Conference.

[435]
Biodiversity and Conservation (2006) 15:1497–1508
Springer 2006
DOI 10.1007/s10531-005-2356-1

Do Orthoptera need human land use in Central

Europe? The role of habitat patch size and linear
corridors in the Białowie_za Forest, Poland

JÖRN THEUERKAUF1,*, SOPHIE ROUYS2

1
Museum and Institute of Zoology, Polish Academy of Sciences, Wilcza 64, 00-679 Warsaw, Poland;
2
Marine Laboratory for Environmental Resource Studies, University of New Caledonia, BP 4477,
98847 Nouméa Cedex, New Caledonia; *Author for correspondence: BP 2549, 98846 Nouméa
Cedex, New Caledonia (e-mail: jtheuer@miiz. waw.pl; fax: +687-254473)

Received 13 January 2005; accepted in revised form 8 August 2005

Key words: Blattodea, Dermaptera, Habitat size, Road corridors, Herbivores, Marsh corridors,
Orthoptera

Abstract. We studied Orthoptera, Dermaptera, and Blattodea of the Białowie_za Forest (Poland) in
order to assess (1) the minimum patch size of open habitat necessary for each species, (2) the role of
linear corridors as habitat, and (3) the impact of herbivores on diversity by comparing the fauna at
periods of different ungulate densities. Many species occurred in the farthest clearings from the
forest edge to arable land. Two third of species occurred in clearings smaller than 10,000 m2. Dry
linear corridors of 10–40 m width and wet linear corridors of 100–200 m width had a species
richness that corresponded to that of clearings of about 10,000 m2. Four species disappeared from
the Białowie_za Forest when ungulate density decreased from 20 individuals/km2 (3000 kg/km2
biomass) at the beginning of the 20th century to 10 individuals/km2 (1000 kg/km2) at the end of the
20th century. We conclude that most Orthoptera, Dermaptera, and Blattodea species could survive
in Central Europe if human land use was replaced by intensive grazing and browsing by wild
herbivores.

Introduction

Closed forest has long been regarded as the natural vegetation of most parts
of Central Europe (Birks 2005). As a result, it is widely believed that there
was little space available to species associated with open landscape until
humans cleared the forest and created meadows and heathland. This image of
closed forest as the natural vegetation has been discussed over the last decade
(e.g. Svenning 2002). Vera (2000) argued that natural forests in the lowlands
of Central Europe were rather park-like landscapes that were shaped and
maintained by herbivores. This ‘‘wood-pasture’’ hypothesis is in opposition
to the ‘‘high-forest" hypothesis (Bradshaw et al. 2003), which has been
favoured throughout the 20th century and that considers closed forest as the
climax for Central Europe but does not acknowledge the major influence of
herbivores. Maybe because the concept of forest has long been that of a
closed canopy without larger clearings, forests are generally not considered
an important habitat for Orthoptera, which are considered indicators of

[437]
1498

grassland naturalness (Báldi and Kisbenedek 1997). Consequently, little is

known of the distribution and habitats favoured by Orthoptera before human
actions (mainly pastoralism and agriculture) shaped the landscape of Central
Europe.
As traditional land use is being abandoned throughout Europe, there are
increasing concerns that man-made habitats should be maintained in order to
preserve open land species (e.g. Firbank et al. 1994). As the majority of
Orthoptera are open land species (Ingrisch and Köhler 1998), the importance
of openings in forests is recognised (Shure and Phillips 1991, Clayton 2002,
Bouget and Duelli 2004) and forest clearings are sometimes maintained for the
conservation of Orthoptera (Kati et al. 2003). However, the minimal size of
openings in forests is still unknown for most species and it is still not clear
which species would actually disappear on a large scale when habitats return to
a non-managed state. Some types of forest cover may even be favourable to
thermophilous species (Liana 1981, Bönsel and Runze 2000). On the other
hand, forest can represent a barrier for species of open habitats and Orthoptera
usually prefer using habitat corridors for dispersal (Collinge 1998, Berggren
et al. 2002, Jordán et al. 2003), although some species are able to cross long
distances through unsuitable habitats: in a German forest 13 Orthoptera spe-
cies occurred on newly created clearings (Laußmann 1993). The knowledge of
the necessary patch size and the role of corridors is crucial for conservation. An
important factor that determines the necessary patch size for a species is the
shading effect of the forest edge (Bieringer and Zulka 2003). The relationship
between patch size and Orthoptera biodiversity has been studied in Hungarian
steppe patches (Báldi and Kisbenedek 1999) and in North American forests
(Shure and Phillips 1991). We studied Orthoptera, Dermaptera, and Blattodea
in a Central European forest in order to assess the minimum patch size of open
habitat necessary for each species and the role of linear corridors as habitat.
The results are intended to help assess which species actually depend on human
land use under the current ecological situation and which species would occur
in Central Europe if most of its surface was closed forest or if it was a park-
land.

Study area and methods

The study area lies in the Polish lowlands on the border to Belarus and includes
the Polish side of the Białowie_za Forest (600 km2) and its surroundings (Figure
1). The Białowie_za Forest is a forest complex of 1450 km2 (5230¢–5300¢ N,
2330¢–2415¢ E) that straddles the Polish–Belarussian border. The forest is a
mosaic of deciduous, coniferous, and mixed tree stands where large ungulates
such as the European bison (Bison bonasus), moose (Alces alces), red deer
(Cervus elaphus), roe deer (Capreolus capreolus), and wild boar (Sus scrofa)
occur. The Polish side of the Białowie_za Forest consists of the Białowie_za
National Park and a commercial forest (480 km2), in which timber harvest,

[438]
 1499

Figure 1. Sample sites (open circles) in the Białowie_za Forest (light grey) and the near sur-
roundings (sample sites at the Bug river not within the ranges of the figure) in 1997–2000. Open
land (white), strict reserve of the Białowie_za National Park (dark grey), rivers and lakes (black lines
and area), state border (dashed and dotted line).

reforestation, and hunting take place. Fifty km2 of the Białowie_za National
Park have been protected as a strict reserve since 1921. No hunting or forestry
is permitted in the strict reserve. The vegetation structure of the strict reserve
and of some places in the commercial forest is little influenced by humans.
Most of the study area is covered by closed forest (Figure 1). Open habitats
that occur within the forest are clearings of natural origin (usually few square
meters but sometimes up to several thousand square meters), young pine re-
growth areas (either plantations or natural, usually dry and sunny), sandpits
(mainly small-scale with bare sandy parts and older parts with vegetation),
mesophilic forest meadows (usually covered by high grasses), and sandy
meadows (mostly dry and with low vegetation). Linear corridors in the forest
consist in roads and railway lanes (10–40 m wide) and open marshes along
rivers, which are semi-natural as the use of meadows and reed in the marsh
almost completely stopped in the 1950s, so the marsh is now often covered by
reed and willow shrubs. The open land around the village Białowie_za is

[439]
1500

connected with the open land outside the Białowie_za Forest by a large river
marsh (Narewka) and a railway lane 30–40 m wide. The open land outside the
Białowie_za Forest is mainly non-intensive arable land, but also includes
sandpits, fallow land, dry and wet meadows.
Koźmiński (1925) recorded 32 species of Orthoptera, 1 species of Dermap-
tera and 1 species of Blattodea (see Table 1) in the Białowie_za Forest at a
period just after the total density of ungulates (including cattle) was of 20
individuals/km2 and of a crude biomass of 3000 kg/km2 (Je˛drzejewska et al.
1997). The forest still bears the signs of the almost medieval use that persisted
until recently (mid 20th century) as clearings were used for hay making, sand
and stone quarries were created throughout the forest, cattle grazed extensively
in the forest whilst some game species were protected. The impact of herbivores
was therefore much higher. Human use of the forest has changed, forestry
exploitation is now being promoted and ungulate densities are being kept low
by hunting. In addition, the European bison are being fed in winter, which
reduces their impact on forest re-growth, As a result, the total ungulate density
was around 10 individuals/km2 for a biomass of 1000 kg/km2 at the time of
this study (Je˛drzejewska et al. 1997).
On 150 days from April to October in 1997–1999 and in September 2000, we
recorded Orthoptera, Blattodea, and Dermaptera at 187 sites in the Białowie_za
Forest and its surroundings (Figure 1). The furthest 2 sites in the surroundings
were 50 km south-west from the Białowie_za Forest: dunes in the Kozki Nature
Reserve 5 km south of the town Siemiatycze and sandy meadows along the
Bug river east of the town Drohiczyn. As we were primarily interested in the
minimum patch size and the total numbers of species in each patch size class, a
standardised sampling effort or a complete species list of each site were not
necessary. However, as we recorded species everywhere where we encountered
them, the sample size for each patch size class was finally comparable con-
sidering that larger plots need a larger sample size (Table 1), except for the size
class from 0.1–1 km2. The small sample size of this size class was related to the
little diversity of habitats in these clearings (mainly fields and meadows for
cattle grazing). All size classes included sites in dry and wet habitats and from
all months to avoid any influence of habitat or season on the patch size
analyses. We visited most sites only once and searched for animals until we
could no longer find new species. However, if we expected species to exist in a
site but could not find them during our first visit, we usually returned at least
once to the site (287 samples on the 187 sites).
Because of earlier experience in identifying Orthoptera in the field, we were
able to identify species in the field acoustically (Bellmann 1985; Bellmann
2004) or by their morphology (Harz 1957) using magnifying glasses. Identi-
fication of Orthoptera using combinations of stridulation and morphological
characteristics are not only ethically preferable but also more reliable than
retrospective identifications of dead individuals. When we were not sure about
a field identification of Blattodea or Dermaptera, we collected the animals and
identified them later with a stereoscope and a key (Harz 1957). We used a

[440]
 Table 1. Frequency of occurrence (in % of sites) of Orthoptera, Dermaptera, and Blattodea at 187 sites in 6 size classes of clearings in the Białowie_za Forest, on linear corridors
(roads of 10–40 m width, river marshes of 100–200 m width), in the glade of Białowie_za (106–107 m2) and in the open land within 50 km around the Białowie_za Forest ( > 108 m2).
The maximal distance to forest edge was measured from sites to the forest edge with agricultural land outside the Białowie_za Forest or the glade of Białowie_za (max. possible: 10 km).

Species Size of open habitat patch (m2) Max distance

0–10 10–10 2 2
10 –10 3 3
10 –10 4
Roads Marsh 4
10 –10 5 5
10 –10 6 6
10 –10 7 8 to forest
> 10
edge (km)
Number of sites 12 (12) 23 (23) 12 (12) 28 (39) 34 (43) 9 (18) 23 (44) 6 (7) 21 (55) 19 (34)
(number of samples)

Orthoptera
Podisma pedestris *
Stenobothrus lineatus* +
Psophus stridulus* +
Stenobothrus stigmaticus* 51
Aiolopus thalassinus 51
Chorthippus vagans 51
Omocestus rufipes 111
[441]

Meconema thalassinum 5 5
Decticus verrucivorus* 10 21
Tettigonia viridissima* 4 29 53 6.2
Gryllotalpa gryllotalpa* 11 9 19 5 2.0
Phaneroptera falcata 11 5 11 3.1
Conocephalus dorsalis* 22 9 33 19 11 8.7
Chorthippus montanus* 3 22 9 33 14 16 5.8
Stethophyma grossum* 4 22 13 50 14 16 4.6
Conocephalus discolor 4 10 4.1
Barbitistes constrictus* 4 5 2.8
Gryllus campestris* 7 3 17 17 5 16 6.4
Chrysochraon dispar* 7 9 44 9 17 43 11 8.7
Tetrix tenuicornis* 11 9 17 10 5 6.3
Chorthippus mollis 14 11 13 33 19 47 9.8
Metrioptera bicolor 14 13 14 84 9.8

1501
Omocestus haemorrhoidalis* 18 6 17 50 10 37 9.8
Euthystira brachyptera 7 9 11 9 9.8
 Table 1. (Continued).

1502
Species Size of open habitat patch (m2) Max distance
2 2 3 3 4 4 5 5 6 6 7 8 to forest
0–10 10–10 10 –10 10 –10 Roads Marsh 10 –10 10 –10 10 –10 > 10
edge (km)
Number of sites 12 (12) 23 (23) 12 (12) 28 (39) 34 (43) 9 (18) 23 (44) 6 (7) 21 (55) 19 (34)
(number of samples)

Gomphocerippus rufus* 7 5.5

Chorthippus parallelus* 8 3 11 4 38 42 3.2
Metrioptera brachyptera* 8 4 3 17 + 8.4
Chorthippus albomarginatus 17 11 33 22 17 38 37 9.8
Oedipoda caerulescens* 25 29 11 30 33 10 47 9.8
Myrmeleotettix maculatus* 17 14 3 11 13 17 5 42 9.7
Chorthippus brunneus* 42 46 3 11 30 17 24 47 9.8
Chorthippus dorsatus* 25 21 6 44 26 67 38 58 9.8
Chorthippus biguttulus* 50 43 18 33 39 50 29 68 9.8
Tetrix bipunctata* 25 11 18 4 5 7.7
[442]

Chorthippus apricarius* 9 50 54 12 44 52 50 33 26 9.8

Omocestus viridulus* 22 17 46 12 56 52 67 33 53 9.8
Metrioptera roeselii* 26 67 61 15. 56 57 67 43 63 9.8
Chorthippus pullus* 4 8 25 21 9 9.6
Tettigonia cantans* 8 61 58 50 15 56 43 17 48 11 9.8
Pholidoptera griseoaptera* 25 26 33 43 29 22 30 5 16 9.8
Tetrix subulata* 33 13 8 4 6 22 17 29 26 9.2
Dermaptera
Forficula auricularia* 14 5
Labia minor 5
Apterygida albipennis 11 5 3.1
Chelidurella acanthopygia 8 6 5.4
Blattodea
Ectobius lapponicus* 8 4 8.4
Ectobius silvestris 33 4 7 3 4 5 8.9

*species found by Koźmiński (1925) in the Białowie_za Forest, +found by Bönsel and Runze (2000) 100 km north-west of the Białowie_za Forest, 1found only on dunes in the Kozki
Nature Reserve and on sandy meadows along the Bug river 50 km south-west of the Białowie_za Forest. Scientific names of Orthoptera follow Heller et al. (1998) and of Blattodea
and Dermaptera follow Harz (1957).
 1503

geographic information system to calculate distances from a clearing to the

next forest border and the total number of species found in a given patch or
corridor size class as measure for species richness. We plotted the numbers of
species against the mean size of clearings (log transformed) for each size class
and tested with SPSS 11.0 for Windows which regression model (linear, log-
arithmic, inverse, quadratic, cubic, power, compound, S-curve, logistic,
growth, and exponential) best explained the increase in species numbers.

Results

We found 44 species (38 Orthoptera, 4 Dermaptera, 2 Blattodea) in our study

area. Of these, 16 species occurred in closed forest, natural clearings or in the
strict reserve of the national park. Many species occurred in clearings that were
the furthest from the forest edge to arable land or the glade of Białowie_za
(Table 1). Most species already occurred in clearings < 0.01 km2. Six species
only occurred outside the Białowie_za Forest but the total number of species
outside was lower than the number of species living in the Białowie_za Forest.
No species occurred exclusively in closed forest.
Although all logarithmic models were also significant, power models best
explained the increase of the mean, total and cumulative numbers of species
(Figure 2). Linear corridors were habitat to 22 (roads) and 21 (marsh) species,
respectively. The numbers of species found in corridors correspond to the
number of species found in clearings of about 10,000 m2.

Discussion

We found 44 species of Orthoptera, Dermaptera and Blattodea in the

Białowie_za Forest and its surroundings. This list includes 9 species that
Koźmiński (1925) did not report for the area. We did not find one species of
Dermaptera (Labidura riparia) and one species of Blattodea (Ectobius
erythronotus) that have been found in the Białowie_za Forest (Liana 2001) and
that probably still exist. Only few other species might occur in and around the
Białowie_za Forest, but even if some other species exist in the area, they would
not have greatly influenced the general results of this study.
Four species (Podisma pedestris, Psophus stridulus, Stenobothrus lineatus and
Stenobothrus stigmaticus), which were common in the study area when the
ungulate density was twice as high and ungulate biomass 3 times higher
(Koźmiński 1925) than at the time of this study, disappeared from the
Białowie_za Forest. We however found one of these species, Stenobothrus
stigmaticus, at a study site 50 km south-west of the Białowie_za Forest in dunes
of the Bug river (Table 1). All four species, which also became rare in other
parts of Central Europe (Harz 1957, Liana 1981) and of which the first two
stand on the Polish red list (Liana 1992), probably disappeared due to changes

[443]
1504

Figure 2. Best fit models (power regression) of mean numbers of species (regression line for single
sites and 95% confidence intervals for sites of each size class) and of total numbers of species (solid
regression line with closed circles for each size class, dotted regression line with open circles for
cumulative number of species) of clearings in the Białowie_za Forest and the open land around the
forest.

[444]
 1505

in land use such as the disappearance of forest pasture and the abandonment of
heathland. In contrast, Liana (1981) found that dry habitats in forested areas
were richer in Orthoptera species than those in the open land. Five spe-
cies (Barbitistes constrictus, Chorthippus pullus, Euthystira brachyptera,
Gomphocerippus rufus, Metrioptera brachyptera) were even restricted to
forested areas. Psophus stridulus occurred in the Biebrza area (north-east
Poland) only on a dune surrounded by trees, which had a warmer micro
climate than the surroundings (Bönsel and Runze 2000). It is likely that
clearings in forested areas are better habitats for thermophilous species due to
thermal advantages. Dragonflies also exhibited a similar dependence to forest
in the Białowie_za Forest where the most thermophilous species reproduced
only in ponds of forest clearings but not in the surroundings (Theuerkauf and
Rouys 2001). Forests might therefore play an important role in maintaining
Orthoptera diversity in Central Europe.
Although forest can improve the habitat quality for thermophilous species, it
can also reduce the survival chances of species that depend on relatively large
open habitat patches. Bieringer and Zulka (2003) found that shading affects the
occurrence of Orthoptera up to 30 m from the forest edge. Clearings under
1000 m2 are therefore almost entirely affected by shading from the forest edge,
and indeed we found only about one third of species on these clearings. Two
third of species already occurred in the next size class (0.01–0.1 km2), which is
probably linked to a reduced shading effect. Those species that disappeared
from the Białowie_za Forest, however, probably needed a larger surface of
habitat. High density of cattle and wild ungulates might have provided the
necessary extension of habitat for these thermophilous Orthoptera species at
the beginning of the 20th century, but the current grazing intensity did not
allow these species to persist. We think that these species probably disappear in
Central Europe where human land use is discontinued and not taken over by
intensive ungulate grazing. In the Bieszczady National Park (Southeast
Poland), Psophus stridulus and Aiolopus thalassinus probably disappeared after
human land use (cattle grazing) was discontinued 60 years ago (Theuerkauf
et al. 2005).
Habitat corridors are known to be important for the dispersal of Orthoptera
(Collinge 1998; Berggren et al. 2002; Jordán et al. 2003). However, our study
indicated that linear corridors in forests are also an important habitat for
Orthoptera. Both corridors in river marshes and along forest roads had
numbers of species that corresponded to clearings of about 10,000 m2. Whilst
the number of species on the wet marsh corridors can be explained by their
width of 100–200 m, the number of species on the road corridors of 10–40 m
was much larger than might be expected by the patch size. Obviously, the
shading effect was not an important limitation in species numbers on these dry
corridors. From an evolutionary perspective, it is possible that species of drier
habitats must be able to exist on smaller habitat patches than those of wet
habitats. The reason might be that river marshes are open on a larger scale than
dry clearings due to regular flooding or activity of beavers (Castor fiber). Dry

[445]
1506

clearings on the other hand can be maintained by ungulates but first need to be
created (for example by wind), which occurs probably on a smaller area.
We conclude that Central Europe might loose a few Orthoptera species
(e.g. the first 7 species in Table 1) if human land use was replaced by grazing by
wild animals. However, under the current ecological situation with low her-
bivore densities and only a few species in most regions, it would take consid-
erable management measures to re-establish natural grazing communities.
Many species of Orthoptera and even some Dermaptera species would prob-
ably disappear from regions where human land use is discontinued and not
replaced by natural grazing. Rather than keeping forest clearings open for the
conservation of Orthoptera (Kati et al. 2003), we argue that the concept of
species conservation in open habitats needs to be reconsidered to include a full
array of herbivores at densities that might appear high but that were common
in historical times (Beutler 1996).

Acknowledgements

We thank C. Okołow, the director of the Białowie_za National Park, for per-
mission to work in the strict reserve, B. Jaroszewicz for providing us with
information during the study, P.-M. A. Dettinger-Klemm and two anonymous
reviewers for useful comments.

References

Báldi A. and Kisbenedek T. 1997. Orthopteran assemblages as indicators of grassland naturalness

in Hungary. Agric. Ecosyst. Environ. 66: 121–129.
Báldi A. and Kisbenedek T. 1999. Orthopterans in small steppe patches: an investigation for the
best-fit model of the species-area curve and evidences for their non-random distribution in the
patches. Acta Oecologica 20: 125–132.
Bellmann H. 1985. Die Stimmen der heimischen Heuschrecken [the calls of native orthoptera].
Verlag J. Neumann-Neudamm, Melsungen [audio tape].
Bellmann H. 2004. Heuschrecken – Die Stimmen von 61 heimischen Arten [Orthoptera – Calls of
61 Native Species]. Ample Edition, Musikverlag. [audio CD].
Berggren Å., Birath B. and Kindvall O. 2002. Effect of corridors and habitat edges on dispersal
behavior, movement rates, and movement angles in Roesel’s bush-cricket (Metrioptera roeseli).
Conserv. Biol. 16: 1562–1569.
Beutler A. 1996. Die Großtierfauna Europas und ihr Einfluß auf Vegetation und Landschaft [The
megafauna of Europe and its impact on vegetation and landscape]. In: Gerken B. and Meyer C.
(eds), Wo lebten pflanzen und Tiere der Naturlandschaft und der frühen Kulturlandschaft Eu-
ropas? [Where did Plants and Animals Live in the Natural and Early Cultural Landscapes of
Europe?]. Natur- und Kulturlandschaft 1, Höxter, pp. 51–106. [In German]
Bieringer G. and Zulka K.P. 2003. Shading out species richness: edge effect of a pine plantation on
the Orthoptera (Tettigoniidae and Acrididae) assemblage of an adjacent dry grassland. Biodiv.
Conserv. 12: 1481–1495.

[446]
 1507

Birks H.J.B. 2005. Mind the gap: how open were European primeval forests? Trends Ecol. Evol. 20:
154–156.
Bönsel A. and Runze M. 2000. Ein Habitat der Rotflügeligen Schnarrschrecke (Psophus stridulus
L., 1758) im nordöstlichen Polen [A habitat for Psophus stridulus L., 1758 in northeastern
Poland]. Articulata 15: 49–62. [In German with English abstract]
Bouget C. and Duelli P. 2004. The effects of windthrow on forest insect communities: a literature
review. Biol. Conserv. 118: 281–299.
Bradshaw R.H.W., Hannon G.E. and Lister A.M. 2003. A long-term perspective on ungulate-
vegetation interactions. Forest Ecol. Manage. 181: 267–280.
Clayton J.C. 2002. The effects of clearcutting and wildfire on grasshoppers and crickets (Orthop-
tera) in an intermountain forest ecosystem. J. Orthoptera Res. 11: 163–167.
Collinge S.K. 1998. Spatial arrangement of habitat patches and corridors: clues from ecological
field experiments. Landscape and Urban Planning 42: 157–168.
Firbank L.G., Telfer M.G., Eversham B.C. and Arnold H.R. 1994. The use of species-decline sta-
tistics to help target conservation policy for set-aside arable land. J. Environ. Manage. 42: 415–422.
Harz K. 1957. Die Geradflügler Mitteleuropas [The Orthoptera of Central Europe]. Gustav
Fischer, Jena. [In German]
Heller K.G., Korsunovskaya O., Ragge D.R., Vedenina V., Willemse F., Zhantiev R.D. and
Frantsevich L. 1998. Check-list of European Orthoptera. Articulata Beiheft 7: 1–61.
Ingrisch S. and Köhler G. 1998. Die Heuschrecken Mitteleuropas [The Orthoptera of Central
Europe]. Neue Brehm-Bücherei Bd. 629. Westarp Wissenschaften, Magdeburg Germany. [In
German]
Je˛drzejewska B., Je˛drzejewski W., Bunevich A., Miłkowski L. and Krasiński Z. 1997. Factors
shaping population densities and increase rates of ungulates in Białowie_za Primeval Forest
(Poland and Belarus) in the 19th and 20th centuries. Acta Theriologica 42: 399–451.
Jordán F, Báldi A, Orci K.-M., Rácz I. and Varga Z. 2003. Characterizing the importance of
habitat patches and corridors in maintaining the landscape connectivity of a Pholidoptera
transsylvanica (Orthoptera) metapopulation. Landscape Ecol. 18: 83–92.
Kati V., Dufrêne M., Legakis A., Grill A. and Lebrun P. 2003. Conservation management for
Orthoptera in the Dadia reserve, Greece. Biol. Conserv. 115: 33–44.
Koźmiński Z. 1925. Ökologische Untersuchungen an Orthopteren des Urwalds von Białowie_za
[Ecological studies of Orthoptera in the Białowie_za Forest], Bulletin de l’Académie Polonaise des
Sciences et des Lettres – Classe des Sciences Mathématique et Naturelles – Série B: Sciences.
Naturelles 1925: 447–475. [In German]
Laußmann H. 1993. Die Besiedlung neu entstandener Windwurfflächen durch Heuschrecken
[Colonisation of newly created wind-fall sites by Orthoptera]. Articulata 8: 53–59. [In German
with English abstract]
Liana A. 1981. Prostoskrzydłe (Orthoptera) w siedliskach kserotermicznych Pojezierza Mazur-
skiego [Orthoptera in xerothermic habitats of the Mazurian lakeland]. Fragmenta Faunistica 25:
479–510. [In Polish with Russian and French abstracts]
Liana A. 1992. Owady prostoskrzydłe Orthoptera [Orthopteroid insects]. In: Głowaciński Z. (ed),
Czerwona lista zwierza˛t gina˛cych i zagro_zonych w Polsce [Polish Red List of Threatened Ani-
mals]. Polish Academy of Sciences, Nature Protection Research Centre, Kraków, pp. 85–91. [In
Polish with English abstract]
Liana A. 2001. Orthoptera-Blattodea. In: Gutowski, J.M. and Jaroszewicz B. (eds), Katalog Fauny
Puszczy Białowieskiej. [Faunal catalogue of the Białowieza Forest]. Instytut Badawczy
Leśnictwa, Warszawa, pp. 92–93.
Shure D.J., Phillips D.L. 1991. Patch size of forest openings and arthropod populations. Oecologia
86: 325–334.
Svenning J.-C. 2002. A review of natural vegetation openness in north-western Europe. Biol.
Conserv. 104: 133–148.
Theuerkauf J., Rouys S. 2001. Habitats of Odonata in the Białowie_za Forest and its surroundings
(Poland). Fragmenta Faunistica 44: 33–39.

[447]
1508

Theuerkauf J., Rouys S. Grein G. Becker A. 2005. New records of Orthoptera in the Bieszczady
Mountains (Southeast Poland) with special regard to the genus Isophya. Fragmenta Faunistica
48: 9–14.
Vera F.W.M. 2000. Grazing Ecology and Forest History. CABI Publishing, Wallingford.

[448]
Biodiversity and Conservation (2006) 15:1509–1527
Springer 2006
DOI 10.1007/s10531-005-2632-0

-1

Species Composition, diversity and local uses

of tropical dry deciduous and gallery forests
in Nicaragua

BENIGNO GONZÁLEZ-RIVAS1, MULUALEM TIGABU2,*,

KARIN GERHARDT3, GUILLERMO CASTRO-MARÍN1 and
PER CHRISTER ODÉN2
1
Facultad de Recursos Naturales y del Ambiente, Universidad Nacional Agraria Apartado Postal 453,
Managua, Nicaragua; 2Department of Forest Genetics and Plant Physiology, Tropical Silviculture
and Seed Laboratory, Faculty of Forest Sciences, Swedish University of Agricultural Sciences,
SE-901 83 Umeå, Sweden; 3Department of Plant Ecology and Evolutionary Biology Centre, Uppsala
University, Villavägen 14, SE-752 36 Uppsala, Sweden; *Author for correspondence (e-mail:
[email protected], [email protected]; phone: +46-90-786-83-19; fax: +46-90-
786-58-96)

Received 6 December 2004; accepted in revised form 15 August 2005

Key words: Biodiversity, Central America, Endangered species, Floristic composition

Abstract. The floristic composition and diversity of tropical dry deciduous and gallery forests were
studied in Chacocente Wildlife Refuge, located on the Pacific coast in Nicaragua during 1994 and
2000. Density, dominance and frequency as well as species and family important values were
computed to characterize the floristic composition. A variety of diversity measures were also
calculated to examine heterogeneity in each forest community. A total of 29 families, 49 genera and
59 species were represented in 2 ha dry deciduous forest. In the gallery forest, the number of
families, genera and species recorded in 2000 inventory was 33, 48 and 58, respectively and slightly
higher than the 1994 inventory. The number of stems ‡10 cm dbh varied from 451 to 489 per
hectare in the deciduous forest, and from 283 to 298 per hectare in the gallery forest. The basal area
was much larger for species in the gallery than dry deciduous forest. Fabaceae, sub family
Papilionoideae, was the most specious family in the deciduous forest while Meliaceae was the
dominant family in the gallery forest. Similarity in species composition and abundance between
deciduous and gallery forests was low. In terms of species diversity, the gallery forest was found
more diverse than the deciduous forest using Fisher’s diversity index. Both forest communities were
characterized by a typical inverse J shape. Therefore, emphasis should be given to the protection of
rare species, i.e. as the forests are still under continued human pressure, an immediate action should
be taken to conserve the remaining flora.

Introduction

Dry forests once covered more than 40% of the total area of tropical forests
(Murphy and Lugo 1986). They are considered to be one of the most threa-
tened of all the major tropical forest habitats and are argued to deserve a high
priority for conservation (Janzen 1988; Gillespie et al. 2000). According to the
Holdridge system of life zone classification, dry tropical and subtropical forests
and woodlands occur in frost-free areas with a mean annual temperature

[449]
1510

higher than 17 C, a mean annual rainfall between 250 and 2000 mm, and an
annual ratio of potential evapotranspiration to precipitation exceeding unity
(Murphy and Lugo 1995).
The area of natural forests in Central America is estimated to be 190,000 km2,
representing ca. 15% of the total land cover. In addition, some 130,000 km2
deforested land is considered suitable only for forestry, adding to 24% of the total
area (Segura et al. 1997). The deforestation rate in Central America is estimated
as 0.5 km2 per year (Roldan 2001). The tendency of human populations to
concentrate in drier climates is hastening the rate of dry forest degradation
(Murphy and Lugo 1995) and deforestation has increased dramatically with
population growth during the last century. Large areas are cleared for grazing
and agriculture and only fragments of dry forests remain (Gerhardt 1994).
Nicaragua has 2500 km2 of tropical dry forests, representing ca. 2% of the
total forest cover (Harcourt and Sayer 1996). The dry forests are found mainly
on the Pacific coast where ca. 50% of the population also lives. Nicaraguan dry
forests have been intensively exploited for commercial timber production. The
major commercial timber species are Swietenia humilis, Cedrela odorata, Bom-
bacopsis quinata, Dalbergia retusa and Guaiacum sanctum (Sabogal 1992). The
extraction of valuable commercial trees for export started in early 1900 (Tercero
and Urrutia 1994), and continued for decades, resulting in considerable
reduction of commercially important species. The Nicaraguan Pacific railway
was constructed in the 1950s and most of the railway sleepers used was extracted
from the dry forest in Chacocente (Tercero and Urrutia 1994). The dry forests
are still major sources of wood for fire, poles and timber, and provide oppor-
tunities for hunting and collection of other important non-timber forests
products (NTFP). In addition to cutting of trees for wood and related products,
the major causes of deforestation have been conversion of dry forests into coffee
plantations, crop fields and ranches (Roldan 2001).
Chacocente National Wildlife Refuge was established in 1983 to protect the
nesting beach of marine turtles and the last area of the tropical dry forest due to
the social, economic, ecological and scientific relevance of this type of ecosys-
tem. During the Sandinista Revolution big ranches were expropriated and be-
came property of the state. In 1990 this land was given to peasant cooperatives.
By 1998, the land tenure changed very rapidly since land was being sold and
cooperative land was converted into private land (Anonymous 2002). Today,
the Chacocente National Wildlife Refuge consists mainly of private farms (84
owners), although some are quite small. The only state land in the refuge is a
small property donated to Ministry of Natural Resources and Environment
(MARENA) by the International Fauna and Flora Organization. The refuge is
not fully protected against human impact and is utilized both legally and ille-
gally by the local people living inside as well as outside the refuge. Anthropo-
genic disturbances such as burning, grazing, wood collection and illegal cutting
are factors affecting plant population density (Gillespie et al. 2000).
The effect of this land use dynamics and forest fragmentation on biological
diversity in Chacocente is not well documented (Sabogal and Valerio 1998). An

[450]
 1511

assessment of species composition and diversity provide information for

developing guidelines for conservation priorities in the region since few com-
parative or quantitative studies in remaining forest fragments in Central
America have been made (Gillespie et al. 2000). In this study, we described the
floristic composition and species diversity of two tropical dry forest types, dry
deciduous and gallery forests, at two different times. The vegetation description
presented will hopefully contribute to a better understanding of the floristic
composition and diversity of the tropical dry forests. Given the threatened
status of dry forests throughout the tropics, particularly in Nicaragua, and the
fact that dry forests are less studied than moister tropical forest types, this
study will provide important baseline data for the region.

Materials and methods

Study area

This study was carried out in Chacocente Wildlife Reserve (1136¢–1130¢ N

and 8608¢–8615¢ W) located on the Pacific coast in the department of Carazo,
Nicaragua (Figure 1). The refuge consists of closed deciduous forest (1099 ha),
gallery forest (471 ha), open low forest (1842 ha), fallows area (554 ha), annual
crops (311 ha), grassland (294 ha), and beach area (71 ha) (Anonymous, 2002).
Chacocente has a dry period of 7 months with less than 50 mm precipitation
per months, and during the rainy season (June–October) rainfall is irregular
with many days without rainfall (Anonymous 2002). The mean annual pre-
cipitation during the last 13 years has been 1422 mm, with a maximum in 1995

Figure 1. Distribution of tropical dry forests in Nicaragua and location of the study site,
Chacocente.

[451]
1512

(1962 mm) and the minimum in 1991 (991 mm). During October 1998,
hurricane Mitch passed over the area and the precipitation that month was as
high as 775 mm. The average annual temperature is 26 C (Anonymous 2002).
The gallery forest, defined as narrow patches along the fringes of semi-
permanent watercourses (Lamprecht 1989), occurs along the main water
course, the Rı́o Escalante. It has a different species composition, structure and
stand density than the more common deciduous forest. The vegetation is
mostly evergreen, the trees are tall and the majority of the trees have a diameter
exceeding 35 cm at breast height. The deciduous forest trees totally or partially
shed their leaves during the dry season.

Sampling and data analyses

Two permanent plots in each forest type were established by Universidad

Nacional Agraria, Managua, Nicaragua in 1989. The area of each permanent
plot was 1.0 ha and subdivided into 25 subplots of 20 · 20 m. Each plot was
systematically surveyed by identifying, measuring, and tagging all trees with
diameter at breast height (dbh) ‡10 cm. The inventories were made in 1994 and
2000. In addition, local names were recorded and information about uses of the
tree species was gathered by consulting a Nicaraguan forest use specialist
(Claudio Calero, personal communication) and relevant literature (Salas 1993;
Stevens et al. 2001). All scientific names were thoroughly cross-checked in the
TROPICOS nomenclatural database (http://mobot.mobot.org/W3T/search/
Vast.html) of the Missouri Botanical Garden.
The importance value index (IVI) and family importance value (FIV) were
used to describe the species composition of the plots. IVI of a species is defined
as the sum of its relative dominance, its relative density and its relative
frequency, which in turn are calculated as follows:
Relative dominance¼ total basal area for a species/total basal area for all
species
Relative density¼ number of individuals of a species/total number of
individuals
Relative frequency¼ frequency of a species/sum frequencies of all species

The frequency of species is defined as the number of subplots (20 · 20 m) in

which the species is present. The theoretical range for relative dominance,
relative density and relative frequency is 0–100%, thus IVI of species may vary
between 0 and 300%. The FIV was computed in the same way as IVI except
that relative frequency was replaced by the relative diversity, computed as the
number of species in a family/total number of species. All species encountered
during both inventories were clustered into three groups based on the mean
number of individuals of a species per hectare as rare (4 individuals per

[452]
 1513

hectare), intermediate (4–24 individuals per hectare) and abundant

(>24 individuals per hectare). According to Duque and Cavelier (2003), a
species with two or fewer individuals in 2.16 ha is considered as locally rare.
A variety of commonly used diversity indices were computed in order to
permit a more precise comparison of the alpha diversity between the two forest
communities. These indices were Margalef’s species richness index, Shannon’s
measure of evenness, Shannon–Wiener’s diversity index, Simpson’s diversity
index and Fisher’s diversity index. These indices are widely employed to
measure biological diversity (Magurran 2004). In addition, the species-abun-
dance patterns in each forest community were plotted. Floristic similarity
between forest communities was assessed using Jaccard’s coefficient of simi-
larity, based on the presence/absence of the species, and Morisita’s index of
similarity, based on number of individuals per species. Jaccard’s coefficient of
similarity and Morisita’s index vary between 0 and 1 and a value close to 1
indicates greater similarity between forest communities (Krebs 1999). The
conservation status of species encountered in our plots was assessed based on
the 2004 IUCN Red List of Threatened Species directory (IUCN 2004).
In addition, candidate species for future IUCN listing were identified based on
their rarity and regional distribution based on Flora of Nicaragua (Stevens
et al. 2001) and the Missouri Botanical Garden’s TROPICOS database.

Results

Floristic composition

A total of 29 families, 49 genera and 59 species were found in the dry deciduous
forest during both inventories (Table 1). While the stem density slightly in-
creased in 2000 inventory, the basal area was relatively less compared to the
inventory made in 1994. Fabaceae, sub-family Papilionoideae was the most
specious family with higher FIV (Table 2). Other families (sub-families) with
‡4 species were Caesalpinioideae and Boraginaceae. Hernandiaceae, though
represented by one species (Gyrocarpus americanus), had the second and third
higher FIV in 1994 and 2000 inventories, respectively owing to the large stem
density per hectare (62 individuals/ha in 1994 and 37 individuals/ha in 2000).
Gyrocarpus americanus stood out as the most abundant species during both
inventories in terms of basal area, relative dominance, relative frequency and
IVI (Table 3). While Tabebuia ochracea was the second most abundant species
during both inventories, Lonchocarpus minimiflorus and Myrospermum
frutescens were more abundant in 1994 and 2000 inventories, respectively. The
rarest species during both inventories were Celtis caudata and Zanthoxylum
caribaeum (Table 4). Four species, Acacia costaricensis, Ficus obtusifolia,
Pithecellobium saman and Trichilia hirta, recorded in 1994 inventory were not
encountered in 2000, but four other species, Adelia barbinervis, Casearia

[453]
1514

Table 1. Summary of floristic composition and structure of trees ‡10 cm dbh in dry deciduous and
gallery forests inventoried in 1994 and 2000.

Forest types-Inventory time Families Genera Species Stem densitya Basal areab

Deciduous-94 29 49 59 451 31.5

Deciduous-00 29 49 59 489 29.0
Gallery-94 33 47 55 298 45.3
Gallery-00 33 48 58 283 49.3
a
stem density = Number of individuals ha
1.
b
basal area (m2 ha
1).

Table 2. The ten most important families (sub-families) in the dry deciduous and gallery forests of
Chacocente in 2000 inventory according to decreasing order of family importance value (FIV).

Forest type Family Genus Species N/ha IFV

Deciduous Papilionoideae 8 10 82 58.3

Caesalpinioideae 3 5 45 29.3
Hernandiaceae 1 1 37 23.3
Mimosoideae 2 3 19 20.7
Bignoniaceae 1 1 40 19.2
Achatocarpaceae 1 1 23 14.8
Boraginaceae 1 4 13 14.0
Apocynaceae 1 1 30 12.7
Spindaceae 2 2 17 10.4
Tiliaceae 1 1 12 7.2
Gallery Miliaceae 3 4 39 40.0
Capparidaceae 1 1 14 17.6
Sapindaceae 1 1 13 17.3
Hernandaceae 1 1 7 15.5
Sterculiaceae 2 2 8 11.6
Annonaceae 2 2 15 11.4
Boraginaceae 1 4 13 14.0
Simaroubaceae 1 1 16 10.7
Apocynaceae 1 1 8 10.0
Rhamnaceae 2 2 8 8.5

corymbosa, Cordia dentata and Trema micrantha, were found in 2000 inventory
(Appendix).
In the gallery forest, the number of families, genera and species encountered
in 1994 inventory were 33, 47 and 55, respectively while 48 genera and 58 species
were recorded in 2000 inventory (Table 1). The total stem density (298 indi-
viduals/ha) was comparatively higher in 1994 inventory than in 2000 inventory
while the basal area was relatively larger in 2000 than in 1994 inventory. In both
inventories, Meliaceae was the most specious family with higher IFV (Table 2).
Most of the important families were represented by 1 or 2 species. The most
abundant species, in terms of basal area, relative dominance, and IVI was
Pithecellobium saman, followed by Trichilia hirta (Table 5). In both inventories,
Cordia alliodora was the rarest species, followed by Hymenaea courbaril in the

[454]
 1515

Table 3. The ten most abundant species in the dry deciduous forest of Chacocente in 1994 and
2000 inventories according to decreasing order of importance value index (IVI) together with
structural characteristics.

Species 1994 Inventory

Basal Relative Relative Relative IVI

area (m2/ha) dominance (%) density (%) frequency (%)

Gyrocarpus americanus 6.5326 20.72 13.57 8.12 42.4

Tabebuia ochracea 2.0714 6.57 8.57 7.48 22.6
Lonchocarpus minimiflorus 2.1939 6.96 9.79 5.56 22.3
Stemmadenia obovata 1.2935 4.10 9.23 6.84 20.2
Caesalpinia exostemma 1.8645 5.91 7.45 4.70 18.1
Myrospermum frutescens 1.3467 4.27 6.01 5.34 15.6
Lysiloma divaricatum 2.0737 6.58 3.34 4.70 14.6
Achatocarpus nigricans 1.3346 4.23 5.78 1.92 11.9
Gliricidia sepium 1.8495 5.87 2.78 2.78 11.4
Luehea candida 0.7214 2.29 2.78 4.06 9.1

2000 Inventory
Gyrocarpus americanus 3.548 12.23 9.41 8.28 29.9
Tabebuia ochracea 2.1671 7.47 10.04 7.63 25.1
Myrospermum frutescens 1.4233 4.91 6.52 6.10 17.5
Caesalpinia exostemma 1.8082 6.23 7.28 3.92 17.4
Stemmadenia obovata 1.0028 3.46 7.53 5.88 16.9
Achatocarpus nigricans 2.0956 7.23 5.90 3.05 16.2
Lonchocarpus minimiflorus 0.9979 3.44 7.90 4.79 16.1
Gliricidia sepium 1.8574 6.40 2.38 2.83 11.6
Luehea candida 0.6862 2.37 3.14 3.92 9.4
Allophylus psilospermus 0.5432 1.87 2.76 2.40 7.0

1994 inventory, and Triplaris melaenodendronin the 2000 inventory (Table 6).
Four species, Casearia tremula, Pithecellobium dulce, Piper aduncum and Randia
nicaraguensis recorded in 1994 inventory were missing in 2000 inventory while
seven other species were encountered in 2000 inventory; namely, Acacia cos-
taricensis, Bursera simaruba, Caesalpinia exostemma, Caesalpinia violacea,
Coccoloba sp., Licania arborea and Tabebuia rosea (Appendix).
As a whole, the number of species recorded in dry deciduous and gallery
forests was nearly the same. However, the stem density in the dry deciduous
forest was twice higher than the gallery forest while the basal area was much
bigger in the latter. It was found that the similarity in species composition
between the two forest communities in both inventories was very low, as shown
by low Jaccard’s (0.27) and Morisita’s (0.35) similarity indices.

Species diversity

The species-abundance patterns of dry deciduous and gallery forests displayed

a typical inverse J-distribution or the log series distribution (Figure 2). The

[455]
1516

Table 4. The ten rarest species in the dry deciduous forest of Chacocente in 1994 and 2000
inventories according to increasing order of IVI together with structural characteristics.

Species 1994 Inventory

Basal Relative Relative Relative IVI

area (m2/ha) dominance (%) density (%) frequency (%)

Zanthoxylum caribaeum 0.0154 0.05 0.11 0.21 0.37

Celtis caudata 0.0305 0.10 0.11 0.21 0.42
Ficus obtusifolia 0.0366 0.12 0.11 0.21 0.44
Coursetia elliptica 0.0266 0.08 0.22 0.43 0.73
Diospyros nicaraguensis 0.0566 0.18 0.22 0.43 0.83
Dalbergia retusa 0.0137 0.04 0.44 0.64 1.12
Calycophyllum candidissimum 0.0113 0.04 0.56 1.07 1.67
Swietenia humilis 0.278 0.88 0.33 0.64 1.85
Chomelia spinosa 0.0715 0.23 0.67 1.07 1.97
Haematoxylon brasiletto 0.6026 1.91 0.56 1.07 3.54
2000 Inventory
Celtis caudata 0.0167 0.06 0.13 0.22 0.40
Zanthoxylum caribaeum 0.0191 0.07 0.13 0.22 0.41
Malpighia stevensii 0.0216 0.07 0.13 0.22 0.42
Coursetia elliptica 0.0281 0.10 0.25 0.44 0.78
Dalbergia retusa 0.0097 0.03 0.50 0.65 1.19
Chomelia spinosa 0.0216 0.07 0.63 0.87 1.57
Diospyros nicaraguensis 0.0923 0.32 0.75 0.65 1.72
Calycophyllum candidissimum 0.0113 0.04 0.63 1.09 1.76
Haematoxylon brasiletto 0.1863 0.64 0.50 0.87 2.02
Swietenia humilis 0.3244 1.12 0.50 0.87 2.49

majority of the species in both forest communities were represented by few

individuals while few species in both forests were represented by many indi-
viduals. In the dry deciduous forest, out of the 59 species recorded during both
inventories, 37 species were considered as rare (<4 individuals/ha), 15 species
as intermediate (4–24 individuals/ha) and 7 species as abundant (>24 indi-
viduals/ha). Of all the species recorded in the gallery forest during both
inventories, 43 species was considered as rare (<4 individuals/ha), 8 species as
intermediate (4–24 individuals/ha) and 4 species as abundant (>24 individuals/
ha). The various diversity measures for each forest community are presented in
Table 7. Although the total number of species recorded in both forest com-
munities was very close, the number of individuals in 2 ha plot was much
higher in the dry deciduous than gallery forests. In terms of numerical species
richness (S/N), the two forest communities differed slightly (cf. 0.1 for gallery
and 0.06 for deciduous forest). According to Margalef’s index of species
richness (DMg), which combines mathematically number of species (S) and
numerical species richness (S/N), the gallery forest (in 2000 inventory) was
found to be more diverse than the dry deciduous forest. Shannon’s measure of
evenness did not differ much between and within forest communities. The
Shannon–Wiener diversity index, which combines species richness and

[456]
 1517

Table 5. The ten most abundant species in the gallery forest of Chacocente in 1994 and 2000
inventories according to decreasing order of importance value index together with structural
characteristics.

Species 1994 Inventory

Basal Relative Relative Relative IVI

area (m2/ha) dominance (%) density (%) frequency (%)

Pithecellobium saman 12.462 27.49 0.76 2.34 30.6

Trichilia hirta 4.2525 9.38 7.89 8.31 25.6
Thouinidium decandrum 2.9502 6.51 5.43 6.23 18.2
Simarouba glauca 1.5827 3.49 2.97 9.09 15.6
Capparis pachaca 0.8255 1.82 4.75 6.49 13.1
Gyrocarpus americanus 2.4077 5.31 2.21 4.16 11.7
Astronium graveolens 1.6976 3.74 1.87 5.97 11.6
Stemmadenia obovata 0.7411 1.63 3.99 4.16 9.8
Guarea glabra 0.9649 2.13 2.97 4.16 9.3
Trichilia moschata 0.8863 1.96 2.04 4.68 8.7

2000 Inventory
Pithecellobium saman 12.7787 25.89 1.22 2.07 29.2
Trichilia hirta 2.4232 4.91 15.29 8.55 28.8
Capparis pachaca 1.3614 2.76 13.15 7.51 23.4
Thouinidium decandrum 2.9474 5.97 9.63 6.74 22.3
Gyrocarpus americanus 4.6233 9.37 4.43 3.63 17.4
Simarouba glauca 2.01 4.08 4.89 8.29 17.3
Annona reticulata 1.3844 2.80 4.13 6.74 13.7
Astronium graveolens 2.0686 4.19 3.21 5.44 12.8
Stemmadenia obovata 0.8589 1.74 6.57 4.15 12.5
Guarea glabra 1.2341 2.50 5.35 4.40 12.3

evenness into a single value, declined over time within each forest community,
and identified the dry deciduous forest as more diverse than the gallery forest.
The complement of Simpson’s index, which attaches more weight to the
abundance of the most common species, also identified the dry deciduous
forest as more diverse than the gallery forest. Fisher’s diversity index, the most
widely recommended measure of diversity, revealed that the galley forest is
more diverse than the dry deciduous forest.

Local uses

Although we did not make a systematic ethno-botanical study, the local uses
of the tree species in both forest communities were identified based on
information gathered from the local people, expert consultation and existing
literature. Accordingly, ten major use categories were identified (Figure 3). It
was found that the largest number of species in both forest communities (53%
of the total species) was used for firewood, followed by timber extraction
(35%), rural construction (27%) and charcoal production (23%). Interest-

[457]
1518

Table 6. The ten rarest species in the gallery forest of Chacocente in 1994 and 2000 inventories
according to increasing order of IVI together with structural characteristics.

Species 1994 Inventory

Basal Relative Relative Relative IVI

area (m2/ha) dominance (%) density (%) frequency (%)

Cordia alliodora 0.0174 0.04 0.1 0.26 0.38

Hymenaea courbaril 0.0287 0.06 0.2 0.52 0.75
Karwinskia calderonii 0.0613 0.14 0.2 0.52 0.83
Triplaris melaenodendron 0.0931 0.21 0.2 0.52 0.90
Guaiacum sanctum 0.0257 0.06 0.3 0.78 1.09
Ceiba pentandra 0.5064 1.12 0.1 0.26 1.46
Cordia gerascanthus 0.09 0.20 0.3 1.04 1.58
Albizia caribaea 0.7178 1.58 0.1 0.26 1.92
Sterculia apetala 0.4611 1.02 0.3 1.04 2.40
Cedrela odorata 1.0955 2.42 0.6 1.82 4.83
2000 Inventory
Cordia alliodora 0.0209 0.04 0.15 0.26 0.45
Triplaris melaenodendron 0.0607 0.12 0.15 0.26 0.53
Hymenaea courbaril 0.0264 0.05 0.31 0.52 0.88
Karwinskia calderonii 0.0821 0.17 0.31 0.52 0.99
Cedrela odorata 0.0095 0.02 0.46 0.78 1.26
Guaiacum sanctum 0.2346 0.48 0.31 0.52 1.30
Cordia gerascanthus 0.0804 0.16 0.46 0.78 1.40
Albizia caribaea 0.7854 1.59 0.15 0.26 2.00
Ceiba pentandra 0.8202 1.66 0.31 0.52 2.49
Sterculia apetala 0.4898 0.99 0.61 1.04 2.64

ingly, 24% of the species are not currently under any kind of use. The
abundance of species by use group was also examined for each forest com-
munity (Figure 3). Given the large number of species used for firewood, the
overall abundance was also high. The most interesting part of this result is
that the abundance of species used for firewood and timber declined from
1994 to 2000 in both forest communities. Although the abundance of the ‘‘not
used’’ species in the gallery forest showed an increasing tendency, the reverse
held true in the deciduous forest.

Species with high conservation importance

Most of the species in our plots were represented by few individuals (Figure 2).
Some of the rarest species were already short-listed in IUCN red list directory
as threatened species. Among these threatened species, five species were cate-
gorized as vulnerable and six species as endangered (Table 8). Bombacopsis
quinata, considered as vulnerable, was not encountered in our plots. We also
identified seven candidate species that could be included in IUCN red list
directory in the future (Table 8).

[458]
 1519

125 125
Deciduous-94 Deciduous-00
100 100
Abundance

Abundance
75 75

50 50

25 25

0 0
0 10 20 30 40 50 0 10 20 30 40 50
Species rank Species rank

100 100
Gallery-94 Gallery-00
75 75
Abundance

Abundance
50 50

25 25

0 0
0 10 20 30 40 0 10 20 30 40
Species rank Species rank

Figure 2. Species abundance plots for dry deciduous and gallery forests inventoried in 1994 and
2000.

Discussion

The number of families, genera and species reported in the present study lies
within the range reported earlier in most Neotropical dry forests. For example,
Gentry (1988) reported 35 families and 55 species per hectare in a gallery forest
in Guanacaste, Costa Rica and Sabogal and Valerio (1998) reported on
average 44 species per hectare in Chacocente dry deciduous forest in Nicara-
gua. The most common family in the deciduous forest was Fabaceae/Papilio-
noideae with 10 species, a pattern common in most Neotropical dry forests
(Gentry 1988). This result also coincides with a study carried out in Central
America where Fabaceae was found to be the dominant tree and shrub family
in six of seven sites studied (Gillespie et al. 2000). Gillespie et al. (2000) made
an inventory in Chacocente and found the same common species as in the
present study. However, the present study found L. minimiflorus and C. exo-
stemma as common species. In tropical dry forest across the north central
Yucatan, the following important natural forest species were reported: Bursera
simaruba, Caesalpinia gaumeri, Gymnopodium floribundum and Piscidia piscip-
ula (White and Hood 2004), which are also encountered in our study. It was
observed that some species recorded in the first inventory (1994) were missing
in the subsequent inventory (2000) while new species were encountered in the
second inventory. Given the large number of species with 1 or 2 individuals,

[459]
1520

unknown
carvings
ornamental
timber
fruits
construction
fodder
firewood
live fence
Charcoal

0 10 20 30 40 50
No. species

1994 unknown
2000 carvings
ornamental
timber
fruits
construction
fodder
firewood
live fence
Deciduous Charcoal

0 50 100 150 200 250 300

Abundance (individuals/2ha)

1994 unknown
2000 carvings
ornamental
timber
fruits
construction
fodder
firewood
live fence
Gallery Charcoal

0 20 40 60 80 100 120 140 160

Abundance (individuals/2ha)
Figure 3. Local uses of tree species and their abundance in dry deciduous and gallery forests
in Chacocente, Nicaragua.

[460]
 1521

Table 7. Diversity measures for trees ‡10 cm dbh in the dry deciduous and gallery forests
inventoried in 1994 and 2000 on 2 ha plots.

Diversity Measures Deciduous Gallery

1994 2000 1994 2000

No. of individuals recorded in 2 ha plots 902 979 597 566

Total number of species recorded 59 59 55 58
Rate of species increase per individual enumerated (S/N) 0.065 0.060 0.092 0.102
Margalef ’s index of species richness (DMg = (S
1)/ln N) 8.52 8.42 8.45 8.99
Shannon’s measure of evenness (J¢ = H¢/lnP S) 1.16 1.15 1.17 1.10
Shannon–Wiener’s diversity index (H¢ =
pi log2pi) 4.71 4.69 4.69 4.48
The reciprocal of Simpson’s index (1/D) 17.1 16.9 14.6 14.5
Fisher’s index of diversity (a = N(1
x)/x) 14.1 13.8 14.8 16.2

Table 8. List of threatened species and suggested candidate species for future IUCN listing.

Species Status

Bombacopsis quinata Vulnerable

Cedrela odorata Vulnerable
Dalbergia retusa Vulnerable
Maclura tinctoria Vulnerable
Swietenia humilis Vulnerable
Esenbeckia litoralis Endangered
Guaiacum sanctum Endangered
Lonchocarpus minimiflorus Endangered
Lonchocarpus phlebophyllus Endangered
Platymiscium pleiostachyum Endangered
Zanthoxylum belizense Endangered
Albizia caribaea Candidate
Celtis caudate Candidate
Diospyros nicaraguensis Candidate
Hymenaea courbaril Candidate
Jacquinia aurantiaca Candidate
Manilkara achras Candidate
Phyllostylon brasiliense Candidate

illegal cutting might have caused the disappearance of this species. However,
the plausible explanation for the appearance of species in the second inventory
could be ascribed to the transition from seedling class in the 1994 inventory to
higher class (trees ‡ 10 cm dbh) in the subsequent inventory.
Tree species richness is difficult to compare for different sample sizes and
geographical variation (Murphy and Lugo 1995). Dry forest at Palo Verde on
the Pacific side of Costa Rica had approximately 52 tree species per hectare
(Murphy and Lugo 1995). Lower values have been found in the drier areas and
particularly in insular forests, such as in Southwestern Puerto Rico near
Guanica where 30–50 tree species per hectare were found. Gentry (1995) re-
ported an average of 65 tree species per ha in 23 Neotropical dry forests, which

[461]
1522

is considerably higher than the number of species found in the deciduous forest
of Chacocente. However, Gentry’s data set included individuals with dbh
‡2.5 cm. In addition, dry forests of Chacocente have a history of severe
selective logging which may be the main factor causing the low number of
species in this forest. Gillespie and Jaffré (2003) compared species richness in
seven different countries using 1000 m2 area and found that species richness is
high in Chamela-Mexico (89), Quiapaca (86) and Chaquimayo-Bolivia (79).
The lower species richness was found in Mudumalai (India) with 15 species.
The number of species for trees ‡10 cm dbh ranged from 3 to 28 species with a
mean value of 16 species per hectare in the Vindhyan dry tropical forest of
India (Sagar and Singh 2005). As a whole, the total number of species recorded
in the present is comparable with other tropical dry forests.
With regard to stem density and basal area, our result lies with the range of
values reported earlier for other tropical dry forests, and in some cases com-
parably higher. For example, Sabogal and Valerio (1998) reported 389 trees/ha
with a basal area of 14.48 m2/ha in the Chacocente dry deciduous forest,
Rundel and Boonpragob (1995) reported 20–88 trees/ha and a basal area
ranging from 7 to 42 m2/ha for tropical dry forest in Thailand. For tropical dry
forest at the north central Yucatan, White and Hood (2004) documented the
basal area in two sites as 20.7 m2/ha and 28.4 m2/ha. Gillespie and Jaffré (2003)
inventoried two tropical dry forests of New Caledonia and found the following
basal area per hectare for each site: Ouen-Toro 32.7 m2/ha and Pindai 32.3 m2/
ha. Gillespie and Jaffré (2003) also pointed out that tropical dry forests in the
Neotropics have greater structural similarity. In the present study, the similarity
in species composition and abundance between dry deciduous and gallery for-
ests was low. The stem density was much higher in the dry deciduous forest
while the basal area was much greater for the gallery forest. This could be
related to better soil moisture condition in the latter than the former, as
moisture is the major environmental factor limiting tree growth in dry areas.
A variety of diversity measures were computed to describe the heterogeneity
of the two forest communities, and it was Simpson’s index and Fisher’s a that
consistently differentiated the two communities. According to Simpson’s
dominance index the dry deciduous forest is more diverse than the gallery
forest. This could be related to the relatively large number of abundant species
in the deciduous forest than the gallery forest (cf. 7 species in dry deciduous
and 4 species in gallery forests with abundance >24 individuals/ha). In 1994
and 2000 inventories of the dry deciduous forest, Gyrocarpus americanus,
Lonchocarpus minimiflorus, Stemmadenia obovata, Tabebuia ochracea and
Caesalpinia exostemma represented 49% and 42% of the total individuals per
hectare, respectively. While Trichilia hirta, Capparis pachaca, Thouinidium
decandrum and Stemmadenia obovata represented 44% and 45% of the total
individuals per hectare found in the gallery forest during 1994 and 2000
inventories, respectively.
Fisher’s diversity index showed that the gallery forest is more diverse than the
dry deciduous forest. One implication of this finding would be the majority of the

[462]
 1523

species inventoried have irregular and clumped spatial distribution in the

deciduous forests, and therefore the gallery forest is characterized by high alpha
diversity. The Shannon–Wiener diversity index is usually found to fall between
1.5 and 3.5 and only rarely surpasses 5.0 (Magurran 2004). The values of
Shannon–Wiener index for the deciduous and gallery forests falls within the
expected range. Gentry (1988) reported values of Shannon–Wiener diversity
index for two sites of 0.1 ha in Nicaragua; Cerro Olumo with 5.80 and Cerro El
Picacho with 5.22 (Clould forests). These values are higher than the present
study; however, the precipitation and altitude of these sites are higher than
Chacocente. Hence, the larger diversity of these sites may be due to higher
precipitation and low temperature favoring growth and survival of more species.
Almost 50% of Nicaragua’s population lives around tropical dry forest and
in spite of the partial protection of Chacocente, human pressure on the
remaining forests are obvious. The result from the present study provides
evidence that the abundance of species used for firewood and timber declined
from 1994 to 2000 in both forest communities. For example Cordia alliodora,
though very common elsewhere in Nicaragua; was one of the rarest species in
Chacocente. Even the abundance of the ‘‘not used’’ species in the deciduous
forest showed a decreasing tendency, indicating an on-going disturbance
(anthropogenic and natural) in the forest reserve and loose protection of the
forest reserve. It is also important to note that species with high commercial
values, such as Swietenia humilis, Cedrela odorata, Dalbergia retusa and
Guaiacum sanctum were among the rarest species in our study. Illegal logging
of commercially valuable species is still a common problem in the whole
country. Apparently, the continuing loss of biodiversity is attributed to mainly
deforestation, as in the case of many tropical dry forests (Thiollay 2002). If this
anthropogenic disturbance is not curved as early as possible, the tree species,
particularly commercial species that appear almost absent in the forest, will
become locally extinct (e.g., Bombacopsis quinata). As both forest communities
are characterized by many species with few individuals, active management,
such as reintroduction of threatened species (Table 8) on private and com-
munity lands is highly desirable to maintain viable populations. Simulta-
neously, an immediate action should be taken to assist the natural regeneration
process to restore species diversity of the remaining relics of tropical dry forest
in the country. We further recommend a detailed assessment of the suggested
candidate species for IUCN listing to determine their conservation status.

Acknowledgements

We thank Claudio Calero for his support during the fieldwork. Also he made a
great contribution with his knowledge about species uses. Thanks also to Ali
Water and Alvaro Noguera for their help during the fieldwork. The Ministry of
Natural Resources and Environment kindly allowed us to carry out this
research at Chacocente National Wildlife Refuge. The study was financed by

[463]
1524

the Swedish International Development Cooperation Agency (UNA-SLU PhD

Program). Finally, the anonymous reviewer is highly appreciated for the
valuable and constructive comments.

Appendix

Appendix. List of tree species ‡10 cm dbh recorded in 1994 and 2000 inventories in dry deciduous
and gallery forests in Chacocente Wildlife Refuge, Nicaragua together with their uses (CH –
charcoal; LF – live fence; FW – fire wood; FOD – fodder; RC – rural construction; FRU – fruit; W
– timber; P – pole; O – ornamental; HC – handicrafts and carvings).

Species Family Deciduous Gallery Uses

1994 2000 1994 2000

Acacia costaricensis Mimosoideae * * FW, CH

Achatocarpus nigricans Achatocarpaceae * * FW, RC
Adelia barbinervis Euphorbiaceae * NO
Albizia caribaea Mimosoideae * * FW, P, CH
Allophylus psilospermus Sapindaceae * * FW, CH, RC
Annona reticulata Annonaceae * * FRU
Ardisia revolute Myrsinaceae * * NO
Astronium graveolens Anacardiaceae * * * * P, W
Bixa orellana Bixaceae * * NO
Brosimum alicastrum Moraceae * * W, FW, CH, HC
Bunchosia cornifolia Malpighiaceae * * FW, RC, CH
Bursera simaruba Burseraceae * * * LF, P
Caesalpinia coriaria Caesalpinoideae * * FW
Caesalpinia exostemma Caesalpinoideae * * * NO
Caesalpinia violacea Caesalpinoideae * * * FW, LF, CH
Calycophyllum candidissimum Rubiaceae * * * * FW, CH, O
Capparis odoratissima Capparidaceae * * FW
Capparis pachaca Capparidaceae * * * * NO
Casearia corymbosa Flacourtaceae * FW
Casearia tremula Flacourtaceae * * * FW
Cecropia peltata Cecropiaceae * * NO
Cedrela odorata Meliaceae * * W, FW, HC
Ceiba pentandra Bombacaceae * * W
Celtis caudata Ulmaceae * * * * NO
Chomelia spinosa Rubiaceae * * NO
Coccoloba caracasana Polygonaceae * * NO
Coccoloba floribunda Polygonaceae * * NO
Coccoloba sp. Polygonaceae * FW
Cordia alliodora Boraginaceae * * * * P, HC, W
Cordia collococca Boraginaceae * * * * CH, HC, W
Cordia dentata Boraginaceae * * * P, LF
Cordia gerascanthus Boraginaceae * * * * CH, HC, W
Coursetia elliptica Papilionoideae * * NO
Croton niveus Euphorbiaceae * * FW
Dalbergia retusa Papilionoideae * * FW, W, FOD

[464]
 1525

Appendix. Continued

Species Family Deciduous Gallery Uses

1994 2000 1994 2000

Diospyros nicaraguensis Ebenaceae * * FW, RC

Erythroxylum havanense Erythroxylaceae * * FW
Esenbeckia litoralis Rutaceae * * * * FW
Ficus obtusifolia Moraceae * O
Gliricidia sepium Papilionoideae * * FW, FOD, W
Guaiacum sanctum Zigophyllaceae * * P, HC, W

Guarea glabra Meliaceae * * CH, P, HC, W

Guazuma ulmifolia Sterculiaceae * * * * FW, CH, FOD
Gyrocarpus americanus Hernandiaceae * * * * NO
Haematoxylon brasiletto Caesalpinoideae * * FW
Hippocratea rovirosae Hippocrataceae * * W, RC
Hymenaea courbaril Caesalpinoideae * * HC, W
Inga sp. Mimosoideae * *
Jacquinia aurantiaca Theophrastaceae * * FW
Karwinskia calderonii Rhamnaceae * * * * W, RC
Licania arborea Chrysobalanaceae * W, HC
Lonchocarpus minimiflorus Papilionoideae * * FW, FOD, W
Lonchocarpus phlebophyllus Papilionoideae * * FW, CH, W
Lonchocarpus sp Papilionoideae * * NO
Luehea candida Tiliaceae * * RC, W
Luehea seemannii Tiliaceae * * FW, CH, P, W
Lysiloma divaricatum Mimosoideae * * FW, CH, RC
Lysiloma sp Mimosoideae * * FW, CH, RC
Machaerium biovulatum Papilionoideae * * FW
Maclura tinctoria Moraceae * * FW, CH, P, W
Manilkara achras Sapotaceae * * FW, P, W
Malpighia stevensii Malpighiaceae * * FW, LF
Myrospermum frutescens Papilionoideae * * * * W, RC
Pithecellobium dulce Mimosoideae * * * FW, CH
Pithecellobium saman Mimosoideae * * * FW, FOD, HC, W
Phyllostylon brasiliense Ulmaceae * * FW, CH
Piper aduncum Piperaceae * NO
Pisonia macranthocarpa Nyctaginaceae * * NO
Platymiscium pleiostachyum Papilionoideae * * FW
Pterocarpus rohrii Papilionoideae * * FW, LF, W
Randia armata Rubiaceae * * FW
Randia cookii Rubiaceae * * FW
Randia nicaraguensis Rubiaceae * FW
Sapranthus nicaraguensis Annonaceae * * * * NO
Senna atomaria Caesalpinoideae * * * * FW, CH
Simarouba glauca Simaroubaceae * * W, CH
Spondias purpurea Anacardiaceae * * * * NO
Spondias sp. Anacardiaceae * * NO
Stemmadenia obovata Apocynaceae * * * * FW
Sterculia apetala Sterculiaceae * * FW
Swietenia humilis Meliaceae * * W, CH, HC

[465]
1526

Appendix. Continued

Species Family Deciduous Gallery Uses

1994 2000 1994 2000

Tabebuia ochracea Bignonaceae * * * * FW, LF, W

Tabebuia rosea Bignonaceae * FW, LF, W
Terminalia oblonga Combretaceae * * FW, HC, W
Tetrorchidium rotundatum Euphorbiaceae * * NO
Thouinidium decandrum Sapindaceae * * * * FW, CH, HC
Trema micrantha Ulmaceae * FW, CH, P
Trichilia hirta Meliaceae * * * W, FW
Trichilia moschata Meliaceae * * * * P, W
Triplaris melaenodendron Polygonaceae * * P, LF
Ximenia americana Olacaceae * * * * NO
Zanthoxylum belizense Rutaceae * * W, RC
Zanthoxylum caribaeum Rutaceae * * * * NO
Ziziphus guatemalensis Rhamnaceae * * W, FW, P
Zuelania guidonia Flacourtiaceae * * NO

References

Anonymous 2002. Plan de manejo del refugio de vida silvestre Rı́o Escalante- Chacocente. Min-
isterio de Recursos Naturales y del Ambiente. Ramboll/Posaf. Managua, Nicaragua.
Duque A. and Cavelier J. 2003. Strategies of tree occupation at a local scale in terra firme forests in
the Colombian Amazon. Biotropica 35: 20–27.
Gentry A.H. 1988. Changes in plant community diversity and floristic composition on environ-
mental and geographical gradients. Ann. Missouri Bot. Garden 75: 1–34.
Gentry A.H. 1995. Diversity and floristic composition of Neotropical dry forest. In: Bullock S.H.,
Mooney H.A. and Medina E (eds), Seasonally Dry Tropical Forests. Cambridge University
Press, Cambridge, UK, pp. 9–34.
Gerhardt K. 1994. Seedling Development of Four Tree Species in Secondary Tropical Dry Forest
in Guanacaste, Costa Rica. PhD thesis, Uppsala, Sweden.
Gillespie T.W., Grijalva A. and Farris C.N. 2000. Diversity, composition, and structure of tropical
dry forests in Central America. Plant Ecol. 147: 37–47.
Gillespie T.W. and Jaffré T. 2003. Tropical dry forests in New Caledonia. Biodiv. Conserv. 12:
1687–1697.
Harcourt C.S. and Sayer J.A. 1996. The Conservation Atlas of Tropical Forest. The Americas,
Simon & Schuster.
IUCN. 2004. 2004 IUCN Red List of Threatened Species. www.iucnredlist.org. Downloaded on 14
December 2004. International Union for Conservation of Nature and Natural Resources,
Cambridge, UK.
Janzen D. 1988. Management of habitat fragments in a tropical dry forest: growth. Ann. Missouri
Bot. Garden 75: 105–116.
Krebs C.J. 1999. Ecology Methodology. 2nd ed. Addison-Wesley Educational Publishers. Inc.
Lamprecht H. 1989. Silviculture in the Tropics: Tropical Forest Ecosystems and their Tree Species.
GTZ, Eschborn.
Magurran A.E. 2004. Measuring Biological Diversity. Blackwell Publishing, Malden, Oxford and
Victoria.
Murphy P.G. and Lugo A.E. 1986. Ecology of tropical dry forest. Ann. Rev. Ecol. Syst. 17: 67–88.

[466]
 1527

Murphy P.G and Lugo A.E. 1995. Dry forest of Central America and the Caribbean. In: Bullock
S.H., Mooney H.A and Medina E (eds), Seasonally Dry Tropical Forests. Cambridge University
Press, Cambridge, England, pp. 9–34.
Roldan H. 2001. Recursos forestales y cambio en el uso de la tierra, Republica de Nicaragua.
Santiago, Chile.
Rundel P.W. and Boonpragob K. 1995. Dry forest of Central America and the Caribbean. In:
Bullock S.H., Mooney H.A. and Medina E (eds), Seasonally Dry Tropical Forests. Cambridge
University Press, Cambridge, England, pp. 93–119.
Sabogal C. 1992. Regeneración de bosques secos naturales en Centro América, con ejemplos de
Nicaragua. J. Veget. Sci. 3: 407–416.
Sabogal C. and Valerio L. 1998. Forest Composition, Structure and Regeneration in a dry forest of
the Nicaraguan Pacific Coast. In: Dallmeier F and Comiskey J.A. (eds), Forest Biodiversity in
North Central and South America, and the Caribbean: Research and Monitoring. Man and The
Biosphere Series, Vol. 21. UNESCO, New York pp.187–212.
Sagar R. and Singh J.S. 2005. Structure, diversity, and regeneration of tropical dry deciduous forest
of northern India. Biodiv. Conserv. 14: 935–959.
Salas J.B. 1993. Arboles de Nicaragua. Instituto Nicaragüense de Recursos Naturales y del Am-
biente. Servicio Forestal Nacional., Managua, Nicaragua.
Segura O., Kaimowitz D. and Rodrı́guez J. 1997. Polı́ticas forestales en Centro América: Análisis
de las restricciones para el desarrollo del sector forestal. IICA-Holanda/LADERAS.
Stevens W.D., Ulloa C.U., Pool A. and Montiel O.M. 2001. Flora de Nicaragua. Missouri
Botanical Garden Press.
Tercero M.G. and Urrutia G.S. 1994. Caracterización florı́stica y estructural del bosque de galerı́a
en Chacocente, Carazo, Nicaragua. Trabajo de diploma. Universidad Nacional Agraria,
Managua, Nicaragua.
Thiollay J.M. 2002. Forest ecosystems: threats, sustainable use and biodiversity conservation.
Biodiv. Conserv. 11: 943–946.
White D.A. and Hood C.S. 2004. Vegetation patterns and environmental gradients in tropical dry
forests of the northern Yucatán Peninsula. J. Veget. Sci. 15: 151–160.

[467]
Biodiversity and Conservation (2006) 15:1529–1543
Springer 2006
DOI 10.1007/s10531-004-6678-1

Identifying conservation priority zones for effective

management of tropical forests in Eastern Ghats
of India

B. BALAGURU1, S. JOHN BRITTO, S.J.2,*, N. NAGAMURUGAN3,

D. NATARAJAN4 and S. SOOSAIRAJ5
1
Department of Botany, Jamal Mohamed College, Tiruchirappalli 620 020, Tamil Nadu, India;
2
Rapinat Herbarium, St. Joseph’s College (Autonomous), Tiruchirappalli 620 002, Tamil Nadu,
India; 3Department of Biotechnology, Kurinji College of Arts and Science, Tiruchirappalli 620 024,
Tamil Nadu, India; 4Department of Microbiology, Kandaswami Kandar’s College, P. velur Namakkal
638 182, Tamil Nadu, India; 5Department of Botany, St. Joseph’s College (Autonomous), Tiruchi-
rappalli 620 002, Tamil Nadu, India; *Author for correspondence (e-mail: sjcbritto@rediffmail.com;
phone: +91-431-2721-304; fax: +91-431-2701-501)

Received 3 December 2003; accepted in revised form 2 November 2004

Key words: Conservation priority zones, Eastern Ghats, Endemism, Geographical information
system, Red listed plants, Shervarayan hills, Tropical forest

Abstract. There are thousands of protected forest areas existing on earth, yet the deforestation rate
continues unabated both inside and outside the protected areas especially in the tropical forests. It
identifies the less effectiveness of the current conservation strategies, which is normally oriented
around the forest area cover rather than the quality of the protected areas. This calls for realistic
and effective management strategies for forests. Based on the drawbacks the present study aims at
identifying conservation priority sites within the protected areas (Reserved Forests) of Shervarayan
hills, Eastern Ghats of Tamil Nadu, India. The remnant forest patches having less effective man-
agement/protection is identified and analysed for its qualitative contribution to the ecosystem.
Quadrats of 20 · 20 m were laid in different vegetation based on the percentage of forest cover and
assess the species diversity pertaining the richness, Endemism and Red list categories. Thematic
layers (maps) such as vegetation type, floristic species richness, floristic endemism, and red list flora
are created and categorized according to their weightage classes and overlaid in GIS domain to
demarcate the Conservation Priority Zones (CPZ). The CPZ are classified according to the priority
status i.e., high, moderate and low based on the contributing species richness, levels of endemism
and concentration of Red listed plants.

Introduction

The present global biodiversity is diminishing at an accelerated pace (Myers

1980; Wilson 1988) especially in the tropical countries (Hamilton 1984; Bowles
et al. 1998; Malcolm and Ray 2000) where the biodiversity is concentrated. The
current status of our forest resources has called for conservation planning
(Mooney and Chapin 1994; Western and Wright 1994; Calridge and
O’Callaghan 1997; O‘Neill et al. 1997; Bawa and Seidler 1998) which seeks to
identify spatial options for the preservation of biodiversity (Williams et al.
1996). The ultimate purpose of conservation is to inform and affect the

[469]
1530

conservation policy (Robertson and Hull 2001). Within the realm of conser-
vation measures, the forest strategists have identified and conserved large tracts
of vegetation as protected areas (Gaston et al. 2002; Margules et al. 2002). Still
the deforestation rate has markedly increased (Downton 1995) and has spread
to the protected areas of tropical region too (Hamilton 1984; Howard 1991;
Redford 1992) rendering ineffectiveness in arresting it. Ecologists nowadays are
on the consensus that biological diversity is not effectively conserved by re-
serves alone (Wilcove 1989). Various quantitative methods that allow relatively
expeditious identification of conservation-priority areas have been proposed in
recent years and these approaches include identification of hotspots of biodi-
versity (Myers 1988, 1990; Dobson et al. 1997), rapid biodiversity assessment
(Oliver and Beattie, 1993 and 1996), identification of indicator and surrogate
species (Curnutt et al. 1994), development or rarity and complementary sets
(Williams et al. 1996), identification of key eco-region (Olson and Dinerstein
1998), and cost-minimizing or land-values analyses (Ando et al. 1998).
This may be due to the very size of the forest tract whereby the porosity of
the protected forest will lead to its ineffective management. Now, it is better to
identify the quality of the vegetation in the protected and non protected areas,
rather than the area size for effective conservation management (Sheil 2001).
Most often we had adopted the conventional approach to maintain biological
diversity by following a protocol based on species by species and threat-by-
threat approach, but it too has its own detriments i.e., the financial drawbacks,
inaccurate complicated database of the forest community (Hutto et al. 1987;
Scott et al. 1987, 1991; Margules 1989, Noss 1991) etc.
In recent years the focus for conservation has shifted from single species
management approach to protection of biodiversity in the aggregate i.e., to
maintain the native plant species in extensive natural landscapes (habitats)
restricting to a minimal size factor, that are sufficiently linked (i.e. corridors) to
allow interaction and genetic interchange among distinct populations (Noss
1983). This approach requires a cohesive and representative system of areas to
be managed for the maintenance of biodiversity. Hence there is a need to
prioritize only those areas, which are considered most essential for conserva-
tion, which are termed as biodiversity priority areas (Olson and Dinerstein
1998). The procedures involve scoring and ranking, which make priority setting
more systematic and explicit (Margules et al. 2002). Prioritization of strategies
is essential to ensure that efforts at conservation yield best possible results and
undesirable side effects, such as the alienation and impoverishment of local
communities can be avoided (Singh and Taneja 2000). Prioritization of sites for
conservation also needs to be done with reference to the (often least studied)
vegetation type (Williams et al. 2002), species richness (Terborgh and Winter
1983; Scott et al. 1987), endemism based on Kier and Barthlott (2001) and
concentration of red listed plants (Ahmedullah 2000; Kumar et al. 2000).
The methods for identifying priority areas vary with the entity selected for
the overall biological conservation planning and management (Margules et al.
2002; Gaston et al. 2002) and for example Ramesh et al. (1997) have suggested

[470]
 1531

conservation priority based on the biodiversity gaps by considering the vege-

tation uniqueness, species richness, endemic flora and endemic fauna in Wes-
tern Ghats, whereas on the other hand Menon et al. (2001) and Amarnath et al.
(2003) have identified conservation priority zones based on the land use
changes, vegetation patch characteristics, phytosociological data, topographic,
bioclimatic and disturbance level in wet evergreen forests of Western Ghats in
Tamil Nadu.
Thus the present study has deviated from the approaches described above
and have considered a new concept with the vegetation type, species richness,
endemic and IUCN red listed plants as base for identifying Conservation
Priority Zones (CPZ) in GIS (Geographical Information System) domain.

Study area

The Shervarayan hills (a part of Eastern Ghats) are located in the northern
part of Salem city, Tamil Nadu, South India and with an area of 469.9 km2.
The study area lies between latitudes 1143¢00¢¢ to 1200¢00¢¢ N and longitudes
of 7800¢00¢¢ to 7822¢30¢¢ E (Map 1) and falls in the Survey of India toposheets
(SOI) 581/1, 2, 5 and 6 (i.e., 1: 50,000 scales). The mean annual rainfall at the
upper hill reaches is 1638 mm and 850 mm at the foothills. The temperature

Map 1. Study area.

[471]
1532

ranges from 13–29C on the hill plateau to 25C and 40C at the foothills. The
soil is red loamy and lateritic. The area is made up of Archaean crystalline rock
like amphibolites, leptinites, garnetiferous granites and charnockites. Bauxite
and Magnesite are the chief mineral resources in the Shervarayan hills. There
are 71 villages, which are administrated by two taluks (political unit equivalent
of an English county) i.e., Yercaud and Omalur. Most of the hill plateau is in
private ownership, which includes coffee estates, villages and their agricultural
lands. Colonial planters had been maintaining and harvesting the coffee estates
till the time of independence of the country and later, the ownership has been
entrusted to the natives. There are 45 reserved forests, which are administered
by the Salem Forest Division. Almost all the reserved forest area is on the outer
slopes of the hill tract facing the human habitats on the fringing foothills
thereby enhancing the proneness to deforestation and very much is the evident
fact.

Methodology

Mapping vegetation type

Vegetation type map of Shervarayan hills (Balaguru et al. 2003) is used which
covers nearly half (49.50%) of the hill area (23260.76 ha) under reserved forests
comprising about six major forest types - evergreen (111.33 ha), semi evergreen
(1057.67 ha), riparian (1145.15 ha), dry mixed deciduous (10179.10 ha),
southern thorn scrub (10735.70 ha), and evergreen scrub (31.81 ha), respec-
tively (Map 1). To evolve potential conservation priority elements, the virgin
and primary forest patches comprising the evergreen, semi evergreen, riparian,
and dry mixed deciduous forests are used as the base, while the evergreen scrub
and southern thorn scrub forests are excluded due to their highly degraded
nature. The scores for each forest type are attributed according to the species
concentration (Figure 1) and substituted to all the representing polygons
accordingly.

Mapping floristic richness

Representative polygons for each forest type are analysed for assessing species
richness contribution by adopting quadrat method (20 · 20 m) (CES 1998;
Ferreira and Prance 1998). This study has taken optimum sampling quadrats
to cover all variations within each type of the vegetation and the number of
quadrats for each forest type is based on the area percentage of the forest cover
(>1000 ha area 0.5%; 1000–2000 (0.5%) and >2000 (0.01%). All living plant
species within the quadrat are identified and the number of species in each
forest type is summed and represented by species richness values and these

[472]
 1533

Figure 1. Conceptual diagram illustrating the building identification of conservation priority

zone.

values are attributed or extrapolated to all such polygons representing the

respective forest type. To produce the species richness maps the ensuing
polygons are regrouped and classified into categories of low, medium and high
according to the ranges of species richness values.

Mapping floristic endemism

The plant species thus collected in the quadrats are identified with the endemic
flora of Peninsular India as enlisted by Ahmedullah and Nayer (1986). Pro-
cedures for deciding on CPZ need more systematic and explicit approach for
priority setting wherein multiple criteria are given scores. These scores are then
combined and ranked accordingly and priority (high, moderate or low) is given
to those areas (Margules et al. 2002). The number of endemic species are
allocated to the respective scores/classes based on their significant status in

[473]
1534

Indian context (Roy 1999; Ajith Kumar et al. 2000) i.e. individual species
endemic to India is considered as ‘Indian Endemic’ in distribution, hence they
received low score (1), similarly individual species endemic to peninsular India
is considered as ‘Regional Endemic’ (2) and species endemic to Eastern Ghats
(including Shervarayan hills) is ‘Local Endemic’ and received the highest score
(3). The number of species and their scores in each of the polygon is then
summed up and values attributed as described for species richness. To produce
endemic species map, the polygons are finally regrouped/reclassified into low,
medium and high degree of endemism according to the summed values
attributed to each polygon (see Figure 1), the polygon with the highest score
had the high degree of endemism and likewise.

Mapping floristic red listed plants

The methodology to map red listed plants is the same as for the endemic plants
map or species richness map. The ensuing plant species in the quadrats are
evaluated based on version 3.1; IUCN/SSC (1999) criteria and cross checked
with Indian Red Data books (Nayer and Sastry 1987–1990) and other relevant
literature Kumaravelu and Chaudhuri 1999). The red listed categories and their
scores are classified into (a) Critically Endangered (CE)
5; (b) Endangered
(En)
4; (c) Vulnerable (VU)
3; (d) Lower risk (LR)/Least Concern (Lc)
2;
(e) Data Deficient (DD)
1 (Table 2). To produce the red listed plant species
map, the polygons are finally grouped into low, medium and high wherein the
scoring is similar to the one adopted for the endemic classification.

Modelling conservation priority zone (CPZ)

The components of various units (classes) from the thematic maps like the
vegetation type, floristic richness, endemism and red lists with their respective
weightages (Figure 1) are essential to develop conservation priority zones for
this study. Considering the conservational importance and status for each class
(unit) of the respective thematic maps, the classes are given weightages to
designate and identify the CPZ. Overlay or superimposition creates a com-
posite output GIS file by combining a number of input GIS files based on the
minimum or maximum values of the input files (Murthy 2000). To prepare the
CPZ map, the respective thematic maps (species richness map, red list map and
finally endemism map) are overlaid on the vegetation type map, which com-
prised the lowermost tier (the base map) using a remote sensing and GIS
software (ERDAS imagine). The model maker (a tool in the ERDAS software)
is used to analyse the overlays, wherein the different features of the thematic
layers are intersected/extracted and new class values are attributed to the
resulting polygons. The polygons are classified according to the conservation
priority status and finally integrated (union criteria in model maker) to generate

[474]
 1535

the CPZ. The authenticity of the areas/zones proposed for conservation pri-
ority is confirmed with ground truthing.

Results

Totally 322 species are recorded from the Shervarayan hills (based on the
quadrats studied in the study area), of which 24 species are endemic (Table 1)
and 23 species are red listed (Table 2). The floristic richness (Map 2) are re-
grouped/reclassed into high (>80 species), medium (40–80 species) and low
(>40 species) rich areas respectively. The endemism and red listed species are
grouped into three zones based on the number of contributing species. The
CPZ map (Map 3) is generated with three classes according to the criteria
described before, based on the combination of scores – high, moderate and low
priority zones.

High priority zone

High priority zone is distributed in five sites with moderate to high species
richness. This zone accounts for 1582.53 ha (6.80%) of the total hill forest area.
The priority sites are authenticated with the presence of select/target species
(under different criteria) like Rubia cordifolia, Crotalaria shevaroyensis, Litsea
oleoides, Smilax zeylanica, Ixora notoniana, Neolitsea scrobiculata, Psychotria
octosulcata, Randia candolleana var. candolleana, Peperomia dindigulensis,
Celastrus paniculatus and Nothopegia colebrookiana in the evergreen forests.
The riparian forests comprise Terminalia arjuna, Mangifera indica, Ficus mi-
crocarpa and Syzygium cumini and on the other hand the semi evergreen forests
is represented mainly by Nothopegia colebrookiana, Celastrus paniculatus, De-
calepis hamiltonii, Santalum album, Naravelia zeylanica, Gymnema sylvestre,

Table 1. Endemic plant species and their distribution status.

Distribution Species Name

Local Endemic Crotalaria shevaroyensis

(endemic to Eastern Ghats)
Regional Endemic Peperomia dindigulensis, Vaccinium neilgherrense Miliusa
(endemic to Peninsular India) eriocarpa, Litsea oleoides, Neolitsea scrobiculata, Curcuma
neilgherrensis, Eranthemum capense, Dolichandrone arcuata,
Indian Endemic Neonotonia wightii, Elaeagnus indica, Decalepis hamiltonii,
(endemic to India) Jasminum trichotomum, Ixora notoniana, Pavetta blanda,
Psychotria octosulcata, Randia candolleana var. candolleana,
Wendlandia angustifolia, Mallotus stenanthus, Tetrastigma
sulcatum, Pamburus missionis, Leucas diffusa, Shorea
roxburghii, Chionanthus mala-elengi.

[475]
1536

Table 2. Red listed plant species and their status.

Species Name Red listed categories

Buchanania lanzan Lower risk

Celastrus paniculatus Vulnerable
Cycas circinalis Threatened
Decalepis hamiltonii Endangered
Gloriosa superba Lower risk
Nothopegia colebrookiana Data Deficient
Pseudarthria viscida Lower risk
Santalum album Endangered
Sapindus emarginata Lower risk/Least concerned
Smilax zeylanica Vulnerable
Terminalia arjuna Lower risk
Gardenia gummifera Endangered
Michelia champaca Vulnerable
Symplocos cochinchinensis Lower risk
Rubia cordifolia Critically endangered
Gnetum edule Endangered
Naravelia zeylanica Vulnerable
Hemidesmus indica Vulnerable
Withania somnifera Vulnerable
Stephnia japonica Vulnerable
Evolvulus alsinoides Lower risk
Gymnema sylvestre Vulnerable
Vernonia arborea Endangered
Polystachya concreta Endangered

Ixora notoniana, Pseudarthria viscida, Buchanania lanzan, Hemidesmus indicus

and Sapindus emarginata.
The above described forests also are characterized with the occasional
presence of some priority species like Withania somnifera, Hemidesmus indicus,
Celastrus paniculatus, Cycas circinalis and Symplocos cochinchinensis with en-
demic constraints.

Moderate priority zone

The zone occupies an area of about 6282.4 ha (27%) enclosing parts of ever-
green forests and dry mixed deciduous forests with species richness ranging
from moderate to low. The evergreen forest under this class includes the en-
demic and red listed species like Symplocos cochinchinensis, Vaccinium neil-
gherrense, Gnetum edule, Rubia cordifolia, Peperomia dindigulensis, Elaeagnus
indica and Curcuma neilgherrensis. The endemic and IUCN red listed plant
species in the dry mixed deciduous forests has both moderate and high richness
and the representing species are Withania somnifera, Naravelia zeylanica,
Dolichandrone arcuata, Hemidesmus indicus, Sapindus emarginatus,

[476]
 1537

Map 2. Species richness map of Shervarayan hills.

Pseudarthria viscida, Nothopegia colebrookiana, Pamburus missionis and

Evolvulus alsinoides.

Low priority zone

This zone with moderate to low species richness occupies an area of about
4524.92 ha (19.45%) of the total forest area and the zone comprises mostly of
the dry mixed deciduous types and to a lesser extent the riparian forests. The
red listed plant species in dry mixed deciduous forests species are Celastrus
paniculatus, Nothopegia colebrookiana, Pseudarthria visida and Hemidesmus
indicus and the select endemic species like Mallotus stenanthus, Pamburus
missionis, Shorea roxburghii and Pavetta blanda The riparian forest has only
one endemic and endangered plant species i.e. Cycas circinalis,

[477]
1538

Map 3. Conservation priority zones.

Discussion

Most of the forests on the outer slopes and plateau of Shervarayan hills are still
facing the wrath of deforestation in spite of its protected status. There are
multidimensional reasons to it and the size of the protected area is the first
detriment rendering the very base of protection as ineffective. Secondly it is
followed by easy accessibility to the forest patches by the illegal loggers wherein
the dense network of the footpaths crisscrossing the forest patches confirm the
same. Thirdly the ineffectiveness of the protection status is the poor knowledge
of conservation prior sites within the protection realms of the forest.
This study also identifies a similarity in species contribution between the
evergreen forests of Shervarayan hills (a part of Eastern Ghats) with that of the
evergreen forests of Western Ghats. Species like Chionanthus ramiflorus,

[478]
 1539

Ligustrum perrottetii, Olea paniculata, Vaccinium neilgherrense, Viburnum

punctatum, Gnetum edule, Elaeocarpus serratus, Syzygium cumini, Memecylon
edule, Symplocos cochinchinensis and Litsea deccanensis are common only in
Western Ghats, but are present in Shervarayan hills too (a trait unique to this
hill when compared to other hills in the Eastern Ghats) (Balaguru 2002). The
vegetation types like evergreen, riparian and semi-evergreen are potentially
most vulnerable owing to their proximity to the surrounding anthropogenic
environment (mining, coffee estates and human habitation) and are designated
with high conservation value, so as to effectively conserve the remnant forest
patches within the realm. These areas as discussed before harbor a number of
red listed and endemic species of conservation importance. Some of the ever-
green and semi-evergreen forests are inadequately represented on the outer
slopes whereas the dry mixed deciduous in the same zone is well represented
i.e., rich species diversity. However widespread logging in these areas may
deplete (in future) the existing forest cover and add to the deforestation extents.
Hence conservation of such areas too is included in conservation priority.
The CPZ map thus generated will help to concentrate the protection strategy
to the zones thus demarcated and help the forest department to have an
effective approach to conserve and maintain the virgin forests – a positive
approach which can be adopted elsewhere in similar forests. This study effec-
tively defends the sole purpose of selecting the virgin forests on Shervarayan
hills for conservation priority zone and its mapping for effective conservation
strategies. The present study identifies itself with similar studies by Menon
et al. (2001) wherein it is discussed that the conservation priorities require the
conservation value of an area and its vulnerability (proximity to human
interference in this case) towards deforestation. Fixing biodiversity priorities
(CPZ in this case) are necessary but in themselves are not sufficient for the
long-term maintenance. Biological diversity requires other tools, and ap-
proaches such as sustainable development (Peters et al. 1989; Hartshorn 1995)
and management prescriptions to minimize the risk of extinction of local plant
population, which have to be focused more sharply in such CPZ. More
effective strategy involves people’s participation, while realizing and ensuring
their domestic needs (fuel wood, fodder, minor forest produce including the
medicinal plants) (Margules et al. 2002). This will enforce a harmonious facet
to the whole process leading to the success of the strategy adopted (Serrao and
Homma 1993; Dawson 1996). What is required therefore is an appropriate
developmental paradigm that can provide a more relevant perception and an
interpretative framework from which such conservation strategies may emerge
(Upreti 1994). Such planning for the stabilization of natural ecosystem is
essential and this will reduce the pressure on the natural forests and prevent
further loss of biodiversity and in the longer run will reestablish the lost forest
stand. The development plans with Sustainable forest management would
enable the effective management of biodiversity in Tropical forests and is
effectively adopted in most revised cases for most of the policies and strategies
associated with forest.

[479]
1540

The potential utility of remote sensing and GIS to identify the CPZ in this
study and culmination of all aspects dealing with the sole purpose of conser-
vation has been effective and reliable (based on the ground truth and field
checks). The resultant maps gives a picture of the CPZ providing a birds eye
view of the areas thus identified. The accessibility to the zones thus identified
can be deciphered and planned, finally paving way for better and effective
conservation.

Conclusion

For identifying priority areas, there must be acceptable ways of measuring

biological diversity, a way of determining an acceptable level of representation
of that diversity in conservation areas. Having set that goal, it is necessary that
a cost effective way of allocating limited resources should be a thought of
criteria. The methods outlined in this paper have made the most effective use of
available field data with the remotely derived satellite data and involves
innovative scoring and ranking procedure that is developed and improved in
this study. As a result, priority setting has been approached systematically and
explicit tolerance. Though the conservation priority areas are geared towards
the future, the forest departments should advocate an alternate approach to
protected area management that would integrate biodiversity conservation
with social development. Such an approach would entail an improved under-
standing of the local pattern of resource use. As a result, the contemplated
conservation strategies would benefit the local population to enable security to
their local livelihood and the base of conservation.

Acknowledgements

The authors are grateful to D. Stoms, California University, Santa Barbara

and to Dr. B. R. Ramesh, Director of Research, French Institute, Pondicherry
for their expertise and assistance. We would also like to thank the Tamil Nadu
forest department and the committed officers of Salem Forest Division for the
permission to carry out study in the Shervarayan hill forests.

References

Ahmedullah M. and Nayer M.P. 1986. Endemic plants of the Indian Region (Vol. I) – Peninsular
India. Deep printers, New Delhi.
Ahmedullah M. 2000. Prioritisation of endangered plants of India. In: Singh S., Sastry A.R.K.,
Mehta R. and Uppal V. (eds), Setting Biodiversity Priorities for India, vol. 2. World Wide Fund
for Nature-India, New Delhi, pp. 442–459.

[480]
 1541

Ajith Kumar, Walker S. and Molur S. 2000. Prioritization of Endangered species. In: Singh S.,
Sastry A.R.K., Mehta R. and Uppal V. (eds), Setting Biodiversity Priorities for India. World
Wide Fund for Nature-India, New Delhi, pp. 341–425.
Amarnath G., Murthy M.S.R., Britto S.J. and Dutt C.B.S. 2003. Diagnostic analysis of conser-
vation zones using remote sensing and GIS techniques in Wet evergreen forests of the Western
Ghats- An ecological hotspot, Tamil Nadu, India. Biodiv. Conserv. 12(12): 2331–2359.
Ando A., Camm J., Polasky S. and Solow A. 1998. Species distribution, land values and efficient
conservation. Science 279: 2126–2128.
Balaguru B., John Britto S., Nagamurugan N., Natarajan D., Soosairaj S., Ravipaul S. and
Arockiasamy D.I. 2003. Vegetation mapping and slope characteristics in Shervarayan hills,
Eastern Ghats using Remote Sensing and GIS. Current Science 85(5): 645–653.
Balaguru B. 2002. Plant Diversity Assessment and its Conservation in Shervarayan Hills Eastern
Ghats of Tamil Nadu Using Remote Sensing and Geographical Information System. Ph.D
thesis, Bharathidasan University, Tiruchirappalli, Tamil Nadu, India.
Bawa K.S. and Seidler R. 1998. Natural forest management and conservation of biodiversity in
tropical forests. Conserv. Biol. 12(1): 46–53.
Bowles I.A., Rice R.E., Mittermeier R.A. and da Fonseca G.A.B. 1998. Logging and tropical forest
conservation. Science 280: 1899–1900.
Calridge G. and O’Callaghan B. (eds) 1997. Community Involvement in Wetland Management:
Lessons from the Field. Incorporating the proceeding of workshop 3: wetlands, local people, and
development, of the international conference on wetlands, Kuala Lumpur, Malaysia, pp. 19–24.
CES (Centre for Ecological Sciences) 1998. Monitoring Biodiversity – A manual of methods. IISC,
Bangalore.
Curnutt J., Lockwood J., Luh H., Nott P. and Russell G. 1994. Hotspots and species diversity.
Nature 367: 326–327.
Dawson K.J. 1996. Interior and exterior land use as factors in landscape biodiversity from case
studies. In: Szaro R.C. and Johnston D.W. (eds), Biodiversity and managed landscapes-Theory
and Practice. Oxford University Press, Oxford, pp. 514–530.
Dobson A.P., Rodriguez J.P., Roberts W.A. and Wilcove D.S. 1997. Geographic distribution of
endangered species in the United States. Science 275: 550–553.
Downton M.W. 1995. Measuring tropical deforestation: Development of the methods. Environ.
Conserv. 22(3): 229–240.
Ferreira L.V. and Prance G.T. 1998. Species richness and floristic composition in four hectares in
the Jau National Park in upland forests in Central Amazonia. Biodiv. Conserv. 7: 1349–1364.
Gaston K.J., Pressey R.L. and Margules C.R. 2002. Persistence and vulnerability: retaining bio-
diversity in the landscape and in protected areas. J. Biosci. 27(4): 361–384.
Hamilton A.C. 1984. Deforestation in Uganda. Oxford University Press and The African Wildlife
Society, Nairobi, Kenya.
Hartshorn G.S. 1995. Ecological basis for sustainable development in tropical forests. Ann. Rev.
Ecol. Systemat. 26: 155–175.
Howard P.C. 1991. Nature Conservation in Uganda’s Tropical Forest Reserves. Publication 3.
Forest Conservation Programme, World Conservation Union, Gland, Switzerland.
Hutto R.L., Heel S. and Landres P.B. 1987. A critical evaluation of the species approach to
biological conservation. Endangered species Updata 4(11): 1–4.
IUCN/SSC criteria Review working group. 1999. IUCN Red list criteria review provisional report;
draft of the proposed changes and recommendations. Species 31 & 32: 43–57.
Kier G. and Barthlott W. 2001. Measuring and mapping endemism and species richness: a new
methodological approach and its application on the flora of Africa. Biodiv. Conserv. 10: 1513–
1529.
Kumar A., Walker S. and Molur S. 2000. Prioritisation of endangered species. In: Singh S., Sastry
A.R.K., Mehta R. and Uppal V. (eds), Setting Biodiversity Priorities for India. World Wide
Fund for Nature – India, New Delhi, pp. 341–386.

[481]
1542

Kumaravelu G. and Chaudhuri K.K. 1999. Proceedings of a National Seminar on Endemic and
Endangered Plants and Animal Species of Eastern and Western Ghats. Tamil Nadu Forest
Department, Chennai.
Malcolm J.R. and Ray J.C. 2000. Influence of timber extraction routes on central African small-
mammal communities, forest structure and tree diversity. Conserv. Biol. 14(6): 1623–1628.
Margules C.R. 1989. Introduction to some Australian developments in Conservation evaluation.
Biol. Conserv. 50: 1–11.
Margules C.R., Pressey P.L. and Williams P.H. 2002. Representing biodiversity: data and proce-
dures for identifying priority areas for conservation. J. Biosci. 27(4): 309–326.
Menon S., Gil Pontius J.R.R., Rose J., Khan M.L. and Bawa K.S. 2001. Identifying conservation –
Priority areas in the tropics: A land - use change modeling approach. Conserv. Biol. 15(2): 501–
512.
Mooney H.A. and Chapin F.S.III 1994. Future directions of global change research in terrestrial
ecosystems. Trends Ecol. Evol. 9: 371–372.
Murthy K.S.R. 2000. Ground water potential in a semi-arid region of Andhra Pradesh - a geo-
graphical information system approach. Int. J. Remote Sensing 21(9): 1867–1884.
Myers N. 1980. Conversion of Tropical Moist Forests. National Academy of Science, Washington,
DC, USA, pp. 250 (illus).
Myers N. 1988. Threatened biotas: ‘‘hotspots’’ in tropical forests. The Environmentalist 8: 1–20.
Myers N. 1990. The biodiversity challenge: expanded hotspots analysis. The Environmentalist 10:
243–256.
Nayer M.P. and Shastry A.R.K. (eds) 1987–1990. Red Data Book on Indian Plants 1,2,3. BSI,
Calcutta.
Noss R.F. 1983. A regional landscape approach to maintain diversity. Bioscience 33: 700–706.
Noss R.F. 1991. From endangered species to biodiversity. In: Kohm K. (ed), Balancing on the
Brink of Extinction the Endangered Species Act and Lessons for the Future. Island Press,
Washington, DC, pp. 227–246.
O’Neill R.V., Hunsaker C.T., Jones K.B., Ritters K.H., Wickham J.D., Schwartz P.M., Goodman
I.A., Jackson B.L. and Bailargeon W.S. 1997. Monitoring environmental quality at the landscape
scale. Bioscience 47: 513–519.
Oliver I. and Beattie A.J. 1993. A possible method for the rapid assessment of biodiversity.
Conserv. Biol. 7: 562–568.
Oliver I. and Beattie A.J. 1996. Designing a cost-effective invertebrate survey: a test of methods for
rapid assessment of biodiversity. Ecol. Appl. 6: 594–607.
Olson D.M. and Dinerstein E. 1998. The global 200: a representation approach to conserving the
Earth’s most biodiversity valuable ecoregions. Conserv. Biol. 12: 502–515.
Peters C.M., Gentry A.H. and Mendelson R.O. 1989. Valuation of an Amazonian rain forest.
Nature 339: 655–656.
Ramesh B.R., Menon S. and Bawa K.S. 1997. A vegetation based approach to biodiversity gap
analysis in the Agasthyamalai region, Western Ghats, India. Ambio 26(8): 529–536.
Redford K.H. 1992. The empty forest. Bioscience 22: 412–422.
Robertson D.P. and Hull R.B. 2001. Beyond biology: toward a more public ecology for conser-
vation. Conserv. Biol. 15(4): 970–979.
Roy P.S. 1999. Remote sensing application in forestry and environment: retrospective and per-
spective. In: Proceedings of ISRS National Symposium on Remote Sensing Applications for
Natural Resources Retrospective and Perspective. Bangalore, pp. 169–185.
Scott J.M., Csuti B., Jacobi J.D. and Estes J.E. 1987. Species richness: a geographic approach to
protection future biological diversity. Bioscience 37: 782–788.
Scott J.M., Csuti B., Smith K., Estes J.E. and Caicco S. 1991. Gap analysis of species richness and
vegetation cover; an integrated biodiversity conservation strategy. In: Kohm K. (ed), Balancing
on the Brink of Extinction the Endangered Species Act and Lessons for the Future. Island Press,
Washington, DC, pp. 282–297.

[482]
 1543

Serrao E.A.S. and Homma A.K.D. 1993. Sustainable Agriculture and the Environment in the
Humid Tropics. National Academy Press, Washington, DC, pp. 265–351.
Sheil D. 2001. Conservation and Biodiversity monitoring in the Tropics: Realities, Priorities, and
distractions. Conserv. Biol. 15(4): 1179–1182.
Singh S. and Taneja B. 2000. Prioritisation for biodiversity conservation: values and criteria. In:
Singh S., Sastry A.R.K., Mehta R. and Uppal V. (eds), Setting Biodiversity Priorities for India.
World Wide Fund for Nature – India, New Delhi, pp. 28–33.
Terborgh J. and Winter B. 1983. A method of sitting parks and reserved with species reference to
Colombia and Ecuador. Biol. Conserv. 27: 45–58.
Upreti G. 1994. Environmental conservation and sustainable development require a new devel-
opmental approach. Environ. Conserv. 21(1): 18–29.
Western D. and Wright R.M. 1994. Natural Connections: Perspective in Community Based
Conservation. Island press, Washington, DC, pp. 581.
Wilcove D.S. 1989. Protecting biodiversity in multiple-use lands: Lessons from the U.S. Forest
Service. Trends Ecol. Evol. 4: 385–388.
Williams P.H., Margules C.R. and Hilbert D.W. 2002. Data requirements and data sources for
biodiversity priority area selection. J. Biosci. 27(4): 327–338.
Williams P.H., Gibbons D., Margules C.R., Rebelo A., Humphires C. and Pressey R. 1996. A
comparison of richness hotspots, rarity hotspots, and complementary areas for conserving
diversity of British birds. Conserv. Biol. 10: 155–174.
Wilson E.O. (ed) 1988. Biodiversity. National Academy press, Washington, DC.

[483]
Biodiversity and Conservation (2006) 15:1545–1575
Springer 2006
DOI 10.1007/s10531-005-2930-6

Comparison of bird communities in primary vs. young

secondary tropical montane cloud forest in
Guatemala

SWEN C RENNER1,*, MATTHIAS WALTERT2 and

MICHAEL MÜHLENBERG2
1
Conservation and Research Center, Smithsonian Institution, 1500 Remount Road, Front Royal, VA
22630, USA; 2Centre for Nature Conservation, Georg-August University of Göttingen, Von-Siebold-
Straße 2, 37075 Göttingen, Germany; *Author for correspondence (e-mail: [email protected])

Received 10 August 2004; accepted in revised form 14 February 2005

Key words: Avian species richness, Conservation, Deforestation, Diversity, Guatemala, Land-use

Abstract. Cloud forests in central Guatemala are fragmented and decreasing in area due to slash-
and-burn agricultural activities. We studied bird species composition, abundance, guild composi-
tion, and site tenacity of a 102 ha plot located in a cloud forest region of the Sierra Yalijux in
Guatemala, half of which was primary forest and half young secondary forest (<7-years-old). Of
the 100 species present 14 were restricted to the Endemic Bird Area ‘Northern Central American
highlands’ (i.e. 66% of a total of 21 endemics). Five of the 100 analysed species, including one of
the restricted-range species (Troglodytes rufociliatus), had a significantly different abundance in
primary and secondary forests. Theoretical analysis suggests that seven species out of a community
comprised of 141 bird species are already extirpated and only three out of the 14 present restricted-
range species might survive the current state of deforestation. Insectivores were the dominant guild
on the plot in terms of numbers of species, followed by omnivores, frugivores and granivores.
However, in terms of individuals, omnivores made up nearly half of the bird individuals in primary
forest, but declined by 44% in secondary forest, whereas granivores more than doubled in this
habitat type. Numbers of species per guild were not significantly different between habitats, while
numbers of individuals per guild were significantly different. In general, individuals per species are
significantly different in the two habitats. Results suggest that most of the species that are currently
surviving in the remnant forests of the Sierra Yalijux might be fairly well adapted to a range of
forest conditions, but that populations of a number of restricted-range species might be small. Even
generalists species like the Common Bush Tanager (Chlorospingus ophthalmicus) are less abundant
in secondary vegetation than in primary forest of the study plot.

Introduction

Annual deforestation in Central America is very high, especially in El Salvador

(4.6%), Nicaragua (3.0%), and Guatemala (1.7%) (World Bank 2001; cf. FAO
2001; World Conservation Monitoring Centre 1992). One potential conse-
quence of deforestation is loss of species (Bierregaard 1990; Bierregaard and
Stouffer 1997; Myers et al. 2000; World Bank 2001), mainly affecting forest
specialists (e.g., McGowan and Gillman 1997; Stattersfield and Capper 2000).
Expanding subsistence agriculture to meet the needs of an increasing human
population (e.g., Máñez-Costa and Renner 2005) threatens biodiversity in

[485]
1546

forest remnants world-wide (e.g., Hughes et al. 2002; Schulz et al. in press) and
especially in Latin America (Markussen 2004). In central Guatemala, natural
habitats, such as primary cloud forests, are often converted and used as part of
the milpa system (slash-and-burn agriculture) (Renner 2003; Markussen 2004;
Markussen and Renner 2005). It is questionable whether secondary forest re-
growth after farming can preserve numbers of species comparable to natural
habitats in tropical landscapes over the long term (Hughes et al. 2002; Renner
2003). In addition, the time required for secondary forest biodiversity to ap-
proach that of primary forest is not known (e.g., Shankar-Raman et al. 1998;
Terborgh 1999; Shankar-Raman 2001; Shankar-Raman and Sukumar 2002).
Cloud forest areas in many parts of the tropics are increasingly deforested,
and many forest remnants have likely already lost a large proportion of their
original avifauna (Kappelle and Brown 2001). Some sites are presumed to have
lost forest bird populations, e.g., the Brazilian Atlantic forests. Other sites may
still retain endemic forest species because of persisting habitat heterogeneity
(Marsden et al. 2004). Nevertheless, fragmentation and loss of forest habitats
are important causes of regional species extinction (Marsden et al. 2004). The
province Alta Verapaz in central Guatemala is located within a Biodiversity
Hotspot (Myers et al. 2000) and an Endemic Bird Area (Stattersfield et al.
1998), and is therefore of high importance for conservation (Veblen 1976;
Islebe 1995; Stattersfield et al. 1998; Markussen and Renner 2005). This region
contains primary forests subject to decreasing area and increasing fragmen-
tation that currently contain endemic and specialised bird species.
We examined species richness, abundance, site tenacity and body mass dis-
tribution of birds in primary cloud forest compared to neighbouring young
secondary forest in the highlands of central Guatemala near the village of
Chelemhá. We used within-one-habitat recaptures as a measure of habitat
quality. Special emphasis was placed on restricted-range species and Central
American cloud forest specialists.

Study area

The Sierra Yalijux where the study site was located, belongs to the north-
easternmost slopes of the northern Cordillera of Guatemala (Municipio Tu-
curú, Alta Verapaz). The total forest cover above 1,800 m (ca. the altitudinal
border of cloud forests in Guatemala) is estimated at 5,500 ha (Markussen
2004). In this region, the climate is tropical to sub-tropical (MAGA 2001) and
precipitation is high, up to 4000 mm per year (Markussen 2004).
We chose a 102 ha study plot in the Sierra Yalijux near the settlement of
Chelemhá (Figure 1). The geographic plot co-ordinates are 9004¢ W and
1523¢ N. It is situated at an elevation between 1,980 m and 2,550 m and
contains approximately 51 ha of primary forest and 51 ha of secondary
forest. The primary forest in the region is mainly mixed oak and pine forest
with Pinus maximinoi as the dominant species (Veblen 1976; Moziño 1996).

[486]
 1547

Figure 1. (a) Location of the study plot (n) in Guatemala and Endemic Bird Area ‘Central
American highlands’ (Stattersfield et al. 1998) (- - - -). Grey indicates areas in Guatemala with
elevations above 1000 m. (b) The study plot (bold line — ) with main trails for transect counts (thin
line —), mist net sites (black dots •) and vegetation classification (n grey: primary forest, white:
secondary habitats of all kinds). Dotted line indicates the Rı́o Chelemhá; arrow on map resembles
approximately 400 m in nature.

The canopy exceeds 35 m and there are generally three vegetation strata, i.e.
understorey <7 m, mid-story 7–20 m, and overstorey >20 m. Stem diameter
at breast height for overstorey trees exceeds 1 m. Young secondary forest
(<10 years-old), is – aside from primary forest – the most abundant vege-
tation in the study region and probably also within the whole of Alta Ve-
rapaz. Secondary forests >10 years cover <5 ha within the community of
Chelemhá. Young secondary vegetation is the result of slash-and-burn agri-
culture (milpa system), and the secondary forest we studied had an age of less
than 7 years. After corn and bean cultivation, shrubs can grow up to 10 m in
height during a fallow period of 7 years. However, a fallow period of this
duration is exceptional occurring only in the study plot. Generally in the
community of Chelemhá there is a fallow period of <2 years. The shrubby
habitat is composed of many different tree species, including oaks (Quercus
sp.); but in contrast to primary forest the pines (Pinus sp.) are rare and the
shrubs form only a single vegetation layer.
For both the understorey and overstorey, the estimated tree height, diameter
at breast height and density were measured for each 25 m section of the bird
count transects.

Methods

Bird surveys employing audio/visual counts as well as standardised mist net-

ting were conducted during the region’s main breeding season between March
and September 2001 and 2002 in both primary forest and adjacent secondary
forest.

[487]
1548

In primary and secondary forest, an existing trail system was used to count
birds by means of regular transect counts in order to assess population den-
sities for different habitats of the plot (Bibby et al. 1995; Krebs 1999; Gilbert
et al. 2000). A total of 3,300 m of counts were conducted along the trails (no
trail intersections allowed, all trails were separated at least by 50 m), 50% in
each vegetation type. A series of three visits per trail were conducted in 2002
between 0530 h and 0900 h. A maximum trail length of 450 m was visited per
day. All sight records and singing individuals were recorded while slowly
walking along the trail. The trails of the entire plot were divided into 22 trail
sections of 150 m each. Eleven sections were situated in primary forest, and 11
in secondary forest. All displaying birds within 100 m distance to both sides of
the trail were recorded, so that each trail section represented a 3-ha-sized
rectangular strip (150 · 2 · 100 m = 3 ha). Each record contained the per-
pendicular observation distance, as well as information on species, sex, age,
and number of individuals. Density estimates were derived from the maximum
number of birds recorded on each 3-ha strip. For some cryptic species, (e.g.
Grallaria guatimalensis) or species with low voices (Buarremon brunneinucha,
Atlapetes gularis), the strip width was reduced to 50 m and the corresponding
observations within 50 m from the transect for the density calculations. Cya-
nocorax melanocyaneus and Cyanolyca pumilo were estimated conservatively by
directly counted individuals. Most flocks (especially the frequent Chlorospingus
ophthalmicus flocks) disbanded early in the breeding season and could be noted
as pairs. Singing and territorial display behaviour of males was counted as a
breeding unit for all other species, except for species where females also display
territorial behaviour. Trochilidae were treated in the same way, except for trap-
liners where only females (if distinctive) were recorded as a breeding unit.
Identification was made using Land (1970), Howell and Webb (1995), and
Edwards (1998). Taxonomy and systematic order follows the checklist of the
American Ornithologists’ Union (1998) with the relevant supplements
(American Ornithologists’ Union 2000, 2002).
Transect and mist netting procedures are often combined in tropical forest
bird community studies since the two methods select different portions of the
community, thus achieving a more accurate assessment of community com-
position (Poulsen 1994; Remsen 1994; Remsen and Good 1996). Twelve mist
net lines were established, six in natural and six in secondary forest. The net
lines were distributed randomly at existing tracks to avoid pseudo-replication
(Hurlbert 1984). Each net line consisted of eight nets of 12 m each and was
opened for 8.5 h per capture day. Capturing was conducted for 2 days at each
line in 2001, and for 4 days in 2002 with a total of 4,896 net hours (12 m net).
Each captured bird was marked and body mass determined with spring bal-
ances.
Effects of land-use were measured as differences in bird community struc-
ture, including changes in species richness, species composition, numbers of
individuals (trail data) as well as in individual turnover, body size and body
mass (mist net data) between primary forest and secondary forest.

[488]
 1549

Statistics

Bird density, differences in species numbers and comparisons of species simi-

larity between habitats were calculated. Similarity measures are common
methods for distinguishing between entities (Magurran 1988; Rosenzweig 1995;
Krebs 1999; Gaston and Blackburn 2000). The Sørensen and Morisita-Horn
indices are more useful than Jaccard or other indices (Magurran 1988; Krebs
1999), and similarity analyses based on Sørensen were used (in Colwell 2000:
Sørensen-Inc.). For each combination of the 12 mist net lines and for each trail
section, Sørensen values were computed and arranged in a dissimilarity matrix.
Ordination of the samples (trail sections, mist net lines) using multi-dimen-
sional scaling (MDS) was carried out.
To estimate the influence of land-use on bird abundance at species level,
observation frequencies were tested using the v2-test with primary forest as
the ‘expected variable’ and secondary forest as the ‘observed variable’. The
Mann-Whitney-U test was used to compare abundance between primary and
secondary forest.

Estimation of total species richness

Species richness is normally not detected completely, that is neither all species,
nor all individuals are detected in natural environments (e.g., Rosenzweig 1995;
Begon et al. 1996; Krebs 1999). For the purpose of estimation of total species
richness, several indices were established with different purposes (e.g., Chao
and Lee 1992; Colwell and Coddington 1994; Rosenzweig 1995). These esti-
mators and indices give an approximate number of species that might be de-
tected when making repeated observations (Magurran 1988; Rosenzweig 1995;
Krebs 1999). Different estimators are of different value according to their
purpose. Here we used the Bootstrap estimator due to the large sample (Ma-
gurran 1988). Otherwise the Jackknife estimator must to be used.
Based on the trail data, we first described species accumulation from the
pooled data of the whole study plot (Sobs in Colwell 2000). Then differences
between primary forest and secondary forest were established from the sepa-
rate samples. In addition, primary and secondary forest were compared with
regard to species composition. To describe the community, observed species
numbers (Sobs), observed individuals numbers (N), and indices were calculated,
and Bootstrap (SBoot), Evenness (E), and the Abundance-based Estimator of
Species Richness (ACE) were used. For calculation and formulas of the esti-
mators see Magurran (1988) and Colwell (2000). See Krebs (1999) and Renner
(2003) for a complete summary.
Body mass analyses required splitting of sexes for certain species. A group is
either a species or one sex of a species. For Diglossa baritula, Lampornis
amethystinus, L. viridipallens, Lamprolaima rhami, and Turdus infuscatus, a

[489]
1550

sexual separation for analysis is essential due to significant body mass differ-
ences between the sexes.

Recapture data

Impact on tropical avifaunas is frequently estimated with mark-recapture

methods (e.g., Lambert 1992; Johns 1992; Holbech 1996; Dranzoa 1998;
Waltert and Mühlenberg 2001). Optimal habitats are usually correlated with
high captures and/or recaptures, and pessimal habitat quality with low cap-
tures and/or recaptures (Matthysen et al. 1995; Reitsma et al. 2002; and oth-
ers). Winker et al. (1995) presumed that intra-specific competition in territorial
birds forces (behaviourally) sub-dominant individuals to occupy sub-optimal
or pessimal habitats with increasing population density. Mainly territorial and
dominant individuals should occupy optimal habitats. In this study, the re-
capture rate Rt was defined as the ratio of first captures Fc to recaptures Rc
(Rt = Rc/Fc), with all same-day-recaptures excluded, and each recaptured
individual being counted only once even when recaptured twice or more times.
In total, each net line was sampled two times in 2001 and four times in 2002
(see above).

Results

Vegetation

Mean distance between trees with diameter at breast height of <0.20 m of the
Chelemhá Plot was 2.94 m±1.77 in primary forest and 1.16 m±0.82 in sec-
ondary forest. Mean height of overstorey trees was 25.9 m±6.6 in primary
forest, 4.9 m±1.9 for understorey trees in primary forest, and 2.9 m±1.9 for
trees in secondary forest (see Table 1).

Table 1. Parameters of the vegetation structure.

Overstorey Understorey

dbh h d dbh h d

Primary forest Mean 84.3 25.9 294.6 5.1 4.9 164.5

±s.d. 53.4 6.6 177.2 2.8 1.9 114.2
n 142 142 142 145 145 145
Secondary forest Mean 59.2 11.67 233.3 2.1 2.6 112.8
±s.d. 74.5 2.58 182.6 1.5 1.9 81.2
n 6 6 6 67 67 67
dbh: diameter at breast height (cm), h: height (m), d: density (cm). Further details on methods are
given in the chapter ‘Study Area and Study Plot’.

[490]
 1551

Avifauna of the study area

On the 102-ha plot, a total of 100 avian species was found to be present of
which 75 are presumably residents. Sixty-six of the 100 avian species recorded
were detected by mist netting, 75 from regular counts along trails, and eight by
incidental records. Forty-seven species were recorded by both mist netting and
transect counts, 17 by mist netting alone and 28 by transect counts alone. Out
of the 100 species, 75 species were presumed resident breeding species, hereafter
treated as ‘residents’. This number is an estimate and includes incidental re-
cords of species which are believed to breed but were only rarely recorded (1–2
observations) by regular transect counts (Coragyps atratus, Lophostrix cristata,
Glaucidium gnoma, Streptoprocne zonaris, Sclerurus guatemalensis, Tachycineta
bicolor, Saltator caerolescens, Dives dives). The total species numbers for the
entire plot was estimated (Colwell 2000) at 61 (Sobs) and 62.75±0.00 (Boot-
strap±s.d.).
Further results on the bird community structure and comparisons to other
tropical and temperate studies are drawn in Renner (2003).

Effects of land-use on the bird community structure

Based on the maximum abundance of each species at each trail section, a total
of 1405 individuals was estimated to be present on the plot. While, in total,
more species were observed in secondary forest (51 and 55 observed species in
primary forest and secondary forest, respectively), more individuals were ob-
served in primary forest (753) than in secondary forest (652). The difference in
overall population size is significant (v2 = 7.41, pdf1<0.01), representing a
decrease of 14% between primary and secondary forest. The higher observed
species numbers in secondary forest is also reflected in higher estimated total
species richness: the Bootstrap estimator in primary forest amounted to 52.91
(±0.77 s.d.) and in secondary forest to 57.80 (±0.00). This suggests that only a
few more species remained undetected (Figures 1 and 2). Diversity statistics of
birds are summarised in Table 2.

Table 2. Summary of diversity statistics (calculated with Colwell 2000) depending on habitat
compiled from transect count data.

Primary forest ±s.d. Secondary forest ±s.d.

Sobs 51 – 55 –
Individuals 1563 – 1766 –
Singletons 0.0 0.00 0.0 0.00
Doubletons 0.0 0.00 0.0 0.00
ACE 51.0 0.00 55.0 0.00
Bootstrap 52.9 0.77 57.1 0.00

[491]
1552

Figure 2. Expected species E (transect counts data from 2002) in primary forest and secondary
forest. Error bars were left out for better illustration. Data were calculated with Rarefaction 1.3
(Holland 2003).

Average pair-wise similarity of bird species composition (mean Sørensen ±s.d.)

of the 22 three-ha sections was nearly constant for both within and between
habitat comparisons. It amounted to 0.61±0.11 (mist netting: 0.65±0.11)
between the 11 primary forest samples, 0.58±0.11 between secondary forest
samples (mist netting 0.66±0.34) and 0.61±0.11 (mist netting 0.63±0.16)
between primary and secondary forest samples. When ordinating samples from
transect counts using non-linear multidimensional scaling the two groups of
sites did largely overlap (Figure 4). A one-way MANOVA of the sample scores
extracted from the two-dimensional ordination revealed no significant differ-
ence between the two groups of sites (Rao’s R2.19 = 1.18, p = 0.33). However,

Figure 3. Observed species Sobs (transect count data from 2002) in primary forest and secondary
forest in the Chelemhá Plot. Error bars were left out for better illustration. Calculated with Colwell
(2000).

[492]
 1553

Figure 4. Non-linear multidimensional scaling (MDS) plot of avifaunal similarity based on

Sørensen-incidence values for 22 transect counts section of 3 ha. Lines connect study sites
belonging to the same habitat category. Habitats: PF – primary forest, YSF – young secondary
forest. Note different scaling on axes.

using the bird data from the twelve net lines, there are two distinct groups
(primary, secondary forest) forming the ordination plot (Figure 5), with
sample scores of trapped bird assemblages showing significant differences
between habitats (Rao’s R2,9 = 8.27, p<0.01).

Figure 5. Non-linear multidimensional scaling (MDS) plot of avifaunal similarity based on

Sørensen-incidence values for 12 mist net lines, each of 102 m length. Lines connect study sites
belonging to the same habitat category. Habitats: PF – primary forest, YSF – young secondary
forest. Note different scaling on axes.

[493]
1554

Forty-three species had individuals in both habitats; three species had similar
mean individual numbers, 17 had higher mean individual numbers in primary
forest and 23 had higher mean individual numbers in secondary forest. Five
species showed significant differences in abundance in the two habitats
(Table 6). Zimmerius villisimus, Troglodytes rufociliatus, Atthis ellioti,
Myadestes unicolor, and Atlapetes gutteralis had significantly different indi-
vidual numbers in primary forest and secondary forest. C. ophthalmicus (p =
0.08), Henicorhina leucophrys (p = 0.06), and Sclerurus mexicanus (p = 0.06)
had almost significant differences.
The most common species, C. ophthalmicus, was comparatively less abun-
dant in secondary forest than in primary forest (Table 6). Three species
(Z. vilissimus, T. rufociliatus, and M. unicolor; Tables 3 and 7) were signifi-
cantly less abundant in secondary forest compared to primary forest. While 23
species are less abundant in primary forest than in secondary forest, at least 22
have 50% more individuals in primary forest.
At the species level, five species were found in significantly different numbers
in the two habitats (Table 3). This represents a minor part of the 100 species
observed, and should be treated with caution. Given that from an accepted
significance level of 5% (p = 0.05) and 61 analysed species with any N to
perform the Mann–Whitney-U test out of the 100 residents (see Table 6), al-
ready 3.05 species (61 · 0.05) could be expected to show significant differences
in abundance between habitat types by chance alone.
Within the Chelemhá Plot there are three species influenced significantly
negatively and two positively in accordance to individual numbers. Several pri-
mary forest birds (Table 7) and especially endemic primary forest birds of the
Central American highlands are still present in the respective parts of this plot.

Guild composition

Guild composition was significantly different between habitats concerning

numbers of individuals but not concerning numbers of species (Table 4). In

Table 3. Species showing significant differences in abundance between habitat types. Abundance
given as mean numbers of individuals calculated from the maximums recorded on 3-ha transect
strips. p: p-level (corrected) based on the Mann-Whitney-U test. Exclusively species with significant
difference (p<0.05) are listed.

Species p Primary forest Secondary forest

Mean ±s.d. Mean ±s.d.

Atthis ellioti 0.07 0.00 – 1.20 0.45

Zimmerius vilissimus 0.04 2.36 1.57 2.17 1.94
Troglodytes ruficiliatus <0.01 1.56 0.73 1.00 0.00
Myadestes unicolor 0.02 2.73 1.42 1.36 0.92
Atlapetes gutteralis 0.03 4.25 4.03 4.20 3.65

[494]
 Table 4. Numbers of species and individuals per guild (transect count data). Primary forest represents expected and secondary forest observed frequency for
v2-test. The first number is total N, the second the relative portion. Differences to Table 2 occur to the fact that here counted and there estimated individuals
are given. As indicated in Table 6, several species and individuals were accounted for more than one guild when clear nutrition preferences were not given.

Exclusiona Insectivores Omnivores Frugivores Nectarivores Granivores Carnivores Total (N) v2 (p)

PF Species NO 24 9 6 5 5 2 51 8.4 (0.20)

% NO 0.46 0.18 0.13 0.11 0.09 0.04 df = 5
SF Species NO 19 11 5 8 10 2 55
% NO 0.34 0.20 0.09 0.14 0.19 0.03
PF Individuals NO 230 346 42 76 57 2 753 144.3 (<0.01)
% NO 0.31 0.46 0.06 0.10 0.08 0.00 df = 5
SF Individuals NO 219 195 46 67 123 2 652
[495]

% NO 0.34 0.29 0.07 0.10 0.19 0.00

PF Individuals O 230 137 42 76 57 2 544 99.7 (<0.01)
% O 0.42 0.25 0.08 0.14 0.11 0.01 df = 5
SF Individuals O 219 83 46 67 123 2 540
% O 0.41 0.15 0.09 0.12 0.23 0.01
PF Individuals G 230 346 42 76 4 2 700 998.1 (<0.01)
% G 0.33 0.49 0.06 0.11 0.01 0.01 df = 5
SF Individuals G 219 195 46 67 65 2 594
% G 0.37 0.33 0.07 0.11 0.11 0.01
a
Exclusion of the most abundant species from analysis. NO: no exclusion, O: Exclusion of the three omnivores Chlorospingus ophthalmicus, Myadestes
unicolor, and Cyanolyca pumilo, G: Exclusion of the three granivores Atlapetes gutteralis, Columba fasciata, and Carduelis notata. Further explanations and
justification see text (‘Results’ and there ‘Guild composition’).
PF: primary forest, SF: secondary forest

1555
1556

Table 5. Body mass changes in recaptured individuals. BM 1: body mass at first capture, BM 2:
body mass at final recapture in the same breeding season. PF: primary forest, SF: secondary forest.

Species Habitat BM 1[g] BM 2[g] Change[g] Relative change

Asphata gularis SF 52.50 55.25 2.75 0.05

Atlapetes gutteralis SF 33.50 42.00 8.50 0.20
Atlapetes gutteralis SF 33.00 32.00
1.00
0.03
Basileuterus belli PF 10.00 10.25 0.25 0.02
Basileuterus belli PF 10.00 9.50
0.50
0.05
Basileuterus belli PF 9.75 9.50
0.25
0.03
Basileuterus belli SF 10.75 11.25 0.50 0.04
Buarremon brunneinucha PF 42.25 42.00
0.25
0.01
Buarremon brunneinucha SF 38.00 38.25 0.25 0.01
Catharus frantzii PF 19.00 33.00 14.00 0.42
Catharus frantzii PF 19.00 30.50 11.50 0.38
Catharus frantzii PF 22.00 39.00 17.00 0.44
Catharus frantzii PF 28.50 27.75
0.75
0.03
Catharus frantzii PF 31.00 30.50
0.50
0.02
Catharus frantzii SF 28.50 29.25 0.75 0.03
Chlorospingus ophthalmicus PF 16.75 17.75 1.00 0.06
Chlorospingus ophthalmicus PF 9.25 19.70 10.45 0.53
Chlorospingus ophthalmicus PF 18.75 15.25
3.50
0.23
Chlorospingus ophthalmicus PF 21.25 19.50
1.75
0.09
Chlorospingus ophthalmicus SF 17.50 21.50 4.00 0.19
Chlorospingus ophthalmicus SF 14.50 18.50 4.00 0.22
Chlorospingus ophthalmicus SF 17.25 18.00 0.75 0.04
Chlorospingus ophthalmicus SF 18.50 19.50 1.00 0.05
Chlorospingus ophthalmicus SF 19.25 18.00
1.25
0.07
Chlorospingus ophthalmicus SF 17.75 17.00
0.75
0.04
Diglossa baritula SF 9.25 9.75 0.50 0.05
Diglossa baritula SF 9.25 9.00
0.25
0.03
Diglossa baritula SF 9.50 8.25
1.25
0.15
Diglossa baritula SF 9.75 9.25
0.50
0.05
Henicorhina leucophrys PF 12.75 13.75 1.00 0.07
Henicorhina leucophrys PF 15.00 13.25
1.75
0.13
Henicorhina leucophrys SF 15.00 15.50 0.50 0.03
Lampornis amethystinus PF 5.10 5.50 0.40 0.07
Lamprolaima rhami PF 5.75 6.25 0.50 0.08
Lamprolaima rhami PF 5.60 5.70 0.10 0.02
Lamprolaima rhami PF 8.30 8.40 0.10 0.01
Lamprolaima rhami PF 6.40 6.50 0.10 0.02
Lamprolaima rhami PF 8.50 8.60 0.10 0.01
Lamprolaima rhami PF 8.20 8.40 0.20 0.02
Lamprolaima rhami PF 9.10 8.00
1.10
0.14
Lamprolaima rhami PF 8.30 8.20
0.10
0.01
Myadestes occidentalis PF 35.50 36.75 1.25 0.03
Myadestes occidentalis SF 24.75 33.25 8.50 0.26
Myadestes unicolor PF 38.00 36.50
1.50
0.04
Myioborus miniatus SF 8.25 7.25
1.00
0.14
Troglodytes ruficiliatus SF 11.75 12.00 0.25 0.02
Wilsonia pusilla SF 9.00 7.75
1.25
0.16
Xiphorhynchus erythropygius PF 37.25 45.00 7.75 0.17
Total individuals: 48 Mean: 18.10 19.74 1.64 0.04

[496]
 Table 6. Abundance for forest species, given as maximum records on 51 ha primary forest (PF) and 51 ha secondary forest (SF) and population estimates
(for 102 ha and 5500 ha, the latter represents the natural pine–oak forests of the Sierra Yalijux). All species potentially abundant in the Sierra Yalijux are
listed. However, guild, habitat, etc. is only listed for the relevant species definitely observed. Taxonomy and systematics follows American Ornithologist’s
Union (1998, 2000, 2002).

Family Species Exp.a Guildb Habitatc Methodd Abundancee Estimate for the Sierra Yalijux
(PF)

PF SF Ind./102 ha Ind./5,500 ha

Cracidae
Ortalis vetula H,C F TC 2 0 4 216
Penelope purpurascens H
Oreophasis derbianus H F –
Penelopina nigra H,C F PF TC 16 16 31 1725
Phasianidae
Dendrortyx leucophrys H,C G TC 1 4 2 108
[497]

Odontophorus guttatus H
Dactylortyx thoracicus H
Cyrtonyx ocellatus H G TC
Ardeidae
Ardea herodias C
Bubulcus ibis C
Cathartidae
Coragyps atratus H,C Ca I
Accipitridae
Chondrohierax uncinatus H,C C TC
Elanoides forficatus C
Accipiter chinogaster H,C
Asturina nitida C
Buteo platypterus C
Buteo solitarius H
Buteo jamaicensis H,C

1557
 Table 6. (Continued).

1558
Family Species Exp.a Guildb Habitatc Methodd Abundancee Estimate for the Sierra Yalijux
(PF)
PF SF Ind./102 ha Ind./5,500 ha
Falconidae
Micrastur ruficollis H,C
Falco sparverius H,C
Scolopacidae
Bartramia longicauda C
Columbidae
Columba livbia H
Patagioenas fasciata H,C F,G TC 9 3 18 971
Zenaida asiatica H
Zenaida macroura C
Columbina inca H
[498]

Columbina passerina H
Claravis mondetoura H F,G N
Leptotila verreauxi C
Geotrygon albifacies H,C F,G N
Psittacidae
Bolborhynchus lineola H,C G,F TC 2 1 4 216
Cuculidae
Piaya cayana H
Tytonidae
Tyto alba H
Strigidae
Megascops trichopsis H
Megascops barbarus H
Lophostrix cristata H C I
Glaucidium gnoma H,C C I
Ciccaba virgata H,C C TC
Strix fulvescens superspecies varia H,C C TC 0 1 0 0
 Asio stygius H
Aegolius ridgway H
Caprimulgidae
Chordeiles acutipennis H
Caprimulgus vociferus) arizonae H,C I TC 1 0 2 108
Apodidae
Cypseloides niger H,C I TC
Streptoprocne rutila H,C
Streptoprocne zonaris H,C I I
Chaetura vauxi H,C
Aeronautes saxatalis H,C I TC
Trochilidae
Campylopterus hemileucurus H,C N N
Colibri thalassinus H,C N N,TC 0 1 0 0
Abeillia abeillei H N N,TC 0 2 0 0
Hylocharis leucotis H,C N N
[499]

Amazilia cyanocephala H N N
Lampornis viridipallens H,C N N,TC 0 2 0 0
Lampornis amethystinus H,C N N,TC 46 45 90 4961
Lamprolaima rhami H,C N N,TC 11 7 22 1186
Eugenes fulgens H,C N N,TC
Doricha enicura H N N
Tilmatura dupontii H,C N TC
Atthis ellioti H,C N N,TC 0 6 0 0
Trogonidae
Trogon collaris H I,F TC 2 0 4 216
Trogon mexicanus H,C I,F TC 11 7 22 1186
Pharomachrus mocinno H,C F N,TC 17 13 33 1833
Momotidae
Aspatha gularis H,C F,I N,TC 3 7 6 324
Ramphastidae
Aulacorhynchus prasinus H,C F N,TC 7 5 14 755

1559
 Table 6. (Continued).

1560
Family Species Exp.a Guildb Habitatc Methodd Abundancee Estimate for the Sierra Yalijux
(PF)
PF SF Ind./102 ha Ind./5,500 ha
Picidae
Melanerpes formicivorus H
Melanerpes aurifrons H,C
Sphyrapicus varius C
Colaptes auratus H,C I N,TC 1 3 2 108
Picoides villosus H,C I N,TC 10 8 20 1078
Piculus rubiginosus H
Furnaridae
Anabacerthia variegaticeps H,C
Automolus rubiginosus H,C I N,TC 5 2 10 539
Sclerurus mexicanus H,C I TC 5 2 10 539
[500]

Sclerurus guatemalensis - I I
Dendrocolaptidae
Xiphorhynchus promeropirhynchus H
Xiphorhynchus erythropygius H,C I N,TC 1 1 2 108
Lepidocolaptes affinis H,C I N,TC
Formicaridae
Grallaria guatimalensis H I N
Tyrannidae
Camptostoma imberbe H I N
Elaenia frantzii C
Zimmerius vilissimus H,C I SF N,TC 26 13 51 2804
Mitrephanes phaeocercus H,C I TC 12 12 24 1294
Cantopus pertinax H I TC
Cantopus sordidulus C
Contopus virens C
Cantopus cinereus C
Contopus borealis C
 Empidonax flaviventris C
Empidonax virescens C
Empidonax minimus C
Empidonax hammondii C
Empidonax oberholseri C
Empidonax affinis H I N
Empidonax flavescens H,C I SF N,TC 0 1 0 0
Empidonax fulvifrons H,C
Sayornis nigricans H,C
Pachyramphus major H
Pachyramphus aglaiae H,C
Vireonidae
Vireo plumbeus H,C I N,TC 7 0 14 755
Vireo huttoni H,C
Vireo gilvus C
Vireo leucophrys C
[501]

Vireo philadelphicus C
Cyclarhis gujanensis C
Corvidae
Cyanocitta stelleri H
Cyanocorax melanocyaneus H,C O N,TC 1 3 2 108
Cyanolyca pumilo H,C O N,TC 16 8 31 1725
Aphelocoma unicolor H
Corvus corax H O TC
Hirudinidae
Tachycineta bicolor H,C I I
Tachycineta thalassina H I TC 0 1 0 0
Notiochelidon pileata H,C I TC
Petrochelidon pyrrhonota C
Hirundo rustica C
Certhididae
Certhia americana H

1561
 Table 6. (Continued).

1562
Family Species Exp.a Guildb Habitatc Methodd Abundancee Estimate for the Sierra Yalijux
(PF)
PF SF Ind./102 ha Ind./5,500 ha
Troglodytidae
Thryothorus modestus H,C I PF N,TC 13 13 25 1402
Troglodytes musculus H,C I N,TC 7 10 14 755
Troglodytes rufociliatus H,C I N,TC 14 2 27 1510
Henicorhina leucophrys H,C I N,TC 35 29 69 3775
Turdidae
Sialia sialis H,C
Myadestes occidentalis H,C O N,TC 15 16 29 1618
Myadestes unicolor H,C O N,TC 30 15 59 3235
Catharus aurantiirostris H,C O N,TC 4 5 8 431
Catharus frantzii H,C O N,TC 35 31 69 3775
[502]

Catharus mexicanus H
Catharus dryas H
Catharus ustulatus C
Catharus guttatus C
Turdus infuscatus H,C O N,TC 15 8 29 1618
Turdus plebejus H,C O N,TC 5 8 10 539
Turdus grayi H,C O N,TC 35 33 69 3775
Turdus rufitorques H,C O TC 1 4 2 108
Mimidae
Dumatella carolineus C
Melanotis hypoleucus H,C I,F SF N,TC 11 13 22 1186
Ptilogonatidae
Ptilogonys cinereus H,C
Peucedramidae
Peucedramus taeniatus H,C
Parulidae
Vermivora chrysoptera H I PF N
 Vermivora peregrina C
Parula superciliosa H,C I N,TC 3 7 6 324
Dendroica pensylvanica C
Dendroica coronata C
Dendroica virens C
Dendroica townsendi C
Dendroica occidentalis C
Dendroica fusca H,C I N,TC
Dendroica graciae H
Mniotilta varia C
Seiurus noveboracensis C
Seiurus motacilla C
Oporornis tolmiei H,C I N,TC 5 1 10 539
Geothlypis poliocephala H,C I N
Wilsonia pusilla H,C I N,TC 4 0 8 431
Wilsonia canadensis C
[503]

Ergaticus versicolor H,C I N

Myioborus miniatus H,C I N,TC 6 4 12 647
Basileuterus rufifrons H,C I N
Basileuterus belli H,C I N,TC 55 54 108 5931
Thraupidae
Chlorospingus ophthalmicus H,C O N,TC 163 89 320 17578
Piranga flava H
Emberezidae
Volatinia jacarina C G N
Sporophila torqueola H,C G N,TC
Tiaris olivacea H,C G N,TC 0 2 0 0
Haplospiza rustica H I TC 0 3 0 0
Diglossa baritula H,C N N,TC 2 6 4 216
Atlapetes gutteralis H,C G PF N,TC 17 42 33 1833
Buarremon brunneinucha H,C G SF N,TC 45 52 88 4853
Melozone biarcuata H G N,TC 0 3 0 0

1563
 Table 6. (Continued).

1564
Family Species Exp.a Guildb Habitatc Methodd Abundancee Estimate for the Sierra Yalijux
(PF)
PF SF Ind./102 ha Ind./5,500 ha
Aimophila rufescens H,C G N
Zonotrichia capensis H,C G N,TC 3 5 6 324
Spizella passerina H G N
Cardinalidae
Saltator caerolescens C G I
Saltator atriceps H G TC 3 2 6 324
Pheucticus ludovicianus C
Icteridae
Dives dives H,C G,I I
Quiscalus mexicanus H,C O TC 1 4 2 108
Molothrus aeneus H
[504]

Icterus chrysater H,C

Icterus galbula C
Fringillidae
Euphonia elegantissima H,C F TC 9 4 18 971
Chlorophonia occipitalis H,C F,G,I TC 1 0 2 108
Carduelis atriceps – G N
Carduelis notata H,C G N,TC 4 13 8 431
Coccothraustes abeillei H,C
:extinct species (not considered as resident)
a
Species expected according to H: Howell and Webb (1996), C: found in Chicacnab 12 km west of Chelemhá (Eisermann 2000). List is derived with the
following criteria: species are resident in the northern mountain ridge of Guatemala and forest species above 1,800 m.
b
Key for relevant species: Ca: carrion F: frugivore G: granivore I: insectivore O: omnivore C: raptor, carnivore N: nectarivore
c
Key: PF: more than 75% of detections in primary forest SF: more than 75% of detections in secondary forest
d
TC: transect counts, N: mist netting, I: incidental record, blank field: not recorded in the Chelemhá Plot.
e
Adapted from transect count data, mist netting data was not used here.
 Table 7. Abundance and estimated population size for threatened forest birds of the Sierra Yalijux. The first 21 entries listed are restricted-range species of
the Northern Central American Highlands (Stattersfield and Capper 2000).

# Species Global statusa Endemicb Mean individuals 102 ha (primary Sierra Yalijux (primary
in primary forest forest) ind./102 ha forest) ind./5500 ha
ind./51 ha*

1 Podilymbus gigas Extinct X NO

2 Oreophasis derbianus Vulnerable X Extinct (app. 1990) 0 0
3 Cyrtonyx ocellatus Near threatened X Non in PF <1 max. 55
4 Otus barbarus Near threatened X NO
5 Strix fulvescens Least concern X Non in PF 0 0
6 Campylopterus rufus Near threatened X NO
7 Lampornis viridipallens Least concern X Non in PF 0 0
8 Lampornis sybillae Least concern X NO
[505]

9 Doricha enicura Least concern X Non in PF 0 0

10 Atthis ellioti Least concern X Non in PF 0 0
11 Asphata gularis Least concern X 3 6 330
12 Notiochelidon pileata Least concern X Non in PF 0 0
13 Xenotriccus callizonus Near threatened X NO
14 Troglodytes rufociliatus Least concern X 14 28 1540
15 Melanotis hypoleucus Least concern X 11 22 1210
16 Turdus rufitorques Least concern X 1 2 110
17 Tangara cabanisi Endangered X NO
18 Ergaticus versicolor Vulnerable X Non in PF 0 0
19 Icterus maculialatus Least concern X NO
20 Carduelis atriceps Least concern X Non in PF 0 0
21 Cyanocorax melanocyaneus Least concern X 1 2 110

1565
 1566
Table 7. (Continued).

# Species Global statusa Endemicb Mean individuals 102 ha (primary Sierra Yalijux (primary
in primary forest forest)ind./102 ha forest) ind./5500 ha
ind./51 ha*
Additional species of conservation concernc
22 Pharomachrus mocinno Near threatened 17 34 1870
23 Penelopina nigra Near threatened 16 32 1760
24 Chlorospingus ophthalmicus 163 336 18,480
[506]

25 Zimmerius vilisimus 26 52 2860

a
adapted from Stattersfield and Capper (2000).
b
Endemics classified as endemic and part of the Endemic Bird Area ‘Central American highlands’ after Stattersfield et al. (1998).
c
Forest species that are found to be affected by land use or have low population sizes based on data of this study.
*NO: no records, neither in the Chelemhá Plot, nor in Sierra Yalijux.
In total Guatemala, 1 species is extinct, 4 are endangered, 2 vulnerable, 12 near threatened and 1 lacks sufficient data according to Stattersfield and
Capper (2000).
PF: primary forest.
 1567

both habitats, the largest group in terms of numbers of species was insectivores
(24 sp. in primary vs. 19 sp. in secondary forest), followed by omnivores (9 vs.
11 sp.). However, in primary forest omnivores were much more abundant than
insectivores. This was mainly a consequence of the large numbers of the
tanager C. ophthalmicus in primary forest which is responsible for roughly half
(45% drop in C. ophthalmicus from primary to secondary forest) of the 56%
drop in individual numbers of omnivores in secondary forest. M. unicolor and
Cyanolyca pumilo are the second and third species with the largest drop of
individuals between primary and secondary forest for omnivores with 8% and
4%, respectively. If these three species (C. ophthalmicus, M. unicolor, and
C. pumilo) are excluded from the analyses, omnivores have a percentage of
25.2% individuals in primary forest and 15.4% in secondary forest. When
excluding the three species, insectivores stay the largest group but granivores
double their relative proportions (10.5% vs. 22.8%) (Table 4).
The only other guild that showed remarkable differences in abundance
between habitats was granivores. Granivores were more than twice as abun-
dant in secondary forest (219% of primary forest) than in primary forest, a
result largely caused by high numbers of the emberizids Atlapetes gutteralis
(responsible for 34% of the increase) and Carduelis notata (11%), but also of
the dove Columba fasciata (2% decrease). When excluding the three species,
granivores are represented with 10.9% in secondary forest and 0.6% in pri-
mary forest.
Numbers of individuals and species of frugivores, nectarivores and carni-
vores are each represented in similar proportions in both habitats. When
excluding these six species as mentioned before the differences in individuals
per guild are still significantly different with p<0.01 (Table 7).

Recaptures

There were 25 species with at least one recapture, of which seven had recap-
tures in both habitats. Twelve species were recaptured exclusively in primary
forest and six exclusively in secondary forest. However, using different species
as sampling units, the numbers of recaptures for all 25 species were not sig-
nificantly different between the two habitats (Wilcoxon test, T=125.0,
Z=0.714; p = 0.48).
Members of five species endemic to Central American highland forests (not
to be confused here with the Endemic Bird Area; Stattersfield et al. 1998) were
recaptured in primary forest (Abeillia abeillei, D. baritula, Empidonax flaves-
cens, Asphata gularis, Melanotis hypoleucus). The recapture rate Rt is £ 0.01,
except for E. flavescens with a recapture rate of Rt = 0.20 in primary forest.
Leaving species level and going one step beyond to the individual level, of all
140 recaptures (plus 28 excluded same-day recaptures) 51 individuals were
recaptured at the same net line (Table 5). Except for three individuals recap-
tured at the same net line, all marked individuals were recaptured at different

[507]
1568

localities within the same habitat, i.e. at one of the six net lines in the same
habitat but differing from the original first-capture net line. Individual recap-
tures show that just three out of 180 were recaptured in a habitat different from
previously captures during the study period in 2001 and 2002. Two of the three
were first captured in primary forest and one moved into primary forest after
being marked in secondary forest.
Two habitat-switching individuals had a higher body mass in secondary
forest than in primary forest. A female L. amethystinus first captured in pri-
mary forest on 20 March 2002 was recaptured in secondary forest on 12 April
2002 (5.5 g fi 6.5 g). C. frantzii first captured in primary forest on 17 March
2002 was recaptured in secondary forest on 17 May 2002 (26.0 g fi 28.3 g).
C. ophthalmicus banded in secondary forest on 17 July 2001 was recaptured in
primary forest on 17 April 2002 with dramatically decreased body mass
(16.5 g fi 8.5 g).
Of all recaptures, 28 individuals were recaptured twice, one individual of
B. belli three times within three months in primary forest at neighbouring mist
nets.

Body mass

Individual recaptures (within same habitat and within the same season, i.e.
within the 2001 or 2002 netting period) indicate that 10 individuals lost body
mass in secondary forest and 12 in primary forest. While eight species were
heavier in primary forest, 14 had higher body mass in secondary forest. For all
species with both more than five captures in each habitat
(Nprimary forest  5 \ Nsecondary forest  5), the species’ body mass distribution was
not significantly different between habitats (v2-test, p>0.05). Forty-eight
individuals were recaptured and body mass was determined (Table 5).
Thirteen of the 23 groups (for definition of ‘group’ see methods) with at least
two individuals in each habitat had a higher mean body mass in secondary
forest. Eight groups had higher body mass in primary forest. The body mass
differences are not significant (MANOVA, variable: ‘habitat’, p = 0.21, Post-
hoc: Newman-Keuls test).
Body mass distribution did not indicate that primary forest represented a
different habitat quality compared to secondary forest in the Sierra Yalijux
(Table 5). Also, the total biomass was generally similar.

Restricted-range species

From the 21 restricted-range species of the Central American Highlands, 10 are

considered to be mainly or exclusively forest birds (Stattersfield et al. 1998).
Within the study plot, 14 range restricted species were recorded (Table 7) of
which one, the Mountain Guan Oreophasis derbianus, might be regarded as

[508]
 1569

extinct in the recent past (Table 6). Eight of the remaining 13 restricted-range
species were exclusively observed in secondary forest in the Sierra Yalijux (see
also Table 7). Of the remaining five species, four did not reveal important
differences in abundance between primary and secondary forest. One re-
stricted-range species, T. rufociliatus, was significantly more abundant in pri-
mary than secondary forest (Table 3).

Discussion

According to the species–area relationship, it is to be expected that defores-

tation in the study region (Sierra Yalijux) should result in loss of both bird
species and individuals. Even if deforestation stopped immediately and the
remaining primary forest area was preserved for the future, the 10 forest
species currently present at the study site in low numbers (Table 7) have an
increased risk of extinction (see also Máñez-Costa and Renner 2005). Species
and area are related according to the formula S = cAz, where S = number of
species, A = area, and z and c are empirically proven constants (MacArthur
and Wilson 1967; Rosenzweig 1995; Waltert et al. 2004; Brooks et al. 1997,
1999a, b and c). Assuming the bird’s sub-populations of the Sierra Yalijux have
limited genetic exchange with other sub-populations in the vicinity (e.g., the
Sierra de las Minas beyond the Polochic valley, 20 km south of the Sierra
Yalijux), one can set Asurviving to 5500 ha as the remaining primary montane
cloud forest (Mühlenberg et al. 1989) with Aoriginal = 16,500 ha for the Sierra
Yalijux (Mühlenberg et al. 1989). The original area of EBA # 018 Central
American highlands is Aoriginal = 1,500,000 ha (Stattersfield et al. 1998). Out
of the 21 species endemic to the Central American highlands (EBA # 018
‘Central American highlands’), ten species are found in primary forest and
oak–pine forests (Stattersfield et al. 1998). Applying the species area relation-
ship it is theoretically expected that only 2.68 endemic species will remain
(Ssurviving) in the 5500 ha of forest remnants in the Sierra Yalijux. In contrast,
13 endemic species were still present, more species than might be carried by
primary forest on basis of the species–area equilibrium. Even if the forest area
is not decreasing and the deforestation rate immediately halted, an optimistic
estimate of six to seven (or more) out of the observed 10 primary forest
highland endemics might not survive.
In order to obtain information on potential historical bird species loss from
the study plot before fieldwork, we established a list of bird species potentially
occurring in the Sierra Yalijux. One-hundred-forty-one species are expected to
occur above 1980 m in northern central Guatemalan highlands (Howell and
Webb 1995), including residents but no migrants as classified by Howell and
Webb (1995). We can assume that of the 141 species expected to occur, nearly
41 species have already become extinct or were not recorded at our study site.
However, it is most likely, that we did not record all species of the Sierra
Yalijux within the 102 ha, because (i) area and species are related by S = cAz

[509]
1570

(see above) and (ii) most species are patchily distributed and/or rare (Thiollay
1994a and b). Even in 100 ha of tropical landscapes ca. 95% of species are
recorded with the methods we used according to Terborgh et al. (1990). The
second aspect would suggest 133 recorded species at the Sierra Yalijux (95% of
141 species), the missing 33 species may be the result of the area–species-
relationship and extinction. The area–species-relationship suggests the presence
of 107 species at our study site, assuming here Aoriginal = 5500 ha, Asurviving =
102 ha, Soriginal = 133, and a mean empirically proven z-value = 0.26 for
tropical bird communities (Brooks et al. 1997, 1999a, b and c). Therefore seven
species are most likely already extinct. Due to the fact that this calculation is
not a species-specific analysis, the nomination of the seven species is impossi-
ble. Nevertheless, based on the previous analysis, it appears that at least seven
species are extirpated, and more species are likely to vanish from the study plot.

Land-use effects on the bird community

Differences at the community level between primary and secondary forest were
generally small. Species richness and species similarity were comparable be-
tween habitat types. Only the numbers of individuals revealed a significantly
higher bird abundance in secondary than in primary forest. While multi-
dimensional scaling of the community data yielded significant differences when
using only mist net captures, the differences for transect counts were not sig-
nificant. This might be the result of a collapsed stratification of the bird
community trapped in secondary forest, where even canopy species occur at
levels that make them vulnerable to capture.
Accurate comparisons of the importance of one habitat vs. another requires
measurement of individual fitness. The question of whether species of con-
servation interest (e.g., endemics and specialists, see Table 7) are able to find
suitable conditions to reproduce in secondary forest as well as they can in
primary forest is of considerable interest. Reproductive success for P. mocinno
is most likely directly linked to primary forest (Renner 2005). The species
breeds in primary forest and finds stable breeding sites only there (Renner
2003; Renner 2005; compare also Gillespie 2001). The same potentially applies
for other forest species and endemics listed in Table 7 (cf. Renner 2003).
Some authors affirm that young secondary forest and old secondary forest
are suitable to preserve a degree of diversity and species richness comparable to
primary forests (Shankar-Raman et al. 1998; Shankar-Raman 2001; Shankar-
Raman and Sukumar 2002). Older secondary forest (10+ years) might be
suitable for preservation of biodiversity and a high degree of species richness,
but unfortunately, at least in the study region of central Guatemala, secondary
forests older than 7 years are rare, even more rare than primary forest.
Therefore, primary forest must be preserved as long as data are lacking to
indicate that young secondary forest might also be suitable for conservation of
biodiversity. Currently and most likely in the future also, young secondary

[510]
 1571

forest is more frequent than old secondary forest and primary forest in the
Sierra Yalijux. The relative proportion of the different types of forest areas is a
consequence of expanding slash-and-burn agriculture and ongoing agricultural
activities in secondary re-growth areas (Markussen 2004; Markussen and
Renner 2005).

Guild composition

Guild structure is different for the two habitats, indicating that the resources
are not equal. While the percentage of insectivore bird individuals did not differ
significantly between primary forest and secondary forest, there is a higher
portion of insectivore species in primary forest than in secondary forest,
probably an indication of a more complex vegetation structure and absolute
numbers show significant differences (see results). Nectarivore species numbers
differ slightly and granivore species numbers are almost doubled in secondary
forest. This indicates that the food availability for species and for individuals
per species is higher in secondary forests.
One striking difference between primary and secondary forest is the large
amount of granivores in the latter, twice that of primary forest. This is
attributed to the larger supply of granivore food. Omnivores are more abun-
dant in primary forest, which superficially implies that there are less specialised
resources and species present. However, most omnivores in primary forest are
ground dwellers, which is obvious because the vegetation structure near ground
in secondary forest is much denser than in primary forest (see ‘Vegetation’ and
compare Renner 2003).

Recaptures

Impact on tropical avifaunas is frequently estimated with recaptures (e.g.,

Lambert 1992; Johns 1992; Holbech 1996; Dranzoa 1998). Optimal habitats
are usually correlated with high captures and/or recaptures and pessimal
habitat quality with low captures and/or recaptures (cf. Reitsma et al. 2002; for
exceptions see Winker et al. 1995).
There were no significant differences between recaptures in the two surveyed
habitats of the Sierra Yalijux. This might be due to the fact that the amount of
immature individuals (as far as recognisable for species, e.g. for C. ophthal-
micus 1:0.5 in primary forest before and 1:1.5 in secondary forest after fledging)
is larger in secondary forest than in primary forest. This implies that immatures
move to sub-optimal or pessimal habitats before potentially occupying an
optimal habitat in primary forest (compare Winker et al. 1995; Waltert and
Mühlenberg 2001; Reitsma et al. 2002).
Twelve species were recaptured exclusively in primary forest and six exclu-
sively in secondary forest, indicating that primary forest might be a ‘better’

[511]
1572

habitat for birds and could contain more territories per area (cf. Renner 2003;
Matthysen et al. 1995; Reitsma et al. 2002).

Conclusions

Results indicate that, in the cloud forests of the Sierra Yalijux, specialised
forest species are extinct or at least diminished in remaining primary forest and
mostly habitat generalists have survived which also inhabit secondary forest.
Even abundant species that are recorded in both habitats (C. ophthalmicus)
show lower populations in secondary forest (Table 7). Based on our calcula-
tions of population sizes, the forest-dependent species P. mocinno, P. nigra and
Z. vilissimus are threatened since they already have small populations.
P. mocinno depends on nesting sites (nest holes in rotting trees), only found in
primary forest. This species will not survive complete deforestation, even when
few breeding holes remain in old secondary forest for approximately 25 years
(Renner 2005). P. nigra is feeding in old secondary forest also, but not in slash-
and-burn agriculture or young secondary forest. Both species are still abun-
dant, but they will disappear when primary forests vanish. Z. vilissimus has
twice as many individuals in primary forest than in secondary forest and will
suffer from deforestation and further fragmentation. All forest endemics from
the Central American highlands (Table 7) will be affected by definition. Con-
trastingly, generalists and non-forest birds and non-forest breeders will gain
from deforestation or at least will not be influenced.

Acknowledgements

This study was supported by the German Research Foundation (DFG) and
was part of the DFG GK 642/1 ‘‘Valuation and Conservation of Biodiversity
in Guatemala’’ and partly by the Gesellschaft für Tropenornithologie. John
Rappole and two unknown reviewers gave valuable comments and Aerin Jacob
was helpful with English. This study was performed with the current laws of
Guatemala and CONAP authorised the study (No. 139–2001). We also would
like to thank the local people of all nationalities and ethnic groups in the
community of Chelemhá supporting our fieldwork and discussing with us
interesting aspects of conservation and ‘development’.

References

American Ornithologists’ Union 1998. The Checklist of North American birds. American Orni-
thologists’ Union, Washington DC.
American Ornithologists’ Union 2000. 42nd supplement to the American Ornithologists’ Union
check-list of North American birds. Auk 117: 847–858.
American Ornithologists’ Union 2002. 43rd supplement to the American Ornithologists’ Union
check-list of North American birds. Auk 119: 897–906.

[512]
 1573

Begon M., Harper J. and Townsend C. 1996. Ecology. Individuals, Populations, and Communities.
Blackwell, London.
Bibby C.J., Burgess N.D. and Hill D.A. 1995. Bird Census Techniques. Neumann, Radebeul.
Bierregaard R. 1990. Avian communities in the understorey of Amazonian forest fragments. In:
Keast A. (ed.), Biogeography and Ecology of Forest Bird Communities. SPB Academic Pub-
lishing, Den Haag, pp. 333–343.
Bierregaard R. and Stouffer P. 1997. Understorey birds and dynamic habitat mosaics in Amazo-
nian rainforests. In: Laurence W. and Bierregaard R. (eds), Tropical Forest Remnants. Uni-
versity of Chicago Press, Chicago, pp. 139–155.
Brooks T.M., Pimm S.L. and Collar N.J. 1997. The extent of deforestation predicts the number of
threatened birds in insular south-east Asia. Conservation Biology 11: 382–394.
Brooks T.M., Tobias J. and Balmford A. 1999a. Deforestation and bird extinctions in the Atlantic
forest. Animal Conservation 2: 211–222.
Brooks T.M., Pimm S.L., Kapos V. and Ravilious C. 1999b. Threat from deforestation to montane
and lowland forest birds and mammals in insular south-east Asia. Journal of Animal Ecology 68:
1061–1078.
Brooks T.M., Pimm S.L. and Oyugi J.O. 1999c. Time lag between deforestation and bird extinction
in tropical forest fragments. Conservation Biology 13: 1140–1150.
Chao A. and Lee S. 1992. Estimating the number of classes via sample coverage. Journal of the
American Statistical Association 87: 210–217.
Colwell R. 2000. EstimateS 6.0: Statistical estimation of species richness and shared species from
samples. Version 6.0. http://viceroy.eeb.uconn.edu/estimates.
Colwell R. and Coddington J. 1994. Estimating terrestrial biodiversity through extrapolation. Philo-
sophical Transactions of the Royal Society of London. Series B: Biological Sciences 345: 101–118.
Dranzoa C. 1998. The avifauna 23 years after logging in Kibale National Park, Uganda. Biodi-
versity and Conservation 7: 777–797.
Edwards E. 1998. A Field Guide to the Birds of Mexico and Adjacent Areas. Belize, Guatemala,
and El Salvador. University of Texas Press, Austin.
Eisermann K. 2000. Avifaunistisch-ökologische Untersuchungen in der Sierra Caquipec, Coban,
Alta Verapaz, Guatemala. Fachhochschule Eberswalde, Eberswalde.
FAO 2001. State of the Worlds Forests. FAO, Roma.
Gaston K.J. and Blackburn T. 2000. Pattern and Process in Macroecology. Blackwell, London.
Gilbert G., Gibbons D.W. and Evans J. 2000. Bird Monitoring Methods. A Manual of Techniques
for Key UK Species. A & C Black Publishers.
Gillespie T.W. 2001. Application of Extinction and Conservation Theories for Forest Birds in
Nicaragua. Conservation Biology 15: 699–709.
Holbech L. 1996. Faunistic diversity and game production contra human activities in the Ghana
high forest zone with references to the western region. Dissertation, University of Copenhagen,
Copenhagen.
Holland S. 2003. Analytic rarefaction 1.3. http://www.uga.edu/ strata/software/.
Howell S.N.G. and Webb S. 1995. A Guide to the Birds of Mexico and Northern Central America.
Oxford University Press, Oxford.
Hughes J., Daily G. and Ehrlich P. 2002. Conservation of tropical forest birds in countryside
habitats. Ecology Letters 5: 121–129.
Hurlbert S.H. 1984. Pseudoreplication and the design of ecological field experiments. Ecological
Monographs 54: 187–211.
Islebe G.A. 1995. High elevation coniferous vegetation of Guatemala: A phytosociological
approach. Vegetatio 116: 7–24.
Johns A. 1992. Vertebrate responses to selective logging: implications for the design of logging
systems. Philosophical Transactions of the Royal Society of London. Series B: Biological Sci-
ences 335: 437–442.
Kappelle M. and Brown A.D. 2001. Bosques Nublados Del Neotrópico. Instituto Nacional de
Biodiversidad, San José.

[513]
1574

Krebs C. 1999. Ecological Methodology. Addison-Welsey Educational Publisher, Menlo Park.

Lambert F. 1992. The consequences of selective logging for Bornean lowland birds. Philosophical
Transactions of the Royal Society of London. Series B: Biological Sciences 335: 443–457.
Land H. 1970. Birds of Guatemala. Livingston Publishing Company, Wynnewood.
MacArthur R.H. and Wilson E.O. 1967. The Theory of Island Biogeography. Princeton University
Press, Princeton, New Jersey.
MAGA 2001. Mapas digitales a escala 1:250,000 elaborados por el Sistema de Información Ge-
ográfica (SIG-MAGA). Ministerio de Agricultura, Ganaderı́a y Alimentación, Guatemala.
Magurran A. 1988. Ecological Diversity and its Measurement. Princeton University Press,
Princeton.
Máñez-Costa M.A. and Renner S.C. 2005. Incentives for conservation. The importance of envi-
ronmental measures in farming systems for bird populations in a fragmented landscape. A case
study in Guatemala. In: Markussen M., Marggraf R., Buse R., Garrelts H., Máñez-Costa M.A.
and Menzel S. (eds), Valuation and Conservation of Biodiversity – Interdisciplinary Perspectives
on the Convention on Biological Diversity. Springer Verlag, Berlin, Heidelberg, pp. 343–356.
Markussen M. 2004. Waldkonversion und Bodendegradation in Bergnebelwaldgebieten Guatem-
alas (Alta Verapaz) - Untersuchungen zur Biodiversitätsforschung unter besonbderer Ber-
ücksichtigung der landschaftsökologsichen Kompartimente in sensiblen Ökosystemen. Ecoregio
11. Shaker Verlag, Berlin.
Markussen M. and Renner S.C. 2005. Biotic and abiotic aspects as a base for the conservation of
biodiversity in a tropical montane cloud forest (Guatemala). In: Markussen M., Marggraf R.,
Buse R., Garrelts H., Máñez-Costa M.A. and Menzel S. (eds), Valuation and Conservation of
Biodiversity – Interdisciplinary Perspectives on the Convention on Biological Diversity. Springer
Verlag, Berlin, Heidelberg, pp. 357–372.
Marsden S.J., Whiffin M., Galetti M. and Fielding A.H. 2004. How well will Brazil’s system of
Atlantic forest reserves maintain viable bird populations? Biodiversity and Conservation XX:
1–19.
Matthysen E., Lens L., van Dongen S., Verheyen G., Wauters L., Adriansen F. and Dhont A. 1995.
Diverse effects of forest fragmentation on a number of animal species. Belgium Journal of
Zoology 125: 175–183.
McGowan P. and Gillman M. 1997. Assessment of the conservation status of partridges and
pheasants in south East Asia. Biodiversity and Conservation 6: 1321–1337.
Moziño J.M. 1996. Flora de Guatemala. Ediciones Doce Calles, Madrid.
Mühlenberg M., Hovestadt T. and Unger D. 1989. Schutz des Nebelwaldes in Guatemala.
Ökologische Station der Universität Würzburg, Würzburg.
Myers N., Mittermeier R., Mittermeier C., da Fonseca G. and Kent J. 2000. Biodiversity hotspots
for conservation priorities. Nature 403: 853–858.
Poulsen B. 1994. Mist netting as a census method for determining species richness and abundances
in an Andean cloud forest bird community. Le Gerfaut 84: 39–49.
Reitsma L., Hunt P., Burson S. and Steele B. 2002. Site fidelity and ephemeral habitat occupancy:
Northern Waterthrush use of Puerto Rican Black Mangroves during non-breeding season.
Wilson Bulletin 114: 99–105.
Remsen J. 1994. Use and misuse of bird lists in community ecology and conservation. Auk 111:
225–227.
Remsen J. and Good D. 1996. Misuse of data from mist net captures to assess abundance in bird
populations. Auk 113: 381–398.
Renner S.C. 2003. Structure and diversity of cloud forest bird communities in Alta Verapaz,
Guatemala, and implications for conservation. Niedersächsische Staats- und Universitätsbib-
liothek, (http://webdoc.sub.gwdg.de/diss/2003/renner/index.html).
Renner S.C. 2005. The Resplendent Quetzal (Pharomachrus mocinno) in the Sierra Yalijux, Alta
Verapaz, Guatemala. Journal of Ornithology 146: 79–84.
Rosenzweig M.L. 1995. Species Diversity in Space and Time. Cambridge University Press, Cam-
bridge.

[514]
 1575

Schulze C.H., Waltert M., Kessler P., Pitopang R., Shahabudin, Veddeler D., Steffan-Dewenter I.,
Mühlenberg M., Gradstein S., Leuschner C. and Tscharntke T. 2004. Biodiversity indicator
groups of tropical land-use systems: comparing plants, birds and insects. Ecological Applica-
tions. Vol. 14(5): 1321–1333.
Shankar-Raman T. 2001. Effect of slash-and-burn shifting cultivation on rainforest birds in Mi-
ozoram, north-east India. Conservation Biology 15: 685–698.
Shankar-Raman T., Rawat G. and Johnsingh A. 1998. Recovery of tropical rainforest avifauna in
relation to vegetation succession following shifting cultivation in Mizoram, north-east India.
Journal of Applied Ecology 35: 214–231.
Shankar-Raman T. and Sukumar R. 2002. Responses of tropical rainforest birds to abandoned
plantations, edges and logged forest in the Western Ghats, India. Animal Conservation 5: 201–
216.
Stattersfield A. and Capper D. 2000. Threatened Birds of the World. BirdLife International, Lynx
Ediciones, London, Barcelona.
Stattersfield A., Crosby M., Long A. and Wege D. 1998. Endemic Bird Areas of the World.
Priorities for Biodiversity and Conservation. Bird Life International, London.
Terborgh J., Robinson S.K., Parker T.A., Munn C. and Pierpont N. 1990. Structure and orga-
nization of an Amazon forest bird community. Ecological Monographs 60: 213–238.
Terborgh J. 1999. Requiem for Nature. Island Press, Washington DC.
Thiollay J.M. 1994a. Disturbance, selective logging and bird diversity: a Neotropical forest study.
Biodiversity and Conservation 6: 1155–1173.
Thiollay J.M. 1994b. Structure, density and rarity in an Amazonian rainforest bird community.
Journal of Tropical Ecology 10: 449–481.
Veblen T. 1976. The urgent need for forest conservation in highland Guatemala. Conservation
Biology 4: 141–154.
Waltert M., Langkau M., Maertens M., Härtel M., Erasmi S. and Mühlenberg M. 2004. Predicting
losses of bird species from deforestation in Central Sulawesi. In: Gerold G., Guhardja E. and
Fremerey M. (eds), Land Use, Nature Conservation and the Stability of Rainforest Margins in
Southeast Asia. Springer, Berlin, Heidelber, pp. 327–338.
Waltert M. and Mühlenberg M. 2001. Zur Beziehung zwischen Abundanz und Habitatqualität:
Fangraten territorialer Gelbbartbülbüls Andropadus latirostris. In: Gottschalk E., Barkow A.,
Mühlenberg M. and Settele J. (eds), Naturschutz und Verhalten. Umweltforschungszentrum
Halle Leipzig, Leipzig, pp. 103–110.
Winker K., Rappole J.H. and Ramos M. 1995. The use of movement data as an assay of habitat
quality. Oecologia 101: 211–216.
World Bank. 2001. World development indicators 2001. Worldbank, Washington, DC.
World Conservation Monitoring Centre. 1992. Global Biodiversity. Status. Chapman & Hall,
London.

[515]
Biodiversity and Conservation (2006) 15:1577–1607
Springer 2006
DOI 10.1007/s10531-005-2352-5

-1

Effects of habitat structure and adjacent habitats

on birds in tropical rainforest fragments and shaded
plantations in the Western Ghats, India

T.R. SHANKAR RAMAN

Centre for Ecological Research and Conservation, Nature Conservation Foundation, 3076/5 IV Cross,
Gokulam Park, Mysore 570002, India; Address for correspondence: Rainforest Restoration Research
Station, 8/364 Cooperative Colony, Valparai 642127, Tamil Nadu, India (e-mail: podocarp@
vsnl.net; phone: +91-4253-221527; fax: +91-821-2513822)

Received 31 August 2004; accepted in revised form 4 August 2005

Key words: Agriculture, Anamalai hills, Bird communities, Cardamom, Coffee, Countryside bio-
geography, Fragmentation, Habitat structure and floristics, Landscape matrix, Tropical wet
evergreen forest

Abstract. As large nature reserves occupy only a fraction of the earth’s land surface, conservation
biologists are critically examining the role of private lands, habitat fragments, and plantations for
conservation. This study in a biodiversity hotspot and endemic bird area, the Western Ghats
mountains of India, examined the effects of habitat structure, floristics, and adjacent habitats on
bird communities in shade-coffee and cardamom plantations and tropical rainforest fragments.
Habitat and birds were sampled in 13 sites: six fragments (three relatively isolated and three with
canopy connectivity with adjoining shade-coffee plantations and forests), six plantations differing
in canopy tree species composition (five coffee and one cardamom), and one undisturbed primary
rainforest control site in the Anamalai hills. Around 3300 detections of 6000 individual birds
belonging to 106 species were obtained. The coffee plantations were poorer than rainforest in
rainforest bird species, particularly endemic species, but the rustic cardamom plantation with
diverse, native rainforest shade trees, had bird species richness and abundance comparable to
primary rainforest. Plantations and fragments that adjoined habitats providing greater tree canopy
connectivity supported more rainforest and fewer open-forest bird species and individuals than sites
that lacked such connectivity. These effects were mediated by strong positive effects of vegetation
structure, particularly woody plant variables, cane, and bamboo, on bird community structure.
Bird community composition was however positively correlated only to floristic (tree species)
composition of sites. The maintenance or restoration of habitat structure and (shade) tree species
composition in shade-coffee and cardamom plantations and rainforest fragments can aid in
rainforest bird conservation in the regional landscape.

Introduction

Fragmentation of tropical forests creates landscapes that typically contain

many habitat remnants surrounded by relatively inhospitable environments.
Studies of tropical forest fragmentation are providing increasing evidence that
the ability of species to survive in fragments may depend on surrounding
habitats and whether the species uses such habitats (Stouffer and Bierregaard
1995a, b; Laurance et al. 1997; Renjifo 2001). This landscape approach to

[517]
1578

conservation of rainforest fragments requires an assessment of the conserva-

tion values per se of lands outside conservation reserves that may include
habitat fragments, secondary forests, private lands, and countryside habitats
such as agricultural plantations (Brown and Lugo 1990; Turner and Corlett
1996; Daily 2001). Approaches that supplement habitat protection by con-
serving wildlife habitats adjoining protected areas, increasing landscape-level
connectivity of patches, and restoring degraded areas are thus gaining ground
as a means to conserve biological diversity (Laurance et al. 1997).
Such an approach to conservation is pertinent to the Western Ghats hill
ranges of India. This hill chain is recognised as a global biodiversity hotspot
(Myers et al. 2000), one of the Global 200 most important eco-regions (Olson
and Dinerstein 1998), and an Endemic Bird Area (Stattersfield et al. 1998). The
Western Ghats faces severe threats from human disturbance due to defores-
tation, developmental activities, conversion to plantations, and habitat frag-
mentation (Nair 1991). Between 1920 and 1990, forest cover in the region
declined by 40%, resulting in a four-fold increase in the number of fragments,
and an 83% reduction in size of forest patches (Menon and Bawa 1997). This is
partly due to this region being one of the hotspots with the highest human
population densities (Cincotta et al. 2000).
One of the major causes of forest fragmentation in the Western Ghats is the
spread of plantations, particularly tea, coffee, and Eucalyptus. The area under
plantations is large and growing. Tea plantations in the south Indian states
increased by 17.7% in the period 1987–1998 from 74,765 to 87,993 ha (Tea
Board 2002). Large areas of Eucalyptus plantations also occur with tea as it is
used as fuel-wood for tea-curing in the factories. Similarly, during 1999–2000,
the US$ 447 million Indian coffee industry had plantations of about
340,306 ha, almost entirely in the Western Ghats region of southern India,
having increased in area coverage by 25.7% from 270,821 ha in 1990–1991
(Coffee Board 2001). These coffee plantations, particularly where grown tra-
ditionally under the shade of native forest trees, form a substantial area of
forest canopy cover in the Western Ghats.
Although shade-coffee plantations are known to support many forest
species, especially birds (Estrada et al. 1997; Greenberg et al. 1997a, b;
Perfecto and Vandermeer 2002; Tejeda-Cruz and Sutherland 2004), their
conservation value is still debated (Rappole et al. 2003). This is partly because
there is considerable variation in bird species richness and composition across
coffee plantations differing in habitat structure and shade tree mixtures
(Greenberg et al. 1997a, b; Wunderle 1999; Perfecto et al. 2003; Tejeda-Cruz
and Sutherland 2004). Although it has been suggested that such plantations
may act as effective refuges and buffer habitats in fragmented landscapes
(Shahabuddin 1997), the landscape-level influence of forest fragments and
plantations on each other is poorly understood (Perfecto et al. 2003). This
study examines bird communities in a fragmented landscape of plantations and
rainforest remnants in the Anamalai hills, southern Western Ghats of India.
The study had two main objectives:

[518]
 1579

1. What is the influence of the landscape matrix adjacent to or surrounding

fragments on tropical rainforest birds? This question is explored by com-
paring fragments that adjoin shade-coffee plantations or other forest types
with those that are ‘‘isolated’’ being surrounded by tea plantations, a rela-
tively inhospitable habitat for birds.
2. What is the influence of the composition of shade tree species mixture in
coffee estates on the bird community? Rainforest sites were compared with
plantations with a near-monoculture of exotic shade tree species, with mixed
canopy, and with diverse, native rainforest tree canopy to test the hypoth-
esis that greater similarity in tree species composition with rainforest
supports a more rainforest-like bird community.

Study area

The Western Ghats is a 1600 km long chain of hills running along the west coast
of the Indian Peninsula (8–21 N) and is recognised as a unique biogeographic
province (Mani 1974). Moist forests, including tropical wet evergreen rainforest,
are found largely south of 16 N, particularly south of the Palghat Gap at 11 N,
a region often called the southern Western Ghats (Pascal 1988). The Anamalai
hill ranges are a major conservation area in the southern Western Ghats con-
taining mid-elevation rainforest in the Indira Gandhi Wildlife Sanctuary
(958 km2, 1012¢ N to 1035¢ N and 7649¢ E to 7724¢ E) and in private-owned
fragments on the Valparai plateau (Figure 1). The natural vegetation of this
region, receiving around 3500 mm of rainfall annually, particularly during the
southwest monsoon (June–September), is classified as mid-elevation tropical
wet evergreen forest of the Cullenia–Mesua–Palaquium type (Pascal 1988).
The Valparai plateau contains around 220 km2 of tea, coffee, and cardamom
plantations surrounded by Wildlife Sanctuaries, National Parks, and Reserved
Forest. Clearing of primary rainforest for plantations began in 1896 and was
mostly complete by the 1930s, although some clearing and conversion of coffee
and cardamom to tea plantations continues to the present day (Congreve 1942,
Raman and Mudappa 2003a). The plateau has one town (Valparai) and a
population of over 106,000 people (1991 Census), mostly estate labourers,
scattered across the town and estates. At least 25 rainforest fragments have
been identified so far in and around the Valparai plateau (Umapathy and
Kumar 2000) and additional sites do exist. Besides two large fragments (2000
and 2600 ha) within the Indira Gandhi Wildlife Sanctuary, the remaining
rainforests all occur as fragments of 0.3–650 ha in size, much of which is on
private land. These fragments are vital for conservation as they contain sig-
nificant proportions of the native fauna (Umapathy and Kumar 2000) and
provide landscape-level connectivity between patches critical for wide-ranging
species such as the Great Hornbill (Buceros bicornis, Raman and Mudappa
2003b), Asian elephants Elephas maximus, tigers Panthera tigris, and wild dogs
Cuon alpinus (Kumar et al. 2002).

[519]
 1580
[520]

Figure 1. Map of the Indira Gandhi Wildlife Sanctuary showing location of the plantation areas in the Valparai plateau (within dashed lines), some
rainforest fragments (light grey), and reservoirs (stippled) on the Valparai plateau.
 1581

Avifauna

The Western Ghats contains 16 species of restricted-range birds including 12

of near-threatened conservation status (Collar et al. 1994; Stattersfield et al.
1998) and one endangered, three vulnerable, and seven near-threatened bird
species, of which all but two inhabit tropical rainforests (BirdLife Interna-
tional 2001). A number of other rare species with range largely restricted to
the Western Ghats and other mountain ranges in the Indian peninsula, Sri
Lanka, or the Himalaya, also occur in these tropical rainforests: e.g. Mal-
abar Trogon Harpactes fasciatus, Asian Fairy Bluebird Irena puella,
Mountain Imperial Pigeon Ducula badia, Jerdon’s Baza Aviceda jerdoni,
Black-crested Baza A. leuphotes, Sri Lanka Frogmouth Batrachostomus
moniliger. Of the 230 bird species identified in the Anamalai hills, around 90
are typical rainforest birds, including 13 endemic species (Kannan 1998;
Raman 2001).

Selection of study strata and sites

Thirteen sites were selected for vegetation and bird sampling: six rainforest
fragments, six plantation sites, and a ‘control’ or reference site containing a
large and relatively undisturbed tract of tropical rainforest in the same ele-
vation range (Table 1). Of the six fragments, three were relatively isolated as
they occurred within tea plantations that had a very sparse canopy of pruned,
non-indigenous silver oak (Grevillea robusta) trees planted at 12 m · 12 m
spacing. The remaining three fragments adjoined shade-coffee estates with
extensive canopy cover. The five shade-coffee plantations differed in the
canopy tree species composition and the cardamom plantation, maintained
by a local tribal settlement, contained a canopy entirely of native rainforest
tree species (rustic cardamom, see Table 1). Three plantation sites adjoined
continuous forest tracts within the Indira Gandhi Wildlife Sanctuary,
whereas the remaining three sites only adjoined smaller fragments that were
in private lands. The control site was an approximately 2600 ha tract of
tropical rainforest (Iyerpadi–Akkamalai complex) within the Indira Gandhi
Wildlife Sanctuary adjoining the plantations on the Valparai plateau. All sites
were within a restricted elevation range of 900–1400 m containing mid-ele-
vation tropical wet evergreen forest vegetation (Pascal 1988). Sites (except
largest fragments and plantations) were mapped by walking around them
with a hand-held GPS (Garmin 12 XL) with the track option activated. The
larger fragments and plantation sites were digitised using a combination of
Survey of India 1:50,000 topographic sheets, GPS tracking, and satellite
imagery. Maps were prepared and areas of sites estimated using MapInfo
Professional software (version 7.0). Areas of plantation sites are approximate
as exact areas and boundaries were not indicated by all the private
companies.

[521]
1582

Table 1. Study sites selected for vegetation and bird sampling in the Valparai plateau and Indira
Gandhi Wildlife Sanctuary (IGWLS), Anamalai hills; RF = rainforest fragment, P = plantation.

Code Site (Stratum, Description

adjacent habitat
canopy connectivity)

A Iyerpadi–Akkamalai 2600 ha primary rainforest in IGWLS with very

(Control, High) low disturbance levels
B Andiparai (RF, Low) 200 ha fragment surrounded largely by tea estates,
connected by narrow corridor to the control site
C Injipara (RF, Low) 18 ha fragment, an abandoned cardamom plantation,
with many exotic shade trees in the canopy, surrounded by
tea estates and a logged Eucalyptus fuel clearing area
D Korangumudi 56 ha fragment, an abandoned cardamom plantation
(RF, Low) with highly disturbed rainforest canopy, more isolated since
conversion of adjoining coffee to tea plantation in 2000
E Manamboli (RF, High) 200 ha primary rainforest fragment within IGWLS continuous
with large area of forest
F Puthuthottam Fragment 92 ha fragment, an abandoned cardamom plantation, with
(RF, High) highly disturbed rainforest canopy
G Tata Finlay Fragment 33 ha moderately disturbed fragment adjoining coffee estates
(RF, High) with mixed native and exotic tree canopy
H Puthuthottam Coffee c. 50 ha, mixed canopy dominated by the exotics Maesopsis
(P, Low) emenii, Erythrina indica and Eucalyptus sp., with few native
trees such as Artocarpus heterophyllus; adjoins Puthuthottam
fragment
I Siva Coffee (P, Low) 45 ha, dominantly exotic tree canopy of silver oak Grevillea
robusta; adjoins Korangumudi fragment
J Tata Finlay Coffee c. 100 ha, mixed tree canopy of exotics such as Erythrina
(P, Low) indica andMaesopsis emenii with existing native species such
as Cullenia exarillata, Mesua ferrea, and Palaquium ellipticum;
adjoins Tata Finlay fragment
K Surulimalai Coffee c. 50 ha, dominantly exotic tree canopy of Erythrina indica,
(P, High) continuous with Manamboli and forests in IGWLS
L Old Valparai Coffee c. 80 ha, mostly exotic tree canopy of Erythrina indica and
(P, High) Maesopsis emenii, continuous with rainforest fragments and
forests in IGWLS
M Sankarankudi c. 60 ha, established by clearing ground vegetation under a
Cardamom (P, High) completely native canopy of rainforest tree species, continuous
with rainforest fragments and forests in IGWLS

Methods

Vegetation sampling

In 12 sites, densities of trees greater than 30 cm girth at breast height (GBH at

1.3 m) were estimated using 15–25 point-centred quarter plots (PCQ sample of
60–100 trees/site, Krebs 1989). In the Siva Coffee plantation it was not possible
to do PCQ plots and hence 5 m radius circular plots (N = 25 plots and 52

[522]
 1583

trees) were laid and an additional 28 random trees were identified to species.
Tree species were identified using available guides (Gamble and Fischer 1935;
Pascal and Ramesh 1997). Shrubs (or in plantations: coffee bushes, cardamom
plants) were counted in 25 plots of 2 m radius and the presence of cane
(Calamus sp.), bamboos, and lianas was recorded within 5 m radius of the
centres of these 25 plots. Elevation was noted at these points using an altimeter.
Canopy variables (height, cover, stratification) and leaf litter depth were
measured at 25 points, evenly spaced 25 m apart as described elsewhere
(Raman and Sukumar 2002).

Bird sampling

I attempted to sample all sites in a relatively uniform and efficient manner over
the winter and breeding season (December to May) when both migrants and
residents were present in the study area. Point counts (Verner 1985; Bibby et al.
1992; Ralph et al. 1995) were used for bird surveys. Point count surveys of
5 min duration were carried out during the first three hours after sunrise when
bird activity was highest (see Raman 2003 for further details). Densities were
estimated using a fixed radius (50 m) approach as they are known to be highly
correlated to variable-radius point count estimates across species (Raman
2003). As some degraded fragment and plantation sites contained relatively
more open vegetation some bias due to detectability differences may have
existed and the results can only be taken as a conservative assessment of the
effects of fragmentation and plantations.
In each site, 30 point count surveys (25 in Sankarankudi cardamom and 26
in Manamboli) were carried out yielding 173–308 detections and an estimated
321–633 individual birds per site. I attempted to ensure independence of data
points to the extent possible by spacing out points and survey days. Successive
points sampled in any day were at least 100 m apart to avoid overlap and
intermediate points were sampled on different days. Although points sampled
on different days overlapped to some extent, the procedure followed ensured
uniform coverage of the site.

Data analysis

For each site, tree density and basal area were calculated using the PCQ
method (Krebs 1989). Average values across replicate samples in each site were
calculated for other habitat variables. Vertical stratification (average number
of strata with foliage) and its coefficient of variation (indexing horizontal
heterogeneity) were calculated following Raman et al. (1998). Tree species
richness was indexed by the number of tree species recorded in the PCQ plots.
The vegetation data was summarised by principal components analysis to
determine fewer uncorrelated components. The factor matrix was rotated by

[523]
1584

the Varimax method to assist in interpretation and display of the results

(Norušis 1990).
The 106 bird species recorded during the study were classified into rainforest
and open-forest (non-rainforest) birds. The rainforest species included birds
that normally occurred even in mature undisturbed rainforests in the southern
Western Ghats (Ali and Ripley 1983; Raman 2001). Open-forest birds occurred
only in disturbed rainforest fragments or in naturally drier and open habitats
and never in mature, undisturbed tropical rainforest. Analyses were performed
using all species, only rainforest species, and only open-forest species.
Bird species richness (per point and cumulative list) and bird abundance per
point were major parameters of interest. Flock sizes for aural detections of
birds were randomly selected from flock-size distribution data of each species
(Raman 2003). As only 25 point count surveys were carried out in one site, I
obtained rarefaction estimates of bird species richness for all sites for 24
sampled points using the program EstimateS (Colwell and Coddington 1994;
Colwell 1997; 100 permutations, sampling without replacement). Cumulative
species richness and abundance (individuals/ha) of birds belonging to different
species categories was also estimated: Western Ghats endemics, priority spe-
cies, migrants (all, rainforest, and open-forest migrants). Priority species were
defined as birds of restricted-range (Stattersfield et al. 1998), discontinuous
distribution (in rainforests of southwest India, Sri Lanka, and northeast India;
Ali and Ripley 1983), or near-threatened (Collar et al. 1994) and excluded
endemics.
The effects of stratum (rainforest fragment, plantation, or control), adjacent
habitat (with high or low tree canopy connectivity) and point count (repeated
measure) on bird richness and abundance were assessed in a multivariate
analysis of variance (MANOVA) as an alternative to repeated-measures
analysis of variance (Zar 1999). To jointly examine the influence of area,
adjacent habitat, and vegetation structure, multiple linear regression was used
to examine relationships between bird community parameters and four inde-
pendent variables: site area, level of canopy connectivity in adjacent habitat
(scored as 1 – low, 2 – medium, and 3 – high as in Table 1), and scores of two
vegetation components extracted by the principal components analysis (PC1
and PC2, see Results). The analysis was carried out on the mean values of bird
community parameters estimated for the N = 13 sites and a backward step-
wise selection procedure was used to identify statistically significant variables
(Zar 1999). As the nature (positive versus negative) and strength of influence of
these variables was of primary interest, only the standardized regression
coefficients (b, with significance levels from t-tests) are presented and ANOVA
and other regression diagnostics are omitted.
The bird species-abundance data was used to estimate similarities between
sites in bird community composition using the Morisita index (Wolda 1981).
Structural dissimilarity (using distribution of foliage in different vertical strata)
and floristic dissimilarity (tree species composition data) were estimated as
1–Morisita index (Raman et al. 1998). Partial Mantel tests were used to assess

[524]
 1585

the influence of floristic and structural dissimilarity on change in bird com-

munity composition with 10,000 permutations to assess statistical significance
(Hemelrijk 1990; Manly 1994). Analysis of similarities (ANOSIM) with the
Bray–Curtis index was used to assess significance of bird community compo-
sitional change between strata (Clarke and Warwick 1994; Clarke and Gorley
2001).
For analysis of habitat use by species, a simple deviation index (D) was
computed for each habitat stratum (rainforest control, fragment, shade-coffee,
and cardamom) as: Dij = (Obs xij
Exp xij)/(Obs xij + Exp xij), where Obs
xij= the average detections of the species i across replicate sites in stratum j,
and Exp xij = (nj/N)* ni, where ni=number of detections of species i,
nj=number of bird detections in stratum j, and N = total number of detec-
tions. Values of the deviation index ranged from
1 (avoidance) to +1
(preference). Values £
0.25 or ‡0.25 were considered to indicate significant
avoidance or usage of the habitat stratum.

Results

Vegetation structure in fragments and plantations

Foliage profile and vegetation attributes showed distinct differences across

sites. The foliage profile data on frequencies of foliage presence at different
vertical strata showed highly significant variation across the four habitats:
primary rainforest control, fragment, coffee, and cardamom (v2 = 89.92,
df = 21, p < 0.001, Table 2). At the highest strata (>32 m), shade-coffee
plantations practically lacked canopy foliage whereas other sites had 20–24%
of the points with foliage. The shade-coffee plantations tended to have foliage
concentrated in the 0–2 m interval (coffee shrubs) and 8–16 m interval (shade
trees), with less foliage in other strata. The cardamom plantation had foliage
concentrated in the 0–2 m interval (cardamom plants), and in the two highest
strata (tall rainforest tree species canopy), whereas intermediate strata
(2–16 m) had less foliage than rainforest as small trees and shrubs had been
Table 2. Variation in foliage distribution at different vertical levels in plantations, rainforest
fragments and primary rainforest control sites in the Anamalai hills.

Vertical strata (m) Control Fragment Coffee Cardamom

>32 20.0 20.0 – 24.0

24–32 45.0 46.0 8.0 76.0
16–24 80.0 50.7 45.6 72.0
8–16 65.0 60.7 83.2 56.0
4–8 80.0 62.0 59.2 48.0
2–4 75.0 58.0 48.0 32.0
1–2 85.0 76.7 94.4 72.0
0–1 90.0 96.0 95.2 88.0
Tabled values are means of estimates across sites in each stratum.

[525]
1586

cleared for planting cardamom (Table 2). Tree density, canopy cover, and litter
depth were highest in primary rainforest, intermediate in fragments, and least
in plantations. Canopy height and basal area were highest in the cardamom
plantation, and shrub density was highest in fragments (Table 3). Notably,
bamboos, canes, and lianas were absent from the plantation sites due to their
elimination during plantation maintenance and weeding.
Principal components analysis of the eleven vegetation variables extracted
two components (PC1 and PC2) accounting for 76.8% of the total variation in
the data-set (Table 4). PC1 was significantly positively correlated to variables
reflecting the density of woody plants, cane, and bamboo. PC2 was positively
correlated to vertical structure and canopy closure and negatively to hori-
zontal patchiness of sites (Table 4). The ordination of the sites using PC scores
shows the primary rainforest and two large fragments (Manamboli and
Andiparai) forming a cluster at the top right indicating a well developed
woody plant community, vertical structure, and canopy closure (Figure 2).
The plantation sites aggregate on the left indicating poorer development of

Table 3. Vegetation characteristics of the study strata in the Anamalai hills.

Parameter Rainforest Plantation

Control Fragment Coffee Cardamom

Number of sites (N) 1 6 5 1

Tree density (no./ha) 583.00 349.00 236.60 229.00
SE 57.16 62.78
Basal area (m2/ha) 61.38 62.28 27.83 89.54
SE 12.76 5.74
Altitude (m) 1304 1065 1118 1074
SE 64.6 27.8
Canopy cover (%) 99.40 88.71 68.79 87.72
SE 4.38 8.72
Canopy height (m) 26.05 23.28 15.25 28.52
SE 1.86 1.55
Shrub density (no./plot) 9.75 13.42 0.51 0.00
SE 2.65 0.45
Cut trees (no./plot) 0.40 0.74 0.00 0.00
SE 0.15 0.20
Vertical stratification (no. of strata) 5.35 4.71 4.14 4.68
SE 0.29 0.51
Horizontal heterogeneity 6.25 5.00 4.92 5.34
Leaf litter depth (cm) 7.33 4.31 3.13 2.52
SE 0.62 0.19 0.27
Bamboo prevalence (proportion of plots) 0.05 0.05 0.00 0.00
Cane prevalence (proportion of plots) 0.40 0.12 0.00 0.00
Liana prevalence (proportion of plots) 0.45 0.24 0.00 0.00
Coffee density (no./plot) 4.88 0.00
SE 0.37 0.00
Cardamom density (no./plot) 0.00 6.36
SE 0.00 0.61
Tabled values are means and standard errors (SE) across sites in each stratum.

[526]
 1587

Table 4. Principal components analysis of vegetation variables: correlations of original variables

with extracted components.

Variable PC1 PC2

Tree density 0.801*** 0.198

Basal area 0.682** 0.393
Shrub density 0.741** 0.152
Litter depth 0.849*** 0.213
Bamboo prevalence 0.845*** 0.032
Cane prevalence 0.956*** 0.088
Liana prevalence 0.968*** 0.131
Canopy cover 0.455 0.874***
Canopy height 0.442 0.569*
Vertical stratification 0.268 0.938***
Horizontal heterogeneity 0.321
0.821***
Eigenvalue 5.519 2.929
Cumulative variance explained (%) 50.17 76.79
*p < 0.05, **p < 0.01, ***p < 0.005.

Figure 2. Ordination of primary rainforest control (n), fragment (m), and plantation (d) sites in
and around the Valparai plateau using principal components analysis of vegetation variables. Site
codes as in Table 1.

woody plants, but varied in vertical structure from poorly developed

(Surulimalai and Siva coffee low on PC2) to well developed (Old Valparai
coffee and Sankarankudi cardamom). The remaining four fragments were
structurally interspersed among plantation sites, with the more disturbed ones
(Korangumudi and Injipara) being low on PC2 due to poorly developed
vertical structure (Figure 2).

[527]
1588

Bird community structure in plantations and fragments

Point count sampling yielded 3299 bird detections with an estimated 5987
individuals belonging to 106 species across the 13 sites. Of the 106 bird species,
70 (66%) were rainforest birds and 36 (34%) were open-forest (non-rainforest)
birds. Birds detected at least thrice during sampling comprised 78 species,
including 57 (73%) rainforest and 21 (27%) open-forest species. The total
numbers of bird species seen in the four main habitat strata were: 43 (primary
rainforest control – one site), 95 (rainforest fragments – five sites), 76 (shade-
coffee plantations – five sites), and 49 (cardamom plantation – one site). The
percentage of rainforest bird species was highest in the primary rainforest
control (95.3%) and the cardamom plantation under natural shade (89.8%).
More open-forest birds had infiltrated into rainforest fragments and shade-
coffee plantations and the percentage of rainforest bird species in the com-
munity was lower at 70.5% and 59.2%, respectively, in these strata. These
differences in the number of rainforest vs. open-forest species across the four
habitat strata were statistically significant v2 = 34.1, df = 3, p < 0.001).

Bird species richness and abundance: effects of stratum and adjacent habitat

Both stratum (control vs. rainforest fragment vs. plantation) and adjacent
habitat (with low vs. high canopy connectivity) had significant effects on bird
species richness and abundance (MANOVA, p < 0.001, Table 5). The repeated
measure (point), which represented the replicate samples taken within each site,
had no significant direct effect or 2-way or 3-way interactions with the other
main effects (stratum and connectivity) in the multivariate analysis (Table 5).
However, there was a significant 2-way interaction between stratum and
connectivity (p < 0.001, Table 5).

Table 5. Results of multivariate analysis of variance (MANOVA) on the effects of habitat stratum
and connectivity on total, rainforest, and open-forest bird species richness and abundance in point
count samples in the Anamalai hills.

Main Effects Wilks’ k F Hypothesis df Error df P

Intercept 0.097 529.42 4 228 0.000

Stratum 0.679 26.93 4 228 0.000
Adjacent habitat 0.785 7.32 8 456 0.000
Point 0.665 0.85 116 908.6 0.871
2-way interactions
Stratum · Adjacent habitat 0.903 6.12 4 228 0.000
Stratum · Point 0.694 0.75 116 908.6 0.973
Connectivity · Point 0.458 0.85 232 913.3 0.935
3-way interaction
Stratum · Adjacent habitat · Point 0.647 0.91 116 908.6 0.747

[528]
 1589

When individual variables were examined, habitat stratum had a significant

effect (p < 0.001) on rainforest bird species richness with plantations having
per-point richness values about one-half to a third lower than in the rainforest
control. Rainforest fragments had up to one-fourth lower rainforest bird
species richness than control sites but the means were higher than in planta-
tions (Figure 3). The presence of adjacent habitats with higher canopy con-
nectivity also resulted in higher rainforest bird species richness and abundance
in plantations and fragments (p < 0.001). The repeated-measure (point) did not
have a significant effect on species richness (p = 0.564) but was significant for
rainforest bird abundance (p = 0.008). There were no significant 2-way or
3-way interactions except for an interaction between connectivity and the
repeated measure factor (point) for rainforest bird abundance (p = 0.002).
Stratum and adjacent habitat had significant (p < 0.001) but reverse effects
on open-forest birds with the primary rainforest control site having fewest,
fragments intermediate, and plantations highest richness and abundance
(Figure 3). Fragments that adjoined shaded plantations with higher canopy
connectivity had lower open-forest bird richness and abundance. For planta-
tion sites, there was little effect of adjoining low and high canopy connectivity
habitats; thereby resulting in a significant interaction between connectivity and
stratum for open-forest bird species richness and abundance (p < 0.001). Open-
forest bird species richness and abundance showed no significant direct or
interaction effects with the repeated-measure factor (point).
As in the case of rainforest birds, richness and abundance of all birds
combined showed significant effects of stratum (p < 0.001) and adjacent hab-
itat (p < 0.009). In fragments, however, adjoining habitats with better canopy
connectivity had little effect on all birds taken together, as the positive effect on
rainforest birds was compensated by the negative effect on open-forest birds
(Figure 3). In plantation sites, total bird species richness and abundance was
enhanced in sites that adjoined rainforest fragments mainly due to the positive
effects it had on rainforest birds. There was thus a significant interaction effect
of stratum and adjacent habitat on total bird species richness (p < 0.001) and
abundance (p = 0.001). There was no significant direct or interaction effect of
the repeated measure (point) on total bird species richness. The repeated
measure factor had a significant direct (p = 0.006) effect, two-way interaction
with adjacent habitat (p = 0.001), and 3-way interaction for total bird
abundance.

Bird community composition: effects of stratum and adjacent habitat

The effects of habitat stratum and adjacent habitat were tested using analysis of
similarities (ANOSIM, using the Bray–Curtis similarity index between sites
and 1000 random permutations). As sufficient replicate sites were unavailable
for a simultaneous two-way analysis of both factors, each factor was analysed
separately. Bird community composition differed significantly between the two

[529]
1590

Figure 3. Effects of habitat stratum (rainforest fragment, primary rainforest control, and plan-
tations) and adjacent habitat (with low vs. high canopy connectivity) on bird community variables
in the Anamalai hills. Figures illustrate estimated means per point and their standard errors for bird
species richness (panels on the left) and bird abundance (panels on the right).

[530]
 1591

habitat strata, rainforests vs. plantations (global R = 0.365, p = 0.023).

Differences between sites differing in adjacent habitats’ canopy connectivity
were not significant (R = 0.111, p = 0.162) when only two categories of
connectivity was considered (low vs. high, where the latter includes the primary
rainforest control site). The analysis was repeated with four categories: rain-
forests adjoining habitats with low and high canopy connectivity (RL, RH),
and plantations adjoining habitats with low and high canopy connectivity (PL,
PH). Bird community composition varied significantly across these four cate-
gories (global R = 0.314, p = 0.042). Pair-wise comparison of categories
revealed significant differences only between RH and PL (R = 0.87,
p = 0.029, 1 of 35 possible permutations) and near significance between RL
and RH (R = 0.296, p = 0.086, 3 of 35 possible permutations) and between
RL and PL (R = 0.519, p = 0.10, 1 of 10 possible permutations). Differences
in bird community composition between rainforest sites and plantation sites
that adjoined rainforest fragments with high canopy connectivity (PH) were
not statistically significant (R < 0.26, p > 0.17).

Bird community structure: correlations with vegetation structure and floristics

The species richness and abundance of bird species categories (all, rainforest,
open-forest, endemics, priority species, rainforest migrants, and all migrants)
was compared between rainforest and plantation sites. Except for open-forest
birds and all migrants, values were lower in plantations than rainforests
(Table 6). Species richness and density of rainforest birds was 43–47% lower in
plantations than in rainforest sites. Although the richness of priority species
and rainforest migrants was not significantly different between the strata, the
abundance of these categories was significantly lower in plantations (Table 6).
Multiple regression analysis indicated a significant positive influence of PC1 on
rainforest and rarefaction bird species richness as well as on the abundance of
all birds, rainforest birds, endemic birds, and priority species (Table 6). The
results indicated that woody plant variables, cane, liana, and bamboo, repre-
sented in PC1 had a generally positive effect on rainforest birds, endemic and
priority species and a negative effect on open-forest bird species richness.
Canopy structural variables represented on PC2 had significant positive effects
on species richness of all birds taken together and for the subset of rainforest
birds as well as on the abundance of priority species and rainforest migrants.
PC2 had a negative effect only on open-forest bird abundance (Table 6). The
level of canopy connectivity in adjacent habitat was significantly positively
related only to richness of priority species. Site area did not appear to have a
significant influence on any of these variables barring a weak positive effect on
migrant species richness (Table 6).
Mantel tests were used to examine the effects of dissimilarity in vegetation
structure and floristics (tree species composition) on bird community dissimi-
larity between sites. Bird community dissimilarity was strongly positively

[531]
1592

Table 6. Cumulative species richness and abundance (individuals/ha) of various categories of

birds in rainforest and shade-coffee and cardamom plantation sites in the Anamalai hills and
significant coefficients from multiple regression with area, adjoining habitat connectivity level, and
habitat components (PC1 and PC2) as independent variables.

Rainforests Plantations Mann– Standardized regression coefficient (b)

Whitney

Mean SE Mean SE U Area Adjoining PC1 PC2

habitat

Bird species richness

Alla 28.6 0.97 21.2 2.17 3.5** – – – 0.554*
Rainforesta 24.7 0.99 14.0 2.11 2.5** – – 0.676* 0.489*
Open-foresta 3.9 1.37 7.2 1.40 11.0 – –
0.728** –
Rarefaction estimate 38.6 1.69 28.1 3.43 6.0* – – 0.503+ 0.433+
Endemic 5.0 0.31 3.2 0.65 6.0* – – 0.586* –
Priority 11.9 0.91 8.8 1.72 10.5 – 0.621* – –
Rainforest migrants 5.4 0.53 5.0 0.26 18.0 – – – –
All migrants 7.4 0.69 8.3 0.49 15.5 0.545+ – – –
Bird abundance
All 22.7 0.94 15.8 1.48 3.0** – – 0.683** –
Rainforest 20.3 1.37 11.1 2.10 4.0* – – 0.761** –
Open-forest 2.5 0.92 4.7 0.89 13.0 – – –
0.659*
Endemic 2.4 0.43 0.9 0.43 5.0* – – 0.666* –
Priority 5.8 0.58 3.3 0.78 6.0* – – 0.644** 0.507*
Rainforest migrants 2.4 0.16 1.6 0.20 5.0* – – – 0.578*
All migrants 3.1 0.29 2.8 0.47 14.0 – – – –
+
p £ 0.10, *p £ 0.05, **p £ 0.01
a
Includes only species detected >2 times in the pooled point count samples in each site. Mean and
standard error (SE) were calculated from 7 rainforest and 6 plantations sites.

correlated to floristic dissimilarity (Mantel test, Kr = 346, p < 0.001) and

weakly to structural dissimilarity (Kr = 101, p = 0.092). When the effects of
the positive correlation between structural and floristic dissimilarities were
controlled using partial Mantel tests, only floristic dissimilarity was positively
correlated to bird community dissimilarity (partial T = 0.409, p < 0.001) and
structural dissimilarity had a non-significant effect (partial T =
0.057,
p = 0.73). Identical results were obtained when only the six plantation sites
and the primary rainforest control site were included in the analysis (N = 7
sites, 21 pair-wise similarities).

Bird species distributions in rainforests and plantations

Analysis of species habitat use with the deviation index showed that most
open-forest species (residents and migrants) occurred more often than expected
in shade-coffee plantations, and less often than expected in rainforest control,

[532]
 1593

fragments, and cardamom plantations (Appendix). A majority of rainforest

residents showed no significant preference or avoidance of rainforest control
and fragment sites, whereas their abundance tended to be negatively influenced
by plantations. There were a number of exceptions, however, and many rain-
forest birds persisted in shade-coffee plantations, being absent or found less
often than expected in rainforest control and cardamom sites (possibly because
of the lower sampling intensity in these two sites). Three rainforest priority
species showed significant preference for rainforest control sites (Yellow-bro-
wed Bulbul, Large Woodshrike, and Malabar Trogon, see Appendix for sci-
entific names). The latter species along with Little Spiderhunter and Mountain
Imperial Pigeon also occurred more often than expected in the cardamom
plantation (Appendix). Among rainforest migrants, positive preference was
shown by Grey and Forest Wagtails to shade-coffee, by Rusty-tailed
Flycatcher to cardamom, and by Large-billed Leaf Warbler to the rainforest
control site. Of the ten species of endemics recorded, the Crimson-backed
Sunbird and White-bellied Blue Flycatcher appeared to prefer the rainforest
control and cardamom plantation sites. In addition, the Nilgiri Flycatcher and
Black-and-Orange Flycatcher chiefly used the rainforest control site. Shade-
coffee plantations were mostly used less by endemics, except for the Malabar
Grey Hornbill and Rufous Babbler. The three endemics not recorded from the
control site during sampling were typically lower-elevation birds (Malabar
Grey Hornbill, Malabar Parakeet) and forest edge species (Rufous Babbler).
These three species have been recorded from the control site during
supplementary observations.

Discussion

Bird community change: effects of stratum and adjacent habitats

An increasing number of studies show the importance of habitat in the sur-

rounding landscape matrix on bird communities of tropical forest fragments
(Stouffer and Bierregaard 1995a, b, Daily et al. 2001; Renjifo 2001; Luck and
Daily 2003). These studies have shown that structurally complex matrices
have greater potential for supporting populations of forest birds than open
areas such as pastures. In the present study, this general pattern is supported
in the comparison of forest fragments that adjoin sites with better canopy
cover and connectivity (shade-coffee plantations) versus more isolated sites
(mostly surrounded by relatively treeless tea plantations). The major differ-
ence in bird community composition was between rainforest fragments that
adjoined sites with high canopy connectivity and plantations that adjoined
sites with low canopy connectivity. Obviously, conversion to plantations
followed by isolation from forests has a greater effect on bird community
change than either factor alone. While the effects of adjacent habitat were

[533]
1594

found to be significant in addition to habitat stratum for a suite of bird

community variables in the MANOVA, this effect did not emerge as very
significant in multiple regression analyses that included vegetation structure
components (PC1 and PC2) as independent variables. This suggests that the
observed effects of adjacent habitat are mediated mainly by changes in veg-
etation structure, for instance, through the presence or absence of hard edges
or the distribution of rainforest versus open-forest species in the adjacent
habitat. Although the benefit to rainforest birds by increased canopy con-
nectivity in adjoining sites is easily understood, the factors inhibiting open-
forest birds in fragments adjoining sites with better canopy connectivity is not
directly apparent. Possible explanations include better-developed habitat
structure in such fragments due to absence of hard edges, competitive effects
of persisting rainforest birds, or the paucity of adjoining open habitats that
can act as source pools for open-forest birds. Among the suite of variables
that may influence bird community structure in these sites, habitat structure
and floristics (tree species composition) appear to have a particularly strong
influence.

Bird community change: effects of habitat structure and tree species composition

The lower abundance and richness of rainforest birds in plantations and

fragments is partly attributable to alteration of habitat structure and the tree
canopy. Shade-coffee plantations were structurally and floristically poorer than
the other habitats with most foliage concentrated in the relatively uniform
coffee-shrub and canopy shade-tree layers. The cardamom plantation under
diverse natural shade was floristically similar to primary rainforest in canopy
trees but had a higher basal area and canopy height possibly due to the greater
stature developed by shade trees after release from competition by removal of
understorey and mid-storey vegetation (Parthasarathy 2001; Raman and
Sukumar 2002). This is not, however, a general feature of cardamom planta-
tions in the study area because other cardamom plantations had a monoculture
canopy of Eucalyptus sp.
The potential of shade-coffee and other plantations to support a high
diversity or abundance of birds relative to forest has been attributed to factors
such as forest stature and vertical stratification (Daniels et al. 1992; Thiollay
1995; Greenberg et al. 1997a; Shahabuddin 1997) and food availability
(Johnson and Sherry 2001). In the Eastern Ghats of India, Beehler et al. (1987)
suggested structural and floristic complexity and availability of remnant forest
patches as key ingredients for supporting and sustaining forest bird popula-
tions. These studies did not directly assess the influence of habitat structure and
floristics in consonance with landscape variables across replicate sites. In the
present study, woody vegetation variables associated with PC1 (density, basal
area, leaf litter depth, cane, bamboo, liana prevalence) had the strongest
influence on the species richness and density of rainforest birds, particularly

[534]
 1595

endemic species. Bird community composition, however, was significantly

influenced mainly by floristics (tree species composition). This pattern of
woody plant variables influencing rainforest bird species richness and abun-
dance and of tree species composition influencing bird community composition
has also emerged strongly in studies across man-altered habitats in other
Indian rainforests (Raman et al. 1998; Raman and Sukumar 2002). Trees,
shrubs, and lianas contribute to the bulk of the physical vegetation substrate
within rainforest and rainforest birds, being adapted to these substrates for
foraging and nesting sites, respond positively to increasing woody vegetation
density. For vertical stratification, strong effects have been noticed in other
studies where more drastically altered habitats such as open habitats (Terborgh
1985; Bowman et al. 1990), shade-less tea plantations (Raman 2001), or other
kinds of plantations (Daniels et al. 1992) are included in the reckoning. In the
present study, canopy cover and vertical stratification (on PC2) negatively
influenced open-forest bird abundance, and positively influenced species rich-
ness (all birds and rainforest birds) and abundance of priority species and
rainforest migrants. In some cases, these effects supplemented the effects of
PC1 thereby reinforcing the view that bird community structure is strongly
influenced by habitat structure in these rainforests and plantations. Area of the
sites did not appear to have a significant influence on most community mea-
sures, possible because no small sites were sampled during the study (all sites
>18 ha). In an earlier study of the effects of fragmentation on birds, rainforest
bird species richness was found to be positively correlated to fragment area, but
substantial decrease in richness occurred only below a fragment size of 10 ha
(Raman 2001). In addition to factors indicated earlier and below, others such
as tree species composition, availability of food resources, and variation in
environmental conditions may also play important roles in bird community
structuring.

Biological infiltration: conservation value of plantations for birds

A simple assessment of conservation value for birds based on species richness

or abundance in plantations is insufficient. From a conservation perspective, it
is important to analyse persistence and usage of these habitats by rainforest
birds (‘forest-affiliated avifauna’, Daily et al. 2001) vis-a-vis open-forest spe-
cies. That only 59–67% of the avifauna in shade-coffee plantations were
rainforest species is significant in this respect. Many typical rainforest bird
species such as the Malabar Trogon, Brown-cheeked Fulvetta, White-bellied
Blue Flycatcher, and Common Flameback were absent in coffee plantations,
whereas others such as Great Hornbill and Dark-fronted Babbler were scarce
or noted only in sites adjoining rainforest fragments. Among migrants, some
species frequently noted in rainforests were absent (Western Crowned Warbler,
Large-billed Leaf Warbler, Rusty-tailed Flycatcher) or scarce (Indian Blue
Robin) in shade-coffee. On the other hand, habitat openings and estate roads

[535]
1596

in these plantations benefited two species of migrant wagtails (Grey and Forest
Wagtails).
The process of incursion of a large number of open-forest species into
rainforest fragments (‘biological infiltration’, Raman 2001) is explained by
alteration of habitat structure as well as the influence of the surrounding
landscape. Fragments, particularly the more-disturbed ones with their patchier
canopy and denser shrubbery, allow the persistence of species that thrive in
open areas and weedy vegetation (e.g., Common Tailorbird, Red-whiskered
Bulbul) due to changes in microhabitat and microclimate. Such open-forest
species derive from the deciduous and thorn forests of the region and such
infiltration into disturbed rainforests is known from many other rainforest
regions (Leck 1979; Daniels et al. 1990; Raman 2001). Dense thickets of
invasive weeds such as Lantana camara within these fragments, with more light
and warmth due to canopy openness, allows the persistence of such species that
do not occur in the cool, dark, evergreen vegetation of undisturbed rainforest
understorey.
Another aspect of conservation relevance to consider is whether these plan-
tations provide resources for birds throughout the year. The data from this
study over the main breeding season only indicate usage as it was not possible to
establish breeding. Plantations that consist mainly of exotic trees may offer few
resources for frugivorous and nectarivorous birds (Greenberg et al. 1997a).
Such plantations may only temporarily support frugivores such as the hornbills
and Pompadour Pigeon that visit the scattered fruiting trees (e.g., Ficus sp.,
Litsea glabrata, and Actinodaphne angustifolia). Nectar-seeking birds such as
sunbirds, drongos, orioles, and Vernal Hanging-Parrot visited flowers of exotics
such as Eucalyptus sp., Grevillea robusta, Erythrina sp., and even coffee bushes
(Crimson-backed Sunbird) in season, besides flowers of native plants. In the
more open coffee plantations, however, an open-forest species (Purple Sunbird)
was often more abundant than the endemic Crimson-backed Sunbird.

Endemic and priority species

The responses of endemic and priority species to habitat alteration were

mostly idiosyncratic. In general, alteration in habitat structure, particularly
woody plant variables, affected their richness and abundance, suggesting the
importance of rainforest structural niches and floristic attributes for the
persistence of these species, as noted in the Agasthyamalai region of the
Western Ghats (Raman and Sukumar 2002). As a consequence, many
endemic and priority species avoided shade-coffee plantations, such as
Malabar Trogon, Mountain Imperial Pigeon, Yellow-browed Bulbul, and
Common Flameback. The occurrence of Malabar Grey Hornbills in shade-
coffee estates can be attributed to the retention of fruit-trees in the canopy
that provide food for hornbills (Raman and Mudappa 2003b). Forest-edge
species, including the endemic Rufous Babbler, also persist in shade-coffee as

[536]
 1597

they contain suitable habitat. Continuous rainforest emerges as an important

habitat for priority species such as Malabar Trogon and Large Woodshrike
and the endemic flycatchers. However, a number of priority species and
endemics showed strong avoidance (deviation index) of the rainforest control
site (Appendix). This was mainly because these species were lower-elevation
birds that occurred in good numbers in relatively undisturbed lower elevation
rainforest but were scarce or absent in the control site available for this
study.

Conservation implications

This study shows that the nature of adjoining habitats affects rainforest bird
communities in tropical rainforest fragments. Specifically, having shade-coffee
rather than tea plantations adjoin fragments has beneficial effects on rainforest
birds, and these effects are probably mediated through the influence of habitat
structure and canopy tree species composition on bird community structure.
Such plantations can thus promote the persistence in the entire landscape of
larger populations of rainforest birds (Beehler et al. 1987; Renjifo 2001).
Individual rainforest birds resident or dispersing from such sites can reduce
the likelihood of chance extinction in fragments through recolonization
(‘rescue effect’, Brown and Kodric-Brown 1977). Increased canopy connec-
tivity in adjoining habitats also has value for shade-coffee plantations. Again,
benefits accrue mainly to rainforest birds: more species and individuals were
supported per unit area in plantations that adjoined continuous forest.
Although there were no completely isolated (surrounded by tea estates) shade-
coffee plantations in the study area, it seems likely that such plantations will
be more depauperate in rainforest birds than those adjoining fragments or
continuous forest. If increased canopy connectivity in adjoining forest habitats
benefits rainforest birds it may also benefit coffee plantation owners. The bird
populations may enable ‘eco-friendly’ or ‘bird-friendly’ coffee certification
(Sherry 2000; Rappole et al. 2003) and reduce insect attack on coffee leaves
(Guatemala: Greenberg et al. 2000), while bees from adjoining forest may
enhance pollination of coffee plants (Costa Rica: Ricketts 2004, Ricketts et al.
2004).
In the study region, ongoing conversion of shade-coffee to tea plantations
driven largely by market forces, is therefore of conservation concern because
tea plantations represent a poorer habitat for rainforest birds and because of
the fallouts for fragments in the landscape. Efforts should be made to halt such
changes while encouraging landowners through tax and other incentives to
promote the relatively more benign form of land use represented by shade-
coffee. Schemes, currently non-existent in India, to certify shade-coffee plan-
tations that are good for birds (Smithsonian Migratory Bird Center 1999) need
to be explored as a means to promote conservation while directly extending
benefits to landowners (Venkatachalam 2005). Such certification should

[537]
1598

incorporate recognition of steps taken by landowners to promote the diversity

of native tree species used as shade trees in coffee plantations. As many bird
species are clearly dependent on rainforests in the landscape, it needs to be
emphasised that efforts to prevent land-use conversion and promote use of a
diversity of native shade tree species can only supplement and not replace
conventional conservation efforts that aim to protect and restore primary
rainforests and fragments.

Acknowledgements

This research was financially supported by the Wildlife Conservation Society,

USA. I am also grateful to Netherlands Committee for IUCN, Tropical
Rainforest Programme, and Barakat Inc., USA, for follow-up support. I
thank the Tamil Nadu Forest Department, especially Dr. Sukhdev and
Mr. V. Ganesan, for permits and support. A number of plantation company
managers allowed me to work on their property for which I am grateful. I
thank Divya Mudappa for help in designing the study, being a sounding-
board and critic of the ideas developed, and for helpful comments on the
manuscript. My colleagues at NCF, especially M.D. Madhusudan, infused
additional enthusiasm into this work through their support and encourage-
ment. I thank R. Raghunath and Pavithra Sankaran for assistance with
fieldwork, Vena Kapoor for help with data entry, and A. Silamban for being
a reliable field assistant.

[538]
 Appendix. List of birds detected in point count sampling in rainforest and plantation sites within and adjoining the Indira Gandhi Wildlife Sanctuary in the
Anamalai hills, Western Ghats. The total number of detections of each species, average detections per 25 point counts and deviations from expected detections
(significant values in bold) are presented. Category codes: Res = resident, Mig = Migrant, Pri = priority species, End = endemic to Western Ghats;
Habitat codes: OF = open-forest, RF = rainforest.

S. No. Habitat Species Category Detections Detections/25 point counts Deviations

Control Fragment Coffee Cardamom Control Fragment Coffee Cardamom

1 OF Ashy Woodswallow Artamus fuscus Res 1 – 0.14 – –

2 OF Asian Koel Eudynamus scolopocea Res 8 – 0.14 1.00 –
1.00
0.61 0.48
1.00
3 OF Black-headed Cuckooshrike Coracina Res 2 – 0.14 0.14 –
melanoptera
4 OF Black-rumped Flameback Dinopium Res 6 – 0.28 0.57 –
1.00
0.22 0.37
1.00
benghalense
5 OF Blue-winged Leafbird Chloropsis Res 1 – – 0.14 –
cochinchinensis
[539]

6 OF Brown-capped Pygmy Woodpecker Res 6 – 0.57 0.29 –
1.00 0.12 0.04
1.00
Dendrocopos
nanus
7 OF Chestnut-headed Bee-eater Merops Res 10 0.83 0.14 1.00 1.00 0.05
0.68 0.39 0.11
leschenaulti
8 OF Chestnut-tailed Starling Sturnus Res 2 – 0.28 – –
malabaricus
9 OF Common Hoopoe Upupa epops Res 2 – 0.14 0.14 –
10 OF Common Iora Aegithina tiphia Res 10 – 0.85 0.29 2.00
1.00 0.07
0.21 0.43
11 OF Common Tailorbird Orthotomus Res 62 – 2.13 6.71 –
1.00
0.37 0.42
1.00
sutorius
12 OF Greater Coucal Centropus sinensis Res 23 – 0.71 2.57 –
1.00
0.41 0.43
1.00
13 OF Grey-bellied Cuckoo Cacomantis Res 1 – 0.14 – –
passerinus
14 OF Grey-breasted Prinia Prinia Res 18 – 0.85 1.71 –
1.00
0.22 0.37
1.00
hodgsonii

1599
15 OF Jungle Myna Acridotheres fuscus Res 2 – – 0.29 –
 Appendix. (Continued).

1600
S. No. Habitat Species Category Detections Detections/25 point counts Deviations

Control Fragment Coffee Cardamom Control Fragment Coffee Cardamom

16 OF Kestrel Falco tinnunculus Res 1 – – 0.14 –

17 OF Large-billed Crow Corvus macrorhynchus Res 44 – 3.13 3.00 1.00
1.00
0.02 0.21
0.56
18 OF Oriental Magpie Robin Copsychus saularis Res 27 – 1.14 2.71 –
1.00
0.27 0.39
1.00
19 OF Pied Cuckoo Clamator jacobinus Res 1 – – 0.14 –
20 OF Plum-headed Parakeet Psittacula Res 34 – 2.70 2.14 –
1.00 0.04 0.18
1.00
cyanocephala
21 OF Purple Sunbird Nectarinia asiatica Res 27 – 0.57 3.29 –
1.00
0.56 0.47
1.00
22 OF Red-whiskered Bulbul Pycnonotus Res 151 – 7.81 13.29 3.00
1.00
0.18 0.33
0.60
jocusus
23 OF Shikra Accipiter badius Res 5 – 0.57 0.14 –
1.00 0.21
0.21
1.00
24 OF Small Minivet Pericrocotus cinnamomeus Res 4 – 0.28 0.29 –
[540]

25 OF Spotted Dove Streptopelia chinensis Res 2 – 0.14 0.14 –

26 OF Streak-throated Woodpecker Picus Res 1 – 0.14 – –
xanthopygaeus
27 OF White-breasted Waterhen Amaurornis Res 2 – – 0.29 –
phoenicurus
28 OF White-throated Kingfisher Halcyon Res 4 – 0.14 0.43 –
smyrnensis
29 OF Asian Brown Flycatcher Muscicapa Mig 11 – 0.85 0.71 –
1.00 0.02 0.19
1.00
dauurica
30 OF Black-naped Oriole Oriolus chinensis Mig 1 – – 0.14 –
31 OF Blue-capped Rock Thrush Monticola Mig 1 – – 0.14 –
cinclorhynchus
32 OF Blyth’s Reed Warbler Acrocephalus Mig 206 3.33 12.64 14.86 9.00
0.65
0.09 0.24
0.29
dumetorum
33 OF Brown Shrike Lanius cristatus Mig 14 – 0.99 1.00 –
1.00
0.02 0.24
1.00
34 OF Common Rosefinch Carpodacus Mig 9 – 0.57 0.71 –
1.00
0.08 0.29
1.00
erythrinus
 35 OF Eurasian Golden Oriole Oriolus oriolus Mig 10 – 0.28 1.14 –
1.00
0.44 0.44
1.00
36 OF Red-throated Flycatcher Muscicapa parva Mig 1 – – 0.14 –
37 RF Asian Paradise-Flycatcher Terpsiphone Res 15 0.83 0.99 1.00 –
0.16
0.05 0.20
1.00
paradisi
38 RF Bar-winged Flycatcher-Shrike Hemipus Res 23 5.83 1.85 0.29 1.00 0.54 0.04
0.56
0.29
picatus
39 RF Besra Accipiter virgatus Res 4 0.83 0.43 – –
40 RF Black-lored Tit Parus xanthogenys Res 23 – 2.27 0.71 2.00
1.00 0.14
0.17 0.05
41 RF Black-naped Monarch Hypothymis azurea Res 31 5.00 2.41 – 8.00 0.36 0.03
1.00 0.53
42 RF Blue-bearded Bee-eater Nyctyornis athertoni Res 1 – 0.14 – –
43 RF Bronzed Drongo Dicrurus aeneus Res 18 – 1.99 0.57 –
1.00 0.20
0.16
1.00
44 RF Brown-cheeked Fulvetta Alcippe Res 48 10.00 4.69 – 3.00 0.47 0.14
1.00
0.12
poioicephala
45 RF Crested Serpent Eagle Spilornis cheela Res 3 – 0.14 0.29 –
46 RF Drongo Cuckoo Surniculus lugubris Res 1 – 0.14 – –
47 RF Emerald Dove Chalcophaps indica Res 6 – 0.85 – –
1.00 0.32
1.00
1.00
0.01
1.00
[541]

48 RF Eurasian Blackbird Turdus merula Res 16 0.83 1.42 0.71 – -0.19 0.09
49 RF Golden-fronted Leafbird Chloropsis Res 7 – 0.28 0.71 –
1.00
0.29 0.40
1.00
aurifrons
50 RF Greater Flameback Chrysocolaptes lucidus Res 25 1.67 2.13 0.86 2.00
0.07 0.07
0.13 0.00
51 RF Greater Racket-tailed Drongo Dicrurus Res 29 0.83 2.70 0.14 8.00
0.45 0.11
0.80 0.55
paradiseus
52 RF Grey Junglefowl Gallus sonneratii Res 7 – 0.57 0.29 1.00
1.00 0.05
0.04 0.29
53 RF Grey-headed Canary Flycatcher Culicicapa Res 55 13.33 3.41 – 15.00 0.52
0.09
1.00 0.55
ceylonensis
54 RF Hill Myna Gracula religiosa Res 62 5.83 3.98 1.86 14.00 0.11
0.07
0.19 0.48
55 RF Indian Scimitar Babbler Pomatorhinus Res 73 5.00 5.97 3.29 2.00
0.05 0.05 0.01
0.49
horsfieldii
56 RF Orange-headed Thrush Zoothera citrina Res 47 1.67 4.26 1.86 2.00
0.36 0.10
0.06
0.30
57 RF Oriental Honey-Buzzard Pemis Res 1 – 0.14 – –
ptilorhynchus
58 RF Oriental White-Eye Zosterops palpebrosus Res 119 14.17 7.67 6.57 2.00 0.22
0.07 0.11
0.65

1601
 Appendix. (Continued).

1602
S. No. Habitat Species Category Detections Detections/25 point counts Deviations

Control Fragment Coffee Cardamom Control Fragment Coffee Cardamom

59 RF Pompadour Green Pigeon Treron Res 17 0.83 0.85 0.57 6.00
0.22
0.19
0.14 0.63
pompadora
60 RF Puff-throated Babbler Pellorneum ruficeps Res 35 3.33 3.27 0.71 3.00 0.11 0.12
0.37 0.04
61 RF Red Spurfowl Galloperdix spadicea Res 9 0.83 0.99 0.14 – 0.10 0.20
0.47
1.00
62 RF Scarlet Minivet Pericrocotus flammeus Res 70 2.50 5.82 2.71 7.00
0.36 0.06
0.06 0.11
63 RF Velvet-fronted Nuthatch Sitta frontalis Res 66 5.00 5.54 2.29 5.00 0.00 0.06
0.12
0.02
64 RF Asian Fairy Bluebird Irena puella Pri 38 4.17 2.56 0.29 13.00 0.18
0.05
0.71 0.62
65 RF Black Bulbul Hypsipetes leucocephalus Pri 33 9.17 2.98 – 1.00 0.57 0.10
1.00
0.45
66 RF Black-crested Bulbul Pycnonotus Pri 12 – 1.42 0.14 1.00
1.00 0.23
0.58 0.02
melanicterus
67 RF Common Flameback Dinopium javanense Pri 6 0.83 0.57 – 1.00 0.29 0.12
1.00 0.35
[542]

68 RF Crimson-fronted Barbet Megalaima Pri 20 – 1.42 1.43 –
1.00
0.02 0.24
1.00
rubricapilla
69 RF Dark-fronted Babbler Rhopocichla Pri 21 2.50 2.27 0.14 1.00 0.22 0.19
0.73
0.25
atriceps
70 RF Dollarbird Eurystomus orientalis Pri 1 – – 0.14 –
71 RF Great Hornbill Buceros bicornis Pri 2 0.83 – 0.14 –
72 RF Heart-spotted Woodpecker Hemicircus Pri 4 0.83 0.14 – 2.00
canente
73 RF Large Woodshrike Tephrodornis gularis Pri 17 3.33 0.85 0.86 1.00 0.44
0.19 0.07
0.15
74 RF Little Spiderhunter Arachnothera Pri 69 3.33 7.10 0.14 14.00
0.22 0.16
0.91 0.44
longirostra
75 RF Malabar Trogon Harpactes fasciatus Pri 5 0.83 0.43 – 1.00 0.37 0.07
1.00 0.43
76 RF Malabar Whistling Thrush Myophonus Pri 83 8.33 7.10 2.00 9.00 0.14 0.07
0.29 0.15
horsfieldii
77 RF Mountain Imperial Pigeon Ducula Pri 22 2.50 1.42 0.14 8.00 0.20
0.07
0.74 0.64
badia
78 RF Plain Flowerpecker Dicaeum concolor Pri 290 16.67 20.74 15.71 14.00
0.14
0.02 0.10
0.24
 79 RF Rufous Woodpecker Celeus brachyurus Pri 2 – 0.14 – 1.00
80 RF Vernal Hanging Parrot Loriculus Pri 70 6.67 4.83 3.00 7.00 0.11
0.03
0.01 0.11
vernalis
81 RF White-bellied Woodpecker Dryocopus Pri 1 – 0.14 – –
javensis
82 RF White-cheeked Barbet Megalaima Pri 125 11.67 8.95 5.86 7.00 0.10
0.02 0.03
0.17
viridis
83 RF Yellow-browed Bulbul Iole indica Pri 102 15.83 9.38 1.00 10.00 0.34 0.11
0.64 0.10
84 RF Ashy Drongo Dicrurus leucophaeus Mig 35 – 2.98 2.00 –
1.00 0.07 0.13
1.00
85 RF Brown-breasted Flycatcher Muscicapa Mig 3 – 0.14 0.14 1.00
muttui
86 RF Forest Wagtail Dendronanthus indicus Mig 24 – 0.57 2.71 1.00
1.00
0.51 0.44
0.31
87 RF Greenish Warbler Phylloscopus Mig 306 23.33 21.73 15.71 15.00 0.00
0.02 0.08
0.24
trochiloides
88 RF Grey Wagtail Motacilla cinerea Mig 15 – 0.28 1.86 –
1.00
0.59 0.47
1.00
89 RF Indian Blue Robin Luscinia brunnea Mig 48 0.83 5.40 0.71 4.00
0.63 0.21
0.50 0.02
[543]

90 RF Indian Pitta Pitta brachyura Mig 1 – – 0.14 –

91 RF Large Hawk Cuckoo Hierococcyx Mig 2 – 0.14 0.14 –
sparverioides
92 RF Large-billed Leaf Warbler Phylloscopus Mig 100 17.50 9.52 – 12.00 0.39 0.13
1.00 0.20
magnirostris
93 RF Pied Thrush Zoothera wardii Mig 2 – 0.28 – –
94 RF Rusty-tailed Flycatcher Muscicapa Mig 19 – 2.27 – 3.00
1.00 0.24
1.00 0.33
ruficauda
95 RF Verditer Flycatcher Eumyias Mig 1 – 0.14 – –
thalassina
96 RF Western Crowned Warbler Mig 4 0.83 0.43 – –
Phylloscopus occipitalis
97 RF Black-and-Orange Flycatcher End 10 5.83 0.28 – 1.00 0.77
0.44
1.00 0.11
Ficedula nigrorufa

1603
 Appendix. (Continued).

1604
S. No. Habitat Species Category Detections Detections/25 point counts Deviations

Control Fragment Coffee Cardamom Control Fragment Coffee Cardamom

98 RF Crimson-backed Sunbird Nectarinia End 156 22.50 12.36 2.29 26.00 0.31 0.03
0.50 0.35
minima
99 RF Grey-headed Bulbul Pycnonotus End 1 – 0.14 – –
priocephalus
100 RF Malabar Grey Hornbill Ocyceros End 33 – 2.41 2.14 1.00
1.00
0.01 0.19
0.45
griseus
101 RF Malabar Parakeet Psittacula End 30 – 3.27 0.71 2.00
1.00 0.19
0.30
0.09
columboides
102 RF Nilgiri Flycatcher Eumyias albicaudata End 29 6.67 2.84 0.14 – 0.50 0.14
0.80
1.00
103 RF Rufous Babbler Turdoides subrufus End 10 – 0.85 0.57 –
1.00 0.07 0.13
1.00
104 RF White-bellied Blue Flycatcher Cyornis End 21 3.33 1.42 – 7.00 0.35
0.04
1.00 0.62
[544]

pallipes
105 RF White-bellied Treepie Dendrocitta End 3 – 0.43 – –
leucogastra
106 RF Wynaad Laughingthrush Garrulax End 2 – 0.14 – 1.00
delesserti
 1605

References

Ali S. and Ripley S.D. 1983. Handbook of the Birds of India and Pakistan. Compact edition,
Oxford University Press, Delhi.
Beehler B.M., Raju K.S.R.K. and Ali S. 1987. Avian use of man-disturbed forest habitats in the
Eastern Ghats, India. Ibis 129: 197–211.
Bibby C.J., Burgess N.D. and Hill D.A. 1992. Bird Census Techniques. Academic Press, London.
BirdLife International 2001. Threatened Birds of the World. Lynx Edicions and BirdLife Inter-
national, Barcelona and Cambridge.
Bowman D.M.J.S., Woinarski J.C.Z., Sands D.P.A., Wells A. and McShane V.J. 1990. Slash-and-
burn agriculture in the wet coastal lowlands of Papua New Guinea: response of birds, butterflies
and reptiles. J. Biogeogr. 17: 227–239.
Brown J.H. and Kodric-Brown A. 1977. Turnover rates in insular biogeography: effect of immi-
gration on extinction. Ecology 58: 445–449.
Brown S. and Lugo A.E. 1990. Tropical secondary forests. J. Trop. Ecol. 6: 1–32.
Cincotta R.P., Wisnewski J. and Engelman R. 2000. Human population in the biodiversity hot-
spots. Nature 404: 990–992.
Clarke K.R. and Gorley R.N. 2001. Primer v5: User Manual/Tutorial. PRIMER-E, Plymouth.
Clarke K.R. and Warwick R.M. 1994. Change in Marine Communities: An Approach to Statistical
Analysis and Interpretation. Plymouth Marine Laboratory, Plymouth.
Coffee Board 2001. Database on Coffee. Coffee Board, Bangalore.
Collar N.J., Crosby M.J. and Stattersfield A.J. 1994. Birds to Watch 2: The World list of Threa-
tened Birds. BirdLife International, Cambridge.
Colwell R.K. 1997. EstimateS: Statistical Estimation of Species Richness and Shared Species from
Samples. Version 6.01b. Internet URL: http://viceroy.eeb.uconn.edu/EstimateS.
Colwell R.K. and Coddington J.A. 1994. Estimating terrestrial biodiversity through extrapolation.
Philos. Trans. Roy. Soc. London Series B 345: 101–118.
Congreve H.R.T. 1942. The Anamalais. Associated Printers, Madras.
Daily G.C. 2001. Ecological forecasts. Nature 411: 245.
Daily G.C., Ehrlich P.R. and Sánchez-Azofeifa G.A. 2001. Countryside biogeography: use of
human-dominated habitats by the avifauna of southern Costa Rica. Ecol. Appl. 11: 1–13.
Daniels R.J.R., Hegde M. and Gadgil M. 1990. Birds of man-made habitats: the plantations. Proc.
Ind. Acad. Sci. (Anim. Sci.) 99: 79–89.
Daniels R.J.R., Joshi N.V. and Gadgil M. 1992. On the relationship between bird and woody plant
species diversity in the Uttara Kannada District of south India. Proc. Nat. Acad. Sci. (USA) 89:
5311–5315.
Estrada A., Coates-Estrada R. and Meritt D.Jr. 1997. Anthropogenic landscape changes and avian
diversity at Los Tuxtlas, Mexico. Biodiv. Conserv. 6: 19–43.
Gamble J.S. and Fischer C.E.C. 1935. Flora of the Presidency of Madras. Reprint, 3 volumes.
Bishen Singh Mahendra Pal Singh, Dehra Dun.
Greenberg R., Bichier P., Angon A.C. and Rietsma R. 1997a. Bird populations in shade and sun
coffee plantations in central Guatemala. Conserv. Biol. 11: 448–459.
Greenberg R., Bichier P. and Sterling J. 1997b. Bird populations in rustic and planted shade coffee
plantations of Eastern Chiapas, Mexico. Biotropica 29: 501–514.
Greenberg R., Bichier P., Angon A.C., MacVean C., Perez R. and Cano E. 2000. The impact of
avian insectivory on arthropods and leaf damage in some Guatemalan coffee plantations.
Ecology 81: 1750–1755.
Hemelrijk C.K. 1990. Models of, and tests for, reciprocity, unidirectionality and other social
interaction patterns at a groups level. Anim. Behav. 39: 1013–1029.
Johnson M.D. and Sherry T.W. 2001. Effects of food availability on the distribution of migratory
warblers among habitats in Jamaica. J. Anim. Ecol. 70: 546–560.
Kannan R. 1998. Avifauna of the Anaimalai hills (Western Ghats) of southern India. J. Bombay
Nat. Hist. Soc. 95: 193–214.

[545]
1606

Krebs C.J. 1989. Ecological Methodology. Harper and Row, New York.
Kumar M.A., Singh M., Srivastava S.K., Udhayan A., Kumara H.N. and Sharma A.K. 2002.
Distribution patterns, relative abundance and management of mammals in Indira Gandhi
Wildlife Sanctuary, Tamil Nadu, India. J. Bombay Nat. Hist. Soc. 99: 184–210.
Laurance W.F., Bierregaard R.O.Jr., Gascon C. and others 1997. Tropical forest fragmentation:
synthesis of a diverse and dynamic discipline. In: Laurance W.F. and Bierregaard R.O.Jr. (eds),
Tropical Forest Remnants: Ecology, Management, and Conservation of Fragmented Commu-
nities. University of Chicago Press, Chicago, pp. 502–514.
Leck C.F. 1979. Avian extinctions in an isolated tropical wet forest preserve. Auk 96: 343–352.
Luck G.W. and Daily G.C. 2003. Tropical countryside bird assemblages: richness, composition,
and foraging differ by landscape context. Ecol. Appl. 13: 235–247.
Mani M.S. (ed) 1974. Ecology and Biogeography of India. Dr. W. Junk Publishers, The Hague.
Manly B.F.J. 1994. Multivariate Statistical Methods: A Primer. 2nd edn. Chapman and Hall,
London, UK.
Menon S. and Bawa K.S. 1997. Applications of geographical information systems, remote sensing
and a landscape ecology approach to biodiversity conservation in the Western Ghats. Curr. Sci.
73: 134–145.
Myers N., Mittermeier R.A., Mittermeier C.G., da Fonseca G.A.B. and Kent J. 2000. Biodiversity
hotspots for conservation priorities. Nature 403: 853–858.
Nair S.C. 1991. The Southern Western Ghats: A Biodiversity Conservation Plan. INTACH, New
Delhi.
Norušis M.J. 1990. SPSS/PC+ Statistics 4.0. SPSS Inc., Chicago.
Olson D.M. and Dinerstein E. 1998. The global 200: a representation approach to conserving the
Earth’s most biologically valuable ecoregions. Conserv. Biol. 12: 502–515.
Parthasarathy N. 2001. Changes in forest composition and structure in three sites of tropical
evergreen forest around Sengaltheri, Western Ghats. Curr. Sci. 80: 389–393.
Pascal J.P. 1988. Wet Evergreen Forests of the Western Ghats of India: Ecology, Structure, Flo-
ristic Composition and Succession. Institut Français de Pondichéry, Pondicherry.
Pascal J.P. and Ramesh B.R. 1997. A Field Key to the Trees and Lianas of the Evergreen Forests of
the Western Ghats (India). Institut Français de Pondichéry, Pondicherry.
Perfecto I. and Vandermeer J. 2002. Quality of agroecological matrix in a tropical montane
landscape: ants in coffee plantations in southern Mexico. Conserv. Biol. 16: 174–182.
Perfecto I., Mas A., Dietsch T. and Vandermeer J. 2003. Conservation of biodiversity in coffee
agroecosystems: a tritaxa comparison in southern Mexico. Biodiv. Conserv. 12: 1239–1252.
Ralph C.J., Sauer J.R. and Droege S. (eds) 1995. Monitoring Bird Populations by Point Counts.
USDA, Forest Service, Pacific Southwest Research Station, Albany, California.
Raman T.R.S. 2001.Community Ecology and Conservation of Tropical Rainforest Birds in the
Southern Western Ghats, India. Ph.D. thesis, Indian Institute of Science, Bangalore.
Raman T.R.S. 2003. Assessment of census techniques for inter-specific comparisons of tropical
rainforest bird densities: a field evaluation in the Western Ghats, India. Ibis 145: 9–21.
Raman T.R.S. and Mudappa D. 2003a. Bridging the gap: sharing responsibility for ecological
restoration and wildlife conservation on private lands in the Western Ghats. Social Change 33:
129–141.
Raman T.R.S. and Mudappa D. 2003b. Correlates of hornbill distribution and abundance in
rainforest fragments in the southern Western Ghats, India. Bird Conserv. Int. 13: 199–212.
Raman T.R.S., Rawat G.S. and Johnsingh A.J.T. 1998. Recovery of tropical rainforest avifauna in
relation to vegetation succession following shifting cultivation in Mizoram, north-east India.
J. Appl. Ecol. 35: 214–231.
Raman T.R.S. and Sukumar R. 2002. Responses of tropical rainforest birds to abandoned plan-
tations, edges and logged forest in the Western Ghats, India. Anim. Conserv. 5: 201–216.
Rappole J.H., King D.I. and Rivera J.H.V. 2003. Coffee and conservation. Conserv. Biol. 17:
334–336.

[546]
 1607

Renjifo L.M. 2001. Effect of natural and anthropogenic landscape matrices on the abundance of
subandean bird species. Ecol. Appl. 11: 14–31.
Ricketts T.H. 2004. Tropical forest fragments enhance pollinator activity in nearby coffee crops.
Conserv. Biol. 18: 1–10.
Ricketts T.H., Daily G.C., Ehrlich P.R. and Michener C.D. 2004. Economic value of tropical forest
to coffee production. Proc. Nat. Acad. Sci. (USA) 101: 12579–12582.
Shahabuddin G. 1997. Preliminary observations on the role of coffee plantations as avifaunal
refuges in the Palni hills of the Western Ghats. J. Bombay Nat. Hist. Soc. 94: 10–21.
Sherry T.W. 2000. Shade coffee: a good brew even in small doses. Auk 117: 563–568.
Smithsonian Migratory Bird Center 1999. Certified shade-grown coffees. <http://www.si.smbc/
coffee/cafelist.htm>.
Stattersfield A.J., Crosby M.J., Long A.J. and Wege D.C. 1998. Endemic Bird Areas of the World:
Priorities for Biodiversity Conservation. Birdlife International, Cambridge.
Stouffer P.C. and Bierregaard R.O.Jr. 1995a. Use of Amazonian forest fragments by understory
insectivorous birds. Ecology 76: 2429–2445.
Stouffer P.C. and Bierregaard R.O.Jr. 1995b. Effects of fragmentation on understory humming-
birds in Amazonian Brazil. Conserv. Biol. 9: 1085–1094.
Tea Board 2002. Tea Board home page. <http://www.teaindia.org/curr_statistics.html>.
Terborgh J. 1985. Habitat selection in Amazonian birds. In: Cody M.L. (ed.), Habitat Selection in
Birds. Academic Press, Inc., Orlando, pp. 311–340.
Tejeda-Cruz C. and Sutherland W.J. 2004. Bird responses to shade coffee production. Anim.
Conserv. 7: 169–179.
Thiollay J.M. 1995. The role of traditional agroforests in the conservation of rain forest bird
diversity in Sumatra. Conserv. Biol. 9: 335–353.
Turner I.M. and Corlett R.T. 1996. The conservation value of small isolated fragments of lowland
tropical rain forest. Trends Ecol. Evol. 11: 330–333.
Umapathy G. and Kumar A. 2000. The occurrence of arboreal mammals in the rain forest frag-
ments in the Anamalai hills, south India. Biol. Conserv. 92: 311–319.
Venkatachalam L. 2005. Coffee: need to achieve market balance. The Hindu Survey of Indian
Agriculture 2005: 113–115.
Verner J. 1985. Assessment of counting techniques. Curr. Ornithol. 2: 247–302.
Wolda H. 1981. Similarity indices, sample size and diversity. Oecologia 50: 296–302.
Wunderle J.H.Jr. 1999. Avian distribution in Dominican shade coffee plantations: area and habitat
relationships. J. Field Ornithol. 70: 58–70.
Zar J.H. 1999. Biostatistical Analysis. 4th edn. Prentice-Hall, New Jersey.

[547]