BIOTROPICA 50(3): 442–454 2018 10.1111/btp.

12552

Leafing patterns and leaf exchange strategies of a cerrado woody community

Maria Gabriela Gutierrez de Camargo1, Gustavo Henrique de Carvalho1, Bruna de Costa Alberton1,2,4 , Paula Reys1,3, and
Leonor Patr
ıcia Cerdeira Morellato1
1
^ncias, Departamento de Bota
UNESP—Universidade Estadual Paulista, Instituto de Biocie ^nica, Laborato

rio de Fenologia, Rio Claro, Sa
~o
Paulo, Brasil
2
^ncias, Programa de Po
UNESP—Universidade Estadual Paulista, Instituto de Biocie
s-graduacßa
~ o em Ecologia e Biodiversidade, Rio Claro,
~o Paulo, Brasil
Sa
3

s, Brasil
Universidade de Rio Verde, Campus I, Rio Verde, Goia

ABSTRACT
The deciduousness of tropical trees and communities depend on ecosystems characteristics such as plant species diversity, and strength
of the dry season. Based on seven years of phenological observations, we provide the first long-term description of leafing patterns of a
woody cerrado community, aiming to investigate (1) the leaf exchange strategies considering the interannual variation in the degree of
deciduousness of individuals and species and quantify the community deciduousness; (2) the relationship between interannual patterns
of leaf fall and leaf flush according to the species’ leaf exchange strategies and climate; (3) the onset of cerrado growing season and its
relation to climate seasonality. To detect seasonality and leafing onset we applied circular statistics and to understand the relationships
between environmental predictors and leaf exchange strategies, we used generalized additive models. From 106 species observed, we
classified 69 as deciduous (26 species), semi-deciduous (25) or evergreen (18) and defined the studied cerrado as a semi-deciduous vege-
tation. Leaf phenology was markedly seasonal and similar among years. Leaf fall peaked in the dry season, and leaf flush in the dry-to-
wet transition. Leaf fall patterns related to temperature and leaf flush to day length and rainfall. Semi-deciduous and deciduous species
were more constrained by climate than the evergreen ones. The cerrado growing season started in the dry-to-wet season transition.
Interannual variations in rainfall and temperature affected the individuals’ and, consequently, species’ degree of deciduousness,
highlighting individual and species variability, and suggesting that cerrado leafing patterns are likely susceptible to future climate change
scenarios.

Abstract in Portuguese is available with online material.

Key words: climatic drivers; deciduous; evergreen; generalized additive models; leaf phenology; savanna; seasonality; semi-deciduous.

DECIDUOUSNESS, DEFINED AS EITHER A PARTIAL OR A FULL LOSS OF studies of the vegetative phenology in these ecosystems have a
leaves, is a key functional trait dependent on ecosystems charac- global reach.
teristics such as soils and species diversity, essential for estimating The Brazilian Cerrado, a Neotropical savanna, is the second
dry season length and intensity in tropical ecosystems (Condit largest vegetation type in the Brazilian territory and presents a
et al. 2000). Leaf development and senescence, as well as leaf gradient of physiognomies, ranging from grasslands to tall wood-
exchange strategies of plant species, are recognized to have lands (Coutinho 2006, Overbeck et al. 2015, Durigan & Ratter
important effects at the ecosystem scale because they are linked 2016). The cerrado harbors the world’s highest woody plant
to a range of processes, including carbon uptake, water and diversity among savannas (Sarmiento 1984, Ratter et al. 1996,
energy fluxes, resource availability and nutrient cycling (Reich Munhoz & Felfili 2006) and is generally under a seasonal climate
1995, Lavorel et al. 2007, Polgar & Primack 2011, Richardson with well-defined wet and dry seasons (Ratter et al. 1996). The
et al. 2013). Additionally, the degree of deciduousness of a com- seasonal climate is known to influence the phenology of plant
munity is an indicator of shifts in the intensity and timing of spe- populations and communities (Monasterio & Sarmiento 1976,
cies activities due to interannual shifts in the weather; for Morellato et al. 2013, 2016). To reduce transpiration and avoid
instance, leaf loss can be more intense during severe dry seasons plant collapse related to water status, plants subjected to seasonal
(Reich 1995, Williams et al. 1997, Condit et al. 2000). As savannas climates tend to lose their leaves during the dry season, facilitat-
are the world’s second most important type of vegetation, con- ing the rehydration process for the growing season (Reich &
tributing to carbon flux and storage (Archibald & Scholes 2007), Borchert 1984, Franco et al. 2005). Most cerrado trees produce
new leaves and flowers at the end of the dry season, just before
the start of the rainy season, indicating that individuals have
Received 31 May 2017; revision accepted 9 February 2018. access to water at greater depth to permit their maintenance even
4
Corresponding author; e-mail: [email protected] when rainfall is low (Monasterio & Sarmiento 1976, Williams
442 ª 2018 The Association for Tropical Biology and Conservation
 Leafing Patterns of a Cerrado Community 443

et al. 1997, Jackson et al. 1999, Batalha & Mantovani 2000, from October to March (Fig. 1). Average annual total rainfall was
Batalha & Martins 2004, Lenza & Klink 2006, Tannus et al. 1524 mm, and the mean annual temperature was 20.7°C between
2006, Pirani et al. 2009). 1972 and 2011 (Camargo et al. 2011). The vegetation is a woody
Few studies have addressed leaf exchange patterns in cer- savanna classified as a cerrado sensu stricto (Coutinho 1978), with
rado communities (Miranda 1995, Munhoz & Felfili 2005, Lenza a discontinuous woody layer reaching 6–8 m high and a continu-
& Klink 2006, Pirani et al. 2009, Alberton et al. 2014). Most of ous herbaceous stratum. A woody species survey in the area indi-
them either focused on reproductive phenophases or in describ- cated that Myrtaceae, Fabaceae and Malpighiaceae are the richest
ing general aspects of the community phenology, providing only families and Bauhinia rufa, Xylopia aromatica, Miconia rubiginosa, Vir-
inferences about the relationship between leafing patterns and ola sebifera and Myrcia guianensis the most abundant species (Reys
environmental cues (e.g., Mantovani & Martins 1988, Batalha et al. et al. 2013).
1997, Batalha & Mantovani 2000, Bulh~ao & Figueiredo 2002,
Batalha & Martins 2004, Damascos et al. 2005, Tannus et al. PHENOLOGICAL DATA.—To describe and analyze woody species
2006, Dalmolin et al. 2015). Others, however, have found nega- leafing patterns, we used a phenological dataset of monthly sur-
tive correlations between leaf fall and precipitation (Munhoz & veys conducted from January 2005 to December 2011. During
Felfili 2005, Pirani et al. 2009). The investigation of vegetative this period, we directly observed leaf fall and leaf flush (following
phenology, its causes and the quantification of leaf exchange Morellato & Leit~ao-Filho 1989) in 2112 marked individuals with
strategies at a community level in the cerrado are, thus, still miss- circumference ≥ 3 cm at 30 cm from stem ground base, sampled
ing. In addition, leaf exchange strategies in communities are over 36 transects (25 m 9 2 m) (see Reys et al. 2013, Vogado
mainly defined based on the leafing patterns of few species and et al. 2016). We took circumference and height measurements of
in short-term phenological observations (Franco et al. 2005, all individuals sampled.
Lenza & Klink 2006), which may include bias related to some To estimate leaf fall and leaf flush phenophases, we
punctual plant response to local weather conditions. Therefore, monthly recorded the phenological activity of each individual in
the use of long-term phenology leafing data series is essential to a qualitative binary scale, with ‘0’ indicating absence of the phe-
the definition of the species’ leaf exchange strategies. nophase and ‘1’ the phenophase presence. We also estimated
Here, we describe, for the first time, the seven-year long leaf the intensity of leaf fall and flush based on a semiquantitative
phenology of a cerrado community, aiming to (1) identify the leaf scale with three classes in which ‘0’ indicates the absence of a
exchange strategies considering the interannual variation in the phenophase, ‘1’ the presence of a phenophase with low to inter-
degree of deciduousness of individuals and species and quantify mediate intensity (up to 50% of the crown flushing new leaves
the community deciduousness; (2) investigate the relationship or of the crown with no leaves or senescent leaves) and ‘2’ the
between interannual patterns of leaf fall and leaf flush according presence of a phenophase with high intensity (more than 50%
to the species’ leaf exchange strategies and the climate conditions; of the crown flushing new leaves or with senescent leaves)
(3) uncover the onset of the cerrado growing season and the rela- (Opler et al. 1980, Morellato et al. 2000, Vogado et al. 2016).
tion to climate seasonality. Despite the use of a monthly frequency of observations to
We expect to find a dominance of leaf exchange species (de- record and classify the leaf exchange strategies in this study, it
ciduous and semi-deciduous) and a higher deciduousness degree is unlikely that we have missed the occurrence of the total
over years related to the length and severity of the dry season absence of leaves in an individual (a 100% deciduous event, see
(Reich & Borchert 1984, Reich 1995, Williams et al. 1997, Condit below) or a leaf fall increase from one month to another that,
et al. 2000, Franco et al. 2005). Based on the few short-term phe- of course, we could also infer from the coverage of newly
nology studies describing leaf exchange strategies of woody cer- emerged leaves. We recorded for all years of observations when
rado species (Lenza & Klink 2006, Pirani et al. 2009, Alberton individuals were totally leafless (100% of leaf fall). The defini-
et al. 2014), we expected leaf fall during the dry season and leaf tion of 100 percent fall category also helped us to define the
flush peaking in the transition between dry and wet seasons, occurrence of leaf flush, since sometimes one month of leaf fall
defining the onset of the growing season. We also expected rain- was followed by new young leaves on the subsequent month.
fall and day length as the main environmental factors related to From March 2007 to December 2008, we did not score leafing
leafing seasonality, as previously suggested for tropical seasonal intensity; we only recorded phenological activity using the quali-
vegetations (Condit et al. 2000, Borchert & Rivera 2001). tative scale. The qualitative and semiquantitative individual
observations were independently collected and provide comple-
METHODS mentary information which improve the analyses of phenological
patterns (Bencke & Morellato 2002).
STUDY SITE.—We developed our study in an area of woody cer- To calculate the intensity index, we adapted the Fournier
rado in the municipality of Itirapina, S~ao Paulo state, southeastern index (Fournier 1974), considering ‘2’ as the class of maximum
Brazil (22°100 31″ S; 47°520 26″ W). The study site is located intensity (Vogado et al. 2016). We choose to use only three inten-
770 m above sea level and covers an area of 260 ha. The climate sity classes instead of five, as suggested by Fournier (1974), and
is characterized by two markedly regular seasons: a dry (winter) commonly used in tropical plant phenology, to facilitate the defi-
season from April to September and a wet (summer) season nition of phenological changes at monthly intervals.
444 Camargo et al.

FIGURE 1. Climate between 2005 and 2011 in Itirapina, S~ao Paulo, Brazil. The rainfall as daily (circles) and monthly (bars) accumulated precipitation. Tempera-
ture as monthly mean (line) and daily means (black circles). Day length is shown as daily values and monthly difference.
 Leafing Patterns of a Cerrado Community 445

LEAF EXCHANGE STRATEGIES AND THE DEGREE OF (Zar 1999), to infer the beginning of the growing season, we
DECIDUOUSNESS.—We classified the individual and species leaf applied circular statistics to the individuals’ onset month of leaf
exchange strategy for each of the five years we scored leaf fall flush following Morellato et al. (2000, 2010). To do that, first we
intensity (2005, 2006, 2009 to 2011). For each individual, we clas- define the start of leafing season of each individual and year as
sified as deciduous the individual recorded with no leaves the first month we recorded leaf flush in each year. Then, we
(100% leaf fall) in at least one month, in a year. We classified as divided the 360° of a circumference by 12 and each month of
semi-deciduous the individual recorded with more than half of the year is represented by a range of 30° starting on January
the branches without leaves (intensity score ‘2’) in at least one (midpoint 15°, see Morellato et al. 2010, for details). Then, we
month in a year. Finally, we classified as evergreen the individual calculate the angle or date (month) correspondent to the most
observed with up to half of the branches without leaves (‘0’ or frequent leaf flush onset (circular mode), the mean angle or onset
‘1’ intensity score) throughout a given year. To estimate the date (the mean angle or mean onset data of the frequency distri-
degree of deciduousness, we calculated the proportion of decidu- bution of individuals around the circle). Finally, we tested whether
ous, semi-deciduous and evergreen individuals in the community the mean onset data are significantly concentrated around the
in each of the five years we evaluated leaf fall intensity. mean angle or date, as a proxy of seasonality by applying the
We classified all species with more than five individuals sam- Rayleigh test (Z) as suggested by Morellato et al. 2000, 2010,. If
pled as deciduous, semi-deciduous or evergreen based on the the mean onset data are significantly concentrated around the
above individual’s annual leaf exchange strategies. A species was mean angle or date, we consider the leafing pattern seasonal and
classified as deciduous if more than half of all individuals sam- the degree of synchrony or seasonality is represented by the
pled were classified as deciduous in at least one of the five years; length of the vector r. The vector r has no unity and ranges from
semi-deciduous if more than half of all individuals were classi- 0 (no seasonality or synchronicity) to 1 (perfect seasonal or syn-
fied as semi-deciduous in at least one of the five years; and as chronous; Morellato et al. 2010). We performed circular analyses
evergreen all the remaining species. per study year, for all the community, and by leaf exchange
strategies using Oriana 4.0 (Kovach Computing Services).
DATA ANALYSES.—As we scored the intensity of phenology using
an ordinal scale ranging from 0 to 2, we applied ordinal multino- RESULTS
mial regressions to assess the drivers of leaf fall and flush. We fit-
ted two models (leaf fall and flush) for species grouped according We sampled 2112 individuals and 106 species (Table S1—Online
with leaf exchange strategies, analyzing the number of individuals Supplementary Material). Over the seven years of observations,
with 0, 1, 2 and 100 percent of leaf fall intensities and 0, 1 and 2 some individuals died or were lost, leading to the final sample of
of leaf flush intensities. We also fitted multinomial models for the 1651 individuals and 102 species. The community phenological
whole community, regardless of leaf strategies. All models had patterns were described based on the total number of individuals
monthly rainfall, mean temperature, monthly day length differ- observed per month, in each year.
ence (longest minus shortest day within the month), and an inter- Of the 106 species initially surveyed, 69 species presented
action between rainfall and temperature as predictors. Monthly more than five individuals and were classified as deciduous (26
rainfall, maximum, mean and minimum temperatures for the species; total basal area = 1.46 m2), semi-deciduous (25 species;
study period (January 2005 to December 2011) were obtained total basal area = 6.10 m2) and evergreen (18 species; total basal
from the climatic station at Estacß~ao Climatol
ogica do Centro de area = 2.29 m2) (Fig 2, Table S1). Species and individual’s
Recursos H
ıdricos e Ecologia Aplicada (CRHEA—Universidade degree of deciduousness varied among years (Fig. 2). Only five
de S~ao Paulo), 4 km from the study site. We used day length data species presented the same deciduousness degree over the five
for the latitude 22°170 S; 47°870 W (maptools package for R). years: Jacaranda caroba, which reached deciduity in all the years
To better understand the nature of the relationships between (most of individuals with no leaves or 100% leaf fall) and,
predictors and leaf exchange strategies, we further fitted general- Anadenanthera peregrina var. falcata, Byrsonima coccolobifolia, Campo-
ized additive models (GAM) using intensity data. These models manesia pubescens and Dalbergia miscolobium, which reached semi-
allowed us to identify threshold regions in environmental vari- deciduity in all the years (most individuals lose more than half
ables and long-term trends for leaf fall (number of individuals of their leaves). Individuals and consequently, the species, rarely
with 100% leaf fall) and presenting leaf flushing. We fitted all lose all their leaves, maintaining at least half of the leaves in
additive models with a correlation structure based on years. We most of the years (Fig. 2). We observed the lowest deciduous-
evaluated the ordinal models by looking at the significance of ness degree in the community in 2005 and the highest one in
each variable and the generalized additive models with R2 and 2010 (Fig. 2). The number of deciduous individuals responded
variable significance. We carried out all analyses in R (R Core to both mean temperature and total rainfall when intra and
Team 2017) using base packages ordinal (Christensen 2016) for interannual seasonality components were considered (R2 = 0.70;
the ordinal regressions and mgcv (Wood 2011) for the general- P < 0.01).
ized additive models. Leafing patterns were seasonal and similar among years and
Considering that recurrent temporal events, such as pheno- within leaf exchange strategies (Figs. 3 and 4). Leaf fall predomi-
logical data, are circular in nature, with no true starting point nated in the dry season, peaking in July (Fig. 3), while leaf flush
446 Camargo et al.

FIGURE 2. Percentage of species (A) and individuals (B) per deciduousness degree in each year according to the leaf fall intensity and the percentage of species
per leaf exchange strategy (C) in a cerrado woody community, southeastern Brazil. For the definitions of individual and species leaf exchange strategy and degree
of deciduousness refer to the Methods section.

was concentrated in the transition between dry and wet seasons, between evergreen and semi-deciduous species, but higher for
peaking in September (Fig. 4). We observed an unexpected deciduous species (Table 1). Leaf fall patterns of all strategies
increase in leaf fall on December 2009 and, on 2010 mainly in were negatively related to the rainfall and temperature, and posi-
the middle of the wet season, related to a larger number of indi- tively related to the day length difference (Table 1). The strongest
viduals losing leaves but at a low intensity (Fig. 3). relations were among leaf fall and temperature and temperature-
Leafing patterns according to the leaf exchange strategies rainfall interaction (Table 1).
were similar among years, overlapping during the periods of When analyzing the variables separately with generalized
higher activity and intensity in the community (Figs. 3 and 4). additive models, we confirm a higher climatic influence and sea-
Deciduous species presented a seasonal leaf exchange patterns sonality in the deciduity of semi and, mainly, deciduous species
highly constricted to the one period of the year, concurrent with (Fig. 5). All leaf fall models had high coefficients of determina-
the highest phenological activity and intensity in the community tion (evergreen R2 = 0.53; semi-deciduous R2 = 0.70; deciduous
(Figs. 3 and 4). During the peaks of leaf fall and leaf flush in the R2 = 0.77). Similarly, all variables but rainfall in the evergreen
community, deciduous species presented the highest synchrony model presented a significant nonlinear effect on leaf fall. A
among individuals and phenophases intensity followed by semi- higher proportion of individuals lost leaves when the rainfall was
deciduous and evergreen species. Conversely, semi-deciduous, and around 50 mm, the temperature between 16 and 19°C and the
mainly evergreen species, presented more individual’s activity out- day length between 11 and 11.5 h (Fig. 5), which corresponds to
with the community peaks of leaf exchange (Figs. 3 and 4). the dry season conditions (see Fig. 1). Individuals deciduity
In general, for both phenophases, leaf fall and flush, the decreased when precipitation was around 100 mm, the tempera-
effects of abiotic variables increased from evergreen to deciduous ture 22°C and the day length 12 h (Fig. 5), corresponding to the
species (Table 1). The exceptions were the influence of the day transition from dry to wet seasons.
length difference on the leaf fall, almost the same for all leaf Leaf flush patterns were positively related to the day length
exchange strategies; and the rainfall on the leaf flush, similar difference, but also to the rainfall, and negatively related to the
 Leafing Patterns of a Cerrado Community 447

FIGURE 3. Leaf fall phenology on percent of individuals (A) and percent of intensity (B) in a cerrado woody community according to the species’ leaf exchange
strategy: deciduous (26 species), semi-deciduous (25 species) and evergreen (18 species). The grey bars indicate the wet season.

temperature (Table 1). Even with a stronger effect of abiotic trees. The community growing season started in early September
variables on deciduous and semi-deciduous species (Table 1), in four of the seven years and otherwise in late August, with the
when analyzed separately with generalized additive models, indi- mean angle between 20 August and 10 September (Table 2).
viduals of species with different leaf exchange strategies When analyzed by leaf exchange strategy, the beginning of the
responded in a similar way to the environmental variables, show- growing season was observed mainly on September for deciduous
ing seasonality over the 7 yr of observations. Leaf flush for all species and between August and September for semi-deciduous
strategies responded nonlinearly to the abiotic variables (ever- and evergreen species (Table S2). Individuals synchronicity or
green R2 = 0.55; semi-deciduous R2 = 0.56; deciduous concentration around the mean onset date were higher for decid-
R2 = 0.62; Fig. 6). We observed the first increase in leaf flush uous species, with r vector around 0.82, followed by semi-decid-
activity when precipitation was around 100 mm, and the day uous and evergreen species, with the length of vector r around
length about 11.5 h, in the transition from dry to wet season. 0.69 and 0.62, respectively (Table S2).
The proportion of individuals flushing peaked at around
200 mm, 19°C and a day length close to 12 h (Fig. 6), which DISCUSSION
corresponds to the middle of the wet season, when the differ-
ences in the amount of sunlight are highest (see Fig. 1). Particu- Cerrado is the most species-rich savanna in the world and sec-
larly, rainy months (>300 mm) also seem to trigger flushing, ond largest Brazilian vegetation type, after Amazonia (Coutinho
although months with such amounts of accumulated rain are 2006, Overbeck et al. 2015, Durigan & Ratter 2016). However,
rare (Fig. 6A). the studies investigating cerrado seasonal patterns are still
We determined the leaf flush onset for half of the individu- sparse, representing less than 10 percent of the phenological
als observed (Table 2), with no evident onset for all remaining studies surveyed for Neotropical region (Mendoza et al. 2017),
448 Camargo et al.

FIGURE 4. Leaf flush phenology on percent of individuals (A) and percent of intensity (B) in a cerrado woody community according to the species’ leaf
exchange strategy: deciduous (26 species), semi-deciduous (25 species) and evergreen (18 species). The gray bars indicate the wet season.

and less than half of those include leafing phenology (Morel- number of individuals putting out new leaves at this time in
lato et al. 2013, unpublished data). Based on our unique seven- the community.
year phenological data, we showed the predominance of decid- In the cerrado community studied, the predominance of
uous and semi-deciduous leaf exchange strategies and estab- deciduous or semi-deciduous species (74%), in relation to the
lished the woody cerrado as an overall seasonal, semi- evergreen ones (26%), was related to the climate seasonality with
deciduous vegetation. The individual’s degree of deciduousness a marked dry season and has also been associated with soil prop-
was highly explained by the climate conditions, with individuals erties and species’ leaf-economy strategies (Franco et al. 2005,
rarely reaching the full deciduity. The cerrado leaf exchanging Souza et al. 2015a). Studies in cerrado sensu stricto sites in Brazil
patterns were consistently seasonal and similar among the Central also found a high proportion of deciduous and semi-
years, confirming the expected seasonality described by short- deciduous woody species (Pirani et al. 2009, Rossatto et al. 2009),
term studies on savannas and other tropical seasonal vegetation while in West African and Australian savannas the strategies are
(Miranda 1995, Williams et al. 1999, Munhoz & Felfili 2005, more evenly distributed in the community (Eamus & Prior 2001).
Lenza & Klink 2006, Pirani et al. 2009). Leaf fall and flush of Semi-deciduous species predominate in Australia, and evergreen
cerrado species were related to all abiotic variables, but mainly and semi-deciduous species in West Africa, (Williams et al. 1997,
to temperature, rainfall-temperature interaction and day length De Bie et al. 1998). Different definitions have been used to clas-
difference, and semi-deciduous and deciduous species were the sify leaf exchange strategies, constraining the comparisons regard-
most constrained by the climate. The onset of the community ing to the amount of deciduity of cerrado savanna and other
growing season, here defined based on a subset of the seasonal dry vegetation (e.g., Reich 1995, Eamus & Prior 2001).
species, started in the transition from dry to wet season A common definition brings deciduous, brevi-deciduous (differ-
(August–September), and it is sustained by the overall larger ing in the length of total leafless period) and evergreen with short
 Leafing Patterns of a Cerrado Community 449

mostly the interannual rainfall variability. That variability is

TABLE 1. Results of the ordinal multinomial models for the relation between leaf fall
expected for seasonally dry forest and savannas (Reich 1995,
and leaf flush (individuals’ activity) and total rainfall, mean temperature,
Eamus & Prior 2001, Goldstein et al. 2008), but has never been
monthly day length difference and rainfall–temperature interaction, according
demonstrated for tropical savannas based on individual long-term
to species leaf exchange patterns. The Estimate value indicates the intensity a
leaf exchange patterns. Therefore, it is necessary exercise caution
variable influences a given phenophase. All the P values were <0.001.
when classification of species leaf exchange strategies is derived
Leaf exchange Standard from short-term observation of vegetative phenology, as it can be
pattern Models Estimate error Z value underestimated whether the survey is conducted in a ‘wet’ year.
Considering the number of leaf exchangers species, their basal
Leaf fall
area and on the fact that individuals’ deciduousness degrees rarely
Evergreen Total rainfall 0.031 0.009 3.313
reached 100 percent leaf fall, we classified the studied cerrado
Mean temperature 0.271 0.009 30.755
community as a semi-deciduous vegetation and that assumption
Day length difference 0.067 0.006 10.655
might apply for general cerrado woodlands (cerrado sensu stricto).
Rainfall 9 temperature 0.258 0.010 25.372
The consistent seasonal leafing patterns observed over seven
Semi-deciduous Total rainfall 0.151 0.008 19.896
years for the studied cerrado community substantiated the pat-
Mean temperature 0.352 0.007 49.048
terns described for other cerrado vegetation based on short-term
Day length difference 0.051 0.005 9.957
observations, with leaf fall largely in the middle of the dry season
Rainfall 9 temperature 0.375 0.008 45.301
and the leaf flush in the transition from dry to wet season, begin-
Deciduous Total rainfall 0.287 0.010 28.350
ning at the end of dry season, after the first shower rains, but
Mean temperature 0.416 0.010 43.700
before the period of high rainfall intensity (Lenza & Klink 2006,
Day length difference 0.068 0.007 10.040
Pirani et al. 2009, Rossatto et al. 2009, Silv
erio & Lenza 2010,
Rainfall 9 temperature 0.469 0.011 42.940
Alberton et al. 2014). This leafing pattern is also found for the
Leaf flush
Australian savannas (Williams et al. 1997, Prior et al. 2004). The
Evergreen Total rainfall 0.135 0.010 13.010
observed peak of leaf fall in the dry season is expected for sea-
Mean temperature 0.086 0.010 8.780
sonally dry vegetations and species, as it is a strategy to improve
Day length difference 0.317 0.008 42.023
water use and avoid water losses by transpiration (Monasterio &
Rainfall 9 temperature 0.095 0.012 8.001
Sarmiento 1976, Reich & Borchert 1984, Reich 1995, Pedroni
Semi-deciduous Total rainfall 0.093 0.009 10.265
et al. 2002, Lenza & Klink 2006, Goldstein et al. 2008). Other
Mean temperature 0.290 0.009 33.285
effects acting as drivers of leaf fall such as seasonal dust accumu-
Day length difference 0.424 0.007 62.288
lation and herbivory were not assessed in this study. Despite this
Rainfall 9 temperature 0.075 0.010 7.185
being a less explores topic, few studies investigated the effects of
Deciduous Total rainfall 0.351 0.012 28.760
biomass harvesting by leaf-cutter ants, showing partial or total
Mean temperature 0.385 0.012 32.030
defoliation in some cerrado’s plant species (Costa et al. 2008,
Day length difference 0.567 0.010 55.570
Mundim et al. 2012). Also, the damage by insects was also more
Rainfall 9 temperature 0.307 0.015 20.220
intense in young leaves during the late dry season, corresponding
to the main period of leaf flushing in the cerrado (Marquis et al.
2001). Therefore, herbivory might change leaves exchanges pat-
or long leaf flush strategies or low or no deciduity (Williams et al. terns by accelerating leaf aging due to damage or reducing photo-
1997, Franco et al. 2005, Lenza & Klink 2006), but their analyses synthetic activity. Moreover, insect-related damage and parasitism
are usually performed contrasting deciduous and evergreens. have negative effects on plant reproduction (Franco 1998, Mar-
Reich (1995) propose deciduous with synchronous leaf flush, quis et al. 2002), factors that could lead to changes on cerrado
deciduous with asynchronous leaf flush and evergreens. We chose phenological dynamics.
the large accepted, more inclusive, deciduous class including both, The peak of leaf flush preceding the start of the rainy sea-
brevi- and deciduous species, all losing all leaves or 100 percent son, observed in cerrado woody plants, is still an intriguing pat-
regardless the length of leafless phase and leaf flushing at the end tern. The predominance of species flushing at the end of the dry
of dry season; a semi-deciduous strategy that implies in the loss season supports the hypothesis that cerrado trees have constant
of a representative amount of leaves, never above a 50 percent access to the water in the soil due to the deep root system of
canopy cover, and evergreens that retain all or almost all leaves most species and that the preceding leaf fall contributes to rehy-
year round (Eamus & Prior 2001). That choice facilitates our dration and the early production of new leaves (Eamus & Prior
comparisons and fit well to our monthly direct observations (see 2001, Goldstein et al. 2008). The positive correlations with the
also Williams et al. 1997). rainfall and day length difference may indicate the favorable con-
Here, we demonstrated, based on our long-term phenology ditions in the transition season to the production of cerrado new
data, that independently of the leaf exchange strategy, individuals’ leaves. Day length difference was the main factor positively pre-
deciduousness degree and, consequently, species’ deciduousness dicting the leaf flushing pattern of cerrado, mainly for deciduous
degree vary among years according to the climate conditions, species, and explains the certainty of the leafing season. For
450 Camargo et al.

FIGURE 5. Fitted generalized additive models’ responses of the proportion of individuals without leaves to the abiotic variables—rainfall (A), temperature (B)
and day length difference (C)—and since the first phenological observation in the field (D).

tropical seasonal dry vegetation, day length is suggested as the species, with evergreens significantly, but less constrained, by the
main trigger for leaf flush, independently of the leaf exchange climate and day length difference, mainly for leaf fall. Leaf fall
strategy (Borchert 2000, Borchert & Rivera 2001, Rivera et al. restricted to the dry season has been described for several short-
2002, Higgins et al. 2011, Rossatto et al. 2013, but see Reich term savanna studies (e.g., Williams et al. 1997, Lenza & Klink
1995, Goldstein et al. 2008). Flushing before the more intense 2006, Pirani et al. 2009). Rossatto et al. (2013) also observed a
rainfall can also avoid loss of nutrient by leaching during the leaf coincident leaf flush onset in cerrado species, but with an
development and favor nutrient uptake before the rainy season, extended leaf flush into the wet season for cerrado evergreen
the flowering and fruiting time of many species, due to the high species. At some savannas, the evergreen species seem to have a
luminance and photosynthetic activity of emerging new leaves more diversified leaf exchange strategy, some flushing new leaves
(Sarmiento et al. 1985, Rossatto et al. 2009). all year long while others concentrating the leaf production in the
Deciduous species presented leaf fall mostly constrained to dry-to-wet season transition (Reich 1995, Franco et al. 2005).
the middle of the dry season and leaf flush to the transition Independent of the leaf strategy, if we simulate the GAM
between dry and wet seasons. Although semi-deciduous and ever- analyses with abiotic variables constant over time, the cerrado
green species also presented a seasonal leafing pattern, some indi- leafing pattern is still seasonal (Figs. 5D and 6D). This resilient
viduals showed leaf exchange activity out of these periods. These seasonal pattern in addition to the high correlation with all the
results agree with the individual’s response to abiotic variables, as abiotic variables makes it hard to establish the triggers and con-
we found a gradual response from deciduous to evergreen straints of vegetative patterns in cerrado, as for other savanna
 Leafing Patterns of a Cerrado Community 451

FIGURE 6. Fitted generalized additive models’ responses of the proportion of individuals flushing new leaves to the abiotic variables—rainfall (A), temperature
(B) and day length difference (C)—and since the first phenological observation in the field (D).

TABLE 2. Circular statistics analyses of the leaf flush onset, per year, in the individuals of a cerrado woody community, southeastern Brazil. Individuals with marked onset: number of
individuals, from the total of observed individuals, for which we could define the leaf flush onset date and include in the circular analyses; (n) number of individuals which
presented the leaf flush onset in the indicated month; onset mean angle: mean angle of data distribution around the circle; onset mean date: month correspondent to the mean
angle; *P < 0.05.

2005 2006 2007 2008 2009 2010 2011

Observed individuals 2112 2111 2100 2046 1979 1947 1891

Individuals with marked onset 1300 1384 1000 876 816 778 994
Most frequent onset month (n) Sep (665) Aug (537) Sep (447) Aug (423) Sep (348) Sep (290) Aug (359)
Onset mean angle (l) 236.4° 229.6° 249.8° 238.3° 245.8° 248.4° 236.0°
Onset mean date Aug Aug Sep Aug Sep Sep Aug
Length of mean vector (r) 0.69 0.64 0.59 0.82 0.76 0.72 0.74
Rayleigh test (Z) 617.21* 560.45* 346.05* 583.44* 468.37* 399.08* 550.24*
452 Camargo et al.

communities (Higgins et al. 2011). Some authors suggest that the the farm. We also thank all the former and actual members of
predominance of woody deciduous or semi-deciduous species, Phenology Laboratory for their invaluable contribution to pheno-
that seasonally change completely or partially their leaves, is an logical data collection in the field and for the many fruitful dis-
adaptation to the seasonal water restrictions, not directly related cussions. Our research was supported by S~ao Paulo Research
to the rainfall but also to the soil water availability and air humid- Foundation (FAPESP) grants #2007/59779-6 and #2009/54208-
ity (Borchert 1994a, 1994b, Reich 1995, Eamus & Prior 2001, 6; FAPESP-Microsoft Research #2013/50155-0 and #2010/
Franco et al. 2005). Apart from abiotic factors, the cerrado 52113-5; FAPESP-VALE-FAPEMIG #2010/51307-0. LPCM
woody species leafing patterns have been related to herbivory holds a research productivity fellowship from CNPq (National
avoidance strategies and plant physiology, according to the differ- Council for Scientific and Technological Development). BA
ent leaf exchange strategies or for aluminum accumulating or (#2014/00215-0 and #2016/01413-5), MGGC (#2010/01762-3
non-accumulating species (Franco et al. 2005, Borges & Prado and #2015/10754-8) and GHC (#2015/14292-9) receive fellow-
2014, Souza et al. 2015b). ships from FAPESP.
We defined the community growing season onset as the dry-
to-wet season transition (August–September). This finding was DATA AVAILABILITY
based on a subset of the community’ species showing a well-defined
rest season and flushing onset, but also in the overall community Data available from the Dryad Digital Repository: https://doi.
significant seasonality and, increase in leaf flushing on early org/10.5061/dryad.ck21rk8 (Camargo et al. 2018).
September. The observed decrease in the leaf fall activity, mainly
for deciduous and semi-deciduous species, and the peak of leaf SUPPORTING INFORMATION
flush activity, for all strategies, when the climatic conditions corre-
sponds to the dry-to-wet season transition also indicates the end of Additional Supporting Information may be found online in the
the cerrado resting season and start of the growing season (Table 1; supporting information tab for this article:
Fig. 5). The absence of a marked resting season for half of the spe-
cies was related to the fact that, even for deciduous species, individ- TABLE S1. Plant species of a cerrado woody community included in the
uals rarely lose all their leaves (100% leaf fall), and that leaf fall, for phenological observations, with the respective: number of individuals sampled
some semi-deciduous and mainly evergreen species, is observed in per species the basal area (m2), leaf exchange strategy and deciduousness
small intensities all year long. Another point is that the absence of a degree in each year of intensity data.
community-wide, well-defined leaf flush onset for all species is not TABLE S2. Circular statistics analyses of the leaf flush onset, per year,
expected for tropical communities, due to the high diversity of spe- in the individuals of a cerrado woody community according to their species
cies and phenological patterns (Morellato et al. 2013, 2016), even leaf exchange strategy.
for seasonal dry vegetations. Alberton et al. (2014), working with
digital cameras daily repeated photographs or near surface phenol- LITERATURE CITED
ogy at the same cerrado site, detected an identical timing of the
growing season as revealed in the present study. Their cross-valida- ALBERTON, B., J. ALMEIDA, R. HELM, R. S. TORRES, A. MENZEL, AND L. P. C.
MORELLATO. 2014. Using phenological cameras to track the green up
tion with the direct monthly observations rules out any inherent in a cerrado savanna and its on-the-ground validation. Ecol. Inform.
subjectivity of our leaf exchange direct observations. 19: 62–70.
The predominance of a distinct leaf flushing season was ARCHIBALD, S., AND R. J. SCHOLES. 2007. Leaf green-up in a semi-arid African
demonstrated for some seasonal dry vegetations around the savanna–separating tree and grass responses to environmental cues. J.
world, the marked dry seasons favoring fast response, oppor- Veg. Sci. 18: 583–594.
BATALHA, M. A., S. ARAGAKI, AND W. MANTOVANI. 1997. Phenological varia-
tunistic species flushing new leaves at the end of dry season, after tions of the cerrado species in Emas – Pirassununga, SP. Acta Bot.
the first rains (Vico et al. 2015). We argue that cerrado could inte- Brasilica 11: 61–78.
grate the ecohydrological framework proposed by Vico et al. BATALHA, M. A., AND W. MANTOVANI. 2000. Reproductive phenological pat-
(2015) for seasonally dry ecosystems. Considering the interannual terns of cerrado plant species at the P
e-De-Gigante Reserve (Santa
variability in the degree of deciduousness of cerrado individuals Rita Do Passa Quatro, SP, Brazil): a comparison between the herba-
ceous and woody floras. Rev. Bras. Biol. 60: 129–145.
and, consequently species over this 7-year study, the importance BATALHA, M. A., AND F. R. MARTINS. 2004. Reproductive phenology of the
of temperature and rainfall as drivers of leaf fall and flush, and cerrado plant community in Emas National Park (Central Brazil).
the similarity to other seasonally dry ecosystems, the cerrado leaf Aust. J. Bot. 52: 149–161.
exchange phenology is likely susceptible to future climate changes BENCKE, C. S. C., AND L. P. C. MORELLATO. 2002. Comparison of two meth-
scenarios of raising temperature and reduced rainfall (Reich 1995, ods of plant phenology estimation, their interpretation and representa-
tion. Rev. Bras. Bot. 25: 269–275.
Eamus & Prior 2001, IPCC 2004, Vico et al. 2015). BORCHERT, R. 1994a. Soil and stem water storage determine phenology and
distribution of tropical dry forest trees. Ecology 75: 1437–1449.
ACKNOWLEDGMENTS BORCHERT, R. 1994b. Water status and development of tropical trees during
seasonal drought. Trees 8: 115–125.
We thank the owners of Fazenda S~ao Jos
e and the Instituto BORCHERT, R. 2000. Organismic and environmental controls of bud growth in
tropical trees. In J. D. Viemont, and J. Crabbe (Eds.). Dormancy in
Arruda Botelho (IAB) for allowing us to conduct fieldwork at
 Leafing Patterns of a Cerrado Community 453

plants: from whole plant behavior to cellular control Garden, pp. 87– JACKSON, P. C., F. C. MEINZER, M. BUSTAMANTE, G. GOLDSTEIN, A. FRANCO, P.
108. CAB International, Wallingford. W. RUNDEL, L. CALDAS, E. IGLER, AND F. CAUSIN. 1999. Partitioning of
BORCHERT, R., AND G. RIVERA. 2001. Photoperiodic control of seasonal devel- soil water among tree species in a Brazilian Cerrado ecosystem. Tree
opment and dormancy in tropical stem-succulent trees. Tree Physiol. Physiol. 19: 717–724.
21: 213–221. LAVOREL, S., S. D
IAZ, J. H. C. CORNELISSEN, E. GARNIER, S. P. HARRISON, S.
BORGES, M. P., AND C. H. B. A. PRADO. 2014. Relationships between leaf MCINTYRE, J. G. PAUSAS, N. P. CATHERINE, AND R. CARLOS. 2007.
deciduousness and flowering traits of woody species in the Brazilian Plant functional types: are we getting any closer to the Holy Grail?
neotropical savanna. Flora Morphol. Distrib. Funct. Ecol. Plants 209: In J. G. Canadell, D. E. Pataki, and L. F. Pitelka (Eds.). Terrestrial
73–80. ecosystems in a changing world, pp. 149–164. Springer, Berlin, Hei-
~ , C. F., AND P. S. FIGUEIREDO. 2002. Phenology of leguminous trees in
BULHAO delberg.
an area of cerrado in the northeast of Maranh~ao. Rev. Bras. Bot. 25: LENZA, E., AND C. A. KLINK. 2006. Phenological behavior of woody species
361–369. in a “cerrado” sensu stricto of Bras
ılia, DF. Brazilian J. Bot. 29: 627–
CAMARGO, M. G. G., G. H. de CARVALHO, B. C. de ALBERTON, P. REYS, AND L. 638.
P. C. MORELLATO. 2018. Data from: Leafing patterns and leaf exchange MANTOVANI, W., AND F. R. MARTINS. 1988. Phenological variations in species
strategies of a cerrado woody community. Dryad Digital Repository. of the “cerrado” of the Moji Guacßu Biological Reserve, S~ao Paulo
https://doi.org/10.5061/dryad.ck21rk8. State. Revta. Bras. Bot 11: 101–112.
CAMARGO, M. G. G., R. M. SOUZA, P. REYS, AND L. P. C. MORELLATO. 2011. MARQUIS, R. J., I. R. DINIZ, AND H. C. MORAIS. 2001. Patterns and correlates
Effects of environmental conditions associated to the cardinal orienta- of interspecific variation in foliar insect herbivory and pathogen attack
tion on the reproductive phenology of the cerrado savanna tree Xylo- in Brazilian cerrado. J. Trop. Ecol. 17: 127–148.
pia aromatica (Annonaceae). An. Acad. Bras. Cienc. 83: 1007–1019. MARQUIS, R. J., H. C. MORAIS, AND I. R. DINIZ. 2002. Interactions among cer-
CHRISTENSEN, R. H. B. 2016. Regression Models for Ordinal Data [R package rado plants and their herbivores: unique or typical. In P. S. Oliveira,
ordinal version 2015.6-28]. and R. J. Marquis (Eds.). The cerrados of Brazil, pp. 306–328. Colum-
CONDIT, R., K. WATTS, S. A. BOHLMAN, R. PE
REZ, R. B. FOSTER, AND S. P. bia University Press, New York, NY.
HUBBELL. 2000. Quantifying the deciduousness of tropical forest cano- MENDOZA, I., C. A. PERES, AND L. P. C. MORELLATO. 2017. Continental-scale
pies under varying climates. J. Veg. Sci. 11: 649–658. patterns and climatic drivers of fruiting phenology : a quantitative
COSTA, A. N., H. L. VASCONCELOS, E. H. M. VIEIRA-NETO, AND E. M. BRUNA. Neotropical review. Glob. Planet. Change 148: 227–241.
2008. Do herbivores exert top-down effects in Neotropical savannas? MIRANDA, I. S. 1995. Phenology of the arboreal stratum of a savanna commu-
Estimates of biomass consumption by leaf-cutter ants. J. Veg. Sci. 19: nity at Alter-do-Ch~ao, PA. Rev. Bras. Bot. 18: 235–240.
849–854. MONASTERIO, M., AND G. SARMIENTO. 1976. Phenological strategies of plants
COUTINHO, L. M. 1978. The concept of Cerrado. Rev. Bras. Bot. 1: 17–23. species in the tropical savanna and semi-deciduous forest of the
COUTINHO, L. M. 2006. The biome concept. Acta Bot. Brasilica 20: 13–23. Venezuelan Lianos. J. Biogeogr. 3: 325–356.
DALMOLIN, A.^ C., F. de ALMEIDA LOBO, G. VOURLITIS, P. R. SILVA, H. J. DAL- MORELLATO, L. P. C., L. F. ALBERTI, AND I. L. HUDSON. 2010. Applications of
MAGRO, M. Z. ANTUNES, AND C. E. R. ORT
IZ. 2015. Is the dry season circular statistics in plant phenology: a case studies approach. In I. L.
an important driver of phenology and growth for two Brazilian Hudson, and M. R. Keatley (Eds.). Phenological research, pp. 339–
savanna tree species with contrasting leaf habits? Plant Ecol. 216: 359. Springer Netherlands, Dordrecht.
407–417. MORELLATO, L. P. C., B. ALBERTON, S. T. ALVARADO, B. BORGES, E. BUISSON,
DAMASCOS, M. A., C. H. B. A. PRADO, AND C. C. RONQUIM. 2005. Bud com- M. G. G. CAMARGO, L. F. CANCIAN, D. W. CARSTENSEN, D. F. E. ESCO-
position, branching patterns and leaf phenology in Cerrado woody BAR, P. T. P. LEITE, I. MENDOZA, N. M. W. B. ROCHA, N. C. SOARES, T.
species. Ann. Bot. 96: 1075–1084. S. F. SILVA, V. G. STAGGEMEIER, A. S. STREHER, B. C. VARGAS, AND C.
DE BIE, S., P. KETNER, M. PAASSE, AND C. GEERLING. 1998. Woody plant phe- A. PERES. 2016. Linking plant phenology to conservation biology. Biol.
nology in the West Africa savanna. J. Biogeogr. 25: 883–900. Conserv. 195: 60–72.
DURIGAN, G., AND J. A. RATTER. 2016. The need for a consistent fire policy MORELLATO, L. P. C., M. G. G. CAMARGO, AND E. GRESSLER. 2013. A review
for Cerrado conservation. J. Appl. Ecol. 53: 11–15. of plant phenology in south and central America. In M. D. Schwartz
EAMUS, D., AND L. PRIOR. 2001. Ecophysiology of trees of seasonally dry trop- (Ed.). Phenology: an integrative environmental science, pp. 91–113.
ics: comparisons among phenologies. Adv. Ecol. Res. 32: 113–197. Springer Netherlands, Dordrecht.
FOURNIER, L. 1974. Un m
etodo cuantitativo para la medici
on de caracter
ısticas MORELLATO, L. P. C., AND H. F. LEITAO ~ -FILHO. 1989. Estrat
egias fenol
ogicas
fenol
ogicas en
arboles. Turrialba 24: 422–423. de esp
ecies arb
oreas em floresta mes
ofila na Serra do Japi, Jundia
ı,
FRANCO, A. C. 1998. Seasonal patterns of gas exchange, water relations and S~ao Paulo. Rev. Bras. Biol. 50: 163–173.
growth of it Roupala montana, an evergreen savanna species. Plant MORELLATO, L. P. C., D. C. TALORA, A. TAKAHASI, C. C. BENCKE, E. C.
Ecol. 136: 69–76. ROMERA, AND V. B. ZIPPARRO. 2000. Phenology of Atlantic rain forest
FRANCO, A. C., M. BUSTAMANTE, L. S. CALDAS, G. GOLDSTEIN, F. C. MEINZER, trees: a comparative study. Biotropica 32: 811–823.
A. R. KOZOVITS, P. RUNDEL, AND V. T. R. CORADIN. 2005. Leaf func- MUNDIM, F. M., E. M. BRUNA, E. H. M. VIEIRA-NETO, AND H. L. VASCON-
tional traits of Neotropical savanna trees in relation to seasonal water CELOS. 2012. Attack frequency and the tolerance to herbivory of
deficit. Trees - Struct. Funct. 19: 326–335. Neotropical savanna trees. Oecologia 168: 405–414.
GOLDSTEIN, G., F. C. MEINZER, S. J. BUCCI, F. G. SCHOLZ, A. C. FRANCO, AND MUNHOZ, C. B. R., AND J. M. FELFILI. 2005. Phenology of the herbaceous
W. A. HOFFMANN. 2008. Water economy of Neotropical savanna trees: layer in a campo sujo community in the Fazenda Agua
Limpa, Federal
six paradigms revisited. Tree Physiol. 28: 395–400. District, Brazil. Acta Bot. Brasilica 19: 979–988.
HIGGINS, S. I., M. D. DELGADO-CARTAY, E. C. FEBRUARY, AND H. J. COMBRINK. MUNHOZ, C. B. R., AND J. M. FELFILI. 2006. Phytosociology of the herb-sub-
2011. Is there a temporal niche separation in the leaf phenology of shrub layer in an area of campo sujo, Distrito Federal, Brazil. Acta
savanna trees and grasses? J. Biogeogr. 38: 2165–2175. Bot. Brasilica 20: 671–685.
IPCC. 2004. Climate change 2014: impacts, adaptation, and vulnerability. Part OPLER, P. A., G. W. FRANKIE, H. G. BAKER, S. JOURNAL, N. MAR, H. G.
A: global and sectoral aspects. Working Group I Contribution to the BAKER, D. C. RICA, C. UNIVERSITARIA, S. JOSE, AND C. RICA. 1980.
Fifth Assessment Report of the Intergovernmental Panel on Climate Comparative phenological studies of treelet and shrub species in tropi-
Change. Cambridge University Press, Cambridge, United Kingdom cal wet and dry forests in the lowlands of Costa Rica. J. Ecol. 68:
and New York, NY, USA. 167–188.
454 Camargo et al.

OVERBECK, G. E., E. VE
LEZ-MARTIN, F. R. SCARANO, T. M. LEWINSOHN, C. R. ROSSATTO, D. R., W. A. HOFFMANN, AND A. C. FRANCO. 2009. Differences in
FONSECA, S. T. MEYER, S. C. MULLER € , P. CEOTTO, L. DADALT, G. DURI- growth patterns between co-occurring forest and savanna trees affect
GAN, G. GANADE, M. M. GOSSNER, D. L. GUADAGNIN, K. LORENZEN, the forest-savanna boundary. Funct. Ecol. 23: 689–698.
C. M. JACOBI, W. W. WEISSER, AND V. D. PILLAR. 2015. Conservation in SARMIENTO, G. 1984. The ecology of neotropical savannas. Harvard University
Brazil needs to include non-forest ecosystems. Divers. Distrib. 21: Press, Cambridge, MA.
1455–1460. SARMIENTO, G., G. GOLDSTEIN, AND F. MEINZER. 1985. Adaptive strategies of
PEDRONI, F., M. SANCHEZ, AND F. A. M. SANTOS. 2002. Phenology of copa
ıba woody species in neotropical savannas. Biol. Rev. 60: 315–355.
(Copaifera langsdorffii Desf. – Leguminosae, Caesalpinioideae) in a semide- SILVE
RIO, D. V., AND E. LENZA. 2010. Phenology of woody species in a typical
ciduous forest, southeastern Brazil. Rev. Bras. Bot. 25: 183–194. cerrado in the Bacaba Municipal Park, Nova Xavantina, Mato Grosso,
PIRANI, F. R., M. SANCHEZ, AND F. PEDRONI. 2009. Phenology of a tree com- Brazil. Biota. Neotrop. 10: 205–216.
munity in a cerrado sensu stricto, Barra do Garcßas, Mato Grosso SOUZA, M. C., P. C. P. BUENO, L. P. C. MORELLATO, AND G. HABERMANN.
state, Brasil. Acta Bot. Brasilica 23: 1096–1110. 2015a. Ecological strategies of Al-accumulating and non-accumulating
POLGAR, C. A., AND R. B. PRIMACK. 2011. Leaf-out phenology of temperate functional groups from the cerrado sensu stricto. An. Acad. Bras.
woody plants: from trees to ecosystems. New Phytol. 191: 926–941. Cienc. 87: 813–823.
PRIOR, L. D., D. M. J. S. BOWMAN, AND D. EAMUS. 2004. Seasonal differences SOUZA, M. C., A. C. FRANCO, M. HARIDASAN, D. R. ROSSATTO, J. F. ARAUJO
, L.
in leaf attributes in Australian tropical tree species: family and habitat P. C. MORELLATO, AND G. HABERMANN. 2015b. The length of the dry
comparisons. Funct. Ecol. 18: 707–718. season may be associated with leaf scleromorphism in cerrado plants.
RATTER, J. A., S. BRIDGEWATER, R. ATKINSON, AND J. F. RIBEIRO. 1996. Analysis of An. Acad. Bras. Cienc. 87: 1691–1699.
the floristic composition of the Brazilian cerrado vegetation II: compar- TANNUS, J. L. S., M. A. ASSIS, AND L. P. C. MORELLATO. 2006. Reproductive
ison of the woody vegetation of 98 areas. Edinb. J. Bot. 53: 153–180. phenology in dry and wet grassland in an area of Cerrado at South-
R Core Team 2017. A language and environment for statistical computing. R eastern Brazil, Itirapina - SP. Biota. Neotrop. 6: bn02806032006.
Foundation for Statistical Computing, Vienna, Austria. VICO, G., S. E. THOMPSON, S. MANZONI, A. MOLINI, J. D. ALBERTSON, J. S.
REICH, P. B. 1995. Phenology of tropical forests: patterns, causes, and conse- ALMEIDA-CORTEZ, P. A. FAY, X. FENG, A. J. GUSWA, AND H. LIU. 2015.
quences. Can. J. Bot. 73: 164–174. Climatic, ecophysiological, and phenological controls on plant ecohy-
REICH, P. B., AND R. BORCHERT. 1984. Water stress and tree phenology in a drological strategies in seasonally dry ecosystems. Ecohydrology 8:
tropical dry forest in the lowlands of Costa Rica. J. Ecol. 72: 61–74. 660–681.
REYS, P., M. G. G. CAMARGO, M. T. GROMBONE-GUARATINI, A. D. P. TEIXEIRA, VOGADO, N. O., M. G. G. CAMARGO, G. M. LOCOSSELLI, AND L. P. C. MOREL-
M. A. ASSIS, AND L. P. C. MORELLATO. 2013. Structure and floristic LATO. 2016. Edge effects on the phenology of the guamirim, Myrcia
composition of a Cerrado sensu stricto and its relevance to ecological guianensis (Myrtaceae), a cerrado tree, Brazil. Top. Conserv. Sci. 9: 291–
restauration. Hoehnea 40: 449–464. 312.
RICHARDSON, A. D., T. F. KEENAN, M. MIGLIAVACCA, Y. RYU, O. SONNENTAG, WILLIAMS, R. J., B. A. MYERS, W. MULLER € , G. A. DUFF, AND D. EAMUS. 1997.
AND M. TOOMEY. 2013. Climate change, phenology, and phenological Leaf phenology of woody species in a north Australian tropical
control of vegetation feedbacks to the climate system. Agric. For. savanna. Ecology 78: 2542–2558.
Meteorol. 169: 156–173. WILLIAMS, R. J., B. A. MYERS, D. EAMUS, AND G. A. DUFF. 1999. Reproductive
RIVERA, G., S. ELLIOTT, L. S. CALDAS, G. NICOLOSSI, V. T. CORADIN, AND R. phenology of woody species in a North Australian Tropical Savanna.
BORCHERT. 2002. Increasing day-length induces spring flushing of trop- Biotropica 31: 626–636.
ical dry forest trees in the absence of rain. Trees-Struct. Funct. 16: WOOD, S. N. 2011. Fast stable restricted maximum likelihood and marginal
445–456. likelihood estimation of semiparametric generalized linear models. J. R.
ROSSATTO, D. R., W. A. HOFFMANN, L. C. R. de SILVA, M. HARIDASAN, L. S. L. Stat. Soc. Ser. B. 73: 3–36.
STERNBERG, AND A. C. FRANCO. 2013. Seasonal variation in leaf traits ZAR, J. H. 1999. Biostatistical analysis, 4th edn. Pearson Education India,
between congeneric savanna and forest trees in Central Brazil: implica- Delhi.
tions for forest expansion into savanna. Trees 27: 1139–1150.