See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/234150538

Relationships Among Flowers, Fruits, And Seeds

Article  in  Annual Review of Ecology and Systematics · November 2003

DOI: 10.1146/annurev.es.18.110187.002205

CITATIONS READS

361 20,421

1 author:

Richard Primack
Boston University
347 PUBLICATIONS   20,900 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Diversity, Equity and Inclusion in Ecology and Conservation View project

BACE - The Boston Area Climate Experiment View project

All content following this page was uploaded by Richard Primack on 04 January 2016.

The user has requested enhancement of the downloaded file.

 ANNUAL
REVIEWS Further
Quick links to online content
Ann. Rev. Ecol. Sysl.1987. 18:409-30
Copyright © 1987 by Annual Reviews Inc. All rights reserved

RELATIONSHIPS AMONG FLOWERS,

FRUITS, AND SEEDS
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.
Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

Richard B. Primack

Department of Biology, Boston University, Boston, Massachusetts 02215

INTRODUCTION

The development of diverse reproductive structures has been one of the major
factors in the evolution of the angiosperms (11, 78, 90). This diversity of
reproductive structures, especially in flowers, is the primary means of
identifying and classifying flowering plants. Systematists, ecologists, and
evolutionary biologists have traditionally treated reproduction in flowering
plants as constituted by three distinct phases, each associated with a large
literature: flowers and pollination (21, 47, 68), fruits and seed dispersal (20,
43, 60, 88, 91 ) , seed and seedling establishment (39, 42).
Specialized studies that consider only one phase of reproduction are typical
in the botanical literature. As an example of the viewpoint that justifies
specialized studies, Cronquist (16, p. 79) says 'The 'morphological integra­
tion' which often permits students of vertebrates to reconstruct an entire
animal from a few bones simply does not exist among the angiosperms.
Another facet of this same situation is that higher plants, being non-motile and
lacking a nervous system, do not have their structure so precisely prescribed
by their environmental requirements as do the animals." This viewpoint that
characters of flowering plants are not closely integrated in the whole plant is
echoed in the writings of virtually all other systematic botanists and is found
in current introductory biology textbooks. At the same time, field researchers
in reproductive ecology are beginning to extend the scope of their studies into
a broader context, blurring the edges of the three phases of reproduction. For
example, in a study of the lily, Clintonia borealis, floral ecology is shown to
have significant effects on both fruit set and seed size (25). Alpine biennial
and perennial gentians differ in aspects of their pollination ecology and

409
0066-4162/8711 120-0409$02.00
 410 PRIMACK

flowering phenology, which in turn are related to seed production and viabil­
ity (76).
The purpose of this paper is to present arguments supporting the idea that
there are close relationships among flowers and pollination, fruits and seed
dispersal, and seed and seedling establishment. These arguments are sup­
ported with diverse evidence drawn from the literature. Finally, six selected
published monographs of plant genera are examined to provide specific
examples.
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

RELATIONSHIP BETWEEN FLOWERS AND FRUITS

Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

A morphological continuity exists among flowers, fruits, and seeds. After all,
it is the ovary of the flower that develops into the fruit, which in turn contains
the seeds. At the extreme are naked pistillate flowers, such as the oak
(Quercus) and many species of Moraceae, such as Antiaris, in which most of
the biomass of the pistillate flower is in fact ovary tissue. Flower parts, in
particular the sepals of many species in the Rhizophoraceae and Dipterocarpa­
ceae, remain attached to the developing fruit and may function in increasing
the photosynthate available to the fruit (8). In addition, the sepals often fold
over the enlarging ovary and function as an additional layer of protection for
the immature seeds. In many species with inferior ovaries, the outer layer of
fruits develops from floral tissue derived from the sepals, petals, and fila­
ments (11).
In general, the individual flower parts form an integrated structure that
must be effective in pollen donation and receipt. In any comparison across
either a phylogenetic grouping or a community, species with large flowers
possessing large petals also tend to have large sepals, filaments, anthers,
ovary, stigma, and style. Positive correlations among stigma depth, style
length, and pollen grain size have also been noted (17, 92). One simple
explanation for many of these correlations is pleiotropic effects (31), possibly
for genes controlling increases in cell size and cell number in floral parts. The
resources (i.e. carbohydrates, proteins, minerals, etc) contained within the
ovary are incorporated into the developing fruit after fertilization. It is reason­
able that species with large flowers and associated ovaries will tend to have
large fruits, because it is unlikely that the fertilized ovary will get smaller
during fruit development. If resources from other flower parts, in particular
the sepals, are translocated jnto the developing fruit following fertilization,
again it is reasonable that species with large flowers will tend to have large
fruits. This can be stated in the following way:

Prediction 1 Species with large flowers will almost always have large fruits.
Species with small flowers may have either small or large fruits
(Figure 1).
 FLOWERS, FRUITS, AND SEEDS 411

ESTABLISHMENT IN SHADY,
UNDISTURBED HABITATS SPRING FLOWERING

I LARGE FRUIT
I
LARGE SEEDLINGS DISPENSERS EXTENDED TIME

I /
I FOR FRUIT MATURATION
LARGE SEEDS

OR '- I
SUMMER FRUITING

"
MANY SMALL SEEDS � LARGE FRUITS

,
/ I
LARGE LEAVES
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

I
LOW FRUIT SET
� THICK TWIGS
Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

MAINLY
LARGE FLOWERS
\
\
-"
CROSS-POLLINATING / / TALL PLANTS

ABUNDANT LARGE
-
FLORAL POLLINATORS
REWARDS

ESTABLISHMENT IN SUNNY. FLOWERING AND FRUITING AT

DISTURBED CONDITIONS TIME OF PEAK RESOURCE
PASSIVE
I AVAILABILITY

I
SEED DISPERSAL

t
SMALL SEEDLINGS

I
�
BRIEF UIT

/
SMALL SEEDS MATURATION TIME

I SMALL FRUITS
'
fill'

THIN TWIJGS I
MANY SEEDS ALL LEAVES

HIGH FRUIT SET

I SMALL FLOWERS , ,
'
MAINLY
� ..;
SHORT PLANTS

I
SELF-POLLINATING �

LIMITED SMALL
FLORAL REWARDS - POLLINATORS

Figure I Relationships among reproductive and vegetative characters during selection for larger
sizes of structures (above) and smaller sizes of structures (below). A thick line predicts a strong
relationship between pairs of characters, while a thin line represents a weaker predicted relation­
ship.

RELATIONSHIP AMONG FRUITS AND SEEDS

A major component of many fruits is the seeds contained inside. While in

certain species, such as blueberries (Vaccinium). the seeds only constitute a
small percent of the final fruit volume, in many other species of nuts, drupes,
and achenes, such as chestnut (Castanea) and sunflower (Helianthus), the
seed forms the majority of the final fruit volume and weight. Consequently,
fruit size will immediately be affected by any selective forces acting on seed
size. While it may be rather obvious to say, for a particular species fruit size
 412 PRIMACK

can never be less than seed size. Species with small seeds can have either
small fruits with one or a few seeds or fruits ranging up to larger sizes with
numerous seeds. Species with large seeds, such as mango (Mangifera),
avocado (Persea), and rambutan (Nephelium) have large fruits with resulting
specialized modes of dispersal. This is illustrated by a survey of tropical
bird-dispersed fruits, demonstrating that birds with specialized feeding habits
tend to feed on large fruits containing large seeds while birds with a greater
dietary range tend to feed on smaller fruits containing smaller seeds (55, 75).
This can be stated as follows:
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.
Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

Prediction 2 Species with large fruits will either contain one large seed or
larger numbers of smaller seeds per fruit. Species with small
fruits will always have one or a few small seeds per fruit (Figure
I).

RELATIONSHIP BETWEEN SEED SIZE AND SEED

NUMBER

An individual plant would achieve its maximum fitness by producing an

infinitely large number of big seeds. The availability of photosynthate, miner­
als, protein, and water limits the resources that can be devoted to reproduc­
tion. The total amount of resources available for reproduction is used in the
production of a certain number of seeds. Increases in seed size may come at
the expense of producing fewer seeds per fruit and fewer fruits per plant (61,
65, 72).

Prediction 3 An evolutionary trade-off exists between seed size and seed

number (Figure 1).

Seed size and subsequent seedling size are important determinants of

success in seedling establishment (39, 53). As a general rule, larger seeds will
have larger cotyledons and nutritional reserves (23, 44). Comparisons of
habitats have shown that plant species of stable, shady, and moist habitats
generally have larger seeds than plant species of sunny, disturbed, and dry
habitats (3, 72). Also, species groupings based on growth form, i.e. herb,
shrub, and tree, show increased average seed weight with plant size in tropical
floras (24, 70). Analysis of covariance using 203 woody species from a
Peruvian forest show that species that establish in canopy gaps tend to have
smaller seeds than species that establish in the shade, though tree height,
dispersal syndrome, and growth form also influence seed size (24). The
importance of seed size in determining establishment success is suggested by
its lower phenotypic and genetic variability in comparison to many other plant
characters less closely associated with fitness (39, 64).
 FLOWERS, FRUITS, AND SEEDS 413

Prediction 4 Species with small seeds tend to occupy habitats that are more
sunny, dry, and disturbed than are those of species with large
seeds (Figure I).

RELATIONSHIP BETWEEN FLOWERS AND MODE OF

POLLINA TlON

Each species of plant possesses a syndrome of characteristics that are adapted

to a particular mode of pollination, such as by birds, moths, bees, flies, bats,
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

butterflies, or wind. These syndromes have been extensively described (21),

Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

and they explain variation both across communities and within individual
plant families such as the Polemoniaceae (35) and Rhizophoraceae (85).
Descriptions of syndromes emphasize flower color, odor, shape, and timing
of anthesis. Clear differences in flower size also exist among these pollination
syndromes. In general, the flowers visited by large, energy-demanding ver­
tebrates, such as bats and birds, tend to be large and produce copious amounts
of food reward in the form of either nectar or pollen. Flowers visited by small
insects such as flies and small bees tend to be small in size and produce only
small amounts of food reward. Wind-pollinated species typically have very
small flowers in which elements of the perianth have been reduced or even
lost during evolution. These are general tendencies, though exceptions do
occur. Species that have small flowers grouped into dense inflorescences that
mimic large flowers, such as many members of the Compositae, are obvious
exceptions to the generalization. Even within categories of pollinators, varia­
tion in pollinator size is correlated with flower size. For example, bumblebee
species with longer tongues tend to visit larger flowers than do species with
shorter tongues (38). Large species of Australian honeyeaters tend to forage
more often than do smaller honeyeater species on plant species with large
flowers containing abundant nectar (58).

Prediction 5 Species with large flowers tend to be visited by larger pollina­

tors than those that visit species with small flowers (Figure 1).

RELA TlONSHIPS BETWEEN FRUITS AND DISPERSAL

SYNDROME

Fruit size, shape, and color are related to the type of dispersal syndrome (26 ,

45). At the extremes, fruits that are wind-dispersed tend to be small and
light-weight (2), while fruits eaten and dispersed by vertebrate animals tend to
be much larger. Within the category of animal dispersal, fruit characteristics
of color, nutrient content, odor, size, and general morphology have been
related to the specific type of vertebrate disperser (60, 75). Detailed animal!
 414 PRIMACK

fruit surveys from rainforest communities show that fruit characteristics tend
to match the size, visual abilities, and gape width of the vertebrates (45, 87).
In general, large, husked fruits are often dispersed by mammals, mainly
monkeys and bats, while birds tend to disperse smaller fruits that are not
protected by a husk. Only large vertebrates are strong enough to pull apart the
spiny valves of large, husked fruit, such as durians (Durio) and to tear apart
the rind in huge leathery fruits such as jack-fruit and chempedak (Artocarpus)
(12, 63).
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

Prediction 6 Vertebrate-dispersed fruit types tend to be larger than wind­

Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

dispersed fruits. Species with large, fleshy fruits tend to be

dispersed by vertebrate dispersers larger than those that disperse
species with smaller, fleshy fruits (Figure I).

TIMING OF FLOWERING, FRUITING, AND SEED

GERMINATION

The selective forces that affect flowering phenology have been debated
extensively in the literature (6, 62, 67). Particularly controversial has been the
idea that the flowering phenology of individual species is linked to the
abundance of pollinators and competition for those pollinators with other
plant species (69). Several examples in which the succcssive flowering of
species appears to be controlled by competition for pollinators have been
shown to be indistinguishable from random patterns (62, 66, 67). Also, the
fruit set of many temperate woodland herbs is not limited by pollinator
visitation rates (54), suggesting that pollinator activity is not the factor
determining flowering phenology at the present time.
The timing of fruit maturation often appears to be closely linked to syn­
dromes of dispersal. For example, the maturation of fleshy fruited plants in
the temperate zone appears to be closely linked to the seasonal abundance and
nutritional requirements of bird dispersers (83). In southern Spain, species
with watery fruits commonly mature in the hot summer when birds require
more water, while species with high-lipid fruits typically mature in the winter
when the caloric requirements of the birds may be high and the relative
availability of food is low (41). These correlations readily suggest that species
with bird-dispersed fruits have come under selection to match their time of
ripening to the activities of the birds. Physiological factors, such as water
availability, have often been cited as influencing the fruiting patterns of
tropical plants, particularly in seasonally dry habitats (6, 62).
Such explanations for flowering and fruiting times may be incomplete as a
result of a tendency to focus on one phase of reproduction in isolation. The
major factor determining flowering time in the temperate zone may simply be
 FLOWERS, FRUITS, AND SEEDS 415

the size of the fruit. In the flora of Britain, large-fruited species flower
(62).
significantly earlier in the growing season than do small-fruited species
The fall-flowering witch hazel (Hamamelis virginiana), which has large
fruits, at first appears to be an exception to this generalization. However, this
species does not mature its fruits until the following year, and it appears to be
derived from spring-flowering ancestors under strong selection to flower even
earlier. In the flora of South Florida, large-fruited species take longer to
mature their fruits than do small-fruited species (62). Fruit development in
seasonal climates must take place within a growing season defined by such
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

limitations as the periods free from killing frosts and from drought conditions.
Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

The differences among families in flowering sequence described by Kochmer

& Handel (49) need to be reanalyzed to determine if differences in fruit size
among these families can explain a significant proportion of the pattern. The
tendency of wind-pollinated trees to nower in the early spring, usually
explained by the lack of leaves on the trees (89), may also be partially
explained by the tendency of these species to have large fruits that require
many months to mature. The small fruit and seed size of many weeds is
probably related to the fact that their unpredictable life history places a high
premium on rapid fruit maturation as well as on high seed numbers. When
selective pressure does operate for particular fruiting times, as seems probable
for many temperate bird�dispersed fruits, then selection will indirectly act on
flowering time since these two characters are linked due to the more-or-less
fixed time required for fruit maturation.

Prediction 7 Species with large fruits will require a greater period of time for
fruit maturation than will species with smaller fruits. In the
temperate zones, large-fruited species will be forced to flower
early in the spring in order to have sufficient time for fruit
maturation before the onset of cold weather in the fall. Small­
fruited species may flower anytime during the growing season
but will tend to flower and fruit at the times when carbohydrate
levels in the plant are highest (Figure 1).

In species-rich tropical rain forests, the main factor determining flowering

and fruiting times may be the need of individuals within a population to
coordinate their reproduction in order to attract pollinators to effect cross­
pollination, attract fruit dispersers, and satiate seed predators (1, 46, 62).
Physiological mechanisms, such as drought-stress and the onset of the rainy
season, are the proximate mechanisms controlling the flowering response but
may not be the driving selective force themselves (l). It would be valuable to
know for a community of species if flower buds and immature fruits can
remain in an arrested state of development for many months once reproduc­
tion has begun. It seems reasonable that flower and fruit development should
 416 PRIMACK

proceed without interruption in order to minimize the time of exposure to

flower and seed predators. Such an uninterrupted reproductive development
would more closely link seed germination with the other phases of reproduc­
tion.

Prediction 8 Fruit development will normally be as rapid as possible to

minimize exposure to seed predators and to minimize metabolic
cost. This prediction could be tested either by comparing related
species which differ in their rate of fruit maturation or by
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

manipulating experimentally the rate of fruit maturation.

Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

The factors affecting flowering and fruiting times might also be extended to
include the best time for the seed to land on the ground. Factors of weather,
leaf litter, and pests in the soil might affect the viability of seeds in the soil
and determine the best time for seed dispersal in temperate species. As an
example of this complex linkage, the red maple (Acer rubrum) and the sugar
maple (A. saccharum) in New England could be cited (73a). The red maple
f10wcrs in April; the samaras then develop rapidly and are shed in June. The
samaras lack dormancy and germinate immediately, with several months
remaining in the growing season. The sugar maple flowers a few weeks later
in May when the danger of frost is past. Fruits are matured over a several­
month period and are finally shed in September. The sugar maple seeds have a
chilling requirement and do not germinate until the following spring. This
example illustrates the relationship among flowering and fruiting times and
germination requirement.
Variation in seed size, physiology, timing of germination, and morphology
among species allows each species to specialize in its germination require­
ments, and such variation hence contributes to the maintenance of species
diversity (37, 95). Among species in the British flora, larger seed size is
associated with lower rates of germination and lower relative growth rates of
the seedlings (36). Families differ in their rates of germination, and this
suggests that phylogenetic constraints are important (36). A tendency for
immediate germination and a corresponding lack of dormancy is typical of
many large-seeded tropical tree species which establish in shady conditions
(23, 57). Selection on the optimum time for seed germination might be
particularly important in many habitats, such as the seasonally dry forests in
the western Ghats of India. Large-seeded, indigenous species that lack any
seed dormancy, such as jack-fruit (Artocarpus heterophyllus). may be under
selection to disperse their fruits at the time of year most suitable for the
germination of these vulnerable seeds.

Prediction 9 The time of year that is best for seed germination will influence
fruiting and flowering times, particularly in species lacking
 FLOWERS, FRUITS, AND SEEDS 417

seed dormancy and in tropical habitats with a pronounced dry

season.

CONTINUITY AMONG FLOWERS, FRUITS, AND SEEDS

Anatomical Evidence

The vascular connections of the fruit to the stem are generally fully formed in
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

the ovary at the time of anthesis. Vascularization of individual flower parts

Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

can be related to the size of the parts as well as their subsequent development
or abscission during fruit development ( II). Following anthesis, fruit de­
velopment consists primarily of cell expansion rather than cell division in
most species (19). Obviously, a flower with a well-developed vascular system
in the ovary and the pedicel will be able to develop rapidly into a fruit with a
well-developed vascular system.
The anatomy of the flower and fruit are used to support various in­
terpretatio ns of the evolution of the angiosperm flower (11, 18, 19). Flower
parts may be viewed as modified leaves, some of which became specialized
for support (sepals), attraction of pollinators (petals), presentation of pollen
(stamens), and protection of the ovules (carpels). The leaflike qualities of
these structures, particularly in primitive angiosperms, as well as the vascula­
ture can be used to support this viewpoint. An alternative explanation is that
leaves as well as individual flower parts are derived from reduced branches.
Whichever explanation is correct, a close anato mical affinity exists between
these structures, in terms of vascularization and surface features, which
suggests but does not prove that the structures are homologous.
The flower, fruit, and seeds also have an anatomical and morphological
continuity in protective mechanisms. Resin canals afld 'latex ducts in the ovary
and ovules are laid down at the stage of flower formation, and they simply
enlarge during fruit development. Secondary compounds, such as alkaloids,
resins, and aromatic oils, which may serve to protect the flowers against
herbivores, will remain throughout the fruit maturation phase and often be
incorporated into the mature seed. These protective compounds may be
broken down or removed from the fleshy tissue of animal-dispersed fruits
during the final days of the ripening process. Floral glands and protective
hairs on the ovary wall or a persistent calyx will also remain on the developing
and even mature fruit (10). As another example, the inferior ovary has been
interpreted as an a da pta tion to protect the ovary from i nju ry during pollination
(33), Wrapping the ovary in additional layers of floral tissue may give the
developing fruit added protection and allow correlated evolutionary changes
to occur in the morphology of the final fruit. It would be in teresting to
determine if the evolution of the inferior ovary is associated with greater
 418 PRIMACK

ovary size at the time of anthesis, more ovules per flower, and more rapid
fruit development.

Genetic Evidence

Pleiotropic effects may link variation in reproductive characters (14, 31 , 34,

78). For example, the S gene in Nicotiana tabacum affects the size and shape
of the leaves, petals, and capsule (77). Such genes may operate through
effects on ce\1 size, possibly associated with changes in the number of
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

chromosomes. If the number of ce\1s per part is constant, then increases in

Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

chromosome number and cell size may result in larger flowers, fruits, and
seeds (48). Another mode of action is for increases in the number of cells in
all flower parts as a result of increased numbers of cell divisions in the flower
primordium. A doubling of the cell number in the flower primordium will
result in larger perianth and androecium parts as well as a larger ovary, which
would develop into a larger fruit with larger seeds. Larger numbers of cells in
the ovary might also allow greater numbers of ovules to be formed. Through
such pleiotropic effects, selection for either increased or decreased size at one
phase in reproduction would have corresponding effect� on the other phases of
reproduction. The effects of pleiotropy may be particularly strong among
reproductive characters, since, as mentioned earlier, all floral parts are prob­
ably homologous. Similarly, the major constituent of many seeds is the
cotyledons, which are modified leaves. Consequently, many of the genes
controlling the development of these structures must be the same.

Physiological Evidence

The same essential physiological factors affect the three phases of reproduc­
tion. Certain of these factors are external, such as temperature, water, and
light. Internal factors, which may be regulated by the external factors, include
carbohydrate and nitrogen levels, and plant hormones. The same basic plant
hormones play a dominant role in regulating plant reproduction and indicate a
physiological continuity in reproduction (50). Auxin, gibberellin, cytokinin,
and ethylene have all been implicated in the initiation of flowering and flower
development (50). Auxin produced by germinating pollen grains stimulates
early fruit development. Auxin, cytokinin, and gibberellin produced by the
developing ovules further stimulate fruit development following fertilization.
Fruit maturation is induced by the ethylene, with the other hormones probably
also playing significant roles. Seed germination is stimulated by the pro­
duction of gibberellin and cytokinin and is inhibited by the production of
abscissic acid. Early seedling leaf development and phototropic responses
are regulated by auxin and gibberellic acid. At each successive stage in
development, these same hormones interact with each other as well as in-
 FLOWERS, FRUITS, AND SEEDS 419

ternal levels of carbohydrates in complex ways that are often specific to each
plant species.
The fruit wall may determine many aspects of seed germination behavior
(42). The fruit wall may split open at maturity in capsules, pods, and follicles;
it may remain as a hard protective layer in nuts, caryopses, and achenes; or it
may envelop the seeds in a soft nutritious flesh in berries and pomes. The
remnants of fruit structures that remain associated with the seed will de­
termine the immediate physical and chemical environment of the seed. At one
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

extreme, seeds released from a capsule are free of the fruit tissue, while seeds
Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

enclosed in nuts and achenes are still completely surrounded by maternal

tissue which can have a dominant effect on germination behavior. The rate at
which this fruit wall deteriorates and releases the seed will play a dominant
role in determining the timing of seed germination. In many weedy and
herbaceous species, the green tissues of the sepals and ovary walls that remain
around the seed even after dispersal determine the light requirements for
germination of the developing seeds (15). Species with a high chlorophyll
content in the tissues surrounding the seeds typically cannot germinate in the
dark, while species with a low chlorophyll content in these tissues are capable
of germination in the dark. The seed coat itself, which is so important in seed
germination, has its origins in the integuments of the ovule, which is floral
tissue (86). So, the tissues of the flower, fruit, and seed are often de­
velopmentally and physiologically integrated in controlling germination.
Resources are necessary for construction of flowers and production of
nectar for pollination, construction of fruit tissues to aid in seed dispersal, and
provisioning the seed for germination and establishment in a new habitat.
Each of these three reproductive phases places demands on the resource
budget of the adult plant. The resources may be measured in terms of such
constituents as carbohydrates, minerals, proteins, and water (30), though
calories are the most universal unit and probably the most appropriate (9). The
usual method of measuring reproductive cost by weighing individual parts
typically overestimates reproductive costs, since the flower and the develop­
ing fruit are photosynthetic and produce a substantial proportion of their own
calories (8). In any case, each phase in plant reproduction does have some
cost, which places limits on reproduction. A plant must devote resources to
growth and maintenance, so only a limited amount of resources remains
available for reproduction. A trade-off between flowering, fruiting, and
seedling establishment may occur, leading to the most efficient and maximal
overall reproduction. If too much of the resources are used in producing
numerous, large flowers, then not enough resources will be available to
mature the developing fruit (79). Alternatively, if only a few, small flowers
are produced, a plant will not be able to mature many fruits and the resources
available for reproduction will go unused. A key factor may be the relative
maternal investment in the ovary at the time of anthesis (52, 92).
 420 PRIMACK

Within the fruit itself, there may be a trade-off between the resources
devoted to seed formation and resources devoted to dispersal processes (76a).
At one extreme are capsules and achenes in which only a small proportion of
the fruit is devoted to dispersal, while in other fruits, such as berries and
pomes, most of the fruit is devoted to dispersal, with only a small seed
volume. Natural selection should act on the relative proportions of resources
devoted to flowers, fruit tissues, and seeds, depending on the relative advan­
tages to the plant of high levels of outcrossing, dispersal of the seeds to new
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

sites far from the parent plant, and the production of numerous, large seeds.
Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

EMPIRICAL EVIDENCE FROM SIX LARGE GENERA

Selection for optimal characteristics within an evolutionary grouping may be

constrained by phylogenetic factors, such as shared morphology, anatomy,
and life history characteristics that limit the potential for natural selection and
speciation (22, 32). Phylogenetic constraints affect such angiosperm charac­
teristics as flowering time (49) and leafing phenology (51). Studies of changes
in size and shape using statistical techniques such as allometry have been
useful in determining constraints on animal evolution (40, 59) and the process
of speciation in plant species (37a). These statistical techniques can be applied
to variation in reproductive characters among plant species within genera. Six
large genera were examined in detail to determine evolutionary correlations
among the sizes of flowers, fruits, and seeds (R. B. Primack, unpublished).
Allometric analysis was used to determine how natural selection operated on
correlated plant characters in order to determine whether there are any de­
monstrable limitations on evolutionary changes.
The raw data for this study were extracted from published monographs (R.
B. Primack, unpublished). The genera selected were all relatively large and
included a mixture of tropical and temperate genera as well as, woody and
herbaceous genera. For each species, data were extracted on the length of
flower parts, fruits, and seeds as well as other reproductive and vegetative
characters. When a petal length for a species was listed as "3 to 7 mm in
length," then 5 mm was recorded. Unusual extreme values were not included
in the recorded values.
Spearman Rank Correlation was used to examine pairs of variables. Analy­
sis was used to determine if the relationship between pairs of variables was
significantly allometric (59). The developmentally earlier stage was always
considered the independent variable. For example, petal length would be
considered as the independent variable in a relationship with fruit length.
Each species is treated as a separate data point in this analysis. Such an
assumption may lead to an overestimate of the sample size if the evolutionary
changes in the studied characters were fewer than the number of species (40).
 FLOWERS, FRUITS, AND SEEDS 421

Only a detailed numerical taxonomy study of each genus could possibly

reveal such information. For our present purposes, highly significant rela­
tionships between pairs of characters repeated in many genera are accepted as
valid evidence.

Genera Examined
I. Calophyllum (Guttiferae) is a genus of about 175 tree and shrub species,
mainly found in the lowland rain forests of Southeast Asia (80), character­
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

ized by relatively small white flowers with numerous stamens and an

Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

olive-like drupe with a fleshy rind covering a large stone.

2. Agave (Agavaceae) is a genus of 300 species of woody herbs, known as
"century plants," distributed throughout arid and often tropical regions of
the Americas (27). Plants have a basal rosette of large, stiff, spiny leaves,
tall inflorescences of funnelform flowers, and an oblong capsule contain­
ing numerous flattened seeds.
3. Astragalus (Leguminosae) is a cosmopolitan genus of about 2000 annual
and perennial species, known as "locoweeds," found usually in open xeric
habitats (4). Species have pinnately compound leaves, flowers in spikes or
racemes, and a legume pod containing ovoid seeds.
4. Lesquerella (Cruciferae) is a genus of about 80 primarily herbaceous,
annual, biennial, and perennial species, known as "cloth of gold," occur­
ring mainly in open, dry habitats in the western United States and Mexico
(71). A basal rosette of leaves produces inflorescences of yellow flowers.
The globous, two-locular silique contains round seeds.
5. Phacelia (Hydrophyllaceae) is a genus of about 200 species of annual,
biennial, and perennial herbs from the Americas known as "scorpion
weeds" (56). The stem has simple or pinnate leaves and campanulate
flowers in scorpiod cymes, followed by ovoid capsules.
6. Asclepias (Asclepiadaceae) is a large genus of perennial herbs from the
Americas and Africa, known as "milkweeds" because of the white latex
(94). The complex flowers are produced in umbels, followed by follicles
which split at maturity to release the numerous, flattened seeds, each
containing a tuft of hair to aid in wind-dispersal. The milkweeds are
particularly interesting here because they have been used extensively in
studies of sexual selection in plants (93).

Variation Among Species Within a Genus

A significant positive correlation exists between petal length and fruit length
among species within each of the six genera (Table I), and the correlation is
very·highly significant in four of the six genera (R. B. Primack, unpublished).
The correlation coefficients range from a high of 0.59 in Agave to a low of
0.28 in Lesquerella. The high values for several of the genera show that
 422 PRIMACK

PHACElIA
10
. / ,.
E P. mitior
�

./
:J::
I-
"
Z
'"
..J
5
'"
..J
�
rn y = 0.33 X' I. 9 I
0-
""
0
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

0
0 15 30
Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

PET AL LENGTH (mm)

Figure 2 Relationship between petal length and capsule length among species in the genus
Phacelia. Each dot represents a species.

evolutionary changes in flower size are closely associated with changes in

fruit size.For example, species of Phacelia with small flowers have smaller
frui ts than species with larger flowers (Figure 2).
A highly significant positive relationship exists between fruit size and seed
size in four of the six genera (Table 1). The very high correlation coefficient
in Calophyllum is not surprising considering that the single seed forms a large
proportion of the drupe volume . However, there is also a high correlation
coefficient for the genus Agave, in which there are many seeds per capsule
(Figure 3). Seed size and fruit size have a nonsignificant, negative correlation
in the genera Asclepias and Phacelia. In the genus Asclepias, the main
constituent of the follicle is the hairs attached to the seeds, and hair l ength is
correlated with fruit length (R. B. Primack, unpublished).
Regression analysis was undertaken to determine if these relat ionshi ps are
si mp le linear relationships or if changes in relative sizes have occurred during
speciation (R. B. Primack, unpublished). In all genera except Agave, there is

AGAVE

�� ;/
IZ

/
Y = 0.11 x ' I' '
E

.!: : �
.....
:I: : .

.'
l- :
e> '
z 6 : .
w
'. . .. /:::.
"
..J
o
.
"
/."
.
.

W
W ./ ..: .
'"

o �I __----�I ----------�I----

o 40 BO

.
CAPSULE LENGTH (mm)

Figure 3 Relationship between capsule length and seed length among species of Agave. Each
dot represents a spec ies
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.
Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

Table 1 Spearman Rank correlations (r) and allometric values (b) between pairs of characters among species within a genus.··b

Leaf Length with:

Petal Length Fruit Length Petal Fruit Seed
with Fruit Length with Seed Length Height Length Length Length
Genus NC r b r b r r r r

Calophyllum 1 27
64
0. 56**-
0.59***
0 . 70*- 0.94***
0.79***
0 . 9 1 **
0.62***
0 . 29***
0.66***
0.45*** 0.40*** 0. 39*** 5
Agave 0.97 0. 33*** 0. 33 *** 0.43*** �
0. 49*** 0.62*** 0. 50***
tTl
Astragalus 236 0. 1 9**- 0. 37*** 0.29*** 0 . 21** �
Lesquerella 62 0.28* 0.45*-* 0 . 27* 0.26*** 0 . 5 1 *** -0.08 0.17 -0.03
[n

Phacelia 67 0. 57**" 0. 51 *** - 0 . 14 -0.47**" 0. 33 ** 0.45***

81 0. 30** 0 . 24*-* - 0.06-*- 0. 30*-
0 . 08
�
Asclepias - 0 . 07 0 . 3 1 *** 0 . 20 0 . 25 * c:

aThe value b is reported from the equation y = ax", where x is the independent variable, y is the dependent variable, and a and b are constants. Each value a
of b was tested for significant differences from I.
>
"One, two, and three stars indicate significance at the 0.05, 0.01, and 0.001 levels, respectively. z
C For each genus, the minimum number of species (N) for each correlation is given. o
[n

�
en

.j::.
IV
W
 424 PRIMACK

a significant allometric relationship between petal length and fruit length with
a slope varying from 0.24 to 0.70 (Table 1). This result suggests that
evolutionary changes in fruit length approach a maximum value as flower size
continues to increase over evolutionary time. In the genera Calophyllum,
Agave, Astra!?aius, and Lesquerella, the relationships between fruit length
and seed length are significantly al10metric with a slope less than 1, suggest­
ing that evolutionary changes in seed length approach a maximum value as
fruit size continues to increase over evolutionary time (Table 1). These
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

allometric analyses demonstrate that changes in the relative size of flowers,

Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

fruits, and seeds are associated with speciation even though these characters
arc closely correlated.

Prediction 10 The evolutionary relationships of flowers, fruits, and seeds

among species within a genus are allometric, presumably as a
result of greater constraints on one character than another.

RELATIONSHIPS OF REPRODUCTIVE AND

VEGETATIVE CHARACTERS

Plant architecture examines the physiological, morphological, and biome­

chanical factors influencing plant form and function (29, 84). The emphasis
of this field has been almost entirely on the vegetative morphology of the
plant. Yet, plant reproduction is an important component in the life history of
the plant that must also have an effect on plant architecture. At the least,
flowers must develop at positions where they can make a suitable display and
be located by pollinators. Further, these positions must be appropriate for the
type of fruit dispersal found in that species. The net result is that the types of
pollination and fruit dispersal must be related as a result of position effects on
the plant.
The characteristics of the flowers, fruits, and seeds must also be closely
coordinated with the vegetative characters of the plant. Studies mentioned
previously have shown that tree species tend to produce larger seeds than herb
species (3, 70). As Corner (13) has described in one of his imaginative
articles: "A big natural genus has massively constructed species with big
buds, thick twigs, large leaves and large flowers, fruits, and seeds. It has,
also, the slender which are small in all their primary parts." This idea was
tested using the 6 genera mentioned previously (R. B. Primack, unpublished).
Within each genus, the correlation of plant height with leaf length is very
highly significant. Tall species have longer leaves than short species (Table
1). Vegetatively larger species also have bigger flowers and seeds in five of
the six genera (Table 1). Only in the genus Lesquerella is this pattern not
seen. In the genus Lesquerella, considerable variation in growth habits and
 FLOWERS, FRUlTS, AND SEEDS 425

breeding systems occurs (71) which apparently obscures the trends observed
in the other genera. In three of the six genera, leaf length has a significant
positive correlation with fruit length.
These relationships of vegetative and reproductive characters have a
mechanical as well as a natural history basis. Thick twigs are necessary to
support large flowers and heavy fruits without being bent or broken. In
addition, large flowers that are visited by large climbing and perching ver­
tebrate pollinators need thick twigs to support the weight of the animal. For
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

example, African species of Erica that are pollinated by perch ing birds have
Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

thicker twigs than related species pollinated by insects (74). A striking

contrast is also seen among bird-pollinated species of the same genus occur­
ring in the Americas and Australasia-for example, the genus Fuchsia.
American species, such as F. simplicicaulis, are pollinated by hovering
hummingbirds and have flowers produced outside of the foliage on slender
twigs. In ntrast, Australasian species, such as F. excorticata, are pollinated
co

by larger, perching honeyeaters and have flowers produced inside the foliage
layer on thick twigs. The thickness of the supporting twig can also be related
to the size of the fruit. In particular, large fruits must be produced on thick
twigs in order to support the weight of the fruit. The twig must also be able to
support the weight of the disperser, in the case of vertebrate-dispersed fruits.
For example, in the genus Artocarpus in Borneo, A. kemando has relatively
small fruits (4 x 2.5 cm) produced at the ends of slender twigs (2-3 mm
thick) and is dispersed by birds. A. odoratissimus has medium-size fruits (10
x 8 cm) produced at the end of thick twigs (4-10 mm thick) and A. integer
,

has large fruits (25 x 12 cm), produced on the trunk and main branches, and
is dispersed by large mammals (12, 63).
In addition, specie s with large flowers, fruits, and seeds tend to have large
leaves. Several potential reasons for this could be advanced. First, the genes
controlling size may affect all of these structures, i.e. pleiotropy. Second,
selection for large fruit size might require simultaneous selection on leaf size
in order to support the fruit photosynthetically. And third, selection for
thicker twigs to support bigger fruits might result in indirect selection on leaf
size; thicker twigs are typically not as branched, so that the fewer leaves
would have to be larger in order to maintain the same photosynthetic area.
Consequently, fruit and leaf characters might be linked indirectly via twig
thickness.

Prediction 11 Vegetative characters, such as leaf size, twig thickness, and

plant height, in addition to being positively correlated among
species within a genus, are also evolutionarily linked with
reproductive characters through a network of positive correla­
tions (Figure I).
 426 PRIMACK

BREEDING SYSTEMS AND FRUIT SET

The intense interest in mating patterns and sexual selection in plants has led to
the observation that dioecious species have a tendency toward fleshy dispersal
units in the angiosperms (5) and the gymnosperms (28). In tum, breeding
systems, fruit set, life fonn, and fruit type have been shown using large data
sets to be interrelated (8 1, 82). These studies demonstrate that fruit set is
higher for species having the characteristics of self-compatibility, dioecy,
annual habit, temperate distribution, and "inexpensive" fruits (i.e. capsules,
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

follicles, achenes), in comparison with species that display self-incompatibili­

Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

ty, hermaphroditism, perennial habit, tropical range, and "expensive" fruits

(i.e. fleshy fruits, nuts) (81). In general, "expensive" fruits appear to be larger
and heavier than "inexpensive" fruits. The correlations of breeding system
with many plant characteristics suggest that the flower, fruit, and seed
relationships presented already might also be correlated with breeding sys­
tems and fruit set. Such correlations are indicated by the well-known tendency
of self-compatible species and populations to have smaller flowers than
related self-incompatible species and popUlations (73). These characters are
probably further linked with seed set, the proportions of ovules that develop
into seeds(7). The low fruit set in certain species with small flowers and large
fruits, such as mangoes (Mangifera), could come about as a consequence of
violating Prediction 1, that species with small flowers tend to have small
fruits.

Prediction 12 Self-fertilizing species with high fruit set will be found to have
smaller flowers, fruits, and seeds, in comparison with out­
crossing species with low fruit set (Figure 1). Probably the
most critical determinant of fruit set is the relative dry weight
of the ovary in relation to the dry weight of the fruit. Where
the ovary-to-fruit ratio approaches
1, the average fruit set will
approach 100% and fruit maturation will be rapid. Where the
ovary-to-fruit ratio approaches 0, the average fruit set will
approach 0% and fruit maturation will be slow. Absolute
ovary size and fruit size should also be considered as variables
in a multivariate test of this prediction.

CONCLUSIONS

This study has presented arguments for close ecological and evolutionary
relationships among pollination, seed dispersal, and seedling establishment.
These arguments are supported by empirical data from six large genera
showing patterns of positive correlations among flowers, fruits, and seeds
among species. The only exceptions to this pattern are negative correlations
 FLOWERS, FRUITS , AND SEEDS 427

between seed size and seed number in some groups. Vegetative characters are
also frequently linked to this variation in reproductive characters . The evolu­
tionary implications of these arguments and results are obvious. First , it
should be apparent that systematic studies of plant reproductive characteristics
should not be too narrowly focused. For example, studies of floral evolution
and pollination ecology within a taxonomically variable group, such as the
Polemoniaceae (35) and the Rhizophoraceae (85), should include parallel
studies of fruit, seed, and vegetative evolution to determine the totality of the
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

evolutionary change. Second, community surveys of plant reproductive ecol­

Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

ogy and phenology should not examine only one aspect of reproduction if the
three phases of reproduction are linked. For example, a survey of fruit types
and dispersal syndromes at the community level in conjunction with studies of
pollination ecology and seed establishment ecology may provide additional
insights into the evolution of fruit characters and dispersal syndromes. It is
quite possible that selective pressures for fruit size and type of dispersal are
largely determined by selection on flower size and seed size. And third, the
approach of determining the linear and allometric relationships among flow­
ers , fruits, and seeds allows predictions to be made of limits to natural
selection. These predictions can provide a valuable starting point in further
field studies and in investigations of plant evolution. Particularly valuable
areas of further investigation would be to examine simultaneous evolutionary
trends for flowers , fruits, and seeds among genera within large families and to
examine parallel evolutionary trends within two or more large , closely related
genera. The exceptions to these general trends and predictions might give
significant insights into the process of speciation and the formation of higher
taxonomic categories.

ACKNOWLEDGMENTS

The most sincere appreciation is expressed to Kamal Bawa and David Lloyd,
who unselfishly guided me throughout this project. Valuable advice was also
given by Joseph Travis (on allometric analysis), Peter Stevens, Mary Willson,
Judy Dudley, Elizabeth Shaw, Reed Rollins, Peter Ashton, Pamela Hall,
Hyesoon Lee, and Mark Leighton. Paul Alloca, Sherry Booth , and Barbara
Dorritie assisted in the data analysis.

Literature Cited

I . Augspurger, C. K. 1 98 1 . Reproductive 3. Baker, H. G. 1972. Seed weight in rela­

synchrony of a tropical shrub: Ex­ tion to environmental conditions in Cali­
perimental studies on effects of pollina­ fornia. Ecology 53:997- 1 0 1 0
tors and seed predators on Hybanthus 4 . Barneby, R. C. 1 964. Atlas of North
prunifolius (Violaceae). Ecology 62: American Astragalus. Mem. N. Y. Bot.
775-88 Gard. 1 3 : 1-J l 88
2. Augspurger. C. K. 1 986. Morphology 5. Bawa. K. S. 1980. Evolution of dioecy
and dispersal potential of wind-dispersed in flowering plants . Ann. Rev. EcoI.
diaspores of neotropical trees. Am. 1. Syst. I I : 1 5-39
Bot. 73:353-63 6. Bawa, K. S. 1983. Patterns of flowering
 428 PRIMACK

i n tro pica l plants. See Ref. 47, pp. 394- forest trees: A review and synthesis.
410 Bot. Rev. 52:260-99
7 . Bawa, K . S . , Webb, C . 1 . 1 984. Flow­ 24. Foster, S. A . , Janson, C. H . 1 98 5 . The
er, fru it and seed abortion in tropical relationship between seed size and es­
forest trees: Impl ications for the evolu­ t ab l is hment conditions in tropica l woody
t ion of paternal and maternal reproduc­ plants. Ecology 66:773-80
tive patterns. A m . 1. Bot. 7 1 :736-5 1 25. Galen , C . Plowright, R. C . , Th om son ,
.

8. Bazzaz, F. A . , Carl son, R. W . , Harper, 1. D. 1 985. Floral biology and regula­

J . L. 1 979. Contribution to reproductive tion of seed set and seed size in the lily.
effon by photosynthesis of flowers and Clilltollia borealis . Am. 1 . Bot.
fruits. Nature 279:554-55 72: 1 544-52
9. Bazzaz, F. A . , Reekie, E. G. 1 98 5 . The 26. Gautier-Hion, A . , Duplantier, J . -M . ,
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

meaning and measurement of reproduc­ Quris, R . Fear, F . , Sourd : c . , et al .

.

tive effort in plants. In Studies on Plant 1 98 5 . Fnlit characters as a basis of fruit

Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

Demographv: A Festschrift for John L. choice and seed dis pers al in a tropical
Harper. ed. J. White, pp. 373-87. Lon­ forest vertebrate community. Gecolagia
don: Academic Press 65:324-37
1 0 . Berg , R. Y. 1 985. Gynoecium and de­ 27 . Gentry, H. S. 1 982. Agaves of Con­
velopment of embryo sac, endosperm, tinemal North America. Tucson: Univ.
and seed in Phofistoma (H ydroph yl­ Ariz. Press
l aceae) relative to taxonomy. Am. 1. 28. Givinish, T. 1 980. Ecological con­
Bot. 72: 1 775-87 straints on the evolution of breedin g sys­
I I . Carlquist, S. 1 969. Toward acceptable tems in seed plants: Dioecy and dispersal
evolutionary interpretations of floral in gymnosperms. Evolution 34:959-
anatomy. Phytomorphology 1 9 :332-62 72
12. Corner , E. J. H . 1 95 2 . Wayside Trees of 29. Givinish , T. , ed . 1 986. 0n the Economv
Malaya . Singapore: The Government of Plam Form and Functioll . Cam­
Printer bridge: Cambridge Univ. Press
13. Comer, E. J. H. 1 967. On thinking big. 30. Goldman, D. A . , Willson, M . F. 1 986.
Phvtomorphologv 1 7:24-28 Sex allocation in functionally hermaph­
1 4. Coyne J. A . Lande, R. 1 98 5 . The ge­
, . roditic plants: A review and critique.
netic basis of species differences in Bot. Rev. 52: 1 57-94
plants . Am. Nat. 1 26: 1 4 1-45 3 1 . Gottlieb, L. D. 1 984. Genet ics and mor­
15. C resswel l , J. E . , Grime, J. P. 1 98 1 . phologic al evolution in plants. A m . Nat.
Induction of a light requirement during 1 23:68 1 -709
seed development and its ecological con­ 32. Gould, S. J . , Lewontin, R. C. 1 979.
sequences. Nature 29 1 :583-85 The spandrels of San Marco and the Pan­
16. Cronquist, A. 1 968 . The Evolutioll alld glossian paradigm: A critique of the
Classification ofFlowerillg Plams. New adaptationist programme. Proc. R. Soc.
York: Houghton Mifflin LOlld. B. Bioi. Sci. 205:58 1-98
17. Cruden, R. W . , Lyon, D. L. 1 985. Cor­ 33. Grant, V. 1 950. The protection of
relations among stigma depth , st y l e ovules in flowering plants. Evolution
length, and pollen grain size: Do they 4 : 1 79-201
reflect function or phylogeny? Bot. Gaz. 34. Grant, V. 1 97 5 . Gelletics of Flowerillg
1 46: 1 43-49 Plants. New York: Columbia Univ.
18. Doyle, J. A . , Donoghue, M. J. 1 986. Press
Seed plant phylogeny and the origin of 35. Grant, V . , Grant, K. 1 965 . Pollillation
the angiosperms: An experimental ill the Phlox Familv. New York: Colum­
cladistic approach . Bot. Rev. 52:3 2 1 - bia Univ. Press .
43 1 36. Grime, 1. P . , Mason, G . , Curtis, A . V . ,
19. Esau, K. 1 965 . Plam Allatomv. . New Rodman , J . , Band, S . R . , et a t . 1 98 1 . A
York: Wiley comparative study of germination char­
20. Estrada, A . , Fleming, T. H . , eds. 1 986. acteristics in a local flora. 1. Ecol.
Frugivores alld Seed Dispersal. Dor­ 69: 1 0 1 7-59
drecht: Junk 37. Grubb, P. J. 1 977. The maintenance of
21. Faegri, K. , Pijl , L. van der. 1 979. The species-richness in plant com munitie s:
Principles of Pollina tion Ecology. Ox­ The imponance of the regeneration
ford: Pergamon niche. BioI. Rev. 52: 1 07-45
22. Felsenstein, J. 1 985. Phylogenies and 37a. Guerrant, E. O. 1 982. Neotenic evolu­
the com parat ive method. Am. Nat. tion of Delphinium lIudicaule (Ranuncu­
1 25 : 1 - 1 5 laceae): A hummingbird-pollinated lark­
23. Foster, S . A , 1 986, On the adaptive spur. Evolution 36: 699- 7 1 2
value of large seeds for tropical moist 38. Harder, L. D. 1 985. Morphology as a
 FLOWERS, FRUITS, AND SEEDS 429

predictor of flower choice by bumble sources limit the fecundity of three

bees. Ecology 66: 1 98-2 1 0 woodland herbs. Oecologia. 7 1 :43 1 -
3 9 . Harper, J . L. 1 977 . Population Biology 35
of Plants. New York: Academic Press 55. McKey , D . 1975. The ecology of
40. Harvey, P. H . , Mace, G. M. 1 982. coevolved seed dispersal systems. In
Comparisons between taxa and adaptive Coevolution of Animals and Plants. ed.
trends: Problems of methodology . In L. E . Gilbert, P. H. Raven, pp. 1 59-9 1 .
Current Problems in Sociobiology, ed. Austin: Univ. Texas Press
King's College Sociobiology Group, pp. 56. Munz, P. A . 1 968. A California Flora
343-61 . Cambridge: Cambridge Univ. and Supplement. Berkeley: Univ. Calif.
Press Press
4 1 . Herrera, C. M. 1 982. Seasonal variation 57. Ng, F. S . P . 1 978. Strategies of es­
in the quality of fruits and diffuse tablishment in Malayan forest trees. In
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

coevolution between plants and avian Tropical Trees as Living Syslems, ed. P .
Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

dispersers . Ecology 63:773-85 B . Tomlinson, M . H . Zimmermann, pp.

42. Heydecker, W . , ed. 1 972. Seed Ecolo­ 1 29-62. New York: Cambridge Univ.
gy. University Park, PA: Penn. State Press
Univ. Press 5 8 . Paton, D. C. 1986. Honeyeaters and
43. Howe, H. F. , Smallwood, J. 1 982. their plants in south-eastern Australia. In
Ecology of seed dispersal. Ann . Rev. The Dynamic Partnership: Birds and
/:,'col. Syst. 1 3 : 20 1 -28 Plants in Southern A ustralia. ed. H . A .
44 . Howe , H . F . , Richter, W. M. 1982. Ford, D . C . Paton, p p . 9- \9. Adelaide:
Effects of seed size o n seedling size in Gov!. Printer
Virola surinamensis: A within and be­ 59. Peters, R. H. 1 983. The Ecological Im­
tween tree analysis. Oecologia (Berlin) plications of Body Size. Cambridge:
5 3:347-5 1 Cambridge Univ. Press
45 . Janson, C. H. 1 983. Adaptation of fruit 60. Pijl, L. van dec. 1 983. PrinCiples of Dis­
morphology to dispersal agents in a persal in Higher Plants. Berlin: Spring­
neotropical forest. Science 2 19 : 1 87-89 er-Verlag. 3rd ed.
46. Janzen, D. H. 1 967. Synchronization of 6 1 . Primack, R. B . 1 979. Regulation of seed
sexual reproduction of trees within the yield i n Plantago. J. Ecol. 66:835-47
dry season in Central America. Evolu­ 62. Primack, R. B . 1 98 5 . Patterns of flower­
tion 2 1 :620-37 ing phenology in communities, pop­
47. Jones, C. E . , Little, R. J. 1 983. Hand­ ulations individuals, and single flowers.
book of Experimental Pollination Biolo­ In Population Structure of Vegetation .
gy. New York: Van Nostrand ed. J. White, pp. 571-93. Dordrecht:
48 . Kenton, A. Y . , Rudall, P. J . , Johnson, Junk
A. R . 1 986. Genome size variation in 63 . Primack, R . B . 1 985. Comparative stud­
Sisyrinchium L. (lridaceae) and its ies of fruits in wild and cultivated trees
relationship to phenotype and habitat. of chempedak (Artocarpus integer) and
Bot. Gaz. 1 47:342-54 terap (Arlocarpus odoratissinus) in Sara­
49. Kochmer, J. P . , Handel, S . N . 1 986. wak, East Malaysia with additional in­
Constraints and competition in the formation on the reproductive biology of
evolution of flowering phenology, Ecol. the Moraceae in Southeast As ia . Malay .
Mon. 56:303-25 Nat. J. 39: 1-39
50. Kramer, P. J . , Kozlowski, T. T. 1 979. 64. Primack , R . B . , Antonovics, J. 1 98 1 .
Physiology of Woody Plants. New York: Experimental ecological genetics i n
Academic Press Plantago . V . Components of seed yield
5 1 . Lechowicz, M. J. 1 984. Why do in the ribwort plantain Plantago lan­
temperate deciduous trees leaf out at dif­ ceolata L. Evolution 35 : 1 069-79
ferent times? Adaptations and ecology of 65 . Primack, R. B . , Antonovics, J. 1982.
forest communities. A m . Nat. 1 24 : 8 2 1 - Experimental ecological genetics in
42 Plantago. VII. Reproductive effort in
52. Lloyd, D. G. 1980. Sexual strategies in populations of P . lanceolata L . Evolu­
plants. 1. An hypothesis of serial adjust­ tion 36:742-52
ment of maternal investment during one 66. Rabinowitz, D . , Rapp, 1. K . , Sork, V .
reproductive session. New Phytol. L . , Rathcke, B . J . , Reese, G . A . , Wea­
86:69-79 ver, J. C. 1 98 1 . Phenological properties
5 3 . Marshall, D. L. 1 986. Effect of seed of wind- and insect-pollinated prairie
size on seedling success in three species plants. Ecology 62:49-56
of Sesbania (fabaceae). A m . J. Bot. 67. Rathcke, B . , Lacey , E. P. 1 985 . Pheno­
73:457-64 logical patterns of terrestrial plants. Ann.
54. McCall, c . , Primack, R. B . 1 987. Re- Rev. Ecol. Syst. 1 6: 1 79-2 1 4
 430 PRIMACK

68. Real , L . , ed. 1983. Pollination Biology. tiferae) . J. Arnold Arbor. Harv. Univ.
Orlando: Academic Press 2 : 1 1 7-699
69. Robertson, C. 1 895 . The philosophy of 81. Sutherland, S . 1986. Patterns of fruit­
flower seasons, and the phaenological set: What controls fruit-flower ratios in
relations of the entomophilous flora and plants? Evolution 40: 1 1 7-28
the anthophilous insect fauna. Am. Nat. 82. Sutherland, S . , Delph, L. F. 1 984_ On
29:97- 1 1 7 the importance of male fitness in plants:
70. Rockwood, L. L. 1985. Seed weight as Patterns of fruit-set. Ecology 65: 1 093-
a function of life form , elevation, and 1 1 04
life zone in Neotropical forests. Biotrop­ 83. Thompson, N_ J . , Willson , M_ F . 1979.
ica 1 7:32-39 Evolution of temperate fruit/bird in­
7 1 . Rollins, R. c . , Shaw, E. A. 1973. The teractions: Phenological strategies.
Genus Lesquerella (Cruciferae) in North Evolution 33 :973-82
Access provided by CONRICYT EBVC and Econ Trial on 09/23/15. For personal use only.

America. Cambridge: Harvard Univ. 84. Tomlinson, P. B . 1987 . Architecture of

Annu. Rev. Ecol. Syst. 1987.18:409-430. Downloaded from www.annualreviews.org

Press tropical plants. Ann. Rev. Ecol. Syst.

72. Salisbury, E. J. 1942. The Reproductive 1 8 : 1-21
Capacity of Plants. London: Bell 85. Tomlinson , P. B . , Primack, R.- B . ,
73. Schoen , D. J. 1977. Morphological, Bunt, J . 1 979. Preliminary observations
phenological , and pollen-distribution of floral morphology in mangrove Rhi­
evidence of autogamy and xenogamy in zophoraceae . Biotropica 1 1 :256-77
Gillia achilleifolia (Polemoniaceae). 86. Westoby, M . , Rice, B. 1982. Evolution
Syst. Bot. 2:280--86 of the seed plants and inclusive fitness of
73a. Schopmeyer, C. S. coordinator. 1974. plant tissues. Evolution 36:7 1 3-24
Seeds of Woody Plants in the United 87. Wheelwright, N. T. 1 98 5 . Fruit size,
States. Agric. Handbk. 450. Forest gape width, and the diets of fruit-eating
Service. US Dep. Agric. birds. Ecology 66:808- 1 8
74. Siegfried, W. R . , Rebela, A. G . , Prys­ 88. Wheelwright, N . T . , Orians, G . H _
Jones, R. P. 1985. Stem thickness of 1982. Seed dispersal b y animals: Con­
Erica plants in relation to avian pollina­ trasts with pollen dispersal , problems of
tion. Oikos 45:1 53-55 terminology, and constraints on evolu­
7 5 . Snow, D. W. 1 98 1 . Tropical tion_ A m _ Nat_ 1 1 9:403- 1 3
frugivorous birds and their food plants. 89_ Whitehead, D _ R _ 1969. Wind pollina­
Biotropica 1 3 : 1-14 tion in the angiosperms: Evolutionary
76. Spira, T. P . , Pollak, O . D . 1986. Com­ and environmental considerations_
parative reproductive biology of alpine Evolution 23:28-35
biennial and perennial gentians (Gen­ 90 , Willson, M, F_ 1 98 3 , Plant Reproduc­
tiana: Gentianaceae) in California. Am. tive Ecology_ New York: Wiley-Intersci .
J. Bot. 73:39--47 91. Willson, M _ F _ 1 986_ Avian frugivory
76a. Stapanian, M_ A_ 1 982_ Evolution of and seed dispersal in Eastern North
fruiting strategies among fleshy-fruited America_ Curr _ Ornithol. 3 :223-79
plant species of Eastern Kansas. Ecology 92_ Willson, M _ F _ , Burley , N_ 1 98 3 _ Mate
63: 1 422-3 1 Choice in Plants: Tactics, Mechanisms
77. Stebbins, G _ L. 1 959. Genes, chromo­ and Consequences . Princeton, NJ: Prin­
somes, and evo,lution. In Vistas in ceton Univ_ Press
Botany. ed. W. Turrill. London: Per- 93 . Willson, M. F" Price, P. W. 1980. Re­
gamon . source limitation of fruit and seed pro­
7 8 . Stebbins , G. L. 1974. Flowering Plants: duction in some Asclepias species. Can.
Evolution Above the Species Level. J, Bot. 58(2):229--3 3
Cambridge: Belknap 94. Woodson, R. E_ Jr. 1 954. The North
79_ Stephenson, A. G. 1 98 1 . Flower and American species of Asclepias L. Ann.
fruit abortion: Proximate causes and ul­ Mo. Bot. Gard. 4 1 : 1-2 1 1
timate function. Ann. Rev. Ecol. Svst_ - 95. Garwood. N . C . 1983. Seed germination
1 2: 253-79 in a seasonal forest in Panama: A com­
80. Stevens, P. F. 1 980. A revision of the munity study . Eco!.. Monogr. 53 : 1 59-
Old World species of Calophyllum (Gut- 81

View publication stats