T
hree-quarters of flowering plant species have hermaph-roditic flowers; however, the remaining species illustrate a tremendous variety of breeding systems. These include not only separate sexes (dioecy), and popu-lations with unisex individuals and hermaphrodites, but also sep-arate unisex and hermaphroditic flowers on the same individual. This diversity has fascinated evo-lutionary biologists for a long time. Darwin1noted that reduced investment in one reproductive function could be compensated for if it left additional resources for the other sexual function. This idea of a trade-off in resource use was elaborated in evolution-arily stable strategy (ESS) mod-els developed by Charlesworth and Charlesworth2, Charnov3andLloyd4. According to this theory, the ESS sex allocation depends on the shape of fitness gain functions relating male and female fertilities to the proportion of resources invested in male function (Box 1). One major dichotomy addressed by the theory is the distinction between dioecy and hermaphroditism. Although it was well understood that separation of sexes could be favored by high selfing and inbreeding depression, this model illustrated another selective mechanism that would also apply to outcrossing species. Accelerating fitness returns on investment of re-sources in a single sexual function would favor separate sexes, whereas diminishing returns on investment would favor hermaphroditism. The theory had a major impact on plant reproductive biology, because it could explain diversity of not only breeding systems but also resource allocation within a population of hermaphrodites.
However, testing the theory proved difficult because the parameters are notoriously hard to estimate. Until re-cently, most tests relied on qualitative comparisons of species. For example, selfing species were predicted to show lower investment in male function than outcrossing species5,6. Furthermore, sex allocation was often expected to depend on the size of an individual plant7.
Recent studies illustrate how to test the model experi-mentally by estimating the fitness gains themselves. Devel-opments in paternity analysis8now permit estimation of relative male fertility, and studies of seedling stages permit more accurate estimation of realized female fertility9. These advances are leading to the first attempts to quan-tify fitness gain functions. At the same time, quantitative genetics has been applied to testing the basic assumption of a trade-off in allocation. This article revisits the evidence for sex-allocation theory, as applied to outcrossing plants, now that these new approaches are available.
Estimating the parameters of fitness gain functions The first choice in describing fit-ness gain functions is deciding how to measure resource allo-cation (r). In the classic form of the model, rrefers to the proportion of resources invested by the plant in stamens rather than in pistil or seed maturation2,3. More recent elaborations separate allocation at the time of flowering from seed production, to allow for differ-ences in the timing of invest-ment10 and/or to include allo-cation to attractive structures, such as the corolla11. Furthermore, allocation can differ depending on whether it is measured in units of biomass, carbon or nutrients, and on whether it takes into account photosynthesis and respiration by reproductive structures12,13. The shapes of the fitness gain functions will differ depend-ing on the currency chosen, unless different currencies have a linear relationship. Thus, it is problematical to measure fitness gains as a function of one trait, such as flower number, and use that function to suggest the shape of fitness gains for other traits, such as biomass allocation.
This property should not interfere with the model’s ability to predict an ESS for the currency chosen by the investigator, provided the underlying assumption of a trade-off in allocation is met. However, different currencies could lead to conflicting predictions about the stability of a breeding system. To describe sex allocation, an appro-priate measure is one comparing allocation to male struc-tures (e.g. stamens) with total allocation by the plant to female and male structures, using a currency, such as biomass, that shows a trade-off.
Another difficulty is presented by the need to estimate male and female fitness in the field. Here, it is assumed that investigators start by observing sex allocation in a mature plant. For such a plant, production of offspring surviving to reproductive maturity provides a reasonable index of fit-ness, particularly for species without variation in age at reproduction. In natural populations, female fitness is usu-ally estimated by seed production. However, the shape of the female gain curve might be influenced by differential success of seed crops that vary in size3. If seeds compete with their siblings for a limited number of safe sites, this local resource competition will saturate female fitness gains (Box 2). The process requires low seed dispersal. Density-dependent population growth is not sufficient to cause diminishing fitness gains (Box 2). To test for the importance of this phenomenon we need to follow the suc-cess of offspring from individual plants and determine how many survive to reproduce.
Experimental tests of sex-allocation
theory in plants
Diane R. Campbell
A general explanation for diversity in plant breeding systems is offered by sex-allocation
theory. This theory assumes a trade-off between allocation of resources to the two sexual functions. It explains the high frequency
of hermaphroditism in angiosperms by diminishing fitness returns on investment of
more resources in a single function. Recent experimental studies provide tests of this theory by measuring male and female fitness
gains, and examining the trade-off assumption. These studies show how fitness
responds to shifts in allocation. Allocation traits often show heritable variation, but
support for a trade-off remains weak.
Diane Campbell is at the Dept of Ecology and Evolutionary Biology, University of California, Irvine,
Three recent studies provide some clues. In the orchid Tolumniathe number of seedlings recruited to the popu-lation correlated positively with the number of fruits pro-duced experimentally by augmenting pollination9. In the thistle, Cirsium canescens, plants setting more seed because of lower insect herbivory also had more surviving offspring14. Finally, in the borage Cynoglossum officinale, biomass of surviving seedlings increased linearly with seed production15. Only the last case provides sufficient information to describe the shape of the fitness function, but for all three species local resource competition was not so extreme that the advantage to producing more seed was eliminated.
Local seed competition will not affect the male gain curve, at least not for outcrossers that disperse pollen to a large set of mates16. This makes the number of seeds sired by an individual (male fertility) an excellent measure of male fitness. Recent years have seen the development of reliable analytical methods for estimating male fertility from molecular data (Box 3). The latest methods use maximum
likelihood to estimate male fertilities from the set of offspring genotypes and genotypes of potential par-ents8,17,18. The goal is to ob-tain estimates of relative male fertilities that have low bias and low variance. Al-though adding sufficient loci to exclude all but one genet-ically possible father of a seed is appealing19, sampling more offspring is usually easier and leads to a greater reduction in sampling vari-ance for a fixed total number of assays20. One caution is that with low exclusion prob-abilities, the likelihood esti-mation tends to underesti-mate the fitness of highly successful males8. It remains unclear how much this bias alters the apparent shape of the male gain curve. Studies taking this approach should, therefore, attempt to esti-mate the bias through com-puter simulation21,22. Another potential source of bias that could flatten the male gain curve is cryptic gene flow17. This refers to siring of seeds by a plant from outside the reference population that goes undetected because a genetically possible father also exists inside the popu-lation. It is possible to elimin-ate bias as a result of cryptic gene flow by simultaneously estimating gene flow and male fertilities23.
Once estimates of male fertilities have been obtained, regression methods could be used to fit a function re-lating male fitness to sex allocation. Alternatively, it is possible to integrate estimation of parameters describing a fitness function into the maximum-likelihood model18. This method (Box 3) has statistical properties making it especially suitable for describing male gain curves.
The shapes of fitness gain functions
During the past 15 years, attention to the male side of re-production has led to increasingly sophisticated studies of plant fitness. However, only recently have attempts been made to estimate either male or female fitness gains as a function of sex allocationper se. Three studies explored fit-ness gains as a function of a trait thought to reflect sex allo-cation and, in so doing, tested aspects of the theory. Elle and Meagher (unpublished) used state-of-the-art paternity analysis18 to estimate male fertility in the andromon-oecious horsenettle Solanum carolinenseand found that it increased with an increasing proportion of male flowers on the plant, while female fertility decreased24. This result supports a basic assumption of the theory: male fitness Box 1. The classic model of sex allocation for outcrossing species
Male fertility (m) (solid line) and female fertility (f) (broken line) are considered power functions of the proportion of resources put into male function (r):
where kmand kfare constants. An exponent (bor c) ,1 generates a function with diminishing returns, whereas an exponent .1 generates accelerating returns. These exponents determine the evolutionarily stable strategy (ESS) breeding system. (a) Both exponents bandc,1, and the ESS is hermaphroditism. Plotting mversus fwould lead to a convex (bowed out) fitness set. (b) Both band c.1 corresponding to accelerating fitness gains, and the ESS is dioecy.
The fitness of a rare mutant iis the average of its relative female and male fertilities, and can be written as37:
where fand mare average female and male fertilities in the whole resident population. Substituting the power functions into this expression yields:
The ratio rib/rbexpresses the fraction of seeds in the population as a whole that are sired by the mutant. Following standard techniques4the ESS value for ris b/(b1c). The condition for hermaphroditism to be an ESS is: b1c.2bc.
These results mean that the ESS breeding system is determined by just the exponents of the fitness gain func-tions and not the scaling factors kmand kf. Similar methods can be used to model sex allocation in partial selfers2,6.
wi=fi+fmi
m
wi =kf(1−ri)c+kmri b
kmr
bkf(1−r)
c
Trends in Ecology & Evolution
b = 0.4, c = 0.6 b = 1.5, c = 2.5
kf km kf km
0.0 0.2 0.4 0.6 0.8 1.0
F
er
tility
r r
0.0 0.2 0.4 0.6 0.8 1.0
Proportion of resources to male function
f m f m
(a) (b)
(Online: Fig. I)
increases with additional al-location to male function.
The two other studies also support this assumption and, in addition, provide in-formation on shapes of the gain functions. For the bumblebee-pollinated herm-aphroditic herbs C. officinale (hound’s tongue) and Echium vulgare (viper’s bugloss), Rademaker and de Jong25 defined allocation (r) as the ratio of flowers to seeds, a choice justified by the physiological trade-off be-tween these traits. This study is unusual in con-sidering how local resource competition influences the female gain curve. Female fit-ness was estimated by com-paring seedling growth and survival around plants that varied in the number of seeds produced15. Male fit-ness was estimated by mecha-nistic studies of pollen export and, because these species are partially selfing, of selfing rate and inbreed-ing depression. Estimated pollen export increased with flower number26, supporting the idea that investing more in male function benefits male fitness, provided pollen export represents male fer-tility. The male gain curves showed linear or diminish-ing returns and plottdiminish-ing male versus female fitness pro-duced convex fitness sets as predicted in the case when hermaphroditism is the ESS (Ref. 25).
A study of humming-bird-pollinated scarlet gilia (Ipomopsis aggregata) is un-usual in measuring sex
allo-cation in units of biomass, allowing exploration of gain curves as specified in the classic theory22. Sex allocation was defined as the ratio of biomass in the stamens versus the pistil and seeds. Male fertility, estimated with a genetic paternity analysis, increased with increasing allocation to male func-tion. When fitted to a gain function the relationship was best described by diminishing returns (Fig. 1), although with considerable scatter. Female fitness, estimated by seed production, showed accelerating returns on invest-ment. Taken together, the shapes of the two fitness func-tions were consistent with the expectation for a hermaph-rodite (Fig. 1). However, the ESS allocation to male function was lower than observed in nature, and pheno-typic selection through male and female functions did not cancel as expected if the population is at an ESS (Ref. 27). This discrepancy is explainable if variation in seed quality or local resource competition cause a flattening of the
female gain curve so that the number of seeds is a poor measure of female fitness, or if no trade-off in biomass allocation exists.
In summary, these initial studies suggest that male fit-ness responds to increased allocation to male function. In addition, fitness gains for hermaphroditic herbs are quali-tatively consistent with theoretical expectations for that breeding system, but there are quantitative discrepancies between observed and expected sex allocation that remain to be explained.
Is there a trade-off in allocation?
Theory of sex allocation assumes a trade-off in allocation of resources, such that plants investing more in male func-tion retain less for female funcfunc-tion. The classic model is even more restrictive in assuming a linear trade-off be-tween resources invested in the two sexual functions. If no Box 2. Local resource competition and female fitness gains
Local resource competition can cause the female gain curve to saturate4. The process requires low seed dispersal
so that seeds compete with siblings. Density-dependent survival or reproduction is not sufficient.
The figure contrasts the case of low (a) and high (b) seed dispersal in populations with density-dependent vival. From left to right are the spatial distribution of seeds in a population, the spatial distribution of progeny sur-viving to reproduce, and female fitness plotted against allocation to male function (r ). Seeds are produced by two genotypes differing in allocation: females (closed symbols) and hermaphrodites (open symbols). Progeny of three individuals of each genotype are shown. The hermaphrodites allocate half their resources to male function (r 50.5) and produce half as many seeds as females, which allocate no resource to male function (r 50). Survival is density-dependent; the high-density population produces more seeds than the low-density population, but no more seeds survive to reproduce.
(a) Seed dispersal is low so that seeds lie in clusters of siblings and progeny compete only with siblings to yield a single survivor. Even though one genotype makes more seed it does not have more surviving offspring and relative fitness measured by the number of surviving offspring is one. The function relating female fitness to allocation levels off (solid line). The diminishing returns on allocation make hermaphroditism more likely to be evolutionarily stable.
(b) Seeds are dispersed further so that they also compete with nonsiblings. Relative female fitness of the two genotypes estimated by surviving progeny is two, and the fitness gain function is linear.
Although only two extreme genotypes are shown here, the same principles apply in a population of hermaphro-dites differing quantitatively in sex allocation. Cases (a) and (b) could be separated using genetic markers to follow the success of progeny produced by plants with different sized seed crops.
Trends in Ecology & Evolution
(a) Low seed dispersal
High density
Low density
High density
Low density
(b) High seed dispersal
r
0 1
Fitness estimate
Fitness estimate
0 1
2 Surviving
offspring
Seeds
r
0 1
0 1 2
Surviving offspring or seeds Female fitness gain Surviving offspring
Seeds
trade-off exists, hermaphro-ditism could be favored even when gain functions do not show diminishing returns3.
A trade-off should gener-ate a negative genetic corre-lation between allocation by a whole plant into two sexual functions28. However, plants can also vary genetically in their ability to acquire re-sources, which tends to gen-erate positive correlations in allocation characters. The sign of the net genetic corre-lation will depend on both sources of genetic vari-ation29. The few quantitative genetic studies of sex allo-cation, in which male and female function were meas-ured in a common currency, have all uncovered some genetic variation (Table 1). However, three of the four studies found positive gen-etic correlations between allocation to male versus female function. Positive correlations are not surpris-ing given genetic variation in acquisition of resources used for reproduction. If a trade-off still exists, it should show as a negative genetic corre-lation when the influence of total reproductive biomass is factored out to find a partial correlation between allo-cation to male versus female function29. Two studies30,31 have used multivariate meth-ods to adjust estimated correlations for other traits and still found zero or posi-tive genetic correlations (Table 1). These initial results suggest that genetic variation in sex allocation is often small compared with vari-ation in traits related to re-source acquisition and vigor31, perhaps because flowers can draw on different resource pools for male and female parts, as suggested by physiological studies13. Future tests of genetic cor-relations between allocation traits should benefit from the potentially powerful ap-proach of artificial selec-tion32,33. An open question is whether negative corre-lations are stronger for certain currencies, such as nitrogen or biomass.
Fig. 1.Estimated fitness gain functions in scarlet gilia (Ipomopsis aggregata)22. An increase in r (biomass
allo-cation to stamens versus pistil and seeds) led to a significant increase in relative seeds sired per flower (a) and a significant decrease in relative seeds produced per flower (b). Total seed production was linearly related to seeds per flower, giving female gain curves, based on both measures, similar shapes. Power functions were fitted using nonlinear regression to obtain the exponents that determine ESS sex allocation in the classic model. The expo-nents b50.48 (95% CI 50.07–0.89) and c53.69 (95% CI 52.82–4.55) make hermaphroditism the ESS because b1c .2bc, with an expected allocation of r50.12. The observed allocation of 0.28 (95% CI 50.25–0.31) is less strongly female biased. Reproduced, with permission, from Ref. 22.
Trends in Ecology & Evolution
r r
0.0 0.2 0.4 0.6 0.8 1.0
0.0 1.0 2.0 3.0 4.0
m = 1.93 r0.48 f = 3.02 (1 – r)3.69
0.0 0.2 0.4 0.6 0.8 1.0
0.0 1.0 2.0 3.0 4.0
Proportion of resources to male function
(a) (b)
Relativ
e seeds sired per flo
w
er (
m
)
Relativ
e seeds produced per flo
w
er (
f)
Box 3. Estimating male fitness gains using paternity analysis
Current methods for determining male fertility from genetic data make use of likelihood estimation. These meth-ods can be used with dominant markers (RAPDs) or codominant markers (isozymes or microsatellites). The dis-cussion here assumes codominance. Multilocus genotypes are obtained for a sample of progeny, their known maternal parents and all possible fathers in the population. The probability of observing a particular offspring geno-type, given the genotypes of its mother and a candidate father, depends on mendelian segregation. Exclusion of all but one possible father means this probability is zero for all other candidates. Complete exclusion for all progeny is typically not possible. Instead, the procedure estimates the vector of male fertilities for all putative fathers in the population, which maximizes the likelihood for obtaining the entire array of offspring genotypes8,17.. This requires
either assuming that relative male fertilities are the same for all female mates or assaying a sufficient sample of progeny per female to estimate fertilities separately for each seed crop. Recently, Smouse et al. showed how to incorporate estimation of a regression parameter relating male fertility to a phenotypic trait into the likelihood analysis18. This method has low bias and could be applied to determine the shape of a male gain curve by estimating
linear and quadratic terms for male fertility regressed on sex allocation.
Table 1. Genetic correlations between sex-allocation traits obtained from crossing studies
Heritable Genetic
Species Traits variation? correlation Refs
Zea mays Tassel mass and seed mass Yes 20.83a 38
Lythrum salicaria Stamen mass and pistil mass In short-styled 0.87a 39
morph only
Begonia semiovata Male flower mass and female Yes 0.88a 40
flower mass
Ipomopsis aggregata Stamen mass and pistil mass Yes 0.40aand 0.38a,b 30
Mimulus guttatus Pollen number and ovule number Yes 20.32, 0.28, 0.36cand 31
20.27c a
P ,0.05. b
Partial genetic correlation factoring out variation in flower number. c
Prospects: designing a study of fitness gains Describing how sex allocation influences fitness gains in plants has only just begun. To date, no single study has combined a measure of sex allocation directly related to the theory, an estimate of female fitness that includes vari-ation in survival of seeds to reproductive maturity and an accurate estimate of male fitness. How can future studies be designed to do so efficiently?
In a natural population it is difficult to estimate male fertility with the precision necessary to describe a gain function. A more powerful approach would involve con-structing an artificial isolated population in which both genotype and biomass allocation to male versus female function can be manipulated. Such an approach has many advantages that offset a probable cost to realism. An iso-lated population would be free from problems introduced by cryptic gene flow. The investigator could choose multi-locus genotypes to maximize statistical power in estimat-ing male fertility. The ability to manipulate allocation would eliminate the possibility that apparent fitness gains are the result of a trade-off with an unmeasured character with a hidden effect on the shape of the gain curve. Because flower number is particularly easy to manipulate, such a study would be easiest to perform in species where variation in sex allocation results partly from variable flower number26. In other species, manipulation might be possible using selective application of soil nutrients34or hormonal treatments to adjust within-flower sex allo-cation. To measure female fitness as the number of seed offspring surviving would require the use of parentage analysis to identify the mother and father of seedlings35, or setting up a separate study using genetic markers to exam-ine the relationship of seed production to realized female fitness (Box 2).
Such experimental studies of fitness gains will be far more informative if they are combined with efforts to determine the mechanisms underlying them. Several hy-potheses have been proposed to explain saturating male gain curves and thus the prevalence of hermaphroditism in animal-pollinated flowers. One idea is that presenting more pollen leads to diminishing returns on pollen removal and export to other flowers, particularly if pollinators become saturated with pollen or lose much pollen to grooming. In addition, increasing pollen production can increase the fraction of pollen moved to other flowers on the same plant (geitonogamous selfing) at the expense of outcrossing26,36. Alternatively, if pollen dispersal is highly localized, pollen grains might show local mate competition with sib pollen that lands on the same stigma, again generating diminish-ing fitness returns4. One way to separate these hypotheses is to measure fitness gains separately during pollen re-moval, pollen export and post-pollination. Effects on pollen removal are probably the most easily addressed, because they can be studied by presenting plants differing in sex allocation to captive pollinators. Effects on pollen export could be separated from post-pollination effects by simul-taneously tracking pollen dispersal and estimating male fertility. Comparing fitness gains based on these two measures would indicate the stage at which any curvature is produced.
Perhaps the most fruitful systems for such experimen-tal tests will be genera in which there are two or more related species that differ in breeding systems. Ascertain-ing whether the shapes of fitness gain curves differ be-tween those species in the direction expected, and the key step in the life cycle at which this happens, would provide a powerful test of sex-allocation theory.
Acknowledgements
I am grateful to Tom J. de Jong and Peter G.L. Klinkhamer for many helpful discussions. I also thank Ann K. Sakai, Nickolas M. Waser and Stephen G. Weller for encouraging me to pursue this topic and for comments on this article.
References
1 Darwin, C. (1859) On the Origin of Species, John Murray
2 Charlesworth, D. and Charlesworth, B. (1981) Allocation of resources to male and female function in hermaphrodites. Biol. J. Linn. Soc. 15, 57–74
3 Charnov, E.L. (1982) The Theory of Sex Allocation, Princeton University Press
4 Lloyd, D.G. (1984) Gender allocations in outcrossing cosexual plants. In
Perspectives in Plant Population Ecology (Dirzo, R. and Sarukhan, J., eds), pp. 277–300, Sinauer
5 Charnov, E.L. (1987) On sex allocation and selfing in higher plants.
Evol. Ecol. 1, 30–36
6 de Jong, T.J. et al. (1999) How geitonogamous selfing affects sex allocation in hermaphroditic plants. J. Evol. Biol. 12, 166–176
7 Klinkhamer, P.G.L. et al. (1997) Sex and size in cosexual plants. Trends Ecol. Evol. 12, 260–265
8 Smouse, P.E. and Meagher, T.R. (1994) Genetic analysis of male
reproductive contributions in Chamaelirium luteum (L.) Gray (Liliaceae). Genetics 136, 313–322
9 Ackerman, J.D. et al. (1996) Seedling establishment in an epiphytic orchid: an experimental study of seed limitation. Oecologia 106, 192–198
10 Seger, J. and Eckhart, V.M. (1996) Evolution of sexual systems and sex allocation in plants when growth and reproduction overlap. Proc. R. Soc. London Ser. B 263, 833–841
11 Lloyd, D.G. (1987) A general principle for the allocation of limited
resources. Evol. Ecol. 2, 175–187
12 Goldman, D.A. and Willson, M.F. (1986) Sex allocation in functionally
hermaphroditic plants: a review and critique. Bot. Rev. 52, 157–194
13 Ashman, T. (1994) A dynamic perspective on the physiological cost of reproduction in plants. Am. Nat. 144, 300–316
14 Louda, S.M. and Potvin, M.A. (1995) Effect of inflorescence-feeding insects on the demography and lifetime fitness of a native plant. Ecology 76, 229–245
15 Rademaker, M.C.J. and de Jong, T.J. (1999) The shape of the female
fitness curve for Cynoglossum officinale: quantifying seed dispersal and seedling survival in the field.Plant Biol. 1, 351–356
16 de Jong, T.J. and Klinkhamer, P.G.L. (1994) Plant size and reproductive success through female and male function. J. Ecol. 82, 399–402
17 Roeder, K. et al. (1989) Application of maximum likelihood methods to population genetic data for the estimation of individual fertilities. Biometrics 45, 363–379
18 Smouse, P.E. et al. (1999) Parentage analysis in Chamaelirium luteum
(L.) Gray (Liliaceae): why do some males have higher reproductive contributions? J. Evol. Biol. 12, 1069–1077
19 Krauss, S.L. (1999) Complete exclusion of nonsires in an analysis of paternity in a natural plant population using amplified fragment length polymorphism (AFLP). Mol. Ecol. 8, 217–226
20 Milligan, B.G. and McMurry, C.K. (1993) Maximum likelihood analysis of male fertility using dominant and codominant genetic markers. Mol. Ecol. 2, 275–283
21 Morgan, M.T. (1998) Properties of maximum likelihood male fertility
estimation in plant populations. Genetics 149, 1099–1103
22 Campbell, D.R. (1998) Variation in lifetime male fitness in
Ipomopsis aggregata: tests of sex allocation theory. Am. Nat. 152, 338–353
23 Sork, V.L. et al. (1999) Landscape approaches to historical and contemporary gene flow in plants. Trends Ecol. Evol. 14, 219–224
24 Elle, E. (1999) Sex allocation and reproductive success in the
andromonoecious perennial, Solanum carolinense (Solanaceae). I. Female success. Am. J. Bot. 86, 278–286
25 Rademaker, M.C.J. and de Jong, T. Testing sex-allocation theory:
flowers vs. seeds in hermaphrodite plants. Oikos (in press)
26 Rademaker, M.C.J. and de Jong, T.J. (1998) Effects of flower number on estimated pollen transfer in natural populations of three hermaphroditic species: an experiment with fluorescent dye. J. Evol. Biol. 11, 623–641
27 Morgan, M.T. (1992) The evolution of traits influencing male and
R
ecently, the field of commu-nity ecology has integrated the notions of ‘top-down’ and ‘bottom-up’ influences on com-munity organization1–5. According to this framework, populations oc-cupy positions in a food web and their abundance or biomass can be controlled by populations at higher trophic levels (e.g. top-down effects of predators on prey), lower trophic levels (e.g. bottom-up effects of biotic resources on consumers) or the same level (more traditional competitive in-teractions). The top-down bottom-up approach is sympathetic to the notion that interactions between populations might be either direct (e.g. a predator controlling prey density) or indirect (e.g. a primary producer enhancing a parasitoid population by increasing popu-lation growth of a herbivore host).Recently, temporal fluctuations in the strengths of inter-actions among species have been of great interest to ecolo-gists6, but these fluctuations have not been integrated into the top-down bottom-up paradigm2.
Many terrestrial (and aquatic) ecosystems are charac-terized by pulsed resources – the temporary availability of dramatically higher than normal levels of resources, which then become depleted with time. Examples include mast fruiting by trees and herbs7,8, periodic irruptions of palat-able insects9, and storm-induced transport of marine re-sources (e.g. whale carcasses) to terrestrial systems10or of terrestrial resources (organic nitrogen or phosphorus) to aquatic systems11. Determining how the effects of pulsed
resources permeate through food webs is a major challenge for community ecology. Tracing the impacts of pulsed resources on communities incorporates both bottom-up and top-down influ-ences, but requires ecologists to incorporate the concepts of disturbance, time delays and mobility of organisms into con-siderations of how populations affect one another.
Conceptual underpinnings Recently, far-reaching effects of pulsed seed production have been documented in several ter-restrial habitats. Mast seeding occurs in many terrestrial ecosys-tems, including boreal, temperate and tropical forests, as well as grasslands and deserts7,12–14. If the trees, shrubs or herbaceous plants that synchronously pro-duce large fruit or seed crops (i.e. mast) are highly abun-dant, even dominant, members of their communities, the result is an extraordinary flush of nutritious foods for con-sumers (Box 1). Pulses of seed production can be elicited by regional or global climatic fluctuations, including El Niño southern-oscillation (ENSO) events, which can cause unusually high rainfall or solar radiation (potentially important abiotic resources). In arid ecosystems, plant communities typically respond to heavy rainfall with explosive production of seeds and vegetative tissues15–17. In moist, tropical forest systems, trees might respond to either unusually heavy rain or unusually dry, sunny conditions with heavy fruit production18–20 (Table 1).
Pulsed resources and community
dynamics of consumers in
terrestrial ecosystems
Richard S. Ostfeld and Felicia Keesing
Many terrestrial ecosystems are characterized by intermittent production of abundant resources for consumers, such as mast seeding
and pulses of primary production following unusually heavy rains. Recent research is revealing patterns in the ways that consumer communities respond to these pulsed resources.
Studies of the ramifying effects of pulsed resources on consumer communities integrate
‘top-down’and ‘bottom-up’approaches to
community dynamics, and illustrate how the strength of species interactions can change
dramatically through time.
Richard Ostfeld is at the Institute of Ecosystem Studies, Millbrook, NY 12545, USA (rostfeld@ecostudies.org); Felicia Keesing is at Siena
College, Loudonville, NY 12211, USA, and the Institute of Ecosystem Studies, Millbrook, NY 12545,
USA (fkeesing@siena.edu).
28 Charlesworth, D. and Morgan, M.T. (1991) Allocation of resources to
sex functions in flowering plants. Philos. Trans. R. Soc. London Ser. B 332, 91–102
29 de Jong, G. (1993) Covariances between traits deriving from
successive allocations of a resource. Funct. Ecol. 7, 75–83
30 Campbell, D.R. (1997) Genetic correlation between biomass allocation
to male and female functions in a natural population of Ipomopsis aggregata. Heredity 79, 606–614
31 Fenster, C.B. and Carr, D.E. (1997) Genetics of sex allocation in
Mimulus (Scrophulariaceae). J. Evol. Biol. 10, 641–661
32 Meagher, T.R. (1994) The quantitative genetics of sexual dimorphism
in Silene latifolia (Caryophyllaceae). II. Response to sex-specific selection. Evolution 48, 939–951
33 Mazer, S.J. et al. (1999) Responses of floral traits to selection on primary sexual investment in Spergularia marina: the battle between the sexes. Evolution 53, 717–731
34 Lau, T. and Stephenson, A.G. (1993) Effects of soil nitrogen on pollen
production, pollen grain size, and pollen performance in Cucurbita pepo (Cucurbitaceae). Am. J. Bot. 80, 763–768
35 Meagher, T.R. and Thompson, E.A. (1987) Analysis of parentage for
naturally established seedlings of Chamaelirium luteum. Ecology 68, 803–812
36 Harder, L.D. and Barrett, S.C.H. (1995) Mating cost of large floral
displays in hermaphrodite plants. Nature 373, 512–515
37 Morgan, M.T. and Schoen, D.J. (1997) The role of theory in an emerging
new plant reproductive biology. Trends Ecol. Evol. 12, 231–234
38 Garnier, P. et al. (1993) Costly pollen in maize. Evolution 47, 946–949
39 O’Neil, P. and Schmitt, J. (1993) Genetic constraints on the
independent evolution of male and female reproductive characters in the tristylous plant Lythrum salicaria. Evolution 47, 1457–1471
40 Agren, J. and Schemske, D.W. (1995) Sex allocation in the monoecious
