ARTICLE IN PRESS Perspectives in Plant Ecology, Evolution and Systematics Perspectives in Plant Ecology, Evolution and Systematics 9 (2007) 29–35 www.elsevier.de/ppees
Hierarchical pre-dispersal fitness assessment in a Mediterranean shrub plant Jose´ M. Serrano, Francisco Lo´pez, Juan A. Delgado, Sara G. Fungairin˜o, Francisco J. Acosta Departamento de Ecologı´a, Facultad de Biologı´a, Universidad Complutense de Madrid, 28040 Madrid, Spain Received 10 July 2007; accepted 11 July 2007
Abstract Under the concept of modularity, it is possible to recognise how seed production, as well as any other process affecting it, are hierarchically structured within fruits, within individual plants and within populations. In this work, we analysed the effects of pre-dispersal seed predation by insects upon a set of hierarchical levels in a population of the Mediterranean shrub plant Cistus ladanifer (‘‘rock rose’’) throughout a complete fruit-producing season (which takes place during the summer months). Almost all individual plants were predated, which implies that the effects of predation at the population level (regardless of the extent of predation within each individual) were virtually uniform. Within the individuals, however, the predation rate was close to a proportion of 0.5 (half of the fruits of each individual were predated), which indicates that this hierarchical level is likely to be subjected to a differential action of selection. Predation rates within the fruits showed an intermediate value (lower than that observed at the population level but higher than that at the individual level). According to these results, the pressure of phenotypic selection may therefore give rise to greater variation among fruits of the same individual than among seeds of the same fruit. In terms of the temporal patterns observed there was a large variation in the increments of predation along the fruiting season, which implies a high degree of heterogeneity in the temporal distribution of the effects of predation pressure on fitness. Besides its use in the specific example of the plant species studied in this work, the methodological procedure presented in this paper (integration of the temporal changes of different hierarchical levels) might be foreseen, in fact, as a useful tool for analysing the hierarchical structuring of fitness in modular organisms in general. This procedure allows to discriminate and integrate selection pressures and their effects across different phenotypic levels, from the infraindividual ones up to the population level. r 2007 Ru¨bel Foundation, ETH Zu¨rich. Published by Elsevier GmbH. All rights reserved. Keywords: Cistus ladanifer; Phenotypic selection; Plant modularity; Seed predation
Introduction Plant fecundity, as expressed by the number of seeds produced via the female function (Devlin and StephenCorresponding author. Tel.: +34913945085; fax: +34913945081.
E-mail address:
[email protected] (J.M. Serrano).
son, 1987), is the most direct – and common-way of measuring plant reproductive fitness. The consequences for fitness of variations in resource availability (nutrients, carbon dioxide, light, and pollen), and plant resource allocation and partitioning strategies (timing of first reproduction, reproductive vs. vegetative structures, size and number of inflorescences) are typically
1433-8319/$ - see front matter r 2007 Ru¨bel Foundation, ETH Zu¨rich. Published by Elsevier GmbH. All rights reserved. doi:10.1016/j.ppees.2007.07.002
ARTICLE IN PRESS 30
J.M. Serrano et al. / Perspectives in Plant Ecology, Evolution and Systematics 9 (2007) 29–35
assessed in terms of total seed production per individual plant (Crawley, 1986; Lovett Doust and Lovett Doust, 1988; Fenner, 1985, 1992). Although fitness values are usually given as overall output figures per individual, the modular structure of plants (Silvertown, 1989) – which obviously includes the reproductive parts that constitute the generative basis of the fitness outcome – gives rise to a potential hierarchical fitness structure within the individual plant (Tuomi and Vuorisalo, 1989a, b; Acosta et al., 1993; Pedersen and Tuomi, 1995; Lo´pez et al., 2002; de Kroon et al., 2005). Under the concept of modularity, each individual plant is made up by groups of iterative, inter-related shoot modules. All modules are semiautonomous units that contribute to the processes of growth and propagation of genetic individuals (Harper and Bell, 1979; White, 1979; Tuomi and Vuorisalo, 1989a, b; Silvertown, 1989; Schmid, 1990; Weiner, 2004; de Kroon et al., 2005). From this point of view, the fitness of an individual plant (or, in more general terms, that of any modular organism) can be expressed as the additive integration of the contribution to fecundity of all the subunits that comprise it. If we consider plant fruits as reproductive modules and assess their individual contribution to fitness, it is possible to recognise how seed production – as well as any other process affecting it – are hierarchically structured within fruits, within individual plants and within populations. This distinction is not a vague structural description but it rather opens up new possibilities for a better understanding of reproduction and fitness-related phenomena in plants. Not only the genetic individuals but also any structural or functional subunits (from shoot modules to higherlevel functional aggregations) can be significantly heterogeneous in terms of seed production and can therefore act as potential levels of phenotypic selection (Acosta et al., 1993). Pre-dispersal seed predation (mainly by insects) is one of the factors that most severely affects seed survival (Janzen, 1971a; Heithaus et al., 1982; Zammit and Hood, 1986; Garcı´ a, 1998; Norambuena and Piper, 2000; Serrano et al., 2001; Bas et al., 2005). Given that it can significantly reduce reproductive output (typically under the influence of density-dependent mechanisms), its magnitude must be analysed in order to obtain a more accurate assessment of plant reproduction performance (Janzen, 1971a; Zimmerman, 1980; Crawley, 1992; Hulme, 1998; Ehrle´n, 2002). For all the above reasons, in the same way that seed production (as an initial measure of fitness) can be estimated in order to analyse different significant potential levels of phenotypic selection (Acosta et al., 1993), seed predation (as a selection pressure affecting seed output in different ways) can also be analysed at different levels in order to understand in greater detail how it affects plant fitness.
In this study, we analysed pre-dispersal seed predation by insects in a natural population of a woody perennial plant species (‘‘rock rose’’, Cistus ladanifer L., 1753) on a set of hierarchical levels in the plants and throughout a complete fruiting period (during the summer every year). This analysis has been carried out through of an ad hoc methodological procedure (valid for any modular organism with a hierarchical structure) that properly allows to study the pre-dispersal predation process taking into account the modular structure of plants and its implications for selection. Seed predation was assessed by means of the degree of infestation of different units within the modular hierarchy: (i) plants infested by insects in the population, (ii) infested fruits per infested plant, and (iii) seeds destroyed by insect predation in each infested fruit. The time functions of seed predation (temporal increase of infestation along the reproductive period) at the different hierarchical levels were integrated in order to obtain the final effects of seed predation by insects on the fitness of individual plants and on the reproductive output of the whole population.
Material and methods Study site and species The study area is located near the town of Tres Cantos, 20 km North of Madrid (Central Spain). All fieldwork was carried out in a natural patch of C. ladanifer containing some scattered trees of Mediterranean Holm-oak (Quercus rotundifolia Lam. 1785). The species of the genus Cistus show an obligatory seeding strategy that is an adaptive feature to the typical recurrent fires that affect many Mediterranean shrub communities. They thus depend upon seeds available in the soil for post-fire seedling establishment. C. ladanifer fruits are lignified globular capsules internally divided by 7–10 valvae that delimit the same number of compartments (known as locules). Seeds (500–1000 per fruit) remain inside the fruits during a period of several weeks, before being released around late August–early September every year. There is no specialised dispersal mechanism and fruits simply open naturally releasing the seeds that fall to the ground. This species is subjected to sizeable seed losses as a result of pre-dispersal insect predation. Some insect species pierce the external wall of the fruit and feed on the seeds, drilling through the valvae to gain access to seeds from adjacent compartments. In most cases, all seeds from infested compartments are eaten and any remaining untouched seeds rot away by the action of microorganisms. In our study area, the almost exclusive seed predator (4 95% of infested fruits) is the larval stage of
ARTICLE IN PRESS J.M. Serrano et al. / Perspectives in Plant Ecology, Evolution and Systematics 9 (2007) 29–35
Cleonymia (Serryvania) yvanii (Duponchel, 1833) (Lepidoptera, Noctuidae). Since the flowering period of C. ladanifer is very short (only a few days in May), virtually all fruits are synchronously produced and thus simultaneously exposed to insect predation.
Sampling and analysis The sampling work was carried out in the summer of 1996. Three hundred C. ladanifer individual plants were randomly selected from the studied population and tagged in the field. In order to analyse the temporal changes in the infestation process, data were collected in 6 consecutive subsamples (50 randomly selected individuals per subsample) at 15-day intervals. The data (see below) from the first subsample were collected in the last week of June, when most fruits were nearly fully grown (with an already partially lignified wall), and those corresponding to the last subsample were taken in the first week of September, coinciding with the opening of fruits and the dispersal of seeds. The sampling procedure in each subsample was as follows: (i) All the fruits of each individual plant were externally inspected to assess whether or not they were infested (this can be easily and confidently determined by simple visual inspection because of the presence of holes caused by the insects in the fruits). (ii) In order to obtain a representative estimate of the degree of internal fruit predation, half of the infested fruits from each individual plant were randomly selected and opened by hand, recording the number of infested compartments per fruit. This information was then used to assess the relative effect of seed predation by insects at all levels of the hierarchical arrangement of seed production in the C. ladanifer population studied: individual plants within the population (level I), fruits within the individual plants (level II) and compartments within the fruits (level III). Seed predation was estimated at level I as the proportion of infested individual plants within the population (i/p), calculated as the number of plants containing at least one infested fruit divided by the total number of plants (a single value per temporal subsample). The estimate of predation at level II was calculated as the number of infested fruits divided by the total number of fruits in every infested individual plant (f/i). These values were averaged in each temporal subsample to obtain a mean external predation rate per infested plant.
31
Finally, seed predation at level III was estimated in each randomly selected fruit as the number of infested compartments divided by the total number of compartments in the fruit (c/f). These values were averaged in each temporal subsample in order to obtain a mean internal predation rate per infested fruit. Temporal changes in seed predation at the three different hierarchical levels considered were analysed throughout the fruiting season by using the information obtained from the six consecutive temporal subsamples. Since predation levels increased to a maximum asymptotic value in all cases, these data were summarised by means of the statistical fitting to suitable asymptotic functions, using SYSTAT statistical software (Wilkinson, 1987). According to the observed temporal change of the data gathered, we used functions of the general type: y ¼ ai ð1 expðbi tÞÞ, where ai (asymptote of the curve) is, respectively, the final proportion of infested individual plants within the population of C. ladanifer (level I), the final value of the mean external predation rate per infested plant (level II), or the final value of the mean internal predation rate per infested fruit (level III). bi is a parameter determining the rate of approach of y (predation rate for each hierarchical level) to its asymptote along time (t). Given that the variables employed to estimate the effects of seed predation at the three different hierarchical levels are independent, the integration of the dynamic effects of predation on pre-dispersal fitness across these levels can be obtained by simply multiplying the temporal functions. This provides a continuous account of the temporal increase in predation rates in the whole population by integrating the different infestation processes taking place at different hierarchical levels in the plants.
Results The total number of plants and fruits studied for the assessment of predation rates at the different hierarchical levels are summarised in Table 1. The analyses carried out (see below) were performed using data from 300 individual plants, which yielded a total of 24 439 fruits. One-third of these fruits (8166) were infested. Half of the predated fruits (4083) were studied in order to assess internal predation. The fitted functions showing the temporal changes in seed predation at the three different hierarchical levels considered are represented in Fig. 1. The maximum likelihood estimation of both parameters (ai and bi) for all fitted functions gave considerably high r2 values (in the order of 0.99) (Fig. 1).
ARTICLE IN PRESS 32
J.M. Serrano et al. / Perspectives in Plant Ecology, Evolution and Systematics 9 (2007) 29–35
Table 1. Total numbers of plants and fruits studied (with number of infested units in parenthesis) in each temporal subsample (t1–t6). Internal predation rates were estimated using a random sample of half the infested fruits from each temporal subsample t1
t2
t3
t4
t5
t6
Total
Plants
50 (27)
50 (36)
50 (45)
50 (43)
50 (47)
50 (48)
300 (246)
Fruits External predation Internal predation
4100 (496) 248 (248)
4086 (1000) 500 (500)
4124 (1614) 807 (807)
3994 (1563) 781 (781)
4148 (1694) 847 (847)
3987 (1799) 899 (899)
24439 (8166) 4083 (4083)
Fig. 1. Effects of insect predation throughout the fruiting season at the different hierarchical levels considered in the population of C. ladanifer. Level I: Individual plants in the population. Level II: Fruits in an individual plant. Level III: Compartments in a fruit. For details on predation estimates at each level, see Material and methods. All fitting curves are asymptotic functions of the type y ¼ ai(1exp(bit)), where ai (asymptote value) is the final proportion of infested units per level.
In the C. ladanifer population studied, the increase in predation rates at the individual level (level I) was considerably high, affecting over 50% of the plants already in the first period (15 days after fruit formation),
around 70% in the following 15 days, and reaching a stabilised value of over 90% after only 45 days from the beginning of the fruiting period (Fig. 1). Fruit predation within individual plants (external insect predation rate
ARTICLE IN PRESS J.M. Serrano et al. / Perspectives in Plant Ecology, Evolution and Systematics 9 (2007) 29–35
per plant; level II) followed a very similar pattern, but with lower values: only about 20% of the fruits were infested in the first 15 days, about 35% after 30 days, and the final value of predation was around 45% of the total fruit yield of the individual plants (Fig. 1). Internal insect predation rate (level III), although showing a similar trend, was higher than that observed at the other levels: in the first 15 days insect predation had destroyed, on average, approximately 45% of the compartments of each infested fruit and became eventually stabilised at a value of around 65% of the total content of the fruit in only 30 days (Fig. 1). Fig. 2 shows the resulting compound function that incorporates the simultaneous temporal changes in predation effects on C. ladanifer pre-dispersal fitness at different hierarchical levels. It summarises the total proportion of predated seeds (as measured by infested compartments) in the population. This synthetic function shows that pre-dispersal seed predation by insects on the C. ladanifer population studied did not produce severe overall seed losses (only around 30%). Most seed predation took place during a period comprised in the first third of the fruiting season – when fruits were still only partially lignified – and that level of predation remained virtually constant during the last third of the season. As a result, the changes in seed predation over time followed a sigmoid function, showing no increase during a considerable interval of time of several weeks before the end of the fruiting season.
Fig. 2. Effects of insect predation throughout the fruiting season on the population of C. ladanifer in terms of proportion of predated seeds (as measured by infested compartments) in the population. This compound function summarizes the simultaneous changes in predation at the different hierarchical levels in the population, and it was obtained by multiplying the functions in Fig. 1.
33
Discussion The intensity of pre-dispersal seed predation by insects in different plant species has been reported to show ample variation in both space and time (Janzen, 1971b; De Steven, 1983; Duggan, 1985; Wright, 1990; Crawley, 1992). Its role as a constraining factor on plant population dynamics has often been considered as paramount, significantly limiting both the recruitment and the resulting abundance of individual plants (De Steven, 1983; Youtie and Miller, 1986; Ehrle´n, 2002, and references therein). In our case, several factors are likely contributing to the predation pattern observed in C. ladanifer. The stabilisation of seed predation, even when much of the seed yield is still potentially available to pre-dispersal seed predators, can easily be accounted for by the developmental phenology of the noctuid larvae. The time period for the completion of their within-fruit developmental stages is shorter than the C. ladanifer fruiting season and thus the cumulative predation rate will therefore become stabilised over the fruiting season as the number of active predators declines. The low predation rate observed in this study could be accounted for by the phenomenon known as predator satiation (larger seed production output than what predator density and activity can affect). There are different ways, however, in which this ‘satiation strategy’ for overcoming insect predation can be structured in the plant population. It could be a consequence of different combinations of (i) a large number of plants producing fruits (relative to the current predation pressure), (ii) a large number of fruits produced per plant, and (iii) a large number of seeds produced per fruit. The simultaneous analysis of predation at different hierarchical levels within the plant population is thus a useful method for ascertaining the relative importance of each one of these components. These contributions are manifested in the temporal dynamic changes in the relative values of (i) the proportion of infested plants within the population, (ii) the proportion of infested fruits per plant, and (iii) the proportion of infested compartments per fruit. This distinction adds a new functional and explanative dimension to the assessment of pre-dispersal fitness, both in terms of the temporal dynamics of the infestation process and in relation to the hierarchical structure of the fitness outcome in the plant population. This provides a significant insight into the details of the process of predation, because an equivalent value of final pre-dispersal fitness can be attained with many different combinations of predation rates (e.g., a high proportion of infested individual plants in the population and a low proportion of infested fruits per plant or vice versa) and also through many different temporal pathways (faster or slower increases in predation rates,
ARTICLE IN PRESS 34
J.M. Serrano et al. / Perspectives in Plant Ecology, Evolution and Systematics 9 (2007) 29–35
reaching identical values at the end of the fruiting season). The relevance of the analysis of these processes lies in the fact that it reveals (i) the magnitude of the predation pressure on reproductive fitness across the organisational levels and developmental stages of the population, and (ii) the degree of homogeneity or heterogeneity in the distribution of this pressure. This heterogeneity could be measured, for example, with the use of the commonly employed Shannon–Wiener index: H¼
s X
pi ln pi ,
i¼1
where s is the number of different classes and pi the relative frequency of the ith class. Since the units within each hierarchical level can only present one of two possible states (classes) – predated or not predated (s ¼ 2) – the maximum heterogeneity would be reached when half of the units are predated and the other half remains intact (Fig. 3). Thus, in terms of predation rate (variable from 0 to 1), a value of 0.5 would indicate maximum heterogeneity in the selective effects of predation at a certain hierarchical level. Any other value above or below this one (greater (40.5) or smaller (o0.5) magnitude of predation pressure) would yield a lower heterogeneity. In the case of the asymptotic functions that describe the change over time in the C. ladanifer population studied, this value (predation rate) is summarised by the parameter ai. The comparison of the magnitude of predation rates among organisational levels in a population thus indicates which levels are subjected to greater losses in terms of the proportion of units affected. Besides this, the analysis of the degree of heterogeneity in the predation pressure across these same levels provides a comparative measure of its selective effects. Regardless of the magnitude of the predation rates, the less heterogeneous the distribution of the effects on a level, the less affected by this pressure as a true selective force
Fig. 3. Graphic representation of the distribution of heterogeneity values (H; estimated with the Shannon–Wiener index) as a function of predation rate (in percentage).
that level would be. In the case of the C. ladanifer population studied, the different individual plants (level I, HI ¼ 0.183) would be similarly affected by the predation pressure, whereas the different fruits within each individual, although suffering lower losses, would be subjected to a more differential action of selection (level II, HII ¼ 0.688). The predation rate within the fruits (level III, HIII ¼ 0.649) showed an intermediate value between these two, which implies that the phenotypic selection pressure of insect predation will give rise to greater variation among fruits than among seeds of the same fruit. In relation to the temporal dynamics of seed predation, and regardless of the final values of infestation, the speed of the progression of this process over the fruiting season also provides a measure of the heterogeneity or homogeneity in the temporal distribution of the effects of predation pressure on fitness. Within the limits of variation of predation rates during the time interval of the fruiting season, a progression through a straightline function would indicate maximum temporal homogeneity, since the increments in predation would be constant. Any deviations from this ideal straight line, both above (asymptotic functions) and below (exponential functions) would increase heterogeneity. In the C. ladanifer population studied, this process is clearly heterogeneous, being skewed towards the initial stages, as indicated by the asymptotic functions (see Fig. 1; degree of temporal heterogeneity summarised in the parameter bi). The integration of the effects of predation across the different hierarchical levels through the overall temporal function (see Fig. 2) provides a dynamic composite pattern of the changes in fitness for the whole population. This procedure can be applied to any number of levels and for different types of mathematical functions by simply multiplying them, as the estimates of predation rates for each level are independent. The organisational hierarchy of the fitness output and the predation selective pressures acting upon it can thus be analysed in order to detect and assess quantitative differences among levels. Different functional responses in the seed production/predation process can be directly compared among years in a population, among several populations or even among populations of different species, provided that the same hierarchy of organisational levels is considered. This functional analysis can be performed by contrasting the magnitude of heterogeneity in predation pressure with that of seed production across the levels. This provides an integrated picture of the interacting processes that determine the pre-dispersal fitness outcome. Two of the levels studied in the present investigation have been reported to be significantly heterogeneous in terms of seed production: fruits and individual plants (Acosta et al., 1993). As we have
ARTICLE IN PRESS J.M. Serrano et al. / Perspectives in Plant Ecology, Evolution and Systematics 9 (2007) 29–35
shown above, however, they differ markedly in the degree of heterogeneity in predation pressure. Whereas the differences in seed production per individual plant will remain after seed predation (due to the homogeneity of the latter factor at this level), those existing among fruits could be attenuated or reinforced depending on which fruits are predated. We believe that the procedure presented in this paper constitutes a useful tool for the assessment and analysis of the hierarchical structuring of fitness. It allows distinguishing and integrating several selection pressures and their effects across different levels, from the infraindividual phenotypic organism, through the genotypic one, to the population level.
Acknowledgements This study was partly funded by The Spanish InterMinisterial Commission of Science and Technology (Research Project AMB-0777-C02-02). We thank Dr. Jose´ L. Yela for the identification of the lepidopteran larvae and E. Ortega, M. Castro, and J.M. Barandica for their help during the fieldwork.
References Acosta, F.J., Serrano, J.M., Pastor, C., Lo´pez, F., 1993. Significant potential levels of hierarchical phenotypic selection in a woody perennial plant, Cistus ladanifer. Oikos 68, 267–272. Bas, J.M., Go´mez, C., Pons, P., 2005. Fruit production and predispersal seed fall and predation in Rhamnus alaternus (Rhamnaceae). Acta Oecol. 27, 115–123. Crawley, M.J., 1986. Plant Ecology. Blackwell, Oxford. Crawley, M.J., 1992. Seed predators and plant population dynamics. In: Fenner, M. (Ed.), Seeds: The Ecology of Regeneration in Plant Communities. C.A.B. International, Wallingford, pp. 157–191. De Kroon, H., Huber, H., Stuefer, J.F., van Groenandel, J.M., 2005. A modular concept of phenotypic plasticity in plants. New Phytol. 166, 73–82. De Steven, D., 1983. Reproductive consequences of insect seed predation in Hamamelis virginiana. Ecology 6, 89–98. Devlin, B., Stephenson, A.G., 1987. Sexual variations among plants of a perfect-flowered species. Am. Nat. 130, 199–218. Duggan, A.E., 1985. Pre-dispersal seed predation by Anthocaris cardamines (Pieridae) in the population dynamics of the perennial Cardamine pratensis (Brassicaceae). Oikos 44, 99–106. Ehrle´n, J., 2002. Assessing the lifetime consequences of plant–animal interactions for the perennial herb Lathyrus vernus (Fabaceae). Perspect. Plant Ecol. Evol. Syst. 5/3, 145–163. Fenner, M., 1985. Seed Ecology. Chapman & Hall, New York. Fenner, M., 1992. Seeds: The Ecology of Regeneration in Plant Communities. C.A.B. International, Wallingford.
35
Garcı´ a, D., 1998. Interaction between juniper Juniperus communis L. and its fruit pest insects: pest abundance, fruit characteristics and seed viability. Acta Oecol. 19, 517–525. Harper, J.L., Bell, A.D., 1979. The population dynamics of growth form in organism with modular construction. In: Anderson, R.M. (Ed.), Population Dynamics. Blackwell, Oxford, pp. 29–52. Heithaus, E.R., Stashko, E., Anderson, P.K., 1982. Cumulative effects of plant–animal interactions on seed production by Bauhinia ungulata, a neotropical legume. Ecology 63, 1294–1302. Hulme, P.E., 1998. Post-dispersal seed predation: consequences for plant demography and evolution. Perspect. Plant Ecol. Evol. Syst. 1(1), 32–46. Janzen, D.H., 1971a. Seed predation by animals. Annu. Rev. Ecol. Syst. 2, 465–492. Janzen, D.H., 1971b. Intra- and interhabitat variations in Guazuma ulmifolia (Sterculiaceae) seed predation by Amblycertus cistelinus (Bruchidae) in Costa Rica. Ecology 56, 1009–1013. Lo´pez, F., Fungairin˜o, S.G., Serrano, J.M., de las Heras, P., Acosta, F.J., 2002. Analysing hierarchically structuredfitness and modular dynamics in plants: integration of concepts from population dynamics. Perspect. Plant Ecol. Evol. Syst. 5 (2), 123–129. Lovett Doust, J.L., Lovett Doust, L.L., 1988. Plant Reproductive Ecology: Patterns and Strategies. Oxford University Press, New York. Norambuena, H., Piper, G.L., 2000. Impact of Apion ulicis Forster on Ulex europaeus L. seed dispersal. Biol. Control 17, 267–271. Pedersen, B., Tuomi, J., 1995. Hierarchical selection and fitness in modular and clonal organisms. Oikos 73, 167–180. Schmid, B., 1990. Some ecological and evolutionary consequences of modular organization and clonal growth in plants. Evol. Trends Plants 4, 25–34. Serrano, J.M., Delgado, J.A., Lo´pez, F., Acosta, F.J., Fungairin˜o, S.G., 2001. Multiple infestation by seed predators: the effect of loculate fruits on intraspecific insect larval competition. Acta Oecol. 22, 153–160. Silvertown, J., 1989. The paradox of seed size and adaptation. Trends Ecol. Evol. 4, 24–26. Tuomi, J., Vuorisalo, T., 1989a. What are the units of selection in modular organisms? Oikos 54, 227–233. Tuomi, J., Vuorisalo, T., 1989b. Hierarchical selection in modular organisms. Trends Ecol. Evol. 4, 209–213. Weiner, J., 2004. Allocation, plasticity and allometry in plants. Perspect. Plant Ecol. Evol. Syst. 6 (4), 207–215. White, J., 1979. The plant as a metapopulation. Annu. Rev. Ecol. Syst. 10, 109–145. Wilkinson, L., 1987. SYSTAT: The System for Statistics. Evanston, Illinois. Wright, S.J., 1990. Cumulative satiation of a seed predator over the fruiting season of its host. Oikos 58, 272–276. Youtie, B.A., Miller, R.F., 1986. Insect predation on Astragalus filipes and A. purshii seeds. Northwest Sci. 60, 42–46. Zammit, C., Hood, C.W., 1986. Impact of flower and seed predators on seed-set in two Banksia shrubs. Aust. J. Ecol. 11, 187–193. Zimmerman, M., 1980. Reproduction in Polemonium: predispersal seed predation. Ecology 61, 502–506.
