Quantifying the importance of local niche-based and stochastic processes to tropical tree community assembly

Although niche-based and stochastic processes, including dispersal limitation and demographic stochasticity, can each contribute to community assembly, it is difficult to quantify the relative importance of each process in natural vegetation. Here, we extend Shipley's maxent model (Community Assembly by Trait Selection, CATS) for the prediction of relative abundances to incorporate both trait-based filtering and dispersal limitation from the larger landscape and develop a statistical decomposition of the proportions of the total information content of relative abundances in local communities that are attributable to trait-based filtering, dispersal limitation, and demographic stochasticity. We apply the method to tree communities in a mature, species-rich, tropical forest in French Guiana at 1-, 0.25- and 0.04-ha scales. Trait data consisted of species' means of 17 functional traits measured over both the entire meta-community and separately in each of nine 1-ha plots. Trait means calculated separately for each site always gave better predictions. There was clear evidence of trait-based filtering at all spatial scales. Trait-based filtering was the most important process at the 1-ha scale (34%), whereas demographic stochasticity was the most important at smaller scales (37-53%). Dispersal limitation from the meta-community was less important and approximately constant across scales (-9%), and there was also an unresolved association between site-specific traits and meta-community relative abundances. Our method allows one to quantify the relative importance of local niche-based and meta-community processes and demographic stochasticity during community assembly across spatial and temporal scales.     

Size-mediated non-trophic interactions and stochastic predation drive assembly and dynamics in a seabird community

Theoretical and empirical evidence suggests that body size is a major life-history trait impacting on the structure and functioning of complex food webs. However, long-term analyses of size-dependent interactions within simpler network modules, for instance, competitive guilds, are scant. Here, we model the assembly dynamics of the largest breeding seabird community in the Mediterranean basin during the last 30 years. This unique data set allowed us to test, through a "natural experiment," whether body size drove the assembly and dynamics of an ecological guild growing from very low numbers after habitat protection. Although environmental stochasticity accounted for most of community variability, the population variance explained by interspecific interactions, albeit small, decreased sharply with increasing body size. Since we found a demographic gradient along a body size continuum, in which population density and stability increase with increasing body size, the numerical effects of interspecific interactions were proportionally higher on smaller species than on larger ones. Moreover, we found that the per capita interaction coefficients were larger the higher the size ratio among competing species, but only for the set of interactions in which the species exerting the effect was greater. This provides empirical evidence for long-term asymmetric interspecific competition, which ultimately prompted the local extinction of two small species during the study period. During the assembly process stochastic predation by generalist carnivores further triggered community reorganizations and global decays in population synchrony, which disrupted the pattern of interspecific interactions. These results suggest that the major patterns detected in complex food webs can hold as well for simpler sub-modules of these networks involving non-trophic interactions, and highlight the shifting ecological processes impacting on assembling vs. asymptotic communities.     

Distinct community assembly processes underlie significant spatiotemporal dynamics of abundant and rare bacterioplankton in the Yangtze River

• Season and landform influenced spatiotemporal patterns of abundant and rare taxa. • Different stochastic processes dominated abundant and rare subcommunity assembly. • River flow and suspended solids regulated assembly processes of rare taxa. The rare microbial biosphere provides broad ecological services and resilience to various ecosystems. Nevertheless, the biogeographical patterns and assembly processes of rare bacterioplankton communities in large rivers remain uncertain. In this study, we investigated the biogeography and community assembly processes of abundant and rare bacterioplankton taxa in the Yangtze River (China) covering a distance of 4300 km. The results revealed similar spatiotemporal patterns of abundant taxa (AT) and rare taxa (RT) at both taxonomic and phylogenetic levels, and analysis of similarities revealed that RT was significantly influenced by season and landform than AT. Furthermore, RT correlated with more environmental factors than AT, whereas environmental and spatial factors explained a lower proportion of community shifts in RT than in AT. The steeper distance–decay slopes in AT indicated higher spatial turnover rates of abundant subcommunities than rare subcommunities. The null model revealed that both AT and RT were mainly governed by stochastic processes. However, dispersal limitation primarily governed the AT, whereas the undominated process accounted for a higher fraction of stochastic processes in RT. River flow and suspended solids mediated the balance between the stochastic and deterministic processes in RT. The spatiotemporal dynamics and assembly processes of total taxa were more similar as AT than RT. This study provides new insights into both significant spatiotemporal dynamics and inconsistent assembly processes of AT and RT in large rivers.     

A description of a highly diverse soft-bottom community in Kingston Harbour,Jamaica

A highly diverse soft-bottom community in Kingston Harbour, Jamaica is described to provide additional information on the structure of tropical benthic communities. It is shown that the community has a level of diversity unique to stable tropical environments, that dominance by a few species is greatly reduced,and that there are no true parallels to this community in similar environments. The community described seems to fit Sanders' stability-time hypothesis. The density and standing crop of the community are low, and possible explanations for this are discussed. Continuous year-round breeding, short life-spans, and fast turnover rates are suggested as the most important factors regulating biomass in the community described.     

Host plant quality and local adaptation determine the distribution of a gall-forming herbivore

Herein we report results of transplant experiments that link variation in host plant quality to herbivore fitness at the local scale (among adjacent plants) with the process of local (demic) adaptation at the landscape scale to explain the observed distribution of the specialist gall former Belonocnema treatae (Hymenoptera: Cynipidae) within populations of its host plant, Quercus fusiformis. Field surveys show that leaf gall densities vary by orders of magnitude among adjacent trees and that high-gall-density trees are both rare (< 5%) and patchily distributed. B. treatae from each of five high-gall-density trees were reared on (1) the four nearest low-gall-density trees, (2) the four alternative high-gall-density trees, and (3) their natal trees (control). Each treatment (source X rearing site) was replicated three times. Nine components of performance that sequentially contribute to fitness were evaluated with over 21000 galls censused across the 25 experimental trees. When reared on their natal trees and compared with low-gall-density neighbors, transplanted gall formers had higher gall initiation success (P < 0.05), produced more (P < 0.001) and larger galls (P < 0.001), and produced a higher proportion of galls that exceeded the threshold size for natural enemy avoidance (P < 0.05). Comparison of gall-former performance on natal vs. alternative high-gall-density trees demonstrated significant (P < 0.001) differences in six performance measures with five differing in the direction predicted by the hypothesis of local adaptation. Overall, these linked experiments document direct and indirect effects of host plant variation on gall-former performance and demonstrate convincingly that (1) high-gall-density trees equate to high-quality trees that are surrounded by trees of relatively lower quality to the herbivore and (2) gall-former populations have become locally adapted to individual trees.     

Disentangling the effects of heterogeneity, stochastic dynamics and sampling in a community of aquatic insects

A dynamic and heterogeneous species abundance model generating the lognormal species abundance distribution is fitted to time series of species data from an assemblage of stoneflies and mayflies (Plecoptera and Ephemeroptera) of an aquatic insect community collected over a period of 15 years. In each year except one, we analyze 5 parallel samples taken at the same time of the season giving information about the over-dispersion in the sampling relative to the Poisson distribution. Results are derived from a correlation analysis, where the correlation in the bivariate normal distribution of log abundance is used as measurement of similarity between communities. The analysis enables decomposition of the variance of the lognormal species abundance distribution into three components due to heterogeneity among species, stochastic dynamics driven by environmental noise, and over-dispersion in sampling, accounting for 62.9, 30.6 and 6.5% of the total variance, respectively. Corrected for sampling the heterogeneity and stochastic components accordingly account for 67.3 and 32.7% of the among species variance in log abundance. By using this method, it is possible to disentangle the effect of heterogeneity and stochastic dynamics by quantifying these components and correctly remove sampling effects on the observed species abundance distribution.     

Asymmetry in community regulation: effects of predators and productivity

Eutrophication and predator additions and extinctions are occurring in ecosystems worldwide. Although theory predicts that both will strongly alter the distribution of biomass in whole communities, empirical evidence has not been consolidated to quantitatively determine whether these theoretical predictions are generally borne out in real ecosystems. Here we analyze data from two types of trophic cascade studies, predator removals in factorial combination with fertilization and observed productivity gradients, to assess the role of top-down and bottom-up forces in structuring multi-trophic communities and compare results from these analyses to those from an extensive database of trophic cascade studies. We find that herbivore biomass declines and plant biomass increases in the presence of predators, regardless of system productivity. In contrast, while plants are increased by fertilization, this effect does not significantly increase herbivores in either the presence or absence of predators. These patterns are consistent among marine, freshwater, and terrestrial ecosystems and are largely independent of study size and duration. Thus, top-down effects of predation are transferred through more trophic levels than are bottom-up effects of eutrophication, showing strong asymmetry in the direction of control of biomass distribution in communities.     

Maintenance of a diverse harpacticoid copepod community in microcosm culture

The design and construction of a closed system circulating sea-water tidal estuarine sediment microcosm is described. The microcosm was used to culture mixed species populations of the Harpacticoida (Crustacea: Copepoda) and the evolution of the community and the successional changes are detailed. The communities that developed in each of four replicate microcosms were statistically indistinguishable with respect to the species richness, the numbers of individuals recorded, the log series diversity and the dominance diversity statistics. The most abundant species were the same in each microcosm with the same rank. A diverse community was maintained that did not tend to monoculture; however, the successional changes observed were not apparent in the estuarine communities from which the culture sediments was collected. Rather than a functional reorganisation of the community effecting this, gross environmental changes resulting from a decreased depth of sediment matrix are implicated. The success in maintaining a diverse community is attributed to a provision of a number of nutrient sources.     

Forest age influences oak insect herbivore community structure, richness, and density.

Plant succession is one of many factors that may affect the composition and structure of herbivorous insect communities. However, few studies have examined the effect of forest age on the diversity and abundance of insect communities. If forest age influences insect diversity, then the schedule of timber harvest rotation may have consequent effects on biodiversity. The insect herbivore community on Quercus alba (white oak) in the Missouri Ozarks was sampled in a chronoseries, from recently harvested (2 yr) to old-growth (approximately 313 yr) forests. A total of nine sites and 39 stands within those sites were sampled in May and August 2003. Unique communities of plants and insects were found in the oldest forests (122-313 yr). Density and species richness of herbivores were positively correlated with increasing forest age in August but not in May. August insect density was negatively correlated with heat load index; in addition, insect density and richness increased over the chronoseries, but not on the sunniest slopes. Forest structural diversity (number of size classes) was positively correlated with forest age, but woody plant species richness was not. In sum, richness, density, and community structure of white oak insect herbivores are influenced by variation in forest age, forest structure, relative abundance of plant species, and abiotic conditions. These results suggest that time between harvests of large, long-lived, tree species such as white oak should be longer than current practice in order to maintain insect community diversity.     

The role of asteroid predators in the organization of a sessile community on pier pilings

The diets of 4 species of asteroids were observed for `9 months on the piles of a pier at Rapid Bay, South Australia.Coscinasterias calamaria (Gray) fed on molluscs, crustaceans and moribund items.Patiriella brevispina H.L. Clark fed on compound ascidians, sponges and moribund items.Tosia australis Gray andPetricia vernicina (Lamarck) ate mainly sponges, bryozoans and compound ascidians. A short-term (6 months) caging experiment showed that 5 common species of sessile animals-3 sponges and 2 compound ascidians-were significantly reduced by seastar predation. Nevertheless, the amount eaten, considered with seastar densities and distributions, indicates that the seastars are unimportant in influencing the utilization of space by sessile fauna. From comparing caged and uncaged controls, we tentatively suggest that other predators are also unimportant, and that competition for substrate space and for access to the water column is (and, has been during the evolution of these sessile organisms) far more influential.     

Demands of the herbivore community on phytoplankton production in the Celtic Sea in August

The apodous detritivore holothurian Leptosynapta galliennei (Herapath, 1865) was collected from the intertidal zone (Brittany, France) in spring 1983, and the effect of short-term starvation on the biochemical composition of different body parts was studied. Variations in the levels of the main organic components of the somatic tissues did not correspond to the observed ponderal growth of the gonad in either males or females. Part of the increase in the energetic value of the testis was due to carbohydrates. Accumulation of carbohydrates, proteins, and especially lipids occurred in the ovary. Glycogen, often considered as the most readily mobilized energy source, varied little in concentration or in quantity in any body part. Lipids were not mobilized. No organ of this thin-skinned holothurian functions as a site of rapidly usable food reserves. The organic nitrogen and carbon necessary to compensate for metabolic loss and to permit gonadal growth during shortterm starvation may be provided by dissolved organic matter. An efficient transintegumentary uptake of dissolved organic matter, perhaps mediated by associated subcuticular bacteria, and/or intestinal uptake are hypothesized.     

The relative importance of host-plant genetic diversity in structuring the associated herbivore community

Recent studies suggest that intraspecific genetic diversity in one species may leave a substantial imprint on the surrounding community and ecosystem. Here, we test the hypothesis that genetic diversity within host-plant patches translates into consistent and ecologically important changes in the associated herbivore community. More specifically, we use potted, grafted oak saplings to construct 41 patches of four saplings each, with one, two, or four tree genotypes represented among the host plants. These patches were divided among two common gardens. Focusing first at the level of individual trees, we assess how tree-specific genotypic identity, patch-level genetic diversity, garden-level environmental variation, and their interactions affect the structure of the herbivore community. At the level of host-plant patches, we analyze whether the joint responses of herbivore species to environmental variation and genetic diversity result in differences in species diversity among tree quartets. Strikingly, both species-specific abundances and species diversity varied substantially among host-tree genotypes, among common gardens, and among specific locations within individual gardens. In contrast, the genetic diversity of the patch left a detectable imprint on local abundances of only two herbivore taxa. In both cases, the effect of genetic diversity was inconsistent among gardens and among host-plant genotypes. While the insect community differed significantly among individual host-plant genotypes, there were no interactive effects of the number of different genotypes within the patch. Overall, additive effects of intraspecific genetic diversity of the host plant explained a similar or lower proportion (7-10%) of variation in herbivore species diversity than did variation among common gardens. Combined with the few previous studies published to date, our study suggests that the impact of host-plant genetic diversity on the herbivore community can range from none to nonadditive, is generally low, and reaches its most pronounced impact at small spatial scales. Overall, our findings strengthen the emerging view that the impacts of genetic diversity are system, scale, and context dependent. As the next step in community genetics, we should then start asking not only whether genetic diversity matters, but under what circumstances its imprint is accentuated.     

Planktic community assembly in flowing water and the ecosystem health of rivers

Some of the modern criteria for assessing ecosystem health are compared with current understanding of ecosystem function in rivers. Owing to the predominance of catchment imports over autochthonous primary production, most rivers are naturally heterotrophic. This does not make them unhealthy but the pristine condition is that much harder to determine. The case is put for an index of ecosystem health and sustainability that takes into account the system's capacity for processing its resources, the species richness and its interdependence and its resilience to external forcing. Although these are not easily quantified, the qualitative indicators of healthy ecosystem function are easily checked. The sensitivity of organisms in suspension to fluvial flow may seem to counter the suitability of plankton as a reliable state indicator of river health. On the other hand, the rules governing the assembly of planktic communities in rivers are often strict and quantifiable: this makes them attractive candidates to act as indicators of the ecological condition of rivers.     

Differences in herbivore preferences, phlorotannin production, and nutritional quality between juvenile and adult tissues from marine brown algae

 Juvenile and adult marine organisms differ in their morphology, chemistry, physiology, behavior, and ecology. Because juvenile algae are thinner, smaller, and have more delicate tissues than adults, they are often assumed to be more susceptible to grazers. We examined within-species food preferences of four common generalist herbivores for juvenile and adult tissues of eight common brown algae in two-choice laboratory food-preference experiments. Our results showed that juvenile algae did not tend to be a preferred food of herbivores. Juvenile tissues were significantly preferred over adult tissues in only four of the 32 combinations of algae and herbivores tested. In 12 experiments, adult tissues were preferred over juvenile tissues, and no choice occurred in the remaining 16 experiments. When sea urchins exhibited a preference, it was always for adult tissues. The other three herbivores, an isopod and two snails, were more variable in their choices, sometimes preferring juveniles, sometimes adults, and sometimes having no preference. We measured nitrogen and phlorotannin concentrations in adult and juvenile seaweeds to see whether these parameters were correlated with herbivore food preferences. Nitrogen levels were similar in juveniles and adults of three algal species and were higher in juveniles of two. Phlorotannin levels were higher in juveniles of four species and lower in juveniles of one. The other three species showed no differences in phlorotannin levels. Phlorotannin concentrations decreased with increasing juvenile size in three species and increased with increasing size in one species. Neither nitrogen nor phlorotannin concentrations explained overall herbivore food preferences for algae of different stages. Our results suggest that preferences of certain grazers for juvenile algae are not as strong as previously assumed and are dependent on herbivore species. Preferences between juveniles and adults are likely to be determined by a combination of morphological and chemical features of the tissues and the unique responses of herbivore species to those features. Received: 10 April 2000 / Accepted: 19 November 2000     

Diel trophic interactions between vertically-migrating zooplankton and their fish predators in an eelgrass community

Diel changes in the composition of crustacean zooplankton and the diets of fish predators from an intertidal eelgrass flat were monitored concurrently. The zooplankton is characterized by two major components. The obligate zooplankters (holoplanktonic calanoid copepods and meroplanktonic decapod larvae) appear to exhibit vertical migration, being present in higher densities near the surface of the water column at night. The facultative zooplankton (amphipods and ostracods) are benthic during the day, but move up into the water column at night. Planktivorous midwaterdwelling fish consume calanoid copepods and decapod larvae during the day and cease feeding or switch their diet to amphipods at night. Benthic-dwelling fish consume some amphipods during both day and night. The factors important in prey selection by fish and the functional significance of vertical migration in both components of the zooplankton are discussed in the light of the changing patterns of fish predation.This paper is Publication No. 183 in the Ministry for Conservation of Victoria, Environmental Studies Series.     

Plant-mediated and nonadditive effects of two global change drivers on an insect herbivore community

Warmer temperatures can alter the phenology and distribution of individual species. However, differences across species may blur community-level phenological responses to climate or cause biotic homogenization by consistently favoring certain taxa. Additionally, the response of insect communities to climate will be subject to plant-mediated effects, which may or may not overshadow the direct effect of rising temperatures on insects. Finally, recent evidence for the importance of interaction effects between global change drivers suggests that phenological responses of communities to climate may be altered by other drivers. We used a natural temperature gradient (generated by elevation and topology), combined with experimental nitrogen fertilization, to investigate the effects of elevated temperature and globally increasing anthropogenic nitrogen deposition on the structure and phenology of a seminatural grassland herbivore assemblage (lepidopteran insects). We found that both drivers, alone and in combination, severely altered how the relative abundance and composition of species changed through time. Importantly, warmer temperatures were associated with biotic homogenization, such that herbivore assemblages in the warmest plots had more similar species composition than those in intermediate or cool plots. Changes in herbivore composition and abundance were largely mediated by changes in the plant community, with increased nonnative grass cover under high treatment levels being the strongest determinant of herbivore abundance. In addition to compositional changes, total herbivore biomass more than doubled under elevated nitrogen and increased more than fourfold with temperature, bearing important functional implications for herbivores as consumers and as a prey resource. The crucial role of nonnative plant dominance in mediating responses of herbivores to change, combined with the frequent nonadditive (positive and negative) effects of the two drivers, and the differential responses of species, highlight that understanding complex ecosystem responses will benefit from multifactor, multitrophic experiments at community scales or larger.     

A strong test of a maximum entropy model of trait-based community assembly

We evaluate the predictive power and generality of Shipley's maximum entropy (maxent) model of community assembly in the context of 96 quadrats over a 120-km2 area having a large (79) species pool and strong gradients. Quadrats were sampled in the herbaceous understory of ponderosa pine forests in the Coconino National Forest, Arizona, U.S.A. The maxent model accurately predicted species relative abundances when observed community-weighted mean trait values were used as model constraints. Although only 53% of the variation in observed relative abundances was associated with a combination of 12 environmental variables, the maxent model based only on the environmental variables provided highly significant predictive ability, accounting for 72% of the variation that was possible given these environmental variables. This predictive ability largely surpassed that of nonmetric multidimensional scaling (NMDS) or detrended correspondence analysis (DCA) ordinations. Using cross-validation with 1000 independent runs, the median correlation between observed and predicted relative abundances was 0.560 (the 2.5% and 97.5% quantiles were 0.045 and 0.825). The qualitative predictions of the model were also noteworthy: dominant species were correctly identified in 53% of the quadrats, 83% of rare species were correctly predicted to have a relative abundance of < 0.05, and the median predicted relative abundance of species actually absent from a quadrat was 5 x 10(-5).     

Priority effects and species sorting in a long paleoecological record of repeated community assembly through time

We studied the relative roles of environmental species sorting and priority effects in the assembly of ecological communities on long time scales, by analyzing community turnover of water fleas (Daphnia) in response to strong and recurrent environmental change in a fluctuating tropical lake. During the past 1800 years, Lake Naivasha (Kenya) repeatedly fluctuated between a small saline pond habitat during lowstands and a large freshwater lake habitat during highstands. Starting from a paleoecological reconstruction, we estimated the role of priority effects in Daphnia community assembly across 16 of these habitat turnovers and compared this with the response of the community to reconstructed changes in three environmental variables important for species sorting. Our results indicate that the best predictor of Daphnia community composition during highstands was the community composition just prior to the transition from lowstands to highstands. This reflects a long-lasting priority effect of late lowstand communities on highstand communities, arising when remnant lowstand populations fill newly available ecological space in the rapidly expanding lake habitat. Species sorting and priority effects had a comparable but relatively small influence on community composition during the lowstands. Moreover, these priority effects decayed rapidly with time as Daphnia communities responded to environmental change, in contrast with the highstand communities where priority effects lasted for several decades.     

Lifetime reproductive success and composition of the home range in a large herbivore

The relationship between individual performance and nonrandom use of habitat is fundamental to ecology; however, empirical tests of this relationship remain limited, especially for higher orders of selection like that of the home range. We quantified the association between lifetime reproductive success (LRS) and variables describing lifetime home ranges during the period of maternal care (spring to autumn) for 77 female roe deer (Capreolus capreolus) at Trois-Fontaines, Champagne-Ardenne, France (1976-2000). We maintained population growth rate (adjusted to account for removals of non-focal animals) near rmax, which enabled us to define the fitness-habitat relationship in the absence of density effects. Using a negative binomial model, we showed that a roe deer's incorporation into its home range of habitat components important to food, cover, and edge (meadows, thickets, and increased density of road allowances) was significantly related to LRS. Further, LRS decreased with increasing age of naturally reclaimed meadows at the time of a deer's birth, which may have reflected a cohort effect related to, but not entirely explained by, a decline in quality of meadows through time. Predictive capacity of the selected model, estimated as the median correlation (rs) between predicted and observed LRS among deer of cross-validation samples, was 0.55. The strength of this relationship suggests that processes like selection of the site of a home range during dispersal may play a more important role in determining fitness of individuals than previously thought. Individual fitness of highly sedentary income breeders with high reproductive output such as roe deer should be more dependent on home range quality during the period of maternal care compared to capital breeders with low reproductive output. Identification of the most important habitat attributes to survival and reproduction at low density (low levels of intraspecific competition) may prove useful for defining habitat value ("intrinsic habitat value").     

Systemically-induced response of cabbage plants against a specialist herbivore, Pieris brassicae

Summary. Plant responses to herbivory might directly affect the herbivore (“direct” defences) or might benefit the plant by promoting the effectiveness of natural antagonists of the herbivores (“indirect” defences). Brussels sprouts attacked by Pieris brassicae larvae release volatiles that attract a natural antagonist of the herbivores, the parasitoid Cotesia glomerata, to the damaged plant. In a previous study, we observed that feeding by caterpillars on the lower leaves of the plant triggers the systemic release of volatiles detectable by the parasitoids from upper leaves of the same plant.?The role of these systemically induced volatiles as indirect defence and the dynamics of their emission were investigated in wind-tunnel dual choice tests with C. glomerata. The systemically induced emission of volatiles varied depending on leaf age and on plant age. Systemic induction affected parasitoid effectiveness, as induced plants could be more easily located by parasitoids than non-induced ones.?The role of the systemic induction as a direct defence was investigated through behavioural and feeding tests with P. brassicae. In dual choice assays, 1st instar larvae preferred to feed and fed more on systemically induced than on non-induced leaves. In single choice assays, the leaf area consumed by caterpillars was larger on systemically induced leaves than on non-induced control leaves. However, caterpillars fed on systemically induced leaves attained the same weight as those feeding on non-induced controls. In addition, P. brassicae pupae whose larvae were fed on systemically induced leaves had longer developmental times than those of larvae fed on non-induced leaves. Adult oviposition behavior was not influenced by systemic induction.?We conclude that systemically induced responses in cabbage might reduce P. brassicae fitness both directly, by affecting their development and feeding behavior and indirectly by making caterpillars and pupae more vulnerable to attack by carnivores. The occurrence of a possible relationship between direct and indirect defence is discussed. Received 24 January 2001; accepted 3 May 2001.