Portal de Periódicos da CAPES

Alternative community compositional and dynamical states: the dual consequences of assembly history

2011; Wiley; Volume: 80; Issue: 3 Linguagem: Inglês

10.1111/j.1365-2656.2010.01799.x

1365-2656

Lin Jiang, Hena Joshi, Sooyung K. Flakes, Yeonjin Jung,

Evolutionary Game Theory and Cooperation

Journal of Animal EcologyVolume 80, Issue 3 p. 577-585 Free Access Alternative community compositional and dynamical states: the dual consequences of assembly history Lin Jiang, Corresponding Author Lin Jiang Correspondence author. E-mail: lin.jiang@biology.gatech.eduSearch for more papers by this authorHena Joshi, Hena Joshi School of Biology, Georgia Institute of Technology, 310 Ferst Drive, Atlanta, GA 30332, USASearch for more papers by this authorSooyung K. Flakes, Sooyung K. Flakes School of Biology, Georgia Institute of Technology, 310 Ferst Drive, Atlanta, GA 30332, USASearch for more papers by this authorYeonjin Jung, Yeonjin Jung School of Biology, Georgia Institute of Technology, 310 Ferst Drive, Atlanta, GA 30332, USASearch for more papers by this author Lin Jiang, Corresponding Author Lin Jiang Correspondence author. E-mail: lin.jiang@biology.gatech.eduSearch for more papers by this authorHena Joshi, Hena Joshi School of Biology, Georgia Institute of Technology, 310 Ferst Drive, Atlanta, GA 30332, USASearch for more papers by this authorSooyung K. Flakes, Sooyung K. Flakes School of Biology, Georgia Institute of Technology, 310 Ferst Drive, Atlanta, GA 30332, USASearch for more papers by this authorYeonjin Jung, Yeonjin Jung School of Biology, Georgia Institute of Technology, 310 Ferst Drive, Atlanta, GA 30332, USASearch for more papers by this author First published: 12 January 2011 https://doi.org/10.1111/j.1365-2656.2010.01799.xCitations: 19AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinked InRedditWechat Summary 1. Much work on ecological consequences of community assembly history has focused on the formation of history-induced alternative stable equilibria. We hypothesize that assembly history may affect not only community composition but also population dynamics, with assembled communities differing in species composition potentially residing in different dynamical states. 2. We provided an empirical test of the aforementioned hypothesis using a laboratory microcosm experiment that manipulated both the colonization order of three bacterivorous protist species in the presence of a protist predator and environmental productivity. 3. Both priority effects and random divergence emerged, resulting in two different community compositional states: one characterized by the dominance of one prey species and the other by the extinction of the same prey. While communities in the former state exhibited noncyclic dynamics, the majority of communities in the latter state exhibited cyclic dynamics driven by the interaction between another prey and the predator. 4. Temporal variability of total prey community biovolume consequently differed among communities with different histories. 5. Changing productivity altered priority effects on the structure and dynamics of communities experiencing only certain histories. 6. Our results support the dual (compositional and dynamical) consequences of assembly history and emphasize the importance of incorporating the dynamical view into the field of community assembly. Introduction Ecologists have long recognized the stochastic nature of species colonization during community assembly and the potential significance of resulting differences in assembly history for ecological communities (Gleason 1927; Sutherland 1974; Diamond 1975). Nevertheless, ecological consequences of assembly history are still not fully understood. One aspect of historical effects, concerning its role in shaping community structure, has been the major focus of past and current assembly research. This research has revealed that variation in the order and timing of species arrivals can result in divergent communities differing in species composition and abundance (reviewed by Samuels & Drake 1997; Chase 2003a), often labelled as alternative stable equilibria (sensuLewontin 1969). Further work on this topic seeks to understand mechanisms underlying variation in the ability of history in producing such divergent communities (e.g. Drake 1991; Chase 2003a, 2007, 2010; Fukami 2004; Ejrnaes, Bruun & Graae 2006; Louette & De Meester 2007; Jiang & Patel 2008; Chase et al. 2009). However, although these investigations have produced valuable insights into the structuring role of assembly history, their focus on the formation of alternative stable equilibria and therefore assumption of stable species composition and abundance at the end of assembly mean that dynamical consequences of assembly history, if any, may have been overlooked. In contrast with the equilibrium view adopted by many assembly studies, ecological communities frequently exhibit more complex dynamics (Turchin 2003). In particular, populations of a substantial number of species exhibit cyclic dynamics in nature (Kendall, Prendergast & Bjornstad 1998), and much effort has been dedicated to elucidating causes of these population cycles and sources of variation in the cyclic patterns (e.g. Turchin 2003). Here, we suggest that population dynamics in local communities sharing the same regional species pool may be influenced by their assembly history. The sensitivity of population dynamics to assembly history has been demonstrated in an empirical system comprising an insect host and two of its natural enemies (a parasitoid wasp and a viral pathogen) (Sait et al. 2000). However, the observed alternative dynamical states (cf. Ives et al. 2008; see also Schroder, Persson & De Roos 2005 for a discussion on the related concept of alternative attractors) associated with different introduction orders of the two consumer species were characterized by the same species composition with all three species present and linked to the peculiar biology of the species involved. History could more likely impact population dynamics when communities undergoing different assembly sequences attain different species compositions but not necessarily alternative stable equilibria. In such cases, variation in assembly history may lead to alternative dynamical states where the assembled communities exhibit different dynamical behaviours, with the type of dynamics in each community determined by the interactions of its member species. It is notable that a significant number of theoretical (e.g. Hastings & Costantino 1987; Briggs, Nisbet & Murdoch 1999; Boer, Kooi & Kooijman 2001; De Roos & Persson 2003; Wearing et al. 2004) and experimental (Henson et al. 1999, 2002; McCauley et al. 1999; Zamamiri, Birol & Hjortso 2001; Ives et al. 2008) studies have documented alternative dynamic states, involving communities comprising either the same or different species composition; however, as noted by Sait et al. (2000), none of these studies have considered communities assembled through different species colonization histories. In addition, alternative dynamical states are only implicitly assumed, but rarely directly addressed, in assembly models based on the identification of community permanence states (sensuLaw & Morton 1993). Given the frequent dependence of community composition on assembly history found in empirical systems (Samuels & Drake 1997; Chase 2003a), alternative dynamical states that accompany history-induced alternative compositional states could be potentially common. This dual consequence of assembly history, however, has not been empirically demonstrated. Productivity has been recognized as a key force shaping a variety of community attributes (Polis & Strong 1996) and is also thought to influence community assembly. Both conceptual models (Chase 2003a) and a host of mathematical models, such as those depicting apparent competition (Holt, Grover & Tilman 1994; Chase & Leibold 2003), intraguild predation (Holt & Polis 1997; Diehl & Feissel 2000) and changes in prey vulnerability through ontology (Chase 1999; De Roos & Persson 2002), have suggested that the likelihood of alternative compositional states varies among habitats with different productivities. The predicted role of productivity in modulating community assembly, however, has not been extensively tested (but see Chase 2003b, 2010; Ejrnaes, Bruun & Graae 2006). Further, none of the studies on this topic have considered possible dynamical consequences of assembly history, and how productivity influences the ability of assembly history in generating alternative dynamical states remains an open question. We conducted a laboratory protist microcosm experiment, in which we manipulated both species colonization history and productivity, to test the hypothesis that communities experiencing different histories may attain alternative community states that differ in both species composition and population dynamics. Our experiment used a simple multitrophic system containing one predatory protist species and three bacterivorous protist species as its prey, reflecting trophic interactions as common components of both assembly models and natural communities. The use of fast-growing protists (generation time <1 day for each species), coupled with the relatively long duration of the experiment (over 80 days), allowed the collection of community dynamics data over many generations that are essential for distinguishing different dynamical patterns, and at the same time minimized transient dynamics that may complicate the interpretation of alternative community states (Connell & Sousa 1983). Materials and methods Each microcosm consisted of a 250-mL glass bottle containing 100 mL of aqueous medium, which provided substrates and nutrients for the growth of bacteria and higher trophic level protists, plus wheat seeds that acted as an agent of slow carbon release. Bottles, medium and wheat seeds were autoclaved before the medium was inoculated with three bacterial species (Serratia marcescens Bizio, Bacillus cereus Frankland & Frankland and Bacillus subtilis [Ehrenberg] Cohn), which served as food resources for the bacterivorous protists. Introduction of the first protist species took place 24 h after bacterial inoculation, after bacteria had grown to appreciable abundance. Our experiment used a two-way factorial design with two levels of productivity (low vs. high) crossed with six assembly sequences involving the three bacterivorous protist species, each replicated three times. As in previous studies (e.g. Jiang & Krumins 2006), we manipulated productivity by controlling medium concentration. Medium in the low- and high-productivity treatments contained 0·2 and 0·6 g protozoan pellet (Carolina Biological Supply, Burlington, NC, USA) per litre of deionized water, respectively. Low- and high-productivity microcosms also received one and three wheat seeds, respectively. Previous work has shown that this difference in productivity led to approximately twofold difference in bacterial abundance and 10-fold difference in bacterivore biomass (Jiang & Krumins 2006). Protist communities in our experiment comprised one predatory ciliate, Dileptus monilatus Stokes, and three bacterivorous ciliates Colpidium striatum Stokes, Halteria grandinella Muller and Tetrahymena pyriformis Ehrenberg as its prey. In stock cultures, we grew the three prey species separately on the same three bacterial species used in our experiment and grew Dileptus on a mixture of the three prey species. Community assembly involved the sequential colonization of microcosms by the three prey species in all possible orders, while keeping the timing of predator colonization constant (see Olito & Fukami 2009 for a protist microcosm study of predator timing effects). This resulted in six assembly sequences [CHT (Colpidium first, Halteria second and Tetrahymena last), CTH, HCT, HTC, TCH and THC]. The three prey species were introduced into microcosms at weekly intervals, while Dileptus was introduced 2 days after the first prey introduction. The size of the inoculum was c. 200 individuals for each prey species, and 20 individuals for Dileptus, to reflect the typically lower abundances of predators than their prey. These individuals were all taken from stock cultures of the same age (7 day old) to minimize variation in cell physiological states between inoculation events. Microcosms were maintained in an unlighted incubator at 22 °C. Starting from 3 days after predator introduction (designated as day 0), we sampled microcosms every 2–3 days to estimate population density of each protist species. Sampling involved withdrawing a small sample (0·3–0·4 mL) from each microcosm, measuring the weight (≈ volume) of the sample on an analytic balance and enumerating the number of individuals of each species under a dissecting microscope. When necessary, we diluted samples that contained dense populations before counting. Sampling continued until day 79, and during this period, partial (7% of the volume) medium replacement was conducted weekly on each microcosm. We recorded population densities as the number of individuals per ml. We also obtained total prey community biovolume by summing population biovolume across all prey species; population biovolume of each species was calculated by multiplying its population density by its cell size, with the latter drawn from our laboratory data base. Data on population density and community biovolume were log-transformed (log10[x + 1]) prior to analysis. We use three complementary methods to assess the likelihood of alternative compositional states among the assembled communities, based on the data on prey community structure (i.e. the abundances of the three prey species) averaged over the last ten sampling dates. Our analyses focused on prey communities that have been subjected to different histories; results remained qualitatively the same when predator abundance data were also considered. We first performed hierarchical clustering on prey community structure, with the understanding that communities grouped into different major clusters may be indicative of alternative compositional states. The clustering procedure was based on Ward's minimum-variance method, and the number of major clusters was determined by inspecting changes in cubic clustering criterion, pseudo F and t2 statistics as clusters joined together. Although cluster analysis is effective at identifying the presence and number of potential alternative compositional states, it was not straightforward to quantify the effects of assembly history and productivity solely based on clustering results. We thus ran two-way manova to directly assess how assembly history, productivity and their interactions affected prey community structure. As another way of assessing the productivity effect, we compared between-community similarities, measured by the abundance-based Bray–Curtis similarity index (Bray & Curtis 1957), between the two productivity levels using a randomization test (based on 1000 permutations). The Bray–Curtis index ranges from 0 (two communities share no species in common) to 1 (two communities with identical species composition and abundances); low values may implicate the presence of alternative compositional states. The randomization test was chosen over a traditional t-test because pairwise similarity values involving the same communities were not independent, violating the assumption of the latter test. We used three methods to assess the likelihood of alternative dynamical states among the assembled communities. First, we conducted spectral analysis on the time series of populations of the three prey species to distinguish cyclic from noncyclic dynamics at the population level. Cyclic fluctuations are indicated by the presence of large-amplitude spectral densities bordered by low-amplitude ones at surrounding frequencies, whereas the absence of such spikes in spectral densities indicates noncyclic dynamics (Bjornstad et al. 1996). Second, we used Fisher's exact test to assess the separate effect of assembly history and productivity on the frequency of cyclic dynamics. We originally performed logistic regression, which also allowed the test of the interaction terms, for the same purpose; the presence of quasi-complete separation in the data, however, prevented maximum likelihood parameter estimation in this analysis. Third, we ran two-way anova, followed by a Tukey's HSD test, to assess the effects of assembly history and productivity on prey temporal variability at the community level, measured as the standard deviation of log-transformed total prey community biovolume over the last 60 days of the experiment. This metric of temporal variability conveys similar information as the coefficient of variation (CV) of untransformed data, but tends to be less influenced by skewed distributions (McArdle, Gaston & Lawton 1990); results based on the two measures were nevertheless similar. Results Differences in assembly history resulted in marked difference in species composition and abundances among the assembled communities. The cluster analysis grouped these communities into two large clusters (Fig. 1): one in which Colpidium was numerically dominant (Fig. 2a–d; Table 1) and one in which Colpidium went extinct (Fig. 2e,f; Table 1). This grouping, alone, accounted for 66·3% of the variance, suggesting the presence of two alternative compositional states. In the cluster of communities dominated by Colpidium, the predator Dileptus attained low densities and persisted until the end of the experiment in some, but not all, microcosms (Fig. 2a–d; Table 1). In the cluster characterized by Colpidium extinction, the majority of communities were composed of Tetrahymena coexisting with Dileptus, which almost always persisted until the end of the experiment and reached substantial peak densities (Fig. 2e,f; Table 1). Three communities in this cluster collapsed when Dileptus eliminated all three prey species before going extinct itself (Table 1). Figure 1Open in figure viewerPowerPoint The result of Ward's minimum-variance hierarchical clustering of the assembled communities, based on the abundances of the three prey species averaged over the last 10 sampling dates. The two major clusters account for 66·3% of the variance in the data. Low: low productivity, high: high productivity. Numbers after treatments indicate replicate numbers. For example, CHT_low1 corresponds to replicate 1 of the low-productivity CHT treatment. Figure 2Open in figure viewerPowerPoint Exemplary population dynamics of protist species in selected treatments. Each panel corresponds to one replicate. Population density was recorded as the number of individuals per ml and plotted on the log scale. Table 1. Final species composition and the type of population dynamics (cyclic vs. noncyclic) in each of the assembled communities. In the species composition column, species are ranked from high to low abundance; species composition for communities in which Colpidium went extinct is marked in boldface. Cyclic dynamics are underlined in the last column Assembly history Productivity Replicate Species composition Cyclic dynamics CHT Low 1 C, H, T, D No CHT Low 2 C, H, T No CHT Low 3 C, H, T No CHT High 1 C, H, T No CHT High 2 C, H, T No CHT High 3 C, H, T No CTH Low 1 C, H, D No CTH Low 2 C, H, T, D No CTH Low 3 C, H, T, D No CTH High 1 C, H, T No CTH High 2 C, T, H No CTH High 3 C, T, H No HCT Low 1 C, H No HCT Low 2 C, H, T No HCT Low 3 C, T, D No HCT High 1 All extinct No HCT High 2 T, D No HCT High 3 All extinct No HTC Low 1 T, D T: yes HTC Low 2 T, D T: yes HTC Low 3 T, D T: yes HTC High 1 C, H No HTC High 2 C, H No HTC High 3 All extinct No TCH Low 1 T, D No TCH Low 2 T, D T: yes TCH Low 3 T, D T: yes TCH High 1 T, D T: yes TCH High 2 T, D T: yes TCH High 3 T,D T: yes THC Low 1 T, D No THC Low 2 C, T, D T: yes THC Low 3 C, T T: yes THC High 1 T, D T: yes, C: yes THC High 2 T, D T: yes THC High 3 T, D T: yes Priority effects contributed predominantly to the aforementioned pattern, resulting in a significant history effect in the manova on prey community structure (Wilk's λ = 0·07, F15, 61·134 = 6·61, P < 0·0001). Colpidium dominated in all microcosms in which it was introduced first (CHT and CTH treatments; Fig. 2a,b; Table 1), and Tetrahymena survived as the only prey species in all but two microcosms in which it was introduced first (TCH and THC treatments; Fig. 2f; Table 1). No Halteria dominance, however, was observed in microcosms in which Halteria was the first colonizer (HCT and HTC treatments). Instead, Colpidium dominance, Tetrahymena dominance or the collapse of the whole system was observed (Fig. 2c–e; Table 1). Here, priority effects manifested through the second colonizing species: Colpidium dominated in the low-productivity HCT treatment (Fig. 2c,d; Table 1) and Tetrahymena persisted as the only prey in the low-productivity HTC treatments (Fig. 2e; Table 1). These secondary priority effects, however, disappeared at the higher productivity in both HCT and HTC treatments, resulting in a significant history and productivity interaction term in the manova (Wilk's λ = 0·225, F15, 61·134 = 2·92, P = 0·0016). The elimination of priority effects at higher productivity in those treatments (i.e. HCT and HTC), combined with strong priority effects at both productivity levels in most other treatments (i.e. CHT, CTH, and TCH), resulted in a nonsignificant main productivity effect in the manova (Wilk's λ = 0·80, F3, 22 = 1·80, P = 0·1761). Moreover, similarity among communities subject to different histories, represented by the Bray–Curtis index, did not differ between the two productivity levels (randomization test: P = 0·199); the index values were low at both productivity levels (mean ± SE: 0·2551 ± 0·0255 for low productivity, 0·2247 ± 0·0258 for high productivity). Note that although results were highly consistent across replicates in treatments exhibiting priority effects, random divergence among replicates was observed for treatments (HCT and HTC at high productivity and THC at low productivity) that exhibited no priority effects. Random divergence, however, did not result in novel community types beyond the two community clusters. For example, Tetrahymena was the sole surviving prey in one replicate in the low-productivity THC treatment, whereas it was subordinate to Colpidium in the other two replicates (Table 1). Community dynamics differed among the assembled communities characterized by different species compositions. Noncyclic dynamics were prevalent in communities dominated by Colpidium; spectral analysis detected only two cases of cyclic dynamics, for Tetrahymena in two replicates (replicates 2 and 3) of the low-productivity THC treatment where the species underwent small-amplitude population fluctuation (Table 1). Cyclic dynamics were common in communities with Tetrahymena as the only surviving prey (11 of 14 cases; Table 1), where the species exhibited large-amplitude population cycles driven by its interaction with Dileptus (e.g. Fig. 2e,f). In one replicate (replicate 1) of the high-productivity THC treatment, Colpidium populations also exhibited cyclic fluctuations, apparently caused by its interaction with Dileptus, before being driven to extinction (Table 1). Fisher's exact tests revealed significant differences between assembly history treatments (P < 0·0001), but not between the two productivity treatments (P = 1·0000), in the frequency of cyclic dynamics. The differences in population dynamics translated into significant differences in the temporal variability of total prey community biovolume (Fig. 3). Similar to the manova results on community structure, anova on prey community temporal variability revealed a significant effect of assembly history (F5, 24 = 26·95, P < 0·0001), a nonsignificant effect of productivity (F1, 24 = 2·34, P = 0·1396) and a significant interaction term between the two (F5, 24 = 14·79, P < 0·0001). Figure 3Open in figure viewerPowerPoint Temporal variability of total prey community biovolume in communities with different assembly histories and productivities. Temporal variability was measured as log-transformed standard deviation of total prey community biovolume. Treatments sharing the same letter are not significantly different from each other in a Tukey's HSD test. The numbers above each bar represent the number of replicates exhibiting cyclic dynamics, out of the number of all replicates within each treatment. For instance, 0/3 for the low-productivity CHT treatment indicates none of the three replicates in this treatment showed cyclic dynamics. Discussion Our results clearly support the hypothesis that alternative dynamical states may accompany alternative compositional states that arise from differences in assembly history. Two major classes of communities, differing in both community structure and dynamics, emerged in our experiment. The first class of communities was characterized by the dominance of Colpidium and the lack of cyclic dynamics, reflecting small impacts from the predator that never attained large population size in these communities. The second class of communities was characterized by the extinction of Colpidium and cyclic populations of Tetrahymena that frequently persisted as the only prey species for the predator which reached large peak densities. The role of assembly history frequently manifests itself through priority effects, which may arise from a variety of mechanisms (e.g. resource competition: Tilman 1982; apparent competition: Holt, Grover & Tilman 1994; intraguild predation: Holt & Polis 1997; ontological change in prey vulnerability: Chase 1999; phenotypic plasticity: Hoverman & Relyea 2008; adaptive evolution: Loeuille & Leibold 2008). Strong priority effects, which were most striking between treatments with Colpidium and Tetrahymena as the first colonizers, played a dominant role in causing the compositional and dynamical differences among the assembled communities. We suggest that apparent competition of different intensities, associated with differential vulnerability of the prey species to the predator, may be responsible for the observed priority effects. Although our experimental data indicate that Dileptus can inflict mortality on each of the three prey species, a separate experiment growing Dileptus on separate prey species revealed their different vulnerability to the predator. That experiment showed that whereas Dileptus attained large populations on Tetrahymena, it was unable to establish sustainable populations on either Colpidium or Halteria alone (Fig. S1, Supporting Information). In this experiment, the most vulnerable Tetrahymena, when introduced first, also supported Dileptus populations of considerable size. In most of the TCH and THC microcosms, apparent competition quickly eliminated Halteria populations and caused the later extinction of Colpidium populations, and trophic interactions between Dileptus and Tetrahymena led to cyclic dynamics of their populations that persisted through the end of the experiment. On the other hand, the much less vulnerable Colpidium failed to support large Dileptus populations in microcosms where it attained substantial abundance as the dominant prey species. As a result, apparent competition was of minor importance and interspecific competition largely structured the prey communities in the CHT and CTH treatments. Note that here Dileptus functioned as a keystone predator preying upon prey species with differential vulnerability. It is tempting to suggest that history-induced alternative compositional and dynamical states may be common in natural communities often featuring keystone predators (e.g. Leibold 1996), an idea that needs to be evaluated by future work. In contrast to the prevalence of priority effects in microcosms in which Colpidium or Tetrahymena was introduced first, Halteria never attained dominance when it was the first colonizer. When considering the introduction order of the second and third species, however, the assembly process was still, to a certain degree, historically contingent, although these secondary priority effects operated only at low productivity (Table 1). One possible explanation for the dominance of Colpidium in the two replicates of the high-productivity HTC treatment, but in none of the replicates of the low-productivity HTC treatment, is that Colpidium body size tends to increase with increasing productivities (Jiang & Morin 2005), which may have allowed the species to enjoy a size refuge against predation at the high productivity. However, this explanation cannot account for Colpidium dominance in the low, but not high, productivity HCT treatments. It also cannot explain Colpidium dominance in the two replicates of the low-productivity THC treatment, but in none of the replicates of the high-productivity THC treatment. Further investigation will be needed to uncover mechanisms behind these results. One interesting pattern emerging from treatments exhibiting no historical effects is that the assembling communities did not converge towards similar states as generally expected when history plays little role (e.g. Fukami 2004; Jiang & Patel 2008), rather results from replicates in the same treatments diverged from one another (Table 1). As random divergence in the trajectory of community assembly has not been observed in comparable protist experiments lacking predators (e.g. Fukami 2004