A gradient analysis of multiple interacting guilds highlights the role of biotic interactions in structuring communities

Background/Question/Methods

The extent to which ecological communities are coherent entities as opposed to mere intersections of individual species distributions has long been one of the fundamental questions of ecology. Gradient analysis is one commonly used tool for addressing this question; however, all such studies have used organisms from a single taxon or guild. This risks missing important connections due to non-competitive interactions such as mutualism, predation, parasitism, commensalism and amensalism, which should be more likely to occur between members of different guilds. Such organisms are unlikely to compete for resources and can have complementary niches that promote non-competitive interactions.

We examined the abundances of taxa in four interacting guilds (seed plants, soil basidiomycetes, fungivorous collembolans and predatory mesostigmatid mites) along an elevation gradient in a forest in the southern Appalachian mountains. A causal discovery algorithm was used to investigate the relative frequencies of interguild and intraguild interactions. We also used elements of metacommunity structure (EMS) analysis to study the extent to which communities are coherent entities made up of non-independently distributed species and tested the hypothesis that combinations of two or more interacting guilds exhibit more coherence than single guilds.

Results/Conclusions

When the taxonomic richness of the four guilds was taken into account, inter- and intraguild interactions were approximately equally common.

In the EMS analysis, the basidiomycete subcommunity was classified as Clementsian, plants and mesostigmatids were quasi-Clementsian (exhibiting continuous species distributions and a high degree of boundary conjunction but not statistically significant amounts of species turnover), and collembola were random, lacking either continuous species distributions or a systematic checkerboard pattern. When sets of multiple guilds were examined, Clementsian and quasi-Clementsian structures predominated. We also compared boundary conjunction, measured as Morisita's index (MI) for these sets of guilds to the weighted average of the MI values of the guilds composing them. This revealed a striking pattern in which only sets of directly interacting guilds had higher-than-baseline boundary conjunction values, and such boundary conjunction values are found in all but one set of directly interacting guilds.

Our results highlight the importance of inter-guild interactions in structuring patterns of cooccurrence. Trophic interactions and plant-fungus symbioses (mutualistic and/or pathogen-host) appear particularly important. Future research should include multiple guilds, and communities must be seen as systems governed by a variety of interaction types.