Tuesday, August 3, 2010 - 10:50 AM

OOS 9-9: Trophic theory of island biogeography

Dominique Gravel, Université du Québec à Rimouski, François Massol, CEMAGREF, Elsa Flore Canard, Institut des Sciences de l'Evolution de Montpellier, David Mouillot, Université Montpellier 2, and Nicolas Mouquet, Université Montpellier 2, CNRS.

Background/Question/Methods

MacArthur and Wilson’s theory of island biogeography (TIB) is a cornerstone in ecology. It contributed to our understanding of the species-area relationship, one of the most fundamental and documented patterns in ecology. Its elegance comes from its simplicity; the species richness on an island is the result of a dynamic equilibrium between immigration events that are randomly sampling the regional species pool, and extinction events. Two major weaknesses however arise from this simplicity: the TIB i) does not consider explicit ecological interactions between species, and ii) does not predict the species identity on islands. Here we are interested to understand how trophic interactions could affect this sampling of the regional species pool and thus the food web structure on islands. Our main objective is to expand the classic TIB to include trophic interactions.
Results/Conclusions

We hypothesized that when a consumer colonizes an island, it needs at least one prey, otherwise it goes extinct from starvation. We expanded the model of the classic TIB to a Trophic-TIB model to include this assumption. The model predicts that the food web structure of the regional species pool has strong effects on the species-area relationship. The selection process in favor of the most connected species also generates a complexity-area relationship. Given an appropriate definition of the food web structure in the regional species pool, the Trophic-TIB model also predicts the occurrence probability of each species on an island. With these predictions in hand, we analyzed the food web properties of a unique dataset on the spatial distribution of pelagic organisms in 50 lakes of the Adirondacks to random samples of equivalent species richness. There is clear indication that community composition in the lakes is not simply a random draw from the regional species pool, as predicted by the classic TIB. We found that consumer species present in a lake have more preys than the null expectation of random samples. We compared the classic-TIB and the Trophic-TIB by AIC and found strong support in favor of the Trophic-TIB. Our results suggest that a selective sampling process of the regional species pool organizes food webs on islands. The Trophic-TIB provides a mechanistic explanation for the variation in local food web structure and explicit predictions on the species distribution.