Towards an integrated theory of Biogeography

Submitted by editor on 8 January 2016. Get the paper!

By Kevin Cazelles, Nicolas Mouquet, David Mouillot and Dominique Gravel

Debates are going on the extent to which ecological interactions spread over spatial scales. The absence of a clear answer to this particular problem casts doubts on the very popular Species Distribution Models (hereafter SDMs) and their ability to predict current and future species geographic ranges. Indeed, classical SDMs focus on the correlation between a set of environmental variables and the presence of a given species. By doing so, they assume that species are distributed independently from each other to predict changes in future distribution of biodiversity. Although new approaches strive to overcome this limitation, the theoretical gap still remains to be bridged.

Schematic representation of our model: In the regional pool, species are considered dependent from each other as they are embedded within a network of ecological interactions (A). Species colonize new localities (B) according to their physiological abilities (C). For instance, species (1) has a wide niche along the environmental gradient whereas species 5 and 6 have narrow niches in high and low values of the gradient, respectively. Species interact locally, which may increase or decrease their probability of going extinct (D).

We went back to the classical Theory of Island Biogeography to investigate the potential impacts of ecological interactions at broad geographical scales. We expanded the work of MacArthur and Wilson to integrate environmental constraints and ecological interactions. To do so, we considered that interactions impact extinction rates, while environmental conditions impact colonization rates. A key to achieve the integration of ecological interactions was to compute the probabilities of assemblages instead of single species occurrence probabilities. By doing so, we were able to evaluate species richness and network properties along an environmental gradient. Hence, we built an integrated stochastic model mixing (1) colonization/extinction dynamics together with (2) ecological interactions and (3) environmental constraints.

Our theoretical studied provides a framework that sheds light on poorly addressed questions at large spatial scales. For instance, we analyzed how increasing the connectance of ecological networks affects the distribution of species richness along an environmental gradient. We show that while positive interactions enhance richness even for harsh abiotic conditions, negative ones prevent the diversification of local communities. Therefore, we illustrated how different aspects of community ecology and biogeography may interplay and shape biodiversity distribution. In our changing world, such integration is urgently needed. Indeed, species are currently tracking shifting climate suitability and we still wonder to what extent the failure of one species will affect the distribution of other species with which it interacts.

Connectance and species richness along an environmental gradient: Using our stochastic model, we were able to compute species richness along an environmental gradient for (A) negative interactions (B) positives interactions and (C) a mixture of positive and negative interactions. Here, colonization rates for all species are Gaussian-shaped with optima randomly distributed along the gradient. We found that increasing connectance strongly affects the species richness in different ways: (A) a decrease, (B) an increase or (C) a flatter profile, pointing out potential impacts of ecological interactions at large scale.

Solving the fundamental issue of the geographical scaling of ecological interactions doubtlessly requires further theoretical advancements. However, as a first major insight, we demonstrate in our paper the strength of focusing on occurrences of assemblages rather than occurrences of single species to integrate ecological interactions. We hope our work will promote uses and developments of methods that integrate information on multi-species occurrences data to improve the quality of biodiversity forecasts (such as the Joint Species Distribution Model does).