Scaling from functional trait interactions to communities and ecosystems

Background/Question/Methods  - Ecological communities and ecosystems are complex, and there have been a proliferation of frameworks and models designed to make predictions about this complexity. Here, I will describe how evolutionary game theory can provide another possible framework for community and ecosystem ecology. Evolutionary games are ideal for such a framework because they include population growth of any number of species, bounded by interactions among species where the interactions are explicitly mediated by the functional traits of species in the community. The population growth is therefore frequency and density dependent, and both ecological and evolutionary changes are predictable. To demonstrate the usefulness of this framework for community ecology, I present the results of two simple studies. First, using a simple Lotka-Volterra competition game, I show how this framework can be used to generate a simple community structured entirely by limiting similarity that is evolutionarily and ecologically stable. Here, we can predict the functional traits in the community, the abundances of each species, and the entire phylogenetic history of the community is known because of the evolutionary dynamic. Second, I describe a simple game of plant-plant competition that can be used to predict net primary productivity (NPP) at a global scale.

Results/Conclusions - First, the Lotka-Volterra game shows how the balance between convergent and divergent selection can shape the diversity of the community. Furthermore, it shows how: 1) limits in the functional traits that are evolutionarily possible can constrain species diversity via limiting similarity and; 2) limits in the minimum level of competition among members of the community also constrain species diversity. However, this model is difficult to test because most of the interactions are phenomenological. Second, the model of NPP predicts root, stem and leaf allocation independently and offers a model that can be explicitly tested against observed data. The model is parameterized from the literature to constrain trait space to what has been observed. I then compare the predictions to two independent datasets: FLUXNET and MOD17. The model predicts each dataset as well as they are able to predict each other, suggesting that uncertainties in the model are no worse than measurement error. In sum, evolutionary game theory provides one possible framework for prediction in community and ecosystem ecology that I believe is underappreciated.