Multimodel Inference and the Understanding of Complexity, Discontinuity, and Normadism

Abstract

Nomadism has received surprisingly little attention in the ecological literature, and further work in this area is needed. The results of Woinarski’s reanalysis of our research findings are broadly similar to our own, and they support our original interpretation. However, his presentation is confusing and difficult to interpret.We used an information-theoretic approach to multimodel selection. We a priori defined plausible candidate models relating the variables described in our original paper or Woinarski’s reanalysis to the phenomenon of nomadism. We tested models that investigate nomadism as a function of nectivory, granivory, diet diversity, mixed diet, distance to body mass aggregation edge, mass, interactions between distance to edge and nectivory, distance to edge and mass, mass and nectivory, or mass and the interaction of edge and nectivory. There is consistency in our results across all sets of models, suggesting that mass, distance to body mass aggregation (scale breaks), and diet (nectivory) are all important factors in determining nomadism. In no case was nectivory or any other diet variable supported in a single-variable model. Given the same data and similar results, we and Woinarski reach fundamentally different conclusions. Woinarski views nomadism as an easily understandable result given knowledge of the proper single mechanistic variable, and he discounts interactions with structural features of the landscape and scaling. We conclude that nomadism is a fundamentally complex phenomenon without a single source of causation, and that it is the interaction of species, species attributes, and landscapes that is responsible for nomadic behavior.