IMPLICATIONS OF DYNAMICALLY VARIABLE TRAITS FOR IDENTIFYING, CLASSIFYING, AND MEASURING DIRECT AND INDIRECT EFFECTS IN ECOLOGICAL COMMUNITIES

This article discusses how the presence of dynamic change in traits af fecting interspecific interactions changes the ways in which interacti ons between species are distinguished, classified, and measured. The p revalence of models lacking any trait dynamics has led to methods of i dentifying and measuring indirect effects that are not valid when phen otypically or evolutionarily plastic traits affect interspecific inter actions. When traits are dynamic, the number of links in a dynamic mod el can no longer be used to distinguish direct and indirect effects, a nd classifying interactions by a single sign denoting effect on equili brium density becomes problematic. The presence of trait dynamics also changes the interpretation of manipulative experiments that have been used to measure indirect effects. Because both traits and population densities can transmit indirect effects, a given ordered series of spe cies will often transmit several indirect effects, which may have oppo site signs. Because some traits can change very rapidly, dynamic equat ions describing population growth rates may often include the densitie s of species that interact indirectly with the given species. Some of the conceptual problems in comparing magnitudes of direct and indirect effects are illustrated by simple models of a three-species food chai n. Different methods of measuring effect magnitudes can give different conditions for when direct effects are larger than indirect effects. Other terminology related to indirect effects, including interaction m odification and higher-order interaction, is critically discussed. Giv en the present paucity of information about trait dynamics, it may be premature to attempt to compare magnitudes of direct and indirect effe cts.