Food-web analysis through field measurement of per capita interaction strength

Abstract

THE idea that the connections between species in ecological assemblages are characterized by a trait called 'interaction strength' has become a cornerstone of modern ecology. Since the classic paper of Watt¹, ecologists have acknowledged the importance of distinguishing pattern (static community features) from process (dynamically based mechanisms), which can determine the immediate observed details. Food webs display pathways of implied dynamics, and thus potentially unite pattern and process in a single framework^2,3. Most analyses^4–8 have focused entirely on web topology and the derived descriptive properties. By contrast, attempts to generalize how natural communities are organized^9,10 or summary statements about whole communities^6,11–13 have emphasized critical processes. Elton's² insights and May's³generalizations and analyses have stimulated current developments. In May's approach, the idea of interaction strength is precise, reflecting coefficients in a jacobian matrix associated with a community dynamics model. He found striking dependence of community stability both on web complexity and on the number and strength of interactions. By contrast, empiricists have usually determined relative interaction strength from single-species removals analysed by multivariate statistics^14,15.I report here the first experimental study designed to estimate interaction strengths in a species-rich herbivore guild, documenting on a per capita basis mainly weak or positive interactions, and a few strong interactions, a pattern which has profound implications for community dynamics.