PARENTS, AND NETWORKS ORDINANCES: DETECTION OF STRUCTURE IN THE INTERACTIVE

Abstract

In this paper we present a comprehensive approach to detect structural patterns in interactive communities of plant and animal species, linked by ecological processes such as pollination, frugivory or herbivory. Simple structural models can reveal gradient, compartmented or nested patterns of interaction; intermediate patterns between a gradient and compartments are also possible. Of special potential interest is a combined model, in which nested structures are embedded within compartments. Interaction patterns can be visualized and analyzed in different ways, either as matrices, as bipartite networks or as multivariate sets through correspondence analysis or other ordination procedures. We also propose that differences among patterns represent outcomes of distinct evolutionary and ecological processes that will be especially relevant in highly diversified communities. In general, compartmentation should reflect coevolutionary histories and constraints, whereas differences in species abundances or dispersal rates may generate nestedness. Hence, instead of choosing one model a priori, to be empirically verified, community structure should be probed for a suite of patterns. The comprehensive approach for detecting community structure that we advocate should help to improve ecological hypotheses on compositional patterns in interactive communities, as well as their attendant empirical tests in actual communities.