Greater Resolution of Distributional Complementarities by Controlling for Habitat Affinities: A Study with Bahamian Lizards and Birds

Abstract

We present a new multivariate method that controls for habitat affinities when testing for distributional complementarities between species, and we apply it to 40 lizard and bird species on 521 islands of the Bahamas The method uses occurrence (presence or absence) data for the species of interest as dependent variables and continuously varying habitat or island characteristics (e g., island area) as independent variables. Technically, it combines a logistic regression approach with a multiway-contingency-table approach (Nerlove and Press 1973). The method thus allows simultaneous evaluation of habitat-by-species relations and species-by-species relations (i.e., species interactions). We find that (1) species interactions overall become substantially more negative once habitat relations are taken into account; (2) habitat relations are statistically more significant than species interactions; (3) three-species interactions tend to be more often negative than two-species interactions; and (4) raw (real) ...