Estimation of habitat quality based on plant community, and effects of isolation in a network of butterfly habitat patches

Minimum viable site networks are crucial for many threatened species but their design is always a difficult procedure. The present study investigated methods to estimate habitat quality of patches, in an ecological network for five butterfly species (Brenthis ino, Clossiana selene, Lycaena helle, Lycaena hippothoe, Proclossiana eunomia) inhabiting wet meadows. Abundance predictions were performed on the basis of botanical relevés with a multiple-species approach combining canonical correspondence analysis and multiple regression. Models are defined on a reference site set and are evaluated on a test site set. All the fitted models predicted the abundance of the butterfly species considerably well (from 61 to 86% of the variation). We evaluated the potential consequences of isolation on local populations, by comparing predicted and observed abundance. It was expected that greater differences would be observed when sites were more isolated. On the test set, differences between predicted and observed abundance were largely correlated to site isolation for L. helle and P. eunomia. The most isolated sites had greater chances to be empty, even if they had high-quality habitat. Reciprocally, when the sites were less isolated, the abundance observed was always greater than predicted, seeming to confirm the role of rescue effects. © 2003 Éditions scientifiques et médicales Elsevier SAS. All rights reserved