The matrix-tolerance hypothesis: an empirical test with frogs in the Atlantic Forest

Abstract

The matrix-tolerance hypothesis suggests that the most abundant species in the inter-habitat matrix would be less vulnerable to their habitat fragmentation. This model was tested with leaf-litter frogs in the Atlantic Forest where the fragmentation process is older and more severe than in the Amazon, where the model was first developed. Frog abundance data from the agricultural matrix, forest fragments and continuous forest localities were used. We found an expected negative correlation between the abundance of frogs in the matrix and their vulnerability to fragmentation, however, results varied with fragment size and species traits. Smaller fragments exhibited stronger matrix-vulnerability correlation than intermediate fragments, while no significant relation was observed for large fragments. Moreover, some species that avoid the matrix were not sensitive to a decrease in the patch size, and the opposite was also true, indicating significant differences with that expected from the model. Most of the species that use the matrix were forest species with aquatic larvae development, but those species do not necessarily respond to fragmentation or fragment size, and thus affect more intensively the strengthen of the expected relationship. Therefore, the main relationship expected by the matrix-tolerance hypothesis was observed in the Atlantic Forest; however we noted that the prediction of this hypothesis can be substantially affected by the size of the fragments, and by species traits. We propose that matrix-tolerance model should be broadened to become a more effective model, including other patch characteristics, particularly fragment size, and individual species traits (e.g., reproductive mode and habitat preference).