Using binary and probabilistic habitat availability indices derived from graph theory to model bird occurrence in fragmented forests

Loss of connectivity is one of the main causes of decreases in habitat availability and, thus, in species abundance and occurrence in fragmented landscapes. It is therefore important to measure habitat connectivity for conservation purposes, but there are several difficulties in quantifying connectivity, including the need for species movement behavioral data and the existence of few consistent indices to describe such data. In the present study, we used a graph theoretical framework to measure habitat availability, and we evaluate whether this variable is adequate to explain the occurrence pattern of an Atlantic rainforest bird (Pyriglena leucoptera, Thamnophilidae). The playback technique was used to parameterize the connectivity component of habitat availability indices and to determine the presence or absence of the study species in forest patches. Patch-and landscape-level habitat availability indices were considered as explanatory variables. Two of these were landscape-level indices, which varied in terms of how inter-patch connections are defined, using either a binary or probabilistic approach. This study produced four striking results. First, even short open gaps may disrupt habitat continuity for P. leucoptera. Second, the occurrence of P. leucoptera was positively affected by habitat availability. Third, proper measures of this explanatory variable should account for the landscape context around the focal patch, emphasizing the importance of habitat connectivity. Finally, habitat availability indices should consider probabilistic and not binary inter-patch connections when intending to explain the occurrence of bird species in fragmented landscapes. We discuss some conservation implications of our results, stressing the advantages of an ecologically scaled graph theoretical framework.