A fundamental tenet of modern ecology and conservation science is that species occurrence in habitat patches can be determined by patch area and isolation. But such island biogeographic models often poorly predict actual species occurrences in structurally complex landscapes that typify most ecosystems. Recent advances in circuit theory have enhanced estimates of species dispersal, and through integration with island biogeography, can provide powerful ways to predict landscape-scale distribution of species assemblages. Applying such an integrative analytical framework to 43 bird species in Tanzania improved model fit by an average of 2.2-fold over models where patch isolation was estimated without accounting for landscape matrix heterogeneity. This approach also allowed us to assess species-specific dispersal rates and quantify differences among land cover types in their permeability to animal movement. These results reaffirm the utility of foundational island biogeographic principles, yet with an important caveat. Two-thirds of the variance in species occurrence in habitat fragments can be explained simply by patch area and isolation, conditional on isolation explicitly accounting for the spatial configuration of different land cover types in the landscape matrix.