Incorporating behavior-based indices of connectivity into spatially explicit population models

Measuring connectivity in fragmented landscapes remains a central problem in ecology. Connectivity metrics range from descriptors of landscape structure to direct observations of a species’ ability to move to and colonize a forest patch. We constructed individual-based spatially explicit population models for a guild of forest rodents in Indiana to test the ability of structural and actual, or behavioral, measures of connectivity to predict patch and landscape occupancy and abundance. Model accuracy was assessed using comparisons with data from trapping studies. Predicted abundances within patches correlated with empirical data for five out of six species, but predicted patterns of patch occupancy corresponded with observations for only one species. Discrepancies may be due to inaccurate parameter values or the absence from the models of ecological processes such as conspecific attraction and competition. Nonetheless, the models demonstrated the utility of patch immigration as a measure of connectivity in explaining population abundance in fragmented landscapes. We discuss potential methods of collecting these behavior-based data.