Despite the recognition that habitat quality is an important predictor of insect demography, patch-based population models predominately focus on patch geometry (i.e., patch size and connectivity) as the primary factor explaining patch occupancy. Using a long-term (10 years) census of a spatially structured population of an erebid caterpillar, we generated a series of statistical models to test the influence of habitat quality, a novel connectivity metric, total annual precipitation and their interactive effects on patch occupancy, abundance, colonization, and extinction. Our measure of connectivity accounts for nonrandom dispersal behavior, which better reflects the natural history of the system and provided a better fit to the data than more conventional connectivity metrics that assume dispersal occurs randomly.
Results/Conclusions
Patch quality was the strongest predictor of the observed dynamics, but this was mediated by precipitation: in dry years, patch occupancy, abundance and colonization were low, but extinction was high, while the reverse was true in wet years. Models including connectivity as well as its interaction with both habitat quality and precipitation received some support (AICc <2): higher connectivity increased the probability of patch occupancy and abundance for low-quality but not high-quality patches; higher connectivity was associated with a higher probability of patch colonization but only in wet years. This work indicates that the assumption of patch homogeneity is invalid for some systems and suggests researchers should strive to incorporate a patch quality metric. Demonstrating the strong effect of precipitation on local dynamics and its interaction with patch quality and connectivity provides evidence that metapopulation models that do not include climatic variables may provide poor predictions under future climatic regimes.