Effects of geographic variation in host plant resources for a specialist herbivore's contemporary and future distribution

Abstract

Species distributions are driven by abiotic and biotic factors, but the importance of variation in the availability and quality of critical resources is poorly understood. Disentangling the relative importance of these factors—abiotic environment, availability of critical resources, and resource quality—will be important to modeling species current distributions and responses to projected climate change. We address these questions using species distribution models (SDMs) for the western monarch butterfly population (Danaus plexippus), whose larvae feed exclusively on Asclepias species known for their heterogeneous distribution and variation in host quality. We modeled the distribution of 24 Asclepias species to compare three monarch distribution models with increasing levels of complexity: (1) a null model using only environmental factors (climate envelope model), (2) a model using environmental factors and Asclepias availability estimated as species richness, (3) and a model using environmental factors and Asclepias’ availability weighted by host plant quality as assessed through a greenhouse bioassay of larval performance. Asclepias models predicted that half of the Asclepias species will expand their ranges and shift toward higher latitudes, while half will contract. These patterns were uncorrelated with host plant quality. Among the three monarch models, the climate envelope model was the poorest performing. Models accounting for host plant availability performed best, while accounting for host plant quality did not improve model performance. The climate envelope model estimated more restrictive contemporary and future monarch ranges compared to both host plant models. Although all three models predicted future monarch range expansions, the projected future distributions varied among models. The climate envelope model predicted range expansions along the Pacific coast and contractions inland. In contrast, the host plant availability and quality models predicted range expansions in both of these regions and, as a result, 14% and 19% increases in distribution (respectively) relative to the climate envelope model. These models do not include other factors affecting monarch persistence. Nevertheless, our findings suggest that accounting for information on host plant availability and response to climate change is necessary to predict future species distributions, but that variation in the quality of those critical resources may be of secondary importance.