Process-based and correlative modeling of desert mistletoe distribution: a multiscalar approach

Abstract

Because factors affecting distributional areas of species change as scale (extent and grain) changes, different environmental and biological factors must be integrated across geographic ranges at different resolutions, to understand fully the patterns and processes underlying species' ranges. We expected climate factors to be more important at coarse resolutions and biotic factors at finer resolutions. We used data on occurrence of a parasitic plant (Phoradendron californicum), restricted to parts of the Sonoran and Mojave deserts, to analyze how climate and mobility factors explain its distributional area. We developed analyses at five spatial resolutions (1, 5, 10, 20, 50 km) within the distributional area of the disperser species, and compared ecological niche models from three commonly used correlative methods with a process-based model that estimates colonization and extinction rates in a metapopulation framework. Correlative models improved when layers associated with hosts and disperser were used as predictors, in comparison with models based on climate only; however, they tended to overfit to data as more layers were added. Dispersal-related parameters were more relevant at finer resolutions (1–5 km), but importance of extinction-related parameters did not change with scale. We observed greater coincidence between correlative and process-based models when based only on dimensions of the abiotic niche (i.e., climate), but a clearer and more comprehensive mechanistic understanding was derived from the process-based algorithm.