Print PageWe are interested in understanding what factors determine the area of distribution of a species at different spatial scales and in distinguishing how patterns at one scale are manifestations of processes operating at other scales. One of the main problems in modeling distributions of species is that factors affecting them are dependent on the scale of analysis, which makes our capacity to extrapolate and interpret the models a confusing and difficult task. Correlative and mechanistic models of species’ ecological niches and geographic distributions offer two contrasting ways of predicting the requirements and processes on which species depend for survival. Using occurrence data retrieved from the field and remotely sensed, we compared these two methodological approaches for the distribution of the desert mistletoe (Phoradendron californicum) in a region encompassing the semiarid region of the Southwest in the United States at different spatial scales (1 to 100 km). A mechanistic model was fit to incorporate balancing colonization and extinction for the desert mistletoe, including spatially explicit information on the host trees, the abundance of the disperser bird species and climatic conditions. Correlative models were carried out using suites of bioclimatic variables and two commonly used algorithms (GARP and Maxent).
Results/Conclusions
Results show there is a different degree of coincidence between these two modeling approaches, with a larger coincidence for larger scales. In general, correlative models improved when the modeling included biologically meaningful variables (i.e., host trees and disperser); accordingly, mechanistic models show a larger amount of sensitivity in relation to the number of parameters added and their predictions are better for finer scales.
