Scale dependence and scale independence in habitat associations of small mammals in southern temperate rainforest

Although field biologists commonly employ methodologies that either are tim e-tested or are believed (usually for very good reason) to be suitable to t he conditions that are encountered, it is rare that results generated using differing methods can be compared on a single fauna. We reanalyze data fro m three studies conducted on the small mammal fauna of Valdivian temperate rainforests in southern Chile. By restricting our analyses to a relatively homogeneous habitat type we avoid many problems associated with studies ove r large spatial scales, such as increased species richness via beta diversi ty. The studies involve a temporal transect, an elevational transect, and a latitudinal transect, and comprise a spatially nested series. Because thes e studies differed in their initial objectives, they employed different met hodologies, both in the field and during subsequent data analysis. At all s cales of analysis there is notable overlap in habitat use by all species. A lthough we observed some seasonal differences in habitat use by species, th ese were not substantial and involved a small subset of the species. At a l ocal scale, numerically dominant species selected non-random microhabitats but co-occurred with other species more frequently than expected by chance. At a larger spatial scale species appeared to select large subsets of the macrohabitat (i.e., "mesohabitat"), providing patterns suggestive of appare nt competition among species. At an even larger spatial scale, that of temp erate southern South America, we predict that species assemblages will agai n exhibit positive associations. Thus, views on the importance of competiti on and other biological processes depend critically on the spatial and temp oral scales over which observations are made.