Defining environmentally heterogeneous sites when faced with conservation urgency and scarce in situ data

We apply an environmental domains approach to identify environmentally heterogeneous characteristics defining a landscape matrix. We built environmental layers for national, regional, and local scales, considering the different scales studies can have. We used a numerical classification of explicit spatial layers and performed a multivariate classification. Based on the domains obtained, we mapped the landscape's climatic heterogeneity and identified a comprehensive set of environmental variables that defined the landscape matrix at each scale. We specifically tested our approach for its suitability to define a sampling strategy for a landscape genetics study, using as focal species the rodent Heteromys pictus. Namely, from the domains obtained at the local scale, we selected sampling localities that comprised the broadest habitat heterogeneity, which we corroborated in the field. The landscape matrix thus generated was used with genetic data previously obtained for H. pictus. Our approach allowed identification of environmental variables significantly associated with dispersal (gene flow) of H. pictus individuals in their natural habitat. We demonstrate its adequacy to efficiently determine sampling localities -or landscape sites- that encompass the highest environmental heterogeneity, in explored and unexplored landscapes, enabling rapid identification of localities and their environmental characteristics where in situ information is scarce.