Calculations on the back of a climate envelope: Addressing the geography of species distributions

Calculations on the back of a climate envelope: Addressing the geography of species distributionsClimate envelope modeling is an approach that identifies association between species distribution and climate, an association that often is strong at broader spatial scales.Beale et al.(1) discuss limitations of climate envelope modeling and propose use of null models to test relationships between species distributions and climate.Their null models maintain the prevalence of observations and spatial structure of species data while altering the geographic location of the distribution.Other qualities of geographic distributions and variability demonstrate a need for improved understanding and management of geographic concepts in climate envelope modeling.Preserving prevalence and spatial structure for distributions is necessary, but not sufficient, to describe a geographic null distribution.Geographic location is more than a referencing frame for registering datasets.Location provides place-and context-specific information, reflecting variability, heterogeneity, and neighborhood effects at many scales (2, 3).The geography of topographic and other influences ensure that climate displays spatial and geographic differences across any study area (4).This local and contextual geography is not included in the null models provided, which ignore subregional variability, increasingly recognized as important in modeling (5).Beale et al. match incomplete null geographic representations for species distributions against nonstationary spatial patterns of climate data leading to results that sometimes support envelope models and sometimes do not.Additionally, analysis of null species distribution models against climate geography may suggest that envelope models are statistically insignificant, but this does not invalidate the observed record of association identified in a climate envelope model.There is scope for further development of suitable null geographic models.