Spatial autocorrelation analysis and the identification of operational units for conservation in continuous populations

Despite recent advances in the identification of genetic population structure through molecular-marker technology, the definition of intraspecific units for conservation remains problematic, particularly when genetic or phenotypic variation is continuously distributed in geographic space. We show that spatial autocorrelation analysis, applied to phenotypic or molecular data, can be used to describe the geographic structure and therefore can help define optimum strategies for conserving genetic variability within species. We propose that the intercept of a spatial correlogram can be an indication of the minimum distance between samples that can conserve and assess genetic diversity with maximum efficiency at lower costs. This parameter can be used both to define units and to establish sampling strategies for conservation programs. We illustrate the utility of this approach by autocorrelation analyses applied to three data sets: isozyme variability among Eugenia dysenterica populations in Brazilian Cerrado and within populations of Adenophora glandiflora in Korea, and microsatellite variation among Ursus arctos populations in North America. Our results suggest that the intercept of spatial correlograms is a useful parameter for establishing operational units for intraspecific conservation in continuous populations, based on overall genetic or phenotypic variability, by defining the minimum geographic distance at which samples are independent.