Representing genetic variation as continuous surfaces: an approach for identifying spatial dependency in landscape genetic studies


  • Melanie A. Murphy,

  • Jeffrey S. Evans,

  • Samuel A. Cushman,

  • Andrew Storfer

M. A. Murphy ( and A. Storfer, School of Biological Sciences, Dept of Biology, Washington State Univ., Pullman, WA 99164, USA. – J. S. Evans, USDA Forest Service, Rocky Mountain Research Station, 1221 S. Main, Moscow, ID 83483, USA. – S. A. Cushman, USDA Forest Service, Rocky Mountain Research Station, Missoula, MT 59801, USA.


Landscape genetics, an emerging field integrating landscape ecology and population genetics, has great potential to influence our understanding of habitat connectivity and distribution of organisms. Whereas typical population genetics studies summarize gene flow as pairwise measures between sampling localities, landscape characteristics that influence population genetic connectivity are often continuously distributed in space. Thus, there are currently gaps in both the ability to analyze genotypic data in a continuous spatial context and our knowledge of expected of landscape genetic structure under varying conditions. We present a framework for generating continuous “genetic surfaces”, evaluate their statistical properties, and quantify statistical behavior of landscape genetic structure in a simple landscape. We simulated microsatellite genotypes under varying parameters (time since vicariance, migration, effective population size) and used ancestry (q) values from STRUCTURE to interpolate a genetic surface. Using a spatially adjusted Pearson's correlation coefficient to test the significance of landscape variable(s) on genetic structure we were able to detect landscape genetic structure on a contemporary time scale (≥5 generations post vicariance, migration probability ≤0.10) even when population differentiation was minimal (FST≥0.00015). We show that genetic variation can be significantly correlated with geographic distance even when genetic structure is due to landscape variable(s), demonstrating the importance of testing landscape influence on genetic structure. Finally, we apply genetic surfacing to analyze an empirical dataset of black bears from northern Idaho USA. We find black bear genetic variation is a function of distance (autocorrelation) and habitat patch (spatial dependency), consistent with previous results indicating genetic variation was influenced by landscape by resistance. These results suggest genetic surfaces can be used to test competing hypotheses of the influence of landscape characteristics on genetic structure without delineation of categorical groups.