Genetic and environmental drivers of migratory behavior in western burrowing owls and implications for conservation and management

Abstract

Migration is driven by a combination of environmental and genetic factors, but many questions remain about those drivers. Potential interactions between genetic and environmental variants associated with different migratory phenotypes are rarely the focus of study. We pair low coverage whole genome resequencing with a de novo genome assembly to examine population structure, inbreeding, and the environmental factors associated with genetic differentiation between migratory and resident breeding phenotypes in a species of conservation concern, the western burrowing owl (Athene cunicularia hypugaea). Our analyses reveal a dichotomy in gene flow depending on whether the population is resident or migratory, with the former being genetically structured and the latter exhibiting no signs of structure. Among resident populations, we observed significantly higher genetic differentiation, significant isolation-by-distance, and significantly elevated inbreeding. Among migratory breeding groups, on the other hand, we observed lower genetic differentiation, no isolation-by-distance, and substantially lower inbreeding. Using genotype-environment association analysis, we find significant evidence for relationships between migratory phenotypes (i.e., migrant versus resident) and environmental variation associated with cold temperatures during the winter and barren, open habitats. In the regions of the genome most differentiated between migrants and residents, we find significant enrichment for genes associated with the metabolism of fats. This may be linked to the increased pressure on migrants to process and store fats more efficiently in preparation for and during migration. Our results provide a significant contribution toward understanding the evolution of migratory behavior and vital insight into ongoing conservation and management efforts for the western burrowing owl.

Keywords: genetic connectivity; genomics; genotype–environment associations; inbreeding; migration.

FIGURE 1

Map of sampling strategy and primary genetic structure results for the western burrowing owl. (a) Plot of NGS‐Admix result at K = 5. Migrant sires were reduced to three individuals each to facilitate analyses. Icons next to sample site names correlate with sites on map in (b) and PCA in (c). Samples are arranged by resident, resident sites that were not genetically distinct from the migrant (“RES‐MIG”), and migrant breeding sites. (b) Map of western burrowing owl breeding range (dotted outline) as predicted based upon eBird data. Sample sites are indicated by icons. Shading indicates cluster membership for the five genetically structured resident breeding sites using a kriging of NGS‐Admix results. (c) PCA on all samples across ~3 M variants using single read sampling. Five resident breeding sites are clearly differentiated. The “RES‐MIG” and migratory breeding sites exhibit no differentiation and hence generally overlap at the origin of the PCA.

FIGURE 2

Comparisons of F statistics between BUOW migratory and resident breeding sites. Exception resident sites that are not structured from the migrant sites are grouped with resident breeding sites. (a) Residents are significantly more differentiated from one another than migrants (W = 26, p < 0.001). (b) Residents exhibit significant isolation‐by‐distance (r = 0.67, p = 0.004) while migrants do not (r = −0.04, p = 0.58). (c) Inbreeding is significantly higher in residents than migrants (W = 1285, p < 0.001).

FIGURE 3

PCA (a) and map (b) portraying gene–environment correlations associated with migratory behavior across the BUOW range. Colors are based upon 10,000 random points across the breeding range, but is restricted to the U.S. due to the availability of the landcover data. (a) PCA of climate variables with PC scores associated with sample sites indicated with symbols that match Figure 1. Arrows indicate the loadings of top‐ranked variables identified by gradient forest analysis. (b) Map of projected GEA correlations across the BUOW range and sample sites indicated as in Figure 1.