Population genetics of the European rabbit along a rural-to-urban gradient

Abstract

The European rabbit (Oryctolagus cuniculus) is declining in large parts of Europe but populations in some German cities remained so far unaffected by this decline. The question arises of how urbanization affects patterns of population genetic variation and differentiation in German rabbit populations, as urban habitat fragmentation may result in altered meta-population dynamics. To address this question, we used microsatellite markers to genotype rabbit populations occurring along a rural-to-urban gradient in and around the city of Frankfurt, Germany. We found no effect of urbanization on allelic richness. However, the observed heterozygosity was significantly higher in urban than rural populations and also the inbreeding coefficients were lower, most likely reflecting the small population sizes and possibly on-going loss of genetic diversity in structurally impoverished rural areas. Global F_ST and G'_ST-values suggest moderate but significant differentiation between populations. Multiple matrix regression with randomization ascribed this differentiation to isolation-by-environment rather than isolation-by-distance. Analyses of migration rates revealed asymmetrical gene flow, which was higher from rural into urban populations than vice versa and may again reflect intensified agricultural land-use practices in rural areas. We discuss that populations inhabiting urban areas will likely play an important role in the future distribution of European rabbits.

Figure 1

Map showing the location of our eight study sites in and around Frankfurt am Main. BV, Bad Vilbel; FH, Flörsheim; K, Kriftel; OP, Ostpark; RP, Rebstockpark; IGG, innerer Grüngürtel; Oskar, Oskar-von-Miller Straße; BB, Bundesbank. Map modified, © OpenStreetMap contributors, CC BY-SA 2.0. https://creativecommons.org/licenses/by-sa/2.0/.

Figure 2

Scatterplot showing the relationship between urbanity indices and inbreeding coefficients (F_IS) in our eight study populations. The lines shows the predicted relationship (linear regression, ***p < 0.001) and the shaded areas the corresponding 95% confidence intervals.

Figure 3

(a) Bayesian Inference Criterion (BIC) values versus numbers of clusters (K), suggesting that K = 8 was the most likely number of genetically distinct clusters. (b) Discriminant analysis of principal components (DAPC) scatterplot: 20 PC axes were retained, cumulatively explaining more than 95% of the total variance. Eigenvalues of the analysis are displayed in the inset. Each individual is represented as dots, circles, triangles, or squares, while the suggested clusters are shown as ellipses.

Figure 4

(a) Scatterplot showing the genetic distinctiveness among the N = 139 individuals based on the first two principal components in our DAPC analysis with prior population assignment. Each individual is represented as dots, circles, triangles, or squares, and the suggested clusters as ellipses. (b) Bar plots showing membership probabilities, assessed as the proximity of individuals to different genetic clusters, grouped by population ID.

Figure 5

Whisker-box plots of migration rate estimates among rural sites, among urban sites, from rural to urban sites, and from urban to rural sites. n.s., not significant; *p < 0.05.

Figure 6

Relationship between pairwise genetic distances, expressed as F_ST/(1 − F_ST) and (a) log-transformed geographic distance and (b) environmental distances (differences in urbanity). We further show the relationships between recent migration rates (estimated using BayesAss) and (c) log-transformed geographic distances and (d) environmental distances. Lines show predicted relationships (linear regressions) and shaded areas the corresponding 95% confidence intervals. n.s., non-significant; *p < 0.05.