A longitudinal genetic survey identifies temporal shifts in the population structure of Dutch house sparrows

Abstract

Dutch house sparrow (Passer domesticus) densities dropped by nearly 50% since the early 1980s, and similar collapses in population sizes have been reported across Europe. Whether, and to what extent, such relatively recent demographic changes are accompanied by concomitant shifts in the genetic population structure of this species needs further investigation. Therefore, we here explore temporal shifts in genetic diversity, genetic structure and effective sizes of seven Dutch house sparrow populations. To allow the most powerful statistical inference, historical populations were resampled at identical locations and each individual bird was genotyped using nine polymorphic microsatellites. Although the demographic history was not reflected by a reduction in genetic diversity, levels of genetic differentiation increased over time, and the original, panmictic population (inferred from the museum samples) diverged into two distinct genetic clusters. Reductions in census size were supported by a substantial reduction in effective population size, although to a smaller extent. As most studies of contemporary house sparrow populations have been unable to identify genetic signatures of recent population declines, results of this study underpin the importance of longitudinal genetic surveys to unravel cryptic genetic patterns.

Figure 1

Map showing the seven sampling locations in the Netherlands: Amsterdam (Am), Berg-en-Terblijt (B&T), Leiden (Le), Twello (Tw), Voorschoten (Vo), Wilp (Wi) and Zoetemeer (Zo). In gray are urbanized areas extracted from the CORIN land cover.

Figure 2

Comparison of pairwise F_ST in pre- and post-decline samples of P. domesticus. The identity line represents the case where pre- and post-decline pairwise F_ST would be equal.

Figure 3

Genetic structure plot of (a) pre-decline and (b) post-decline samples as inferred from Bayesian genetic clustering for K=2. Each bar represents an individual partitioned according to its probability of assignment to a cluster.

Figure 4

Estimates of effective population size of seven Dutch P. domesticus populations. The left side of each plot corresponds to pre- and post-decline estimates based on three single sample methods (N_eSib, N_eABC and N_eLD). Percentages of change in N_e for each method are represented below the post-decline estimates. The right side of each plot corresponds to the variance effective population size estimates (on another scale) based on three temporal methods (TempoFS, MLNE and Moment-based method; CONE estimates were excluded as they were all infinite). Error bars show 95% confidence intervals. Negative, infinite and large estimates are replaced by ‘n.a.'. All estimated values are reported in Supplementary Appendix 5.