Widespread genetic connectivity of feral pigeons across the Northeastern megacity

Abstract

Urbanization may restrict, facilitate, or have no effect on gene flow, depending on the organism and extent of urbanization. In human commensals, with high dispersal ability, urbanization can facilitate gene flow by providing continuous suitable habitat across a wide range. Additionally, suburban or rural areas with lower human population density may act as a barrier to gene flow for these human commensals. Spatial population genetic approaches provide a means to understand genetic connectivity across geographically expansive areas that encompass multiple metropolitan areas. Here, we examined the spatial genetic patterns of feral pigeons (Columba livia) living in cities in the eastern United States. We focused our sampling on the Northeastern megacity, which is a region covering six large cities (Boston, Providence, New York City, Philadelphia, Baltimore, and Washington, DC). We performed ddRAD-Seqon 473 samples, recovered 35,200 SNPs, and then used multiple evolutionary clustering analyses to investigate population structuring. These analyses revealed that pigeons formed two genetic clusters-a northern cluster containing samples from Boston and Providence and a southern cluster containing all other samples. This substructuring is possibly due to reduced urbanization across coastal Connecticut that separates Boston and Providence from New York and mid-Atlantic cities. We found that pairs of pigeons within 25 km are highly related (Mantel r = 0.217, p = .001) and that beyond 50 km, pigeons are no more related than they would be at random. Our analysis detected higher-than-expected gene flow under an isolation by distance model within each city. We conclude that the extreme urbanization characteristic of the Northeastern megacity is likely facilitating gene flow in feral pigeons.

Keywords: Columba livia; Northeastern megacity; Rock Dove; pigeon; population genetics; urban evolution.

FIGURE 1

Maps showing sampling location including (a) cities where samples were collected, (b) National Land Cover Data (NLCD v.2011; Homer et al., 2015), and (c) light intensity at night (NASA Earth Observatory, 2017). Sample locations are shown as teal circles. Note that while National Land Cover Data (NLCD) is often used to indicate urbanization, NASA’s Earth Observatory images of artificial light at night shows the interconnectedness of urbanization that is missed by landcover data. A more colorblind friendly version of this figure appears in the supplement (Figure S1)

FIGURE 2

Discriminant analysis of principal components (DAPC) for 35,200 genome‐wide SNPs recovered from pigeons. This graph recapitulates geography, showing separation of sampled collected in the northern part of the megacity (Boston and Providence) from samples collected in more southern parts of the megacity along the 1st PC axis (x‐axis) and New York City/Philadelphia samples separating from the Baltimore/Washington DC samples along the 2nd PC axis (y‐axis)

FIGURE 3

Bar plots showing our ADMIXTURE v1.3.0 (Alexander et al., 2009) results at K = 2, 4, and 9. At K = 2, samples from the northern region of the Northeastern megacity (Boston) show a different shared ancestry from samples collected in more southern regions of the megacity. At K = 4 and K = 9, samples from Boston continue to show a different shared ancestry than samples from the rest of the megacity. Cross‐validation indicated that K = 2 is the most well‐supported K value based on analysis of 35,200 SNPs genome‐wide SNPs

FIGURE 4

(a) Scatterplot of the natural log of Euclidean geographic distance between samples and pairwise genetic distance showing weak isolation by distance relationship. The trend line (shown in red) depicts the linear relationship calculated by a simple Mantel test (Mantel r = 0.132, p = .001). (b) Mantel correlogram showing the relationship between genetic distance and geographic distance within each 25 km distance class. Each circle represents a distance class, and filled circles represent significant associations (α = 0.05)

FIGURE 5

Estimated effective migration surface (EEMS) for pigeons in the Northeastern United States. Coloring of the map represents relative effective migration rates ranging from higher‐than‐average (blue) to lower‐than‐average (red) historic gene flow with isolation by distance represented as the null (white). Circles represent the approximate sampling range for each city. Within many cities, there is high‐than‐average gene flow and isolation by distance, but there is lower‐than‐average gene flow between cities. While this map shows higher‐than‐average gene flow to the west, we are unable to draw conclusions from this area since it was not sampled. Inset map shows sample locations in black