Introduction
Colonization of newly available habitats has led to some of the most conspicuous cases of biological diversification known in natural systems (e.g. Carson & Kaneshiro 1976; Lerner et al. 2011; Sato et al. 2001), raising questions about the relative roles of random founder effects and selection in population divergence (Carson 1975; Mayr 1963; Price et al. 2010; Templeton 1980). Population divergence following colonization can occur due to both local adaptation to novel selective pressures (Rundle & Nosil 2005; Schluter 2000), and to stochastic effects derived from rapid changes in effective population size (Charlesworth 2009; Lande 1980; Simpson 1953). Founder effects and directional selection may result in rapid shifts in allele frequencies and accelerated rates of trait evolution, as shown by instances of rapid evolution following colonization events at time scales of thousands to a few million years (e.g. Lerner et al. 2011; Millien 2006; Satoet al. 2001; Seehausen 2006; Wessel et al. 2013), to scales as short as a few decades (e.g. Chen et al. 2018; Jensenet al. 2017; Mathys & Lockwood 2011; Sendell-Price et al.2020). In the context of global environmental change, contemporary colonization of modified habitats provides the opportunity to study evolution in progress (Colautti & Lau 2015; Huey et al. 2000; Johnson & Munshi-South 2017; Perrier et al. 2020; Reznick & Ghalambor 2001; Salmón et al. 2021), and to understand the relative contributions of directional selection, phenotypic plasticity, gene flow and demography in driving rapid divergence in newly established populations (Campbell-Staton et al. 2020; Szulkinet al. 2020b).
A potential case of contemporary colonization is provided by a suburban population of the Oregon junco (Junco [hyemalis ]oreganus ), a small passerine that typically inhabits mixed-coniferous forests of western North America (Miller 1941; Nolanet al. 2002). In the early 1980s, a group of Oregon juncos became naturally established on the campus of the University of California at San Diego (UCSD) (McCaskie 1986; Rasner et al. 2004; Yeh 2004), on the coast of the Pacific Ocean. More recently, other Oregon junco populations have been identified in suburban areas of southern California, where they inhabit a mixed landscape of buildings and park areas. The early colonists from UCSD continued breeding and formed a resident population of about 70 pairs that has remained stable (Atwellet al. 2014; Yeh & Price 2004). Mark-recapture studies have shown that they are year-round residents at UCSD, but between September and May they are joined by small flocks of wintering Oregon juncos coming from other populations (Fudickar et al. 2017; Unitt 1984; Yeh 2004).
The Oregon junco is composed of various geographic forms that originated recently as part of the postglacial radiation of the dark‐eyed junco (Junco hyemalis ), following the northward recolonization of the North American continent as ice sheets receded after the Last Glacial Maximum (LGM), ca. 18,000 years ago (Friis et al. 2016; Miláet al. 2007). The Oregon junco has been traditionally divided into at least seven geographically structured forms showing considerable phenotypic and habitat variation (Dwight 1918; Friis et al. 2018; Miller 1941; Nolan et al. 2002). These forms include, from south to north: townsendi from the San Pedro Martir mountains in northern Baja California, Mexico; pontilis from the Sierra Juarez mountains, also in Baja California; thurberi from the mountains of southern to northern California; pinosus from the coastal region of central California; montanus from the interior regions of Oregon, Washington and British Columbia; shufeldti from coastal regions of Oregon and Washington; and oreganus from coastal British Columbia and southern Alaska (Miller 1941; Nolanet al. 2002) (Fig. 1).
Previous studies on the ecology and evolution of the UCSD junco population have generally assumed that it originated from the nearest breeding population of Oregon juncos, located in the Laguna Mountains, 70 km east of UCSD (Fig. 1), based on the appearance of birds sampled during the non-breeding season. This population belongs to the subspecies thurberi and inhabits elevations above 1,500 m, a montane habitat with a more extreme temperature regime than the milder Mediterranean climate of the UCSD campus (Unitt 1984; Yeh & Price 2004). Studies published to date have focused on the evolution of social signaling traits during colonization of a novel environment (Priceet al. 2008; Reichard et al. 2020; Yeh 2004); the role of plasticity in population persistence during the early stages of colonization (Price et al. 2008; Yeh & Price 2004); patterns of morphological and genetic variation in comparison with other California populations, using both microsatellites (Rasner et al. 2004) and MHC loci (Whittaker et al. 2012) and hormonal changes underlying shifts in phenotypic and life-history traits during adjustments to urban environments (Atwell et al. 2012; Atwell et al. 2014; Fudickar et al. 2017). Based on phenotypes of over-wintering non-resident birds and proximity to potential sources, Yeh (2004) proposed that the population of UCSD juncos had likely originated from the thurberi population in the nearby mountains as a result of overwintering birds remaining to breed. Simulation-based analyses using microsatellite data reported in Rasner et al. (2004) were congruent with the hypothesis that the UCSD population experienced a founder event, with the Ne of the founding population ranging from 7 to 70 birds. Fudickar et al. (2017) used genome-wide SNP data to distinguish between year-round residents and wintering visitors at UCSD, and revealed marked differentiation of UCSD residents with respect to other Oregon junco populations. Divergence levels were in fact comparable to those found among dark-eyed junco lineages originated during the post-glacial radiation of the complex (Friis et al.2016; Friis et al. 2018; Friis & Milá 2020), suggesting that high-throughput SNP data may provide a level of phylogenetic resolution not afforded by the more limited genetic sampling of previous studies.
Here we use genome-wide SNP data and extensive geographic sampling of Oregon junco forms across the region to reconstruct the evolutionary and demographic history of the UCSD junco population. First, we use phylogenetic and co-ancestry analyses to identify the source junco population from which the UCSD population originated. We then apply demographic modelling to test whether the UCSD population is more likely to have resulted from a founder event ca. 30 generations ago, as previously assumed based on the date of appearance at UCSD, or instead diverged elsewhere for a longer period before colonizing the campus. We also use feather isotope ratios to determine whether birds that appear to be winter visitors at UCSD based on genetic data have indeed bred at a different latitude as opposed to being recent recruits into the resident breeding population. Finally, to infer potential selective pressures involved in rapid genetic divergence, we implement a genotype-environment association (GEA) analysis, and take advantage of a newly annotated version of the Junco hyemalis reference genome (Feng et al. 2020; Friis et al. 2018) to identify candidate loci under divergent selection.