Introduction
Invasive populations are useful systems to investigate responses to novel environments, providing insight into mechanisms underlying invasion success and native species’ capacity to adapt to a changing world (Moran & Alexander 2014). Despite this opportunity, these studies often examine only one introduction, reducing their power to draw robust conclusions that are broadly applicable (Packer et al. 2017). For this reason, there is a growing interest in studying invasive species that have been introduced to multiple geographically and environmentally diverse localities (Kueffer et al. 2013; Packer et al. 2017). In this respect, the European starling (Sturnus vulgaris) is an excellent system to investigate evolutionary responses to a wide range of introduced environments, from tropical Fiji to temperate Argentina (Pinto et al. 2005).
European starlings are native to the Palearctic but repeatedly have been introduced to novel environments, flourishing in their invasive ranges (Long 1981). Starlings have now been introduced to every continent barring Antarctica (Rollins et al. 2006, Figure 1). Their invasion success likely results from a suite of life-history and behavioral traits that may facilitate ecological flexibility. For example, they are often classified as diet generalists, preferring insects, but they will eat most other foods depending on availability of resources (Cabe, 1993). Another feature that likely plays a role in European starlings’ ability to persist in new localities is their flexibility in patterns of seasonal migration. Although not all starling populations are migratory (e.g. in Australia and New Zealand, Higgins et al. 2006), in populations that are migratory, there is a great deal of individual variation in migratory behavior (i.e. individuals can be differentially migratory from year to year; Blem 1981; Feare, 1984). Some research suggests that seasonal migration may be an adaptive strategy in response to seasonality, therefore migratory flexibility in starlings may allow them to persist in seasonal environments and facilitate range expansion (Winger et al., 2019). This trait may contribute to differences in population structure across introductions. 
European starlings were introduced to North America in 1890 as part of an American Acclimatization Society initiative to populate Central Park with the birds from Shakespeare’s plays (Cooke 1928; Phillips 1928). The initial introduction consisted of approximately 60 individuals released in 1890 and 40 more in 1891, leading to a total of ~100 individuals released into Central Park in New York City (Cabe, 1993). From this founding population, starlings have expanded their range across all of North America where their current population exceeds 200 million individuals, over one-third of the global population of this species (Feare, 1984). This range expansion has taken place in the last 130 years, demonstrating their ability to persist in a heterogeneous novel environment. Given the diverse environments colonized by starlings in North America, it is interesting that nuclear markers indicate that little population structure exists (allozymes, Cabe, 1998; single nucleotide polymorphisms, Hoffmeister et al. 2019).
Other starling introductions from the 19th century have been previously studied, including the mid-19th century Australian introductions (Rollins et al, 2009; 2011; 2016) and the late 19th century South African introduction (Berthouly-Salazar et al, 2013). In Australia, up to sixteen different introduction attempts have been made with birds originating from the United Kingdom, from 1856-1881, with only two resulting in recorded established populations from ~165 original birds (Higgins et al 2006; Long 1981). Nuclear and mitochondrial markers identified concurrent population structure across the Australian range, and nuclear polymorphisms were associated with environmental variables in that population (e.g. aridity; Cardilini et al. 2020, Rollins et al. 2009, 2011). In contrast to the high levels of propagule pressure in Australia, only one introduction to South Africa of ~18 birds originating from Britain in or around 1897 has been recorded (Winterbottom and Liversidge 1954). That introduction enables a powerful comparison with the North American introduction because of similarities in timing of these events (1897 and 1890, respectively). Both the Australian and South African introductions have reduced mitochondrial genetic diversity in comparison to the native source population in the UK (Berthouly-Salazar et al. 2013; Rollins et al. 2011).
Founding population sizes during introduction are often small, resulting in genetic bottlenecks and lower genetic diversity than in the native range (Baker & Stebbins 1965; Nei et al, 1975). However, numerous insights from studies of other invasions suggest that decreased genetic diversity at introduction may not hinder these species’ ability to become established in novel environments (Frankham 2005; Dlugosch et al, 2015). Factors such as the number of introduction attempts, the timing of these attempts, dispersal patterns in the introduced range and the rate of population expansion may play a larger role in shaping patterns of genetic diversity and ultimately contributing to successful colonization. A wide body of evidence suggests that adaptation in introduced ranges occurs rapidly, and this does not appear to be reliant on genetic diversity (Rollins et al. 2013).
Here, we use mitochondrial control region sequence data to examine starling population structure in North America and compare mitochondrial genetic diversity in populations from the native-range and from three established invasions: North America, Australia, and South Africa. Although the limitations of using mitochondrial DNA in population genetic analyses have been well characterized (Ballard and Whitlock 2004; Bazin et al, 2006), there are several benefits associated with its use. First, previous studies of starlings in Australia, South Africa and the UK used mitochondria control region sequence data, so the comparative strength of our study is predicated on using the same marker. Second, Australian studies that have compared population structure using mitochondrial sequence data to that of microsatellite (Rollins et al. 2011) and single nucleotide polymorphism data (Cardilini et al. 2020) found similar patterns, supporting the validity of our approach. Third, mitochondrial DNA is still one of the most reliable sources of DNA that can be extracted from historical museum specimens (Ramakrishnan et al., 2009; Mason et al., 2011; Guschanski et al., 2013), and population analyses using historical specimens rely on comparable datasets from modern birds, such as this. Finally, although mitochondrial DNA cannot provide a complete evolutionary picture, it is especially useful as evidence to clarify recent changes in a population (Zink and Barrowclough, 2008). This is especially true of the non-coding control region, which has high nucleotide diversity (Saccone et al., 1991).
In this study, we use this unique biological system that features multiple, independent, and documented introductions to investigate how propagule pressure (e.g. the number of introductions), environmental factors, and the expansion rate in introduced ranges influence contemporary population structure and genetic diversity. Based on previous research using nuclear markers, we predict low levels of population structure within North America. We predict that the mitochondrial diversity of the North American population will be lower than that of Australia, where multiple introductions were made (Jenkins 1959), and these occurred prior to and had a greater number of propagules than the New York introduction (Australian introductions started in 1854; Jenkins 1959). Further, we predict similar levels of genetic diversity in South Africa and North America, due to similarities in timing of introductions and propagule pressure. We discuss microevolutionary changes that have occurred since the introduction of these populations across the world.