Introduction
When wild organisms independently colonize ecologically similar environments, their descendants can be studied as replicated evolutionary experiments. This permits us to test for parallel or non-parallel changes in genotypes and phenotypes in nature, and thereby assess the predictability of evolution outside of laboratory settings. Several classic studies that have taken this approach provide powerful evidence of parallel evolution (see examples in stickleback and lizards (Colosimo, Hosemann et al. 2005, Mahler, Ingram et al. 2013)) . Conversely, some lab experiments have found that even under perfectly replicated conditions, evolution can be highly contingent on population ancestry and/or stochastic processes. For example, minor differences in standing genetic variation can determine whether or not an adaptive trait will evolve under relatively strong selection spanning thousands of generations, when using E.coli (Blount, Lenski et al. 2018). Other theoretical studies also find that population history affects the degree of repeatability and propensity of parallel evolution in different populations (Barghi, Hermisson et al. 2020, Otte, Nolte et al. 2021).
Feral study systems are well suited for testing fundamental evolutionary questions (Gering, Incorvaia et al. 2019, Mabry, Rowan et al. 2021), including the degree to which responses to recent selection are predictable and reversible in separate populations. When domesticated animals re-colonize the wild, they confront natural selection pressures that were often relaxed in their recent (captive) ancestors; feralisation can therefore be expected to drive rapid evolution in populations that harbour sufficient additive genetic variation in traits undergoing selection (Henriksen, Gering et al. 2018, Gering, Incorvaia et al. 2019, Gering, Incorvaia et al. 2019). Since independently feral populations have often undergone similar environmental change (e.g. increased social competition for territories and/or mates, requirements to seek out food and shelter, and interactions with naturally-occurring pathogen communities), the descendent populations can be compared to learn if parallel environmental changes drive the evolution of overlapping sets of genes and/or gene functions. These contrasts are especially informative (with respect to general features of feralisation) where colonization timelines and invaded habitats are similar among the allopatric populations that are being compared.
Another compelling feature of feral study systems is that focal populations often originate from relatively well-documented starting points. The domesticated sources of feral populations are often well studied and relatively well-known genetically, certainly in comparison to the actual wild progenitors of domesticated species that existed several thousands of years ago. Of course, independent feral populations of a given domesticated species may also be non- identical because of factors such as divergent founder (domesticated) source populations, random genetic drift, and/or localized admixture between feral, domesticated and wild relatives (Gering, Incorvaia et al. 2019). This makes feral populations excellent models for studying whether, and how, colonizing populations’ sources impact contemporary evolution.
A final benefit of studying feralisation is that evolutionary changes accompanying domestication have been the subject of intensive recent study (Wright 2015, Wright, Henriksen et al. 2020). This background knowledge permits testing, across a wide array of organisms, whether genes and/or functions that were previously modified under artificial selection undergo further change when domesticated taxa recolonize the wild (Johnsson, Williams et al. 2016).
Prior studies of feralisation have often focused on individual population case studies. For example, genomic analyses of feral chickens (Gallus gallus ) on Kauai Island, Hawaii, recently found evidence of rapid recent evolution at loci controlling traits that were also modified under domestication (e.g. genes that regulate behavior, reproduction, and growth) (Johnsson, Gering et al. 2016). Despite this functional overlap, selective sweeps found in Kauai’s feral fowl were largely different from known G. gallus domestication genes and improvement genes that were identified by selective sweep analyses (Johnsson, Gering et al. 2016). Strong selective sweeps in modern domesticated birds are more likely to represent ‘improvement’ genes that have been selected relatively recently during intensive modern breeding for layer and broiler chickens. For example, a coding change in the geneTSHR is found in virtually all modern domesticated birds (Rubin, Zody et al. 2010), but this mutation is largely lacking in archaeological domesticated chickens (Girdland Flink, Allen et al. 2014, Loog, Thomas et al. 2017). Given this distinction between selective sweep regions found in Kauai’s feral chickens vs. recent domestication/ improvement selective sweeps found in captive G. gallus , it is now important to compare additional (independently feral) gene pools. This will help determine how replicable feralisation processes are, and whether they are a model for parallel evolution.
The sources of feral populations are obviously of importance when comparing their gene pools. In the specific case of Kauai’s chickens, admixture between wild-living Red Junglefowl and escaped domesticated birds appears to have capacitated recent adaptation (Gering, Johnsson et al. 2015, Johnsson, Gering et al. 2016). This admixture most likely followed the releases of large numbers of domesticated birds during hurricanes Iniki (1992) and Iwa (1982) into the wild, as reported by locals and supported by chromosomal painting techniques (Martin Cerezo, Lopez et al. 2023). This timeframe also coincided with exponential growth of the feral population’s density on Kauai Island (Gering, Johnsson et al. 2015). Chickens were first brought to the Hawaiian Islands (including Kauai) by Polynesian settlers (circa 400 to 1200), and were thought to be Red Junglefowl imported for ritual and/or cockfighting purposes (though may also have been partially domesticated) (Kirch 2011, Thomson, Lebrasseur et al. 2014). Subsequent introductions of domesticated G. gallus to Hawaii followed European contact that began centuries later (1778) and involved domesticated chicken breeds of largely western origins. For example, large numbers of animals have been imported to the Hawaiian Islands via a local hatchery (Asagi) that was established in 1935 and specializes in Cornish Rocks (a broiler breed), White Leghorns (a layer breed), and smaller numbers of heritage breeds (www.asagihatchery.com). Importation and release of live G. gallus from commercial sources also coincided with introductions of non-commercial ‘backyard’ chickens kept by residents for egg production, meat production, and cockfighting which, though illegal in the US, remains highly popular on Kauai and throughout the Pacific region (Young 2014). Contemporary genomes of feral Kauai chickens are consistent with recent interbreeding among these diverse and asynchronously introduced source populations (Martin Cerezo, Lopez et al. 2023).
The clearest distinction between the histories of feral chickens on Kauai vs. Bermuda is that Red Junglefowl were never imported and released into Bermuda. This is largely because ancient Polynesians did not venture into the Atlantic region. Additionally, the landraces and ‘backyard’ varieties found in the two regions are likely non-identical, because the European explorers and slaves who colonized Bermuda beginning 1505 were closely connected (via culture and commerce) to European, African, and Caribbean locales whereas Hawaii was connected via more global recent trade routes and fashions (including Asian, Pacific, American, and European). Finally, there are no large commercial poultry operations in Bermuda and chickens are largely kept by Bermuda residents as fancy pets (vs. for food and cockfighting). The Bermuda Poultry Fancier’s Society was formed there more than a century ago, and current members reported to us that diverse domesticated breeds have long been imported to the island, including breeds developed in Europe (e.g. d’Uccles, Orpingtons, Faverolles, Friesians, Old English Game) and the Americas (e.g. Plymouth Rock, Brahma), from diverse stock sources (personal communication Ronnie Lopes, of Bermuda Poultry Fancier’s Society and US Master’s Cup Poultry Show). Escaped domestic G. gallus have been highly successful in colonizing the tiny 53.2 km2 Bermudian archipelago, reaching an estimated size of >30,000 individuals by 2011 and displaying a panoply of phenotypes seen in fancy chickens (e.g. rose, walnut, and duplex combs, polydactyly, barred and silver plumage patterns, and bantam morphology) (Government of Bermuda 2013). Similar to Kauai, Bermuda’s feral G. gallus densities have only recently reached their current high densities, with local authorities hypothesizing that damage from Hurricane Emily (1987) kickstarted exponential growth (Government of Bermuda 2013).
Feral populations that lack recent Red Junglefowl ancestry (like Bermuda) could be predicted to show evolutionary divergence from Kauai’s admixed fowl in their contemporary gene pools. On the other hand, feralisation might filter input sources and/or select for parallel genomic and phenotypic changes if a common feralisation ‘syndrome’ is adaptive in both regions. To evaluate these predictions, we examined genomic and phenotypic data from feral chickens of Bermuda and compared them to feral chickens from the Hawaiian island of Kauai. Kauai and Bermuda chickens have colonized strikingly similar environments. Both locales have subtropical climates in which feral chickens inhabit a variety of urban, suburban, rural, and wild habitats. Both Kauai and Bermuda ecosystems lack native terrestrial carnivores and, perhaps consequentially, have sustained unusually large and dense populations of feral G. gallus for decades (average group size for wild Red Junglefowl is around 6-8 birds, whilst groups on these islands can be as large as 30 or more birds (personal observation)). In the present study, feral birds from Bermuda (n=21) and Kauai (n=25) were sampled, phenotyped, and genotyped (10X density). They were then used to identify candidate feralisation selective sweeps, with the sweep regions further characterised using chromosomal painting techniques (see methods).