Inter- and intra-archipelago dynamics of population structure and gene flow in a Polynesian bird

Abstract

Islands are separated by natural barriers that prevent gene flow between terrestrial populations and promote allopatric diversification. Birds in the South Pacific are an excellent model to explore the interplay between isolation and gene flow due to the region’s numerous archipelagos and well-characterized avian communities. The wattled honeyeater complex (Foulehaio spp.) comprises three allopatric species that are widespread and common across Fiji, Tonga, Samoa, and Wallis and Futuna. Here, we explored patterns of diversification within and among these lineages using genomic and morphometric data. We found support for three clades of Foulehaio corresponding to three recognized species. Within F. carunculatus, population genetic analyses identified nine major lineages, most of which were composed of sub-lineages that aligned nearly perfectly to individual island populations. Despite genetic structure and great geographic distance between populations, we found low levels of gene flow between populations in adjacent archipelagos. Additionally, body size of F. carunculatus varied randomly with respect to evolutionary history (as Ernst Mayr predicted), but correlated negatively with island size, consistent with the island rule. Our findings support a hypothesis that widespread taxa can show population structure between immediately adjacent islands, and likely represent many independent lineages loosely connected by gene flow.

Introduction

Islands are relatively simple, natural theaters in which to study patterns and processes of diversification (Carlquist, 1974, Darwin, 1859, Grant, 1998, Jønsson et al., 2017, Mayr and Diamond, 2001, Wallace, 1881). Expanses of deep, open water act as natural barriers to gene flow for terrestrial taxa and thus generally promote isolation and differentiation of insular populations. The Southwest Pacific is an ideal system to study diversification processes (Mayr and Diamond, 2001) because the region comprises a myriad of oceanic islands of varying size, age, isolation, and topographic complexity (Hall, 2002). These islands host lineages with varying dispersal abilities (Weeks and Claramunt, 2014), each with different modes of evolution (Brown et al., 2013, Flannery, 1995, Mayr and Diamond, 2001, Steadman, 2006, Thibault and Cibois, 2017). For example, community assembly of insular floras and faunas can stem from colonization by mainland ancestors (i.e., the classic MacArthur and Wilson, 1963, MacArthur and Wilson, 1967); nonadaptive, in-situ diversification within archipelagos (Andersen et al., 2015, Darwell et al., 2019, Oliver et al., 2018, Sarnat and Moreau, 2011); and back-colonization from island to mainland (Filardi and Moyle, 2005). Indeed, this region has contributed disproportionately to the study of speciation theory and community assembly (Diamond, 1977, Diamond, 1970a, Diamond, 1970b, Diamond et al., 1976, Diamond and Mayr, 1976, MacArthur and Wilson, 1967, MacArthur and Wilson, 1963, Mayr, 1963, Mayr, 1942, Wilson, 1961, Wilson, 1959).

The evolutionary history of Pacific island lineages is complex, particularly in avifaunal communities. Birds pose a conundrum to the paradigm of allopatric speciation (Dobzhansky, 1937, Mayr, 1963, Mayr, 1942); namely, their dispersal ability should promote gene flow across oceanic barriers and thus limit the effects of geographic isolation. Yet, many lineages are well-differentiated among islands and archipelagos (Mayr and Diamond, 2001). For example, Ceyx and Todiramphus kingfishers (Andersen et al., 2015, Andersen et al., 2013), Zosterops white-eyes (Moyle et al., 2009), and Pachycephala whistlers (Andersen et al., 2014b, Jønsson et al., 2014) are classic examples of polytypic, geographic radiations whose complex patterns of phenotypic and genetic differentiation suggest that—at least in these lineages—gene flow between island populations ceases after colonization. This pattern describes the paradox of ‘great speciators’ whereby some lineages have the capacity to colonize archipelagos broadly and yet still give rise to ever more species despite their ability to colonize (Diamond et al., 1976). The great speciator hypothesis has received renewed attention with more refined methods to calculate diversification rates (Andersen et al., 2015, Irestedt et al., 2013, Moyle et al., 2009, Pedersen et al., 2018).

Alternatively, the supertramp hypothesis posits that vagile, r-selected species repeatedly repopulate remote, species-poor islands and thus do not speciate due to repeated colonization and rampant gene flow (Diamond, 1974). In addition to the stringently defined supertramps, other categories of tramp species follow similar assumptions but persist on a range of islands that vary in species-richness and size (Diamond, 1975, Mayr and Diamond, 2001). For example, Mayr and Diamond (2001) define a class “D-tramp” that is widespread on small islands and also occupies the interior of larger islands. Despite their theoretical importance and geographic ubiquity in the avifauna of the Southwest Pacific, the widespread supertramps and tramps remain relatively understudied with molecular methods (but see Cronk et al., 2005, Linck et al., 2016, Pepke et al., 2019, Toussaint et al., 2013). As such, major questions remain about the evolutionary origin of and patterns of differentiation within tramp taxa. For example, do tramps evolve from a sedentary, allopatric lineage or do they represent an ancestral state of high dispersal ability that is lost (Wilson, 1961, sensu taxon cycles; Wilson, 1959)? Furthermore, do isolated tramp populations represent distinct lineages or remain connected by gene flow due to their high dispersal ability?

Interestingly, there is some evidence that populations of tramp species are indeed variable across their range. This pattern may represent independent, incipient lineages, or simply the propensity for widespread taxa to show marked, random variation. For example, Mayr, 1932, Mayr, 1940 noted a high degree of seemingly random body size variation in the Polynesian wattled honeyeater (Foulehaio carunculatus), which he acknowledged was “inconvenient to the taxonomist” (Mayr 1940, p.264). Recognizing the pattern was intriguing beyond taxonomic conundrums, Mayr posited two hypotheses to explain its occurrence. The first, presented in his initial description of the variation in the species (Mayr, 1932), was that it was positively correlated with distance from the equator, in accordance with Bergmann’s Rule. The second hypothesis attributed this variation as largely meaningless and due simply to genetic drift in small, insular populations (Mayr, 1940). We used Mayr’s hypotheses as a framework to explore diversification, gene flow, and morphological variation in Foulehaio honeyeaters. We tested if latitude, island size, or phylogeny correlated with size by leveraging Mayr’s wing length data, which he found to be an effective proxy for size (Mayr, 1932).

The wattled honeyeater species complex (genus Foulehaio) is an ideal system in which to study within-species diversification dynamics and the origin of tramp lineages. The group is sister to Meliphacator provocator of Kadavu, Fiji, from which it diverged about three million years ago (Marki et al., 2017) and is otherwise closely related to, and nested phylogenetically within, a clade of deeply divergent, single-island endemics, each of which is sympatric with one of three Foulehaio species across Fiji and Samoa (Andersen et al., 2019, Andersen et al., 2014a, Marki et al., 2017). Two additional single-island endemics that are sister to this clade inhabit interior, highland forests of the Solomon Islands. The three species of Foulehaio span >1200 km across Central Polynesia, including Fiji, Tonga, Wallis & Futuna, Samoa, and American Samoa (Higgins et al., 2008, Mayr, 1945, Pratt, 1987). They occur on all of the main islands and many of the smallest islands in the region, including those separated by shallow-water barriers and those isolated by more than one thousand kilometers of open ocean. Two of the three recognized Foulehaio species are seemingly sedentary, allopatric lineages that are restricted to one or two large Fijian islands: F. procerior occurs on Viti Levu and its satellite islands, including the Yasawa Group and Ovalau and Vatulele islands, whereas F. taviunensis is found on Vanua Levu, Taveuni, and its satellite islands, including Kiowa and Qamea (Fig. 1). Conversely, F. carunculatus is widespread throughout Central Polynesia, occurring on dozens of islands in Fiji’s Moala Group and Lau Archipelago, Tonga, Wallis & Futuna, and the Samoan islands (Pratt, 1987). Curiously, F. carunculatus exhibits a checkerboard pattern in that it is present on some islands but absent on other, often neighboring, islands and seemingly in a random fashion (Andersen pers. obs.). The overall phylogenetic pattern of the Pacific honeyeaters (Andersen et al., 2019, sensu 2014a)—whereby single-island endemics that typically occur in highland, interior forests from the Solomon Islands to Samoa branch sequentially until the origin of the widespread tramp Foulehaio carunculatus—makes this study system particularly attractive to address questions regarding the derivation of tramp lineages.

Foulehaio honeyeaters have been studied in the context of higher-level phylogenetics, but only with limited genetic data and limited geographic sampling (Andersen et al., 2014a, Marki et al., 2017). The three allopatric species are up to 8% divergent at the mitochondrial ND2 locus, with little mitochondrial divergence within F. carunculatus (Andersen et al., 2014a, Modak, 2011). The widespread distribution of F. carunculatus hints at tramp-like behavior; however, there are phenotypic differences within this species, suggesting some degree of isolation. For example, F. carunculatus in Fiji has four distinct size classes that average larger in the south (Ono-i-Lau) and smaller northwards (Northern Lau) with intermediate birds in Southern Lau and Matuku (Mayr, 1932). Moreover, observed, yet unquantified, plumage differences in specimens from across the distribution of F. carunculatus are notable and further suggest within-species differentiation in this group (Mayr, 1932).

Few studies have leveraged modern DNA datasets to explore ongoing, within-species dynamics on Southwest Pacific archipelagos (but see Andersen et al., 2017, Cowles and Uy, 2019, Cronk et al., 2005, Darwell et al., 2019, Gyllenhaal et al., 2020; Linck et al., 2016, Toussaint et al., 2013), and no study has focused on the entire Foulehaio radiation within a population-genomic framework. Here, we assessed patterns of diversification and gene flow among island populations in the Foulehaio species complex. We used genome-wide data to explore the evolutionary history between island populations of Foulehaio honeyeaters, employing both species tree and population genetic frameworks. We sampled broadly across the distribution of Foulehaio and analyzed thousands of single nucleotide polymorphisms (SNPs) derived from target capture of ultraconserved elements (UCEs). We also used trait data—namely linear wing measurements as a proxy for body size—to test for a correlation between genetic and phenotypic differentiation in a widespread tramp species. The goal of our study was to explore the level of population structure and gene flow within Foulehaio across ocean barriers. We investigated questions pertaining to the evolutionary origin of this widespread tramp, evaluated the genetic divergence among isolated populations of Foulehaio, and assessed potential drivers of morphological differentiation among these populations. Our study contributes to the broader study of island biogeography by establishing the degree to which widespread island taxa have differentiated.