Population differentiation in the white-fronted chat (Epthianura albifrons) at a continental scale: implications for dispersal, biogeography and conservation

Introduction

Habitat fragmentation is major cause of species decline, producing isolated local populations that suffer from the negative effects of inbreeding and demographic stochasticity (Haddad et al. 2015; Frankham et al. 2019). Strong dispersal ability, which allows species to cross barriers, is therefore an important species attribute that provides resilience to local population extinction and ultimately species decline (Niebuhr et al. 2015). Because of the power of flight, birds are generally less susceptible to the demographic and genetic consequences of habitat fragmentation than less vagile taxa such as reptiles or mammals (Frankham et al. 2010). However, historic geographical barriers have still been important in avian speciation, and isolation of island populations has resulted in numerous examples of morphological differentiation (Mayr and Diamond 2001).

Determining the dispersal ability of animals is not easy, although advances in telemetry have made great inroads into this important ecological discipline (Kays et al. 2015). Historically, dispersal in birds has been inferred from appearance and disappearance of birds in particular regions between seasons and has provided a broad understanding of movements associated with migration and nomadism (Keast 1968). Mark/recapture studies have been important in refining understanding of these patterns of movement, and also in identifying local movements (Paradis et al. 1998; Fandos et al. 2023). However, recaptures at large spatial scales occur at extremely low probability, and issues with tracking devices for very small individuals means that our knowledge of dispersal in small passerines is limited.

Morphological differences between populations have often been inferred to reflect isolation and thus the presence of barriers to dispersal (Fan et al. 2024). In particular, many insular populations of birds are recognised as subspecies on this basis (Mayr and Diamond 2001). By extension, a lack of morphological difference may indicate the absence of barriers and the capacity for dispersal over distances narrower than the extent of the barrier.

For example, the white-fronted chat (Epthianura albifrons) is considered to have the capacity to disperse across potential oceanic barriers because of an absence of morphological variation in island populations, reinforced by punctuated observation records that imply nomadism (Schodde and Mason 1999). However, the determination of the absence of morphological variation is equivocal, with more recent data (Major 2012) supporting earlier measurements that had historically been used to infer subspeciation (Mathews 1912; Keast 1958). Resolving this uncertainty is important, because the white-fronted chat is a threatened species (NSWSC 2010a, 2010b) with naturally fragmented saltmarsh as its core habitat (Ashcroft and Major 2013), such that understanding its dispersal capacity is a fundamental issue for conservation efforts to prevent further decline.

Robust information for identifying barriers to movement and hence dispersal ability in a species can also be obtained from the analysis of genetic variability between locations across its range (Haig et al. 2011). Furthermore, a population genetic approach has the benefit of establishing whether movements actually contribute to the breeding population, which is not necessarily the case (Coulon et al. 2010). Moreover, a genetic approach has the additional benefit of directly assessing the extent of any inbreeding, which is one of the deleterious effects of habitat fragmentation that conservation management seeks to address (Frankham et al. 2010). The aim of this study is to use highly variable genetic markers (microsatellites) to identify any population discontinuities in the distribution of the white-fronted chat, and to determine the extent to which the species is threatened by anthropogenic landscape modification.

Methods

Study sites and species

The white-fronted chat is a ground-foraging insectivore in the honeyeater family (Meliphagidae). It is distributed across southern mainland Australia from 200 km north of Perth in the west to 500 km north of Sydney in the east, as well as the continental islands of Tasmania, King Island, Flinders Island, Kangaroo Island and Rottnest Island (Higgins et al. 2001). White-fronted chats prefer open habitats, particularly salt marsh vegetation, and they are not found in forest or woodland (Higgins et al. 2001). The species is non-territorial and gregarious, foraging in flocks and often nesting in loose colonies (Major 1991a, 1991b). They are considered to be nomadic in some parts of their range (Keast 1958), although they are also known to be sedentary in other areas, particularly where they inhabit coastal salt marsh (Major 1991a; Higgins et al. 2001).

E. albifrons was sampled from 15 sites distributed across the ~3000 km known range of the species. These sites were selected based on recent records in eBird (Sullivan et al. 2009) and chosen to measure genetic variability associated with geographical distance and isolation by potential barriers to dispersal (Fig. 1). To investigate the effect of oceanic barriers, sites were chosen on Rottnest Island (18 km from mainland), King Island (90 km from mainland) and two locations in Tasmania (130 km from mainland), as well as sites in proximal mainland locations (Mandurah, Torquay, Laverton). To investigate the effect of urban barriers, sites were chosen in two nature reserves that were surrounded by urban development in Sydney, at Homebush Bay and Towra Point. To investigate the effect of continuous forest as a barrier, sites were chosen to the east (Swan Bay, Nowra) and west of the Great Dividing Range (Macquarie Marshes, Lake Cargelligo, Hay, Bungendore), and on the eastern (Freycinet) and western (Montagu) sides of Tasmania.

Fig. 1.

Location records (grey circles) and position of study sites (black triangles) for DNA collection from 15 populations of white-fronted chats spanning continental Australia and adjacent islands. Potential dispersal barriers (Nullarbor, Bass Strait and Great Dividing Range) are indicated by dashed lines. Study site names are subsequently referred to by three-letter abbreviations: ROT, Rottnest Island; MAN, Mandurah; KIN, King Island; MON, Montagu; FRE, Freycinet; TOR, Torquay; LAV, Laverton; BUN, Bungendore; NOW, Nowra; TOW, Towra Point; HOM, Homebush Bay; SWA, Swan Bay; MAC, Macquarie Marshes; CAR, Lake Cargelligo; HAY, Hay.

Results

Clear population structuring was evident in all Structure models, with five distinct genetic populations (K = 5) best explaining the dataset as determined by both maximum mean log-likelihood (Fig. 2a) and rate of change in log-probability between successive K values (Fig. 2b).

Fig. 2.

Determining the number of genetic populations from (a) the maximum (least negative) mean log-likelihood value, and (b) maximum second order rate of change of the likelihood function (Δ K). Models are derived from eight simulations of each of 15 model population structures ranging between 1 and 15 genetic populations. Using both criteria the ‘best’ models are those with K = 5. Error bars represent standard deviations (n = 8).

The two populations sampled on the island of Tasmania (Montagu and Freycinet) showed similar genetic composition but were distinct from the remaining 13 populations (Fig. 3). The other two oceanic island populations (Rottnest Island and King Island) also had unique population signatures. Of the mainland populations, Homebush Bay and Towra Point, the two populations surrounded by urban development, had distinct population signatures but the remaining mainland populations showed little variation among each other, while being distinct from all island populations (Fig. 3).

Fig. 3.

Population genetic structure of 264 white-fronted chats grouped into 15 sampling locations (see Fig. 1 for full location names) identified using the program Structure. Each bar represents an individual bird and shows the percentage contribution of five genetically distinct populations (represented by different colours) to its genetic composition.

Genetic diversity measured by both allelic richness (F = 6.76, P < 0.01) and heterozygosity (F = 5.33, P < 0.01) varied significantly among populations (Fig. 4). The mean effective number of alleles per locus ranged between 3.0 (Towra Point) and 5.5 (Macquarie Marshes) with the two populations surrounded by an urban matrix having significantly lower allelic richness than the other mainland populations, whereas the ocean islands had intermediate values. Pairwise comparisons indicated that the population on the smallest oceanic island (Rottnest) had significantly lower allelic richness than most mainland populations (Appendix 1). Similar patterns were apparent for mean unbiased heterozygosity, which ranged between 0.63 (Towra Point) and 0.79 (Macquarie Marshes).

Fig. 4.

Variation in genetic diversity amongst 15 populations of white-fronted chats represented by (a) mean (±s.e.) number of effective alleles per locus, and (b) mean (±s.e.) unbiased heterozygosity. Blue bars denote ocean-island populations; black bars denote urban-island populations and red bars denote open mainland populations.

The genetic distances between all island and mainland populations were significantly different, including mainland pairwise F_ST comparisons with ocean island populations and with urban island populations (Table 1). Additionally, all pairwise F_ST comparisons among islands were highly significant. Pairwise differences between mainland populations were not significant with the exception of the NOW/CAR, NOW/BUN and TOR/HAY pairs (Table 1). D_est and F_ST values were highly correlated (R² = 0.91) and accordingly, pairwise D_est showed the same pattern (Appendix 2). Ordination and clustering of the D_est matrix showed that the distances among island populations were large, and that the mainland populations clustered together (Fig. 5).

Fig. 5.

Genetic differentiation amongst 15 populations of white-fronted chats represented by (a) ordination and (b) group-averaged clustering of the pairwise D_est matrix using PRIMER. The ocean-island (FRE, MON, KIN, ROT – blue) and urban-island (TOW, HOM – black) populations differed significantly from each other and from all the open mainland populations (NOW, MAN, TOR, BUN, LAV, SWA, MAC, CAR, HAY – red) as determined by pairwise F_ST. Refer to Appendix 2 for pairwise D_est matrix.

There was no evidence of genetic isolation by distance across mainland populations (Fig. 6) despite the large geographic distances compared with the proximity of island populations (Mantel R² = 0.154).

Fig. 6.

Correlation between genetic and geographic distances of open mainland sites, demonstrating an absence of isolation by distance as determined by Mantel test (R² = 0.154, P > 0.5).

Discussion

The white-fronted chat is a threatened species that may be susceptible to population decline due to habitat fragmentation, yet conservation efforts are hindered by a limited understanding of the species’ dispersal capacity and genetic connectivity. Sampling across the ~3000 km breadth of the species’ geographical range, genetic analysis revealed a lack of isolation by distance, and no population differentiation in white-fronted chats across mainland Australia, apart from two populations surrounded by an urban matrix that have been described previously (Major et al. 2014). In contrast, the populations on three oceanic islands were genetically distinct from the mainland populations and from each other, even though in one case the extent of the oceanic barrier from the mainland was only 18 km. These findings are highly relevant for understanding the species’ dispersal capacity, biogeographic history, and requirements for conservation.

While the north–south range of the white-fronted chat is continuous (Fig. 1) the east–west distribution is separated by both the Great Dividing Range and the Nullarbor Plain, with only scattered records of the species across the southern coastal fringe of Australia. This effective discontinuity has previously been postulated as a biogeographical barrier to the eastern and western distribution of the species, with a western subspecies, E. a. westralensis, distinguished by its paler colouration, separated from the eastern E. a. albifrons (Mathews 1912). However, this morphological variation has been disputed, ultimately resulting in the rejection of this subspecies (Keast 1958). Our genetic analysis supports this decision. In contrast, significant genetic structure has been identified by microsatellite analysis over much smaller spatial scales for other small passerines such as the eastern yellow robin (Eopsaltria australis), which exhibits strong north/south population structuring across a 1500 km continuous extent of suitable habitat in eastern Australia (Pavlova et al. 2013). Moreover, this species does not span the east/west extent of the continent, with the congeneric western yellow robin (E. griseogularis) replacing it in the west. Several other genera of small passerine have distinct eastern and western counterparts, e.g. eastern (Acanthorhynchus tenuirostris) and western (A. superciliosus) spinebills (Christidis and Boles 2008). The contrasting absence of any population structuring for the white-fronted chat both along latitudinal and longitudinal gradients suggests that the species has a strong capacity for dispersal, at least across non-urbanised land. White-fronted chats have been recorded in a range of open habitats including saltmarsh, chenopod shrublands and farmland but are generally absent from forests and woodlands (Higgins et al. 2001). Because these open habitats are relatively continuous, unlike southern Australian woodlands that have suffered severe habitat fragmentation, isolation by resistance or isolation by distance is less likely to be observed in white-fronted chats than in similar-sized woodland passerines including confamilial honeyeaters (Amos et al. 2014). In contrast, the open-habitat frequenting tawny-crowned honeyeater (Gliciphila melanops) shows no structure between south-eastern and south-western populations (Dolman and Joseph 2012). The lack of any isolation by distance observed in white-fronted chats is also consistent with earlier description of the species as being nomadic (Keast 1958).

Differences in nomadic behaviour, and by inference mobility, have historically been considered an important driver of subspeciation in Epthianura (Keast 1958) and for the lack of subspeciation in mainland E. albifrons. However, Keast (1958) identified a significant difference in bill length in Tasmanian E. albifrons and considered Bass Strait to be an effective barrier to gene flow, supporting the subspecies E. albifrons tasmanica described by Mathews (1912). Taxonomic delineation of Tasmanian E. albifrons has since been rejected (Schodde and Mason 1999) with the assertion that there were no morphological differences and presumably at least sporadic movements of this nomadic species across Bass Strait. Similar-sized passerines, e.g. flame robin (Petroica phoenicea), silvereye (Zosterops lateralis) and grey fantail (Rhipidura albiscapa), are known to cross Bass Strait (Garnett et al. 1991), making it reasonable to infer, prior to this study, that white-fronted chats might do likewise. However, unlike these recognised migrants, white-fronted chats have ‘short, broad, somewhat rounded wings’ (Higgins et al. 2001) rather than the longer, pointed wings more typical of migrants (Vágási et al. 2016).

Morphometrics of the same individuals as those sampled in the current study revealed significant differences in body size between island and mainland populations (Major 2012). This was interpreted as indicative of a restricted gene flow across oceanic barriers, challenging the assumption of movement across Bass Strait. The highly distinct genetic structures of the mainland, Tasmanian and geographically intermediate King Island populations observed in the present study support this challenge and perhaps justify reassessment of subspecies status. Regardless, the logical implication of these collective data is that white-fronted chats are not as highly mobile as has been interpreted from their oft-stated nomadic status: their movements, or the breeding success of migrants, are insufficient to prevent genetic drift or local adaptation. This establishes a paradox in that the lack of genetic differentiation across the mainland compared with species like E. australis is explained by high mobility, but that the differentiation between islands and the mainland is explained by a lack of it.

A plausible explanation of this paradox is that nomadism and flight capability may not be correlated. While chats can be nomadic, flying between areas of transient high food availability (Major 1991b), between drought and rainfall areas (Williams and Main 1976), or between patches of discontinuous habitat (Lindsay et al. 2015), this does not mean that they are capable fliers over long distances. They may have the predilection to move, but limitations in flight morphology may mean that they require rest stops, preventing them from traversing inhospitable habitat in which they cannot land. In contrast, species such as Z. lateralis, R. albiscapa and P. phoenicea are migrants with strong flight capability (Garnett et al. 1991), but they are also extremely territorial with high breeding-site fidelity, facilitating genetic differentiation. With weak flight capability, chats may not readily be able to traverse an oceanic barrier where resting opportunities are unavailable. Given that chats are never found in urban habitats and are particularly sensitive to human disturbance (Jenner et al. 2011), urban areas may also form a hostile barrier leading to genetic differentiation among relict populations surrounded by urban encroachment.

Our microsatellite data do not have the resolution to determine how the differentiation of the island populations has evolved, and future research measuring gene flow using an SNP array would be rewarding. However, it seems most likely that the island populations are relicts of a more continuous population. In the case of the urban islands (Homebush Bay and Towra Point), museum records indicate that prior to urbanisation, birds were found in wetlands that linked the remnant saltmarshes, creating a continuous distribution across what is now a major urban centre (Jenner et al. 2011). In the case of the oceanic islands (Rottnest Island, King Island and Tasmania), all three were joined to the mainland before sea-level rise over the last 7000–12,000 years. Genetic drift and/or local adaptation are likely to have occurred since separation, with some evidence of loss of genetic diversity in these smaller populations, and more recently in the small urban isolates.

Although the white-fronted chat is relatively abundant and remains secure in the more southern part of its range, a lack of strong dispersal ability preventing it from crossing oceanic and urban barriers invokes lessons for future conservation. The two urban populations in this study are listed as Endangered, and the species as Vulnerable in New South Wales (NSWSC 2010a, 2010b). Saltmarsh is the preferred habitat for the species and the natural fragmentation of this specialist habitat is being further fragmented by anthropogenic development and climate change, particularly in the coastal zone (Laegdsgaard et al. 2009; Ashcroft and Major 2013). Genetic structuring has also been identified among populations of the congeneric and Critically Endangered yellow chat (Epthianura crocea) (Houston et al. 2017), another habitat specialist with highly disjunct distributions (Keast 1958). This study demonstrates that it would be foolhardy to consider that the nomadic disposition of chats insulates them from habitat fragmentation. Populations isolated by hostile barriers such as expanses of water (e.g. Rottnest Island, King Island, Tasmania) or urbanised land (e.g. Homebush Bay, Towra Point) should be considered isolated populations and managed accordingly.

Supplementary material

Supplementary material is available online.