Seasonal migrations between geographically distant breeding and non-breeding grounds are common for threatened marine animals including turtles (Hays, 2008), whales (Mate et al., 2015), seabirds (Pinet et al., 2011), and sharks (Werry et al., 2014). These migratory behaviors can be complex, varying spatiotemporally and by sex and age (Druskat et al., 2019; Blevins et al., 2022). Despite the challenges of tracking migratory species, animal movement data are critical to our understanding of the species’ spatial ecology, population structure, vulnerability to anthropogenic threats and ability to adapt to environmental change (Block et al., 2005; McCauley et al., 2015; Paiva et al., 2015). Changes in ocean circulation and increased temperatures due to climate change are impacting marine species, causing alterations to species distributions, migration routes and gene flow (Robinson et al., 2009; Bates et al., 2014; Donelson et al., 2019). The impact may be more profound along subtropical Western Boundary Currents (WBCs), where the rate of sea surface warming is two to three times faster than the global mean (Wu et al., 2012). The East Australian Current (EAC) is a WBC that brings warmer waters from the tropics in Queensland to the temperate regions of New South Wales (NSW) (Wu et al., 2012) and it has experienced a poleward shift extending 92 km south over the past three decades (Li et al., 2022). Shifts in marine species distributions and assemblages as well as changes to ecological interactions between macroalgae, herbivores and higher-order predators have been increasingly documented due to the warming EAC (Thompson et al., 2009; Wernberg et al., 2011; Carroll et al., 2016; Vergés et al., 2016). Understanding these changes is critical to adaptive wildlife management frameworks to increase species resilience to climate change and reduce cumulative threats to the marine ecosystem (Allen and Singh, 2016; Ogburn et al., 2017).

Marine turtles are long-lived, circumglobally distributed reptiles that undertake large-scale migrations between breeding grounds and feeding areas (Hamann et al., 2010). Many marine turtle populations have experienced severe population declines as a result of anthropogenic threats, including climate change, unsustainable harvesting, pollution, marine debris, habitat degradation and disease (Wallace et al., 2011). The six species of turtles occurring in Australian waters are protected under both state and national legislations and international agreements (Commonwealth of Australia, 2017). Green turtles (Chelonia mydas) are listed as Vulnerable in Australia under the Environment Protection and Biodiversity Conservation Act 1999 (EPBC Act, 1999) and Endangered globally by the International Union for Conservation of Nature (IUCN Red List, Seminoff, 2004), for which nine genetically distinct genetic stocks (i.e. populations, Management Units [MUs]) have been identified in Australia and at least 25 in the greater Indo-Pacific region (Jensen et al., 2016b). Despite long-term research programs (Limpus, 2008), the demographics and geographic boundaries of many genetic stocks in the Indo-Pacific region are not well understood. To assess the impact of threatening processes on these genetic stocks, a greater understanding of migration patterns between breeding and foraging grounds is critical to develop effective conservation management strategies that mitigate threats at the appropriate geographic and population scale (Hamann et al., 2010; Komoroske et al., 2017). This is especially the case in NSW, which has been identified as an important foraging site for green turtles (Commonwealth of Australia, 2017) but where multiple threats to turtles are known to occur (de Gouvea Pedroso et al., 2020; Peddemors and Dalton, 2022). Further, green turtles are now more frequently sighted in southern NSW waters compared to 30 years ago ( Figure 1 ; Atlas of Living Australia), likely due to the southern extension of the EAC and increasing sea surface temperatures in this region.

Research using tagging, satellite telemetry and genetic techniques have greatly improved our knowledge of population structure, spatiotemporal migration patterns and foraging ecology of green turtles globally. Green turtles undergo ontogenetic habitat and diet shifts during their life histories (Arthur et al., 2008). After hatching, green turtles migrate to pelagic environments and are influenced by oceanic currents for ca. 3 - 5 years (Musick and Limpus, 1996; Reich et al., 2007). Juvenile green turtles recruit to marine and estuarine habitats, feeding primarily on seagrass and macroalgae (Limpus, 2008; Esteban et al., 2020). As juveniles through to adults, green turtles spend a large proportion of their lives on foraging grounds and can show high site fidelity to specific foraging areas (Broderick et al., 2007; Dutton et al., 2019; Siegwalt et al., 2020). Adult green turtles temporarily return to breeding grounds, which based on global datasets, can be over 2,800 km (average ~ 800 km) away from foraging grounds (Hays and Scott, 2013). These adult migrations are significantly less than the mean distances for post-hatchlings migrating away from the natal nesting sites during their pelagic juvenile development stage (~ 4,500 km, Hays and Scott, 2013). Migration distances and foraging locations can vary considerably among individuals of the same genetic stock (Read et al., 2014; Tanabe et al., 2023) and green turtle foraging grounds are often found to support multiple stocks (Komoroske et al., 2017).

Estimating the contribution of source populations to foraging aggregations using Mixed Stock Analysis (MSA) has been widely used to understand population boundaries and complex migration patterns of marine megafauna, including turtles (Bowen et al., 2007; Vargas et al., 2008; Garofalo et al., 2013) and whales (Schmitt et al., 2014; Brüniche-Olsen et al., 2018). MSA has also been instrumental in improving fisheries management outcomes by identifying biologically relevant population units (Dahle et al., 2018; Fields et al., 2020; Euclide et al., 2021). Because green turtle genetic stocks have generally been delineated using the maternally inherited mitochondrial DNA (mtDNA) control region, mtDNA haplotype frequencies can be used to assess the mixed stock composition at foraging grounds across their global distribution (Proietti et al., 2013; Dutton et al., 2019; Jensen et al., 2019; Piovano et al., 2019; Jensen et al., 2020). MSA can also contribute to our understanding of the impact of climate change on green turtles (Jensen et al., 2018). A combination of factors, including ocean current patterns, geographic distance, and relative genetic stock size have been found to significantly influence the composition of green turtle foraging grounds (Lahanas et al., 1998; Jensen et al., 2020; Kynoch et al., 2022; Stahelin et al., 2022). Findings of foraging turtles that originate from distant rookeries (Naro-Maciel et al., 2014; Read et al., 2015; Jones et al., 2018) and size class segregation at different foraging grounds (Hamabata et al., 2018) indicate that additional drivers, such as active swimming, geomagnetic imprinting and environmental conditions, may influence foraging habitat selection by juveniles (Lohmann et al., 2008; Naro-Maciel et al., 2014; Read et al., 2014; Putman and Mansfield, 2015).

This study aimed to assess the genetic diversity and genetic stock origins of green turtles stranded along the NSW coast in southeastern Australia. The adults, sub-adults and juveniles sampled were presumed to represent the foraging turtles of coastal NSW based on diet analysis conducted on a subset of samples (Glen, 2021) and the rarity of nesting events in NSW (NSW Department of Planning and Environment). Identifying the connectivity between genetic stocks and NSW foraging grounds will increase our understanding of factors influencing migratory patterns in this species and how threatening processes in NSW are impacting green turtle genetic stocks in the greater Indo-Pacific region.

Results of the MSA showed that green turtle foraging grounds in NSW support multiple genetic stocks but are dominated by those from the southern GBR, and to a lesser extent, New Caledonia. The similarity of estimates using uniform priors and priors weighted by stock size and geographic distance, and the corresponding CIs suggests that the MSA results are robust. The importance of these findings, based on opportunistically collected samples from stranded animals, is two-fold. Firstly, the quantification of stock composition provides critical insights into the migration patterns and post-migration residence of foraging green turtles at the southern-most extent of the species’ distribution in Australia, in a region where very little research has been conducted. From a broader perspective in the Indo-Pacific, the MSA of stranded turtles indicating their origins from southern GBR and New Caledonia supports limited tag recovery data from southern GBR turtles (Limpus, 2008), and documents the proportional representation of these two stocks/MUs in the temperate waters of NSW, thus better informing the way this species should be managed in NSW.

Finding that over ~ 80% of adult-sized, sub-adult and juvenile turtles in NSW originate from the closest genetic stock in the southern GBR provides strong evidence that geographic distance influences stock composition in NSW. Similar to other turtle foraging aggregations in the Indo-Pacific (Read et al., 2015; Jensen et al., 2016b; Jones et al., 2018; Piovano et al., 2019), the estimated composition of the NSW foraging aggregations was unlikely influenced by genetic stock size based on similar estimates obtained using uniform priors versus weighted priors that incorporated stock size. For instance, there was minimal contribution from northern GBR stock and estimates using the stock size-weighted model were only slightly higher (< 2.5%) compared to uniform priors, even though this genetic stock is over four times larger than the southern GBR stock. Major ocean surface currents also appear to be influencing the location of recruitment to neritic foraging sites in this region ( Figure 2 ). There is compelling evidence that ocean currents influence the early life stage migration and movements of green turtles and other marine turtle species globally (LaCasella et al., 2013; Proietti et al., 2014; Jensen et al., 2020). In the southwest Pacific, green turtle post-hatchlings can remain in the pelagic zone for five years before recruiting to foraging sites (Chaloupka, 2002) and are likely to spend this time in the EAC’s offshore warm water eddies (Boyle, 2006). This behavior may be represented by the post-hatchling turtles in this study with CCL < 38 cm. The prevailing EAC is likely facilitating the movement of post-hatchlings from southern Queensland to NSW, while the South Equatorial Current, which bifurcates into the North Caledonian Jet (NCJ) and South Caledonian Jet (SCJ), facilitates movement from New Caledonia to NSW together with the EAC ( Figure 2 ). These currents may explain the lack of turtles in NSW that originated from the Coral Sea and northern GBR stocks, as well as the higher contribution of the New Caledonia stock in northern NSW compared to central NSW ( Figure 3 ). The influence of these currents has been proposed to explain stock composition of foraging green turtles at southern foraging grounds in Queensland, which are very similar in composition to NSW foraging aggregations (Jensen et al., 2016b; Jones et al., 2018). This is further supported by long-term tagging studies that documented green turtle breeding migrations, in both directions, between New Caledonia breeding grounds and foraging grounds in southern and northern GBR (Read et al., 2014).

Greater variation in the natal origin of post-hatchlings was observed compared to adult-sized, subadult and juvenile turtles that had presumably recruited to NSW foraging areas, with higher estimated contribution of the Coral Sea and ‘other’ stocks for post-hatchlings. The estimated stock composition for post-hatchlings were comparable to a foraging population in the northern GBR (Jones et al., 2018). While the large 95% CIs indicate uncertainty around these estimates, Boyle (2006) similarly estimated a 28% contribution of the Coral Sea stock for stranded post-hatchlings in southern Queensland and northern NSW. Significant differences in stock composition for loggerhead turtles (Caretta caretta) in different ontogenetic phases has also been documented (Bowen et al., 2005). In that study, haplotype frequency data indicated that pelagic post-hatchlings in the North Atlantic had highly mixed origins, while subadults generally recruited to neritic foraging grounds that were geographically closer to their natal rookery (Bowen et al., 2005).

A small proportion of adult-sized, subadult and juvenile turtles stranded in NSW were also estimated to originate from more distant genetic stocks based on the haplotypes they possessed ( Tables 2 , 3 ), however, stock assignment was not possible due to the presence of these haplotypes across multiple stocks ( Supplementary Table 1 ). Similarly, eight of the orphan haplotypes identified in this study have also been documented in more distant foraging sites including in the Torres Strait, Palmyra Atoll and Ecuador ( Table 2 ). Interestingly, one juvenile female from central NSW was found with haplotype CmP4.6, a common haplotype documented for the ‘black’ green turtle morphotype breeding in the north-central/eastern Pacific Ocean around the Galapagos Islands, Michoacán and Costa Rica approximately 12,000 km from Australia (Dutton et al., 2014). While the color and morphology of the stranded turtle’s carapace was not recorded, juvenile turtles possessing this haplotype have been found in mixed foraging aggregations sympatrically with the ‘yellow’ morphotype of the south-central/western Pacific genetic lineage (e.g. New Zealand; Godoy et al., 2016; Álvarez-Varas et al., 2021). Central/eastern Pacific juvenile turtles have also infrequently been recorded foraging on the GBR in Australia (Limpus et al., 2005) and the genetic stock origin has been confirmed by genetic analysis (FitzSimmons, unpublished data). The presence of central/eastern Pacific turtles in the western Pacific Ocean foraging grounds is likely to occur due to transoceanic dispersal events by post-hatchlings along the warm waters of the westward-flowing North and South Equatorial Currents (Limpus et al., 2005; Nishizawa et al., 2014; Álvarez-Varas et al., 2021). Oceanic currents driving long-distance migrations of post-hatchlings have similarly been suggested to explain unexpected contributions from distant genetic stocks to green turtle foraging grounds in Malaysia (Jensen et al., 2016a).

The broad confidence levels, uncertainty around the contributions of more distant stocks and the presence of orphan haplotypes in a small percentage of individuals on foraging grounds in NSW and Queensland (Jensen et al., 2016b; Jones et al., 2018) highlights the need for more complete genetic sampling of green turtle genetic stocks in the Indo-Pacific region. Whole mtDNA genome sequencing (Frandsen et al., 2020) would improve the delineation of some green turtle stocks and improve the precision around MSA estimates to enable individual assignments to specific rookeries in the future. While management implications would likely remain unchanged for NSW, greater statistical power for MSA would further our understanding of the complex, long-distance dispersal, and migration patterns for this species.

Latitudinal changes in stock composition have been observed along nine major foraging grounds in the northern and southern GBR region in eastern Queensland, with gradual shifts in haplotype frequencies and stock composition observed from north to south (Jensen et al., 2016b; Jones et al., 2018). A higher estimated contribution from New Caledonia to the northern NSW aggregation compared to central NSW indicated that latitudinal shifts may continue down to the southern-most limit of the species’ distribution. Similarly, a latitudinal decline in haplotype and nucleotide diversity was documented in Queensland, with higher genetic diversity generally found in northern Queensland foraging grounds compared to the southern GBR region (Jensen et al., 2016b; Jones et al., 2018). Higher levels of genetic diversity in northern NSW compared to central NSW also supports anecdotal evidence of the progressive expansion of green turtle foraging grounds to southern NSW over recent decades. A reduction in genetic diversity at increasing distances from the origin of expansion has similarly been observed for many species undergoing range expansions (Peter and Slatkin, 2013). While the pattern of decreasing genetic diversity at decreasing latitudes was not maintained when including NSW sites as there were higher genetic diversity levels observed in northern NSW compared to southern Queensland, the geographic scale of sampling differed between these studies. Higher levels of genetic diversity in NSW could be attributed to the larger geographic sampling scale that potentially included multiple foraging grounds within the northern and central NSW aggregations. An increased understanding of green turtle foraging movements in NSW environments would assist in future assessments of genetic diversity patterns along the east coast of Australia.

Understanding the sex ratio of mixed foraging aggregations in relation to the operational and primary (hatchling) sex ratios at green turtle rookeries can also improve our knowledge of population dynamics, including the impact of climate change. A moderate female-biased sex ratio of samples (1M:2F) was observed in this study, indicating that greater numbers of female green turtles may be stranding in NSW compared to males. While the sex-ratio of stranded animals and free-ranging turtles foraging in NSW is currently unknown, similar or even more strongly female-biased foraging aggregations have been documented across the species' distribution (Limpus et al., 2009; Allen et al., 2015; Godoy et al., 2016; Jensen et al., 2016a; de Almeida et al., 2021). Sex-specific differences in the natal origins of green turtles were also observed in NSW. The estimated contribution of the southern GBR stock was similar between sexes, but all of the estimated contribution from New Caledonia were female, with no males assigned to this stock. Since nesting sand temperatures are known to determine green turtle sex ratios, with more female offspring produced at higher temperatures and more males produced at lower temperatures (Standora and Spotila, 1985), the female-biased sex ratio observed from the New Caledonia stock may have arisen due to a long-term shift in the percentage of females hatching in this region. Although hatchling sex ratios for this stock are unknown, higher sand temperatures have been documented for loggerhead turtle breeding sites in New Caledonia compared to southern GBR sites (Read et al., 2013). Similarly, for a GBR foraging aggregation, an extreme female bias was found for juveniles and subadults (> 99%) originating from the northern GBR genetic stock and a less-pronounced female bias across all size classes for the cooler breeding grounds in the southern GBR (Jensen et al., 2018). In light of trends for increasing feminization of green turtle populations (Jensen et al., 2018) and evidence of temporal changes to stock composition at other foraging grounds due to changes in population dynamics of genetic stocks (Jensen et al., 2016b; Stahelin et al., 2022), future sampling of green turtles in NSW of all sex and size classes over a temporal period will further our understanding of the impact of climate change on green turtles in the Indo-Pacific.

The data presented in the study are deposited in the GenBank repository (https://www.ncbi.nlm.nih.gov/genbank/), accession numbers: PP278330- PP278333.

The animal study was approved by Taronga Conservation Society Australia Animal Ethics Committee. The study was conducted in accordance with the local legislation and institutional requirements.

JD: Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Project administration, Writing – original draft. JH: Data curation, Writing – review & editing. KR: Data curation, Writing – review & editing. KVH: Data curation, Writing – review & editing. DM: Data curation, Writing – review & editing. OP: Data curation, Writing – review & editing. NF: Writing – review & editing, Formal analysis, Methodology. LH: Data curation, Writing – review & editing. KM: Writing – review & editing, Data curation. SI: Methodology, Writing – review & editing. PM: Conceptualization, Data curation, Project administration, Writing – review & editing.