This study was carried out at two hawksbill turtle nesting locations in PNG, separated by an approximate distance of 920 km ( Figure 1 ).

Conflict Group of Islands (10° 46’ 10.992’’ S; 151° 45’ 26.2512’’ E) is located approximately 150 km southeast of Alotau in the Louisiade Archipelago within MBP. The Conflict Group consists of 21 uninhabited islands, except for a small resort on Panasesa. The main islands include Irai, Gabugabutau, Tupit (Tobiki), Panarakuum, Kolavia, Muniara, Aroroa, and the Reef Islands. Both satellite tracking and genetic sampling were conducted at the Conflict Group during the nesting season (November–March) from 2017 to 2020.

Kavieng (2° 34’ 25” S; 150° 47’ 43” E) is located at the northern tip of the New Ireland Province. Since 2013, Ral island, Edmago island and the larger Lissenung Island, have been monitored for turtle nesting by the owners of Lissenung Resort. These islands are located to the west of the Kavieng township and remain unoccupied, except for a small resort on Lissenung. From mid-September to the end of March, a conservation project relocates marine turtle eggs to incubate in the resort hatchery. From 2015–2021, genetic samples from one hatchling per nest were collected and donated to this study.

At the Conflict Group, a total of 39 tissue samples were collected from nesting female hawksbill turtles across multiple islands from 2017 to 2020. Skin samples (<0.5 cm2) were collected using a scalpel blade from the front or rear trailing flipper and stored in 2 ml cryo-vials in >70% ethanol. At the Conflict Group, all nesting hawksbills were tagged with a titanium tag (Limpus, 1992), and measured for curved carapace length (CCL) before being released.

At Kavieng, 56 tissue samples were collected from hawksbill hatchlings on Ral, Edmago, and Lissenung Islands from 2015 to 2021. No nesting females were encountered at the Kavieng sites, but nests were translocated to a protected hatchery. One tissue sample was collected from a single hatchling from each nest and stored in 2 ml cryo-vials in >70% ethanol. All samples were stored (-20°C) and transferred to Griffith University in Australia for long-term storage and analysis.

All research was conducted under PNG’s National Research Institute approval. All genetic samples were transported to international universities for analysis under CITES permits 018140 and 022064 (export) and WT2019–000439 and PWS2021-AU-001689 (import).

Genomic DNA was obtained from all samples using the salting out extraction method described in Jensen et al. (2013). The control region (or d-loop) of the mitochondrial genome was amplified to generate about 800 base pair sequences using LCM-15382 (5’ GCT TAA CCC TAA AGC ATT GG 3’) and H950g (5’ GTC TCG GAT TTA GGG GTT TG 3’) primers (Abreu-Grobois et al., 2006), PCR reactions were carried out in a 25 μl reaction volume. The PCR cycling parameters were as follows: an initial 4 min of DNA denaturation at 95°C, followed by 36 cycles of 25 s at 95°C, 25 s at the annealing, temperature, and 30 s at 72°C, followed by a final 2 min of extension at 72°C. The annealing temperature consisted of 2 cycles at 56°C, 2 cycles at 54°C and 33 cycles at 52°C. Each PCR setup included negative controls to detect contamination, and the PCR products were visualized on 1.2% agarose gels stained with SYBR Safe DNA Gel Stain (Thermo Fisher Scientific). The PCR products were purified, and sequenced by Macrogen Inc. (Seoul, Korea).

Raw sequences were edited using the software Geneious 6.1.7. (https://www.geneious.com). (Applied Biosystems, Foster City, California, USA). Each sequence was manually inspected for uncalled and miscalled bases, and all variable positions were confirmed by comparing sequences from the forward and reverse strands. Haplotypes were assigned by comparing aligned sequences to the ShellBank database (global marine turtle reference database; www.shellbankproject.org), which contains a reference library of published hawksbill haplotype sequences, and by searching the GenBank database (http://www.ncbi.nih.gov) The standardised nomenclature was used to name the new haplotypes using the EiIP prefix for Indo-Pacific hawksbill haplotypes, followed by the next number in order. New haplotypes were submitted to GenBank (http://www.ncbi.nih.gov). The sequences were then aligned, edited, and cropped at a standard cropping site of approximately 770 bp using Geneious Prime.

We analysed all 39 Conflict Group hawksbills confirmed as independent nesting individuals through flipper tags. For Kavieng, a total of 52 samples were sequenced, but after excluding hatchling samples likely to be from the same mother, only 42 samples remained and were used for the final analysis. When the mother could be sampled (and tagged), one hatchling from each nest may represent the nesting female since they carry the same copy of mtDNA. However, since hawksbill turtles may lay 1–5 nests within a season, it is important to avoid pseudo-sampling (i.e. sampling the offspring from the same mother in consecutive nests). To ensure individual hatchlings from different mothers were selected, we applied a conservative 10 to 18-day window, the renesting interval for hawksbill turtles when hatchlings with the same haplotype could originate from the same mother. This was assumed for up to four consecutive nests. This let us rule out any hatchlings born during that time with the same haplotype, increasing the chance that samples represented individual nesting individuals ( Table S1 ). Although we do not know the remigration interval for hawksbill turtles in PNG but likely 2-9 years (Limpus, 2009), and given the harvest pressure, the likelihood of re-sampling the same mother is reduced but pseudo-sampling across years cannot be ruled out. Furthermore, we observed no substantial differences in haplotype frequencies between the strategies (not removing samples and removing samples). Given these results, we are confident that regardless of the strategy applied, it will not significantly affect the study’s outcomes.

In addition to samples from rookeries in Kavieng and the Conflict Group we included data from nesting hawksbill turtles in Thailand at Kram Island (Wahidah and Syed Abdullah, 2009), Malaysia at Malaka Island, Redang Island, and Sabah Turtle Islands (Wahidah and Syed Abdullah, 2009; Nishizawa et al., 2016, Vargas et al., 2016), in Australia at Western Australia, northeast Arnhemland in the Northern Territory and north Queensland (Vargas et al., 2016; Bell and Jensen, 2018; LaCasella et al., 2021), and in the Solomon Islands (Vargas et al., 2016; LaCasella et al., 2021) ( Table S2 ). We first tested for any significant genetic structure across years and studies to combine various published datasets from the region. These included five data sets from Sabah Turtle Islands in East Malaysia (Wahidah and Syed Abdullah, 2009; Nishizawa et al., 2016; Vargas et al., 2016), three datasets from Malaka (Wahidah and Syed Abdullah, 2009; Nishizawa et al., 2016) and three datasets for Redang (Nishizawa et al., 2016; Vargas et al., 2016) in Peninsular Malaysia, two datasets from Milman Island (Vargas et al., 2016; LaCasella et al., 2021) and two datasets from Arnavon, Solomon Islands (Vargas et al., 2016; LaCasella et al., 2021) ( Table S2 ). All subsequent analyses were performed on pooled datasets across locations and years were appropriate.

Population structure was tested using pairwise F_ST comparisons and exact tests of population differentiation with the software Arlequin v 3.5.1.2 (Excoffier and Lischer, 2010). To determine significance, a probability level of P > 0.05 was used. A first round of pairwise comparisons tested temporal trends within previously published rookeries collected across different years and studies to determine if these could be combined or if haplotype composition had changed significantly over time. A second set of pairwise comparisons tested previously identified genetic stocks against the two new sites at Conflict group and Kavieng to assess the overall stock structure. Significance values for F_ST were obtained from 10,000 permutations. Exact tests of population differentiation were conducted with 100,000 permutations and 10,000 dememorization steps (Raymond and Rousset, 1995). Finally, haplotype (Hd) and nucleotide (π) diversity were calculated for each Management Unit using Arlequin v 3.5.1.2 (Excoffier and Lischer, 2010). Haplotype diversity was assessed following Nei (1987), and nucleotide diversity was computed using Tamura and Nei. (1993).

Sixteen Argos satellite tags (SPOT6, Wildlife Computers, Seattle, Washington, USA) were attached after oviposition to nesting hawksbill turtles found on various islands of the Conflict Group during the 2017-2018 and 2018-2019 nesting seasons ( Figure 2 ; Table S4 ). As per Godley et al. (2002), satellite tags were attached anteriorly to the hawksbill turtle’s carapace and were released when the epoxy (Sika AnchroFix -3+) had completely cured to avoid turtles rubbing off the tag. The tags were additionally painted with International Micron 66 anti-fouling paint to prevent algal growth and other fouling organisms. All turtles curved carapace length (CCL) were measured and individually numbered titanium flipper tags were applied prior to release.

The Argos satellite system (http://www.argos-system.org/) was used to relay location data, and the Wildlife Computer Portal was used to store the received data. Filtering of all tracks and analysis was conducted in R Statistical Software (V4.0.0; R Core Team, 2021). Argos locations were first filtered using the R package SDLfilter (Shimada, 2016) to remove duplicates and positions with location errors of > 1km, i.e., only Argos classes 0, 1, 2 and 3 were retained. Locations between which swimming speeds were deemed implausible (> 5 km/hr as per Shimada et al., 2012) were also removed. The remaining track locations were intersected with spatial polygons of land masses, and any fixes that overlapped were discarded. We visually assessed the trajectories of each turtle to classify track segments as either inter-nesting, migratory or foraging behaviours. The start of the migration period was defined as the point when individuals showed directed and continuous movement away from nesting areas. We considered foraging behaviour had begun following a significant decrease in overall swim speed and the cessation of directed movement.

To describe migration, we calculated swimming speed, distance, and total duration for each turtle. For turtles that arrived in foraging areas (n = 10), we quantified foraging activity by estimating the 50 and 95% utilisation distributions (UDs) using minimum convex polygons implemented using the adehabitatHR package (Calenge, 2006; Calenge, 2015). We assumed the 50% UD contour to represent each turtle’s core centre of activity and that the 95% UDs represented overall foraging areas. Turtles were classified into foraging groups based on the spatial clustering of their foraging home ranges.

Sequence data from ~770 bp fragments of mtDNA was used from 39 nesting females at the Conflict Group and from 42 hatchlings from Kavieng ( Table S2 ). We identified a total of 6 haplotypes from the Conflict Group. The most common was haplotype EiIP-33 (84.6%), followed by EiIP-34, EiIP-59, EiIP-93, EiIP-09 and EiIP-114 all found a low frequency (<8%), with no new haplotypes identified. From Kavieng, we identified a total of 5 haplotypes, with the most common being haplotype EiIP64 (39%), followed by EiIP-33 (29%), EiIP-59 (24%), EiIP-39 (2%) and one newly discovered haplotype, EiIP-150 (5%) (Genbank Accession: to be added) ( Table S2 ).

Given that multiple studies have been conducted from these sites over many years, the first step was to test for temporal variation in rookeries with data from several studies collected across multiple years. Both pairwise F_ST and exact tests showed no significant differentiation across studies or years within the same location except for samples from Sabah Turtle Islands from 2014 (Nishizawa et al., 2016) which were significantly different from other years within Sabah Turtle Island (F_ST = 0.07817 - 0.15951, P < 0.05), and exact test (P < 0.05) ( Table 1 ) . As a result, all samples within rookies were combined with the exception of Sabah Turtle Islands, which were separated into East Malaysia “Old” (1998-2008) and East Malaysia “New” (2014) in subsequent analysis. A similar result was found in Nishizawa et al. (2016).

The haplotype frequencies of samples from the Conflict Islands was significantly different from those of the Kavieng rookeries for both the exact test (P< 0.0001) and F_ST (F_ST = 0.270, P ≤ 0.01; Table 1 ). Moreover, the Conflict Group and Kavieng samples each differed significantly from the Asia-Pacific hawksbill stocks in Malaysia, Australia and the Solomon Islands ( Figure 1 ; Table 1 ). Both the exact test and F_ST indicate high significance differentiation, except for Milman Island and Arnhem Land, which showed significant differences for the exact test (P = 0.0084) but not for the F_ST test (F_ST = 0.0059, P = 0.2129) ( Table 1 ). Due to breeding seasonality (predominantly summer vs winter nesting), these genetic stocks have already been deemed separate MUs (Vargas et al., 2016). This represents at least nine hawksbill MUs now characterised in the Asia-Pacific region ( Figure 1 ).

Estimated nucleotide diversity ranged from very low in Peninsular Malaysia (0.00027) to high in Milman Island (0.01994), with intermediate levels at Conflict Group (0.00263) and Kavieng (0.00138) ( Table S3 ). Haplotype diversity was highest at Kavieng (h = 0.7120 ± 0.0346) relatively low at Conflict Group (0.4062 (±- 0.0969) and lowest at Peninsular Malaysia (0.0901 (± 0.0421) ( Table S3 ).

All turtles tagged were “primary” or “within season recaptured” turtles (meaning caught for the first time or within the nesting season), with a mean CCL of 80.9 (s.d.=,2.60, range = 77.5 to 86.3 cm, n = 16) and mean curved carapace width (CCW) of 71.9 (s.d. = 3.24, range = 66.1 to 76 cm, n = 8) ( Table S4 ). Fifteen hawksbill turtles were tracked for a total of 4476 days ranging from 53 to 747 days (mean ± SE = 298 ± 66). The sixteenth hawksbill stopped tracking within 27 days and did not leave the vicinity of the initial tagging location before tag failure. Upon tag assessment, no apparent cause to the failure could be attributed, including poaching. This turtle was excluded from the rest of the analyses.

The aim of this study was to assess if hawksbill turtles at Conflict Islands and Kavieng were individual genetic stocks or if they were connected to each other or other western Pacific stocks in the Solomon Islands or Australia. This study’s findings emphasise that the two newly described PNG genetic stocks need to be managed as demographically separate MU’s, but their migratory and foraging life history also links them to Australia’s management and conservation action. The importance of Australia’s habitat in supporting multiple (mixed stock) hawksbill turtle populations has been highlighted by other recent studies (neQld, Madden Hof et al., 2023; Solomon Island, Hamilton et al., 2021; Torres Strait, Barr et al., 2021; Vanuatu, Jim et al., 2022). This highlights the need for greater national and regional cooperation, particularly given the decline in the neQld hawksbill stock (Bell et al., 2020) and the likely decline of other western Pacific hawksbill populations (Pilcher, 2021). Also, sea turtle harvesting for meat and eggs in Australia and PNG is a traditional fishery undertaken by traditional owners in coastal and islander communities. Within PNG, shells are also used to make utilitarian items such as needles and limes spatulas, decorations such as earrings and bracelets, and for ceremonial purposes or trade (Kinch and Burgess, 2009). Despite efforts to track harvest levels in PNG, accuracy is complicated due to the remote nature of harvesting locations and limited government presence (Kinch, 2020a). Small-scale artisanal fisheries (Eley, 1988; Kwan, 1991) and market surveys (Hirth and Rohovit, 1992; Kinch and Burgess, 2009) have reported evidence of hawksbill turtles (with particular concern raised for MBP) being heavily targeted by local inhabitants of the Louisiade Archipelago (Kinch, 2020b). These findings align with an assessment of exploitation in PNG, which identified MBP as one of three targeted harvest locations in need of intervention (Opu, 2018).

Ineffective, unregulated or complete lack of management and/or protection at hawksbill rookeries or foraging grounds is likely to negatively impact the PNG population and other western Pacific stocks (Bell and Jensen, 2018; Madden Hof et al., 2023). In the context of PNG, no marine turtle, with the exception of the leatherback turtle, is protected by PNG’s legislation [e.g. Fauna (Protection and Control), 1976 (Kinch, 2006)]. There are no laws, regulations, or quota limits to harvesting hawksbill turtles or to sell, offer or consign sale or be in possession. The International trade (Fauna and Flora) Regulation, 2014 does prescribe that it is illegal to trade to and from PNG in CITES-listed fauna (which includes hawksbill turtles). The taking of hawksbill turtles by Papua New Guineans within Australia’s EEZ (i.e. the Torres Strait Protected Zone) is also allowed under the Torres Strait Treaty, 1985 as long as they are traditional inhabitants of ‘Treaty’ Villages conducting “traditional fishing”. However, each Party should to its best endeavours identify and protect fauna that are or may become threatened with extinction (Article 14(1)(a) Torres Strait Treaty, 1985; Kinch, 2020a). However, inhabitants assignment, regulation and enforcement are limited (Busilacchi et al., 2018).

There are, however, many other laws that could support sea turtle conservation in PNG (refer, Kinch, 2006; Kinch, 2020a) including the Organic Law on Provincial Governments and Local-level Governments, 1997, which allows for the development of local-level conservation laws (under Sections 42 and 44) and could potentially be used to ban or limit hawksbill turtle and egg take, and establish nesting beach closures. Additionally, the Fisheries Management Act, 2016 provides the framework to regulate hawksbill turtles as a sustainable fishery. While all potential options require scientifically based assessments of PNG’s hawksbill turtle population status and trajectory that are informed by annual harvest rates, applying the precautionary principle could at least ensure a level of protection as an intermediary step. Initial protection could be afforded through the uplisting of hawksbill turtles as a ‘Protected Species’ under the Flora and Fauna Protection and Control Act, 2014, alongside a review and strengthening of the Torres Strait Treaty, 1985. In acknowledging the role of marine turtles in local communities, it is suggested the strengthening of these policy and legislative options be considered and planned in consultation with and in recognition of communities and their rights to sea turtle resources. This should be done alongside improvements to the local community economy and provision of alternatives to the reliance on hawksbill turtles for subsistence, trade and culture. Ongoing research by CICI and the World Wide Fund for Nature is already underway to fill these critical assessment and research gaps (CICI and Coral Islands Ltd., 2021; Madden Hof et al., 2022).

In eastern Australia, where most of the Conflict Group hawksbill turtles forage, hawksbill turtles will also require further national and transboundary management and protection. Hawksbill turtles are protected and managed across a raft of Commonwealth and State legislation, policy and recovery strategies in Australia. Still, harvesting can occur by Australian Aboriginal and Torres Strait Islanders under the Native Title Act, 1994. Under this Act and other frameworks, (e.g. Torres Strait Turtle Fishery), there are no legislative or regulated quota limits to harvesting. Instead, communities are encouraged to self-manage and permit their catches through their local traditional LORE, and as such, there are no quantifiable harvest rates. While there are few geo-political strategies in place to protect transboundary marine turtle populations in this region (Bell and Jensen, 2018), given PNG shares its hawksbill populations with Australia (and beyond), joint and concerted co-management efforts should now be explored to ensure future conservation and protection across PNG’s hawksbills full life history. This study will help facilitate these discussions and underpin the development of conservation strategies to protect hawksbill turtles throughout their life history and the western Pacific range.