Evidence of long-distance movement of green sawfish (Pristis zijsron) in Western Australia

Ethics approval

Handling and sampling of sawfish was conducted under Murdoch University Animal Ethics Approvals RW2397/11 and RW3191/19, Western Australian Government Department of Primary Industries and Regional Development Fisheries Exemption numbers 3378 and 3553, Department of Fisheries Regulation 178 (SPA 11-11), and Department of Environment and Conservation Permit SF007889.

Sample collection

Pristis zijsron tissue samples were collected from individuals at eight sites in Western Australia (Fig. 1, Table 1). Live P. zijsron were captured with monofilament gillnets (see Morgan et al. 2015) or cast nets. Upon capture, P. zijsron were placed onto their backs in the shallows with their gills submerged, inducing a state of tonic immobility, and then measured for stretched total length (STL) and sexed. Tissue samples in the form of fin clips or hole punches were taken from either the pelvic or dorsal fins and immediately preserved in 100% ethanol. Captured sawfish were then released at the site of capture. Samples were collected from 137 live juvenile P. zijsron (Table 1) between 2003 and 2022 (651–3195 mm STL), and 1 carcass from Cable Beach in 2022, measured at 875 mm STL, which is consistent with the reported size at birth for this species (i.e. 750–900 mm) (see Morgan et al. 2015; Lear et al. 2023). Most of these were from the Ashburton River estuary and adjacent tidal creeks, which had been included in previous kinship investigations (see Ingelbrecht et al. 2024), with the present study focusing instead on individual relationships between different populations of P. zijsron along the north-western Australian coastline.

Fig. 1.

Sampling locations for green sawfish (Pristis zijsron) in Western Australia, illustrating the distance between capture locations of half siblings (arrows) captured at Cape Keraudren in 2008 and the Ashburton River in 2014, or the Ashburton River in 2014 and Cable Beach in 2022.

Genotyping and data filtering

Tissue samples were sectioned into ~10-mg pieces and submerged in ~100 μL of 70% ethanol in an Eppendorf Twintec polymerase chain reaction (PCR) extraction plate, in preparation for DNA extraction and genotyping at Diversity Arrays Technology (DArT, DArT Pty Ltd, Canberra, ACT, Australia). Extraction of DNA was performed with a NucleoMag magnetic bead-based extraction kit (Macherey–Nagel, Berlin, Germany), by using the methods described by Ingelbrecht et al. (2024). Genotyping was performed using Pristis DArTseq (DArT Pty Ltd), with a high-density assay and sequencing effort of 2.5 million reads. Single-read sequencing was performed on a Novaseq X high-throughput sequencer (Illumina, San Diego, CA, USA), following standard protocols described by Kilian et al. (2012). To assess the quality of markers and consistency of allele calling (i.e. reproducibility), 30% of sampled individuals were genotyped in technical replication.

Data for 130 individuals, corresponding with 35,210 SNPs, were generated by DArT Pty Ltd. Samples from eight P. zijsron failed sequencing and were therefore not included. Data filtering was implemented in R (ver. 4.2.3, R Foundation for Statistical Computing, Vienna, Austria, see https://www.r-project.org/) using the package dartR (ver. 2.7.2, see https://cran.r-project.org/package=dartR; Gruber et al. 2018). For statistical inference of kinship, the use of at least 500 SNPs with a moderately high minor allele frequency (MAF) (i.e. ≥0.2) is recommended for maximising pedigree assignment rate, while minimising the rate of errors (Huisman 2017; Premachandra et al. 2019). Therefore, SNPs were retained according to call rate (≥95%), reproducibility (i.e. the proportion of identical genotypes calculated from technical replicates; ≥ 95%), and read depth (lower/upper thresholds = 5/40). Because allele frequency affects the accuracy and rate of kinship inference (see Huisman 2017; Premachandra et al. 2019), four different MAF lower thresholds were used (≥0.1, 0.2, 0.3 or 0.4). Furthermore, because of the impacts of MAF filtering on heterozygosity, different individual sample heterozygosity lower/upper thresholds were applied depending on MAF, as follows: 0/0.42 (MAF ≥ 0.1), 0/0.47 (MAF ≥ 0.2), 0/0.52 (MAF ≥ 0.3), 0/0.55 (MAF ≥ 0.4), generating four filtered SNP datasets. Monomorphic loci or loci showing significant departure from Hardy–Weinberg equilibrium (HWE) (at a significance threshold of α = 0.05, using a Bonferroni correction for multiple comparisons) were removed.

Kinship analysis

Kinship was inferred using the following three independent analyses: Sequoia (ver. 2.5.3, see https://cran.r-project.org/package=sequoia; Huisman 2017), COLONY (ver. 2.0.6.8, see https://www.zsl.org/about-zsl/resources/software/colony; Jones and Wang 2010b), and genomic relatedness probabilities calculated in dartR. Kinship assignments through Sequoia and COLONY were performed by full-pedigree reconstructions, by using filtered datasets of SNPs and two genotyping error rates: 1 or 5%. Sequoia and COLONY pedigrees were reconstructed using the same parameters as described previously for P. zijsron (see Ingelbrecht et al. 2024), but using a threshold log-likelihood ratio (LLR) ‘Tfilter’ of −4 in Sequoia.

Genomic relatedness probabilities, calculated from all available SNPs, were used to corroborate kinship assignments and distinguish between first- and second-degree relatives. Pairwise genomic relatedness probabilities between 0.41–0.59 and 0.18–0.31 were taken to indicate first- and second-degree relationships respectively, on the basis of 95% confidence intervals for relatedness reported by Speed and Balding (2015). The most likely genealogical relationships were assembled manually, using combined information from Sequoia and COLONY pedigrees, and estimates of pairwise genomic relatedness. To account for false positive inferences, an assignment was accepted (see reported assignments) only if it was supported by both Sequoia and COLONY in at least one pedigree, or if these analyses were not in congruence, by one of these analyses and pairwise genomic relatedness probabilities. False-negative assignments were accounted for by reconstructing pedigrees from multiple SNP datasets with varying MAF values (≥0.1, 0.2, 0.3 or 0.4). Furthermore, an assignment was deemed to be supported by COLONY only if it was accompanied by a confidence probability of ≥90%. Acceptance of full-sibling assignments through COLONY was based on exclusive probabilities (i.e. the probability that all individuals within a family are full siblings and that no other individuals are full siblings with that family), whereas acceptance of half siblings was based on confidence probabilities of assigned dyads. For young-of-the-year (YOY) P. zijsron, capture location was assumed to be proximal to the location of birth. Age classes were estimated using growth rate information for P. zijsron in Western Australia (Lear et al. 2023). Full siblings estimated to be of the same age were assumed to be littermates.

Sibling pairs

Overall, 62 P. zijsron were assigned to 25 litters of full siblings, with litter size ranging from two to five pups. Furthermore, 76 P. zijsron were assigned to 96 half-sibling pairs (i.e. pairwise relationships). Two full-sibling litters and six half-sibling pairs involved at least one P. zijsron sampled outside of the Ashburton River estuary and adjacent tidal creeks (Table 2). Of these, one individual was a half sibling of two other P. zijsron, which in turn were full siblings. These three individuals were captured together as neonates (804–843 mm STL) in the Bay of Rest, Exmouth Gulf, and are most likely to be littermates. Furthermore, one full-sibling pair was captured at Cape Keraudren in 2008, as YOY (i.e. 0+) (1000 and 1070 mm STL), providing the first evidence of P. zijsron parturition in the Exmouth Gulf and Cape Keraudren (Table 2).

Distribution of kin over long distances

Four half-sibling pairs involved sawfish captured over 500 km apart (Table 2). Of these, one pair consisted of an individual captured in the Ashburton River estuary in 2014, as a neonate (846 mm STL), and its half sibling that was sampled as a recently deceased neonate (875 mm STL), ~870 km to the north-east at Cable Beach, Broome, in 2022 (i.e. ~8 years separating their births) (Fig. 2). The remaining three pairs (involving five juvenile P. zijsron) each consisted of one individual captured at Cape Keraudren in 2008, and their half siblings that were captured ~530 km to the south-west in the Ashburton River estuary in 2014, including one full-sibling pair from Cape Keraudren that shared the same half sibling from the Ashburton River (Table 2). These five individuals were estimated to be 2 years old or younger at the time they were captured (≤1620 mm STL), with ~6–7 years separating half-sibling births (Lear et al. 2023).

Fig. 2.

Images of individual green sawfish (Pristis zijsron) sampled for kinship investigations. (a) 2022 #38, young-of-the-year (YOY) P. zijsron (875-mm stretched total length, STL) sampled at Cable Beach, Broome, recently deceased, which was a half sibling of a sawfish captured in 2014 in the Ashburton River estuary. (b) 2014 #02, YOY P. zijsron (874 mm STL) sampled in the Ashburton River estuary, which was a half sibling of two full siblings captured at Cape Keraudren in 2008. (c) 2014 #05, ~2-year-old juvenile P. zijsron (1620 mm STL) sampled in the Ashburton River estuary, which was the half sibling of a sawfish captured at Cape Keraudren in 2008.

Dispersal

This is the first study to provide evidence of long-distance movement for the green sawfish, Pristis zijsron, which we define as any movement from one site to another where the distance between those sites exceeds 100 km. Telemetry data indicate that the movement of juvenile P. zijsron is generally confined to small core-activity spaces, concentrated in shallow waters, with home range increasing with growth (Morgan et al. 2017). Considering that all P. zijsron included in the kinship analyses were juveniles, the differences in capture locations between half siblings sampled >500 km apart are likely to be due to long-distance, parental movement between sites.

Previous genetic studies have reported that both female and male P. zijsron are regionally philopatric, with restricted gene flow at large spatial scales in northern Australia indicating limited dispersal in both sexes (Phillips et al. 2017; Ingelbrecht et al. 2024), although the degree to which philopatry occurs can vary among individuals or populations of the same species (Trenham et al. 2001). Furthermore, male-biased dispersal has been reported for the largetooth sawfish, Pristis pristis (Linnaeus, 1758), and the narrow sawfish, Anoxypristis cuspidata (Latham, 1794) (Phillips et al. 2017; Green et al. 2018), which may suggest that half siblings sampled more than 500 km apart are more likely to be related paternally. However, movement over similar distances is not exclusive to males for all Rhinopristiformes; Barnett et al. (2024) tracked a mature female giant shovelnose ray, Glaucostegus typus (Anonymous [Bennett], 1830) (~2490 mm TL), over 700 km south from its initial tagging location in Abbot Bay, Queensland. Thus, additional telemetry or genetic work is required to determine whether long-distance dispersal is sex-biased in P. zijsron, or related to reproduction or resource availability.

Pristis zijsron is legally protected within Australian Commonwealth waters under the Environment Protection and Biodiversity Conservation Act 1999, having been listed as Vulnerable in 2008 after satisfying Criterion 1 (i.e. decline in numbers) of the Environment Protection and Biodiversity Conservation Act eligibility requirements (Threatened Species Scientific Committee 2008; Department of the Environment 2015). In Western Australia, P. zijsron is also protected under the Wildlife Conservation Act 1950 and the Fish Resources Management Act 1994, which prohibits the take, possession, sale, purchase or consignment of P. zijsron and its products within the state (Department of the Environment 2015), although even with this protection, P. zijsron remains threatened by habitat degradation, gillnet and trawl fisheries in Australian waters (Kyne et al. 2021; Harry et al. 2024). A national recovery plan is, however, in place for sawfishes listed under the Environment Protection and Biodiversity Conservation Act, which includes a series of objectives and actions to improve outcomes for sawfishes (see Department of the Environment 2015). This work directly contributes to one of these objectives; specifically, objective 9b, which lists ‘the identification of important habitats for all life stages of sawfish species, including connectivity between regions’ as Priority 1 (i.e. prompt action necessary to mitigate key threats and provide valuable information to help identify long-term population trends) (Department of the Environment 2015). This study revealed the presence of half siblings between the Ashburton River and Cable Beach, thereby providing genetic evidence of movement and population connectivity between the Pilbara and Kimberley regions. Dispersal of individuals over large spatial scales reduces the risk of localised extinction for P. zijsron, as it suggests that assemblages could be replenished from elsewhere in the event of population decline. However, long-distance coastal movement of P. zijsron could be impeded, should major structures be established along the shoreline, with previous research demonstrating that movement of smaller juveniles between the Ashburton River and adjacent tidal creeks was constrained following construction of an offloading facility (see Lear et al. 2024). The potential impediment to long-distance movement should be taken into consideration in any coastal development plans. By providing important insight into the movement behaviour of P. zijsron, this research will contribute to more effective conservation strategies for this threatened species.