Weak population structure of the Spot‐tail shark Carcharhinus sorrah and the Blacktip shark C. limbatus along the coasts of the Arabian Peninsula, Pakistan, and South Africa

Abstract The increase in demand for shark meat and fins has placed shark populations worldwide under high fishing pressure. In the Arabian region, the spot‐tail shark Carcharhinus sorrah and the Blacktip shark Carcharhinus limbatus are among the most exploited species. In this study, we investigated the population genetic structure of C. sorrah (n = 327) along the coasts of the Arabian Peninsula and of C. limbatus (n = 525) along the Arabian coasts, Pakistan, and KwaZulu‐Natal, South Africa, using microsatellite markers (15 and 11 loci, respectively). Our findings support weak population structure in both species. Carcharhinus sorrah exhibited a fine structure, subdividing the area into three groups. The first group comprises all samples from Bahrain, the second from the UAE and Yemen, and the third from Oman. Similarly, C. limbatus exhibited population subdivision into three groups. The first group, comprising samples from Bahrain and Kuwait, was highly differentiated from the second and third groups, comprising samples from Oman, Pakistan, the UAE, and Yemen; and South Africa and the Saudi Arabian Red Sea, respectively. Population divisions were supported by pairwise FST values and discriminant analysis of principal components (DAPC), but not by STRUCTURE. We suggest that the mostly low but significant pairwise FST values in our study are suggestive of fine population structure, which is possibly attributable to behavioral traits such as residency in C. sorrah and site fidelity and philopatry in C. limbatus. However, for all samples obtained from the northern parts of the Gulf (Bahrain and/or Kuwait) in both species, the higher but significant pairwise FST values could possibly be a result of founder effects during the Tethys Sea closure. Based on DAPC and FST results, we suggest each population to be treated as independent management unit, as conservation concerns emerge.


| INTRODUC TI ON
With an increasing number of conservation challenges and species under threat, population genetics offer a noninvasive tool to uncover otherwise unattainable information (Allendorf & Waples, 1996;Van Wijk et al., 2013). The identification of genetic structure is fundamental in determining the extent of reproductive isolation between populations (Hartl, 1988) and can have direct implications in designing effective protection plans.
In sharks, studies of genetic structure have shown subdivision on different geographic scales, ranging from small-scale genetic structure across less than hundreds of kilometers (Gaida, 1997), to largescale genetic structure between regions separated by ocean basins (Benavides et al., 2011;Daly-Engel et al., 2012;Duncan, Martin, Bowen, & De Couet, 2006;Portnoy, McDowell, Heist, Musick, & Graves, 2010;Schultz et al., 2008), to worldwide panmixia (Castro et al., 2007;Hoelzel, Shivji, Magnussen, & Francis, 2006). The genetic structure observed in different shark species is believed to depend on hard and soft barriers to gene flow. Hard barriers result from ancient events creating a physical landmass barrier to oceanic gene flow (e.g., the terminal Tethyan Event and the Isthmus of Panama, which separated the Indian and Atlantic Oceans and the Pacific and Atlantic Oceans, respectively). Soft barriers to gene flow are those related to a species' biology and behavior or invisible physical factors such as water currents or temperature (Cowman & Bellwood, 2013).
In sharks, biological and behavioral factors reported to influence genetic structure are vagility and reproductive behavior. Vagility is associated with body size, and a positive correlation has been found between body size and dispersal range (Mejía-Falla & Navia, 2011). This is supported by the finding that large species [>3 m total length (TL)] often have circumglobal distributions, for example, the whale shark Rhincodon typus (Castro et al., 2007)
The Arabian region has long been recognized as a global hotspot of marine biodiversity (Renema et al., 2008) and might be of particular importance to the diversity of elasmobranchs. For example, one of the world's least recorded carcharhinids, the smoothtooth blacktip shark Carcharhinus leiodon, is found in the Arabian/Persian Gulf (hereafter referred to as The Gulf) (Moore, White, Ward, Naylor, & Peirce, 2011). Furthermore, many of the shark taxa in the Arabian region are genetically distinct from their closest relatives in neighboring areas (Naylor et al., 2012) and the wider Indo-Pacific region (e.g., Corrigan et al., 2017;Delser et al., 2016;Haseli, Malek, & Palm, 2010;Naylor et al., 2012;Vignaud, Mourier, et al., 2014;White, Last, Naylor, Jensen, & Caira, 2010). This distinctiveness might have been enhanced by the geological event that resulted in the closure of the Tethys Sea, a major seaway connecting the Atlantic and the Indian Ocean via the Mediterranean Sea and The Gulf (Lambeck, 1996). During this event, approximately 23-15 million years ago, small isolated water pools formed along The Gulf's seafloor, which are thought to have had an important effect on the origin, dispersal, and speciation of several elasmobranch groups (Last, Matsumoto, & Moore, 2012;Musick, Harbin, & Compagno, 2004).
Carcharhinus sorrah and C. limbatus are requiem sharks that reach a maximum total length of 160 and 250 cm, respectively.
Throughout the Indo-west Pacific, they generally occur along continental and insular shelves, over coral reefs and muddy bottoms (Ebert, Fowler, & Compagno, 2013). Based on the International Union for Conservation of Nature (IUCN) Red List criteria, both species are listed as Near Threatened globally (Burgess & Branstetter, 2009;Pillans, Stevens, & White, 2009) and as Vulnerable regionally (Jabado et al., 2017).
Carcharhinus sorrah has been shown to exhibit a significant genetic structure over stretches of deep water (Giles et al., 2014).
Based on mitochondrial ND2 sequences, substantial genetic divergence was found between individuals from the Timor Sea/Gulf of Carpentaria and those from Borneo, the South China Sea, Thailand, and India (Naylor et al., 2012). Genetic studies of C. sorrah across northern Australia, in contrast, have suggested a panmictic population structure (Lavery & Shaklee, 1989;Ovenden, Kashiwagi, Broderick, Giles, & Salini, 2009). Although the species can move long distances (>1,000 km), almost 50% of tagged individuals in a tracking study were recaptured within 50 km of their tagging site (Stevens, West, & McLoughlin, 2000). This suggests that movement of most individuals is limited, probably resulting in little mixing between sites.
Carcharhinus limbatus is known to travel distances of over 2,000 km, with movements being influenced by seasonal changes in surface water temperatures (Kohler & Turner, 2001). The species uses shallow coastal waters as nurseries where juveniles spend the first months of their lives (Heupel & Simpfendorfer, 2002;Simpfendorfer & Milward, 1993).
Evidence of genetic structure was found between nurseries in North America, the Gulf of Mexico, and the Caribbean . Females were hence suggested to disperse nonrandomly and to exhibit philopatric behavior (Keeney, Heupel, Hueter, & Heist, 2003). Pronounced structuring was detected between African (KwaZulu-Natal and Sierra Leone) and Indo-Pacific populations and those of the eastern Atlantic based on mitochondrial DNA (mtDNA) (Keeney & Heist, 2006). However, this analysis did not include any South American populations, which were tested later and revealed that C. limbatus from northern Brazil is genetically distinct from the previously studied populations (Sodré et al., 2012). The aim of this study was to unravel patterns of connectivity among stocks of these two commercially exploited species along the coasts of the Arabian Peninsula, Pakistan, and KwaZulu-Natal, South Africa (hereafter referred to as South Africa), to facilitate regional conservation and management.

| Sample collection and laboratory procedures
Fin clips or gill slit samples of C. sorrah were obtained from local landing sites in Bahrain, Oman, the UAE, and Yemen and of C. limbatus from Bahrain, Kuwait, Oman, Pakistan, Saudi Arabia (Red Sea), South Africa, the UAE, and Yemen ( Figure 1, Table 1). Samples from F I G U R E 1 Sample locations for Carcharhinus sorrah and C. limbatus. Numbers correspond to landing site locations in Table 1 TA B L E 1 Landing sites sampled between May 2011 and July 2013 and respective sample sizes by country. Number in brackets corresponds to sampling locations in Figure 1 Country South Africa originated from sharks caught in mesh nets as part of a bather protection program (Dudley & Cliff, 1993). All samples were preserved in 96% ethanol.
Sharks landed along the coasts of the Arabian Peninsula were assumed to originate from fleets operating within a restricted range.
To ensure that the origin of the collected specimens was accurately represented by their landing sites, fishermen were asked to report their approximate fishing grounds and trip lengths. Moreover, observations on boat length, design, and engine power were made whenever possible to verify the reported fishing range. Not included in the study were samples originating from boats with offshore operating capacities, that is, medium-sized boats (>15-18 feet), characterized by a deep-V hull design, portable fuel gallons, and an engine >400 horse power. Tissue sampling was randomized by collecting no more than ten samples of each species on the same day. The only exception was Pakistan where landings of C. limbatus only occurred on the last day of fieldwork (n = 57). A breakdown of sex and size composition for all samples is available in Supporting Information Table S1. In addition, samples of 18 pregnant C. sorrah females with a total of 78 pups were collected from Deira fish market, Dubai, the UAE. These samples were not included in the population structure analysis but were instead used to detect null alleles by checking for genotype mismatches (i.e., genotypes that do not share a common allele) between pups and their known mothers (Marshall, Slate, Kruuk, & Pemberton, 1998).
Total genomic DNA from Red Sea samples was extracted following the protocol described in . DNA from all other samples was extracted using an adjusted glass milk protocol (Boom et al., 1990). Samples were incubated overnight in lysis solution (10 mM Tris-HCL (pH 8.0) and 1 mM EDTA, 1% SDS, and 50 μg/ ml proteinase K. Samples were then centrifuged, the supernatant was transferred to a new tube with sodium iodide (NaI), and 10 μl of glass milk solution were added. The DNA was washed with 500 μl of a solution that comprised of 100 mM NaCl, 1 mM EDTA and 10 mM Tris and 50% ethanol). Pellets were dried and then washed with 500 μl of 1× TE solution (500 μl of 10 mM Tris, 100 μl of EDTA, and 49.4 ml of distilled water). The extracted DNA was eluted into a new tube in 1× TE. Finally, the quality and quantity of the extracted DNA was checked from a random subset of the extracted samples using a NanoDrop spectrophotometer, ND-1000 Serial 7749, device (Thermo Scientific, UK).
For C. limbatus, 11 loci were used of which ten were species specific (Supporting Information Table S2c,d) .
Amplification was performed using the Qiagen multiplex PCR kit (Qiagen, Redwood, California). Multiplex PCRs were carried out in a total volume of 10 μl, containing approximately 20 ng of genomic DNA, 5 μl multiplex master mix solution, 1 μl primer mix, and 2 μl of RNase-free water. For each species, primers were organized into two sets of primer mix (Supporting Information

| Genetic diversity
Alleles were scored using the program GENEMAPPER (v3.7; Applied Biosystems Furthermore, we checked for Mendelian-inconsistent errors by determining mismatch error rates between mother and pup samples. Error rates were calculated for each locus by dividing the number of mismatched genotypes by the total number of genotypes (Marshall et al., 1998). The latter analysis was only performed for C. sorrah due to the unavailability of matched mother and pup samples for

| Population structure
The degree of genetic differentiation among sampling sites and locations was estimated using corrected pairwise F ST measured in GenoDive (v2.0; Meirmans & Van Tienderen, 2004). Pairwise F ST was tested for significance at level 0.05 with 10,000 permutations.
Multiple comparison p values were corrected with false discovery rate (FDR) adjustment in R (v.2.7.2; R Team 2015). Neighbor-joining trees using pairwise F ST between different locations were constructed using the adegenet package in R (Jombart, 2008 iteration of 100,000 steps. The length of the burn-in period was verified by ensuring that the Ln P(D) and the likelihood of the runs had stabilized. A correlated allele frequency model was used with sampling site as location prior and admixture were assumed, as recommended when population structure is likely to be subtle (Falush et al., 2007;Hubisz, Falush, Stephens, & Pritchard, 2009 Yemen. Allelic richness ranged from 3.5 ± 0.4 (Bahrain) to 3.9 ± 0.4 (the UAE and Yemen). The value of F IS , an inbreeding coefficient measure that calculates the proportion of the variance in the subpopulation contained in an individual (Raymond & Rousset, 1995), was small at all locations ranging from −0.01 ± 0.01 (Oman) to 0.01 ± 0.02 (the UAE) ( Table 2).
Mismatches between reference genotypes and regenotyped replicates were also low, with only one locus (CS55) displaying a high rate (≥5%) of genotyping error (Supporting Information Table S4), due to incorrect allele scoring. Two loci (CS40, CS55) showed higher rates of genotypic mismatch between mothers (n = 18) and pups (n = 78) than the rest of loci (Supporting Information Table S5). These two loci were consistent in their unreliability across the genotyping error tests and thus were considered unreliable and were excluded from further analysis.
Pairwise F ST values were low but mostly significant (Table 3).
Samples from Bahrain showed higher and significant differentiation from all other locations (F ST = 0.03, p < 0.001 for all comparisons) ( Table 3). The probability support produced by STRUCTURE for a range of Ks (1-10) was highest for K = 1, indicating a single population (Supporting Information Figure S1a).
The DAPC scatterplot supported weak fine-scale genetic differentiation into three groups. The first group comprises all samples from Bahrain, the second from the UAE and Yemen, and the third from Oman (Figure 2a). A neighbor-joining tree also illustrated limited gene flow between Bahrain and all other locations (F ST = 0.01, p < 0.001) (Supporting Information Figure S2). A Mantel test indicated no significant IBD pattern (p = 0.622). All pairwise comparisons involving Bahrain showed high genetic distance, irrespective of geographic distance (data not shown).

| Carcharhinus limbatus
Summary statistics averaged across all loci indicated relatively high levels of heterozygosity across all sampling locations (Table 2).
Numbers of mismatches between reference genotypes and regenotyped replicates were also low (Supporting Information Table   S7), with only one locus (AC 17) showing a high rate of genotyping error (≥5%), caused by an allele scoring error. High genotyping error in other markers (AC 60, AG 2) was attributed to failure of amplification. These loci also deviated from HWE, suggesting that failure of amplification might be caused by allele dropout. These loci were hence excluded from further analysis.
Pairwise F ST values were mostly low but significant (Table 4).
However, samples from Bahrain and Kuwait showed low differentiation from each other but were highly differentiated from all other locations (F ST = 0.13-0.19, p < 0.001) ( Table 4). The probability support produced by STRUCTURE for a range of Ks (1-10) was highest for K = 1, indicating a single population (Supporting Information Figure S1b).
The DAPC scatterplot also supported population subdivision between three groups. The first group comprises all samples from Bahrain and Kuwait, the second from Oman, Pakistan, the UAE and Yemen, and the third from South Africa and the Saudi Arabian Red Sea (Supporting Information Figure 2b). While the second and third groups

| Sex-biased dispersal
The frequency distribution of AI c for C. sorrah differed slightly among sexes (Figure 3a). Males had more positive values, while females had more negative values. Mean AI c values were lower for females (−0.07 ± 0.2 cf. 0.12 ± 0.2 (±SE)) (Figure 3a), yet a Wilcoxon's rank-sum test between sexes was not significant (W = 17,484, p = 1) (Supporting Information Figure S4a).

| D ISCUSS I ON
This study presents a regional analysis of the genetic population structure of two potentially overexploited shark species Spaet et al., 2016) along the coasts of the Arabian Peninsula, Pakistan, and South Africa.
Overall, our findings support three populations for both species.
Population subdivision was supported by pairwise F ST and DAPC, but not by STRUCTURE. The failure of STRUCTURE to identify genetic heterogeneity might be attributed to (a) a variation in sample size among sampling locations (n = 51-96) (Kalinowski, 2011;Puechmaille, 2016) or (b) the complexity and discontinuity of the data space (e.g., multimodality) (François & Durand, 2010;Gilks, 2005) or (c) limited genetic differentiation among populations (Latch, Dharmarajan, Glaubitz, & Rhodes, 2006). In situations of weak genetic differentiation, DAPC has proven to be a powerful tool in detecting fine-scale structure (Novembre et al., 2008; O'Connor et al., 2015; Patterson, Price, & Reich, 2006). We hence believe that for our dataset, maximizing variance between predefined clusters, while minimizing variance within clusters as employed by DAPC (Jombart et al., 2010), was the more sensitive and therefore more appropriate approach to illustrate the observed fine-scale differences.
Philopatry of C. sorrah and C. limbatus around the Arabian Peninsula was not supported by  based on nuclear and mtDNA. Yet, mtDNA variation observed by  might have been insufficient to detect the possible genetic heterogeneity. Past studies detecting philopatry in C. limbatus either showed higher mtDNA haplotype and nucleotide diversity  than that observed in  or focused their sampling on neonates collected from nursery grounds . Failure to detect possible philopatry due to low mtDNA diversity has previously been observed in sharks Male vs female assignment bias Alc (b) C. limbatus (Martin, Naylor, & Palumbi, 1992;Portnoy et al., 2016), as well as in yellowfin tuna, Thunnus albacares (Ely et al., 2005).
We tested for evidence of sex-biased dispersal of C. sorrah and  (Portnoy et al., 2010), Lemon shark (Negaprion ssp) (Schultz et al., 2008), and across ocean basins in the shortfin mako shark Isurus oxyrinchus (Schrey & Heist, 2003 Table S10). Elsewhere, changes in sea surface temperature have been shown to influence the movement of sharks (Keeney & Heist, 2006). In particular, the offspring of C. limbatus migrates from nursery grounds to offshore wintering grounds when temperatures drop below 21°C (Castro, 1996). This Future research to understand the role of philopatric behavior in generating fine-scale structure in shark populations (Momigliano et al., 2017;Pazmiño et al., 2018;Portnoy et al., 2015) in the Arabian region is warranted. Particular focus should be placed on a comparison of geographic scales of heterogeneity partition produced by neutral (both microsatellite and mtDNA) vs. non-neutral markers. In addition, it would be interesting to assess whether genetic heterogeneity is structured at non-neutral markers among nursery grounds.

| CON CLUS ION
Findings of this study have contributed to our understanding of the population structure of C. sorrah and C. limbatus along the coasts of the Arabian Peninsula, Pakistan, and South Africa. Based on the nuclear markers used in this study, we suggest that both C. sorrah and C. limbatus exhibit populations subdividing the area into three groups. In C. sorrah, the first group comprises all samples from Bahrain, the second from the UAE and Yemen, and the third from Oman. In C. limbatus, the first group comprises samples from Bahrain and Kuwait, the second from Oman, Pakistan, the UAE, and Yemen, and the third from South Africa and the Saudi Arabian Red Sea. The generally weak population structure observed in this study may possibly be due to the effect of sex-biased dispersal (i.e., through site fidelity or philopatry), which could promote population closure on finer geographic scales. The distinctiveness of all samples from Bahrain and Kuwait from all other sampling locations could be the result of founder effects during the Tethys Sea closure. Overall, our study suggests that conservationists and resource managers should treat each of the three mentioned groups as separate conservation units.

ACK N OWLED G M ENTS
This work was undertaken with the financial support from Kuwait Foundation for the Advancement of Science (KFAS), whose assistance we gratefully acknowledge. We are very thankful to all the fishermen that we have interacted with. We are mostly grateful to Mr Yaslam Saed (Ministry of Fish Wealth, Yemen) for making the sampling process at Yemen easier than expected, even in a state of war. We would also like to thank Mr Lateef and Mr Mohammed Belal (Deira fish market, the UAE), Mr Akbar Bhai (Pakistan), Mr.
Mohammad Yusur (Socotra, Yemen), and Mr Bo Khalid (Oman) for their valuable guidance and advise on sampling local fish markets and landing sites. We thank Operations and Research staff of the KZN Sharks Board for the provision of blacktip sharks and the collection of tissue samples. Lastly, we would like to pass a special thank you to two anonymous reviewers for their insightful and constructive comments, which significantly improved the manuscript.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
Dareen Almojil designed the study, collected samples for C. limbatus and C. sorrah from Kuwait, Bahrain, UAE, Oman, Yemen, and Pakistan, analyzed and interpreted the data, and wrote the manuscript. Julia Spaet provided tissue samples for C. limbatus from Saudi Arabia and critically reviewed and edited the manuscript. Geremy Cliff provided tissue samples for C. limbatus from South Africa. All authors approved the final version of the manuscript.

DATA ACCE SS I B I LIT Y
The entire dataset used in this study has been deposited in the Dryad Repository, https://doi.org/10.5061/dryad.811nr62.