Distribution and site fidelity of four endemic catshark species in Walker Bay, South Africa

Catsharks (family Scyliorhinidae) and the recently reclassified deepwater catsharks (family Pentanchidae) are two of the largest families of elasmobranchs and include species that function as important mesopredators in almost all marine ecosystems. This study focuses on four species endemic to the coast of southern Africa: the puffadder shyshark Haploblepharus edwardsii, dark shyshark H. pictus, leopard catshark Poroderma pantherinum and pyjama catshark P. africanum. Similar to most catsharks, these four species are underrepresented in chondrichthyan research. Our investigation aimed to gain insight into the distribution and site fidelity of the focal species through mark-recapture efforts in Walker Bay on the southwest coast. The use of 95% minimum convex polygons indicated large overlaps in distribution among the species as well as between sexes, except for H. edwardsii. Site fidelity was universally low (0.005) at three key sample sites, although travel distances between sites averaged 3–5.5 km across all species. The results suggest that sexual segregation is not present for the studied catshark species, with the possible exception of H. edwardsii, which had a low capture rate. The low levels of site fidelity and movement also indicated significant levels of site interconnectivity between the three commonly sampled sites as they fell within the same 5-km2 region of the bay. From the present findings, Walker Bay could be considered an area of interest for conservation with respect to the four species, allowing for further study of their population dynamics and the influence of the local marine protected area.


Introduction
Distribution and site fidelity of four endemic catshark species in Walker Bay, South Africa TL Johnson 1✝ * , JC de Bresser 1✝ , E Cottrant 1,2 , NJ Drobniewska 1 , TG Paulet 1,3  and LG Underhill 2,4 whereas P. pantherinum and P. africanum have greater variation in size, attaining 77 cm and 109 cm, respectively (Ebert et al. 2021b).There is a significant data gap concerning the movement ecology of these four endemic catsharks, with only one study using the mark-recapture technique in the Eastern Cape Province, South Africa, on the movement patterns of H. edwardsii, P. pantherinum and P. africanum, as well as the brown shyshark H. fuscus (Escobar-Porras 2009).In that study, the catshark species showed high levels of site fidelity and travel distances of >150 km across the study region, a pattern reflected in other catshark species, such as S. canicula (Sims 2003;Jacoby et al. 2012) and the nursehound shark S. stellaris (Compagno 1984).
This study aims to build on the research of Escobar-Porras (2009) on the ecology of catsharks endemic to southern Africa.At the time of that publication, H. edwardsii was classified as Near Threatened (Escobar-Porras 2009); however, the species was reevaluated as Endangered in 2019, thus increasing the importance for providing new data to assist in its conservation.We describe the distribution area of the four species, as well as their site fidelity and travel distances across different regions of Walker Bay on the south coast of South Africa.The findings may help in formulating well-substantiated management strategies to benefit the studied species and their respective ecosystems.

Data collection and study area
Data were collected in Walker Bay, South Africa (34°30′ S, 19°20′ E), over 8 years, from January 2014 to June 2022.Primary data collected by the South African Shark Conservancy (SASC) were supplemented by data submitted by anglers to the Oceanographic Research Institute (Jordaan andMann 2021, 2022).Sampling was weather-dependent and no sampling was conducted from March 2020 to January 2021 owing to lockdowns during the COVID-19 pandemic.A total of 27 named fishing locations were sampled, in addition to a further 23 unnamed fishing locations (Figure 1).Sharks recaptured outside the sampling area were also included to calculate travel distances.This investigation was conducted under research permits issued by the Department of Forestry, Fisheries and the Environment (nos.RES2014-35, RES2016-23, RES2017-31, RES2018-59, RES2019-61, RES2020-16 and RES2022-67) and ethics clearance (2021/V15/LU) was issued by the Faculty of Science Animal Ethics Committee of the University of Cape Town.
Walker Bay is characterised by three distinct habitat types: rocky reefs, kelp forests and sand.It is a biodiversity hotspot due to the nearby mixing of the cold Benguela upwelling system and warm Agulhas Current, which meet at the Cape of Good Hope, ~80 km northwest of the bay (Osgood et al. 2019).Since 2001, a small region of Walker Bay (45 km 2 ) has been designated as a seasonal Marine Protected Area (MPA), from 1 July to 30 November, to protect the cetacean population, and the remaining 60 km 2 of the bay is designated as a restricted area (RSA 2001).The three most-sampled sites were Old Harbour, New Harbour and Rietfontein (Figure 1).Old Harbour and Rietfontein are characterised by shallow (<10 m) kelp forests with significant cover of sea bamboo Ecklonia maxima and patches of upright wrack Bifurcariopsis capensis, whereas New Harbour is characterised by shallow (<10 m) sandy habitat.

Mark-recapture
Sharks were caught either by fishing (handline, rod and reel, or longline) or by hand while snorkelling.Handlines comprised 1.2-mm nylon monofilament line and de-barbed circle hooks ranging in size from 6/0 to 12/0 to target various shark species and sizes.Hooks were baited with sardine Sardinops sagax, Pacific saury Cololabis saira, common octopus Octopus vulgaris or Cape Hope (chokka) squid Loligo reynaudii.Bait canisters containing 1 kg of defrosted, crushed sardine were also deployed to attract specimens.For longline fishing, a demersal longline 300 m in length was deployed.The longline had weights of 10 kg attached to each end which served as anchors, and a surface buoy.Twenty gangions with circle hooks (sizes 8-12/0) were attached 15-m apart along the main line and baited with the same bait as used for handlines.Longlines were set for 1.5-2 h.Hand capture while freediving was conducted in water <8 m deep.Captured sharks were held underwater in a soft mesh cage (30-mm mesh size, 600-mm diameter and 250-mm height) for the duration of the dive session (~1 h).Retention in the mesh cage was deemed ethically suitable as catsharks are buccal pumpers and thus lack of movement does not cause metabolic stress (Ebert et al. 2021b).Sharks with a lateral span smaller than the mesh size were taken directly to the laboratory to prevent escape and held in tanks until processing after the dive session.Fishing and snorkelling took place primarily during the day between 07:00 and 16:00.
Captured sharks were placed on a measuring mat and had their eyes covered with a wet cloth to minimise stress.The total length (TL in cm) was recorded and sex was determined by the presence or absence of claspers.
Individuals were tagged using external dart tags (Dunlop et al. 2013).The tag number of recaptured and newly tagged individuals was recorded, together with morphometric data; sharks <30 cm TL were not tagged because of their small size.
The recapture rate (R) for each species was calculated using the equation: where C is the total number of capture events including captures and recaptures, and N is the total number of individuals captured.

Data analysis
All analyses were conducted using RStudio with R 4.2.0 (RStudio Team 2022).All statistical analyses were performed using nonparametric tests by reason of a non-normal distribution of the data (Shapiro-Wilk: p < 0.05).

Minimum convex polygons
All single capture and recapture locations of individuals were used to determine the species' distribution.For analysis of the distributions of the four species, minimum convex polygons (MCP) were constructed using the R package 'adehabitatHR' (Calenge 2006) and then illustrated through maps with the packages 'ggmap' (Kahle and Wickham 2013) and 'rstudioapi' (Ushey et al. 2020).Any data-collection errors were accounted for using 95% polygons, as was the bias associated with MCPs (Burgman and Fox 2003;Nilsen et al. 2008).Nevertheless, the outer 5% of the capture points were also shown for additional information and interpretation.MCPs were constructed for each species to analyse their general distribution across Walker Bay and were also separated by sex.The area of each polygon was calculated and then corrected to exclude landmass using the polygon tool in Google Earth Pro 7.3.4.8642.
Using polygon areas, the percentage overlap between sexes or species was derived according to the formula described by Attwood and Weeks ( 2003 where area αβ is the shared area of both groups or species, and distribution area α and distribution area β are the individual polygon areas of sex/species α and β, respectively.Directional overlap percentages between sexes were determined by calculating each part of the above equation separately, as follows: This resulted in two overlap percentages per species and by sex, instead of one, such as with the Attwood and Weeks formula.Directional overlap percentages were used as extended information, yet the Attwood and Weeks (2003) formulas were used for the interpretation of the results in the Discussion.

Site fidelity and travel distance
Site fidelity (SF) was calculated as per the methods described by Tschopp et al. (2018) from species permanence (IT) and species periodicity (It): where F i is the proportion of time in the study area given by the time between the first and last capture (days); F is the length of the sampling period (days); c is a binary value indicating capture ('1') or no capture ('0') of an individual I on sampling occasion j; and T is the number of sampling occasions.If recapture did not occur, then IT, It and SF were set to 0. SF was measured on a scale of 0-1, where 0 equates to no fidelity and 1 to absolute fidelity.All values were calculated for every individual tag number, and the mean and standard error were further calculated for each species per location and sex to provide a population average.For further details on the method refer to Tschopp et al. (2018).Only Old Harbour, New Harbour and Rietfontein (Figure 1) were used in this analysis to provide a baseline for site fidelity over a suitable sample size.Because of heavy zero-weighting from single captures (Tweedie 1984), each index was compared per location and sex using a Tweedie general linear model with log link (T) produced using the 'tweedie' function from the package 'statmod' (Dunn 2022).
To determine straight-line travel distances of individual sharks, the Haversine formula was used to calculate distances between capture sites (Robusto 1957).The maximum and mean travel distances were then recorded for each species and sex.Mann-Whitney tests were used for each species to determine any significant differences in travel distance between sexes.

Dataset overview
The dataset contained 2 461 capture events, representing 1 950 individual sharks.The most common shark to be captured was H. pictus, with 1 295 capture events

Distribution area Overall species distribution
When considering all individuals of each species, the 95% MCP showed that H. pictus had the largest estimated overall distribution range across Walker Bay at 133.1 km 2 when corrected for landmass (Figure 2).The secondlargest distribution was represented by P. pantherinum with 114.5 km 2 , followed by H. edwardsii with 98.0 km 2 .The smallest estimated distribution range was shown by P. africanum with 61.2 km 2 .
Inspecting the overlap between the four shark species showed an area that included the four distributions.This area of overlap was measured to be 46.7 km 2 .Haploblepharus pictus enveloped P. pantherinum and P. africanum completely in the distribution area, with a calculated shared overlap of 92% and 67%, respectively.In turn, P. pantherinum enveloped P. africanum completely in the distribution, with a calculated shared overlap of 73%.The lowest shared percentage overlap was found between P. pantherinum and H. edwardsii, with 44% overlap.Furthermore, H. edwardsii shared their distribution area with H. pictus with 61% overlap and with P. africanum with an overlap of 62%.Directional overlap is summarised in Supplementary Table S1.
Overlap between groups based on sex showed that in three of the four shark species (i.e.H. pictus, P. pantherinum and P. africanum) the group with the smaller distribution was completely enveloped by the group with the larger distribution.Specifically, H. pictus males, P. pantherinum females and P. africanum females were completely enveloped by their male or female counterparts.The largest level of shared overlap between males and females was found in P. africanum at 93%, followed by P. pantherinum at 82% and H. pictus at 55%.Haploblepharus edwardsii showed the smallest level of overlap, with 21% shared area between males and females.Directional overlap between males and females per species is summarised in Supplementary Table S2.
Inspection of all capture points showed that the group with the largest 95% distribution area was not consistent with the farthest capture/recapture coordinates for each species.Specifically, H. pictus females had the largest 95% distribution, whereas males were represented in the outermost points of overall distribution.Similarly, P. africanum males represented the largest 95% distribution, whereas females showed the largest overall distribution in Walker Bay.For P. pantherinum, there were no noticeable differences in total distribution between males and females, although males showed a larger 95% distribution.Finally, the 95% distribution for H. edwardsii females was clearly larger than that for males yet the difference in overall distribution was less marked.4).With respect to all four shark species, no significant interaction models were found using the Tweedie GLM (p ≥ 0.05), and thus both sexes and all locations could be combined into a single model for each species.Haploblepharus edwardsii and P. africanum expressed no significant difference in SF between sexes or locations in any instance (p ≥ 0.05).Haploblepharus pictus expressed significantly greater SF in males than in females, as well as at Old Harbour compared with at New Harbour and Rietfontein (p < 0.05).However, no such significant difference was noted between New Harbour and Rietfontein (p ≥ 0.05).Poroderma pantherinum expressed a significant difference in SF only between Old Harbour and New Harbour (p < 0.05) with no other significant differences present for sex or the remaining pairwise locations (p ≥ 0.05).All test statistics are summarised in Supplementary Table S3.

Travel distance
Of 318 individuals that were recaptured at least once, only 93 instances of travel were recorded from 83 individual sharks.They included 34 H. pictus, 23 P. pantherinum Females of P. africanum were found to travel the farthest, with a maximum recorded travel distance of 38.55 km (Table 1).There were no significant differences in the mean travel distance (km) between males and females of any species (p ≥ 0.05).

Discussion
The goal of this study was to gain insight into the distribution and site fidelity of four endemic catshark species through mark-recapture efforts in Walker Bay, South Africa.Relatively large areas of overlap were found between the sexes for H. pictus, P. africanum and P. pantherinum, but not for H. edwardsii.Overlap between species was found to be relatively large as well.Travel distances between sites averaged between 3 and 5.5 km across all species.Nevertheless, site fidelity (SF) was universally low (<0.005) for all key locations in this study.
Haploblepharus pictus dominated the majority of captures of the four target species, contributing to >50% of all captures, as previously reflected in an occurrence analysis using baited remote underwater videos (Osgood et al. 2019).Considering that most fishing took place inshore, a shallow topography was expected.This was especially prevalent at the most common fishing sites, Old Harbour, New Harbour and Rietfontein, which did not exceed 10 m in depth.Therefore, it was likely that depth preference played a major role in the number of captures since H. pictus is rarely found deeper than 35 m depth (Pollom et al. 2019a;Ebert et al. 2021b), whereas the remaining three species can be found in depths exceeding 100 m (Pollom et al. 2020a(Pollom et al. , 2020b(Pollom et al. , 2020c)).Although this reasoning is also likely consistent when considering the small total number of captures for H. edwardsii, it should also be considered that the Endangered status of this species played a credible role in its relative rarity (Pollom et al. 2020c).Specifically, if the population status of H. edwardsii did not play a role, the large difference in capture numbers when compared with P. pantherinum and P. africanum would probably not have been observed, considering that all three species are found across a similar depth range (Pollom et al. 2020a(Pollom et al. , 2020b(Pollom et al. , 2020c)).
All four species demonstrated moderate to high levels of spatial overlap with each other (>40%), suggesting significant levels of co-occurrence and spatial interaction.Evidence provided by both Compagno (1984) and Escobar-Porras (2009) suggests levels of microhabitat separation between H. edwardsii, P. pantherinum and P. africanum, as well as with the brown shyshark H. fuscus, which has a similar habitat preference to H. pictus (Human 2007;Ebert et al. 2021b).This could suggest that, whereas the species studied here have spatial overlap at the coarse scale, they likely segregate across microhabitats, although further studies that investigate the reasoning for segregation would be recommended.The lowest level of overlap with other species was expressed by H. edwardsii, at 44-62%.While the low capture numbers of this species likely impacted this result (Pollom et al. 2020c), it is possible that H. edwardsii might be outcompeted by the other species of interest and thus seeks refuge in areas across greater distances.Therefore, it is likely that intraspecific competition, alongside other factors such as trawling bycatch, targeted recreational fishing and habitat degradation, is driving H. edwardsii to extinction (Attwood et al. 2000(Attwood et al. , 2011;;da Silva et al. 2015).
When considering potential sexual segregation, results from this study do not align with previous research on other catsharks (Richardson et al. 2000;Wearmouth and Sims 2008;Wearmouth et al. 2012;Riesgo et al. 2019;Vásquez-Castillo et al. 2021).Often, sexual segregation is initiated owing to the harassment of females by males (Wearmouth et al. 2012).This may mean that harassment is less common in the catshark species in this study.An exception to this was H. edwardsii, which did show a substantial difference between male and female distribution, with only 21% overlap.The most probable explanation lies in the low number of individuals of H. edwardsii and consequently less weight of the MCP directional overlap being pulled towards the three main fishing locations (Burgman and Fox 2003;Nilsen et al. 2008).An alternative explanation might be that sexual segregation is present within this species and that females reside in shallower areas to avoid male harassment.This has been observed in females of other catshark species, where additional benefits include limiting energy-costly mating or staying close to their eggs in shallower water (Sims et al. 2001).In addition to resulting in potentially fewer mating events, sexual segregation could impact the fitness of H. edwardsii, as females would be more susceptible to being included as bycatch in area-focused fisheries (Wearmouth et al. 2012).Lower fitness caused by sexual segregation could explain why H. edwardsii is less abundant and more susceptible to disturbance than ecologically similar species such as H. pictus.
We acknowledge the potential for overestimation or underestimation of MCP size in this study, as a result of sample size and effort bias (Burgman and Fox 2003;Nilsen et al. 2008).Nevertheless, most of the possible limitations are buffered by focusing on 95% of the data while still including the total distribution in the results.This way, the illustrated area can be used with greater confidence for possible management strategies, as the MCPs show the area where most catshark species were observed.Samplesize sensitivity did not play a role in the MCPs derived for H. pictus, P. africanum and P. pantherinum, whereas the same cannot be said for H. edwardsii, which future research should take into account.Nevertheless, not much is known about this Endangered species, and MCPs remain of great importance for possible conservation in highlighted areas.
The overall SF was minimal in all species, failing to surpass a mean value of 0.006 in any instance.The low overall SF does not align with the results of Escobar-Porras ( 2009) that suggested high levels of SF.While this could suggest large degrees of variance in SF between different catshark subpopulations, the method of calculation was unclear in the study of Escobar-Porras (2009) and was also likely impacted by the small sample size.While the low SF would at first appear contradictory to the small mean travel distances, as well as the small number of individuals recorded to have travelled, it is important to consider that travel distance measurements relied heavily on the probability of recapture, which was small to moderate at most and was heavily weighted towards Old Harbour, New Harbour and Rietfontein.The low SF values would suggest that every shark species travels more than the travel distances suggest.Catsharks are not strongly adapted to swimming because of their soft fin structure and a reliance on buccal pumping (Thomson and Simanek 1977;Ebert et al. 2021b), characteristics that seemingly run counter to the higher-than-expected travel distances.However, there are potential behavioural explanations for the low levels of SF, such as nocturnal feeding behaviour (Ebert et al. 2021b) or post-capture avoidance of the fishing sites.Further studies to measure these behavioural responses are recommended.It is also recommended that a larger number of sites are sampled in subsequent studies to gain a greater representation of SF in Walker Bay as a whole.
Despite low levels of SF, it was apparent that H. pictus showed greater SF at Old Harbour than at New Harbour and Rietfontein, and similarly P. pantherinum showed greater SF at Old Harbour than at New Harbour.Considering that Old Harbour and Rietfontein share the same habitat type, it is unlikely that the presence of kelp played a key role in habitat preference in the case of H. pictus.Kelp habitats are typically considered more suitable for catsharks compared with sand habitats since they provide a greater selection of prey and shelter from predators, as well as suitable sites for egg deposition (Pretorius and Griffiths 2013;Osgood et al. 2019;EC unpublished data).While this could explain the greater SF for P. pantherinum, other environmental factors need to be considered for H. pictus, such as food availability, intraspecific competition or recreational fishing pressure.However, without quantification of these variables it is difficult to ascertain the cause of the significant difference between catsharks at Old Harbour and Rietfontein.Future research should include more-diverse fishing locations for capture data.This would not only avoid effort bias for MCP, but also provide more information on potential drivers such as depth distribution, habitats and anthropogenic disturbance.With the results of this study, potentially supplemented with additional data in the future, more efficient management strategies can be formed.This will be beneficial for the conservation of all four catsharks and H. edwardsii especially.

Figure 2 :
Figure 2: Total distribution of four endemic catsharks obtained from 95% minimum convex polygons in Walker Bay, southwest coast of South Africa

Figure 3 :
Figure 3: The 95% distribution areas of the studied catsharks in Walker Bay, South Africa, based on sex variance: (a) Haploblepharus edwardsii, (b) Haploblepharus pictus, (c) Poroderma pantherinum, and (d) Poroderma africanum Mean site fidelity of males and females of Haploblepharus edwardsii, Haploblepharus pictus, Poroderma pantherinum and Poroderma africanum at Old Harbour, New Harbour and Rietfontein in Walker Bay, South Africa.Bars represent standard error

Table 1 :
The mean and maximum travel distances of recaptured catshark individuals across Walker Bay, South Africa.Haploblepharus edwardsii was excluded as no individual was found to have travelled between the sites sampled