Overfishing and climate change elevate extinction risk 1 of endemic sharks and rays in the southwest Indian 2 Ocean hotspot 3

56 57 The southwest Indian Ocean (SWIO) is a hotspot of endemic and evolutionarily distinct 58 sharks and rays. We summarise the extinction risk of the sharks and rays endemic to 59 coastal, shelf, and slope waters of the SWIO (Namibia to Kenya, including SWIO islands). 60 Thirteen of 70 species (19%) are threatened: one is Critically Endangered, five are 61 Endangered, and seven are Vulnerable. A further seven (10%) are Near Threatened, 33 62 (47.1%) are Least Concern, and 17 (24.2%) are Data Deficient. While the primary threat 63 is overfishing, there are the first signs that climate change is contributing to elevated 64 extinction risk through habitat reduction and inshore distributional shifts. By backcasting 65 their status, few species were threatened in 1980, but this changed soon after the 66 emergence of targeted shark and ray fisheries. South Africa has the highest national 67 conservation responsibility, followed by Mozambique and Madagascar. Yet, while 68 fisheries management and enforcement have improved in South Africa over recent 69 decades, drastic improvements are urgently needed elsewhere. To avoid extinction and 70 ensure robust populations and future food security, there is an urgent need for the strict 71 protection of Critically species and sustainable 72 management of all species, underpinned by species-level data collection and bycatch 73 reduction. 74 Red 78 80 81 82


Introduction
The southwest Indian Ocean and the adjacent Benguela Current (hereafter SWIO) have 115 one of the most distinctive shark and ray faunas globally, comprised of high richness and 116 endemicity with a large number of evolutionarily distinct species 11,12 . This area exhibits a 117 rich diversity of over 250 species from at least 47 families, in part due to the variety of 118 habitats, including warm-water tropical coral reefs, kelp and mangrove forests, and warm-119 temperate and cool-water rocky reefs and sand flats 12,13 . This biogeography is influenced scale fisheries catch is estimated to be as much as three times that of the industrial 133 sector 16 . Several nations in the SWIO face significant socio-economic challenges and 134 rank in the lowest quartile of the Human Development Index (HDI) 17 , limiting their ability 135 to manage marine resources effectively. This includes sharks and rays, which are subject 136 to generally unregulated take in parts of the SWIO, particularly in artisanal fisheries. 137

138
Here, we provide an assessment of extinction risk status of 70 sharks and rays endemic 139 to the SWIO. Specifically, we: (1) assess the extinction risk of these sharks and rays using 140 the IUCN Red List Categories and Criteria, (2) compare the change in extinction risk over 141 ~40 years against a retrospective assessment for 1980 using the Red List Index, and (3) 142 determine the countries with the most significant national conservation responsibility. 143 Finally, we propose some general policies that, if implemented, will help to safeguard 144 shark and ray populations in the SWIO. 145

148
Taxonomic diversity and species richness 149 150 This study comprised 70 endemic species (38 sharks and 32 rays, the latter comprising 151 guitarfishes, electric rays, and skates) from 7 orders, 20 families, and 39 genera (Table  152 1). Families with the highest species richness were Rajidae (hardnose skates, n = 12, 153 17.1% of all species) and Pentanchidae (deepwater catsharks, n = 15, 21.4%), 154 collectively comprising more than a third (38.5%) of the regional endemic fauna. Species 155 richness was greatest along the South African and southern Mozambican coastlines, with 156 a maximum number of 19 species occurring in each country (Fig. 1a). The richness of 157 threatened (Critically Endangered, CR; Endangered, EN; or Vulnerable, VU) shark and 158 ray species also suggested an inverse latitudinal gradient (n = 13; Fig. 1b). The high 159 concentration of threatened endemics in South Africa is driven by the threatened sharks 160 (n = 8), whereas threatened rays (n=5) were more disparately distributed across the 161 region ( Fig. 1c & d). Nearly one-fifth (n = 13, 19%) of assessed endemic sharks and rays in the region are 166 threatened with extinction (Table 1). One species, the Shorttail Nurse Shark 167 (Pseudoginglymostoma brevicaudatum), is CR and at an extremely high risk of extinction. 168 It is assessed under Criterion A2cd as it has undergone a suspected population reduction 169 of >80% over the past three generations (30 years) due to a decline in habitat quality and 170 actual and potential levels of exploitation. Five species (7%) are EN and face a very high 171 risk of extinction, and seven species (10%) are VU, facing a high risk of extinction ( is insufficient information to accurately assess their extinction risk (i.e., data are so sparse 201 for these species that assessors were not able to determine whether they are CR, LC, or 202 somewhere between). Three of the six species of guitarfishes from the family 203 Rhinobatidae require further information to assign a risk category. One of five endemic 204 scyliorhinid catsharks and three of 15 pentanchid catsharks are DD. There are fewer data 205 available regarding the status of rays overall, and nearly one-third are DD ( (Fig. 4).  Table 1). When disaggregating 269 the RLI down to country-level, the most significant decline in RLI is from 1980 to 2005 in 270 greatest decline in country-level RLI occurred in Madagascar (0.672 to 0.558; Fig. 5c). 272 Nine range countries bear some responsibility for conserving the 70 endemic SWIO 273 species that have been assessed using the IUCN Red List Categories and Criteria (Table  274 3). Consistent with the inverse latitudinal richness trend, South Africa had the highest 275 national conservation responsibility (NCR) of all nine range countries (NCR = 1), followed Here, we provide the first comprehensive reassessment of extinction risk in sharks and 282 rays that are endemic to waters of the SWIO. Of 70 species herein assessed for the IUCN 283 Red List, nearly one-fifth are threatened and thus have a high to extremely high risk of 284 extinction (1 CR, 5 EN, 7 VU). Despite a lack of data from parts of the region, it is clear 285 that excessive fishing activity and limited management capacity are substantial barriers 286 to ensuring robust shark and ray populations into the future. A further quarter of species 287 are DD and could potentially be listed as threatened as additional data become available. 288 Furthermore, this assessment of endemic species belies the overall status of sharks and 289 rays in the region. If we include the wider-ranging coastal, pelagic, and deepwater 290 species, there are 75 additional globally threatened sharks and rays that occur in the 291 region, including 26 that are EN and 12 that are CR. Including these groups would also 292 add another 21 DD species. We next (1) compare these findings to threat patterns globally 293 and in other regions, and identify measures to (2) avoid extinctions, (3) ensure 294 sustainability, (4) maintain robust functional populations, (5) drive down data deficiency 295 gaps, and (6) cope with prevalent and emerging threats. 296

297
The percentage of threatened endemic species in this region (19%) is considerably lower 298 than that observed globally (37%) 1 . At the regional level, 42% of species (n=50) are 299 threatened or predicted to be threatened in the Northwest Atlantic and two-thirds of 300 species (67%, n=48) in the Mediterranean Sea 9 . A regional assessment (including all 301 species, not only endemics) of the Arabian Sea and its adjacent waters region found 302 50.9% of species are threatened 19 . Although we find extinction risk in SWIO to be lower 303 than in these regions, many of the most threatened families found in this region are not 304 included in this assessment, including the sawfishes, wedgefishes, hammerheads, and 305 thresher sharks 8,10,20,21 , and if non-endemics are included, there are 82 of 227 species 306 (36% threatened). In any case, there are at least six endemic species that are endangered 307 (EN or CR) and require urgent conservation action to prevent further declines and 308 extinction.  25 , 321 although since this assessment, its range has been extended to include Mozambique 26 . 322

323
We recommend that governments implement management interventions for CR and EN 324 species without delay. These interventions should involve strict prohibitions on landings 325 where they are not yet in place and capacity for enforcement of laws. Highly impactful 326 fishing gears, such as large-mesh (shark-directed) gillnets and longlines, should be 327 regulated, and legislation against destructive fishing practices such as reef nets and blast 328 fishing, which damage habitats such as coral reefs, should be enforced to ensure the 329 continued presence of these threatened species in the wild. If threats are not mitigated 330 rapidly, species such as the Shorttail Nurse Shark could become extinct in the very near 331 future. This situation could follow that of at least one, possibly two, sawfish species which 332 are already considered locally extinct in South Africa (Largetooth Sawfish Pristis pristis and Green Sawfish P. zijsron). Although they are the first rays protected in the region, 334 protection was implemented too late, two years before the last sighting of a sawfish 27 . 335

336
For species in this region that are VU due to small geographic range sizes, occurring in 337 few locations, and inferred to have declining populations (e.g., Flapnose Houndshark 338 Scylliogaleus quecketti, Natal Shyshark), there is an opportunity to implement spatial 339 closures of important habitat to complement catch and fishing effort reduction 340 approaches. Establishing closures will require the identification of overlap between the 341 existing protected area network and key habitat features and understanding movement 342 behaviour and potential aggregation sites 28,29 . Marine Protected Areas might prove to be 343 a suitable approach for conserving threatened endemic sharks 30 . Even a modest 344 expansion of the protected areas network has significant potential to contribute to the 345 conservation of these species 31 . and post-release mortality is unknown, even in South Africa, where data collection is 362 relatively robust 32 . 363 Encouragingly, almost half of the species assessed here are LC, which means their 365 populations are stable or declining slowly such that population reduction thresholds are 366 not triggered. In many cases, these species are not exposed to the pressures to which 367 threatened species are. For example, the geographic or bathymetric ranges of some 368 species mean they are sparsely or never fished. Even when such a species is fished, 369 resilience to this pressure is indicated by relatively stable population trends over time. fisheries, but given their increase in abundance, they appear robust to moderate levels of 374 fishing activity (< 50 t per annum), although further management measures will be needed 375 to ensure sustainability if catches increase. Some other targeted or retained bycatch 376 species (e.g., Bluntnose Spurdog, Slime Skate Dipturus pullopunctatus) also exhibit some 377 level of resilience to fishing pressure. However, it is essential that these LC species be 378 monitored in terms of abundance and catch to maintain robust, ecologically functional 379 populations that yield ecosystem services to humanity and contribute to food security. 380

381
A quarter of the species assessed had insufficient data available for an accurate 382 assessment and were evaluated as DD. Many countries are still reporting catches as 383 "sharks", and species-level monitoring of rays has been particularly neglected in the 384 region. Catch reconstructions reveal serious discrepancies where reported catches are 385 far lower than the reconstructions, around 200% in Madagascar and Mauritius 34,35 , and 386 >75% in Tanzania 36 . While there has been progress in assessing the species composition 387 and monitoring of fisheries, there remains a lack of species-specific population trend and 388 time-series data, particularly in countries other than South Africa. The lack of species-389 specific fisheries data means that declines in sensitive species (e.g., angelsharks, 390 guitarfishes) could go unnoticed 27,37 . More information may reveal other species that are 391 threatened. Further, more detailed information will be needed to provide effective spatial 392 planning and fisheries management while minimizing impacts and conflicts with resource having undergone reductions of more than 50% over the past three generations due to 401 deepwater trawl and longline fisheries operating within their ranges. These declines will 402 continue if deepwater fisheries are further developed in the absence of management 38 . 403 We caution that as deepwater fisheries increase, particularly in Mozambique, 404 Madagascar, and Tanzania, including fishing by distant water nations, many of the 405 deepwater LC species may be at greater risk of extinction 39 . Monitoring fisheries 406 expansions into deeper or more remote waters overlapping with the geographic ranges 407 of deepwater LC species will be important (along with species-level catch data). 408

409
Although declines in VU species are mainly due to fishing, one of these species, the Tiger 410 Catshark is unique in that it has undergone a population reduction (including a reduction 411 in area of occupancy) that is at least partially related to an ecological shift in currents due 412 to climate change 18,40 . For this catshark, mortality due to fisheries does not appear 413 substantial enough to be the only factor causing this reduction, highlighting the 414 importance of considering climate change in future Red List assessments of sharks and 415 rays. As species distribution models for sharks and rays become available 41 , future 416 assessments could consider using climate projections. Trait-based approaches are 417 already available to evaluate the potential risk of climate change and will be helpful for 418 future reassessment 42,43 . 419 420 Conclusion 421 Here, we find that 13 of the 70 (19%) endemic shark and ray species in the SWIO are 422 threatened with an elevated risk of extinction. There is thus a need for a collaborative 423 regional improvement in shark and ray conservation to reduce risk for these endemic 424 species. However, this limited species sample belies the actual risk to the elasmobranch 425 fauna in the SWIO region, as some of the most highly threatened cosmopolitan 426 elasmobranch groups, including sawfishes, wedgefishes, and hammerhead sharks, are not endemic to this region and were thus not included in this study (the vast majority of 428 which are CR globally). There is a great urgency to act to avoid further extinctions, ensure 429 sustainability, maintain robust functional populations, reduce data deficiency, and secure 430 livelihoods and food security for coastal people. On-going monitoring and data collection 431 at the species level are essential, particularly for threatened and NT species. Species-432 specific annual fisheries-independent population monitoring needs to take place. In the 433 absence of such data, species-specific monitoring of catches and landings (taking into  44,45 . This study provides evidence that extinction risk has increased in the 443 SWIO region due to overfishing and climate change and that action is needed to bend the 444 curve for elasmobranchs there. 445

446
Methods 447 We first describe the geographic and taxonomic scope of the regional endemic 448 elasmobranch extinction risk assessment, followed by the application of the IUCN Red 449 List Categories and Criteria, species mapping and spatial analyses, and the calculation 450 of a Red List Index. The categories CR, EN, and VU are collectively termed 'threatened' categories. A species 497 qualifies for one of the three threatened categories by meeting the quantitative threshold 498 for that category within one of the five criteria (A-E). The NT category is applied to species 499 that come close to, but do not meet, a threshold for a threatened category. The LC 500 category is applied to species that have been assessed against the Red List criteria but 501 do not qualify for CR, EN, VU, or NT. There were two ways species were assessed as 502 LC: i) data show that the species has a stable or increasing population size over 3 GL, or 503 ii) because they inhabit remote or deepwater areas that are not subject to threats and 504 therefore it can be inferred that the population is not undergoing reduction. The DD 505 category is applied to a species when there is inadequate information to make a direct or 506 indirect assessment of the risk of extinction based on its distribution and/or population 507 status 50 . The Red List assessment process includes a structured approach to classifying 508 threats into 11 primary classes, such as human intrusions and disturbance, pollution, 509 biological resource use, and climate change and severe weather 51  Posterior Density interval was used as the interval estimator of 95% credible intervals.

552
While there are many demographic approaches to calculating generation length 50 , these 553 are generally data-intensive and have been applied to relatively few sharks and rays. 554 Therefore, to derive generation length (GL), a simple measure that requires only female 555 age-at-maturity and maximum age was used: 556 GL = maximum age+([maximum ageage-at-maturity]*z)), 557 where z depends on the mortality rate of adults and is typically around 0.3 for mammals 558 but we assume z is 0.5 to account for the truncation of age-structure due to overfishing 559 and underestimateion of age in chondrichthyans. This value represents the median age 560 of parents of the current cohort. To derive population reduction over 3 GL, the 561 proportional decline over the x years of available catch rate or landings datasets was 562 calculated, and this was used to calculate annual proportional change, which was then 563 scaled across the 3 GL period. 564 565 If a species qualified for a change in conservation status from a previously published 566 assessment (a 'downlisting or 'uplisting' in status), changes were classified as either 567 Genuine or non-genuine changes. Genuine changes are assigned due to actual 568 increases or decreases in the level of extinction risk that a species faces based on change 569 in the threatened processes. In contrast, non-genuine changes are assigned due to new 570 information, taxonomic changes, and/or errors in the application of criteria or incorrect 571 data used in the previous assessment 50 . reviewed and validated by regional experts and taxonomists. The final distribution maps 585 were prepared using ArcGIS 10.6. The ranges of each species were clipped to their 586 known depth range based on the highest-resolution bathymetry dataset available across 587 the region (15 arc seconds) 63 . One species, Kaja's Sixgill Sawshark (Pliotrema kajae), 588 was excluded from all spatial analyses, as it was not possible to map its range due to a 589 lack of data. 590 591 Red List Index. We derived retrospective assessments for two earlier periods, 2005 and 592 1980 (with the current assessments set at 2020), in order to calculate a Red List Index 593 (RLI) 64 . Before this current reassessment, all except 15 newly described species had 594 assessments published on the IUCN Red List. All changes in Red List category except 595 one were considered non-genuine changes due to new information 50 Table 3. Nine range countries in the southwest Indian Ocean and their national conservation responsibilities for all 70 endemic shark and ray species across the region, where Red List statuses are known. Responsibility for each country is calculated based on the numbers of species occurring in the country's Exclusive Economic Zone (EEZ), the most recent Red List assessment category, and the proportion of each species' range area occurring in the EEZ (values were normalized to range from 0 to 1).