Remote reefs and seamounts are the last refuges for marine predators across the Indo-Pacific

Abstract

Since the 1950s, industrial fisheries have expanded globally, as fishing vessels are required to travel further afield for fishing opportunities. Technological advancements and fishery subsidies have granted ever-increasing access to populations of sharks, tunas, billfishes, and other predators. Wilderness refuges, defined here as areas beyond the detectable range of human influence, are therefore increasingly rare. In order to achieve marine resources sustainability, large no-take marine protected areas (MPAs) with pelagic components are being implemented. However, such conservation efforts require knowledge of the critical habitats for predators, both across shallow reefs and the deeper ocean. Here, we fill this gap in knowledge across the Indo-Pacific by using 1,041 midwater baited videos to survey sharks and other pelagic predators such as rainbow runner (Elagatis bipinnulata), mahi-mahi (Coryphaena hippurus), and black marlin (Istiompax indica). We modeled three key predator community attributes: vertebrate species richness, mean maximum body size, and shark abundance as a function of geomorphology, environmental conditions, and human pressures. All attributes were primarily driven by geomorphology (35%-62% variance explained) and environmental conditions (14%-49%). While human pressures had no influence on species richness, both body size and shark abundance responded strongly to distance to human markets (12%-20%). Refuges were identified at more than 1,250 km from human markets for body size and for shark abundance. These refuges were identified as remote and shallow seabed features, such as seamounts, submerged banks, and reefs. Worryingly, hotpots of large individuals and of shark abundance are presently under-represented within no-take MPAs that aim to effectively protect marine predators, such as the British Indian Ocean Territory. Population recovery of predators is unlikely to occur without strategic placement and effective enforcement of large no-take MPAs in both coastal and remote locations.

Fig 1. Indo-Pacific sampling efforts and frequency distribution of predator attributes.

(A) Map of deployments (n = 1,041) with protection level and numbers of deployments per region, unprotected (outlined in blue), partially protected or small no-take MPAs (outlined and filled in pink), and large no-take MPAs (>1,000 km², outlined and filled in green). (B) Frequency distributions of vertebrate species richness, (C) mean maximum body size (cm), and (D) shark abundance (sum of MaxN) across all deployments. The numerical values for B, C, and D can be found in S1 Data. (E) Shark abundance (log[SumMaxN + 1]) in each region (same color scale as for A). (F) Key to EEZs within the Indo-Pacific. EEZ from https://rosselkhoznadzor.carto.com/tables/world_maritime_boundaries_v8. Some EEZs are contested. 1, BIOT (UK); 2, Maldives; 3, Sri Lanka; 4, Cocos (Keeling) Island (Aus); 5, Malaysia; 6, Christmas Island (Aus); 7, Indonesia; 8, Australia; 9, Palau; 10, Papua New Guinea; 11, Micronesia; 12, Solomon Island; 13, Nauru; 14, New Caledonia (Fr); 15, Vanuatu; 16, Norfolk Island (Aus); 17, Marshall Islands; 18, Kiribati; 19, Fiji; 20, Tuvalu; 21, Kermadec Island (NZ); 22, Wallis and Futuna (Fr); 23, Samoa; 24, Tonga; 25, Howland and Baker Island (US); 26, Tokelau (NZ); 27, Phoenix Island Group; 28, Niue (NZ); 29, American Samoa (US); 30, Palmyra Atoll (US); 31, Cook Island (NZ); 32, Jarvis Island (US); 33, Line Island Group (US); 34, French Polynesia (Fr); 35, Pitcairn (UK). EEZ, Exclusive Economic Zones; MPA, marine protected area.

Fig 2. Examples of midwater predators surveyed by the BRUVS.

(A) Blue shark (Prionace glauca). (B) Rainbow runner (Elagatis bipinnulata). (C) Mahi-mahi (Coryphaena hippurus). (D) Black marlin (Istiompax indica). BRUVS, baited remote underwater video system.

Fig 3. Drivers and patterns of vertebrate species richness in the Indo-Pacific.

(A) Relative contribution of main drivers explaining variations in species richness were generated from 100 iterations of BRTs. (B, C) Partial dependence plot (lines), observed values (dots), and 95% confidence intervals for distance to the coast (B) and SST. (D) Predictions of species richness (top 5% values, >3.8, in red). The numerical values for (A) can be found in S2 Data. BRT, boosted regression tree; dist coast, distance to nearest coast; dist CoralTri, distance to the Coral Triangle; dist seamount, distance to nearest seamount with summit depth <1,500 m; SST, sea surface temperature.

Fig 4. Drivers and patterns of mean max body size in the Indo-Pacific.

(A) Relative contributions of the main drivers explaining variation in body size were generated from 100 iterations of BRTs. (B,C) Partial dependence plot (lines), observed values (dots), and 95% confidence intervals for SST (B) and distance to nearest market (thresholds represented by breaking point [C]). (D) Prediction values of body size (top 5% values, >108 cm, in red). The numerical values for (A) can be found in S2 Data. BRT, boosted regression tree; Dist market, distance to nearest market; Dist pop, distance to nearest population; Dist seamount, distance to nearest seamount with summit depth of <1,500 m; SST, sea surface temperature.

Fig 5. Drivers and patterns of shark abundance in the Indo-Pacific.

(A) Relative contributions of drivers explaining variations in shark abundance (log[sumMaxN + 1]) were generated from 100 iterations of BRTs. (B,C) Partial dependence plot (lines), observed values (dots), and 95% confidence intervals for distance to nearest market (B) and seabed depth (C) and thresholds represented by breaking point (C). (D) Predicted values of shark abundance and hotspots (top 5% values, >0.54, in red). The numerical values for (A) can be found in S2 Data. BRT, boosted regression tree; Chla, chlorophyll-a concentration; Dist coast, distance to nearest coast; Dist CoralTri, distance to the Coral Triangle; Dist market, distance to nearest market; Dist seamount, distance to nearest seamount with a summit depth of <1,500 m; SST, sea surface temperature.

Fig 6. Frequency distributions of predator attribute values predicted to occur under different spatial management regimes in the Indo-Pacific.

(A) Vertebrate species richness, (B) body size, and (C) shark abundance (log[sumMaxN + 1]) across the entire unprotected Indo-Pacific, inside partially protected or small MPAs and inside large no-take MPAs (>1,000 km²). Vertical lines and values are associated medians. MPA, marine protected area.

Fig 7. Predicted shark abundance and occurrence along a gradient of human pressures (Distance to Market) and habitat suitability (Depth).

Values are segregated according to protection levels and whether they are hotspots (>.95 quantiles) or not (NA).