3D ocean assessments reveal that fisheries reach deep but marine protection remains shallow

Global conservation efforts are poised to expand under the Kunming-Montreal Global Biodiversity Framework, which targets 30% ocean protection by 2030 and recognizes the role of protected areas and other effective area-based conservation measures (OECMs). The conclusion of the High Seas Treaty (BBNJ) further broadens opportunities offshore and in areas beyond national jurisdiction. However, conservation planning has largely remained two-dimensional, influenced by terrestrial paradigms and a historical focus on shallow waters, despite the inherently three-dimensional nature of marine ecosystems. Deep and vertically complex habitats are under increasing pressure, yet assessments of human use and conservation achievements remain 2D. A critical gap exists in understanding which depths fisheries target, although sensitivity to fishing varies strongly with depth. The authors propose shifting to a 3D representation of ecosystems, human use, and impacts. They develop and apply a framework overlaying benthic and pelagic depth realms onto marine ecoregions to assess 3D ecological representativeness of conservation and the depth distribution of fishing effort, and to evaluate alignment between protection and fishing pressure in 3D, thereby informing global conservation policy.

Prior research has highlighted that deep marine ecosystems (mesopelagic, deep reefs, seamounts) are increasingly pressured and require targeted conservation, yet global assessments of conservation and fisheries footprints have remained largely 2D. Global cumulative impact maps have been instrumental for policy but lack depth resolution. Some recent prioritization efforts consider 3D or even 4D concepts for climate-smart design, but a comprehensive 3D framework for identifying conservation gaps across depths has been missing. Fisheries science documents fishing “down the deep,” with expansion offshore and deeper as shallow stocks decline; deep-sea fisheries often show poor profitability without subsidies and pose sustainability challenges, including high bycatch and long-lasting habitat impacts. Governance fragmentation and the historical bias toward shallow, nearshore research and management contribute to the persistence of 2D approaches. The literature also notes that depth influences ecosystem sensitivity, vertical ecological connectivity transmits impacts across depths, and that OECMs (often vertically zoned) can complicate achieving holistic biodiversity outcomes if depth structure is ignored.

The study constructs a global 3D ocean zoning by overlaying depth realms (benthic and pelagic) onto established marine ecoregions. Ecoregions: The authors use the global map of coastal and pelagic realms from Spalding et al. (2007, 2012), accessed via UN-WCMC, at 0.01° precision. Coastal ecoregions extend to 200 nautical miles (or to the 200-m isobath if farther), while “pelagic” ecoregions represent off-shelf waters; to avoid confusion with water-column terminology, these are termed off-shore ecoregions. Depth realms: Benthic zones include euphotic (0–30 m), upper mesophotic (30–60 m), lower mesophotic (60–150 m), rariphotic (150–300 m), upper bathyal (300–1000 m), lower bathyal (1000–3500 m), abyssal (3500–6000 m), and hadal (>6000 m). Pelagic zones include epipelagic (0–200 m), mesopelagic (200–1000 m), bathypelagic (1000–3500 m), abyssopelagic (3500–6000 m), and hadopelagic (>6000 m). Conservation data: The World Database on Protected Areas (WDPA) and the World Database on OECMs are processed following recommended methods to ensure reliable area and protection-level estimates. Only MPAs with known spatial boundaries are retained; terrestrial parts are clipped; data are simplified to 0.01° resolution. IUCN management categories are used as a proxy for protection levels, grouping “Not Applicable/Not Reported/Not Assigned” as “Unknown.” To avoid double-counting overlapping designations, the authors retain the highest protection level in overlapping polygons via a stepwise subtraction of higher-category areas from lower ones. Bathymetry: GEBCO 2020 grid provides depth, enabling assignment of each protected cell to a benthic realm and, for pelagic protection, inclusion of all pelagic realms from surface to seabed for that cell. Cell areas are computed accounting for latitude-dependent cell width in EPSG:4326; results are validated against the terra package’s cellSize() function. Protection coverage per realm is calculated as protected area divided by total area of that realm, reported for benthic and pelagic. Fishing data: Global Fishing Watch (GFW) Fleet Daily Fishing Activity v2.0 at 0.01° resolution is used for 2018–2020. GFW compiles AIS and VMS data, primarily covering large commercial vessels (capturing an estimated 50% of EEZ fishing and 80% of high-seas fishing). The 2019 dataset (n≈205.7 million fishing events) is the reference year in figures, with 95% CIs across 2018–2020. Each fishing event is geolocated, attributed to an ecoregion and the bathymetric depth beneath it, and joined to gear information via MMSI. Two approaches are used: (1) Depth impacted: assumes vertical connectivity and counts all realms from surface to seabed under the fishing location as impacted. Fishing pressure (hours km²) per benthic realm and per 3D realm (depth realm within ecoregion) is computed by summing hours and dividing by spatial extent. (2) Depth targeted: discriminates pelagic vs benthic gears (possible for ≈55% of hours). Benthic gears are assumed to target the benthic realm matching seafloor depth at the location. Pelagic gears are assigned to pelagic realms within their typical operating depth ranges (from literature/technical sources and expert validation) constrained by local bathymetry. Ambiguous gear types (≈45% of hours) are labeled “unspecified.” Statistical analyses and prioritization: The authors relate protection coverage (log-transformed) of 3D realms to fishing pressure (log-transformed). They define conservation priority profiles by crossing fishing pressure (below/above median) with protection progress (behind/past halfway from 2020 to 2030 coverage targets), applied both to all MPAs+OECMs and to high protection (IUCN Ia/Ib) only. Software: Analyses conducted in R using sf, terra, ggplot2, tidyr; QGIS used for visualization; code available via Zenodo. Data are open-access via sources listed in Methods.

• Protection across depths is uneven. The euphotic realm (0–30 m) is best protected with approximately 15% coverage and 1.2% in high protection (IUCN Ia/Ib), and it is most widely represented across ecoregions. The abyssal realm has the smallest overall protection coverage (~5.8%) and minimal high protection (~0.6%), with 75% of abyssal area in areas beyond national jurisdiction, suggesting potential gains via the BBNJ Treaty. Within EEZ-dominated depths, lower mesophotic and rariphotic realms are among the least protected globally and across most coastal ecoregions. Across depths, most coverage is IUCN VI or unknown, and high protection (Ia/Ib) averages only ~0.7% globally; one-third of depth-by-ecoregion units have <0.1% Ia/Ib. The hadal realm shows the greatest Ia/Ib coverage (~3%), with higher high-protection fractions in the Eastern Indo-Pacific (≈10%) and polar ecoregions (≈1–2%).
• Fisheries are vertically extensive. Fishing pressure and gear types are structured by bathymetry: areas over euphotic to upper bathyal depths have above-average pressure dominated by trawlers; areas over lower bathyal to hadal depths have lower average pressure dominated by drifting longlines. Average fishing pressure declines with increasing depth, with a sharp six-fold decrease beyond ~1500 m; however, total fishing effort is similar over abyssal and mesophotic areas. Benthic fishing effort is greatest in euphotic and mesophotic and persists to upper bathyal, occurring down to lower bathyal. Pelagic mid-water trawls and drifting longlines are common over lower bathyal to abyssal depths. Overall, about 37% of total fishing effort overlies depths >300 m, directly or indirectly affecting deep ecosystems.
• Misalignment of protection with fishing pressure. Protection coverage of 3D realms is negatively correlated with fishing pressure, indicating a bias toward least-impacted areas. For all MPAs+OECMs: p = 0.021, R^2 = -0.22; for high protection (Ia/Ib) only: p < 0.001, R^2 = -0.33. Ninety percent of 3D realms have not achieved 5% high protection, and 47% of 3D realms combine low coverage with high fishing pressure, falling into the highest priority category for conservation action.
• Priority areas and policy implications. Highest conservation priorities—low protection and high fishing pressure—concentrate in mesophotic, rariphotic, and upper bathyal realms across ecoregions. Lowest priorities—high protection and low pressure—tend to occur at lower bathyal and abyssal depths in coastal ecoregions. To maximize ecological benefits, future MPAs/OECMs should avoid residual placement in low-use, already well-represented areas and instead target underrepresented depths and high-pressure realms; high-seas action is needed for lower bathyal and abyssal gaps. Depth regulations (e.g., trawl depth limits) can help curtail unsustainable deep-sea fishing.

By constructing and applying a 3D framework that overlays depth realms onto marine ecoregions, the study directly addresses the central question of how well marine conservation represents biodiversity across depth and how it intersects with the depth distribution of fishing. The findings reveal systemic underprotection at mesophotic, rariphotic, bathyal, and abyssal depths, while fisheries operate throughout the water column and down to deep benthic realms. The strong negative relationship between protection and fishing pressure demonstrates residual conservation patterns, where protected areas are preferentially placed where they conflict least with human use, limiting conservation effectiveness. The results underscore that conservation targets defined and tracked in 2D can mask 3D representation gaps and overlook vertical connectivity, which transmits impacts from pelagic to benthic systems. The study argues for surface-to-seafloor protection as a default to capture vertical ecological connections and to avoid loopholes created by vertically zoned OECMs/MPAs. It emphasizes the need to expand high or full protection (akin to IUCN Ia/Ib) to achieve ecological, social, and climate co-benefits, particularly where fishing pressure is high and in underrepresented depth realms. The BBNJ Treaty is highlighted as a mechanism to address abyssal and off-shore gaps. Incorporating depth into planning and monitoring frameworks can guide climate-smart conservation by providing a portfolio across depth gradients and accommodating species’ depth shifts under climate change.

This work introduces the first global 3D assessment framework integrating depth realms with ecoregions to evaluate conservation representativeness and the depth distribution of fishing. It shows that while fisheries reach deep—impacting benthic and pelagic realms throughout the vertical ocean—marine protection remains shallow, with high protection severely underrepresented across depths and biased toward low-pressure areas. The framework identifies priority 3D realms—especially mesophotic, rariphotic, and upper bathyal—for immediate conservation action and shows the need for high-seas measures to close lower bathyal and abyssal gaps. The authors recommend: adopting surface-to-seafloor protection as standard; incorporating depth-based indicators into global and national conservation targets and tracking; increasing the fraction of high or full protection to at least 10% as recommended by the scientific community and policy strategies; and aligning protection with areas of high fishing pressure to realize net biodiversity gains. Future research and implementation should integrate additional human pressures (renewables, hydrocarbons, potential deep-sea mining), account for climate-driven shifts in species and fishing effort, and improve fisheries data to distinguish pelagic vs benthic activities and gear depth ranges at higher spatial resolution.

Key limitations include data gaps in fishing records: approximately 45% of GFW fishing activity lacks clear pelagic vs benthic gear discrimination, and AIS-based datasets underrepresent small-scale fisheries, particularly in regions such as the Caribbean, Southwest Pacific, and Indian Ocean, leading to underestimation of fishing pressure in shallower realms. Bathymetry is used as a proxy to infer vertically impacted depths, which, while grounded in evidence of vertical connectivity, remains an indirect measure of targeted depths. Conservation prioritization is based on two variables—protection coverage and fishing pressure—excluding other growing pressures (e.g., offshore energy, hydrocarbon extraction, deep-sea mining) and not explicitly modeling future climate-driven redistributions of species and fishing effort, which could refine spatial priorities at finer resolutions. Despite these constraints, multi-year stability in fishing-bathymetry patterns supports robustness of main conclusions.