Projecting contributions of marine protected areas to rebuild fish stocks under climate change

Climate change is altering ocean conditions (warming, deoxygenation, acidification, sea level rise, more frequent/intense marine heatwaves), driving species range shifts, phenology changes, and ecosystem restructuring. Overall, fisheries yields have been negatively impacted, though with regional variability. Many stocks remain over-exploited, with notable regional contrasts across European waters. Rebuilding over-exploited stocks is a core management objective and legal requirement in several jurisdictions (e.g., EU CFP). While reducing fishing pressure and spatial/temporal closures are common tools, climate change may hinder rebuilding to sustainable targets. Prior work suggests that under 1.5–2.6 °C warming, a nontrivial fraction of ecoregions show no sign of rebuilding at MSY fishing levels, implying the need for conservation-focused reductions. MPAs can increase biomass, diversity, and sizes within boundaries, and benefits can spill over. Yet, the effectiveness of MPAs under climate change, including the hypothesized “Protection Paradox,” remains uncertain. This study asks how no-take MPAs, alongside effort controls, could contribute to climate-resilient rebuilding in the Northeast Atlantic. It hypothesizes: (1) no-take MPAs increase stock biomass under climate change; (2) larger MPA coverage combined with conservation-focused management more effectively rebuilds biomass; (3) MPAs improve catches for over-exploited stocks via spillover.

The paper synthesizes prior evidence that no-take MPAs are generally more effective at rebuilding biomass than partially protected areas, with documented spillover benefits to fisheries. Meta-analyses and modeling studies have explored social-ecological trade-offs and climate-ready fisheries strategies. Previous global simulations show that under warming, fishing at MSY may not suffice to rebuild stocks, and more conservative strategies are needed. The study also situates findings within debates on the “Protection Paradox,” whereby protected, less-disturbed systems may harbor species more sensitive to climate stressors, as well as the importance of governance and stakeholder engagement for MPA effectiveness.

Study area and stocks: 739 exploited stock units of 231 species (202 fishes, 29 invertebrates) across eight Northeast Atlantic marine ecoregions (FAO Area 27) with reconstructed catches (2000–2019) from Sea Around Us were modeled. Stocks are species-by-ecoregion units with estimated catches. Model: A linked climate–fish–fisheries dynamic bioclimate envelope model (DBEM) simulated annual abundance, biomass, and potential catches on a 0.5° grid for surface and bottom habitats. Environmental drivers from GFDL-ESM4 included sea surface/bottom temperature, dissolved oxygen, salinity, vertically integrated net primary production, sea ice extent, and surface advection. Habitat suitability per cell was derived from environment, bathymetry, and habitat features. Carrying capacity scaled with suitability and primary production. Population dynamics used logistic growth with fishing mortality and movement: larval dispersal via advection-diffusion (based on currents and pelagic larval duration), adult movement along density gradients modulated by species mobility. Biomass derived from abundance and size; growth responses to temperature and oxygen were included via a generalized von Bertalanffy submodel. A 100-year spin-up (1951–2000 climatology) established equilibrium. Ocean acidification effects were not included due to uncertainty. Scenarios: Climate followed SSP1-2.6 (strong mitigation) and SSP5-8.5 (high emissions) from 1950–2100. Analyses expressed outcomes versus global mean atmospheric warming levels to reduce model structural variability. Fishing scenarios assumed constant F/FMSY across time: 0.5, 0.75, 1.0, 1.5 (under-, conservative, MSY, and over-exploitation). MPA scenarios designated no-take coverage of 5%, 15%, and 30% of FAO Area 27. The 5% scenario reflects current MPA coverage mapped to the model grid (~5.5% due to resolution). Larger coverages were created by randomly adding protected grid cells such that 5% and 15% are nested subsets of 30%. Mean patch sizes were 1246±1056 km² (5%), 1555±1364 km² (15%), and 1646±1510 km² (30%). Fishing displacement: Protected grid cells had F=0. Immediate neighboring cells were labeled “surrounding” and received proportionally increased F to simulate displaced effort: prop = 1 + 1/(number of surrounding cells). Surrounding-cell fishing mortality was multiplied by prop; unprotected, non-surrounding cells retained baseline F. Statistical analysis: Relationships between biomass (and catches) and warming were evaluated at stock and community levels. Linear mixed-effects models (lme4 in R) assessed contributions of fishing (factor F/FMSY), MPA coverage (factor; effects measured relative to 5% coverage), and global warming level (continuous GWL) to biomass and catch potential. For biomass, four model structures were compared by AIC; Model 4 (Biomass ~ factor(F/FMSY)GWL + factor(MPA) + GWL + (1|stock)) had the lowest AIC. For catch, models included an over-exploited categorical term (F/FMSY>1) interacting with MPA; Model 5 (Catch ~ factor(F/FMSY)*GWL + factor(MPA)*factor(over-exploited) + GWL + (1|stock)) minimized AIC. Analyses were also repeated substituting MPA coverage with the proportion of each stock’s geographic range falling within MPAs (continuous). Sensitivity analysis: Three alternative random MPA allocations were generated for each coverage under SSP5-8.5 with F/FMSY=1 for 10 representative species; an additional mixed model tested the effect of MPA locations (Biomass ~ factor(MPA) + GWL + factor(locations) + (1|stock)).

• Biomass-warming relationship: Total biomass across all stocks declines by approximately −8% per °C of atmospheric warming at F/FMSY=1; under overfishing (F/FMSY=1.5), the rate is ~−4% per °C. At the stock level, 84% of stocks show significant biomass declines with warming; 9% show significant increases. Median stock-level decline is −5.4% per °C relative to biomass at MSY.
• Effects of MPA coverage: Larger no-take MPA coverage increases biomass. Relative to a 5% MPA baseline, 15% coverage increases biomass by ~2.5%; 30% coverage roughly doubles the positive effect relative to 15%. At the stock level, biomass increases by ~3.4% for each additional 10% of the stock’s geographic range protected.
• Range-based protection vs area-based: Protecting 30% of a stock’s range increases biomass by 10.2±0.6%, compared to 5.4±0.4% when 30% of the region is protected without regard to stock biogeography.
• Fishing reductions: Reducing fishing by 25% from MSY (F/FMSY=0.75) increases biomass by ~29–30% on average. However, warming reduces biomass by 12.6±0.7% per °C at that fishing level.
• Offsetting climate impacts: Under anticipated 21st-century warming of 2.6–2.9 °C, warming would reduce biomass by ~31–39%. Combining F/FMSY=0.75 (≈+29%) with 30% MPAs (≈+5.4% over 5% baseline) can nearly compensate for projected climate-driven biomass declines.
• Catch potential: For over-exploited stocks (F/FMSY=1.5), 15% and 30% MPAs increase catch potential by ~6% and ~10%, respectively, via spillover. Warming reduces catch potential by ~6–8% per °C (≈15–25% under 2.6–2.9 °C). For sustainably managed stocks (F/FMSY=1), MPAs reduce catch potential by ~3–5% due to reduced fished area, despite biomass gains. Catch is more sensitive to climate when biomass is higher (lower F, larger MPA).
• Drivers: Warming elevates sea surface/bottom temperatures and reduces dissolved oxygen and net primary production, aligning with the modeled biomass declines. Primary production responses bifurcate between SSP pathways around ~2 °C warming.
• Sensitivity: Alternative random MPA locations did not significantly alter biomass outcomes (p>0.05).

The results support all three hypotheses. No-take MPAs increase biomass of over-exploited stocks under climate change, larger coverage combined with conservative fishing targets can rebuild biomass to management goals while offsetting warming impacts, and MPAs can enhance catches of over-exploited stocks via spillover. Mechanistically, climate-driven warming and deoxygenation constrain metabolic scope, while declining primary production reduces energy input, jointly depressing biomass and catch potential. MPAs serve as refugia (“fish banks”) that maintain higher biomass and facilitate spillover to adjacent fisheries, helping adaptation to climate impacts. However, trade-offs emerge: for sustainably managed stocks, MPAs may reduce catch due to smaller fished areas; local access, livelihoods, and cultural dimensions must be considered in MPA design. Evidence of a “protection paradox” appears: as stocks rebuild (lower F), density-dependent processes and possibly greater climate sensitivity of larger-bodied individuals can reduce marginal fishery benefits and increase climate sensitivity of the stock. Governance effectiveness, equitable stakeholder engagement, and integration with broader management measures are critical for realizing MPA benefits under climate change.

Achieving international biodiversity targets (e.g., conserving 30% of ocean areas) with effective no-take MPAs, alongside reducing fishing intensity, can rebuild over-exploited fish biomass and provide fisheries benefits in the Northeast Atlantic under projected 21st-century warming. No-take MPAs should be integrated into climate-resilient rebuilding plans to meet conservation and seafood production objectives, while addressing social and governance considerations. The study indicates that strategic protection of a substantial portion of stock ranges, paired with conservative fishing mortality, can nearly offset anticipated climate-driven biomass declines.

• Modeling scope: DBEM does not include interspecific trophic interactions, evolutionary adaptation, or explicit socio-economic dynamics; acidification effects were omitted due to uncertainty.
• MPA design: MPAs were randomly allocated and assumed to be fully effective no-take zones with perfect compliance; actual MPAs often allow some fishing and exhibit variable governance and enforcement.
• Effort displacement: Fishing effort redistributed proportionally to immediately surrounding cells; real-world patterns may be heterogeneous or non-redistributed, affecting outcomes and costs.
• Scenario set: Only two climate pathways (SSP1-2.6, SSP5-8.5), fixed fishing levels, and three MPA coverages were explored; results may vary under alternative pathways or adaptive management.
• Single ESM forcing: Projections used GFDL-ESM4; multi-model ensembles could better capture structural climate uncertainties.
• Extreme events and range shifts: Effects of marine heatwaves, detailed network connectivity, and explicit analysis of species’ shifting ranges interacting with MPA placement were not fully evaluated.
• Data uncertainties: Sea Around Us catch reconstructions carry spatial uncertainties affecting stock identification (though relative change projections are less sensitive).