Ecosystem-based fisheries management forestalls climate-driven collapse

The study addresses how fisheries management can adapt to climate-driven changes in marine ecosystems, focusing on whether ecosystem-based fisheries management (EBFM) confers resilience relative to non-EBFM approaches. Marine systems are experiencing rapid warming, sea-ice loss, and biophysical shifts, with significant implications for fish populations and dependent economies. The eastern Bering Sea (EBS) supports major US fisheries (walleye pollock, Pacific cod) managed under EBFM, including a system-wide 2 million ton annual groundfish catch cap (2 MT cap). Recent unprecedented warming and marine heatwaves have already driven species redistributions and recruitment declines, raising concerns that current policies—largely climate-naive—may be inadequate under sustained change. The research evaluates how existing EBFM policies perform under projected climate scenarios, whether they reduce risk of declines, and identifies thermal tipping points for fishery collapse.

The paper synthesizes evidence that marine ecosystems are reorganizing under climate change, with broad documentation of biological responses and growing calls for climate-adaptive management. Prior assessments advocate ecosystem management to enhance resilience, yet few studies explicitly test EBFM performance under climate change. Earlier EBS simulations often focused on single-species or did not include explicit climate effects on population dynamics, limiting their utility for assessing policy performance. Conflicting projections exist regarding potential yield changes under climate change, with some broader analyses suggesting possible gains, while species- and region-specific studies for the Bering Sea often project declines. The study builds on integrated modeling frameworks and management strategy evaluation (MSE) approaches to assess ecosystem-informed harvest policies under multiple climate trajectories.

The authors implement a management strategy evaluation (MSE) within the Alaska Climate Change Integrated Modeling (ACLIM) framework, coupling climate, ecosystem, multispecies assessment, and socioeconomic allocation models.

• Climate forcing and downscaling: Three CMIP5 global climate models (GFDL-ESM2M, CESM1, MIROC-ESM) under RCP 4.5 and RCP 8.5 drive high-resolution regional projections via the Bering 10K ROMS coupled with the BESTNPZ lower-trophic model (ROMSNPZ). Projections span 2006–2099; for CESM RCP 4.5, 2080–2100 conditions are held at 2080. Survey-replicated indices (summer bottom temperatures, cold-pool extent, seasonal zooplankton abundances) are extracted and bias-corrected to 2006–2017 hindcasts using a delta method with variance scaling.
• Multispecies assessment (CEATTLE): A climate-enhanced, statistical age-structured multispecies model is fit (1979–2017) for walleye pollock, Pacific cod, and arrowtooth flounder. Temperature influences weight-at-age and predation via a bioenergetics submodel; recruitment uses climate-enhanced spawner-recruit functions with covariates (spring/fall large zooplankton, bottom temperature, cold pool). Parameters are estimated by penalized maximum likelihood. Future projections (2018–2100) are driven by ROMSNPZ indices affecting growth, predation, and recruitment. Uncertainty is represented via process error in recruitment (random draws around fitted relationships) and scenario uncertainty across GCMs and RCPs.
• Harvest policy scenarios: Three scenarios are evaluated: (1) no fishing (F=0), (2) single-species harvest with a Tier-3 sloping harvest control rule (HCR) producing Acceptable Biological Catch (ABC) without a cap, and (3) as in (2) but constrained by the EBFM system-wide 2 million ton cap on total groundfish TAC across species (the 2 MT cap) applied via the ATTACH model. The HCR uses climate-naive B0 and B40% reference points with an ecosystem cutoff at B20%; Ftarget is derived iteratively to produce 40% B0 in 2095–2099 for pollock and cod, and historical average F for arrowtooth.
• Socioeconomic allocation and catch (ATTACH): TAC and realized catch are estimated from ABC using an ensemble of log-linear and SUR regressions calibrated to 1992–2017 policy regimes, with an explicit constraint that the sum of TACs does not exceed 2 MT. Catch is removed from populations annually and simulations advance one year.
• Risk and tipping point analyses: Performance is summarized as relative changes in catch and biomass versus a persistence baseline (2006–2017 climate held constant). Risks of decline (>10%), severe decline (>50%), and collapse (>80%) are computed over moving periods. Thermal tipping points are identified by relating proportional change in catch to bottom temperature using GAMs with bootstrapped confidence intervals; inflection points are where the second derivative differs most from zero. Replicates per scenario: 100 for no-fishing and no-cap, 30 for 2 MT cap.
• Error structure: The MSE includes process error in recruitment and implementation via ATTACH but does not include observation or estimation error; CEATTLE serves as both operating and estimation model given high historical precision. Non-focal species ABCs are held at historical averages in ATTACH.

• Environmental change: Summer bottom temperatures in the EBS increase under all scenarios, with RCP 8.5 yielding +2 to +4.5 °C by 2100; RCP 4.5 shows +1 to +2.5 °C, with variability across GCMs.
• Unfished spawning biomass (relative to a persistence baseline): By 2075–2100, pollock declines on average 47% (RCP 4.5) and 70% (RCP 8.5); Pacific cod declines 23% (RCP 4.5) and 41% (RCP 8.5); arrowtooth flounder increases 7% (RCP 4.5) and declines 6% (RCP 8.5). Under RCP 8.5, >33% of simulations show >90% declines in pollock unfished SSB by century’s end.
• EBFM 2 MT cap effects: The cap stabilizes pollock catch up to mid-century and reduces climate-driven declines in pollock biomass and catch across scenarios; benefits diminish after mid-century with strong warming. The cap has little effect on Pacific cod (managed near target F), but reduces exploitation at high biomass for pollock and arrowtooth, conferring notable benefits to arrowtooth catch, especially under RCP 8.5. In fastest-warming trajectories, collapse can occur before 2050 even with the cap.
• Risk of fishery collapse: Under RCP 8.5, by 2075–2100 pollock catch collapses (>80% decline vs persistence) in >70% of simulations and exhibits severe declines (>50%) in ~90%. Pacific cod catch collapses in roughly one-third of simulations and shows severe declines in ~65%. The 2 MT cap reduces near-term risk of declines for pollock and arrowtooth, with marginal effects for cod; risk reductions wane over time and are negligible after ~2080 in warm scenarios.
• Thermal tipping point: A summer bottom temperature of approximately 2.1–2.3 °C is identified as a tipping point where pollock and cod catch shift from stable/increasing to rapidly declining. Multiyear warm stanzas (>5 consecutive years above 2.1 °C) become common after ~2033 under RCP 8.5 (and in two GCMs under RCP 4.5), rare historically.
• Management implications: EBFM (via the 2 MT cap and reduced exploitation at high biomass) provides near-term buffering and stability but cannot fully offset large climate-driven declines anticipated after mid-century under high emissions. Stabilizing policies may increase end-century hyperstability risk (rapid catch collapse following stable catches), underscoring the need for biomass monitoring.

The findings demonstrate that current EBFM policies in the EBS, notably the 2 MT cap and associated reductions in exploitation at high stock sizes, can delay and reduce climate-driven declines in pollock and stabilize arrowtooth catches in the near term. However, species-specific climate sensitivity and projected warming beyond mid-century limit the adaptive capacity of existing measures, especially under RCP 8.5. The identified 2.1–2.3 °C bottom temperature threshold provides a practical, climate-informed indicator for anticipating fishery performance changes and could be integrated with forecasting to guide adaptive responses. The 2 MT cap’s benefits are asymmetrical: it confers little advantage for Pacific cod, reflecting historical allocation priorities and management near target F, while strongly aiding pollock and arrowtooth under moderate warming. As warming intensifies, risks of severe declines and collapse rise sharply, and EBFM’s stabilizing effect may inadvertently mask underlying biomass erosion, increasing the risk of abrupt catch collapse (hyperstability). Effective climate-resilient management will likely require a portfolio approach blending fixed caps with adaptive, climate-informed single-species rules, embracing variability to maintain safe operating spaces, and enhancing early warning through fishery-independent biomass indicators.

EBFM in the eastern Bering Sea provides measurable near-term resilience to climate change by stabilizing catches and reducing exploitation at high stock sizes, particularly for pollock and arrowtooth flounder. Nonetheless, under high emissions, projected warming pushes the system beyond the capacity of current policies after mid-century, with high probabilities of severe declines and collapse by century’s end for pollock and significant risk for Pacific cod. A thermal tipping point near 2.1–2.3 °C offers an actionable benchmark for management. Future work should evaluate adaptive or climate-informed alternatives to static ecosystem caps, incorporate more species and food-web interactions, test dynamic reference points (e.g., climate-informed B0 and B40%), and conduct full MSEs that include observation and estimation error. Broader climate mitigation (e.g., achieving pathways akin to RCP 4.5) materially improves outcomes relative to RCP 8.5, highlighting the dependence of fishery resilience on global emissions trajectories.

• Scope of species: Only three focal groundfish species were explicitly modeled; other species under the 2 MT cap were held at historical average ABCs, potentially omitting important interactions and allocation shifts.
• Error representation: The MSE included process error in recruitment and scenario uncertainty but did not include observation or estimation error; CEATTLE served as both operating and estimation model.
• Socioeconomic assumptions: ATTACH assumes current policies and priorities persist; TAC allocation is simulated to meet the 2 MT cap without optimization and may not reflect future adaptive policies.
• Climate inputs and bias correction: Projections rely on three CMIP5 GCMs with downscaling and bias correction; CESM RCP 4.5 lacks variable forcing after 2080 (held constant at 2080), and calibration strategies introduce uncertainty.
• Management design: The evaluated EBFM policies were not designed specifically for climate adaptation; benefits may differ under explicitly climate-informed rules. Stabilizing policies can mask biomass declines, raising hyperstability risks.
• Generalizability: Results pertain to the EBS context and management institutions; transferability to other systems requires caution.