Global fishery prospects under contrasting management regimes

We estimate that global $MSY$ is 98 million metric tons (MMT) (after scaling up to account for data gaps), which is substantially larger than the 80 MMT reportedly caught across the globe in recent years (12), but is consistent with MSY calculations in ref. 6 (83–100 MMT) and ref. 5 (95 MMT). Our estimate reduces $MSY$ of forage fish and assumes no unreported historical catch. Incorporating a recent global estimate of illegal fishing (23%) (13), our estimate of $MSY$ would rise to 121 MMT. Our estimates of each fishery’s current $B/B_{MSY}$ and $F/F_{MSY}$ are plotted in Fig. 1 for all global fisheries (Fig. 1A) and for three illustrative oceanic regions (Figs. 1 B–D). The Northeast Pacific is known to be well managed—our estimates accord with this (Fig. 1B). Many fisheries in the Northeast Atlantic are in poor condition, although 56% are on a path toward sustainability, as fishing pressure is below $F_{MSY}$ (Fig. 1C). As a final illustration, we estimate very low biomass and continued high fishing mortality for the bulk of fisheries in the Western Central Pacific (Fig. 1D). The worse a fishery’s current status, the larger the potential gains from reform. Overall, we estimate the global median fishing mortality is $F/F_{MSY}=1.5$ (overfishing is occurring) and biomass is $B/B_{MSY}=0.78$ (stocks are overfished); these are consistent with refs. 4, 5, and others.

Taking estimated current fishery status as a starting point, for each of the three future policies (BAU, $F_{MSY}$, RBFM), we consider two scenarios. The first scenario applies the policy only to stocks of “conservation concern” (i.e., the 77% of stocks for which we estimate $B/B_{MSY}<1$ and/or $F/F_{MSY}>1$, which roughly corresponds to the FAO definition of “fully exploited” or “over-exploited”). Under that scenario, stocks not of conservation concern are assumed to maintain current biomass forever. The second scenario applies the policy to all stocks. Because the conservation concern scenario provides a less optimistic estimate of global benefits of reform, we adopt it as the default. There, we find that simultaneous gains for catch, profit, and biomass (RBFM relative to BAU) is a likely outcome for the majority of stocks (56%; Fig. 1), and countries (23 of the top 30 in harvest) of the world.

Focusing on the 10 countries with the greatest potential absolute increase in fishery profits relative to BAU also shows substantial gains in both conservation and catch (Fig. 2). China’s immense fisheries show enormous potential (Fig. 2A); other prominent fishing nations in Asia, such as Indonesia, India, Japan, and Philippines, also secure large reform benefits along all three dimensions (Fig. 2B). Seven of these 10 countries derive >50% of profit increases from the NEI species groups (red, Fig. 2).

Although results thus far suggest that nearly every country in the world stands to gain from fishery recovery regardless of its objectives, some important distinctions emerge across recovery policies. Although all reasonable recovery policies are expected to give rise to increases in biomass of fish, albeit to different levels, we expect the RBFM class of policies, which focuses on economics returns, to achieve the highest levels of biomass (14). The $F_{MSY}$ policy will have lower fish biomass, lower profits, and slower recovery times, but upon complete recovery, will ultimately generate the largest catch. The aggregated catch, profit, and biomass under each policy scenario illustrates that trade-offs across policies are small for the aggregated global fishery (Fig. 3).

Applying the stated policy only to stocks of conservation concern, the RBFM policy gives rise to annual increases of 2 MMT in catch, $31 billion in fisheries profit, and 388 MMT in biomass compared with current levels (Fig. 3). These values rise substantially (to 16 MMT, $53 billion, and 619 MMT) when all fisheries, not just those of current conservation concern, are managed more effectively and when comparing to BAU (Fig. 3); this accounts for fishing more aggressively on currently underexploited stocks. Each of these values could reasonably be adjusted upward by an additional 28% because our database covers only the 78% of reported catch with adequate data. The largest gains accrue from fisheries with the most depleted status, although these gains may be limited if depleted fisheries have crossed tipping points beyond which recovery may not be possible (15). Even targeting reforms at just fisheries of conservation concern, global fishery profits are 29% higher under RBFM than under $F_{MSY}$ (Fig. 3); this wedge grows to 64% when applying policies to all fisheries (Fig. 3). This increase in profits under RBFM relative to $F_{MSY}$ has two components: an “optimization effect” (68% from optimizing the harvest policy) and a “pecuniary effect” (32% from price increases and cost decreases). Although we have not explicitly modeled effects of fishery reform on consumers, they are likely to benefit from the catch increases (and price decreases) that arise from fishery recovery. Consumers may also benefit from higher quality product under RBFM, albeit at a commensurately higher price.

Although the most suitable institutional reforms to achieve recovery will depend on social, economic, and ecological objectives and conditions, various approaches such as cooperatives (16, 17), territorial rights (18, 19), or individual transferable quotas (8, 20) could be used to improve economic results under a range of harvest policies. Although these all fall under the umbrella of RBFM, each will bring different benefits in different settings that must be weighed against the costs of reform. Although these costs have not been explicitly modeled here, experience from countries such as Iceland, New Zealand, and Australia suggest that they are likely to be only a fraction of the potential benefits identified here (21). Our results suggest that some of the greatest economic improvements in fisheries may come more from improving institutions than from improving the status of fished stocks. Furthermore, these gains in profit can occur quickly following institutional reforms, because they do not exclusively rely on stock recovery. Such rapid economic gains can help offset many of the necessary short-term costs associated with stock recovery when catches must temporarily decline to enable recovery.

To ensure model tractability and to apply it at a global scale, we made a number of simplifying assumptions. SI Appendix contains an extensive description and set of robustness checks and sensitivity analyses; we note a few here. First, because our entire analysis is built on estimates of the current fishery status, it is natural to ask how sensitive our results are to these uncertain estimates. We performed numerous routines to estimate our model’s ability to predict out of sample, broken out by region and fishery size. Results suggest that our methods for estimating $B/B_{MSY}$ and $F/F_{MSY}$ are more robust for fisheries in the developed world that are not extremely overfished. The presence of low-to-moderate unreported fishing does not bias these estimates (although it will bias $MSY$ estimates downward). Second, predictions of absolute economic effects rely on estimates of economic parameters (demand and costs). To this end, we built a price database derived from export values and other data. We transformed the data to ex-vessel equivalents and modeled demand with a constant elasticity of −1.15, consistent with refs. 22 and 23. We model costs by identifying the unassessed fisheries estimated to be in bionomic equilibrium and backing out the cost that drives profit to zero under open-access equilibrium. This procedure results in costs per MT that are consistent with those in ref. 24 and generates mean cost/revenue ratios of 67% in the current fishery. We conducted numerous sensitivity analyses examining the effects of a range of these biological and economic parameters; although these affect our quantitative predictions, our qualitative findings are largely unaffected. We also performed a historical analysis where catches for all assessed stocks from 1980 to 2012 were predicted using the model with data up to 1980 only. Actual catch compares extremely well with the model’s prediction (correlation, 0.99; value of P < 1%).

Our model allows us to make novel predictions of the timing of fishery status into the future under alternative management approaches (Fig. 4). If the BAU policy is applied to all fisheries, the proportion below a recovery target of 0.8$B_{MSY}$ (see ref. 25 and SI Appendix) rises from 53% today to 88% in 2050 and the proportion experiencing fishing pressure above $F_{MSY}$ rises from 64% to 84%. These values are consistent with those of Quaas et al. (26), who estimate declines under BAU for all stocks studied, with an estimated biological decline of about 77%. We find that, if reform efforts are put in place now, the median time to recovery would be just 10 y, and by midcentury, the vast majority (98%) of stocks could be biologically healthy and in a strong position to supply the food and livelihoods on which the world will increasingly rely.

We developed a novel approach combining several sources of data and models to conduct fishery-level analyses. Our database of 4,713 fisheries is drawn from both the RAM Legacy Stock Assessment database (27) and the FAO marine capture databases (species–country–FAO region triples) (28). Excluding fisheries that failed to meet minimum criteria (SI Appendix), this process accounts for 78% of reported global catch. Because data availability varies across these fisheries and by region, we used surplus production models consistently across all fisheries to estimate MSY-related parameters and predict future trajectories of biomass.

We extracted biomass and fishing mortality directly from stock assessments for the 397 fisheries included from the RAM database (43% of catch in our database). Status of the remaining 4,316 “unassessed” fisheries are estimated using a novel two-step process involving global regression analysis and a structural fisheries modeling approach (merging methods in refs. 4 and 10; SI Appendix), if available, or extracted from FAO State of World Fisheries and Aquaculture database (12). The structural modeling approach, catch-MSY (10), assumes stationarity of $MSY$, $F_{MSY}$, and $B_{MSY}$ parameters; see SI Appendix for further details including sensitivity analyses. The set of fisheries categorized as NEI by the FAO were handled with a two-step process using an appropriate comparison set of fisheries from the global database and imputing values to the NEI stock (SI Appendix); this could be improved with better reporting of catches of these species. Although trophic effects are not explicitly modeled, they may play a role in the potential benefits of rebuilding fisheries. In reality, each stock we have modeled exists within a larger ecosystem context that influences the dynamics of each species within it (e.g., ref. 15). Recovering predatory species increases predation on prey species. Consequently, reform projections from single-species models may be optimistic for ecosystems in which many piscivorous species are recovered. To partially account for this, we conservatively reduced MSY of all forage fish stocks, which provide a food source for many higher–trophic-level fish and mammals, by 25% (29).

We assessed the future trajectories of all individual fisheries up to midcentury (2050) using a Pella–Tomlinson surplus production model (30) with fishery-specific life-history parameters, described as follows:

where $\left(\left(\varphi +1\right)/\varphi \right)g$ is the intrinsic rate of the growth for the species, K is the carrying capacity, and $H_t$ is the harvest in year t. We assume stationarity in g and K parameters. We set the growth curve parameter ϕ at 0.188, per refs. 31 and 32, such that $B_{MSY}$ occurs at 40% of K; alternatives are analyzed in SI Appendix. Estimates of $MSY$ and g were also derived from the RAM database for assessed fisheries. In cases where only some of these values are provided, we used available data to calculate missing values (e.g., estimating $MSY$ from $Catch$, $B/B_{MSY}$ and $F/F_{MSY}$, and g as $MSY/B_{MSY}$) or performed a two-step process described above.

For each fishery, we modeled a range of fishing policies, each of which assigned a scaled fishing mortality rate tailored to that specific stock (i.e., a “control rule”) for all possible realizations of biomass. Profit in a period is revenue (price times catch) minus the cost of fishing, which is an increasing function of the fishing mortality applied. Prices are adjusted each year from a global seafood demand curve with constant elasticity of −1.15. Our base case scenario assumed zero discounting (for which profit is optimized dynamically); higher discount rate scenarios are reported in SI Appendix. This allowed us to predict the annual catch, profit, and biomass of fish in the ocean across a range of alternative harvest policies, as reported above.

We are grateful to National Center for Ecological Analysis and Synthesis for computation support. We acknowledge financial support from the David and Lucile Packard Foundation, Waitt Foundation, and Helmsley Charitable Trust.