Multispecies Portfolios of U.S. Marine Fisheries: Ecosystem‐Based Fisheries Management Reduces Economic Risk

Abstract

INTRODUCTION

Traditional approaches to fisheries management have inherent uncertainties that undermine social, economic, and ecological outcomes, which will only get worse as climate continues to change and other ocean uses continue to perturb marine ecosystems (c.f., Link 2010, 2018; Barange et al. 2018; Free et al. 2019; Plagányi 2019). Fisheries management often focuses on single species or populations. For example, in classical fisheries management, management is based on individual stock dynamics with limited or no consideration of the entire fishery system (Browman and Stergiou 2004). Although this kind of approach has yielded an enormous body of scientific research and practical application (Lynch et al. 2018), and has resulted in positive outcomes (Methot et al. 2014; Hilborn et al. 2015, 2020; Melnychuk et al. 2021), it can be risky due to uncertainties in our understanding of the drivers of individual population dynamics, multispecies interactions, and our limited ability to accurately predict future environmental conditions (Link 2010; Skern‐Mauritzen et al. 2016; Marshall et al. 2019). More so, the risks extend into economic, social, and even governance considerations (Link 2018). In the present context the main risks are presented as loss in revenue, but there are also other risks to consider (e.g., overfishing, fishery efficiency, market shifts, catchability, climate change impacts, socio‐cultural impacts), which we do not address herein.

The various possible mixes of a portfolio of fishes caught in a fishery can provide a range of revenues, risks and efficiencies. Image credit: Jason Link. Photos courtesy of Shutterstock.

Fishery managers are tasked with making numerous decisions about living marine resources, including harvest rates, biomass targets, and the spatial distribution of protections. The management of fisheries has generally focused on biological features, factors, and parameters. The economic facets of fisheries management, though important parts of the discussion, often are resultant outcomes but are not typically management targets (c.f., Methot et al. 2014; Pascoe et al. 2023). This reality translates into missed opportunities, or remaining challenges, for the economic productivity associated with fisheries, which are often performing suboptimally. Moreso, fishery management is largely uncoordinated, at least with respect to all the fish species and fisheries in a given ecosystem. There is a growing body of evidence that classical approaches to fishery management can yield suboptimal outcomes, especially as environmental conditions continue to change (Fogarty 2014; Lynch et al. 2018). There is also evidence that these classical, single‐species management approaches, which remain uncoordinated across all the fisheries in a region, can result in notable foregone yield (Fogarty 2014; Link 2018; Ye and Link 2023). Moreso, it has been noted that to meet all the legal mandates for all managed fisheries, an ecosystem approach using what is documented herein is not only allowable, but advisable (Murawski 1991; Link 2010). All these factors result in fisheries being prosecuted and managed with more risk than they need to be (Edwards et al. 2004; Sanchirico et al. 2008; Jin et al. 2016; Link 2018; Carmona et al. 2020).

To mitigate these risks and consider economic performance of fisheries more directly, portfolio approaches and theory have been explored in the context of multispecies marine fisheries (e.g., Edwards et al. 2004; Sanchirico et al. 2008; Rădulescu et al. 2010; DuFour et al. 2015; Schindler et al. 2015; Jin et al. 2016; Carmona et al. 2020; Lopetegui and del Valle, 2022). This approach represents a systemic treatment of multiple stocks in a fishery ecosystem and focuses on the aggregate dynamics of a group of management units (e.g., many species or populations). Such approaches (i.e., portfolio optimization across multiple stocks) are now used everywhere in financial portfolio decision making, especially in the US$9 trillion of defined‐benefit pension plans in the United States and $7 trillion in the defined‐contribution retirement funds. Overall, this approach is accepted as the gold standard for portfolio construction for much of another $26 trillion of corporate equities and mutual‐fund shares held by investors outside of retirement plans (Markowitz 1952; Sexauer and Siegel 2024). As with a financial stock portfolio (Markowitz 1952; Roy 1952), the emergent properties of a diverse portfolio of target species will be more stable than any one stock on its own (Doak et al. 1998; Sanchirico et al. 2008; Brown et al. 2016; Link 2018). When examined theoretically, empirically, experimentally, or via simulations, portfolio management consistently produces better outcomes, including increasing the status and value of the resource, reducing risk, and improving buy‐in from stakeholders and resource users (Edwards et al. 2004; Sanchirico et al. 2008; Schindler et al. 2015; Jin et al. 2016; Link 2018). Additionally, there are recognized benefits regarding the status of stocks, economic valuation, and business, stock and regulatory stability, all within current legal mandates. Furthermore, when adopted, this approach has headed off other challenges and mitigated other pressures that were not always foreseen, such as not overlooking the potential value of lower abundance stocks in the context of a larger group of species (Schindler et al. 2015; Link 2018). These fisheries portfolio studies have entailed looking at an entire group of species as a set of stocks within a portfolio and seeing how they collectively perform relative to economic returns and economic risk (Edwards et al. 2004; Sanchirico et al. 2008; Rădulescu et al. 2010; Jin et al. 2016). This is distinct from biological risk and yield, though it is related, and provides a unique look at how well fisheries perform.

Key metrics from such portfolio approaches include the portfolio frontier—the envelope of what is theoretically (optimally) possible—and the frontier “risk gap”—the difference between any given, realized value and the frontier—as well as estimates of risk and valuation of an entire suite of targeted stocks in a group of fisheries (Edwards et al. 2004; Sanchirico et al. 2008; Rădulescu et al. 2010; Jin et al. 2016; Carmona et al. 2020). These are rarely if ever reported upon to gauge fisheries performance in an operational fisheries context. Several works have explored fisheries portfolios (Edwards et al. 2004; Sanchirico et al. 2008; Rădulescu et al. 2010; DuFour et al. 2015; Schindler et al. 2015; Jin et al. 2016; Carmona et al. 2020; Lopetegui and del Valle 2022); the collective outcomes of these analyses are generally that portfolios demonstrate that there are many economic inefficiencies and less realized value for fisheries than there could be (Edwards et al. 2004; Sanchirico et al. 2008; Rădulescu et al. 2010; Jin et al. 2016; Carmona et al. 2020). That is, theoretical studies demonstrate that the further away from the “efficient frontier” that a set of aggregated landings is, the more risk is incurred for a given level of revenue (Figure 1; Rădulescu et al. 2010; Jin et al. 2016; Carmona et al. 2020). The risk gap (risk difference between point a’ and b; or a; Figure 1) can be used as an ecosystem‐level indicator to inform managers of the unnecessary risk being assumed for a desired level of revenue from a portfolio of fisheries stocks.

Figure 1.

A schematic of a portfolio efficiency frontier, as modified from Jin et al. (2016). It demonstrates the portfolio efficient frontier (curved red or blue lines) and the risk gap. R represents a given level of total revenue and SD is the standard deviation associated with a specific revenue; F and F′ are two efficient frontiers (single‐species [SS] and ecosystem‐based fisheries management [EBFM]); b denotes an actual, realized portfolio; a and a′ denote the optimal portfolios on F and F′. The distance between b and a or a′ represents the risk gap.

Open in new tabDownload slide

Thinking of fisheries in a regional ecosystem as a portfolio can help implement ecosystem‐based fisheries management (EBFM). Examining the composite behavior of a suite of fisheries and how they perform as a system (i.e., correlation, which captures interactions) is an essential part of EBFM and at the core of portfolio analysis (Link 2018). Applying portfolio analytic approaches to regional fisheries ecosystems has many potential benefits, again, including increased stability of the portfolio relative to an individual stock, decreased risk at the aggregate level, or conversely increased value for the same risk, and acknowledgement of inefficiencies and trade‐offs in how a management target is actually obtained. It also has the advantage of seeing how close a group of fisheries landings is to a desired outcome relative to risk and value. The additional benefit is recognizing that an ecosystem‐based approach likely results in more desirable outcomes than just the summation of single stock approaches. In short, portfolio analyses can ascertain, and reinforce, the economic benefits of EBFM. Yet despite the growing body of evidence for the value of multispecies portfolio approaches to fisheries management, there are relatively few examples of its application in actual, operational practice and consideration (Link 2018). Exploring portfolio configurations and options would not only help to facilitate the implementation of EBFM (Sanchirico et al. 2008; Link 2010; Jin et al. 2016; NMFS 2016; Link 2018; Carmona et al. 2020), it would also demonstrate the power of such an approach to meet the regulatory mandates of maintaining stock status (Lynch et al. 2018) and highlight the value of realized benefits sought for expanding the blue economy generally and seafood in particular (Bennett et al. 2019).

Most EBFM and fisheries management efforts in general have understandably focused on the biological status of a targeted group of stocks (Link 2010; Lynch et al. 2018). Most portfolio analyses have been a bit more theoretical, with few applied to actual fisheries data and fewer still presented with real‐time, inflation‐adjusted current data, and rarely as compared across ecosystems (Edwards et al. 2004; Sanchirico et al. 2008; Rădulescu et al. 2010; Jin et al. 2016; Carmona et al. 2020). Here, we aim to estimate and evaluate fisheries portfolios for commercial fisheries in all the major Fishery Management Council regions in the USA (https://bit.ly/4bsUGNl), providing aggregated (EBFM) and single‐species portfolio frontiers, values over time, and risk gaps. Our hypotheses, consistent with portfolio theory, are that the EBFM frontier will necessarily outperform single‐species applications due to the consideration of system interactions, that there are gaps between actual (or realized) revenue and risk (mean standard deviation) values compared to either the EBFM or single‐species efficiency frontiers, and that even in well‐managed systems the performance gap may be larger than is intended relative to fisheries management policies.

METHODS

Data Sources and Portfolio Selection

Data for the portfolio analyses of each region were obtained from the publicly available National Marine Fisheries Service Landings database (https://bit.ly/3zsxce9). These data provide coverage across geography, taxonomy, and fishing sector, reporting on value, amount (by mass), and related information. Here, we primarily emphasize both commercial landings and value over the full time series, by landed taxa, for all major regions in the USA. These publicly available data have been quality‐assurance/quality‐control checked prior to their public release. Data from 1990 to 2020 for each region was used for the analysis. The landings values are reported in pounds, and we retain that here for familiarity and ease of use (1,000 pounds = 0.454 metric tons). The revenue values were deflated to 2020 real dollars to adjust for inflation using the Bureau of Labor Statistics Producer Price Index monthly data for processed and frozen foods, as that is a widely used inflationary adjustment index (https://bit.ly/3W9F39l).

For each region, we included all the taxa in a portfolio, which comprised the top 25 species by landings revenue. These data were reviewed to ensure that there was adequate coverage of the main taxa landed, that the majority of fisheries were represented from a region, and that there was mix of taxa that were not all caught concurrently. Though some species aggregation, removal, truncation, interpolation, or addition of zeroes to missing values were feasible and recommended (Brewster et al. 2023a, 2023b), the data were largely only aggregated in a few instances and mostly were treated as is from the database. These covered approximately the top 50–75% of landings revenue in each region. The Pacific Islands region was not assessed due to questions around representativeness of the data, given it predominantly stemmed from a single port (Honolulu) and was composed of a handful of tuna Thunnus spp. and related highly migratory species. Although not presented here, we explored other candidate portfolios for each of these regions with a variety of ~20–30 species. Though the results from these other possible portfolios did differ in terms of magnitude and scope of the portfolio frontiers and their associated metrics, the main, fundamental features were generally consistent. Further details are explored in related work in more detail (Brewster et al. 2023a, 2023b).

Portfolio Analyses and Risk Gap

To estimate the portfolio of a given suite of taxa for a specific region, we used the methods generally characterized in Edwards et al. (2004), Sanchirico et al. (2008), and Jin et al. (2016), which in effect applies the classic modern portfolio approach (Markowitz 1952) to living marine resources. In short, the approach is summarized as minimizing the variance (i.e., risk) of harvesting a set of stocks to obtain a certain amount of revenue (i.e., value of fisheries landings). The methodology generally follows the approach of Sanchirico et al. (2008) and Jin et al. (2016), a more detailed description is found in Brewster et al. (2023a, 2023b), and we briefly describe the approach here.

Because we applied methods used in finance to fisheries stocks portfolios, adjustments to the value at risk model are necessary to account for ecological and policy constraints and variability of fisheries stocks. Minimum and maximum revenue weights should be set to reasonable levels based on historical patterns in revenues and policy constraints. For example, allowing the minimum revenue weight (w_i,t) of a stock to be 0 would be equivalent to allowing the fishery for that species to be closed. In finance, a buyer can borrow money to buy shares of an asset (stock, bond, etc.) such that revenue weights derived from optimization can exceed historic weights. An analogous increase in revenue weights for harvest fisheries species is unlikely to be sustainable, so a sustainability parameter is used to constrain the maximum revenue weights in the optimization. Finally, external environmental conditions influencing fishery stock production that existed in the past may have changed in the present, thus past revenues in a portfolio should be down‐weighted for the optimization.

To account for these differences between finance and fisheries, we set biological constraints (i.e., minimum and maximum harvest weights to constrain the revenue weights for each species at time t) and included a sustainability parameter (ɣ) to be used in setting the maximum weight for species i at t. We set ɣ = 1, but it could be lowered by fisheries management to control harvest levels. We also included a decay factor (λ) to down‐weight earlier data in the time series. We used λ = 0.741, which results in an observation retaining 5% of its weight after 10 years.

RESULTS

Fisheries Landings Data

The landings for the top 25 taxa by landed value in all regions exhibited three main patterns (Figure 2). First, most regions had a combined landings of the top valued taxa on the order 0.5–5.0 billion pounds, or 0.3–3.0 million metric tons. Second, most regions had one to three dominant taxa for biomass of landings, usually a species of small pelagic fish. And finally, most regions exhibited dynamics in the biomass caught among a suite of individual taxa, but most exhibited stability in the total amount landed, within one‐quarter of an order of magnitude. The exceptions to this latter point were the Mid‐Atlantic and South Atlantic, which exhibited declines of more than half an order of magnitude over the past 15 years. The West Coast exhibited the most interannual variability, reflective of the upwelling dynamics of the ecosystem and dominance by small pelagics, but was again consistent between 0.75 and 1.25 billion pounds for the past 20 years. Similar patterns were seen for revenue of the top 25 taxa in each region (Figure 3). Declines were observed in the Alaskan, Gulf of Mexico, and South Atlantic regions for the past ~15 years, but they have been stable since the early 2000s and the remaining regions have been relatively stable over the past 30 years. Values ranged from $100 million (10⁸) to $3 billion (10⁹) in any given year, with the Alaskan, New England, and Gulf of Mexico regions being the most valuable, in the billions of dollars in most years. However, the contributions were spread more evenly across taxa than the weight of the landings, with multiple fish, invertebrate, and other taxa groups contributing to the overall revenue.

Figure 2.

Landings by weight for the top 25 species by revenue in each region. (A) Alaska, (B) Gulf of Mexico, (C) Mid‐Atlantic, (D) New England, (E) West Coast, (F) South Atlantic. Note, units are in pounds.

Open in new tabDownload slide

Figure 3.

Landings by adjusted revenue for the top 25 species by revenue in each region. (A) Alaska, (B) Gulf of Mexico, (C) Mid‐ Atlantic, (D) New England, (E) West Coast, (F) South Atlantic. All values are in US$, adjusted to 2020 dollars.

Open in new tabDownload slide

Frontiers and Risk Gaps

The portfolio efficiency frontiers for all the major regions (Figures 4 and 5) exhibited patterns as theoretically expected (Figure 1). As shown in Jin et al. (2016), the frontier corresponding to EBFM (i.e., correlations across species are accounted for) has lower risk for a given revenue level than the frontier corresponding to single species management (i.e., where species interactions are not considered). Both the Sanchirico et al. (2008; Figure 4) and Jin et al. (2016; Figure 5) plots demonstrated lower risk for a given target revenue or more revenue for a given risk in the EBFM frontier compared to the single‐species frontier. The difference between the EBFM and single‐species frontiers, though the same response in all regions, was greater in some regions than others. For instance, the difference between the two frontiers was relatively larger in the East Coast regions (Figure 4), whereas it was less distinguishable in Alaska or the West Coast. The yearly distinctions were less pronounced (Figure 5), though the West Coast and mid‐Atlantic did have some notable divergences in some years.

Figure 4.

The single species and ecosystem‐based fisheries management efficiency frontiers for the top 25 of all landed species (by revenue) for all regions, depicted in the Sanchirico et al. (2008) style plots. The actual portfolio values are from the past 15 years (not all 30 years in the time series) and the frontiers are for the terminal year. (A) Alaska, (B) Gulf of Mexico, (C) Mid‐Atlantic, (D) New England, (E) West Coast, (F) South Atlantic.

Open in new tabDownload slide

Figure 5.

The single species and ecosystem‐based fisheries management efficiency frontiers for the top 25 of all landed species (by revenue) for all regions, depicted in the Jin et al. (2016) style plots. The actual portfolio values and frontiers are for each year of the 30‐year time series. (A) Alaska, (B) Gulf of Mexico, (C) Mid‐Atlantic, (D) New England, (E) West Coast, (F) South Atlantic.

Open in new tabDownload slide

In each region for each year, the annual revenue–risks (mean–variance) observed is below the frontier (Figure 5), and at least always below the EBFM frontier in aggregate years (Figure 4). Of note, Alaska had higher expected values than other regions, with values on the order of $1 billion (10⁹), consistent with prior observations (Link and Marshak 2019, 2021) and observed revenues (Figure 3). The risks for all regions were on the magnitude of $10s of millions (10⁷) to $100s of millions (10⁸). New England and the Gulf of Mexico had expected revenues on the order of $1 billion (10⁹), although the values for revenue in those regions were lower than Alaska. All the other regions had expected revenues, and actual annual values, on the order of $100s of millions (10⁸). In the 15‐year composite plots (Figure 4), some actual values exceeded the single‐species frontier but were still below the EBFM frontier. This can be due to several reasons, including how optimal fisheries were executed relative to all the species caught, but largely as any observed value compared to a terminal year value may be relative to a different annual maximum, and the frontiers reflected conditions more heavily weighted towards the end of the time series. Nevertheless, these composite frontier plots show that in some locations (e.g., New England, many years in Alaska, some in the Gulf of Mexico and the West Coast), all of the realized values were much lower than either frontier. In addition, revenues were always less than what could be expected from an EBFM portfolio frontier for the same level of risk. Collectively from these observations, we can state there was additional risk assumed for the observed revenue realized by the portfolio. The analysis also illustrates that, for the same amount of variability, revenue that could have been realized by the fisheries was instead left in the ocean and, hence, was not realized in the fisheries.

Presented in absolute terms, the risk gaps from any given period for any region ranged from the low of $10s of millions (e.g., South Atlantic in recent years) to more than $300 million (New England, mid‐2000s; Figure 6). That is, there was additional risk assumed for the actual level of revenue realized by the portfolio. Most regions exhibited an absolute risk gap on the order of $20–50 million. The Gulf of Mexico, Alaskan, and New England regions had higher absolute risk gaps, associated with higher total revenues (Figure 3), often approaching or exceeding $100s of millions. The past 5 years show an average risk gap of $5 million (South Atlantic) to ~$90 million (Alaska; Table 2). New England, the Gulf of Mexico, and the West Coast have had an average absolute risk gap of $45 million, $36 million, and $31 million, respectively for that time. Collectively, the average absolute risk gap across all six regions has been over $250 million. The cumulative risk gap for the past 5 years has ranged from $25 million (South Atlantic) to nearly $5oo million (Alaska; Table 2). Cumulatively, the absolute risk for the past 5 years has exceeded $1 billion.

Figure 6.

The risk gap (standardized per dollar [gray bar] and absolute [red line] over time for all regions). Dashed gray line represents the average standardized risk gap for all regions over the period. (A) Alaska, (B) Gulf of Mexico, (C) Mid‐Atlantic, (D) New England, (E) West Coast, (F) South Atlantic.

Open in new tabDownload slide

Comparing the risk gap relative to the value of the revenue, the median normalized risk gap (i.e., per dollar of revenue) across all regions for this period was ~0.05 (Figure 6). That is, for each dollar of revenue from a regional portfolio, 5 cents were at risk unnecessarily. Though presenting some variability across time, the magnitudes of the normalized risk gaps are relatively stable for the Alaskan and New England region (Figure 6, with a couple years exception in New England), centered around the median normalized risk gap. The Gulf of Mexico has had an increase in the normalized risk gap over this period, whereas the South Atlantic has declined. The Gulf of Mexico and West Coast have been above the median normalized gap in recent years and have been more variable than the other regions.

DISCUSSION

Portfolio theory predicts that managers can decrease exposure to risk by considering system interactions: a result also consistent with the tenets of EBFM (Markowitz 1952; Roy 1952; Link 2018). The examples shown here for U.S. regions demonstrate that empirical results are consistent with portfolio theory, with patterns that indicate that there is unnecessary risk incurred for the historical fisheries revenue that has been generated from each region studied. Conversely, these examples also show that additional revenue could have been generated for the level of risk fisheries managers are incurring. Every time portfolio frontiers have been examined, the patterns remain consistent (Edwards et al. 2004; Sanchirico et al. 2008; Rădulescu et al. 2010; DuFour et al. 2015; Schindler et al. 2015; Jin et al. 2016; Link 2018; Carmona et al. 2020; Lopetegui and del Valle 2022), indicating that there is clear room for improvement in the economic performance of fisheries, both theoretically and practically. One way to think about it is that holding the target revenue constant and minimizing the variance leads to the result that for the same returns less risk could have been incurred. Alternately, one could have equivalently defined the problem to be the optimization of returns given the constraint on the system variance–covariance, which would be interpreted as increasing returns for the same level of risk. Combining both risk and revenue considerations simultaneously represents the trade‐off space for EBFM and reiterates outcomes from prior fishery portfolio work. These results argue that the economic performance of fisheries in the USA could be improved.

These results also show that taking into account multispecies considerations (i.e., the EBFM approach; Skern‐Mauritzen et al. 2016; Marshall et al. 2019; Link and Marshak 2021; Karp et al. 2023) provides more desirable results than solely focusing on a set of single‐species of stocks. The value of managing for EBFM is that one has better yields and lower risk (Sanchirico et al. 2008; Jin et al. 2016; Link 2018; Carmona et al. 2020), as seen by the categorically higher revenue values of the EBFM curves compared to the single‐species efficiency curves. The single‐species frontiers are categorically also riskier than the EBFM frontier efficiency curves. This can be due to many reasons, but in the portfolio context here is largely related to accounting for the covariance among species that helps lower risk. It is unlikely due to any inherent ecosystem constraints, as those were essentially the same for either set of efficiency frontiers. The approach to EBFM can also lead to increased stability (c.f., Fogarty 2014; Link 2018), implicit in most of our risk gap results. The key point is that it is unrealistic and untenable to continue managing fisheries assuming that they are caught in isolation and do not interact with other species in the ocean and seafood markets.

Our results indicate that in any given year for any given region, the system is not optimized with respect to either single‐species or EBFM portfolios. The single species divergence could be a result of risk policies to avoid overfishing, which are different than maximum economic yield goals (Grafton et al. 2012); or it could be a result of other constraints placed on the fishery, some of which may have to do with operational exigency and efficiency of fleets, particularly of vessels, gear, permits, fuel efficiency, weather, locating fish; or perhaps even as a consequence of the fact that economic yield is not the primary goal typically managed for in U.S. fisheries systems (Methot et al. 2014; Lynch et al. 2018). Certainly, constraints also still exist in an EBFM context, but the benefits of considering stocks as a composite portfolio would not be realized. In both cases, a more nuanced accounting of revenue compared with net revenue might explain part of the risk gap. In any case, the economic performance of fisheries has potential room for improvement. We tend to think of that in terms of increasing economic revenue. But it could also be maintaining economic revenue with a lot less risk, or variability, incurred. Certainly, fisheries management in most U.S. regions tend to be generally performing well relative to biological objectives (Methot et al. 2014; Hilborn et al. 2015, 2020). Though all regions have some overfished stocks (NMFS 2023a, 2023b), the fisheries in the USA are generally recognized as being well managed (Melnychuk et al. 2021). Yet even in these relatively well managed regions, however, there remains room for economic improvement as the risk gaps still are much greater than zero. That would most likely apply to the other regions of the world as well.

The absolute portfolio risk gaps have been highly variable for many regions over time. This is likely indicative of several challenges facing fisheries management that have been exacerbated over the past several decades, which include managing fisheries on a stock‐by‐stock basis and ignoring the broader systematic patterns; changes in fisheries composition due to climate change, distribution shifts, or changes in productivity; chronic overfishing; over‐capitalized fleets; lack of coordination across managed taxa; or other factors (Methot et al. 2014; Link 2018; Lynch et al. 2018). Regardless of the causality of the observed risk gaps, they demonstrate the utility of examining a suite of fisheries for a region at once. At the very least, these portfolio gaps and related portfolio information can be an important diagnostic that illustrates the economic performance of a fishery in a region. The median standardized risk gap (per dollar of revenue) across all regions over the past 30 years was ~0.05. That is, for each dollar of revenue from a regional portfolio, 5 cents were at risk unnecessarily. The range was anywhere from 1 cent to ~20 cents per dollar. While 5 cents or even 10 cents on the dollar may seem small, considering revenues in the range of hundreds of millions to billions of dollars, the risk gap then indicates that hundreds of millions of revenue dollars are potentially being risked unnecessarily.

Some regions are closer to their efficiency frontiers than others. These regions include the Mid‐Atlantic, West Coast, and South Atlantic. To be clear, this does not necessarily translate into each and every one of these stocks or groups of stocks doing well relative to biological reference points or other measures of stock status (Lynch et al. 2018); nor does it imply any statement as to the efficacy of the management choices in each region, and certainly does not provide background of the many considerations that went into those decisions. Obviously, each region faces different factors that have influenced fisheries management decisions over the past decades (Link and Marshak 2021). The spikes and declines seen in a given region over time in the portfolio gaps can be linked back to the landings, and ultimately what management conditions the fisheries in a region were operating under. A full comparison of stock status and portfolio performance would necessitate another thread of research beyond the scope of the present work. These fisheries vary across a wide range in terms of biological diversity and taxonomic composition, habitat structure, ecosystem productivity, number and types of fleets, number of times ecosystem considerations have been considered, use of fishery ecosystem plans, total economic value of their fisheries, and related considerations (Link and Marshak 2021). Yet, we can generally state that fisheries that are performing relatively well in relation to their portfolio frontiers are in regions that tend to be managed in the aggregate, tend to have fewer fisheries management plans, and tend to have less overfished stocks than other regions (Link and Marshak 2019, 2021). However, these are also some of the regions with higher proportions of recreational fisheries that are not accounted for in this analysis. This would also include the Pacific Islands region, which was not included in this study due to data limitations. It is probable that without these other (recreational) landings data, the results presented for regions with large recreational fisheries are conservative and these regions might be more similar to some of the other regions. Nationally it is unsurprising that there are regional differences in these fisheries portfolios, given the distinctions in volume, value, and taxa caught in the different fisheries (Link and Marshak 2021).

The absolute value of the revenue gap across these U.S. regions indicates that collectively there are hundreds of millions of dollars at unnecessary risk in U.S. marine capture fisheries, on average over $250 million each of the past 5 years. If that level of inefficiency occurred in other economic sectors, market forces would be expected to quickly correct any such significant efficiency gaps (Boettke 2010). That has not occurred for marine capture fisheries due to the unique market pressures this industry faces. Fisheries policies and management that have focused on single stocks that ignore species interactions also likely contribute to these inefficiencies. Given the value of fisheries in the USA and globally, and the importance of fisheries for food security, food provisioning and the post‐vessel seafood industry in the country, this result is problematic (Bennett et al. 2019; FAO 2022; NMFS 2023a). We argue it is time for U.S. fisheries managers to monitor these risk and revenue gaps in their performance measures.

Calculating the risk gap for these U.S. fisheries was a reasonably straightforward task. A revenue gap (the vertical difference between a realized portfolio and a frontier in Figure 1) indicator would also be a useful indicator, though the task would be somewhat less straight forward given potential convergence considerations at the upper end of the efficiency frontiers. To use risk (or revenue) gaps as an ecosystem‐level indicator would require additional thought and planning in the selection of which stocks should be considered as part of a regional fisheries portfolio. To begin considering these risk gaps in a more operational context, we are not suggesting that they be used in place of means to establish quotas or similar biological reference points. Rather, they could serve as indicators of fisheries performance to evaluate how well fisheries objectives were obtained, with implied suggestions for altering a portfolio mix and associated management decisions to better optimize a portfolio in the future. We also evaluated other candidate portfolios, which covered approximately the top 50–75% of landings revenue in each region, and a protocol to do so most appropriately for each region would need to be adopted. Further work is warranted to better elucidate these details of each region, as notable data decisions and other considerations are necessary to understand the nuances in each location. In unpublished data, the results differ but generally confirm that EBFM is less risky than single‐species frontiers, that is risk gaps (though of varying magnitudes) tend to be smaller in more diverse portfolios, and that there is still room for economic improvement in fisheries revenue. Though the general patterns and overall tenet of the results are robust to the major drivers of fisheries revenue in a region, careful attention would be required to select candidate portfolios with component species/species complexes that make the most sense for a given situation. Hence, we recommend that in any future operational application of the portfolio approach elucidated here that management bodies would have to purposefully select appropriate portfolios. In addition, they would have to select appropriate constraints for the optimization problem. For simplicity and demonstration purposes in this manuscript, the constraints were based on maximum historical landings weights for each species. One might consider using multispecies models (Karp et al. 2023) with harvest control rules implemented to empirically estimate appropriate, other types of stock constraints.

There are a number of decision points employed in this research that would need to be tailored to regional contexts before this portfolio theory‐based approach could be considered for operational use to implement ecosystem indicators and thereby potentially advance EBFM. First, the portfolio selection here was based purely on the revenue value of stocks in a region. Other cross‐regional or international markets may also need to be considered (e.g., Sguotti et al. 2023). When implemented as a regional ecosystem indicator, management bodies would need to develop criteria for considering a stock as part of the portfolio. This might logically align with existing fisheries management plans. Second, constraints used here for the optimization problem were simplified, and a more thorough review of regional management practices and species interactions would need to be considered. Data limitations would need to be considered. For example, in this study we excluded presenting the Pacific Islands as landings in Hawaii had relatively few stocks with complete time series of landings and revenue for the period of 1990–2020. Additionally, as noted above, recreational and artisanal fisheries were not considered in this study; frontier analyses were based on commercial revenue, excluding recreational landings and ignoring costs. However, doing so would be necessary in regions or with stocks where commercial fisheries are not the predominant source of landings and revenue (e.g., Gulf of Mexico, South Atlantic), and would likely alter the interpretation of if not the actual magnitude of the risk gaps. Incorporating recreational fisheries would also help to elucidate trade‐offs between commercial and recreational fisheries. Furthermore, the treatment of discards also warrants some attention, particularly in regions with a high volume of recreational fisheries. In some instances, the lack of considering discards can provide a different perspective on optimized landings (e.g., Shertzer et al. 2024). Finally, climate change will likely disrupt the historical patterns in the data used for this type of analysis (Karp et al. 2019). Improving our understanding of climate impacts on fish stocks and incorporating those impacts into this analysis would be important for a more robust view of economic risks to fisheries (c.f., Crowe and Parker 2008; Ando and Mallory 2012; for examples from other sectors). Obviously more detailed analyses and examinations in any given region in the USA, or elsewhere in the world with comparable data, would need to be executed before these results could be operationally adopted. But we assert that continued exploration to address these caveats should provide robust results and generally improve the utility of the risk gap as an ecosystem indicator.

Fisheries are prosecuted with some risk. It not feasible to think that full optimization to the point of zero risk (i.e., minimal variance) can occur for fisheries. Our results need to be interpreted in that context. Theoretically, one would not assume that any given realized portfolio value would routinely occur at the portfolio efficiency frontier (Edwards et al. 2004; Sanchirico et al. 2008; Rădulescu et al. 2010; DuFour et al. 2015; Schindler et al. 2015; Jin et al. 2016; Carmona et al. 2020; Lopetegui and del Valle 2022). Thus, the question becomes, how much risk gap would be expected, or what would be tolerable, in a fisheries context? We pose the question, but don't presume to know the answer beyond general directionalities and ordinal suggestions. More importantly, we propose the calculation and monitoring of this suite of portfolio‐related metrics to ensure that the discipline and practice of fisheries science and management begins to wrestle with this question. Perhaps exploring the level of risk tolerance would be an important policy choice that each management jurisdiction would find beneficial.

If a fisheries manager recognizes that there are risk gaps, then what can they do about it? We posit that simply monitoring these portfolio‐related indicators is a useful start. Beyond that, one can begin to explore a range of management decisions that balance the trade‐offs across species to better optimize portfolio revenue return and to minimize portfolio risk. For example, more flexible access, mitigating bycatch constraints, and promoting new markets help to allow fishers to target species that are currently productive. Specific actions could be determined for a given set of species in a particular portfolio for a region with a specific history, policy, and fishing context. Portfolio‐associated indicators can serve as a diagnostic for overall and specifically fisheries economic performance, and tracking them over time can inform whether fisheries management decisions are having the desired outcomes.

From our examination of prior studies, simulations, and observations (Edwards et al. 2004; Sanchirico et al. 2008; Rădulescu et al. 2010; DuFour et al. 2015; Schindler et al. 2015; Jin et al. 2016; Link 2018; Carmona et al. 2020), coupled with our results here, we assert that by using a systems‐based portfolio approach, fisheries management decision makers (such as regional fisheries management councils, state fisheries commissions, regional fisheries management organizations) can explore how to mitigate the risks facing living marine resource management and achieve better outcomes. Considering fisheries portfolios would escalate the rate of these more systemic evaluations, and further advance EBFM and its associated benefits.

ACKNOWLEDGMENTS

These analyses were partially funded by the Lenfest Ocean Program (Pew Charitable Trusts Grant Agreement 00035254), the Walton Family Foundation (Grant 00109688), and the NOAA Cooperative Institute for the North Atlantic Region (NA19OAR4320074). We want to thank the Fisheries Portfolio Project steering committee—Lisa Kerr, Scott Crosson, John Walden, Chip Collier, Doug Lipton, Mike Ruccio, Karen Abrams, Chris Dumas, Jeff Buckel, and Rob Griffin—which provided sage advice, helpful suggestions, and sound counsel as we developed and applied these methods. We also acknowledge the late Steve Edwards, who first introduced the portfolio concept to us and facilitated interactions between economists and ecologists. We thank reviewers of prior versions of the manuscript, including Scott Crosson, Rob Griffin, and Doug Lipton for their useful comments. There is no conflict of interest declared in this article.

DATA AVAILABILITY STATEMENT

All data and R scripts used to produce results in this manuscript are available online (https://bit.ly/3RSF3Ic).

REFERENCES