Integrating marine historical ecology into management of Alaska’s Pacific cod fishery for climate readiness

Abstract

The Pacific cod (Gadus macrocephalus) fishery was closed in 2020 after a rapid decline in biomass caused by the marine heat waves of 2014–2019. Pacific cod are exceptionally thermally sensitive and management of this fishery is now challenged by increasingly unpredictable climate conditions. Fisheries monitoring is critical for climate readiness, but short-term monitoring data may be inadequate for recognizing and anticipating change under rapid climate changes. We propose an interdisciplinary, marine historical ecology framework that looks to long-term records (local and traditional knowledge, history, archaeology, and paleoclimatology) to capture a long range of ecological variability and provide historical context for management. In order to connect to contemporary fisheries management, this framework must be built on a common vocabulary and an understanding of the key metrics used in fisheries stock assessments. Here, we propose metrics derived from Pacific cod stock assessment and synthesize information relevant to understanding the effects of past warming periods on cod populations across the Gulf of Alaska and Bering Sea. This case study provides a framework for thinking about how to use these historical records in the context of fisheries management under rapidly changing climate conditions.

Introduction

Rapid ocean warming over the last two decades has resulted in changes to primary producers, animal biogeography, and ecosystem organization, and has presented significant challenges to fisheries management on a global scale (Lee et al. 2023). Across global ocean environments, warming has also manifested as extreme marine heat waves (MHWs), when temperatures are warm enough to alter the structure of marine ecosystems across multiple timescales and trophic levels (e.g. Mantua and Hare 2002, Hobday et al. 2016, Reid et al. 2016, Smale et al. 2019). From January 2014 to January 2017, an unprecedented MHW, nicknamed “the Blob,” occurred in the North Pacific Ocean, influencing sea surface and deep-water temperatures over a vast area (Bond et al. 2015). Ecological impacts generated by the warm Blob were particularly pronounced in the Gulf of Alaska, where food webs are prone to dramatic reorganization (Gaichas et al. 2015). The MHW led to several converging changes: decreasing prey availability and increasing metabolic demands were coupled with increased competition among fish, bird, and mammal predators with overlapping diets (Barbeaux et al. 2020b). Poor quality zooplankton, shifts in the food web, and toxic algal blooms were all blamed for unusual mortality events among whales and seabirds, as well as disastrous declines in fisheries productivity (Cavole et al. 2016, Suryan et al. 2021).

Chief among these disasters was the collapse and eventual closure of the Pacific cod (Gadus macrocephalus) fishery from 2018 to 2020 (Williams et al. 2021). The crash of this iconic fishery has been described as one of the “most dramatic documented changes in a sustainably managed marine fishery” (Laurel et al. 2023) and has fundamentally challenged Alaska’s fisheries management strategies (Barbeaux et al. 2020b). Pacific cod are exceptionally thermally sensitive, and changes in thermal habitat during the MHW have been linked to lower egg hatch rates and higher mortality for juveniles (Laurel et al. 2021, 2023). These factors led to a drop in spawning biomass and eventually to the closure of one of the region’s most valuable fisheries, with losses of up to $100 million annually (Alaska Seafood Marketing Institute 2023).

As increasingly unpredictable marine conditions threaten to destabilize well-established fisheries management frameworks, climate readiness is critical (Burden and Fujita 2019, Barbeaux et al. 2020b). Pacific cod are a sentinel of ecosystem change because of their sensitivity to thermal shifts; consequently, localized monitoring may help anticipate climate stress and recruitment failure (Laurel et al. 2023). However, existing annual or decadal data (e.g. Litzow et al. 2022) may be inadequate for recognizing and anticipating change under increasingly rapid climate change scenarios (Beller et al. 2020). Long-term perspectives are critical to avoid the shifting baseline syndrome (Pauly 1995) and to prevent the gradual decline of fisheries resources, a phenomenon that has been recognized for three decades. A shifted baseline leads to a failure to recognize larger past population sizes and the erroneous assumption that recent population sizes provide appropriate reference points for management.

While the omission and truncation of past data have quantifiable impacts on conservation and management targets for fisheries, including critical measures of stock size, carrying capacity, and maximum sustainable yield (McClenachan et al. 2012, Schijns and Pauly 2022), historical data are rarely used in Alaska’s fisheries assessment. The development of the field of marine historical ecology over the last two decades has highlighted the wealth of long-term data sources that exist (Lotze and McClenachan 2013, Thurstan 2022, Rick 2023, Moore et al. 2024). A broad knowledge base exists for the use of historical data in marine fisheries management (Engelhard et al. 2016); yet it is rare to integrate these diverse perspectives on long-term change within the context of commercial fisheries assessment and management (McClenachan et al. 2024). Historical ecology is essential in an era of changing climate, but this approach is challenging because it requires integrating historical data, which are typically focused on past fishing, and climate history, which focuses on environmental variables.

In this article, we explore how marine historical ecology can enhance climate resilience in management of the Pacific cod fishery and whether the recent crash may have precedents in history. This fishery is well-situated to integrate marine historical ecology for three reasons: (i) thermally sensitive Pacific cod offer an opportunity to examine how this fish has responded to past periods of warming; (ii) the commercial Pacific cod fishery is data-rich; and (iii) Pacific cod annual assessments increasingly recognize the potential value of historical dynamics. Here, we review current Pacific cod research to illustrate how marine historical ecology can inform models used to set reference points for management. This review builds on previous marine historical ecology work to propose a framework for interpretation that draws on the parameters and metrics used in commercial fisheries stock assessment. These parameters provide a common vocabulary that connects marine historical ecology directly to contemporary fisheries management and anchors the convergence of diverse lines of evidence.

The Pacific cod fishery

The Pacific cod is a cold-water, demersal fish that is broadly distributed throughout the North Pacific Ocean from 34^oN to 63^oN latitude (Fig. 1; Hulson et al. 2022). In Alaska, Pacific cod are found along the continental shelf from the Gulf of Alaska, the Aleutian Islands, and into the Bering Sea, and they may be found from 25 to 550 m deep depending on the season and their age (Shimada and Kimura 1994). Tagging studies suggest that some cod make significant seasonal migrations across the region for winter spawning and summer foraging (e.g. Shimada and Kimura 1994), while others remain in one location year-round (e.g. Savin 2007).

Pacific cod are known to be extremely thermally sensitive and recent studies suggest they are vulnerable during the earliest life stages (e.g. Laurel et al. 2023). Temperature is closely related to ontogeny and reflected in cod movement in the water column: eggs are released at the bottom, larvae migrate to the surface, and mature fish move back to deeper, cooler water. After biomass decreased dramatically during the recent MHW, observational data were used to examine how changes in thermal habitat influence cod across multiple life stages. The results indicate that cod track cold water masses for spawning, and temperatures warmer or colder than 0–4°C in the earliest stages of life can result in reduced recruitment success because of thermal stress on eggs and starvation in larvae (Laurel et al. 2008, 2009, Laurel and Rogers 2020, Laurel et al. 2021, Litzow et al. 2022, Laurel et al. 2023). Thermal shifts also lead to metabolic stress later in life (Barbeaux et al. 2020b). This evidence suggests that natural mortality and biogeographic shifts will increase during periods of warming temperatures.

Pacific cod have been a mainstay of fishing in the Gulf of Alaska and Bering Sea for at least 6000 years and likely longer. While Pacific salmon are central to life along this coastline and have received significant attention in the archaeological, historical, and ethnographic literature (e.g. Campbell and Butler 2010, Colombi and Brooks 2012), the archaeological record suggests cod were as important to ancestral Aleut (Unanga|$\hat{\rm x}$|⁠) and Alutiiq (Sugpiaq) people (Crockford et al. 2004, West 2009). Cod were caught using hook and line from a kayak near shore or at depths up to 300 m (Fig. 2; Holmberg 1985, Stanek 2000, Berezkin et al. 2012). Fish were taken singly or two to three at a time using a rig spreader (Knecht 1995), and were eaten fresh, dried, and smoked (Partlow and Kopperl 2011). The remains of this harvest can be found in the zooarchaeological—or animal bone—record across the Gulf of Alaska, Aleutian Islands, and into the Bering Sea.

Little is known about the role of Pacific cod in the Russian colonization of Alaska in the 18th century (but see Margaris et al. 2015), though this was a period of great disruption to traditional subsistence as Aleut and Alutiiq men were conscripted to hunt for marine mammals for the Russian American Company (Reedy-Maschner 2010). Pacific cod were harvested commercially beginning in 1863 during the transition of Russian to American control of Alaska (Cobb 1927) when cod were fished by American schooners with dories and handlines, and many Scandinavian fishermen moved north from Washington and California for the fishery (Reedy and Maschner 2014). Shore stations were established in the 1880s and were concentrated in the Shumagin and Sanak Islands, where cod occupy the nearby banks year-round (Fig. 1; Reedy and Maschner 2014, Mackovjak 2019). Aleut fishermen participated in this commercial fishery and were employed at the shore stations (Mackovjak 2019). The fish were exported primarily as salt cod until the 1950s when the focus was on using harvested cod as bait, and Pacific cod harvest increased dramatically in the 1970s after the “Americanization” of the fishery and the establishment of the Magnuson-Stevens Fishery Conservation and Management Act (Mackovjak 2019).

Since the 1970s, the federal Pacific cod fishery has been managed by the National Oceanic and Atmospheric Administration (NOAA) Fisheries and the North Pacific Fishery Management Council using groundfish management plans for the three major stocks: Bering Sea, Aleutian Islands, and Gulf of Alaska. From a biological perspective, Alaska’s fisheries are generally considered well managed (Hilborn et al. 2020) and the cod fishery has long been described as sustainable (Laurel et al. 2023). However, in 2020, the spawning stock biomass was determined to have dropped below an acceptable level and the fishery was closed; it reopened in 2021 after severe economic losses (Williams et al. 2021). Recent stock assessments by the Alaska Fisheries Science Center acknowledge that current dynamics should be understood in a longer historical framework; however, it is not clear how historical data might be used in a practical way in fisheries stock assessment (e.g. Hulson et al. 2022). Here, we propose that historical data may be used to understand cod population dynamics across the Gulf of Alaska and Bering Sea over multiple scales; to apply these to fisheries management requires aligning multidisciplinary approaches to data collection and analysis.

Linking management metrics and historical datasets

Individual lines of historical evidence can be used to examine changes in past fish abundance and population dynamics, fishing activity, and environmental conditions (e.g. Scarborough et al. 2022, Moore et al. 2024). However, historical evidence can be difficult to communicate in a way that is useful for management: disciplinary boundaries and the variable nature of historical evidence have created different ways of collecting, assessing, and presenting data (e.g. Froyd and Willis 2008, Rick and Lockwood 2013, Thurstan 2022). As a result, these lines of evidence may be inconsistent with the needs of fisheries assessment or may be limited to retrospective descriptions of fisheries. To address these limitations, we assembled the Pacific Cod Historical Ecology Working Group with two goals: (i) to establish a convergent research team of specialists in cod assessment and management, archaeology, cultural anthropology, marine historical ecology, paleoclimatology, paleoecology, and genomics; and (ii) to generate a common vocabulary based on parameters and metrics used in fisheries stock assessment that can be used to link contemporary management to historical lines of evidence (Table 1).

Stock assessment vocabulary frames our analysis of the Pacific cod fishery and is used here to align different disciplines. Annual assessments form the foundation for fishery management by providing current stock status estimates based on population dynamics models. Fisheries managers use these estimates to set sustainable harvest policies based on both ecosystem and economic needs. Assessments are intended to prevent overfishing while also maximizing the number of fish that can be caught every year. Assessment models draw on decades of data about fish abundance, life history, and catch to estimate biological reference points such as stock size, spawning biomass, and initial fishing mortality, while environmental and genetic data are used to contextualize the fishery (e.g. Barbeaux et al. 2020b, 2022; Table 1).

For Pacific cod, assessment data are drawn from observer reports, catch reports, vessel surveys, and electronic monitoring data across fishery types (commercial, recreational, and subsistence) and gear types (pot, longline, trawl, and jig; Barbeaux et al. 2022, Hulson et al. 2022, Spies et al. 2022). As described in Table 1, data employed by the annual cod stock assessments include multiple categories of information that may come from the fishery itself (fishery-dependent data) or from outside the fishery (fishery-independent data). Here, we have distilled these data into six categories: (i) environmental conditions; (ii) selectivity; (iii) catch; (iv) body size; (v) life history; and (vi) stock delineation. These categories are derived from assessment literature (i.e. DeAlteris and Skrobe 2000, Cooper and Weir 2006) and were chosen because these concepts are fundamental to fishery assessment, and they can be aligned with long-term records (Table 1 and Fig. 3). We provide justification for these categories, their role in both modern and ancient fisheries, and evidence to support the application of long-term data to assessment.

Environmental conditions

Environmental variables consist of anything that influences fish life history, including water temperature, nutrient availability and primary productivity, and predation. While the confluence of these variables is complex, research on Pacific cod recruitment suggests that temperature and nutrient availability (Table 1) are most critical to hatch success, growth, and survival (Laurel et al. 2023); these variables support primary productivity and favorable foraging conditions (Whitney 2015). Until recently, these environmental conditions have essentially been held constant in assessment and when determining maximum sustainable yield and other catch specifications (Cooper and Weir 2006).

Given that Pacific cod are susceptible to thermal variation, the recent MHW has forced assessors to consider new ways to incorporate environmental linkages in fisheries stock assessments (e.g. Karp et al. 2019, Punt et al. 2024). NOAA defines a MHW as an anomalously warm water event when sea surface temperatures (SSTs) exceed a seasonal threshold for at least 5 consecutive days. The threshold for a severe MHW is typically above the 90th percentile (Hobday et al. 2016). The baseline period for defining and identifying these events in Alaska is based on a 30-year period from 1982 to 2011, and analysis of this period suggests heat waves have increased in both frequency and intensity since 1982 (Environmental Protection Agency 2024). Pacific cod stock assessments for the Gulf of Alaska (Hulson et al. 2022) and Aleutian Islands (Spies et al. 2024) now incorporate thermal shifts—or a Marine Heatwave Index—into models that estimate natural mortality, and heat wave conditions invariably result in increased fish mortality. However, there are limited data to address whether these events have historical precedent. To contextualize changes in the Pacific cod fishery, “hindcasting” may be used to examine how fish respond to climate change and for modeling future shifts (Froyd and Willis 2008). To be useful in a commercial fisheries management context, long-term environmental records must be able to: (i) provide detailed information on temperature and primary productivity and (ii) be analogous to contemporary conditions. Paleoclimatology has drawn on terrestrial (e.g. Anderson et al. 2016) and marine sediment cores (e.g. Addison et al. 2013), tree rings (Wiles et al. 2014), glacial chronologies (Barclay et al. 2009), and sclerochronology (Hallmann et al. 2011) to establish a pattern of high-frequency climate shifts across the Gulf of Alaska, which includes temperature and productivity data that may be useful for hindcasting.

Global records suggest a period of accelerated warming in the northern hemisphere from the 1890s to the 1940s, or the period of Early Twentieth Century Warming (Fig. 3). This warming occurred primarily across higher latitudes in North America, Greenland, Western Europe, and Russia (Frye 1983, Delworth and Knutson 2000, Yamanouchi 2011, Hegerl et al. 2018), and there is evidence of periodic change in both SST and primary productivity. In the Gulf of Alaska, historical atmospheric data available in the Simple Ocean Data Assimilation (Carton et al. 2000) indicate this warming event manifested as anomalously cool SSTs in the period before 1920 and anomalously warm SST after 1925 (McClenachan et al. n.d.; Fig. 4). To explore whether these changes are consistent with MHW conditions, we extracted observational temperature data from the Simple Ocean Data Assimilation—sparse input version 3 (SODAsi.3; Giese et al. 2016)). These data span the period 1850–2013 and consist of monthly mean values on a 0.4° × 0.25° longitude/latitude grid and on 40 vertical levels. To analyze temporal trends, we focused on SST anomalies around the Shumagin Islands in the western Gulf of Alaska (bounded by 54 N, 197 W; 52 N, 199 W; 55 N; S205 W; and 57 N, 203 W), where the commercial fishery has been active since the 1860s (Mackovjak 2019). Intervals of excessive heat were calculated by performing the area average temperature. Following conventions adopted by NOAA (Hobday et al. 2016), these averages include only grid points for which the monthly temperatures meet the definition of a MHW (above the 90% threshold). Intervals below the 10% threshold correspond to marine cold waves (MCWs) (Fig. 4).

The results of this analysis indicate there has been considerable temperature variability over the course of the last 100 years (∼1865–1950, which aligns with the reconstructed catch record—see the “Catch” section), both in magnitude and frequency (Fig. 4). Of particular significance to Pacific cod stocks are the increased prevalence of MHW events after 1915, when the temperature data switch from a system that favors frequent MCW to one that includes periods of sustained MHW conditions. Similar fluctuations are seen in dendrochronological and meteorological records from coastal Alaska, with warmer temperatures reflected in tree ring growth in the early 20th century (Wiles et al. 1998).

The observational record of environmental variables exists only back in time to ∼1850 ce; beyond that time frame, paleoclimate methods are required to reconstruct past environmental conditions. As described earlier, a key element of Pacific cod stock modeling includes a term for primary productivity, which can be inferred from paleoclimate proxy analyses using geochemical markers (e.g. Addison et al. 2012, 2013) or micropaleontological information (e.g. Barron et al. 2016). However, a major limitation is time and/or sample resolution: to be useful at the annual time scales that stock assessment models require for hindcasting approaches, paleoclimate records need to be developed from rare high-sedimentation-rate depositional environments (e.g. >0.2 cm/year) using ultra-high sampling resolution techniques, such as with core-scanning X-ray fluorometry (XRF; nondestructive sample analysis as fine as every 200 µm; Richter et al. 2006).

The temperate fjords along the margin of the Gulf of Alaska contain some of the highest biologically driven sediment accumulation rates in the world and low-oxygen bottom waters that enhance organic matter preservation (Skei 1983, Syvitski et al. 1987, Burrell 1989). A subannually resolved scanning XRF geochemical analysis was conducted on a sediment core record recovered from Katlian Bay near Sitka, Alaska (Fig. 1), and compared to a suite of modern instrument observations to generate a model of mean summer net primary productivity (NPP; Fig. 5; Addison et al. 2013). It was found that the geochemical ratio of bromine (Br)/chlorine (Cl), which is a proxy for sedimentary organic matter accumulation (Ziegler et al. 2008, Leri et al. 2010), has a highly significant correlation with mean summer NPP (Fig. 5a), making the geochemical Br/Cl ratio viable for reconstructing Gulf of Alaska primary productivity over longer timescales (Fig. 5b). When the multidecadal variance of this dataset is calculated (Fig. 5c) and compared against the generalized Late Holocene climate of the high-latitude Pacific region (Fig. 5d), both the magnitude and variability seen in the most recent ∼150 years are dwarfed by changes seen during earlier time periods. For example, there are several periods during the Neoglacial phase that exhibit Br/Cl Z-scores that are extremely high (>3) or extremely low (<3) relative to the values seen since 1900 ce (max 1.8; min −1.5); these same intervals are also associated with some of the highest variance observed throughout the last 2000 years. These data highlight that paleoclimate variability commonly exceeds the instrumental record of observations for the most recent ∼150 years.

Taken together, the data presented in Figs 4 and 5 indicate that the magnitude of change in past environmental conditions—specifically SST and NPP—exceeds those seen during the last few decades. Because modern fisheries assessment models are based on only decades of data, we suggest past environmental conditions may be applied to cod management modeling in two ways: first, at the most basic level, these documented periods of change in temperature and productivity improve our understanding of the range of natural variation in the system (e.g. Froyd and Willis 2008, Jiao 2009). Natural variation at centennial or millennial scales has not been incorporated into fisheries assessment in coastal Alaska; however, records of historical climate fluctuations in the North Pacific have been connected to changes in Pacific salmon (Oncorhynchus sp.; Beamish and Bouillon 1993), northern anchovy (Engraulis mordax) and Pacific sardine (Sardinops sagax; Lindegren et al. 2013), and now Pacific cod production (McClenachan et al. n.d.; see the “Catch” section). In the North Atlantic, climate shifts have been compared to changes in fish population size over hundreds of years (e.g. Engelhard et al. 2014) and have led to changes in long-term management plans and estimates of fishing mortality (e.g. Clarke et al. 2011).

Second, reconstructed thermal shifts (Fig. 4) have the potential to support emerging predictive tools for improving decision-making in the context of changing climate and environmental conditions. Barbeaux et al. (202b) predict major declines in Pacific cod mortality during future periods of warming: e.g. based on recent temperature data, the Bering Sea stocks are predicted to experience significant decline during the next heat wave. Assessment models that capture the relationship between temperature and fish mortality (e.g. Spies et al. 2024) must be informed by historical data to illustrate the range of environmental variability possible, to improve projections, and to assess risk (Free et al. 2023). Periods of change may be divided into time blocks representing past thermal shifts based on temperature reconstructions (e.g. Fig. 4), which provides context for understanding the magnitude of contemporary temperature change and fluctuations in fish catch that might be expected from future heat waves. Historical and paleoclimate data are particularly well suited to inform this approach because they underscore nonstationarity over long time periods, and illustrate a relationship between temperature and fish mortality (see the “Catch” section; Punt et al. 2024).

Selectivity

Gear selectivity is the capacity of a gear type to capture fish of a certain age or size. For example, mesh size, hook size, trap style, and fishing location all influence the size of the fish caught. Gear selectivity is described by size-based probability curves based on harvested fish length (Hovgård and Lassen 2000). As described earlier, cod are harvested by a multiple-gear fishery that includes pots, longlines, trawls, and jigs. Each gear type collects different size fish: the longline and pot fisheries catch larger fish on average (>64 and >40 cm, respectively), while the trawl fishery collects a wider size distribution, catching fish as small as 10 cm in length. This information is used to determine different catch allowances for each gear type on an annual basis.

To make ancient, historical, and modern catch (see the “Catch” section) and fish size (see the “Body size” section) data comparable, gear selectivity must be standardized. Historical information on selectivity in the past is available in multiple sources, as descriptions of the fishery include information on gear types and changes over time (Muir 2023). Similarly, the archaeological record contains information on gear selectivity that is both indirect and direct: gear selectivity can be surmised from fish body size distributions (Sanchez 2020, West et al. 2020), and the artifact record provides a direct measure of the limitations of specific ancient gear types, including hooks and nets (Stewart et al. 2021, Salmen-Hartley and McKechnie 2023). In the North Pacific, Indigenous fishers developed sophisticated fishing technologies that are generally considered highly size selective and may have been designed to avoid mortality among specific age classes (Salmen-Hartley and McKechnie 2023).

There has not yet been an analysis of the effects of gear selectivity on ancient cod harvest. However, the studies described in this paper assume that the selectivity of handheld gear used by ancestral Alutiiq and Aleut fishermen (Partlow and Kopperl 2011) is comparable to the longline and jig fisheries today (Burger et al. 2006, West et al. 2020). This assumption deserves further scrutiny: ancestral Alutiiq and Aleut fishermen used hook and line to harvest cod from kayaks (Fig. 2), and archaeological collections from coastal Alaska are rich in cod harvesting technology. These objects have been identified based on oral history and ethnographic descriptions (e.g. Steffian et al. 2015). For example, fishing line, bone hooks, fishing weights, and rig spreaders are preserved in museum collections and tell a detailed story of fishing strategies in the past (e.g. Knecht 1995, Berezkin et al. 2012). Using studies on Pacific halibut (Hippoglossus stenolepis) ancient gear selectivity as a model (e.g. Stewart et al. 2021, Salmen-Hartley and McKechnie 2023), cod hook metrics may be used to estimate the selectivity of Aleut and Alutiiq cod fishing gear, which is fundamental for interpreting the range of fish sizes represented in the archaeological record. Together with size (see the “Body size” section) and age (see the “Life history” section) data derived from the archaeological record, this technology can be used to assess the size selectivity of ancient fisheries relative to modern fisheries.

Catch

Catch is a broad and complex category that quantifies the number of fish caught and the amount of fishing effort in commercial fisheries. In the United States, catch data for Pacific cod come from a variety of sources, including federal, state, and noncommercial fishery catch, as well as foreign fishery reports and incidental catch in non-cod fisheries. Catch biomass, or the total weight of the harvested fish, is recorded in tons of fish per year by gear type (e.g. Barbeaux et al. 2020a: table 2.2). Catch data are used in conjunction with other data (e.g. vessel surveys, information on environmental, technological, and biological factors) to estimate abundance (von Szalay et al. 2007). When combined with effort data, catch-per-unit effort can also be used to derive relative abundances or stock densities (Gulland and Rosenberg 1992). These abundance estimates are used to establish fundamental targets for management, to determine stock response to fishing, and are critical for understanding stock response to recent environmental changes (Stevenson and Lauth 2019, Barbeaux et al. 2020b).

The baseline for Pacific cod assessment was established in 1977 when the fishery transitioned to US data collection and management under the Magnuson Steven Act of 1976. As described earlier, this coincided with a period of high productivity in the cod fishery, which has been linked to a shift to cool conditions in the central North Pacific Ocean and a regime shift in the marine ecosystem (e.g. Anderson and Piatt 1999, Hare and Mantua 2000, Litzow 2006). The scope of the 1976/1977 regime shift has been studied widely and catch data suggest that groundfish, which includes Pacific cod, increased in abundance (Litzow 2006). After this shift, the argument was made that historical data are necessary to understand both regime shifts and MHW patterns across the North Pacific, though these have only extended over a period of decades and current trends have been described as unprecedented (e.g. Francis and Hare 1994, Litzow et al. 2020, Williams et al. 2021). Here, we present local and traditional knowledge (LTK), historical records, and archaeological data that extend this record: if the catch serves as a proxy for the size of the cod population, these data indicate that cod have gone through alternating boom and bust periods over the last 4000 years, and there have previously been periods of severe decline. Furthermore, these periods of change can be seen at multiple scales (decadal, centennial, and millennial) and are likely connected to climate change and changing marine temperatures.

LTK is cumulative observational and experiential knowledge about environmental factors by local peoples who rely on fish, animals, and plants for survival (Berkes and Folke 1998, Thornton and Scheer 2012). On a global scale, LTK has been shown to improve both socioecological understandings of management practices (e.g. Molnar and Babai 2021) and monitoring programs for wildlife comanagement (e.g. Peacock et al. 2020). In Alaska, Indigenous and local residents are deeply reliant on and interested in the health of local resources, which support both economic and cultural security. As a result, LTK has been used in Alaska for co-management, knowledge co-production, and resource monitoring (Robards et al. 2018). However, particularly in regions where fisheries are data-rich and commercially valuable, historical constraints prioritize western and natural science data (Usher 2000, Trisos et al. 2021) and Indigenous and local perspectives are rarely applied (Carothers et al. 2021).

Qualitative evidence for fluctuations in cod catch comes from fishing communities in Alaska’s Western Gulf and Aleutian Islands. Cod has been an essential fishery in many Alaska Native Aleut communities for thousands of years, and local fishermen are profoundly aware of the volatility of fisheries (Reedy 2019). The Aleut word for Pacific cod is atxida or atxidan, which has been translated as “the fish that stops” (Bergsland 1994). This linguistic history, which is rooted in relationships with lands, waters, and resources, underscores the natural temporal and spatial variability in fish populations (e.g. Betts et al. 2011, Reedy and Maschner 2014). Like today, historical accounts document rapid changes in the abundance of fish in different locations (Shields 2001) and through time (Opheim 1994). For example, fisheries closures in 2020 reminded Aleut fishermen of the decades of “missing” cod in the late 1930s to 1960s when cod catch declined dramatically (Fig. 3; Shields 2001, Reedy and Maschner 2014). Aleut elders describe a longing for cod during this time and some local fishermen today explain this disappearance as “mother nature” and overfishing by “old timers” (Jacka 1999, Reedy and Maschner 2014, Reedy 2015). These declines in the 1930s have been discussed by fishermen for almost a century and are recorded in local accounts, fishing industry journals, oral histories, and ethnographies (e.g. Opheim 1994, Mackovjak 2019, Reedy 2019).

In an effort to interrogate the historical observations of cod decline, McClenachan and members of the Pacific Cod Historical Ecology Working Group (McClenachan et al. n.d.) created a century-long catch reconstruction for the Gulf of Alaska (Fig. 6). This record spans the period from 1864 to 1950 and was generated using archival records related to the American schooner fishery and observed catch rates, effort, and estimates of catch derived from the Pacific Fisherman journal and government reports (McClenachan et al. n.d.: supplementary material). In the Gulf of Alaska, catch data exist beginning in 1863 with the onset of the commercial fishery (e.g. Cobb 1927) and the reconstruction in Fig. 6 shows an increase in cod catch before a peak in 1915, followed by a decline after 1925. Potential drivers of change include economic, political, and social variables, such as competition with the Atlantic cod fishery and low demand (McClenachan et al. n.d.), and a shift in the focus of fishing across the region to salmon (Roppel 1986, Beamish and Bouillon 1993). However, McClenachan et al. (n.d.) argue that a thermal shift best explains this trend: the period of decline in the 1930s corresponds to the period of warming anomalies described in the “Environmental conditions” section that began in 1916 when water warmed in seasons and depths critical to cod life history (e.g. Laurel et al. 2023; Fig. 6). This catch reconstruction confirms the significant catch decline in the 1930s that has been experienced and discussed by local fishermen for almost a century (e.g. Opheim 1994, Mackovjak 2019, Reedy 2019), and provides evidence that this decline was likely related to a significant thermal shift.

This catch record can be extended over thousands of years in the Gulf of Alaska and Aleutian Islands, where cod have made up a significant portion of the diet for millennia. In the early 19th century, Veniaminov (1984: 39) described Aleut fishermen in the eastern Aleutian Islands catching hundreds of cod a day from kayaks in the spring (Fig. 2). This bounty is reflected in the zooarchaeological record, which is rich in cod remains (Moss and Cannon 2011, West et al. 2020). Based on the relative abundance of cod bones in archaeological samples, archaeologists argue there was a peak in cod harvest from 1550 to 550 bce (3500 to 2500 bp) on Sanak Island (Betts et al. 2011; Fig. 3), when cod were larger in body size (see the “Body size” section; West et al. 2020). Datasets from both Sanak and Kodiak Islands indicate a decrease in cod harvest after 550 bce at the same time that there was a decrease in body size (Fig. 3; Kopperl 2003, West et al. 2020). These data suggest it is possible that cooler conditions from 1550 to 550 bce (Fig. 4; Mann et al. 1998) were favorable for cod recruitment, while the relative decrease in cod abundance after this time could be due to poor growth or reduced marine productivity (Laurel et al. 2008, Helser et al. 2018). While the resolution is much coarser than the historical record, quantification of cod bones in the archaeological record across space and through time suggests cod catch fluctuated across the Gulf of Alaska over the last 4000 years (West et al. 2020).

Anchored by concepts derived from assessment—including environmental parameters and gear selectivity—LTK, historical, and archaeological catch data may be used together to contextualize the targets for cod management in two ways: first, local stakeholders have both historical and real-time information about the social and economic effects of fluctuating fish catch and fisheries management decisions on coastal communities (Robards et al. 2018). In the Gulf of Alaska, Aleutian Islands, and Bering Sea, local and Indigenous fishermen rely on both subsistence and commercial fishing to sustain their communities, and community survival requires the acknowledgment that flexibility and anticipation are critical in this volatile environment (Reedy 2024).

Second, risk assessment plays an important role in estimates of the acceptable biological catch and may be informed by long-term records that describe catch and climate conditions (Barbeaux et al. 2022). Risk assessments provide a range of harvest alternatives in various contexts, including future climate conditions (Barbeaux et al. 2022). These harvest alternatives may be informed by retrospective biomass estimates, environmental considerations, and trends in stock abundance, which together determine the current fishing risk level and recommendations for fishing. Current targets generally rely on short time series (e.g. Hilborn 1992) and regional-scale data, but LTK, historical, and archaeological records can extend these targets with evidence for changes in cod catch during past thermal shifts. For example, the International Pacific Halibut Commission suggests that historical data may be included in models to avoid stock sizes that were known to have led to poor fishery performance (Stewart and Martell 2015). LTK, historical, and archaeological data tend to be geographically specific, which can highlight local depletions that may justify fishing limitations in the short-term (e.g. Penney et al. 1999) and localized assessment strategies designed to anticipate depletions (e.g. Barbeaux et al. 2018). Together with catch data from the recent MHWs, long-term trends in catch shift the expectations of future productivity in the fishery and support a management approach that embraces nonstationarity (Punt et al. 2024) and locally relevant fisheries management (e.g. Ban et al. 2017).

Body size

In combination with other life history data, the size of the fish (both weight and length) is used in stock assessment to model the effects of harvest on the health and reproductive success of fish populations (Cadrin and Secor 2009). Fish length is often used to estimate the effects of harvest pressure on fish stocks as the largest fish are removed (Shin et al. 2005), but measured fish length may also be affected by age, predation, water temperature, season of harvest, location and depth of harvest, food availability, and gear type. Together with weight and age, fish length is used in a series of growth models and is expressed as a length-frequency distribution for each gear type (DeAlteris and Skrobe 2000; see the “Selectivity” section). In the Pacific cod assessment, fish length distributions are represented by the number of fish that fall into size category bins, and these data are available for different gear types in every year beginning in 1977 (e.g. Hulson et al. 2022).

Until the recent MHW, the Pacific cod fishery was considered sustainable and cod generally responded well to management decisions (Williams et al. 2021). However, in 2015 at the onset of the MHW, concerns were raised about the steep declines in average cod length (Barbeaux et al. 2020a). Later assessment revealed this was likely due to increasing temperatures as whole year classes were either lost or moved away, driving down the average length of the cod population. The relationship among cod fishing pressure, ontogenetic temperature requirements, and movement is complex and poorly understood, and the drivers behind fish size are difficult to tease apart (Li et al. 2022).

Historical and zooarchaeological data both provide evidence that may be useful for teasing apart these drivers. Historical data provide evidence of cod body size changes, as catch was often reported both in total numbers and total weight. Based on these reported catch statistics, there is strong evidence that the average size of cod caught in the Alaska salt cod fishery declined in the years 1922–1931 (e.g. Rathbun 1894), associated with the decline in overall catch and period of warming illustrated in Fig. 6. Similarly, zooarchaeology—the analysis of animal bones from archaeological sites—may be used to estimate past fish body size and to test for the effects of both harvest pressure and climate change (e.g. Butler 2001, McKechnie 2007, Betts et al. 2011).

In the North Pacific, archaeological size data exist for cod over a 6000-year period that spans ancient, historical, and modern time periods (West et al. 2020). West et al. (2020) estimated length distributions in archaeological collections from the Kuril Islands to Southeast Alaska using existing measurements of archaeological cod skeletal elements (e.g. Orchard 2003), and extracted modern cod measurements from the Alaska Fisheries Science Center North Pacific (NORPAC) database. These datasets were rendered comparable in two ways: first, modern fish length data were constrained by fishing location and fishing depth to accommodate the harvest areas for both the modern and ancient fisheries. Harvest areas were identified using a combination of archaeological site location data and ethnographic descriptions of fishing activity (e.g. Jochelson 2002), and then converted to metric depth and distance data for the modern fishery. Second, the modern dataset was limited to hook-and-line fishery to control for gear selectivity by approximating the methods used by ancient fisheries.

The results of the zooarchaeological analysis indicate two trends: first, cod sizes estimated from the zooarchaeological record show a cline consistent with modern cod body sizes across this vast geographic area. Today, the largest cod can be found in the central and eastern Aleutian Islands, with the average length growing smaller to the west and east; remarkably, archaeological estimates across this region mirror this pattern over 6000 years (West et al. 2020). Second, analysis of individual fishing locations reveals some evidence of change in fish size through time. There is no clear evidence for fishing pressure before commercial fishing, which would be represented by sharp declines in cod relative abundance and body size in the archaeological record (see Broughton 1997). However, in locations where commercial fishing has the longest history—the eastern Aleutian Islands, including Sanak and Unalaska Islands—there is a statistically significant decrease in the size of the largest fish after the introduction of commercial fishing (West et al. 2020).

These results suggest both basin-wide stability and localized fluctuations in cod size. Despite the climate and cod abundance changes described in sections “Environmental conditions” and “Catch,” the consistent size cline over 6000 years suggests some aspect of stability in the cod population over centuries to millennia across the North Pacific basin (see sections “Environmental conditions” and “Catch”). Slight, but statistically significant, reductions in cod size in specific locations suggest that Pacific cod may respond to overfishing depending on geography, similar to Atlantic cod (Planque and Fredou 1999, Laurel et al. 2023). Given these trends, current broad management regions may mask variability in individual cod stock responses to fishing (see the “Stock delineation” section; Spies 2012).

Life history

Fish life history includes fish age, growth rate, movement, and environmental conditions. These variables are fundamental for understanding both recruitment and mortality in a fish stock, whether a stock may be vulnerable to overfishing, and the effects of thermal shifts on recruitment. As described earlier, in comparison to other commercially important fish in the North Pacific (Litzow et al. 2020) and to the closely related Atlantic cod (Righton et al. 2010), Pacific cod are extremely vulnerable to temperature change, particularly during early life stages (Laurel et al. 2023).

The ancient and historical cod fisheries selected for adult fish (West et al. 2012; McClenachan et al. n.d.), so it is not possible to compare detailed early life history data using standard zooarchaeological methods or historical records. However, both stock assessments and historical studies incorporate biological and life history data from otoliths. Otoliths are accretionary calcium carbonate (CaCO₃) structures in the inner ear that grow throughout the lifetime of the fish (Campana 1999). Modern otoliths are used to age and reconstruct growth rates of individual fish (Thorrold et al. 1997, Ashworth et al. 2017), movement, and migration recorded in trace element ratios (Sr:Ca and Ba:Ca; Elsdon and Gillanders 2004), water temperature and salinity recorded in stable isotope values (δ¹⁸O and δ¹³C; Kastelle et al. 2022), and metabolic demand and hypoxic tolerance (Deutsch et al. 2020, Smoliński et al. 2021). Historical and archaeological collections often include otoliths, which can be identified to species morphologically (e.g. Harvey et al. 2000; Table 1) and analyzed using the same set of methods applied to modern fish.

While studies from the North Pacific are limited, Atlantic cod life histories have been reconstructed in detail using archaeological otoliths. These studies demonstrate connections between Atlantic cod growth rate and marine temperatures (Olafsdottir et al. 2017), length-at-age reconstructions (Pedersen et al. 2022), and changes in migration during periods of climate change (Edvardsson et al. 2022). The initial study of archaeological otoliths from the Gulf of Alaska and Aleutian Islands shows promise: otoliths from the last 600 years have been used to reconstruct both life history and to assess fish responses to late Holocene climate change (Fig. 3). Analysis of δ¹⁸O in archaeological cod otoliths demonstrated ontological change and movement through the water column (Young-Boyle 2015, Helser et al. 2018), as well as nearshore temperature changes over the last 600 years (West et al. 2012, Helser et al. 2018). Based on their comparison of δ¹⁸O in archaeological and contemporary otoliths, Helser et al. (2018) suggest there has been stability in cod ontogenetic migratory life history over 200 years. This record should be expanded to other local environmental variables that are recorded in archaeological fish bone and otolith chemistry (δ¹³C, δ¹⁵N, δ¹⁸O, amino acid δ¹³C), which can be analyzed to characterize geographic locations/habitats used by fish in the past (e.g. Elliott Smith et al. 2023), changes in temperature and metabolic demand (e.g. Helser et al. 2018), and NPP and foraging activity (Misarti et al. 2009).

Stock delineation

A fish stock is different from a fish population, as it may not be based on intrinsic biological differences among groups, but instead may be defined by harvesting boundaries determined by management or politics (Cooper and Weir 2006). In the United States, the cod resource is divided into the three broad, geographic stocks or management units: the Bering Sea, Aleutian Islands, and Gulf of Alaska. Each region has its own stock assessment and harvest specifications (see Barbeaux et al. 2022, Hulson et al. 2022, Spies et al. 2022).

While these broad stock delineations persist, genetic data reveal that cod population structure is more complicated than these units suggest. Across this region, cod population structure is likely the result of deep passes and current patterns that have limited cod movement (Cunningham et al. 2009) and resulted in a lack of gene flow among these regions (Spies 2012). From a management perspective, different groups of cod may be responding differently to harvest pressure and local environmental change, and management could allocate catch to reflect these populations (Spies 2012). As described in the “Body size” section, defined management units may result in unintentional depletion (e.g. Spies and Punt 2015, West et al. 2020).

Critically, at the beginning of the last MHW, Pacific cod were observed moving north from their historical range in the southern Bering Sea, likely during summer feeding migrations (Spies et al. 2020). After the end of the last MHW, local fishermen reported cod returned from the north, and were “big” and found in the same spots they had been before the heat wave. However, the time depth of these movements is unknown: genomic analysis of archaeological cod (Table 1) can be used to assess whether modern stock delineations reflect long-term genetic structure for this species. Studies on Atlantic cod demographics (Martinez-Garcia et al. 2022) and herring spawning groups (Atmore et al. 2022) demonstrate the influences of both fishing pressure and climate change on fish populations and their distribution, and these long-term data are being used to help set policy about fishing restrictions (Engelhard et al. 2014). If Pacific cod spawning location or migration activity was influenced by past MHWs or warm periods, genomic analysis spanning critical climatic fluctuations in the past can test this assumption over long time periods. Historical stock structure provided by ancient DNA (aDNA) can inform management units: these data will provide insight into whether population structure has changed over time and whether this might change in the future if cod shift their spawning areas as MHWs increase (Rand et al. 2023).

Discussion and conclusions

The Blob extreme warming event shocked resource managers with its severity, extent, and devastating effects on fisheries in the Gulf of Alaska and Bering Sea (Barbeaux et al. 2020b). In the aftermath of the Blob, managers have begun to consider ecosystem-based management strategies, climate modeling, and using historical data (e.g. Litzow et al. 2021, Free et al. 2023, Spies et al. 2024). Like many fisheries around the world, the cod fishery relies on just decades of fishery and environmental data to inform management. History asks us to confront our baseline understandings of the Pacific cod fishery and to acknowledge that changes in catch are an expected result of environmental shifts.

In this paper, the members of the Pacific Cod Historical Ecology Working Group have outlined how marine historical ecology provides an opportunity to contextualize dramatic climate events like the Blob and the environmental and biological effects of such extremes on thermally sensitive Pacific cod. We have described the kinds of qualitative and quantitative data that are available in historical datasets and have proposed ways to align long-term data with the concepts and data requirements of contemporary fisheries management. To offer a path forward for aligning these datasets, we have presented a common vocabulary that relies on the central concepts employed by commercial fisheries managers and can be used by marine historical ecologists to link long-term quantitative data, long-term qualitative data, and short-term management data. We propose that defining key metrics that can be compared across time is a critical first step to successfully integrating the varied datasets seen in other historical ecology projects with real assessment and management concerns and goals, and is the first step toward applying these long-term datasets to adaptive fisheries management approaches.

While research on the history and archaeology of the Pacific cod fishery is limited relative to the Atlantic cod fishery (e.g. Rosenberg et al. 2005, Alexander et al. 2009, Engelhard et al. 2014), examples provided in this paper draw on other case studies and management concepts to suggest ways that historical Pacific cod datasets might be aligned with management concerns. Natural variability characterizes both the North Pacific climate and the cod resource: while the recent MHW caused fluctuations in cod harvest, we have explored the possibility that these fluctuations also occurred in the past. Linguistic, historical, and archaeological data illustrate that cod populations have been variable historically and over millennia, showing periods of boom and bust over multiple scales. Climatological, paleoenvironmental, and archaeological data provide details about significant change in marine conditions in the past over similar scales, and the extent to which cod experienced those changes. Together, these datasets suggest that, like today, cod populations responded to periods of warming in the past with lower recruitment and slower growth, and we hypothesize they may have changed their migratory patterns.

Despite these periods of boom and bust, both genetic data and body size analyses underscore cod resilience over centuries and millennia. In contrast to the Atlantic cod fishery, historical and archaeological records suggest that Pacific cod was relatively lightly fished historically (West et al. 2020; McClenachan et al. n.d.). In the case of Atlantic cod, high historical fishing mortality may have induced earlier maturation and age truncation in some populations (Hutchings and Rangeley 2011). There is little evidence for similar heavy fishing pressure on Pacific cod populations. Genetic data indicate that cod have never been driven to extinction in their historical range and populations persist, despite being variable (Canino et al. 2010). Similarly, the consistent size distribution and ontogenetic life history described in the “Body Size” and "Life History" sections are an indication of population stability over centuries to millennia. More information about historical stock structure (e.g. Martinez-Garcia et al. 2022) could inform management unit structuring and lend insight into whether cod will shift their spawning areas as MHWs increase. If cod have successfully adapted to MHW conditions in the past, then the expectation is that they should recover from MHWs if fishing is limited.

As the North Pacific Ocean continues to experience both gradual warming and punctuated MHWs, marine scientists and managers suggest that we must learn from past heat waves to provide insights into how to prepare for future heat waves (e.g. Free et al. 2023, McClenachan et al. 2023). While data from the most recent MHW are rich and detailed, this approach offers little time depth to address the effects of significant warming on the complex interaction of fish populations, environmental conditions, economic systems, and cultural systems. Establishing the adaptive response necessary to weather future MHWs around the world may begin with a long-term view of these dynamics. Furthermore, such data are relevant to Aleut and Alutiiq communities that depend on Pacific cod as a subsistence and commercial resource. These communities recognize the potential scientific application and cultural significance of historical and archaeological information about their own regions and histories. As keen project participants sharing their observations and experiences, these combined data can assist in their own efforts to retain sustaining roles in this complex commercial Pacific cod fishery. The methods and metrics described here represent useful findings for contextualizing contemporary Pacific cod assessment, challenging the use of baselines, and providing a shared vocabulary that is a major step toward successful integration across disciplines. This convergence provides a way forward for marine historical ecology in the North Pacific region, offers pathways for future research in historical and archaeological contexts, and can be applied to fisheries and other marine resources around the world.

Acknowledgments

This paper is the product of meetings of the Pacific Cod Historical Ecology Working Group, funded by the National Science Foundation Navigating the New Arctic (RISE 2220552, 2220553, 2220554) and the Boston University Pardee Center for the Study of the Longer-Range Future. Thanks to Torben Rick, Trevor Lamb, John Barron, and three anonymous reviewers for comments that dramatically improved this paper, and to the community of Sand Point, Alaska, for their support of this project. J.A.A. was supported by the US Geological Survey Land Change Science Program. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the US Government.

Author contributions

Catherine F. West, Loren McClenachan, and Steven J. Barbeaux (Funding acquisition, Conceptualization, Methodology, Formal analysis, Writing—original draft, Writing—review & editing) and Jason A. Addison and Bruce T. Anderson (Funding acquisition, Methodology, Formal analysis, Data curation, Visualization, Writing - review & editing) and Ingrid Spies, Courtney A. Hofman, Katherine L. Reedy, Emma A. Elliott Smith, Michael A. Etnier, Thomas E. Helser, and Bruce P. Finney (Funding acquisition, Methodology, Writing—review & editing)

Conflict of interest

None declared.

Data availability

Sediment core geochemical data are archived at the World Data Center Pangaea (https://doi.pangaea.de/10.1594/PANGAEA.980873). Simple Ocean Data Assimilation—sparse input version 3 (SODAsi.3) data are available upon request ([email protected]).

References