Climate and competition influence sockeye salmon population dynamics across the Northeast Pacific Ocean

Climate change is having a profound effect on the world’s ocean ecosystems, where it poses challenges to species persistence and fisheries sustainability (Cheung et al. 2010; Gattuso et al. 2015; Pörtner et al. 2019). As the world’s oceans warm and experience more frequent extremes (e.g., marine heat waves; Di Lorenzo and Mantua 2016; Cheng et al. 2019), species may be negatively or positively impacted depending on their distribution, and these effects can vary strongly across geographic gradients (Hoegh-Guldberg and Bruno 2010; Pinsky et al. 2013). In addition to direct effects on both the abundance and distribution of species, ocean climate may mediate the consequences of ecological processes (e.g., interactions among species) and be filtered through the life history characteristics of individual populations (Blois et al. 2013). As a result, the consequences of changing climate on marine species, and the economies and ecosystems that depend on them, can be difficult to predict and manage.

In addition to impacts from climate change, large increases in abundance of some fish species in the North Pacific Ocean over the past several decades have altered the dynamics of a wide variety of species from phytoplankton and zooplankton to seabirds and salmon (Springer and van Vliet 2014; Batten et al. 2018). Indeed, more Pacific salmon (Oncorhynchus spp.) have returned to fresh water from the North Pacific Ocean in recent years than at any time in the previous 90 years (Ruggerone and Irvine 2018). Pink salmon (Oncorhynchus gorbuscha) numerically dominate the abundance of salmon in the North Pacific (average 500 million adults per year), and ∼22% of salmon in recent decades (40% of biomass) are from salmon released by hatcheries in Asia and Alaska (primarily pink and chum (Oncorhynchus keta), which account for the majority of biomass; Ruggerone and Irvine 2018). These salmon are broadly distributed and overlap in the North Pacific Ocean (Myers et al. 1996; Beacham et al. 2014), where they can compete for common prey with other pelagic consumers (Kaeriyama et al. 2000; Bugaev et al. 2001; Davis et al. 2005; Johnson and Schindler 2009). Increased hatchery production in recent years has fueled debate about the potential for adverse effects on North Pacific ecosystems and has led to calls for international cooperation among North Pacific nations to reduce total releases of hatchery-reared juvenile salmon into the North Pacific Ocean (Holt et al. 2008; Debertin et al. 2017).

While the effects of climate change and increased competition among salmon for limited food resources have each been documented, the joint effects of these stressors on salmon productivity are poorly understood (but see Debertin et al. 2017; Cunningham et al. 2018). In particular, there has been little analysis of the potential mediating effect of ocean climate on density-dependent interactions across geographic gradients. Such mediating effects may occur, for example, as a result of climate-induced reductions (or increases) in growth during early life phases leading to increased (or decreased) sensitivity to density-dependent effects during later life phases. Alternatively, ocean climate may mediate the effects of density dependence within the same life phase. Here, we capitalize on a large dataset of sockeye salmon (Oncorhynchus nerka) populations from across the eastern North Pacific (Fig. 1a), along with information on ocean climate conditions and indices of potential salmon competitors at sea (Fig. 1b) to quantify the combined effects of ocean warming and increasing competition on sockeye salmon across their range.

Numerous marine species exhibit latitudinal gradients in responses to a warming ocean and, consistent with previous work (e.g., Mueter et al. 2002; Litzow et al. 2019), we found similar evidence for this in sockeye salmon productivity (Table 1). At the southern end of the sockeye salmon range (West Coast here), a warmer ocean during early marine life was related to reduced productivity, but in the middle and northwestern end of their range a warming ocean was associated with increased productivity (Figs. 2a and S1¹). The effect of a warming ocean was estimated to be 2.5 times stronger at the northwestern end of their range (23% increase in recruits per spawner per standard deviation unit (SDU) increase in SST; ≈1.5 °C) than in the middle (9% increase), whereas sockeye in the southern portion of their range were predicted to experience a 12% reduction in productivity (Table 1). The range of ocean temperatures encountered by sockeye salmon during early marine life are well within their physiological limits, suggesting that processes correlated with SST (e.g., stratification, phenology of spring bloom, advection affecting delivery of nutrients or zooplankton to coastal areas, or fish growth energetics) as opposed to direct temperature effects, drive these SST–sockeye productivity relationships.

Increasing competitor abundance was negatively associated with sockeye productivity at the southern end of their range, where a 1 SDU increase in competitor abundance (≈119 million salmon) was predicted to result in a ∼21% reduction in recruits-per-spawner (Table 1; Figs. 2b and S1¹). In contrast, we found evidence of a weaker negative association between competitor abundance and sockeye productivity in the Gulf of Alaska and northwestern end of the sockeye range (∼9% reduction in both regions; Table 1; Figs. 2b and S1¹).

The combined effects of a warming ocean and increasing salmon competitor abundance (and their interaction) across the North Pacific shifted from negative to positive across the sockeye range from south to north. At the southern end of their range our analysis predicts a 30% reduction in recruits produced per spawner for every increase of 1.5 °C in SST and 119 million salmon competitors (Figs. 2d and S1¹). The combined SST and competitor effects were highly variable in the middle of the sockeye range (Fig. 2d), but positive at the northwestern end of their range where the negative influence of competition was offset by a stronger positive influence of ocean temperature such that the same magnitude of increases in ocean temperature and competitors as above were predicted to result in a 19% increase in the number of recruits produced per spawner.

We found weak evidence that the effect of increasing ocean temperature during early marine life mediates the consequences of competition later in marine life at the northwestern end of the sockeye range (i.e., a positive interaction term whose credible intervals still overlap zero; Table 1; Fig. 2c). In contrast, the interaction terms for the other two ocean regions were smaller and more centered around zero. The weakly positive interaction term for the Bering Sea stocks suggests that as the ocean warms, the predicted effect of competition becomes weaker (i.e., it is antagonistic). We hypothesize that this may occur because a warming ocean during early marine life that increases sockeye productivity and perhaps growth at the northwestern end of their range may also make sockeye less sensitive to density-dependent interactions later in marine life.

We conducted sensitivity analyses to further examine the evidence for competition stemming from the combined abundances or biomass of pink, sockeye, and chum salmon in addition to climate effects, at both a North Pacific and North American scale. Inferences from these analyses were broadly similar to those presented here for pink salmon, though the estimated effect of the interaction between climate and competition for West Coast stocks and of competition for Gulf of Alaska and Bering Sea stocks varied to some degree depending on which competitor index was used (Supplemental Information, Table S5¹ and Figs. S2–S6¹). In addition, when our analysis was repeated with all available brood years of data, including those before the 1976–1977 ocean regime shift that strongly influenced both sockeye and pink salmon abundances, we found that the evidence for competition effects on productivity in the Bering Sea and the Gulf of Alaska, but not West Coast, declined while the effect of SST remained similar (Table S5¹ and Fig. S9¹). Other studies have found some evidence for nonstationarity in salmon – ocean climate relationships (Malick 2020), including a weaker negative effect of SST on sockeye productivity in West Coast region stocks after a pronounced decline in Aleutian Low variance in 1988–1989 (e.g., Litzow et al. 2019). We found no evidence to support nonstationarity in climate effects when we repeated our analysis on a dataset that was truncated to only consider the time period after 1988–1989 (Supplemental Information, Table S5¹ and Fig. S8¹).

Our findings are supported by previous research on Bristol Bay (Bering Sea) sockeye salmon that indicated the survival benefits from greater early marine growth offset the adverse effects of pink salmon on sockeye salmon during late marine life. Abundances of both pink and sockeye salmon in the North Pacific doubled after the 1977 ocean regime shift, and greater productivity of Bristol Bay sockeye salmon is associated with greater early marine growth (Ruggerone et al. 2007). Sockeye salmon originating from the Bering Sea interact with relatively few pink salmon during early marine life and numerous pink salmon during subsequent years when they are distributed farther west, leading to reduced growth, survival, and abundance of sockeye salmon (Ruggerone et al. 2003). Pink salmon effects on sockeye salmon are expressed by strong biennial patterns that cannot be explained by ocean climate. However, when examining adult returns per parent spawner, the complex life history of Bering Sea sockeye salmon (multiple years of residence in freshwater and ocean habitats) may make the detection of these biennial patterns more difficult. In other words, progeny from each brood year interact with both odd-year (abundant) and even-year (less abundant) pink salmon, potentially dampening the pink salmon effect on returns from the brood year. In contrast, sockeye salmon from the southern region, such as Fraser River, have relatively simple life histories dominated by a single age class (age-1.2) that maintains the biennial pattern in population characteristics (Ruggerone and Connors 2015).

Though a growing body of evidence suggests that competition among salmon at sea can influence salmon growth, maturity, and productivity, the potential for food resources to limit salmon production across the North Pacific continues to be vigorously debated (Amoroso et al. 2017; Shuntov et al. 2017). While the majority of salmon production is from wild populations, hatchery production increasingly contributes to the number of salmon at sea. For example, the abundance of hatchery pink salmon during 2005–2015 (82 million adults per year, or 17% of all pink salmon) exceeded the abundance of wild chum salmon and was equal to the abundance of wild sockeye salmon over the same time period (Ruggerone and Irvine 2018). In addition, there are strong geographic differences in hatchery production. For example, Alaskan hatchery production of pink salmon represented 18%–49% of total annual pink salmon produced in Alaska from 2005 to 2015.

Our analyses allow us to quantify what the potential consequences of hatchery production may be for sockeye productivity across their range. Using the parameter estimates from eq. 1 (Table 1), we estimate that total hatchery production of pink salmon has reduced sockeye productivity at the southern end of their range by ∼15%, on average, over the past decade (2005–2015; Supplemental Information¹). This suggests that hatchery production has contributed to the depressed productivity of sockeye salmon in British Columbia, some of which have recently been assessed as at risk of extinction (COSEWIC 2017). In contrast, above-average SST conditions in the Gulf of Alaska and Bering Sea regions over the past decade are estimated to have largely offset the negative effects of hatchery production on sockeye productivity. In the Gulf of Alaska, hatchery pink salmon production is estimated to have reduced sockeye productivity by ∼5%, on average, over the past decade, while in the Bering Sea the positive influence of above-average SST has led to an increase in productivity of ∼5%, on average, compared with an increase of 10%, on average, if no hatchery production had occurred (Supplemental Information¹).

Sockeye exhibit a remarkable degree of variation in life histories (Quinn 2018). This life history diversity (e.g., variable age at ocean entry or maturity), which can dampen the effects of a variable environment on salmon survival and abundance (Moore et al. 2014), may also moderate the effects of ocean climate and competition on sockeye by spreading the consequences of adverse climate and competition across multiple life histories within a cohort. This buffering effect may be particularly important for moderating the effects of competition because of the high-frequency variation in competitor abundance from year to year due to the fixed 2-year life cycle of pink salmon. The loss of life history diversity, for example due to climate warming (Cline et al. 2019), has the potential to increase the vulnerability of sockeye populations to the adverse effects of variable environmental conditions and reduce the stability of these populations and the fisheries that depend on them. As such, future research should seek to better understand how life history diversity mediates the consequences of a warming ocean and density-dependent interactions among salmon at sea on salmon dynamics.

Increasing abundances of salmon across the North Pacific, and in particular pink salmon, have been linked to a trophic cascade in epipelagic waters, leading to fewer zooplankton, reduced growth, survival and delayed maturation of salmon, reduced reproductive success of seabirds, and perhaps reduced foraging efficiency of southern resident killer whales (Orcinus orca) (Springer and van Vliet 2014; Ruggerone and Connors 2015; Batten et al. 2018; Ruggerone et al. 2019). Nonetheless, some jurisdictions (e.g., Alaska and Russia) continue to allow increasing hatchery production of pink and chum salmon with minimal consideration of adverse effects on distant salmon populations. Our findings highlight the importance of international cooperation to consider and potentially constrain the number of hatchery salmon released into the ocean to help Pacific salmon adapt to a warming and increasingly uncertain future.

We are grateful for the efforts of the many technicians and biologists who gathered and processed the salmon data used in this manuscript and to Jeanette Clark, Rich Brenner, and Stephanie Freund for compiling it. Ann-Marie Huang and three anonymous reviewers provided detailed and constructive feedback on an earlier version of the manuscript. This project was supported in part by the National Center for Ecological Analysis and Synthesis and the State of Alaskan Salmon and People project.