Widespread thiamine deficiency in California salmon linked to an anchovy-dominated marine prey base

Significance

California’s anadromous salmonids face chronic stressors that include habitat loss and degradation, invasive species, hatchery influences, and climate change. We show that thiamine deficiency in California's salmonids is linked with an anchovy-dominated forage base and diet. Anchovy carry an enzyme (thiaminase) that degrades thiamine in the gut of consumers; they are also low in thiamine and high in lipids, which may promote thiamine deficiency by elevating oxidative stress. Thiamine deficiency will likely persist in California salmonids as long as anchovy dominate their ocean forage base and diet, further threatening culturally, ecologically, and economically valuable populations that are already threatened. Effective treatments are available for some salmon populations, but short-term treatment benefits may come with long-term fitness risks.

Abstract

Thiamine (vitamin B₁) deficiency in marine systems is a globally significant threat to marine life. In 2020, newly hatched Chinook salmon (Oncorhynchus tshawytscha) fry in California’s Central Valley (CCV) hatcheries swam in corkscrew patterns and died at unusually high rates due to a lack of this essential vitamin. We subsequently investigated the impacts and causes of thiamine deficiency in California’s anadromous salmonids. Our laboratory studies defined the relationship between thiamine concentrations in Chinook salmon eggs and early life-stage survival in offspring; we used these data to develop a model that estimated 26 to 48% thiamine-dependent fry mortality across consecutive years (2020–2021) for winter-run Chinook salmon. We established an egg surveillance effort that found widespread thiamine deficiency in CCV Chinook salmon in 2020 and 2021, and emerging thiamine deficiency in Klamath River and Trinity River coho salmon (Oncorhynchus kisutch) in 2021. We determined that thiamine injections into adults raised egg thiamine concentrations above levels found to impact early life-stage survival and swimming behavior. Ocean surveys, prey nutrition, salmon gut contents, and stable isotope data link thiamine deficiency to an ocean diet dominated by a booming population of northern anchovy (Engraulis mordax). This forage fish had low thiamine, high lipid, and high thiaminase activity levels consistent with both a thiaminase and oxidative stress hypothesis for causing thiamine deficiency in California salmon. Our research suggests California’s already stressed anadromous salmonids will continue to be impacted by thiamine deficiency as long as their ocean forage base and diet are dominated by northern anchovy.

Thiamine (vitamin B1) is essential to almost all life on Earth as a cofactor for many foundational enzymes (1). As such, thiamine is critical in energy production, nucleic acid synthesis, production of certain amino acids, and other aspects of energy metabolism. Thiamine is produced by plants and microbes at the base of food webs (2), and most organisms obtain thiamine through their diet. Thiamine is water-soluble, is not readily stored in tissues, and has a relatively short half-life in most organisms (3). Thiamine deficiency complex (TDC) is a nutritional deficiency of thiamine that causes a range of morbidities including neurological problems and death; it is paradoxically both an ancient human malady and an emerging global threat to fish and wildlife populations (4). Thiamine deficiency, called ‘beriberi’ in humans, has a long history; it was first described between 300–2700 BCE in The Yellow Emperor’s Classic of Medicine (5) and was causally linked to TDC in 1897 CE by Christiaan Eijkman (6). TDC was linked with diseases in fish and wildlife in the early 20th century (7, 8), and is a possible driver of population declines for diverse taxa including fish, reptiles, mollusks, and birds (4).

TDC—also known as early mortality syndrome (EMS) and M74—has been most widely diagnosed and studied within populations of anadromous fishes, where it is considered a threat to global salmonid fishery stability. TDC has affected salmonid populations in the Laurentian Great Lakes, Lake Champlain, the New York Finger Lakes, and the Baltic Sea (9–13). It is characterized by a thiamine deficiency in eggs due to poor maternal transfer resulting in high offspring mortality from yolk-sac to swim-up stages (14).

TDC in fish, wildlife, and humans can occur through different mechanisms. TDC in livestock (sheep, cattle, chickens), farmed fur-bearing animals (mink and foxes), aquaculture, and captive zoological animals can be caused by low thiamine diets, but the dominant causative factor for TDC in these cases was consumption of food that was rich in thiaminase (1, 15). Thiaminases are enzymes present in certain microbes, plants, insects, invertebrates, and even vertebrate species, including many fishes (16–19). Importantly, prey high in thiaminases degrade thiamine in the gut of consumers during digestion. This linkage between a thiaminase-rich diet and TDC has occurred in humans (20), livestock (21), fur farm animals (7, 22, 23), Pacific harbor seals (Phoca vitulina) (24), and fish (8, 14). Widespread TDC in Great Lakes salmonines has been linked to consumption of invasive, thiaminase-rich alewife (Alosa pseudoharengus) (25–27) while studies of TDC in Baltic Sea Atlantic salmon implicated diets that are low in thiamine and high in unsaturated fat, which presumably depletes thiamine stores by elevating oxidative stress (28–30).

TDC was diagnosed for California Chinook salmon (Oncorhynchus tshawytscha) in 2020, the first documented case for Pacific salmon in western North America. Although low egg thiamine was reported earlier in Alaska Chinook salmon, no fry mortality was reported (31). Chinook salmon fry at multiple hatcheries in California’s Central Valley (CCV), showed clinical signs of TDC including anorexia and sporadic spinning behavior prior to death between hatching and first feeding. TDC was confirmed when affected fry were treated in a thiamine bath and within hours recovered normal swimming and feeding behavior (32).

Here, we report key findings from an interdisciplinary study launched in spring 2020 to characterize the mechanisms, impacts, and mitigation measures for TDC and early life stage mortality in California’s anadromous salmon (33). Our first objective was to quantify relationships between egg thiamine concentration, fry morbidity, and mortality. Next, we tested the efficacy of thiamine supplementation to mitigate the impacts of TDC on fry. We then paired the egg thiamine-fry survival relationship with egg thiamine surveillance data to characterize the prevalence and severity of TDC across California’s salmon populations, including different seasonal run-types and rivers. We quantified the stable isotope ratio of nitrogen (δ¹⁵N) of each egg sample for evidence of a diet link to egg thiamine concentration at spawning. Finally, we conducted a food web and nutritional analysis of salmon diets and the forage base of the California Current Ecosystem, the marine feeding ground for California salmon. Our goal was to determine whether a diet switch by Chinook salmon to thiaminase-rich or high-fat and low thiamine prey was related to TDC as documented in other cases reviewed above.

Results

Impacts of Thiamine Deficiency on California Salmon.

Offspring survival and egg thiamine concentrations.

Unfertilized eggs and embryos were collected from 135 spawning Chinook salmon to generate data linking egg thiamine concentrations to impacts on fry behavior and survival. TDC-related clinical signs and mortality were observed and recorded for four spawning groups from 80 days postfertilization (dpf), approximately the swim-up fry stage, to ~120 dpf. The 2020 winter-run Chinook salmon group was split into treatment and control subgroups, wherein 4–10 weeks prior to spawning adult females in the treatment group received thiamine injections and those in the control group received a sham injection of saline solution (Table 1). The fry mortality rate in the control group (n = 28) averaged 23% and ranged from 0 to 90% for individual families; there were no fry mortalities in the thiamine-treated families (n = 29).

Table 1.

Summary of eggs and embryos collected for analysis of fry mortality rates used in the development of our thiamine-dependent fry mortality (TDFM) model

Source	Brood Year	Fry rearing location	Embryos per female collected	Unfertilized egg thiamine concentration (mean ± SD) nmol g−1	Observed fry mortality rate	Observed fraction of families with TDC symptoms
Livingstone Stone Hatchery winter-run (untreated)	2020	UC Davis	90 to 135 each from 28 spawners	4.9 ± 4.2	23 ± 31%	50%
Livingstone Stone Hatchery winter-run (adult females were given pre-spawn thiamine injections)	2020	UC Davis	90 to 152 each from 29 spawners	34.9 ± 7.3	0%	0%
Feather River Hatchery spring-run (untreated)	2020	UC Davis	100 each from 30 spawners	9.2 ± 4.8	14 ± 15%	Data were not collected
Feather River Hatchery fall-run (untreated)	2021	Salmon in the Classroom	35 to 51 each from 11 spawners	6.4 ± 3.4	11 ± 23%	64%
Coleman Fish Hatchery late-fall-run (untreated)	2022	UC Davis	200 each from 37 spawners	4.9 ± 1.2	14 ± 16%	16%

At the earliest age of 84 dpf and the latest age of 117 dpf, fry in control-group families began to exhibit clinical signs of TDC, including lethargy and corkscrew swimming, while no fry from the treated families did. Clinical signs of TDC were observed for 50% of control-group winter-run Chinook salmon, 64% of fall-run Chinook salmon, and 16% of late fall-run Chinook salmon families, with all but one of the affected fry coming from egg sample groups having thiamine concentrations below 8 nmol g⁻¹ (Fig. 1A and Table 1).

Fig. 1.

(A) Observed egg thiamine concentrations and associated fry survival rates to 80 to 120 d postfertilization (dpf) for Chinook salmon samples summarized in Table 1 (tan dots). Data points having a black dot in their center indicate families having fry that showed clinical signs of thiamine deficiency complex (TDC). The best fit logistic model is indicated by the bold red line. Posterior 95% confidence intervals on the best fit model are indicated with the gray shading. Inset provides a closer view of survival at concentrations <10 nmol g⁻¹. (B) Egg thiamine concentrations for control and treatment groups. Left panel shows results from Livingstone Stone Hatchery winter-run Chinook (LSH Wr-Ck) salmon prespawn injections in 2020 and 2021. Right panel shows results for Feather River Hatchery spring-run Chinook salmon (FRH Sp-Ck) prespawn injections.

We fit a dose–response model that revealed a clear relationship between egg thiamine concentration and percentage of fry surviving to ~120 dpf (Fig. 1A). The model predicted the following fry survival rates at the listed egg thiamine concentration values: 95% at 7.7 nmol g⁻¹ [95% CI = (6.2–9.1)], 90% at 5.9 nmol g⁻¹ [CI = (4.8 to 6.9)], and 50% at 2.7 nmol g⁻¹ [CI = (2.3 to 3.1)]. We used these estimates to define the following impacts criteria for measured egg thiamine concentrations: severe for ≤ 2.7 nmol g⁻¹, impacted for >2.7 and < 5.9 nmol g⁻¹, likely impacted for 5.9 - 7.7 nmol g⁻¹, and not likely impacted for >7.7 nmol g⁻¹.

Efficacy of thiamine treatments.

Thiamine injections into adult female Chinook salmon 4 to 10 wk prior to spawning time resulted in dramatically higher egg thiamine concentrations at time of spawn when compared with untreated eggs (p < 0.001, Kruskal–Wallis rank sum test). For winter-run Chinook salmon in 2020, mean egg thiamine concentration at spawn was 5.2 ± 4.4 nmol g⁻¹ for the control group and 34.9 ± 7.3 nmol g⁻¹ in the thiamine injection group (Fig. 1B). Similarly, winter-run and spring-run Chinook salmon in 2021 had much higher mean total thiamine concentrations in the thiamine injection groups (31.9 ± 13.9 nmol g⁻¹ and 24.0 ± 8.4 nmol g⁻¹) than in the control groups (3.1 ± 1.5 nmol g⁻¹ and 9.9 ± 5.0 nmol g⁻¹), respectively.

Egg thiamine surveillance.

We collected unfertilized (and untreated) eggs from adult female salmonids (targeting 30 females per sampling event) at spawning to assess thiamine concentrations from 13 populations across eight hatcheries and four run timing seasons (two coho salmon populations; one late-fall-run, one winter-run, two spring-run, and seven fall-run Chinook salmon populations) in 2020 and 2021 (Fig. 2). We documented thiamine deficiency (defined here as eggs having a total thiamine concentration ≤7.7 nmol g⁻¹) in all CCV spawning groups, but only some of the Klamath and Trinity River spawning groups. The proportion of thiamine-deficient CCV Chinook salmon families within a population ranged 7–91% in 2020 and 45–100% in 2021 (Fig. 3, Bottom panel). Egg thiamine concentrations for all CCV salmon families ranged from 2.5 to 28 nmol g⁻¹ in 2020, and 2.0 to 23 nmol g⁻¹ in 2021. Among CCV samples, winter-run Chinook salmon had the lowest population mean values in each year (5.2 ± 4.4 nmol g⁻¹ in 2020, 3.1 ± 1.4 nmol g⁻¹ in 2021; one population), fall-run Chinook salmon had the highest mean value in 2020 (11.6 ± 4.7 nmol g⁻¹; five populations), and spring-run Chinook salmon had the highest mean value in 2021 (9.9 ± 5.0 nmol g⁻¹; one population) (Table 1).

Fig. 2.

Salmon hatchery and ocean prey sampling locations. Hatcheries are indicated with red dots, and hatchery name abbreviations are shown in parentheses. Hatchery name is indicated by 3-letter codes: Livingston Stone Hatchery (LSH), Coleman National Fish Hatchery (CNH), Feather River Hatchery (FEH), Nimbus Fish Hatchery (NIH), Mokelumne Fish Hatchery (MOH), Merced Fish Hatchery (MER), Iron Gate Hatchery (IGH), and Trinity River Fish Hatchery (TRH).

Fig. 3.

Center panel shows violin plots with dashed lines for potential health risk categories for untreated eggs sampled at hatcheries: severe for ≤2.7 nmol g⁻¹, impacted for >2.7 and <5.9 nmol g⁻¹, likely impacted for 5.9 to 7.7 nmol g⁻¹, and not likely impacted for >7.7 nmol g⁻¹. Klamath/Trinity River (Top panel) and California Central Valley (Bottom panel) populations ordered by adult freshwater entry timing. Species are indicated by two-letter codes: WR = winter-run Chinook salmon; FR = fall-run Chinook salmon; SR=spring-run Chinook salmon; LF = late fall-run Chinook salmon; Co = coho salmon. The proportion below 7.7 nmol g⁻¹ for each sampling group is in the column at Left. Right panel shows the mean model predicted thiamine dependent fry mortality rate (TDFM) rate for each sampling group. TDFM was not predicted for coho salmon populations because the relationship between thiamine concentration and survival developed here was specific to Chinook salmon.

In the Klamath and Trinity rivers thiamine deficiency was rare in Chinook and coho salmon eggs from hatchery samples in 2020, but more prevalent in 2021 (Fig. 3, Top panel). In 2020, 0% of Chinook salmon (three populations that included one spring-run and two fall-runs) and 3% of the coho salmon (two populations) families from the Klamath and Trinity River hatcheries were thiamine deficient, while those rates ranged from 0–17% for Chinook salmon in 2021 and 57 to 63% for coho salmon in 2021. Thiamine concentrations for Klamath and Trinity River Chinook salmon eggs sampled in these years (11.6 ± 4.7 nmol g⁻¹ in 2020, 13.3 ± 3.4 nmol g⁻¹ in 2021) were generally higher than those for CCV Chinook salmon eggs (SI Appendix, Table S1); this was also true for Klamath and Trinity River coho salmon eggs in 2020 (17.0 ± 5.0 nmol g⁻¹), but not in 2021 (7.7 ± 3.0 nmol g⁻¹).

Model estimates for thiamine-dependent fry mortality in California Chinook salmon populations.

Using observed egg thiamine concentrations for Chinook salmon in 2020 and 2021, our model predicted that thiamine-dependent fry mortality (TDFM) was highly variable across populations and years. Winter-run Chinook salmon were predicted to be most impacted by TDFM each year, with mean fry mortality values of 26% [95% CI = (0 to 61%)] in 2020 and 48% [95% CI = (6 to 88%)] in 2021. (CNH) late-fall-run Chinook salmon in 2020 and fall-run Chinook salmon in 2021 had predicted mean TDFM values of 23% [95% CI = (2 to 60%)] and 22% [95% CI = (4 to 61%)], respectively. FEH spring-run Chinook salmon in 2020, FEH fall-run Chinook salmon 2021, and (MEH) fall-run Chinook salmon in 2021 were predicted to have mean TDFM values between 11–13%. The remainder of the populations were predicted to have mean TDFM < 10%, with MEH fall-run Chinook salmon in 2020 having the lowest mean predicted TDFM of 2%. Model-based TDFM estimates for all Klamath/Trinity Chinook salmon samples in 2020 and 2021 had mean values of 0 to 3% (Fig. 2B).

Ocean-Caught Adult Chinook Salmon Diets, Prey Community, and Prey Nutrition.

Ocean-caught Chinook salmon gut contents.

We found that northern anchovy dominated the gut contents of ocean-caught Chinook salmon in June–October in 2020 (98% biomass), 2021 (89% biomass), and 2022 (100% biomass) (Fig. 4A and SI Appendix, Table S4). Eleven prior years (between 1955 and 2007) of Chinook salmon gut content data collected in May–August in the same region had much greater diet diversity, with northern anchovy volume fractions averaging 49% and ranging 13 to 81% (34) (Fig. 4A). Krill (Euphausiid spp.) had the second highest relative biomass of prey in 2020 (2%), while market squid (Doryteuthis opalescens) had the second highest relative biomass of prey in 2021 (11%). Northern anchovy dominated the gut content biomass in all months sampled in each year (June–October 2020, June–September 2021 and 2022; SI Appendix, Fig. S2). By month, the relative biomass of krill in gut contents was greatest in samples collected in June 2020 (13%), and the relative biomass of market squid was greatest in samples collected in August 2021 (16%). Overall prey diversity in the gut contents from 2020–2022 was extremely low, with only one to four prey types having a biomass fraction > 1% in all months sampled.

Fig. 4.

(A) Chinook salmon gut contents for northern anchovy (Engraulis mordax), young-of-year (YOY) rockfish (Sebastes spp.), krill (Euphasiid spp.), sardines (Sardinops sagax), market squid (Doryteuthis opalescens), and other YOY fish % volume from (34) and % biomass for 2020–2022 (SI Appendix, Table S4). Gaps between the stacked bars and 100% indicate % biomass of other prey taxa. (B) Core area % biomass in trawl catch from the springtime Rockfish Recruitment and Ecosystem Assessment survey (RREAS) (35).

Shifts in the prey community over time.

The proportion, by weight, of epipelagic micronekton (those taxa commonly preyed on by salmon) captured in the springtime mid-water trawl Rockfish Recruitment and Ecosystem Assessment Survey (RREAS) indicate the 2019–2023 central California marine forage community was unique with respect to the dominance of northern anchovies and the scarcity of other forage taxa compared to that from 1990–2018 (Fig. 4B and SI Appendix, Fig. S3). The proportion of krill in trawl catches was especially low from 2019–2021 and 2023, reaching values previously seen only in 1998.

The proportion, by weight, of coastal pelagic fish species collected from trawl gear during the summertime California Current Ecosystem Survey (CCES) also demonstrates both the shift to and northward expansion of an anchovy-dominated forage base from 2017 to 2021 (Fig. 5). In 2017 and 2018, anchovy-dominated trawls extended as far north as San Francisco, CA. In 2019, the northern boundary of anchovy-dominated trawls extended another ~150 km northward to Pt. Arena, CA, and in 2021 an additional ~150 km northward to Cape Mendocino, CA. Estimated biomass of the central stock of northern anchovy in the California Current Ecosystem increased dramatically after 2015, reaching a near record high in 2021 (SI Appendix, Fig. S4) (36, 37).

Fig. 5.

Proportion of northern anchovy biomass in trawl clusters indicated in green-shaded pie charts from the summertime California Current Ecosystem Survey (CCES) (36). Black dots indicate zero catch, while empty circles indicate 0% proportion of anchovy biomass in trawl clusters having nonzero catch.

Salmon prey nutrition.

We found distinct differences in three dimensions of the nutrition of five leading Chinook salmon prey items for this region (Fig. 6 and SI Appendix, Table S5). Anchovy had the highest and most variable thiaminase activity levels, with a mean value of 35 nmol g⁻¹ min⁻¹ and a maximum value of 206 nmol g⁻¹ min⁻¹. Of the remaining prey species analyzed, only herring had thiaminase levels high enough to potentially be problematic. In contrast, mean thiamine concentrations were highest in krill, followed by young-of-the-year (YOY) rockfish (Sebastes spp.), and while market squid samples were lower on average than YOY rockfish they spanned a larger range of values than all other prey items sampled. Herring (Clupea pallasii) had the second lowest thiamine concentrations, and anchovy had the lowest mean thiamine concentrations. Lipid content was especially high and variable in anchovy followed by herring, while YOY rockfish, market squid, and krill had the lowest lipid content.

Fig. 6.

(Top panel) Prey thiaminase activity (nmol min⁻¹ g⁻¹), (Middle panel) thiamine concentration (nmol g⁻¹), and (Bottom panel) lipid content (%) for the 5 main Chinook salmon (Oncorhynchus tshawytscha) prey items northern anchovy (Engraulis mordax), Pacific herring (Clupea pallasii), krill (Euphasiid spp.), young-of-year rockfish (Sebastes spp.), and market squid (Doryteuthis opalescens).

Salmon diet link to egg thiamine concentration.

There was a strong inverse relationship between δ¹⁵N of Chinook and coho salmon eggs and egg thiamine concentration in 2020 and 2021 (Fig. 7). CCV Chinook salmon populations had higher δ¹⁵N and lower egg thiamine concentrations than Klamath/Trinity Chinook salmon population samples. Winter-run CCV Chinook salmon had the lowest egg thiamine concentrations and highest δ¹⁵N values, while Klamath and Trinity River Chinook salmon had the highest egg thiamine concentrations and lowest δ¹⁵N values (SI Appendix, Table S3).

Fig. 7.

Means and standard deviations for thiamine concentrations and nitrogen stable isotopes (δ¹⁵N in ‰) for Chinook and coho salmon egg samples in 2020 (Left panel) and 2021 (Right panel). CCV population groups are indicated by filled symbols, and Klamath/Trinity population groups are indicated by open symbols. Mean values for Chinook salmon groups are shown with squares and coho salmon groups are shown with circles.

Discussion

Offspring Survival Rates Versus Egg Thiamine Concentrations.

Our results suggest that CCV Chinook salmon (EC50 = 2.70 nmol g⁻¹) are more sensitive to low egg thiamine concentrations than Lake Ontario Chinook salmon (EC50=2.06 nmol g⁻¹) but less sensitive than Lake Ontario coho salmon (EC50 = 2.95 nmol g⁻¹) and steelhead O. mykiss (EC50 = 6.54 nmol g⁻¹) (38). Fry survival studies linking health impacts to thiamine status for west coast steelhead and coho salmon could help fill important gaps in understanding species-, region-, and population-specific fry survival sensitivity to thiamine concentration. Likewise, these fry survival studies could be paired with a broader scale egg thiamine surveillance program to better understand the scope of thiamine deficiency and its impacts on west coast steelhead and coho salmon populations.

Benefits and Risks of Thiamine Treatments.

Supplementation of thiamine via injection in female Chinook salmon during maturation was effective in raising egg thiamine concentrations for early migrating (winter-run and spring-run) Chinook salmon. Although population-level treatments are likely not feasible for most naturally spawning Chinook salmon, it may be possible to effectively treat some portion of a population of naturally spawning early-migrating ecotypes that are trapped either during or soon after completing their return migration and then released to natural spawning areas. Introduced coho salmon in the Platte River (Michigan, USA) which received a thiamine injection prior to initiation of upstream migration, at a dose comparable to that used in this study, exhibited significantly higher survival and tissue-thiamine concentration when compared to unsupplemented individuals (39).

Thiamine supplementation has benefits and costs. While supplementing thiamine via injection in adult female Chinook salmon prior to spawn is highly effective at increasing egg thiamine concentration and potentially provides metabolic benefits during migration, this could mask any preexisting thiamine deficiency and hamper managers’ abilities to estimate its population level impacts. Supplementation may also circumvent selection for traits or genes that promote resilience to low thiamine conditions. For naturally spawning fish, one should also consider the population-level risks and benefits associated with handling stress and migration delays. If sensitivity to low thiamine varies within a population and has a genetic basis, natural selection might drive adaptation to low thiamine conditions, but adaptation could be subverted by broad application of thiamine supplementation. For instance, Atlantic salmon from the eastern United States and Canada with high thiaminase I in their natural diet exhibited reduced impact of a high thiaminase I diet (40). When evaluating trade-offs for affected populations, one should consider risks and benefits associated with both demographics and genetics (41).

Egg Thiamine Surveillance Results.

There are only a few published studies on the topic of thiamine deficiency in Pacific salmon in their natural range. One study did not find thiamine deficiency in four species of Fraser River (British Columbia) anadromous salmonids from samples taken in 2015 (and sockeye salmon in 2014), wherein average total thiamine concentrations from eggs taken on the spawning grounds exceeded 10 nmol g⁻¹ for all species (42). However, their sample sizes were small (ranging from 2 to 20 samples per species) and limited in time and space. In contrast, thiamine deficiency (total thiamine concentrations <7.7 nmol g⁻¹) was found in about 50% of the steelhead egg samples collected from three Oregon coast hatcheries in 2019 and one hatchery in 2020 (43). A majority of Chinook salmon egg samples from western Alaska in 2012 (74%), 2014 (70%), and 2015 (58%) had total thiamine concentrations <8.0 nmol g⁻¹, with 1 to 5% of samples having a concentration <1.5 nmol g⁻¹ (31, 44). It is possible that thiamine deficiency is more widespread in Pacific salmon and steelhead than previously reported in the literature simply because samples have not routinely been collected and analyzed.

Our egg thiamine data contained some interesting patterns of variation among populations, regions, and years. Our data strongly suggest that these patterns are due to variable ocean diets and that ocean diets are linked to spatial-temporal overlap with different prey. Specifically, thiamine deficiency was widespread for CCV Chinook salmon in 2020 and 2021, substantial for Klamath and Trinity River coho salmon in 2021, but relatively rare for Klamath and Trinity River Chinook salmon in these years. These differences may reflect the different “last feeding” seasons and/or different ocean foraging areas.

Egg thiamine concentrations in our samples mostly declined from 2020 to 2021, and the percent of thiamine deficient samples from the CCV, Klamath, and Trinity rivers increased (Fig. 3). This temporal pattern may reflect increased exposure to anchovy-dominated foraging areas that expanded northward to Cape Mendocino in summer 2021 and increased the overlap with historical summertime spatial distributions for all of California’s Chinook salmon (45) and coho salmon populations (46).

Salmon egg δ¹⁵N values at spawning are likely related to diet during the final few months of ocean foraging. Our data show that populations with eggs with enriched δ¹⁵N values had consistently lower thiamine concentrations. Eggs from CCV winter-run Chinook salmon had the largest fraction of thiamine deficient samples and the highest δ¹⁵N values in both years (SI Appendix, Table S3). This population is known to have the most southerly ocean distribution of California’s Chinook salmon (47, 48), where their foraging areas were dominated by northern anchovies in 2019–2021. The higher δ¹⁵N values are consistent with a diet rich in anchovy, which have higher δ¹⁵N values and trophic position (TP ~ 3.1) relative to krill (TP ~ 2.2) in Central California (49). The more northern Klamath and Trinity River Chinook and coho salmon populations have lower δ¹⁵N indicating a diet composed of lower trophic level prey, consistent with greater dietary contributions of items such as krill, juvenile rockfish, and crab megalopae. As these northern populations are farther removed from the distribution of anchovy, the patterns observed in the δ¹⁵N data and inferred diet help explain the spatial and temporal variability in Chinook and coho salmon egg thiamine concentration. While changes in δ¹⁵N baselines across the region and between 2020 and 2021 may have occurred due to factors such as variability in upwelling and nitrate utilization (49–51), it is unlikely that these shifts are great enough to confound our interpretation that the relationship between δ¹⁵N and egg thiamine concentrations are due to diet differences. Future research integrating stable isotope analyses in archival tissues and compound specific approaches can provide further insights into temporal and spatial linkages between shifts in specific salmon prey and egg thiamine.

Egg thiamine concentrations for different populations may also reflect population-specific exposures to stressors that vary by year, watershed, or season (52). For instance, Yukon River Chinook salmon egg thiamine concentration has been reported to decline with migration distance (31), and CCV and Klamath and Trinity River salmonids use different estuaries and river migration corridors that can present a variety of potentially important stressors that may impact thiamine status at spawning time. Additionally, recent research has indicated that river gravels could be a source of microbially derived thiamine, and differences in microbial communities between watersheds or spawning areas may result in thiamine availability changes to salmon (53).

Salmon Diets, Ocean Prey Fields, and Prey Nutrition.

Multiple lines of evidence point to an unusual dominance of northern anchovy in Central California’s forage fish community and in Chinook salmon diets in 2019–2022. Anecdotal reports indicated anchovy-dominated Chinook salmon diets for fish caught in central California’s 2019 ocean fisheries. Gut content data from 2020–2022 ocean fisheries were also anchovy-dominated, a diet composition that reflects the lowest diversity in prey items ever reported for California Chinook salmon. From 1955 through 2007, northern anchovy was an important prey for California Chinook salmon, but other important prey items included YOY rockfish, krill, Pacific sardine, Pacific herring, market squid, and crab megalopae; there was also a decline in prey diversity with an increasing importance of northern anchovy and Pacific sardine up to the last sampling year in 2007 (34).

While our gut content data are compelling, they are also limited in space and time compared with the ocean distributions for the populations of interest. For instance, we have no gut content samples from November through April, and we are comparing our sampling with historical data collected from May to August. Our gut content sampling is likely dominated by the numerically dominant CCV fall-run Chinook salmon population complex that is known to aggregate in the Central Coast sampling areas and seasons from which our data were collected.

Ecosystem surveys indicated a dramatic shift in the marine forage base dating back to 2014, with a rapid expansion in the biomass and extent of northern anchovy dominance in 2014–2021 (36) (SI Appendix, Fig. S2). A decline in forage base diversity was also notable in springtime RREAS trawl catch from 2018–2021 (54). 2019–2021 was distinct from an earlier period of high coastal pelagic species abundance and very low abundance of other forage taxa (krill, cephalopods, and YOY groundfish) during the delayed seasonal upwelling conditions of 2005–2006 (SI Appendix, Fig. S3). A key difference between these two periods was the relatively high abundance of both anchovies and sardines during the earlier event (54, 55), whereas in the recent event sardines were at very low levels and anchovy abundance was far greater.

It is also notable that two years in the California Current System impacted by major El Niño events had key elements of the 2019–2022 ocean food-web. Specifically, Chinook salmon gut content data were anchovy-dominated in 1983 (with very low diet diversity), and the 1998 RREAS survey data had especially low krill proportions and low prey diversity, though in 1998 sardines and anchovies were about equally represented. The 2019–2022 period did not include a strong El Niño event; it began with a modest intensity El Niño event that ended in mid-2019, then had La Niña conditions from mid-2020 through the end of 2022 (54).

An anchovy-dominated diet could explain thiamine deficiency in California salmon. Our nutritional analysis found that anchovy samples were exceptional for having both the highest lipid fraction, lowest thiamine, and highest thiaminase activity levels of the five prey species sampled. The average anchovy thiaminase activity of 35 nmol g⁻¹ min⁻¹ in our samples was more than sufficient to lower thiamine concentrations to critical levels in its consumers when it was the dominant prey item (14). Moreover, the effects of a thiaminase-rich anchovy-dominated diet on thiamine deficiency may have been amplified by the low thiamine and high lipid content of anchovy. For example, a low thiamine and high lipid sprat (Sprattus sprattus)-dominated diet has been correlated with thiamine deficiency in Baltic Sea Atlantic salmon (28). The combination of a lipid-rich and thiamine-poor diet can lead to oxidative stress that depletes thiamine, as thiamine can act as an antioxidant (28). For thiamine deficiency in California Chinook and coho salmon, the high thiaminase activity, low thiamine, and high lipid levels we found in anchovy are thus consistent with both the thiaminase and oxidative stress hypotheses. Thiamine deficiency in Yukon River Chinook salmon may be linked to a combination of warming river temperatures that increase metabolic demand for thiamine during spawning migrations and increased feeding on Bering Sea or Subarctic North Pacific thiaminase-positive prey (31, 44). High latitude Chinook salmon thiaminase-positive prey include herring, rainbow smelt, and capelin (31).

In contrast, the low-lipid, low thiaminase activity, and higher thiamine concentrations we measured in krill, herring, juvenile rockfish, and market squid indicate that a lack of these historically frequent prey items in California Chinook salmon diets may also contribute to thiamine deficiency. The results of our stable isotope analysis were consistent with this possibility, showing that individuals feeding on lower trophic level prey (like krill, crab megalopae, and juvenile rockfish) did indeed have higher egg thiamine concentrations. Thiamine degradation by the thiaminase I enzyme can be inhibited by some cosubstrates (56), and it is possible that other important nutrients are provided by prey taxa that have been scarce in recent salmon diets.

Management Implications.

Like many populations of Pacific salmon and steelhead along the US West Coast, California Chinook and coho salmon have suffered from myriad and well-known impacts and stressors to their populations and habitats (57). Numerous management actions have been taken to mitigate or reverse these impacts via a variety of recovery plans and conservation legislation. The recent emergence of thiamine deficiency is a worrisome addition to the stressor list. CCV winter-run and spring-run Chinook salmon, listed as Endangered and Threatened, respectively, are protected by the US Endangered Species Act due to their elevated risk of extinction (58). If high levels of thiamine deficiency persist for a prolonged period, populations of these protected species could be extirpated or the conservation unit (e.g., winter-run Chinook salmon Evolutionarily Significant Unit) could go extinct (58). California’s fall-run Chinook salmon are a major component of US West Coast salmon fisheries, and a prolonged decline in their productivity is already having serious impacts on these fisheries (59). Thiamine deficiency differs from other stressors afflicting salmon in several ways that are directly and easily observable: Affected fry exhibited unusual behavior and subsequent rapid mortality in hatcheries (32), and fishery surveyors saw highly unusual numbers of dead and moribund fry in downstream migrant traps.

Large-scale adaptive management experiments could yield valuable insights into the impacts of thiamine deficiency on salmon population dynamics and the efficacy of treatments. Specifically, using different coded-wire tag codes or parentage based genetic tags that allow for distinguishing among thiamine treatment and control release groups would allow for evaluating impacts on differential egg to adult survival rates. One such experiment was initiated at Coleman National Fish Hatchery (CNH) with their late-fall-run Chinook salmon population in brood year 2024. Because most CCV late-fall-run Chinook salmon mature at ages 3 and 4, most adult returns (and ocean captures in fisheries) will happen in 2027 and 2028. Such experiments could be done with multiple hatchery and natural populations over multiple years to better understand the impacts of thiamine deficiency and treatments on adult-to-adult salmonid productivity rates.

Field studies and captive rearing studies could yield important insights into mechanisms and impacts of thiamine deficiency in salmonids. For instance, nonlethal ocean sampling (to collect gut contents, and muscle plugs or blood samples for thiamine analysis), tagging, and releasing Chinook salmon caught in the ocean, then tracking their return migration with acoustic receivers could yield insights into links among ocean diet, thiamine status, and adult migration movement and survival rates. A captive feeding experiment using a variety of diet treatments could inform the diet link to egg thiamine status for Chinook salmon the way such experiments have for captive lake trout (14).

Early warning of imminent thiamine deficiency could help guide interventions. Systematic monitoring of egg thiamine levels in fisheries targeting maturing adults near mouths of rivers to which they are returning might provide such a warning. If the anchovy-diet hypothesis holds up, existing seasonal surveys that document the forage community composition and distribution, such as NOAA’s springtime Rockfish Recruitment or Ecosystem Assessment survey or summertime California Current Ecosystem Survey, could alert managers to conditions favorable to thiamine deficiency in Chinook salmon. Routine monitoring of salmon stomach contents from fish caught in the ocean also provides valuable information, albeit only when ocean fisheries are open.

Because thiamine is needed for growth, maturation, and neurological coordination, thiamine deficiency may also have more subtle, sublethal effects that warrant further consideration and investigation. For example, models used to forecast abundance of salmon are typically based on relationships between the numbers of salmon caught or returning at different ages that depend on maturation and survival rates (60). TDC-driven changes in these rates could degrade the performance of forecasts and potentially lead to overfishing. Furthermore, prolonged increases in natural mortality caused by reduced disease resistance or increased vulnerability to predators due to reduced swimming performance, as has been observed in Atlantic salmon (40) and in models of lake trout (61), could necessitate changes in fishery management reference points to ensure fishery sustainability.

Ecosystem-based management actions could conceivably mitigate thiamine deficiency in salmon, but we need to learn much more about its proximate and ultimate causes. If the proximate cause is a diet composed predominantly of thiaminase-containing anchovy, management actions aimed at increasing prey diversity might be warranted. Directed actions aimed at controlling anchovy biomass or distribution have been considered and judged unlikely to be wise or successful owing to factors that include the unknown controls on past anchovy boom-bust cycles, the important role that anchovy play in ocean food-webs, stakeholder opposition, and a lack of fishery-capacity for responding to rapid anchovy abundance expansions and collapses (62).

The ultimate cause of thiamine deficiency may lie in changes closer to the base of the food chain, perhaps driven by physical or chemical changes in the California Current Ecosystem. For example, the shift in dominance between Pacific sardine and anchovy within the California Current is thought to be linked to shifts in the size structure of plankton assemblages which are in turn driven by shifts in wind-driven upwelling (63), with some evidence for similar shifts in the pelagic ocean microbiome (64). While it is hard to imagine how managers could address such changes, a better understanding could address the question of whether thiamine deficiency is likely to be ephemeral or persistent and perhaps increase in severity and spatial extent.

Finally, thiamine deficiency could be affecting other taxa that feed extensively on anchovy. Some piscivores, like marine mammals and birds, are both susceptible to thiaminase-driven thiamine deficiency and display clinical signs of thiamine deficiency, and could therefore be easily noticed by researchers and the public (52). At the time of this writing, we are not aware of reports noting thiamine deficiency impacts on West Coast species other than salmonids.

Across multiple systems, prey quality and diet shifts are common factors linked with thiamine deficiency in predators. However, there may be different combinations of factors in different systems affecting thiamine supply, transfer or demand. In the Laurentian Great Lakes, the proximal explanation for thiamine deficiency in salmonids points to a shift to an alewife-dominated (and thiaminase rich) prey base (25, 27). Alternatively, many Baltic Sea researchers suggest that a shift to a diet that is low in thiamine and high in lipids reduces the thiamine supply and increases thiamine depletion via increased oxidative stress in Atlantic salmon (28, 30). Thiamine deficiency in Yukon River Chinook salmon may be linked to a combination of warming river temperatures that increase metabolic demand for thiamine during spawning migrations and increased ocean feeding on thiaminase-positive prey (31, 44). Our results link thiamine deficiency to an anchovy-dominated diet that features high thiaminase, high lipids, and low thiamine. Moreover, the switch to an anchovy-dominated diet coincided with the rapid emergence of an anchovy-dominated forage base in the Central California Current System. Alternatively, research has identified that thiamine is associated with benthic microbial communities in aquatic ecosystems, that net production varies across space and time, and that net production is likely modulated by anthropogenic factors (53). Furthermore, trophic transfer of thiamine through food webs shows evidence of biodilution (65) and may be disrupted by filamentous cyanobacteria (66). Collectively, the evidence from these studies point to at least three possible mechanisms of thiamine deficiency in wildlife: i. changes in prey quality (e.g., prey containing higher thiaminase or lipid content, or lower thiamine), ii. increased metabolic demand for thiamine due to altered stressors, and iii. changes in thiamine production and transfer within food webs and ecosystems. Thiamine deficiency as a consequence of multiple stressors deserves further research.

The observations reported here, as well as those from the Baltic Sea and North America’s Great Lakes and Finger Lakes, raise numerous questions whose answers could influence the management response to this phenomenon. We hope to pursue some of them in the context of Pacific salmonids in California: How much thiamine is available to the food web as aqueous and particulate-bound forms, and how does that vary spatially and temporally? Does the thiamine produced by aquatic microbial communities influence levels of thiamine in salmon? When and where are salmon acquiring thiamine deficiency? What are its sublethal effects on growth, maturation, reproduction, and susceptibility to disease and predation? How long do transgenerational effects last and how do they manifest? Does susceptibility to thiamine deficiency vary among populations and species of Pacific salmonids, or over time due to mitigating or aggravating factors, and if so, which factors? By better understanding the causes and consequences of thiamine deficiency in California salmon, we hope to gain insight into whether this is an ephemeral phenomenon or an emerging chronic stressor.

Materials and Methods

Thiamine Deficiency Impacts on Salmon.

Early fry mortality experiments.

We generated data linking egg thiamine concentrations to fry behavior and survivorship with laboratory studies using eggs and embryos from four CCV Chinook salmon populations. Unfertilized and fertilized Chinook salmon eggs were collected from CCV spring-run and winter-run Chinook salmon in 2020, fall-run in 2021, and late-fall-run in 2022 (Table 1). A subset (10 g) of unfertilized eggs were immediately frozen on dry ice to preserve thiamine and were kept frozen at −80 °C until shipped on dry ice to State University New York (SUNY) Brockport for thiamine analysis (SI Appendix). Fertilized eggs were incubated in heath trays. Upon reaching the eyed-egg stage, embryos were counted by hand and, for all but fall-run Chinook salmon, transferred to University of California Davis for evaluation of fry survival and behavioral signs of TDC from ~80 to 120 d postfertilization (dpf). Fall-run Chinook salmon embryos were transferred to high school classrooms for student evaluation of fry survival and behavioral signs of TDC to ~120 dpf as part of the Spinning Salmon in the Classroom project (67). Embryos were transferred to flow-through aquaria with a temperature-controlled water bath. Embryo mortalities were removed and not considered in our fry survival calculations. Posthatch, fry mortalities were counted and removed from each tank and any observations of behaviors associated with TDC were recorded regularly. These egg thiamine concentration and fry behavior and survival data were then used to define TDC impacts criteria and develop a model for predicting TDFM for California Chinook salmon of all run types. Additional details of the egg collections, fry rearing, and data collection methods are provided in the SI Appendix.

Thiamine treatment experiments.

We evaluated the efficacy of thiamine supplementation via adult prespawn injections (Table 2). Prespawn adult injections were given to Livingston Stone (winter-run Chinook salmon) females in 2020 and 2021, and FEH (spring-run Chinook salmon) males and females in 2021. Details of the thiamine supplementation methods are provided in the SI Appendix.

Table 2.

Central Valley Chinook Salmon (Oncorhynchus tshawytscha) thiamine supplementation and control groups

Population Source	Brood Year	Treatment group	Control group
Livingston Stone Hatchery winter-run	2020	adult females were given a prespawn thiamine injection (n = 33)	adult females were given a prespawn saline injection (n = 28)
Livingston Stone Hatchery winter-run	2021	adult females were given a prespawn thiamine injection (n = 49)	Untreated (no saline injections; n = 31)
Feather River Hatchery spring-run	2021	adults (n = 4582, males and females) were given a prespawn thiamine injection, 2424 returned, 30 females were sampled	adults (n = 211, males and females) were given a prespawn saline injection, 115 returned, 30 females were sampled

Egg thiamine surveillance at hatcheries.

Unfertilized (and untreated) eggs were collected from adult salmonids at spawning to assess thiamine concentrations from 16 different populations at seven different hatcheries from January 2020 to December 2021. CCV hosts winter-run, spring-run, fall-run, and late-fall-run Chinook salmon (68). The Klamath River and Trinity River host fall-run and spring-run Chinook salmon, and coho salmon. Ecotypes are named for the season in which adults return to freshwater (68). Chinook salmon eggs were collected at Iron Gate Hatchery (IGH), Trinity River Hatchery (TRH), Livingston Stone National Fish Hatchery (LSH), CNH, Feather River Fish Hatchery (FEH), Nimbus Fish Hatchery (NIH), Mokelumne River Hatchery (MOH), and Merced River Hatchery (MEH), all located in California (Fig. 2 and SI Appendix, Table S1). Coho salmon eggs were collected from TRH and IGH each year. During spawning, ~10 grams of eggs were collected from approximately 30 females from each hatchery population group in each year. Eggs were immediately frozen on dry ice to preserve thiamine and were kept frozen (−80 °C) until shipped on dry ice to SUNY Brockport for thiamine concentration analysis.

Approach for estimating TDFM: Population level impacts of TDFM in California’s Chinook salmon were estimated using a simulation-based approach as field-based observations were not available. First, we fit a dose–response model (described below) to the previously described laboratory data linking egg thiamine concentration to fry survival. Second, we applied the dose–response model to measured egg thiamine concentrations from each hatchery sampling event. We assumed that the egg thiamine samples from each hatchery population provided a reasonable surrogate for the companion natural and hatchery spawning populations in the watershed that brood year.

Input data for the dose–response model consisted of ~18,000 embryos spawned from 135 females whose unfertilized eggs were measured for thiamine concentrations as outlined above. The outcome of interest was the proportion of embryos surviving in each group to the study endpoint as fry at 80 to 120 dpf. We used a 4-parameter sigmoid dose–response model to represent the fry survival rate response to egg thiamine concentration:

$$Yi=U+[L-U]/(1+[C/EC50i{]}^S,$$

where: Y_i is fry survival (proportion) for family group i; U and L are the upper and lower limits of survival (proportion), respectively; C is the observed thiamine concentration (nmol g⁻¹); EC50 is the thiamine concentration at which the proportion of survival is estimated to be ½ of U for family group i, and S is the slope of the relationship.

The dose–response model was fit in a Bayesian framework. As the outcome of interest was the proportion of embryos surviving to the study endpoint, we assumed a binomial likelihood that was estimated using a Metropolis Markov Chain Monte Carlo sampling routine. As survival is expected to be zero when thiamine concentration is zero, we further constrained the 4-parameter dose–response model to a 3-parameter model fitting U, EC50, and S. To reflect the assumption that different egg families may have different sensitivity to thiamine with regard to survival and account for overdispersion, we allowed EC50 to be represented by a distribution rather than a single value. Specifically, we used a hierarchical model and assumed EC50 was normally distributed, with mean μ and standard deviation σ. We only represented EC50 with hyperparameters (μ and σ) as including additional parameters (e.g., to allow for among-family variation in U) resulted in model convergence issues. Each parameter was fit assuming noninformative uniform priors over 4 chains, each 100 K in length. Convergence was assessed using the Gelman–Rubin diagnostic (i.e., R hat ≤ 1.01) after a 10% burn-in period and a thinning interval of 2. All analysis was performed in R v4.3.2 (69) using the RJAGS package 4-14 (70). Additional details describing our modeling approach are provided in the SI Appendix.

We estimated TDFM in untreated spawning populations (either hatchery or natural) in the same brood year by applying the fitted dose–response model to measured egg thiamine concentrations for a given Chinook salmon hatchery sampling event. Specifically, the empirical egg thiamine concentrations for each sample group (identified by the combination of hatchery, run type, and year) were repeatedly and randomly sampled with replacement 10,000 times to represent the variability in egg thiamine concentrations. These samples of egg thiamine concentration were then applied to the dose–response model under conditionally resampled model parameters 10,000 times. These thiamine and model parameter samples were then used to estimate and provide uncertainty around TDFM predictions in terms of percentage mortality [i.e., TDFM = (1 − proportion survival) × 100].

Ocean salmon prey and nutrition.

Ocean-caught salmon diets.

Stomachs from adult Chinook salmon caught in the ocean between Monterey, CA, and Point Arena, CA were collected by commercial and recreational fishers in 2020 (June–October, N = 357), 2021 (June–September, N = 247), and 2022 (June–September, N = 153). Salmon stomachs were removed intact and frozen soon after capture. These were later thawed and contents removed, rinsed, examined, and sorted (34). Stomach contents were identified to the lowest taxonomic level possible. Individual prey item counts, lengths of intact specimens and total wet mass of each taxon was measured and recorded. We calculated percent biomass of prey, by species, for each sampling month and year. For comparison, percent volume of prey in Chinook salmon stomachs collected in central California ocean fisheries in (May–August) 1955, 1980–1986, and 2005–2007 were obtained from (34).

Shifts in prey community over time and nutritional status of salmon prey.

We used data from two biological surveys conducted in the California Current Ecosystem to quantify the composition and distribution of the ocean forage base for California salmon. We obtained summertime small pelagic fish abundance and distribution from 2017–2021 collected on NOAA’s California Current Ecosystem Survey from (36). This survey targets coastal pelagic species (such as northern anchovy, Pacific sardine, Pacific mackerel, jack mackerel, and Pacific herring) using a large surface trawl towed at high speeds in summer and fall seasons. We report overall biomass trends, and the proportion of anchovy biomass, by weight, for haul clusters along cross-shelf transect lines.

We collected data on springtime forage community structure and obtained prey samples from the RREAS, a midwater trawl survey for pelagic YOY rockfish (Sebastes spp.) and other forage species that has been conducted annually off of central California since 1983, and coastwide along the US West Coast since the early 2000s (35, 71). The RREAS occurs annually from mid-April to mid-June and uses a modified Cobb midwater trawl with a fine mesh codend liner to more efficiently sample smaller forage taxa such as pelagic juvenile fishes and krill. Relative biomass indices and associated uncertainty estimates were developed using a model-based approach from net haul data (54, 71). Additional details for the prey sampling and analysis are provided in the SI Appendix.

Nutritional analysis.

Thiamine (including free thiamine—TH, thiamine monophosphate—TMP, and thiamine pyrophosphate—TPP) was extracted in duplicate from salmon eggs and prey using the same methods (described in SI Appendix, section S1). We limited our prey nutritional analysis to five of the most frequently observed taxa from an investigation of California Chinook salmon gut contents from 1955–2007: euphausiids (krill, of which the most frequently occurring species are Thysanoessa spinifera and Euphausia pacifica) market squid (Doryteuthis opalescens), northern anchovy, YOY rockfish (Sebastes spp.), and Pacific herring (Clupea pallasii) (34).

Thiaminase activity in prey items was measured at the U.S. Geological Survey, Columbia Environmental Research Center in Columbia, Missouri, using an indirect method that relies on measurement of a nucleophilic cofactor, 4-nitrothiophenol (4-NTP) (72). Briefly, samples were homogenized on dry ice, 0.5 g of homogenate was suspended in phosphate buffer (1.25 mL of 100 mM, pH 6.5), vortexed, and centrifuged (16,000 x g for 10 min). Supernatants were assayed in Tris(2-carboxyethyl) phosphine hydrochloride (TCEP, 10 mM) buffer, 4-NTP (80 µM), both with and without thiamine (400 µM) in 96-well microtiter plates. Enzymatic reactions were initiated by the addition of the sample, monitored for absorbance at 411 nm, and run 60 min in a Synergy HT Multi-Detection Microplate Reader (BioTek Instruments, Winooski, VT). Activity was determined by linear regression (Microsoft Excel software) of the change in absorbance over time, corrected for nonthiaminase activity by the difference in slopes (with and without thiamine added), converted to concentration using the molar extinction coefficient for 4-NTP (13,650 M⁻¹cm⁻¹), and reported as nmol/min/g-sample. All assays were performed in duplicate. Solution blanks and positive control samples (Paenibacillus thiaminolyticus) were run with each plate of samples as a QA/QC check.

Lipids were extracted from each prey sample (1 g) in 2:1 chloroform/methanol solvent containing 0.01% butylated hydroxytoluene, which was used as an antioxidant. The organic solvent was evaporated under nitrogen gas and total lipid content was determined gravimetrically (73). Total lipid contents are expressed in % of wet weight.

Stable isotope analysis.

Stable isotope analysis was done on subsamples of Chinook and coho salmon eggs run for thiamine analysis (SI Appendix, Table S3). Samples were freeze dried, pulverized, encapsulated, and analyzed for nitrogen isotopes using a Costech ECS 4010 Elemental Analyzer interfaced with a Thermo Delta V Advantage Continuous Flow Isotope Ratio Mass Spectrometer at the Idaho State University Stable Isotope Laboratory. The data are reported in standard delta notation (δ¹⁵N) relative to the atmospheric N₂ reference standard. Analytical precision, calculated from analysis of standards distributed throughout each run, was ≤±0.2‰. Trophic enrichment (Δ¹⁵N) for Chinook salmon was recently established to be 3.5 ± 0.3‰ for muscle (74). For capital breeders such as Pacific salmon (75) it has been suggested that the δ¹⁵N of eggs should be similar to long-term, slow turnover tissues such as muscle (76). Isotope data for Pacific salmon eggs are sparse, but relatively small δ¹⁵N changes during migration and spawning are supported by the available data: Values of “fresh” fish, carcasses, and eggs are similar, within 1‰ (77).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Egg thiamine concentration data have been deposited in EDI (https://doi.org/10.6073/pasta/ce132c1ef916900b5b50650388aa2266). All study data are included in the article and/or supporting information. Previously published data were used for this work (78).