Pacific Cod metabolism and swimming performance are similar across temperatures following prolonged thermal acclimation

Abstract

Warming ocean temperatures can increase the metabolic rates of fishes, potentially contributing to changes in their growth and survival to recruitment age. During prolonged marine heatwave conditions in the Gulf of Alaska between 2014 and 2019, Pacific Cod (Gadus macrocephalus) metabolic rates may have increased, but little is known about the relationship between metabolism and temperature for immature individuals of this species. We examined the effect of prolonged temperature exposure (~1 year) on the performance (standard, routine, and maximum metabolic rates; critical swimming speed; and aerobic scope) and swimming efficiency (cost of transport and optimal swimming speed) of age-1 Pacific Cod during two laboratory experiments across a range of temperatures (Expt. 1: 2°C, 4°C, 6°C and 8°C; Expt. 2: 6°C, 10°C and 14°C). We also explored relationships between performance and additional body state variables (e.g. condition and growth) and environmental variables (e.g. photoperiod and salinity). Temperature did not influence baseline metabolic performance (standard and routine metabolic rates) in either experiment. However, we observed significantly higher baseline metabolic rates in Expt. 2 compared to Expt. 1, even at the same temperatures. In contrast, maximum performance metrics (e.g. maximum metabolic rate and critical swimming speed) were significantly influenced by temperature. These patterns in performance were generally explained by differing costs of transport and rates of oxygen consumption during swimming trials between the two experiments. Further, body state variables and environmental variables were poorly correlated with performance, even when combined in a multivariate framework. Together, these findings suggest that other factors, such as season, oceanographic conditions early in life, year-class effects, or epigenetic effects, may influence Pacific Cod metabolism more than temperature or measured body state variables and environmental variables following prolonged thermal acclimation.

Introduction

Global climate change can lead to long-term and rapid changes to marine ecosystems through increasing ocean temperatures and more frequent and intense marine heatwaves, which are defined as prolonged periods of anomalously warm regional water temperatures (Perry et al., 2005; Nagelkerken and Connell, 2015; Hobday et al., 2016; Suryan et al., 2021). Under the tenets of the metabolic theory of ecology (Gillooly et al., 2001; Brown et al., 2004), the metabolic rates of marine ectotherms will increase with temperature because these animals cannot internally regulate their body temperature. Consequently, these animals must either acclimate to the changing thermal environment (Crozier and Hutchings, 2014) or relocate to more suitable habitat (Habary et al., 2017; Howard et al., 2020) to reduce thermal stress. For marine fishes, elevated metabolism may contribute to accelerated maturation at smaller sizes (Cheung et al., 2013), changes in reproductive output (Pettersen et al., 2018), energetic tradeoffs in foraging (Bruno et al., 2015) and shifts in biogeographic range limits (Slesinger et al., 2024), each of which has potential to influence recruitment and survival. However, metabolic responses to temperature can be highly variable across species (Killen et al., 2017), life stages (Gauldie, 1996) and life history strategies (Johansen et al., 2021), and there can be substantial individual variability (Burton et al., 2011). Thus, it can be challenging to interpret changes to fish metabolic rate holistically in the context of climate change.

Metabolic rates of fishes can be estimated using oxygen consumption rates (Chabot et al., 2016; Svendsen et al., 2016), which can then be used to calculate a variety of performance metrics (Table 1). These metrics can vary in their thermal sensitivity and provide information on how fish perform across a range of temperatures (e.g. reaction norms; Kingsolver et al., 2004; Hurst et al., 2012). Baseline measures of metabolic performance (Chabot et al., 2016) quantify the minimum oxygen consumption necessary to maintain cellular homeostasis (standard metabolic rate [SMR]) or low levels of activity (routine metabolic rate [RMR]), while measures of maximum performance, such as maximum metabolic rate (MMR) and critical swimming speed (U_crit), approximate the upper performance limits of an individual (Killen et al., 2017; Cano-Barbacil et al., 2020). In response to thermal stress, both baseline and maximum performance tend to increase with temperature up to a thermal optimum (Bruno et al., 2015; Schulte, 2015; Tirsgaard et al., 2015). However, metrics of maximum performance generally increase more rapidly with temperature than baseline metabolism (Norin et al., 2014). Changes to both the metabolic “floor” and “ceiling” at elevated temperatures can influence aerobic scope, which is the amount of energy above the baseline that is available for foraging, growth, reproduction and other daily activities (Clark et al., 2013). Absolute aerobic scope (AAS), representing the net difference between maximum and baseline performance, generally increases with temperature to a maximum and then declines as the circulatory and ventilatory systems reach their capacity to deliver oxygen to respiring tissues (Pörtner and Knust, 2007; Pörtner et al., 2017; but see Clark et al., 2013). In contrast, factorial aerobic scope (FAS), which represents the ratio of maximum to baseline performance, generally declines as temperatures increase (Halsey et al., 2018; Deutsch et al., 2020).

Table 1.

Definitions, abbreviations, and units for terms used in this study

Term	Definition
Oxygen consumption (MO2)	Oxygen consumption rate of an individual, calculated as the slope of the linear decrease in oxygen content over time in a swim tunnel respirometer (Equation 1)
Q 10	The factorial change in performance for a 10°C change in temperature, calculated from Equation 6
Performance metric	A term that broadly encompasses the specific metabolic and swimming traits explored in this study that relate to an individual’s ability to perform during a swimming trial, including SMR, RMR, MMR, critical swimming speed, FAS, AAS, COT and optimal swimming speed
Baseline performance
SMR	Minimum MO2 required for a fish to maintain homeostasis. SMR is calculated as the y-intercept of the linear model log(MO2) = β0 + β1(swim speed) during the swim trial, representing theoretical oxygen uptake at zero speed. SMR values were excluded if the R2adj from the model was <0.4
RMR	Oxygen consumption for a fish during a period of low activity, calculated as the 20% quantile for all MO2 measurements during the overnight period prior to the swim tunnel trial
Maximum performance
MMR	Maximum MO2 rates attained by an individual, calculated as the mean of the three highest MO2 measurements during the swim trial
Critical swimming speed (Ucrit)	The point in a swim trial at which a fish can no longer swim against the current, calculated from Equation 2
Aerobic scope
FAS	MMR/RMR. The ratio of the difference between maximum and minimum oxygen consumption for an individual
AAS	MMR-RMR. The net difference in maximum and minimum oxygen consumption for an individual
Swimming efficiency
COT	An index of swimming efficiency, calculated as the energy required to swim at a certain speed. Lower values indicate more efficient swimming
Optimal swimming speed (Uopt)	The speed at which a fish swims most efficiently, calculated as the minima of the relationship between COT and swimming speed using Equation 4
Body state variables
HSI	Body condition metric, measured as [liver wet weight (g)]/[whole body wet weight (g)] * 100
Liver lipids (mg g−1)	Body condition metric, measured as the milligram of lipid per gram of liver tissue. Only available for Expt. 1
Muscle lipids (mg g−1)	Body condition metric, measured as the milligram of lipid per gram of muscle tissue. Only available for Expt. 1
Absolute growth rate (mm d−1)	The difference in standard length at the time of swim trial and standard length approximately 4–7 months prior (Expt. 1) or 3–6 months prior (Expt. 2), divided by the number of days elapsing between the two measurements
GSI	Relative gonad mass, measured as [gonad wet weight (g)]/[whole body wet weight (g)] * 100
Parasite wet weight (g)	Wet weight of parasites (in grams) if present in an individual. Parasite wet weights were sampled opportunistically and only quantified in Expt. 2
Days since last feeding	Number of days between last feeding and the day of the swim trial (ranges between 2 and 5 days)
Environmental variables
Rearing temperature (°C)	Temperature (in °C) of rearing tank and swim tunnel, averaged over the final 3 months of the rearing period
Salinity ratio	Ratio of salinity on the day of the swim trial, divided by the average salinity experienced over the prior 3 months
Photoperiod	Number of hours of daylight for the rearing environment on the day the fish was acclimated to the swim tunnel. All fish were reared under a natural photoperiod for Kodiak Island, AK

Most fish metabolic studies occur after short periods of exposure (often less than 30 days) to experimental temperatures (Paul et al., 1988; Claireaux et al., 2000; Tirsgaard et al., 2015). However, new techniques for estimating metabolic rates in the field, such as the use of δ¹³C values in otolith carbonate (Chung et al., 2019; Martino et al., 2020; De Groot et al., 2024), often integrate metabolic rate over long periods of time (e.g. 2–12 months), necessitating a better understanding of how fish metabolism may shift in response to prolonged periods of thermal acclimation. Both the long-term warming associated with anthropogenic climate change and the distinctive periods of warming associated with marine heatwaves can lead to prolonged periods of elevated temperatures in the field, particularly because both long-term warming and marine heatwaves can lead to an earlier onset of summer conditions in the spring and the persistence of warm temperatures into the fall (Li and Donner, 2022; Wang et al., 2021). Some studies have shown that the baseline metabolic rates of fishes can change over time, with SMR and RMR decreasing in response to prolonged periods of thermal stress (Sandblom et al., 2016; Archer et al., 2021). A decline in baseline metabolic rate after prolonged exposure to elevated temperatures may be an acclimatory response to protect the organism’s aerobic scope, stabilizing AAS and FAS (Norin et al., 2014). Acclimation can occur over multiple generations or within a single generation (Donelson et al., 2012; Pilakouta et al., 2020), possibly in as little as 3–5 weeks following exposure (Johansen et al., 2021). Investigating fish responses to prolonged temperature exposure is critical for understanding acclimation capacity and accurately interpreting time-integrative metabolic proxies such as otolith δ¹³C.

Following two prolonged periods of marine heatwave conditions in the Northeast Pacific between spring 2014 and winter 2016 and for nearly a full year in 2019, the Gulf of Alaska Pacific Cod (Gadus macrocephalus) population declined by more than 75% in adult biomass, leading to the first-time closure of the federal fishery in 2020 (Peterson Williams et al., 2022). This population likely experienced severe metabolic stress during these events (Barbeaux et al., 2020; Laurel et al., 2021), with adults needing to consume an estimated 6–12% more food per day to maintain growth (Barbeaux et al., 2020). Elevated metabolism during the marine heatwaves may have also led the Gulf of Alaska population to move into the Bering Sea (Stevenson and Lauth, 2019; Spies et al., 2020) and potentially as far north as the Alaskan Arctic. Pacific Cod were reported in the Chukchi Sea for the first time in 2018, as the Gulf of Alaska and Bering Sea populations moved north in response to elevated temperatures throughout both regions (Cooper et al., 2023). To date, only two studies have examined the temperature dependence of adult Pacific Cod metabolic rate in the laboratory, with inconsistent results. Paul et al. (1988) demonstrated an approximately linear relationship between metabolism and temperature (3.5–12°C), while Hanna et al. (2008) found no relationship in a similar temperature range (4–11°C). Importantly, these studies did not examine metabolic rates following prolonged exposure (greater than 8 weeks) to temperatures consistent with those experienced during marine heatwaves in the Gulf of Alaska (e.g. 13–15°C; Thalmann et al., 2024).

In this study, we estimated the thermal sensitivity of performance metrics (Table 1), including measures of baseline metabolism (SMR and RMR), maximum performance (MMR and U_crit), aerobic scope (FAS and AAS) and swimming efficiency, including cost of transport (COT) and optimal swimming speed (U_opt) for age-1 Gulf of Alaska Pacific Cod following a prolonged acclimation (~1 year) to experimental temperatures. We ran two independent laboratory experiments extending over a temperature range of 2°C to 8°C (Expt. 1) and 6°C to 14°C (Expt. 2), with both experiments sharing a 6°C temperature treatment, to explore the impacts of prolonged acclimation across a range of temperatures. Due to high levels of individual variation in performance metrics, we also explored potential relationships between performance and other body state variables, such as condition and growth, and environmental variables, such as photoperiod and salinity. We hypothesized that all performance metrics would exhibit some degree of thermal sensitivity across the range of temperatures experienced in these two experiments, with elevated metabolic performance and decreased swimming efficiency at higher temperatures.

Materials and Methods

Fish collection and transport

Age-0 Pacific Cod were collected by demersal beach seine (36-m bag with 5-mm mesh; 1 m × 2.25 m wings with 13-mm mesh) from the nearshore waters of Cook Bay, Kodiak Island, AK, in mid-July 2019 (Expt. 1) and late August 2021 (Expt. 2) by the National Oceanic and Atmospheric Administration (NOAA) Alaska Fisheries Science Center (AFSC) Fisheries Behavioral Ecology Program (Laurel et al., 2007). Summer temperatures in these nearshore habitats were approximately 13.2 ± 1.1°C in 2019 and 10.6 ± 1.2°C in 2021. Annual collection permits were issued by the Alaska Department of Fish and Game to NOAA AFSC staff and collaborators. Fish were packed in 8-L plastic bags containing 2 L of seawater with 100% oxygen atmosphere (n = 3 fish/container) and shipped overnight in insulated boxes to the AFSC laboratory in Newport, OR. All animal procedures were approved by Oregon State University’s Institutional Animal Care and Use Committee (IACUC-2022-0262) and followed recommendations from the ARRIVE guidelines (Kilkenny et al., 2010).

Fish rearing

Pacific Cod were introduced to 380-L tanks and acclimated to experimental temperatures at a rate of 1°C per day for both experiments. At this time, all individuals were measured (standard length, mm, and wet weight, g) and tagged with 8-mm internal Passive Integrated Transponder (PIT) tags (BioMark). Fish were anaesthetized prior to tagging by immersion in seawater with 50 ppm MS-222 for ~1.25 minutes, and tags were inserted in the body cavity via needle injection. For Expt. 1, Pacific Cod were reared at four target temperatures (2, 4, 6, 8°C, n = 11 fish/tank) with one replicate tank per temperature (Table 2). For Expt. 2, fish were reared at three target temperatures (6, 10, 14°C, n = 10 fish/tank) with two replicate tanks per temperature. Fish were fed to satiation 2×/wk (Expt. 1) or 3×/wk (Expt. 2) using a gelatinized combination of squid, krill, herring, commercial fish food, amino acid supplements and vitamins (“gel food”; 52% protein, 14% fat), and fish were measured every 4 (Expt. 2) or 6 (Expt. 1) months. Swim trials for Expt. 1 occurred between December 2020 and March 2021, 12–15 months after introduction into the rearing tanks, while swim trials for Expt. 2 occurred between April and September 2022, 6–11 months after introduction into the rearing tanks. For both experiments, tanks were maintained on filtered, flow-through (~150 mL s⁻¹) seawater pumped from Yaquina Bay, OR, chilled or heated to within 0.5°C of target temperatures. All fish were reared under a seasonal photoperiod representative of Kodiak Island, AK, throughout each experiment.

Table 2.

Expt. 1 and Expt. 2 were characterized by several differences in season, oceanographic conditions and fish age, as summarized here

Expt.	Experimental temperatures	Replicate tanks/treatment	Collection year	Oceanographic conditions	Season of trials	Dates of trials	Acclimation (days)	Age at trials (year)	Body mass (g) at trials
1	2°C, 4°C, 6°C, 8°C	1	2019	Marine heatwave	Winter	Dec 3, 2020, to Mar 5, 2021	375–467	1.5–2	95.0 ± 22.6
2	6°C, 10°C, 14°C	2	2021	Cooler conditions	Summer	Apr 15, 2022, to Sept 19, 2022	176–333	1–1.5	88.7 ± 15.5

Several differences were inherent between the two experiments despite attempts to maintain methodological consistency (Table 2). Expt. 1 swim trials occurred in winter (December–March), while Expt. 2 swim trials occurred in spring and summer (April–September). Fish in Expt. 1 were unintentionally collected during marine heatwave conditions in the Gulf of Alaska in 2019, with temperatures greater than 3°C above average for the coastal nearshore (Amaya et al., 2020), while the fish in Expt. 2 were collected during a cooler year, 2021. Further, to test fish at similar sizes (between 170 and 210 mm) across experiments, the fish in Expt. 1 required longer acclimation periods (375–467 days) than the fish in Expt. 2 (176–333 days). As such, Expt. 1 fish were ~1.5–2 years old at the start of swimming trials, while Expt. 2 fish were ~1–1.5 years old at the start of swimming trials. The relatively long acclimation periods in both experiments were necessary to limit the effect of body size on metabolic rate and swimming ability and to allow adequate material to accrete onto the otolith for a companion study examining the use of δ¹³C as a potential proxy for metabolic rate for this species (e.g. Chung et al., 2019; Martino et al., 2020). Age-1 Pacific Cod otoliths grow relatively slowly (Expt. 1: 4.7 ± 2.8 μm daily; Expt. 2: 6.1 ± 1.8 μm daily); therefore, an approximately yearlong rearing period was necessary to obtain enough material for otolith analyses, similar to other otolith δ¹³C studies on subarctic species (Chung et al., 2019).

Intermittent-flow respirometry

Intermittent-flow respirometry was used to measure fish oxygen consumption rates at the conclusion of their rearing period. For both experiments, swim trials occurred when fish were ~180-mm Standard Length (SL)to limit the effect of size on metabolic rate. Prior to respirometry, fish fasted for at least 48 hours to limit the influence of digestion on metabolic rate (Chabot et al., 2016). In Expt. 1, fish in the 8°C treatment reached the target size fastest and swam first, followed sequentially by the lower temperature treatments. Growth rates were more similar between temperature treatments in Expt. 2, so, to minimize bias associated with size variability, we alternated temperature treatments over the course of the respirometry period such that only a portion of the fish in a temperature treatment swam consecutively. For Expt. 2, we ran duplicate swim trials for two fish from each temperature treatment (n = 6) to compare metabolic metrics for the same individuals across two swim trials. After the first trial, these fish were returned to their rearing tanks and the second trial occurred approximately 1 week later. We calculated percent differences in swimming metrics between these two trials for the same individual. For these individuals, we used the second of the duplicate trials to provide performance metrics for subsequent analyses so that measurements at dissection (e.g. body condition, gonad development, etc.) were taken at the same time that performance was measured. However, the use of either the first or second of these duplicate runs did not significantly impact performance at any temperature treatment.

We used a one-chamber, acrylic 10-L, 110-V swim tunnel (Loligo Systems Product #SW10110) for respirometry. Prior to swim trials, we calibrated the water flow into the swimming chamber using a flow meter (Flowtherm NT, Loligo Systems), and fish were maintained at their designated swim speeds during trials via a motorized propeller incorporated into the chamber. Fish oxygen uptake (mg O₂ L⁻¹ s⁻¹) was measured over the entire period that the fish was held in the swim tunnel with AutoResp (v 2.3.0) software. We used a submersible, in-line recirculation pump (Eheim 1250), connected to the swim chamber via gas impermeable Tygon® tubing, to circulate dissolved oxygen within the respirometry chamber during the flush period of each measurement cycle. Water temperature in the chamber was maintained at experimental temperatures and was monitored using a Pt100 temperature sensor (Loligo Systems) connected to a Fibox 3 instrument, and oxygen in the chamber was monitored using a 100-mm dipping probe oxygen mini sensor (Loligo Systems). After approximately every 7–10 swim trials, oxygen probes were recalibrated, and the swim tunnel apparatus was drained and cleaned to reduce the impact of microbial respiration on swim trials. Periodic measures of background oxygen consumption in the empty respirometer were <1% of the fish’s RMR and were therefore considered negligible in both experiments.

For each swim trial, an individual fish was transferred from their holding tank into the swim tunnel and allowed to acclimate overnight for approximately 12 to 17 hours. During the overnight acclimation period, fish swam at a constant speed of 0.5 body lengths (BL) s⁻¹ to maintain a steady swimming rate. At the beginning of the swim trial, we increased flow rates to 0.6 BL s⁻¹, followed by incremental flow rate increases of 0.2 BL s⁻¹ increments for each speed thereafter. Swim trials were divided into 10-minute respirometry cycles consisting of a 4-minute flush period where oxygenated water was pumped through the system and a 6-minute measurement period where the pump was off and oxygen consumption in the chamber was measured. The first minute of the measurement period was omitted from subsequent calculations to buffer against potential lags in oxygen levels following the flush period. Fish swam at each speed for three 10-minute cycles before their swimming speed was increased (30 minutes total), and additional cycles were added if the fish exhibited any unusual behaviour, such as erratic movement or rotating in the tank. To minimize disturbance to the fish, the swimming apparatus was covered in insulated foam with a small window for observation. The swim tunnel trials were complete when the fish reached exhaustion, which was determined when the fish stopped swimming against the induced current and rested on the back wall of the swim tunnel for 10 seconds. At this point, the fish was removed from the apparatus and euthanized through an overdose of buffered MS-222 (500 ppm), following American Veterinary Medical Association Guidelines on Euthanasia (https://www.avma.org/sites/default/files/2020-).

Analysis of performance

Oxygen consumption rates (MO₂; mg O₂ kg⁻¹ h⁻¹) were calculated using the equation:

(1)

where δO₂/δt represents the slope of the linear decrease in oxygen content in the chamber per second for the final 5 minutes of the measurement period for each cycle, V_resp is the total respirator volume in litres, M_fish is the mass of the fish in kilograms, and 3600 is the constant used for unit conversions (Svendsen et al., 2016). Fish were assumed to be neutrally buoyant; therefore, density measurements of individuals were not used in the above equation (Svendsen et al., 2016). Following procedures by Chabot et al. (2021), we applied a 29-second rolling mean to all δO₂/δt measurements and retained slopes with an R²_adj ≥ 0.8.

We characterized metabolic performance for individuals in both experiments using a variety of performance metrics (Table 1). We calculated two measures of baseline performance: SMR and RMR. SMR values, representing oxygen consumption extrapolated to zero speed, were excluded from subsequent analysis if the R²_adj from the linear model between MO₂ and swimming speed was less than 0.40. These excluded fish represented 28% of all individuals in Expt. 1 (n = 12) and 30% of all individuals in Expt. 2 (n = 17). RMR was calculated as the lowest 20% quantile of oxygen consumption measurements during the overnight period prior to the swim trial (Chabot et al., 2016). RMR, as defined in this study, is calculated using methods used to determine SMR from resting animals in Chabot et al. (2016) and Clark et al. (2013). Our RMR values were approximately 31.7 ± 11.6% higher than our y-intercept-calculated SMR values across all temperature treatments in this study, which likely reflects the increase in metabolism from fish with slow swimming speeds compared to fish that are assumed to not swim at all. We also found similar variability in the SMR and RMR metrics when assessed across the individuals that were measured multiple times, suggesting that variation in RMR did not supersede variability in SMR. Measurements associated with the first 3 hours of the overnight acclimation period were omitted from RMR calculations due to variability in oxygen consumption as fish adjusted to the swimming chamber.

We also calculated two metrics of maximum performance: MMR, representing the mean of the three highest MO₂ values during the swim trial, and U_crit, representing the point at which the fish could no longer swim against the current. U_crit was calculated for all fish using the equation:

(2)

where U₁ represents the final successful swim speed, U₂ represents the swim speed at which the fish reached exhaustion, T₁ represents the time into the final swim speed increment at which the fish reached exhaustion and T₂ represents the total time of the swim speed increment (30 minutes) including the flush and the measurement periods (Brett, 1964).

Finally, we calculated two measures of aerobic scope: FAS, calculated as MMR divided by RMR, and AAS, calculated as MMR minus RMR (Clark et al., 2013). We used our estimates of RMR to calculate aerobic scope metrics instead of our estimates of SMR so that we could obtain aerobic scope values for all fish, including the individuals that were excluded from the SMR calculations. We acknowledge that this differentiation will influence our estimates of aerobic scope, leading to approximately 28.1 ± 14.3% lower AAS and 37.3 ± 18.7% lower FAS when using RMR instead of SMR in aerobic scope calculations (Clark et al., 2013). However, because our RMR estimates are calculated using established methods to determine minimum oxygen consumption for individuals swimming at slow speeds, the use of RMR should not lead to unreasonable conclusions regarding aerobic scope. Further, age-1 Pacific Cod are likely to swim constantly in the field, so aerobic scope measurements using RMR may be more ecologically relevant for this species.

To evaluate swimming efficiency, we calculated COT (J km⁻¹ kg⁻¹) at each swimming speed by converting MO₂ values to J km⁻¹ kg⁻¹ using an oxycalorific equivalent (3.24 calories mg⁻¹ O₂) and converting calories to joules (4.18 J cal⁻¹). This value was then divided by the corresponding swimming speed (Claireaux et al., 2006). For each temperature treatment within each experiment, we examined the relationship between COT and swimming speed by fitting nonlinear models with the equation:

(3)

where U is the swimming speed, and a, b and c are model coefficients (Claireaux et al., 2006). Optimal swimming speed (U_opt), defined as the minima of the relationship between COT and swimming speed where swimming was most efficient, was calculated using parameters from Equation 3 and the equation:

(4)

Although Pacific Cod were similar sizes in both experiments, body mass accounted for a significant source of variation in metabolic traits in Expt. 2. This pattern was primarily driven by an inverse relationship between performance and body mass in one of the tanks (14A) for the 14°C treatment (two-way ANOVA; P = 0.02, Fig. S1). Therefore, for both experiments, we normalized all calculations of performance to the mean body mass for that experiment (Expt. 1: 95.0 g; Expt. 2: 88.7 g) using the equation:

(5)

where MM_expected is the expected value for a performance metric at the mean body mass for the experiment, MM_actual is the known value of that performance metric for the individual, Mass_actual is the known body mass of the individual, Mass_mean is the mean body mass for that experiment and b is the mass scaling coefficient (−0.25; e.g. Gillooly et al., 2001; Kleiber, 1932). Mass scaling was computed within the Respirometry package in R v. 4.2.2 (Birk, 2024).

To evaluate the influence of temperature on each performance metric (SMR, RMR, MMR, U_crit, FAS, and AAS), we fit linear models to predict performance based on the fixed effects of temperature, experiment and their interaction. Models failed to converge when a random effect of tank was added, so we used one-way ANOVAs to determine whether there were significant differences among replicate tanks in Expt. 2, when two tanks per temperature treatment were used (Table S1). Tanks did not significantly impact performance at any temperature treatment; therefore, we pooled replicated tanks for subsequent analysis.

We also calculated the temperature coefficient Q₁₀ for our performance metrics in both experiments using the equation:

(6)

where T₁ is the lowest temperature, T₂ is the highest temperature and MM is the mean value for the performance metric at each of those two temperatures. Q₁₀ is a measure of how much the rate of a biological process changes for every 10°C that the temperature changes. For example, a Q₁₀ of 2 indicates that a reaction rate doubles with a 10°C increase in temperature (Brown et al., 2004; Elias et al., 2014), while a Q₁₀ of 1 indicates that there is no change in reaction rate with temperature.

Finally, to predict oxygen consumption rates in relation to swimming speed, temperature, and experiment, we examined scaled MO₂ across all swimming speeds using a linear mixed effects model fit with restricted maximum likelihood using the R package lme4 (Bates et al., 2015). For this analysis, we used an average MO₂ value for each of the three replicates per swim speed for each fish such that each fish only had one MO₂ value per swim speed. We also included a random model intercept of fish ID. MO₂ was log transformed and swim speed was scaled by subtracting the mean and dividing by the standard deviation to aid model interpretation. We extracted marginal means using R packages effects (Fox and Weisberg, 2019) and ggeffects (Lüdecke, 2018), which average specific model variables of interest over the values or levels of the remaining nonfocal predictors in the model, to interpret significant interactions between variables (Lüdecke, 2018). All analyses were run in R version 4.2.2. and visualized in ggplot2 (Wickham, 2016).

Body state and environmental variables

We collected information on body state and environmental variables for both experiments, including growth (mm d⁻¹), gonad development (gonadosomatic index [GSI]), body condition (hepatosomatic index [HSI]), salinity, photoperiod, and days since last feeding (Table 1). Growth rates were calculated from the difference in standard length at the time of swim trial and standard length approximately 4–7 months prior (Expt. 1) or 3-6 months prior (Expt. 2), divided by the number of days elapsing between the two measurements. GSI was calculated as the wet weight of the gonads divided by the wet weight of the fish, and HSI was calculated as the wet weight of the liver divided by the wet weight of the fish. For Expt. 1, we also collected information on liver and muscle lipid concentration (mg lipid g⁻¹) to provide more information on individual condition that could affect MO₂. Lipids were extracted in chloroform/methanol according to Parrish (1987) using a modified Folch procedure (Folch et al., 1957). The total lipid weight for both tissues was determined gravimetrically in tin weighing pans using a microbalance. Extracted lipid samples (250 μl) were evaporated for 1 hour and then reweighed every 5 minutes until weights were stable. HSI was highly correlated with liver lipids in Expt. 1 (Pearson’s correlation, r = 0.85), and Pacific Cod primarily allocate energy to lipid storage in muscle and liver tissues as temperatures fall below 5°C (Copeman et al., 2017). Therefore, we did not quantify liver or muscle lipids for Expt. 2. In Expt. 2, the presence and wet weights of parasites such as tapeworms, nematodes, and liver worms were noted if obvious during dissections.

Due to high levels of variation in performance metrics within tanks and temperature treatments, we explored the potential for these body state and environmental variables to influence performance. We computed tank-specific Pearson’s correlation coefficients to explore relationships between performance metrics and body state and environmental variables within each tank. We observed few patterns between any performance metrics and any single body state or environmental variable. Therefore, we used principal components analysis (PCA) to visualize potential relationships between performance metrics and the combined suite of all body state and environmental variables measured in the two experiments (Bakker, 2023). We explored potential relationships between each performance metric and the first two principal components of the PCA using Pearson’s correlation coefficients.

Results

Influence of temperature on performance

Temperature did not influence SMR, RMR, or FAS in either experiment (Fig. 1, Table 3, and Table S2). However, temperature significantly affected MMR, U_crit, and AAS, with a clearer temperature effect in Expt. 1 than in Expt. 2. Q₁₀ values ranged from 1.02 to 1.44 across all swimming metrics in both experiments, indicating that there was little change in those metrics with a 10°C increase in temperature (Table 4). Although Q₁₀ values were slightly higher in Expt. 1 than in Expt. 2 for most performance metrics, no metrics showed consistent patterns in Q₁₀ across both experiments.

Figure 1.

Temperature had variable influence on age-1 Pacific Cod performance, but all performance metrics except critical swimming speed differed between the two experiments. Performance metrics include: (a) SMR, (b) RMR, (c) MMR, (d) critical swimming speed, (e) FAS, (f) AAS (n = 71 for SMR; n = 100 for all other panels). Small symbols represent individual fish, and large symbols represent the tank means ± standard deviations. Circlesrepresent Expt. 1, and triangles represent Expt. 2. Error bars indicate 95% confidence intervals.

Table 3.

Sample sizes (n), tank means and tank coefficients of variation for all age-1 Pacific Cod performance metrics, body state variables and environmental variables

Expt.	Tank temperature	n All fish	n SMR calc.	SMR (mg O2 kg−1 h−1)	RMR (mg O2 kg−1 h−1)	MMR (mg O2 kg−1 h−1)	U crit (mm s−1)	FAS (mg O2 kg−1 h−1)	AAS (mg O2 kg−1 h−1)	COT (J kg−1 km−1)	HSI	Liver lipids (mg g−1)	Muscle lipids (mg g−1)	Growth (mm d−1)	GSI	Parasite weight (g)	Days since feeding (d)	Salinity ratio	Photo-period (h)
	Tank means
1	2	10	6	40.51	61.94	126.60	293.24	2.05	64.66	1713.17	2.87	176.01	4.67	0.15	0.30	–	3.5	0.98	9.9
1	4	11	8	48.45	61.92	136.40	342.18	2.23	74.47	1676.47	2.79	179.03	4.43	0.15	0.18	–	3.5	0.99	9.3
1	6	11	9	40.78	59.02	160.49	370.18	2.76	101.47	1683.48	2.19	137.98	4.41	0.20	0.12	–	3.5	0.91	9.0
1	8	11	8	47.95	62.88	143.29	347.55	2.33	80.40	1701.63	1.96	118.86	4.43	0.23	0.00	–	3.6	0.97	9.0
2	6	9	7	74.84	95.25	161.01	346.97	1.75	65.76	2135.46	3.42	–	–	0.35	0.27	0.05	2.9	1.01	13.7
2	6	9	7	75.36	90.54	169.63	343.21	1.90	79.09	2212.87	2.75	–	–	0.27	0.26	0.05	2.8	1.01	14.1
2	10	10	6	76.76	101.13	173.07	415.61	1.73	71.94	1996.90	2.53	–	–	0.40	0.11	0.16	2.8	0.99	14.6
2	10	10	5	76.87	101.93	165.91	372.73	1.61	63.98	2241.46	2.96	–	–	0.28	0.33	0.03	2.8	0.98	14.8
2	14	10	7	90.73	102.02	177.77	379.78	1.94	75.75	2413.64	3.47	–	–	0.33	0.06	0.02	2.5	0.98	13.8
2	14	9	8	79.22	91.56	162.69	406.26	1.94	71.13	1945.54	2.91	–	–	0.41	0.15	0.00	2.9	1.03	14.6
	Coefficients of variation by tank
1	2	10	6	0.24	0.17	0.19	0.17	0.36	0.44	0.19	0.33	0.25	0.08	0.17	1.09	–	0.31	0.03	0.02
1	4	11	8	0.20	0.13	0.14	0.19	0.23	0.27	0.14	0.34	0.23	0.08	0.17	1.74	–	0.35	0.02	0.03
1	6	11	9	0.28	0.10	0.10	0.10	0.15	0.15	0.19	0.44	0.35	0.09	0.12	2.37	–	0.32	0.04	0.00
1	8	11	8	0.31	0.17	0.14	0.14	0.16	0.21	0.16	0.29	0.47	0.24	0.15	–	–	0.31	0.01	0.00
2	6	9	7	0.14	0.12	0.11	0.12	0.12	0.28	0.13	0.43	–	–	0.30	1.30	1.98	0.36	0.04	0.04
2	6	9	7	0.13	0.09	0.09	0.07	0.08	0.13	0.14	0.32	–	–	0.21	0.77	2.05	0.30	0.04	0.02
2	10	10	6	0.14	0.15	0.15	0.13	0.19	0.37	0.15	0.36	–	–	0.34	1.63	1.11	0.28	0.06	0.03
2	10	10	5	0.17	0.10	0.17	0.16	0.20	0.46	0.14	0.15	–	–	0.22	0.55	2.99	0.33	0.04	0.04
2	14	10	7	0.22	0.29	0.16	0.17	0.47	0.43	0.20	0.29	–	–	0.30	2.23	3.16	0.28	0.06	0.07
2	14	9	8	0.34	0.41	0.22	0.13	0.35	0.38	0.28	0.42	–	–	0.26	1.20	–	0.32	0.08	0.08

Performance metrics included SMR, RMR, MMR, critical swimming speed (U_crit), FAS, AAS, and COT. SMR values were calculated only for individuals with an R²_adj >0.4 for the relationship between oxygen consumption (MO₂) and swimming speed. Body state variables included HSI, liver lipids, muscle lipids, GSI, parasite wet weights, and days since feeding. Environmental variables included the ratio of the salinity on the day of the swimming trial to the average salinity during the rearing period, and the photoperiod.

Table 4.

Temperature coefficient (Q₁₀) values for metabolic performance metrics

Performance metric	Q 10 Expt. 1	Q 10 Expt. 2
SMR	1.32	1.17
RMR	1.03	1.05
MMR	1.23	1.04
Critical swimming speed (Ucrit)	1.33	1.18
FAS	1.24	1.08
AAS	1.44	1.02
COT	0.99	1.00

Metrics are calculated for each of the two experiments using Equation 6.

Influence of experiment on performance

All performance metrics except U_crit significantly differed between the two experiments, particularly in the shared 6°C temperature treatment (Fig. 1 and Table S2). Expt. 2 fish exhibited 77.3% higher SMR, 58.5% higher RMR, 18.7% higher MMR, and 27.6% higher COT across all temperatures. Aerobic scope measurements were higher in Expt. 1, with 22.7% higher FAS and 13.2% higher AAS.

Oxygen consumption and cost of transport

Rates of oxygen consumption (MO₂) were well explained by swimming speed, experiment, and temperature (R²_conditional = 0.78), with these fixed effects accounting for 57% of the variability (R²_marginal = 0.57). MO₂ was influenced by a significant interaction between swimming speed and experiment, which was primarily explained by fish in Expt. 1 consuming more oxygen at faster speeds across all temperature treatments compared to Expt. 2 (Fig. 2 and Table 5).

Figure 2.

Age-1 Pacific Cod oxygen consumption generally increased with swimming speed for each temperature treatment and experiment. However, the rate at which oxygen consumption increased with swimming speed differed between the two experiments across all temperatures. Data points represent individuals (n = 100). Error bands represent 95% confidence intervals.

Table 5.

Linear mixed effects model results for predicted age-1 Pacific Cod oxygen consumption rates in response to swimming speed, experiment, temperature and their interactions (n = 100)

Fixed effect	Estimate	Std. error	t value	P value
Intercept	4.61	0.041	113.29	<0.0001
Swimming speed	0.26	0.017	15.69	<0.0001
Expt. 2	0.19	0.046	4.18	<0.0001
Temperature	0.04	0.040	1.10	0.274
Swimming speed × Expt. 2	−0.12	0.019	−6.72	<0.0001
Swimming speed × temperature	0.01	0.018	0.78	0.435
Temperature × Expt. 2	−0.04	0.046	−0.99	0.320
Swimming speed × temperature × Expt. 2	−0.03	0.019	−1.30	0.194

Fish ID was included in the model as a random effect.

COT was lower for Expt. 1 fish across most swim speeds, indicating more efficient swimming for individuals in this experiment (Fig. 3 and Table 6). U_opt ranged from 256.75 to 316.18 mm s⁻¹ for Expt. 1 and was highest for fish at 4°C. U_opt was 34% higher at 6°C in Expt. 2 compared to Expt. 1. However, COT models would not converge for the 10°C and 14°C temperature treatments in Expt. 2, likely due to a continued decline in COT at higher swimming speeds. For these individuals, we did not observe the expected upturn in the COT curve as swimming speed increased, suggesting that swimming remained efficient for Expt. 2 fish even at higher swimming speeds (Claireaux et al., 2006).

Figure 3.

Age-1 Pacific Cod COT generally decreased with swimming speed to an optimal swimming speed (U_opt) where COT was minimized for each temperature treatment and experiment. However, fish in the 10°C and 14°C temperature treatments in Expt. 2 did not show an increase in COT following an optimal swimming speed, and nonlinear models fit from Equation 3 for these two temperature treatments would not converge at these temperatures. Data points represent individuals (n = 100). Error bands represent 95% confidence intervals.

Table 6.

Nonlinear regression equations and R² values for predicted age-1 Pacific Cod COT in response to swimming speed, calculated using Equation 3

Expt.	Temperature	a	b	c	R 2	U opt  (mm s−1)	COT @ U opt  (J g−1 km−1)
1	2°C	236 700 ± 12 860	0.005264 ± 0.01792	3.047 ± 0.5861	0.42	256.75	1372.29
1	4°C	270 000 ± 11 850	0.02111 ± 0.07534	2.746 ± 0.6085	0.57	316.18	1343.02
1	6°C	244 700 ± 8204	0.001349 ± 0.002814	3.271 ± 0.3538	0.44	260.55	1352.69
2	6°C	359 000 ± 8973	0.04063 ± 0.09504	2.645 ± 0.3984	0.75	350.31	1647.78
1	8°C	270 100 ± 9954	0.008995 ± 0.0264	2.893 ± 0.4973	0.57	308.24	1339.15
2	10°C	–	–	–	–	–	–
2	14°C	–	–	–	–	–	–

Optimal swimming speed (U_opt) was calculated using Equation 4, and COT at the optimal swimming speed (COT @ U_opt) was calculated using Equation 3 and parameter values from this table. Separate models were run for each temperature treatment in the two experiments. Nonlinear models did not converge for the 10°C and 14°C temperature treatments in Expt. 2.

Influence of individual variability on performance

Variability among individuals was high within tanks for most performance metrics, body state variables, and environmental variables (Table 3). We also observed high variability in metabolic performance for the six individuals for which we ran duplicate swim trials in Expt. 2 (Table S3). Within individuals, relative differences between the two trials varied for SMR and RMR by 35.79 ± 28.54% and 35.72 ± 31.04%, respectively. However, we observed consistently lower within-individual variability in MMR and U_crit compared to other performance metrics (percent differences of 6.46 ± 7.71% and 8.04 ± 6.25%, respectively). These patterns suggest that baseline metabolic metrics may be more variable than maximum performance metrics within individuals.

Influence of body state and environmental variables on performance

Relationships between performance, body state variables, and environmental variables were inconsistent across tanks and experiments (Table S4). HSI and liver lipids were positively correlated with SMR and RMR and negatively correlated with U_crit, and the salinity ratio was negatively correlated with FAS across all tanks for Expt. 1, but these trends were not present in Expt. 2. Mean growth rates across all tanks were 88.8% higher in Expt. 2, generally increased with temperature, and were positively correlated with U_crit in all tanks for both experiments (Fig. S2). We also observed more than double the number of individuals with gonad development in Expt. 2, even though fish in Expt. 2 were approximately 6 months younger than those from Expt. 1. No other body state or environmental variables (days since feeding, muscle lipids, or photoperiod) showed consistent positive or negative trends with performance metrics across tanks within an experiment. For Expt. 2 where parasite wet weight was quantified, 67% of all fish with parasites were excluded from SMR calculations because of poor model fit.

The first two principal components of the PCA comprised 54.3% of the variation in body state and environmental variables among individuals for Expt. 1 and 50.0% for Expt. 2 (Fig. 4 and Table S5). However, even though the combined body state and environmental variables accounted for more than half of the overall variability between fishes, we observed only moderate to weak correlations between any performance metric and axis 1 of the PCA (Fig. 5). In Expt. 1, relationships were strongest between PC1 and FAS (r = 0.56), AAS (r = 0.56), U_crit (r = 0.55), and MMR (r = 0.49). These trends appeared to be primarily driven by differences in body state and environmental variables between the two coldest temperatures and the two warmest temperatures in this experiment. Correlations between performance and PC1 were weaker in Expt. 2 across all metrics (Fig. 5). In addition, most of the relationships between performance and the first two principal components were inconsistent across temperatures (Table S6). U_crit was the only variable to exhibit a positive correlation with PC1 across all temperatures in both experiments.

Figure 4.

PCA of body state and environmental variables from the two experiments explained 54.3% of the variability between age-1 Pacific Cod in Expt. 1 (panel a) and 50.0% of the variability in Expt. 2 (panel b). Vectors indicate correlations between the body state and environmental variables and the first two principal axes. Colours indicate temperature treatments. Separate models were run for each of the two experiments (Expt. 1; n = 43; Expt. 2; n = 57).

Figure 5.

Age-1 Pacific Cod performance metrics were weakly to moderately correlated with the first principal component of the PCA containing body state and environmental variables for both experiments. Pearson’s correlations and sample sizes are reported for performance metrics for Expt. 1 (top 2 rows) and Expt. 2 (bottom 2 rows). Performance metrics include SMR, RMR, MMR, critical swimming speed (U_crit), FAS, and AAS. Colours indicate temperature treatments. Separate models were run for each of the two experiments (Expt. 1; n = 43; Expt. 2; n = 57).

Discussion

As the world’s oceans continue to warm, there is a growing need to understand how fish will respond and potentially acclimate. We acclimated fish to temperature treatments for extended periods (9–15 months) to examine the impact of prolonged temperature exposure on metabolic performance, which differs from many other laboratory experiments on metabolic thermal reaction norms where metabolism is measured after a short-term acclimation to temperature (e.g. 1 week to 1 month; Soofiani and Priede, 1985; Paul et al., 1988; Tirsgaard et al., 2015). Under these longer-term acclimation conditions, we found no influence of temperature on SMR, RMR, or FAS for age-1 Gulf of Alaska Pacific Cod, and a moderate influence of temperature on other performance metrics. However, we did find substantial differences in performance between the two experiments and high individual variability that could only be partially explained by other body state variables or environmental variables such as feeding schedule, body condition, salinity, and somatic and reproductive growth. These results suggest that the duration of temperature acclimation, individual variability, and potentially other factors that were not directly tested in this study, such as seasonal variability in photoperiod or cohort effects related to oceanographic conditions at time of hatch, can influence performance, even in a controlled laboratory environment.

The Q₁₀ temperature coefficient values hovered near 1 across all performance metrics, suggesting that temperature was not a primary driver of metabolic rate for this study, particularly for baseline metabolic metrics (SMR and RMR). In response to acute thermal exposure (less than 4 weeks), other studies have demonstrated that adult Pacific Cod (Paul et al., 1988), juvenile Atlantic Cod (Soofiani and Priede, 1985) and adult Atlantic Cod (Claireaux et al., 2000; Tirsgaard et al., 2015) exhibit a positive linear increase in metabolic rate with temperature, with MMR generally increasing more rapidly than SMR and Q₁₀ values often greater than 2. In contrast, Hanna et al. (2008) demonstrated no differences in either baseline or maximum metabolic performance for adult Pacific Cod acclimated to temperature for 8 weeks. Taken together, previous findings and the present study show that both adult and age-1 Pacific Cod have the capacity to acclimate to chronically elevated temperatures, potentially within as little as two months, which is comparable to studies of other fishes demonstrating similarly rapid acclimation of metabolic performance to elevated temperatures (Donelson et al., 2012; Johansen et al., 2021). In response to prolonged exposure to elevated temperatures, fish may upregulate mitochondrial densities in their tissues, increase the number of red blood cells in their bloodstream, and increase gill surface area to maximize oxygen diffusion (Johansen et al., 2021). Typically, acclimation to elevated temperature involves a decrease in baseline metabolic rate such that aerobic scope is preserved (Sandblom et al., 2016), so perhaps in this study, baseline metabolic rates were downregulated at the higher temperatures over time. However, plasticity in these traits may influence an individual’s ability to acclimate, particularly at high temperatures, and a delayed ability to alter these parameters may contribute to increased mortality or reduced fitness.

Pacific Cod exhibited higher variability across most performance metrics at 10°C and 14°C compared to other temperatures. This variability was particularly evident for RMR, where we observed both the highest and the lowest individual RMR values across the entire study. Body state variables appeared to play a role in these divergent patterns of baseline metabolic performance with higher-performing individuals typically in better condition (e.g. higher HSI values). It is also notable that most fish with low RMRs at 14°C were such poor swimmers that a corresponding SMR value could not be calculated from the relationship between oxygen consumption and swimming speed for these individuals. While we had only one mortality (in the 14°C treatment) over the course of either experiment, the high variability in metabolism at both 10°C and 14°C suggests that these temperatures may be approaching the limits of the adaptive capacity for age-1 Pacific Cod. This pattern was also observed in Laurel et al. (2016), which found that the growth rates of age-0 Pacific Cod began to decline at 14°C. During recent marine heatwaves in the Gulf of Alaska, temperatures exceeded 14°C in the nearshore for several months during summer, and there was evidence for size-selective mortality on juveniles, with smaller individuals virtually absent in the nearshore by the end of summer during marine heatwave conditions (Thalmann et al., 2024). Potentially, individuals that were less able to acclimate to these high temperatures may have been more susceptible to starvation or predation, further contributing to population declines during marine heatwaves. Further, limited capacity to acclimate to high temperatures may have triggered movement to cooler waters in the Bering Sea (Spies et al., 2020) and, potentially, the Alaskan Arctic (Cooper et al., 2023) during the marine heatwaves, particularly since performance generally remained high even at the coolest temperatures examined in this study.

We observed striking differences in performance between our two experiments, including higher SMR and RMR values, and no apparent increase in COT at faster swimming speeds for Expt. 2 fish. Oxygen consumption rates also increased more slowly at faster swimming speeds for Expt. 2 fish, and we observed several swimmers in Expt. 2 with U_crit values greater than 450 mm s⁻¹ at 10°C and 14°C. One potential explanation is that some individuals in Expt. 2, particularly in the 10°C and 14°C temperature treatments, used anaerobic metabolism at faster swim speeds, which would have decreased the apparent influence of aerobic metabolism on swim performance. At 10°C and 14°C, oxygen consumption rates were much more variable at faster swimming speeds than for other temperature treatments, and we observed several individuals whose oxygen consumption rates declined before they reached their critical swimming speed, suggesting that these individuals may have switched to primarily anaerobic metabolism at faster swimming speeds. We also observed a continued decline in COT at 10°C and 14°C, even at the highest swimming speeds, and an absence of a clear U_opt at these temperatures. This pattern was not observed in the other temperature treatments or in the 6°C treatment for Expt. 2. Gadids, like other white-muscled finfish, have high glycolytic potential, with substantial individual variability in anaerobic performance (Reidy et al., 2000) and a relatively fast transition between aerobic and anaerobic metabolism at higher temperatures (Kunz et al., 2018). During periods of intense exercise or hypoxia, fish can produce energy in the absence of oxygen by undergoing anaerobic glycolysis, where glucose is converted to lactate to produce adenosine triphosphate (Plambech et al., 2013; Ejbye-Ernst et al., 2016). A transition to anaerobic metabolism for some individuals in 10°C and 14°C temperature treatments further suggests that these fish may be responding somewhat differently to elevated temperatures even after a yearlong acclimation period.

Variability in baseline metabolic performance (SMR and RMR) that we observed between the two experiments could be related to differences in the season in which the experiments were run. Age-1 Pacific Cod may exhibit higher baseline metabolic rates in the summer, a result similar to other studies (Gatten et al., 1988; Toledo et al., 2008). Throughout these experiments, we varied photoperiod, but not temperature, based on seasonal conditions for Kodiak Island, AK. However, other studies on ectothermic metabolism (e.g. sharks, Wheeler et al., 2022; and lizards, Gatten et al., 1988) that specifically consider seasonal variability in both temperature and photoperiod have found that photoperiod may be a stronger regulator of metabolic rate than temperature, suggesting that seasonal variability in photoperiod may have contributed to some of the patterns that we observed in this study. In summer, age-1 Pacific Cod in the Gulf of Alaska use coastal habitats (~15 m deep) for foraging and shelter before moving further offshore in the winter (Laurel et al., 2009; Abookire et al., 2024). Higher baseline metabolism in the summer associated with high prey availability in these coastal habitats may allow for increased anabolism of tissues (e.g. the increased input hypothesis), potentially increasing growth and, ultimately, reproductive yield (Bennett and Ruben, 1979; Pettersen et al., 2018). In addition to higher baseline metabolic rates that may be associated with their swim trials occurring in summer, Expt. 2 fish had higher somatic growth and gonadal development rates than Expt. 1 fish, even though these fish were approximately 6 months younger than Expt. 1 individuals (Table 2). This suggests that the higher baseline metabolic rates for Expt. 2 individuals may reflect additional physiological processes (i.e. growth, reproduction) that would be expected during the summer in their natural environment. In contrast, there may be a fitness advantage in maintaining a lower baseline metabolism in winter (e.g. the allocation hypothesis; Pettersen et al., 2018; Steyermark, 2002). During winter, temperate and subarctic fishes such as Pacific Cod must be able to survive cold temperatures and limited food availability (Norcross et al., 2001; Hurst, 2007). Downregulating metabolic rate during winter may be an adaptation to conserve fat stores and protect fish from starvation risk and stress associated with low temperatures (Crawshaw, 1984; Larivée et al., 2010). However, the absence of seasonal temperature variation in our experimental set-up, with only one experiment run during each season, does not permit us to fully explore the potential for seasonal variation in Pacific Cod metabolic rate in this study.

The variability between the two experiments may also be due to persistent effects imposed by the different oceanographic environments in which the fish hatched and developed. Expt. 1 fish were collected from the field in the summer of 2019 (mean temperature: 13.2 ± 1.1°C), which was an anomalously warm marine heatwave year in the Gulf of Alaska, associated with low hatch success (Laurel and Rogers, 2020), a shift to earlier hatching (Almeida et al., 2024) and reduced abundance in the nearshore (Abookire et al., 2021; Thalmann et al., 2024). In contrast, Expt. 2 fish were collected in the summer of 2021 (mean temperature: 10.6 ± 1.2°C), a substantially cooler year with approximately double the density of age-0 individuals in the coastal nursery habitat than in 2019 (Hulson et al., 2023). We observed lower SMR and RMR for individuals born into marine heatwave conditions than the cohort born in cooler conditions. This pattern is consistent with lower SMR observed in Sticklebacks (Gasterosteus aculeatus; Pilakouta et al., 2020), Brown Trout (Salmo trutta; Archer et al., 2021) and Mosquitofish (Gambusia holbrooki; Seebacher et al., 2014) hatched into warm conditions in the field. At a molecular level, it is hypothesized that fish born into cooler conditions (such as in 2021 in the Gulf of Alaska) upregulate specific rate-limiting steps of metabolism to compensate for low water temperatures, resulting in higher baseline metabolic rates (Seebacher et al., 2014). Conversely in warm conditions, fish reduce mitochondrial function to suppress baseline metabolic rates (Seebacher et al., 2010; Norin et al., 2014). It is likely that the lower SMR and RMR observed during Expt. 1 were influenced by both the season of the experiment and the anomalously warm oceanographic conditions in which the fish hatched, highlighting a need for future studies to more comprehensively explore these relationships.

Body state and environmental variables had mixed influence on metabolism and performance, accounting for only ~50% of the variability between individuals even when combined in a multivariate framework. When combined, body state variables were more highly correlated with maximum performance metrics than baseline performance metrics, suggesting that body state variables were perhaps more informative of maximum performance. Growth rates were generally well correlated with metabolism, with a positive relationship observed for MMR, FAS, and AAS. Other studies have also found strong relationships between growth and metabolism (Rosenfeld et al., 2015; Archer et al., 2021; White et al., 2022), and it is hypothesized that elevated metabolism is required to synthesize new tissue during the growing process, even when the animal is at rest. However, we did not observe consistent patterns between growth rate and metabolism across all tanks used in either experiment. Age-1 Pacific Cod are at a pivotal phase of early reproductive development, with some individuals devoting more energy to somatic growth and others to reproductive growth. We observed inverse relationships between somatic growth and GSI in our PCA, which may have obscured the apparent relationship between growth and metabolism. Measures of condition, including HSI and liver lipids, were the only body state variables to consistently show positive relationships with baseline metabolism across all tanks in Expt. 1, with faster metabolism correlated with higher HSI values. At the 14°C treatment in Expt. 2, we observed two distinct clusters of individuals, with animals in better condition generally outperforming those in poorer condition, which suggests that those individuals may be at an advantage when navigating extreme temperatures (e.g. Abookire et al., 2024). Individual variability in metabolism and swimming efficiency may have also been influenced by behavioural traits that were not directly measured in this study, such as aggressiveness (e.g. Lahti et al., 2022; Metcalfe et al., 2016). Overall, the limited and inconsistent patterns between metabolism and body state variables, even when combined, suggest that genetic and behavioural variation, season, year-class and oceanographic conditions at hatch may be more important in driving variation in metabolic rate for age-1 Pacific Cod.

Conclusions

Metabolic rate can be a powerful indicator of shifts in the foraging environment, habitat, and reproductive potential of fishes in response to changing ocean temperatures. As more studies focus on long-term, integrative measures of metabolic rate for wild fishes, we recommend several considerations for studies focused on fish metabolism under warming ocean conditions.

Thermal Acclimation. The relationship between metabolism and temperature for fishes can change over time, particularly if the temperature is within the range of acclimation capacity for the species. Further, a fish’s ability to acclimate to temperature will also be influenced by its unique intrinsic attributes, including genotype, parental effects and epigenetic variability. We demonstrate no relationship between temperature, SMR, and RMR after 1 year of prolonged acclimation to experimental temperatures, which is a typical period over which metabolic proxies, such as otolith δ¹³C, integrate. Thus, acclimation capacity over a prolonged time period represents an important consideration in interpreting patterns derived from integrative metabolic proxies and for evaluating an individual’s ability to adapt to warming. However, it is unlikely that, even during marine heatwave conditions, temperatures in the field would remain constant over a full-year period, representing a challenge for interpreting the consequences of long-term thermal acclimation on fish metabolism in the wild.

Seasonality and prior thermal experience. While this study was not designed to explicitly test for these effects, our results point to the potential importance of seasonal and cohort effects in influencing baseline metabolism for Pacific Cod. Both laboratory and field estimates of metabolic rate can reflect variability associated with season, thermal conditions experienced early in life, and cohort effects, providing different insights depending on the oceanographic conditions or time of year in which the study is conducted. Although they cannot be fully controlled for a priori, an understanding of seasonal and oceanographic trends at the time of fish collection may be particularly important for interpreting variability in their metabolic rate.

Body state and environmental variables. In this study, where all fish were similar sizes, we observed limited influence of body state and environmental variables (e.g. somatic and reproductive growth, body condition and parasite presence) on metabolism. While different species may exhibit different relationships between body state variables and metabolism based on their life stage, life history strategy, and activity levels, we hypothesize that body state variables may be less important than other factors when considering primary drivers of variability in metabolic rate.

Overall, we find that the relationships between temperature and metabolism for fishes can be substantially more complex than simple expectations of temperature dependence derived from measurements following short-term acclimation. In this study, age-1 Pacific Cod exhibit the capacity for thermal acclimation in the laboratory as evidenced by similar metabolic rates across a wide range of temperatures, which is a pattern that is also observed for adult Pacific Cod (Hanna et al., 2008). However, at our highest temperature treatments (10°C and 14°C), we observed increased variability in metabolism and a decline in COT at faster swimming speeds, suggesting that some individuals may not be able to acclimate as effectively or may be more reliant on anaerobic metabolism to maintain performance at elevated temperatures. In the field, these individuals may have been at increased risk of starvation or predation, particularly during stressful conditions such as those occurring during marine heatwaves. Further, while harder to control for, factors such as season, thermal history and epigenetics may play substantial roles in metabolic variability between individuals, as evidenced by the more than 60% lower metabolic rates observed in our first experiment. A better understanding of the influence of these factors on metabolism, and ultimately, their influence on survival trajectories, is needed for predicting relationships between temperature, recruitment, and field metabolic rates in a warming climate.

Author contributions

H.L.T., B.J.L., M.B.R.H. and J.A.M. conceived and designed the study; H.L.T. and M.B.R.H. collected the data; H.L.T. conducted the analyses; H.L.T., B.J.L., M.B.R.H., E.S. and J.A.M. interpreted the data; H.L.T. wrote the manuscript; all authors edited and approved the manuscript.

Conflicts of interest

The authors declare no competing interests.

Funding

This work was supported by the NOAA Office of Education Educational Partnership Program Experiential Research and Training Opportunity (NERTO) through the Living Marine Resources Cooperative Science Center (NA16SEC4810007 to H.L.T.); the National Science Foundation Graduate Research Fellowship Program (1840998 to H.L.T.); the Cooperative Institute for Marine Ecosystem and Resources Studies (CIMERS); and a North Pacific Research Board Graduate Student Research Award (to H.L.T.). Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Oceanographic and Atmospheric Administration or the National Science Foundation.

Data availability

Data and code are available in a GitHub repository: https://github.com/hthalmann/Thalmann_et_al_Pacific_Cod_Thermal_Metabolic_Variation

Supplementary material

Supplementary material is available at Conservation Physiology online.