The effects of warm thermal variability on metabolism and swimming performance in wild Atlantic salmon (Salmo salar)

Abstract

Warmer and more variable temperatures have been implicated in the recent decline of Atlantic salmon (Salmo salar) in Eastern Canada. To date, we know little on how ecologically relevant thermal fluctuations affect swimming performance in fishes. The goal of this study is to determine the effects of warm versus cool diel thermal variability on swimming efficiency and the speed limit for sustainable aerobically fueled swimming. We acclimated wild S. salar juveniles to a cool and a warm ecologically realistic diel thermal profile (16–21 and 19–24°C), and then tested individuals over a common acute change in temperature (16–24°C). We measured metabolic rate and swimming kinematics at a range of swimming speeds, at five temperatures (16, 18, 20, 22, and 24°C) and calculated swimming efficiency. Our temperature acclimation did not appear to significantly affect energetic and kinematic swimming efficiency, but acute exposure to high temperature did increase overall metabolic rate. It appears that wild S. salar can swim efficiently and sustainably during both acute cool and warm exposures, and after acclimation to diel thermal variation of 16–21 or 19–24°C.

1. INTRODUCTION

Temperature has pervasive effects on the physiology of fishes. Fishes can modify cellular reaction rates, whole‐animal metabolic rates, and performance after prolonged exposure to a different temperature (Fry & Hart, 1948; Lee et al., 2003; MacNutt et al., 2006; Sappal et al., 2015). Recently, there has been increasing focus on understanding how fishes cope with natural diel thermal variations, especially given that climate models project increases in thermal variability (Morash et al., 2018; Moyen et al., 2020; Niehaus et al., 2012; Vasseur et al., 2014). Diel thermal variations/cycles in freshwater systems can be large, and even in temperate rivers, the daily change in water temperature can span 3–15°C (T_max−T_min) (Lynch et al., 1984; Malcolm et al., 2004). Occasionally, these daily changes in temperature may reach a level that could lead to physiological stress and decreases in performance, and even death in extreme cases. Changes in temperature could impact fish's metabolic rate and capacity for energy production, which, in turn, can impact their growth, reproduction, swimming performance, etc. (Gillis et al., 2023; Neubauer & Andersen, 2019).

Swimming requires sustained effort and significant amounts of adenosine triphosphate (ATP) to fuel contracting muscles. Energy in the form of ATP production is also critical for such things as growth, basal metabolism, maintaining energy reserves, and gonad development (Crossin et al., 2016; Hagelin et al., 2016; Strople et al., 2018), and available energy must be portioned among these activities. The amount of energy fish can conserve depends, at least partly, on the efficiency at which they swim (Rand & Hinch, 1998). An adequate ability to swim aerobically also benefits the fish by curtailing unnecessary bouts of unsustainable anaerobic swimming, which can take from 12 to 24 h of recovery time (Hvas et al., 2020; Hvas et al., 2021; Powell et al., 2009; Zhang et al., 2018). The capacity for non‐exhaustive sustainable aerobic swimming may also vary by test temperature (Fry & Hart, 1948; Jones et al., 2008; Tudorache et al., 2010) or acclimation temperature (Dickson et al., 2002; Fry & Hart, 1948; Hvas et al., 2017), as has been found in exhaustively swum fish. Notably, most studies that have measured sustainable aerobic swimming have done so at different stable acclimation temperatures, and how this important metabolic metric changes with diel thermal cycling remains untested.

The capacity for sustainable swimming is thought to be temperature‐dependent in accordance with the power output of supporting aerobic red muscles (Bone et al., 1978; Rome, 2007). Without adequate aerobic power from the red muscles, the “fast‐twitch” anaerobic white muscles are recruited (Booth et al., 1997). White muscle recruitment is apparent through rapid tail beats and subsequent burst‐forward motion of the fish (Booth et al., 1997; Jayne & Lauder, 1994) and has been found to occur at ~80% of the U _crit (Beddow & McKinley, 1999; Hvas & Oppedal, 2017). This anaerobic activity quickly depletes limited supplies of glycogen, making it unsustainable for any length of time (Kieffer, 2000; Sänger & Stoiber, 2001). The onset of unsustainable bursts at the gait transition speed (U _gt) has been identified as the de facto speed limit for sustainable aerobic swimming (U _sus) (Ejbye‐Ernst et al., 2016; Marras et al., 2013; Svendsen et al., 2015). The effect of temperature on U _gt is ambiguous. U _gt can increase with cooler thermal acclimation or acute warming as both increase red muscle power output (Rome et al., 1992; Sisson & Siddell, 1987). In some cases, however, U _gt decreases with acute warming (brook charr, Salvelinus fontinalis Richardson 1836; Tudorache et al., 2010). Performance at warmer conditions could become limited by the capacity to deliver oxygen to generate metabolic use (Farrell, 2007; Jones et al., 2008). At higher temperatures, the proportion of the maximum oxygen consumption or metabolic rate (MMR) used for basal metabolism (routine metabolic rate [RMR]) increases (Eliason et al., 2011; Lee et al., 2003). This can reduce the difference (MMR − RMR) or aerobic scope (AS) that supports aerobic activities (denoted as AS_sus for sustainable aerobic swimming; Ejbye‐Ernst et al., 2016). Despite our understanding of the effects of stable, warm temperatures on sustainable aerobic swimming capacity and aerobic scope in fishes, there is still a gap in knowledge regarding the influence of ecologically relevant warm diel thermal cycles on these biomechanical and metabolic variables.

It is also unclear how warm thermal cycles will affect swimming efficiency—the amount of work/energy (input) needed for a particular swimming speed (output). Work done by the tail to swim depends on the distance and frequency of its lateral excursions (Gibouin et al., 2018), both of which can change with acute warming and acclimation temperature (Lea et al., 2016; Nudds et al., 2020; Stevens, 1979). Temperature can also affect the efficiency of mitochondrial respiration, ATP production, and red muscle contractions (Ferguson et al., 2002; Gerber et al., 2020; Rome, 2007), which could influence the cost of swimming or its metabolic rate (MO_2swim). Changes in swimming metabolic rate ultimately affect the cost of transport (COT, J/km). It has not yet been determined how the swimming efficiency of fish changes within warm diel thermal cycles and across different acclimation thermal cycles.

To better understand sustainable aerobic swimming capacity and efficiency under natural diel thermal cycles, we used non‐reproductively active adult fresh water, wild Atlantic salmon (Salmo salar Linneaus 1758) caught as parr from the Miramichi River as a model species/river system. This expansive river system harbors at least 78 species of fish (McKenzie, 1959) and hosts one of the largest migrations of S. salar (Cunjak & Newbury, 2005). The S. salar play a key cultural role in many communities and are involved in building and maintaining local economy and ecology; however, wild populations have declined sharply in abundance over the past ~30 years (Dadswell et al., 2021; DFO, 2019). Declines have been attributed, at least partly, to the warmer summer temperatures in freshwater environments (COSEWIC, 2010). S. salar are an excellent model for studying the impacts of warm, thermal variability on swimming performance, as their life history requires them to be able to sustain aerobic swimming for migration, as well as frequently use anaerobic swimming for territorial defence, predator avoidance, passing river obstacles, etc. (Thorstad et al., 2012). In addition, in the Miramichi River, the daily maximum temperature exceeds 23°C as often as 50–60 days in a year, occasionally rising above 29°C in the summer (Caissie et al., 2013). Temperatures above 23°C can cause S. salar to aggregate at cool water refugia (Breau et al., 2007), and this temperature has been identified as a thermal threshold, beyond which S. salar are considered stressed (Breau, 2013; Breau et al., 2011). Indeed, chronic exposure to temperatures above 23°C results in significant mortalities in hatchery A. salmon (Gamperl et al., 2020). Here we compared wild S. salar acclimated to diel thermal cycles reflective of cool and warm streams (i.e., 16–21 and 19–24°C respectively, see Caissie et al., 2013) typical of the Miramichi River during the summer months. We also acutely warmed these fish from 16 to 24°C at ecologically relevant warming rates to understand within‐day variations in swimming performance. We hypothesized that acute warming and the warmer acclimation thermal cycle would alter tail beat kinematics and, consequently, the work exerted, which, together with changes in physiology, could alter swimming efficiency. We predicted the speed limit for sustainable swimming would increase with cool acclimation and acute warming based on red muscle power output, but at suboptimally warm temperatures (>23°C), performance will become oxygen limited and decline.

2. METHODS

2.1. Animal collection, rearing, and holding

We collected S. salar (age 1+ and 2+) from the Rocky Brook (Miramichi) in October 2018 using electrofishing for 2 days. We transported the fish to the Crabtree Aqualab at Mount Allison University (NB, Canada) using a 750‐L tank filled with river water, which was kept oxygenated by periodically injecting oxygen. Upon arrival, we moved the fish into 300‐L circular fibreglass tanks (60 cm tall, 92 cm diameter). We provided tanks with recirculating fresh water and subjected them to a 12:12 light–dark photoperiod. We fed the fish to satiation once daily with commercial fish pellets (EWOS Enviroclean pigmented pellets, EWOS Canada limited). Fish used in this experiment were grown in two different tanks with 16–21 and 19–24°C thermal cycle regime for >1 year. Diel thermal cycles had a temperature change rate of ~0.42°C h⁻¹, a temperature maximum at 7:00 p.m. and minimum at ~7:00 a.m. During the growth period, parr and smolts were subjected to various physiological measurements as detailed in Andrew et al. (2024). They subsequently desmolted to a freshwater non‐reproductively active phenotype, and these fish were used in this experiment. All care and subsequent experimental procedures were approved by Mount Allison Animal Care Committee following guidelines from Canadian Council on Animal Care (protocol number: 101929).

2.2. Swimming experiment timeline and setup

Each experiment with one fish lasted 3 days (e.g., 16–21°C shown in Figure 1). Each fish experienced varying water temperatures (1624°C) and velocities (24–73.5 cm/s), where we measured their RMR, active metabolic rate (AMR), and swimming footage for analysis on behavior and tail beat kinematics.

FIGURE 1.

Thermal profile in the swim tunnel for an Atlantic salmon (Salmo salar) acclimated to the 16–21°C diel thermal cycle. Temperature (°C) is the mean recorded during the measurement phase for intermittent respirometry. On day 1, salmon rested at its acclimation diel thermal cycle (16–21°C). On days 2 and 3, the temperature changed in 1°C increments every 80 min, and after every 2°C increase, we measured standard metabolic rate (SMR) and active metabolic rates (AMRs).

On the morning of day 1, we transferred a 12‐h fasted fish into a 30‐L Loligo swim tunnel (14 × 14 × 46 cm; W × H × L) for a 24‐h recovery period prior to measuring RMR. During this time, the swim tunnel was flushed for 140 s every 480 s and sealed in the remaining 340 s. The sealed period consisted of a 60‐s wait phase followed by a 280‐s measurement phase, during which the rate of decline in oxygen levels was used to calculate metabolic rate. Ambient water flow was 17.5 cm/s (~0.6 BL/s). Water temperature stayed within ~0.2°C of the desired set point by controlling warm water influx (~35°C) from a ~ 100‐L reservoir that opposed a constant cooling action of a chiller. The swim tunnel was illuminated dimly throughout the experiment except in the third day when full light conditions were used for swim trials. In the first 30 min, fish acclimated to 16–21 and 19–4°C experience 16 and 19°C, respectively, then stepwise warming (0.42°C/60 min) until 21 and 24°C in the afternoon (7:00–8:00 p.m.). After 1 h, the temperature was decreased hourly by 0.67 and 0.42°C (for each respective acclimation groups) until 16°C was reached. This process mimics the changes in their acclimation tanks to ensure consistent exposure to the same thermal regime for which they are acclimated.

On the morning of day 2, we measured the RMR at 16°C and as the temperature increased (+1°C/80 min) from 16 to 24°C (Figure 3). At each of 16, 18, 20, 22 and 24°C, we measured 9–10 metabolic rates and averaged the lowest three as RMR. After staying at 24°C for 80 min, temperature was decreased by 1°C every 80 min until ~16°C.

FIGURE 3.

MO _2swim (active metabolic rate – standard metabolic rate [SMR]) as a function of swim speed in Atlantic salmon (Salmo salar) acclimated to 16–21°C (turquoise; n = 6) and 19–24°C (pink; n = 6) diel thermal cycles (a), then swum at 16, 18, 20, 22, and 24°C (dark‐to‐light blue color scale; [b]). Each small panel shows data of a fish acclimated to either 16–21°C (top row) or 19–24°C (bottom row). Lines depict robust mixed model predictions (detailed in Supporting Information); bands reflect 95% CIs. Outliers downweighed by the robust mixed model are displayed as less‐visible points (visibility ∝ weight). NS denotes no significant (p > 0.05) effect of temperature treatments.

On the morning of day 3, the fish experienced the same warm temperature ramp as day 2, from 16 to 24°C. At the five temperatures of interest (16, 18, 20, 22, and 24°C), the fish experienced a stepped swimming protocol modified from Jain et al. (1997). Water velocity increased by 4.5 cm/s every 6 min from the ambient 17.5 cm/s (~0.6 BL/s). At each velocity, we measured metabolic rate using 6‐min loop intermittent respirometry (with 100, 60, and 200 s of flush, wait, and measurement phases, respectively) and collected video footage to record tail beat kinematics, including tail beat frequency (TBF) and tail beat amplitude (TBA), which were captured aerially using a Nikon D5300. Additionally, we quantified the duration of intense bursting behavior, characterized by sustained swimming effort against the grid in front of the swim tunnel (Zhang et al., 2019) and vigorous tail beats. Intense bursting generally occurred at speeds faster than U _gt, likely where the fish transition from bursts to full anaerobic sprinting. When intense burst duration exceeded 36 s (10% of the 6 min increment), at speeds presumably above that of interest (U _gt), the swim trial was terminated. Thus, U _gt represents the speed where salmon begin to display unsustainable anaerobic bursts. After the last swim test, fish were weighed and measured for fork length, total length, height at girth, and width at girth (Table 1), then returned to their original tanks.

TABLE 1.

Atlantic salmon (Salmo salar) morphometrics.

	Acclimation thermal cycle
	16–21°C	19–24°C
Morphometrics
Mass (g)	361 ± 83	344 ± 48
Fork length (cm)	30.5 ± 1.6	29.6 ± 1.3
Total length (cm)	31.7 ± 1.9	30.9 ± 1.1
Height at girth (cm)	6.0 ± 0.6	5.9 ± 0.5
Width at girth (cm)	4.1 ± 0.6	3.8 ± 0.3
Condition factor	1.3 ± 0.2	1.3 ± 0.1
SBE correction factor a	1.1 ± 0.0	1.1 ± 0.0
N	6	6

Note: Values represent means ± SD. Morphometrics between acclimation temperature cycles are all statistically similar (p > 0.05) according to the two‐tailed t‐test.

^a Correction factor for solid blocking effects (SBE) were calculated according to the procedure in Bell and Terhune (1970).

2.3. Video analysis

We tracked fishtail and snout co‐ordinates using an artificial intelligence (AI) deep learning algorithm in DeepLabCut (version 2.1.9) (Mathis et al., 2018). First, we selected all videos with fish swimming at a common representative temperature (20°C). From these videos, we extracted 500 frames (selected by k‐means method in DeepLabCut), then in each frame, we located and labeled the fish snout and tail tip. Out of all labeled frames, 95% trained the AI, whereas the remaining 5% validated AI‐tracking accuracy. AI‐tracked and human‐labeled co‐ordinates differed by 5.03 pixels or 0.30 cm on average. AI was then used to quantify the co‐ordinates of the snout and tail tip in every frame (Figure S1).

We documented fish swimming behavior using an event‐logging software, BORIS (Behavioural Observation Research Interactive Software; Friard & Gamba, 2016). We counted or timed three distinct behaviors: swimming, burst, and intense bursting. Fish were swimming when they actively beat their tail, with bellies above the substrate. Burst occurred when fish exhibited irregular tail beats and lunged forward in the swim tunnel (Figure S2). Fish were intense bursting when they exhibited vigorous tail beats and exert sustained swimming effort against the front grid of the swim tunnel.

3. CALCULATIONS

3.1. Swim speed

We first corrected swimming speed increments (U _inc) for solid blocking effects (i.e., acceleration of water flow due to fish body blocking) using correction factor ∈_S in Equation (1) (Bell & Terhune, 1970). We then divided the corrected swim speeds (U _corr) with fork length (in centimeters) to get relative swimming speed (U _rel) in body lengths (BL s⁻¹) using Equation (2)

$$U_{\text{corr}}=U_{\mathrm{inc}}\left(1+{\in }_S\right)$$ (1)

$$U_{\mathrm{rel}}=\frac{U_{\text{corr}}}{\text{Fork Length}}$$ (2)

3.2. Metabolic rates

We calculated mass‐specific metabolic rates from the rates of oxygen decline (β ₁) during the measurement phase in respirometry using Equation (3) (with V _sys and V _fish representing volume of the system and fish, respectively). For every 9–10 metabolic rates measured at each temperature of interest, on resting fish (day 2), we averaged the lowest three as RMR. All metabolic rates measured during swimming experiments (on day 3) were treated as AMRs. To ensure that AMR accurately represents metabolic rate when the fish is actively swimming, we omitted AMRs obtained in speed increments where fish spent less than 60% of the time swimming. The remaining AMR values (323 out of 496) have an average percentage swim time of 94%. AMRs were then subtracted from the corresponding fish’ RMR measured at the corresponding temperature, which yields MO_2swim—the metabolic rate attributed exclusively to swimming (Equation 4).

$$\mathrm{MR}=\frac{{\beta }_1·\left(V_{\mathrm{sys}}-V_{\text{fish}}\right)}{\text{Fish Mass}}$$ (3)

$$\mathrm{M}{\mathrm{O}}_{2\text{swim}}=\mathrm{AMR}-\mathrm{RMR}$$ (4)

We assessed energetic swimming efficiency using MO_2swim, and the COT calculated as the energy (in joules) spent per kilometer traveled. We calculated COT_net (i.e., transport costs attributed to swimming) and COT_gross (i.e., cost attributed to swimming and living) using Equations (5) and (6), respectively; the conversion ratio 0.0143 kJ mgO₂ ⁻¹ was obtained from Videler (1993) and Ohlberger et al. (2006) and has been used for S. salar (Hvas et al., 2017).

$$\mathrm{CO}{\mathrm{T}}_{\mathrm{net}}=\frac{\mathrm{M}{\mathrm{O}}_{2\text{swim}}}{U_{\text{corr}}}·\frac{0.0143\mathrm{kJ}}{\mathrm{mg}{\mathrm{O}}_2}·\frac{3600\mathrm{s}}{\mathrm{h}}·\frac{10^5\mathrm{cm}}{\mathrm{km}}$$ (5)

$$\mathrm{CO}{\mathrm{T}}_{\text{gross}}=\frac{\mathrm{AMR}}{U_{\text{corr}}}·\frac{0.0143\mathrm{kJ}}{\mathrm{mg}{\mathrm{O}}_2}·\frac{3600\mathrm{s}}{\mathrm{h}}·\frac{10^5\mathrm{cm}}{\mathrm{km}}$$ (6)

We calculated absolute sustainable aerobic scope (AAS_sus) using the MO_2swim at U _gt. MO_2swim at U _gt was estimated using Equation (4). We then calculated AMR_sus (Equation 7) and the factorial sustainable aerobic scope (FAS_sus; Equation 8).

$$\mathrm{AM}{\mathrm{R}}_{\mathrm{sus}}=\mathrm{AA}{\mathrm{S}}_{\mathrm{sus}}+\mathrm{RMR}$$ (7)

$$\mathrm{FA}{\mathrm{S}}_{\mathrm{sus}}=\frac{\mathrm{AM}{\mathrm{R}}_{\mathrm{sus}}}{\mathrm{RMR}}$$ (8)

3.3. U _gt

To determine U _gt, we first calculated burst rate (burst s⁻¹) at each swim speed by dividing the corresponding burst count with swimming duration. Burst rate (response variable) and swim speed (U _rel) were then related using a piecewise/segmented mixed effects model (Muggeo et al., 2016) where the inflection point is U _gt. (Example model fits are displayed as black lines in Supporting information Figure S2). The model is structured as follows (Equation 9): an intercept (β _0j ) and a slope coefficient (β _1j ) that varies as a random effect (denoted by the subscript j) between each “run” or swim trial nested within individual, another slope coefficient (β _2j ) that permits a steeper increase in burst rate after U _gt. U _gt may vary between each run via κ _0j and between swim temperatures (T _{swim_cat}; categorical fixed factor) via κ ₁. Because including T _acc as a factor in the model leads to convergence issues, fish acclimated to 16–21 and 19–24°C had separate models. We omitted data from four swim trials due to unclear burst rate–speed relationship (erratic/wild burst rate observations); this step is important to get a good model fit.

$$\begin{array}{c}\mathbf{\text{Burst Rate}}={\beta }_{0j}+{\beta }_{1j}{U}_{\mathrm{rel}}+{\beta }_{2j}\max \left\{U_{\mathrm{rel}}-{\mathit{U}}_{\mathbf{gt}},0\right\}\\ ={\beta }_{0j}+{\beta }_{1j}{U}_{\mathrm{rel}}+{\beta }_{2j}\max \left\{U_{\mathrm{rel}}-\left({\kappa }_{0j}+{\kappa }_1{T}_{\text{swim\_cat}}\right),0\right\}\end{array}$$ (9)

3.4. Tail beat kinematics

We calculated TBF, TBA, and tail thrust from tail co‐ordinates obtained from DeepLabCut. Co‐ordinates were first rescaled (pixels:cm) with the ratio 166 px/7 cm based on the width of the swim tunnel in pixels and centimeters. We then selected a 10‐s period when fish swam steadily with little‐to‐no ground speed. Throughout, we considered the lateral motion of the tail tip and semiautomatically identified its “turning points” (TPs)—the local maxima and minima in the time series. We averaged lateral distances (Δy) between subsequent TPs (within the 10‐s period) as one TBA value. Over the same 10 s, we calculated one TBF (Hz) using Equation (10), and one tail thrust (in arbitrary units or a.u.) using Equation (11) (Gibouin et al., 2018).

$$\mathrm{TBF}=\frac{0.5{\overbrace{\left(\mathrm{TP}\text{count}-1\right)}}^{\text{count of completed half}-\text{cycles}}}{10\mathrm{s}}$$ (10)

$$\text{Tail thrust}=\mathrm{TB}{\mathrm{A}}^2·\mathrm{TB}{\mathrm{F}}^2$$ (11)

3.5. Statistical analysis

3.5.1. Kinematics and energetic swimming efficiency

We conducted all statistical analysis in R (version 3.6.2). TBA, TBF, tail thrust, and MO_2swim were considered as response variables, and swim speed (U _rel or U _inc; predictors), swim temperature, and acclimation temperature as factors (Supporting information provides the exact model specification). Then, we assessed if the slope (β _slope) of the relationship or its intercept (β _int) depends significantly (α = 0.05) on swim temperature (T _swim; continuous variable), acclimation temperature (T _acc; categorical variable), and/or their interactions, using ANOVA (type III). Models that contained multicollinear predictors were reduced (partly to avoid strong multicollinearity) by assuming that T _swim had no effect on the intercept (for TBA, TBF, tail thrust, MO_2swim models) and no speed‐varying effects (for COT models); visually (based on Figures 4, 5. Models were fitted using function lmer (Bates et al., 2015), rlmer (Koller, 2016), or lme (Pinheiro et al., 2016), depending on residual distributions (further discussed in Supporting information).

FIGURE 4.

Cost of transport (COT) as a function of swimming speed in Atlantic salmon (Salmo salar) acclimated to 16–21°C (turquoise; n = 6) and 19–24°C (pink; n = 6) diel thermal cycles ([a]; [COT_net] and [b] [COT_gross]), swum at 16, 18, 20, 22, and 24°C (dark‐to‐light blue color scale; (c) (COT_net); and (d) COT_gross). Observations are shown as open circles; filled squares represent predictions from robust mixed models (details in Supporting Information); NS denotes no significant (p > 0.05) effect of temperature treatments.

FIGURE 5.

Gait transition speed (U _gt) in Atlantic salmon (Salmo salar) acclimated to 16–21 (turquoise; n = 6; [a]) and 19–24°C (pink; n = 6; [b]) diel thermal cycles, then swum at 16, 18, 20, 22, and 24°C. Each observation is represented by a colored translucent point; points belonging to the same fish are joined by translucent lines; black points represent means; opaque colored points represent estimated marginal means with 95% CIs (bars). Gray horizontal lines mark every 10% decline in U _gt from its maximum value for each acclimation group.

To assess the link between kinematic and energetic swimming efficiency, we correlated fish‐specific β _slope from tail thrust and MO_2swim models. We assessed their Pearson's correlation coefficient significance (α = 0.05) using two‐tailed t‐test.

3.5.2. Sustainable swim speed and metabolism

We related U _gt, AAS_sus, AMR_sus, and RMR (response variables) to swim temperature (T _{swim_cat}) and acclimation cycling temperature (both categorical predictors) using mixed effect models (function lme or lmer, details in Supplemental Materials). Then, we tested the statistical significance (α = 0.05) of T _{swim_cat}, T _acc, and their interactions using ANOVA (type III).

4. RESULTS

4.1. Standard and maximum metabolic rate

Chronic acclimation to cool and warm temperature cycles did not affect S. salar RMR or AMR (acclimation temperature; p = 0.53 and p = 0.802, respectively). However, both RMR and AMR increased with increased acute swimming temperatures and RMR more than doubled every 8°C of acute warming (test temp; p < 0.001; Table S1; Figure 2) from an average of 80 to 172 mgO₂ kg⁻¹ h⁻¹ at 16 and 24°C, respectively. Based on these averages, RMR is expected to increase by 2.6‐fold for every 10°C of acute warming (Q₁₀ = 2.6). Mean RMR (averaged across test temperatures) of 16–21 and 19–24°C fish was 124 and 118 mgO₂ kg⁻¹ h⁻¹, respectively. Similarly, both AAS_sus and FAS were not impacted by acclimation temperature, but there was a significant effect of acute swimming temperature (swim temperature; p < 0.001) and an interaction between acclimation cycling temperature and acute swimming temperature on both AAS_sus and FAS (acclimation temperature: swim temperature; p < 0.001 and p = 0.014, respectively).

FIGURE 2.

Oxygen consumption in Atlantic salmon (Salmo salar) acclimated to 16–21°C (turquoise; n = 6) and 19–24°C (pink; n = 6) diel thermal cycles, then swum at 16, 18, 20, 22 and 24°C. Active metabolic rate (AMR_sus; squares) and standard metabolic rate (SMR; triangles; [a]) Sustainable absolute aerobic scope (AAS_sus; [b]), factorial aerobic scope (FAS_sus; [c]). Each observation is represented by a colored translucent point; points belonging to the same fish are joined by translucent lines; black points represent means; opaque colored points represent estimated marginal means with 95% CIs (bars).

4.2. Swimming oxygen consumption

Oxygen consumption rates to support swimming alone (MO_2swim = AMR − RMR; Figure 3) were unaffected by acclimation cycling temperature (swim speed: acclimation temperature; p = 0.476; Table S1) and acute warming (swim speed: swim temperature; p = 0.525; Table S1). MO_2swim only significantly increased with swimming speed (swim speed; p < 0.001) and appears to be resistant to changes in temperature when the thermal effects on RMR are removed.

4.3. Cost of transport

COT_gross, which accounts for energetic costs attributed to both swimming and living, increased with acute warming across the different swim speeds (Figure 4d; swim temperature; p < 0.001) but was not affected by acclimation cycling temperature. However, when we calculate COT_net (Figure 4a,c), which is attributed to swimming metabolism only (MO_2swim), it was unaffected by acclimation cycling temperature (acclimation temperature; p = 0.12; Table S2) and acute warming (swim temperature; p = 0.50). The net cost of transport was relatively stable within aerobic sustainable swimming speeds tested here.

4.4. Sustainable swimming speed and aerobic scope

U _gt varied with acute warming differently for S. salar at different acclimation cycling temperatures (acclimation temperature: swim temp; p < 0.001; Table S1). Fish acclimated to 16–21°C exhibited a linear increase in U _gt from 16 to 24°C (Figure 5a). In contrast, along the same thermal range, U _gt was roughly constant for fish acclimated to 19–24°C (Figure 5b).

4.4.1. Swimming tail beat kinematics

Tail beat amplitude and frequency (Figure 6a–d) were not affected by acclimation cycling temperature (acclimation temperature; p > 0.05, and acclimation temperature: swim speed; p > 0.05; Table S3) but were significantly affected by acute warming (swim temperature; swim speed; p = 0.001). Tail beats widened and became less rapid from 16 to 24°C, more so at relatively high swimming speeds (Figure 6b,d). The total tail thrust (TBA²TBF²) remained unaffected by swim temperature (swim temperature: swim speed; p = 0.30) and acclimation regime (acclimation temp; swim speed; p = 0.42), indicating unchanged tail mechanical effort for swimming.

FIGURE 6.

Tail beat amplitude (TBA), tail beat frequency (TBF), and tail thrust (TBA²TBF²) in Atlantic salmon (Salmo salar) acclimated to 16–21°C (turquoise, n = 6) and 19–24°C (pink; n = 6) diel thermal cycles ([a], [c], [e]), then swum at 16, 18, 20, 22, and 24°C (dark‐to‐light blue color scale; [b], [d], [f]). Each small panel shows data from one fish acclimated to either 16–21 (top row) or 19–24°C (bottom row). Lines depict mixed model predictions (equations in Supporting Information); bands reflect 95% CIs; NS denotes no significant (p > 0.05) effect of temperature treatments.

5. DISCUSSION

Due to the ecological relevance and biological influence of diel thermal cycles, experimental studies examining temperature effects on fishes have now begun to incorporate thermal variation. Here, we investigated how chronic acclimation to a cool and warm diel thermal cycle and acute warming affect the swimming kinematics, energetics, basal metabolism, and sustainable aerobic swimming capacity of wild S. salar. Contrary to our predictions based on the putative thermal threshold for wild S. salar (Breau, 2013; Breau et al., 2011), we determined that fish acclimated to 16–21 or 19–24°C did not exhibit any differences in RMR or AAS_sus, nor were there differences in swimming efficiency, as measured through MO_2swim and COT_net. Swimming kinematics were also unchanged by acclimation to either thermal cycle, and therefore, tail thrust and the U _gt were also similar between the two thermal profiles. However, we did determine that acute warming, as expected, increased the RMR and, thus, the absolute cost of swimming but did not affect the overall swimming efficiency. Interestingly, even at 24°C, which is a temperature beyond where wild S. salar show behavioral stress responses (Breau, 2013; Breau et al., 2011), we saw no impairments in any measures of metabolism, swimming kinematics, or overall aerobic sustainable swimming performance, at least over the acute measurement period. Our data suggest that acute exposure to 24°C after acclimation to 16–21°C or 19–24°C thermal cycles may not be stressful for these S. salar, at least in terms of their swimming capacity, and underscore the importance of incorporating variable thermal history for understanding thermal effects on wild fish.

5.1. Diel thermal acclimation and metabolic rate

As expected from the thermodynamic effects of temperature, acute warming increased RMR in our S. salar like in other salmonids (Gilbert et al., 2020; Poletto et al. 2017; Steinhausen et al., 2008) but was unaffected by thermal acclimation to either diel cycle. We know that thermal compensation of metabolic rate can counter acute increases in RMR through warm acclimation by subsequently decreasing RMR to limit energy expenditure (Kochhann et al., 2015; McDonnell & Chapman, 2015). For example, warm acclimated sculpins (Myoxocephalus scorpius L.) demonstrate full thermal compensation, completely losing the initial increase in RMR after transfer from 10 to 16°C (Sandblom et al., 2014). In S. salar acclimated to stable temperatures (18 vs. 23°C; Hvas et al., 2017), RMR increased only by 1.3‐fold every 10°C (Q₁₀), significantly less than the typical Q₁₀ of 2–3 for this fish under acute warming (Bowden et al., 2018; Penney et al., 2014; Tunnah et al., 2017). Interestingly, the compensatory effect of acclimation on RMR appears absent in our 19–24°C fish where average RMR is only 6 mgO₂ kg⁻¹ h⁻¹ lower than those in 16–21°C fish. There are few studies investigating metabolic thermal compensation in aquatic animals acclimated to cooler versus warmer variable temperatures; however, those that do seem to also suggest a lack of thermal compensation of routine metabolic rate under variable thermal conditions (subtropical goby [Bathygobius cocosensis Bleeker 1854] da Silva et al., 2019; striped marsh frog [Limnodynastes peronii Duméril and Bibron 1841] Niehaus et al., 2011; lake whitefish [Coregonus clupeaformis Mitchill, 1818] Eme et al., 2018; abalone [Haliotis discus hannai Ino 1953] Kang et al., 2019; Littorina saxatilis Olivi 1792 and L. obtusata L. McMahon et al., 1995). Without compensation of RMR, fish at warmer thermal cycles (e.g., 19–24°C) would incur higher energetic costs of living (i.e., RMR), which could potentially limit their aerobic scope and capacity for daily life‐history activities like growth, swimming, and reproduction. Indeed, lake whitefish exposed to thermal variability during development are smaller and have a higher mortality rate (Eme et al., 2018). Despite the lack of metabolic thermal compensation during the acclimation process in thermally variable environments, there are undoubtedly other physiological parameters that are changing to support life in these environments. However, given the increase in intensity, frequency, and duration of more extreme and variable weather events, we need to understand the ability of fishes to thermally compensate increases in RMR in warm, thermal fluctuations and if this does, in fact, create a greater energetic cost.

The AAS_sus in these fish was also unaffected by acclimation to thermal cycles, as the acute effects of temperature increased aerobic swimming capacity and AMR to a similar extent. The speed limit for sustainable swimming U _sus, and its proxy U _gt, is expected to initially increase with acute warming based on the greater power output of red muscles (Rome et al., 1992; Swank & Rome, 2001) until aerobic scope becomes limiting causing U _sus/gt to decline. Here, U _gt represents the speed where burst‐type swimming begins (Figure S2). We found that U _gt only increased with acute warming in 16–21°C fish, whereas fish at 19–24°C maintained a constant U _gt. In the Pacific salmon (Oncorhynchus nerka Walbaum 1792), several studies have shown that the speed threshold called U _crit for prolonged swimming (sustainable for 20 s–200 min; Brett 1964) increases then decreases considerably with rising temperatures (Chen et al., 2015; Eliason et al., 2011; Lee et al., 2003). The decrease at higher temperatures has been attributed to a reduction in the aerobic scope due to limitations of the heart (Eliason et al., 2013; Farrell et al., 2008; Steinhausen et al., 2008). Here, our wild S. salar, AAS_sus, and U _gt stayed high (> 90% of the maximum) at 24°C. The robustness of swimming performance to warm temperatures seems consistent with other studies on S. salar acclimated to a stable 23°C (U _gt and U _crit > 90% of the maximum; Hvas et al., 2017) or acutely warmed to 25°C (U _crit > 90% of the maximum; Zhang et al., 2018). Studies have shown that S. salar show behavioral signs of stress at 23°C (Breau, 2013; Breau et al., 2011), and mortality increases in hatchery fish (Gamperl et al., 2020); however, we determined that swimming performance is maintained at 24°C in wild S. salar acclimated to these diel thermal cycles. Moreover, aerobic scope for S. salar has been predicted to decline from 18°C and reach null at 25°C. This prediction is based partly on knowledge of O. nerka (Breau, 2013) but lacks empirical support. Our data, and that of others highlighted earlier, suggest that S. salar can sustain aerobic scope at warm temperatures when acclimated to ecologically relevant thermal cycles or to constant thermal acclimations of 23°C (Hvas et al., 2017), which supports maintenance of swimming performance at these warmer temperatures.

5.2. Metabolic/energetic swimming efficiency

Due to the effects of temperature on red muscle and general swimming performance, we predicted that our warm acclimation diel thermal cycle would affect S. salar MO_2swim or COT_net, but this was not the case. Warm acclimation has previously been shown to compensate for the reductions in efficiency of mitochondrial processes that occurs with acute warming (Baris et al., 2016; Chung & Schulte, 2015; Roussel & Voituron, 2020). For example, S. salar acclimated to 20°C showed a 20%–30% increase in mitochondria efficiency compared to the 12°C acclimation group tested at a common temperature (Gerber et al., 2020). Additionally, warm acclimation can alter the contraction efficiency–velocity curves for red muscle (Rome, 2007), potentially affecting the swimming efficiency of fishes. It is not clear why we did not observe an acclimation effect of diel thermal cycle on MO_2swim, but it is tempting to speculate that this lack of compensation may be a result of the thermally variable nature of the acclimation environment. Notably, fish in diel thermal cycling environments are thought to potentially lose some capacity to acclimate (Seebacher et al., 2012; James & Tallis, 2019; Brown et al., 2024). Rapid diel thermal variations may obscure the cue for long‐term seasonal thermal acclimation (Angilletta, 2009; Gabriel et al., 2005).

Like acclimation temperature, we did not find a significant overall effect of acute warming on swimming efficiency. In fishes, the speed of swimming has been known to correlate with red muscle contraction speed (Rome, 2007; Rome et al., 1990). Notably, the most efficient contraction speed shifts to faster domains with acute warming (Ferguson et al., 2002), meaning improved efficiency at faster swim speeds and reduced efficiency at lower speeds with acute warming, as observed. Although not the focus of our study, our data may provide the first evidence that the speed‐specific temperature effect on red muscle contraction efficiency translates to whole fish swimming efficiency. Still, the biological implication of this effect seems small with the largest COT_net variation in only ~10% over 8°C at 1.8 BL/s. More importantly, we showed for the first time that S. salar swimming efficiency is temperature insensitive, in the face of acute warming or distinct chronically diel cycling warm temperatures (16–21 vs. 19–24°C), at least under our thermal conditions. Other studies showed that the energy use while swimming, specifically the AMR, increased with acute warming in juvenile S. salar (Alexandre & Palstra, 2017) and adults (Lennox et al., 2018), but it was not indicated why AMR increased in this study. This increase could be attributed to greater swimming cost (MO_2swim), which relates to poorer swimming efficiency, or to an increased cost of living (RMR). To the best of our knowledge, there are yet to be any other studies investigating swimming performance/efficiency of fish acclimated to thermally variable environments, and this will be an interesting area of research going forward.

5.3. Tail beat kinematics

S. salar TBF and TBA were surprisingly unaffected by the different acclimation thermal cycles, which likely contributed to the lack of effect in swimming efficiency. Fish exposed to different stable temperatures have been shown to possess modified red muscle through hypertrophy or changes in maximum shortening velocity (Egginton & Sidell, 1989; Rome, 2007; Shuman & Coughlin, 2018; Gamperl & Syme 2021), which could potentially influence tail beat kinematics. Stevens (1979) is one of the few studies that have demonstrated an effect of different stable acclimation temperatures on TBF in fishes (rainbow trout [Oncorhynchus mykiss Walbaum1792] and largemouth bass [Micropterus salmoides Lacépède 1802]). It is possible that the diel thermal cycling acclimation condition in our experiment reduced the acclimation response on TBF and TBA, although the exact reason is not known, and this area needs further study.

In contrast to the lack of effect of acclimation to our diel cycles, S. salar TBF and TBA varied with acute warming from 16 to 24°C. Changes in tail beat kinematics could result in different mechanical work required/exerted to swim at different temperatures, but an effect of temperature on the amount of tail thrust at a given speed was not found. Salmonids subjected to acute warming have consistently shown alterations in tail beat kinematics although the direction of effect has been variable (Stevens, 1979; Nudds et al., 2014; Nudds et al., 2020; Lea et al., 2016). In our S. salar, acute warming widened tail beats, implying that greater lateral work is being done by the tail, which suggests an increase in tail thrust (∝ TBA²TBF²; Gibouin et al., 2018). However, TBF (notably also a common predictor for MO_2swim; Steinhausen et al., 2005; Zupa et al., 2021) is simultaneously reduced with acute warming, indicating less work done per unit time and may be a result of the impact of temperature on myosin ATPase activity (Watabe, 2002). Thus, tail thrust was observed to remain constant regardless of acute warming or acclimation temperature. Fish exposed to different stable acclimation temperatures have previously been shown to be morphologically different, which could result in different body drag and the amount of tail thrust needed for swimming (Nesteruk et al., 2014; Tytell et al., 2010; Zhong et al., 2019). However, differences in body shape and condition factor did not occur in our experiment (Table 1). The swimming angle of attack (Liao, 2007) and head yaw (Ding et al., 2022) have also been known to influence body drag and presumably mechanical swimming efficiency and required tail thrust, but these were not studied in this experiment. Ultimately, however, no temperature effect was found on the S. salar tail thrust in individuals acclimated to diel thermal cycles.

5.4. Conclusion

Previous understanding based on stable, warm acclimation temperatures suggests temperatures beyond 23°C are stressful for S. salar. However, our data demonstrate that swimming performance of wild fish is not affected beyond these temperatures, at least when acclimated to diel thermal cycles. Overall, we found no clear negative effect of warm temperature (24°C) on S. salar acclimated to ecologically relevant diel thermal cycles, except for the uncompensated increase in RMR. Energetic swimming efficiency, as assessed through MO_2swim and COT_net, was not affected by temperature. However, RMR increased with acute warming, without compensatory decreases from warm acclimation in the 19–24°C group. This suggests S. salar exposed to 19–24°C may have greater living cost, which is a potential concern for energy supply and survival. U _sus and AAS_sus did not fall below 90% of the maximum at 24°C after acute warming from 16 to 24°C for both 16–21 and 19–24°C acclimated fish. We do note that the warmest temperature used here (24°C) is less than the maximum temperature recorded in the Miramichi (varies across sites from 25.2 to 32.9°C) and its average summer maximum of 25.9°C (Caissie et al., 2013). As we used wild fish in this experiment, who regularly would be exposed to temperatures higher than 24°C, it is possible that these chosen thermal cycles would not have an effect. In contrast, many past studies have used cultured fish, which can exhibit much different responses than the same species captured from natural habitats (Morgan et al., 2022). Our use of wild fish may explain some of the differences between our study and past studies on S. salar using cultured fish. The effects of climate change on thermal variability and extreme acute changes in temperature in freshwater habitats are likely to impact all life stages and physiological aspects of fish. The data presented here will be crucial for a holistic understanding of the biology and ecology of threatened fish species like S. salar (Gillis et al., 2023).

AUTHOR CONTRIBUTIONS

A.J. Morash, S. Currie, and S. Andrew conceived the study design. S. Andrew conducted all the laboratory experiments, statistical analysis, and the first draft of the manuscript. All authors contributed to writing the final draft of the manuscript.

FUNDING INFORMATION

This work was funded by the Atlantic Salmon Research Joint Venture (A.J. Morash and S. Currie), NSERC Discovery Grants (A.J. Morash and S. Currie), and the New Brunswick Innovation Foundation (S. Andrew).

CONFLICT OF INTEREST STATEMENT

The authors declare that there are no competing interests.

Supporting information

Data S1.

Andrew, S. , Currie, S. , & Morash, A. J. (2025). The effects of warm thermal variability on metabolism and swimming performance in wild Atlantic salmon (Salmo salar). Journal of Fish Biology, 106(3), 893–907. 10.1111/jfb.15996

DATA AVAILABILITY STATEMENT

Data generated or analysed during this study will be available in the Borealis Data Repository upon acceptance. https://doi.org/10.5683/SP3/8U7G03.