Temperature-dependent plasticity mediates heart morphology and thermal performance of cardiac function in juvenile Atlantic salmon (Salmo salar)

Anthropogenic climate change is projected to have extensive impacts on biodiversity, as temperature highly constrains the physiological processes of ectothermic organisms (Dillon et al., 2010; Kappelle et al., 1999). As average global temperatures increase, it is also projected that the frequency and severity of transient heat waves will increase (IPCC, 2021). Organisms will therefore have to contend with warmer temperatures throughout their lives, as well as unprecedented acute temperature events. Over short time frames, the ability of organisms to cope with these thermal challenges will depend predominantly on their current proximity to upper thermal limits and their capacity to respond to warmer temperatures through phenotypic plasticity (Crozier and Hutchings, 2014; Huey et al., 2012; Pörtner and Peck, 2010; Somero, 2010). Phenotypic plasticity in response to temperature can be partitioned into developmental plasticity and reversible thermal acclimation. Developmental plasticity refers to the shaping of phenotypic traits by early life environments, which can occur across developmental stages (Angilletta and Angilletta, 2009). Developmental plasticity can have profound, and often permanent, effects on traits expressed throughout an organism's life. In contrast, thermal acclimation occurs within a single life stage, often in a reversible manner (Angilletta and Angilletta, 2009). In the face of rapid climate warming, it is important to understand the capacity for plasticity in ectotherms, as well as the proximate mechanisms through which plasticity in thermal tolerance is achieved.

Aerobic demands increase with temperature in ectotherms and, as such, cardiorespiratory traits are thought to play an important role in the response to warming. In fish, increasing heart rate is one of the primary mechanisms for meeting increased tissue oxygen demands during acute warming (Eliason and Anttila, 2017; Farrell, 2009). Heart rate increases rapidly with temperature until an inflection point known as the Arrhenius breakpoint temperature (T_AB), after which temperature-dependent increases in maximum heart rate (f_H,max) begin to slow (Anttila et al., 2013a; Casselman et al., 2012). With further increases in temperature, the heart eventually becomes arrhythmic at the cardiac arrhythmia temperature (T_arr), which represents a collapse of normal cardiac function (Anttila et al., 2013a; Casselman et al., 2012). T_AB has been proposed to serve as a measure of the optimal temperature for cardiac performance, and T_arr as a measure of the upper limit for cardiac function (Anttila et al., 2013a; Casselman et al., 2012). In Pacific salmon, these cardiorespiratory measures have been used to predict temperature-related mortality in the wild and assess vulnerability to environmental change (Cooke et al., 2012; Farrell et al., 2008; Muñoz et al., 2015). Acute warming experiments have also demonstrated that cardiac output (the product of heart rate and stroke volume) peaks and subsequently declines at temperatures below the upper critical temperature in most fish species (Clark et al., 2008; Ekström et al., 2016; Eliason et al., 2013; Farrell et al., 1996; Gollock et al., 2006; Mendonça and Gamperl, 2010; Penney et al., 2014; Sandblom and Axelsson, 2007). Cardiac performance has therefore been implicated as an important contributor to upper thermal limits in fishes (Farrell, 2009; Pörtner and Knust, 2007).

Traits conferring increased cardiac capacity have been linked to improved upper thermal limits in many fishes. In Atlantic salmon (Salmo salar), Chinook salmon (Oncorhynchus tshawytscha), redband trout (Oncorhynchus mykiss gairdneri) and the common killifish (Fundulus heteroclitus), individuals with higher upper thermal limits were shown to have higher peak heart rate (Anttila et al., 2014b; Chen et al., 2018; Muñoz et al., 2015; Safi et al., 2019). Relative ventricle size, which correlates with stroke volume (Franklin and Davie, 1992), was positively correlated with upper thermal tolerance in both Atlantic salmon (Anttila et al., 2013b) and European sea bass (Dicentrarchus labrax; Ozolina et al., 2016). Oxygen supply to the heart itself is also thought to limit cardiac performance and thermal tolerance in fishes (Clark et al., 2008; Ekström et al., 2017; Farrell, 2009). In over half of all teleost species, the ventricular wall is composed solely of a spongy myocardium that receives its oxygen supply from the oxygen that remains in the luminal venous blood after the oxygen demands of systemic tissues have been met (Davie and Farrell, 1991). In athletic teleosts, such as salmonids, the spongy myocardium is encased by a compact myocardium which receives an oxygen-rich coronary blood supply from the gills (Davie and Farrell, 1991). The relative thickness of the compact myocardium appears to vary with the metabolic requirements of a fish's life history, both interspecifically and intraspecifically (Davie and Farrell, 1991; Eliason et al., 2011; Graham and Farrell, 1992; Pombo et al., 2012; Santer and Walker, 1980). In Oncorhynchus mykiss, Graham and Farrell (1992) found that anadromous fish had thicker compact myocardial layers compared with lake-dwelling conspecifics – presumably to help meet the greater aerobic demands associated with migration. Similarly, Eliason et al. (2011) demonstrated that sockeye salmon (Oncorhynchus nerka) with more metabolically challenging migration routes possessed larger ventricles with improved coronary blood supplies. The pyramidal shape of the salmonid ventricle is also pivotal to optimal cardiac performance, with ventricular roundness negatively correlating with cardiac output and critical swim speed (Claireaux et al., 2005). Altogether, there is a clear link between heart morphology, cardiac performance and aerobic capacity in active fishes such as salmonids.

In recent years, extreme heat events have had catastrophic impacts on migration success and survival across species of salmonids (Baisez et al., 2011; Farrell et al., 2008; Hinch et al., 2012; Martins et al., 2011; Martins et al., 2012; Munoz et al., 2020), which numerous studies have attributed to cardiorespiratory limitations at elevated temperatures (Crossin et al., 2008; Eliason et al., 2011; Farrell et al., 2008; Martins et al., 2012). While thermal tolerance in salmonids tends to be locally adapted to the historical conditions of ancestral rivers (Chen et al., 2015; Eliason et al., 2011; Gradil et al., 2016; Poletto et al., 2017; Verhille et al., 2016), experimental manipulations have demonstrated a significant role for phenotypic plasticity in determining the thermal performance of cardiac function. Acclimation to a common temperature can erase population differences in upper thermal tolerance (Anttila et al., 2014b), while developing under warmer thermal conditions can increase the optimal temperature for cardiac performance (Muñoz et al., 2015). In some fish species, it is thought that thermal acclimation during later life stages may underestimate the organism's full plastic potential by excluding persistent effects of developmental plasticity (Crozier and Hutchings, 2014). As rates of climate warming threaten to outpace adaptation (Morgan et al., 2021; Muñoz et al., 2015), it is important to understand how the cardiorespiratory system of fishes can respond to a warmer environment via phenotypic plasticity, including both developmental plasticity and reversible acclimation effects.

In Atlantic salmon, it remains unresolved whether exposure to warmer average temperatures starting in early development can maintain or improve cardiac performance when juvenile fish are exposed to short-term heat stress. The purpose of this study was to assess the capacity for phenotypic plasticity in the acute thermal tolerance of juvenile Atlantic salmon, while investigating potential mechanistic underpinnings in heart morphology and function. To this end, we reared Atlantic salmon under control (+0°C) or elevated (+4°C) temperature conditions from fertilization. In juvenile salmon, we measured the response of f_H,max to warming using a non-invasive Doppler echocardiograph system (Muir et al., 2021), in order to determine cardiorespiratory thresholds for thermal performance (T_AB, T_arr) in the two temperature treatments. In addition to measures of f_H,max, we simultaneously measured blood flow at the atrioventricular valve. Stroke distance, the distance travelled by the blood during a cardiac cycle, was used as a proxy for how stroke volume responds to acute warming (as volume is calculated by multiplying stroke distance by valve area). Measures of f_H,max and stroke distance were also used in conjunction with estimates of atrioventricular valve area to calculate derived cardiac output (Q_D) as a function of temperature (Muir et al., 2021). Using histology, we examined morphological characteristics of the heart that were previously associated with improved aerobic performance in salmonids, comprising the proportion of compact myocardium in the ventricle, ventricular roundness and relative ventricle size. We additionally examined relative ventricle size via measures of relative ventricular mass (RVM). If there is phenotypic plasticity in the thermal performance of heart function, we would expect fish raised at elevated temperatures to have higher upper thermal limits for cardiorespiratory performance. We predicted that plasticity in upper thermal limits would be associated with differences in maximum cardiovascular capacity. Furthermore, we predicted that plasticity in thermal performance would be associated with morphological changes in the heart that enhance aerobic capacity in salmon (e.g. larger, more pyramidal ventricles with higher proportions of compact myocardium). The overall goal was to elucidate how thermal conditions experienced from the start of development affect thermal performance phenotypes later in life and provide insight into the mechanistic drivers of improved thermal tolerance at the level of the heart.

Atlantic salmon, Salmo salar Linnaeus 1758, descended from the LaHave River population (Nova Scotia, Canada; 44.4°N, 64.5°W) were obtained as fertilized eggs from the Ontario Ministry of Natural Resources and Forestry (OMNRF) Normandale Research Facility (Vittoria, ON, Canada) in the autumn of 2016 and 2017. This population has been reared in OMNRF hatcheries for 6 generations, and in a recent study continued to show local adaptation to the thermal conditions of the LaHave River (Gradil et al., 2016). Eggs were transported to a hatchery facility at the University of Western Ontario 24 h after fertilization and housed in egg trays corresponding to one of two thermal treatments: control (+0°C) or elevated (+4°C). The rearing temperatures selected for early development were 7°C for the control treatment and 11°C for the elevated treatment, both of which fall within the natural range for the autumn spawning season in the LaHave River (Harding Gradil, 2015). We selected a 4°C temperature differential to follow a previous assessment of plasticity in thermal tolerance in juvenile Chinook salmon (Muñoz et al., 2015). OMNRF recommends that Atlantic salmon eggs are incubated between 7 and 10°C in aquaculture (OMNRF, 2010). We therefore selected 7°C as the egg incubation temperature for the control treatment, as this would allow for a 4°C temperature differential between treatments, while keeping the elevated temperature group close to the recommended incubation range (11°C).

Rearing temperatures were held constant within 0.5°C of the specified temperatures. Daily and seasonal fluctuations in water temperature were excluded in order to separate potential effects of seasonal temperature variation from the effects of average rearing temperature. Atlantic salmon develop in a temperature-dependent manner, such that fish raised in warmer temperatures complete endogenous feeding and emerge from the substrate earlier than those reared at lower temperatures (approximately 4 months post-fertilization for control fish and 2.5 months for elevated temperature fish). For this reason, experiments were staggered between temperature treatments based on the difference in time to exogenous feeding (Muñoz et al., 2015). Given this difference in developmental rate, allowing rearing temperature to fluctuate seasonally could have contributed to confounding effects on thermal tolerance based on when trials were performed. Instead, fry were transferred to 40 l tanks at the conclusion of endogenous feeding and the rearing temperatures were raised by 4°C to mimic the seasonal increase in temperature that cues the transition to exogenous feeding. This resulted in juvenile rearing temperatures of 11°C for the control treatment and 15°C for the elevated temperature treatment.

Juvenile fish were provided with pelleted feed ad libitum for the remainder of the rearing period (EWOS Commercial Feeds, Bergen, Norway), and experiments were performed in mixed-sex 1+ yearlings (∼1 year post-fertilization for elevated temperature fish and ∼1.5 years post-fertilization for control fish). All experiments were carried out according to Western University Animal Care protocol 2018-084.

Blood flow at the atrioventricular valve was measured in juvenile Atlantic salmon using the Indus Doppler Flow Velocity (DFV) System, as previously described by Muir et al. (2021). Briefly, test fish were anaesthetized using 100 mg l⁻¹ of MS-222 buffered with 200 mg l⁻¹ sodium bicarbonate, and body mass was then measured. Anaesthetized fish were stabilized in a weighted sling (ventral-side up) within a holding reservoir to ensure correct orientation for placement of the Doppler probe. Individuals were ram-ventilated and maintained at their respective rearing temperatures (11°C or 15°C) with recirculating water from a temperature-controlled water bath (VWR, Edmonton, AB, Canada) containing a maintenance dose of anaesthetic (75 mg l⁻¹ of MS-222 buffered with 150 mg l⁻¹ sodium bicarbonate). Reservoir water temperature was monitored with a digital thermometer (Omega, St-Eustache, QC, Canada).

Following a 15 min stabilization period in the holding reservoir, the 20 MHz transducer probe was held perpendicular to the ventral side of the fish, posterior to the gills, such that the direction of the probe's ultrasound beam was parallel to the direction of blood flow through the atrioventricular valve of the heart. Signals were digitized and displayed as continuous, real-time grayscale DFV spectrograph waveforms in the Doppler Signal Processing Workstation software (Indus Instruments, Houston, TX, USA). Baseline spectrographs were visually examined to ensure normal heart function before starting the thermal performance trial. Fish were then pharmacologically stimulated to reach f_H,max using intraperitoneal injections of 1.2 mg kg⁻¹ atropine sulphate (Sigma-Aldrich, St Louis, MO, USA) and 4 μg kg⁻¹ isoproterenol (Sigma-Aldrich) dissolved in 0.9% NaCl, following Casselman et al. (2012). Atropine sulphate was administered to block vagal tone, while isoproterenol was used to fully stimulate adrenergic β-receptors. Each injection was followed by a 15 min equilibration period, then a stepwise temperature increase was applied in 1°C increments every 6 min (Casselman et al., 2012). After each 1°C increase, we ensured that water in the holding reservoir had stabilized at the desired temperature for a few seconds before recording spectrograph waveforms of atrioventricular blood flow in 30 s intervals (10 per temperature). Once a shift from rhythmic to arrhythmic heart beats appeared in the spectrogram, the fish was removed from the apparatus and euthanized by lethal overdose of MS-222. In a subset of fish (N=6 per treatment), the ventricle was excised, rinsed twice with PBS to remove excess blood, and weighed to determine RVM. RVM was calculated using the formula RVM=(M_V/M_B)×100%, where M_V is ventricle mass (g) and M_B is body mass (g). The remaining fish were prepared for histological sectioning, described in detail below.

Velocity spectrographs were later processed in Doppler Signal Processing Workstation to calculate f_H,max and total stroke distance at each temperature, following Muir et al. (2021). Measures of f_H,max and stroke distance were subsequently used to calculate derived cardiac output across temperature, in conjunction with estimates of atrioventricular valve area (described in detail below). From these data, we determined peak f_H,max, stroke distance and derived cardiac output for each fish, and identified the temperatures that these peak measures occurred at (T_peakfH, T_peakSD and T_peakQD, respectively). Given the difference in growth rate between the temperature treatments, our initial sampling design resulted in the control fish being substantially smaller than the elevated temperature fish. To address a potential confound between body size and the experimental treatments, our analyses included only the subset of fish which fell within the overlapping range of body sizes between treatments (16–35 g; N=13 for +0-treatment, N=16 for +4-treatment). An additional 15 fish were removed because the atrioventricular flow signal was of poor quality, or the pharmacological injections induced an immediate cardiac arrythmia.

Measures of f_H,max from Doppler spectrographs were also used to determine T_AB of f_H,max and T_arr. T_AB was calculated in SigmaPlot 13.0 (Systat Software, San Jose, CA, USA) as described by Muñoz et al. (2015). Briefly, an Arrhenius plot was generated for each fish by plotting the natural logarithm of f_H,max as a function of the inverse of temperature (in K). A piecewise, two-segment linear equation was applied to fit a biphasic line to the data using the software's ‘Dynamic Fit Wizard’ tool. The point at which the slope changed, representing the point at which temperature-induced increases in f_H,max shift to a lower exponent, was identified as T_AB (Yeager and Ultsch, 1989). T_arr, which signifies the upper thermal limit for cardiac performance, was identified as the temperature at which arrythmias appeared in the spectrograph waveforms.

To ensure that hearts were in diastole prior to fixation, an intraperitoneal injection of 100 mg kg⁻¹ KCl was administered immediately prior to euthanasia. Euthanized fish were then fixed in 10% neutral buffered formalin (VWR, Radnor, PA, USA). After 48 h, the midsection of the fish was trimmed (anterior to the eyes and posterior to the pectoral fins) and returned to 10% neutral buffered formalin for an additional 72 h. Following serial dehydration, the dissected specimen was stored in 70% ethanol and sent to the Robarts Molecular Pathology Facility (London, ON, Canada) for embedding and histological sectioning. Serial sections (10 µm thickness) of the whole fish were taken through the sagittal plane, traversing the heart, and stained with Haematoxylin and Eosin (H&E). Two individuals were excluded from the analysis because of poor slide quality.

Sections were imaged using an Olympus SZX9 dissecting microscope equipped with an OPTIKA C-B5 camera and measurements were taken in OPTIKA PROView (OPTIKA Srl, Ponteranica, BG, Italy). The diameter of the atrioventricular valve opening was measured in serial sections to determine the maximum diameter (see Figs S1 and S2). To ensure that measures of ventricular morphology were taken at a consistent location within the heart across individuals, only sections that passed through both the atrioventricular channel and the bulbus arteriosus were considered for measures of ventricular area, ventricular roundness and thickness of the compact myocardium, and only fish that fell within an overlapping size range between treatments were used for morphological comparisons (N=5 per treatment). From the subset of slides passing through the atrioventricular channel and the bulbus arteriosus, subsequent measures of ventricle morphology were taken for each fish on the slide where atrioventricular valve diameter was maximized. Area of the ventricle was measured by tracing the outer edges of the ventricle chamber using the ‘Polygon’ setting in OPTIKA PROView. Ventricular roundness was measured as the length to width (L:W) ratio. Ventricle length was measured from the dorsal edge of the bulbus arteriosus to the tip of the ventricular apex. Ventricular width was measured at the midpoint of ventricular length using a perpendicular line in the ‘Vertical>Four Points’ setting in OPTIKA PROView. Thickness of the compact myocardium was measured at the outer edges of this ventricular midpoint line, and converted to a percentage of total ventricular width. Morphological measurements were adapted from Anttila et al. (2015), Claireaux et al. (2005) and Johnson et al. (2014). Representative sections showing morphological measurements are shown in Fig. S1.

To calculate Q_D, we first estimated the radius of the atrioventricular valve for each fish using a linear regression of the relationship between body mass and observed atrioventricular radius (R²=0.59, P<0.001) measured in histological sections for a subset of fish (N=18; Fig. S2). Rearing temperature was not included as a predictor in this analysis, as its inclusion did not improve model fit. The area of the atrioventricular valve was then estimated assuming a circular shape using the formula πr², where r represents valve radius. Q_D (ml min⁻¹ kg⁻¹) was calculated by multiplying stroke distance (cm) by atrioventricular valve area (cm²) and f_H,max (beats min⁻¹) at a given temperature, and divided by body mass (kg).

Body mass was compared between treatments using an unpaired t-test with Welch's correction, and RVM was compared using an unpaired t-test. Measures of thermal performance (T_AB, T_arr, T_peakfH, T_peakSD and T_peakQD), peak cardiac function (peak f_H,max, peak stroke distance and peak Q_D) and histological measures of heart morphology (ventricle size, ventricular L:W ratio, percentage compact myocardium) were examined using one-way ANCOVA that included rearing treatment (+0, +4) as a fixed factor and body mass as a covariate. To examine how cardiac function parameters responded to increasing temperature ‒ and test whether these responses differed between rearing treatments ‒ f_H,max, stroke distance and Q_D were analysed across temperature using a linear mixed model with AR (1) as the covariance structure, measurement temperature as the repeated measure, body mass as a covariate, and rearing treatment, measurement temperature and their interaction as fixed factors. Statistical analyses were performed in SigmaPlot 13.0, GraphPad Prism 6 (GraphPad Software, La Jolla, CA, USA), or SPSS v. 27 (IBM, Chicago, IL, USA).

Only fish within an overlapping size range across treatments were included in comparisons of thermal performance, cardiac function and heart morphology (16–35 g). Within this subset, there was no significant difference in body mass between fish in the +4-treatment (26.2±5.4 g) and fish in the +0-treatment (22.3±5.3 g; t₂₇=2.0, P=0.054). Body mass was included as a covariate in all of our analyses. Body mass was not a significant covariate for peak f_H,max (F_1,26=2.2, P=0.15), but was associated with the temperature at which peak f_H,max occurred (T_peakfH; F_1,26=9.0, P=0.01). Body mass was not a significant covariate for the other thermal performance metrics associated with f_H,max: T_AB (F₁,₂₆=1.4, P=0.24) and T_arr (F_1,26=4.1, P=0.05). Body mass was not a significant covariate for peak stroke distance (F_1,26=1.6, P=0.21), or the temperature at which peak stroke distance occurred (T_peakSD; F_1,26=0.07, P=0.79). Body mass significantly affected peak Q_D (F_1,26=7.9, P<0.01), but did not affect the temperature at which peak Q_D occurred (T_peakQD; F_1,26=3.0, P=0.10).

Within the subset of fish used for histological analysis of the heart, there was no significant difference in body mass between fish in the +4-treatment (26.1±5.5 g) and fish in the +0-treatment (21.3±5.4 g; t₈=1.4, P=0.20). Across this size range, ventricle area varied with body mass (F_1,7=6.2, P=0.04), but did not vary with ventricular L:W ratio (F_1,7=0.96, P=0.36) and percentage compact myocardium (F_1,7=0.01, P=0.94).

To examine the thermal sensitivity of cardiac performance in the two temperature treatments, we examined the responses of f_H,max, stroke distance and Q_D to warming using linear mixed models. Q_D increased with warming in both groups, before reaching a peak and subsequently declining (Temperature effect: F_18,285=3.2, P<0.001; Fig. 1A). Increases in Q_D during warming were mediated by increases in f_H,max (Temperature effect: F_18,278=29.4, P<0.001; Fig. 1B), while stroke distance tended to decline with warming in both groups (Temperature effect: F_18,284=2.1, P=0.008; Fig. 1C). We did not detect a significant treatment effect on Q_D (Treatment: F_1,39=0.18; P=0.7; Fig. 1A), f_H,max (Treatment: F_1,32=2.3; P=0.14; Fig. 1B) or stroke distance (Treatment: F_1,42=0.07; P=0.8; Fig. 1C), nor was there a significant interaction between the effects of rearing treatment and acute warming on Q_D (Temperature×Treatment effect: F_11,280=0.96, P=0.47; Fig. 1A), f_H,max (Temperature×Treatment effect: F_11,274=0.35, P=0.97; Fig. 1B) or stroke distance (Temperature×Treatment effect: F_11,277=0.56, P=0.86; Fig. 1C).

We further examined the thermal performance of cardiac function by comparing key temperatures for f_H,max, stroke distance and Q_D between temperature treatments. The Arrhenius breakpoint temperature for f_H,max (T_AB) was significantly higher for fish from the +4-treatment than for those from the +0-treatment (F_1,26=26.7, P<0.001; Fig. 2A, Table 1). Similarly, +4-treatment fish attained their peak f_H,max at higher temperatures (T_peakfH) than did +0- treatment fish (F_1,26=7.5, P=0.01; Table 1) and maintained cardiac function to higher temperatures before arrythmias occurred (T_arr; F_1,26=12.2, P<0.01; Fig. 2, Table 1). Peak stroke distance also occurred at significantly higher temperatures in +4-treatment fish (F_1,26=26.3, P<0.001; Table 1). Finally, Q_D tended to peak at higher temperatures in the +4-treatment group, but T_peakQD did not differ significantly between treatments (F_1,26=3.6, P=0.07; Table 1).

Despite differences in the thermal performance of cardiac function, we did not detect significant differences in peak cardiovascular capacity between temperature treatments (Table 2). While fish from the +4-treatment maintained rhythmic heart function to higher temperatures on average, peak f_H,max did not differ significantly between temperature treatments (F_1,26=0.7, P=0.40), nor did peak measures of stroke distance (F_1,26=1.5, P=0.23) and Q_D (F_1,26=0.1, P=0.78).

In addition to measures of cardiac function, we investigated heart morphology as a potential mechanistic driver of improved thermal performance in the +4-treatment fish. RVM did not differ between treatment groups, with a mean RVM of 0.13±0.02% in +0-treatment fish and 0.12±0.03% in +4-treatment fish (t₁₀=0.8, P=0.44). Similarly, ventricle area did not differ significantly between rearing treatments when controlling for body mass (F_1,7=0.3, P=0.59; Table 3). Ventricle shape was also similar between treatments, with no significant difference in ventricular L:W ratio (F_1,7=2.8, P=0.14; Table 3). However, fish raised in the +4-treatment displayed significantly higher percentages of compact myocardium within their ventricles than fish raised in the +0-treatment (F_1,7=12.9, P=0.01; Table 3).

In fishes, the heart is typically one of the first organs to fail during thermal stress and has therefore been positioned as an ‘ecological thermometer’ (Iftikar and Hickey, 2013). In the present study, we found a remarkable degree of plasticity in the thermal performance of cardiac function in juvenile Atlantic salmon. Fish reared under warmer temperatures (+4°C) displayed a 5.3°C increase in the mean optimal temperature for cardiac performance (T_AB) and maintained rhythmic heart function (T_arr) to temperatures 1.9°C higher than control fish. However, contrary to our prediction, shifts in the thermal limits for cardiorespiratory performance were not accompanied by treatment-level differences in peak cardiac function. As we discuss below, plasticity may be acting on traits that preserve cardiac function under high temperature conditions, rather than traits relating to maximum cardiovascular capacity.

As aerobic demands increase with temperature, heart rate and stroke volume represent the fundamental components of cardiac output that can be altered to increase oxygen delivery to tissues. In the present study, increases in derived cardiac output during acute warming were driven by increases in maximum heart rate ‒ a seemingly universal finding in fishes (Eliason and Anttila, 2017; Farrell, 2009). Concomitantly with increases in heart rate, we observed declines in stroke distance – a proxy for stroke volume – as temperatures increased. This finding agrees with numerous studies in fishes that have documented an inverse relationship between maximum heart rate and stroke volume during warming (Eliason et al., 2013; Farrell et al., 1996; Keen and Farrell, 1994; Morgenroth et al., 2021). As the frequency of heart contractions increases, diastolic filling between contractions and the systolic force of the contractions are reduced (Driedzic and Gesser, 1988; Farrell et al., 1996; Shiels et al., 2002). Oxygen limitation is also known to reduce cardiac contractility at high temperatures, further constraining stroke volume during acute warming (Ekström et al., 2016, 2017). Because we measured blood flow at the atrioventricular valve, decreases in ventricular inflow may therefore reflect a reduction in cardiac filling time or an increase in the end-systolic volume remaining from preceding contractions (i.e. a decreased ejection fraction resulting from reduced contractility). Though not statistically significant, stroke distance tended to decline more rapidly with warming in +0-treatment fish than in +4-treatment fish (Fig. 1C). Indeed, stroke distance peaked at much lower temperatures in the +0-treatment (Table 1). A preservation of stroke volume during rapid increases in heart rate could allow fish to increase cardiac output more rapidly during warming, though we did not detect a significant interaction between measurement temperature and temperature treatment in these fish.

Indicators of maximum cardiovascular capacity, such as peak heart rate and relative heart size, have previously been linked to increased upper thermal tolerance in fishes. However, estimates of peak Q_D did not differ between our rearing treatments. Peak stroke distance, used here as a proxy for stroke volume, was also unaffected by rearing temperature. This finding is perhaps unsurprising given that fish from the two treatments did not display differences in relative ventricle size or ventricular shape. While fish raised at elevated temperatures displayed higher thermal limits for cardiac function and reached peak f_H,max at significantly higher temperatures, peak f_H,max did not differ between treatment groups. Although increases in maximum heart rate are often observed with warm acclimation in juvenile and adult fish (Anttila et al., 2014b; Safi et al., 2019), exposure to elevated temperatures starting in early development can have varied effects on heart rate. In a Chinook salmon study, peak f_H,max was significantly higher in fish reared under elevated temperatures from fertilization, but there was a lack of plasticity in T_arr (Muñoz et al., 2015). In a sockeye salmon study, a 4°C increase in egg incubation temperature resulted in higher peak heart rate in one population, lower peak heart rate in another population, and had no effect on peak heart rate in the remaining two populations (Chen et al., 2013). Increased egg incubation temperature also had mixed effects on juvenile thermal tolerance across populations in this study. These patterns demonstrate the variation in plastic capacity that exists across fish species, populations and life stages, as well as the diversity in proximate mechanisms through which thermal plasticity is achieved.

In many fish species, chronic changes in temperature elicit shifts in the relative proportion of spongy to compact myocardium within the ventricle (Keen et al., 2017). Here, we observed considerable plasticity in ventricular myoarchitecture, as the percentage of compact myocardium was 70% greater in fish reared at elevated temperatures. Similar plasticity in the cardiac phenotype has been observed following warm acclimation in a number of salmonids, including rainbow trout (Keen et al., 2017; Klaiman et al., 2011), Arctic char and Atlantic salmon (Anttila et al., 2015). In parallel with the present findings, Anttila et al. (2014a) observed a greater proportion of compact myocardium in Atlantic salmon parr with higher optimal temperatures for cardiac function (T_AB) and better swimming performance. These trends toward higher percentages of compact myocardium among more thermally tolerant salmonids may imply an increased importance of coronary circulation under high-temperature conditions. While the compact myocardium's coronary blood supply is not required for maintaining routine cardiac output, it is thought to increase in importance as the heart's luminal oxygen supply becomes constrained during exercise, warming or hypoxia (Davie and Farrell, 1991; Morgenroth et al., 2021). Indeed, previous studies in fish have demonstrated that acute thermal tolerance is reduced when coronary blood flow is blocked via surgical ligation (Ekström et al., 2017; Ekström et al., 2019; Morgenroth et al., 2021). Similar to the present findings in +0-treatment fish, reduced thermal performance following coronary ligation was characterized by a failure to maintain stroke volume with warming and an early onset of heart rate collapse (Morgenroth et al., 2021). Our findings lend support to the notion that acute thermal tolerance in fish may be limited by oxygenation of the heart itself, whereby function of the oxygen-deprived myocardium declines at high temperatures and compromises oxygen delivery to systemic tissues (Clark et al., 2008; Ekström et al., 2017; Farrell, 2009). Further investigations into the degree of capillarization in the compact myocardium of these fish will be fruitful.

As the climate warms, one of the most noticeable impacts on salmonids will be acute thermal stress in freshwater environments (Crozier et al., 2008). Streams and rivers are less thermally stable than marine environments, making salmon particularly vulnerable to high water temperatures during freshwater stages. In juvenile salmonids, high river temperatures have been hypothesized to create aerobically constrained environments and limit survival during this critical life stage (Crozier and Zabel, 2006; Gradil et al., 2016; Muñoz et al., 2015). Acute cardiorespiratory responses to elevated temperature are therefore ecologically relevant in juvenile salmonids, and will only increase in importance as climate change increases the frequency and severity of extreme heat events (IPCC, 2021). In a previous study by Gradil et al. (2016), juveniles from this hatchery-bred population were found to be narrowly adapted to the summer temperature conditions of their native environment. In that study (Gradil et al., 2016), upper thermal limits for cardiac performance (T_arr) showed close alignment with peak summer temperatures in the LaHave River, suggesting a vulnerability to extreme heat events in this population. In the present study, we demonstrate a significant capacity for plasticity in juveniles from this population, such that exposure to chronically warmer temperatures starting during early development can increase acute thermal tolerance at the level of the heart. Thus, increases in average temperature may help buffer juvenile salmon against the thermal stress imposed by summer heat events in freshwater habitats.

In this study, Atlantic salmon were reared in one of two thermal regimes across the embryonic, alevin and juvenile stages of their life cycle before thermal performance of cardiac function was examined. This study design has been used previously to examine phenotypic plasticity in thermal performance in Chinook salmon (Muñoz et al., 2015). However, as fish from the two temperature treatments were not returned to a common rearing temperature prior to experiments, this study design limits our ability to distinguish between lasting developmental effects and reversible thermal acclimation within the juvenile stage. To distinguish between these two processes would require an experimental design that includes an additional two treatments in which individuals from each temperature are switched to the opposite temperature following development. Interestingly, developmental plasticity and reversible thermal acclimation are not mutually exclusive ‒ as early life environments can have lasting effects not only on an animal's thermal performance but also on their future acclimation capacity (Beaman et al., 2016). In zebrafish, exposure to elevated temperatures during embryonic development was shown to increase thermal acclimation capacity in later life and improve thermal hardiness (Scott and Johnston, 2012). In Drosophila melanogaster, heat tolerance is mediated by an interaction between current acclimation temperature and embryonic temperature (Willot et al., 2021). Thus, the phenotypic changes observed in this study are likely the result of exposure to elevated temperatures across multiple life stages.

A common trade-off associated with developing at higher temperatures is an accelerated developmental rate, but smaller juvenile body size. However, this pattern was not observed in the present study. On the contrary, +4-treatment fish tended to be larger than +0-treatment fish, likely because they began feeding approximately 6 weeks earlier, as a result of their accelerated rate of development. Additionally, temperature differences between treatments were maintained after development, and temperature generally shows a positive association with post-development growth rate in Atlantic salmon (Handeland et al., 2008). Interestingly, egg incubation temperature has also been shown to affect later growth performance in Atlantic salmon (Finstad and Jonsson, 2012), such that fish exposed to elevated temperatures during embryogenesis displayed faster growth rates as juveniles (particularly at warmer temperatures). Given the observed association between temperature and growth, we used a size-matching approach to ensure that there was no significant difference in body size between treatments for the fish used in our analyses. This approach adds a potential confound in that fish included from the +4-treatment were either younger or relatively slower growing on average, while fish included from the +0-treatment were either older or relatively faster growing on average. However, these size and growth differences are unlikely to have had a major influence on our conclusions, as we observed only weak associations between body size and cardiorespiratory measures, and the overall trends were similar when we included the smallest +0-treatment fish and largest +4-treatment fish. The thermal plasticity observed in this study appears to be adaptive, but future investigations into possible performance trade-offs associated with developmental temperature would be fruitful.

In summary, we demonstrated that exposure to higher average water temperatures starting early in development significantly increased the cardiorespiratory thresholds for thermal performance in a hatchery-bred population of Atlantic salmon. Contrary to our prediction, this plasticity in thermal tolerance was not accompanied by plasticity in maximum cardiovascular capacity, as peak f_H,max and relative ventricular volume did not differ between treatments. Rather, fish from the warm treatment displayed higher proportions of compact myocardium in their ventricular walls, a trait commonly associated with increased aerobic demands. As the compact myocardium represents the portion of the ventricle receiving a coronary blood supply, this change in myoarchitecture may improve oxygen supply to the heart itself under high temperature conditions (Anttila et al., 2013b; Ekström et al., 2017). In conclusion, this study contributes to our understanding of how average thermal conditions can affect the response to acute warming and identifies a morphological strategy which may improve the cardiac performance of fish during thermal stress.

The results in this paper are reproduced from the PhD thesis of C.A.M. (Muir, 2022).