An unusually high upper thermal acclimation potential for rainbow trout

We address a crucial question relevant to global climate change and of general interest to ecological and conservation physiologists: does a fish species normally regarded as a cold-loving species have sufficient standing intra-specific variation to form a warm-tolerant population that could perhaps thrive in a warmer world?

Abstract

Thermal acclimation, a compensatory physiological response, is central to species survival especially during the current era of global warming. By providing the most comprehensive assessment to date for the cardiorespiratory phenotype of rainbow trout (Oncorhynchus mykiss) at six acclimation temperatures from 15°C to 25°C, we tested the hypothesis that, compared with other strains of rainbow trout, an Australian H-strain of rainbow trout has been selectively inbred to have an unusually high and broad thermal acclimation potential. Using a field setting at the breeding hatchery in Western Australia, thermal performance curves were generated for a warm-adapted H-strain by measuring growth, feed conversion efficiency, specific dynamic action, whole-animal oxygen uptake (ṀO₂) during normoxia and hypoxia, the critical maximum temperature and the electrocardiographic response to acute warming. Appreciable growth and aerobic capacity were possible up to 23°C. However, growth fell off drastically at 25°C in concert with increases in the time required to digest a meal, its total oxygen cost and its peak ṀO₂. The upper thermal tipping points for appetite and food conversion efficiency corresponded with a decrease in the ability to increase heart rate during warming and an increase in the cost to digest a meal. Also, comparison of upper thermal tipping points provides compelling evidence that limitations to increasing heart rate during acute warming occurred well below the critical thermal maximum (CT_max) and that the faltering ability of the heart to deliver oxygen at different acclimation temperatures is not reliably predicted by CT_max for the H-strain of rainbow trout. We, therefore, reasoned the remarkably high thermal acclimation potential revealed here for the Australian H-strain of rainbow trout reflected the existing genetic variation within the founder Californian population, which was then subjected to selective inbreeding in association with severe heat challenges. This is an encouraging discovery for those with conservation concerns for rainbow trout and other fish species. Indeed, those trying to predict the impact of global warming should more fully consider the possibility that the standing intra-specific genetic variation within a fish species could provide a high thermal acclimation potential, similar to that shown here for rainbow trout.

Introduction

Fishes, being the most specious vertebrate taxa, have a diverse physiology that has allowed them to exploit almost every aquatic habit, including those with water temperatures ranging from −1.9°C to over 40.0°C (Davenport and Sayer, 1993). All the same, certain species are being displaced from their historical biogeographic ranges during the current era of global warming because thermal performance ranges of fishes are species specific and more biogeographic shifts are predicted for a warmer future (Cheung et al., 2006; Pörtner and Farrell, 2008; Pereira et al., 2010). Specifically, the geographic ranges of fishes occupying narrow and warm thermal niches (i.e. warm-water thermal specialists) have contracted or shifted as a result of climate warming and other similar species are at risk of displacement as our climate continues to warm. Thus, global warming is clearly a conservation concern. However, the models predicting these biogeographic shifts rarely consider the potential for a fish species to acclimate or adapt to new thermal conditions—possibilities that are becoming evident among fishes as evidenced by intra-specific local adaptations of thermal tolerance and thermal performance (McKenzie et al., 2021).

Good examples of intra-specific local adaptations of thermal tolerance and thermal performance include killifish (Fundulidae spp.) populations along the east coast of North America (Fangue et al., 2006), Lake Magadi tilapia (Tilapia grahami) in Africa (Reite et al., 1974), Atlantic cod (Gadus morhua) populations in the North Atlantic Ocean (Grabowski et al., 2009), Chinese minnow (Rhynchocypris oxycephalus) populations in Northeast coast of China (Yu et al., 2018), sockeye salmon (Oncorhynchus nerka) populations in the Pacific Northwest (Eliason et al., 2011), perch (Perca fluviatilis) populations in Sweden (Sandblom et al., 2016) and redband trout (Oncorhynchus mykiss gairdneri) populations in Idaho (Rodnick et al., 2004; Chen et al., 2018). Nevertheless, intra-specific variability in thermal tolerance is not evident in all fish species. For example, a northern and a southern population of Atlantic salmon (Salmo salar) did not show local adaptation of their heat tolerance (Anttila et al., 2014), but instead the two populations possessed a considerable potential for thermal acclimation (i.e. phenotypic thermal plasticity), the ability to maintain or improve performance after prolonged exposure to a new temperature. Consequently, population-based modelling will need reliable information on local intra-specific adaptation and the acclimation potential of their upper thermal ceiling (Sandblom et al., 2016). Indeed, Lefevre et al. (2017) have already suggested that reliable predictions about fish populations under climate change will need to consider the underlying physiological mechanisms, some of which we sought here for temperature acclimation.

In the present study, we focussed on the thermal acclimation potential of rainbow trout (O. mykiss) in part because of its successful introductions from the endemic habitat in the Pacific Northwest onto every continent, except Antarctica. Yet, rainbow trout are considered a ‘cold-water’ species, one that prefers temperatures <20°C (MacCrimmon, 1971). Indeed, historically, rainbow trout spread southwards in the Pacific Northwest as a post-glacial colonizer ~10 000 years ago. Furthermore, the US Environmental Protection Agency (USEPA) recommends a 7-day average daily mean maximum of 18°C for the management of their watersheds in the US Pacific Northwest (US Environmental Protection Agency, 2003). Nevertheless, population differences exist for the thermal performance of rainbow trout (Myrick and Cech, 2000, 2004; Rodnick et al., 2004; Verhille et al., 2016) and warm acclimation can increase their critical thermal maximum (CT_max; Becker and Wolford, 1980; Myrick and Cech, 2000).

Of interest to us was the H-strain of rainbow trout, which has been genetically isolated and inbred for stocking natural lakes for over 20 generations at the Pemberton Freshwater Research Centre (PFRC), a hatchery in arid Western Australia. Over this time, the genetic diversity of the H-strain became reduced compared with the founder population that originated from Sonoma Creek in the San Francisco Bay area, USA, likely as a result of inbreeding and periodic thermal selection events over the decades (i.e. natural summer temperature extremes >27°C in a reduced water supply from the Lefroy Brook that caused large mortality events among the fish stock: 66% in 1961 and >98% in 1970) (Ward et al., 2003; Molony et al., 2004). When acclimated to only 15°C, the acute thermal tolerance of the H-strain is already impressive compared with Pacific Northwest populations. For example, their critical thermal maximum (CT_max) is an impressive 29°C and when acutely warmed to 25°C, they maintain 43% of their maximum aerobic capacity (Chen et al., 2015). However, no one has tested the hypothesis that warm acclimation of this H-strain of rainbow trout can further improve their upper thermal performance.

To test this hypothesis and to seek mechanistic insights into thermal performance, we comprehensively assessed the cardiorespiratory phenotype of the H-strain of rainbow trout at the PFRC field setting. By using 2°C increments for the six acclimation temperatures (from 15 to 25°C), we could more finely resolve thermal optima compared with 3–5°C temperature increments (e.g. Christensen et al., 2020) that are more typically used (Farrell, 2016) and focus on upper tipping points of the thermal performance curves. Specifically, we measured growth, whole-animal aerobic capacity, the energetic cost to digest a meal and their CT_max, as well as hypoxia tolerance (because hypoxia tolerance and CT_max have been previously linked among other strains of rainbow trout in the Pacific Northwest; Zhang et al., 2018). At each acclimation temperature, we also measured the response of maximum heart rate (f_Hmax) to acute warming as an increase in heart rate (f_H) is the primary cardiac response to acute warming and warm acclimation can increase the peak f_Hmax reached with acute warming (Eliason and Anttila, 2017; Farrell, 2016; Vornanen, 2017).

Methods

The broodstock of H-strain rainbow trout has been maintained for almost 50 years at the PFRC (Molony, 2001), which is at a remote location south of Perth, Western Australia. We used fish from the general hatchery population (approximately 100 parents) that were bred in spring 2018 and then raised for 9 months in aerated ponds that received water from the Big Brook dam on the Lefroy Brook, Western Australia. During rearing, water temperature fluctuated daily (solar effect) and seasonally according to ambient temperature fluctuations. Experimental procedures and protocols were approved by the University of British Columbia Animal Care Committee (A18-0340).

For the present experiment 3000 rainbow trout were transferred into indoor, 250-l fibreglass holding tanks (50–80 fish per tank), where they were held before any experimentation for a minimum of 4 weeks at one of the six acclimation temperatures (15, 17, 19, 21, 23 or 25°C) with a 12:12 h light/dark cycle. The water temperatures were kept constant and did not incorporate variation that would occur in the natural water supply from the Big Brook Dam. Each tank contained a biological filter to remove particulates and an immersion heater to raise the water temperature above that of the ambient receiving water, as needed.

Experimentation started in February 2019 and lasted through to June 2019. A growth trial was performed in triplicate at each acclimation temperature (50 fish in each tank; 18 tanks in total). A further 80 fish were held at each acclimation temperature in additional, separate tanks for exclusive use with the physiological tests. The periodic removal of fish for physiological measurements, therefore, did not disturb the growth trial. Each fish was used for one experiment at each acclimation temperature except that six fish were reused after the aerobic scope tests to examine the response of f_Hmax to acute warming. Mortality was monitored daily in each tank. Fish mass is reported in Supplemental Table S1.

Growth trial

Fish were fed to satiation with commercial freshwater trout feed (Skretting, Cambridge, Tasmania, Australia) three times daily, except on the day preceding and the day of weighing for growth measurements. Individual fish mass and length were measured at weeks 0, 2 and 4 of the growth trials. The amount of food added to each tank was recorded daily, which allowed a calculation of food conversion efficiency for the total mass of fish in the tank (i.e. food conversion efficiency was measured in triplicate at the tank level for each acclimation temperature).

Respirometry trials to determine the respiratory phenotype

Established protocols and analytical procedures (Chabot et al., 2016b; Zhang et al., 2016; Zhang et al., 2017) were used to monitor oxygen uptake (ṀO₂) of individual fish at their acclimation temperature over a period of 2 days. Each trial monitored eight fish simultaneously and separately in eight respirometers (760 ml), which were all submerged in a 200-l water reservoir supplied with a recirculating chiller and immersion heaters to maintain water at the required acclimation temperature. A circulation loop for each respirometer continuously mixed the water inside respirometry chamber and fibre optic oxygen probe (Firesting O₂, PyroScience GmbH, Aachen, Germany) contained in this loop recorded the water oxygen level (as % of air saturation; % air sat.) inside the respirometer every 1 s. The optodes were calibrated using aerated water (100% air saturation) and nitrogen bubbled water containing sodium sulphite (0% saturation). Water oxygen levels were converted to water PO₂ (mm Hg) and dissolved oxygen concentration (mg O₂ l⁻¹) for data analysis, as needed.

All respirometers had computer-controlled flush pumps (Compact Pump 1000, Eheim, Germany) and relays (AquaResp, University of Copenhagen, Denmark) that controlled the intermittent replenishment of the respirometer with fresh water between each ṀO₂ measurement period. Each 10-min cycle consisted of a 60-s flush period, a 60-s stabilization period and a 480-s period when ṀO₂ was measured. During this ṀO₂ measurement period, ṀO₂ was calculated as the rate of the decrease in dissolved oxygen over time (Chabot et al., 2016a). A UV-sterilizer light was placed in the water reservoir to reduce microbial growth. All ṀO₂ measurements were corrected for the background ṀO₂, which was measured in each respirometer without a fish (for 30 min) both before and immediately after every trial. The entire respirometry system was thoroughly disinfected with 10% bleach for 1 h and thoroughly cleaned between trials.

The respiratory phenotype of individual rainbow trout was measured for 16 fish per acclimation temperature, i.e. two trials with 8 fish for each acclimation temperature. Each trial was preceded by a 48-h fast and a measurement of body mass (Table S1) before securing a fish inside a respirometer typically between 12.00 h and 14.00 h. Each fish was individually agitated inside the respirometer (to exhaustion but no longer than 10 min) to elicit a maximum rate of oxygen uptake (ṀO_2max), which was assigned to the highest ṀO₂ value observed while the fish recovered (Zhang et al., 2020). Each fish was then left undisturbed for 48 h, during which time ṀO₂ was continuously monitored (~150 ṀO₂ values were generated per fish). From these ṀO₂ measurements, standard metabolic rate (SMR) was estimated by applying a 20th quantile algorithm (q0.2; Chabot et al., 2016b), while absolute aerobic scope (AAS) was calculated as the difference between ṀO_2max and SMR (Fry, 1971; Claireaux et al., 2005) and factorial aerobic scope (FAS) was calculated as the quotient of ṀO_2max to SMR (Clark et al., 2005). The respirometry trial ended with a hypoxia challenge test that lasted 2–3 h, during which nitrogen gas was bubbled through a ceramic micro-bubble diffuser into the water reservoir such that the % air saturation progressively decreased by 0.3% min⁻¹ until the % air saturation was so low that the fish lost its dorsal-ventral equilibrium. This ambient level of % air saturation was recorded and converted to an incipient lethal oxygen concentration (ILOC, mg O₂ l⁻¹) or incipient lethal oxygen partial pressure (ILOP, mmHg). Fish were immediately revived in aerated water. The % water saturation when SMR could not be maintained (either C_crit, mg O₂ l⁻¹ or P_crit, mmHg) was interpolated by graphing ṀO₂ as a function of % water saturation during progressive hypoxia (Schurmann and Steffensen, 1997; Claireaux and Chabot, 2016).

Individual feeding trials to determine specific dynamic action

Fish were individually fed to satiation in the acclimation holding tank, weighed and transferred immediately into a respirometer (as described above) at their acclimation temperature with minimal air exposure (maximum 20 s). Thus, continuous ṀO₂ measurements started within 20 min after feeding ceased and continued for 48–72 h depending on when ṀO₂ returned to SMR (q0.2 of ṀO₂ recordings) for 10–15 measurements per acclimation temperature (see Table S1 for body mass). The SMR of fish in the feeding trial turned out to be similar (t₁₀ = 0.62, P = 0.55) to that of fish in the respirometry trials described above. Fish were euthanized at the end of the specific dynamic action (SDA) trial (80 mg l⁻¹ MS222 buffered with 160 mg l⁻¹ NaHCO₃) before being weighed. Analysis of SDA between the start of feeding and the return to SMR (SDA _duration, h) started with an inspection of individual ṀO₂ traces to ensure adherence to established analysis criteria (Chabot et al., 2016a). Short periods of activity were smoothed with a moving average algorithm, and a non-parametric quantile regression algorithm was used to describe the SDA curve. The total amount of oxygen consumed during SDA (SDA magnitude, mg O₂ kg⁻¹) was determined by an integral of the area below the fitted SDA curve above the SMR and the peak ṀO₂ on the fitted curve was assigned as SDA_peak (mg O₂ h⁻¹ kg⁻¹).

Response to acute warming: CT _max  and maximum heart rate (f_Hmax) measurements

CT_max was measured using an established methodology (Beitinger et al., 2000) by rapidly transferring 10 fish after a 24-h fast into a single 200-l aerated tank at 12°C. They habituated to the tank for 1 h before temperature was incrementally increased at a rate of 0.3°C min⁻¹ to 22°C using two heating rods (100 W Titanium Heater, Aquatop, Brea, CA, USA). Subsequently the heating rate was reduced to 0.1°C min⁻¹, until a temperature was reached that caused individual fish to lose equilibrium (their CT_max), whereupon the individual was immediately removed and revived in a recovery bath at their acclimation temperature.

Similarly, an established methodology (Casselman et al., 2012) was used to measure f_Hmax during an acute, incremental warming protocol. From each respirometry trial, 6 of the 10 fish were placed in a recovery tank for 4 h, using a total of 12 fish per acclimation temperature. Briefly, a fish was anaesthetized (80 mg l⁻¹ MS222 buffered with 160 mg l⁻¹ NaHCO₃) and transferred to a sling immersed in a 12°C water bath where its gills were continuously irrigated with recirculating water containing a maintenance dose of anaesthetic (65 mg l⁻¹ MS222 buffered with 130 mg l⁻¹ NaHCO₃). An electrocardiograph (ECG) was recorded on-line using two stainless steel electrodes, one inserted below the muscle layer near the heart and the second below the pelvic fin as a reference. The ECG signal was amplified and digitized (Animal Bio Amp and a Powerlab 8/30, ADInstruments Inc., Bella Vista, NSW, Australia), and f_H was extracted from the ECG signal using Labchart software (ADInstruments). To obtain f_Hmax, vagal tone to the heart was blocked with an intra-peritoneal injection of atropine sulphate (1.8 mg kg⁻¹) and cardiac adrenergic β-receptors were maximally stimulated with an injection of isoproterenol (6 μg μg kg⁻¹) (Sigma Chemicals, Perth, Western Australia). The incremental heating protocol began after a 30-min period at 12°C, which provided time for the pharmaceutical agents to have their effects and ample time for the fish to reach thermal equilibrium with the ambient water temperature. By incrementally heating the water bath by 1°C every 6 min, we ensured that f_Hmax had stabilized at each temperature increment. Warming continued until a temperature was reached when f_Hmax peaked (T_peak). Further incremental warming was terminated at the temperature that first generated a cardiac arrhythmia (T_arr), at which point the fish was euthanizing with a lethal dose of anaesthetic. The temperature quotient (Q₁₀) for f_Hmax during acute warming (Casselman et al., 2012) was calculated as the ratio of f_Hmax values for 15°C and 25°C at each acclimation temperature.

Statistical analysis

Data analysis was conducted in R (v.3.6.2, R Core Team, Austria) and Prism (v.8, GraphPad Software, USA). All values are presented as means ± standard error of the mean (sem) with statistical significances assigned when α = 0.05. For the growth study, a mixed effects analysis was performed on fish mass and length (tank ID and time as fixed effects, acclimation as a random effect, mass and length were fit with a Gompertz growth non-linear regression with a least square fit) while feeding was examined with a two-way analysis of variance (ANOVA) followed by a Tukey’s post hoc-test (Table S2). Differences in acclimation temperatures were examined for each variable using a one-way ANOVA followed by a Tukey’s post-hoc test (Table S3). To account for allometric scaling, oxygen consumption data (SMR, ṀO_2max, AAS) were adjusted to the average body mass (16.23 g) by summing the residuals for the linear regression between mass and oxygen consumption with the predicted value at the average body mass (16.23 g). This method was applied to oxygen consumption data for SDA values (SMR, peak and net peak; average body mass 16.17 g). Regression analysis was performed between the performance parameter and acclimation temperature to determine the line of best fit (quadratic polynomial for SMR and ILOS, Gaussian function for ṀO_2max, AAS and FAS). Differences in CT_max and T_arr were tested using a two-way ANOVA followed by a Tukey’s post-hoc test (Table S2), while T_peak, peak f_Hmax and f_Hmax at 15°C were tested using a one-way ANOVA followed by a Tukey’s post-hoc test (Table S3). A Pearson’s correlation coefficient was determined for the relationship between individual f_Hmax and individual ṀO_2max values at each acclimation temperature.

Results

During the 4-week growth trial, fish gained body mass and length at all six acclimation temperatures, with both metrics increasing significantly by week 2 (F_{1.3, 14.9} = 3690, P < 0.0001 and F_{1.2, 14.2} = 4249, P < 0.0001, respectively; Fig. 1A). By week 4, however, significant differences in growth were evident among the acclimation temperatures (F_5,11 = 144.6, P < 0.0001). While body mass increased by 3.1- to 4.4-fold after 4 weeks at acclimation temperatures from 15°C to 23°C, the increase in body mass was only 1.7-fold at 25°C (Fig. 1A, B).

Figure 1.

Mass, length, average daily feed, food conversion ratio (FCR), specific dynamic action (SDA) duration and peak SDA, standard metabolic rate (SMR) and net peak SDA values for PFRC H-strain rainbow trout (O. mykiss). (A) Body mass (g), (B) fish length (mm) and (C) average daily feed ration (total per tank and averaged to number of rainbow trout in tank and per day) for each temperature acclimation group held in triplicate tanks (n = 3) and measured at weeks 0, 2 and 4. (D) FCR of rainbow trout as function of acclimation temperature (n = 3). (E) Peak SDA, measured as the highest oxygen uptake rate (ṀO2) value following feeding, SMR, measured as ṀO2 and net peak SDA measured as the difference between peak SDA and SMR, all presented as a function of acclimation temperature (n = 12–16). (F) SDA duration measured as a function of acclimation temperature (n = 12–16). Data points not sharing letters indicate significant differences between acclimation groups for that week. Asterisk indicates significant differences of all acclimation groups between two consecutive sampling time points. All values are presented as means ± sem. Mass, length and feed were assessed using a linear mixed effects model while FCR, SDA duration, peak SDA, SMR and net peak SDA were assessed using a one-way ANOVA with Tukey’s post-hoc tests.

Appetite increased as fish grew in size as evidenced by an increase in total daily feed intake over time at acclimation temperatures from 15°C to 23°C (Fig. 1C). However, appetite was significantly suppressed at 25°C for the entire growth trial. The food conversion ratio (FCR; the inverse of the efficiency of feed utilization) was highest (by 30–50%) for the 23°C-acclimated fish compared with the other acclimation temperatures (Fig. 1D). Thus, the significantly lower appetite of the 23°C-acclimated fish (Fig. 1C) was somewhat offset in terms of growth by a low FCR, which was similar for acclimation temperatures between 17°C and 23°C (Fig. 1D). Consequently, peak growth performance occurred over the acclimation temperature range of 17–23°C.

The feeding energetics trial measured the oxygen cost (ṀO₂) of digesting a meal (the SDA), which reached a peak value (SDA_peak) and then decline towards the post-prandial SMR that was also determined. Both SDA_peak and post-prandial SMR increased significantly as a function of acclimation temperature (F_5,71 ≥ 59.8, P < 0.0001), both peaking at an acclimation temperature of 25°C (F_5,71 = 13.08; Fig. 1E). Furthermore, net SDA_peak during digestion (the difference between SDA_peak and post-prandial SMR) increased significantly with acclimation temperature (P ≤ 0.042; Fig. 1E), more than doubled, from 87 mg O₂ kg⁻¹ h⁻¹ at 15°C to 183 mg O₂ kg⁻¹ h⁻¹ at 25°C. Thus, the allocation of oxygen specifically for digestion increased significantly with temperature acclimation up to 25°C. SDA duration also varied significantly with acclimation temperature, but decreased significantly with acclimation temperature (F_5,71 = 7.9,P < 0.0001; Fig. 1F), perhaps in part because appetite decreased at the warmest acclimation temperatures (F_5,36 = 168.1, P < 0.0001; Fig. 1C). As a result, the total oxygen cost of digesting a meal (SDA magnitude) was lowest at 21°C and highest at 25°C (F_5,71 = 5.4, P < 0.0001; Fig. 1C) despite the much lower appetite of the 25°C-acclimated fish (Fig. 1).

The respirometry trial, which used unfed fish, provided another independent measurement of SMR (Fig. 2A). The two independent SMR measurements were statistically indistinguishable (t₁₀ = 0.62, P = 0.55). Again, SMR increased significantly with acclimation temperature (F_5,56 = 26.2, P < 0.0001). Maximum oxygen uptake (ṀO_2max) peaked at acclimation temperatures between 17°C and 23°C, being statistically indistinguishable at these four acclimation temperatures (F_{5, 77} = 4.91, P ≥ 0.19; Fig. 2A), but numerically highest at 19°C (771 ± 16 mg O₂ kg⁻¹ h⁻¹). Consequently, the capacity to increase oxygen uptake (AAS = ṀO_2max − SMR) similarly peaked at acclimation temperatures between 17°C and 21°C (F_5,77 = 2.47, P ≥ 0.19; Fig. 2A) and, again, was numerically highest at 19°C (639 ± 18 mg O₂ kg⁻¹ h⁻¹). Both ṀO_2max and AAS decreased significantly for the 25°C-acclimation temperature when compared with the 19°C-acclimation temperature (Fig. 2A; ṀO_2max: F_5,77 = 4.9, P = 0.0012; AAS: F_5,77 = 1.6, P < 0.0001; Table S3). FAS (the ratio of ṀO_2max and SMR) was greatest at acclimation temperatures between 15°C and 19°C, but decreased significantly at acclimation temperatures ≥21°C (Fig. 2B). Thus, while ṀO₂ could be increased beyond SMR by about 6-fold at acclimation temperatures between 15°C and 19°C, the 25°C-acclimated fish could increase in ṀO₂ above SMR by less than 4-fold, still sizeable albeit reduced aerobic scope. Notably, when SDA_peak was expressed as a ratio of SMR, it was never greater than twice SMR and virtually independent of acclimation temperature (Fig. 2C), suggesting that peak ṀO₂ may be mechanistically linked with SMR in a manner largely independent of the acclimation temperature.

Figure 2.

Standard metabolic rate (SMR), maximum oxygen uptake (ṀO_2max), absolute aerobic scope (AAS) and factorial aerobic scope (FAS) of PFRC H-strain rainbow trout (O. mykiss) as a function of acclimation temperature. (A) SMR, ṀO_2max and AAS (difference between ṀO_2max and SMR) as a function of acclimation temperature (n = 12–16). SMR (blue square) was fitted with a polynomial quadratic function and one-way ANOVA (Y = 135.0 + 10.5^*X + 1.35^*X², R² = 0.67, F = 31.14, P < 0.0001), ṀO_2max (red circle) and AAS (green triangle) were fit with a Gaussian distribution curve and one-way ANOVA (Y = 760.2^*exp(−0.5^*((X-19.6)/9.89)2), R² = 0.22, F = 4.91, P = 0.0006; Y = 634.7^*exp(−0.5^*((X-18.6)/7.37)2), R² = 0.42, F = 15.40, P < 0.0001). ṀO_2max and SMR values were corrected to mean body mass of acclimation group, presented per unit kg. (B) Factorial of ṀO_2max/SMR (FAS) and peak SDA/SMR (n = 12–16). FAS (green circle) was fitted with a Gaussian distribution curve and one-way ANOVA (Y = 1.9^*exp(−0.5^*((X-2.0)/1.7)2), R² = 0.73, F = 43.41, P < 0.0001). (C) Critical oxygen level to maintain SMR (P_crit or C_crit) and oxygen level when fish is loss of whole animal equilibrium (ILOP or ILOC), as a function of acclimation temperature. P_crit (or C_crit) and ILOP (or ILOC) are presented as both water PO₂ (mm Hg; left y-axis) and dissolved oxygen concentration (mg O₂ l^-1; right y-axis). Curve was fit with polynomial quadratic function and one-way ANOVA (Pcrit: Y = 26.3 + 1.60^*X + 0.19^*X²), R² = 0.57, F = 22.19, P < 0.0001; Ccrit: Y = 1.54 + 0.062^*X + 0.0095^*X²), R² = 0.40, F = 11.58, P < 0.0001; ILOP: Y = 23.3 + 0.80^*X + 0.061^*X²), R² = 0.28, F = 6.41, P < 0.0001; ILOC: Y = 1.37+ 0.021^*X + 0.0031^*X²), R² = 0.57, F = 1.91, P = 0.10, (n = 12–16). Data points not sharing letters indicate significant differences between acclimation groups for that variable. All values are presented as mean ± sem and were assessed using a one-way ANOVA with Tukey’s post-hoc (α = 0.05).

Warm acclimation significantly reduced hypoxia tolerance of the H-strain, but the extent of the decrease in hypoxia tolerance depended on the endpoint used and its dimension (Fig. 2C). One hypoxia tolerance endpoint was the threshold water oxygen level to maintain SMR (either P_crit, mm Hg or C_crit, mg O₂ l⁻¹; Table S1). Both P_crit and C_crit increased significantly with acclimation temperature (P_crit, mm Hg: F_5,69 = 22.19, P < 0.0001; C_crit, mg O₂ l⁻¹: F_5,69 = 11.58, P < 0.0001). The other hypoxia tolerance endpoint was the threshold water oxygen level for the loss of the righting reflex (either ILOP, mm Hg, or ILOC, mg O₂ l⁻¹). While ILOP increased significantly with acclimation temperature (F_5,73 = 6.41, P < 0.0001), it did so to a lesser degree than P_crit. Also, ILOC was independent of acclimation temperature (F_5,73 = 1.91, P = 0.10; Fig. 2C). Since ILOP considers the oxygen partial pressure gradient for diffusion at the gills, while ILOC considers the amount of oxygen contained in the water moved over the gills, our results suggest that at 25°C the need for a sufficiently high oxygen partial pressure gradient to oxygenate blood at the gills is perhaps mechanistically more important than the temperature-dependent decrease in water oxygen at this water temperature.

Independent of the acclimation temperature, acute warming always increased f_Hmax until it reached a peak value (Q₁₀ values ranged from 1.70, 1.69, 1.81, 1.92, 1.93 and 1.79 with ascending acclimation temperatures; Fig. 3A). As expected, warm acclimation incrementally reset f_Hmax to a lower value when a comparison was made at the common test temperature of 15°C (F_5,52 = 23.11, P < 0.0001). For example, at the 15°C test temperature, f_Hmax was 22 bpm lower for 25°C-acclimated fish compared with 15°C-acclimated fish (58 bpm vs 80 bpm; F_{1.83, 32.6} = 5.96, P = 0.0017; Fig. 3A; Table S3). Warm acclimation also significantly increased the temperature for peak f_Hmax (T_peak; F_5,56 = 5.34, P = 0.0004). For example, T_peak was 2.7°C higher for 23°C-acclimated fish compared with 15°C-acclimated fish (27.6 ± 0.4°C vs 24.9 ± 0.3°C). Thus, warm acclimation, by resetting f_Hmax and increasing T_peak, preserved some scope to increase f_H in response to acute warming. Even so, peak f_Hmax was independent of acclimation temperature between 15°C and 23°C (ranging from 166 ± 4 to 179 ± 4 bpm; Table S1), and peak f_Hmax actually decreased significantly for the 25°C-acclimated fish (F_5,56 = 5.3, P ≤ 0.01).

Figure 3.

Maximum heart rate (f_Hmax) and critical thermal maximum (CT_max) in response to acute warming of PFRC H-strain rainbow trout (O. mykiss) as a function of acclimation temperature. (A) Mean f_Hmax for each acclimation temperature group during acute warming (n = 12). Acclimation temperatures were 15°C (), 17°C (), 19°C (), 21°C (), 23°C () and 25°C (). Data points are connected by a solid line provided all fish in an acclimation group retained a rhythmic heartbeat at that temperature. If not a broken line connects the average f_Hmax for the fish the still retained a rhythmic heartbeat, with the percentage count remaining indicated in the inset. All individual fish had developed an arrhythmic heartbeat by 31°C and by 28°C for those at the two coldest acclimation temperatures. (B) Temperature at which each acclimation group reached its CT_max (n = 10), cardiac arrhythmia temperature (T_arr) and maximum f_Hmax (T_peak). Amber area labels the difference between whole-organism and tissue thermal tolerance and the differences between CT_max and T_arr at 15°C, 23°C and 25°C acclimation temperature are denoted. Two-way ANOVA performed indicates T_arr is significantly different from CT_max (F = 213.2, P < 0.0001). (C) Peak f_Hmax reached at each acclimation temperature (n = 12). All values are presented as means ± sem, except right y-axis in (A) presented as count in %. (sem may be hidden by the symbol). Dissimilar letters represent statistically significant differences among mean values for that variable.

A cardiac arrhythmia was triggered whenever a fish was acutely warmed beyond T_peak (i.e. at T_arr; Fig. 3A inset). Like T_peak, warm acclimation significantly increased T_arr by 3.4°C when 15°C-acclimated fish were compared with 23°C-acclimated fish (25.7 ± 0.5°C vs 29.1 ± 0.5°C; F_5,56 = 6.29, P = 0.0002), but this benefit was lost at a 25°C-acclimation temperature because T_arr (27.2 ± 0.6°C) decreased significantly (F_5,56 = 6.29, P = 0.0001; Fig. 3B).

Warm acclimation significantly increased the temperature at which acute warming of fish in normoxic water lost their righting reflex, increasing CT_max from 29.0 ± 0.4°C for 15°C-acclimated fish to 31.1°C for 23°C-acclimated fish (F_5,53 = 95.5, P < 0.0001). In contrast to the other physiological impairments seen at the 25°C-acclimation temperature, CT_max was unchanged (31.2°C; Fig. 3B).

Discussion

Our discovery of an unexpectedly high thermal acclimation potential for the H-strain of rainbow trout found in Western Australia, as compared with other studied strains of rainbow trout (Fig. 4), has an important ecological implication. It suggests that sufficient standing genetic variation could exist naturally within existing rainbow trout populations to provide sufficient intra-specific thermal acclimation potential to deal with a certain degree of global warming. Furthermore, by examining a broad range of thermal performance curves for a wide range of physiological function, we provided insights into the physiological mechanisms that underlie transition temperatures, thermal optima and tipping points (when a performance metric no longer increased or decreased). Importantly, and despite performing these physiological studies at a geographically remote field site, the accuracy of interpolating upper thermal tipping points was improved by using 2°C increments in acclimation temperature rather than more typical increments of 3–5°C in previous studies and focusing on temperatures at 15°C and beyond. Indeed, our results, which were obtained at two levels of biological organization, consistently showed upper thermal tipping points at acclimation temperatures of 21°C and 23°C.

Figure 4.

Maximum oxygen uptake (ṀO_2max), absolute aerobic scope (AAS) and factorial aerobic scope (FAS) as a function of acclimation temperature. The data are synthesized from literature for geographically distant rainbow trout (O. mykiss) strains. ṀO_2max; (A), AAS (B) and FAS (C) over a range of testing temperatures in various studies on rainbow trout (O. mykiss) strains. The data are presented in either Gaussian or linear regressions models as mean ± 95% C.I. Figure legends state the acclimation temperature, body mass, location, strains, original studies and whether standard metabolic rate (SMR) or resting metabolic rate (RMR) are measured. The studies had different mass, testing protocols and analytical techniques, as well as acclimation temperatures (indicated), but not all of the differences in the curves can be attributed to these. A red vertical dash line marks 18°C as the 7-day average of the daily maxima criterion for the management of rainbow trout habitat in the Pacific Northwest (U.S. Environmental Protection Agency, 2003). Rainbow trout (O. mykiss) and redband trout (O. mykiss gairdneri) are in blue and green, respectively, where the aerobic capacity thermal performance curves are measured by acute exposure. The performance curves measured by the previous acclimation study are in orange, and the data of the present acclimation study are in red. Patterns of lines differentiate the strains. The range of the curve is constrained by testing temperature ranges. This is modified from McKenzie et al. (2021) by adding data from Dickson and Kramer (1971; a 95% CI for FAS was unavailable) and excluding data from Poletto et al. (2017).

Intra-specific differences in thermal acclimation potential for growth performance are already known for rainbow trout. For example, Myrick and Cech (2000) compared two Californian strains of rainbow trout, the Mount Shasta and Eagle Lake strains. Peak growth rate occurred between 19°C and 22°C for the Mount Shasta strain, while the Eagle Lake strain grew best at a cooler 19°C-acclimation temperature. Nevertheless, the Mount Shasta strain slowed growth at a 22°C-acclimation temperature and ceased growing and lost weight at the 25°C-acclimation temperature. The Eagle Lake strain resembled other rainbow trout strains that have peak growth performance at acclimation temperatures of 15.0–18.6°C (Hokanson et al., 1977), the data used by the USEPA for rainbow trout management in the Pacific Northwest (U.S. Environmental Protection Agency, 2003). The H-strain of rainbow trout has a higher thermal growth performance than other strains thus far examined by still growing significantly at an acclimation temperature of 25°C (a 1.7-fold increase over 4 weeks), even though growth performance started to fall off at an acclimation temperature of ≥21°C.

Suppression of appetite at supra-optimal temperatures is a well-established phenomenon in fishes (Jobling, 1981; Myrick and Cech, 2000). The decline in growth potential when the H-strain was acclimated 25°C was certainly associated with a suppression of appetite (the lowest feed intake) and the lowest feed conversion efficiency (the highest FCR, which was optimal between 17°C and 23°C). An aerobic scope protection hypothesis has been advanced to explain appetite suppression at high temperature (Jutfelt et al., 2021), one that proposes that meal size is reduced to limit postprandial SDA and ‘protect’ residual aerobic scope. But how fishes might achieve and regulate such a protection is unknown. Our data SDA determinations provide new insights. As expected, ṀO₂ peaked during digestion (Fu et al., 2007; Eliason et al., 2008; Pang et al., 2011; Eliason and Farrell, 2014) and SDA_peak increased significantly as a function of acclimation temperature, even at 25°C. SMR also increased with acclimation temperature, as expected (e.g. Sandblom et al., 2014; Claireaux et al., 2000), which would add to the SDA during digestion. To remove this effect, we calculated net SDA_peak, which still increased with acclimation temperature up to 25°C. Thus, neither SDA_peak nor net SDA_peak appeared to have a ceiling at the supra-optimal acclimation temperature of 25°C, as might be expected for the aerobic scope protection hypothesis. Instead, we found considerable appetite suppression in 25°C-acclimated rainbow trout, an ample aerobic scope to buffer SDA_peak (which was about half of FAS), but costly food processing of a reduced meal size. Moreover, appetite suppression was triggered at acclimation temperatures cooler than 25°C when FAS was higher and SDA_peak was lower. SDA duration was shortest at acclimation temperatures between 19°C and 25°C, and SDA magnitude was lowest at 21°C. Consequently, we compared the ratio of SDA_peak and SMR with the ratio of ṀO_2max and SMR (i.e. FAS) for each acclimation temperature (this comparison was possible because SMR for the feeding and aerobic scope trials were indistinguishable). SDA_peak/SMR was almost independent of acclimation temperature and, importantly, was never more than twice SMR because the increases in SDA_peak and SMR with acclimation temperature closely matched each other. How such a balance of metabolic rates might be regulated is unclear from the present work, but a similar discovery was made in lionfish (Pterois spp.) (Steell et al., 2019). Thus, while our findings are consistent with the aerobic scope protection hypothesis in that appetite falls off seeming in parallel with FAS, the entire FAS was never fully exploited by SDA_peak and a variable aerobic scope buffer exists for different levels of appetite suppression as a function of acclimation temperature. How these buffer capacities might be regulated and what determines their magnitude will require further study, as will our discovery that digestion of a small meal became less efficient at a supra-optimal acclimation temperature.

Work with low-altitude or high-altitude populations of killifish (genus Aphysosemion) in equatorial West Africa previously showed SDA performance declined at acclimation temperatures outside of the ecologically relevant temperature (25°C) for this genus (McKenzie et al., 2013). Specifically, the low-altitude population processed a meal much faster than a high-altitude population when tested at the supra-optimal acclimation temperature of 28°C. At a sub-optimal acclimation temperature of 19°C, the converse was true. However, the reduced digestive performance in killifish was not associated with any change in SDA_peak (McKenzie et al., 2013), unlike the present study where acclimation temperature affected the speed with which the H-strain of rainbow trout processed a meal, with the nadir in SDA duration occurring at acclimation temperatures of 21°C and 23°C.

The unusually high upper thermal acclimation potential of H-strain rainbow trout is highlighted by comparing our ṀO_2max and AAS curves with literature values for this genus (Fig. 4). (Note: Our AAS for the H-strain rainbow trout at an acclimation temperature of 15°C compares favourably an early measurement at this acclimation temperature: 544 mg O₂ kg⁻¹ h⁻¹ for Chen et al., 2015 and 558 mg O₂ kg⁻¹ h⁻¹ for the present study.) Our results reflect that the H-strain of rainbow trout has the potential to acclimate to temperatures more comparable with those for the redband rainbow trout, a subspecies of rainbow trout that successfully adapted to desert habitats after becoming geographically isolated from the Columbia River in Idaho and Oregon, USA. Redband rainbow trout routinely experience highly variable and hot summer temperatures, e.g. 19–29°C for 12-mile Creek (Rodnick et al., 2004). Furthermore, the H-strain of rainbow trout has a higher aerobic performance and thermal optimum for ṀO_2max compared with two southern Utah strains of rainbow trout (Fig. 4; Hull-Erickson strain and DeSmet strain; Dickson and Kramer, 1971). Furthermore, the aerobic benefit of the H-strain’s ability to warm acclimate to 23°C can now be quantified by comparing the ṀO_2max after acute warming to 23°C (Chen et al., 2015) with that after acclimation to 23°C. Acutely warming to 23°C resulted in a 30% lower ṀO_2max and a 19% lower AAS than measured after acclimation to 23°C. Thus, the remarkable phenotypic plasticity for thermal acclimation of the H-strain rainbow trout provides a clear and substantial benefit to aerobic capacity at an acclimation temperature of 23°C. While another Californian population of rainbow trout can maintain 95% of its peak aerobic capacity when acutely warmed to 24.5°C (Verhille et al., 2016), its thermal acclimation potential is unmeasured. A pressing and unanswered question, therefore, is whether or not sufficient standing genetic variation exists within natural rainbow trout populations to acclimate to temperatures well above 20°C and deal with a certain degree of global warming. The Californian founder population of the H-strain and the redband trout certainly do.

The primary cardiac response of a fish to acute warming is to increase f_H (e.g. Sandblom and Axelsson, 2007; Steinhausen et al., 2008; Ekstrom et al., 2016), which has an upper limit (Farrell, 2016; Gilbert et al., 2019). Indeed, increasing f_H is the primary mechanism to meet an inexorable increase in tissue oxygen demand during acute warming when the mitochondrial efficiency of using oxygen to generate ATP sometimes decreases (Little et al., 2020). Even when fish are swimming, acute warming increases f_H up to a peak value (Steinhausen et al., 2008; Eliason et al., 2011, 2013). Our data for the effect of acute warming on f_Hmax are entirely consistent with this literature. Furthermore, warm acclimation of the H-strain helped retain a scope to increase f_H during acute warming, as shown previously for other fishes, which lowered intrinsic f_H and increased T_peak, T_arr and peak f_H (Farrell, 1991, 2016; Farrell and Smith, 2017). Warm acclimation of the H-strain of rainbow trout reset their f_Hmax (22-bpm lower in 25°C-acclimated than the 15-acclimated fish when tested at 12°C), increased their T_peak and T_arr (by 2.7°C and by 3.4°C, respectively, for 15°C- and 23°C-acclimated fish) but did not increase their peak f_Hmax. Similar to the earlier acute warming study of the H-strain rainbow trout acclimated to 15°C (Chen et al., 2015), our T_peak for f_Hmax for 15°C-acclimated fish (24.7°C) is similar to their value (23.5–24.0°C), as is the T_arr value (25.5°C vs 24.7–25.7°C). Peak f_Hmax (170–180 bpm) of the H-strain stood out compared with lower values for f_Hmax in other salmonids (e.g. Steinhausen et al., 2008; Eliason et al., 2011, 2013; Gilbert et al., 2019) including rainbow trout (≤130 bpm; Wood et al., 1979). The mechanism behind the faster heartbeat of the H-strain may be an important factor for their high thermal tolerance given that excessive warming is thought to eventually trigger conduction failure and cardiac arrhythmias in fish (Anttila et al., 2014; Farrell, 2016; Vornanen, 2017).

Any cardiac benefits of warm acclimation were clearly lost for the H-strain of rainbow trout acclimated to 25°C because peak f_Hmax (151 ± 6 bpm), T_peak (26.2 ± 0.8°C) and T_arr (27.2 ± 0.6°C) all decreased significantly. Indeed, their scope to increase f_Hmax during acute warming was small given that peak f_Hmax was reached at 26.2°C. In view of this and because resting and maximum f_H tend to converge at warm temperatures in trout and salmon (Steinhausen et al., 2008; Farrell, 2009; Eliason et al., 2011; Eliason et al., 2013), we propose a novel mechanism to explain both the appetite suppression and post-handling mortality in 25°C-acclimated H-strain trout that we observed. Increasing routine f_H is the primary cardiac response of fishes to meet the elevated ṀO₂ associated with SDA and dealing with handling stress (Clark et al., 2008; Eliason et al., 2008; Chen et al., 2018). Clearly, the capacity to mount such increases became very limited at the 25°C acclimation temperature. Lacking a capacity to increase f_H for digestion certainly aligns with the hypothesis purporting that cardiac function is a potential weak link for upper thermal tolerance of fishes (Wang and Overgaard, 2007; Pörtner and Farrell, 2008; Farrell, 2009). This mechanism would also align with the aerobic scope protection hypothesis (tissue oxygen extraction and cardiac stroke volume can still potentially increase somewhat).

CT_max is likely the most commonly used indicator of upper thermal tolerance for fish and rainbow trout as a species appear to have an upper ceiling for its upper thermal tolerance as measured by CT_max. Our CT_max value for 15°C-acclimated fish (29.0°C) compares well with an earlier determination for the H-strain acclimated to 15°C (28.1–29.7°C; Chen et al., 2015) and for a 15°C-acclimated Japanese strain of rainbow trout, which has been intensely selected for upper thermal tolerance over multiple generations and originally from California (29.7°C; Ineno et al., 2005). The CT_max for redband trout (28.8–29.8°C; Rodnick et al., 2004; Chen et al., 2018) is close to the upper daily water temperature recorded for their stream habitat (Rodnick et al., 2004). In contrast, CT_max values ranging from 21°C to 26.5°C are much lower for steelhead trout (O. mykiss) from the Columbia River and North Santiam River (Oregon) (Coutant, 1970; Redding and Schreck, 1979) than the 25°C-acclimated H-strain rainbow trout, which maintained CT_max around 31°C. Yet, this maintained upper thermal tolerance clearly contrasted with other physiological performance variables that had failed at this acclimation temperature, i.e. growth, appetite, FCR, the oxygen cost of digesting a meal and an increasing heart rate with acute warming. Likewise, the similar CT_max values for the Californian Eagle Lake (30.6°C) and Mount Shasta (30.0°C) strains of rainbow trout do not reflect intra-specific differences in their thermal growth performance curves (Myrick and Cech, 2000). Therefore, the merit of CT_max as a useful ecological indicator of upper thermal tipping points for life-supporting physiological functions must be questioned, despite the ease with which the measurement can be made. This concern is readily seen by using a thermal tolerance polygon (Fry, 1971) to directly compare CT_max, T_arr and T_peak (Fig. 3B; Table S2), physiological indices that were determined with a similar acute warming protocol. Independent of the acclimation temperature, T_arr and T_peak are consistently lower than CT_max (minimally by 2°C for T_arr and even more for T_peak), but the magnitude of this difference varies with acclimation temperature, being greatest amount at 15°C and 25°C, e.g. CT_max was 5°C higher than T_peak (26.2°C) at 25°C. This comparison provides compelling evidence that limitations to increasing f_H during acute warming can occur well below CT_max and that the faltering ability of the heart to deliver oxygen at different acclimation temperatures is not reliably predicted by CT_max for the H-strain of rainbow trout.

In conclusion, our upper thermal performance curves for the H-strain rainbow trout suggest that its Californian founder population had a suite of genes that were preferentially selected over many generations of inbreeding in an Australian desert climate to generate a strain with an unusually high upper thermal acclimation potential. Indeed, the H-strain seems most similar to a cousin, the redband trout, which has naturally adapted to desert streams that can reach 29°C. Despite this encouraging discovery for those with conservation concerns for rainbow trout and other fish species, a 25°C-acclimation temperature may not be sustainable for the H-strain rainbow trout given their slow growth rate, reduced appetite and aerobic scope, inefficient food conversion and processing, limited scope to increase f_H during warming and limited ability to tolerate handling stress. As seen for rainbow trout populations introduced to Japan and Argentina (Ineno et al., 2005; Crichigno et al., 2018) and much like the Swedish perch in the warm-water discharge enclosure of a nuclear power plant, which have a higher thermal tolerance than the endemic and founding perch population (Sandblom et al., 2016), the standing variation of a founder population may be an undiscovered but crucial source for biogeographic redistributions of certain fish species. Intra-specific differences in upper thermal performance now need to be sought in populations living at the southern end of their biogeographic range and, if discovered, will need to be considered when managing and modelling fish habitats for conservation purposes. Whether convergent evolution of high upper thermal performance respiratory phenotypes has a common or different suite of physiological mechanisms remains to be explored.

Funding

This project was supported by funding from Natural Sciences and Engineering Research Council of Canada to A.P.F., who holds a Canada Research Chair, and Mitacs Globalink Research funding to O.A.A. Travelling Fellowships from the Company of Biologists subsidized logistic costs of Y.Z. Y.Z. held Elizabeth R Howland Fellowship & Pei-Huang Tung and Tan-Wen Tung Graduate Fellowship.

Author Contributions

O.A.A. helped design the project, performed all the experimental work, performed data analysis, wrote the initial draft of the manuscript and performed subsequent edits. C.S.L., M.S. and A.P.F. conceived and designed the experiments and contributed to the writing of the manuscript. Y.Z. helped with setting up the experimental work, contributed to the data analysis, synthesized the thermal performance curves of aerobic capacity and assisted in the writing of the manuscript. M.J.H.G. contributed to the cardiac methodology, data analysis, editing of the manuscript.

Competing Interest

The authors declare no competing interests.

Supplementary material

Supplementary material is available at Conservation Physiology online.