Effects of acclimation temperature on the thermal physiology in two geographically distinct populations of lake sturgeon ( Acipenser fulvescens )

in of 2.1 – 3.4 ◦ C in Manitoba, sturgeon, fulvescens , from both northern and southern populations in Manitoba were acclimated to 16, 20 and 24 ◦ C for 30 days, after which critical thermal maximum (CT max ) trials were conducted to investigate their thermal plasticity. We also examined the effects of temperature on morphological and physiological indices. Acclimation temperature significantly influenced the CT max , body mass, hepatosomatic index, metabolic rate and the mRNA expression of transcripts involved in the cellular response to heat shock and hypoxia ( HSP70 , HSP90a , HSP90b , HIF-1 α ) in the gill of lake sturgeon. Population significantly affected the above phenotypes, as well as the mRNA expression of Na + /K + ATPase - α 1 and the hepatic glutathione peroxidase enzymeactivity.ThesouthernpopulationhadanaverageCT max at 20 and 24 ◦ C, respectively. Immediately following CT max trials, mRNA expression of HSP90a and HIF-1 α was positively max r = 0.7, r = 0.62,respectively; P 0.0001). Lake sturgeon acclimated to 20 and 24 ◦ C hepatosomatic indices and 244% reduction, P < 0.0001) and metabolic suppression and 42.1% reduction, respectively; P < 0.05) to sturgeon acclimated 16 ◦ C, of population. mRNA relative to the southern population. Acclimation to 24 ◦ C also induced mortality in both populations when compared to sturgeon acclimated to 16 and 20 ◦ C. Thus, increased temperatures have wide-ranging population-specific physiological consequences for lake sturgeon across biological levels of organization.


Introduction
Across Canada, mean annual and seasonal temperatures have increased between 1.7 and 2.3 • C from 1948 to 2016, with the largest increases occurring in northern Canada (Zhang et al., 2019;Vincent et al., 2015). By 2050, mean annual water temperatures within Manitoba, where several endangered populations of lake sturgeon exist, are projected to increase by 2.1-3.4 • C (Manitoba Hydro, 2015). As lake sturgeon are a long-lived species, which may require as long as 18-28 years to mature (COSEWIC, 2006;Scott and Crossman 1998), individuals in Manitoba today may live to see the effects of increased environmental temperatures, such as those projected for 2050. Furthermore, their progeny will most certainly experience elevated temperatures. As temperature is a critical factor impacting fish development (Aloisi et al., 2019;Piper et al., 1982), it is important to understand how future increased environmental temperatures will affect the development and physiology of this iconic species.
Lake sturgeon populations in Manitoba range throughout the province, from the Winnipeg River in the south, and northward to the outlet of the Nelson and Churchill Rivers into the Hudson Bay (Manitoba Hydro, 2016). Historical barriers exist in waterways between populations, which have limited gene flow (McDougall et al., 2017) and contributed to genetically distinct populations. These latitudinally separated riverine environments likely have differing temperature profiles that may influence population-specific life-history traits like the growth and thermal plasticity of these genetically distinct populations (Pollock et al., 2015). Since the mid-1990s, stocking of hatchery-reared lake sturgeon has been conducted to bolster the remaining wild populations in Manitoba (McDougall et al., 2014). At the lake sturgeon hatchery in Grand Rapids, Manitoba, located at the North end of Lake Winnipeg, northern fish are typically reared at 16 • C due to hatchery heating limitations, but the hatchery intends to increase rearing temperatures in the future with system upgrades. However, we know little regarding the potential effect that increased rearing temperatures may have on thermal physiology of lake sturgeon throughout development and post-release.
Many cellular, physiological and behavioural changes made by fishes as a result of increased environmental temperatures are assessed by changes in mRNA expression of genes associated with acute thermal stress, resultant cellular damage and acclimation such as heat-shock proteins (HSP70, HSP90a, HSP90b) (Fangue et al. 2006;Komoroske et al., 2015;Lund et al., 2002;Shi et al., 2015). Additionally, changes in genes associated with hypoxia tolerance (HIF-1α) and osmoregulation (Na + /K + ATPase-α1) have implicit roles in the cellular response to thermal stress (Jeffries et al., 2014;McBryan et al., 2013;Portner, 2010;Vargas-Chacoff et al., 2018). Furthermore, changes in condition factor, hepatosomatic index (HSI), metabolic rate and glutathione peroxidase (GPx) activity have been shown to relate to increasing environmental temperatures. Condition factor can fluctuate with temperature changes as well as seasonally due to abiotic and biotic factors (Giosa, et al., 2014;Mazumder et al., 2016). Thus, HSI can be a useful additional indicator of changes in body condition and metabolism as well as glycogen and lipid reserves (Chellappa et al., 1995) that are likely to change in the liver with variation observed in natural populations once temperature thresholds and upper physiological limits have been reached (Purchase and Brown, 2001;Morrison et al., 2020). Similarly, metabolic rates (both routine and maximum) are additionally affected by environmental temperature in many fishes (Norin and Clark, 2015) including sturgeons (Yoon et al., 2018; and likely play a role in the distribution of a species (Payne et al., 2015). Hepatic glutathione peroxidase (GPx) detoxifies oxidatively damaging peroxides formed as a result of acute and chronic thermal stress (Halliwell and Gutteridge, 1999) with increased mRNA expression and enzyme activity demonstrated with increased temperatures (Almroth et al., 2015;Dorts et al., 2012). The ability of lake sturgeon to make physiological changes to acclimate to their warming environment in response to thermal stress, i.e. phenotypic plasticity, is crucial in an ever-changing environment and may be a key predictor for a species future success (Rodgers et al., 2018;Somero, 2010;Seebacher et al., 2015;Gabriel et al., 2005). Divergent populationspecific responses of these physiological parameters may be anticipated in sturgeon populations from differing thermal environments as have been observed in other geographically separated populations (Fangue et al., 2006;Geerts et al., 2014;Pereira et al., 2017;Yampolsky et al., 2014).
The aim of this study was to use measurements of key physiological and molecular variables to evaluate the thermal plasticity of juveniles from different populations of lake sturgeon in Manitoba, Canada (Fig. 1). We acclimated lake sturgeon from both northern (Burntwood River-BR) and southern (Winnipeg River-WR) populations to three environmentally relevant thermal regimes of 16, 20 and 24 • C. As these two geographically distinct populations of lake sturgeon in Manitoba have independent environmental and genetic histories, we hypothesized that they would exhibit divergent population-specific responses to acclimation and acute thermal stress. We predicted that the southern WR population of lake sturgeon, with greater thermal variation and range, would demonstrate increased thermal plasticity compared to their northern BR counterparts as demonstrated in other species (Fangue et al., 2006;Geerts et al., 2014;Pereira et al., 2017;Yampolsky et al., 2014). Additionally, we predicted that these differences in plasticity would be apparent in the critical thermal maximum (CT max ) of lake sturgeon, and the differential expression of mRNA transcripts important in the response to heat shock, hypoxia and osmoregulatory disruption (HSP70, HSP90a, HSP90b, HIF-1α and Na + /K + ATPase-α1). Furthermore, we predicted temperature-dependent population-specific responses in body size, HSI, metabolic rate and GPx activity as observed in other species under thermally stressful environmental conditions.

River temperatures
River water temperatures were measured in both the WR and BR in 2019 by a DigiTemp SDI-12 submersible temperature sensor (Forest Technology Systems; Victoria, British Columbia, Canada) and a series 500 SDI-12 transducer (TE Connectivity; Schaffhausen, CH), respectively. Water temperature measurements in the WR were recorded downstream of the Pointe du Bois Generating Station where spawning individuals were caught for this study (50 • 17 52 N, 95 • 32 5 W). Water temperature measurements in the BR were recorded at the Miles Heart Bridge (55 • 45 12 N, 97 • 50 30 W) ∼74 air km southwest of the sturgeon spawning site (56 • 02 46.5 N, 96 • 54 18.6 W) on the same river. Measurements were taken at midnight (0:00:00 am) for both locations therefore they likely represent the lower range of daily temperatures in both rivers. Temperature comparisons between the WR and BR were based on days where data is

Lake sturgeon husbandry
In May and June of 2019, gametes from wild-caught female and male lake sturgeon were harvested from individuals at both the Pointe du Bois Generating Station on the WR (50 • 17 52 N, 95 • 32 51 W), and below first rapids on the BR (56 • 02 46.5 N, 96 • 54 18.6 W). Eggs and sperm from the WR were transported to the University of Manitoba animal holding facility where fertilization took place, whilst those from the BR population were fertilized at the Grand Rapids Fish Hatchery (53 • 09 25.9 N, 99 • 17 21.9 W). Individuals from the WR population were the product of fertilization of eggs from two females with the sperm from two males (two maternal families). Individuals from the BR population were the product of the fertilization of eggs from one female with the sperm from six males (one maternal family). Spawning individuals, particularly in the northern BR population, were limited due to the remote location of the spawning site and also the declining population, with a total adult population estimated between 112 and 579 sturgeon fish to sturgeon (Lacho and Hrenchuk, 2018). Additionally, the number of families in the southern WR population was limited due to the number of wild-caught spawning females that were able to be sampled in 2019. Post-fertilization, embryos from both populations were de-adhered by submerging embryos in a clay solution (Fullers Earth; Earhart et al., 2020) and gently stirred by hand for 1 h. Embryos were then rinsed with de-chlorinated freshwater and incubated in tumbling jars at 12 • C until hatch, which occurred at ∼9 days post- fertilization (DPF). Post-hatch, larvae of equal numbers from each maternal family were transferred to a total of four 9 L flow-through aquaria, two tank population −1 , with aeration and bio-balls as substrate. The temperature was maintained at 12 • C until 13 DPF, after which temperature was increased at 1 • C day −1 until 16 • C to match hatchery rearing conditions. Freshly hatched artemia (Artemia International LLC; Texas, USA) was provided as a starting diet at 19 DPF, prior to complete yolk-sac absorption, after which tank substrate was removed over a 7-day period. Lake sturgeon were fed to satiation three times daily on a diet of artemia throughout the experiment. All animals in this study were reared and sampled under guidelines established by the Canadian Council for Animal Care and approved by the Animal Care Committee at the University of Manitoba under Protocol #F15-007.

Acclimation
Beginning at 30 DPF, lake sturgeon from each population were acclimated to treatments of 16, 20 and 24 • C at a density of ∼ 70 sturgeon 9 L aquaria −1 (Fig. 2). As there were more families, and thus more individual sturgeon from the WR population, lake sturgeon from the WR were reared in four tanks, whilst BR sturgeon were reared in two tanks for each acclimation treatment, to keep stocking density equal across populations throughout acclimation. In the 20 and 24 • C treatments, the water temperature was increased from 16 • C at a rate of 1 • C day −1 until the desired acclimation temperature was reached. Lake sturgeon remained at these acclimation temperatures for 30 days, and the temperature was recorded every 15 min by HOBO Water Temperature Pro v2 Data Loggers (Onset Computer Corporation; Bourne, MA, USA). Mortality as well as temperature via thermometer was recorded at least three times daily. After 29 days of acclimation, fish were fasted for a 24-h period prior to beginning CT max trials (Downie and Kieffer 2016;Downie et al., 2018;Lee et al., 2016).
At the end of the acclimation, prior to CT max trials, eight fish from each treatment were haphazardly selected and euthanized by immersion in an overdose of tricaine methanesulfonate solution (250 mg L −1 ; MS-222, Syndel Laboratory, Vancouver, Canada) buffered with an equal volume of sodium bicarbonate. Gill tissue was then extracted, preserved in RNAlater (Thermo Fisher Scientific, Waltham, USA), and stored at −80 • C prior to the quantification of mRNA transcripts. An additional 10 fish were haphazardly selected and euthanized from each acclimation treatment; body mass (weighed to 0.0001 g) and total length (measured to nearest 1 mm) were recorded for each individual, as well as liver wet mass (weighed to 0.0001 g) which was used to calculate the HSI as the ratio of the wet mass of the liver (W liver ) to the wet mass of the body (W body ):

Critical thermal maximum trials
On the day of the CT max trials, eight fish were haphazardly selected from acclimation tanks and placed individually in experimental units (∼200 ml of water volume and 9.5 cm long × 5 cm across) in an aerated recirculating water bath initially set to the acclimation temperature of the treatment being tested. Experimental units had mesh-screened sides to allow for water flow through each unit. Eight fish were tested trial −1 , with three to four trials conducted over two consecutive days for each experimental treatment. CT max trials were conducted at the same time every day, between 9 am and 12 pm to avoid any confounding effects of diurnal shifts on physiology and gene expression (Lankford et al., 2003;Somero, 2020). The temperature of the water bath was regulated by an Isotemp recirculating heater (Fisher Scientific; Hampton, USA) whilst water temperature was constantly recorded by a temperature probe placed in the centre of the experimental setup (Witrox Oxygen probe, Loligo Systems; Viborg, Denmark). Fish were held in these experimental units for 1 h prior to the CT max trials to reduce the potential effects of handling stress. After 1 h, trials began by increasing the temperature of the water bath by 0.3 • C min −1 until fish were unable to right themselves after a physical disturbance (Bard and Kieffer, 2019;Beitinger et al., 2000;Yoon et al., 2019;Yusishen et al., 2020). When fish were unable to right themselves, the final CT max temperature was recorded, the fish was euthanized, mass and length was recorded and the gill tissue was removed and preserved as previously described. Liver tissue was then sampled, immediately flash-frozen in liquid nitrogen, and stored at −80 • C until use for measuring GPx activity. An additional eight fish from a single trial were placed individually into 9 L tanks in a Multi-Stressor Unit (AquaBioTech; Coaticook, Quebec, Canada) at their respective acclimation temperature after their CT max was reached and allowed to recover for 3 days before tissue sampling. During this 3-day recovery period, lake sturgeon were fed and were observed actively feeding on freshly hatched artemia three times daily. From average CT max values, an acclimation response ratio, the rate an organism increases their CT max in response to acclimation, was calculated for each population, subtracting the average CT max of the 16 • C acclimation treatment (CT max16 • C) from that of the 24 • C treatment (CT max24 • C) and dividing by the change in acclimation temperature between treatments ( • C):

Metabolic rate
Measurements of whole-body metabolic rate (ṀO 2 ) were taken 3 days after each CT max trial for a given acclimation treatment, using intermittent flow respirometry (Loligo Systems, Viborg, Denmark) as previously described (Yoon et al., 2019) with some modifications. In brief, fish were fasted for at least 12 h prior to experimentation. Flow within the chambers [volume: 43.40 +/− 4.32 (mean +/− S.D.) ml] was maintained at a low level to not cause any physical stress, but sufficient for water exchange and accurate measurement ofṀO 2 . The intermittent respirometry cycle was variable for each temperature treatment to ensure a linear decline in oxygen saturation more than 10%, but not below 70% as metabolic suppression was reported at this point in age 1+ lake sturgeon (Svendsen et al. 2014). For 20 and 24 • C, the parameters were 360 s of flush followed by 60 s of wait and 300 s of measurement. For 16 • C, 360 s of flush followed by 60 s of wait and 900 s of measurement was used. The respirometry chambers were surrounded by a black curtain to minimize any visual disturbance to the fish during each trial. Routine metabolic rate (RMR) was assessed for 2 h following a 4-h period in the metabolic chamber to minimize the effects of transfer. After RMR was assessed, fish were removed from the chambers and a standardized chase protocol was performed for 15 min. Then, fish were immediately returned to the same chambers, andṀO 2 was measured for the following two measurement cycles. Biological oxygen demand (BOD) was measured before and after each experiment. Assuming a linear increase, pre-and post-experiment BOD data points were used to linearly interpolate BOD over the experiment period and all MO 2 data was corrected by the corresponding BOD. Only slopes of oxygen decline with R 2 ≥ 0.9 were used for data analysis. RMR was calculated by averaging MO 2 measured for 2 h whilst the maximum metabolic rate (MMR) was chosen as the highestṀO 2 after the chase protocol. Factorial aerobic scope (FAS) was calculated by dividing MMR by RMR. There was no 24 • C acclimation treatment available for the BR population for MMR, RMR or FAS as there were insufficient numbers of lake sturgeon remaining to conduct these trials.

RNA extraction, cDNA synthesis and qPCR
Gills from lake sturgeon were homogenized in 500 μl of lysis buffer, for 10 min at 50 Hz using a TissueLyser II (Qiagen; Germantown, MD, USA). Total RNA was extracted from gill homogenates using a PureLink RNA Mini Kit (Invitrogen; Ambion Life Technologies) following the manufacturer's instructions. Total RNA purity and concentration were evaluated for all samples using a NanoDrop One (Thermo Fisher Scientific) followed by gel electrophoresis to assess RNA integrity. Post-extraction, RNA samples were stored at −80 • C. cDNA was synthesized from 1 μg of DNAse treated total RNA using a qScript cDNA synthesis kit following the manufacturer's instructions (Quantabio; Beverly, Massachusetts). Synthesis was conducted using a SimpliAmp Thermal Cycler (Thermo Fisher; Waltham, Massachusetts) with cycling conditions of 1 cycle of 22 • C for 5 min, 1 cycle of 42 • C for 30 min and 1 cycle of 85 • C for 5 min and hold at 4 • C. Following synthesis, cDNA samples were stored at −20 • C.
Real-time quantitative polymerase chain reaction (RT-qPCR) for each gene of interest, HSP70, HSP90a, HSP90b, HIF-1α and Na + /K + ATPase-α1, was conducted using 5 μl of Bio-Rad SsoAdvanced Universal SYBR Green Supermix, 0.1 to 0.04 μl of 100 μM primers, 2 μl of diluted cDNA per sample and nuclease-free water adjusted for each assay to bring the total volume of each well to 10 μl (Table 1). For all experimental assays except HIF-1α, each well contained 0.025 μl forward and 0.025 μl reverse primer, whilst this was doubled to 0.05 μl forward and 0.05 μl reverse for each reference gene. For HIF-1α, 0.02 μl forward and 0.02 μl reverse primer were used. The cDNA of all samples was diluted 1:10 with nuclease-free water for all RT-qPCR assays. All primers were designed based on sequences from an annotated transcriptome produced by the pyrosequencing of a lake sturgeon ovary (Table 1; Hale et al., 2009). The expression of the genes of interest was normalized to the relative expression of reference genes RPS6 and RPL7 and then analyzed after applying the 2 -Ct method described by Livak and Schmittgen (2001). Expression of all genes was then normalized to that of the WR 16 • C acclimation treatment prior to CT max trials in order to make comparisons between populations. The WR 16 • C acclimation treatment was chosen as a reference based on hatchery rearing conditions and its relatively low levels of mRNA expression across acclimation treatments, time points and genes of interest.

Glutathione peroxidase activity assays
Initial extraction of GPx from the lake sturgeon livers was conducted by homogenizing tissues in ice-cold homogeniza-  Target genes were chosen based on their roles in the response to heat shock, cellular stress, hypoxia and osmoregulatory disruption. RPS6 and RPL7 were used as reference genes and showed stable expression across treatments. Efficiencies are listed as a percentage tion buffer (50 mM Tris-HCl, 5 mM EDTA and 1 mM DTT; pH 7.5). Briefly, 50 μl of homogenization buffer was added to each vial containing tissue, which was then freeze-thawed in liquid nitrogen three consecutive times. Tissues were then centrifuged (accuSpin Micro 17R, Fisher Scientific) for 10 min at 10000 RPM at 4 • C. The supernatant was removed and stored at −80 • C prior to assay measurement, which occurred between 1 and 3 days post-extraction. Concentrations of GPx were measured using a commercially available enzyme assay kit in 96-well plates (Cayman Chemical; Ann Arbor, Michigan, USA). Samples were diluted based on the manufacturer's instructions, where the concentration of GPx in the sample decreased the absorbance of the well to between 0.02 and 0.135 min −1 . Twenty microliters of the sample was added to each well on the assay plate, followed by 50 μl assay buffer, 50 μl co-substrate mixture and 50 μl NADPH. Sample wells were compared to the background wells (70 μl assay buffer, 50 μl co-substrate mixture and 50 μl NADPH) and positive control wells (50 μl assay buffer, 50 μl co-substrate mixture, 50 μl NADPH and 20 μl GPx control) included on each plate. Absorbance was measured at 340 nm using a plate reader (PowerWave XS2, BioTek), and a linear decrease in NADPH was observed. Backgrounds, positive controls and samples were run in duplicate. The intra-and inter-assay variations were 6.05 and 13.83%, respectively. Data was not corrected for inter-assay variability.

Statistical analysis
Differences in mortality between treatments and populations were analyzed via Cox proportional hazards model in R v3.6.2 (R core Team, 2013) using the 'survival' and 'survminer' packages (Kassambara et al., 2019;Therneau, 2015). Assumptions for all Cox proportional hazard models were assessed using the 'cox.zph' function included in the 'survival' package, to ensure that residuals were independent of time. First, a Cox proportional hazards model was run with only the effect of temperature included in the model, to isolate the effect that different thermal environments had on lake sturgeon regardless of population. Next, in addition to a Cox proportional hazards model with covariates of tempera-ture and population included, the 'pairwise_survdiff' function from the 'survminer' package with the same covariates and a Bonferroni correction was used to compare the mortality across treatments and populations of lake sturgeon.
CT max data was analyzed using non-parametric statistical tests as it could not be transformed to adequately pass the Levene's test. Thus, Kruskal-Wallis multiple comparison tests and a Bonferroni correction were used to determine significance within populations and across acclimation treatments, using the 'dunn.test' package (Dinno, 2017). Finally, the Wilcoxon signed-rank test was used to determine significance between populations, within a single acclimation treatment. As CT max data violates assumptions of parametric tests, Spearman's correlations were used to analyze the relationship between expression of the mRNA transcripts and GPx activity at each relevant time point to the CT max of individual lake sturgeon across acclimation treatments and populations. Only significant correlations are reported.
Morphometrics including mass, length, condition factor and HSI, as well as measurements of metabolic rate were analyzed using a two-factor ANOVA with population and acclimation treatment and their interaction included in the model as fixed effects. Three-factor ANOVAs were used to analyze GPx activity and mRNA expression of HSP70, HSP90a, HSP90b, HIF-1α and Na + /K + ATPase-α1 with population, acclimation treatment and time as well as their interactions included in the model as fixed effects.
For all ANOVAs, Shapiro-Wilks and Levene's tests were used to assess the normality of data and homogeneity of the variance, respectively. Normality was also visually inspected using fitted residual plots. If assumptions of normality or homogeneity were violated, either a ranked, log or square root transformation was applied to the data set. Additionally, for each ANOVA, the effect of the rearing tank was assessed and found to be not significant; therefore, it was not included in final models. Detailed results can be found for all statistical analyses (see Supplementary Material). Following ANOVAs, multiple comparison tests were performed and corrected with Tukey's honestly significant difference tests from the 'mult- comp' package (Hothorn et al., 2008). All statistical analyses were performed with a significance level of 0.05.

River temperatures
In 2019, the WR temperatures exceeded 20 • C on 17.6% of measured days, with 36.7% above 16 • C, and 63.2% below 16 • C. In contrast, the BR was never recorded above 20 • C, with temperatures above 16 • C recorded on 14.7% of days and 85.3% of days below 16 • C (Fig. 3). Days exceeding 16 • C were consecutive for both populations, above this threshold in the WR for 100 days from 20 June to 28 September and in the BR for 40 days from 11 July to 20 August. Temperatures in the WR were above 20 • C for 51 days from 8 July to 28 August. Throughout the summer, 21 June to 21 September, when larval lake sturgeon are developing, the BR was on average 4.6 ± 0.8 • C colder than the WR.

Mortality
Lake sturgeon reared at 24 • C had elevated mortality compared to those at 16 and 20 • C, with a hazard ratio of 5.37 (P < 0.0001). The BR and WR lake sturgeon had increased mortalities as temperatures increased, with 4.8, 6.9 and 25.5% mortality and 3.3, 3.3 and 15% (P < 0.05; n = 145 treatment −1 ) in 16, 20 and 24 • C, respectively.

Morphometrics
There were significant effects of population and acclimation treatment on mass and length of lake sturgeon (P < 0.001) as well as an interaction of population and acclimation treatment for mass (P < 0.05) and a near interaction for length (P < 0.06). Lake sturgeon from the BR population had a body mass of 0.62 g in 16 • C increased to 0.92 g in 20 • C and were larger when compared to WR fish in both treatments (P < 0.01; Table 2). The BR lake sturgeon had significant increases in mass in 20 (47.1%) and 24 • C (46.9%) when compared to 16 • C (P < 0.001). In contrast, lake sturgeon from the WR had increased mass with each acclimation treatment, with individuals in 20 • C heavier than the those in 16 • C (P < 0.05) and those in 24 • C heavier than both lower treatments, 84.3 and 32.8%, respectively (P < 0.001).
Sturgeon from the BR population had increased body length of 15.4% in the 20 • C acclimation treatment when compared to WR lake sturgeon (P < 0.001; Table 2). The BR lake sturgeon had significant increases in length in 20 and 24 • C treatments when compared to 16 • C (P < 0.0001). In contrast, sturgeon from the WR had increased length with each increasing acclimation treatment. Individuals at 20 • C were 8.5% longer than those at 16 • C whilst those at 24 • C were longer than both lower treatments, 23.4 and 13.7% respectively (P < 0.005). There was a significant interaction between population and acclimation treatment on condition factor in lake sturgeon (Table 2 (10) Significance was determined by a two-factor ANOVA (P < 0.05) followed by Tukey's honestly significant difference post hoc test. Asterisks represent significant differences between populations at each acclimation treatment. Letters represent significant differences within populations, across acclimation treatments. Morphometric data are expressed as mean +/− SD (n = 10-16 per treatment-indicated by parentheses) lake sturgeon had condition factors 13.8% greater than BR sturgeon from the same acclimation treatment (P < 0.05; Table 2).
There was an effect of both population and acclimation treatment on the HSI of lake sturgeon ( Fig. 4; P < 0.001). Sturgeon from both the WR and BR acclimated to treatments of 16 • C had increased HSI when compared to fish at 20 and 24 • C (P < 0.0001). The WR lake sturgeon acclimated to 16 • C had hepatosomatic indices 244 and 93% higher than those acclimated to treatments of 20 and 24 • C, respectively (P < 0.0001), with similar results in the BR population (P < 0.05). The BR lake sturgeon HSI was increased when compared to the WR across all acclimation treatments (P < 0.05).

Critical thermal maximum
Acclimation treatment and population influenced the CT max with each subsequent acclimation treatment increasing the CT max for each population (P < 0.05; Fig. 5). At both 20 and 24 • C, the WR population had increased CT max compared to the northern BR counterparts by 0.71 and 0.45 • C, respectively (P < 0.05). Comparisons of CT max show that the two populations of lake sturgeon have differing acclimation response ratios. The WR lake sturgeon acclimation response ratio was 0.41, whilst their BR counterparts were 0.34 over the same acclimation treatments.

Metabolic rate
RMR was significantly affected by both population and acclimation temperature (P < 0.0001). There were differences Figure 4: Hepatosomatic index of WR and BR lake sturgeon, Acipenser fulvescens, after 30 days of acclimation to 16, 20 and 24 • C. Significance was determined by a two-factor ANOVA (P < 0.05) followed by Tukey's honestly significant difference post hoc test. Asterisks represent significant differences between populations at each acclimation treatment. Lowercase letters represent significant differences between acclimation treatments, within populations. Box plots represent the mean and 25th and 75th percentiles, whilst whiskers indicate the minimum and maximum values. Dots represent individual data points (n = 10-13 per treatment) in RMR in the WR population between 16 and 24 • C as well as 20 and 24 • C (P < 0.01; Fig. 6A). In the WR 24 • C acclimation treatment, there was a reduction in RMR compared to 16 and 20 • C. In the BR population, there was : CT max of WR and BR lake sturgeon, Acipenser fulvescens, after 30 days of acclimation to 16, 20 and 24 • C. Significance differences within populations, across acclimation treatments, was determined by Kruskal-Wallis multiple comparisons with P-values adjusted with the Bonferroni method. Significant differences across populations, within a single acclimation treatment, was determined by Wilcoxon signed-rank test. Asterisks represent significant differences between populations at each acclimation treatment. Lowercase letters represent significant differences between acclimation treatments, within populations. Box plots represent the mean and 25th and 75th percentiles, whilst whiskers indicate the minimum and maximum values. Dots represent individual data points (n = 24-32 per treatment) a 26% decrease in RMR between 16 and 20 • C (P < 0.05). Between populations, the WR population had 28 and 70.1% higher RMR than BR in 16 and 20 • C, respectively (P < 0.01). The MMR was significantly affected by population and acclimation temperature (P < 0.01). There were differences in MMR in the WR population between 16 and 24 • C, as well as 20 and 24 • C (P < 0.05; Fig. 6B). Between populations, the WR population had 29.8% higher MMR than the BR in 20 • C (P < 0.05). In the WR population, acclimation to 24 • C led to a reduction of MMR compared to the 16 and 20 • C. There was an effect of acclimation temperature on FAS (P < 0.05; Fig. 6C). In the BR population, FAS was 29.6% higher at 20 • C than 16 • C (P < 0.01).

mRNA transcript expression
There was an interaction between population, acclimation treatment and time on the mRNA expression of HSP70 (P < 0.05; Fig. 7A). Immediately post-CT max trials, mRNA expression of HSP70 was elevated in every treatment across populations compared to pre-trial levels, and in the WR population 1.7-fold compared to the BR population at 24 • C (P < 0.05). Three days post-CT max , the BR population expressed HSP70 mRNA 2.9-fold and 4.6-fold higher at 16 • C than lake sturgeon from 20 and 24 • C, respectively (P < 0.05). and 24 • C. Significance was determined by a two-factor ANOVA (P < 0.05) followed by Tukey's honestly significant difference post hoc test. Asterisks represent significant differences between populations at each acclimation treatment. Lowercase letters represent significant differences between acclimation treatments, within populations. Box plots represent the mean and 25th and 75th percentiles, whilst whiskers indicate the minimum and maximum values. Dots represent individual data points (n = 7-8 per treatment) There was also an interaction between population, acclimation treatment and time on the mRNA expression of HSP90a (P < 0.0001; Fig. 7B , (E) Na + /K + ATPase-α1, (F) liver glutathione peroxidase (GPx) enzyme activity of lake sturgeon, Acipenser fulvescens, acclimated to 16, 20 and 24 • C, pre-, immediately post-and 3 days post-CT max trials. Asterisks represent significance between WR and BR populations of lake sturgeon. Lowercase letters a and b represent significance between acclimation treatments. Lowercase letters x, y and z represent significance between time points (P < 0.05; three-factor ANOVA). Data are expressed as mean +/− SEM [HSP70 n = 6-8, HSP90a n = 6-8, HIF-1α n = 5-8, HSP90b n = 5-8, Na + /K + -α1 n = 5-8, glutathione peroxidase enzyme activity n = 6-8] cant differences in HSP90a mRNA expression across acclimation treatments for both populations. In the WR population, mRNA expression of HSP90a decreased with increasing acclimation temperature with the 16 • C treatment demonstrating expression 5.3-fold higher than the 24 • C acclimation treatment (P < 0.05). In contrast, the BR population increased expression in the 24 • C treatment when compared to 16 and 20 • C. These opposite patterns of HSP90a expression resulted in differences between populations within acclimation treatments of 16 and 24 • C (P < 0.05). Within the 16 • C acclimation treatment, the WR population had a 6.7-fold higher expression compared to that of the BR. In the 24 • C treatment, the BR population increased expression 3.3-fold when compared to that of the WR.

Conservation Physiology • Volume 08 2020
Research article of HSP90a immediately following the CT max trials was positively correlated with individual CT max of lake sturgeon across acclimation treatments and populations (ρ = 0.7; P < 0.0001).
Three days post-CT max trials, mRNA expression of HSP90a was significantly elevated in the BR 16 • C acclimation treatment 3.1-and 2.4-fold compared to 20 and 24 • C treatments, respectively (P < 0.05). There were also elevated levels of HSP90a expression in the WR 24 • C acclimation treatment (P < 0.05), and near significantly elevated levels in the 20 • C acclimation treatment (P < 0.06), when compared to the BR population.
There was an effect of treatment (P < 0.0001) and a trend towards an effect of the population (P < 0.051) on the expression of HIF-1α mRNA (Fig. 7C). Pre-CT max , expression was elevated in the WR 20 • C acclimation treatment 5.2-fold compared to individuals from the same population in the 16 • C treatment (P < 0.05). Immediately following CT max , 24 • C treatments for both populations demonstrated elevated HIF-1α mRNA expression of 5.6-and 3.6-fold relative to WR and BR lake sturgeon in 16 • C, and the WR 20 • C treatment had 4.3-fold higher expression than 16 • C. The expression of HIF-1α mRNA immediately following CT max was positively correlated with individual CT max of lake sturgeon across acclimation treatments and populations (ρ = 0.62; P < 0.0001).
There was an effect of time (P < 0.005) and a near significant effect of the population (P = 0.09) on the mRNA expression of HSP90b mRNA (Fig. 7D).
There was an effect of population and time on the expression of Na + /K + ATPase-α1 mRNA (P < 0.05; Fig. 7E), which demonstrated consistently elevated expression in the BR compared to the WR population across acclimation treatments and time points.

Glutathione peroxidase activity
There was an interaction between population, acclimation treatment and time on hepatic GPx activity (P < 0.05; Fig. 7F). Pre-CT max , the BR population demonstrated a 1.5fold increase in GPx activity when compared to their WR counterparts at 20 • C (P < 0.05). Three days post-CT max trials, the BR population showed a 1.8-fold increase in GPx activity compared to their WR counterparts at 16 • C (P < 0.05). There was also a significant trend in GPx activity in the BR population 3 days post-CT max , where the 24 • C acclimation treatment had activity 2.4-fold higher than the 20 • C acclimated lake sturgeon from the same population (P < 0.05). The same 20 • C acclimated BR treatment also showed lower GPx activity 3 days post-trial compared to pre-and post-trial levels (P < 0.05).

Discussion
In the current study, we demonstrated changes in diverse physiological and molecular phenotypes in response to accli-mation temperature in juvenile lake sturgeon from a northern and southern population in Manitoba. In addition, lake sturgeon populations were affected differently, with alteration of many physiological and molecular characteristics being population dependent. To the best of our knowledge, this is the first study to investigate the thermal tolerance of sturgeons across populations, sturgeon thermal tolerance at multiple levels of biological organization and lake sturgeon thermal tolerance at the molecular level.

Physiological responses to acclimation
Mortality was elevated in 24 • C acclimation treatments, when compared to 16 • C and 20 • C treatments suggesting that this treatment was stressful to lake sturgeon, regardless of population. These increased rates of mortality may be indicative of increased metabolic constraints and cellular stress due to chronic thermal stress potentially leading to decreased pathogen tolerance as observed in Atlantic cod, Gadus morhua, and three-spined sticklebacks, Gasterosteus aculeatus, during acclimation to thermally stressful conditions (Larsen et al., 2018;Dittmar et al., 2013). Additional signs of metabolic stress are apparent in the decrease in both RMR and MMR at 24 • C in the WR population compared to 16 and 20 • C. In the BR population, RMR was lower than WR sturgeon and decreased between 16 and 20 • C; whilst in the WR, it remained consistent between 16 and 20 • C. These differing thresholds for metabolic suppression between populations may be influenced by their population-specific thermal histories with the BR population experiencing lower yearly temperatures and decreased metabolic rates at lower acclimation temperatures. However, FAS increased with acclimation temperature in both populations, indicating a greater separation between MMR and RMR, potentially resultant from higher rates of metabolic development under increased temperatures. In green sturgeon, until a limiting factor such as food availability is reached, growth increases with environmental temperature (Poletto et al., 2018). In the current study, increased mass of BR lake sturgeon, when compared to WR sturgeon at 16 and 20 • C, may be indicative of countergradient variation (e.g. Fangue et al., 2009) as lake sturgeon from the northern BR population possibly grow faster to take advantage of shorter growing seasons. However, at 24 • C, the BR population demonstrated no further increase in either mass or length suggesting an upper thermal limit for growth in this population (e.g. Koskela et al., 1997;Oyugi et al., 2012). In contrast, the WR population had increased in size with each temperature. These results suggest the presence of population-specific upper thermal thresholds for lake sturgeon in Manitoba.
Investigation of HSI indicated an additional influence of elevated temperatures on the liver size in lake sturgeon with decreased HSI apparent in acclimation temperatures above 16 • C for both populations. Decreases in HSI have been linked to diminishing glycogen reserves in three-spined stickleback Gasterosteus aculeatus (Chellappa et al., 1995) and white sturgeon (Hung et al., 1990). These differences in HSI may

11
be the result of a trade-off between whole-body RMR and liver function. A reduction in RMR of BR lake sturgeon may facilitate decreased energy consumption and result in an increase of hepatic glycogen and lipid reserves that were not evident in WR lake sturgeon. In the northern part of their range, lake sturgeon must survive through an extensive overwintering period wherein lipid stores likely play a critical role in survival (Byström et al., 2006;Yoon et al., 2019). Coldadapted populations of fish may be better able to accumulate energy stores, especially glycogen and lipids, as observed in Atlantic cod from different thermal environments (Purchase and Brown, 2001). Thus, lower metabolic rates and higher HSI may be potential evidence of how the northern BR population copes with prolonged sub-Artic winters (Lotka, 1922;Schaefer and Walters 2010), with the adoption of an energy storage maximization strategy (Post and Parkinson, 2001). Consequently, rearing lake sturgeon for prolonged periods at temperatures above 16 • C may not allow them to accrue these necessary glycogen and lipid reserves, thereby decreasing their ability to survive this overwintering period if released prior to winter. However, further research is necessary to confirm this observation.

Molecular responses to acclimation
In addition to physiological responses, there were population differences in the molecular responses of lake sturgeon to the acclimation treatments. At 30 days of acclimation and pre-CT max trials, lake sturgeon from the WR and BR populations exhibited opposite patterns of HSP90a expression with a significant threshold for both populations between 20 and 24 • C. In the BR population, mRNA expression of HSP90a is induced in 24 • C, relative to the 20 and 16 • C, whilst the WR population demonstrated suppressed mRNA expression in 24 • C, relative to 20 and 16 • C. These opposite patterns in HSP90a mRNA expression may be indicative of differing mechanisms used to handle chronic thermal stress as observed in redband trout, Oncorhynchus mykiss gairdneri, from different environments (Narum et al., 2013). Additionally, mRNA expression of HIF-1α in the WR 20 • C acclimation treatment increased, relative to 16 • C, but this increase is not observed at 24 • C, potentially demonstrating a temperature-dependent threshold in this population, similar to what has been observed in Crucian carp, Carassius carassius (Rissanen et al., 2006). HIF-1α expression peaking pre-CT max , in an inverted U-shape may be indicative of sublethal response thresholds that could be predictive of long-term impacts Jeffries et al., 2018). It is energetically costly to produce and activate HSPs (Heckathorn et al., 1996;Sanchez et al., 1992) and HIF-1α likely has metabolic influences (Pelster and Egg, 2018;Richards, 2009). These population-specific metabolic, growth, HSI and molecular response patterns across acclimation temperatures likely indicate that northern and southern populations differently handle constrained metabolic budgets due to elevated temperatures (Somero, 2020;Tomanek and Somero, 1999). These physiological factors likely play a role in observed behavioural differences in seasonal movement patterns between the northern and southern populations of lake sturgeon across their geographic range, as fish from warmer southern climates move less in the summer than their northern counterparts (Moore et al., 2020).

CT max
In the present study, CT max across treatments and populations ranged from 32.2 to 35.4 • C, which is within the range reported for lake sturgeon (Wilkes, 2011;Yusishen et al., 2020). The CT max results demonstrated that lake sturgeon from the WR had increased acclimation response ratios when compared to their BR counterparts. However, both populations had acclimation response ratios higher than most species from similar and lower latitudes (Gunderson and Stillman, 2015) which indicates increased thermal plasticity, potentially based on the large genome size of lake sturgeon (Ellis et al., 2014;Fontana et al., 2004). Differences in CT max between populations, 0.45 to 0.71 • C, are potentially indicative of population-level effects and not just family effects, as families of lake sturgeon from the same river systems have exhibited ±0.18 • C in CT max (95% confidence interval; six families; six individuals per family; range 32.4-32.9 • C; Deslauriers et al., in revision). As greater climate variability affects thermal plasticity in animals (Rohr et al., 2018;Seebacher et al., 2015;Somero, 2010), the increased acclimation response ratio in the southern population of lake sturgeon is potentially linked to the more variable thermal environment and greater thermal range that this population would experience.

Molecular and enzymatic responses to CT max
Plasticity in mRNA expression of transcripts involved in response to thermal and hypoxic stressors, HSP90a and HIF-1α, was observed immediately following CT max trials. HSP90 is a highly constitutively expressed heat-shock protein that also functions to protect cells from thermal stress by aiding in substrate recognition, cellular signalling and refolding of misfolded proteins (Iwama et al., 2004;Li et al., 2012;Mahanty et al., 2017). In the current study, there was a positive relationship between acclimation temperature and HSP90a mRNA expression post-CT max in both populations. The mRNA expression of HSP90a was the most highly induced of all the genes measured immediately following CT max trials (11.3-to 163-fold increase, relative to pre-trial levels). In contrast, expression of the constitutive isoform, HSP90b, did not display a similar plastic response following CT max trials. Similar to HSP90a, a plastic response is observed in mRNA expression of HIF-1α, a transcription factor involved in the response to hypoxia and regulated by temperature in Crucian carp (Rissanen et al., 2006;Wenger, 2002). In the current study, immediately following CT max trials, lake sturgeon from both populations at 24 • C increased mRNA expression of HIF-1α relative to those acclimated to 16 • C, and this expression was also correlated to individual CT max across acclimation treatments and populations. The

12
WR population exhibited an increase in HIF-1α mRNA expression between 16 and 20 • C, continuing at 24 • C. However, this same increase was delayed in the BR population, only occurring at 24 • C, possibly indicative of differing thermal thresholds for expression induction and the role that this protein may play in cross-tolerance to thermal stress (Maloyan et al., 2005), although this was not specifically addressed in this study.
The mRNA expression of HSP70 showed elevation in response to CT max trials, with increases in expression ranging from 3-to 10-fold across acclimation treatments and populations. Changes in gill HSP70 mRNA expression can subsequently lead to much higher levels of protein expression as observed in the gills of the goby, Gillichthys mirabilis (Buckley et al., 2006;Somero, 2020). Under normal conditions in eukaryotes, HSP70 functions as a required chaperone for protein assembly (Lindquist, 1992;Roberts et al., 2010). Under times of thermal stress, HSP70 can act to bind to and stabilize proteins against misfolding and prevent intracellular aggregation, serving as an indicator of stress severity (Logan and Somero, 2011;Tomanek and Somero, 1999;Welch and Feramisco, 1985;Yamashita et al., 2010). Immediately following the CT max trials, there was an increase in mRNA expression of HSP70 in the WR 24 • C treatment, compared to their BR counterparts. This relative increase in expression may be indicative of the WR population's greater ability to acutely upregulate mRNA expression of HSP70 after acclimation to thermally stressful temperatures as reported in rainbow trout, O. mykiss (Currie et al., 2005). This increased response was not observed in the BR population and may play a role in the increased CT max of the WR population in comparison to the BR in this acclimation treatment. Three days post-CT max , expression of HSP70 returned to near baseline levels in all treatments, except for BR 16 • C, which remained elevated compared to 20 and 24 • C. Similarly, 3 days post-CT max , HSP90a was elevated in the BR 16 • C treatment as compared to 20 and 24 • C and depressed in the 24 • C treatment compared to pre-trial levels, but neither of these observations was true for the WR population. Elevated mRNA expression of HSP70 and HSP90a 3 days post-CT max in the BR 16 • C in comparison to 20 and 24 • C acclimation treatments was possibly due to increased cellular damage, slower rates of cellular repair and delayed recovery that may impact long-term individual fitness (Jeffries et al., 2018;Tomanek and Somero, 2000), whilst decreases in HSP90a in 24 • C could be indicative of further metabolic changes. Additionally, a trend of elevated GPx activity, a family of antioxidant enzymes key in eliminating reactive oxygen species that may form as a result of thermal stress (Do et al., 2019), was observed in the BR population with significantly altered expression pre-trials and 3 days post-CT max . Acute exposure to elevated temperatures has been demonstrated to increase GPx activity in the bald notothen, Pagothenia borchgrevinki (Almroth et al., 2015) and the European bullhead, Cottus gobio (Dorts et al., 2012). Thus, increased GPx activity in the BR population was most likely a result of higher levels of oxidative stress when compared to WR lake sturgeon. Increased mRNA expression of HSP70, HSP90a and enzyme activity 3 days post-CT max in the BR 16 • C treatment, but not at 20 and 24 • C or in the WR population, further demonstrate population-specific responses to thermal stress and the ability of acclimation to decrease potentially resultant cellular consequences.
Gill mRNA expression of Na + /K + ATPase-α1 revealed further differences between populations. Na + /K + ATPase-α1 makes up the functional pumping subunit of the heterodimeric protein complex (Hu et al., 2017;Wong et al., 2016) that actively exchanges Na + and K + ions to maintain cellular ion balance (Ito et al., 2010). At all sampling points, mRNA expression of Na + /K + ATPase-α1 was elevated in BR lake sturgeon, when compared to their WR counterparts. Expression of Na + /K + ATPase-α1 mRNA is altered in hightemperature acclimation treatments in three different species of Pacific salmon (Jeffries et al., 2014;Tomalty et al., 2015) as well as in Atlantic salmon (Vargas-Chacoff et al., 2018). This gene has also been implicated as a driver of population differences across different salinities in the semi-anadromous Sacramento splittail, Pogonichthys macrolepidotus (Mundy et al, 2020). Na + /K + ATPase-α1 is under hormonal control (Feraille and Doucet, 2001;Ito et al., 2010;Therien and Blostein, 2000), and isoforms of it can be upregulated in the presence of cortisol and growth hormone (McCormick et al., 2013). Thus, increases in mRNA expression observed in the current study may have multiple explanations. Na + /K + ATPase-α1 is not well studied in freshwater fish in response to thermal stressors. In contrast to the above salmon literature, in the current study, there was no evidence of an effect of temperature on the mRNA expression of Na + /K + ATPase-α1. These differences in Na + /K + ATPase-α1 mRNA expression may be evidence of further differences between populations, or varying effects of stress and osmotic disruption induced in BR lake sturgeon.

Conclusions
Conservation hatcheries continue to rear and release lake sturgeon to enhance endangered wild populations throughout their distribution. In order to ensure the success of these stocking programmes, and to preserve lake sturgeon throughout their natural range, it is necessary to understand the effects of different environmental temperatures on the survival and physiology of sturgeon from diverse populations. This study has demonstrated significant population-specific physiological effects following 30 days of acclimation to relevant environmental temperatures and those that may be anticipated within the lifetime of sturgeon currently being released (Manitoba Hydro, 2015). Prolonged temperatures above 16 • C may not be appropriate for rearing Manitoba populations of lake sturgeon with decreases in HSI potentially impacting their ability to survive overwintering. Whilst lake sturgeon can acclimate to increased environmental temper-

13
atures, increases in mortality as well as wide-ranging physiological consequences including diminished HSI, metabolic depression and alteration of gene expression are evident as a result of chronic thermal stress, which are likely related to environmental and possibly genetic differences between populations (Harder et al., 2019). The numbers of individuals captured for spawning and subsequent mix of genetic variability in the resultant progeny for this study are restricted due to the endangered status of the species in Manitoba (COSEWIC, 2006;COSEWIC, 2017) and thus limit our interpretation; however, the data presented suggest populationlevel responses to increased acclimation temperatures and are supported by a number of studies conducted on other geographically separated populations of the same species. This study addressed the effects of increasing environmental temperatures on developing lake sturgeon that were fed to satiation; however, natural environments represent a complex set of factors that can interactively affect population and community outcomes (Moe et al., 2012). Thus, future studies should examine the interactive effects of multiple stressors on measured physiological variables to fully understand the consequences on lake sturgeon population health under future warming scenarios.