Acute measures of upper thermal and hypoxia tolerance are not reliable predictors of mortality following environmental challenges in rainbow trout (Oncorhynchus mykiss)

Abstract

Anthropogenic climate change threatens freshwater biodiversity and poses a challenge for fisheries management, as fish will increasingly be exposed to episodes of high temperature and low oxygen (hypoxia). Here, we examine the extent of variation in tolerance of acute exposure to these stressors within and among five strains of rainbow trout (Oncorhynchus mykiss) currently being used or under consideration for use in stocking programmes in British Columbia, Canada. We used incipient lethal oxygen saturation (ILOS) as an index of acute hypoxia tolerance, critical thermal maximum (CT_max) as an index of acute upper thermal tolerance and mortality following these two acute exposure trials to assess the relative resilience of individuals and strains to climate change-relevant stressors. We measured tolerance across two brood years and two life stages (fry and yearling), using a highly replicated design with hundreds of individuals per strain and life stage. There was substantial within-strain variation in CT_max and ILOS, but differences among strains, although statistically significant, were small. In contrast, there were large differences in post-trial mortality among strains, ranging from less than 2% mortality in the most resilient strain to 55% mortality in the least resilient. There was a statistically significant, but weak, correlation between CT_max and ILOS at both life stages for some strains, with thermally tolerant individuals tending to be hypoxia tolerant. These data indicate that alternative metrics of tolerance may result in different conclusions regarding resilience to climate change stressors, which has important implications for stocking and management decisions for fish conservation in a changing climate.

Keywords: CTmax; Climate change; ILOS; fish; hypoxia tolerance; inter-individual variation; upper thermal tolerance.

Figure 1

Variation in hypoxia tolerance (ILOS) and upper thermal tolerance (CT_max) within and among strains in the 2017 brood year (experiments 1 and 2). (a) ILOS for fry, (b) ILOS for yearlings, (c) CT_max for fry, (d) CT_max for yearlings. BW, Blackwater strain (in blue); CL, Carp Lake strain (in red); FV, Fraser Valley Domestic strain (in orange). Black bars indicate mean of each strain. For sample sizes, see Table 1. Differences between strains were not statistically compared for fry because the tolerance of each strain was assessed separately in either one or two trials (Table 1). For yearlings, where multiple trials were performed for each strain, data were analyzed using nested linear mixed effect models followed by Tukey pairwise comparisons (α = 0.05). Significant differences are indicated by dissimilar letters. Data for the PN (yearling) are presented in Supplementary Table S5.

Figure 2

Hypoxia tolerance (ILOS) and upper thermal tolerance (CT_max) for the 2018 brood year (experiment 3). (a) ILOS for fry, (b) ILOS for yearlings, (c) CT_max for fry, (d) CT_max for yearlings. BW, Blackwater strain (in blue); CL, Carp Lake strain (in red); HF, Horsefly strain (in yellow). Black bars indicate mean of each strain. For sample sizes see Table 2. All data were analyzed using linear mixed effects models with Tukey pairwise comparisons (α = 0.05). Significant differences are indicated by dissimilar letters. Data for the PN (fry) are presented in Supplementary Table 8.

Figure 3

Percent mortality in rainbow trout that had experienced a hypoxia tolerance trial followed by an upper thermal tolerance trial (experiments 2 and 3). (a) Percent mortality for fry, (b) Percent mortality for yearling. BW, Blackwater strain (in blue); CL, Carp Lake strain (in red); FV, Fraser Valley strain (in orange); HF, Horsefly strain (in yellow); PN, Pennask Lake strain (in green). Values are mean of percent mortality from the replicate trials for each strain/life stage ± SEM. n = number of trials, which varied depending on the strain and life stage from 3 to 15 (Tables 1 and 2). Differences in mortality within a brood year were analyzed by one-way ANOVA with Tukey-HSD pairwise comparisons (α = 0.05). Significant differences between strains are indicated by dissimilar letters, with q-r for the 2017 brood year and a–c for the 2018 brood year.

Figure 4

Correlation between hypoxia tolerance (ILOS) and upper thermal tolerance (CT_max) for the fry life stage for 2018 brood (experiment 3). (a) BW, Blackwater strain (in blue), (b) CL, Carp Lake Strain (in red), (c) HF, Horsefly strain (in yellow); (d) PN, Pennask Lake strain (in green). Sample sizes differ from Tables 1 and 2 due to PIT tag loss in some individuals and are as follows: BW: n = 435; CL: n = 426; HF: n = 453; PN: n = 253. All correlations were analyzed using Kendall rank correlation (α = 0.05). Note that CT_max was determined 3 weeks following the determination of ILOS.

Figure 5

Correlation between hypoxia tolerance (ILOS) and upper thermal tolerance (CT_max) for multiple strains across two brood years (2017 panels a, c and e; 2018 panels b, d and f) at the yearling life stage (experiment 3). (a,b) BW, Blackwater strain (in blue); (c,d) CL, Carp Lake strain (in red); (e) FV, Fraser Valley strain (in orange); (f) HF, Horsefly strain (in yellow). Sample sizes differ from Tables 1 and 2 due to PIT tag loss in some individuals and are as follows: BW (2017):  n = 499; BW (2018):  n = 100; CL (2017):  n = 400; CL (2018):  n = 100; FV:  n = 491; HF:  n = 99. All correlations were analyzed using Kendall rank correlation (α = 0.05). Note that CTmax was determined 2 weeks following the determination of ILOS in 2017 and 3 weeks following the determination of ILOS in 2018.

Figure 6

Trial order effect on hypoxia tolerance (ILOS) and upper thermal tolerance (CT_max). (a) ILOS for Blackwater (BW) and Carp Lake (CL) strains, (b) CT_max for BW and CL. Black bars indicate mean of each strain. For sample sizes, see Table 3. Trial order effect on ILOS and CT_max was analyzed using linear mixed effects models with Tukey pairwise comparisons (α = 0.05). Significant differences between trial orders are indicated by an asterisk (*). Green for both CT_max and ILOS indicates fish that did not undergo a previous experiment; blue indicates fish that did undergo a previous experiment. Note that data for trials with ILOS then CT_max are the same as those shown in Fig. 2.