Cyclical and stochastic thermal variability affects survival and growth in brook trout

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY-NC-ND 4.0 https://creativecommons.org/licenses/by-nc-nd/4.0/ Cyclical and stochastic thermal variability affects survival and growth in brook trout © 2019 Elsevier Ltd. Accepted version (Final draft) Pisano, Olivia M.; Kuparinen, Anna; Hutchings, Jeffrey A. Pisano, O. M., Kuparinen, A., & Hutchings, J. A. (2019). Cyclical and stochastic thermal variability affects survival and growth in brook trout. Journal of Thermal Biology, 84, 221-227. https://doi.org/10.1016/j.jtherbio.2019.07.012 2019 1 Cyclical and Stochastic Thermal Variability1 Affects Survival and Growth in Brook Trout2 3 Olivia M. Pisano a 4 Anna Kuparinen b 5 Jeffrey A. Hutchings a, c 6 7 a Department of Biology, Dalhousie University, 1355 Oxford Street, Halifax, NS B3H4R2,8 Canada; email: [email protected] 10 bDept Biological and Environmental Science, University of Jyväskylä, PO Box 35, FI-4001411 Jyväskylä, Finland; [email protected] 13 c Institute of Marine Research, Flødevigen Marine Research Station, N-4817 His, Norway;14 [email protected] 16 Corresponding author: Jeffrey A. Hutchings, Department of Biology, Dalhousie University, 135517 Oxford Street, Halifax, NS B3H4R2, Canada; [email protected]; tel: +1 902 494 268718 19 Keywords: temperature; common-garden; fluctuation; constancy; fitness; stochasticity20 2 Abstract21 Directional changes in temperature have well-documented effects on ectotherms, yet few studies22 have explored how increased thermal variability (a concomitant of climate change) might affect23 individual fitness. Using a common-garden experimental protocol, we investigated how24 bidirectional temperature change can affect survival and growth of brook trout (Salvelinus25 fontinalis) and whether the survival and growth responses differ between two populations, using26 four thermal-variability treatments (mean: 100C; range: 7-130 C): (i) constancy; (ii) cyclical27 fluctuations every two days; (iii) low stochasticity (random changes every 2 days); (iv) high28 stochasticity (random changes daily). Recently hatched individuals were monitored under29 thermal variability (6 weeks) and a subsequent one-month period of thermal constancy. We30 found that variability can positively influence survival, relative to thermal constancy, but31 negatively affect growth. The observations reported here can be interpreted within the context of32 Jensen’s Inequality (performance at average conditions is unequal to average performance across33 a range of conditions). Projections of future population viability in the context of climate change34 would be strengthened by increased experimental attention to the fitness consequences of35 stochastic and non-stochastic thermal variability.36 3 1. Introduction37 Temperature affects ectotherm physiology (Angilletta et al., 2004; Pörtner and Farrell,38 2008; Farrell, 2009) with consequences for individual fitness and population viability,39 particularly under forecasted changes in climate (Rieman et al., 2007; Wenger et al., 2011a,b). In40 addition to directional shifts, increased thermal variability is predicted to be a concomitant of41 climate change (Hanson et al., 2012; Wang and Dillon, 2014). However, considerably less42 attention has been directed to how variability in temperature affects fitness-related traits43 independently of changes to the mean (Vasseur et al., 2014; Dowd et al., 2015). This represents44 an important knowledge gap, given that thermal variability can represent a central determinant of45 ectotherm responses to environmental change (Colinet et al., 2015; Sinclair et al., 2016).46 Predictions of how temperature variability might affect individuals and populations47 depends on how variability is quantified (Dowd et al., 2015; Bozinovic et al., 2016; Sinclair et48 al., 2016). Thermal variation can be manifest in various ways, e.g., cyclical vs non-cyclical;49 stochastic vs. non-stochastic; high-amplitude vs. low-amplitude cycles. It can also be manifest at50 various temporal scales (e.g., days, weeks, months, years), and at levels considered to be extreme51 in the context of a species’ or population’s thermal performance curve (Sinclair et al., 2016).52 This can make it challenging to study the effects of bidirectional changes in temperature under53 laboratory conditions in a consistent and readily comparable manner both within and among54 species, which might account for the relative paucity of such studies relative to the amount of55 research on directional thermal change.56 Predicted responses to thermal fluctuations will also depend on the degree to which the57 temperature variations encompass the thermal optimum for the species, or population, under58 study (Morash et al., 2018). Here, the application of Jensen’s inequality (Jensen, 1909) has59 4 proven invaluable in predicting and interpreting changes in metrics of individual ‘performance’60 (e.g., metabolic rate, growth rate) resulting from fluctuating changes in temperature (Ruel and61 Ayres, 1999; Denny, 2017).62 Experimental work on thermal variability has largely focused on invertebrates (e.g.,63 Kingsolver et al., 2009; Williams et al., 2012; Colinet et al., 2015). Among vertebrates, there has64 been some work on reptiles (Du and Ji, 2006; Les et al., 2009) and amphibians (Niehaus et al.,65 2012) but comparatively little on fishes (Morash et al., 2018). The effects of thermal variability66 on per capita population growth rate have been modelled for at least one endotherm (black-faced67 spoonbill, Platalea minor; Pickett et al., 2015) and experimentally explored for the green alga68 Tetraselmis tetrahele (Bernhardt et al., 2018).69 Here, we examine the effects of thermal variability on two populations of brook trout70 (Salvelinus fontinalis), a fish widely distributed throughout eastern North America. For guidance71 regarding our laboratory levels of temperature and temperature variability, we examined water72 temperature data for four rivers in close proximity (<100m to 5km) to our study populations to73 ensure that our thermal experimental treatments reflected those likely to be experienced under74 natural conditions. According to Hanson et al. (2012), between the periods of 1951-1980 and75 1981-2010, the standard deviation (σ) of global surface temperatures increased 16% during76 summer (June-August) and 7% during winter (December-February). However, more than 20% of77 the globe experienced an increase of more than 2σ in 2009, 2010, and 2011 (relative to the 1951-78 1980 baseline; Hanson et al., 2012). Given this information, as discussed in more detail in79 sections 2.2 and 2.3, our experimental value of 1.24σ can be interpreted as encompassing an80 empirically defensible increase in thermal variability that trout might be expected to experience81 under climate change.82 5 Our primary objective is to explore how predictably cyclical and stochastic thermal83 variability might affect survival and growth in the early, post-hatching stage of life. Changes in84 water temperature are likely to be particularly important in early development, especially for fish85 such as trout that depend on a yolk sac for nutrition prior to the initiation of exogenous feeding86 (Jensen et al., 2008). Using a common-garden experimental protocol, we address a secondary87 objective of determining whether survival and growth responses to thermal variability are likely88 to differ genetically between populations of the same species.89 90 2. Materials and methods91 2.1. Study populations92 The two study populations of brook trout inhabit Ouananiche Beck (46° 39.0’ N, 53°93 11.0’ W) and Watern Cove River (46° 37.9’ N, 53° 9.5’ W), small rivers on Cape Race,94 Newfoundland, Canada (bounded by 53°16’ W, 46°45’ N, 53°04’ E, and 46°38’ S). This small,95 barren, coastal region is traversed by multiple short (0.27-8.10 km), low-order streams most of96 which contain resident trout populations that are genetically distinct from one another97 (Hutchings, 1993; Belmar-Lucero et al., 2012; Wood et al., 2014). Life-history differences98 among populations are thought to represent adaptive responses to environmentally different99 selective regimes, following habitat fragmentation (Hutchings, 1993, 1996; Wood et al., 2014).100 Phylogeographic work suggests that the populations originated from a common ancestor and101 have been isolated since the Wisconsin deglaciation (Danzmann et al., 1998).102 103 2.2. Temperature104 6 The experimental protocol subjected trout to either a constant (10°C) or variable105 temperature (range: 7° to 13°C; section 2.3), based on an empirically defensible suite of values106 experienced by the two source populations in the wild. The best available temperature data for107 Cape Race brook trout are those measured hourly over a one-year period (October 2009 to108 September 2010) in four separate rivers, using HOBO data loggers (Fig. 1; Table 1): Bristol109 Cove River, Cape Race River, Cripple Cove River, and Whale Cove River. One of our study110 populations (Ouananiche Beck) is a tributary of Bristol Cove River, and the other (Watern Cove111 River) is located 2-5 km from these four rivers. Combining data for all four rivers yields a mean112 of 9.59°C and a σ of 2.41°C for the days between 16 May and 15 June, the approximate time113 frame originally intended for the experiment.114 Based on linear quantile-quantile plots for each dataset, the temperature data are115 distributed normally, meaning that 68.2% of the pooled-temperature values would fall within the116 range of 9.59 + 2.41°C. Put another way, at 1σ of the observed average mid-May to mid-June117 temperatures in 2010, 68.2% of the temperatures experienced by trout would be expected to fall118 between 7.18 and 12.00°C (a range of 4.82°C). For logistical reasons, the actual dates of our119 experiment differed slightly from the planned time period, extending from 26 April to 5 June.120 For these dates, the pooled temperature data for the four Cape Race rivers averaged 8.28°C with121 a σ of 2.40 (Table 1). Under normality, 68.2% of the temperatures in the wild would fall between122 5.88 and 10.68°C, a range of 4.80°C. The temperatures to which the experimental trout were123 exposed ranged between 7 and 13°C. This range (6°C) is 24% greater than that associated with124 1σ for both the mid-May to Mid-June (4.82°C) and late-April to early-June (4.82°C) periods.125 Thus, the range in temperatures in our common-garden experiment can be thought of as126 7 approximating an anomaly of 1.24σ relative to 2010 conditions, an increase that falls well within127 the measurable increase in global surface temperatures documented by Hanson et al. (2012).128 129 2.3. Experimental design130 After one generation in the laboratory, mature adults originating from the two131 populations were reared and spawned at Concordia University, Montreal, in November 2015. For132 each population, 5 males were each crossed with 6 different females, resulting in 30 families per133 population. On 1 February 2016, fertilized eggs and recently hatched individuals were134 transported to the Aquatron Facility at Dalhousie University where they were acclimated to135 laboratory conditions in small, 2.8-litre flow-through aquaria at 5°C. On 26 April, trout were136 subjected to one of four temperature variability treatments: (1) a constant temperature of 10°C;137 (2) a periodic, cyclical fluctuation of 3°C every two days, with temperatures ranging from 7° to138 13°C; (3) a stochastic or random fluctuation of + 3° or 6°C every two days, with temperatures139 ranging from 7° to 13°C, i.e., the ‘low-stochasticity treatment’; and (4) a treatment analogous to140 (iii) but with the stochastic temperature change occurring daily, i.e., the ‘high-stochasticity141 treatment’. The temperatures were achieved by cooling or heating ambient water provided to142 three separate, temperature-controlled header tanks that provided a constant flow (1.5 litres min- 143 1) of water to each of the experiment tank racks. The temperature of the water in each header144 tank was measured daily.145 Fish were randomly selected for each replicate tank from a pool of all available fish in146 each population. There were 5 and 7 replicates for the Watern Cove and Ouananiche Beck147 populations, respectively. Twenty-seven individuals were placed in each replicate tank (all tanks148 were identical) one week before the start of the experiment and subjected to the same149 8 photoperiod, light intensity, water flow, and food (fish were fed daily with an identical mixture150 of live shrimp, Artemia spp., and dry Corey Aquafeeds ® 0.7 mm pellets). The periodic151 treatment followed a cyclical pattern of 7°-10°-13°-10°-7°C. Temperatures associated with the152 stochastic treatments (either 7°, 10° or 13°) were chosen randomly, using a random number153 generator (Fig. 2).154 Three aquarium racks, each supporting sixty 2.8-litre, flow-through tanks, were155 established at one of the three experimental temperatures. The experimental tanks were separated156 by rack, or temperature, and randomly allocated to a location within the rack. Temperature157 changes (i.e., reassignment of tank location among racks) occurred every two days for tanks158 associated with the constant, cyclical, and low-stochasticity treatments, and every day for those159 associated with the high-stochasticity treatment. Tanks were randomly allocated to a position on160 a rack each time a temperature change occurred. Any tank not moved to a different rack on a161 given day (i.e., staying at the same temperature) was randomly re-allocated to a different position162 on the same rack. Tanks associated with the high-stochasticity treatment were re-distributed163 within the same rack if they were subjected to the same temperature for more than two164 consecutive days.165 The duration of the experimental period was 41 days, ending on 5 June 2016. The166 following day, all fish were transferred to the 7°C rack and left undisturbed for 5 weeks. Water167 temperatures were measured daily for each of the experimental racks to compare the nominal168 (intended) temperatures with the actual (measured) temperatures. For logistical reasons, actual169 temperature data were available daily for all three racks from days 11 through 41.170 On 13 July 2016, a post-experimental monitoring period (31 days) was initiated to171 examine whether differences in survival and(or) growth between treatments and(or) populations172 15 Relative to temperature constancy, slower growth has been reported to be associated with309 thermal fluctuations in several fishes (Cox and Coutant, 1981; Chadwick and McCormick, 2017;310 Morash et al., 2018). However, as documented by previous researchers (e.g., Morash et al., 2018;311 Penney et al., 2018), the influence of thermal fluctuations on developmental, physiological, and312 life-history traits can be conditional on factors such as population origin (Oligny-Hébert et al.,313 2015; present study), the degree to which thermal fluctuations occur near Topt (the optimum314 temperature that maximizes performance; Morash et al., 2018), and other abiotic variables315 (Penney et al., 2018).316 Notwithstanding some intriguing results and potential avenues for future research, we317 caution that the creation of thermally variable conditions can take many forms (e.g., cyclical vs318 non-cyclical; stochastic vs. non-stochastic; high-amplitude vs. low-amplitude cycles) and there319 can be logistical challenges in appropriately creating the intended variability. For example, upon320 examination of the actual temperatures experienced by our experimental fish, it was evident that321 trout in the cyclical treatment experienced more changes in temperature (on 20 of 40 days) than322 those in the low-stochasticity treatment (15 of 40 days; for comparison, those in the high-323 stochasticity treatment experienced a change in temperature on 27 of 40 days). As a result, trout324 exposed to the low-stochasticity treatment may have experienced a greater degree of temperature325 consistency than originally anticipated. One example of a logistical difficulty we faced was the326 challenge in creating the high-stochasticity treatment. In addition to experiencing a higher327 temporal level of thermal stochasticity, the tanks in which these fish resided were shifted every328 day rather than every two days. This more frequent change in tank position was necessitated by329 logistical constraints imposed by the temperature-control system in the laboratory. As a330 consequence, we are unable to conclude whether the differences between the trout in the high-331 16 stochasticity treatment and those in the other treatments were associated with differences in332 thermal stochasticity, frequency of tank relocation, or both (although every effort was made to333 shift the tanks as carefully as possible and with minimal movement of water within the tanks).334 Lastly, there can also be non-trivial challenges in replicating levels of thermal variability in the335 laboratory that are empirically defensible under natural conditions (although we have strived to336 do so; cf. Fig. 1).337 338 5. Conclusions339 The present study represents one of few that has explored the effects of thermal340 variability on metrics of fitness in an aquatic vertebrate. Within the context of our experimental341 protocol regarding levels of thermal constancy and variability, our results suggest that: (i)342 temperature variability at some level can positively influence survival relative to thermal343 constancy; (ii) growth rate is negatively affected by temperature variability; and (iii) common-344 garden experiments should incorporate empirically defensible measures of thermal variability as345 the baseline ‘treatment’ ‒ rather than temperature constancy ‒ for examining the effects of346 thermal variability on fitness. Given the challenge in determining the appropriate temporal scale347 at which thermal variability ought to be examined, perhaps an ideal approach would be to348 compare the effects of thermal variability at multiple temporal scales on fitness-related traits in a349 single experiment.350 We conclude that the influence of stochastic and non-stochastic changes in temperature351 on individual fitness are not readily predictable (in part because thermal performance curves are352 not static) and that this field of endeavour warrants considerably more attention than it has353 received to date. Projections of future population viability in the context of climate change would354 17 be strengthened by increased experimental research on the fitness consequences of stochastic and355 non-stochastic thermal variability.356 357 Acknowledgements358 We are grateful to Dylan Fraser for securing the eggs for our use, Jim Eddington for fish-359 husbandry logistics, and Melanie Massey and Bror Jonsson for discussion. We thank four360 anonymous referees for their very helpful comments and criticism. Funding: This work was361 supported by the Natural Sciences and Engineering Research Council of Canada (170146-2013);362 the Academy of Finland; and the European Research Council (COMPLEX-FISH 400820).363 364 References365 Angilletta, M., Steury, T., Sears, M., 2004. Temperature, growth rate, and body size in366 ectotherms: fitting pieces of a life-history puzzle. Integr. Comp. Biol. 44, 498-509.367 Bates, D., Maechler, M., Bolker, B., Walker, S., 2015. Fitting linear mixed-effects models using368 lme4. J. Stat. Softw. 67, 1-48.369 Belmar-Lucero, S., Wood, J.L.A., Scott, S., Harbicht, A.B., Hutchings, J.A., Fraser, D.J., 2012.370 Concurrent habitat and life history influences on effective/census population size ratios in371 stream-dwelling trout. Ecol. Evol. 2, 562-573.372 Bernhardt, J.R., Sunday, J.M., Thompson, P.L., O’Connor, M.I., 2018. Nonlinear averaging of373 thermal experience predicts population growth rates in a thermally variable environment.374 Proc. R. Soc. B. 285, 20181076.375 18 Beauregard, D., Enders, E., Boisclair, D., Kidd, K., 2013. Consequences of circadian fluctuations376 in water temperature on the standard metabolic rate of Atlantic salmon parr (Salmo377 salar). Can. J. Fish. Aquat. Sci. 70, 1072-1081.378 Bozinovic, F., Sabat, P., Rezende, E.L., Canals, M., 2016. Temperature variability and thermal379 performance in ectotherms: acclimation, behaviour, and experimental considerations.380 Evol. Ecol. Res. 17, 111-124.381 Burton, T., Killen, S.S., Armstrong, J.D., Metcalfe, N.B., 2011. What causes intraspecific382 variation in resting metabolic rate and what are its ecological consequences? Proc. R.383 Soc. B 278, 3465-3473.384 Carveth, C.J., Widmer, A.M., Bonar, S.A., Simms, J. R., 2007. An examination of the effects of385 chronic static and fluctuating temperature on the growth and survival of spikedace, Meda386 fulgida, with implications for management. J. Therm. Biol. 32, 102-108.387 Chadwick, J.G., McCormick, S.D., 2017. Upper thermal limits of growth in brook trout and their388 relationship to stress physiology. J. Exp. Biol. 220, 3976-3987.389 Colinet, H., Sinclair, B.J., Vernon, P., Renault, D., 2015. Insects in fluctuating thermal390 environments. Annu. Rev. Entomol. 60, 123-140.391 Cox, D.K., Coutant, C.C., 1981. Growth dynamics of juvenile striped bass as functions of392 temperature and ration. Trans. Am. Fish. Soc. 110, 226-238.393 Danzmann, R.G., Morgan, R.P., Jones, M.W., Bernatchez, L., Ihssen, P.E., 1998. A major sextet394 of mitochondrial DNA phylogenetic assemblages extant in eastern North American brook395 charr (Salvelinus fontinalis): distribution and postglacial dispersal patterns. Can. J. Zool.396 76, 1300-1318.397 Denny, M., 2017. The fallacy of the average: on the ubiquity, utility and continuing novelty of398 19 Jensen’s inequality. J. Exp. Biol. 220, 139-146.399 Dowd, W.W., King, F.A., Denny, M.W., 2015. Thermal variation, thermal extremes and the400 physiological performance of individuals. J. exp. Biol 218, 1956-1967.401 Du, W.-G., Ji, X., 2006. Effects of constant and fluctuating temperatures on egg survival and402 hatchling traits in the northern grass lizard (Takydromus septentrionalis, Lacertidae). J.403 Exp. Zool. 305A, 47-54.404 Ergon, T., Borgan, Ø., Nater, C., Vindenes, Y., 2018. The utility of mortality hazard rates in405 population analyses. Meth. Ecol. Evol. 9, 2046-2056.406 Farrell, A.P., 2009. Environment, Antecedents and climate change: lessons from the study of407 temperature physiology and river migration of salmonids. J. Exp. Biol. 212, 3771-3780.408 Feltz, C. J., Miller, G. E., 1996. An asymptotic test for the equality of coefficients of variation409 from k populations. Stat. Med. 15, 647-658.410 Goel, M.K., Khanna, P., Kishore, J., 2010. Understanding survival analysis: Kaplan-Meier411 estimate. Int. J. Ayurveda Res. 2010 1, 274-278.412 Hanson, J., Sato, M., Ruedy, R., 2012. Perception of climate change, Proc. Nat. Acad. Sci. USA413 109, 14726-14727.414 Hokanson, K.E.F., Kleiner, C.F., Thorslund, T.W., 1977. Effects of constant temperatures and415 diel temperature fluctuations on specific growth and mortality rates and yield of juvenile416 rainbow trout Salmo gairdneri. J. Fish. Res. Board Can. 34, 639-648.417 Hutchings, J.A., 1993. Adaptive life histories effected by age-specific survival and growth rate.418 Ecology 74, 673-684.419 Hutchings, J.A., 1996. Adaptive phenotypic plasticity in brook trout, Salvelinus fontinalis, life420 histories. Ecoscience 3, 25-32.421 20 Jensen, J.L.W.V., 1909. Sur les fonctions convexes et les inégalités entre les valeurs moyenne.422 Acta mathematica 30, 175-193.423 Jensen, L.F., Hansen, M.M., Pertoldi, C., Holdensgaard, G., Mensberg, K.D., Loeschcke, V.,424 2008. Local adaptation in brown trout early life-history traits: implications for climate425 change adaptability. Proc. Roy. Soc. Lond. B 275, 2859-2868.426 Kingsolver, J.G., Ragland, G.J., Diamond, S.E., 2009. Evolution in a constant environment:427 thermal fluctuations and thermal sensitivity of laboratory and field populations of428 Manduca sexta. Evolution 63, 537-541.429 Les, H.L., Paitz, R.T., Bowden, R.M., 2009. Living at extremes: development at the edges of430 viable temperature under constant and fluctuating conditions. Physiol. Biochem. Zool.:431 Ecol. Evol. Approaches 82, 105–112.432 Martin, T.L., Huey, R.B., 2008. Why “suboptimal” is optimal: Jensen’s inequality and ectotherm433 thermal preferences. Am. Nat. 171, E102-E118.434 Marwick, B., Krishnamoorthy, K., 2019. cvequality: Tests for the Equality of Coefficients of435 Variation from Multiple Groups. R software package version 0.1.3. Retrieved from436 https://github.com/benmarwick/cvequality, on 18/02/2019.437 McDermid, J.L., Fischer, F.A., Al-Shamlih, M., Sloan, W.N., Jones, N.E., Wilson, C.C., 2012.438 Variation in acute thermal tolerance within and among hatchery strains of brook trout.439 Trans. Am. Fish. Soc. 141, 1230-1235.440 Morash, A.J., Neufeld, C., MacCormack, T.J., Currie, S., 2018. The importance of incorporating441 natural thermal variation in when evaluating physiological performance in wild species.442 J. Exp. Biol. 221, jeb164673. doi:10.1242/jeb.164673443 Niehaus, A.C., Angilletta, M.J., Sears, M.W., Franklin, C.E., Wilson, R.S., 2012. Predicting the444 21 physiological performance of ectotherms in fluctuating thermal environments. J. Exp.445 Biol. 251, 694-701.446 Oligny-Hébert, H., Senay, C., Enders, E.C., Boisclair, D., 2015. Effects of diel temperature447 fluctuation on the standard metabolic rate of juvenile Atlantic salmon (Salmo salar):448 influence of acclimation temperature and provenience. Can. J. Fish. Aquat. Sci. 72, 1306-449 1315.450 Penney, H.D., Beirão, J., Purchase, C.F., 2018. Phenotypic plasticity during external embryonic451 development is affected more by maternal effects than multiple abiotic factors in brook452 trout. Evol. Ecol. Res. 19, 171-194.453 Pickett, E.J., Thomson, D.L., Li, T.A., Xing, S., 2015. Jensen’s inequality and the impact of454 short-term environmental variability on long-term population growth rates. PLoS ONE455 10, e0136072. https://doi.org/10.1371/journal.pone.0136072.456 Pörtner, H.O., Farrell, A.P., 2008. Physiology and climate change. Science 322, 690-692.457 R Core Team, 2016. R: A language and environment for statistical computing. Vienna (AT): R458 Foundation for Statistical Computing; [accessed 29-09-2018]. https://www.r-project.org.459 Rieman, B.E., Isaak, D.J., Adams, S., Horan, D., Nagel, D., Luce, C., Myers. D., 2007.460 Anticipated climate warming effects on bull trout habitats and populations across the461 Interior Columbia River Basin. Trans. Am. Fish. Soc. 136, 1552-1565.462 Ruel, J.J., Ayres, M.P., 1999. Jensen’s inequality predicts effects of environmental variation.463 Trends Ecol. Evol. 14, 361-366.464 Schneider, C.A., Rasband, W.S., Eliceiri, K.W., 2012. NIH Image to ImageJ: 25 years of image465 analysis. Nat. Methods 9, 671-675.466 Sinclair, B.J., Marshall, K.E., Sewell, M.A., Levesque, D.L., Willett, C.S., Slotsbo, S., Dong, Y.,467 22 Harley, C.D.G., Marshall, D.J., Helmuth, B.S., Huey, R.B., 2016. Can we predict468 ectotherm responses to climate change using thermal performance curves and body469 temperatures? Ecol. Lett. 19, 1372-1385.470 Therneau, T.M., Grambsch, P.M., 2000. Modelling Survival Data: Extending the Cox Model.471 Springer, New York.472 Vasseur, D.A., DeLong, J.P., Gilbert, B., Greig, H.S., Harlet, C.D., McCann, K.S., Savagae, V.,473 Tunney, T.D., O’Connor, M.I., 2014. Increased temperature variation poses a greater risk474 to species that climate warming. Proc. Roy. Soc. B 281, 20132612. doi:475 10.1098/rspb.2013.2612.476 Wang, G., Dillon, M.E., 2014. Recent geographic convergence in diurnal and annual temperature477 cycling patterns flattens global thermal profiles. Nature Climate Change 4, 988-992.478 Wenger, S.J., Isaak, D.J., Dunham, J.B., Fausch, K.D., Luce, C.H., Neville, H.M., Rieman, B.E.,479 Young, M.K., Nagel, D.E., Horan, D.L. et al., 2011a. Role of climate and invasive480 species in structuring trout distributions in the Interior Columbia Basin, USA. Can. J.481 Fish. Aquat. Sci. 68, 988-1008.482 Wenger, S.J., Isaak, D.J., Luce, C.H., Neville, H.M., Fausch, K.D., Dunham, J.D., Dauwalter,483 D.C., Young, M.K., Elsner, M.M., Rieman, B.E. et al., 2011b. Flow regime, temperature,484 and biotic interactions drive differential declines of trout species under climate change.485 Proc. Natl. Acad. Sci. USA 108, 14175-14180.486 Williams, C.M., Marshall, K.E., MacMillan, H.A., Dzurisin, J.D.K., Hellmann, J.J., Sinclair,487 B.J., 2012. Thermal variability increases the impact of autumnal warming and drives488 metabolic depression in an overwintering butterfly. PLoS ONE 7, e34470.489 doi:10.1371/journal.pone.0034470.490 23 Wood, J.L.A., Belmar-Lucero, S., Hutchings, J.A., Fraser, D.J., 2014. Relationship of habitat491 variability to population size in a stream fish. Ecol. Appl. 24, 1085-1100.492 Xu, C.L., Letcher, B.H., Nislow, K.H., 2010. Size-dependent survival of brook trout Salvelinus493 fontinalis in summer: effects of water temperature and stream flow. J. Fish Biol. 76,494 2342-2369.495 496 24 Table 1. Water temperature data (mean; standard deviation, σ), based on hourly measurements497 recorded in 2010, for four rivers on Cape Race, Newfoundland. Proposed experimental dates: 16498 May to 15 June; actual experimental dates: 26 April to 5 June.499 500 Population Temperatures during Temperatures during501 proposed experimental dates actual experimental dates502 503 mean σmean σ504 Bristol Cove River 9.85 2.42 8.30 2.25505 Cape Race River 10.50 2.29 8.67 2.46506 Cripple Cove River 9.51 2.64 8.81 2.68507 Whale Cove River 8.51 2.27 7.34 2.40508 509 Pooled data 9.59 2.41 8.28 2.40510 511 512 513 514