Thermal sensitivity does not determine acclimation capacity for a tropical reef fish


  • Jennifer M. Donelson,

    Corresponding author
    1. ARC Centre of Excellence for Coral Reef Studies, and School of Marine and Tropical Biology, James Cook University, Townsville, Qld. 4811, Australia
    2. Climate Adaptation Flagship CSIRO, Hobart, Tas. 7001, Australia
    Search for more papers by this author
  • Philip L. Munday

    1. ARC Centre of Excellence for Coral Reef Studies, and School of Marine and Tropical Biology, James Cook University, Townsville, Qld. 4811, Australia
    Search for more papers by this author

Correspondence author. E-mail:


1. Short-term measures of metabolic responses to warmer environments are expected to indicate the sensitivity of species to regional warming. However, given time, species may be able to acclimate to increasing temperature. Thus, it is useful to determine if short-term responses provide a good predictor for long-term acclimation ability.

2. The tropical reef fish Acanthochromis polyacanthus was used to test whether the ability for developmental thermal acclimation of two populations was indicated by their short-term metabolic response to temperature.

3. While both populations exhibited similar short-term responses of resting metabolic rate (RMR) to temperature, fish from the higher-latitude population were able to fully acclimate RMR, while the lower-latitude population could only partially compensate RMR at the warmest temperature. These differences in acclimation ability are most likely due to genetic differences between the populations rather than differences in thermal regimes.

4. This research indicates that acclimation ability may vary greatly between populations and that understanding such variation will be critical for predicting the impacts of warming environmental temperatures. Moreover, the thermal metabolic reaction norm does not appear to be a good predictor of long-term acclimation ability.


Understanding the ability of species to cope with higher temperatures is critical for predicting the likelihood of range shifts (Walther et al. 2002; Perry et al. 2005; Hickling et al. 2006; Parmesan 2006) and extinction rates (Thomas et al. 2004) owing to global warming. There is growing evidence that tropical species may be especially sensitive to global warming because they have narrower thermal niches and are currently living closer to their thermal maximum compared with species from higher latitudes (Deutsch et al. 2008; Tewksbury, Huey & Deutsch 2008; Wright, Muller-Landau & Schipper 2009; Neuheimer et al. 2011). These thermal attributes possibly occur because tropical species have evolved in a relatively stable thermal environment. One pathway by which organisms can cope with adverse environmental conditions is through acclimation, a form of plasticity, which involves the modification of behavioural, physiological or morphological characteristics in relation to changes in their environment (Angilletta 2009). However, production of these phenotypic changes is costly, as is maintaining sensory and response pathways for plasticity (Reylea 2002; Hofmann & Todgham 2010). As tropical regions often experience relatively small temperature fluctuations, it may be that plasticity or acclimation capacity for a large range of temperatures would not be preserved in most tropical species. If temperatures are stable and predictable, there would be fitness benefits in fine tuning physiological performance to the thermal environment (Bradley 1982). For example, benefits can occur through reducing the production of various enzyme forms with differing thermal optimums (Heinrich 1977). This could limit plasticity to future temperature change (Stillman 2003) and cause populations to exhibit very different capacities to cope with global warming (Eliason et al. 2011; Kelly, Sanford & Grosberg 2011).

The temperature range experienced by a population can influence the width of the fitness thermal reaction norm (phenotypic expression of a measure of fitness across a temperature range) and the amount of thermal plasticity that is expressed (Berry & Björkman 1980; Via 1993; Angilletta et al. 2006; Cavieres & Sabat 2008; Angilletta 2009). If optimal performance is expected to occur in the temperature range most frequently experienced by a population, then populations from a warmer environment are predicted to perform better in warm conditions than populations from a cooler environment. Furthermore, populations that experience greater environmental variation are predicted to perform well at a greater range of temperatures than populations that experience less environmental variation (Lynch & Gabriel 1987; Huey & Kingsolver 1989; Gilchrist 1995). This leads to an expectation that populations which experience similar temperatures and similar seasonal variation will possess similar fitness thermal reaction norms. However, the thermal reaction norm a population exhibits will depend not only on its thermal history (variation in thermal environmental over the time selection has occurred), but also on other biological attributes that are under selective pressures and any trade-offs among traits that occur as a result of thermal specialization (optimizing performance to particular temperatures) (Conover & Present 1990; Conover & Schultz 1995).

Physiological performance of ectotherms, including fish, is strongly influenced by environmental temperature (Fry 1967; Hazel & Prosser 1974; Houde 1989). Recent research on aquatic ectotherms has proposed metabolic attributes as the key factors in determining persistence in a warming environment (Pörtner & Knust 2007; Pörtner & Farrell 2008). As water temperature increases so does the minimum energy needed to maintain physiological cell functions (RMR, resting metabolic rate) (Bret 1971). The difference between maximal oxygen uptake (MMR, maximum metabolic rate) and RMR defines the amount of aerobic activity which can be undertaken, and consequently influences the capacity for critical activities such as reproduction and growth (Pörtner & Knust 2007; Pörtner & Farrell 2008). At temperatures greater than an optimum, MMR cannot increase at the same rate as RMR, reducing the scope for aerobic activity (Steinhausen et al. 2008; Farrell 2009). The thermal sensitivity and plasticity of these key metabolic traits is poorly understood in tropical compared to temperate species. However, new research indicates that RMR, MMR and aerobic scope are strongly affected by temperature in coral reef fishes (Nilsson et al. 2009; Gardiner, Munday & Nilsson 2010; Donelson et al. 2011; Johansen & Jones 2011). One recent study has found evidence for developmental thermal acclimation in reef fishes, with individuals exhibiting improved metabolic traits when they are reared at elevated temperatures from shortly after hatching (Donelson et al. 2011).

This study aimed to compare metabolic thermal reaction norms and the capacity for developmental acclimation of these reaction norms to future ocean conditions. Recent studies have suggested that populations of coral reef fish are likely to be severely impacted by rising sea temperature because of a rapid decline in the metabolic thermal reaction norm with increasing temperature (Nilsson et al. 2009; Gardiner, Munday & Nilsson 2010; Nilsson, Östlund-Nilsson & Munday 2010). However, whether these short-term (days to weeks) responses to higher temperatures are a good predictor of the potential for longer-term acclimation of the populations to rising temperature is unknown. We tested the hypothesis that metabolic thermal reaction norms of populations provide a good indicator of acclimation ability and thus are useful for predicting likely impacts of global warming. The model species used in this study was the common coral reef damselfish, Acanthochromis polyacanthus, and fish utilized were from two geographically separated populations which differ in maximum and minimum temperatures, but not the range of seasonal or daily temperature variation. The two populations are also well established as being genetically distinct since the early Pliocene (Planes, Doherty & Bernard 2001) and no mixing is possible as they lack a pelagic larval phase (Robertson 1973). Juveniles from each location were reared until maturity at seasonally cycling temperatures matching averages of the sampling locations, and at average temperatures predicted to occur over the next 50–100 years (+1·5 °C and +3·0 °C; Lough 2007; Munday et al. 2009). Metabolic thermal reaction norms of each population were estimated by testing the RMR and MMR, and calculating net aerobic scope. The potential for thermal acclimation in each population was then estimated by comparing the metabolic performance of fish that developed in elevated temperature regimes (+1·5 °C and +3·0 °C), and had the potential for developmental acclimation, to fish reared in present-day temperatures and tested at elevated temperatures.

Materials and methods

Study species and experimental treatments

The tropical coral reef damselfish A. polyacanthus (Bleeker 1855) is widespread in the Indo Pacific (15°N–26°S and 116°E–169°E). Across its range populations experience a total temperature span (inclusive of seasons) of approximately 20–31 °C. Fish were collected from 4 to 8 m depth of the reef slope at two locations on the Great Barrier Reef (GBR), Heron Island (23°27′S, 151°57′E) and the Palm Islands (18°37′S, 146°30′E). These two populations represent the lower extent (Heron Island) and middle (Palm Island) of the species’ range on the GBR. The Palm Island fish were offspring produced by eight wild pairs collected in July–August 2007 and spawned in December 2007–February 2008 (144 juveniles). The Heron Island fish were collected as recently hatched juveniles in February–March 2009 (mean wet weight: 2·5 g) from 10 pairs (120 juveniles). Gametogenesis and embryogenesis for both locations occurred under average temperature conditions for the collection locations. Fish were divided into the three temperature treatments within 3 months of hatching. The ontogenetic stage of individuals collected from the wild was identified by their size and juvenile colour pattern. Fish were reared in 10–20 replicate tanks (40–60 L) per temperature treatment until maturity for up to 28 months (see below for details) in seasonally cycling temperature treatments matching present-day averages for the collection locations and +1·5 °C and +3·0 °C above these averages. The elevated temperature treatments match sea surface temperatures projected to occur on the GBR by 2050–2100 (Lough 2007; Munday et al. 2009).

Experimental temperatures simulated averages for the 10 year period, 1999–2008, from each location (Australian Institute of Marine Science temperature loggers 6–8 m; Temperatures were modified weekly to replicate seasonal temperature changes and daily variation within the treatments was ±0·4 °C. While fish at Heron Island experience cooler temperatures than Palm Island fish (Heron Island mean: summer = 27·0 °C and winter = 21·8 °C; Palm Island mean: summer = 28·5 °C and winter = 23·2 °C), the two locations experience similar seasonal variation in temperature (difference between winter and summer means: Heron Island = 5·2 °C, Palm Island = 5·3 °C; mean daily variation: Heron Island = 1·29 °C, Palm Island = 1·12 °C).

Heron Island fish grew faster and took only 11–14 months to reach maturity and therefore were tested at the same size, but an earlier age than Palm Island fish (24–28 months old) in the present study. The metabolic attributes (trends between treatments) of fish from Palm Island did not differ between 1 and 2 years old (Donelson et al. 2011 and present study).

Metabolic attributes

Respirometry was conducted at the end of summer as per Nilsson et al. (2009) and Donelson et al. (2011). Both RMR and MMR were measured directly, which allowed the calculation of net aerobic scope (MMR-RMR). Net aerobic scope indicates the amount of energy available for aerobic activity once the costs of basic maintenance are removed. For RMR measurements, each fish was allowed to acclimatize in the respirometer (a 4 L Perspex cylinder with 144 mm inner diameter) for 1 h with a constant water flow. Following acclimatization, the chamber was sealed and oxygen concentrations were monitored with an oxygen electrode (WTW OXI 340i or OXI 3310, Weilheim, Germany) for 30 min. Oxygen concentrations remained above 70% of air saturation during trials. Fish were given at least 3 h rest before measuring MMR. For measurement of MMR, the chamber was placed upright creating a circular swimming area (see Nilsson et al. 2007 for details). Water current was created by a 60 mm magnetic stirring bar inside the sealed chamber and the speed was set to the maximum aerobic swimming speed of the fish. Oxygen level in the water was measured for 5–10 min. For both trials, the respirometer was submerged in a temperature-controlled aquarium to maintain a stable temperature. Subsequently, the wet weight of each fish was measured to the nearest mg. From Palm Island 12 present-day, 8 + 1·5 °C and 7 + 3·0 °C individuals were tested and from Heron Island, 10 present-day, 10 + 1·5 °C and 12 + 3·0 °C. Fish tested from Heron Island were 17·5 g ± 4·3 and from Palm Island were 21·3 g ± 5·3 wet weight (mean ± SD).

All fish were tested at the average summer temperature of their developmental treatment giving the response of each population to elevated water temperature when developmental acclimation was possible. Testing temperatures were Heron Island present-day = 27·0 °C, +1·5 °C = 28·5 °C and +3·0 °C = 30·0 °C, and for the Palm Island present-day = 28·5 °C, +1·5 °C = 30·0 °C and +3·0 °C = 31·5 °C. The testing temperature was always within 0·2 °C of the temperatures in the rearing tanks. Fish that were maintained at the present-day summer average, and thus had no opportunity for long-term acclimation (non-acclimation), were subsequently acclimatized at the two higher temperatures predicted for their location for 7 days and tested to investigate the acute effects of temperature on metabolic attributes.


Differences between metabolic attributes (RMR, MMR and net aerobic scope) of non-acclimation and acclimation fish from Heron and Palm Islands were tested with a factorial anova, with location, acclimation treatment and testing temperature all fixed factors. When significant effects of treatment were found, Fisher’s LSD post hoc tests were used to establish where significant differences existed (P < 0·05). RMR data were log10 transformed to satisfy homogeneity of variance assumptions.


Fish from both locations with no opportunity for acclimation exhibited an increase in RMR with increasing test water temperature (Fig. 1a; Table 1). When fish were tested at the same temperatures, RMR was superior (i.e. lower) in Palm Island fish at 28·5 and 30·0 °C (Fig. 1a). MMR did not vary with increasing temperature, but was generally greater in fish from Heron Island across all temperatures (Fig. 1b; Table 1). Elevated MMR consequently caused no significant differences between populations in net aerobic scope at either temperature (Fig. 1c). To investigate whether both populations responded to increases in temperature relative to their local average, non-acclimation fish were compared at their average present-day summer temperatures, +1·5 and +3·0 °C. Heron Island fish achieved a significantly greater net aerobic scope (P < 0·01) than Palm Island fish at their summer average temperature; however, greater aerobic scope of Heron Island fish was not achieved at either +1·5 or +3·0 °C (Fig. 1c).

Figure 1.

 Mean resting metabolic rate (a), maximum metabolic rate (b), net aerobic scope (c) of non-acclimated fish and mean resting metabolic rate (d), maximum metabolic rate (e), net aerobic scope (f) of acclimated Heron (black circles) and Palm Island (grey diamonds) fish (±SE). *Significant difference between populations at testing temperatures. +Significant differences between non-acclimation and acclimation population means. Palm Island non-acclimated: 28·5 °C n = 12, 30·0 °C n = 5, 31·5 °C n = 5; acclimated: 30·0 °C n = 8, 31·5 °C n = 7. Heron Island non-acclimated: 27·0 °C n = 10, 28·5 °C n = 5, 30·0 °C n = 6; acclimated: 28·5 °C n = 10, 30·0 °C n = 12.

Table 1.   Factorial anova comparison of RMR (log10 transformed), MMR and net aerobic scope of fish depending on location (Heron vs. Palm Island), acclimation and temperature treatment (present-day, +1·5, +3·0 °C)
 RMRMMRNet scope
  1. MMR, maximum metabolic rate; RMR, resting metabolic rate.

  2. Significant P values are indicated in bold.

Location F 1,84  = 4·08 P < 0·05 F 1,82  = 37·97 P < 0·001 F 1,71  = 33·97 P < 0·001
Acclimation F 1,84  = 11·97 P < 0·001 F 1,82 = 0·01 P = 0·911 F 1,71 = 3·54 P = 0·063
Temperature F 2,84  = 31·63 P < 0·001 F 2,82 = 0·68 P = 0·509 F 2,71  = 4·02 P < 0·05
Location*acclimation F 1,84 = 2·90 P = 0·092 F 1,82 = 0·192 P = 0·662 F 1,71 = 0·68 P = 0·412
Location*temperature F 1,84 = 0·06 P = 0·946 F 1,82 = 1·29 P = 0·281 F 2,71 = 0·98 P = 0·382
Acclimation*temperature F 2,84  = 5·02 P < 0·01 F 2,82 = 0·06 P = 0·940 F 2,71 = 1·26 P = 0·287
Location*acclimation*temperature F 2,84 = 3·94 P = 0·051 F 2,82 = 0·06 P = 0·943 F 2,71 = 0·21 P = 0·807

Differences in metabolic attributes between acclimated and non-acclimated fish, and thus evidence of developmental thermal acclimation, were identified in both populations, but to differing extents. The Heron Island population exhibited a greater capacity to modify RMR and consequently net aerobic scope, and therefore exhibited a greater capacity for acclimation (Fig. 1d,f; Table 1). In contrast, Palm Island fish were only capable of reducing RMR at 31·5 °C (Fig. 1d). Performance of acclimated Heron Island fish was equal or better than the performance of acclimated Palm Island fish at 28·5 and 30·0 °C as well as +1·5 and +3·0 °C (Fig. 1d–f; P < 0·05). Specifically, net aerobic scope values at 30·0, +1·5 and +3·0 °C were significantly increased (Fig. 1f). In addition, all RMR measures that were originally significant between populations in non-acclimated fish were no longer significant with acclimation (28·5 and 30·0 °C). The only attribute that exhibited no change with acclimation was MMR, which was not strongly influenced by temperature originally (Fig. 1e).


Acclimation is an important mechanism for organisms to cope with a changing environment. We found differences in the ability of two populations of A. polyacanthus to developmentally modify metabolic attributes to future predicted water temperatures, with full compensation possible in Heron Island fish compared to only partial compensation of RMR at +3·0 and 31·5 °C for Palm Island fish. This distinction in developmental acclimation ability suggests that the higher-latitude Heron Island population will not suffer a decrease in performance as ocean temperatures rise, but the mid-latitude Palm Island population is likely to suffer a reduction in metabolic capacity for critical activities such as growth and reproduction. Of course, such differences in performance could be affected by other mechanisms for coping with environmental change, such as parental effects and adaptation (Mousseau & Fox 1998; Galloway & Etterson 2007). In contrast to differences in acclimation capacity, the two populations (non-acclimated) exhibited similar RMR response to acute temperature increases (i.e. their resting metabolic thermal reaction norms at summer averages, +1·5 and +3·0 °C). Consequently, the population differences in acclimation ability were not evident when comparing the thermal sensitivity in RMR of the two populations using standard methods (7 days maintained at new temperature conditions) (Nilsson et al. 2009; Gardiner, Munday & Nilsson 2010). This highlights that the capacity for long-term acclimation is not necessarily evident in short-term measures of metabolic thermal reaction norms.

Improvement in key metabolic attributes through developmental acclimation will reduce the impact of elevated water temperature to coral reef fish. Reductions in RMR should equate to declines in daily energy expenditure, and is likely to provide fish with additional energy for non-critical functions such as reproduction or growth. This decline in RMR is most likely to have occurred through a reduction in mitochondrial density and therefore baseline energetic demands (including proton leakage) (Pörtner 2002). However, a reduction in mitochondrial density could result in functional limitations, including restrictions to maximum aerobic capacity (Pörtner 2001). In addition, acclimated Heron Island fish exhibited similar net aerobic scope across all testing temperatures, while non-acclimated fish exhibited a decline in net aerobic capacity with increasing temperature. This maintained aerobic capacity could result in sustained abundance, growth rate, swimming capacity and foraging ability of this population in the future (Pörtner & Knust 2007; Johansen & Jones 2011).

Evidence for local thermal adaptation was found at the two locations, with both populations achieving similar RMR when comparing performance at their summer averages, +1·5 and +3·0 °C. Additionally, Palm Island fish obtained lower RMR than Heron Island fish at absolute temperatures of 28·5 and 30·0 °C. For A. polyacanthus, local adaptation might be expected owing to the lack of pelagic larvae, limited dispersal (Robertson 1973) and populations being genetically distinct since the early Pliocene (Planes, Doherty & Bernard 2001). Local adaptation of RMR would be beneficial as it defines the amount of energy required at rest and, under food-limited situations, will influence the amount of spare energy for activities such as growth and reproduction (Bret 1971).

In addition to the evidence for local thermal adaptation, there was also evidence for metabolic performance differences between populations unrelated to their thermal environment. Firstly, Heron Island fish achieved significantly higher MMR across all testing temperatures, as seen by Gardiner, Munday & Nilsson (2010) at this location. In addition, fish from Heron Island had greater net aerobic scope at their summer average than Palm Island fish at theirs, owing to the higher MMR but not a lower RMR. There are multiple possible physiological explanations for the ability of Heron Island fish to achieve greater MMR including a greater cardiac output and consequently blood flow across the gills, larger gill surface area, increased oxygen carrying capacity of blood through haemoglobin or red blood cells, or an increased capacity to remove oxygen from the blood by the tissues (Gardiner, Munday & Nilsson 2010; Perry & Gilmour 2010). Regardless of the mechanism it seems that the genetic divergence between populations is the underlying cause of physiological differences, as there is no difference between locations in seasonal or daily variation of thermal environment, but strong differences in genetics (Planes, Doherty & Bernard 2001).

In general, the ability to cope with variation in environmental temperature is thought to be linked to the thermal range or variation a population experiences (Angilletta et al. 2006; Angilletta 2009). However, the present study found differences in acclimation capacity of A. polyacanthus populations regardless of thermal experience. With developmental acclimation, Heron Island fish were able to reduce RMR down to levels equivalent to Palm Island fish across the thermal range investigated and to outperform Palm Island fish in aerobic scope. As well as being genetically distinct, the two populations investigated are believed to have resulted from separate invasion events in the past 5 million years (Planes, Doherty & Bernard 2001). This, as well as both populations experiencing similar thermal variation, suggests that differences in acclimation capacity may be linked to historical differences rather than recent thermal environment, especially as a genetic basis for variation in thermal reaction norms has been established in a range of taxa (Knies et al. 2006; Driessen, Ellers & Van Straalen 2007; Yamihira et al. 2007). Furthermore, it is possible that acclimation capacity has not yet undergone selection as present temperatures have not exceeded the thermal performance breadth of the species.

Knowledge of acclimation capacity is critical to fully understand the impact of predicted temperature increases on local populations. Our results suggest that the metabolic thermal reaction norm does not necessarily provide a good predictor of the potential for developmental acclimation. It is also apparent that not all populations will respond similarly to temperature increases, even with similar thermal histories. There is the possibility that differences observed in metabolic attributes between the two populations are owing to selection during the first months of life which occurred in the wild Heron Island population and not in the laboratory reared Palm Island population. However, this is unlikely because the metabolic measures obtained for the ‘best’ Palm Island individuals were not similar to the ‘best’ Heron Island individuals. Instead, they were similar to the ‘worst’ Heron Island individuals, and thus Palm Island fish possessed poorer metabolic attributes. It is likely that differences in acclimation ability between reef fish populations because of genetic differentiations could be common as self-recruitment, which over time can lead to genetic distinction, has been identified in a number of species (Jones et al. 1999; Almany et al. 2007). The present findings suggest that understanding the capacity of populations to modify key metabolic attributes with developmental acclimation will be critical to predicting their response to future warmer conditions.


This study was supported by the ARC Centre of Excellence for Coral Reef Studies, the CSIRO Climate Adaptation Flagship, the Australian Coral Reef Society and the GBRMPA Science for Management Awards. Thanks to staff at JCU Research Aquarium Facility for logistical support and to N. Gardiner for collection of Heron Island fish. This project was completed under JCU Ethics A1233, A1415 and A1426.