Ecological and evolutionary consequences of metabolic rate plasticity in response to environmental change

Abstract

Basal or standard metabolic rate reflects the minimum amount of energy required to maintain body processes, while the maximum metabolic rate sets the ceiling for aerobic work. There is typically up to three-fold intraspecific variation in both minimal and maximal rates of metabolism, even after controlling for size, sex and age; these differences are consistent over time within a given context, but both minimal and maximal metabolic rates are plastic and can vary in response to changing environments. Here we explore the causes of intraspecific and phenotypic variation at the organ, tissue and mitochondrial levels. We highlight the growing evidence that individuals differ predictably in the flexibility of their metabolic rates and in the extent to which they can suppress minimal metabolism when food is limiting but increase the capacity for aerobic metabolism when a high work rate is beneficial. It is unclear why this intraspecific variation in metabolic flexibility persists—possibly because of trade-offs with the flexibility of other traits—but it has consequences for the ability of populations to respond to a changing world. It is clear that metabolic rates are targets of selection, but more research is needed on the fitness consequences of rates of metabolism and their plasticity at different life stages, especially in natural conditions.

This article is part of the theme issue ‘The role of plasticity in phenotypic adaptation to rapid environmental change’.

1. Introduction

A fundamental animal function is the metabolic conversion of food into a form of energy—ATP—that is usable by the body's cells; the rate of metabolism thus forms a nexus between environmental resources and animal fitness [1]. An animal's metabolic rate is usually recorded in terms of whole-animal oxygen consumption. This is really a proxy for the underlying process of cellular respiration, in which ATP is generated from nutrient molecules. While it is possible to produce ATP in the absence of oxygen (through glycolysis), this yields less ATP per molecule of the energetic substrate than the alternative of oxidative phosphorylation, and so most multicellular organisms produce the majority of their ATP through a process that consumes oxygen. As a consequence, measurements of oxygen uptake by the body (which are relatively easy to make) can give a relative measure of the animal's overall rate of cellular respiration (which is more difficult to quantify).

The minimum level of sustainable metabolism occurs when an animal is non-reproductive, unstressed, inactive and not digesting food; this is termed the standard metabolic rate (SMR) in ectotherms and the basal metabolic rate (BMR) in endotherms if they are within their thermoneutral zone. The highest rate of aerobic metabolism that can be achieved is termed the maximum metabolic rate (MMR), although endotherms have a second form of maximal metabolism, summit metabolism (M-sum), which is the maximum aerobic rate induced by exposure to cold. Relatively little time may be spent operating at these minimal or maximal extremes, but they nonetheless have biological significance: BMR or SMR defines the minimal ‘cost of living’ that cannot be avoided, while maximum rates are measures of the individual's capacity for work (MMR) or heat generation (M-sum). The difference between the minimal and maximal rates (termed the aerobic scope) defines the maximum amount of oxygen available for activities such as muscular work or digestion.

All species appear to show significant among-individual variation in both minimal and maximal rates of metabolism even under standardized conditions [2–5]. The scale of this variation is at first sight puzzling because metabolic rates have fitness consequences [6] but is likely owing to the optimal metabolic rate being context-dependent [2,7]. It should be noted that while the different forms of metabolic rates are often found to be correlated, especially when comparing among species [8,9], minimal and maximal metabolic rates are best treated as independent traits because they are under different selection pressures that may vary in parallel but can be uncoupled [10–12].

Metabolic rates can appear to be (and are often treated as) consistent and repeatable traits of an individual. However, the repeatability of metabolic rates declines over time [13–15] and is weaker in more variable environments [16]. Moreover, metabolic rates are known to show plasticity in response to environmental conditions [17]. This is a topical point given the increasing rate of environmental change, especially in terms of temperature; temperature-induced changes in metabolic rates (principally of ectotherms) have been suggested to be a key likely cause of population failure and local extinction in a warming world [18,19]. Greater plasticity is likely to increase resilience—a concept that has been tested among species [20] but is lacking empirical evidence among individuals of a species. Linked to plasticity is the phenomenon of acclimation—physiological traits such as metabolic rate may show an acute change in response to an environmental perturbation (e.g. in temperature) but this change can reduce with exposure time [21,22].

In this review, we evaluate the evidence that metabolic rates are plastic traits and discuss the environmental features that drive changes in metabolism. We highlight increasing evidence that both populations and individuals within populations differ in the plasticity of their metabolic rates, and consider the physiological and cellular drivers of both intraspecific variation in metabolism and its plasticity. Through consideration of the costs and benefits of having flexible rates of metabolism we then evaluate how this will influence the capacity of species to cope with and adapt to environmental change, highlighting the gaps in knowledge that prevent a full understanding of this important subject.

2. Evidence for plasticity in metabolic rates

Metabolic rate, along with most physiological traits, exhibits phenotypic plasticity in response to changes in either the animal's internal state or its environment [17,23,24]. Alterations to metabolism can be programmed so as to allow the animal to cope with predictable changes in its energetic state or demands, as when BMR is reduced when animals hibernate or aestivate [25], when M-sum is increased prior to migration in birds [17,26], or when SMR is reduced in intertidal animals that shut down when the tide goes out [27]. Different aspects of metabolism can exhibit separate temporal rhythms, indicating independent controlling mechanisms: while both BMR and M-sum may be elevated over winter in small birds, the increase in M-sum has been found to precede that in BMR and it may last longer into the spring [28]. On top of these programmed changes, metabolism can vary in response to a stochastic change in the environment. As an example, metabolic rates of ectotherms increase after an acute rise in temperature but then usually drop again as the animal becomes acclimated to the new temperature in order to reduce maintenance costs [22]. Not all aspects of metabolism respond to the same extent: it has been proposed that the metabolic floor of ectothermic organisms (i.e. their SMR) is more plastic in response to increasing temperature than is their ceiling (MMR) [29], although this is not always the case [30]. Similarly, the BMR of endotherms generally increases after cold-acclimation and decreases after warm-acclimation [17,31], and has been found to be much more flexible in response to temperature than either MMR [32] or M-sum [33]. Metabolic rate also exhibits plasticity in response to changes in food availability, with food restrictions leading to a reduction in SMR or BMR [34–38], but not necessarily in MMR [37].

Although a great deal of attention has been paid to individual variation in whole-animal (body mass-adjusted) metabolic rate [2,4], few studies have investigated variation in metabolic plasticity, but these reveal significant variation in the extent to which animals can vary their metabolism. Plasticity varies across time within individuals (the BMR of rodents is more responsive to temperature in summer than in winter [31]), but it also varies among individuals and populations. At the population level, variation in metabolic rate plasticity occurs in response to temperature, diet quality and season: rufous-collared sparrows (Zonotrichia capensis) from Mediterranean ecosystems exhibit greater BMR flexibility in response to environmental temperature than do members of the same species from desert ecosystems, which have a largely inflexible BMR [39]. However, the BMR of the desert sparrows shows greater flexibility in response to diet than that of sparrows from the Mediterranean [40], emphasizing how conclusions about metabolic rate flexibility depend on the environmental context. Greater plasticity in resting metabolic rate in response to temperature in high- versus low-latitude populations of cane toads (Rhinella marina) has been proposed to facilitate this species' invasion into higher latitude regions of Australia, owing to an enhanced capacity to maintain critical physiological functions in the colder climate [41]. Similarly, cane toads from colder climates also exhibit greater plasticity in their lower temperature tolerance [42]. Across seasons, different subspecies of the stonechat (Saxicola torquata) exhibit differential plasticity in the annual cycle of their BMR when kept in a common environment with annually varying day length but constant temperature, indicating a genetic difference in programmed seasonal change in metabolic rate [43].

Evidence for population differences in metabolic rate plasticity in response to temperature has also been shown to exist at the cellular level: members of a high-latitude subspecies of the Atlantic killifish (Fundulus heteroclitus) increase their mitochondrial volume density and surface area relatively more than their low-latitude counterparts in response to cold-acclimation [44], and these differences are reflected in high-latitude fish having a higher whole-animal metabolic rate [45]. These findings indicate that variation in plasticity may be important for adaptation to a seasonally more variable environment, and possibly to a climatically more variable future as well (keeping in mind that plasticity in killifish has been shown to differ in response to the cold but not the warm [45]). Plasticity of mitochondrial respiration in response to temperature also differs between clones of Daphnia pulex from temperate and subarctic environments, but without showing a clear latitudinal pattern, although clonal differences in mitochondrial function are again more pronounced when assayed in cold conditions [46].

Among individuals, differential flexibility in metabolic rate among fish is linked to their growth rates: brown trout (Salmo trutta) that either increase or decrease their SMR the most in response to increased or decreased food availability, respectively, grow the fastest relative to their less flexible conspecifics [35]. A similar pattern is seen in a different fish species (qingbo, Spinibarbus sinensis) where individuals with the most flexible SMR in response to changing food levels grow the most when food availability is high [47]. Moreover, in response to food scarcity, individual brown trout with the greatest reduction in SMR lose the least amount of fat in a simulated overwintering scenario [36]. If lowering metabolic rate is an adaptive response to food shortage, then this suggests that resource-poor conditions do not preclude an appropriate plastic response (i.e. the cost of plasticity is not too great, despite energetic limitations), as otherwise suggested (reviewed in [48,49]). Phenotypes with more flexible metabolic rates may therefore be at a competitive advantage in an environmentally unstable future with more fluctuating food supplies, as also suggested by Canale & Henry [50].

Individual animals also differ in their metabolic rate flexibility in response to temperature changes, with potentially important life-history consequences. Siberian hamsters (Phodopus sungorus) showed relatively consistent among-individual differences in the plasticity of their BMR (repeatability of BMR plasticity = 0.31), and those hamsters that maintained their BMR level and showed no sign of plasticity in response to cold exposure spent less time in torpor, compared to individuals that increased their BMR in the cold [51]. The fitness consequences of not entering torpor are hard to assess in a laboratory study in which food was readily available. Similarly, while individual zebra finches (Taeniopygia guttata) that exhibited a larger increase in their resting metabolic rate when exposed to the cold were better able to defend their body temperature [52] —suggesting a lower risk of hypothermia-induced mortality in more flexible phenotypes—this also assumes that there is sufficient food available to cover the increased energetic demand. In fish (barramundi, Lates calcarifer), individuals with low SMR, MMR and aerobic scope exhibited a larger increase in these metabolic rates relative to their high-metabolic rate conspecifics when they were faced with elevated temperatures but, when challenged with hypoxia, the metabolic rates of the same individuals hardly changed [11]. Since both rapidly fluctuating temperatures and hypoxia occur in the barramundi's natural environment in tropical Australia and may worsen in a climatically more extreme future, the low sensitivity of MMR and aerobic scope to hypoxia of low-metabolic rate fish could be advantageous as it allows these individuals to maintain their maximum aerobic capacity. However, reduced hypoxia sensitivity may trade off with a larger increase in the ‘cost of living’ (SMR) at warmer temperatures.

3. Physiological/cellular mechanisms underlying (changes in) metabolic rates

In order to interpret variation in metabolism, we need to understand the underlying processes that are responsible for oxygen consumption rates. Not all tissues or organs within the body have the same energy demand: for instance, cells in the brain, liver and kidney are likely to use more ATP per unit mass than skin or connective tissue [53], as will the intestines when digesting food [54]. A number of studies have therefore attempted (with mixed success) to test whether individual variation in whole-animal oxygen consumption can arise from differences in the relative size or activity of these body components. As an example, Vézina et al. [55] found that variation in both BMR and M-sum of red knots (Calidris canutus) was explained by variation in the residual mass of key organs after correction for body mass. Thus, individuals with relatively large hearts and muscles for their size tended to have higher BMRs, while a high M-sum was associated with proportionally large muscles, heart and stomach. A similar study of eels (Anguilla anguilla) showed that the combined relative mass of the heart, liver, spleen and intestine explained 38% of the among-individual variation in SMR, despite these organs only comprising 1.6% of the total body mass [56]. This both highlights the metabolic demands of these organs but also the potential difficulty in detecting their influence on whole-animal metabolic rate, because they can make up a surprisingly small percentage of the total cells (and hence mitochondria) in the body. If there is little variation among individuals in body composition and in relative organ size (as may be the case in laboratory studies if all animals have been kept in the same standardized benign conditions), then it is unlikely that measurements of organ size will explain variation in their whole-organism metabolic rates (but see [57]). This may help explain why other attempts to explain intraspecific variation in minimal or maximal metabolic rates have found weak or no correlations with relative organ size (e.g. [58,59]).

An alternative approach to determining the drivers of whole-animal metabolic rates is to examine variation in the functioning of key organs and tissues, rather than their size. One relevant measure is ‘metabolic intensity’, conceptually defined as the energy consumption per unit mass of tissue but in practice usually measured indirectly as either mitochondrial density or the activity of key rate-limiting mitochondrial enzymes [55]. Variation in both minimal and maximal metabolic rate among individuals has been found to correlate with differences in cytochrome c oxidase and/or citrate synthase activity in their mitochondria [55,59], although these correlations are not always evident [56]. An alternative approach to quantifying mitochondrial function is to measure oxygen consumption rates of either isolated mitochondria or the mitochondria within samples of permeabilized tissue. While care must be taken in the interpretation of these in vitro measurements, they can nonetheless reveal variation in mitochondrial performance that relates to variation in the metabolism of the animal from which they came. Thus, Salin et al. [60] showed that the SMR of individual brown trout was correlated with variation in the ‘leak’ respiration rate of their liver mitochondria, while MMR correlated with variation in the leak respiration of their muscle mitochondria. This leak respiration occurs when the mitochondria actively pump leaked protons back across the inner membrane in order to re-establish the proton gradient necessary for ATP production. Leak respiration is therefore a measure of the inefficiency of the mitochondria in producing ATP. The tissue-specificity of the correlations of mitochondrial leak respiration with SMR and MMR makes functional sense, because the liver is among the most metabolically active tissues under the conditions in which SMR is measured [53], while the muscles may contribute most to MMR [60]. One of the messages of that study is that a high metabolic rate can indicate inefficiency at producing ATP, although there may be a benefit of producing fewer damaging reactive oxygen species (ROS) [61]. There may also be more general cellular drivers of metabolic rate, such as the lipid composition of cellular membranes: comparisons between endotherms and ectotherms, and among species of endotherms, have found that BMR or SMR (after correction for body mass) increases with the degree of poly-unsaturation of cellular membranes (the ‘membrane pacemaker’ hypothesis of metabolism [62]).

Clearly, there are a number of traits that can apparently covary with metabolic rate, but relationships among them can be complex and they do not always vary in parallel: as an example, among-individual variation in the size of energy-demanding organs does not always correlate with variation in their metabolic intensity [55,63]. As a consequence, while these cross-sectional correlational studies can suggest cellular drivers of metabolic rate, they are rarely conclusive. A more convincing approach is to explore relationships in animals whose metabolic rates have been either up- or downregulated (as a result of artificial selection, experimental manipulations or natural changes in environmental conditions). This approach has shown that the link between membrane lipids and metabolism is unlikely to be causal because artificial selection experiments that caused a significant shift in minimal metabolism also altered membrane lipid composition, but in the opposite direction to that predicted by the membrane pacemaker hypothesis [64]. Causality can also be tested through manipulations of supposed cellular drivers of metabolic rate: while dietary manipulations that alter membrane lipid composition have not resulted in changes in minimal metabolism—again contrary to the predictions of the membrane pacemaker hypothesis [65]—manipulations of mitochondrial leak respiration (through use of uncoupling agents) have led to changes in whole-animal metabolic rate, indicating a causal link between mitochondrial and whole-animal respiration rates [66].

While it has usually not been possible to track within-individual changes in the underlying traits (such as organ size, mitochondrial function, etc.) because measurement often requires the animal to be sacrificed, changes in whole-animal metabolic traits can nonetheless prove informative. For example, although there is often a correlation between an individual's BMR and its M-sum [8], Barceló et al. [12] were able to demonstrate through environmental manipulations that BMR and M-sum are under independent control: while cold exposure led to an increase in both the BMR and the M-sum of white-throated sparrows (Zonotrichia albicollis), a diet shift only altered their BMR and had no effect on their M-sum. Exploration of the body composition of these birds showed that in both experimental manipulations, the increase in BMR was related to increases in the relative size of digestive and excretory organs, whereas the increase in M-sum after cold exposure was presumed to be owing to changes in the metabolic intensity of the muscles (because there was no increase in their size) [12], a response that is thought to be, at least in part, driven by changes in gene expression of several key metabolic pathways [67]. BMR was also found to change faster than either M-sum or MMR in birds exposed to an abrupt shift in ambient temperature, possibly because of differences in the relative rates at which organs can change their size versus their metabolic intensity [68].

Within-individual changes in organ size can happen during ontogeny, with consequences for metabolic rate: there is a shift in endotherms from BMR being driven by the fastest-growing organs early in life (when growth is fastest) to it being more influenced by organs with high metabolic intensity later in development [69]. There are also reversible changes in relative organ size (and hence metabolic rate) when animals are faced with major energetic challenges such as long-distance migrations [70] or infrequent but large meals [71,72]. Components of mitochondrial structure and function can also shift in response to changes in ATP requirement [1,73] and/or resource availability [74–76], with the typical response being an increase in the efficiency of ATP production (measured as ATP produced per unit consumption of oxygen) when conditions are more challenging [74,75]. However, mitochondrial responses can differ between organs (and even between muscle types) of the same individual [74,77], and increases in mitochondrial efficiency can come at a cost of increased rates of ROS production, which may explain why ATP production efficiency is not always maximized [76].

The process of acclimation can to some extent provide a buffer against the adverse effects on physiological processes of environmental change: several weeks' exposure to a higher ambient temperature reduces the thermal sensitivity of a range of physiological processes in ectotherms [22]. Temperature acclimation in whole-animal metabolic rate is matched by acclimation in mitochondrial function [78,79] through alterations to mitochondrial membrane fluidity, and cytochrome c oxidase and/or citrate synthase activity, but this capacity for full acclimation may only be over a limited temperature range that corresponds to expected temperatures within the geographical range of the species or population [44,79,80]. This has implications for the ability of organisms to cope with climate change, because the thermal range over which full acclimation can occur may need to evolve in parallel with rises in ambient temperatures.

4. Costs/benefits of metabolic plasticity in response to environmental change

It is possible to identify clear benefits to plasticity in metabolic rates: it has short-term benefits in terms of energy savings when food is scarce and enhances growth when food is plentiful [35,36,47]. Metabolic plasticity is also likely to increase resilience to climate change [22]. Given these benefits, there must be either costs or limitations that prevent metabolic plasticity from being greater than it is. There has been much discussion and speculation on the limits to plasticity in phenotypic traits in general [48,81] and modelling exercises that explore how costs might influence the persistence of plasticity [82], but as yet there is little clear evidence of how significant the costs might be [48,49].

These putative costs can be divided into two types: those of maintaining the potential to adjust metabolism, and those associated with actually undergoing a change in metabolism. Maintaining the capacity for plasticity might be expensive in terms of the machinery needed to monitor the environment (or the organism's state) and to then adjust/regulate the phenotype accordingly [81]. In the case of metabolic rate, it seems unlikely that the monitoring required for the adjustment of metabolism would be greater than that needed to regulate other aspects of nutritional state, suggesting that this cost may not be significant. But the capacity to alter metabolism may be traded off against other traits important for fitness. This has been found in other contexts where behavioural flexibility traded off with foraging performance [83], and there is circumstantial evidence that selection for greater plasticity in metabolic rates in animals invading colder environments has led to reduced burst locomotor performance, suggesting that such trade-offs may exist [41], although there appears not to be any experimental evidence of this to date. It has also been suggested that plasticity in physiological traits may come at the expense of the ability to express an extreme phenotype [84], but this has yet to be demonstrated in the context of metabolic flexibility.

It is possible that there might be pleiotropic effects that link metabolism and other traits, so constraining or imposing indirect costs on metabolic flexibility. However, while the metabolic rate is clearly the result of many complex interacting factors that link mitochondria, tissues and organs, there is little clear evidence of how they individually or collectively may constrain the capacity of the animal to alter its metabolism. Indeed, the extent to which different physiological traits can vary apparently independently of one another is surprising [85], although links between different metabolic attributes within individuals—such as SMR and MMR—may be masked under benign conditions and only revealed in environments where constraints on trait variation are more pronounced [11]. Nonetheless, there may be limits to flexibility—for example, the extent to which mitochondria can ramp up ATP generation owing to the risk of greatly increased ROS production [1]. The costs of altering metabolic processes to cope with a changed environment must also be balanced against the benefits: maintaining the means for extensive up- or downregulation of metabolic physiology is only likely to be beneficial if the animal is likely to encounter significant fluctuations in energy supply or demand (e.g. if adopting a lifestyle of very irregular but large meals [86]). The cost of these metabolic adjustments may be reduced where they are programmed to follow either seasonal or life-history changes in energy demand [43], and where appetite changes in parallel [87,88].

5. Possible evolutionary responses of metabolism to environmental change

How might we expect rates of metabolism to respond to long-term changes in the environment? A recent review found that metabolic rates had an overall average narrow-sense heritability of 0.19 (reported range: 0–0.72), so have the potential to evolve [6]. Indeed, evolutionary changes in metabolic rates have been demonstrated experimentally through selection experiments that have caused marked changes in BMR in 10–25 generations in rodents [89,90], while the rapid evolution of SMR has been demonstrated in natural populations of Trinidadian guppies (Poecilia reticulata) in response to changes in predation pressure [91] and selection on resting metabolic rate has been suggested to drive the evolution of metabolic rate plasticity in this species [92]. SMR has also been found to be under selection in a wild population of snails (Helix aspersa) where individuals with low to intermediate SMR had higher survival, independent of other performance traits (locomotion speed and dislodgement force) [93]. It is not only resting metabolism that may be under selection: low-food environments were recently found to select for higher maximum metabolic rates in juvenile Atlantic salmon (Salmo salar), presumably because of the positive association between metabolic rate and competitive ability in that species [94].

Given this scope for rapid evolutionary changes in metabolic rates, what responses are likely in a warming world? Since minimum levels of metabolism (SMR) have been found to be more plastic than maximum levels (MMR) in ectotherms and exhibit greater thermal compensation (reduction) in response to chronic warming [29], it seems likely that the gradual elevation of metabolic rate incurred by global climate warming will drive an evolutionary reduction of at least ectothermic SMR. On the other hand, because those individuals with a relatively low SMR are faced with a proportionally greater increase in their metabolism when faced with an acute warming event [11], the evolution of minimum metabolic rates in a warming world may depend on the relative frequency of extreme warming events (heat waves), because these may impose different (and potentially opposite) selection pressures from those produced by gradual changes in average temperatures. Along the same lines, because measurements of narrow-sense heritabilities of active metabolic rates are significantly higher than those of resting metabolism [6], this suggests that upper limits to metabolic rate are less dependent on environmental conditions, which reflects the findings of Sandblom et al. that maximum levels were more fixed than resting in a population of fish faced with warming [29]. Interestingly, measured heritabilities of metabolic rate appear to be higher in endotherms than in ectotherms [6], perhaps because the environmental impacts (particularly from temperature) are greater on ectotherm metabolism than that of endotherms.

An increase in the incidence of extreme environmental conditions may also be predicted to lead to an (evolutionary) increase in the use of torpor, aestivation or hibernation, because these are all means to save energy when conditions deteriorate through a controlled reduction in metabolic costs. This is supported by the observation that the highest proportion of species using torpor is found in regions with the most extreme climatic events [50]. Moreover, increased use of torpor occurs after extreme changes in the landscape, such as fire [95], and species using torpor have a lower risk of extinction from environmental stress [96,97]. Thus, the incidence, frequency and duration of torpor events can all be predicted to increase as a result of climate change.

6. Directions for future research

The costs and limits of phenotypic plasticity are still largely unknown, despite its importance for coping with and evolving in new environments [98,99]. Given that among-individual variation is the raw material on which natural selection can operate, more empirical research is needed to investigate how and why plasticity in metabolic rates varies among members of a population. What are the costs of having a flexible metabolic rate, which must be traded off against its more obvious benefits? One possibility is that this plasticity constrains or covaries with plasticity in other key organismal traits [100]. For instance, while a range of physiological traits (metabolic rate, haematocrit, corticosterone and immune function) can apparently change independently of one another [85], metabolic rate may constrain plasticity in behaviour [101]. Given the importance of behaviour for responding to a changing environment [102,103], investigations are needed of the links between metabolic and behavioural plasticity. Ideally, these should be combined with a more integrative approach by which we can determine the mechanisms responsible for (variation in) plasticity, such as variation in the thermal sensitivity of mitochondria [18,19]. The value of understanding this kind of underlying mechanism is that we then have a much greater ability to predict organismal responses to new environmental conditions, beyond those for which we have empirical data—which could prove invaluable in a changing world. New approaches are being developed that allow appropriate samples to be taken repeatedly from the same animal (e.g. [104]), which can greatly aid in our understanding of within-individual changes in the mechanisms underlying metabolic rate plasticity.

We also need to consider the life stage at which plasticity is occurring. Burggren [105] recently suggested that more emphasis should be placed on the phenotypic plasticity of juvenile or developing organisms in response to the more extreme and stochastic weather events associated with climate change, because plasticity in adult individuals is irrelevant if extreme events such as heat waves would have killed off these individuals before they reached maturity. Age- and size-dependent differences in metabolic responses to warming and ocean acidification have indeed been reported for marine molluscs [106], but the majority of studies consider just a single life stage (which is usually not early-stage juvenile).

Extending our research to field conditions may also prove fruitful for a fuller understanding of how and when plasticity is important for responding to environmental change. The continuing development and miniaturization of accelerometers and heart rate tags allow for continuous and long-term monitoring of metabolic rate proxies within individuals, which can provide important information about physiological performance and plasticity in free-roaming animals [107,108]. Field-based studies also have the benefit of testing animal responses in a context where environmental conditions (such as temperature, humidity, salinity and food supply) fluctuate rather than remain artificially constant (as in most laboratory studies). Studies conducted at constant temperature have proved to be inaccurate at predicting responses to fluctuating conditions [109]; moreover, as well as being more natural, these fluctuations in environmental parameters can be more relevant to animal performance than long-term averages [110] and may have profound effects on the ability of animals to acclimate and evolve in a changing world.

Data accessibility

This article has no data.

Authors' contributions

Both authors planned, researched and wrote the article.

Competing interests

We declare we have no competing interests.

Funding

This paper was written while N.B.M. was funded by ERC Advanced Grant 322784 and T.N. was funded by Individual Postdoctoral Grant DFF-4181-00297 from the Danish Council for Independent Research as well as PECRE Grant 37 from the Marine Alliance for Science and Technology for Scotland.