Animal temperature limits and ecological relevance: effects of size, activity and rates of change


*Correspondence author. E-mail:


  • 1Climate change is affecting species distributions and will increasingly do so. However, current understanding of which individuals and species are most likely to survive and why is poor. Knowledge of assemblage or community level effects is limited and the balance of mechanisms that are important over different time-scales is poorly described. Laboratory experiments on marine animals predominantly employ rates of change 10–100 000 times faster than climate induced oceanic warming. To address this failure we investigated differences in individual and species abilities to tolerate warming, and also how rate of warming affected survival.
  • 2This study identifies community level effects of thermal biology by applying a multi-species, multi-trophic level approach to the analysis of temperature limits.
  • 3Within species analyses of 14 species from 6 phyla showed smaller individuals survived to higher temperatures than large animals when temperatures were raised acutely. If this trend continues at slower warming rates, the early loss of larger individuals has marked consequences at the population level as larger individuals form the major reproductive component.
  • 4Between species comparisons showed active species survived to higher temperatures than sessile or low activity groups. Thus active groups (e.g. predators) and juvenile or immature individuals should fare better in rapid warming scenarios. This would be expected to produce short-term ecological imbalances in warming events.
  • 5The rate of warming markedly affected temperature limits in a wide range of Antarctic marine species. Different species survived to temperatures of 8·3–17·6 °C when temperatures were raised by around 1 °C day−1. However they only survived to temperatures between 4·0 °C and 12·3 °C when temperatures were raised by around 1–2 °C week−1, and temperatures of only 1·0–6·0 °C were tolerated for acclimations over periods of months.
  • 6Current models predicting range changes of species in response to climate change are either correlative or mechanistic. Mechanistic models offer the potential to incorporate the ecophysiological adaptation and evolutionary processes which determine future responses and go beyond simple correlative approaches. These models depend on the incorporation of data on species capacities to resist and adapt to change. This study is an important step in the provision of such data from experimental manipulations.


Studies of thermal biology have, in the past, been predominantly single species, or comparisons of a few species, and are mechanistic (e.g. Hardewig et al. 1999; Pörtner et al. 2006; Young et al. 2007). However, ecological processes (e.g. predator/prey interactions and competition) are affected by changes in temperature through differing ecophysiological responses of the species involved. To understand these issues requires a broad multi-species community level understanding of temperature responses and tolerances of species across phyla and trophic guilds. This is a move towards such a community and wider-level temperature biology approach, and is aimed at making macrophysiology scale analyses (Chown et al. 2004; Chown & Gaston 2008) to understand thermal biology and the landscape and ecosystem scale effects of change.

One area of thermal biology that currently elicits large interest is responses to changing environments. Two approaches dominate the investigation of likely species response to change. The first is based on observations centred on identifying animal ranges and comparing these with current and past environmental conditions, especially temperature records, to predict survival capabilities (Walther et al. 2002; Bridle & Vines 2007; Jiguet et al. 2007). The second is based around experiments where animals are held in the laboratory and conditions manipulated (Peck et al. 2004; Pörtner et al. 2006; Peck et al. 2008a). The two approaches produce markedly different predictions. Experimental approaches identify internal mechanisms allowing animals to cope with change, and proximate limits to capacities. They allow a mechanistic understanding of the processes involved when species or populations fail to survive environmental change. However, experiments are perforce over shorter time-scales, the temperature elevations used are much faster than the vast majority of natural environmental change and usually limited in numbers of species studied, providing little community or ecosystem level information. Modern day observational approaches mainly suffer from lack of knowledge of genetic and functional tolerance differences between populations, and adaptation rates. They also suffer because site-specific microclimates often differ significantly from the average environments measured for observed areas (Warren et al. 2001), and because animals can survive short periods outside their calculated range limits and recover when conditions return to those concomitant with long-term persistence. These data are absent from historical observations.

Models predicting responses in organism distributions to changing environments fall into two categories: correlative and based around climate envelopes that relate distribution to physical environment (e.g. Pearson & Dawson 2003) or mechanistic and incorporating ecophysiological factors (e.g. Buckley 2008). There are strengths and problems in both approaches (Angilletta 2009). Correlative models require less information than mechanistic ecophysiological models, and can be constructed using only environmental data from known sites. However, predictions of range shifts to modified climates using this approach often do not match current observed results (Warren et al. 2001; Bridle & Vines, 2007). These problems may occur because correlative models ignore potential evolutionary responses, organismal phenotypic flexibility, acclimation potential and variation in dispersal (Soberón & Peterson 2005). Many organismal capacities and ecological interactions also depend on temperature (e.g. Peck et al. 2004; Poulin 2006). Mechanistic ecophysiological models offer the advantage of going beyond inferring distributional impacts of climate change by including analyses of adaptations to current environments and how these adaptations affect responses to future environments. Crozier & Dwyer (2006) used this approach to incorporate growth, and temperature effects on growth, in models of range changes in a skipper butterfly. Buckley (2008) took this further and produced a dynamic energetic model that coupled animal energetics with population dynamics to predict responses to change that produced markedly different predictions from purely correlative models. To move this approach into community and ecosystem level understanding will require ecophysiological data on population level responses to environmental change for many species, and an understanding of how characters such as feeding rate, growth, capacity to perform work, temperature tolerances etc vary with temperature and between species. The approach here is to start on this path by evaluating temperature responses across ecological guilds and provide early data at the community level comparing ecophysiological responses of species with markedly different feeding and activity levels, and to assess variation within populations. To reach the point of producing community level predictions of response to change incorporating ecophysiological data is a non-trivial task. It does, however, offer the prospect of producing predictions of response to change based on a stronger understanding of the mechanisms dictating success or failure.

There are few avenues of investigation currently being pursued that bridge the gap between the laboratory data produced on organismal capacities and the requirements of mechanistic models. One such is the approach used here, that is to evaluate experimental physiological responses over a range of rates of temperature change. Investigating the effects of different rates of change will allow the identification of where different organismal responses are important in dictating survival. Thus survival of acute change is predicated by resistance mechanisms, whereas survival of slower but long-term change will need acclimation and evolutionary mechanisms to be used.

Analysis of the relationship between upper temperature limit and the rates of temperature change allows an estimation of the longer term survival capacity by extrapolation. This approach allows the average species survival for a community, fauna or ecosystem to be estimated. Applying this to ecologically divergent groups also allows the potential for perturbation of communities to be assessed, for instance if predators on average are more resistant than prey species. We hypothesize that when this approach is taken for a wide range of marine species covering suspension feeders, grazers, scavengers and predators, data will show three major findings: (i) that regardless of phylogenetic differences or trophic guild considerations, smaller individuals within a species will survive to higher temperatures than larger ones; (ii) that more active species survive to higher temperatures in elevated temperature trials; and (iii) that the relationship between rate of change of temperature and upper limit will be curvilinear and will allow estimates of longer term upper survival temperatures to be made. The first hypothesis is based on evidence from the few studies of activity and temperature relationships in marine species (e.g. Peck et al. 2004, 2007). It is consistent with the oxygen limitation hypothesis (Pörtner 2006). The second hypothesis is a consequence of the oxygen limited thermal tolerance hypothesis (Peck et al. 2002; Pörtner 2002a; Pörtner et al. 2006, 2007). The third hypothesis is novel. It is based on combined considerations of the oxygen limitation hypothesis under rapid experimental warming with acclimation and adaptation effects at slower temperature elevations. With rapid warming survival above critical temperatures is limited by tolerance to, and level of, accumulated anaerobic end products (Peck et al. 2002; Pörtner 2002a). Faster warming allows survival to higher temperatures before anaerobic end-product accumulation overcomes resistance capacity. At slow rates of warming (monthly to annual), where upper limits are well-below critical oxygen limits in daily or weekly warming trials, acclimation and adaptation effects become important. The outcome should be a curvilinear relationship between temperature limit and rate of warming in experiments. Survival to higher temperatures in very rapid temperature elevations has also been shown for a few fish species (Mora & Moya 2006) and one insect (Terblanche et al. 2007).


Specimens of all species studied were collected by scuba divers from between 5 and 25 m depth at sites near the British Antarctic Survey Rothera research station (67°34′11″ S, 68°07′88″ W), with the exception of Waldeckia obesa, which were collected using baited traps at 30–50 m depth. After collection animals were held in a throughflow aquarium system at ambient temperature and 12 : 12 h light : dark lighting regime. They were held in these conditions for between 1 and 3 weeks before experimentation to ensure that any animals damaged during collection could be identified and to remove the major effects of extraneous factors such as feeding and capture stress effects on metabolism. The aim was to use animals in a standard metabolic condition to evaluate resilience to elevated temperature.

Temperature control methods and regimes used were similar to those of Peck et al. (2004). Specimens were placed in 75 L internal volume tanks with hollow walls, through which water was pumped from a temperature-controlled unit. The whole system was placed inside a temperature-controlled room. Temperatures in this system could be held at a required set value ±0·1 °C. After transfer to the experimental system animals were allowed 48 h to acclimatize to the new conditions. Thereafter temperatures were raised on average by 1 °C every day. When temperatures were raised they were raised in 0·5 °C steps on the morning and evening of each day. This gave a gradual temperature rise across the day when temperatures were raised. Experiments always started at 0 °C.

Temperature limits were identified using either tactile stimuli (touching or prodding with a seeker), or appropriate behavioural stimuli (e.g. movement of antennae, tube feet or tentacles). When animals were no longer responsive they were deemed to have reached their upper temperature limit. At this point for each individual the temperature was noted, and a measure of size made. For the limpet Nacella concinna, the brachiopod Liothyrella uva, the bivalve molluscs Yoldia eightsi and Laternula elliptica, and the ascidian Cnemidocarpa verrucosa this was the maximum dimension of any linear measurement. For the starfish Odontaster validus and the urchin Sterechinus neumayeri it was maximum diameter and for the brittle star Ophionotus victoriae it was maximum disc diameter. All measures were made using vernier callipers with an accuracy of ±0·1 mm. For the amphipods W. obesa and Paraceradocus gibber, the nemeratean Parborlasia corrugatus, and the anemone Urticinopsis antarcticum the measure of size used was surface dried wet weight. For the gastropod Marseniopsis mollis and the amphipod Cheirimedon femoratus size was measured as dry weight after drying to constant weight in an oven at 60 °C. Weights were measured on a top pan balance with an accuracy of ±0·1 mg.


within species upper temperature limits and individual size

When individuals of 14 species of Antarctic marine invertebrates were warmed acutely there was a wide variation in resilience both between individuals and species (Fig. 1). Thus in the limpet N. concinna, the first individual became unresponsive at 4·8 °C, but the most resilient individuals survived to 13 °C. In the amphipod W. obesa these figures were 12 °C and 18 °C. The widest range was in the bivalve Y. eightsi, where the first individual to become unresponsive occurred at 4 °C, but one individual survived to over 22 °C. In all species, however, small individuals survived to higher temperatures than larger ones. This is evident in distributions in Fig. 1 where, in all species, specimens from the largest 10% of the size distribution failed to survive to the highest temperature. Furthermore in all species except 1 (the amphipod W. obesa) no specimens from the largest 25% of the size distribution survived to the highest temperatures. Thus the upper right edges of these plots show a progressive decrease in maximum size for survival as upper temperature limits are approached. Furthermore, when the size ranges of the species studied are divided into quartiles and means calculated for the upper temperature values for each quartile there are negative relationships between these quartile temperature values and animal size in all species (Table 1).

Figure 1.

Data for 14 species showing upper temperature limits for individuals in relation to their size (diameter, length, height (mm), wet weight (g) or dry weight (g)). Data for all 14 species studied show that smaller individuals have a significantly higher upper temperature limit when either the maximum size of surviving individuals at any temperature is regressed against temperature (in all cases r2 > 0·5, P < 0·05) or if the data are split into quartiles on size and the mean survival temperature for each quartile is compared with upper limit (Table 1)(overall Pearson correlation P < 0·05). Upper temperature limit for each individual was the point where they became unresponsive to external stimuli.

Table 1.  Regression parameters for equations relating mean upper temperature limits for quartiles calculated on size ranges for data shown in Fig. 1. Data were separated into quartiles on the basis of size. Mean upper temperatures were calculated for each size quartile and these were regressed against mean size. The negative slopes show that upper temperature limits decrease with increasing size
Nacella concinna–0·3740·3040·051·510·230
Waldeckia obesa–0·0020·0010·126·110·020
Odontaster validus–2·4503·3640·020·530·470
Ophionotus victoriae–0·1070·7170·010·020·880
Liothyrella uva–1·6540·5290·379·780·006
Sterechinus neumayeri–0·4010·6350·060·400·530
Yoldia eightsi–0·2730·1080·186·440·020
Urticinopsis antarcticum–4·4203·3780·111·710·210
Cnemidocarpa verrucosa–0·1150·2560·010·200·656
Paraceradocus gibber–0·0670·0510·091·700·208
Laternula elliptica–0·0382·1990·269·850·004
Marseniopsis mollis–0·0940·0740·181·630·330
Cheirimedon femoratus–0·2800·2130·201·730·319
Parborlasia corrugatus–0·0260·0030·9671·20·014

activity and upper temperature limits

When all of the species studied here are scored for feeding mode, activity type, level and amount of time spent active each day (Table 2), clear differences are obtained, with active scavenging or predatory species separating well from sedentary suspension feeders, and other guilds lying between the two. The feeding/activity quotients (calculated as the fourth root of the product of the activity scores) are strongly correlated with upper lethal temperatures (Fig. 2, Pearson Correlation Coefficient = 0·734, P = 0·003). Calculating the regression between activity quotient and upper temperature (Mean upper temp (°C) = 6·07 + 2·17 activity quotient, r2 = 0·47, F = 12·6, P = 0·004, n = 14). An analysis of residuals around this line at the phylum level shows no significant effect (Fig. 3, anova, F6,7 = 0·70, P = 0·66).

Table 2.  Scores for feeding mode, movement type, speed and duration during day for species in Fig. 1. The activity quotient is derived as the fourth root of the product of the feeding and activity scores
SpeciesFeeding modeMovement typeMovement speedMovement duration in dayProductFeeding/activity quotientRank
  1. In feeding mode: 1 = passive ciliary; 2 = pumping; 3 = grazing; 4 = capture.

  2. In movement type: 1 = sedentary; 2 = sedentary + muscular activity (e.g. valve closure); 3 = crawling; 4 = burrowing; 5 = walking; 6 = swimming.

  3. In movement speed: 1 = none; 2 = slow; 3 = medium; 4 = fast.

  4. In movement duration: 1 = never; 2 = very rare; 3 = occasional; 4 = sometimes; 5 = often.

Cheirimedon femoratus46454804·681
Waldeckia obesa46443844·432
Paraceradocus gibber46332163·833
Yoldia eightsi34351803·664
Ophionotus victoriae45321203·315
Parborlasia corrugatus4324963·136
Nacella concinna3325903·087
Odontaster validus4323722·918
Sterechinus neumayeri3324722·918
Marseniopsis mollis3323542·7110
Laternula elliptica2423482·6311
Urticinopsis antarcticum4222322·3812
Liothyrella uva122281·6813
Cnemidocarpa verrucosa211121·1914
Figure 2.

Upper temperature limits plotted in relation to activity quotients (calculated for each species studied on the basis of activity type (sedentary, crawling, walking, swimming, Table 2), maximum speed and proportion of time spent conducting activity. Upper temperatures shown are means from data in Fig. 1 for each species. More active species survive to higher temperatures (Pearson correlation coefficient = 0·734, P = 0·003).

Figure 3.

Residuals of fits for data in Fig. 2 (species temperature limit vs. activity quotient) to the overall regression. Data shown (mean ± SE) are calculated at the phylum level. There is no significant phylum effect (anova, F6,7 = 0·70, P = 0·66).


within species upper temperature limits and individual size

Size (and by implication, age and sexual maturity) is not often taken into account in investigations of upper lethal limits (Chown, Gaston & Robinson 2002). In the infaunal bivalve mollusc L. elliptica, however, the loss of ability to rebury in sediment with rising temperature is linked to size and oxygen availability (Peck et al. 2007). Ecologically the loss of large animals before small individuals has consequences for macroecological predictions of responses to climate change, as the impact on the effective reproductive population will be greater than average population limits and recruitment will thus be affected by smaller temperature elevations and to a greater than expected degree (Perrin 1993). One area where the effects of the removal of larger individuals has resulted in a cascade of effects on life-history characters is in the impact of fisheries on exploited populations. Fisheries generally target larger individuals, which favours reproduction at earlier ages and slower growth (Coltman 2008). However, unexpected outcomes have also been identified, with behavioural responses to fishing gear that potentially change recruitment patterns (Uusi-Heikkiläet al. 2008) and the demonstration that, in some species, larvae of older individuals have better survival potential than those of younger ones (Bobko & Berkeley 2004; Birkeland & Dayton 2005). Humans act as a strong evolutionary force that is almost ubiquitous across the globe (Palumbi 2001), and climate change is one aspect of this.

Similar effects from the early removal of larger individuals by a warming environment would be expected to those seen from fishing but, without large effort to characterize them across the ecosystem, they will be unpredictable. However, such data would be of direct relevance and great importance to the analysis, and prediction of responses to climate change.

It should, however, be noted here that all of the analyses on relations between temperature change and size to date have been rapid or very rapid compared to the rate of ecologically important and climate change processes. Of specific importance here is that ‘with slower rates of change’ acclimation and adaptation processes become the main elements of species responses, and these are absent from the currently available data. Following 60 days of acclimation to 3 °C acute temperature trials still showed small individuals surviving to higher temperatures than large ones (Peck, Morley & Clark, unpubl) in 6 of the species of Antarctic marine invertebrate studied here.

Despite the above concerns, it is at least likely that where warming is significant over monthly to annual time-scales then large individuals will be more affected than small ones. There will also be cases where acute warming occurs naturally with a regime shift, change in water mass movement or during or following migration. At these times size effects would be expected to be important.

activity and upper temperature limits

The current paradigm is that in short-term trials upper temperatures for survival in marine species are limited by tissue oxygen supply and aerobic scope (Pörtner 2002a; Peck et al. 2004; Pörtner et al. 2007). Our data support this as more active species have higher aerobic scopes than sessile ones. Across species comparisons show that, in mammals, large species have larger aerobic scopes than small ones (Weibel et al. 2004). Thus large species should do better in our comparisons than small ones. However, in our comparison (Fig. 2) the most active species were small (the amphipods C. femoratus and W. obesa) and three of the four least active species were relatively large (the ascidian C. verrucosa, the anemone U. antarcticum and the bivalve mollusc L. elliptica). A concern may be that this could be a difference caused by cross-phylum comparisons compared to the within phylum analyses of mammals. However, our analysis of residuals showed no significant effect at the phylum level, and a bias of this type is therefore unlikely. Phylogenetic effects on data of this type are strongest at higher taxonomic levels (Chown et al. 2002, 2004), and a lack of effect at the phylum level strongly indicates that the major trends identified here would not be altered by a wider-scale phylogenetically based analysis.

It should be noted here that previous within species analyses have also shown smaller individuals survive to higher temperatures than larger ones in marine species (Peck et al. 2004, 2007). However, as small phyla are often more active than larger ones in marine ectotherms, our data may help to explain why small species have been historically suggested to be less vulnerable to extinction events (Cardillo 2003).

effect of varying rate of warming

The value of short-term experiments in predicting the consequences of climate change on populations and species has been questioned (Barnes & Peck 2008). Short-term data are of value in understanding mechanistic cellular and physiological responses. They may also be of value in quantifying dispersal limits where migrating or drifting individuals cross-thermal thresholds or move to new temperature regimes, phenomena that are common in dispersing organisms, and important for Antarctic marine species colonising island chains away from the main continent. These data are, therefore of great value in identifying limits to dispersal capability. In experiments investigating upper temperature limits, rate of warming is very important. By comparing our data for survival in our acute trials with published data for medium-term warming where temperature is raised around 1–2 °C and animals then allowed to acclimate to the new temperature for several days to a week, and also with data for temperatures that Antarctic marine species have been acclimated to for periods of 1 month or more a relationship between rate of warming and upper temperature limit can be obtained (Fig. 4). The Antarctic species studied survive to temperatures between 8·3 °C and 17·6 °C when temperatures are raised acutely (1 °C day−1 rise), only survive to 4·0–12·3 °C when elevations are weekly and only to 1–6 °C for long-term acclimated treatment. The mean value for acclimated survival of over 1 month is 3·3 °C, where significant numbers of species would be predicted to suffer long-term survival problems. This is 2–3 °C above current summer maximum temperatures in large parts of the southern Ocean. Fitting a regression to the data in Fig 4 after logging both axes (Ln upper temperatures = 2·57 – 0·398 Ln days, r2 = 67·6, F = 59·3, P < 0·001), and extrapolating suggests that average species survival of elevated temperatures for in excess of 1 year would be 1·3 °C. As for the analysis of upper limits and activity, a residuals analysis at the phylum level shows no significant effect (Fig. 5, anova: F6,35 = 0·5, P = 0·81). In some years current summer temperatures already exceed 1·3 °C at Rothera, where this work was done. However, this temperature limit needs very careful interpretation as the 1-year time-scale is well-beyond the current data range. Models do not predict average annual temperatures in Antarctica to rise above 1 °C for many decades and seasonal factors may allow species to survive significantly higher summer temperatures for short periods. Studies that account for seasonal variation in temperature are needed to give a clearer view of the impact of elevated temperature, rather than those based around simple designs with temperature elevation to constant values.

Figure 4.

Mean upper temperature tolerance limits vary exponentially with rate of temperature rise. Data shown are upper temperature limits for species (14 species shown for 1 °C day−1 rise (this study); 7 species for 1–2 °C week−1 rise and 16 for long-term acclimation studies, where species survive for in excess of 1 month. Data from this study and: Somero & DeVries (1967); Peck (1989); Gonzales-Cabrera et al. (1995); Pörtner et al. (1999); Peck et al. (2002); Bailey et al. (2005); Lannig et al. (2005); Lowe & Davison (2005); Seebacher et al. (2005); Brodte et al. (2006); Jin & DeVries (2006); Podrabsky & Somero (2006); Peck et al. (2008). Data showing Ophionotus victoriae can acclimate to +1 °C but not +2 °C, M.S. Clark pers. obs.; data showing Laternula elliptica cannot acclimate to +4 °C but can survive > 1 month at +3 °C, S. Morley, pers. obs. Data showing Cheirimedon femoratus can survive > 1 month at +4 °C, Marseniopsis mollis, Sterechinus neumayeri, Paraceradocus gibber, Yoldia eightsi, and Heterocucumis steineni can survive > 1 month at +3 °C M.S. Clark, S. A. Morley & L.S. Peck pers. obs.

Figure 5.

Plot of residuals of data shown in Fig. 4 after Ln transformation of both variables. Residuals are shown as means ± SE and are calculated at the phylum level. There is no significant effect of phylum (anova: F6,35 = 0·5, P = 0·81).

The long-term limits in Fig. 4 are significantly lower than the medium-term survival values. Where there is data they are also significantly below the temperatures where tissues begin to accumulate anaerobic end-products of metabolism such as succinate (Mark, Bock & Pörtner 2002; Peck et al. 2002; Pörtner et al. 2006). Thus in the bivalve L. elliptica, medium-term temperature limits are around 9 °C and anaerobic end-products accumulate at around 6 °C (Peck et al. 2004; Pörtner et al. 2006). However, its long-term acclimated limit is around 3 °C. The temperatures where anaerobic products begin to accumulate are termed the critical temperatures and have been proposed as the physiological limits for survival (Mark et al. 2002; Pörtner 2002b; Pörtner & Knust 2007). It is possible to explain these data using the oxygen limitation hypothesis where pejus thresholds have been defined as the temperatures where aerobic scope or capacity falls from optimal levels and performance is ‘getting worse’ (Pörtner 2002a, 2006). Although pejus conditions are defined in such a way that the concept of worsening performance must be true when conditions move away from the absolute optimum, this theory could explain the reduction of temperature limits with ex-posure duration seen here. This possibility is also supported by the correlation found between oxygen delivery and environmental temperatures beyond which growth performance and abundance decrease in eelpout (Pörtner & Knust 2007).

Our data can also be explained in a wider ecological/physiological framework. When conditions change rapidly (acute change), resistance mechanisms are important, and oxygen limitation has been demonstrated, and widely accepted, to be the mechanism dictating survival limits. In our study these responses are clear when temperatures are changed daily or weekly. At slower rates of change (monthly to yearly) acclimation and other processes such as rate of utilization of stored reserves become important in dictating survival. In this context the acclimation of many physiological processes is important (e.g. membrane transport, cell homeostatis, neuro-muscular function, locomotory organization, feeding and absorptive processes) and poor acclimation in any could affect physiological and ecological performance resulting in loss of fitness. Here oxygen limitation is one important mechanism, but not the sole mechanism. At climate change relevant rates of warming (annual to decadal or longer) adaptation and ecological mechanisms are major factors dictating survival. Thus changes in food availability, predator/prey interactions and evolution of new characters are of high importance. Physiological limits will still play an important role in dictating ecological balance, but they will be only one of several mechanisms, and oxygen limitation will only be one of several physiological systems under adaptational limitation (Peck 2005; Barnes & Peck 2008). At rates of change predicted by climate models several ecological factors become important that are outside of physiological limitation mechanisms. These include alien invasion, spread of predator ranges, or contraction of predator ranges changing competitor balance and changes in quantity and quality of resource availability such as food, nutrients and available space for settlement and colonization (Walther et al. 2002; Thomas et al. 2004; Barnes & Peck 2008; Chown & Gaston 2008). There is very little information on where physiological limits trade off with adaptational and ecological constraints. Our data, showing acclimation temperatures to be several degrees below critical oxygen limitation temperatures at faster rates of warming suggest that other factors than oxygen limitation have started to become important in monthly time-scale studies. These probably include energy balance or homeostasis mechanisms.

It has been suggested that the ability to acclimate to changing conditions is the most important criterion in dictating which species will survive and which will fail (Stillman 2003). We would argue that many factors are important, including immediate physiological scopes and temperature effects on biotic interactions. Different species and populations will respond in varying ways to change and the crucial factor dictating success will not always be acclimation. Stillman (2003) also argued that tropical species should be more limited in their ability to acclimate than temperate or polar species. Our data indicate Antarctic ectotherms have very poor abilities to acclimate to elevated temperature and are at least as sensitive as tropical marine ectotherms. Peck et al. (2008b) also showed the Antarctic brittle star O. victoriae to be incapable of acclimating to +2 °C, a temperature < 0·5 °C above currently experienced summer maximum temperatures. Furthermore recent data on temperature tolerances for survival and ability to perform activity in congeneric bivalve molluscs from tropical to Antarctic latitudes show both tropical and polar species to be living permanently close to their thermal limits (Morley et al. 2007, unpublished data). Thus tropical species may not be the most limited in their abilities to acclimate.

Outside of considerations of acclimation, tropical species have been shown to be susceptible to change and this is especially true for coral bleaching where short-term tolerances are close to maximum experienced temperatures (Hoegh-Guldberg 1999) and mortality is markedly affected by symbiont type and host–symbiont relations (Sampayo et al. 2008). However, Antarctic marine ectotherms are recognized as among the most stenothermal on Earth (Peck & Conway 2000; Aronson et al. 2007). They are also characterized by slow physiological rates, growth and great age (Peck & Brey 1996; Peck 2002). They often exhibit very slow larval development rates and unusual reproductive strategies (Peck & Robinson 1994, Peck et al. 2006a,b), and are the only species to lack the classic heat-shock response (Clark, Fraser & Peck 2008a,b; Clark et al. 2008c). These factors may also make them less able to cope with or adapt to change than species from other latitudes.

The type of study presented here, evaluating effects of rate of temperature rise and identifying both size and ecotype effects based on activity is a broad scale physiological approach with the potential for comparing large scale patterns of faunas from different regions. This is a macrophysiological approach. It offers one of the few avenues for bridging the gap between ecophysiological experimental and observational environmental envelope approaches to predicting broad scale and ecosystem level effects of climate change. It also provides a route for identifying ecologically relevant species tolerance limits.


The concept for this work was developed through discussions with several people especially David Barnes and Andrew Clarke. Authors thank the dive team at Rothera station for help in collecting animals. Authors also thank the UK National Facility for scientific Diving at Oban for overall diving support. Dr Robbie Wilson gave much constructive criticism of the manuscript during review that has markedly improved the final paper. The work was conducted as part of a NERC IMP.