John S. Terblanche, Centre for Invasion Biology, Stellenbosch University, Private Bag X1, Matieland, Stellenbosch 7602, South Africa. Tel.: +27-21-808-2605; fax: +27-21-808-2405; e-mail: email@example.com
The fundamental equation of the metabolic theory of ecology (MTE) indicates that most of the variation in metabolic rate are a consequence of variation in organismal size and environmental temperature. Although evolution is thought to minimize energy costs of nutrient transport, its effects on metabolic rate via adaptation, acclimatization or acclimation are considered small, and restricted mostly to variation in the scaling constant, b0. This contrasts strongly with many conclusions of evolutionary physiology and life-history theory, making closer examination of the fundamental equation an important task for evolutionary biologists. Here we do so using scorpions as model organisms. First, we investigate the implications for the fundamental equation of metabolic rate variation and its temperature dependence in the scorpion Uroplectes carinatus following laboratory acclimation. During 22 days of acclimation at 25 °C metabolic rates declined significantly (from 127.4 to 78.2 μW; P = 0.0001) whereas mean body mass remained constant (367.9–369.1 mg; P = 0.999). In field-fresh scorpions, metabolic rate–temperature (MRT) relationships varied substantially within and among individuals, and therefore had low repeatability values (τ = 0.02) and no significant among-individual variation (P = 0.181). However, acclimation resulted in a decline in within-individual variation of MRT slopes which subsequently revealed significant differences among individuals (P = 0.0031) and resulted in a fourfold increase in repeatability values (τ = 0.08). These results highlight the fact that MRT relationships can show substantial, directional variation within individuals over time. Using a randomization model we demonstrate that the reduction in metabolic rate with acclimation while body mass remains constant causes a decline both in the value of the mass-scaling exponent and the coefficient of determination. Furthermore, interspecific comparisons of activation energy, E, demonstrated significant variation in scorpions (0.09–1.14 eV), with a mean value of 0.77 eV, significantly higher than the 0.6–0.7 eV predicted by the fundamental equation. Our results add to a growing body of work questioning both the theoretical basis and empirical support for the MTE, and suggest that alternative models of metabolic rate variation incorporating explicit consideration of life history evolution deserve further scrutiny.
Any population exposed to a novel environment is expected, in the longer term, to adapt to that environment, or at the very least, and in the absence of its emigration or extinction, to respond to selection imposed by that environment. Indeed, responses to selection are common both in the laboratory and in the field (e.g. Huey et al., 1991; Gibbs, 1999; Hoekstra et al., 2001; Kingsolver et al., 2001). One unintended consequence of these responses is that organisms held in the laboratory for several generations typically adapt to the laboratory conditions (Harshman & Hoffmann, 2000; Matos et al., 2000; Sgrò & Partridge, 2000, but see also Krebs et al., 2001). Differences between laboratory colonies and field populations have been documented in a wide variety of traits and species. In insects, these include those cited above and cold and heat tolerance in Drosophila (Zatsepina et al., 2001), antennal sensilla chemo- and mechanoreceptors in Hemiptera (Catala et al., 2004), pheromone communication between sexes in the screwworm Cochliomyia hominivorax (Hammack, 1991), and CO2 anaesthesia effects on knockdown and recovery times in cockroaches (Branscome et al., 2005). Such laboratory adaptation can also take the form of a relatively rapid decline in stress resistance. For example, in Drosophila melanogaster, starvation and desiccation resistance declined from LT50 values of 50.1–35.9 h, and 14.3–8.9 h, respectively, over a period of 4 years (Hoffmann et al., 2001). Therefore, it has been suggested that investigations using laboratory selection, which provide a useful and essential complement to comparative studies (Kingsolver & Huey, 1998; Gibbs, 1999; Feder & Mitchell-Olds, 2003), must take due cognisance of laboratory adaptation.
In a similar fashion, holding organisms for substantial periods in the laboratory could give rise to significant, unintended acclimation effects. It is widely appreciated that individual insects can respond rapidly to a given environmental treatment and to its relaxation (e.g. Lee et al., 1987; Hoffmann et al., 2003; Chown & Nicolson, 2004; Rako & Hoffmann, 2006; Terblanche et al., 2006). Such responses form the basis of a large and proliferating physiological field aimed at investigating the nature, time course and mechanistic underpinnings of phenotypic plasticity. What is perhaps less widely acknowledged (although see Spicer & Gaston, 1999) is that unintended acclimation can confound evolutionary investigations of physiological trait variation. One recent demonstration of the significance of this problem is provided by an investigation of the scaling of avian metabolic rate. McKechnie et al. (2006) demonstrated that captive birds have a shallower metabolic rate–body mass relationship than wild birds, because small birds tend to up-regulate basal metabolic rate in captivity, whereas the converse is true in large birds. They attributed this effect to acclimation in captivity. Moreover, they argued that their results do not support the nutrient supply network model which forms the basis of the metabolic theory of ecology (MTE) (Brown et al., 2004).
The fundamental equation of the MTE posits that metabolic rate varies as body mass to the power of three-quarters, as a consequence of fractal-like nutrient supply networks, size invariance of the final branches in the network and the evolutionary minimization of energy required to distribute resources (West et al., 1997). The effects of temperature on biochemical reactions form the second component (termed universal temperature dependence) of the fundamental equation (Gillooly et al., 2001). Thus, the MTE posits that basal (or standard) metabolic rate varies principally as a consequence of organismal body mass and environmental temperature in the following manner (Gillooly et al., 2001):
where Q is the metabolic rate, M is the body mass, b a universal scaling exponent (0.75), b0 a taxon-specific constant, E the mean activation energy of the respiratory complex, k the Boltzmann constant and T the absolute environmental temperature. The proponents of the MTE have argued that with the exception of minimization of transport costs, evolutionary change (including adaptation, acclimatization – a phenotypic change in the natural environment in response to an abiotic factor, and acclimation – a laboratory-induced phenotypic change) accounts for little of the variation in metabolic rate, and that it does so mostly through variation in b0 (e.g. Gillooly et al., 2006). Thus, although metabolic rate may vary for reasons other than those associated with body size and biochemical kinetics (e.g. Brown et al., 2004), these reasons are among ‘the secondary factors required to explain the remaining variation within and between groups’ (Gillooly et al., 2006). Much of what concerns evolutionary physiology (Feder et al., 2000) is therefore apparently relegated to a secondary role in the evolution of metabolic rate.
According to the MTE, variation in metabolic rate can account for most of the variations in population parameters, generation times and the mutation rates of organisms (Ernest et al., 2003; Savage et al., 2004; Gillooly et al., 2005), ultimately explaining variation in speciation rates across the planet and large-scale patterns in biodiversity (Allen et al., 2002, 2006; but see also Thomas et al., 2006). Therefore, much of the life history and other variations seen among organisms is a consequence of evolution only inasmuch as it affects variation in b0 in the fundamental equation of the MTE. This reduced role of evolution in generating variation among organisms runs counter to much of evolutionary biology and life-history theory (e.g. Roff, 2002; West-Eberhard, 2003). Therefore, it is incumbent on evolutionary biology to explore the assumptions and predictions of the fundamental equation of the MTE.
Here we do so using scorpions as model organisms. First, we assess the extent to which acclimation of metabolic rate might influence either E or b. According to the MTE, residual variation about the fundamental equation should typically not be directional (see Chown et al., 2007), and should not affect the parameters of the fundamental equation, except for b0. Therefore, we determine how acclimation affects the repeatability of metabolic rate–temperature (MRT) relationships in the scorpion Uroplectes carinatus (Pocock 1890) (Buthidae), and whether changes in metabolic rate in individuals of this species, associated with laboratory acclimation, affect b. We further test the universal temperature dependence component of the MTE (see Gillooly et al., 2001; Clarke, 2004) by comparing the MRT relationships (which are readily converted to activation energy, E; Cossins & Bowler, 1987) obtained in U. carinatus with those found for other scorpions. The MTE predicts that the mean activation energy for a particular higher taxon should lie between 0.6 and 0.7 eV (Gillooly et al., 2006).
Materials and methods
Study animals, collection sites and laboratory maintenance
Individuals of U. carinatus Pocock (Buthidae) were collected from Musina, Limpopo Province (522 m a.s.l., 23.87°S, 29.45°E) (n = 9) and Sutherland, Western Cape (1450 m a.s.l., 32.20°S, 20.34°E) (n = 14) during the late austral autumn to early summer, placed in individual containers with soil/leaf litter from the local habitat, and shipped immediately in an insulated box to the laboratory. Over the past 30–40 years, Musina has experienced a mean annual rainfall (MAR) of 339 mm, mean annual air temperature (MAT) of 28.6 °C, and mean annual minima (MAmin) and maxima (MAmax) of 15.1 and 30.9 °C respectively. Sutherland is characterized by a MAR of 266 mm, MAT of 29.4 °C, and MAmin and MAmax of 17.3 and 35.5 °C respectively (weather data obtained from South African Weather Service; http://www.weathersa.co.za).
On arrival in the laboratory, scorpions were transferred to individual, nonairtight, labelled containers (20 × 15 × 10 cm3) with clean soil from their natural habitat. Scorpions were given access to water on moist filter paper and held at 25 ± 1.5 °C (mean ± SD) in a climate chamber (Labocon, Pretoria, South Africa; photoperiod: L : D 12 : 12 h) housed within an air-conditioned laboratory. Within 3 days of collection, scorpions were used for respirometry recordings in two experimental blocks of five consecutive days with an acclimation period of 13 days between blocks. During this acclimation period scorpions were left undisturbed, apart from watering and randomization of containers among shelves every 3 days. The total experimental duration was thus 22 days.
Metabolic rates of 21 individuals were recorded as carbon dioxide production using a flow-through respirometry system. A calibrated LI-6262 (LiCor, Lincoln, NE, USA) infra-red gas analyser (IRGA) was connected to an eight-channel multiplexer (Sable Systems International, SSI, Las Vegas, NV, SUA). The multiplexer was housed in a PTC-1 peltier-effect temperature cabinet (SSI). The first seven channels of the multiplexer regulated airflow through respirometry cuvettes containing individual scorpions and the eighth channel was used for baseline CO2 and H2O measurements. All electronic instruments (except for the PTC-1) were controlled by DATACAN V software (SSI) on a desktop computer.
Individual scorpions were weighed (to 0.01 mg; Mettler Toledo AX504, Columbus, OH, USA) and placed in 60-mL respirometry cuvettes for metabolic rate determination. Metabolic rate estimates were obtained using equipment set-up and methods described previously (Terblanche et al., 2005), with the flow rate here set at 100 mL min−1. To ensure the selection of standard (resting) metabolic rates (SMR), an electronic activity detector (AD-1; SSI) was used to monitor a single individual's CO2 patterns. This combined CO2 and activity trace was subsequently compared with the CO2 production patterns from the remaining six channels. Activity was easily detected as a several-fold increase in CO2 production for all individuals and such data were excluded. All data were sampled by the PC using DATACAN V at 1 Hz.
Respirometry recordings were performed at 15, 20, 25, 30 and 35 °C in this sequence, on each day for each individual. Internal cuvette temperature was monitored by means of a 40 SWG Type-T thermocouple, recorded via TC-1000 and Universal Interface (SSI), located inside the first cuvette. Temperature of the PTC-1 was changed manually, monitored on the desktop PC, and recording only commenced if the cuvette temperature was within 1 °C of the target test temperature. Typically, a 20- to 30-min delay occurred while the PTC-1 stabilized at the new test temperature and this allowed some time for the scorpions to adjust to the new temperature before respirometry recording commenced. Inspection of the CO2 traces confirmed that metabolic rate was steady at the new selected temperatures. After the final test temperature's recording had been completed, the scorpions were re-weighed, placed inside their containers, and returned to the climate chamber at 25 °C. Experiments began at similar times of the day and lasted for similar durations (13–14 h) throughout all experiments. The order of individuals and test temperatures was consistent on every day (i.e. not randomized) to enable statistical testing for systematic effects (e.g. time × temperature). Sex was not scored in these scorpions as preliminary experiments showed little effects of gender on metabolic rate once body mass has been accounted for. The entire experimental set-up (including the gas analyser) contributed 0–2% of the variance to the total variance in metabolic rate estimates, as determined by repeated measures of blank cuvettes in the multiplexer system followed by a nested (hierarchical) anova (variance components estimation). In consequence, experimental ‘noise’ was considered to be minimal compared with biological variation.
Data extraction and statistical analyses
DATACAN V software (SSI) was used to extract and analyse SMR data (corrected to standard temperature and pressure). Repeated measures anova and ancova were performed in SAS version 8.0 (Cary, NC, USA) statistical software package whereas all other analyses were performed using Statistica version 7.0 (Statsoft, Tulsa, OK, USA).
The slopes of the MRT regression relationships were used to analyse potential variation in sensitivity to temperature. Log10 conversions of metabolic rate data (in mL CO2 h−1) and body mass (in g) (for analyses of covariance with mass as covariate) were performed prior to the MRT analyses. Residuals were examined for normality using a Shapiro–Wilks test and in most cases log10 transformation corrected, or at least improved the fit to a normal distribution. Mean experimental body mass and mean recorded test temperatures were treated as continuous variables in all analyses. Individuals in which temperature did not elicit an increase in metabolic rate were excluded from the analyses (this occurred in two individuals which had recently moulted). In two cases, individuals from the Musina population died of unknown causes during the acclimation period and were replaced by individuals which had experienced the same conditions (shipping, acclimation and handling) but had not been measured in the field-fresh condition because of equipment constraints. An individual which had begun moulting on the final day of experiments (day 22) was also excluded from respirometry recordings.
After testing for effects of different source populations (see Results), metabolic rate variation within and between days was investigated using repeated measures analyses of variance (anova) and covariance (ancova). The full data set containing metabolic rate at each test temperature and each individual's body mass was used in a repeated measures model using an unstructured covariance matrix in proc mixed with a reduced maximum-likelihood estimation method. These preliminary analyses showed no time × temperature, no body mass × time × temperature, and no individual × time × temperature interaction effects on metabolic rate (repeated measures analyses, P > 0.23). Subsequently, a reduced data set was created containing the MRT relationship slope values and body mass for each individual on each experimental day. This data set was used to test for systematic effects of time and body mass on the MRT relationship slope values by repeated measures anova and ancova. These analyses found no effect of time and no effect of body mass on the slope of the MRT relationships (see Results).
To determine whether changes in the proportion of intra- to interindividual variance of the MRT slope occur with acclimation to constant conditions, repeatability (τ) estimates were made. Repeatability was calculated for the first experimental block of 5 days (field fresh) and compared with the second experimental block of 5 days (acclimated), thus representing repeatability over two short periods (i.e. days 1–5 and days 18–22 in this case). Repeatability and its confidence limits were calculated using the intraclass correlation coefficient approach for MRT slopes from a one-way anova (Lessels & Boag, 1987; Falconer & Mackay, 1996; Krebs, 1999).
To explore the variation induced by different laboratory acclimation states on metabolic rate–body mass scaling relationships a randomization model was derived using a nonlinear equation describing the decline in metabolic rate with acclimation of the form:
where M is the metabolic rate, x is the decay variable based on change in metabolic rate with acclimation, and t is the time in the laboratory. In the first case, this was undertaken for n = 50 individuals, using 1000 random numbers re-sampled with replacement using Microsoft Excel that were constrained to the same acclimation state (i.e. only field collected), assuming metabolic rate = body mass0.82 (a value obtained for arthropod mass-scaling relationships in several studies; see Addo-Bediako et al., 2002; Chown et al., 2007). In the second case, using the same number of individuals and randomizations, each individual was in a random state of acclimation (values generated on a continuous scale). Using these random acclimation times metabolic rates were derived using the experimentally determined exponential decay (in U. carinatus during acclimation y = MR e−0.025t, where t = randomly generated time estimate in the laboratory). These models were used to illustrate the potential confounding effects of an acclimation-induced decline in metabolic rate on mass-scaling exponents and the coefficient of determination (r2).
Interspecific variation in MRT relationships was examined to test the assumption of the MTE that activation energy, E, should lie in the 0.6- to 0.7-eV range. These data were gathered from studies published in the Anglophone literature with particular emphasis over the last 50 years. Metabolic rates were converted to log10μL O2 g−1 h−1 following Lighton (1991) where necessary to enable comparisons among species. In some species the original body mass data were not available, and it was therefore not possible to use mass0.82 for interspecific comparisons. Multiple regressions (ordinary least-squares, Type III) were used to determine slopes of log-transformed mass-specific MRT relationships within species and these were compared using an ancova and Homogeneity of Slopes tests. Data are presented as means ± standard error of the mean (SE) unless otherwise stated and significance was set at P = 0.05.
Body mass and metabolic rate variation
Following the quantitative assessment method outlined in Marais et al. (2005), U. carinatus did not show cyclic or discontinuous gas exchange at rest (Fig. 1), and although the effects of activity were readily discernable in the recorded ventilation traces, the animals generally remained inactive inside respirometry cuvettes. No differences in body mass, metabolic rate at a fixed temperature, slopes of MRT relationships and mass-independent metabolic rates (P > 0.1 in all cases) were found among the two source populations. Therefore, for all further analyses the two source populations were pooled. Among all individuals, VCO2 declined (F9,183 = 2.24; P = 0.021) whereas body mass remained constant over time (F9,187 = 0.01; P = 0.999) during acclimation at 25 °C (Fig. 2). Consequently, mass-corrected VCO2 decreased during acclimation (ancova, F10,381 = 54.09; P = 0.0001).
Variation in MRT relationships
Within individuals, regressions of metabolic rate against temperature were significant, except in the case of recently moulted individuals which were excluded from analyses. Therefore, among all individuals temperature caused a significant increase in metabolic rate (F19,948 = 58.72; P = 0.0001). Body mass was not related to the slope of the MRT relationship (F9,156 = 0.05; P = 0.999), i.e. larger individuals were not more sensitive to temperature, nor vice versa. Among all individuals, slopes of the MRT relationship, which can readily be expressed as eV values (see Cossins & Bowler, 1987; Gillooly et al., 2001), did not change systematically over the acclimation period (F9,156 = 0.39; P = 0.939; Fig. 3).
Generally, the slopes of MRT relationships varied considerably, both within and among individuals in either the field-fresh or acclimated state (Fig. 3). Within field-fresh individuals the repeatability (intraclass correlation coefficient, τ) of the slopes obtained for each individual from the log10 MRT relationships was not significant (F19,78 = 1.345; P = 0.181). By contrast, in acclimated scorpions the within-individual variation in the slopes of MRT relationships was lower and the variation between individuals was therefore significant (F19,78 = 2.449; P = 0.0031). Consequently, acclimation resulted in a fourfold increase in τ from 0.02 (95% CL 0.00–0.09) to 0.08 (95% CL 0.05–0.25). However, the difference between the repeatability values was nonsignificant (P > 0.05).
The randomization models illustrated the potential effects of different acclimation states on mass-scaling relationships (Fig. 4). In particular, a scaling relationship derived from individuals or species which vary in acclimation states (by allowing t to range from field fresh to completely acclimated) resulted in a significant decline in the mass-scaling exponent (t1997 = 61.4; P = 0.0001) and an increase in the variation thereof in contrast to t = 0 (Fig. 4a). Unsurprisingly, this was accompanied by a significantly poorer fit of the linear relationships (t1997 = 86.6; P = 0.0001), i.e. a decline in the coefficient of determination (Fig. 4b).
Among species MRT variation
As expected, temperature had a significant and positive effect on metabolic rate in all scorpion species for which data were extracted from the literature, except one species which was marginal (Table 1). Metabolic rate differed among species at a common temperature and the slopes of MRT relationships also varied significantly among them (Table 2). Mean activation energy, E, expressed in eV, was 0.77 ± 0.11 (95% CI 0.72–0.82) which is significantly higher (P < 0.05) than both 0.6 and 0.7.
Table 1. Scorpion metabolic rate–temperature relationships expressed as the slope of metabolic rate (in log10μL O2 g−1 h−1) against test temperature (in K) calculated for data extracted from the literature.
Slope ± SE
Intercept ± SE
Similar superscript letters indicates homogeneity of slopes by an ancova (see Table 2 for statistics).
*Data taken from the means of acclimated individuals on day 22.
†P. villosus from Robertson et al. (1982) was excluded from this regression relationship to avoid replication of a species.
Table 2. Outcomes of an analysis of covariance (ancova) comparing metabolic rate (A) at the mean covariate (temperature) and (B) homogeneity of slopes of metabolic rate–temperature relationships among 11 scorpion species.
SS, sums of squares; MS, mean squares.
Species × temperature
In U. carinatus, metabolic rates declined significantly with acclimation to constant temperature and fasting conditions in the laboratory, although body mass remained constant during this period. Metabolic rates in this study were at a peak of 127.4 μW (converted to μW following Lighton, 1991) in field fresh individuals on day 1 of the experiment and declined to 78.2 μW by day 22. These values were higher and lower, respectively, than the predicted metabolic rate of 95.6–95.8 μW for these animals at their measured body mass (range day 1–day 22: 367.96–369.11 mg) based on Lighton et al.’s (2001, Eq. 2, p. 610) scaling relationship derived for scorpions. Lighton et al. (2001) acclimated animals in their study for 2–6 weeks (J.R.B. Lighton, pers. comm.) at 22–25 °C, which might explain the similarity between the values for acclimated individuals in this study and their estimates of metabolic rate.
The decline in metabolic rate with acclimation while body mass remained constant is similar to that observed in Damon annulatipes whip-spiders over a 2-week period at a similar acclimation temperature (Terblanche et al., 2004). However, in the present study metabolic rate declined by ∼39%, which is greater than the 16–33% reduction observed in the whip-spiders. Generally, a decline in metabolic rate in the laboratory may be a consequence of a reduction in stress (McRae, 2001), a decline in temperature variation (Harshman & Hoffmann, 2000; Sgrò & Partridge, 2000, 2001; Hoffmann et al., 2001), or a reduction in activity and feeding (Terblanche et al., 2004). Further work is required to disentangle these effects and their relative importance.
Metabolic rate–temperature relationships were highly variable both within and between individuals shortly after capture, evidenced by low and nonsignificant repeatability values observed at this time. By contrast, within-individual variation of the MRT relationships declined with acclimation to constant conditions, indicated by the increase in repeatability over this period. This is likely to be true of other scorpion species too. In a study of scorpion population energetics using acclimated individuals, Lighton et al. (2001) stated that ‘the measured Q10 varied significantly among individuals[…]and each individual showed a consistent and tightly determined temperature response…’. Why acclimation might have this effect is not clear. Rate–temperature relationships represent the net effect of temperature on a complex suite of underlying events (Hochachka & Somero, 2002). However, down-regulation of one or a few biochemical process (at either the cellular or tissue level) may result in a decline in metabolic rate (see discussion in Hawkins, 1995). In turn, this could result in more systematic, and therefore repeatable, MRT relationships at the whole-animal level.
What is clear is that the decline in metabolic rate and change in repeatability of the MRT relationships with acclimation are not strictly in keeping with the fundamental equation of the MTE. Specifically, although the proponents of the MTE acknowledge that some variation around the fundamental equation should be found, caused by adaptation, acclimatization or acclimation (Gillooly et al., 2006), they do not suggest that this variation should be directional, nor do they indicate that it should affect anything other than the b0 term of the equation. In this study, a directional change in the extent of within- and among-individual variability of the MRT relationship took place. More significantly, the randomization model demonstrated that acclimation has a considerable effect on b of the mass-scaling relationships (Fig. 4). In particular, a scaling relationship derived from individuals or species which differ in acclimation state results in a significant decline in the mass-scaling exponent and an increase in its variation. The same kinds of effects were documented by McKechnie et al. (2006). Together, these studies, of two very different groups of organisms, suggest that acclimation is not a ‘secondary factor’ required to explain the variation remaining from the fundamental equation (Gillooly et al., 2006), but has more profound effects on metabolic rate variation.
Interspecific variation in MRT relationships
Whole organismal MRT relationships measured over a wide range of temperatures are invariably positive whether assessed at the intraspecific level (reviewed in Cossins & Bowler, 1987; Chown & Nicolson, 2004) or whether investigated across species from many environments as has been recently done for several groups (see Gillooly et al., 2001; Clarke & Fraser, 2004). The effect of temperature on metabolic rate has been documented for at least 11 scorpion species in the Anglophone literature confirming, unsurprisingly, a positive relationship at the interspecific level for this group. More significantly, the variation amongst species in the slope of this relationship provides evidence against Lighton et al.’s (2001) suggestion that there is little interspecific variation in scorpion MRT relationships. Several reasons exist for the differences in our findings. First, it is not clear from the literature the extent to which activity might have influenced the MRT relationships. At higher temperatures, activity is more likely, and, therefore, if activity is not carefully controlled for, MRT relationships will appear steeper than they really are (see discussions in Chown et al., 2003; Hodkinson, 2003). This effect may be exacerbated if the typical environmental temperatures for the species differ substantially because discomfort and an increase in activity are likely to take place at different temperatures. Thus, interspecific differences might be entirely artefactual. Second, the differences may reflect reality. In this case, differences in environmental conditions might account for variation in the MRT relationship, as has been found for insects (Addo-Bediako et al., 2002) and fish (Clarke, 2006, but see also Gillooly et al., 2006). Quite how such differences might arise, and whether they are adaptive cannot be determined without further investigation. Nonetheless, their existence at the whole organism level is unlikely to be explained simply (Clarke, 2006).
One major implication of this among-species variation, should it turn out not to be an artefact of experimental conditions, is that it provides further support for substantial variation in the temperature component of the fundamental equation of the MTE (Gillooly et al., 2001). Although the proponents of the MTE have acknowledged that some variation should be expected around both components of the fundamental equation, they have never made clear how much variation should be expected. Gillooly et al. (2001) predict that E should have a range similar to that of measured activation energies for metabolic reactions of 0.2–1.2 with a mean value of 0.6. Their empirical data on E from whole organismal metabolic rates varies from 0.41 to 0.74, with a mean of 0.62 eV. They take this as support for their predictions. Later, they argue that their theory for the scaling of metabolic rate ‘predicts that E takes on a limited range of values, 0.6–0.7 eV, with an average of about 0.65…’ (Gillooly et al., 2006, p. 400). Scorpions are characterized by considerable variation in E (0.09–1.14 eV intraspecifically in U. carinatus; 0.20–1.00 interspecifically), with a mean value of 0.77 eV, significantly higher than both 0.6 and 0.7.
By contrast with the MTE, Clarke (2004) (see also Clarke & Fraser, 2004) specifically predicted substantial among-species variation in MRT intraspecific relationships and a difference between these relationships and the broader interspecific MRT relationship found when many species from a range of environments are included in a single analysis. The present data do not directly address the latter idea, but certainly draw attention to the need to investigate Clarke's (2004) hypothesis specifically, and, more generally, the reasons for MRT slope variation (see also Clarke, 2006).
We are grateful to Aimee Ginsburg, Anne Taylor and Jacques Scheepers for help collecting and to Ian Engelbrecht and Ansie Dippenaar for identification of the scorpions. Mike Kenward provided valuable support with the SAS analyses and statistical queries. Ken Storey, Sue Jackson, Elrike Marais, Jaco Klok, Wolf Blanckenhorn and an anonymous referee provided helpful comments on this work. Larissa Heyns and Jaco Klok provided technical support during the early stages of this project. JST and CJ are funded by an NIH grant AI-52456 to E.S. Krafsur and SLC is supported by NRF grant FA2004032000006.