Quantifying the cost of thermoregulation: thermal and energetic constraints on growth rates in hatchling lizards


Correspondence author. E-mail: clbrewst@uark.edu


  1. The optimality model of thermoregulation predicts that as the cost of thermoregulation increases, thermoregulation effort will decrease.

  2. We designed a manipulative experiment to quantify the energetic cost of thermoregulation on growth rates in eastern collared lizards (Crotaphytus collaris) by comparing growth of hatchling lizards from high- and low energetic cost of thermoregulation treatments.

  3. We designed treatments to mimic restricted thermal microenvironments (which require lizards to devote more time and energy to maintain preferred body temperatures) and unrestricted thermal micro-environments (which minimize time and energy needed to maintain body temperature).

  4. Lizards maintained similar body temperature between treatments – contradicting predictions of the optimality model of thermoregulation – but grew more slowly in the high-cost thermoregulation treatment than in the low-cost thermoregulation treatment.

  5. The reduction in growth rates in the high energetic cost thermoregulation treatment was most consistent with animals diverting energy from growth to locomotion for thermoregulation.


Physiological processes function optimally over a narrow range of temperatures (Huey 1991). To maintain body temperatures (Tb) within an appropriate range for physiological processes, many ectotherms thermoregulate behaviourally, controlling heat gain or loss through conduction, convection, evaporation and radiation (Tracy 1976; Angilletta 2009). Despite the ability of many ectotherms to maintain appropriate Tbs in heterogeneous thermal environments, even a careful thermoregulator is limited by available temperatures (Angilletta 2009). Furthermore, as environmental temperatures deviate from the optimal range for physiological function, individuals must devote more time, energy or both to thermoregulation (Huey & Slatkin 1976). The tight linkage between an ectotherm's physiology and its environment causes shifts in thermal habitat to directly impact growth and survival (Sears 2005) and opens the door to many tests of theoretical thermal biology.

Many ectotherms maintain a relatively narrow range of Tb that presumably maximizes overall fitness (Huey 1991; Angilletta 2009). This range of temperatures is believed to reflect the optimal body temperature and is referred to as the set-point range (Tset; Hertz, Huey & Stevenson 1993; but see Martin & Huey 2008). Depending on temperatures available (or the operative environmental temperatures, Te; Bakken, Santee & Erskine 1985), lizards must spend varying amounts of time and energy thermoregulating to maintain Tb within Tset. Huey & Slatkin (1976) formalized a theory that predicts optimal thermoregulatory tactics of individuals by accounting for a trade-off between the energetic costs and benefits of thermoregulation. Tests of predictions from this theory do not provide a clear picture (Grant 1990; Herczeg et al. 2006; Row & Blouin-Demers 2006; Vickers, Manicom & Schwarzkopf 2011). For example, in a laboratory experiment, Withers & Campbell (1985) manipulated the energetic costs of thermoregulation by varying the frequency at which desert iguanas (Dipsosaurus dorsalis) were required to move between basking sites (i.e. a shuttle box). Consistent with the Huey & Slatkin (1976) model, increasing the energetic cost of thermoregulation decreased behavioural thermoregulation (reduced Tb) in the desert iguana. In contrast, a rigorous comparative study of 22 species of lizards found that these species thermoregulated more effectively as the cost of thermoregulation increased (as Te deviated from Tset; Blouin-Demers & Nadeau 2005). Equivocal support might indicate that other important elements have been excluded from the existing optimality model. For example, the spatial structure of the thermal habitat should also influence the costs and benefits of thermoregulation (Angilletta 2009). Alternative habitats with similar frequency distributions but different spatial distributions of Tes are predicted to impose different costs to a thermoregulating ectotherm (indexed by necessary movements to track preferred temperatures; Angilletta 2009; Sears, Raskin & Angilletta 2011). Thus, both the temporal and spatial distribution of Te should influence the net benefit of thermoregulation in a given habitat.

The eastern collared lizard (Crotaphytus collaris) is a saxicolous lizard restricted primarily to xeric habitats in the south-western United States. Arkansas and Missouri represent the north-eastern range limit of this species (Trauth, Robison & Plummer 2004). In these states, collared lizards are restricted to the insular glades of rocky south-facing slopes within the Ozark Mountains (Brisson, Strasburg & Templeton 2003). In the early 1900s, the United States National Forest Service began implementing fire suppression laws to reduce anthropogenic fires in the Ozarks (Saveland 1995; Fralish 2004). From 1940 to 1990, the Ozark forests saw near complete exclusion of wildfires (Verble 2012). Fire suppression allows invasive vegetation, such as eastern red cedar (Juniperus virginiana) to colonize glades, thereby fragmenting and degrading glade habitats for animals that prefer open environments (Templeton et al. 2001). Floristic and biophysical changes in structure are linked with fire suppression (Van Zandt et al. 2005).

In their long-term monitoring of collared lizards, Sexton, Andrews & Bramble (1992) noted that growth rates, body size and activity times of juvenile collared lizards declined in the same fire-suppressed glade between 1960 and 1988. Reduced growth rates correlated with a 50% reduction in the number of females that became sexually mature in their first year. Furthermore, because reproductive output is strongly linked with adult body size in collared lizards (Trauth 1978; Ballinger & Hipp 1985) and in many other reptiles (Fitch 1970; Shine 1980), reduced growth rates can affect lifetime reproductive output. The cumulative effects of these changes on life-history parameters include long-term fitness consequences.

Although the specific mechanisms remain elusive, vegetative encroachment clearly affects body growth. Altering thermal suitability of a habitat could affect growth rates of lizards by a host of indirect and direct mechanisms. Here, we address three specific (albeit nonexclusive) mechanisms. A decrease in thermal suitability could cause: (i) deviation of Tbs from Tset (Tb hypothesis); (ii) diversion of energy from growth to thermoregulation (allocation hypothesis) or (iii) reduction in energy intake (acquisition hypothesis).

Tb hypothesis

An increased energetic cost of thermoregulation could result in a change in mean Tb (Huey & Slatkin 1976; Angilletta 2009). If lizards in thermally constrained habitats spend less time and expend less energy moving to suitable thermal microenvironments, then they could experience a change in mean Tb (Huey 1974) as a result of less accurate thermoregulation (how closely an organism maintains its Tb to Tset is referred to as accuracy of Tb, db; Hertz, Huey & Stevenson 1993). Reptilian growth rates are highly dependent on Tb (Andrews 1982; Buffenstein & Louw 1982; Sinervo 1990; Van der Have & De Jong 1996; Angilletta, Steury & Sears 2004). Thus, a mean Tb outside of Tset (greater or less than) caused by less accurate thermoregulation should negatively affect growth by reducing net energy gain (Dunham, Grant & Overall 1989).

Allocation hypothesis

If lizards in thermally constrained environments thermoregulate accurately, such that they maintain similar body temperatures (Tbs) to lizards in less constrained environments, then they might expend more energy thermoregulating. Under this scenario, thermal constraints impose additional energetic costs by increasing locomotion to track suitable microhabitats, necessarily leaving less energy available for allocation to growth.

Acquisition hypothesis

Finally, slower growth of lizards in thermally constrained environments might result simply from limited food intake. Intake could be reduced through two pathways. First, lizards that spend more time thermoregulating will have less time available for foraging. Second, prey might be limited in thermally constrained habitats because habitat changes decrease its suitability for prey species. Either pathway could reduce energy acquired and hence available for assimilation. The Tb hypothesis and the acquisition hypothesis are not necessarily mutually exclusive (i.e. there could be an interaction between Tb and consumption; see Brett 1971). However, they are both distinguished from the allocation hypothesis by the fact that the latter focuses explicitly on differences in energy expenditure.

We designed a manipulative experiment to address the optimality model of thermoregulation in hatchling collared lizards by manipulating the locomotory cost of thermoregulation. In addition, we designed this experiment to quantify the energetic cost in terms of growth and to distinguish among competing explanations, assuming a measurable difference in growth rate between treatments. We tested for an effect on growth by quantifying growth rates of hatchling collared lizards exposed to two treatments in a laboratory environment that facilitated repeated measurement of Tb and direct comparison of growth rates. We designed these treatments to impose different energetic costs in environments with similar mean Te but different spatial distributions by manipulating the spatial arrangement of thermal microsites. All individuals were fed ad libitum to minimize any potential impact of reduced food availability (the acquisition hypothesis).

Hatchling collared lizards provide an ideal system for testing these hypotheses because they are glade specialists and maintain relatively warm Tbs (field Tb of 38·1 and 38·9 °C for hatchlings and adults, respectively, in the source population of this study; unpublished data) and are likely affected by shifts in the spatial distribution of the thermal habitat, such as vegetative encroachment. Furthermore, hatchlings must maintain high rates of growth to reach critical minimum body size in the limited growing season prior to winter (Sexton, Andrews & Bramble 1992), so this juvenile period is especially sensitive to selective pressure over short periods of time.

Materials and methods

General methods

Using a pole and noose, we collected 17 hatchling collared lizards between late July and early September 2011, from a site in central Arkansas. Upon capture, lizards were toe-clipped for identification, and mass, SVL and Tb were recorded. We chose an SVL of 45 mm as the minimum body length for hatchlings to standardize the size of individuals used in the study. Lizards were transported to the laboratory and allowed to acclimate for 2–4 days before beginning trials. We returned the lizards to their source populations within a week of completing experiments. Consistent with recommendations for wild taxa (Sikes, Paul & Beaupre 2012), we followed taxon-specific guidelines approved by the American Society of Ichthyologists and Herpetologists for the use of wild amphibians and reptiles in research (Beaupre et al. 2004).

We provided food and water ad libitum. Each animal was offered eight adult crickets (Acheta domestica) daily and provided water in a small plastic dish. We also misted the enclosures every 4 days with a water bottle. We injected liquid vitamin supplements into a single cricket and fed these individually to lizards every 4 days. Lizards were housed individually in 2·4 × 0·6 × 0·6 m enclosures (Fig. 1) constructed from plywood and aluminium flashing. Each enclosure included a 7–8-cm layer of sandy gravel as substrate and seven ceramic bricks (~ 22 × 10 × 7 cm) for basking sites (three under each light and one at the centre of each enclosure). Enclosures were erected in two rows of eight, leaving a walkway between and along the ends of the rows. Basking lights (100 watt halogen) were suspended 40 cm above and 20 cm from the ends of each enclosure.

Figure 1.

Enclosure diagram. B, brick (~ 22 × 10 × 7 cm), W, water dish (~10 cm diameter), U = basking unit with three bricks (two bricks on bottom, one on top) and one basking lamp. Enclosure dimensions = 2·4 × 0·6 × 0·6 m, distance from centre brick to centre of basking unit = 1 m, distance from top of basking unit to basking lamp = 0·25 m. Diagram not to scale.

One experimental treatment (ALT) was designed to impose a high energetic cost of thermoregulation and mimic basking opportunities in thermally constrained habitats. This treatment consisted of basking lamps whose illumination cycles alternated every 20 min between ends of enclosures throughout the daily basking period (12 h). Lizards would have to track the alternating light in this treatment to maintain their preferred body temperature. The ALT treatment was designed to increase movement rates and energy spent thermoregulating compared to individuals in the low energetic cost of thermoregulation treatment (FXD). The FXD treatment was designed to mimic a habitat that was less thermally constrained than the ALT treatment. The FXD treatment was physically identical to the ALT treatment except that lights alternated only once (at mid-day) in a 12-h basking period. In principal, a perfect thermoregulator with a preferred Tb range of ~34–40 °C (a hypothetical Tset based on the mean field Tb for hatchlings in the source population of this study, 38 °C) would therefore be required to move at most twice in a 12-h basking period for the FXD treatment versus thirty-seven times in the ALT treatment.

Separate circuits controlled each light treatment. Treatments were randomly assigned to enclosures. Digital timers controlled ambient (room) lighting and basking thermal regimes. Ambient lighting switched on 60 min before basking lights, and switched off 60 min after basking lights, providing 14-h light and 10-h dark photoperiod. The goal of this lighting schedule was to mimic sunrise and sunset periods when little radiant heat is available. Background ambient temperatures were maintained by central HVAC at approximately 21 °C.

Data collection

We recorded Tb, SVL and body mass every 4 days over a 20-day period. A maximum of three lizards were measured in a 1-h block to allow individuals not captured in adjacent enclosures to return to normal activities. Using a pole and noose, we captured individuals as quickly as possible and immediately recorded a cloacal temperature using a digital thermocouple. If we were unable to capture an individual within 20 s, or if the individual moved farther away or closer to the heat source for more than 10 s, we would move to another animal of the same treatment. On measurement days, we systematically rotated the time we recorded Tbs of each lizard (1030, 1300, 1530 or 1800 h) so that we recorded Tb a total of four times (once at each time) for each individual. Up to three lizards were recorded at a given time block, and the treatment to be tested at that time block was randomized. We recorded body length to the nearest 0·5 mm. Care was taken to ensure that lizards were relaxed and in similar postures when measured. We measured each lizard a minimum of three times once captured and used the longest measurement that we obtained consistently in subsequent analyses. We measured mass using a digital scale to the nearest 0·001 g. After we obtained all measurements for an individual, we fed one cricket injected with a liquid vitamin solution to that individual before releasing it back into its enclosure.

We recorded activity times and movement rates of hatchlings throughout their enclosures with cameras suspended above the enclosures. Each individual was recorded in 80-min time blocks during early, middle and late activity times (starting at 1000, 1220 and 1440 h, respectively) for a total of 240 min each. We used these video data to quantify movement rates and time spent thermoregulating (time basking). For movement rates, we recorded the number of movements greater than 10 cm and the total distance moved by each individual. We also recorded the time each individual spent within or outside a 0·5 m radius from the illuminated basking lamp.

We used operative temperature models (OTMs) to estimate the distribution of temperatures in enclosures and to compare thermal availability of the two treatment types. We evaluated several model types to identify a model that most closely matched a hatchling carcass with an ibutton thermocron (Dallas Semiconductor) temperature data logger placed in the body cavity. The OTM that best fit the recordings from the animal carcass was a 1·2-cm PVC cap painted primer grey and sealed with approximately 6 cm of black electrical tape. We distributed models every 20 cm throughout enclosures (65 models total) in each treatment and recorded temperatures every 10 min.

An underlying assumption in this experiment is that if lizards in both treatments are to maintain Tset, then lizards in ALT treatment must spend more energy shuttling to basking sites than lizards in FXD treatments. Although video data can be used to quantify the mean frequency and distance travelled, the most appropriate currency to estimate energetic cost is Joules. We performed a post hoc analysis of the estimated cost of locomotion for collared lizards in both treatments using published data. The cost of locomotion accounts for the resting metabolic rate (RMR) and the metabolic cost of transport (net locomotion cost). Energetic data for C. collaris are unavailable, so we used the net locomotion cost for Dipsosaurus dorsalis, a similar sized Iguanid lizard (2·22 mL O2 g−1*km at 40 °C; John-Alder & Bennett 1981) to estimate transportation costs for collared lizards in each treatment based upon the mean mass and mean distance moved by individuals. We converted the net energetic cost into Joules assuming the standard conversion of 20·2 J mL−1 O2 (Voituron, Herold & Grenot 2000).


We analysed all data with sas 9.2 statistical software (SAS Institute 2011) or R (R version 2.14.1. R Development Core Team 2011). We checked data for normality and homogeneity of variance (Zar 1984). All data met parametric assumptions. Initial mass and initial SVL were analysed by treatment group. We used initial mass or initial SVL as a covariate in ancova when testing for differences between treatments for growth. The slope of relative SVL to time (change in SVL every 4 days, over 20 days) was used to quantify growth in length (SVL). Because change in mass was temporally variable and nonlinear (mean R2 for mass was 0·46), growth in mass is reported as a daily rate (total change in mass over 20 days). We analysed Tb, basking and movement data (number of moves and total distance moved) using mixed models. In these analyses, we assigned time of day and treatment as fixed effects and individual as a random effect. Mixed models were fit using restricted maximum likelihood (REML).


There were no significant treatment–by–body size interactions in growth rate comparisons (SVL F3, 13 = 0·02, P = 0·892; mass F3, 13 = 0·72, = 0·410). Neither initial SVL nor initial mass differed significantly between treatments (F1, 15 = 1·19, = 0·293 and F1, 15 = 0·57, = 0·461, respectively). Lizards in the FXD treatment grew faster in length (0·26 mm day−1) than those in the ALT treatment [0·19 mm day−1 (NFXD = 9, NALT = 8; F2, 14 = 18·50, < 0·001); Fig. 2a] with initial SVL as a covariate. However, growth rates based on mass did not differ statistically (with initial mass as a covariate) from those in the ALT treatment [0·046 g day−1 in the ALT treatment and 0·062 g day−1 in the FXD treatment (F2, 14 = 0·64, = 0·438); Fig. 2b]. Results were unchanged using mass- and length-specific measures of growth as in Sinervo & Adolph (1989) and Sinervo (1990).

Figure 2.

Bar graphs of least-squared means and standard error (bars) of (a) growth rate in length (mm day−1), (b) growth rate in mass (g day−1), (c) number of moves, (d) distance moved (m).

Hatchling Tbs were similar between treatments (35·3 °C and 34·6 °C in the ALT and FXD treatment, respectively; F1, 15 = 0·75, = 0·401) and did not differ significantly by time of day (F3, 45 = 1·89, = 0·143) but did differ temporally depending on treatment (i.e. the treatment–by–time interaction was statistically significant; F3, 45 = 4·50, = 0·008). Lizards in the FXD treatment exhibited a moderate increase in Tb throughout the day, whereas Tb of lizards from the ALT treatment declined slightly over the same time period (Fig. 3). Similarly, lizards in the ALT and FXD treatments did not differ in activity time spent basking (59% and 53%, respectively; F1, 10 = 1·30, = 0·279), but did display both a temporal difference between treatments (F2, 12 = 12·73, = 0·001) and a treatment–by–time interaction (F2, 12 = 42·6, < 0·001). In addition, lizards in the FXD treatment exhibited a broader range of Tb than did lizards in the ALT treatment (Fig. 4). Lizards in the ALT treatment moved more often and farther (29·6 moves and 84 m; Fig. 2c,d) in a 240-min block than lizards in the FXD treatment (13·6 moves and 31·6 m; F2, 9 = 15·91, = 0·003 and F2, 9 = 19·97, = 0·001, respectively). Lizards exhibited a temporal difference between treatments in both distance moved (F2, 14 = 4·09, = 0·040) and number of moves (F2, 12 = 8·72, = 0·005), as well as treatment–by–time interaction (distance moved; F2, 14 = 4·09, = 0·012 and number of moves; F2, 12 = 6·74, = 0·010).

Figure 3.

Temporal variation in body temperatures, Tb, of hatchling collared lizards during the experiment. Tb of lizards in the FXD treatment (black symbols and lines) increased moderately throughout the day, whereas Tb of lizards in the ALT treatment (grey symbols and lines) remained relatively constant. Data symbols represent means ± SE.

Figure 4.

Frequency histograms of body temperatures exhibited by hatchling collared lizards during experimental trials. Lizards from the FXD treatment (dark grey bars) exhibited a broader Tb range than lizards from the ALT treatment (light grey bars).

The mean temperature (Te) from all points within measured enclosures over a 12-h period was 24·4 °C in the ALT treatment and 24·6 °C in the FXD, although access to temperatures at or near the preferred temperature was available at all times during photophase. The mean Te under the illuminated basking light was nearly identical between treatments (34·4 °C in the ALT treatment, and 34·5 °C in the FXD treatment). The mean minimum Te was 19·5 °C for both treatments (centre of enclosures), but the mean maximum Te (on a basking brick directly under heat lamp) was higher in the FXD treatment (46 °C) than the ALT treatment (43 °C) due, presumably, to the fact that the longer ‘on’ time allowed deeper heating of the bricks and surrounding substrate. The mean minimum Te directly under the basking lamp in the ALT treatment during photophase (26 °C) never completely cooled to the mean Te at centre of enclosures, as was the case in the FXD treatment.

We used the equation published by John-Alder & Bennett (1981) to estimate costs of locomotion in lizards from ALT and FXD treatments based on the mean mass of all individuals at end of experiment (8·4 g). We calculated the total distance moved by extrapolating the distance covered in a 4-h time block to the entire 12-h activity period (lizards moved an average of 84 and 31·6 m, respectively, in the ALT and FXD treatments in 4 h). Based upon these data, the estimated net cost of locomotion for individuals in the ALT and FXD treatments were 4·7 and 1·67 mL O2, respectively (Table 1). Finally, we calculated the total cost of locomotion in a 12-h activity period between treatments by adding the predicted resting metabolic rate (RMR) of lizards (1·5*Mass0·80 (12 h); Thompson & Withers 1992) to the net locomotion cost. This calculation yields a locomotion cost of 191 J/12 h for lizards in the ALT treatment, and 133 J/12 h for those in the FXD treatment.

Table 1. Total locomotion cost estimates for a 12-h activity period. Mean distance (km), mean mass (g), net locomotion cost in mL O2 (LC1), net locomotion cost in Joules (LC2), resting metabolic rate in Joules (RMR), % LC2 of RMR (% RMR), total locomotion cost in Joules (TLC)
TreatmentDistanceMassLC1LC2RMR% RMRTLC


The experimental treatments we imposed were sufficient to elicit a biologically relevant response even over the short experimental period of 20 days. The ALT treatment required animals to move more frequently and greater distances each day to maintain Tbs and was coupled with a significant growth penalty. Lizards in the FXD thermal treatment averaged 37% faster growth rates in length and a 31% greater absolute change in length as compared to similar aged individuals in the ALT treatments; a difference that was statistically significant. The pattern was similar for body mass (35% faster growth rate and a 34% greater average total increase in mass of FXD individuals compared to ALT individuals), although the difference was not statistically significant.

The allocation hypothesis provides the best mechanistic explanation for these differences. This hypothesis predicts that differences in growth rate result primarily from diversion of energy from growth to thermoregulation. Lizards in the ALT treatment moved twice as often and almost three times farther than those in the FXD treatment. Because lizards in both treatments maintained nearly identical Tbs, these data suggest that the difference in locomotion was directly tied to thermoregulation. The estimated cost of locomotion in each treatment suggests the ALT treatment imposed a 43% greater energetic cost to maintain Tb as compared to the FXD treatment.

Our results do not support the Tb hypothesis. Lizards spent similar fractions of time basking and maintained similar Tbs in each treatment. This hypothesis would have gained support only if mean Tb had differed between treatments. Such was not the case and video data showed that most individuals quickly abandoned a basking site once the light cycled-off. In many instances, lizards moved within 2 min of the light change and went directly to the opposite basking site. We detected a statistically significant treatment–by–time interaction term, indicating a temporal difference in Tb between treatments. In the FXD treatment, mean Tb increased moderately over the course of a day (from 32·02 °C at 1030 h to ~36 °C at 1800 h), whereas in the ALT treatment mean Tb remained near 35 °C. Consistent with these differences, lizards from the FXD treatment exhibited a broader range of Tbs than those from the ALT treatment. Thus, counter to the predictions of the theory of optimal thermoregulation, lizards in the ALT treatment appeared to thermoregulate more effectively than those in the FXD treatment despite the increased energetic cost in the former. We note, however, that individuals in the FXD treatment were sometimes slow to move to the basking light if it illuminated at the opposite end of the enclosure in the mornings, whereas lizards in the ALT treatments never spent more than 20 min before they were exposed to a heat lamp regardless of the end of the enclosure they happened to occupy overnight. This interesting behaviour also explains the observed temporal difference and the treatment–by–time interaction associated with percent time basking, as well as movement data (distance moved and number of moves).

The last explanation (acquisition hypothesis) predicts that differences in lizard growth are best explained by a reduction in energy intake. Although food was available ad libitum to control for differences in food availability, we cannot exclude the possibility that lizards in the ALT treatment consumed less than those in the FXD treatment. It was not possible to account for total energy ingested and egested because the substrate used in enclosures made finding crickets and faecal pellets difficult. These facts notwithstanding, prey items were regularly provided in quantities that would preclude differences in prey potentially available to the focal animal. Furthermore, video data showed that lizards spent similar fractions of their activity time basking, suggesting that lizards in the ALT treatment did not spend more time thermoregulating than lizards in the FXD treatment. Thus, if lizards in the ALT treatment did in fact consume fewer prey items, it was not due to decreased food or available foraging time as is predicted by the acquisition hypothesis.

We found a considerable difference between treatments in SVL growth rate, but no statistical difference in mass growth rate in this study. Equally surprising, we found no difference in body condition index between treatments at the end of the study (anova comparing residuals from a regression of log10 body mass on log10 SVL; F1, 15 = 0·02, P = 0·88). The lack of statistically significant differences in mass growth rate is due primarily to the much greater within-sample variance in mass values, as reflected by the coefficient of variation in mass (CV = 0·67) relative to the coefficient of variation in SVL (CV = 0·21, see also Fig. 2a,b). A considerable proportion of the variance in mass can be explained by fluctuation in gut content. For example, the mean mass of individual crickets used in this study was 0·35 g, which represents approximately 5% of the average body mass of hatchlings. As the average mass increase for individuals over the course of the 20-day study was approximately 1·1 g, variance in recent consumption could skew growth estimates by one gram or more (on a number of occasions, we observed hatchlings consume 2–3 crickets in a single feeding bout). Similarly, recent defecation could also skew mass estimates. These confounding factors may have been avoided by a 3- to 4-day fasting period at the beginning and end of trials so that gut contents would have been standardized for all individuals, which would yield a more accurate representation of overall change in body mass. However, this technique would have no effect on variance of mass estimates through the middle of experiments (we measured mass every 4 days) when lizards were offered food ad libitum.

This study addresses a key prediction of the optimality model of thermoregulation: a reduction in thermoregulation effort as the energetic cost of thermoregulation increases (Huey & Slatkin 1976; Angilletta 2009). Two studies have supported this prediction in a laboratory environment (Withers & Campbell 1985; Herczeg et al. 2006). Two other studies testing the relationship between energetic costs and thermoregulation have not yielded the predicted patterns (Blouin-Demers & Nadeau 2005; Row & Blouin-Demers 2006), although these studies did not examine the effect of cost independent of differences in the distribution of available temperatures. Our study helps to fill this gap. We found no link between the level of thermoregulation and the energetic cost as predicted by the Huey and Slatkin model. The ALT treatment required a far greater effort for thermoregulation through increased locomotion than the FXD treatment, yet lizards in both treatments maintained similar Tbs. Thus, these data suggest that the relationship between energetic costs and thermoregulation is more complex than originally thought (Angilletta 2009).

Blouin-Demers & Nadeau (2005) speculated that individuals might not decrease thermoregulatory effort when the cost of thermoregulation increases. These authors argued that the physiological consequences of thermoconformity in a low thermal quality environment would outweigh the costs in a high thermal quality environment. For example, digestive efficiency drops off sharply in Dipsosaurus dorsalis and Sceloporus occidentalis when Tbs fall below 28 °C (Huey 1982) and below 30 °C in S. undulatus (Angilletta, Hill & Robson 2002). Thermoconformity in these species exposed to a Te at or below 28 °C would result in digestive failure. As the cost of thermoregulation is predicted to be greater in a low-quality thermal environment, they argued that thermoregulatory effort should increase as the thermal quality of the environment decreases. A major difference in the Blouin-Demers & Nadeau (2005) study and ours is that lizards in the present study were exposed to virtually identical thermal environments (Te of 24·4 °C ALT and 24·6 °C FXD). Instead of testing for the effects of differences in frequency distributions of Te, we directly manipulated the spatial distribution of Te. However, similar to Blouin-Demers and Nadeau's observation, thermoconformity in our study would have imposed a high physiological cost (assuming similar thermal sensitivity of energy assimilation in collared lizards relative to other species), suggesting that although the increased cost of thermoregulation was strong enough to elicit an energetic consequence (in terms of growth rate), it was not strong enough to induce a reduction in thermoregulatory behaviour. We point out that the optimality model predicts that lizards should continue to thermoregulate as long as the benefits outweigh the costs, which appears consistent with our data and the previously mentioned explanation. A particularly interesting point to address in future studies is the influence of energy assimilation in regulating the relative benefits of thermoregulating in costly thermal environments. In short, energy assimilation might serve as an important mediator of thermoregulation.

Our study is similar to Withers & Campbell (1985) in that both studies manipulated the spatial distribution of Te to impose high- and low-cost thermoregulation treatments. However, lizards in Withers & Campbell (1985) reduced behavioural thermoregulation as the energetic costs of thermoregulation increased (i.e. they maintained lower Tb). The conflicting results between Withers & Campbell (1985) and our study are likely due to differences in experimental protocols. First, lizards in Withers & Campbell's (1985) study were tested over short (1-day) treatments rather than chronic exposure to treatments as in our study. Second, the high-cost treatment imposed a substantially higher cost (in terms of locomotor cost) in their study because lizards were required to shuttle 1 m every 30 s to track a basking source rather than 2 m every 20 min in our study. In fact, the low-cost treatment used by Withers & Campbell (1985) required lizards to move 1 m every 180 s, which is greater in terms of locomotor cost than even the high-cost treatment in our study. A third important difference is that lizards in Withers & Campbell (1985) were fasted during trials. If lizards had not been fasted and had unlimited access to food (as were lizards in our study), they might have been more motivated to thermoregulate when the costs increased. Finally, Withers & Campbell (1985) did not test for effects of thermoregulation costs on growth rate.

To our knowledge, this study is the first to experimentally manipulate the spatial distribution of thermal environments to test for the effects of thermoregulation trade-offs on growth rate. The results presented here provide evidence that altering the spatial distribution of available temperatures can dramatically affect growth when lizards are forced to spend more energy thermoregulating. These data quantify the penalty on growth imposed by altering the thermal availability in a manner consistent with ongoing habitat degradation. Although the energetic cost of thermoregulation imposed in our ALT treatment was likely greater than lizards would experience as glades are invaded by vegetation, the strength of the treatment effect over such a short time span makes the allocation hypothesis a likely contributor to reduced growth of individuals and population declines of lizards in degraded glades.


We thank the landowners for allowing us access to the collared lizard source population. This study would not have been possible without their generosity and patience on our many visits to their facilities. We thank Timothy Clay, Nicholas Whitten, Cody Barnes, Melissa Rogalla, Jonathan Gray and the rest of the UALR volunteers who helped in the construction of enclosures, collection visits and lizard care throughout the study. We also thank Steven J. Beaupre, Raymond Huey and two anonymous reviewers for constructive comments that significantly improved the manuscript. This research was supported financially by the University of Arkansas at Little Rock and approved by the Institutional Animal Care and Use Committee (protocols: R-11-03 and R-10-01).