Evaporative cooling via panting and its metabolic and water balance costs for lizards in the American Southwest

Lizards thermoregulate using a suite of behavioral (i.e. shuttling and body positioning) and physiological (i.e. color change, panting, cardiovascular adjustments) mechanisms to control body temperature (T_b) or to avoid critical or lethal extremes (Cowles and Bogert, 1944; Norris, 1967; Bartholomew, 1982). When conditions are conducive to elevated T_b, lizards will often thermoregulate to maximize activity times dedicated to foraging, breeding and territory defense (Huey, 1982, 1991). In a recent study (Loughran and Wolf, 2020), we observed that when operative environmental temperatures [the equivalent T_b that integrates the effects of air temperature (T_a), wind speed and radiative exchange] exceed a lizard's preferred limits, many species can substantially lower their T_b below T_a by evaporative cooling via panting. However, because panting in lizards has historically been viewed as an ineffective thermoregulatory mechanism (Mautz, 1982a), data on metabolic and evaporative water loss (EWL) rates have typically not been reported at temperatures that exceed panting thresholds (T_pant) or approach the critical thermal maxima (CT_max; e.g. Munsey, 1972; Snyder, 1975; Mautz, 1982b; Sannolo et al., 2018; Muñoz-Nolasco et al., 2019; but see Case, 1972; Frappell and Daniels, 1991). Thus, the physiological costs that affect the panting efficacy of species, and the subsequent ecological consequences, remain poorly understood. While the positive relationship between T_a and lizard metabolic rate has been well established (Andrews and Pough, 1985), we have limited insight into the comparative contribution of the van't Hoff effect (often quantified as Q₁₀, which describes the magnitude of change in the rate of a reaction over a 10°C range), and the mechanical costs associated with panting (i.e. increased breathing and heart rates) that increase the overall metabolic rate in panting lizards. Disentangling these factors is key to understanding the metabolic ‘cost’ of panting, as a species' ability to evaporatively cool is closely tied to respiratory rates and depth (Tattersall et al., 2006).

Evaporative cooling potentially has substantial ecological benefits such as increasing surface activity times in hot conditions, providing individuals with a competitive edge against conspecifics, and temporarily maintaining a thermal safety margin equivalent to retreating to thermal refugia (Tracy and Christian, 1986; Sunday et al., 2014). However, the increased water loss rates and increases in metabolism associated with elevated T_a while panting produce cost and benefit tradeoffs such that lizards must recoup the water and energy costs of panting from food items to maintain positive water and energy balance or face reduced condition or dehydration (Bentley and Schmidt-Neilsen, 1966; Huey and Slatkin, 1976; Mautz, 1982a,b; Andrews and Pough, 1985). While panting behavior is occasionally observed in nature, its frequency and utility for evaporative thermoregulation and varying thermal conditions are poorly understood (Furness, 2021; Le Galliard et al., 2021). Thus, understanding the costs and benefits of panting and how these might vary among species and habitats is especially important given the rapid warming and drying of the arid regions of the world, where lizards are often diverse and abundant (Pianka and Vitt, 2003).

In this study, we examined the variation in the evaporative and metabolic costs of cooling via panting and their association with thermoregulatory performance over a range of temperatures for 17 lizard species, representing four families, from the American Southwest. For these species, we previously noted significant interspecific variability in their capacity to lower T_b below T_a while panting, with species showing the greatest panting ability tending to inhabit more arid environments (Loughran and Wolf, 2020). Furthermore, we found that while thermal thresholds such as T_pant and CT_max were phylogenetically conserved, panting ability was not. We concluded that panting ability is likely adaptive, and that the mechanisms that affect it (i.e. metabolic and water loss rates) will track with each species' respective panting abilities (Loughran and Wolf, 2020). Because the species used in our previous study (Loughran and Wolf, 2020) and in this study occur in a wide array of habitats, ranging from desert to montane, we can examine how the mechanisms that influence species’ panting efficiency vary across thermal environments. Additionally, as T_b increases with increasing T_a, we can also estimate the physical costs of panting by separating the van't Hoff effect on metabolism from metabolic increases due to the mechanical costs of ventilation.

Here, we asked the following questions; (1) how are values for T_b depression and rates of EWL correlated within and among species of panting lizards?; (2) what are the mechanical costs of panting, and how do these values compare with increases in metabolism due to the van't Hoff effect with increasing heat loads and T_b?; and (3) what are the potential water and energy balance costs of evaporative cooling and what are the potential implications for lizards in a rapidly warming world?

We sampled a total of 202 lizards, including 13 species from Phrynosomatidae: zebra-tailed lizard (Callisaurus draconoides Blainville 1835), greater earless lizard (Cophosaurus texanus Troschel 1852), Texas horned lizard [Phrynosoma cornutum (Harlan 1825)], greater short-horned lizard (Phrynosoma hernandesi Girard 1858), round-tail horned lizard (Phrynosoma modestum Girard 1852), regal horned lizard (Phrynosoma solare Gray 1845), twin-spotted spiny lizard (Sceloporus bimaculosus Phelan and Brattstrom 1955), Clark's spiny lizard (Sceloporus clarkii Baird and Girard 1852), southwestern fence lizard (Sceloporus cowlesi Lowe and Norris 1956), Yarrow's spiny lizard (Sceloporus jarrovii Cope 1875), crevice spiny lizard (Sceloporus poinsettii Baird and Girard 1852), ornate tree lizard [Urosaurus ornatus (Baird and Girard 1852)] and side-blotched lizard (Uta stansburiana Baird and Girard 1852); one each from Crotaphytidae: eastern collared lizard [Crotaphytus collaris (Say 1822)], and Teiidae: chihuahuan spotted whiptail [Aspidoscelis exsanguis (Lowe 1956)]; and two from Iguanidae: desert iguana [Dipsosaurus dorsalis (Baird and Girard 1852)] and chuckwalla (Sauromalus ater Duméril 1856) (Table 1). These species were targeted based on the wide range of habitats occupied [i.e. lowland desert (D. dorsalis) to arid upland (C. texanus, S. bimaculosus), to montane woodland (S. jarrovii, P. hernandesi)] and disparate ecologies [i.e. sit-and-wait (C. collaris) versus active foraging (A. exsanguis), habitat generalists (S. clarkii, S. cowlesii) versus specialists (S. poinsettii)]. Lizards were captured at multiple localities in New Mexico and Arizona, USA, between May and September of 2017 and 2018, and in May of 2019. Sauromalus ater were captive-hatched individuals from parents originating in Riverside Co., CA, USA, and were kept in semi-naturalistic outdoor enclosures prior to trials.

Animals were captured with a lasso attached to a ∼3.7 m long pole or by hand. Following capture, lizards were held in cloth bags and transported to the University of New Mexico laboratory in a cooler that was kept in a climate-controlled environment (room temperature or cooler) to reduce activity. Animals were held no longer than 72 h prior to trials, with the majority of individuals (>60%) undergoing a trial within 24 h of being captured. Food was withheld a minimum of 6 h prior to trials. Lizards were offered water ad libitum with a wet paper towel or free-standing water prior to trials. Following trials, lizards were euthanized via an intracelomic injection of a 50% mixture of MS-222 and water (Conroy et al., 2009) and deposited in the Museum of Southwestern Biology. Animal care protocols were approved by the Institutional Animal Care and Use Committee of the University of New Mexico (protocol no. 16-200437-MC). Lizards were captured under permits from the New Mexico Game and Fish Department (#3627) and the Arizona Game and Fish Department (#SP510878). Ironwood Forest National Monument also granted permission to collect animals.

T_a and T_b were measured using a thermocouple thermometer (model TC-2000, Sable Systems, Las Vegas, NV, USA) with two Cu–Cn thermocouples (model RET-4, Physitemp, Clifton, NJ, USA) inserted in the animal chamber via a small hole sealed with silicone glue. One thermocouple measured chamber T_a and the second thermocouple measured T_b and was inserted ∼10 mm (up to 20 mm for larger lizards such as chuckwallas) into the lizard's cloaca and held in place with a 1 cm wide piece of vinyl electrical tape.

Metabolic rate and EWL measurements were made over a T_a range of 35 to 50°C using a flow-through respirometry system. The respirometry chamber consisted of a transparent plastic container (1.7 l, 12 cm×8 cm×16 cm for lizards under 50 g; 3.6 l, 20 cm×8 cm×22 cm for lizards over 50 g; Snapware Total Solutions, Pyrex, Greencastle, PA, USA) sealed by a snap-latch lid lined with a silicone gasket. Before measurements, we tested the chambers for ambient air infiltration by comparing CO₂ and H₂O vapor concentrations in the chambers against air that had been scrubbed in a Drierite/Ascarite column. The lizard chamber was placed inside an environmental chamber (model no. 166VL, Percival Scientific, Perry, IA, USA) where temperature was controlled to ±0.5°C.

Incurrent air was provided from a compressed air source pushed through a purge gas generator where air was scrubbed of CO₂ (<1 ppm) and H₂O (dew point <−20°C; model PCRMBX1A##-F, Puregas LLC, Broomfield, CO, USA). Air flow to animal chambers was regulated by mass-flow controllers (MC-5SLPM-TFT/5M, Alicat, Tucson, AZ, USA). Flow rates to the chamber were selected to maintain a low H₂O concentration (<5 ppt water vapor). Depending on H₂O concentration and animal body mass (M_b), flow rates ranged from 0.5 to 5 standard liters per minute (SLM). Excurrent air from the chambers and baseline air were subsampled by a multiplexer (model TR-RM8, Sable Systems) that was electronically programmed to switch between chamber and baseline at 5 min intervals. Subsampled air was directed to a CO₂/H₂O gas analyzer (model LI-840A in 2017, model LI-7000 in 2018 and 2019, LI-COR, Lincoln, NE, USA) at approximately 300 ml min⁻¹. Gas analyzer units were zeroed with Drierite- and Ascarite-scrubbed air and spanned with a dew point generator set at 5°C (model LI-610, LI-COR) and with a known CO₂ gas mixture (1802 ppm). Voltage outputs from the thermocouple and gas analyzer units were digitized using an analog–digital converter (model UI-2, Sable Systems), and were recorded once per second by Expedata (version 1.4.15, Sable Systems). All tubing in the system was ¼ inch (6.35 mm) Bev-A-Line IV tubing (Thermoplastic Processes Inc., Warren, NJ, USA).

Lizards were weighed to ±0.1 g accuracy on a digital scale (model V31XH202, Ohaus, Parsippany, NJ, USA) and placed into the lighted chamber at a T_a of 35°C and were left until T_a and T_b equilibrated, allowing for habituation to the chamber environment. Any fecal material produced prior to or during the trial was removed from the chamber, weighed and subtracted from initial body mass, and the chamber environment was allowed to re-equilibrate. Experimental trials commenced when T_b was equal to T_a (±0.2°C). We started all lizards at 35°C and then increased the temperature to 38°C, followed by increases in 2°C increments until the lizard reached its approximate CT_max. Lizards were held at each new set chamber temperature for approximately 30 min to allow T_b to equilibrate with T_a, before ramping up to the next temperature. Lizard activity was monitored via video camera and activity measurements were categorized as ‘inactive’, ‘brief activity’, ‘continuous activity’ and ‘panting’. It was noted when lizards were continuously active, i.e. when a lizard engaged in running or jumping for longer than 5 s, and those measurements were removed from the data series. Slight intermittent gaping was not categorized as ‘panting’. A detailed description of the heating protocol and the criteria used for measuring panting and CT_max is given in Loughran and Wolf (2020). In brief, a lizard's CT_max was designated when (1) there was a sharp increase in T_b, indicating evaporative cooling was no longer effective, (2) a lizard showed prolonged distress or escape behavior (i.e. continuous running/jumping) for >5 min, (3) chamber CO₂ values fell sharply, indicating heat shock, and (4) a lizard showed a loss of balance or righting response (LRR), including the onset of spasms (OS; Lutterschmidt and Hutchison, 1997; see Loughran and Wolf, 2020, for discussion). We designated the maximum T_a experienced by the lizard at their CT_max as the heat tolerance limit (HTL). Following each trial, the lizard was removed from the chamber, reweighed and allowed to recover in a cotton bag with a wet paper towel until its T_b stabilized to approximately room temperature.

For each trial, we selected the lowest 5 -min average CO₂ readings per recorded temperature. We aimed to select 5 min segments in the last 5–10 min of each 30 min reading, so that the animal's T_b had sufficient time to equilibrate. Segments where ‘continuous activity’ was observed were excluded from the analyses. We used eqns 10.5 and 10.9 from Lighton (2008) to determine the rate of CO₂ production in ml min⁻¹ and H₂O production in mg h⁻¹, respectively. We assumed a respiratory quotient (RQ) of 0.71 for lizards. We used a thermal equivalent of 27.8 J ml⁻¹ CO₂ produced to estimate metabolic heat production (MHP; Walsberg and Wolf, 1995), and the latent heat of water vaporization of 2.41 J mg⁻¹ H₂O to estimate evaporative heat loss (EHL; Tracy et al., 2010). MHP and EHL estimates are reported in milliwatts per gram (mW g⁻¹). We define metabolic and evaporative scope as the ratio of the maximum observed rate (at CT_max) to the minimum observed rate (at 35°C) and define evaporative heat dissipation efficiency (EHD) as the ratio of maximum EHL to maximum MHP.

All statistical analyses were carried out using R software (version 4.0.2, http://www.R-project.org/). To determine the effect of panting on metabolic and evaporation rates, panting status was divided into two categories when making comparisons over a range of T_a values: ‘panting’ and ‘not panting’. We analyzed the effect of panting on MHP and EHL for each species using linear mixed effects models with the nmle package (https://CRAN.R-project.org/package=nmle) with panting status treated as a main effect, T_a treated as a covariate, and individual lizard treated as a random factor.

We used the mean maximum observed MHP, maximum Q₁₀-predicted MHP, maximum EWL and EHL, maximum EHD, and maximum T_a–T_b gradient for each species when conducting regression analyses. We compared maximum observed MHP values with Q₁₀-predicted MHP values using Welch's t-test, as variances between groups were not equal. We used general linear models to analyze the relationships between (1) the cost of panting (i.e. maximum observed−predicted MHP) and maximum EHL, (2) maximum EHL with maximum T_a–T_b gradient, and (3) EHD with evaporative scope. For test 2, leverage values for one species (P. modestum) had a significant outlier effect on the overall species trend due to high leverage. We thus report regression results both with and without P. modestum included. Phylogenetic independence of traits was tested a priori (maximum temperature gradient maintained while panting, as well as the associated traits of maximum MHP and EHL, showed no phylogenetic signal as determined by Blomberg's K and Pagel's λ; Loughran and Wolf, 2020). Because we account for mass-specific rates of MHP and EHL, we did not include M_b as a main effect in those respective model analyses.

There was a significant increase in EWL (mg g⁻¹ h⁻¹) with increasing T_a for all species and a significant effect of panting on water loss rate for 16 species (Fig. 1, Table 1): C. collaris: 24.4-fold increase (F_1,117=148.3, P<0.001), D. dorsalis: 16.2-fold increase (F_1,62=128.9, P<0.001), S. ater: 13.7-fold increase (F_1,49=26.8, P<0.001), C. draconoides: 14.3-fold increase (F_1,54=30.3, P<0.001), C. texanus: 17.5-fold increase (F_1,55=105.0, P<0.001), P. cornutum: 15.9-fold increase (F_1,27=14.9, P<0.01), P. hernandesi: 15.9-fold increase (F_1,70=9.5, P<0.01), P. modestum: 26.2-fold increase (F_1,47=65.8, P<0.001), P. solare: 12.2-fold increase (F_1,8=15.5, P<0.01), S. bimaculosus: 11.9-fold increase (F_1,46=29.3, P<0.001), S. clarkii: 14.2-fold increase (F_1,66=55.5, P<0.001), S. cowlesi: 10.3-fold increase (F_1,50=15.1, P<0.01), S. jarrovii: 10.6-fold increase (F_1,39=31.2, P<0.001), S. poinsettii: 4.6-fold increase (F_1,81=5.1, P=0.02), U. ornatus: 11.7-fold increase (F_1,63=139.2, P<0.001) and U. stansburiana: 5.4-fold increase (F_1,34=25.8, P<0.001). While panting, slopes ranged from 0.015 g h⁻¹ °C⁻¹ for S. poinsettii to 0.18 g h⁻¹ °C⁻¹ for S. ater. Minimum rates of EWL (g h⁻¹) at T_a=35°C varied from an average of 0.01 g h⁻¹ for C. draconoides to 0.09 g h⁻¹ for S. ater, whereas maximum rates of EWL varied from an average of 0.1 g h⁻¹ for U. stansburiana to 1.5 g h⁻¹ for S. ater (Table 1).

Lizards lost an average of 6.4% M_b over the course of a trial, and an average of 2.8% M_b h⁻¹ while panting at peak capacity at the species' respective CT_max. There was a significant negative association between M_b and the percentage M_b h⁻¹ lost while panting (F_1,182=40.4, P<0.001), with a significant interactive effect of lizard family (F_3,182=11.5, P<0.001) accounting for different average body masses of families (i.e. Iguanidae versus Phrynosomatidae). Small-bodied lizards tended to lose a greater percentage body mass at higher rates than large-bodied lizards, with the smallest lizards (U. oranatus and U. stansburiana) losing approximately 2.1% and 4.0% M_b h⁻¹, respectively, whereas the largest lizard (S. ater) lost an average of 1.4% M_b h⁻¹ while panting at maximum capacity (Table 1).

There was a significant increase in MHP (mW g⁻¹) with increasing T_a for all species and the onset of panting had a significant effect on metabolic rate for 11 species from 35°C to the CT_max (Fig. 2, Table 2): C. draconoides: 5.2-fold increase (F_1,54=47.6, P<0.001), C. texanus: 7.2-fold increase (F_1,55=10.6, P<0.01), C. collaris: 5.5-fold increase (F_1,117=32.3, P<0.001), D. dorsalis: 6.8-fold increase (F_1,62=25.1, P<0.001), P. hernandesi: 3.9-fold increase (F_1,70=12.7, P<0.001), S. bimaculosus: 3.7-fold increase (F_1,46=9.9, P<0.01), S. clarkii: 5.0-fold increase (F_1,66=22.8, P<0.001), S. cowlesi: 3.1-fold increase (F_1,50=28.3, P<0.001), S. jarrovii: 3.4-fold increase (F_1,39=8.3, P<0.01), U. ornatus: 4.2-fold increase (F_1,63=96.9, P<0.001) and U. stansburiana: 2.5-fold increase (F_1,34=14.5, P<0.001).

The maximum observed MHP significantly exceeded the maximum predicted MHP estimates from pre-panting Q₁₀ values for 10 species (Fig. 3): C. collaris: 4.3 mW g⁻¹ (123% higher, t=6.4, P<0.001), D. dorsalis: 4.2 mW g⁻¹ (127% higher, t=4.7, P<0.001), C. draconoides: 3.6 mW g⁻¹ (97% higher, t=4.1, P<0.001), C. texanus: 5.4 mW g⁻¹ (109% higher, t=4.6, P<0.001), P. cornutum: 4.1 mW g⁻¹ (120% higher, t=2. 4, P=0.04), P. hernandesi: 2.9 mW g⁻¹ (61% higher, t=3.1, P<0.01), S. bimaculosus: 3.0 mW g⁻¹ (74% higher, t=3.0, P=0.01), S. clarkii: 5.0 mW g⁻¹ (137% higher, t=9.0, P<0.001), S. cowlesi: 3.0 mW g⁻¹ (63% higher, t=4.4, P<0.001) and U. ornatus: 5.6 mW g⁻¹ (117% higher, t=6.9, P<0.001). Phrynosoma solare, S. jarrovii and S. ater had maximum MHP values that exceeded estimates by 3.7 mW g⁻¹ (102% higher), 1.4 mW g⁻¹ (29% higher) and 1.5 mW g⁻¹ (34% higher), respectively, suggesting a metabolic cost for panting, although these values were not statistically significantly different from values expected based on pre-panting Q₁₀ estimates (Table 2). Aspidoscelis exsanguis, P. modestum, S. poinsettii and U. stansburiana had maximum MHP values that were within 1 mW g⁻¹ of Q₁₀ predictions, indicating a minimal effect of panting on the overall increase in metabolic rate.

EHL significantly affected lizard T_a–T_b gradient for each species when tested independently. Across species, there was a significant positive relationship between maximum EHL and maximum T_a–T_b gradient (F_1,15=14.2, P<0.01; R²=0.49; Fig. 4). There was a non-significant positive effect of the metabolic cost of panting (i.e. maximum observed−predicted MHP) on maximum EHL when P. modestum was included in the regression (F_1,15=3.4, P=0.08; R²=0.18) and a significant effect of metabolic cost when P. modestum was excluded (F_1,14=24.1, P<0.001; R²=0.61; Fig. 5). During panting, rates of evaporative heat dissipation efficiency (ratio of maximum EHL to maximum MHP) had a significant positive relationship with evaporative scope (F_1,15=11.2, P<0.01; R²=0.43). Aspidoscelis exsanguis and S. poinsettii had the lowest overall evaporative heat dissipation efficiency while panting, at 1.5±0.1 and 1.9±0.1, respectively, owing to their high MHP (A. exsanguis) or low EHL (S. poinsettii) when exposed to high air temperatures. In contrast, C. collaris, P. solare and P. modestum had the highest evaporative heat dissipation values, at 5.0±0.5, 4.9±1.9 and 6.1±0.9, respectively, as a result of their high EHL values and comparatively low MHP values (Table 2). Interestingly, S. ater, which had a relatively low average EHD value of 2.0±0.2, still managed to significantly lower its T_b an average of 1.7°C below T_a while panting, whereas S. jarrovi, which had a moderate EHD of 2.8±0.2, was only able to lower its T_b an average of 1.0°C below T_a while panting (Table 2; Table S1).

Across our survey of 17 species of lizards we found that panting effort – increases in metabolic rate – directly affected evaporative cooling ability in lizards, where an increased effort resulted in greater rates of EWL. Increasing rates of EHL produced increasing T_b depression, supporting the direct link between evaporative cooling and T_b. We also found that metabolic rate increased approximately 4.7 times across species from T_a of 35°C to the CT_max and approximately 37% of this increase was attributable to the mechanical costs of panting. Modest increases in metabolism produced high rates of EHL with ratios of EHD to MHP (cost of panting values) averaging 5.7 across 14 species. Overall, our data indicate that T_b depression via evaporative cooling has modest metabolic costs, but high costs in terms of water that make it an expensive strategy for controlling T_b under high environmental heat loads. Below, we discuss these results in more detail.

In all species, rates of EWL increased with the onset of panting, although the cooling effects were highly variable among species (Fig. 1). Across species, there was a high correlation between rates of EWL and T_b depression, with species that had the greatest evaporative rates during panting depressing T_b 2–3°C below T_a (Fig. 4). Among the most effective were C. collaris, which evaporated water at a maximum rate of 37.3 mg H₂O g⁻¹ h⁻¹, roughly equivalent to an EHL rate of 25 mW g⁻¹, followed by S. clarkii, with a maximal EWL rate of 36.8 mg H₂O g⁻¹ h⁻¹ (equivalent to 25.6 mW g⁻¹), and P. modestum, which showed a maximal evaporative rate of 55 mg H₂O g⁻¹ h⁻¹ (equivalent to 36.8 mW g⁻¹). In contrast, species such as A. exsanguis and S. poinsettii showed more modest increases in EWL following the onset of panting, with maximal observed EWL rates of 20.1 mg H₂O g⁻¹ h⁻¹ (∼13.4 mW g⁻¹) and 13.1 mg H₂O g⁻¹ h⁻¹ (∼9.3 mW g⁻¹), which resulted in minimal depression of T_b (Table 1; Table S1). Overall, species differences represented a 5-fold variation in EWL and cooling ability, with evaporative scope (magnitude of increase in evaporation from 35°C to CT_max) ranging from roughly 5 at the low end (A. exsanguis, S. poinsettii) to roughly 25 at the high end (C. collaris, P. modestum, S. clarkii; Table 2). In the case of A. exsanguis, high variation in EWL measurements due to the erratic escape behavior may have inflated the estimate of evaporative scope (Fig. 1, Table 2). Nevertheless, the range in evaporative scope represents differences in the ability to depress T_b ranging from 0.7 to 2.7°C on average (Table S1), a trend that potentially tracks the aridity of habitats occupied by species (Duvdevani and Borut, 1974; Loughran and Wolf, 2020).

While there are several studies that have described the EWL rates of some species used in this study, comparative data on EWL rates at T_a that exceed a lizard's T_pant or approach its CT_max are limited (Le Galliard et al., 2021). Dawson and Templeton (1963) measured EWL rates for C. collaris and reported EWL rates of 1.0–6.0 mg H₂O g⁻¹ h⁻¹ at 40–45°C. Chamber flow rates were much lower than those used in the current study (∼125 ml min⁻¹) with chamber humidity ∼30%, which produce conditions that inhibit EWL. Similarly, Templeton (1960) measured EWL rates for D. dorsalis at air temperatures above its T_pant, reporting water loss rates at 3.6 mg H₂O g⁻¹ h⁻¹. It is likely that experimental conditions in these studies explain the overall view at the time that most lizards lack the capacity for evaporative cooling as the conditions were not reflective of hot arid deserts. A seminal study by Case (1972) was apparently the first to recognize and show the capacity for evaporative cooling in lizards residing in hot deserts. Case (1972) measured heating and cooling rates in S. ater [obesus] from the Mojave Desert and their ability to maintain stable T_b at T_a exceeding 44°C. During heating trials, S. ater started panting at T_a 40–44°C and were able to maintain a T_a–T_b gradient of 2–5°C for up to 2 h, which required EWL rates of 13.2–22.8 mg g⁻¹ h⁻¹. Case (1972) also found seasonal differences in cooling ability, with animals tested in the spring having a greater capacity for evaporative cooling compared with those tested in the autumn. In our study, EWL rates in S. ater averaged 13.9 mg g⁻¹ h⁻¹ at their CT_max and lizards were able to maintain an average T_a–T_b gradient of 1.7°C (Table 1; Table S1), which compares well with the values observed by Case (1972).

We sought to quantify the metabolic cost of panting for lizards by separating temperature-induced increases in metabolic rate (i.e. van't Hoff effect) from increases in metabolism due to the physical cost of panting. This allows us to establish how energetically expensive panting is for lizards, as well as better understand the role of T_a in influencing a lizard's metabolic rate when temperatures exceed T_pant. This approach provides some insight into the factors affecting the metabolic costs of panting at high temperatures but does not account for all factors that may influence energy expenditure in nature (i.e. air density, ventilation rates, stress) and should thus be considered with caution. To estimate the temperature-induced increases in metabolism, we calculated a pre-panting Q₁₀ value for each species and used this value to account for thermally induced changes in metabolism above the panting threshold to the CT_max. Our pre-panting Q₁₀ estimated values compare well with literature estimates of Q₁₀ values over this temperature range (e.g. Crawford and Kampe, 1971; Watson and Burggren, 2016). Overall, metabolic rate at the CT_max was from 3-fold (A. exsanguis) to more than 7-fold (C. texanus) higher than values measured at a T_a of 35°C, with species exhibiting the greatest panting effort (i.e. metabolic cost of panting) also showing the largest increases in EWL (Table 2). Panting costs, after removal of the van't Hoff effect contributions, varied widely across species but increased with evaporative cooling ability (Figs 2 and 3). We found that 10 species showed panting costs above the Q₁₀ metabolic baseline, with panting costs ranging from <1 times (P. modestum, S. poinsettii) to roughly 3 times (C. collaris, C. texanus, D. dorsalis) greater than the predicted metabolic rate based on pre-panting Q₁₀ values. These values indicate that given a species' overall panting effort, energetic costs of panting are either equal to or greater than the thermally induced van't Hoff effect but are comparatively small compared with the EHL from this activity (Table 2; Table S2). Consequently, the metabolic heat produced from increasing panting effort does not cancel the cooling effects of EHL.

Species that showed greater increases in metabolism with panting typically had higher EWL rates. Phrynosoma modestum had exceptionally high EWL rates that were supported by low metabolic costs, making it one of the most efficient species at dissipating heat (Fig. 5). This could be due to its small size combined with the low metabolic rate and high heat tolerance characteristic of Phrynosoma, although further investigation is required. One species, A. exsanguis, did not display an increase in metabolic rate above the Q₁₀ estimates with the onset of panting. Rather than panting, this species displayed rapid gular fluttering (opening its mouth and rapidly pulsating the hyoid and buccal membrane) when its T_b exceeded the panting threshold, and increases in metabolism with increasing T_b tracked pre-panting Q₁₀ metabolism estimates (Fig. 3). The gular flutter observed in A. exsanguis was distinct from the panting observed in other species as an evaporative cooling mechanism. With panting, lizards typically increase breathing frequency and decrease breathing depth to rapidly ventilate the upper airways; during gular flutter, air is being rapidly moved over the interior of the buccal membranes, with relatively little movement of air in the upper respiratory passages. For A. exsanguis, the gular flutter we observed did not produce EHL at rates sufficient to provide cooling of the body, but may have served to provide cooling for the brain (Tattersall et al., 2006).

While we have shown that many lizards have a significant capacity for evaporative cooling and T_b depression, water deficits accrued while panting can rapidly account for a significant portion of a lizard's daily water budget. We found that lizards lost anywhere from 70 mg H₂O h⁻¹, or 2% M_b h⁻¹ in the smallest species (U. stansburiana, mean M_b=3.4 g) to >1500 mg H₂O h⁻¹, or 1.3% M_b h⁻¹ in the largest (S. ater, mean M_b=114 g) while panting at their upper heat tolerance limits (Table 1). While larger-bodied species evaporated substantially greater amounts of water while panting, the overall depletion relative to their water budget was lower when compared with that of small-bodied species. The higher rates of EWL for small-bodied species have important consequences for modulating activity times available to individuals, as higher rates of heat and water flux constrain thermoregulatory opportunities when different-sized individuals are placed in similar thermal conditions (Loughran, 2014; Sears et al., 2016). For instance, when Kearney et al. (2020) modeled thermoregulatory behavior of different-sized lizards under the same thermal conditions, they found that a 10 g lizard had to shuttle between microclimates 5 times more frequently than a 1000 g lizard under the same thermal conditions to maintain heat balance. In the context of evaporative cooling, because small-bodied lizards lose water at significantly higher rates than large-bodied lizards, it is costlier to employ panting as a thermoregulatory mechanism, which may lead smaller individuals to favor thermoregulatory shuttling over evaporative cooling. These high rates of water loss across the species' size range suggests that if lizards use panting to modulate their activity periods when exposed to high environmental heat loads they must primarily acquire prey to replace water lost via panting, i.e. they must forage for water.

One of the consequences that climate warming is predicted to have on lizards is the reduction in available activity hours, which are necessary for foraging, breeding and territory defense (Sinervo, et al., 2010). Might species that can effectively depress T_b via panting use panting as a strategy to increase activity periods to counter higher environmental temperatures? We have seen that, especially in smaller species, panting rapidly produces large water deficits that have to be made up. There is some evidence that predators increase pressure on prey to increase water intake (reviewed in McCluney et al., 2012). However, continued drying of the southwest deserts in the coming decades will result in reduced plant and insect biomass, which may in turn result in decreases in lizard population abundance (Archer and Predick, 2008; Flesch et al., 2017). Consequently, evaporative cooling via panting may not be a viable thermoregulatory option for lizards in arid lowland deserts, as increased rates of water loss may be difficult to recover. Nevertheless, panting may confer an advantage for species in higher elevation habitats, where thermal conditions may be more conducive to greater heat dissipation and higher prey capture rates than in low elevation habitats (Chamaillé-Jammes et al., 2006). Furthermore, because rapid upward shifts in elevation have been documented in species distributions, species that are adept at panting may be poised to outcompete species that do not pant as their distributions shift upward (Chen et al., 2011; Wiens et al., 2019; Loughran and Wolf, 2020). Whether extending activity periods via panting might provide a growth or survival benefit is an open question and data on the frequency of panting by lizards in the wild are sparse. This study provides insight into the costs and benefits of evaporative cooling and the potential for physiological thermoregulation to partially release constraints on activity imposed by a rapidly warming world.

We thank Chris G. Anderson, Lauren M. Bansbach, Bruce L. Christman and J. Tom Giermakowski for invaluable assistance in locating and capturing lizards in the field, and Chris R. Tracy for providing the chuckwallas. We thank Denis V. Andrade and one anonymous reviewer for constructive comments that helped to improve the manuscript.