Hotter deserts and the impending challenges for the Spiny-tailed Lizard in India

Foraging activity in the wild and not only thermal variables measured in the lab plays an important role in predicting the impacts of climate change in an herbivorous desert lizard. Abstract Ectotherms are particularly vulnerable to climate change, especially those living in extreme areas, such as deserts, where species are already thermally constrained. Using the vulnerable herbivorous lizard, Saara hardwickii , as a model system, we used a multi-pronged approach to understand the thermal ecology of a desert agamid and potential impacts of rising temperatures. Our data included field-based measures of operative temperatures, body temperatures, and activity, as well as lab-based measures of thermal limits, preferences, and sprint speed. As expected, the temperature dependence of locomotor performance and foraging activity were different, and in the worst-case global warming scenario (SSP5-8.5), potential sprint speed may decrease by up to 14.5% and foraging activity may decrease by up to 43.5% by 2099. Burrows are essential thermal refuges, and global warming projections suggest that S. hardwickii may be restricted to burrows for up to 9 hours per day by 2099, which would greatly limit critical activities, like foraging and seeking mating opportunities. Overall, we show that key information on thermal ecology, including temperature-sensitive behaviours in the wild, is necessary to understand the multiple ways in which increasing temperatures may influence ectothermic vertebrates, especially


Introduction
Climate change is one of the major drivers of biodiversity change in the current century (Sala et al., 2000).With climatic projections forecasting an increase in global temperatures of up to 4°C by the end of this century (IPCC, 2019), nearly one million species of plants and animals are estimated to face the risk of extinction (IPBES, 2019).Ectotherms are particularly susceptible to the impacts of climate change since performance and fitness are highly dependent on environmental temperatures (Abram et al., 2017;Angilletta, 2009).
Many models use physiologically-based thermal tolerance limits as criteria to generate broadscale predictions of ectotherm extinction probabilities under global warming scenarios (Huey et al., 2010;Sinervo et al., 2010).Ectotherms, however, are not completely vulnerable to environmental conditions.Thermoregulation strategies, such as behaviourally seeking refuges or adjusting activity windows, can maintain body temperature (T b ) within preferred ranges, even in sub-optimal conditions (Huey et al., 2003;Logan et al., 2019).Predictive models that exclude coping strategies, such as the use of thermal refugia, the thermal variability of different microhabitats, the capacity to thermoregulate, or activity patterns, may over-or under-estimate the potential effects of climate warming on species (Briscoe et al., 2016;Gunderson & Leal, 2015;Moore et al., 2018).Because climate change is predicted to impact behaviour (Abram et al., 2017) through physiology (e.g., Burraco et al., 2020;Kingsolver et al., 2013), approaches to understand the impacts of climate change requires measurements of the thermal environment that species experience as well as their responsive strategies (Gates, 2012;Gunderson & Leal, 2016).
Lizards are ideal model organisms to study thermal biology and the impacts of climate warming on ectotherms (Pianka, 1986).As with most ectotherms, the relationship between temperature and critical functional processes, such as locomotion and growth, is typically characterized by a non-linear and asymmetrical curve, peaking at an optimum temperature where performance is at its highest level (T opt ).To maintain body temperatures within their preferred range around T opt , lizards utilise a wide range of thermoregulatory strategies, which include metachrosis, postural changes, altering activity times, and shuttling between shaded and open microhabitats (Angilletta, 2009).To prevent overheating when temperatures exceed thermal limits, lizards that are behavioural thermoregulators seek thermal refuge.Using Biology Open • Accepted manuscript refugia, however, has costs as animals usually cannot forage, mate or defend territories during that time.Many have predicted that the temperature increases due to global warming will increase the hours in which lizards are restricted in their refugia.Using hours of restriction (h r ) as a measure, Sinervo et al. (2010) predicted that by 2080, local extinctions of lizards may reach 39%, and total species extinction may reach 20%.Predictive models based on h r that are calculated using thermal tolerance limits (Sinervo et al., 2010), however, might underestimate the activity budgets of lizards because they do not take into account variation in other behaviours that might also be influenced by temperature (Gunderson & Leal, 2016;Kearney, 2013).
Lizards surviving in areas with extreme climates, such as deserts, are even more vulnerable to the impacts of climate change since environmental temperatures often exceed their thermal thresholds (Huey et al., 2009;Ivey et al., 2020;Sunday et al., 2014).
Hardwicke's spiny-tailed lizard, Saara hardwickii is a diurnal herbivorous lizard dwelling in arid and semi-arid areas of India, Pakistan, and Afghanistan.(Das et al., 2013;Dutta & Jhala, 2007;Khan & Mahmood, 2004).This species is under the vulnerable category of the IUCN (Vyas et al., 2022).In addition to severe poaching and habitat loss due to cropland expansion (Dutta & Jhala, 2014), S. hardwickii also faces the thermal challenges of living in a desert ecosystem.The Thar desert of Jaisalmer district in Rajasthan, India, where densities of S. hardwickii are high (Ramesh & Ishwar, 2008), is one of the hottest areas in India (Booth et al., 2014).Air temperatures can reach up to 50 o C in this region.It is in this landscape that we studied the thermal ecology of this agamid and its vulnerability to climate change.From systematic field measures, we obtained data on body temperature (T b ), operative temperatures (T e ), and activity pattern, which enabled us to quantify the effectiveness of thermoregulation and temperature-dependent activity.These data from the wild also enabled us to determine the Voluntary Thermal Maxima (VT max ) or average daily maximum T b , an ecologicallyrelevant measure of a thermal threshold which indicate the temperature at which individuals of the species seek thermal refuge in burrows (Camacho & Rusch, 2017;Ivey, 2020).Using wild-caught lizards brought into a controlled laboratory condition, we measured the Critical Thermal Maxima (CT max ), Critical Thermal Minima (CT min ), preferred temperature (T set ), and thermal performance curve of locomotion.This is the first study of thermal ecology for any lizard in India.With these data, we explore how temperature influences the physiology and behaviour of a desert vertebrate, and evaluate how projections of temperature increases may impact h r , locomotor performance, and foraging activity.With an integrated approach that Biology Open • Accepted manuscript incorporates measures of behaviour in the wild and thermal limits in the laboratory, these data provide a multivariate understanding of the vulnerability of this threatened species.

Study area
All sampling and data collection were conducted at two sites in the Thar desert of Jaisalmer: Sam (26°49'32.5"N70° 30' 42.3"E) and Bedhiya (26°52'07.3"N70°27'36.6"E)near Desert National Park Wildlife Sanctuary, India (See supporting information S1 for images of field sites).These areas are characterized by extremely harsh, arid summers with air temperatures reaching up to 50˚C, and cold winters with temperatures dropping as low as 0 ˚C.Rainfall in the Thar landscape is erratic and typically ranges between 100 and 450 mm per year (Weather station data, Jaisalmer; ~ 45 km from both sites).Data on the thermal ecology of this species were generated between the relatively cooler month (March) and the hottest month (May) within their active season in 2021.

Operative temperatures in the wild (T e )
To quantify the Operative Environmental Temperature (T e ) of lizards in different microhabitats, we deployed biophysical models made from copper in areas that lizards utilise (as per Bakken 1992;Bakken & Angilletta, 2014;Dzialowski, 2005).Copper models were moulded to match a typical adult, including the tail which is large and does not autotomize in this species (400 mm length × 35 mm width).Models were hollow but closed on both sides, with a temperature logger (Maxim Thermochron iButton DS1923; accuracy ±0.5°C) positioned in the middle using a plastic ring.The loggers were set to record operative temperature every 15 minutes for three months.To determine the appropriate size of copper model for Te measurements, we tested two sizes that emulated adult and subadult lizards (See supporting information S2 for more details).We found no significant differences in Te between these copper models and thus only the adult size was used for the rest of the analyses.Copper model temperatures were also calibrated in the laboratory by comparing them with cloacal and skin temperatures of three wild-caught lizards of different size classes (See supporting information, S3).
Saara harwickii in the Thar landscape mainly utilize the open environment and burrows in these study areas.Although trees and shrubs are present at the Bedhiya study site, the lizards do not typically utilize them for thermoregulation.Thus, to capture the variation in Biology Open • Accepted manuscript T e between the most extreme microhabitats, we positioned copper models in the open and within burrows.For T e in the open microhabitats, copper models (moulded with body + legs) were positioned in direct sunlight (n = 2).For T e within burrows, copper models (body only) was placed ~1m inside an inactive burrow (n = 2).Copper models that resembled only the body of a lizard was used within the burrow because burrows are narrow and lizards come into direct contact with the burrow walls.Furthermore, we expect heat transfer within burrows to be via conduction whereas on the surface, heat transfer happens via conduction, solar radiation, radiation from the ground, and convection (Muth 1977).We used only four copper models per site (Sam and Bedhiya) as the study sites were relatively homogeneous in terms of microhabitat structure.Within each site, copper models were placed 500m apart.
To evaluate the factors affecting T e , we used a linear model (LM) with T e as the response variable, and microhabitat (open or burrow), site (Sam and Bedhiya) and month (March, April, and May) as predictor variables with microhabitat and month as interaction terms.We used scatter plots to evaluate linearity and Q-Q plots to check for the normality of residuals.

Field Active Body Temperature (T b ) and effectiveness of thermoregulation
To measure the body temperatures (T b ) of lizards in the field, 25 individuals were captured with a lasso at their burrow entrance and were individually marked using a nontoxic felt-tip permanent marker (Johnson, 2005).Mass (Pesola® 500g precision scale, ± 0.5g), snout-vent length (SVL, ± 0.1 cm), sex, and reproductive status if female (using abdominal palpations) were recorded for all individuals.If the captured lizard was a gravid female, we released it at the burrow from which we had initially captured it.Mass and SVL were used to calculate Body Condition Index (BCI; as per Cecchetto et al., 2020;Peig & Green, 2009).
Each lizard was then fitted with a breathable cotton harness that had a temperature data-logger (Maxim Thermochron iButton DS1921G; accuracy ±0.5°C) stitched into it, such that the sensing surface of the logger touched the skin of the lizard.The combined weight of the logger and the harness was <6g (<5% of lizard body weight), and the harness allowed free movement of the lizard (See supporting information S2 for the image of the harness attached to the lizard).The temperature logger was set to measure skin temperature (T b ) at 15-minute intervals with data ranging from 17 to 63 days between March to May 2021.Field body temperatures (T b ) from 19 individuals were successfully obtained and data were checked manually for aberrant readings, which were removed (e.g.consecutive readings with a 10°C difference suggested detachment of the thermal logger).Lizard skin temperature (T b ) was Biology Open • Accepted manuscript used as a proxy for core temperature, which we validated in the lab (See supporting information, S3).For all lizards (n = 19), Voluntary Thermal Maximum T b (VT max ) was calculated as the average maximum daily T b (Ivey et al., 2020, Brattstrom, 1965).
To evaluate the factors affecting T b , we used a linear mixed model (LMM) with field T b as the response variable, and open T e , BCI, and sex as fixed effects.Individual identity was included as a random variable.Scatter plots and Q-Q plots were used to evaluate linearity and normality of residuals respectively.
To quantify the effectiveness of thermoregulation (E), we calculated the thermoregulatory accuracy (d b ) and the thermal quality of the habitat (d e ).We measured db as the mean of deviations of T b from T set (see below for T set methods).If T b was below the T set range, we subtracted T b from the lower bound of T set range.If T b was above the T set range, we subtracted upper bound of T set from T b .We measured de as the mean of deviations of T e from all microhabitats from T set .If T e was below the T set range, we subtracted T e from the lower bound of T set range.If T e was above the T set range, we subtracted upper bound of T set from T e .

Temperature-dependent activity
To evaluate the direct effects of temperature on the activity of the lizard, we used a combination of focal observations, scan surveys and T e .Behavioural observations were conducted from a hide using binoculars from 0700 to 1900 for 10 days (5 days per site) spread over each month (March-May).Overcast and rainy days were avoided.
We used Temperature-based Activity Estimation algorithm (TBAE; Davis et al., 2008;Moore et al., 2018) that evaluated if marked and tagged lizards (n = 19) were active or inactive, based on the absolute difference between body (T b ) and operative (T e ) temperatures.
If T b was closer to T e of open environment than that of burrow, then the state was classified as 'active', and vice-versa (Moore et al., 2018).Scan sampling were carried out to validate these results, wherein, we recorded the activity of marked and tagged individuals (n = 19) as "active" (in the open) or "inactive" (inside burrow) until the lizard was out-of-sight (Altmann, 1974).We compared the activity budget created by the above algorithm with that obtained from focal observations, and the accuracy of the estimated activity budgets was assessed using Cohen's Kappa test (Davis et al., 2008).

Biology Open • Accepted manuscript
To assess the temperature dependence of foraging activity in the wild, we conducted scan sampling of lizards from 10 active S. hardwickii burrows that were randomly chosen and marked with flags.Scan sampling of lizards emerging from these burrows was carried out at 15-minute intervals from 0700 hrs to 1900 hrs, and proportion of individuals foraging at any scan were computed.We used generalized additive model (GAM) to predict foraging activity as a function of Open T e using the package 'mgcv' (Wood, 2015) in R 4.1.0(R core team, 2020) as the expected relationship between temperature and activity was non-linear and unimodal.We also calculated the range of T b at which the proportion of individuals foraging is greater than or equal to 95% of maximum proportion of individuals foraging (F 95 ).

Housing conditions in the lab
Measures of preferred body temperature (T set ), Critical thermal maxima (CT max ) and minima (CT min ), and locomotor performance were conducted at the field station.For these measurements, adult lizards within the size range of 14.5 to 24.2 cm SVL were captured from the wild and housed in individual terraria (62x62x30 cm) that had a 150W heating lamp and artificial refugia made of untreated wood.During captivity, lizards were provided with ad libitum access to food (fresh and washed Dactyloctenium aegyptium leaves).All experiments described below were carried out in a cool room (22.5°C -26°C) between 0800 to 1900 after acclimation for 24 hours.The order of measurements was the following.On the first day, we conducted the T set experiment and obtained two measures of sprint speed at different temperatures separated by a minimum of 4 hours of recovery (Angilletta et al., 2002).On the second day, two additional measures of sprint speed at different temperatures were taken and CT min was measured.On the last day, one measure of sprint speed was performed, and CT max was measured at the end.Due to permit restrictions for this protected species, the sample sizes for each measurement were limited to the minimum.To reduce the effects of fatigue, not all captured lizards (N = 23) were utilised for both T set and critical thermal measurements.
Capture was staggered such that lizards spent no more than 4 days in captivity.To ensure that time in captivity and the experiments did not adversely affect the performance of lizards, we measured the mass of all lizard before and after the experiments and found no significant changes (T(44) = -0.03,P = 0.97).Thereafter, lizards were released near their respective burrows.

Preferred Body Temperature (T set )
To measure T set , we created a thermal gradient in the lab which comprised of a long and narrow wooden frame (300×25×25 cm) with a sandpaper substrate, and a 150W infrared lamp at one end, which generated a temperature gradient from 25°C to 52°C.Each lizard (N = 7 female, 10 male) was introduced individually to the gradient and allowed to acclimatize for two hours, after which, body (cloacal) temperatures were recorded every 5 minutes for an hour.Cloacal temperature was measured using a temperature logger (Amprobe model TMD-50) that was connected to a k-type thermocouple which was inserted 1 cm into the cloaca of the lizard and kept in place using surgical tape.Temperatures were logged automatically.T set was considered as the 25-75% interquartile range of the cloacal temperatures of the last hour of the 3-hour trial.We excluded T set data for the lizards which had their T set >2 SD away from the mean of the median T set of all the lizards, or those who had failed to move and actively thermoregulate within the gradient (Ivey et al., 2020).A linear model (LM) was used to examine the effects of BCI and sex on T set .We ensured the normality of residuals using the Shapiro-Wilk test (w = 0.44, P = 0.94) and linearity using residuals vs fitted plots.

Critical thermal maxima (CT max ) and minima (CT min )
We measured CT max and CT min on 15 individuals (N = 7 female, 8 male).CT max was the body temperature at which onset of muscle spasms was observed and CT min was the body temperature at which loss of righting response (ability to turn over when placed on its back) was observed .During the trials, cloacal temperature was measured using a temperature logger (Amprobe model TMD-50) that was connected to a k-type thermocouple which was inserted 1 cm into the cloaca of the lizard and kept in place using surgical tape.The initial body temperature of the lizards before the experiments was ~25°C.To determine CT max , each lizard was placed in a metal container with a 150W infrared lamp 40 cm above it, which resulted in a heating rate of approximately 1°C per min for the lizard.Cloacal temperatures and behaviour were constantly monitored, and CT max was recorded as the cloacal temperature at which muscle spasms were first observed (Lutterschmidt & Hutchison, 1997a, 1997b).
Once lizards reached CT max, they were immediately removed from the container and transferred to a pre-cooled chamber to assist in lowering their body temperature.To determine CT min , each lizard was placed in a terrarium in a room at 25°C.Using icepacks, the body temperature of the lizard was lowered at a rate of approximately 0.5°C min -1 .We started testing the righting response of the lizard when the terrarium temperature reached 8°C.Righting response was tested every 15 seconds until the lizard could not right itself Biology Open • Accepted manuscript when flipped on its back (Labra & Bozinovic, 2002;Laspiur et al., 2021).The terrarium temperatures were reduced to as low as 2°C.When CT min was reached, the lizard was placed at room temperature.No lizard suffered mortality during these experiments, and all resumed normal activity (movement and feeding) within 2-5 hours after these experiments.We used linear models (LM) to statistically determine the effects of BCI and sex on CT max and CT min .
We ensured the normality of residuals using the Shapiro-Wilk test (CT min ; w = 0.91, P = 0.14, CT max ; w = 0.95, P = 0.66) and linearity using residuals vs fitted plots.

Locomotor performance
Locomotor performance for each lizard (N = 12 female, 11 male) was measured at five body temperatures between 25°C, and 45°C, with approx.5°C intervals.Body temperatures above 34°C were achieved using the method described for CT max , and a water bath with icepacks was used to reduce body temperature of lizards.Each lizard was kept at the selected temperature for 1 hour before introducing it to a racetrack (250×25×25 cm).The racetrack was made of a wooden frame lined with 260-grit sandpaper to emulate the most common substrate in the wild -fine gravel (Ramesh & Ishwar, 2008).Each lizard was made to run two sprint trials per day separated by > 4 hours.The trial run consisted of five consecutive 0.5m sections, with a single initial stimulus (touching their rear thigh with a finger).This was repeated three times, separated by five-minute rest.Nikon D7000 camera with Tokina 11-16mm lens was used to record all trials at 25 frames per second.All videos were analysed in Avidemux 2.7.8 and the fastest run for each lizard was used for the analysis.
A trial was not included in the analysis if the lizard failed to move.
Optimum temperature (T opt ) was considered to be the T b (treatment body temperatures of the lizards) at which the lizard's speed was maximum (P max ).Performance breath (B 95 ) was defined as the range of T b at which the performance is greater than or equal to 95% of P max (Cecchetto et al., 2020).To calculate these variables, generalized additive mixed models (GAMMs) were fitted to the non-linear data using the package 'mgcv' (Wood, 2015) in R 4.1.0(R core team, 2020) where T b , BCI, sex and site were considered as fixed effects, after accounting for the random variation between individuals which was expected to be significant (Artacho et al., 2013;Cecchetto et al., 2020).We included CT max and CT min data to anchor values for body temperatures, at which sprint speed is 0 m/s (similar to Cecchetto et al., 2020).A coARMA correlation structure with 2 knots were used to construct the curve.

Projecting the impact of climate warming
We projected the impact of climate warming on hours of restriction (h r ), locomotor performance and foraging activity.We used projections based on three future pathways of societal development -Shared Socioeconomic Pathways (SSPs): 1) SSP1-2.6where there would be an increase of 1°C by 2099 (+0.011°C /year; 2099 represented by the time slice 2080-2099); 2) SSP2-4.5 where there would be an increase of 3°C by 2099 (+0.036°C /year) and 3) SSP5-8.5 where there would be an increase of 4°C by 2099 (+0.047°C /year).These SSPs, align with the projections for Jaisalmer based on the ensemble of 27 Coupled Model Intercomparison Project phase 6 (CMIP6) available for temperature (https://esgfnode.llnl.gov/search/cmip6).Before the ensemble, models were re-gridded to 1° × 1° resolution from their original resolution using a bilinear interpolation method (Almazroui et al., 2020).The projected increments in temperature were added to the T e data.
Activity restriction in different microhabitats (open and burrow) was projected using CT max , VT max , B 95 and T set as different thresholds for activity.The number of hours above the different thermal thresholds in each microhabitat were considered as the hours of restriction (h r ), and we projected the h r for each threshold variable by adding projected increase in temperature to current biophysical model data unilaterally.It is likely that different microhabitats might have different rates of temperature increase but this model will provide a broad estimate of how h r might change with climate warming (Brusch et al., 2016).
To project the future reductions in foraging opportunity, the current maximum proportion of individuals foraging (based on the scan survey data) at the associated T e was set to 100%.
By adding the projected temperature increases to the curve of foraging temperaturedependence (constructed using GAM), we calculated the potential decline in foraging opportunity in the future.These predictions are restricted to summer when the temperatures are the highest in the desert ecosystem of Jaisalmer.To project changes in locomotor performance in warming conditions, current performance at optimal temperature (P max ) was set to 100%, and increases in temperature on the thermal performance curve (constructed using GAMM) were used to calculate potential relative decrease in future.
The mean thermoregulatory accuracy (d b ) for all T b data was 3.45 ± 0.01 (ranging from 0 to 20.2), and the mean thermal quality (d e ) of all microhabitats combined was 7.54 ± 0.03 (ranging from 0 to 34.9).Overall effectiveness of thermoregulation (E) of S. hardwickii was 0.54, suggesting that this species is a moderately precise thermoregulator.We found that extent of thermoregulation was highest in May (E = 0.6) followed by April (E = 0.56), and March (E = 0.44) indicating a higher effectiveness of thermoregulation during the hotter months.

Temperature-dependent activity
The results of TBAE showed substantial agreement with our test data (82.7%;K = 0.6; P<0.001).Using TBAE, the proportion of activity by lizards was found to be 6% higher in March compared to April and May.Total proportion of activity was the same during April and May, with a greater proportion of lizards being active in the mornings and evenings compared to the afternoon or night during both those months (See supporting information, Figure S4).
A temperature-dependent activity rate curve based on the scan surveys (N = 1127) and T e in the open microhabitat showed that lizards were active and outside their burrows between the operative temperatures of 31.2°C and 48°C.Activity was highest at 39.6°C

Biology Open • Accepted manuscript
which is only 0.2°C less than the T opt and within the T set range (See supporting information, Figure S5).Foraging in these field sites (N = 258 observations) occurred between 34.1°C to 48.6°C with the highest proportion of individuals observed foraging at 41.7°C (Figure .2).
The range of T b at which the proportion of individuals foraging is greater than or equal to 95% of maximum proportion of individuals foraging (F 95 ) was 40.7°C to 43°C.

Preferred temperature (T set )
Preferred temperature (T set ) of S. hardwickii, from measurements in a thermal gradient in the lab was 39.2°C ± 0.3°C SE and the T set range was 38.2°C ± 0.3 SE -40°C ± 0.4°C SE.There was no significant effect of sex (F(1,14)= 0.04, P = 0.84) or BCI (F(1.14)= 1.19,P = on T set .

Locomotor performance
As expected, the thermal performance curve showed a significant effect of temperature on sprint speed of S. hardwickii (GAMM: F(6.27,132.81)= 73.73,P<0.001; Figure 3).The GAMM with temperature as a fixed effect, 'individual' as random effect, BCI, sex and site as covariates explained 81.5% of the deviance.The effect of BCI (β = 0.004, Biology Open • Accepted manuscript thermal limits (T set , B 95, VT max and CT max ) for 6 hours of the active period of the day (07:00 -20:00) in the hottest month of the year (May), during which S. hardwickii is restricted to their burrows (Figure 4).

Projecting the impact of climate warming
Projection models for h r considering the CT max limit (49 °C), showed that there might We did not use CT min for our predictions since the difference between the average daily minimum T b and CT min were large (14.8°C).

Discussion
Many animals have physiological, behavioral, or life history strategies that could mitigate the impacts of climate change on fitness (Hoffmann & Sgrò, 2011), yet change in global temperature is still expected to be a challenge (Parmesan, 2006).Impacts of climate change are thought to be even more pronounced for ectotherms surviving in thermally extreme areas, such as some deserts, where species are already thermally constrained and where thermal buffers for behavioral thermoregulation, such as vegetation, are limited (Grant , 1988).In our study on the thermal ecology of S. hardwickii, we integrated labbased information on thermal responses and limits with field measures of thermal ecology and behavior to understand how this desert ectotherm thermoregulates and the way in which it may respond to rising temperatures.Our results support the premise that different physiological and behavioural functions have different thermal optima.Congruent with our broad expectation, we find that projected temperature increases will significantly restrict lizards to the thermal refuges of burrows and substantially decrease both locomotor performance and foraging activity by the end of the century.
hardwickii is a moderately effective thermoregulator and uses burrows and open microhabitats to thermoregulate to maintain their average hourly T b within their T set and/or B 95 range.We also saw shifts in effectiveness of thermoregulation (E) as the monthly temperatures increased, suggesting an increase in energy utilized for thermoregulation as the months progressed.As a result, we find that the lizards in these study areas are already restricted to their burrows for 6-9 hours out of 13 daylight hours during the hottest month (May) of the year.Burrows provide an excellent buffer for lizards from temperatures in the open (Heatwole, 1970;Ivey et al., 2020;Moore et al., 2018), and for S. hardwickii in the Thar desert where vegetation is scant and provide little to no shade, burrows are an essential thermal buffer against extreme temperatures throughout the day.We found that S. hardwickii usually emerged out of their burrows after 0800 hrs and were mostly in the open until their body temperature (T b ) reached preferred temperature (T set ).After approximately 1100 hrs, when T e open exceeded VT max , body temperature (T b ) of lizards decreased, which signalled either complete restriction to burrows or higher rates of shuttling between the open microhabitat and burrows (Cowles & Bogert, 1944;pers obs).Saara hardwickii might also be thermoregulating by adjusting body posture (Muth, 1977) or through changes in skin colouration (Dutta & Jhala, 2007) which affects reflectivity and radiant heat gain (Stuart-Fox et al., 2017).Although S. hardwickii shows active thermoregulation to avoid extreme body temperatures, the narrow range of body temperatures that we have measured in the wild will have considerable influence on physiological and behavioural performance.
Effective locomotory performance and foraging behaviour are critical to the survival of species, and both are influenced by temperature.Thermal performance measures of S.
hardwickii indicate their highest sprint speeds at 39.2°C, which is also close to the preferred Biology Open • Accepted manuscript active temperature for this species in the wild.Fastest sprint speed during the highest activity phase that we document here would be beneficial for effective escape from both terrestrial and avian predators, especially since lizards seldom range far from their burrows.High performance capacity, however, might not imply high activity rates for other behaviours (Gunderson & Leal, 2016).While the total activity of S. hardwickii in the wild was highest near its optimal performance temperature (T opt ), as measured by sprint speed in the lab, and was within its thermal preference range (T set ), the highest number of individuals were seen foraging in the wild when temperatures were 2.5°C higher than T opt .Also, lizards were actively foraging within a narrower temperature range and had higher body temperatures while foraging than when at peak locomotor performance.One possible explanation for these differences in thermal sensitivity of foraging and locomotion is diet.Adults of S. hardwickii maintain a primarily herbivorous diet (Dutta & Jhala, 2007) and digestion of plant material is more efficient at higher temperatures (see Qu et al., 2011;Huey, 1982).Uromastyx aegyptei, a lizard closely related to S. hardwickii, does not chew its food and depends entirely on microbial fermentation for digestion (Foley et al., 1992).Similarly, during our scan surveys, we observed that S. hardwickii also does not chew, although information about digestive processes remains unknown.For herbivorous lizards, such as S. hardwickii, the possible benefits of high temperature may be negated if increases in environmental temperature forces body temperatures to be even closer to their upper thermal thresholds, resulting in a risk of overheating and a reduction in available activity hours.
We projected the impacts of rising temperatures on S. hardwickii using a threepronged approach.Firstly, we used the 'activity restriction model' or 'threshold model' posited by Sinervo et al. (2010) to predict the number of hours for which this species would be restricted to their burrows under different climate change scenarios.Our activity restriction models predict that, as temperatures rise, S. hardwickii will continue to lose hours of activity during its active months because operative temperatures (T e ) in the open environment will exceed preferred temperature (T set ) for the species.Further, T e in the open environment will also exceed B 95 (the range of T b at which locomotory performance is greater than or equal to 95% of lizard's maximum speed), and both Voluntary Thermal Maxima (VT max ) in the field and Critical Thermal Maxima (CT max ).Notably, even in the worst-case climate warming scenario, spiny-tailed lizard burrows in the Thar desert do not seem to exceed thermal thresholds and may be the only microhabitat available to the species as a thermal refuge.In fact, our measure of the thermal refugia in burrows may be an underestimate as we placed copper models only up to 1m inside burrows, but temperatures could be even cooler deeper Biology Open • Accepted manuscript within burrows.We included T set and B 95 in our model because outside the critical threshold ranges (CT max ), preference or performance are compromised and may also affect survival probability.This conservative measure allows us to plan conservation efforts before critical thermal limits are reached.During our observations of lizard activity, we recorded lizards using burrow entrances for territory defense and foraging.However, the projected levels of warming might restrict lizards to deeper parts of the burrow for more hours, further reducing their time to forage, defend territories, and obtain mating opportunities which occur in the open environment.Further studies exploring the thermal gradient of the species' burrows are required, but since burrows are not used for active foraging or mating, the climate-induced increases in burrow use will still come at a cost for the species in this landscape.
Rising temperatures will not only influence the number of hours that are thermally suitable for ectothermic species but will affect body performance and temperature-sensitive activity.Consistent with our expectation, we predicted a notable reduction in locomotor performance of S. hardwickii under global warming conditions.Reduction in sprint speed that range from 0.8% to 14.5% by 2099, depending on the climate projection model, might have a significant negative impact on the species' ability to evade predators during their active hours (Ekner-Grzyb et al., 2013;Husak, 2006).Physiological thresholds and performance do not necessarily govern the activity of all individuals, and thus we expect some variation in the number of individuals that are active, depending on environmental temperatures as well as on individual variation (Gunderson & Leal, 2016).Since this study considers the impacts of climate change through the lens of the Bogert effect, our projections, like many others, do not take into account genetic adaption or plasticity of thermal thresholds, which may play an important role increasing the odds of survival for S. hardwickii ( Gunderson & Leal, 2012).
These projections also do not incorporate potential changes in the forage availability or vegetation structure in the environment.When we modelled changes in the proportion of individuals foraging solely as a function of environmental temperature, we found that the reduction in foraging activity was even more dramatic than the reductions in locomotor performance.In the worse-case global warming scenario, the number of individuals foraging is projected to reduce by 43.5%, which might prove to be devastating for an herbivorous lizard in terms of survival.If these temperature-induced reductions in foraging activity are coupled with temperature-induced reduction in forage availability, S. hardwickii becomes even more vulnerable.
Many studies have used physiological parameters, such as thermal thresholds and locomotor performance, to predict the impacts of climate change on ectotherms (e.g.

Biology Open • Accepted manuscript
Gunderson & Leal, 2012;Huey et al., 2009;Ivey et al., 2020;Logan et al., 2013;Wilms et al., 2011).It is also well known that ectotherms can behaviourally thermoregulate to optimize their performance within an optimum temperature range in the wild, despite heterogeneity and change in environmental conditions.Active behavioural regulation can, in fact, constrain rather than drive evolution of thermal traits (i.e., "Bogert effect"; Huey et al., 2003).
However, when the temperature range for activity is narrower than the physiological limits, as we see for S. hardwickii, using physiological constraints and not accounting for activity in the wild might underestimate the impacts of climate change (Gunderson & Leal, 2016).
Using activity in the wild could also overestimate the impacts of climate change as behavior is more plastic than thermal thresholds.However, we found the thermal optima of locomotory performance and the operative temperature at which total activity in the wild is highest is very similar.This suggests that even though the thresholds for activity in the wild might be more plastic, the decline in activity from optimum when environmental temperature increases are still expected.Overall, key data on thermal ecology with information about temperature-sensitive behaviours in the wild is necessary to understand the ways in which increasing temperatures may influence ectothermic vertebrates, especially for vulnerable species such as S. hardwickii that need climate-aware conservation plans and policies for their continued protection.
001) and month (F(1,15483) = 909.2,P<0.001) on T e , with no significant interaction between the Biology Open • Accepted manuscript two variables (F(2,15483) = 2.1, P = 0.14).Above ground environmental temperatures in the open (T e open; β = 36.4,T = 310.6,P<0.001) were significantly higher than temperatures in the burrow (T e burrow; β = 31.8,T = 356.4,P<0.001).T e open at the two study sites were not significantly different (F(1,1266) = 0.1, P = 0.75).Mean environmental temperature (T e open) in March was 33.3°C ± 0.2°C SE compared to 36.4°C ± 0.2°C SE in April and 38.9 °C ± 0.2°C SE in May.Field Active Body Temperature (T b ) and Effectiveness of Thermoregulation (E) A total of 65952 field active body temperature (T b ) data points from 19 individuals were obtained.In March, the mean field T b was 33.6°C ± 0.03°C SE compared to 35.2°C ± 0.01°C SE in April and 36.5°C± 0.01 May.The effect of T e open on field T b was significant be an additional one-hour restriction in activity due to climate warming in the intermediate-(SSP2-4.5) case scenario and a two-hour activity restriction in the worst-(SSP5-8.5)case scenario (Figure5).Projection models for h r , considering the VT max limit (46.3 °C), showed that there might be an additional two-hour activity restriction in the best-(SSP1-2.6)case and intermediate-(SSP2-4.5) case scenarios and an additional three-hour restriction in the worst-(SSP5-8.5)case scenario (Figure5).Considering the upper bound of B 95 range of sprint speed (41.5 °C), projection models for h r showed that there might be an additional one-hour activity restriction in the best-(SSP1-2.6)case and intermediate-(SSP2-4.5) case scenarios and an additional two-hour restriction in the worst-(SSP5-8.5)case scenario (Figure5).Given the upper bound of T set range (40 °C), h r projections suggest an additional one-hour activity restriction in the best-(SSP1-2.6)case scenario, and an additional two-hour restriction in intermediate-(SSP2-4.5) case and worst-(SSP5-8.5)case scenarios (Figure5).