Habitat heterogeneity affects the thermal ecology of an endangered lizard

Abstract Global climate change is already contributing to the extirpation of numerous species worldwide, and sensitive species will continue to face challenges associated with rising temperatures throughout this century and beyond. It is especially important to evaluate the thermal ecology of endangered ectotherm species now so that mitigation measures can be taken as early as possible. A recent study of the thermal ecology of the federally endangered Blunt‐nosed Leopard Lizard (Gambelia sila) suggested that they face major activity restrictions due to thermal constraints in their desert habitat, but that large shade‐providing shrubs act as thermal buffers to allow them to maintain surface activity without overheating. We replicated this study and also included a population of G. sila with no access to large shrubs to facilitate comparison of the thermal ecology of G. sila populations in shrubless and shrubbed sites. We found that G. sila without access to shrubs spent more time sheltering inside rodent burrows than lizards with access to shrubs, especially during the hot summer months. Lizards from a shrubbed site had higher midday body temperatures and therefore poorer thermoregulatory accuracy than G. sila from a shrubless site, suggesting that greater surface activity may represent a thermoregulatory trade‐off for G. sila. Lizards at both sites are currently constrained from using open, sunny microhabitats for much of the day during their short active seasons, and our projections suggest that climate change will exacerbate these restrictions and force G. sila to use rodent burrows for shelter even more than they do now, especially at sites without access to shrubs. The continued management of shrubs and of burrowing rodents at G. sila sites is therefore essential to the survival of this endangered species.


| INTRODUC TI ON
Many organisms are threatened by the projected increase in global temperatures. As ectotherms, reptiles are disproportionately threatened because their body temperatures are dependent on the temperatures of their environment (Aragón et al., 2010). Models estimate that nearly 40% of lizard populations may be extirpated by 2080 (Sinervo et al., 2010), and heliothermic (sun-basking) lizards occupying the hottest habitats on the planet could be at particularly high risk because temperatures are already so high. Field observations of microhabitat use paired with comparisons of animals' field-active and preferred body temperatures to the available microhabitat temperatures can give insight into how an animal uses its thermal landscape (Burrow et al., 2001;Fawcett et al., 2019;Taylor et al., 2021).
Such data can also be used to calculate the population's hours of restriction, or the number of hours per day that temperatures in certain microhabitats exceed the animal's preferred body temperature or their upper thermal tolerance and are therefore undesirable or unavailable for use. This information can be used to identify thermal and ecological parameters that may help conserve threatened reptiles and their communities. For example, shrubs and other vegetation are important contributors to the habitat heterogeneity that provides a mosaic of temperatures for effective thermoregulation by lizards (Basson et al., 2017;Goller et al., 2014), suggesting that shrubs may help buffer reptiles from climate change.
The Blunt-nosed Leopard Lizard (Gambelia sila) (Figure 1a) is an ectotherm that has been listed as federally endangered since 1967 because almost 90% of the species' historical range has been converted into uninhabitable agricultural fields (U.S. Fish & Wildlife Service, 1998). Once ranging across the vast San Joaquin or California Central Valley, G. sila are now restricted to a few small patches of relatively undisturbed San Joaquin Desert habitat. These heliothermic lizards are adapted to the very hot and dry California San Joaquin Desert ecosystem, where already high temperatures are becoming even more extreme (Germano et al., 2011;Ivey et al., 2020). Adult G. sila are primarily only active for a quarter of the year (late April through mid-July) (Germano & Williams, 2005;Montanucci, 1965), during which time they experience high environmental temperatures (Ivey et al., 2020). They feed and breed in this short window, using Giant Kangaroo Rat (Dipodomys ingens) burrows for shelter at night and during the heat of the day , then entirely retreat into the burrows for most of the remaining nine months of the year. Lizards in many populations, but not all, associate with desert shrubs, including the large gymnosperm shrub Ephedra californica. Ephedra californica is a foundation species in the San Joaquin Desert community  and facilitates the presence of community members, including G. sila Lortie et al., 2017;Westphal et al., 2018), which select for shrubs at fine spatial scales (Germano & Rathbun, 2016).
Until recently, technological constraints have prevented researchers from collecting the continuous body temperature data necessary for studying the thermal ecology of a species such as G. sila. Advances in miniaturization and technology of radiotelemetry transmitters now allow for ample data collection on physiological aspects of small animals (Weaver et al., 2021). Ivey et al. (2020) studied the thermal ecology of G. sila at a single site with abundant shrubs in 2018 and found that shrubs appear to serve as an important thermal buffer from the heat of the sun, potentially allowing G. sila to remain aboveground instead of retreating underground where they would be unable to perform necessary daily activities (Ivey et al., 2020;Westphal et al., 2018). To further test this hypothesis, we studied G. sila in 2019 at the same site as Ivey et al. (2020), hereafter called Shrubbed, and added a second nearby site where G. sila had virtually no access to shrubs (Shrubless). This allowed us to further assess the importance of shrubs for thermoregulating G. sila that were experiencing otherwise similar environmental conditions, and therefore understand how important shrubs may be in ensuring this endangered species' survival. If shrubs provide a thermoregulatory benefit to G. sila, then lizards with access to shrubs should be active aboveground longer and use rodent burrows less often during the day than lizards without access to shrubs. Additionally, lizards with access to shrubs should thermoregulate more accurately (i.e., fieldactive body temperatures closer to preferred body temperatures; Hertz et al., 1993) than lizards without access to shrubs, and should have fewer hours of restriction currently and in modeled future scenarios when ambient temperatures will rise. Identifying aspects of the environment, such as shrubs, that may help G. sila thermoregulate more efficiently is important for informing management efforts to protect this species and other sensitive San Joaquin Desert species from rising temperatures in some of the hottest, driest parts of the continent.

| Field sites and study species
A pair of sites, one dominated by E. californica and other smaller perennial shrubs (hereafter named Shrubbed), and the other with no E. californica and very few other shrubs (Shrubless), were selected on the Elkhorn Plain within the Carrizo Plain National Monument in California, USA (Figure 1b). Shrubless was selected because lizards had been seen in the area previously, and it was only 6.5 km away from Shrubbed where we have previously collected data.
The sites are similar in size (400 m 2 ), as well as climate and elevation. Microhabitat use and shrub association of lizards at Shrubbed were studied in 2016 (Westphal et al., 2018), and field-active body temperatures of lizards were regularly recorded there in 2018 (Ivey et al., 2020). Gambelia sila at Shrubbed had access to ample shade provided primarily by the aforementioned large E. californica shrubs.
Shade was also available from smaller perennials such as Isocoma acradenia and Gutierrezia californica and small annual plants such as Amsinckia sp. and nonnative grasses. In contrast, lizards at Shrubless had limited access to aboveground shade, which was provided by very few I. acradenia, G. californica, and Astragalus sp. (mostly A. lentiginosus, sometimes A. oxyphysus), in addition to small annual forbs and grasses. Shrubless had only a few individual perennial shrubs in the entire site, and notably, these shrubs were only used by a total of two lizards whose territories happened to overlap with these shrubs. Therefore, the use of shrub-provided shade by lizards at Shrubless was extremely rare (see Results). Lizards at both sites had access to burrows, which were confirmed to be engineered by D. ingens from 5 nights of trapping with 61 traps at each site in August 2020.
Dipodomys ingens were captured at both sites, with very small numbers of D. nitratoides at Shrubbed exclusively.
We captured twenty lizards at each site (N = 40 total) by handheld lasso over the course of three days in late April 2019, and collected the following data for each lizard: sex, reproductive state in females (gravid or not), snout-vent length (SVL, ±0.5 mm), and mass (±1 g). Lizards were fitted with VHF temperature-sensitive

| Microhabitat use
We tracked G. sila using a VHF receiver (R-1000 Telemetry Receiver, Communications Specialists, Inc., Orange, CA, USA) and 3-element designated as underground in a burrow if they were not visible from the burrow mouth; sometimes, lizards sat close to the entrances of burrows, but this was categorized as the open because most of their body, notably the temperature-sensitive radio collar, was in sunlight.
We then calculated the percent of time G. sila used each microhabitat in May, June, and July at each site. To compare the probability that a lizard would be found underground (in burrows) between the two sites, we ran a mixed-effects logistic regression model in R (R Core Team, 2020;RStudio, 2020, lme4 package v. 1.1-26, Bates et al., 2015 with time as a polynomial, site and month as fixed effects, and lizard ID as a random effect. At the end of the active season, we collected data on D. ingens burrow densities at each site by counting the number of active or recently inactive burrows (Bean et al., 2012) within 10 m along four 100-m randomly placed transects at each site. We compared the burrow densities at the Shrubbed and Shrubless sites with Welch's t-test in R. We also collected data on perennial shrub densities by counting the number of perennial shrubs in a 10-m radius around 16 random points (Zuliani et al., 2021) at each site, and compared the densities with Welch's t-test in R.

| Temperature variables
At the center of each site, we installed a stationary 3-m tall solarpowered (Tycon RemotePro 2.5 W Solar Power System with Vikram Solar Eldora 10P solar panel) omni-antenna (Telonics Model RA-6B) and receiver with data acquisition system (Telonics TR-5 Option 320). We estimated the range for continuous, gap-free data collection with this antenna to be approximately 300 m. About every five minutes, the receiver logged the interpulse interval of the signal from each radio collar in range, and we downloaded these data from the receivers each week. Because the radiotransmitters were externally attached to the lizards, T b values may represent an overestimation of core T b because they can heat rapidly from solar radiation; however, surgical implantation of radiotransmitters is not possible in an endangered species such as G. sila. We used manufacturer-provided calibration curves and the program Vinny Graphics v2.07 to convert the interpulse intervals to field-active body surface temperatures, which act as estimates of lizard body temperature (T b ). Prior to analysis, we removed any outliers greater than two standard deviations away from each lizard's mean T b , as these likely represented glitches in the data acquisition system; such outliers were uncommon (<5% of data points).
To collect data on the environmental temperatures of the three available microhabitats to these lizards for the entirety of the study, we deployed lizard physical models in sunlight, in the shade, and inside burrows, using the same models as Ivey et al. (2020).
Models consisted of copper pipes (2.5 cm diameter and 12 cm long) capped with PVC and spray-painted matte gray and matte tan to resemble the color of the lizards' skin. Models that were placed under shrubs and in the open were given two "legs" in the form of metal wire looped around the pipes so they could be propped up to resemble G. sila resting posture. Each model housed a Thermochron iButton (DS1921G-F5) programmed to record temperature every hour, on the hour. While empty models provide instantaneous operative temperature, we chose to fill the models with water to mimic a body cavity (Dzialowski, 2005) and to replicate the exact methods of Ivey et al. (2020); we also added plumber's tape before screwing on the caps to maintain watertight seals. We placed the models haphazardly at each site (Shrubbed:

| Preferred body temperature and thermoregulatory accuracy
As G. sila aestivation approached in mid-July, we recaptured and reprocessed each lizard and removed their collars. Before returning each lizard to its capture site, we collected data on its preferred body temperature (T set ) in a thermal gradient as described in Ivey et al. (2020). The gradient consisted of 3 lanes (250 × 20 × 25 cm) filled with sand substrate and separated by wood dividers, ranging from 47°C at the hot end to 10°C at the cool end. Three G. sila were placed into the center of the gradient at a time, each in its own lane, with thermocouples (Model 5SRTC-TT-K-40-72; Omega Engineering, UK) in their cloacae recording body temperature every 10 min for three hours. These data were recorded on a data logger (Model RDXL4SD; Omega Engineering, Egham, Surrey, UK), and only the last hour of data was used for analysis.
We calculated average T set for each of the two populations after removing outliers greater than 2 standard deviations away from each lizard's mean, and we used the interquartile range (IQR) of each population as its T set range. Since there was no significant difference in T set between the two populations (see Results), we used the mean T set IQR of all lizards for the following analyses. We calculated lizard thermoregulatory accuracy (d b ) by subtracting the mean T set IQR from each instance of T b in the field (Hertz et al., 1993). When T b fell within T set IQR, d b was zero. Either very high positive or very low negative values of d b represented poor thermoregulatory accuracy because the field-active T b was higher or lower than T set range. Lizard 14.3, 2018).

| Hours of restriction and climatic projections
We compared temperatures from the physical models (T e ) to G. sila T set and T pant each hour of the day for each month to calculate the number of hours in a day that a given microhabitat would be thermally stressful (i.e., exceed either T set or T pant ) for a lizard. We designated hours of restriction as "basking restriction" when temperatures in open sunlight were too hot and lizards therefore must remain in shade or in burrows; "aboveground restriction" when temperatures in the open and shade of large shrubs were too hot and lizards therefore must retreat to burrows (this is only applicable for lizards at Shrubbed); and "total restriction" when all three microhabitats including burrows were too hot (Ivey et al., 2020).
Each of these hours of restriction variables was then recalculated by adding 1°C and 2°C to the T e values for each microhabitat,  Although some lizards at Shrubless found some shade from sparse annual plants and shrubs, they collectively spent very little time in the shade throughout the active season because shade was largely unavailable. In June and July, lizards from Shrubless spent 46% and 57% of their observed time, respectively, inside burrows, compared with 31% and 43% for lizards at Shrubbed. The probability that lizards at Shrubless would be found underground in D. ingens burrows instead of aboveground was higher than that for lizards at Shrubbed (z = 4.35, p < .001) throughout the season. Burrow density was not significantly different between the two sites (Shrubbed: 35.83 ± 4.71 burrows/100 m, Shrubless: 44.67 ± 6.26 burrows/100 m; t = −1.36, p = .23). Lizards at both sites most likely spent all their time in burrows at night.

| Microhabitat use
The woody perennial shrubs most commonly used for shade by lizards at Shrubbed were G. californica, followed by I. acradenia, E. californica, and unidentifiable dead small shrubs, which were likely either I. acradenia or G. californica. One individual had access to and used E. fasciculatum (Figure 2b). In May, when annuals were plentiful, lizards at Shrubbed used the shade of Amsinckia sp. 52% of the time they were in shade, and this decreased to 19% and 10% in June and July, respectively, when lizards started using woody shrubs more often for shade (Figure 2b). Lizards at both sites also used annual or perennial Astragalus sp., as well as the annual forb E. gracillimum and nonnative grasses (primarily Schismus sp. and Bromus sp.) for shade.

| Thermoregulation
The mean T set for G. sila at Shrubbed was 34.1°C with IQR of  Currently, T e inside burrows at both sites never exceeds T set or T pant , and T e under shrubs at Shrubbed never exceeds T pant .

| Hours of restriction and climatic projections
As expected, adding 1°C and 2°C to the T e data resulted in additional projected hours of restriction associated with climate change for both populations in June and July ( Figure 5). Lizards at Shrubbed will be restricted from basking in sunlight and thus staying within their T set range for 9-10 daylight hours, and lizards at Shrubless will be similarly restricted for 8-11 hr. Notably, lizards at Shrubbed should still be able to stay aboveground for several hours under future climate change scenarios because of their access to the shade of a shrub, while lizards at Shrubless do not have this option. Even

| D ISCUSS I ON
As predicted, we found that G. sila that had access to shrubs spent more time aboveground than those that did not, as lizards at Shrubless spent more time inside D. ingens burrows. However, unexpectedly, the presence of shrubs did not give G. sila higher thermoregulatory accuracy. This was mainly because staying inside burrows for longer periods of time actually allowed lizards to remain closer to their preferred body temperature, suggesting a trade-off between thermoregulation and activity aboveground. There was no difference in D. ingens burrow density between the two sites, indicating that the higher frequency of burrow use by G. sila at Shrubless was not the result of more available burrows. Instead, lizards at Shrubless Like Ivey et al. (2020), we found that G. sila will be further constrained from being active aboveground under future climate change scenarios, with temperatures undesirable (above preferred) or unlivable (above thermal maximum) for many hours per day. This constraint, however, is mitigated by shrubs, as lizards with access to shrubs could remain aboveground for several hours longer than lizards with no such access. Taken together, our study shows that shrubs are important in buffering G. sila from the effects of high temperatures, but D. ingens burrows remain the most essential refugia from high temperatures both now and in the future.

| Microhabitat use-activity aboveground
The presence of shrubs allowed G. sila to spend more time aboveground, potentially enabling them to continue patrolling for mates, looking for prey, or engaging in other activities. Although it is unknown whether G. sila can hunt and/or mate underground, typically F I G U R E 4 Thermoregulatory accuracy (d b ) of Gambelia sila at a Shrubbed site (orange) and a Shrubless site (blue) during daylight hours over the course of their 3-month primary active season in 2019, with gray shading representing 1 SEM. The line at zero represents lizards thermoregulating within T set ; positive values mean that lizards are thermoregulating above the upper bound of their T set range; negative values mean that lizards are thermoregulating below the lower bound of their T set range. During the hottest months of June and July, lizards from Shrubbed had poorer thermoregulatory accuracy than lizards from Shrubless F I G U R E 5 Hours of restriction from using specific microhabitats for Gambelia sila at a Shrubbed Site and a Shrubless Site over the course of their 3-month primary active season in 2019, calculated as the number of daylight hours in which microhabitat operative temperature T e exceeds T set (orange) or T pant (green). Current data show estimates from 2019, and +1°C and +2°C data model increases in temperature due to climate change. In general, lizards at Shrubbed experienced about one more hour of restriction than lizards at Shrubless Shrubbed Shrubless May June July heliothermic, diurnal lizards conduct the majority of these behaviors aboveground. A critically endangered lizard in Australia, the Pygmy blue-tongue lizard (Tiliqua adelaidensis) spends the majority of its time underground inside burrows but still needs to exit its burrow to feed (Milne et al., 2003). This lizard has likely evaded extinction thus far due to the tolerable temperatures inside burrows, and artificially added burrows have increased their density (Souter et al., 2004). Burrows constitute crucial thermal refugia for other lizard species inhabiting hot, arid regions worldwide, and their importance is even more critical as temperatures rise (Fenner et al., 2012;Grillet et al., 2010;Moore et al., 2018). Models suggest that lizards will need to go deep into burrows to deal with climate change (Kearney & Porter, 2020). However, aboveground shade may also be critical to facilitate feeding, mating, and other behaviors in species such as G. sila. Crotaphytid lizards hunt their prey using visual cues and lack lingually mediated prey chemical discrimination (Cooper et al., 1996), suggesting that most hunting indeed occurs aboveground. Male crotaphytids rely on bright mating coloration to find mates (Baird, 2004), with chemosensory cues from femoral secretions appearing to play secondary roles such as permitting female assessment of male quality (Baird et al., 2015). Shrubs may therefore play a critical role in allowing G. sila to hunt, find, and court mates, and defend territories, especially as temperatures in the open continue to rise. In our study, we did not examine whether there were consequences for the lizards spending less time aboveground at Shrubless in terms of hunting success or fitness. Such a study would further elucidate the importance of shrubs in allowing aboveground activity in G. sila.

| Microhabitat use-shade
As the season progressed and the temperatures rose, the importance of shade increased for G. sila at both sites ( Figure 2a). Lizards mostly used annuals early in the season when annual cover was thick and then used perennials more often as time went on (Figure 2b).
Dense grasses reduce locomotion speed in lizards (Newbold, 2005), and G. sila prefer open ground (Warrick et al., 1998) and tend to avoid areas with invasive annual grasses Germano et al., 2001;Hacking et al., 2013). However, our study shows that when shrubs are not available, G. sila can use annuals for shade. We did not place models underneath annuals to assess the thermal quality of this microhabitat, an excellent topic for future study. Although many ectotherms avoid areas with invasive grasses, the microhabitats under these grasses may actually be cooler than an undisturbed area and may theoretically provide a better thermal environment as temperatures rise (Garcia & Clusella-Trullas, 2019).
In addition, we observed G. sila climbing annuals including grasses, especially at Shrubless, which could be a way to escape high surface temperatures or to gain a better view of the surroundings. Astragalus sp. were used much more often as shade by lizards at Shrubless than those at Shrubbed (Figure 2b)  Numerous studies at Shrubbed in previous years documented more extensive use of E. californica by G. sila (Ivey et al., 2020;Lortie et al., 2020;Westphal et al., 2018). Westphal et al. (2018) showed that G. sila select for large shrubs such as E. californica more than what would be expected based on shrub density,  showed that G. sila scat is found more frequently under E. californica canopies than in the open. Our study followed a relatively wet winter, and the smaller I. acradenia and G. californica shrubs may not have been as present during the studies conducted in previous years. The understories of E. californica were also smothered with tall nonnative grasses capitalizing on the shade provided by the shrub, which likely prevented G. sila from using them for shade as often as in previous years Ivey et al., 2020;Westphal et al., 2018). This observation suggests that G. sila are flexible and can use shade from any plant, not just E. californica, which is important information for habitat management and restoration efforts.
Qualitatively, from our telemetry observations, the G. sila at Shrubbed seemed to use smaller perennial shrubs such as I. acradenia and G. californica more often than E. californica for thermoregulatory purposes, and instead seemed more likely to retreat to E. californica if they felt threatened. Large, dense shrubs provide lower temperatures than small shrubs (Kerr et al., 2008), but G. sila appear to use burrows when temperatures become really high. Gambelia sila may prefer to use smaller shrubs, when available, for thermoregulatory purposes because they provide cover from solar radiation with less obstruction of surrounding views, allowing these visually oriented lizards to better see prey, predators, mates, and rivals.

| Thermoregulation
Thermoregulatory accuracy was higher for lizards at Shrubless than at Shrubbed, which was unexpected because we predicted that the ability to utilize shrubs would improve the thermoregulatory accuracy of G. sila. However, our result is consistent with the observation that T e in the open was higher at Shrubbed than at Shrubless (Figure 3), even though we chose these nearby sites as "matched" sites. Models inside burrows also warmed up faster in the morning at Shrubbed than at Shrubless in May, but not in June or July (Figure 3).
The temperature variation between sites may reflect soil composition, reflectance, or other variables (Limb et al., 2008). Our results suggest that very small differences in environmental temperatures can impact body temperature and thermoregulatory accuracy in heliothermic lizards, and emphasize the importance of understanding the thermal landscape of a given environment (Milling et al., 2018), which has been shown via models to impact thermoregulation (Sears et al., 2016).
Another contribution to the better thermoregulatory accuracy of G. sila at Shrubless is that they spent more time in burrows during the middle of the day (Figure 2), where T e is closer to T set , while lizards at Shrubbed spent more time aboveground, both in open sunlight and in the extensive shade that is unavailable at Shrubless. It is possible that lizards at Shrubbed were able to risk operating at T b higher than their T set during the day because they have an available aboveground buffer in the form of ample shade, while lizards at Shrubless have to limit their time spent aboveground because they cannot risk becoming too hot before retreating into a burrow. Simulated models indicate that lizards are expected to conserve energy by thermoconforming in more homogeneous landscapes such as Shrubless (Basson et al., 2017); the lizards at Shrubless indeed spent less time in sunlight and therefore were more thermoconforming than lizards at Shrubbed. Notably, our T set values may underestimate the true T set of G. sila, given that we could only measure T set for three hours and could not afford time to allow lizards extensive acclimation inside the gradient.

| Predation risk and other site differences
The lack of shrubs at Shrubless may have consequences that extend beyond thermoregulation. More G. sila at Shrubless (N = 6) were lost to probable predation than at Shrubbed (N = 1). Indeed, there were more confirmed mortalities (dead lizard found with collar) at Shrubless (N = 4) than at Shrubbed (N = 1); some of these lizards had missing limbs, but otherwise, their bodies were mostly intact. Lost collars were likely lizards that were carried away by birds, which are common predators of G. sila (Germano, 2019). In addition, two collars at Shrubless were found with lizard entrails, suggesting that those lizards were killed by avian predators (Germano, 2019;Nelson, 1934). While sample sizes of dead and lost G. sila are too small to draw definitive conclusions, these data suggest that lizards at Shrubless might experience higher predation pressure than those at Shrubbed. Lack of large shrubs such as E. californica may allow birds of prey or other visually oriented predators such as snakes to more easily see and capture lizards on the desert floor. Predation may therefore be an additional reason why G. sila at Shrubless spent more time underground in rodent burrows than those at Shrubbed.
Predator avoidance was found to be an even higher priority for lizards in choosing a microhabitat than thermoregulation in Velvet geckos (Oedura lesueurii, Downes & Shine, 1998), and Mediterranean lizards (Psammodromus algirus) avoided leafless shrubs in early spring because they could not hide from predators as easily (Martín & López, 1998). In accordance with this idea, G. sila were observed using E. californica for predator avoidance in our study and in others Montanucci, 1965;Westphal et al., 2018).
There are many other factors that may contribute to differences in activity and thermoregulation between lizards at our two sites, which we did not explicitly measure for this study. Abundance and composition of small arthropods that serve as the lizards' prey (Germano et al., 2007) may be different between the two sites, especially since the vegetation composition is so different. There is also the possibility that soil composition is different between the two sites; anecdotally, the soil at Shrubbed is rockier than at Shrubless.
This may have contributed to thermal differences on the ground that impacted lizard thermoregulation.

| Hours of restriction and climatic projections
Our analysis of hours of restriction confirms the conclusion of Ivey et al. (2020)  to current T e , whereas certain microhabitats might actually warm at a slower rate, providing thermal buffers (Baust, 1976;Gonzálezdel-Pliego et al., 2020;Scheffers, Brunner, et al., 2014;. A more robust prediction would take into account these differences in warming rate for each microhabitat compared with ambient temperature, which would likely be more favorable for the lizards. Furthermore, we measured burrow T e relatively close to the entrances of D. ingens burrows, and it is likely that temperatures are lower deeper inside these complex burrow networks. The fact that lizard T b was lower than burrow T e at night in May (Figure 3) supports this notion. As the climate warms, lizards may be able to move deeper inside these burrows to continue thermoregulating within their T set .

| CON CLUS ION
We found that G. sila without access to shrubs are not necessarily in greater danger of overheating or losing hours of activity, as lizards at Shrubless thermoregulated closer to their T set than lizards from Shrubbed. While shrubs may play an important role in lizard thermoregulation, lizards at Shrubless spent more time in burrows and thermoregulated more accurately, suggesting that burrows are as important to the thermal ecology of G. sila as shrubs, or likely even more important. There also appears to be a trade-off between more accurate thermoregulation and activity spent aboveground, as implied by the fact that lizards at Shrubless had higher thermoregulatory accuracy and spent more time in burrows. In addition to deploying artificial shade structures , ensuring the continued presence of D. ingens may be essential in securing G. sila persistence. Burrows excavated by ecosystem engineers such as D. ingens are often critical to the survival of other community members (Pike & Mitchell, 2013;. Additionally, our data suggest that shrubs could be important in protecting G. sila from avian predators such as ravens, further underscoring the conclusion that the ideal habitat for G. sila is San Joaquin Desert with D. ingens precincts and shrubs. To ensure that our results are relevant to the conservation of G. sila across California's San Joaquin Desert, expanding our methods to include additional populations of G. sila would provide a management-applicable understanding of how these lizards interact with their thermal landscape on multiple spatial scales (Steen, 2010). Recognizing the importance of water availability, another environmental factor that is becoming more and more limited in the San Joaquin Desert as droughts become more regular will also help us understand constraints faced by G. sila and other desert lizards that are facing similar stressors.

ACK N OWLED G M ENTS
We thank B Axsom, J. Hurl, and B Lindquist of BLM for logistical assistance, K Ivey for leading an original study on which this study was based, and the following people for their support: K Bodwin,

CO N FLI C T O F I NTE R E S T
None declared.