Hot springs, cool beetles: extraordinary adaptations of a predaceous insect in Yellowstone National Park

Abstract

Several metazoans live in extreme environments, but relatively little is known about the adaptations that these extremophiles have evolved to tolerate their conditions. The wetsalts tiger beetle, Cicindelidia hemorrhagica (LeConte) (Coleoptera: Cicindelidae), is found in the western USA, including the active geothermal springs in Yellowstone National Park (YNP). Here, we characterize behavioral, ecophysiological, and morphological traits of adult C. hemorrhagica living on hot springs in YNP compared to adults living in a non-hot spring environment in Idaho. Individuals in YNP behaviorally warmed and cooled themselves at surprisingly different frequencies than those in Idaho, with YNP individuals infrequently cooling themselves even though surface temperatures were greater because of geothermal activity and consequent bottom-up heating of individuals compared to the saline-flat habitat in Idaho. After a series of lethal thermal maxima and internal body temperature experiments, our results suggest that an explanation for the differential behavior is that the adult in YNP has evolved increased heat reflectance on the ventral portion of its abdomen. This increased heat reflectance seems to be caused by a physical feature as part of the exoskeleton’s ventral abdominal plate, which likely protects the beetle by serving as a heat-resistant shield. The extreme conditions in YNP seem to have selected C. hemorrhagica to be among the most thermophilic insects known.

Introduction

Tiger beetles (Coleoptera: Cicindelidae) are insect predators that include many species adapted to high ambient temperatures as adults. The behavioral and ecophysiological traits of these species are adaptive for occupying niches unavailable to other species because of high temperature. The wetsalts tiger beetle, Cicindelidia hemorrhagica (LeConte), is one such warm-adapted species that has been recorded actively hunting and scavenging in active hot springs areas of Yellowstone National Park, USA (YNP) for more than 130 years (Hubbard 1891, Bertholf 1979) (Fig. 1a, g). Hubbard (1891) noted that the environments these tiger beetles inhabited likely exceeded 50 °C, surpassing the upper thermal limits for many known insect species. Although there have been many studies on the thermal features and microbial thermophiles of YNP (Inskeep and McDermott 2005, Boyd et al. 2009, Fouke 2011), C. hemorrhagica in YNP has not been studied until recently (Willemssens 2019, Willemssens et al. 2024).

Cicindelidia hemorrhagica is found in the western USA, and its typical habitat includes sandy, wet substrates that border saline flats or other warm, stagnant water sources (Pearson et al. 2015). Warm-adapted tiger beetles often employ a maxi-thermal strategy, in which their optimal temperature is near their upper lethal limit (Dreisig 1979). In this sense, they live in environments with high-temperature stress, and, therefore, can be considered thermophiles. In YNP, C. hemorrhagica is found exclusively in and near hot springs (Willemssens et al. 2024) and is active in environments not only with high surface temperatures because of insolation at high elevation but also hot geothermal waters containing high concentrations of heavy metals and alkaline or acidic pH that most vertebrates and invertebrates in YNP actively avoid (Inskeep and McDermott 2005, Boyd et al. 2009, Fouke 2011, Adams et al. 2024, Willemssens et al. 2024).

Insects often behaviorally thermoregulate to adjust their internal body temperatures. As ectotherms, these thermoregulatory behaviors often change as the position of the sun shifts during the day (Dreisig 1979, 1981, Pearson and Lederhouse 1987, Schultz and Hadley 1987, Brosius and Higley 2013). Warm-adapted tiger beetles, especially species in environments like those of C. hemorrhagica, behaviorally thermoregulate by stilting, shading, and dipping their abdomens in water to evaporatively cool themselves (Dreisig 1979, 1990, Morgan 1985, Pearson and Lederhouse 1987, Knisley et al. 1990, Brosius and Higley 2013, Willemssens 2019).

With behavioral thermoregulatory cooling at frequencies as much as 50% of mid-day activities, Pearson and Lederhouse (1987) observed a lethal maximum temperature of approximately 48.0 ± 0.1 °C for C. hemorrhagica in Arizona. However, our observations of C. hemorrhagica adults in and around hot springs in YNP suggest that they do not exhibit thermoregulatory behaviors at the same frequencies as those for warm-adapted tiger beetles elsewhere (Willemssens 2019). In YNP, they seem to behaviorally cool themselves at much lower frequencies compared to individuals of the same species elsewhere or other warm-adapted tiger beetle species (Willemssens 2019). At the same time, C. hemorrhagica adults in YNP are present on much hotter surfaces because of geothermally heated water. We did not identify any key characters, including male genitalia, which suggested the individuals in YNP and Idaho were different species. We also conducted a CO1 comparative analysis, which strongly suggests that the YNP and Idaho populations are the same species (99.8% overall similarity) (Supplementary Table S1).

Therefore, our initial behavioral observations lead to the question: Has C. hemorrhagica in YNP evolved adaptations in addition to the species’ known tolerance to high temperatures? To answer this question, we conducted a series of behavioral, ecophysiological, and morphological studies from 2016 to 2020 with C. hemorrhagica adults in geothermal areas of YNP and in a non-geothermal area in Idaho (Supplementary Fig. S1).

Materials and Methods

We conducted research from 2016 through 2020 at a total of 4 geothermal locations in YNP and 1 non-geothermal location in Idaho (Peterson 2022, Adams et al. 2024, Bowley et al. 2024, Willemssens et al. 2024).

Detailed Site Descriptions (see Supplementary Information).

Behavior

During the summer of 2016, initial observations were made of live adult C. hemorrhagica to develop an ethogram (catalog of discrete behaviors) (Supplementary Table S2). We categorized these behaviors by behavioral activities (ie feeding), modifiers (ie feeding on larvae), and states (ie sun). Additionally, we grouped the behaviors by function (ie “reproduction” consists of the behaviors: abdomen probe, mating, mate guarding, and ovipositing) (Supplementary Table S2). Of interest to our research were behaviors and habitat use associated with thermoregulation.

In our observations in 2017 (not reported here) and 2018 at 3 locations in YNP (Sizzling Basin, Angel Terrace, Dragon Spring), and 1 location in Idaho (42.9352, -115.7502) (Supplementary Table S3), we used continuous focal sampling, where we observed 1 adult for 10 min continuously. We observed the beetles with close-focus binoculars while speaking into a digital voice recorder to record behaviors. At the beginning of each recording, the time, location, observer, ambient air temperature, soil conditions, and sex of the adult (if known) were noted. Within each hour, we observed a maximum of 3 adults. If we lost visual of the beetle during our 10-min time window, we noted this in our recording as “out of sight” until the beetle was visible again. If the beetle flew away or remained hidden for longer than 10 s, we chose a new beetle and observed for the remaining time/time lost. About 2 to 3 beetles out of approximately 35 per day went out of sight or flew away during observations.

In 2018, 8 different observers were used to record behavior. Observers worked in teams of 3 with one person observing behavior, one person recording temperatures and other meteorological data, and the third person on wildlife (primarily bear) watch. Observations at all locations were recorded when conditions were sunny and winds were relatively low, (0 to 6 km/h), with some sudden and short-lived wind gusts up to 32 km/h.

Observers were field trained for 1 wk to recognize and report behaviors, and we did not include these data in our data set. Observers were rotated to avoid fatigue and to minimize potential observer bias.

Each hour, we noted the weather conditions with a Kestrel 2500 pocket weather meter to account for behavioral changes due to weather changes. The experimental unit was one adult C. hemorrhagica beetle, where Angel Terrace had 148 replications, Dragon Spring had 91 replications, Sizzling Basin had 118 replications, and Idaho had 80 replications. Differences in replications per location were due to weather, beetle availability, and permit allowance, among others.

The recordings were transcribed using JWatcher (version 1.0) behavioral analysis software. Data were transcribed at least twice to ensure accuracy in transcription. This resulted in construction of a master file containing durations and counts for each behavior by each beetle, and we added the associated temperatures, habitat use, weather, and soil conditions for each beetle. This master file was then used to visualize (GraphPad Prism version 8.0.0 for Windows, GraphPad Software, San Diego, California USA, www.graphpad.com) and analyze [(SAS ver.9.4 (SAS Institute, Cary, NC) software] the percentage of daily activity spent on a specific behavior, thermoregulatory behavior, habitat use, and differences in temperatures, among others. We used an unbalanced t-test to determine the mean differences between the behaviors of the Idaho and YNP populations (= Welsh-Satterthwaite in proc t-test). We chose the Welsh-Satterthwaite unequal variances t-test because we observed more beetles in YNP (359 individuals) than in Idaho (81 individuals). We used the Pearson correlation coefficient to determine a linear correlation between behavior and temperature. We did not assess behavioral differences between sexes because sex was nearly impossible to determine via binoculars unless mating or mate guarding was attempted.

Lethal Thermal Maxima

Lethal thermal maxima were measured using a sous vide (Joule, Seattle, WA, USA). The sous vide allowed ramping up temperatures in a 24.6 L (26 quarts) water bath container (Rubbermaid, Wooster, OH, USA) and displaying the water temperature. We placed single live beetles in an air-filled 20-ml scintillation vial with a leak-proof lid and submerged them in the water; therefore, each vial was considered an experimental unit. A vial containing a thermocouple was placed into the water to measure internal temperatures in the vial. In 2018, different acclimation duration (2, 3, or 7 h) and acclimation temperatures (33 or 40 °C) were used to assess the importance of hours and temperature in influencing lethal limits. The data showed that the acclimation temperature was more important than the acclimation duration in influencing lethal limits. Because of these findings, in 2019, beetles were allowed an acclimation time of 30 min to water of 40 °C before we increased the temperatures. In 2018, water levels in the bath were adjusted to allow a temperature rate of 1°C/min with various starting water temperatures. In 2019, we used a constant water volume and used an auto-transformer voltage regulator to adjust the sous vide output to maintain our 1°C/min, starting at 40 °C and ending when the beetle was considered dead.

Temperatures inside the water and vial were noted when beetles showed distress such as wing pumping, loss of footing, or protruding of genitalia (scored as critical thermal maximum). When the beetle became immobile (scored as lethal thermal maximum), we noted the end temperatures and removed the vial from the water (our experimentally observed differences in critical versus lethal thermal maxima never exceeded 1 °C). The beetle was observed for an additional 15 min, 1 h, and 24 h to ensure expiration/mortality (we deleted the maximum lethal temperature on the rare occasion when a beetle revived). Between each trial, we removed the water, refilled the container, and adjusted water temperatures to 40 °C. The numbers of beetles we collected were limited by YNP research permit requirements and therefore this affected how we conducted the laboratory experiments. In 2018, trials consisted of 4 to 8 vials, each with a single beetle; in 2019, each trial consisted of 5 vials, 4 vials with 1 beetle and 1 vial containing the thermocouple. We classified the beetles by sex and location. For Idaho, we assessed 3 trials with 12 beetles, for Dragon Spring and Rabbit Creek (44.6218, -110.434; 2,230 m elevation; pH 9.2), we assessed 2 trials with 8 beetles per location. The tested beetles had a sex ratio of 50:50. We conducted several experiments with differing pre-conditioning to determine if pre-conditioning tiger beetles (both C. hemorrhagica and C. punctulata) affect their lethal temperature limit (Supplementary Table S4).

To analyze the data, for within-trial analysis, we used simple t-tests to determine significant differences in lethal thermal maxima. Additionally, we used Mixed Models analyses in SAS (Statistical Analysis System) to perform a meta-analysis across experiments. We also used regression analyses to determine the relationship between the conditioning temperature and the lethal thermal maxima of the beetles.

Internal Temperature Comparisons

To characterize potential differences in internal temperatures of the beetles, we conducted a series of laboratory experiments (Supplementary Fig. S1). We collected approximately 25 beetles per site for the 2019 and 2020 field seasons to use for internal temperature comparison and other planned experiments (the numbers of beetles we collected were limited by YNP research permit requirements and this affected how we conducted the laboratory experiments). These collected individuals were then placed in small plastic containers that contained air openings for gas exchange and a damp paper towel to maintain relative humidity. We placed a maximum of 5 beetles in each container to prevent cannibalism and each container was placed in a daypack. Containers were then placed in a YETI Tundra 35 cooler (YETI, Austin, TX, USA) with ice packs as soon as possible after returning to the field vehicle. Coolers were unloaded on arrival at Montana State University and placed in a −20 °C freezer to quickly kill individuals and provide freshly frozen specimens for experimentation.

We created the heated water bath using a 12-L sous vide container (Rubbermaid, Wooster, OH, USA) filled with approximately 8 L of water before each testing period. The temperature of the water bath was controlled and regulated by the sous vide that was connected to a cellphone controller via Bluetooth. We produced thermocouples using 24-American Wire Gauge thermocouple wire connected to miniature T-type thermocouple connecter plugs (OMEGA, Norwalk, CT, USA). Plastic straws with a 6-mm diameter were cut into 10-cm segments to act as straight-framed reinforcement for the thermocouple wire to hang straight down, perpendicular to the water bath. The wire was bent into a 90˚ angle, so the exposed thermocouple wire was parallel to and 2 mm above the surface of the water. Internal temperatures of the beetles were recorded using an SD-947 4-channel thermometer/SD card data logger (REED, Wilmington, NC, USA).

Freshly killed beetles collected from each site within YNP were paired with an Idaho individual of similar size and same sex. We grouped our pairs together based on the original field research site where the YNP beetles were collected (Dragon Spring vs Idaho and Rabbit Creek vs Idaho). We then inserted the exposed end of the thermocouple wire into each beetle through the genital opening, running the wire up the abdomen lengthwise without breaking through exoskeleton or penetrating the thorax (Fig. 2). Consequently, the thermocouple wire was positioned approximately midline between the ventral and dorsal portions of the abdomen. In later experiments, we removed the legs from every individual before experimentation to eliminate the possibility of thermal conduction or evaporative cooling from the water touching specifically the hind and mid pairs of legs.

We conducted 2 water-bath experiments with 8 treatment replicates per experiment over 4 different temperatures. Each replicate consisted of 2 beetles attached to thermocouples and 1 bare thermocouple control for all 8 replicates. Eight beetles from Idaho were paired with 8 from Rabbit Creek for the first experiment, and another 8 beetles from Idaho were paired with 8 from Dragon Spring for the second experiment. Beetles with their legs removed were used in the last 4 pairings of each YNP location and Idaho comparison experiment. We used 32 adults between all 3 populations: 8 from Dragon Spring, 8 from Rabbit Creek, and 16 from the Idaho population (Supplementary Table S5). Three thermocouples were connected to the thermocouple reader at a time and held in place over the water bath by a 2.5-cm thick × 5.1-cm wide wooden plank with notches filed 2.5-cm deep into the side of the board that were designed to hold the reinforced thermocouple wire (Fig. 2). One beetle from a YNP site and the Idaho site were randomly assigned to thermocouples and 1 bare thermocouple was used as the control. Thermocouples that were attached to beetles had alternating positions above the water bath. This ensured that temperatures experienced by the thermocouples were independent of position in the enclosed water bath. The water bath was heated initially to 40 °C for the first trials and increased to 55 °C by 5 °C increments for each trial thereafter. Once the thermocouples were in place and the water bath was heated to the correct temperature, they equilibrated for 2 min. Once the thermocouples equilibrated, we recorded internal temperatures of the beetles once per minute for 10 min. This process was then repeated 8 times for each of 2 experiments (ie 1 beetle pair, 4 temperatures, 8 times).

Cuticular Wax and Internal Temperatures Comparisons

Our findings from the internal temperature comparison experiments between Idaho and YNP individuals suggested that there was a physical aspect responsible for the difference in internal temperatures we observed in the YNP beetles. To further understand this phenomenon, we conducted temperature comparison experiments that determined the contribution of cuticular wax to any difference in internal temperatures that YNP adult C. hemorrhagica experienced (Supplementary Fig. S1). Nine pairs of freshly killed beetles of similar size and sex were selected from Dragon Spring and Rabbit Creek. We did not use Idaho beetles because we had already determined there was strong statistical evidence there was a difference in internal temperatures in the YNP beetles. Therefore, the next question was whether the cuticular wax on the YNP beetles contributed to these differences. One beetle from each pair was placed in a 20-ml scintillation vial and submerged in 15 ml of dichloromethane for 1 min before we removed it, which removed the cuticular wax without compromising the exoskeleton (Supplementary Table S5). We mounted both beetles above the water bath in the same fashion as the previous experiments described above and recorded internal temperatures for only 45 °C and 50 °C. We only used 45 °C and 50 °C because these temperatures produced the largest differences in internal temperatures of the beetles, and our question was whether the cuticular waxes or wax-less exoskeleton was responsible for the differences in temperatures. For all experiments described above, beetles were completely dried in a drying oven and weighed to assess if weight was a contributing factor in our results.

Abdominal Plate Extraction and Temperature Comparison

The results of the internal temperature comparison (including completely dried specimens on thermocouples) and the cuticular wax removal experiments suggested that the mechanism or mechanisms responsible for differences in internal temperatures are caused by a physical characteristic on the ventral face of the YNP beetles (Supplementary Fig. S1). Because the thermocouples were placed in the abdomen, we next focused on the ventral abdominal face. The abdominal ventrites are not connate, but they remain linked together after dissection if carefully handled, forming what will hereafter be referred to as a ventral abdominal plate. Eight individuals from the summer 2020 YNP collection period were used for isolating the ventral abdominal plate. We did not use Idaho beetles because our objective was to determine if the lower internal temperatures in YNP beetles were solely attributed to the ventral abdominal plate. Once we identified and labeled dead specimens by sex, we removed each of their legs. We used a #2 X-Acto knife (X-Acto, Westerville, Ohio, USA) to dissect out the abdominal plate by slicing the body along the midline where the dorsal and ventral sides meet. These cuts were made on both sides of the beetle before forceps were used to lift the ventral plate from the body. The plate was then cleaned using forceps to carefully scrape away viscera and rinsed with ethanol. Once the abdominal plate was prepared, each plate was hot glued to a randomly assigned thermocouple. Another thermocouple had electrical tape cut to the approximate shape of the abdominal plate attached to it using hot glue to act as the control for each experiment. A bare thermocouple was also included to monitor air temperatures above the water bath as in previous experiments (Supplementary Fig. S2). We conducted 8 replications in this experiment using one thermocouple with a YNP ventral plate, 1 with electrical tape, and 1 bare thermocouple. As explained above, we only used 45 °C and 50 °C because these temperatures produced the largest differences in internal temperatures of the beetles, and our question in this case was whether the ventral plate was responsible for the differences in temperatures.

Data Analysis for Thermocouple Experiments

Before analysis, we first had to calibrate thermocouples because not every thermocouple displayed the same temperature measurement as the bare thermocouple control when measuring the ambient air temperature. The difference between the control and thermocouples used for adult beetles only differed by a maximum of 0.2 °C, so thermocouples were calibrated ± 0.2 °C to account for the difference in measured temperature between the bare thermocouple that was used as the control. Each thermocouple was measured against the control before recording for this reason. Corrected temperatures were then compiled into a spreadsheet based on the paired treatment of comparison experiment performed. Initial data visualizations and t-tests were performed for each group using Microsoft Excel. The temperature values of each thermocouple paired treatment with a beetle attached to it were directly compared to the bare thermocouple temperature reading for each minute of recording. This provided us with the following equation:

where $ΔT$ is the difference in temperature between the bare thermocouple and the thermocouple with the beetle. The $ΔT$ values of each YNP and Idaho pairing were totaled and averaged according to the temperature of the water bath to produce one average $ΔT$ per temperature per beetle. These average values were grouped together by location and used to calculate the standard deviation and standard error for each location. The standard error values were later incorporated into visualizations of these temperature data.

The temperature data from the Idaho and YNP comparison experiments and the cuticular wax comparisons were transferred to RStudio version 4.0.2 for further data visualization and analysis (RStudio Team 2021). A linear mixed effects model was developed using the mean $ΔT$ values as the dependent variable for each temperature a beetle that had its temperature recorded, resulting in 4 mean $ΔT$ values per individual beetle in the YNP and Idaho comparisons. The mean $ΔT$ values for 45 °C and 50 °C in the cuticular wax comparison experiments were the only 2 $ΔT$ values used for these individuals. Each beetle was assigned its treatment to differentiate between groups. Beetles from the Idaho and YNP comparison experiments had their location substituted for their treatment. Additional variables included in the model were the presence of legs, air temperature outside the water bath, sex, and dried beetle weight and each were evaluated for their potential influence on the $ΔT$ values of each beetle. Variables not influential to $ΔT$ values were excluded from the final model. Initial diagnostic figures evaluating equal variance among observations and assessing the residual normality of the data were developed for the model to prevent any analysis that carried statistical violations. The model was then evaluated for assessing differences between groups accounting for the repeated measurements conducted on each beetle.

A separate linear mixed effects model was produced for the repeated measurements in the ventral abdominal comparison experiment to assess differences between the cuticular venter and the imitation venter. These values were kept separate from the whole beetle thermocouple experiments because dry weights were negligible and ambient air temperatures were not recorded throughout experimentation. This model was only used to determine the differences in $ΔT$ between treatments and no other variable.

Spectrophotometric Measurements

Six dead adult beetles were analyzed by the Materials Research Laboratory at the University of Illinois at Urbana-Champaign for spectrophotometric performance of the abdominal plate using spectrophotometry. Three beetles from YNP and 3 from Idaho were selected at random for the observational analysis. Each beetle had its ventral abdominal plate dissected from its body with any remaining viscera carefully scraped away to not damage any of the area of interest. Baselines were then established on a Cary 5000 UV-VIS-NIR spectrophotometer (Agilent Technologies, Santa Clara, California, USA). We placed each abdominal plate carefully in receptacles, so they were not accidentally damaged between observational periods. Each plate was placed on a fabricated piece of sheet metal that was specially made for accommodating the passage of light through the abdominal plate. The abdominal plate was also suspended on a glass pillar throughout the diffuse absorption recording to ensure all reflected and transmitted light was captured by the equipment’s sensors. Each beetle was recorded using the same baselines established for diffuse reflection, diffuse absorption, and diffuse transmittance. Data from each of the 3 beetles from each location were combined into mean reflectance measurements and visualized in RStudio.

Scanning Electron Microscope Characterization

Results from our temperature comparison experiments justified the need for further investigation of cuticular microstructures on the abdomens of dead adult C. hemorrhagica. We first examined the exoskeleton of YNP and Idaho individuals using a relatively high-powered dissection microscope (Leica MD80, Mannheim, Germany) and did not observe any outstanding physical differences in cuticular structure. Therefore, we next investigated physical structural differences using scanning electron microscopy (SEM).

We used beetles collected in 2020 from both YNP and Idaho for SEM work because these specimens were the least likely to have a brittle exoskeleton or have been damaged in the freezer over time. Each specimen was rinsed in ethanol and gently wiped clean using kimwipes (Kimtech, Roswell, GA, USA) and allowed to air dry in a petri dish for approximately 2 min. The beetle was then secured with carbon and copper tape on the sample stage. Specimens were then sputter-coated with approximately 25 nm of iridium (Electron Microscopy Sciences, Hatfield, PA, USA). Coated beetles were then imaged using a Zeiss Supra 55VP Field Emission Scanning Electron Microscope. Images recorded were at 500×, 1,000×, 1,500×, and 5,000×, focusing on the first 4 abdominal segments on the ventral side.

We recorded additional SEM imaging using the same procedure on YNP beetles that had their cuticular wax removed using dichloromethane. To determine if there were noticeable qualitative differences in the layering of the ventral plate of YNP C. hemorrhagica, we embedded 2 adult beetles in epoxy resin (Buehler Epoxicure), sectioned the beetles with a low-speed saw under water irrigation, and polished the samples to 0.05-µm grit finish to produce a smooth region of interest for cross sectional analysis. These samples were sputter-coated with iridium to avoid sample charging. We visually inspected all SEM images for qualitative structural differences between YNP and Idaho specimens that could possibly explain the internal temperature differences.

Results and Discussion

Behavioral Differences Between YNP and Idaho Cicindelidia hemorrhagica Adults

The beetles spent the largest proportion of daily activity as “no movement” for both the YNP (68.2 ± 1.1%) and Idaho (54.2 ± 2.6%) populations (Fig. 1b). Tiger beetles are characterized and easily recognized by their “run/stop” movement. While running after prey or a mate, their vision is impaired because they cannot resolve images sufficiently to form a clear image and determine location (Pearson 1988). Because of this, they must stop briefly to refocus and adjust their path according to their prey or mate. The “no movement” represents this short stop-motion, as well as longer durations where the beetle displayed no behavior (ie remains still in a resting habitus) in any given habitat type. “No movement” does not necessarily mean beetles were completely inactive, but rather they may have accumulated sensory input on prey, predators, or potential mates before engaging in active behavior.

We eliminated the “no movement” category to better examine the other behaviors. In Idaho, C. hemorrhagica spent most of its daily activity on feeding (26.3 ± 2.5%), while YNP beetles spent it on movement (8.9 ± 0.5%) (Fig. 1b). Cicindelidia hemorrhagica in YNP spent more time on reproductive behavior (4.4 ± 0.8%) than did the individuals in Idaho (1.0 ± 0.8%). The increased movement in YNP may be due to the more limited food sources and the need to chase after (or from) potential mates.

When comparing heating thermoregulatory behavior (bask, push-up, and shake) and cooling thermoregulatory behavior (stilt, wing pump, and abdomen dip), there was a significant difference between YNP and Idaho adults (P = 0.0002). Idaho adults exhibited more cooling thermoregulatory behavior (6.8 ± 1.6% of the time) (Supplementary Video S1), while YNP adults exhibited more heating thermoregulatory behavior in YNP (5.4 ± 0.6% of the time) (Fig. 1b; Supplementary Video S2). For Idaho adults, there was a positive correlation between temperatures (both ambient air and soil) and thermoregulatory behavior (air temperature: r = 0.276, P < 0.013; soil temperature: r = 0.362, P < 0.0009). This is consistent with increased thermoregulatory behaviors for cooling as air and soil temperatures increase, as is seen with most insects. In contrast, in YNP we observed a highly significant negative correlation between thermoregulatory behavior and temperatures (air and soil) (air temperature: r = −0.224, P < 0.0001; soil temperature: r = −0.153, P = 0.0037). The negative correlation, coupled with observed behavior (Supplementary Figs. S4 and S5), indicates that thermoregulatory behaviors are associated primarily with warming at cooler temperatures and behavioral thermoregulation for cooling is almost non-existent.

This finding aligns with our observations of C. hemorrhagica in Idaho, where the adults did not become active until 9 am and emerged from underneath the dry salt crust to bask. As temperatures rose, the adults moved toward the water to feed, began stilting, abdomen dipping in water, and would become inactive in the shade throughout the hottest part of the day (Supplementary Video S1). In the evening, the adults moved from the water and returned to the dry salt-crusted area, where they remained buried overnight.

For YNP adults, the lack of behavioral thermoregulation with high temperature is surprising given the high air, substrate, water, and steam temperatures to which individuals are exposed. Indeed, we observed individuals active in and immediately adjacent to thermal pools with measured maximum temperatures from 50 to 70 °C (Supplementary Video S2). Rather than the proportions of time shifting behavior throughout the day, as seen in Idaho adults, YNP adults spent a continuously higher proportion of time in the relatively dry and sunny habitat (Supplementary Figs. S4 and S5). The beetles rarely hid in shade, even throughout the hottest parts of the day (Supplementary Video S2). Even though the proportion of adults in wet habitat increased when temperatures rose, the overall proportion was much lower than the Idaho adults, which is unsurprising given the wet habitat in YNP is caustic hot water.

This finding aligns with our observations as we saw YNP adults emerging from the dry substrate in the mornings with minimal basking. As temperatures increased, their activity increased as well; some individuals immediately moved to the water while most remained on the relatively dry soil hunting for prey, interacting with one another, or attempting to mate. YNP adults showed no behavioral differences when approaching hot spots caused by the thermal spring (ie stilting, wing pump, abdomen dip, flying away, or moving to a cooler spot immediately) (Supplementary Video S2). In the evening when the air temperatures cooled, adults moved back under the dry, warm substrate to bury themselves overnight.

When examining the overall proportion of thermoregulatory behavior (6.4 ± 0.6%) and after adding habitat use (no shade, minimal wet habitat) into thermoregulation, we find no support that adults use thermoregulatory behavior to cope with the high temperatures associated with thermal springs. They seem largely indifferent to temperature fluctuations in the field. Although we cannot directly compare the Idaho and YNP adults due to differences in elevation and distance between locations, there is sufficient overlap in temperature by time of day and average temperature to conclude that these populations behave differently, and the differences are not because of differences in air temperatures (Supplementary Table S3, Supplementary Figs. S4 and S5). Although we did not record behaviors from 2019 to 2021, we collected beetles at the Idaho location in late July each of those years. The air and saline-flat water temperatures were approximately 30 °C each day we were there, typical of the average temperatures (Supplementary Table S3). Both years, we observed extensive behavioral cooling from late morning to mid-afternoon that was like the reported 2018 values when the air temperature was higher (Supplementary Video S1). These observations, and the fact that in YNP the adults experience high air, water, steam, and surface temperatures of the geothermal springs, further support that the YNP adults do not behaviorally cool themselves as would be expected of warm-adapted tiger beetles (Supplementary Video S2).

In Idaho, we observed adults at a wet surface temperatures of about 35 °C exhibiting multiple thermoregulatory behaviors. These behaviors for temperature tolerance are well-studied in tiger beetle species and other thermotolerant species such as the Saharan silver ant Cataglyphis bombycina (Hymenoptera: Formicidae) (Dreisig 1979, 1981, 1984, Guppy et al. 1983, Morgan 1985, Pearson and Lederhouse 1987, Schultz and Hadley 1987, Knisley et al. 1990, Gehring and Wehner 1995, Hoback et al. 2000, Smolka et al. 2012, Brosius and Higley 2013). In contrast, our observations of C. hemorrhagica in and near thermal pools of YNP included individuals immediately adjacent to pools and steam as high as 70 °C without any displayed thermoregulatory behavior (Willemssens 2019, Bowley et al. 2024).

Lethal Thermal Maxima of Cicindelidia hemorrhagica in Yellowstone National Park

A possible explanation for YNP C. hemorrhagica not behaviorally cooling themselves at the same or greater frequency compared to the Idaho individuals is that those in YNP have a higher lethal thermal maximum. We found that YNP adults had a mean lethal thermal maximum of 50.4 ± 0.3 °C, with a maximum of 52.6 °C. The mean lethal thermal maxima were 1.1 °C higher than those found in Arizona (Pearson and Lederhouse 1987) and 1.1 °C higher than those found in Idaho (Fig. 1c). This value is a new high temperature record for tiger beetles, surpassing the previous limit of 48.9 °C for Cicindela pimeriana (Pearson and Lederhouse 1987).

The lethal thermal maximum of YNP adults was significantly greater than Idaho adults (P = 0.0023). However, we observed YNP adults at surface temperatures as high as 70 °C (Willemssens 2019, Bowley et al. 2024). This suggests that they likely have mechanisms to prevent internal heating to deleterious levels. As presented above, we know that C. hemorrhagica in YNP infrequently exhibits thermoregulatory cooling behavior and we can, therefore, rule out behavior as a primary mechanism. The increased lethal thermal maximum of YNP adults possibly indicates a slight physiological adaptation, but if this were the sole mechanism, the lethal thermal maximum should be much higher.

Internal Body Temperature Comparisons Between Yellowstone National Park and Idaho Cicindelidia hemorrhagica Adults

Because the lethal thermal maximum of the YNP adults was only modestly higher than the Idaho individuals, we investigated whether internal body temperatures differed in the presence of heating. The simulated environmental conditions of the water-bath comparison experiments using thermocouples to measure internal abdominal temperatures were carried out from 40 °C to 55 °C to compare internal abdominal body temperatures of Idaho and YNP populations. Results indicated that YNP adults had a significantly lower internal temperature than Idaho adults (Fig. 1d). There was little evidence to suggest that ambient air temperature at the time of recording (P = 0.295), sex of the beetle (P = 0.751), and dry weight of the beetle (P = 0.49) influenced the $ΔT$ values (bare thermocouple control minus thermocouple in freshly killed beetle’s abdomen) recorded in the internal temperature comparison experiments with Idaho and YNP adults. Therefore, these variables were excluded from the final model because they did not contribute to strengthening the predictive power of the model. This means that whatever biological differences we observed between the YNP and Idaho individuals were purely attributable to the location from where the beetles were sampled or the treatment that was applied to the YNP adults. Direct comparisons between YNP study location in Table 1 (P = 0.159) and the Tukey group comparison in Table 2 (P = 0.61) suggest that YNP populations are not sufficiently different from one another to be considered different population measurements, effectively simulating a uniform YNP group of $ΔT$ values.

Because our results showed that $ΔT$ values for YNP populations were not significantly different from one another, the comparison between YNP and Idaho adults is of primary interest in determining differences in internal body temperatures between populations. Direct comparisons between Idaho $ΔT$ values in Table 1 (P < 0.001) to Dragon Spring, our first research location in YNP, provide us with very strong evidence that these groups have different $ΔT$ values. In the Tukey comparison in Table 2 comparing Idaho to both Rabbit Creek, our second research location in YNP (P < 0.001), and Dragon Spring (P = 0.002), there is very strong evidence that these population $ΔT$ values differ from one another (Fig. 1d). This relationship was consistent for all temperatures of the water bath experiments with an overall mean difference of 1.03 °C (Supplementary Table S5). The largest mean difference between internal body temperatures was 1.63 °C at 55 °C between the Idaho and Rabbit Creek individuals.

Our results suggest that YNP adults have a greater heat-reflection mechanism along the ventral face of the abdomen than Idaho adults (Fig. 1d). Because we did not observe differences in this heat reflective effect between freshly killed and completely dried specimens, our results suggest that a liquid milieu in the body or physiological process is not involved in this heat reflection. Consequently, this implies the lower internal body temperatures for YNP adults are most likely produced by a physical or morphological mechanism, or both.

Cuticular Wax Internal Temperature Comparisons

Microstructures made of chitin and wax secretions can improve insect survival. Several studies have explored the effects of chitin and wax secretion on light and heat dissipation or water loss in insects (Hadley 1978, 1979, Dreisig 1979, Hadley and Schultz 1987, Schultz and Hadley 1987, McClain et al. 1991, Seago et al. 2009, Shi et al. 2015, Amore et al. 2017, Alves et al. 2018, Pavlović et al. 2018, Cuesta and Lobo 2019, Xie et al. 2019, Zhang et al. 2020).

However, there were no statistical differences between the mean $ΔT$ values of the groups with and without their cuticular waxes (P = 0.193) in our experiments (Table 2). Therefore, it is evident that the cuticular wax directly contributes little to the reflective heat effect we observed with the Idaho and YNP comparisons. In addition, there was no difference in $ΔT$ for individuals with and without cuticular wax from either Rabbit Creek (P = 0.799) or Dragon Spring (P = 0.999) (Table 2). These data therefore suggest that the structure of the cuticle itself has a greater role in the observed difference between YNP and Idaho internal body temperatures.

Ventral Abdominal Plate Temperature Comparisons

Because cuticular wax was only responsible for no, or minor, differences in internal body temperatures between the YNP and Idaho C. hemorrhagica populations when exposed to high surface temperatures, we needed to isolate the structure of the exoskeleton to determine if we could continue to observe the temperature differences. This was done by dissecting the ventral abdominal plate in preparation for more thermocouple comparison measurements. To ensure that the overall shape of the adult beetle was not responsible for the differences, and to determine if contact with the thermocouple created a difference in the temperature reading, we compared the ventral abdominal plate to an imitation plate made of tape and a bare thermocouple. The $ΔT$ results from thermocouple measurements from 8 ventral plates at water bath temperatures of 45 °C and 50 °C indicate statistical differences between thermocouples that had an abdominal plate attached, the electrical tape imitation plate, and the bare thermocouple (P = 0.037) (Table 3) (Supplementary Figs. S2 and S3). This further suggests that contact with the thermocouple did not influence temperature recordings because the ventral abdominal plate group had greater $ΔT$ values than the imitation plate group.

Building off what we observed in our previous experiments, we know that YNP adults show increased heat resistance, and the component responsible is not only present in living or freshly thawed YNP adults but in desiccated specimens as well. We also know that the cuticular wax is not directly responsible for this difference in $ΔT$ measurements. Based on our results, we determined there may be a difference between the microstructures on the ventral face of the abdomen of YNP adults that may help reduce incoming heat from their environmental surfaces rather than a difference in temperature caused by the shape of the beetle or the presence of cuticular wax.

Structural features that aid in thermoregulation are known in insects. Cataglyphis bombycina (Hymenoptera: Formicidae) is the most heat-tolerant insect currently known. These ants possess microscopic triangular setae that aid in light reflection which dissipate heat from incoming sunlight (Shi et al. 2015). The micro setae on the elytra of Neoceramyx gigas (Coleoptera: Cerambycidae) possess corrugated facets, reducing incoming heat on the elytra by providing total internal reflection of light when subjected to various angles of sunlight (Zhang et al. 2020). Rosalia alpina (Coleoptera: Cerambycidae) possess micro-pyramid-shaped structures on their elytra (Pavlović et al. 2018). These structures improve radiative heat exchange between the beetle and its environment, making it possible for the insect to survive in areas with more intense sunlight for longer periods of time (Pavlović et al. 2018).

Spectrophotometry and Scanning Electron Microscopic Analysis

The suggestion that the mechanism for temperature differences is physical or morphological caused us to next employ spectrophotometric measurements on the ventral abdominal plate of C. hemorrhagica. The diffuse reflectance properties in the ultraviolet (UV), visible, and near infrared (NIR) light wavelength ranges of 6 selected individuals from YNP and Idaho were recorded. Our results indicated that YNP adults had a greater reflectance from the 750 nm to 2000 nm wavelength range (Fig. 1e, f). This range lies primarily in the NIR to IR spectrum. These results support our internal temperature results and suggest that YNP adults reflect more radiation that influences body temperature and may possess structures that can offload this radiation by reflecting it toward the environment. Reflective properties of the exoskeleton are important strategies in insect thermoregulation. Both chitinous and wax reflective properties can contribute to improved reflection, which help reduce body temperatures (McClain et al. 1991, Seago et al. 2009, Shi et al. 2015, Alves et al. 2018, Pavlović et al. 2018, Cuesta and Lobo 2019, Wang et al. 2021).

We then examined the ventral abdominal plate surfaces with scanning electron microscopy (SEM). We selected both Idaho and YNP adults for comparisons with the SEM and analyzed them to ensure the same settings were used while imaging individuals from both locations.

Images of ventral abdominal plates with and without the cuticular waxes removed suggested that there were no obvious structural differences between beetles from either location (Supplementary Figs. S6 and S7). The ventral surface appears as an arrangement of polygonal chitin tiles (Supplementary Fig. S7). These structures are unlike those found on the identified heat-reflective structures on C. bombycina, N. gigas, and G. goliatus, which rely on setae-like structures on the dorsal abdominal surface of each species to improve the reflective properties of their exoskeletons from solar insolation (Shi et al. 2015, Xie et al. 2019, Zhang et al. 2020).

Cicindelidia hemorrhagica is subject to NIR and IR radiation in YNP that is both generated by and reflected off the substrate, affecting internal body temperatures (Fig. 1f). Because there are virtually no setae on the venter of the abdomen, setae are not responsible for the heat reflectance we observed. Because we did not observe obvious structural differences on the ventral abdominal plates between YNP and Idaho adults, the heat-resistance mechanism might involve layering of the cuticle. Structural layering can be responsible for the coloring of the insect cuticle, which may in turn affect the reflective and thermoregulatory properties of the exoskeleton (Schultz and Hadley 1987). Elemental and chemical composition of the individual cuticle layers was not determined but would be useful for completing our understanding of how the cuticle influences IR reflectance.

Preadaptations of Cicindelidia hemorrhagica and Further Adaptations in Yellowstone National Park

That populations of C. hemorrhagica behave differently to the same abiotic factor (temperature) is, to our knowledge, unique. If YNP C. hemorrhagica has been selected for high-temperature resistance, as these results indicate, such a substantial adaptation likely occurred in only 5,000 to 15,000 generations. YNP was affected by the Bull Lake and Pinedale Glaciations. The Pinedale Glacier had an average ice thickness of 700 m, was the most recent glaciation in YNP, and started to recede about 15,000 years ago (Pierce 1982). This means that the extremophile organisms living in YNP had fewer than 15,000 years to adapt to living on the geothermal springs. Because most tiger beetle species typically have one generation per year (we confirmed at least a 1-yr generation time for C. hemorrhagica by lab rearing), we know that C. hemorrhagica in YNP had a maximum 15,000 years to adapt to not only the high temperatures associated with thermal springs, but also the high concentrations of heavy metals and extreme pH conditions (Adams et al. 2024). Indeed, 15,000 generations likely represent the maximum number of generations because we observed multiple larval stages at the same time in YNP. Variation in larval stages implies that harsh conditions, such as limited food availability, are likely to extend generation times to 2 or 3 yr (Pearson 1988).

The central portion of YNP is a broad, volcanic plateau approximately 2,500 m above sea level. Mountain ranges at the North, East, South, and Northwest side with ridges surround the plateau with an additional height of 600 m to 1,200 m. The Madison Plateau, located about 2,600 m above sea level, connects to the Southeast side of the central volcanic plateau of YNP. Therefore, based on the topography and thermal activity of YNP from which we can infer limited movement for C. hemorrhagica, the only possible routes for C. hemorrhagica into YNP 15,000 years ago would have been along the Madison Plateau or alongside outward streaming rivers such as the Yellowstone River.

The C. hemorrhagica populations have likely been relocating along with the shifts in thermal features. With their seemingly exclusive association with geothermal springs and high mountain range surroundings, any movement inside and outside of YNP is unlikely. We argue that YNP C. hemorrhagica is thermophilic rather than thermotolerant due to its exclusive association with thermal springs, larval burrows within 5 m of the thermal water, and its lack of cooling thermoregulatory behavior.

The question remains how exactly does YNP C. hemorrhagica cope with high temperatures (and other extreme conditions). There are limited examples of metazoans at temperatures greater than 50 °C and most of these animals become inactive or are sterile (Clarke 2017). Consequently, C. hemorrhagica is one of the most temperature-tolerant sexually reproductive metazoans currently known. Our results suggest that the already warm-adapted C. hemorrhagica resists bottom-up heating from the YNP geothermal springs by reflecting IR radiation. This heat reflection is attributable to the cuticular structure of the exoskeletal ventral sternites of the abdomen, forming a heat-resistant shield. However, we do not yet understand the precise structural feature that affords the greater heat reflection.

Although we have focused here on how C. hemorrhagica survives high temperatures, we have not yet addressed why. Elsewhere (Adams et al. 2024), we explored the effect of heavy metals on the depauperate trophic community associated with YNP hot springs. There, we noted that heavy-metal biomagnification in C. hemorrhagica indicates that only about 50% of its diet is associated with hot-springs community members, and the only available sources of alternative food are insects that are stunned by the heat and fall into the hot springs. This use of stunned insects as food has been noted for other tiger beetle species in thermal areas (Pearson and Vogler 2001). Thus, the thermophilic adaptations of C. hemorrhagica provide access to a unique food resource, in addition to affording thermal protection from extreme cold and reduced frequency of predators.

Acknowledgments

We thank A. Carlson (Yellowstone National Park) for assisting in the permit process and locating research sites. We thank M. Alleyne, J. Soares, M. Ali, and the University of Illinois at Urbana-Champaign’s Materials Research Lab, M. Hofland, A. Lingley, D. Juliano, V. Ferreira, M. Rolston, A. Massey, B. Schwartz, C. Donahoo, A. Piccolomini, and C. Dittemore (Montana State University), P. Higley, M. True, M. Webb, K. Lozano, M. Gardner, J. Campbell, E. Muslic, and E. Patton (University of Nebraska) for technical assistance.

Author contributions

Kelly Willemssens (Conceptualization [Lead], Data curation [Lead], Formal analysis [Lead], Investigation [Lead], Methodology [Lead], Project administration [Supporting], Resources [Supporting], Software [Equal], Supervision [Lead], Validation [Equal], Visualization [Equal], Writing—original draft [Lead], Writing—review & editing [Lead]), John Bowley (Conceptualization [Lead], Data curation [Lead], Formal analysis [Lead], Investigation [Lead], Methodology [Lead], Project administration [Supporting], Resources [Supporting], Software [Lead], Supervision [Lead], Validation [Lead], Visualization [Lead], Writing—original draft [Lead], Writing—review & editing [Lead]), Braymond Adams (Conceptualization [Supporting], Investigation [Supporting], Resources [Supporting], Writing—original draft [Supporting], Writing—review & editing [Supporting]), Monica Rohwer (Investigation [Supporting], Resources [Supporting], Supervision [Supporting], Writing—original draft [Supporting], Writing—review & editing [Supporting]), Miles Maxcer (Investigation [Supporting], Resources [Supporting], Supervision [Supporting], Writing—original draft [Supporting], Writing—review & editing [Supporting]), Chelsea Heveran (Conceptualization [Equal], Formal analysis [Equal], Investigation [Supporting], Methodology [Equal], Resources [Equal], Supervision [Supporting], Visualization [Equal], Writing—original draft [Supporting], Writing—review & editing [Supporting]), David Weaver (Conceptualization [Equal], Formal analysis [Equal], Investigation [Supporting], Methodology [Supporting], Resources [Supporting], Supervision [Supporting], Writing—original draft [Supporting], Writing—review & editing [Supporting]), Tierney Brosius (Investigation [Supporting], Resources [Supporting], Visualization [Equal], Writing—original draft [Supporting], Writing—review & editing [Supporting]), Erik Oberg (Investigation [Supporting], Resources [Supporting], Writing—original draft [Supporting], Writing—review & editing [Supporting]), Leon Higley (Conceptualization [Lead], Data curation [Lead], Formal analysis [Lead], Funding acquisition [Lead], Investigation [Lead], Methodology [Lead], Project administration [Lead], Resources [Lead], Software [Equal], Supervision [Lead], Validation [Equal], Visualization [Equal], Writing—original draft [Lead], Writing—review & editing [Lead]), and Robert Peterson (Conceptualization [Lead], Data curation [Lead], Formal analysis [Lead], Funding acquisition [Lead], Investigation [Lead], Methodology [Lead], Project administration [Lead], Resources [Lead], Software [Equal], Supervision [Lead], Validation [Equal], Visualization [Equal], Writing—original draft [Lead], Writing—review & editing [Lead])

Funding

This research was funded by Therion, LLC, the University of Nebraska-Lincoln, the Montana Agricultural Experiment Station, and Montana State University. This research was performed under U.S. National Park Service/Yellowstone National Park research permits #7092 and #8100.

Conflicts of interest. The authors declare no conflict of interest.

References