Extreme regional heterothermy during flight in diverse wild bats

Summary

Endotherms actively regulate their body temperatures to maintain relatively constant internal temperatures in heterogeneous environments, but experience regional variation in body temperature (heterothermy). This is especially pronounced in bats, which face locomotor and energetic challenges as the only flying mammals: cooling of the wing and its constituent muscles during flight could substantially impact performance and energetics. Here, we characterize thermoregulatory patterns in locomoting bats from 27 tropical and temperate species over a range of air temperatures by measuring core body temperature and wing muscle temperatures along a proximodistal axis. We find that bat core and wing temperatures are highly labile and are impacted substantially by air temperature. Measured wing muscle temperatures demonstrate the largest temperature gradient from core to locomotor muscle for active endotherms reported to date. These results expand our understanding of locomotor performance in bats, and suggest that thermoregulation and environmental conditions together shape locomotor performance in endotherms.

Graphical abstract

Highlights

• •Regional heterothermy affecting wing muscles occurs in tropical and temperate bats
• •Air temperature has a significant effect on wing and core body temperatures
• •Flight results in substantial changes in wing and core body temperatures
• •Bat wing muscles are more vulnerable to air temperatures further from the core

Ecology; Zoology; Animal Physiology

Introduction

Temperature has a pervasive effect on biological systems. Homeothermic endotherms, such as birds and mammals, maintain body temperatures that are relatively constant and higher than those of the surrounding environment via endogenous heat production, which may enable critical physiological systems to function optimally despite unpredictability in external conditions.^1,2 However, heat transfer is an unavoidable consequence of maintaining a body temperature that differs from ambient: warm endotherms typically lose heat to their environments, resulting in regional heterothermy, or differences in temperature among body regions, especially in the periphery of the body vs. the core.^2,3 Insulation by feathers, fur, and/or fat can slow and ultimately reduce heat loss across the skin.^4,5,6 Heat balance can also be altered by actively modifying peripheral insulation, such as by erecting feathers or hair, or by changing patterns of blood flow, thereby decreasing temperature variation in the core.^3,6,7,8,9 For many endotherms, this is most evident in variation in skin temperatures with environmental conditions.^10,11

Bats are small, volant, endothermic mammals. Bat flight is enabled by the wings, which are highly modified forelimbs with a flight surface composed of very thin skin, typically around 10 μm in small bats.¹² The wings are well-vascularized with a large surface area, thus predisposing bats to heat loss and intensifying regional heterothermy in normothermic bats by cooling the wings substantially.¹³ Regional heterothermy in bats is especially remarkable in comparison with that of other endotherms because this peripheral cooling affects critical components of the locomotor system, including the flight muscles located in the arm, forearm, and hand. Temperature differences among body regions may be exacerbated in bats by temporal heterothermy, given that a number of both temperate and tropical bat species use torpor to conserve energy, some seasonally and some for shorter periods.^{14,15,16,17,18,19} Seasonal use of torpor has long been recognized and studied in temperate bat species, which regularly employ bouts of hypothermia accompanied by lowered metabolic rates.^14,20 More recently, a growing body of research demonstrates widespread use of torpor by tropical bats to reduce thermoregulatory burden, associated with bouts of hypo- or hyperthermia, depending on species and ambient conditions.^16,17,21 Bats of some species are flexibly able to initiate and maintain flight over a relatively wide range of core body temperatures and environmental temperatures as well. For example, Eptesicus fuscus can fly at body temperatures <30°C in the lab and the field, while typical mammalian normothermic temperatures are 30°C–41°C.^22,23

Because both core body temperature and peripheral temperature are highly labile in bats, which employ a high-intensity locomotor mode, the study of the effects of environmental temperature on physiology and flight performance offers great potential for new insights into constraints on performance faced by endotherms. In particular, this topic may shed light on the constraints imposed by the temperature sensitivity of components of the locomotor system, which are rarely considered. Moreover, bats are globally distributed, with species ranges extending into high latitudes, while the greatest species diversity is in the tropics.²⁴ Thus, bats of similar anatomy or evolutionary history may contend with radically different environmental conditions around the world, and effective mechanisms for heat dissipation may vary among environmental conditions.

To better understand the interaction of environmental temperature and regional heterothermy in bat wings, we measured core and wing muscle temperatures in bats from diverse species at a range of air temperatures at field sites in southeastern Arizona, USA, and northern Belize. These field sites, one temperate and one tropical, differ substantially in temperature and humidity. We hypothesize that, despite likely differences in how tropical and temperate bats regulate their resting body temperatures, regulation of body temperature during activity will be influenced more by environmental temperature than physiology, due to thermoregulatory constraints inherent to bats’ unique anatomy. We predict that (1) muscles in the forearm and hand are significantly cooler than muscles in or closer to the core during rest and activity; (2) there is a relationship between air temperature and distal muscle temperature, but not between air temperature and proximal muscle temperature; (3) flight cools distal muscles, but not proximal muscles.

Results

We collected temperature data at multiple points along the wing (Figure 1), as well as rectal temperature, a proxy for core temperature, in three different measurement conditions: at rest before flight (pre-flight), immediately after flight (flight), and after resting for 3 min post-flight (post-flight). In total, we sampled 86 individuals (33 in Belize and 53 in Arizona; Figure 2), from 27 species, primarily from the families Vespertilionidae and Phyllostomidae, but also from Molossidae and Mormoopidae. Four individuals are represented in the dataset twice, because they were measured at two different air temperatures. Although data were collected over two-week periods at similar times of year, air temperatures were higher in Belize (mean: 26.3°C, range: 24.3°C–28.7°C) than in Arizona (mean: 14.1°C, range: 9.2°C–22.3°C). The average humidity in Arizona was approximately 28.5%²⁵ across the days and hours we collected data, and in Belize, approximately 85.7% (Rummel, unpublished data). All measurements were taken in ambient conditions, and air temperatures reflect field conditions at the time of measurement, such that the Belizean taxa were evaluated only at air temperatures that do not overlap with the Arizonan air temperatures.

Figure 1.

Bat in flight

Color-coded labels indicate anatomical locations of temperature measurements.

Figure 2.

Evolutionary relationships and sample sizes of the bats in this study

Color coding indicates the location of the field site sampling.

Measurement condition (pre-flight, flight, or recovery), air temperature, body region, and their three-way interaction were all significant predictors of the temperature of both the core and the wings in bats. In each measurement condition, all slopes of the relationship between body region temperature vs. air temperature differed from zero (Figure 3; Table 1, CI: the 95% confidence intervals did not include zero), indicating that air temperature had a significant effect on the temperatures of all wing regions as well as rectal temperature, our proxy for core body temperature. We evaluated whether body regions differed from each other in temperature magnitude and temperature trend by evaluating contrasts among regions within each treatment (Figure 3; Table 2). Rectal and pectoralis temperatures did not differ significantly from one another under any measurement condition, but all other pairwise comparisons between regions differed significantly (Table 2). At 19°C, which was the mean air temperature of the dataset, the biceps and rectal temperature differed by 2°C and the wrist was 11.9°C colder than rectal temperature in the pre-flight condition. This difference increased to 15.3°C in the flight condition (Table 3).

Figure 3.

Body temperature as a function of air temperature

Body temperature plotted against air temperature for (left) pre-flight, (middle) flight, and (right) recovery.

Data points and lines are colored by body region according to Figure 1 and legend at right. Lines represent predictions from the best-fitting model, which includes the interaction of anatomical region, measurement condition (pre-flight, flight, or recovery), and air temperature as fixed effects. Individual was included as a random effect.

Table 1.

Slopes ±standard error of the relationship between body region temperature and air temperature in the three different treatments

	Pre-flight	Flight	Recovery
Rectal	0.151 ± 0.0376	0.320 ± 0.0373	0.261 ± 0.0378
Pectoralis	0.108 ± 0.0371	0.298 ± 0.0373	0.213 ± 0.0376
Biceps	0.202 ± 0.0371	0.394 ± 0.0373	0.325 ± 0.0376
Forearm	0.647 ± 0.0371	1.214 ± 0.0373	0.885 ± 0.0376
Wrist	0.652 ± 0.0371	1.064 ± 0.0373	0.902 ± 0.0377

Table 2.

Comparisons between body regions

	Pre-flight				Flight				Recovery
	estimate	p-value	slope	p-value	estimate	p-value	slope	p-value	estimate	p-value	slope	p-value
Rectal - Pectoralis	0.39	0.511	0.043	0.80	0.329	0.6848	0.022	0.978	0.451	0.3725	0.0485	0.4044
Rectal - Biceps	2.159	<0.001	−0.052	0.66	2.782	<0.001	−0.073	0.307	2.411	<0.001	−0.064	0.480
Rectal - Forearm	7.755	<0.001	−0.500	<0.001	11.449	<0.001	−0.893	<0.001	8.339	<0.001	−0.624	<0.001
Rectal - Wrist	11.847	<0.001	−0.502	<0.001	15.260	<0.001	−0.744	<0.001	12.116	<0.001	−0.641	<0.001
Pectoralis - Biceps	1.771	<0.001	−0.0946	0.09	2.454	<0.001	−0.0957	0.09	1.960	<0.001	−0.112	0.03
Pectoralis - Forearm	7.366	<0.001	−0.539	<0.001	11.120	<0.001	−0.916	<0.001	7.888	<0.001	−0.672	<0.001
Pectoralis - Wrist	11.458	<0.001	−0.544	<0.001	14.931	<0.001	−0.766	<0.001	11.664	<0.001	−0.689	<0.001
Biceps - Forearm	5.596	<0.001	−0.444	<0.001	8.666	<0.001	−0.820	<0.001	5.928	<0.001	−0.560	<0.001
Biceps - Wrist	9.687	<0.001	−0.450	<0.001	12.477	<0.001	−0.670	<0.001	9.705	<0.001	−0.577	<0.001
Forearm - Wrist	4.092	<0.001	−0.005	0.99	3.811	<0.001	0.150	<0.001	3.776	<0.001	−0.0167	0.99

Contrasts between temperature estimates for each body region in each measurement condition, calculated at an air temperature of 19°C; and contrasts between the slopes of the relationship between body region temperature and air temperature. Estimates and slopes were calculated using our best-fit model and the emmeans package in R. p-values less than 0.05 are shown in bold.

Table 3.

Comparisons between measurement conditions

	Pre-flight – flight				Pre-flight – recovery				Flight – recovery
	estimate	p-value	slope	p-value	estimate	p-value	slope	p-value	estimate	p-value	slope	p-value
Rectal	−1.398	<0.001	−0.17	<0.001	−0.703	0.0159	−0.11	0.01	0.695	0.0184	0.056	0.29
Pectoralis	−1.458	<0.001	−0.19	<0.001	−0.640	0.0294	−0.10	0.02	0.818	0.004	0.085	0.07
Biceps	−0.775	0.0051	−0.19	<0.001	−0.451	0.1713	−0.12	0.004	0.324	0.4124	0.068	0.18
Forearm	2.295	<0.001	−0.57	<0.001	−0.119	0.88	−0.24	<0.001	−2.414	<0.001	0.33	<0.001
Wrist	2.015	<0.001	−0.41	<0.001	−0.434	0.199	−0.25	<0.001	−2.449	<0.001	0.16	<0.001

Contrasts between temperature estimates for each body region across measurement conditions, calculated at an air temperature of 19°C, and contrasts between the slopes of the relationship between body region temperatures and air temperature in each condition, preflight, flight, and recovery, for each body region. p-values less than 0.05 are shown in bold.

We tested whether the temperature of each region and its relationship to ambient temperature differed among pre-flight, flight, and recovery conditions by evaluating contrasts between conditions (Figures 3 and 4; Table 3). For all body regions, pre-flight temperatures differed significantly from flight temperatures: flight temperatures were greater than pre-flight temperatures for rectal, pectoralis, and biceps, and cooler for forearm and wrist at the median air temperature, 19°C (Figure 4; Table 3). For all regions, pre-flight temperatures also differed significantly from recovery temperatures: recovery temperatures were greater than pre-flight temperatures for all regions (Table 3). Flight temperatures were significantly lower than recovery temperatures for the distal wing regions (i.e., wrist and forearm) (Table 3). Considering the full range of air temperatures tested, rectal and pectoralis temperatures stayed the same or rose during flights at air temperatures above 12°C. Below 12°C, flight temperatures for these regions declined relative to pre-flight temperatures (Figure 5; Table 3). The biceps, forearm, and wrist experienced in-flight cooling relative to pre-flight temperatures even at much warmer air temperatures: the biceps cooled at air temperatures below approximately 15°C, while the forearm and wrist experienced cooling at air temperatures as high as 24°C (Figures 4 and 5).

Figure 4.

Body region temperatures across measurement conditions

The difference between measured muscle temperatures and rectal temperature (muscle temperature minus rectal temperature) for each body region (panels arranged proximal to distal, left to right) and measurement time point. Within a panel, points from a single individual are connected by lines. Points and lines are color-coded by air temperature according to the scale on the right.

Figure 5.

Pre-flight and flight temperature trends

Superimposition of pre-flight (dashed) and flight (solid) data and best-fit lines for the relationship of body vs. air temperature, illustrating the different relationships of body region temperatures with air temperatures pre-flight and during flight. The temperatures at which the pre-flight and flight lines intersect are hypothesized to be energetic transition points.

Discussion

Our data show that flight and environmental temperature play major roles in dictating the highly variable temperatures of wing muscles in bats. The differences we report between distal muscle temperatures and proximal muscle and core temperatures, at rest and particularly during flight, are extreme relative to those experienced by other endotherms that have been studied. While regional heterothermy has been described in endotherms across a range of sizes and ecologies, temperature differences–e.g., peripheral cooling–typically impact the skin, blubber, or regions of the body that lack musculature, thus reducing risk of compromising muscle function and locomotor performance in variable environments.^10,26,27,28 Our data show that this is not the case for bats, in which wing muscle temperatures drop substantially as environmental temperature falls. We observed temperature differences between the most proximal vs. most distal sites as great as 20°C at the lowest ambient temperatures (Figure 3).

Remarkably, flight changed temperatures in all regions of the wing and increased the difference between body and distal wing temperatures most substantially in colder air. Flight tended to result in warmer temperatures proximally, with increases in pectoralis and rectal temperatures after flight at all but the lowest air temperatures (T_a < 12°C) (Figure 5). The distal muscles of the bat wing were much more strongly affected by ambient temperatures, however. We conclude that bat wing muscles are more vulnerable to environmental temperatures with increasing distance from core. Even when core T_b is tightly regulated, distal muscle temperatures are not, and they fluctuate substantially with environmental temperature.

In-flight thermoregulation and the effect of air temperature

Flight produces a thermal environment that differs substantially from that of terrestrial locomotion, due to forced convection across the wings of flying animals. In addition, radiative heat loss to the night sky, which can cool an object to temperatures below air temperature, can be a major influence on heat balance in bats.^29,30 These avenues of heat transfer likely result in net heat loss across the large surface area of the wings under most conditions in which bats are active. The core-shell paradigm of endothermic thermoregulation says that as ambient temperature falls, peripheral cooling will help limit heat loss and maintain core body temperature at the setpoint.^2,3,31

Our observations that the wing muscle temperatures of diverse species in temperate and tropical environments fall during nocturnal flight are in line with expectations based on typical thermoregulatory patterns in endotherms. Large flight muscles whose mass is primarily situated in the core, such as the pectoralis and latissimus dorsi, generate heat during activity, which warms the core and is then circulated via blood flow to the rest of the body.^32,33 The wings are part of the “shell,” and it is likely to be energetically favorable overall to limit heat loss by allowing wing temperature to differ substantially from that of the core due to the large wing surface area and high degree of vascularization.³⁴ Indeed, heat loss occurs at a rate proportional to the temperature difference between source and sink. Consequently, maintaining all wing muscles at core body temperature would require that bats deliver a great deal of heat to the wing via blood flow, which would require increased heat production in the core. The potential for rapid heat loss across the wing skin, therefore, makes it unlikely that maintaining constant core temperature across the wing would be metabolically sustainable.

The pattern of wing temperatures we measured likely results from a combination of factors. Heat loss across the wing, heat delivery to the wing via blood circulation, and heat generation by the muscles of the arm, forearm, and hand during flight all contribute to muscle temperatures at any given time.^29,35,36,37 Pre-flight and flight fit lines intersect at temperatures that likely indicate the points at which tradeoffs between these factors become energetically significant (Figure 5). We suggest that these intersection points (about 15°C for the biceps, and 23°C–24°C for the forearm and wrist) represent the air temperatures at which it becomes energetically favorable for bats to conserve energy by letting peripheral muscle temperature fall via convective heat loss and vasoconstriction. Conversely, in warmer environmental conditions, heat dissipation may increase in importance and may be enhanced by facilitating the warming of the wings, particularly the handwing, which will increase heat flux from the wing to the environment. The large, mostly hairless surface of the wings thus makes them vital heat exchange organs, but may also preclude daytime flights because of the risk of radiative warming resulting in hyperthermia.^31,38

We found substantial variation in core body temperatures (measured via rectal temperature) among species and measurement time points (Figure 3). Air temperature and the temperature of measurement regions, including rectal temperature, are positively correlated, i.e., bats tended to have higher rectal temperatures at higher air temperatures. At low air temperatures, bats had low rectal temperatures but were still able to initiate and sustain flight (Figures 3 and 4). Indeed, some of the core body temperatures (T_b < 35°C) we observed at low air temperatures are below measured thresholds for torpor onset in some species, though in line with reports of low active T_b in several bat species.^39,40 This may be related to the tendency toward heterothermy among many of the species we were able to evaluate at low air temperatures and their capacity for initiating flight at ambient temperatures that are several degrees below normothermic body temperature.^23,41 While some bats may depress T_b without entering torpor, reduced temperatures in the flight muscles may still impact flight performance and flight initiation^42,43,44 (see later in discussion). At the warmest air temperatures in our dataset, core body temperatures of bats in flight were as high as 42°C, with measured pre-flight rectal temperatures up to 40°C. These rectal temperatures are among the highest reported for bats, though they are similar to T_b seen in other active endotherms in warm environments.^{45,46,47,48,49,50,51}

Implications for flight performance

Some degree of regional heterothermy is likely common among diverse endotherms due to the physical constraints of maintaining a core temperature that differs from environmental temperature.^2,6 However, cool wings likely impose a tradeoff whereby the muscles that power flight and control wing conformation are subjected to highly variable temperatures. Muscle function is highly temperature-sensitive, and contractile rates decline relative to their optimum as temperature decreases.⁵² Bat wing muscles are no exception; their contractile rates drop at temperatures below 37°C.^43,44,53 However, bat wing muscles distal to the shoulder, such as the extensor carpi radialis longus (ECRL) and an interosseous muscle, display lower temperature sensitivity than expected based on other mammalian muscles.^44,54,55 In the tropical bat Carollia perspicillata, these two muscles are significantly less temperature-sensitive than the pectoralis muscle, which suggests that there is some physiological and/or biochemical compensation in muscles that regularly operate at variable or lower temperatures.⁵⁶

In this study, we found that bats, particularly Myotis species and E. fuscus, initiated and maintained normal flights at body temperatures and pectoralis temperatures around 30°C, particularly at low ambient temperatures. Temperate bats have been observed making flights at subzero environmental temperatures in the wild and are known to roost in hibernacula with microclimate temperatures that are close to freezing.^23,57,58,59 In the insectivore E. fuscus, the pectoralis muscle also shows lower temperature sensitivity than expected based on muscles of other mammals, and could be related to their use of torpor and occasional flight at cool normothermic temperatures.^43,55 These observations suggest that bats are able to arouse from torpor and initiate and maintain flight at low air and body temperatures. This likely places intense demands on the locomotor system, because the pectoralis and other flight muscles must generate adequate power, as well as activation and relaxation rates, to initiate and maintain flight.^60,61 Because muscle contractile rates are temperature-sensitive, maintaining pectoralis function at low temperatures likely requires physiological compensation, as suggested by previous work.^43,56,62 The data presented here lend support to the hypothesis that activity at low body temperatures imposes a need for adaptation in the thermal performance of muscles that are important for flight.

In Arizona, where the lowest air temperatures we report occurred, we observed that insect activity declined steeply at around 10°C. Bat activity also decreased at that temperature: bats tended to be torpid or were unwilling to fly even when body temperatures were close to normothermic. Even bats that were apparently torpid would sometimes fly briefly, but at these torpid body temperatures, they crawled more often than they flew. We suggest that there may be a biomechanical as well as an energetic limit to performance in the cold; below a threshold temperature, critical flight muscles cannot generate the power necessary for flight. Conversely, declining muscle function at high temperatures may also contribute to poor flight performance, although systemic effects of overheating may arise before muscle temperature effects become apparent.^47,63

Effectiveness of experimental design in evaluating flight-proxy temperatures

Measuring body temperatures in free-moving animals is challenging, although the use of temperature-sensitive loggers and transmitters inserted subcutaneously or glued onto the skin has become increasingly popular recently.^19,64,65 These methods can be used to measure or approximate core body temperatures, but they do not allow the direct measurement of other body regions that may affect energy balance or locomotor performance in small animals. Rummel et al.¹³ directly and continuously measured muscle temperatures of flying bats in a laboratory setting. Their methodology is difficult to apply outside of the laboratory due to the need for an instrumentation tether for continuous temperature recording, as well as the limited range of species that thrive in captivity. By comparing measurements made immediately after flight bouts to those made pre- and a few minutes post-flight (i.e., after a few minutes of recovery), we effectively assessed the thermal status of the core and flight muscles from numerous animals in performance-relevant conditions.

The temperatures of the more distal regions of the wing were lower than those of the more proximal wing in all bats we studied. These results are consistent with those of temperatures measured continuously during flight.¹³ However, the magnitude of regional heterothermy in controlled wind tunnel experiments was greater than that which we report here. On average, during wind tunnel flights at an environmental temperature of approximately 22°C, the pectoralis was 2.4 ± 0.7°C, the biceps 4.4 ± 1.0°C, and the forearm 11.6 ± 0.4°C cooler than rectal temperature. Estimated muscle–rectal temperature differences during the naturalistic flights observed here, based on our linear mixed model were: pectoralis: 0.4 ± 0.5°C, biceps: 2.6 ± 0.5°C, and forearm: 8.8 ± 0.5°C. This difference could arise from any of several sources. First, the wind tunnel measurements were made only while the bat was in flight, during active convective heat loss. Here, our measurements were made in the field with the bat in hand, before and after active convection. Thus, the in-flight temperatures we report here may be overestimates, and true in-flight temperatures are likely lower than our measurements. It is also possible that an as-yet unknown degree of vasodilation occurs after flight; this would allow some peripheral rewarming, even in the short period between the end of a flight and our measurements.

Additionally, body temperature can be affected by stress-induced hyperthermia (SIH). Handling stress can cause SIH in small bats and birds and is usually attributed to the peripheral vasoconstriction and shivering that result from hypothalamus stimulation.^49,66,67 In silver-haired bats (Vespertilionidae: Lasionycteris noctivagans), SIH may be context-dependent, but in captivity results in an increase in body temperature in bats handled after temperature treatments.⁶⁷ Repeating measurements across conditions (pre-flight, flight, recovery) minimizes the impact of potential hyperthermia due to handling stress. Because our protocol was the same at each time point, handling stress is similar for all conditions, although we cannot rule out all possible effects, and we consider our comparisons to reflect real, biologically relevant differences in body temperature between measurement conditions.

Humidity differed substantially between our sites: our field site in Arizona is situated in the dry American Southwest, while our field site in Belize is a tropical rainforest. Humidity impacts thermoregulation by altering heat dissipation via evaporative water loss. Evaporative water loss is minimized in many bats at moderate T_a, presumably to reduce water loss; this may be especially pronounced in bat species dwelling in arid environments.⁶⁸ Our measured air temperatures were typically under 30°C, even in Belize, and these ambient conditions were not likely to be particularly stressful. This suggests that evaporative heat loss may not be a primary mechanism for maintaining heat balance for the bats in this dataset, and more broadly for bats in nocturnal flight, during which convective and radiative cooling likely predominate.^29,68 Thus, the effect of humidity is likely more important when the bat is at rest. However, it is unclear to what extent and in what contexts bats may dissipate heat via evaporative water loss vs. convection for the species in this study.⁶⁹ Evaporative heat loss may have been less effective in higher humidity conditions, resulting in higher body temperature measurements from Belize vs. Arizona.

Conclusions

Bats experience a range of core and peripheral body temperatures at rest and during flight under naturalistic conditions. Core body and wing temperatures are affected by air temperature, which varies over the course of a season and even a single night. We specifically examined the temperatures of muscles in the wing, which are critical elements of the locomotor system but whose functions are temperature-sensitive. Our findings contribute novel information regarding the complex interplay of bat thermoregulatory patterns, environmental temperatures, and flight, demonstrating that air temperature has a substantial impact on wing muscle temperatures and suggesting that wing temperatures are actively regulated during flight. We suggest that though endotherms are capable of maintaining stable core body temperatures in the face of environmental heterogeneity, environmental temperatures have a substantial impact on body temperature and critical organismal function.

Limitations of the study

To our knowledge, this is the most comprehensive assessment of thermoregulatory patterns in bat wings to date, but we acknowledge limitations that can be addressed in future work. First, we note that our taxon sampling primarily includes representatives from two Chiropteran families. Second, we made discrete measurements of body region temperatures with bats in hand, and we were unable to measure body region temperatures when bats were freely flying. Thermal cameras are a promising avenue for non-invasively measuring body surface temperatures during active locomotion.³⁰ However, high cost and the challenges of achieving adequate resolution when observing flying animals currently limit their use. As instrumentation improves and costs decrease, thermal videography is likely to prove an effective tool for future research. Finally, we do not investigate wing membrane temperatures here, but it is likely that the membrane itself is critically important in thermoregulation and is a rich area for future research.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Andrea D. Rummel (arummel@rice.edu).

Materials availability

No new unique reagents were generated in this study.

Data and code availability

• •The dataset supporting this article has been uploaded as part of the supplemental information.
• •Original code supporting this article has been uploaded as part of the supplemental information.
• •Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.

Acknowledgments

We thank Sakthi Swarrup, Aaron Corcoran, Joao Lima, Kenny Breuer, and Brandon Hedrick for their help collecting data in the field. Arizona-based data collection was funded by NSF IOS-1931135 to S.M.S.

Author contributions

Conceptualization: A.D.R. Methodology: A.D.R. Investigation: A.D.R., B.L.Q., A.D.K., and S.M.S. Formal analysis: A.D.R., B.L.Q., and A.D.K. Writing – original draft: A.D.R. All authors edited the article, gave final approval for publication, and agree to be held accountable for the content of this article.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Deposited data
Raw data and code for analysis	This paper	–
Experimental models: Organisms/strains
Bats	Wild	N/A
Software and algorithms
R version 4.4.1	R Core Team	https://www.r-project.org/

Experimental model and subject details

Bat capture procedures

Bats were captured in mist nets and harp traps at the Southwestern Research Station in Portal, AZ, in late May 2022 and early June 2023, and at the Lamanai Archaeological Reserve and Lamanai Outpost Lodge near Indian Church, Belize, in late April and early May 2024. We included all adult, non-reproductive animals available for study during the period of field residency that made voluntary flights lasting at least 1 min, totaling 86 individuals, of which 15 were female, 53 were male, and 18 were of unknown sex. We attempted to obtain as large a range of species as possible with respect to family, body mass, and feeding ecology.

Ethics

Arizona experiments were approved under Brown University IACUC 19-6-0004 and State of Arizona Game and Fish Department permit numbers SP776270 and SP033903. Belize experiments were covered under Brown University IACUC 21-03-0001 and Rice University IACUC-24-015, and Belize Forestry Department Scientific Research and Collecting Permit WL/1/24 (06). We followed USDA and American Society of Mammologists standards for the capture and handling of wild bats.⁷⁰

Method details

Flight arena

Bats were flown either on night of capture or the subsequent night in one of two flight tents: a large, custom-built tent with dimensions 4.9 × 6 × 3.6 m (l × w × h), and a smaller, commercially available tent with dimensions of 3.6 × 3 × 2.7 m (l × w × h) (Coleman Back Home Screenhouse). Bats were only flown on nights when tent wind speeds were undetectable.

Temperature measurements

We measured muscle temperature, T_m, at the surface of a muscle by pressing a type K thermocouple probe with a 1 mm tip onto the muscle of interest and reading the temperature with a handheld thermocouple thermometer (Omega HH912T, accuracy: 0.04% ± 0.3°). These surface measurements were not substantially different (<1°C) from measurements we made with thermocouple probes inserted into a muscle of interest (see¹³ for details of insertion probes) and so we made only these surface T_m measurements, which are less invasive and less stressful for the animal. We measured temperature at four anatomical locations along the ventral aspect of one wing’s proximodistal axis in the following order for each bat: the wrist, at the origin of several hand muscles; the extensor group of the forearm; biceps; and pectoralis muscles (Figure 1). We measured rectal temperature last by inserting a lubricated thermocouple approximately 1 cm into the rectum. We measured air temperature before each measurement of body temperature; we did not record humidity in the field, but because humidity may impact heat dissipation via evaporative heat loss, we note approximate humidity measurements at the Arizona site based on nearby weather station data for the dates of our data collection and at the Belize site based on data we collected across a similar time frame in 2025 (unpublished data).

Experimental design

We treat temperature measurements made immediately before flight as baseline core and muscle temperatures and designated these as “pre-flight” measurements which are representative of resting conditions. Bats were released into the tent immediately after pre-flight measurements and allowed to fly for at least 1 min and up to 20 min. Mean flight time was 8.4 ± 5.5 min. Bats were captured by hand or in hand nets after the conclusion of their flight, and temperature was measured as soon as possible after flight stopped. We designate these as “flight” measurements. Subjects were then returned to their small cloth holding bags to rest, and temperatures were measured again 3 min after the bat was returned to its bag; we designate these as “recovery” measurements. Because we hypothesized that flight distal muscle temperatures would be lower than pre-flight temperatures, we made this additional set of “recovery” measurements to evaluate whether (1) temperatures after the 3 min of rest better reflect flight or resting conditions and (2) temperatures return relatively quickly to those of resting conditions after recovery from flight. Each set of measurements took less than 2 min.

Quantification and statistical analysis

Phylogenetic signal assessment

We first assessed whether there was phylogenetic signal in our data, given the comparative nature of our study. Because phylogenetic methods typically require one trait value per species, we selected a subset of our data that included the species for which we had measurements at multiple air temperatures (19 of 27 species) (see also Rummel et al.¹³). We then calculated the difference between core temperature (measured as rectal temperature) and body region temperature for each individual within each species, and regressed this difference on air temperature. The slope of these relationships (body region temperatures vs. air temperature) was used to assess phylogenetic signal via Pagel’s lambda for each body region (pectoralis, biceps, forearm, wrist) individually, using the function phylosig in phytools and a tree pruned to our species subset.^71,72,73 We found no significant phylogenetic signal for any of the four body regions (Table S1), so we proceeded without correcting for phylogeny in subsequent analyses.

Linear mixed models

We used linear mixed models to evaluate variation in recorded temperatures using the lmer function in the package lme4 in R.^74,75 Models were fitted via maximum likelihood with combinations of parameters, including air temperature, body mass, body region, field site, and measurement time, and then compared via likelihood ratio tests using the anova function in R. We included individual as a random effect to account for inter-individual variation, as well as intra-individual variation resulting from the repeated measures structure of the data. Our final model included the interaction of air temperature, body region, and measurement time. Including body mass in a four-way interaction resulted in approximately the same AIC value as a three-way interaction without body mass.

When we included field site as an additional predictor variable we found a significant interaction between field site, air temperature, and body region. A post-hoc test to compare the slopes of the relationship between body region temperature and air temperature at each timepoint and body region between the two field sites (emmeans function in R) found that only the slopes of wrist temperatures as a function of air temperature were significantly different between the two field sites at each measurement time point (Figure S1; Table S2). The significant interaction between field site, air temperature, and body region thus seems to be driven solely by muscle temperatures at the wrist. AIC values for the models with and without field site as a predictor differed only marginally (with field site: AIC = 5144.82 and AIC = 5137.37, respectively). We thus chose to use the simpler model without field site as a predictor, reporting here the results for this final model only, which included neither body mass nor field site.

Comparisons across body regions

We evaluated differences in the relationship between air temperature and body region across all three measurement conditions using the final, best-fitting model. We used the emmeans package and pairs function in base R to test pairwise comparisons within body region across measurement conditions and within measurement conditions across body region (with correction for multiple comparisons).⁷⁶ We tested the following hypotheses: (1) forearm and wrist temperatures will be significantly lower than rectal, pectoralis, and biceps temperatures in all measurement conditions (i.e., pre-flight, flight, and recovery); (2) the slope of the relationships between forearm and wrist temperatures and air temperature will be steeper than those of rectal, pectoralis, and biceps temperatures with air temperature; (3) post-flight measurements from the forearm and wrist will be significantly lower than pre-flight or flight temperatures, which will not differ significantly.

Published: September 27, 2025