ABSTRACT
We used thermal imagining and heat balance modelling to examine the thermal ecology of wild mammals, using the diurnal marsupial numbat (Myrmecobius fasciatus) as a model. Body surface temperature was measured using infra-red thermography at environmental wet and dry bulb temperatures of 11.7–29°C and 16.4–49.3°C, respectively; surface temperature varied for different body parts and with environmental temperature. Radiative and convective heat exchange varied markedly with environmental conditions and for various body surfaces reflecting their shapes, surface areas and projected areas. Both the anterior and posterior dorsolateral body areas functioned as thermal windows. Numbats in the shade had lower rates of solar radiative heat gain but non-solar avenues for radiative heat gain were substantial. Radiative gain was higher for black and lower for white stripes, but overall, the stripes had no thermal role. Total heat gain was generally positive (<4 to >20 W) and often greatly exceeded metabolic heat production (3–6 W). Our heat balance model indicates that high environmental heat loads limit foraging in open areas to as little as 10 min and that climate change may extend periods of inactivity, with implications for future conservation and management. We conclude that non-invasive thermal imaging is informative for modelling heat balance of free-living mammals.
INTRODUCTION
Most mammals (69%) are nocturnal (the presumed ancestral state), some (20%) are diurnal and a few are cathemeral (8.5%) or crepuscular (2.5%; Bennie et al., 2014). Nocturnal activity, especially for small species that have a relatively high surface-to-volume ratio and therefore experience potentially high rates of heat exchange with their environment, is one mechanism by which mammals can inhabit environments characterised by limited water availability and high environmental temperatures (Walsberg, 2000; Withers et al., 2004). Nevertheless, some small mammals, even those in hot and arid environments, are diurnal and contend not only with high ambient temperatures but also the considerable heat load imposed by solar radiation. Examples include ground squirrels of Africa (Xerus spp.; Fick et al., 2009) and North America (Ammospermophilus, Otospermophilus and Ictidomys spp.; Chappell and Bartholomew, 1981a,b; Vispo and Bakken, 1993; Walsberg, 2000), spiny mice (Acomys spp.; Levy et al., 2016) of Africa and the Middle East and an Australian marsupial, the numbat (Myrmecobius fasciatus; Friend, 1986). Diurnal activity allows these small mammals to exploit high environmental temperatures to reduce the energetic costs of endothermy (van der Vinne et al., 2015) and some typically nocturnal mammals become diurnal when they are energetically challenged by cold or limited food (Hut et al., 2012; Levy et al., 2019).
Some small diurnal mammals have pelts that facilitate high levels of solar heat gain, presumably so they can maximally exploit solar radiation to reduce their requirement for metabolic heat production (Walsberg and Wolf, 1995; Cooper et al., 2003a). However, environmental heat loads can become extreme at high temperature and intense solar radiation, greatly exceeding metabolic heat production and evaporative heat loss, thereby limiting habitat use and the timing and duration of activity and foraging periods (Chappell and Bartholomew, 1981a,b; Vispo and Bakken, 1993; Fick et al., 2009). Consequently, environmental conditions can have considerable ecological impacts, interacting with competition, predation and resource availability to determine when and where small diurnal animals can be active (Levy et al., 2016, 2019). The unprecedented rates of environmental change now occurring due to human-induced global warming mean that it is critical that we understand the ecological consequences of changes in environmental conditions if we are to predict future patterns of species distribution and abundance, and conserve and manage them (Chown and Gaston, 2008; Williams et al., 2008; Fuller et al., 2010).
Temperature exchange occurs across the body surface of animals via conduction, convection, radiation and evaporation (Porter and Gates, 1969). Endotherms can acutely manipulate this exchange by modifying their metabolic heat production and evaporative water loss, altering their insulation and patterns of blood flow, and changing their posture and microclimate (Tattersall and Cadena, 2010). Core body temperature (Tb) of endothermic mammals and birds can differ substantially from environmental temperature and there is typically a temperature gradient from the endotherm's core across its surface to the environment (Porter and Gates, 1969; Speakman and Król, 2010). Consequently, measuring the surface temperature (Ts) along with microclimate data provides information about thermal exchange between the endotherm and its environment (Tattersall, 2016).
Infrared thermal imaging allows for real-time measurement of Ts for multiple parts of an animal's surface simultaneously from a distance and without contacting or instrumenting the animal. This avoids any potential effects of stress or restraint and allows measurements for wild, free-living animals undertaking normal behaviours in their natural environment (McCafferty et al., 1998; Speakman and Ward, 1998; McCafferty, 2007; Tattersall and Cadena, 2010). All objects, including animals, with a temperature above 0 K emit electromagnetic radiation, the wavelength and intensity of which are related to the object's surface temperature (Speakman and Ward, 1998). Modern thermal imaging cameras and associated analysis software convert this electromagnetic radiation into a thermogram using the entire detected radiation signal along with background and atmospheric temperature (Tattersall, 2016). Thermal imaging is a particularly useful tool for exploring the thermal ecology of species that are logistically difficult to catch and/or maintain in captivity, and those that are sensitive or threatened where there are additional legal and ethical issues associated with invasive procedures, due to its a non-contact, non-invasive nature; it is painless with no harmful effects (Tattersall and Cadena, 2010; Usamentiaga et al., 2014). Thermal imaging, together with measurement of environmental conditions within the animal's microhabitat, also allows for development of heat transfer models that provide real-time estimates of heat exchange (Levy et al., 2019).
Here, we explore the application of infrared thermal imaging and heat balance modelling for examining the thermal ecology of wild, free-living mammals, using the diurnal numbat (Myrmecobius fasciatus) as a model. Numbats are endangered marsupials in the monospecific family Myrmecobiidae. Their thermal biology is of particular interest as they are the only truly diurnal marsupial (Dickman and Calver, 2023), providing unique opportunities for thermoregulation and they have the energetic challenge of a low-energy termitivorous diet (Calaby, 1960; Friend, 1986; Cooper and Withers, 2004a). Their foraging periods need to coincide with the diel window when termites are active (Evans and Gleeson, 2001; Friend, 1986). Numbats behaviourally exploit favourable environmental conditions for foraging and avoid activity during wet, cold weather (Calaby, 1960; Cooper and Withers, 2004b), sheltering at night and during inclement conditions in an insulating nest in a burrow or hollow (Cooper and Withers, 2005) and their pelt has unique biophysical properties that facilitate extremely high solar heat gain (Cooper et al., 2003a). Thermal imaging is an ideal approach to non-invasively explore the numbat's heat exchange, considering its endangered status and the logistical challenges of studying a species that can't be trapped by conventional methods (Connell and Friend, 1985). We examine here the diurnal heat exchange of numbats under varying environmental conditions, heat exchange of various body surfaces and their potential role as a thermal window and possible thermal consequences of their striking black and white posterior dorsal stripes. We also examine the limitation of high environmental heat loads on foraging activity with implications for future conservation and management of numbats and other small diurnal mammals.
MATERIALS AND METHODS
Numbats (Myrmecobius fasciatus Waterhouse 1836) were studied at two Western Australian wheatbelt reserves, Dryandra Woodland (32.8°S, 117.0°E) and Boyagin Nature Reserve (32.5°S, 116.9°E), ∼170 km south-east of Perth, in each of the 12 months of the year, during 2020 and 2021. Numbats were located by driving slowly (5–10 km h−1) around the woodland tracks and we measured their surface temperature (Ts) remotely using infrared thermography (Tattersall and Cadena, 2010; Tattersall, 2016). We obtained thermal videos for 50 individual numbats over 20 different days. It was not generally possible to individually identify numbats by their stripe pattern during or after filming, as the stripes were only visible through the thermal camera and in the resulting images under some (high radiation load) environmental conditions. Consequently, we cannot exclude potentially repeated measurements of some individuals, but recordings were made over a large area in both reserves to minimise the chance of repeated measurement of individuals.
We recorded thermal videos (30 frames s−1) using a FLIR T1050sc infrared camera (Wilsonville, Oregon, USA) with an f 83.4 mm 12 deg telephoto lens, from a distance of 3–35 m. Sharp focus was achieved with a combination of the camera's auto-focus function and manual focus. A Kestrel 5400 (Boothwyn, PA, USA) portable weather station was used to record microclimate variables [Ta, ambient air temperature (°C); RH, relative humidity (%); BP, barometric pressure (kPa); WS, wind speed (m s−1); GT, globe temperature (°C); WBGT, wet bulb globe temperature (°C)]. Numbats were typically found standing on the ground or on fallen logs, so the weather station was placed in a similar microhabitat location under the same conditions of solar radiation as the numbat. Ground temperature (Tg; °C) immediately adjacent to the numbat was obtained from the thermography video recordings.
Heat exchange of numbats was calculated from Ts, environmental parameters and morphological measurements using biophysical modelling (McCafferty et al., 2011; Tattersall, 2016; Tattersall et al., 2018). Researcher IR Max4 software (Wilsonville, Oregon, USA) was used to analyse thermography video recordings after Tattersall et al. (2018) to obtain Ts of the various numbat body parts. For these analyses we assumed an emissivity of 0.95, that atmospheric temperature was equivalent to Ta, and calculated environmental temperature as the mean of Ta and Tg. We used the measured RH and estimated camera to numbat distance. The camera was calibrated against a national standard (FLIR, Oregon, USA). Mean Ts for various body areas (dorsolateral anterior body, dorsolateral posterior body, abdomen, chest, inside and outside leg, tail, face, and nose tip) was calculated for polygons drawn on video frames. We avoided inclusion of edges of body parts to minimise underestimation of Ts due to fur fibre effects (Cena and Clark, 1973a). It was not possible to obtain a Ts for every body part for every numbat, as visibility depended on its orientation and behaviour.
We calculated the environmental variables necessary for the biophysical model as follows. Direct and diffuse radiation were calculated using equations 13 and 14, and atmospheric, ground and reflected ground radiation from equation 17, of Liljgren et al. (2008). Solar zenith angle was determined from date, time and latitude/longitude for each numbat video recording using the NOAA Solar Calculator (https://gml.noaa.gov/grad/solcalc/calcdetails.html). The earth–sun distance was determined from the observation date using the USGS earth–sun distance calculator (https://www.usgs.gov/media/files/earth-sun-distance-astronomical-units-days-year). Solar radiation (S; W m−2) in the horizontal plane was calculated from solar zenith angle, earth–sun distance and the measured microclimate parameters by iteratively adjusting the value of S until GT calculated using equation 17 of Liljgren et al. (2008) equalled the measured GT.
For the morphometric parameters of the biophysical model, surface areas for numbat body parts were calculated from measurements of 7 numbat specimens (Western Australian Museum collection), using standard equations for surface area of cylinders (body, legs, feet, tail) and cones (nose, face). The body cylinder was divided equally into anterior and posterior sections, and dorsal (dorsal and lateral surfaces; 75% of body circumference) and ventral (25%) sections. The posterior dorsal section consisted of black and white stripes (black:white areas 1.2:1). The projected area of body parts (m2) was calculated in the horizontal plane (i.e. ground shadow area) for cylinders after Monteith (1973) and cones after Pennell and Deignan (1989), for the corresponding solar zenith angle and an average azimuth angle of 45 deg for the numbat to the solar radiation.
Biophysical modelling was accomplished in RMarkdown (https://rmarkdown.rstudio.com) using RStudio (http://www.posit.co/) in the R environment (https://www.r-project.org/). We calculated the radiative heat gain for each body part as the sum of direct, diffuse and ground reflected shortwave radiative gains, and the atmospheric/ground longwave gains minus the longwave radiative loss from the body surface, after Tattersall et al. (2018). Total radiative heat gain (HGrad) was calculated as the sum of gains for all body parts. We assumed a pelt reflectance of 7% for black stripes, 32% for white stripes and 18% for the rest of the body after Cooper et al. (2003a). To evaluate the effect of stripe colour on heat balance, the heat exchange model was also run using a reflectance of 18% and assuming the posterior body Ts was the same as that of the anterior body. Convective heat loss was calculated for each body part based on calculated surface area, Ts and geometry (vertical or horizontal cylinder); we assumed that the numbat was facing into the wind. Calculations used functions airtconductivity, airviscosity, Grashof, Nusselforced, Nusseltfree, Prandtl, Reynolds and StephBoltz from R package Thermimage (https://cran.r-project.org/package=Thermimage). Total convective heat loss (HLconv; W) was calculated as the sum of losses for all body parts.
We calculated the time that a 552 g numbat (Cooper and Withers, 2002) could store its total heat gain (HGtotal=HGrad–HLconv) until they reached a critical core body temperature (Tb) of 40°C (the maximum core Tb recorded for numbats implanted with temperature transmitters) from a core Tb of 35°C (at which they became active in the morning; Tb data from Cooper and Withers, 2004c). We assumed a specific heat for numbat tissues of 3.8 J g−1°C−1 (Giering et al., 1996) and metabolic heat production and evaporative heat loss of 3–6 W and 2–4 W respectively (1–2 times the field metabolic rate and water turnover of numbats; Cooper et al., 2003b). We ignored potential respiratory convective heat loss (McCafferty, 2007) as there are no data for expired air temperature for numbats and it is assumed to be low considering the potential for counter-current heat exchange afforded by the long snout.
Data are presented as the mean±s.e.m. unless stated otherwise. Statistical analyses were accomplished using the R package RMarkdown (https://rmarkdown.rstudio.com) and statistiXL (www.statistiXL.com). Lsmeans (https://CRAN.R-project.org/package=lsmeans) with Tukey adjustment was used to compare the marginal mean Ts of each body part at the mean WBGT. The lmer function in lme4 (https://CRAN.R-project.org/package=lme4) and lmerTest (https://CRAN.R-project.org/package=lmerTest) were used to determine linear mixed-effect models and examine the relationship between WBGT or GT and the Ts of different body parts and camera distance; individual nested in day was included as a random factor in the mixed-effects model. The Akaike information criterion (AIC) was used to evaluate the fit of models for WBGT and GT, with a ΔAIC of ≥2 indicating a superior model (e.g. Cavanaugh and Neath, 2019). As there was a significant interaction between WBGT (and GT) with body part Ts, we then investigated the Ts relationship separately for each body part to determine whether the slope was significantly different from 1 (one-sample t-test, using the mean and s.e. of the slope) and identify potential breakpoints in the relationship using a broken stick model (package SiZer; https://CRAN.R-project.org/package=SiZer). Linear mixed effect models, with day as a random factor, were also used to explore the relationship of HGrad, HLconv and HGtotal with WBGT, and to determine if HGtotal was lower for numbats in the shade compared with in the sun. For numbats where the black and white stripes were distinct, a paired t-test was used to determine if black stripes had a higher Ts, HGrad and HLconv than white stripes. A repeated-measures ANOVA and a priori simple contrasts (see Withers and Cooper, 2011) was used to compare the heat exchange calculated for the striped dorsal posterior body with the heat exchange calculated when assuming the area was all black, all white, or the same colour as the anterior body.
RESULTS
The 50 thermal videos of numbats were recorded between 07:30 and 19:11 h during daylight, ranging from 2 h and 11 min after sunrise to 16 min before sunset. Ambient air temperature during observations ranged from 13.8 to 36.8°C, ground temperature from 13.8 to 46°C, GT from 16.4 to 49.3°C and WBGT from 11.7 to 29°C (Table S1). Wind speeds were low, with no wind (<0.1 m s−1) recorded for 56% of observations and a maximum wind speed of 1 m s−1. Radiative heat load ranged from 360 to 966 W m−2 with 38% of observations for numbats in the shade, with a mean radiative load of 425±7.0 W m−2 and the remaining 62% of observations for numbats in the sun, with a mean radiative load of 642±31 W m−2.
Surface areas of various body parts, calculated for cylinders or cones using the dimensions from measurements of museum specimens, ranged from 0.14 cm2 for the nose tip and 28 cm2 for the face to 125 cm2 for the anterior body and 142 cm2 for the posterior body. Stripes on the posterior body had an area of 75 cm2 for black and 67 cm2 for white stripes. Horizontal projected areas of the various body parts varied with solar elevation; they were typically smaller than their actual surface areas (Fig. S1).
Body surface temperatures
Surface temperature measurements (Fig. S2 and Table S2) varied for different body parts. Mean Ts of the face (33.6±0.45°C) was significantly higher (t387-392≥3.33, P≤0.037) than for the other body parts, while the nose tip (24.2±0.45°C) and tail (26.3±0.45°C) were colder (t387-391≥4.76, P≤0.001; Fig. 1). The Ts of the anterior and posterior body did not differ (32.1±0.45°C; t392=0.13, P >0.999), nor did the ventral chest (31.5±0.50°C) or abdomen (30.9±0.52°C; t389=1.10, P=0.990); ventral and dorsolateral surfaces also didn't differ (t390-391≤2.23, P≥0.407; Fig. 1). Black stripes (35.5±0.87°C) were significantly hotter (t26=6.61, P<0.001) than white stripes (33.8±0.81°C; Fig. 1) for the 54% of the thermal images in which the stripes could be discerned.
Surface temperature of various body regions for 50 wild free-living numbats (Myrmecobius fasciatus) filmed with an infrared camera. Marginal mean (±s.e.m.) surface temperatures are shown. Different letters indicate significant differences at P<0.05 between body part surface temperatures, determined by LSmeans with Tukey adjustment at the mean wet bulb globe temperature (WBGT; 19.2°C; dashed line). Black and white stripes are presented and analysed separately as they were only distinguishable at high solar radiation intensities hence a higher mean temperature; *indicates significant difference between black and white stripes at P<0.05. The grey area indicates the core body temperature range of an active numbat at this study site (from Cooper and Withers, 2004c).
Surface temperature of various body regions for 50 wild free-living numbats (Myrmecobius fasciatus) filmed with an infrared camera. Marginal mean (±s.e.m.) surface temperatures are shown. Different letters indicate significant differences at P<0.05 between body part surface temperatures, determined by LSmeans with Tukey adjustment at the mean wet bulb globe temperature (WBGT; 19.2°C; dashed line). Black and white stripes are presented and analysed separately as they were only distinguishable at high solar radiation intensities hence a higher mean temperature; *indicates significant difference between black and white stripes at P<0.05. The grey area indicates the core body temperature range of an active numbat at this study site (from Cooper and Withers, 2004c).
The linear mixed effects models for numbat Ts with both WBGT (F1,48=115, P<0.001) and GT (F1,44=85, P<0.001) were highly significant, with a significant effect of body region (F10,388≥10, P<0.001) and interactions between body region and environmental temperature (F10,388≥3.8, P<0.001). Filming distance (F1,46≤0.005, P≥0.942) and all associated interactions (P≥0.729) were insignificant, so distance was removed from all lmer models. There was also no effect of numbat nested in day (LRT2≤0.02, P≥0.989) but these random factors were retained in the models to account for non-independence of measurements. The linear mixed effects model for WBGT (AIC=2172) was a better fit (ΔAIC=13) for numbat Ts than that for GT (AIC=2185). The significant slopes (F1,45≥45, P<0.001) of the relationships between Ts and WBGT with nose tip (0.92±0.12), tail (0.92±0.13) and posterior body (0.86±0.10) did not differ from 1 (P≥0.168; Fig. 2), while slopes for the other body parts were also significant (F1,26-47≥26, P<0.001) but lower than 1 (range 0.44±0.06 for face to 0.81±0.10 for anterior body; P≤0.036; Fig. 2).
Surface temperature of body regions of wild numbats (Myrmecobius fasciatus) as a function of wet bulb globe temperature (WBGT), determined by linear mixed effect models. Dashed lines are the line of equality (slope=1) for WBGT and the solid lines the observed slopes for the relationships. For the anterior and posterior body regions, there is a significant break point at WBGT=20°C. *Indicates that the overall slope ≠1.
Surface temperature of body regions of wild numbats (Myrmecobius fasciatus) as a function of wet bulb globe temperature (WBGT), determined by linear mixed effect models. Dashed lines are the line of equality (slope=1) for WBGT and the solid lines the observed slopes for the relationships. For the anterior and posterior body regions, there is a significant break point at WBGT=20°C. *Indicates that the overall slope ≠1.
For anterior and posterior body regions, there was a breakpoint in the relationship between WBGT and Ts at WBGT=20.0°C (ranges 18.2–24.3°C and 16.7–24.3°C respectively; Fig. 2), and for GT at 28.7°C (range 22–37°C) and 29.2°C (range 22–37°C) respectively, with the breakpoint 95% confidence limit incorporating ≤47% of the range of temperature data. The slopes relating Ts to WBGT or GT for both anterior and posterior body were significantly >0 below the breakpoint (slope 95% CL excluded 0) but there was no significant effect of WBGT or GT above their breakpoints (slope 95% CL included 0). For other body parts, the breakpoint CL incorporated >50% of the temperature data range and so the breakpoints are not considered meaningful.
Thermal balance
Mean HGrad of the face (0.112±0.112 W) at the mean WBGT (19.2°C) was significantly lower (t343-347≥4.00, P≤0.003) than that of all other body parts other than the nose (0.015±0.112 W; t343=0.818, P=0.996), chest (0.159±0.126 W; t346=0.354, P>0.999) and abdomen (0.290±0.134 W; t347=1.28, P=0.937), whereas HGrad of the posterior body (2.79±0.112 W) was higher than that of all the other body parts (t343-347≥5.15, P≤0.001; Fig. 3). For images where the black and white stripes were visible, HGrad was significantly higher for black stripes (1.63±0.195 W) compared to white stripes (1.08±0.144 W; t49=10.3, P<0.001; Fig. 3). HGrad calculated for a posterior body area without black and white stripes (i.e. assuming the same Ts and reflectance as the anterior body) was 2.77±0.347 W, which was the same as the actual calculated HGrad for the dorsal posterior body (2.77±0.346 W; t49=0.371, P=0.712) but was higher for an all-black (3.07±0.367 W; t49=9.01, P<0.001) and lower for an all-white (2.30±0.308 W; t49≤0.001) posterior body.
Radiative and convective heat gain for various body parts of wild numbats (Myrmecobius fasciatus). Values are mean (±s.e.m) absolute radiative (black bars) and convective (white bars) heat gain. Different letters indicate significant differences at P<0.05 between heat gain for the various body parts. Black and white stripes are presented and analysed separately as they were only distinguishable at high solar radiation intensities; *indicates significant difference between black and white stripes at P<0.05.
Radiative and convective heat gain for various body parts of wild numbats (Myrmecobius fasciatus). Values are mean (±s.e.m) absolute radiative (black bars) and convective (white bars) heat gain. Different letters indicate significant differences at P<0.05 between heat gain for the various body parts. Black and white stripes are presented and analysed separately as they were only distinguishable at high solar radiation intensities; *indicates significant difference between black and white stripes at P<0.05.
Mean HLconv of the face (0.190±0.035 W) at the mean WBGT did not differ from the chest or abdomen (t329≤1.96, P≥0.570), was significantly lower than the legs and body (t325-326≥4.93, P≤0.001) and was higher than the nose and tail (t326-327≥4.23, P≤0.001; Fig. 3). For images where the black and white stripes were visible, HLconv was significantly higher for black stripes (0.230±0.017 W) compared with white stripes (0.182±014 W; t49=8.76, P<0.001; Fig. 3), reflecting their differences in Ts. The actual calculated HLconv for the posterior body (0.395±0.030 W) did not differ from that predicted for a posterior body area without black and white stripes (0.397±0.032 W; t49=0.189, P=0.850) or for an all-white posterior body area (0.388±0.030 W; t49=0.470, P=0.640) but was higher for an all-black posterior body (0.433±0.032 W; t49=2.66, P=0.011).
Both WBGT (F1,40≥5.97, P≤0.019) and body region (F8,327-360≥13.6, P<0.001) had highly significant effects on HGrad and HLconv, with significant interactions between body region and WBGT (F8,327-360≥3.94, P<0.001) but no overall effect of animal nested in day (LRT2≤2.74, P≥0.253). For HGrad, exploratory investigation of each body part separately indicated that the HGrad of both the anterior and posterior body areas was independent of WBGT (F1,74-75≤1.38, P≥0.244) while the relationships between WBGT and other body parts were significant (F1,46-74≥16.7, P<0.001). For HLconv, both dorsal and ventral areas of the body were independent of WBGT (F1,57-74≤2.066, P≥0.158), with HLconv of the extremities driving the observed WBGT effect.
Total heat gain for numbats was generally negative at low WBGT and increased significantly to about 20 W at high WBGT; HGtotal (W)=1.18 WBGT (°C)–18.3 (F1,75=58.1, P<0.001; Fig. 4). The negative HGtotal of −0.696±0.320 W (i.e. net heat loss) for numbats filmed in the shade was significantly lower than the heat gain of 8.89±1.19 W for numbats in the sun (F1,43=7.22, P=0.010; Fig. 4). However, direct solar HGrad accounted for only 18±2.6% of the total HGrad for numbats in the sun (and 4.5±1.7% for numbats in the shade), with atmospheric (20±0.9% sun and 28±0.4% shade) and ground (34±8.1% sun and 47±1.0%) radiation being the most substantial component of radiative load for numbats. For numbats with positive heat gain, we calculated that the time they could sustain their total heat gain (i.e. time until Tb reached 40°C) decreased from over 150 min at low WBGT to only 10 min at WBGT above 23°C (Fig. 4).
Heat exchange and time to reach a critical body temperature of 40°C as a function of wet bulb globe temperature for wild numbats (Myrmecobius fasciatus). Absolute heat exchange (top) and time to reach a critical body temperature (Tb) of 40°C (bottom). Black symbols are for numbats in the shade and white symbols for numbats in the sun.
Heat exchange and time to reach a critical body temperature of 40°C as a function of wet bulb globe temperature for wild numbats (Myrmecobius fasciatus). Absolute heat exchange (top) and time to reach a critical body temperature (Tb) of 40°C (bottom). Black symbols are for numbats in the shade and white symbols for numbats in the sun.
DISCUSSION
Temperature has an all-pervasive influence on animal function (Withers et al., 2016), so understanding the interaction between environmental temperature and thermal balance is essential to understanding the ecology of animals in the field. This has particular relevance for the future conservation and management of species in the face of global warming. Infrared thermography provides a non-contact means of measuring surface temperatures for wild, free-living animals (McCafferty et al., 1998; McCafferty, 2007; Tattersall, 2016) and so it is particularly well-suited to long-term study of difficult to study and/or endangered species. Our model of heat exchange for numbats based on field Ts measurements was remarkably consistent with measured variables for physiological and anatomical traits for this species and other small diurnally active mammals. These thermal models allow us to identify the opportunities and limitations of current and future environmental conditions for management and conservation.
The heat exchange apportioned to different body parts reflected their shape, orientation, and physical and projected surface areas (Fig. 3; Fig. S1). Convective heat exchange was generally negative (i.e. heat loss) and was strongly influenced by body part shape, surface area and orientation, while radiative heat exchange was positive (i.e. heat gain) for all body parts and strongly dependant on projected surface area. The Ts of the numbat's face most closely approximated Tb and was least impacted by WBGT i.e. it was poorly insulated, consistent with the head being a major heat sink for many endotherms (McCafferty, 2007; McCafferty et al., 2011), but its relatively small surface area meant that absolute heat exchange was inconsequential compared to that of the body, despite the body being better insulated (Fig. 3). Likewise, despite the presumed importance of appendages to heat exchange (McCafferty, 2007; McCafferty et al., 2011), the overall heat exchange of the numbat's appendages (legs and tail) contributed little to their total heat exchange.
The numbat's anterior and posterior body, the largest of its anatomical regions, accounted for the greatest heat exchange, and for both there was a breakpoint in the Ts and WBGT relationship (Fig. 2) at a GT (∼29°C) that closely matched the thermoneutral zone measured under laboratory conditions, where there is also a clear breakpoint in thermal conductance (Cooper and Withers, 2002). Below thermoneutrality, numbats piloerect their pelt, which increases pelt resistance (Cooper et al., 2003a) and conserves metabolic heat; consequently, Ts scales with environmental temperature below thermoneutrality. At high environmental temperatures, numbats pilodepress, which decreases pelt resistance and Ts becomes independent of environmental temperature. In this way the numbat's general body area functions as a thermal window i.e. a body surface with a high or variable Ts (Tattersall, 2016). There was no evidence of numbats using other regions such as the legs or ventral surface as thermal windows. The Ts of legs and ventral surface were related to WBGT with a slope <1, indicating that body temperature was also a substantial driver of Ts, but there was no significant break-point that would indicate that they were ‘opened’ to facilitate heat loss at high temperature and ‘closed’ at lower temperatures (Tattersall, 2016). The Ts of the moist nose scaled with WBGT, most closely approximating WBGT of all the body parts, as expected for a moist surface. However, the extremely small area of the nose limits the potential for a major role in overall heat exchange. The Ts of the tail also scaled with WBGT without a break-point, with the thin tail core well insulated by the ‘bottlebrush’ of fur covering it. It is possible that the tail could be used as a shade parasol, as observed for diurnal Cape (Xerus inauris) and antelope (Ammospermophilus leucurus) ground squirrels (e.g. Chappell and Bartholomew, 1981a; Bennett et al., 1984; van Heerden and Dauth, 1987; Fick et al., 2009) as it is well insulated, but our general observations suggest that this is unlikely; communication or predator avoidance are possible alternative roles.
The function of the numbat's striking black and white dorsal stripes is unlikely to be related to heat exchange, in agreement with previous conclusions based on numbat pelt biophysics (Cooper et al., 2003a). Black stripes were up to 6°C hotter than white stripes in images at high environmental temperatures where stripes were visible, as expected from the lower solar reflectance of black compared with white stripes (Cooper et al., 2003a). This difference in Ts for black and white surfaces is commonly observed (e.g. goats, zebras; Cena and Clark, 1973a,b; Finch et al., 1980; Cobb and Cobb, 2019). Although it has been proposed that zebra stripes may facilitate thermoregulation (e.g. Larison et al., 2015) and generate convective air currents to enhance heat loss (Cobb and Cobb, 2019), a thermoregulatory role of black and white stripes remains unresolved (Pereszlényi et al., 2021). For numbats, black stripes had a higher Ts and a higher HGrad and HLconv compared with white stripes, but we also calculated that HGrad and HGtotal were not different for the actual striped posterior body compared with a non-striped posterior body (substituting reflectance and Ts of the anterior body). Indeed, the inter-relationship between colour, structure and environmental conditions on the resistance and solar heat gain of pelts is complex, and thermal exchange between an animal and its environment does not necessarily reflect coat colour (e.g. Burtt, 1981; Walsberg and Schmidt, 1989; Walsberg, 1990; Walsberg and Wolf, 1995). Caro (2009) suggested that the role of the numbat's black and white stripes is most likely pattern blending, enhancing crypsis by merging with dappled light. We concur with this suggestion.
It is not surprising that HGrad dominated thermal exchange with the environment for numbats, as the biophysical properties of their pelt indicate remarkable adaptation for direct solar heat acquisition (Cooper et al., 2003a). Numbats behaviourally exploit solar radiation by timing activity to favourable environmental conditions. They are most active on days with high solar radiation and avoid cold, wet and overcast days (Cooper and Withers, 2004b). Biophysical measurements of numbat pelts indicate that solar heat gain would contribute a heat load to the body core of 50–357% of resting metabolic rate at ambient temperatures of 15–32.5°C (Cooper et al., 2003a), although such calculations can overestimate heat load by 32–48% (Walsberg and Wolf, 1995). Our predicted HGrad was 268–766% of a numbat's resting metabolic rate and 280% of its field metabolic rate (metabolic data from Cooper and Withers, 2002; Cooper et al., 2003b), which are considerably higher than the predictions of Cooper et al. (2003a). However, our calculations for radiative heat gain by wild numbats include the considerable HGrad from other than direct solar radiation. Calculated heat gain from only direct solar radiation of 63–180% of resting metabolic rate is more consistent with predictions of 34–243% of resting metabolic rate from laboratory pelt studies (Cooper et al., 2003a) adjusted for likely overestimation (Walsberg and Wolf, 1995). Our results highlight the importance of long-wave radiative heat gain for heat exchange and thermoregulation. Long-wave atmospheric (19%) and ground (33%) radiation are substantial, about 52% of the total radiative heat gain compared with 29% direct, 5% diffuse and 15% ground reflected short-wave radiation. A similar importance of long-wave radiative heat gain to total radiative heat gain is apparent for other mammals at similar solar heat loads, such as Boran cattle (61%; Finch, 1976) and black goats (62%; Finch et al., 1980).
The mean environmental heat gain for all numbats was 4.5× basal (Cooper and Withers, 2002) and 1.7× field (Cooper et al., 2003a,b) metabolic heat production, reaching 7.6× and 2.5× for numbats in the sun. These very high rates of environmental heat gain are a consequence of the numbat's remarkable pelt structure (Cooper et al., 2003a). They exceed values for wild, free-living antelope ground squirrels, which also have pelts that facilitate high rates of solar heat gain (Walsberg, 1990); their environmental heat gain also dominates heat balance, being approximately 6× basal and 2× active metabolic heat production (Chappell and Bartholomew, 1981a,b). Clearly, it is environmental rather than metabolic heat load that is important for small diurnally active mammals, especially the numbat.
Van der Vinne et al. (2015) suggested that diurnal activity is of particular importance for endotherms that experience energetic limitations and that temporal niche shifts from nocturnality to diurnality can indicate a thermoregulatory or energetic challenge. Numbats are the only marsupial to have completely switched to diurnality from their nocturnal ancestry (Dickman and Calver, 2023). The diurnal timing of their foraging periods reflects when termites are active in shallow sub-surface soil galleries where numbats access their termite prey (Christensen et al., 1984; Evans and Gleeson, 2001; Friend, 1986). The reductions in thermoregulatory energy expenditure afforded by exploiting radiative heat gain during the day and avoiding activity during cold nights are probably important in allowing numbats to rely exclusively on their low energy-density termitivorous diet (Redford and Dorea, 1984; Cooper and Withers, 2004a; Cooper, 2011), especially in arid and semi-arid environments characterised by low primary productivity. Similar energetic advantages are afforded to other small diurnal mammals such as ground squirrels (e.g. Chappell and Bartholomew, 1981a,b). The use of insulating nests within burrows and night-time torpor have also been recognised as energy-conserving traits employed by numbats at low environmental temperatures (Cooper and Withers, 2004c, 2005).
Despite the energetic benefits of diurnal activity for small mammals, there are costs such as altered prey abundance, greater conspicuousness to predators, and extreme environmental heat loads from high air temperatures and solar radiation. For numbats, timing of foraging reflects their termite prey availability, being highest at mid-day as the soil temperature increases in cooler periods but declining at mid-day when termites are less active in hotter periods (Christensen et al., 1984; Evans and Gleeson, 2001). Numbats can withstand the predation pressure of native diurnal predators such as birds of prey and reptiles (and temporally avoid native mammalian predators), although introduced predators such as cats (Felis catus) and European red foxes (Vulpes vulpes) threaten their survival (Friend, 1990; Friend and Thomas, 1995). High daytime environmental temperatures can require increased evaporative heat loss, but the termitivorous diet of numbats has a favourable water economy index (water to energy turnover), facilitating them achieving water balance without access to free water (Cooper and Withers, 2004a). The challenge of high environmental heat gain, especially for numbats in the more arid parts of their range, will be exacerbated by global warming, which not only impacts mean ambient temperature and average annual rainfall but also results in increased frequency, intensity and duration of extreme weather events such as heatwaves (Meehl and Tebaldi, 2004; IPCC, 2014). To persist, they must be able to withstand these extreme periods in addition to a general increase in the average impacts (Walsberg, 2000; Buchholz et al., 2019).
The ambient temperatures at which we observed numbats to be active were consistent with those of other studies for captive numbats (Cooper and Withers, 2004b) and wild numbats in a range of environments, including arid parts of their former natural distribution (Christensen et al., 1984; Bester and Rusten, 2009; Hayward et al., 2015). Ambient temperatures approaching 40°C and ground temperatures approaching 50°C become prohibitive for numbat activity in our and other studies. We calculated that numbats (552 g; Cooper and Withers, 2002) foraging at WBGT temperatures exceeding 23°C would approach a critical body temperature (40°C) from a morning Tb of 35°C in ∼10 min (Fig. 4). We believe that our thermal model for numbats in a semi-arid habitat is equally applicable to numbats in more arid environments. Chappell and Bartholomew (1981a) determined a remarkably similar maximum activity period under intense solar radiation for antelope ground squirrels (Ammospermophilus leucurus) in the open of 8–13 min and we calculated a maximum foraging duration for Cape ground squirrels (Xerus inauris) of 20 min from the data of Fick et al. (2009). Beyond these exposure limits, small diurnal mammals must shift their foraging during mid day to cooler microclimates, such as the shade, become inactive in thermally favourable retreats, such as burrows, or even become cathemeral or nocturnal.
Shaded areas are suitable microclimates for diurnal mammals to forage in when environmental temperatures are high (Fuller et al., 2014; Levy et al., 2019) e.g. ground squirrels, degus (Octodon degus) and sand rats (Psammomys obesus) are small diurnal mammals that use shaded locations for foraging at high ambient temperatures (Chappell and Bartholomew, 1981b; van Heerden and Dauth, 1987; Ilan and Yom-Tov, 1990; Bacigalupe et al., 2003). That numbats foraging in the shade have significantly lower HGtotal than when foraging in the sun is consistent with numbats in arid environments preferentially occupying habitats with higher canopy cover (Vieira et al., 2007; Hayward et al., 2015), more so than sympatric nocturnal marsupials (Berry et al., 2019), gaining the combined benefits of shade and protection from aerial predators. However, limiting activity to shaded areas in hot weather reduces the area available for foraging, especially in arid areas where vegetation cover is sparse and during summer when the solar zenith angle is smallest. Consequently, it is during the hottest and driest conditions when shade is the most essential that it is least available (Chappell and Bartholomew, 1981a; Levy et al., 2019) and this problem is likely to be exacerbated by the increased heat and aridity predicted for most of the world's hot deserts with climate change (Sala et al., 2000; IPCC, 2014). The predominance of atmospheric and ground radiation as a heat load in shade means that even shaded areas may have too high a heat load for small diurnal animals to remain active under conditions of very high environmental temperatures.
Numbats seek shelter in logs and burrows during the day when environmental temperatures are high, with a bimodal activity pattern of foraging in the mornings and late afternoons (e.g. Christensen et al., 1984; Bester and Rusten, 2009), which also reflects the temporal availability of termites in sub-surface soil galleries (Evans and Gleeson, 2001; Friend, 1986). Other small diurnal mammals such as ground squirrels (e.g. Chappell and Bartholomew, 1981a; van Heerden and Dauth, 1987; Long et al., 2005), sand rats (Ilan and Yom-Tov, 1990), golden spiny mice (Levy et al., 2016) and degus (Bacigalupe et al., 2003) similarly cease foraging to seek cool retreats when environmental temperatures are very high. However, ceasing activity for part of the day results in lost foraging opportunity.
Diurnal species have greater flexibility for temporal activity shifts than nocturnal species (Kronfeld-Schor and Dayan, 2008) and moving activity to earlier and later in the day can be taken to the extreme by a shift to nocturnality (Levy et al., 2019). For example, some predominantly diurnal rodents increase nocturnal activity during periods of high heat and aridity (Vieira et al., 2010). Such shifting of activity periods may assist some mammals to withstand the challenges of climate change (Hetem et al., 2012; Fuller et al., 2014). However, shifting diel activity patterns can be ecologically costly for small mammals as there can be considerable day to night differences in environmental conditions, so adaptations to diurnality may reduce fitness at night (Kronfeld-Schor and Dayan, 2003). For numbats, shifts towards nocturnality would require a similar shift in the activity of termites in sub-surface soil galleries where they are accessible to numbats and for numbats to be able to avoid predators and navigate their environment at night. The numbat's visual system is adapted to the high and constant light levels associated with diurnal activity, with a constant pupil size, a cone-dominated retina and a visual field adapted to detection of overhead avian predators (Arrese et al., 2000), consistent with birds of prey being the numbat's predominant natural predators (Friend, 1993; Bester and Rusten, 2009). Nocturnal activity would make numbats particularly vulnerable to both native marsupial and introduced eutherian predators. Consequently, such ecological and morphological constraints may well preclude numbats from shifting to cathemeral or nocturnal activity.
The thermal model of heat exchange that we developed for numbats by thermally imaging free-living numbats in the field demonstrates the importance of heat exchange with the environment, in particular radiative heat exchange, on their thermal biology and ecology. This is also evident for some other small diurnal mammals from more invasive physiological studies. For numbats, thermal modelling is particularly ecologically relevant because the temporal pattern of radiative heat gain is associated with the temporal activity pattern of their termite prey, which is a specialised low-energy density diet. Open spaces are important not only as foraging areas but also for facilitating positive solar heat gain and are considered an aspect of habitat critical for numbat survival (Friend and Page, 2017). However, excessive heat gain at high environmental heat loads becomes limiting and the importance of shaded areas must be considered, particularly with respect to assessing future numbat habitat. Numbats in hot arid environments may be forced to reduce foraging times or extend activity into periods with higher risk of predation. Indeed, a small population of numbats re-introduced into the arid zone of South Australia experienced substantial weight loss and individuals were predated on at dusk, possibly by an owl (Bester and Rusten, 2009). Consequently, we consider availability of shade to be a critical resource in numbat habitats and suggest that translocations to hotter inland arid zones be carefully considered, as these are the populations most potentially vulnerable to the impacts of climate change.
Acknowledgements
We thank Kenny Travouillon for allowing us to measure numbat specimens in the Western Australian Museum collection. Field thermography measurements followed the Australian Code of Practise for the care and use of animals for scientific purposes, approved by the Curtin University Animal Ethics Committee (AEC_2019_23).
Footnotes
Author contributions
Conceptualization: C.E.C., P.C.W.; Methodology: C.E.C., P.C.W.; Software: P.C.W.; Formal analysis: C.E.C., P.C.W.; Investigation: P.C.W.; Resources: C.E.C., P.C.W.; Data curation: C.E.C.; Writing - original draft: C.E.C.; Writing - review & editing: P.C.W.; Visualization: C.E.C.; Project administration: C.E.C.; Funding acquisition: C.E.C., P.C.W.
Funding
This research used equipment purchased by the Australian Research Council's Discovery Projects funding scheme (project DP160103627).
Data availability
Supporting data can be found in the electronic supplementary information.
References
Competing interests
The authors declare no competing or financial interests.