Heating rates are more strongly influenced by near-infrared than visible reflectance in beetles

Reflectance of sunlight is a fundamental property affecting surface heating rates. Widespread associations between reflectance in ultraviolet (UV, 300–400 nm) and human-visible (VIS, 400–700 nm) wavelengths and climate (Bogert, 1949; Delhey, 2017; Gaston et al., 2009; Gloger, 1833) suggest selection on reflectance properties for thermal benefits. However, animal colouration in ultraviolet and human-visible wavelengths is also selected for other functions, such as camouflage or communication, which can override or conflict with selection for thermoregulation (Hegna et al., 2013). By contrast, reflectance in near-infrared (NIR, 700–2500 nm) wavelengths is expected to be primarily selected for thermoregulation rather than visual functions because animals are visually insensitive to this wavelength range (Stuart-Fox et al., 2017). Absorption of NIR light can have a significant effect on heat gain because NIR wavelengths comprise more than half of the energy in sunlight, which spans wavelengths of ∼280–2500 nm (Stuart-Fox et al., 2017). Recent studies have shown a strong correlation between NIR reflectance and climate in birds and butterflies, where species in hot, arid environments have higher NIR reflectivity (Medina et al., 2018; Munro et al., 2019). These studies suggest an important role for NIR in thermoregulation, but direct evidence linking reflectivity in different wavelength ranges to heating rates is currently lacking.

The effects of surface reflectivity on radiative heat gain may be modulated by the surface area exposed to direct sunlight and morphological properties of the integument (Walsberg, 1992). For example, the triangular hairs covering Saharan silver ants enhance both the VIS–NIR reflectivity and mid-infrared emissivity (Shi et al., 2015). Similar effects of surface structures on heat dissipation are observed in butterfly wings (Krishna et al., 2020; Tsai et al., 2020). Surface sculpturing of insect elytra could potentially play a role in thermoregulation (Drotz et al., 2010; Lindroth, 1974) by increasing effective heat transfer area (Chen et al., 2001). However, the possible role of elytral surface sculpturing in thermoregulation has yet to be considered or empirically tested. To affect thermoregulation, heat must be transferred from the surface to the body through an insulating layer, including the cuticle in insects. A thinner cuticle generally has higher transmittance and hence more radiation penetrates through the cuticle (Cuesta and Lobo, 2019). For instance, smaller dung beetles with thinner cuticle warm up faster than larger beetles with thicker cuticle when exposed to the same heat source (Amore et al., 2017). Thus, it is essential to consider both optical and morphological properties of the integument to understand adaptations to modulate heat gain from sunlight.

In this study, we investigated how optical and morphological properties of the integument influence heat transfer in jewel beetles. We chose jewel beetles (Coleoptera: Buprestidae) as a model group because they have highly diverse cuticle properties and are often exposed to direct sunlight. Adults are active in spring and summer and spend most of their time feeding and mating on their host plants. Cuticle properties may be particularly important to enable beetles to reach active temperatures in the early morning when they are nearly immobile. Jewel beetles are well known for their diverse elytron colours and their body size varies by orders of magnitude. Their surface sculpture ranges from a smooth surface to deep grooves or indentations (Fig. 1). The dorsal surface is the major area of beetles exposed to solar radiation; therefore, here we focused on the elytra, which make up the majority of the dorsal area.

To investigate radiative heat gain, researchers typically illuminate the sample with a light source, but artificial light sources generally have a very different spectral power distribution to sunlight, particularly in NIR wavelengths. Here, we used a solar simulator that closely resembles the solar power distribution (Fig. 2B) and an isolated thermal chamber (Fig. 2A) to minimize any effect of conduction and convection. We used optical filters to isolate the effects of UV–VIS and NIR wavelengths on heating rates. This system allowed us to test the effect of cuticle reflectivity, size, thickness and surface sculpturing on heating rate of isolated elytra from 17 jewel beetle species. For two species, we examined how the heating rate of isolated elytra corresponded to body temperature. Our experiments provide evidence for a direct link between NIR reflectivity and heat transfer in jewel beetles.

We used museum specimens of 16 Australian and one Filipino jewel beetle species from the Australian National Insect Collection (ANIC). Species were selected to ensure there was diversity in cuticle microstructure and reflectance in our sample. Only species with minimal patterning on their elytra were selected to accurately relate average reflectance properties to heating rate. We only used species that were well represented in ANIC due to the need for destructive sampling. To connect the results of an isolated elytron with an intact beetle, we collected seven Melobasis propinqua (–37.881355, 144.181317) and 10 Temognatha chalcodera (–31.122308, 120.713545; Department of Biodiversity, Conservation and Attractions, Government of Western Australia, collection permit number, FO25000127) in October 2019 and January 2020, respectively.

We measured the reflectance of three different locations of the beetle elytra (300–2100 nm) and averaged them to represent the reflectance of the species (see Supplementary Materials and Methods 1 for measurement details). Reflectance is the proportion of incident light reflected from a surface at each wavelength interval. Heating rate depends on the proportion of total irradiation that is absorbed by the surface (i.e. neither reflected nor transmitted) and therefore depends on the spectral power distribution of the light source as well as the reflectance spectrum of the surface. To account for this, we calculated reflectivity for each species. Reflectivity is defined as the ratio of total reflected to total incident radiation integrated over the wavelength range of interest (Smith et al., 2016; Eqn 1 below) and therefore varies under different illumination conditions.

We measured the area, thickness and rugosity of the samples to test their correlation with their thermal properties. For the area, we measured the incidental area subjected to illumination (effective area) during the experiment because radiative heat load is a function of surface area intercepting radiation (Walsberg, 1992). We took photographs using a camera (D7200; Nikon, Tokyo, Japan) with a macro lens (60 mm Apo Macro, CoastalOpt; JenOptik, Jena, Germany) positioned at the normal angle and calculated the elytral area from the pictures using ImageJ (v. 1.52a; Schneider et al., 2012). Micro-CT reconstructions of elytra were used to derive thickness and rugosity. Average thickness was measured from 10 transverse sections from the middle regions of the elytra. We defined rugosity as the ratio of the surface area (dorsal side of the elytron) to the projected area (Trasatti and Parsons, 1986). Specifically, we chose a 3×3 mm² region of interest (ROI) at the middle of each elytron for calculation, but the ROIs were decreased to 1.5×1.5 or 2×2 mm² for smaller elytra samples to obtain flat surfaces. Full details of micro-CT scans and rugosity measurements are provided in Supplementary Materials and Methods 2.

We measured heating rate of one elytron (left elytron) for each of 17 species. We assessed repeatability of heating rate estimates by repeating the experiment five times on three different elytron samples (Calodema regalis, Temognatha obscuripennis, Stigmodera gratiosa).

To compare the heating rate of elytra with the body temperature of beetles, we measured the heating rate of recently deceased individuals (seven Melobasis propinqua and 10 Temognatha chalcodera) as well as their removed elytra. These two species differ substaintially in reflectance, surface sculpturing and size. We collected living beetles and froze them at −20°C for 30 min, then thawed them completely at room temperature before the experiment to ensure freshness. We used recently deceased beetles for ethical reasons (discomfort from inserting a thermocouple into the body of a living beetle) and to ensure that our experiments excluded all potential effects on heat transfer apart from direct radiation.

The experiments were conducted in a closed glass thermal chamber (Fig. 2A) surrounded by a flowing water system to control the ambient temperature within the chamber. A transparent window on the top of the chamber with different optical filters covering it allowed us to control the illumination wavelengths in the chamber. The solar simulator was positioned directly above the chamber such that the sample was illuminated from directly overhead. Inside the chamber, samples were placed in the centre on a transparent acrylic platform and connected with a thermocouple to record the temperature change throughout the experiment (see Supplementary Materials and Methods 3 for details). We also placed another thermocouple in the chamber to monitor the air temperature. The thermocouples were connected to the thermometer and the temperatures were recorded once every 20 s.

Each trial comprised a sequence of alternating heating and cooling periods: 5 min cooling, 5 min heating (UV–NIR), 10 min cooling, 5 min heating (NIR), 10 min cooling, 5 min heating (UV–VIS). The first 5 min cooling period without any illumination was to ensure that the sample temperature was not increasing prior to the first heating period. A 10 min cooling period prior a heating period was to ensure samples were similar temperature at the beginning of the heating period. We estimated heating rate (°C min⁻¹) as the maximal change per minute over a 1 min period for each heating treatment – ALL (UV–NIR), UV–VIS and NIR – because this range captured the maximal change per minute. Thus, our measure of heating rate corresponds to the maximum heating rate of the sample due to direct radiation.

We tested whether the elytral properties (reflectivity, effective area, thickness, rugosity) predicted heating rate using generalised linear models (lmer, R package lme4; Bates et al., 2015) followed by model selection. Heating rate for the ALL, UV–VIS or NIR illumination was the response variable and the four elytral properties were the fixed effects. We did not include interaction terms in the models because there was no a priori reason to expect an interaction and to avoid model over-specification given our sample size (n=17). We tested normality, homoscedasticity and collinearity for the explanatory variables before including all four elytral properties in the full models. We did not use phylogenetic correction because our experiment does not test for an evolutionary association. Our experiment is designed to quantify the physical relationship between elytral properties and heating rate, which does not depend on phylogenetic relatedness.

We used corrected Akaike's information criterion (AIC_C) for model selection and model averaging (dredge, R package MuMIn; https://CRAN.R-project.org/package=MuMIn). We included models with ΔAIC_C <4 (cumulative AIC_C weights=0.871–0.908) as our top subset models. We calculated the model averaged coefficients and 95% confidence intervals (CI) along with the relative importance of variables (RIV) for each variable by summing the normalized AIC_C weights of the models in this subset where a given variable was present. We considered a variable to be biologically significant if its 95% CI did not include 0 and it had high RIV (see Supplementary Materials and Methods 4 for model selection details).

We tested the repeatability of the heating rate measurement by conducting a repeatability test on five repeated heating rate measurements of three isolated elytron samples (rpt, R package rptR; Stoffel et al., 2017). The repeatability tests were performed on linear models, in which we set species as a random factor with the heating rate as the dependent variable.

To correlate the heating rate of the recently deceased beetles (bodies) with those of their removed elytra, we ran a linear model (lmer) on the heating rates of two beetle species. In the model, we set the heating rate as the response variable and beetle ID as a random variable. For explanatory variables, we included species, heating rate type (body/isolated elytron) and their interaction to partition interspecific and intraspecific variation. All statistical analyses were performed in R 3.6.3 (https://www.r-project.org/).

Reflectivity had a strong effect on heating rate under NIR illumination but not under UV–VIS illumination (Table 1, Fig. 3). Reflectivity was retained in five out of six models in the top subset (ΔAIC_C <4 models; Table S1–S3) for NIR and had an RIV (sum of the normalized AIC weights) of 0.89, though the 95% CI of the estimate slightly overlapped zero (95% CI −0.024 to 0.001; Table 1). There was a trend towards reflectivity affecting heating rate under ALL illumination, which was clearly driven by NIR reflectivity (Fig. 3). Under ALL illumination, reflectivity had an RIV of 0.58 and was retained in six out of 11 models (95% CI −0.042 to 0.015; Table 1). An increase of 1% reflectivity resulted in a decrease in heating rate of 0.011°C min⁻¹ for NIR and 0.013°C min⁻¹ for ALL (Table 1). Effective area had a strong effect for ALL illumination and was retained in seven out of 11 models in the top subset, with the 95% CI of the estimate slightly overlapping zero (−0.006, 0.002). An increase of 1 mm² effective area resulted in a 0.002°C min⁻¹ decrease in heating rate for ALL. Cuticle thickness and rugosity had no effect under any of the three illumination conditions.

The heating rate measurement was highly repeatable in ALL and NIR illuminations [ALL: R=0.861±0.251 (mean±s.e.), likelihood ratio test P<0.001; NIR: R=0.829±0.258, P<0.001] and moderately repeatable under UV–VIS illumination (R=0.520±0.284, P<0.05) (Harper, 1994).

The heating rate of M. propinqua was significantly higher than that of T. chalcodera for both the isolated elytra and the intact bodies under all three illuminations (ALL: χ²=724.218, P<0.001; NIR: χ²=818.047, P<0.001; UV–VIS: χ²=469.432, P<0.001; Fig. 4). On average, the heating rate of M. propinqua was 1.8 times higher than that of T. chalcodera for the isolated elytra and 4.3 times higher for the intact bodies. The greater thermal mass of the larger T. chalcodera resulted in a greater difference in heating rate between the isolated elytron and body than for the much smaller M. propinqua. Across the three illumination conditions, the average heating rate of the isolated elytra was 3.2 times higher than that of the intact beetles for T. chalcodera, whereas it was only 1.3 times higher for M. propinqua. The difference in the heating rate of isolated elytra and intact bodies was statistically significant under all three illuminations (ALL: χ²=325.716, P<0.001; NIR: χ²=120.528, P<0.001; UV–VIS: χ²=315.842, P<0.001) and the difference was significantly greater for T. chalcodera than for M. propinqua (ALL: χ²=21.105, P<0.001; NIR: χ²=15.658, P<0.001; UV–VIS: χ²=8.021, P<0.01).

We demonstrate that reflectivity and the surface area exposed to direct illumination significantly affect heating rate in jewel beetles, whereas cuticle thickness and rugosity do not have a significant effect. Heating rates were measured using a light source closely resembling the power distribution of natural sunlight and an isolated chamber in which we could examine the effects of specific wavelength ranges on radiative heat gain while controlling and minimizing convection and conduction. Results show that beetle elytra with lower reflectivity have a higher heating rate. This effect is driven by NIR reflectivity, likely because there is more variability in NIR than in UV–VIS reflectivity among our sample species (Fig. 5). We found that larger effective area is associated with lower heating rate, likely owing to higher thermal inertia of larger elytra. By contrast, cuticle thickness and surface rugosity had minimal effects on maximum heating rate of isolated elytra. The difference in elytron heating rate between species corresponds to a difference in body heating rate, suggesting that variation in the NIR reflectivity of elytra will affect the body temperature of beetles. Combined, our results provide robust evidence that reflectance properties, including reflectance in NIR wavelengths invisible to animals (Stuart-Fox et al., 2017), can affect heating rates and may therefore play a significant role in thermoregulation.

The effect of reflectance on thermoregulation has been tested in several insects (Umbers et al., 2013); however, most previous studies have focused on visible colouration and neglected NIR wavelengths (e.g. Hegna et al., 2013; Jong et al., 1996; Outomuro and Ocharan, 2011; Rao and Mendoza-Cuenca, 2016) and/or had limited control of convective or conductive heat transfer (e.g. Amore et al., 2017; Brakefield and Willmer, 1985; Brashears et al., 2016; Hegna et al., 2013; Willmer and Unwin, 1981; but see Schultz and Hadley, 1987). By contrast, we were able to isolate the effect of reflectivity on heat transfer by carefully controlling irradiance, conduction and convection. With the optical filters, we revealed a significant effect of NIR reflectivity on heating rates for biological samples. Differences between the spectral power distribution of artificial light sources and natural sunlight have been a major limitation in the use of laboratory experiments to infer the thermal consequences of different coloured integuments in the field. The spectral power distribution of the solar simulator in our experiments resembles that of natural sunlight, indicating that, all else being equal, the effect on radiative heat transfer detected in our experiment will be similar to natural conditions.

The jewel beetles in our study showed greater variation in NIR reflectivity than UV–VIS reflectivity and, consequently, NIR reflectivity had a stronger influence on heating rate. Unlike UV–VIS colouration, NIR reflectance is free to vary in response to selection for thermal benefits because it is not directly subjected to visual selective pressures; few, if any, animals have visual sensitivity beyond 750 nm (Stuart-Fox et al., 2017). Among the jewel beetles in our study, the variance of NIR reflectivity is more than nine times higher than that of UV–VIS reflectivity, despite the significant variation in visible colour. A similar pattern of higher variance in NIR reflectivity is observed in butterflies (NIR 1.67 times higher), whereas the variance in the two wavelength ranges is approximately equal for birds (in sunbirds, even higher variance in the UV–VIS range) (Medina et al., 2018; Munro et al., 2019; Shawkey et al., 2017). High variance in NIR reflectivity indicates capacity for jewel beetles – and potentially poikilothermic insects more generally – to respond to thermal selection pressures by varying NIR reflectance. This is consistent with a stronger importance of reflectivity for thermoregulation in ectotherms than endotherms (Stuart-Fox et al., 2017). The high variation in NIR reflectance raises intriguing questions regarding underlying mechanisms. Currently, it is unclear how NIR reflectance varies for pigment-based versus structural colours. However, a great diversity of reflectance profiles can be achieved through structural modifications, enabling UV–VIS and NIR reflectance to be uncoupled to some extent (Stuart-Fox et al., 2017). Elucidation of the mechanisms underlying adaptive variation in NIR reflectance is a novel and promising area for future research.

The variance in heating rate we observed among jewel beetle samples is likely large enough to have biological significance. The body temperature of poikilotherms depends on ambient temperature and influences different activities such as foraging and flight. For example, the movement of the peach rootborer (Capnodis tenebrionis, a jewel beetle) tripled when the ambient temperature increased from 25 to 30°C, indicating the transition from non-active to active status (Bonsignore and Bellamy, 2007). Although the latter study examined temperature increases of 5°C increments, smaller increases over this critical range may result in substantial improvements in activity level. The maximum difference in heating rate of elytron samples in our experiments is 1.3°C min⁻¹ (ALL, min.–max.: 2.0–3.3°C min⁻¹). This variation in heating rate is likely to correspond to biologically meaningful differences in body temperature and lead to significant thermal consequences, particularly for small species with low thermal inertia.

The role of reflectance in thermoregulation may be critical to reach active body temperature, particularly in the morning on cloudy, cool days. When air temperature is low, radiative heat gain is essential for poikilotherms to reach active temperature (i.e. basking; Kingsolver, 1983b). Warming up faster enables an earlier access to limited resources (e.g. food, mates) than interspecific and intraspecific competitors and longer activity time, which potentially leads to higher fitness (e.g. longer flight time relates to higher reproductive success for Colias butterflies; Ellers and Boggs, 2004; Kingsolver, 1983a). The thermal benefit of low reflectivity is also suggested in butterflies; species in cooler, temperate environments have lower reflectance of the thorax and basal wings, which may enable them to warm up the flight muscles faster (Kang et al., 2021; Munro et al., 2019). In contrast, high NIR reflectance may help to prevent overheating, allowing beetles to be active for longer when exposed to sunlight. High NIR reflectivity has been shown to be advantageous for birds in hot, dry environments (Medina et al., 2018). However, small organisms such as jewel beetles can relatively easily avoid exposure to sunlight, so behavioural thermoregulation, rather than reflectivity, may be the primary adaptive strategy to prevent overheating. We have observed Temognatha jewel beetles hiding themselves in the shade on host plants at noon when it is extremely hot on summer days (A.M.F., personal observation). It is important to note that the effect of reflectance on body temperature could be overridden by convection under natural environmental conditions (Umbers et al., 2013). Nevertheless, our results suggest that reflectance can influence radiative heat gain, which would be maximized when the beetles are exposed to sun and sheltered from wind.

Our results show strong thermal effects of elytral effective area on the maximum heating rate of isolated elytra under full spectrum illumination. Size varies greatly among our jewel beetle samples and affects the heating rate of isolated elytra, though the effect was only detectable under full spectrum illumination. Our results also indicate that body size affects the rate of heat transfer from the isolated elytra to the body. For the large T. chalcodera, the heating rate of the isolated elytra was 67% higher than the body owing to thermal inertia, whereas for the small M. propinqua, the heating rate of the isolated elytra was only 22% higher than the body. Although body size is also correlated with cuticle thickness, we found no thermal effect of cuticle thickness under any illumination. In dung beetles, thoracic volume (Carrascal et al., 2017) and chitin weight per volume unit (Amore et al., 2017), which are both correlated with body size, have a significant effect on heating rate. Chitin weight per volume unit provides a measure of thermal mass; however, cuticle thickness may not correspond well to thermal mass because the cuticle structure, density and associated air volumes may vary between species. Additionally, the effect of cuticle thickness may be weak relative to effects of elytron reflectivity and effective area.

Surface sculpturing showed no effect on heating rate in our experiment, which could be because any thermal benefit of surface sculpturing is linked to convection (Casey, 1992) or because its function is unrelated to thermoregulation. Complex surface sculpturing increases effective area, which may result in higher radiative heat gain; but our results suggest no significant effect on heating rate. Surface sculptures such as dimples or protrusions affect convective heat transfer (Elyyan et al., 2008) to varying degrees depending on the geometry (Burgess and Ligrani, 2005). For example, the strength and intensity of vortices increase as the dimple depth increases (Burgess and Ligrani, 2005). We excluded airflow in the thermal chamber so very limited convective heat exchange occurred around the surface of elytra. Other functions of surface sculpturing of elytra have been proposed, including sexual recognition (Lindroth, 1974), background matching (Lindroth, 1974), thermal protection (Lindroth, 1974), water/dust repellency (Ball, 1985; Lindroth, 1974) and friction reduction when burrowing (Baehr, 1979), yet few have been tested experimentally. Some of these functions may depend on other cuticle properties. For instance, Namib tenebrionid beetles collect water droplets from the air using bumpy elytra in a combination of hydrophobic and hydrophilic surfaces (Parker and Lawrence, 2001). Though several functions have been proposed, surface sculpturing of elytra has also been hypothesized to be an ancestral trait, having neither specific function nor selective value but being preserved solely because the sculpturing is harmless and not costly (Ball, 1985; Lindroth, 1974).

In conclusion, we provide a rigorous test of how optical and morphological properties of the integument influence the heating rate of jewel beetle elytra. We demonstrate a thermal effect of NIR reflectivity on integument heating rates. Our experiments highlight the importance of NIR reflectivity as a potential adaptation for thermoregulation in beetles and, more broadly, poikilotherms.

We thank Australian National Insect Collection for providing elytron samples and the Melbourne TrACEES (Trace Analysis for Chemical, Earth and Environmental Science) Platform for access to the Phoenix Nanotom M micro-CT scanner. We thank Laura Ospina for help in the experimental setup and protocol, and providing filter transmission spectra.