Cassowary casques act as thermal windows

Many ideas have been put forward for the adaptive value of the cassowary casque; and yet, its purpose remains speculative. Homeothermic animals elevate body temperature through metabolic heat production. Heat gain must be offset by heat loss to maintain internal temperatures within a range for optimal performance. Living in a tropical climate, cassowaries, being large bodied, dark feathered birds, are under thermal pressure to offload heat. We tested the original hypothesis that the casque acts as a thermal window. With infrared thermographic analyses of living cassowaries over an expansive range of ambient temperatures, we provide evidence that the casque acts as a thermal radiator, offloading heat at high temperatures and restricting heat loss at low temperatures. Interestingly, at intermediate temperatures, the casque appears thermally heterogeneous, with the posterior of the casque heating up before the front half. These findings might have implications for the function of similar structures in avian and non-avian dinosaurs.

Results the cassowary has thermal windows. To better understand whether cassowary's use their casque for temperature homeostasis, we conducted infrared thermographic analyses of 20 live cassowaries over an expansive range of ambient temperatures (T a ). Consistent with this prediction, we found that the casque displayed evidence of reactive vasomotion across different heat loads. Indeed, the casque, distal end of the bill, and legs all showed a capacity for thermal adjustment (Fig. 1). These regions showed relatively large differentials between appendage surface temperature and T a at intermediate T a , and smaller thermal contrasts at either end of T a extremes. Conversely, the body was minimally affected by changes in T a , owing to insulation from the feathers. The proximal bill and neck displayed a linear negative relationship with increasing T a suggesting no active regulation of blood flow to these surfaces. This is not unexpected since at a T a of 10 °C, the differential is ~20 °C because the surface temperatures of both regions are ~30 °C. Similarly, at a T a of 30 °C, the differential declines to ~5 °C because the neck and proximal bill are still close to 30 °C. In this way, these regions are not making large thermal adjustments. The capacity for thermal adjustment translated into propensity for heat exchange, although body surfaces varied significantly in their ability to exchange heat with the environment (Fig. 2, Table S1, χ 2 = 741.70, d.f. = 6, P < 0.001). Notably, the surfaces of the cassowary most similar to core body temperature (i.e., the uninsulated eye, neck and proximal bill) were thermal windows at low T a , losing heat to the cold environment. The ability of a body region to serve as a thermal window depended on T a (χ 2 = 420.30, d.f. = 6, P < 0.001). For instance, the body appendages from which the most heat was lost at low T a also lost less heat (per m 2 ) at the highest temperatures. the casque is the most important thermal window at high heat loads. Importantly, at the highest T a , the casque dissipated the greatest amount of heat (per m 2 ) to the environment (Fig. 2). The distal bill and legs likewise increased the amount of heat lost to the environment at increasing T a , at least when represented as a function of total heat exchange (Fig. 3). Moreover, the casque offloaded more heat to the environment than the eye, bill (distal and proximal) and neck at high T a . Indeed, heat loss across the casque accounted for 8% of all heat exchange, a disproportionately high value given the relatively small size of the casque. In this way, the casque served as a substantial thermal window when the animal was under high thermal loads. This process is best illustrated by thermal images of the casque over cold, moderate and hot T a (Fig. 2). At low and high T a , the surface temperature of the casque approximates T a , likely owing to vasoconstriction and vasodilation, respectively, of casque vasculature. The thermal images also revealed a mosaic of thermal adjustments at moderate T a , wherein

Discussion
The function of the casque has remained enigmatic for nearly 200 years 1,29 . Our results confirm our hypothesis that the cassowary can regulate heat exchange via the casque. Indeed, cassowaries use their casque much in the same way as some other birds use their beaks for heat exchange 13,20,30 . Here, cassowaries used the casque as a thermal window when ambient temperature (T a ) was high, but restricting heat loss when T a was low. It is noteworthy that the cassowaries were observed dunking their heads into water at high T a (D.L.E. pers. obs.), perhaps to further increase cooling of the blood in the casque. Interestingly, at intermediate T a (25 °C), the temperature profile of the casque was heterogenous with the posterior of the casque heating up before the front half. Separate vessels may be supplying the anterior and posterior regions, enabling asynchronous dilation and constriction of vessels, as seen in the toco toucan's (Rhamphastos toco) bill 13 . Computed tomographic scanning reveals anterior-posterior differences in the amount of trabeculae in the cassowary casque, with dense trabeculae at the front that becomes increasingly diffuse towards the back 11 . Nonetheless, additional imaging would be useful to reconstruct the arterial and venous pathways of the casque vasculature, as has been used to (virtually) dissect avian vascular anatomy in other species 31 . Body surfaces that track T a are not considered to be actively involved in heat exchange, such as the feathers. The large proportion of total heat exchange by the body is simply a function of its large surface area. The eye, neck, and proximal bill likewise exchange some heat with the environment, but these regions are not specialised for that purpose. Instead, they maintain a consistently high surface temperature independent of environmental variation. Conversely, other body regions vary greatly in temperature over a range of T a , and can therefore be considered specialised for heat exchange 12 . Specifically, the casque, distal bill and legs were thermal windows.
Allen's rule hypothesizes that animals in cooler environments are expected to have smaller appendages, relative to whole-body size 32 . Appendages tend to lose the most heat, thus enlarged appendages will have increased overall heat dissipation, while conservation of heat is associated with reduced appendage size [32][33][34] . This holds true for bill size in many birds, which are often larger in birds from warmer environments than those from cooler environments 20 . Casques and casque-like structures are also present on birds inhabiting warmer habitats, for example hornbills (Bucerotiformes: Bucerotidae) 30 , helmeted guineafowl (Numida meleagris) 35 and the maleo (Macrocephalon maleo) 36 . Interestingly, pterosaurs and some dinosaurs, including members of the Ornithischia and Saurischia had similar cranial adornments [26][27][28] . While the function of these structures is still poorly understood, our results suggest these dinosaurian ornaments may have also served a thermoregulatory role. Hence, findings about the function of the casque in extant species could complement current and future findings in fossilised species with similar structures.

Methods
Collection of data. Live, adult southern cassowaries (Casuarius casuarius; n = 20) were photographed with a hand-held thermal imager (Testo 875i, Testo Ag, Lenzkirch, Germany) in zoological parks ranging from Victoria in southern Australia (outside of their natural range) to Queensland in northern Australia, between April 2014 and September 2015 (austral autumn and spring, respectively). These parks were chosen to encompass a wide thermal range without the need to restrain birds within temperature controlled facilities, which would be logistically unlikely owing to their protected status and dangerous disposition. Meteorological conditions were recorded using a wireless weather station (Vantage Pro 2 Plus, Davis Instruments Australia, Pty. Ltd., Kilsyth, Australia) situated within 10 metres of the cassowary's enclosure. Thermal images were taken from a distance of 0.5-2 metres. The camera had a double lens which produced a thermal image and a digital image at a resolution of 160 × 120 pixels. Images were collected between 0800-1700 h covering a wide range of ambient temperatures (T a ), 5-36 °C. Temperature data from the images were measured using IRSoft 3.1 (Testo Ag, Lenzkirch, Germany). Average surface temperature of the casque, eye, bill (distal and proximal), neck, body (torso) and 'legs' (tarsometatarsus only as the tibiotarsus is covered with feathers) were calculated, assuming an emissivity of 0.96. Mean surface temperatures were measured using either a line (neck, legs, body), point (eye), circle (distal bill; around nostrils, and proximal bill; anterior to eye) or by tracing the area (casque). The casque was further divided into four regions: the distal anterior, proximal anterior, distal posterior and proximal posterior. A total of 2,487 images were used.
The La Trobe University Animal Ethics Committee approved the study (AEC14-11) and all methods were carried out in accordance with the relevant guidelines and regulations.
Calculation of heat exchange. Heat exchange (q; W/m 2 ) estimates were calculated according to previously published studies 37 , and using the Thermimage package (v. 3.1) in R 38,39 . The estimates of heat exchange (loss = negative, gain = positive) were calculated as the sum of the convective and radiative heat exchange from a particular body surface, incorporating local measurements of ambient temperature, solar radiation, wind speed, and relative humidity. Operative temperature experienced by the cassowaries was not estimated due to an inability to safely measure the precise local microenvironment near the bird at the time of image capture. To visually inspect differences among surfaces, we plotted estimates of heat exchange at three values of ambient temperature spanning the ranges of measured ambient temperatures. Total heat exchange (Q) was estimated by multiplying heat exchange by area values (m 2 ) obtained from adult specimens found in museums. To assess the proportional role of the casque in heat loss and gain, we fit linear mixed models of Q casque as a function of Q total , and used the slope estimates as an indicator of the proportion of heat exchanged by the appendage.
statistical analyses. To quantify the influence of ambient temperature on regional differences in casque surface temperature, we fit linear mixed effects models 40 , incorporating the casque surface of interest (broken into 4 quadrants based on proximal-distal and posterior-anterior designations) as a fixed effect and animal identity as a random effect, with image identity nested within animal identity. Following the estimation of body surface specific heat fluxes, we fitted linear mixed effects models of the area specific heat exchanges (W/m 2 ), using body surface of interest (eye, neck, proximal bill, distal bill, mean casque, leg, and body), ambient temperature, and time of day as fixed effects, animal identity as a random effect, with image identity nested within animal identity. Surface and ambient temperature interactions were also assessed in the above model. In all cases, residuals were verified for normality and homoscedasticity. Finally, to estimate the relative contributions of the casque to heat exchange, we compared the model estimates from total heat exchange for the casque to the total body heat exchange estimates, as well as to published estimates of basal heat production in similarly sized emus 41 . We used model fits (±95% confidence interval) as measures of support and to summarise results. P-values were obtained using likelihood ratio tests (Type II Wald's chi-square tests) using the car package in R 42 .

Data Availability
The datasets generated during and/or analyzed during the current study are availble from the corresponding author on reasonable request.