Many birds can defend body temperature (T b ) far below air temperature (T a ) during acute heat exposure, but relatively little is known about how avian heat tolerance and evaporative cooling capacity varies with body mass (M b ), phylogeny or ecological factors. We determined maximum rates of evaporative heat dissipation and thermal end points (T b and T a associated with thermoregulatory failure) in three southern African ploceid passerines, the scalyfeathered weaver (Sporopipes squamifrons, M b ≈10 g), sociable weaver (Philetairus socius, M b ≈25 g) and white-browed sparrowweaver (Plocepasser mahali, M b ≈40 g). Birds were exposed to a ramped profile of progressively increasing T a , with continuous monitoring of behaviour and T b used to identify the onset of severe hyperthermia. The maximum T a birds tolerated ranged from 48°C to 54°C, and was positively related to M b . Values of T b associated with severe heat stress were in the range of 44 to 45°C. Rates of evaporative water loss (EWL) increased rapidly when T a exceeded T b , and maximum evaporative heat dissipation was equivalent to 141-222% of metabolic heat production. Fractional increases in EWL between T a <40°C and the highest T a reached by each species were 10.8 (S. squamifrons), 18.4 (P. socius) and 16.0 (P. mahali). Resting metabolic rates increased more gradually with T a than expected, probably reflecting the very low chamber humidity values we maintained. Our data suggest that, within a taxon, larger species can tolerate higher T a during acute heat stress.

Birds show phylogenetic variation in the relative importance of respiratory versus cutaneous evaporation, but the consequences for heat tolerance and evaporative cooling capacity remain unclear. We measured evaporative water loss (EWL), resting metabolic rate (RMR) and body temperature (T b ) in four arid-zone columbids from southern Africa [Namaqua dove (Oena capensis, ∼37 g), laughing dove (Spilopelia senegalensis, ∼89 g) and Cape turtle dove (Streptopelia capicola, ∼148 g)] and Australia [crested pigeon (Ocyphaps lophotes), ∼186 g] at air temperatures (T a ) of up to 62°C. There was no clear relationship between body mass and maximum T a tolerated during acute heat exposure. Maximum T b at very high T a was 43.1±1.0, 43.7±0.8, 44.7±0.3 and 44.3±0.8°C in Namaqua doves, laughing doves, Cape turtle doves and crested pigeons, respectively. In all four species, RMR increased significantly at T a above thermoneutrality, but the increases were relatively modest with RMR at T a =56°C being 32, 60, 99 and 11% higher, respectively, than at T a =35°C. At the highest T a values reached, evaporative heat loss was equivalent to 466, 227, 230 and 275% of metabolic heat production. The maximum ratio of evaporative heat loss to metabolic production observed in Namaqua doves, 4.66, exceeds by a substantial margin previous values reported for birds. Our results support the notion that cutaneous evaporation provides a highly efficient mechanism of heat dissipation and an enhanced ability to tolerate extremely high T a .

Facultative hyperthermia, the elevation of body temperature above normothermic levels, during heat exposure, importantly affects the water economy and heat balance of terrestrial endotherms. We currently lack a mechanistic understanding of the benefits hyperthermia provides for avian taxa. Facultative hyperthermia has been proposed to minimize rates of water loss via three distinct mechanisms: M1) by maintaining body temperature (Tb) above environmental temperatures (Te), heat can be lost non‐evaporatively, saving water; M2) by minimizing the thermal gradient when Te > Tb, environmental heat gain and evaporative water loss rates are reduced; and M3) by storing heat via increases in Tb which reduces evaporative heat loss demands and conserves water. Although individuals may benefit from all three mechanisms during heat exposure, the relative importance of each mechanism has not been quantified among species that differ in their body size, heat tolerance and mechanisms of evaporative heat dissipation. We measured resting metabolism, evaporative water loss and real‐time Tb from 33 species of birds representing nine orders ranging in mass from 8 to 300 g and estimated the water savings associated with each proposed mechanism. We show that facultative hyperthermia varies in its benefits among species. Small songbirds with comparatively low evaporative cooling capacities benefit most from (M1), and hyperthermia maintains a thermal gradient that allows non‐evaporative heat losses. Other species benefited most from (M2) minimizing evaporative losses via a reduced thermal gradient for heat gain at high Te. We found that (M3), heat storage, only improved the water economy of the sandgrouse, providing little benefit to other species. We propose that differences in the frequency and magnitude of hyperthermia will drive taxon‐specific differences in temperature sensitivity of tissues and enzymes and that the evolution of thermoregulatory mechanisms of evaporative heat dissipation may contribute to differences in basal metabolic rate among avian orders. Understanding the mechanistic basis of heat tolerance is essential to advance our understanding of the ecology of birds living in hot environments that are warming rapidly, where extreme heat events are already re‐structuring avian communities. A plain language summary is available for this article.

Little is known about the phylogenetic variation of avian evaporative cooling efficiency and heat tolerance in hot environments. We quantified thermoregulatory responses to high air temperature () in ∼100-g representatives of three orders, namely, the African cuckoo (, Cuculiformes), lilac-breasted roller (, Coraciiformes) and Burchell's starling (, Passeriformes). All three species initiated respiratory mechanisms to increase evaporative heat dissipation when body temperature () approached 41.5°C in response to increasing , with gular flutter observed in cuckoos and panting in rollers and starlings. Resting metabolic rate and evaporative water loss increased by quantitatively similar magnitudes in all three species, although maximum rates of evaporative water loss were proportionately lower in starlings. Evaporative cooling efficiency [defined as the ratio of evaporative heat loss (EHL) to metabolic heat production (MHP)] generally remained below 2.0 in cuckoos and starlings, but reached a maximum of ∼3.5 in rollers. The high value for rollers reveals a very efficient evaporative cooling mechanism, and is similar to EHL/MHP maxima for similarly sized columbids which very effectively dissipate heat via cutaneous evaporation. This unexpected phylogenetic variation among the orders tested in the physiological mechanisms of heat dissipation is an important step toward determining the evolution of heat tolerance traits in desert birds.

There was an error published in J. Exp. Biol. 219, pp. 2137-2144 The scaling of the y-axes in the insets of Figs 2 and 3 is incorrect. The corrected figures are printed below.We apologise to the authors and readers for any inconvenience this may have caused. We predicted that evaporative cooling in Burchell's sandgrouse (Pterocles burchelli) is highly efficient and provides the basis for tolerance of very high air temperature (T a ). We measured body temperature (T b ), resting metabolic rate (RMR) and evaporative water loss (EWL) at T a between 25°C and ∼58°C in birds exposed to successive increments in T a . Normothermic T b averaged 39.0°C, lower than typical avian values. At T a >34.5°C, T b increased linearly to a maximum of 43.6°C at T a =56°C. The upper critical limit of thermoneutrality (T uc ) was T a =43.8°C, closely coinciding with the onset of panting and gular flutter. Above the T uc , RMR increased 2.5-fold to 2.89 W at T a =56°C, a fractional increase far exceeding that of many other species under comparable conditions. Rates of EWL increased rapidly at T a >42.9°C to 7.84±0.90 g h −1 at T a =56°C, an 11-fold increase above minimal levels. Maximum evaporative cooling efficiency (ratio of evaporative heat loss to metabolic heat production) was 2.03, but could be as high as 2.70 if our assumption that the birds were metabolising lipids is incorrect. Thermoregulation at very high T a in P. burchelli was characterised by large increases in RMR and EWL, and is much less efficient than in taxa such as columbids and caprimulgids.