Temperature Regulation and Heat Dissipation During Flight in Birds

Birds are capable of flying in environments ranging from the hottest deserts to altitudes as high as 11 km (Laybourne, 1974) where the ambient temperature is approximately – 50 °C. Although there have been many studies of thermoregulation in resting birds, almost nothing is known about thermoregulation during flight.

Core or skin temperature during forward non-gliding flight has been measured in only two species. Hart and Roy (1967) reported an average core temperature of 44·5 °C and an average skin (breast) temperature of 43·0 °C in pigeons flying at ambient temperatures between 4 and 30 °C. Aulie (1971) measured core temperatures of 44·1 °C in pigeons and 42·1 °C in budgerigars flying at ambient temperatures of 25 and 29 °C. These are strikingly high temperatures even when compared with resting birds. However, all these measurements were made during flights of less than 10 min and in some cases of less than 2 min. Furthermore, the birds were either trailing wires or carrying a bulky external telemetry transmitter. This paper describes measurements of core and skin temperature made on starlings (Sturnus vulgaris) during flights of one-half to two h, using miniaturized implanted radio transmitters.

All of the experiments described here were carried out in a wind tunnel designed and constructed for bird research. The tunnel consisted of an intake and settling section 2 m long, a 2·2 m working section and a fan driven by a 17 kW DC motor downwind of the working section. The working section was rectangular in crosssection and tapered from 0·9 m wide by 0·5 m high at the front to 0·9 m square at the rear. Birds were prevented from flying out of the upwind end by a barrier composed of 2 mm nylon monofilament lines strung vertically on 13 mm centres, and were kept out of the fan by a net across the downwind end. Wind speed could be adjusted with the motor’s power supply. At the average wind speed used in experiments, 13·5 ms⁻¹ at the upwind end, the turbulence 0·1 m behind the upwind barrier was 5 % and the turbulence decreased to 3 % at a distance of 0·5 m downwind of the barrier. The tapering cross-section of the working section resulted in a gradient of wind speeds from 13·5 ms⁻¹ at the front to 9 ms⁻¹ at the rear. To check for side-to-side uniformity, the wind speed was measured with a hot wire anemometer, at various points in a plane perpendicular to the direction of wind flow and 0·1 m behind the upwind barrier. At all points more than 60 mm from the walls of the wind tunnel the wind speed was uniform to within 5 %.

The air temperature in the tunnel was controlled by opening and closing a pair of doors in front of the intake to admit either outdoor air or room air to the tunnel. An electric heater strung across the intake made it possible to raise tunnel temperature 2–4 °C above outside or room temperature.

During all flights air temperature was measured with a mercury thermometer fastened inside the wind tunnel (systematic error less than 0·2 °C). Absolute humidity (partial pressure of water vapour) could not be controlled in any way, but it was measured with a wet-dry bulb thermometer and recorded for each flight. During these experiments the absolute humidity ranged from 3·7 × 10³ kg/m³ to 1·9 × 10⁻² kg/m³ (4–18 mmHg).

Further details about the construction of the wind tunnel are presented elsewhere (Torre-Bueno, 1975).

All the starlings used in this project were taken with mist nets or traps from farms in Dutchess County, New York, during the autumn when they congregate in large roosts. Birds were kept in a large outdoor aviary. Mealworms and an insectivore diet developed by Lanyon (1971) were available ad lib.

The basic strategy of the training technique was to prevent the birds from doing anything else but fly in the wind tunnel. For this reason all birds had their toes taped in as natural a position as possible before each flight to prevent them from clinging to either the upwind barrier or the downwind net. Any bird which landed in the wind tunnel was immediately chased up by waving or banging a stick on the floor of the tunnel or, if this was ineffective, taken out, untaped and returned to its cage.

By closely watching a flying bird it was possible to detect intention movements made before landing or turning. When a bird in training made an intention movement we would reach into the tunnel and place a stick or hand in its expected path. This was almost always adequate to induce the starling to maintain forward flight.

After three weeks to two months of daily training about one-third of the starlings would reach the point where they would fly for more than an hour without disturbance. They were then ready to be used in experiments. The longest continuous flight made by a starling was 8 h.

During flight a bird’s air speed could be determined by noting the position in the gradient of wind speeds at which it flew. If a bird persistently flew toward or hit the upwind barrier, the motor was adjusted to a higher setting, thus shifting the whole gradient to higher wind speeds. Conversely, if a bird flew just in front of the downwind net the wind speed in the tunnel was lowered. By this means it was possible to determine the air speed which the birds preferred. All starlings used in this study flew at air speeds between 12·5 and 14·5 ms⁻¹ when given a choice. This is within the range of speeds reported by Meinertzhagen (1955) for starlings flying in the wild.

To avoid disturbing the birds with trailing wires, all the temperature data reported here were obtained using a radiotelemetry system. The transmitters consisted of a blocking oscillator which transmitted a tone in the AM band pulsed at a rate proportional to the resistance of a thermistor and therefore to temperature. Transmitters of this type are described in Southwick (1973) and MacKay (1968). The transmitters were encapsulated in layers of epoxy resin, beeswax and medical grade silicone rubber and were about 20 mm long and weighed less than 2 g. Two grams is about 3 % of a starling’s total body weight, or less than its weight changes from eating a meal. Therefore it is unlikely that the weight of the transmitters significantly increased the energetic cost of flight.

The receiving system consisted of a helical antenna surrounding the working section of the wind tunnel, an AM radio receiver which translated the radio frequency pulses into audio frequency pulses, a Schmitt trigger which sharpened and standardized the pulses, and a frequency meter which converted the interpulse interval into a d.c. level suitable for a chart recorder.

Transmitters were calibrated by immersing them in a well-stirred water bath and slowly raising or lowering the temperature of the bath. Interpulse interval was determined by using a digital period meter (Fluke Model 1900A) calibrated against line frequency. The water bath temperature was measured with a mercury thermometer graduated to o-i °C. Once a calibration curve was determined, the same period meter was used with an oscillator to adjust the frequency meter and chart recorder to give full-scale deflexion for the range of temperatures of interest (about 10 °C from 37 to 47 °C).

The overall systematic error of this system as determined during calibration is estimated to be ± 0·1 °C, primarily due to the systematic error in the chart recorder. The recording system was checked periodically for drift using the same oscillator-period meter system which was used to calibrate it. To check for drift in the transmitters, the transmitter used to take most of the data reported was recovered and recalibrated at the end of the experiment. The calibration had changed by less than 0·1 °C.

The time constant of the system was less than 1 min and was due to the thermal inertia of the transmitter. It was measured by exposing the transmitter to a temperature step. Since the ‘thermal time constant’ of a starling, determined from the fastest rate of change of core temperature observed, is more than twice as long, the records are not distorted by the thermal inertia of the transmitter.

For measurements of core temperature, transmitters were implanted in the peritoneal cavity. Surgery was done using a non-sterile technique and local anaesthetic and the incision closed with sutures and collodion. Birds recovered from surgery quickly and were used in experiments on the day following surgery. No evidence of fever was seen and the core temperature during rest was within the range found in healthy birds. Placement of thermistors was checked by X-ray photographs. All data reported here were taken from thermistors lying deep in the peritoneal cavity and near its anterior end. No thermistor was in contact with an air sac.

Skin temperature was measured with transmitters identical to those used for core temperature, except that the thermistor was connected to the transmitter by an electrically insulated lead 100 mm long instead of being mounted inside the transmitter package. Calibration and recording techniques were the same as for core temperature except that the time constant was determined from the manufacturer’s specifications for the thermistors (Thermometric Bio thermobeads) with a diameter of 0’5 mm and a nominal time constant of 0·5 s in air. This is very much faster than is necessary to accurately follow changes in skin temperature. The total systematic error of the transmitters varied from ±0·1 to ±0·5 °C, and was due largely to radiofrequency interference.

The thermistors were implanted under the skin of the breast. Measurements made during dissections showed that at this point the skin is less than 0·05 mm thick in starlings. It was assumed that the temperature measured at this depth is not significantly different from the temperature at the skin surface. To avoid cutting the pennamotor muscles or their innervation, which could abolish local thermo-regulatory feather movements, the thermistors were implanted by making an incision posterior to the sternum and pushing the thermistor forward under the skin with a forceps. The transmitter was then implanted in the peritoneal cavity and the incision sutured and closed with collodion. Thermistors were implanted in four locations in three birds (one bird was used twice, but the measurements were not made simultaneously). The four thermistor locations are illustrated in the insets in Fig. 5. The feather thickness varies over the breast, being thinnest just over the keel and becoming thicker on either side. In a starling held in the hand at room temperature the feather depth measured with a probe ranged from less than 1 mm at the centre of the breast where the keel almost projects through the feather layer to about 2 mm at a distance of 10 mm to the side of the keel (see inset b in Fig. 5). The thermistors referred to as ‘keel’ thermistors are near the keel and therefore under a thinner layer of feathers, while the ‘breast’ thermistors are near the centre of one side of the breast and therefore under a thicker layer of feathers.

Fig. 1 presents a typical record of core temperature during flight. It is from a starling flying at 13·5 ms⁻¹ at an ambient temperature of 3 °C. For the first 46 min the bird flew without any disturbance. Then it began making intention movements to land and it was necessary to reach into the wind tunnel and chase it to make it continue flying.

Several points are worth noting. First, the highest core temperature occurred within a few min of the start of the flight. Afterwards the core temperature declined to a steady-state level. This initial overshoot occurred on most, but not all flights. The peak core temperature was normally less than 1 °C higher than the steady-state temperature, and total duration of the overshoot was usually between five and ten min. This indicates that core temperatures taken during the first few min of flight cannot be accepted as representative of the core temperature during extended flights.

Secondly, after the end of the overshoot the core temperature was steady for as long as the bird would fly without disturbance. Between the 15th and 30th min of flight (in flights which lasted at least 30 min without disturbance) the mean difference between the minimum and maximum core temperatures was only 0·24 °C (s.d. = 0·17, n = 21). For comparing flights at different ambient temperatures the core temperature after 30 min in flights in which the bird flew at least this long without disturbance was defined as the steady-state temperature (i.e. the temperature it is assumed the bird would have maintained on a long flight).

Finally, even slight disturbance, such as waving a hand behind the bird or making a loud noise, was sufficient to substantially change its core temperature. This suggests that measurements made on birds carrying heavy telemetry apparatus or disturbed during flight may not be representative of the conditions in natural flights.

The flight shown in Fig. 1 was made at an ambient temperature well within the range which a flying starling can tolerate. For comparison, Fig. 2 is a record from a starling flying at an ambient temperature of 35 °C. At this temperature the bird had to be chased continuously for the first 15 min of flight. The overshoot in core temperature was both broader and higher than at lower ambient temperatures. After 10 min of flight the core temperature declined, but instead of reaching a steady state it increased again and reached very high levels; 22 min after the start of the flight, when the core temperature exceeded 46 °C, the ambient temperature was reduced to 16 °C. The core temperature rose to 46·8 °C and subsequently declined, within 5 min, to about 43 °C, a temperature typical of other flights at an ambient temperature of 16 °C. The starling showed no adverse effects from the high core temperature it experienced.

A total of 35 undisturbed flights of longer than 30 min were obtained from two starlings. Both were in winter plumage and all data points for each bird were collected within a period of 6 weeks. Alternate flights in warm and cold air were made to avoid confounding the data with seasonal effects.

At ambient temperatures between 0° and 28 °C both birds maintained an elevated and relatively constant body temperature during flight (Fig. 3). The mean 30 min core temperature, for flights between 0 and 28 °C, was 44·0 °C in bird no. 100 and 42·7 °C in bird no. 115. There was no significant correlation between core temperature and date or time of flight in either bird. Both birds were reluctant to fly at ambient temperatures above 28 °C, and did not maintain a steady core temperature when forced to fly at these temperatures. They were also reluctant to fly at ambient temperatures below 0 °C, and only one undisturbed flight of greater than 30 min was made below 0 °C. One bird was forced to fly at an ambient temperature of –18 °C. At this temperature ice formed in its nares within a few minutes and it began to fly with its bill open, presumably in order to breathe. Almost immediately ice began to form in its bill and I had to remove the bird from the tunnel and clear the nares and bill.

A typical record from a ‘breast’ thermistor is shown in Fig. 4. The starling (no. 112) flew at 13 ms⁻¹ without chasing or disturbance for the entire duration of the record. The transmitter used had a systematic error of ±0·1 °C. Ambient temperature was changed four times during the flight. Paradoxically the lowest skin temperature, 42·1°C,.was observed at an intermediate ambient temperature, 18·2 °C, and the highest skin temperature, 43·2°C, was observed at 2·4 °C, the lowest ambient temperature. Skin temperature was higher at 2·4 °C than at 6·4 °C.

After either the beginning of a flight or a change in ambient temperature, the skin temperature became constant more rapidly than the core temperature. I have, therefore, defined the skin temperature after 15 min of flight at one ambient temperature as the steady-state temperature. This made it possible to take several data points during the same flight.

The skin temperature data collected from the four implanted birds are summarized in Fig. 5. In well-feathered areas the skin temperature was close to the core temperature measured in the other starlings. As noted before, the lowest skin temperature in well-feathered areas occurred at intermediate ambient temperatures. Under a thinner layer of feathers the skin temperature declined with decreasing ambient temperatures, but even at an ambient temperature of 0 °C the skin temperature was closer to the core temperature than to the ambient.

At ambient temperatures below 10 °C starlings always flew with the bill closed. Between 10 °C and 15 °C some starlings opened the bill very slightly during flight. With increasing ambient temperature the bill was opened wider until at above 30 °C they always flew with the bill gaped to its fullest extent.

At moderate ambient temperatures most starlings flew with the legs retracted and the feet either hidden beneath the body feathers or just visible at the sides of the body. In all flights above 28 °C starlings extended all three leg joints so that the leg and foot stretched down below the body so as to expose them to the airstream. Continuous extension of the legs was never observed in undisturbed starlings flying at ambient temperatures below 28 °C. It is possible that if the feet had not been taped, intermediate conditions (such as exposing the digits while keeping the rest of the foot and leg covered) might have been observed.

If the core temperatures of 42·7 and 44·0 °C measured in this study actually represent the steady-state core temperatures of starlings during extended flights, they are among the highest temperatures observed in a normal healthy animal. In addition to being higher than is normally observed in most birds, they are 2–4 °C higher than the resting temperature of the starlings. Birds of several species (including Towpees, Western Plumed Pigeons, and House Wrens) all die within a few minutes when their core temperature reaches some level between 46 and 47 °C (Dawson & Schmidt-Nielsen, 1964; Dawson, 1954; Dawson & Bennett, 1973, Baldwin & Kendeigh, 1932). Therefore, the core temperatures of the flying starlings were probably just a few degrees below the lethal temperature. The core temperatures during flight were also high temperatures in a biochemical sense. Morowitz (1968) calculated the denaturation rate for a typical protein as a function of temperature, using data from proteins of various phyla. At 44 °C the calculated denaturation rate is 20 % per day. Either a flying starling is resynthesizing its body protein at a high rate or, more likely, the proteins of birds are more resistant to heat denaturation than those of mammals.

It seems reasonable to ask what function an elevated core temperature during flight might serve. One obvious possibility is that it is a thermoregulatory adjustment to the heat stress of heavy exercise. Hyperthermia appears as a response to heat stress in many species of birds (Calder & King, 1974). By increasing the thermal gradient between the bird and the environment, hyperthermia increases the heat flow to the environment. However, the core temperature of flying starlings does not follow the pattern observed during heat stress. In resting birds hyperthermia does not appear until the ambient temperature reaches an upper critical temperature, after which the degree of hyperthermia is an increasing function of ambient temperature. In the flying starlings, however, the core temperature is independent of ambient temperature over the entire range at which the birds were willing to fly. Since the starlings were reluctant to fly at temperatures below 0 °C and unable to fly at –18 °C, apparently because they were too cold, it is very difficult to believe that they maintained elevated core temperatures at 0 °C to facilitate heat loss.

Another possible function of the hyperthermia during flight is heat storage. Camels and other desert animals can conserve water by allowing the body temperature to rise during the day and eliminating the stored heat by radiation and convection at night (Schmidt-Nielsen, 1964). It seems unlikely, however, that a bird could store significant heat during a flight and get rid of it afterwards. Assuming that a starling has the same specific heat as water it would take approximately 330 joules to increase its body temperature 1 °C. The metabolic rate of a flying starling is approximately 10 watts (Torre-Bueno, 1976), and a flying starling would produce enough heat to raise its body temperature 4 °C in just over 2 min. If the function of the increased temperature during flight is to store heat, it would mean that the starling tolerates an increased body temperature for a flight which may last several hours to store the heat produced during the first few minutes’ of flight. The fact that the temperature settles to a steady-state level, after initially overshooting in the first few min of flight, is further evidence that the starling does not maintain a high core temperature because it is unable to get rid of the heat.

A clue to the possible function of the elevated core temperature during flight is found in studies of the relation of body temperature to exercise in humans. The core temperature during exercise in humans is elevated above the resting temperature. The degree of elevation is independent of the ambient temperature and other environmental conditions but is positively correlated with the rate of exercise (Nielsen, 1970). Asmussen & Böje (1945) have shown that in humans both the maximal work output and the efficiency of the muscles increase with increasing core and muscle temperature. They reported muscle temperatures as high as 40·7 °C during maximal work in humans. If bird muscle responds to increasing temperature in the same way, an elevated core temperature would have two advantages for a flying bird. Increasing the power output per unit mass of the pectoral muscles would enable birds to fly with a smaller and lighter muscle mass than would otherwise be the case. Since the energetic cost of flight is critically dependent on body mass (Tucker, 1973) this would reduce the energetic cost of flight. An increase in the efficiency of the flight muscles would also reduce the cost of transport by allowing a bird to fly at a greater distance for a given amount of fuel. It appears that, as in humans, the high core temperatures observed in flying starlings are not the result of an inadequate thermoregulatory system, but an adaptive response to strenuous exercise.

If increasing the body temperature does increase the efficiency or power of the flight muscles, it would be advantageous to a bird to increase its core temperature as fast at possible upon initiating a flight, even at the expense of causing core temperature to overshoot before it reached the steady state. This is a possible explanation for the overshoot in core temperature observed in most flights in the starlings. Asmussen & Böje (1945) observed a similar overshoot in muscle temperature in humans initiating heavy exercise.

The data on skin temperature provide some insight into the relative importance of the feathers and the body tissues as insulation. In this discussion the insulation or thermal resistance of all structures below the skin surface will be referred to as skin insulation.

The thermal resistance of both the skin and feathers can be adjusted, the former by vasodilation and constriction, and the latter by feather erection. The relative change in these two quantities is probably responsible for the seemingly paradoxical rise of skin temperature at low ambient temperatures. The temperature difference across a structure is proportional to the product of its thermal resistance and the rate of heat flow through it ; ΔT = QR. Since core temperature is independent of ambient temperature; an increase in skin temperature must result in a decrease in the temperature difference between the core and the skin surface. This means that either the thermal resistance of the skin on the breast or the rate of heat flow through the breast decreases at low ambient temperatures. It would indeed be paradoxical if the thermal resistance of the skin were to decrease, but there are reasons to expect a decrease in heat flow. Several parts of the body (such as the bill, face and underwing surface) are poorly feathered. Since the bird’s ability to increase thermal resistance in these regions is limited, the rate of heat loss through them will increase as ambient temperature decreases. Veghte & Herreid (1965) have shown that in birds resting at low ambient temperatures the face and bill are warmer than the outer surface of the feathers in other parts of the body, indicating that they are losing heat faster. If a bird cannot avoid an increase in heat loss from some regions, it must compensate by either increasing its metabolic rate or reducing the heat loss in other regions. It seems reasonable to assume that in well-feathered regions such as the breast the thermal resistance of the feathers can be increased sufficiently to reduce the heat flow even if the ambient temperature decreases. This drop in heat flow will inevitably lead to a rise in skin temperature. Therefore, the increase in skin temperature is not a sign of an abnormality in the thermoregulatory system, but is an adaptive response to an inevitable increase in heat flow elsewhere in the body.

This work was supported by the Rockefeller University and by NIH Research Grant HL-02228 to Knut Schmidt-Nielsen.