ABSTRACT
This study examines variation in thoracic temperatures, rates of pre-flight warm-up and heat loss in the solitary bee Anthophora plumipes (Hymenoptera; Anthophoridae).
Thoracic temperatures were measured both during free flight in the field and during tethered flight in the laboratory, over a range of ambient temperatures. These two techniques give independent measures of thermoregulatory ability. In terms of the gradient of thoracic temperature on ambient temperature, thermoregulation by A. plumipes is more effective before flight than during flight.
Warm-up rates and body temperatures correlate positively with body mass, while mass-specific rates of heat loss correlate negatively with body mass. Larger bees are significantly more likely to achieve flight temperatures at low ambient temperatures.
Simultaneous measurement of thoracic and abdominal temperatures shows that A. plumipes is capable of regulating heat flow between thorax and abdomen. Accelerated thoracic cooling is only demonstrated at high ambient temperatures.
Anthophora plumipes is able to fly at low ambient temperatures by tolerating thoracic temperatures as low as 25°C, reducing the metabolic expense of endothermic activity.
Rates of heat generation and loss are used to calculate the thermal power generated by A. plumipes and the total energetic cost of warm-up under different thermal conditions. The power generated increases with thoracic temperature excess and ambient temperature. The total cost of warm-up correlates negatively with ambient temperature.
INTRODUCTION
The majority of studies of endothermy in bees in temperate and cool climates have been on social species in the family Apidae, particularly in the genera Apis (e.g. Heinrich, 1979; Cooper et al. 1985; Dyer and Seeley, 1987; Coelho, 1991; Underwood, 1991) and Bombus (e.g. Heinrich, 1972a,b, 1976; Prys-Jones, 1986; Surholt et al. 1990; Esch and Goller, 1991). Among solitary bees most studies have been on relatively large species active in warm or tropical climates, particularly the carpenter bees of the genus Xylocopa (Anthophoridae) (e.g. Chappell, 1982; Nicolson and Louw, 1982; Louw and Nicolson, 1983; Baird, 1986; Heinrich and Buchmann, 1986; Willmer, 1988; Surholt et al. 1990). Comparisons across species show that body mass alone is not a good indicator of endothermic ability (May, 1976) and small species adapted to cold thermal regimes are capable of high rates of warm-up and high thoracic temperatures (Stone and Willmer, 1989b). Endothermy is widespread among small solitary bees active in cool climates and is known in the families Andrenidae, Anthophoridae, Colletidae, Halictidae and Megachilidae (Stone and Willmer, 1989b). Do small solitary species have thermoregulatory abilities comparable with the better known social species active in similar environments? How do their responses to changes in ambient temperature, in terms of the thermal power generated and the total energetic cost of warm-up, compare to those of Bombus, the best known endothermic bees? This study addresses these questions through detailed study of a small anthophorid solitary bee, Anthophora plumipes, active in a cold thermal regime.
Anthophora is a large genus of fast-flying, robust bees occurring on all continents except Australia and South America. They are often extremely furry and all members of the genus examined to date are extremely endothermic (G. Stone, in preparation). In Britain, the commonest species is Anthophora plumipes, whose geographic range extends as far as Israel in the east. In Britain, A. plumipes flies from March until May, and in Israel from February until April. Throughout its range it is active in the spring when weather conditions and ambient temperature (Ta) fluctuate widely. This variation creates a situation in which some degree of endothermic thermoregulation has advantages over activity that is governed solely by dependence on unpredictable environmental conditions.
An important variable in studies of thermal physiology is body mass because, for organisms of a constant form, body mass determines surface area to volume ratios and hence the balance between mass-specific rates of heat generation and loss (May, 1976; Bartholomew, 1981; Heinrich and Heinrich, 1983). Across species, body mass is an important variable both in heterothermic insects (Stone and Willmer, 1989b; Coelho, 1991) and heterothermic mammals (Stone and Purvis, 1992). This study examines in detail the role of body size in warm-up rates and body temperatures in a single species.
MATERIALS AND METHODS
Field measurements of body temperature
The ‘grab-and-stab’ techniques used in this study are as described by Stone and Willmer (1989a). Grab-and-stab measurements were made at feeding and nesting sites in the Botanical Gardens, Oxford, and in the grounds of Merton College and University College, Oxford, during 1987, 1988 and 1989. Laboratory measurements of body temperature were made at Oxford University Department of Zoology over the same period and at the Botany Department of the Hebrew University, Givat Ram, Jerusalem, in February and March 1989. In Israel, bees were collected from an artificial mediterranean plant community established at Beit Jala (Har Gilo) in the Occupied Territories of the West Bank to the south of Jerusalem.
Laboratory measurement of warm-up rates and body temperatures
During warm-up and tethered flight, the bee was suspended from a fine thermocouple implanted shallowly in the thoracic flight muscles, as described by Stone and Willmer (1989b). In bumblebees (Heinrich, 1972, 1976) and carpenter bees (Heinrich and Buchmann, 1986), the temperature of the thorax is controlled by regulation of heat transfer from the thorax in the form of hot haemolymph passing down the petiole into the abdomen. At low Ta, Bombus minimises heat loss from the thorax to the abdomen by operation of a countercurrent heat exchange system in the petiole (Heinrich, 1976). Continuous measurement of abdominal temperature (Tab) in A. plumipes was achieved using a flexible copper–constantan thermocouple (diameter 0.1mm) inserted through a small hole in the second abdominal tergite. The thermocouple was inserted dorso-laterally to avoid damage to the dorsal heart, to a depth of approximately 1mm, and secured in place with adhesive. The temperature at which a bee initiated tethered flight is referred to as its voluntary flight temperature (VFT). After flight for a period of 60s or so, body temperature usually stabilised at a value termed the stable flight temperature (SFT) (Stone and Willmer, 1989a). After each experiment, bees were released, apparently unharmed, at the site of capture. Laboratory investigations of thermogenesis were carried out at four ambient temperatures (Ta): 9, 16, 21 and 29°C.
After a number of warm-ups over the full range of thoracic temperature (Tth) from Ta to VFT, bees showed general lowering of warm-up rates. This apparent fatigue could be dramatically ‘cured’ by feeding the bee with a solution of sucrose. Shortly after feeding had been initiated, there was a marked increase in abdominal pumping and a rapid increase in Tth. The bee warmed to levels in excess of those in previous warm-ups and ceased feeding shortly before flight. This increase in apparent thermogenic ability remained for several subsequent warm-ups. The major effect of feeding was an increase in VFT and in the power of tethered flight. This suggests that observed levels of endothermy could be dependent on the energy reserves carried by the bee at the time of capture. Both male and female A. plumipes collect nectar to the exclusion of all other flight activities during the early part of their flying period (Stone, 1989). Bees were therefore collected during the later stages of this period to minimise the probability that they might be limited during warm-up by low levels of nectar in their crops. Each bee was also allowed only five periods of tethered flight before release.
Conductance
In order to exclude the possibility of physiological modification of cooling rates, cooling constants were obtained for freshly killed dead bees. The bee was attached to a thermocouple in the normal way and its thorax heated with a microscope lamp (48W, Vickers, UK) situated 10cm from the bee. Heating of head and abdomen was minimised by shading them with pieces of polished sheet steel, acting both as shading screens and heat sinks. The bee was enclosed in a Perspex chamber to minimise the cooling effects of air currents in the room. When the bee’s thorax had been warmed to, and stabilised at, the required temperature, it was allowed to cool passively until it had equilibrated with room air temperature. Each bee was warmed and allowed to cool three times. These three coolings were used to calculate a mean value of the cooling constant for each individual. Total body mass and thoracic mass were determined on an electric balance (Mettler AE160).
Statistical methods
When analysing the effects of more than one continuous variable (such as Ta and body mass) on another continuous variable (such as Tth), multiple regression has been used. Utilisation of this technique is only valid if all the data are statistically independent. When each data point comes from a different individual, this assumption is probably justified. When each individual contributes a different number of values to the data set it may not be. If there are differences between individuals (such as damage due to insertion of a thermocouple) that are not due to the variables being examined, individuals for which more data were obtained will bias the analysis towards this unknown variable. To control for this effect, each individual must contribute the same weight to the analysis. To achieve this, mean values of the variables being investigated were obtained for each individual and are referred to as individual means (e.g. individual mean warm-up rate, etc.). These values are then used in a normal multiple regression analysis. Comparisons between Israeli and British populations were made by using country of origin as a categorical variable in analyses.
RESULTS
Field measurements of body temperature
During free flight in the field Tth increased from an average of approximately 25°C at Ta=5°C to 39°C at Ta=26°C. The gradient of the least-squares regression of Tth on Ta is 0.62 (95% confidence limits 0.587–0.656; N=81, r2=0.93, P<0.001) and for Tab on Ta it is 0.96 (95% confidence limits 0.916–1.015; N=81, r2=0.92, P<0.001). As Ta increases, the gradients of both Tab and Tth on Ta decrease and a better fit for both Tth and Tab is obtained using polynomial regressions (Fig. 1). There is a strong positive correlation between body mass and Tth, controlling for the effect of Ta (N=80, r2=0.96, P<0.001).
Laboratory measurements of warm-up rate, voluntary flight temperatures and stable flight temperatures
Mean warm-up rates and body temperatures obtained at the different ambient temperatures are summarised in Table 1.
Warm-up at an ambient temperature of 21°C
Acceptable warm-up traces were obtained from 21 male and 28 female A. plumipes at Ta=21°C. A typical warm-up trace is shown in Fig. 2. During warm-up there was no audible buzzing, and no vibration of the thorax was visible through a dissecting microscope (magnification 30X). In almost all cases, the first warm-up from Ta was markedly curvilinear, warm-up rate increasing with Tth and sometimes decreasing slightly just before flight temperatures were reached. Thereafter, rates of warm-up became more linear as a result of increased rates of warm-up at low Tth.
Individual mean warm-up rate correlates positively with both individual mean Tth and body mass (Fig. 3A,B). When individual mean warm-up rate is regressed against both individual mean Tth and body mass, the significant effect of body mass disappears (N=48, r2=0.3, TthP<0.001, body mass NS). There is no significant difference between warm-up rates in males and females once the effects of Tth and body mass have been controlled for. At high Tth, the warm-up rates in large females reached more than 19degreesmin−1, among the highest rates in any heterotherm.
There is a strong positive correlation between the rate of abdominal pumping and Tth, as shown in Fig. 4. Furthermore, at a given Tth, smaller bees have higher rates of abdominal pumping and, at a given temperature and body mass, males have lower rates of pumping than females (multiple regression: N=236, r2=0.67, TthP<0.001, body mass P<0.001, sex P<0.001).
At Ta=21°C, the temperature at which flight was initiated by the bee (VFT) was somewhat higher than the temperature that it sustained during tethered flight (SFT) (Table 1). Larger bees warmed to a higher VFT than smaller bees (Fig. 5A) and maintained a higher SFT (Fig. 5B).
Initial rates of warm-up by all bees at 9°C were very low. Four out of 14 females and 9 out of 17 males failed to reach a Tth high enough for flight. Even brief interruption of flight by the bee, leading to cooling of the thorax by more than 2–3°C, made it impossible for flight to be resumed without warm-up. There was a strong positive correlation between ability to warm up to flight temperatures and body mass (Fig. 6). Despite low initial warm-up rates, A. plumipes generated a thoracic temperature excess (Tex) of 22–24°C at this Ta [individual mean male excess 22.6±0.5°C, (N=8); individual mean female excess 23.7±0.4°C (N=8)], 5–7°C more than the average Tex at take-off at Ta=21°C.
Warm-up rates at Ta=29°C were uniformly high (Table 1), although the maximum recorded warm-up rate (19.2degreesmin−1 for a 201mg female) was not higher than the maximum recorded at Ta=21°C. The Tex at which those A. plumipes that could be induced to fly flew was only 7–8°C (Table 1) and, when tethered flight was initiated, Tth rose rapidly. Although Tth approached an asymptotic SFT value, in only three males and three females were flights of over 15s in duration recorded. Prolonged flight would lead to the generation of far higher Tth values. In all cases, flight ceased once Tth approached 38–40°C.
At Ta=21°C there was no rise in Tab associated with a simultaneous decrease in Tth at cessation of tethered flight for any of the five A. plumipes examined. At Ta=29°C, however, Tth rose rapidly during flight to 38–40°C, and at the end of flight there was a rapid rise in Tab (Fig. 7).
The effect of ambient temperature on warm-up in A. plumipes
Warm-up rates and flight temperatures
Warm-up rate correlates positively with Ta (Fig. 8). Relationships between VFT, SFT and Ta give an indication of the thermoregulatory ability of A. plumipes independent of the conclusions based on ‘grab-and-stab’ data. For both males and females, gradients of both VFT and SFT on Ta are significantly less than 1 (mean ± standard errors): male VFT, 0.16±0.06; female VFT, 0.19±0.03; male SFT, 0.49±0.11; female SFT 0.44±0.03). For all four data sets a better fit is given by polynomial functions (Fig. 9). For both sexes at low Ta, VFT>SFT, while the relationship is reversed at higher Ta. Fig. 9 suggests that no temperature change with initiation of tethered flight (i.e. VFT=SFT) should occur at an ambient temperature of approximately 24.5°C for males and 20.0°C for females.
Time required to complete warm-up
The total time required to complete warm-up to flight temperature in A. plumipes is also dependent on Ta, as shown in Fig. 10. While at Ta=21°C, warm-up took a mean of 1.6±0.1min for both males (N=17) and females (N=11), warm-up at Ta=9°C took a mean of 11.2±0.6min (N=11) for males and 11.7±0.6min (N=9) for females. Larger bees warm up more rapidly than small bees (which reduces warm-up time), but also warm up further than small bees (which tends to extend warm-up). Thus, whether large bees take longer or shorter periods to complete warm-up depends on the balance between these effects. At Ta of both 9°C and 21°C these effects seem to cancel each other out – there was no significant effect of individual mean warm-up rate, body mass or individual mean VFT on warm-up duration.
Conductance
Conductance values were obtained for 14 A. plumipes (8 males and 6 females) at an ambient temperature of 21°C. The rate of heat loss correlates positively with Tex (Fig. 11A). The mean cooling constant for A. plumipes is 0.53min−1, giving a mean thermal conductance of 0.03 Wg−1 degree−1. The cooling constant (and therefore the thermal conductance) of A. plumipes decreases significantly with increasing thoracic mass; small bees lose heat more rapidly than large bees do (Fig. 11B). If total body mass is used rather than thoracic mass, males also lose heat more slowly than females do (N=14, body mass P=0.002, sex P=0.01).
Comparisons between Israeli and British populations of A. plumipes
Warm-up data at an ambient temperature of 21°C were obtained for four male and four female A. plumipes from Jerusalem. Although both males and females of the Israeli population have a lower mean body mass than the British population, they have higher mean VFT and SFT at 21°C, and males have higher warm-up rates than their British counterparts (Table 1). Having controlled for the effects of thoracic temperature and body mass, Israeli bees have a higher individual mean warm-up rate (multiple regression: N=56, r2=0.73, body mass P=0.065, TthP<0.001, country P=0.04) and higher VFT values, having controlled for the effect of body mass (multiple regression: N=49, r2=0.47, body mass P<0.001, country P=0.006).
DISCUSSION
Comparison of grab-and-stab and laboratory data
Table 2 shows mean SFT values at the ambient temperatures used in the laboratory together with mean grab-and-stab estimates for the same ambient temperatures taken from Fig. 2. The two methods agree well in their estimation of Tth in flight at high and at low Ta, although at moderate Ta grab-and-stab estimates are higher than the corresponding SFT values. The reasons why SFT values are probably under-estimates and grab-and-stab over-estimates of true Tth have already been discussed (Stone and Willmer, 1989a). Laboratory and field data agree that 25°C is very close to the minimum Tth at which this species flies and 39°C close to the maximum.
Power generation during warm-up
During warm-up, energy is expended both in movement of the musculature and in generation of heat. Here only the thermal power generated is considered and this will be an underestimate of the actual power generated (Bartholomew, 1981). The thermal power generated at a given Tex is the sum of the rate of heat storage in the thorax and the rate of passive heat loss. This can be expressed as: rate of heat generation at a given Tex = (warm-up rate X specific heat capacity of tissue X thoracic mass) + (passive cooling rate X specific heat capacity of tissue X thoracic mass), if heat losses from other body tagmata are ignored (Heinrich and Bartholomew, 1971). If it is assumed that, during warm-up, all the heat generated is sequestered in the thorax, then rates of passive heat loss depend only on Tex (as long as there is no significant variation in factors such as air movement around the bee). It is therefore possible to calculate such rates of heat loss at any Tex for any Ta from best-fit regressions of rate of cooling as a function of Tex, such as those shown in Fig. 5A. As shown above, there are strong correlations between warm-up rate and Tth at a given Ta. From these relationships it is possible to calculate the mean warm-up rate for A. plumipes at a given Tth at a specific Ta, and thus at a given Tex. From the relationships and quantities described above, the power generated at a specified Tth and Ta can be calculated. Table 3 shows steps in the calculation of the rate of heat production during warm-up for A. plumipes at 9 and 21°C. Power output for males and females as a function of Tex is shown in Fig. 12A.
Power output increases linearly with Tex and at a given Tex power output at 21°C is higher than that at 9°C. Although the rate of heat loss depends only on the Tex, the rate of heat generation depends on Tth (Fig. 12B) (Heinrich, 1975, 1987). At Ta=9°C (Table 3) and with a Tth of 15°C (a Tex of 6°C), the mean power generated by A. plumipes is 0.27W, and the rate of heat loss 0.19W. 70% of the estimated generated heat is lost through passive cooling. At Ta=21°C and with a Tth of 27°C (the same Tex of 6°C), while the rate of passive heat loss is still 0.19W, the total power generated is 0.63W. At the higher Ta, passive heat loss accounts for only 30% of the total power generated.
The power generated during warm-up is strikingly similar to that generated by queen bumblebees (Heinrich, 1975). At a Tex of 20°C, queen Bombus vosnesenskii produce 0.28–0.35 Wg−1 thorax (Heinrich, 1975). Table 3 shows that at Ta=21°C male A. plumipes produce 0.26–1.37 Wg−1 thorax, depending on the Tth. The maximum power output produced by A. plumipes (1.37 Wg−1 thorax at a Tth of 39°C at Ta=21°C) is somewhat higher than the maximum rate reported by Heinrich (1975) for B. vosnesenskii (1.05 Wg−1 thorax) and cuculiinine moths (Heinrich, 1987, maximum rates almost identical to those of B. vosnesenskii). Using the common approximation that 1ml of oxygen liberates 20.1J (e.g. Weis-Fogh, 1967; Casey et al. 1981), A. plumipes uses approximately 247ml O2 g−1 thorax h−1 (or 84.5ml O2 g−1 totalbodymass h−1), some-what higher than rates recorded for other endothermic insects (Kammer and Heinrich, 1974; Bartholomew et al. 1981; Casey, 1981; Casey et al. 1981; Morgan, 1987). The oxygen consumption of A. plumipes is comparable with rates determined for flying honeybees (Hocking, 1953) and bumblebees (Heinrich, 1975; Bertsch, 1984; Surholt et al. 1990; Ellington et al. 1990). This illustrates the fact that, in terms of metabolic power output, high rates of warm-up are compatible with flight. The greater the Tex that the bee is maintaining, the higher its rate of heat production must be to counter heat losses and the greater the required rate of oxygen supply to the tissues (Bartholomew and Barnhart, 1984), explaining the positive correlation between abdominal pumping rate and thoracic temperature. Body mass probably correlates negatively with pumping rate because small bees cool faster and have to expend more energy in maintaining the same Tex (Bartholomew and Casey, 1978).
The total energetic cost of warm-up
A is difficult to measure alone, and it has been suggested (Heinrich, 1975) that it is best to calculate CA using known estimates for the other variables. When a bee is maintaining its Tth at a steady value, dTth/dt=0 and dHp/dt=CATex. When a dead bee is cooling passively, dHp/dt=0 and WSdTth/dt+CATex=0. Because dTth/dt is negative, this can be rewritten as WSdTth/dt =CATex and CA=(WSdTth/dt) divided by Tex. W, S and Tex are easy to measure and dTth/dt is the rate of passive cooling at the Tex that is being maintained, determined from Fig. 11A. For a female A. plumipes, thermoregulating at Tth=36°C at Ta=21°C, CA=(0.065 g×3.4 Jg−1 degree−1×7 degreesmin−1)/(36-21)=0.1 J degree−1 min−1, slightly lower than the value given by Heinrich (1975) for queen Bombus vosnesenskii (0.18 J degree−1 min−1).
With an estimate of CA, the total energy required to warm-up at any Ta can be calculated once the area under the warm-up curve has been measured. Table 4 gives the values required to calculate the total energy required to warm up at four values of Ta. The results for A. plumipes are compared with data obtained in the same way by Heinrich (1975) for queen Bombus vosnesenskii in Fig. 13. It is clear from Table 4 and Fig. 13 that the lower the Ta and the longer it takes to raise Tth to flight temperatures, the more expensive the warm-up becomes. The total energy required by a female A. plumipes is rather less than that required by queen B. vosnesenskii. The mean mass of the thorax of queen B. vosnesenskii is 0.210g, over three times greater than that of female A. plumipes (0.065g), and the bumblebee therefore needs to expend more energy in order to raise the temperature of its thorax by 1°C than A. plumipes does. B. vosnesenskii is also a better thermoregulator before flight than A. plumipes is – at low Ta values, the former generates a rather higher Tex than the latter and reaches a Tth of near 36°C whatever the Ta (Heinrich, 1975). A. plumipes could be said to reduce the total energy required to warm up by sacrificing its ability to regulate Tth within such narrow bounds.
The inability of some of the smaller A. plumipes to warm up at Ta=9°C indicates that the period for which the required metabolic expenditure must be sustained, and the total Tex that must be generated, clearly place constraints on the conditions under which flight is possible.
Variation in the form of warm-up curves at a given ambient temperature
There is a change in the form of the warm-up curve over time at Ta=21°C in A. plumipes from a curvilinear first warm-up to more or less linear warm-ups thereafter. This change suggests that during the first warm-up the bee’s endothermy is in some way modified, resulting in higher metabolic rates at lower Tth during subsequent warm-ups. It is possible that such an effect might be caused by release of a modulating hormone such as octopamine. Octopamine levels have been shown to rise during the first few minutes of flight in locusts (Goosey and Candy, 1980) and cockroaches (Bailey et al. 1983), resulting in increased rates of carbohydrate metabolism (Candy, 1978; Whim and Evans, 1988) and increases in the force generated by the flight muscles (Whim and Evans, 1988). Endothermy occurs in the flight muscles and involves very high rates of metabolism, and octopamine release during the first warm-up could produce the observed changes in patterns of warm-up.
Falls in the VFT and warm-up rates obtained for a given animal at Ta=21°C are probably the results of exhaustion. Wigglesworth (1949) implicated substrate depletion as a cause of flight exhaustion in blowflies, and feeding with glucose rapidly restores flight ability (Hudson, 1958; Clegg and Evans, 1961). In bumblebees, the flight fuels are mono- and disaccharides stored in the gut (Surholt and Newsholme, 1983; Surholt et al. 1988), rather than carbohydrates mobilised from the fat body as in some other endothermic insects (Ziegler and Schultz, 1986; Joos, 1987). Radiolabelled glucose appears in the blood of blowflies less than 30s after ingestion (Clegg and Evans, 1961) and absorbed sugars are rapidly incorporated into the transport sugar trehalose (Treherne, 1957; Clegg and Evans, 1961; Friedman, 1967). It is therefore quite possible that the recovery of endothermic performance shown by Anthophora plumipes in response to feeding was due to availability of renewed energy sources, and that poor performance prior to feeding may have been due to low energy supplies.
How good a thermoregulator is A. plumipes?
If the gradient of the regression of Tth on Ta is used as the indicator of thermoregulatory ability (May, 1976; Stone and Willmer, 1989a), then it is clear that the thermoregulatory ability of A. plumipes depends on whether Tth before or during flight is used. Comparison of VFT on the one hand and SFT and ‘grab-and-stab’ results on the other indicates that regulation of Tth is better prior to flight than it is during flight. Table 5 summarises data for endothermic bees, moths and beetles. In every case where Tth both before and during flight have been recorded, thermoregulation before flight is far better than it is during flight. This result shows that although these endothermic insects are clearly able to regulate Tth in still air before flight, they are unable to compensate entirely either for the high rates of convective cooling during flight at low Ta or for the heat generated by flight muscle activity at high Ta. Achievement of higher thoracic temperatures than can be maintained in flight may have independent selective advantages. If higher muscular power and levels of sensory coordination are required for take-off than for normal flight, then higher values of Tth at take-off may be selected for regardless of the cooling that follows. It is also possible that higher values of Tth in non-flying A. plumipes at low Ta have advantages independent of flight; they may contribute to a greater ability to detect mates or predators in resting males or to digging ability while excavating nests in females (Stone, 1989).
Comparison of the gradients of best-fit regressions of Tth on Ta for insects sampled during flight shows that, compared to other endothermic insects, Anthophora plumipes is an average thermoregulator. It cannot match the levels of thermoregulation shown by bumblebees (Heinrich, 1972a,b, 1975) or by the much larger carpenter bees in the genus Xylocopa (Heinrich and Buchmann, 1986; Baird, 1986; Willmer, 1988). A. plumipes shows a similar gradient to those of Apis mellifera (Cooper et al. 1985; Heinrich, 1979) and the euglossine bees investigated by May and Casey (1983). The gradient of Tab on Ta does not differ significantly from one, indicating that there is no regulation of abdominal temperature. A. plumipes nonetheless manages to fly over a wide range of Ta values and, in particular, down to very low temperatures. This is at least in part due to the low minimum Tth this species will tolerate in flight. Minimum values of Tth in flight of about 25°C are well below the minimum Tth for flight recorded for Apis mellifera (Cooper et al. 1985) and the bumblebee species investigated to date (e.g. 30°C for Bombus edwardsii, Heinrich, 1975). The cuculiinine moths of the genera Lithophane and Eupsilia, which are able to warm up from temperatures near 0°C and have a similar body mass to A. plumipes, maintain a minimum Tth of around 30°C in flight at Ta=0°C (Heinrich, 1987). It appears that, through evolving a low minimum Tth for flight, A. plumipes has compromised on the high cost of maintaining a high Tex at low Ta.
Simultaneous measurement of Tth and Tab in A. plumipes suggests the presence of physiological regulation of heat flow between thorax and abdomen. A petiole countercurrent system is known from larger anthophorid carpenter bees in the genus Xylocopa (Heinrich and Buchmann, 1986). This study extends the evidence for such a physiological mechanism to smaller anthophorids. In A. plumipes flushing of heat to the abdomen appears only to occur at relatively high ambient temperatures; during most of its spring flight period this species rarely experiences such thermal stress.
The role of body mass in thermoregulation
VFT, SFT and temperatures during free flight all correlate positively with body mass in A. plumipes. At low Ta, high body mass is also advantageous in the generation of a sufficient thoracic temperature excess to allow flight. Females have a significantly higher mean body mass [Britain 185±x mg (N=x), Israel 177±5mg (N=6)] than males [Britain 160±x mg (N=x), Israel 141±20mg (N=5)], and these correlations with body mass should therefore generate mean sexual differences in physiological responses to changing Ta. Mass differences probably explain why female A. plumipes are able to fly to lower minimum ambient temperatures than males (Stone and Willmer, 1989b). If both females and males have similar upper tolerance limits for thoracic temperature, females will be forced to cease flight activity at a lower Ta than males. This effect of difference in mass between the sexes is demonstrated by the observation that females cease cooling with the onset of flight at a lower Ta than males do. These conclusions on the role of body mass agree with other studies of hymenopteran thermoregulation (Willmer, 1985a,b; Larsson, 1989a,b; Coelho, 1991). Considerable variation in warm-up rates and body temperature exists between bee taxa that is not explained by variation in body mass (G. Stone and Willmer, 1989b). The differences between Israeli and British populations suggest that physiological adaptation can occur within a species independently of differences in body size. Similar differences between populations of the same species have been demonstrated in Amegilla sapiens, a close relative of Anthophora plumipes (Stone, in preparation). Because no comparable measurements of thermal conductance were obtained for Israeli bees it is impossible to establish whether the higher VFT and warm-up rates of Israeli A. plumipes are due to higher rates of heat production or lower rates of heat loss per unit mass.
ACKNOWLEDGEMENTS
My sincere thanks go to Dr Pat Willmer and Dr Peter Miller for encouragement and support throughout this work. I would also like to thank Dr Steve Simpson, Dr Sally Corbet, Avi Shmida, Chris O’Toole and two anonymous referees for their constructive criticism. This work was supported by an SERC studentship.