ABSTRACT
Energy metabolism at rest and during pre-flight warm-up was measured in a variety of anisopterous dragonflies.
Resting oxygen consumption was similar in its relation to body temperature (Tb) and body mass to that of other insects. At 30 °C, log M = 0·91 log m + 0·44, where M is metabolism (W) and m is body mass (kg).
Metabolism during warm-up was calculated both from measurements of Tb and from oxygen consumption. By the former method, log M = 1·01 log m + 2·22 at the maximum Tb attained during warm-up, and log M= 0·90 log m+ 1·87 at Tb = 30 °C. Oxygen consumption measurements mostly gave values of M about 15% higher.
Total energy cost of warm-up is directly related to mass, thermal conductance and Tb at takeoff, and inversely related to warm-up rate.
INTRODUCTION
In most animals, both resting and active metabolism (M) can be related to whole body mass (m) by empirical equations of the form M = amb. Considerable interest has centred on whether the exponent, b, reflects either proximate or ultimate forces controlling rates of metabolism. As yet, efforts to elucidate this relationship have met with very limited success. Nevertheless, such equations are of considerable value for at least two reasons. First, body mass has such a large effect on whole-body metabolism in most taxa that knowledge of this parameter can permit reasonable estimates of metabolism in untested species once the general relationship is known. Second, by showing how much variation in metabolism is accounted for by size, the residual variation becomes evident and its causes become more accessible to investigation.
The relation of energy utilization to body size is known only for a few groups of insects (Keister & Buck, 1974). Respiratory metabolism of adult Odonata, in particular, is little studied, although the larvae have received some attention (Sayle, 1928; Petitpren & Knight, 1970). The data that do exist refer primarily to insects at rest although many species warm up by synchronous contractions of the thoracic muscles before flight (e.g. Krogh & Zeuthen, 1941; Dorsett, 1962; Kammer & Heinrich, 1974). This phenomenon is well known in moths and bees and also occurs in numerous other groups, but little is known of its relation to body size (Heinrich & Casey, 1973; May, 1976b).
In this paper I attempt to fill in some of these gaps in our knowledge of insect energetics. I have measured the energy requirements of various species of dragonflies covering a wide range of body weights, both at rest and during warm-up. In some cases I have also determined the effect of temperature on metabolism. These data may eventually contribute to determination of complete energy budgets for these insects.
METHODS
Adult dragonflies were collected near Gainesville, Florida, U.S.A. (30 °N, 82 °W), in the Panama Canal Zone (9 °N, 80 °W), and near Urbana, Illinois, U.S.A. (40° N, 88° W). Animals were used on the day of capture in measurement of resting O2 consumption, or one to three days later in other experiments. The masses reported are those of live or freshly killed specimens. Ten male Anax junius and five male Tramea lacerata dried at 90 °C to constant mass had dry mass = 34% ‘wet’ mass (32–36%).
Statistical differences between sample means have been determined using Student’s t test; the criterion for significance is P < 0·05. Other statistical procedures are discussed as appropriate. Grouped data are expressed as mean ± S.D. unless otherwise noted.
Resting O2 consumption
Rate of O2 consumption in dragonflies at rest was determined by standard manometric techniques using a Gilson differential respirometer. Single individuals were introduced into flasks of 20–150 ml, depending on the species, and allowed at least one hour in dim light (an opaque cloth was placed over the respirometer) to acclimate to the flask and to come to thermal equilibrium. Thereafter, measurements were taken at 5–15 minute intervals for at least two hours. I did not observe activity directly. Intervals in which vigorous activity occurred were obvious because of greatly elevated O2 consumption, and these measurements were discarded. Carbon dioxide was absorbed by soda lime in each flask. An energy equivalent of 2·01 × 104 J.l−1 of O2 (4·8 cal.ml−1) was assumed. Experiments were run between 10·00 and 21·00 hours Eastern Daylight Savings Time.
Metabolism during warm-up
Body temperatures were monitored during warming and cooling with copper- constantan thermocouples (0·003 in or 0·002 in lead diameter) implanted into the thoracic muscles through a pinhole in the metepisternum. Thermocouple output was monitored continuously on a strip chart recorder fitted with an appropriate reference junction. Ambient temperature was controlled to ± 1 °C in a Lab-Line constant temperature room or, in Panama, at 25 ± 1 °C with room air conditioning. Variation in Tn during any one experiment was ordinarily less than 1·0°C. Warm-up commenced spontaneously or after I pinched the abdomen of the specimen.
Oxygen consumption during warm-up in three species of dragonflies from Illinois, Anax junius, Boyeria vinosa and Tramea lacerata, was measured in an open flow system. Compressed air was driven successively through a column of Drierite desiccant, a metabolism chamber (a Plexiglass tube 2·54 cm diameter × 13 cm long, volume reduced as appropriate with Styrofoam plugs), a second column of Drierite, and an RGI flowmeter, and into a Beckman F3 O2 analyser, then vented to the atmosphere. Flow rate was about 70 ml.min−1 and was read to the nearest ml.min−1. The flushing time constant (r) for the system was 24–27 s (mean 25·2). This parameter was determined by quickly switching a three-way valve to change the inflow from air to a mixture of 20·31% O2–79·69% N2, thus imposing essentially a step change in O2 concentration. The time required for the change in concentration to reach 63% of its final value is equal to r. There was also an absolute lag of about 30 s between chamber and analyser. Output of the analyser was monitored on a Honeywell Electronik 16 Multipoint chart recorder. Ambient temperature was controlled to ± 1 °C with room air conditioning.
Animals were prepared for measurement by cutting their wings off within 0·5 cm of the base and implanting a thermocouple into their thorax as described above. Air temperature within the chamber was also monitored with a copper-constantan thermocouple. Thermocouple leads were brought through a side arm in the chamber and connected to the chart recorder.
I calculated according to Depocas & Hart (1957), assuming an RQ of 1·0 and an energy equivalent of 2·13 × 104 J.l−1 of O2 (Bartholomew, 1968). Carbon dioxide was not removed from the gas outflow, so, if RQ was actually less than 1·0, was underestimated (about about 10% if RQ = 0·7). The energetic equivalent of a litre of O2 decreases with decreasing RQ, however, so that at the flow rates and approximate levels of involved here the calculated heat production would vary only 2% for any RQ from 0·7 to 1·0. The RQ probably is less than 1·0 (Kallapur & George, 1973) as implicitly assumed in calculating Mr.
Since oxygen concentration changed rapidly and recording was discontinuous (recorder cycle time, 28·5 s), I determined peak by drawing straight lines through the two points immediately before and the two immediately after the interval containing the peak. Minimum oxygen concentration was taken as a point midway between the intersection of these lines and the lowest recorded points. In three cases, simultaneous records on another recorder with a continuous trace gave results within 1 % of the value of calculated as above.
Total oxygen consumption was calculated by graphically integrating oxygen concentration over time, from the time warm-up began until oxygen concentration rose again to resting levels. This quantity was converted to total heat production (E0) as above. A corresponding value of total heat production, Ea, was determined from the record of Tb by measuring R and the median value of Tb— Ta within each interval from the time warm-up commenced until Tb began to drop exponentially. The average rate of heat production was calculated for each interval and summed over all intervals.
Specific heat
I determined s (equation 1) for whole dragonflies by modifying the method of Hart (1951). A calorimeter, consisting of a 450 ml Dewar flask with a Styrofoam lid, contained about 200 ml of water. A thermometer graduated in 0·1 °C and readable with a lens to 0·01 °C extended through a hole in the lid. A plastic screen attached by wooden supports to the lid held the dragonflies beneath the water. The thermal equivalent mass of the assembled calorimeter, including a small, magnetic stirring bar, was determined by rapidly introducing a known quantity of water at a known temperature above that of the flask and determining the equilibrium temperature. Four replications gave an average value equivalent to 18 g of water, and this was added to the actual mass of water in determinations of r.
The contents of the calorimeter were mixed continuously with a magnetic stirrer and were near room temperature, 21–22 °C. I killed and removed the wings from 30–35 dragonflies (Libellula luctuosa, with a few L. pulchella) weighing 10–12 g altogether, and heated them to 37–40 °C in a closed container with moist paper towelling. One specimen had a thermocouple implanted, and its temperature was taken as the average temperature of the entire mass of dragonflies. After the temperature of the dragonflies and of the calorimeter stabilized (the latter within 0·01 °C), the implanted dragonfly was removed and the remainder quickly transferred to the calorimeter. The temperature change was noted to the nearest 0·01 °C at intervals for 30–50 min. The value of r was calculated according to Weber (1941).
RESULTS
Resting metabolism
Table 1 summarizes data on resting metabolism of dragonflies taken during June- August. The table emphasizes the effects of taxonomic differences, size and temperature. Two species of Libellula are treated as one taxon. They occur together and are similar in behaviour and thermoregulatory ability (May, 1976 a), and mass-specific rates of metabolism do not differ significantly at 30 –C (Table 1).
Diel changes were not systematically investigated. All individuals were tested during their normal activity periods except four Libellula, four Pantala and two Tramea that were run after 1900 h. For each combination of species and temperature in which some species were tested relatively early and some relatively late in the day, data on mass-specific metabolism obtained before and after 1500 h were compared. Only in the case of Miathryria marcella at 40 °C was there a significant difference (14·7 ±1·6 W.kg−1 before 1500 h, 19·2 ± 2·6 W.kg−1 after 1500 h, four individuals in each group). Thus all data are lumped without regard for the time of day.
Sex made a significant difference in only one case, Pachydiplax longipennis at 30 °C. Mass-specific metabolism of females was 5·8 ± 0·67 W.kg−1 (n = 8) and of males was 4·7±0·45 W.kg−1(n = 8). At 20, 35 and 40 °C also, female Pacyhdiplax had higher rates of metabolism than males, but the differences were not significant. Data are lumped without regard to sex except for intraspecific comparisons within Pachydiplax.
The effect of Tb on mass-specific resting metabolism of four taxa of Libellulidae appears in Table 1. Values of Q10 mostly vary from 2 to 3 in the range 20–40 °C. All species are similar in temperature sensitivity. Fig. 1 shows in more detail the variation with temperature in male Pachydiplax. Male Pachydiplax were also tested in April and these results likewise appear in Fig. 1. In all cases metabolic rates were higher in April than at the corresponding temperature in June-August, although the difference was significant only at 30 °C. On the other hand, seven male Erythrodiplax connata at 30 °C expended 6·1 ±0·71 W.kg−1 in April, less than in June-August although not significantly so. In April at Gainesville, mean high temperature is 27·4 °C, mean low is 14·3 °C. In July the corresponding range is 32·5–21·7 °C (T. Walker, personal communication).
Fig. 2. shows the effect of body size on mean resting metabolic rate (Mr) at 30 °C. This temperature is near the low end of the normal range of daytime body temperature of many dragonflies in the field (May, 1976a). Least-squares linear regression of the log transformed variables yields log Mr = 0·91 log m +0·44, r = 0·97. Similar regressions at 20 °C and 40 °C give log Mr= 0·92 log +0·079 and log Mr = 0·79 log m + 0·32 respectively, but these are based only on Libellula spp., P. longipeinnis, M. marcella and E. connata.
Corbet (1963) recognized two broad behavioural categories among dragonflies - perchers and fliers. Perchers spend much of their time on stationary perches and make sallies from time to time to feed, defend their territory, etc. Fliers remain almost continuously on the wing during periods of activity. Four of the species shown in Fig. 2 are fliers, the remainder perchers. If these groups are analysed separately the relations of metabolism to mass are: fliers - log Mr = 0·77 log m + 0·062, r = 0·99; perchers - log MT = 0·81 log m + 0·025, r = 0·97 Analysis of covariance using a randomized groups design (Edwards, 1966) indicates that fliers have significantly higher resting metabolism at a given body mass than perchers (P < 0·005).
For four taxa in which oxygen consumption was measured in 12 or more individuals at 30 °C, the relation of individual metabolism to mass was: E. connata - log Mr= 0·99 log m + 0·77, r = 0·78,71 = 14,95% confidence limits of the slope = 0·60–1·39; Libellula - log Mr= 1·73 log m+ 3·05, r = 0·92, n = 16, confidence limits = 1·51 – 1·95; M. marcella -log Mr = 1·60 log m + 3·17, r = 0·94, n = 12, confidence limits = 1·36–1·84; P- longipennis - log Mr= 1·02 log m + 0·80, r = 0·78, n = 16, confidence limits = 0·66–1·38.
Metabolism during warm-up
The relationship to mass of mean Ma calculated from equation 1 is shown for 15 species of dragonflies in Fig. 3. Data on cooling constants, warm-up rates, and maximum Tb in these species have been presented elsewhere (May, 1976 a, b) except for Tetragoneuria cynosura collected in Illinois. In this species at Ta= 25 °C and Tb = 30 °C, mean warm-up rate of six individuals = 6·6 °C.min−1 (5·6–8·6°C. min−1). Mean Tb at takeoff was 35·3 °C (31·2–39·7 °C). In three individuals that warmed to very high Tb, warm-up rate decreased before takeoff so that mean warm-up rate at takeoff was only 5·7 °C.min−1. Mean K = 0·47 min−1, body mass = 0·191 g, and thoracic mass = 0·095 g.
Since warm-up rate was nearly constant except in Tetragoneuria, calculated Ma at takeoff approximates the maximum rate of heat production during warm-up. Its relation to mass (Fig. 3) is log Ma= 1·01 log m+ 2·22, r = 0·91. Large species tend to initiate flight at higher Tb than small species, so Fig. 3 also includes values of Ma calculated at Tb= 29–30 °C. Then log Ma = 0·90 log m+ 1·87, r = 0·94. Miathyria marcella took off at an average Tb of 27·3 °C and so is excluded from the latter regression.
Most workers have assumed that the specific heat of insect tissue is about 3·3 × 10 J(kg. °C) −1 as stated by Krogh & Zeuthen (1941), although these authors did not actually measure it. Three replicate determinations of specific heat of whole dragon flies gave values of 3·47, 3·37 and 3·26 × 103 J.kg−1. °C−1, in excellent agreement with the accepted value, which was therefore used in calculating Ma above. Shinozaki (1957) determined values ranging from 2·8 to 3·5 J.kg−1. °C−1 for four insect species.
Maximal heat production (M0) and cumulative energy expenditure (E0) during warm-up were also estimated independently from O2 consumption. In these experiments, only animals that warmed uninterruptedly, or at most with one or two brief pauses, were used.
Most dragonflies warming in the chamber wing-whirred continuously for a period of several minutes, then apparently tried to fly. Thereafter they usually struggled violently or alternately struggled, wing-whirred and cooled at brief intervals. During these periods might decrease or remain approximately constant. Body temperature rose slowly or fluctuated about a nearly constant level. When struggling ceased, calculated usually dropped to resting levels within 60 s, although in some cases it remained elevated for several minutes. Table 2 summarizes data on M0, Ma, E0 and E from these experiments.
DISCUSSION
Resting metabolism
Resting metabolism of dragonflies is unremarkable in either its magnitude or its relation to size. My results for adult A. junius at 30 °C agree closely with those of Petitpren & Knight (1970). They found a heat production of 6·3 × 10−3 W in this species. Fig. 4 compares heat production at rest in Odonata with Hemmingsen’s (1960) equation for poikilotherms in general as well as that of other adult insects of comparable size. All the data were adjusted to a temperature of 20 °C by assuming a Q10 of 2·5, the average for dragonflies from Table 1. The results for insects are mostly above the ‘standard’ line probably in part because the insects were not starved nor was the possibility of low levels of activity eliminated. The slope of the regression for each of the insect groups is higher than the standard slope of 0·75, although the difference is statistically significant only for Kittel’s data. Nevertheless, the consistency of the differences in slopes suggests that within many insect orders resting metabolism may increase more rapidly with mass than is the case in animals generally.
Fliers have slightly but uniformly higher Mr than perchers of the same size. The fliers represented in Fig. 2 also tend to have smaller thoraxes than the perchers, so the difference in Mr is not due to greater mass of flight muscle in fliers. It is possible, however, that fliers were more active during experiments than were perchers. When fliers and perchers are considered separately the slopes of log Mr vs. log m are lower than that of all species taken as a whole, and nearer 0·75, Nevertheless, if individual species are examined (1 flier, 3 perchers), these slopes are higher than the slope based on the means of all species. The difference is highly significant in the cases of Libellula and Miathyria. The small intraspecific size range greatly increases the possibility for error in the latter cases and makes it unwise to attach much weight to individual slopes. Nonetheless, the occurrence of relatively high slopes in all four species strengthens the suggestion that the relation of metabolism to mass varies between different subgroups within the Odonata.
The response of Mr to temperature also is well within the range of variation shown by other insects (Keister & Buck, 1974). All species show a slight tendency towards lower temperature sensitivity as Tb increases. This may result in some energy saving, since daytime body temperatures in the field are relatively high (May, 1976a), although in summer males of P. longipennis the region of lowest sensitivity corresponds to a temperature range just below that maintained most often in the field.
Fig. 1 suggests seasonal compensation of Mr in male Pachydiplax, although the difference is significant only at 30 °C. A similar compensatory change occurs in the upper limits of temperature tolerance in this species (May, 1976a). Results from E. connata show, however, that if seasonal adjustments occur they are not uniform throughout the order.
Metabolism during warm-up, and metabolic scope
Inaccuracies exist in determining true rates of heat production from either M0 or Ma. The principal of these in the former case results from the lag in the response of the apparatus to changes in O2 uptake. The method of Bartholomew et al. (1979) should largely correct for such errors. Nevertheless, still may be underestimated slightly in most cases, since this parameter changes continuously rather than in a series of steps.
The data on total O2 consumption provide an independent estimate of the error of maximum measurement since total O2 is not subject to underestimation due to system lags. Total heat production as calculated from total O2 consumption should exceed total heat production calculated from Tb by approximately the same amount that true M0 exceeds Ma (see below), provided that true and total O2 consumption are well correlated. A close linear correlation exists between apparent and total 02 consumption (r = 0·84), suggesting that the latter condition is met. Since E0 consistently exceeded Ea by about the same percentage that M0 exceeded Ma (Table 2), M0 is probably a good estimate of true heat production.
Both E0 and M0 exceed Ea and Ma, respectively, by 10–40% (Table 2). The differences apparently result from the fact that equation 1 does not account for evaporative heat loss, heat loss consequent to increased convection during wing movements, or variation in heat loss due to variable circulation from thorax to abdomen during warm-up.
The latter was not monitored during oxygen consumption measurements, but in separate experiments concurrent measurements during warm-up of thoracic temperature and the temperature of the basal abdominal segments indicated very little heat loss via circulation. At takeoff, the average change in thoracic temperature/change in abdominal temperature was: A. junius - 13·6 °C/0·57 °C (N = 17, Ta= 15–30 °C); B. vinosa - 7·1 °C/0·03 °C (TV = 6, Tn= 25 °C). Also, in several other insects (Heinrich & Bartholomew, 1971; Heinrich, 1976), as well as dragonflies (Heinrich & Casey, 1978), heat loss through this channel has proved to be negligible during warm-up.
Wind velocity in the vicinity of the thorax of intact A. junius during wing-whirring was assessed with a hot-bead anemometer (Vogel, 1969). Velocity was less than 0·1 m.s−1 at all points, corresponding to an increase in heat loss of less than 10–12%. Also, warm-up rates of eight A. junius with the wings excised near the base averaged 5·02±0·80 °C.min−1 vs. 4·77± 0·96 °C.min−1 for the same specimens with intact wings. It is not known whether wing removal affects parameters other than convective cooling, however. I estimated evaporative heat loss (Hf) by assuming that the ventilation rate was 1200 1 (kg.h) −1 (Weis-Fogh, 1967), and that dry air entering the tracheae at Ta was saturated at Tb and expired without cooling. Body temperature was taken to be the average Tb during warm-up and struggling. The results appear in Table 3. Estimates of convective and evaporative losses depend on assumptions, noted above, that undoubtedly introduce errors into the calculations. To the extent that they are reliable, however, such estimates suggest that the sum of these two avenues of heat loss is very close to the observed differences between Ma and M0 and between Ea and E0.
On the whole, therefore, data on oxygen consumption are consistent with calculated values of Ma and Ea. The small differences between M0 and Ma are readily explained in four out of five cases on the basis of evaporative cooling and a small increase in convective heat loss due to the vibrating wing stumps. Experiments with B. vinosa at Ta= 25 °C showed greater than usual discrepancies between the two methods of calculating metabolic rate. The reason is not clear, as the difference apparently cannot be attributed to greater rates of evaporation or circulation by B. vinosa. The fact that in this case M0 exceeded Ma by considerably more than E0 exceeded Ea suggests that actually might have been overestimated, since for other species and conditions Rie respective differences were about the same. Such discrepancies necessitate some caution in accepting heat production values determined by either method, but the overall agreement of the techniques supports their reliability and that of the data in Fig- 3.
Based on the data of Table 2, the regression shown in Fig. 3 probably underestimates actual metabolism by roughly 15% irrespective of tn. At Tb= 30 °C and M = 10−3 kg, expected Ma+ 15% is about 32·5 times expected Mr while at m = 10−4 kg, Ma+ 15% is 29·5 times Mr. All the species used in determining Ma were fliers, however, and when Ma is compared to Mr for fliers only, the relative values of Ma + 15% are 29·5 Mr at 10−3 kg and 19·5 Mr at 10−4 kg, suggesting that metabolic scope may decrease with decreasing mass. Bartholomew & Casey (1977) found a similar relationship of metabolic scope to mass in beetles during maximal terrestrial activity.
The relation of Ma to m is also dependent on Tb. In large species with relatively low cooling constants, the contribution to Ma of the K (Tb— Ta) term (equation 1) is small relative to the contribution due to warming rate (R). In small species, however, K(Tb— Ta) may greatly exceed R at high Tb— Ta. In other words, in the latter case the cost of overcoming heat loss is greater than that required for heat storage. Thus, in order to maintain constant R, small species must increase Ma more rapidly with Tb. For example, at the very beginning of warm-up at Tb= Ta = 25 °C, Ma = Rs(m/60) and the regression of Ma on m is log Ma= 1·07 log m +2·26. This implies that metabolic scope and the absolute level of Ma are more temperature sensitive in small than in large species. If the same is true of flight metabolism the consequences for small dragonflies in flight may be profound, since in the field Tb is generally more variable in small than in large species. Heat production probably depends strongly on the rate of muscle firing, however (Heinrich & Bartholomew, 1971; Kammer & Heinrich, 1974) and there is evidence to suggest that in moths wingbeat frequency is less dependent on temperature in flight than during warm-up (Heinrich & Bartholomew, 1971 ; unpublished observations). Wingbeat frequency has not been examined as a function of Tb in dragonflies.
ACKNOWLEDGEMENTS
I am grateful to the Smithsonian Tropical Research Institute, Balboa, Canal Zone, and to the Department of Zoology, University of Florida, for the use of facilities and equipment. Special thanks are due to Drs Ken Prestwich and John Anderson, of the latter department, for their assistance with measurements of resting metabolism. I am also particularly indebted to David Vleck and Dr G. A. Bartholomew for permitting me to use their unpublished technique for calculating instantaneous and to my father, Dr Frank P. May, for additional help with calculations. Dr T. J. Walker kindly supplied climatic data for Gainesville, and Dr T. M. Casey suggested a number of improvements in the manuscript. This work was supported in part by NASA grant NSD 2099 to Dr J. E. Heath.