Oxygen Consumption of Moths During Rest, Pre-Flight Warm-Up, and Flight in Relation to Body Size and Wing Morphology

For almost a century and a half it has been known that some insects during flight experience body temperatures significantly above air temperature and that the heat responsible for this temperature difference is a by-product of the activity of the flight muscles. However, the mechanisms of endothermic temperature control of insects have been examined in detail only during the last decade (see the review of Heinrich, 1974).

Flapping flight is the most energetically demanding mode of animal locomotion. The flight muscles of insects are the most metabolically active of tissues. Because the mechanical efficiency of muscle is approximately 20%, about four-fifths of the large expenditure of energy during insect flight appears as heat, some of which is retained in the thorax and can result in body temperatures as high as or higher than those of birds and mammals. Several studies have reported the metabolic rates of flying moths (see Heinrich, 1974) but only one study has examined the relation of oxygen consumption during flight to body mass and to the components of the flight machinery in moths (Casey, 1976).

Although the heat production that results in elevated body temperature is an obligatory consequence of the activity of flight muscle, in some moths, bees, and beetles elevated body temperature must be attained before flight is possible. Before these insects can become airborne they must undergo a pre-flight warm-up during which the elevator and depressor muscles contract nearly simultaneously rather than alternately, generating much heat and little wing movement.

In this study we (1) report the relationship of oxygen consumption to body mass in heterothermic moths of several families during rest, pre-flight warm-up, and flight,(2)present scaling relationships for components of the flight machinery of sphingids and saturniids and relate their energy metabolism during flight to wing span, wing loading, and wing beat frequency, (3) consider some of the ecological and evolutionary implications of these data, and (4) discuss the convergent patterns of energy expenditure evolved by heterothermic insects and vertebrates.

These studies were carried out during December 1975 and January 1976 at the Barro Colorado Island Station of the Smithsonian Tropical Research Institute in the Panama Canal Zone. Moths were captured at lights at night and measured the following day between 09.00 and 18.00 h. From time of capture until measurement each moth was held in a separate container and kept under continuous illumination to inhibit activity.

Oxygen consumption was determined from the decrease of oxygen in closed chambers of known volume during a measured time interval while the moths were either resting, undergoing pre-flight warm-up, or flying. The chambers were glass jars of approximately 0·2, 0·4, 0·9, 4·0 and 9·01 fitted with airtight lids equipped with three-way valves. The volume of each chamber was measured to the nearest ml by determining the amount of water required to fill it. The chamber used depended on the size of the moth and the activity being measured. All but the smallest moths were flown in either the 4·0 or the 9·01 jars.

Resting metabolism was measured for 20 min to an hour depending on body mass. The moth was then transferred to a second chamber and pinched, touched, or shaken to initiate warm-up. When warm-up began, the chamber was sealed. When the moth started to fly, an air sample was removed and the time was noted with a stop-watch. The animal was transferred to the flight chamber and flown for at least 2 min (3 min when possible). The duration of the flight was recorded and an air sample was taken. The flight chamber was then opened and the sound of the moths’ wing beats was recorded on a tape recorder through a microphone held at the entrance to, or inside, the chamber. The moth was then seized by the wings or shaken into a net and its thoracic temperature determined (see below). Wing beat frequency was subsequently determined by retranscribing the tape at either one-eighth or one-sixteenth normal speed, and counting the beats by ear during an interval timed by stop-watch.

All temperatures were measured with a copper-constantan thermocouple connected to a Bailey Instruments Co. Laboratory Thermometer calibrated with a U.S. Bureau of Standards mercury thermometer. Body temperatures were measured with a thermocouple mounted in a hollow needle 0·2 mm in diameter that was thrust directly through the body wall into the thoracic muscles. Reduction of thoracic temperature by conduction of heat into the thermocouple needle amounted to less than 0·1 °C, and was ignored.

All measurements were made in an air-conditioned room in which air temperatures were 22–24 °C and relative humidities measured by sling psychrometer ranged from 57 to 70%.

After metabolic measurements were completed, each animal was weighed, killed by injection of ethyl acetate and dissected. Total mass and thoracic mass were determined to the nearest 0·1 mg with an analytical balance. Wing length and span were measured to the nearest 0·1 mm. Wing area was measured to the nearest o-oi cm* by cutting off the wings, arranging them on a light-table in the configuration characteristic of flight, tracing their outline on graph paper, and counting the squares enclosed.

The wings of all animals were preserved to confirm identifications. Voucher specimens were preserved for most species. Identifications were made or confirmed by Julian P. Donahue of the Entomology Section of the Natural History Museum of Los Angeles County where all of the moths are deposited. A list of the species studied together with measurements of their oxygen consumption and aerodynamically important aspects of their morphology appear in Table 8.

Both sphingids and saturniids have heavy bodies and are conspicuously endothermic during flight, but they differ markedly in behaviour. Sphingids are adapted for high velocity flight, feed on nectar while hovering, and have a relatively prolonged post-larval life. Adult saturniids do not feed and have short life spans; presumably their flight functions primarily for locating members of the opposite sex and finding appropriate sites for oviposition.

The relationships of aerodynamically important morphometric and functional parameters to body mass in our samples of sphingids and saturniids (Tables 1 and 2) are summarized below.

1. In both families thoracic mass, wing length, wing span, and wing area increase exponentially with total body mass, whereas wing loading and aspect ratio have no significant correlation with body mass.

2. The mass of the thorax on average is relatively greater in sphingids than in saturniids.

3. Wing span is smaller and more closely correlated with body mass in sphingids than in saturniids.

4. Wing area in sphingids is less than half that in saturniids and is less variable, and wing loading (body weight per unit wing area) in sphingids is more than twice that in saturniids.

5. Aspect ratio (span² ÷wing area) averages about 60% higher in sphingids than in saturniids.

Since the thorax is almost completely filled with flight muscles, its mass can be used as an index to the mass of the flight muscles. In both saturniids and sphingids thoracic mass (and by inference, the mass of the flight muscles) scales with the 0 ·8 power of total body mass (Table 1). Consequently, wing length, wing span, and wing area are positively correlated with thoracic mass (Table 3) just as they are with body mass.

Rest. In the daytime, nocturnal moths remain quiescent and their body temperatures approximate to that of their surroundings. At air temperatures of 22 –24 °C the oxygen consumption of sphingids, saturniids and heterothermic moths of other families at rest increased exponentially with body mass (Fig. 1 and Table 8). The slope of the log-transformed values of on body mass for sphingids is slightly more than three-quarters whereas, that for saturniids is approximately two-thirds (Table 4). Both the slopes are significantly different from 1 ·0 (P < 0 ·01), but they do not differ significantly from each other.

The total amount of oxygen consumed during pre-flight warm-up increased exponentially with body mass (Fig. 2). Both the slope and the 1 g intercept of the regression of the log-transformed values of on mass were higher in sphingids than in saturniids (Table 4). Duration of warm-up was not correlated with body mass or thoracic mass.

increased exponentially with total body mass. During hovering flight the regression of the log-transformed values of on body mass for all moths we measured has a slope of 0-82 (Fig. 1). The slope of the regression for sphingids is higher than that for saturniids (0 ·77 v. 0 ·59). The same is true for the intercepts; during flight of a 1 g sphinx moth is 1 ·6 times that of a satuniid of the same size (Table 4). during flight also increases exponentially with thoracic mass; both slope and intercept are greater in sphingids than in saturniids. also increases with wing loading in both families, the exponent being 1·1 in sphingids and, 0·74 in saturniids. However, in both families the coefficients of determination relating ⁠, to wing loading are smaller than those for either total mass or thoracic mass (Table 5).

Wing beat frequency is much higher in sphingids than in saturniids (Table 2) and in both it is inversely related to wing length and wing span. However, it is not strongly correlated with body mass in either family (Table 6). The relation between wing beat frequency and area differs greatly in the two families. In sphingids, it is strongly inversely correlated with wing area (slope ≃ 0·4), whereas in saturniids the slope approximates zero. The slope of the regression of wing beat frequency on wing loading in sphingids approximates zero, whereas in saturniids the relationship between the two is highly variable. Wing beat frequency was more highly correlated with morphological parameters in sphingids than in saturniids.

Mean thoracic temperature measured after 3–5 min of hovering flight in a respirometer was 42·5 °C± 1·9 S.D. in sphingids (N = 27) and 39·031±2·3 in saturniids(N = 33). Both values are somewhat higher than those reported for these families under field conditions (Bartholomew & Heinrich, 1973). Since hovering flight is more energetically demanding than forward flight and is accompanied by less convective cooling, the temperatures reported here are probably higher than those which the same individuals would maintain during forward flight under natural conditions.

At rest, body temperature in heterothermic moths approximates to ambient temperature, and presumably all physiological systems are operating at low levels. During flight a large amount of mechanical work is done by the flight muscles, and metathoracic temperatures become elevated and are held near 40 °C over a wide range of ambient temperatures. Nearly all the oxygen consumed by a moth during flight is accounted for by the flight motor, but at rest the relative contribution of the flight motor to the total oxygen consumed is probably smaller. Nevertheless, the slopes of the log-log regressions of on body mass at rest and during flight are remarkably similar, 0·81 v. 0·77 in sphingids and o-68 v. 0-73 in saturniids (Table 4 and Fig. 1). However, the intercepts of the regressions for rest and flight differ greatly. A 1 g sphinx moth during flight consumes oxygen at about 170 times its resting rate, and a 1 g satumiid moth during flight consumes oxygen at about 125 times its resting rate.

The slope of the log-log regression of oxygen consumption at rest on body mass in our pooled sample of heterothermic moths (Fig. 1) is generally similar to that found in most groups of organisms (Hemmingsen, 1960), i.e. oxygen consumption increases with mass to approximately the three-quarter power. However, comparative data on the dependence of on body mass in adult insects are surprisingly fragmentary. The data for insects as a group (Kittel, 1941; Zeuthen, 1955; Enger & Savalov, 1958; Kiester & Buck, 1964) cannot be characterized by any single value for b in the allometric equation.

There is of course no a priori reason why animals with widely differing patterns of organization should have similar rates of metabolism just because they are the same size. However, a comparison of the allometric relations of oxygen consumption in insects with those in air-breathing vertebrates (Fig. 3 and Table 7) is of both physiological and evolutionary interest. The slope of the log-log regression of resting oxygen consumption of heterothermic moths on body mass in our sample is similar to those for resting reptiles, but the 1 g intercept is much higher for the moths; the value of a for the moths at 23 °C is almost 4 times that for reptiles at 20 °C and i’4 times that of reptiles at 30 °C. The same pattern is apparent when active states are compared. The slopes of the log-log regression of oxygen consumption on body mass are similar in flying moths and in reptiles active at 30 °C, but the value of a for the moths is more than 40 times that for the reptiles. It is particularly striking that the scaling of oxygen consumption of flying moths is almost identical to that for birds (and also bats) during flight. However, the factorial scope of the moths is at least 10 times that of birds. The difference, of course, is a function of the fact that insects are ectothermic when at rest while with few exceptions birds are not. However, this difference between the insects and the higher vertebrates largely disappears when heterothermic members of both groups are compared (Bartholomew & Casey, 1977).

The quantitative relations summarized above allow us to offer an interpretation of the differing patterns for energy utilization found in insects, reptiles, birds, and mammals. The pattern of energy utilization in each group is distinctive. It is a central component of their ecology and has been a major factor in their evolution and adaptive radiation.

Birds and mammals achieve high absolute rates of sustained aerobic activity by virtue of an elaborate circulatory-respiratory complex which requires a substantial and continuous energy expenditure to operate. The continuous operation of a high capacity supply system for aerobic metabolism is correlated with a high basal metabolic rate (BMR). BMR is defined as the metabolic rate of a fasting homeotherm at rest in thermal neutrality. Reptiles are poikilotherms. At rest their rates of energy metabolism vary directly with temperature and at all body temperatures are lower than those of homeotherms. Reptiles do not face the necessity of maintaining a high idling rate of resting metabolism, but they pay a price, they lack the capacity for sustained high levels of aerobic metabolism and must depend heavily on anaerobiosis during activity. Heterothermic insects combine resting levels of oxygen consumption in the reptilian range with the capacity for sustained aerobic metabolism during activity equal to that of flying birds and mammals. This virtuosity is physiologically mediated by the effectiveness of their system for direct delivery of oxygen to the flight motor without the intervention of the complications of a cardiovascular supply system.

Although several investigators have estimated the magnitude of the cost of warm-up from the specific heat of insect tissue, the rate and extent of temperature increase during warm-up, and the coefficient of heat transfer (see for example, Krogh & Zeuthen, 1941; Heinrich & Bartholomew, 1971; Heinrich, 1975), the only prior direct measurements of the energy cost of warm-up in moths are those of Heath & Adams (1967) on the sphingid, Celerio (Hyles) lineata.

From Fig. 2 it is apparent that the total amount of oxygen consumed during warm-up is strongly and positively correlated with size. In sphingids and saturniids we found that approximately 75 % of the variability in total oxygen consumption during warm-up can be accounted for by body mass (Table 4). Nevertheless, previous studies of pre-flight warm-up in moths have indicated either (1) that its rate and duration are not correlated with body size (Heinrich & Bartholomew, 1971 ; Heinrich & Casey, 1973; Bartholomew & Epting, 1975), or (2) that rate is positively, but very weakly, correlated with body size (May, 1976). May has argued on allometric grounds that specific heat loss of insects increases more rapidly with decreasing size than does specific heat production, and that, therefore, their rate of warm-up should increase with increasing size. The data presently available do not appear to us to allow either rejection or acceptance of this hypothesis. Measurements of total oxygen consumption during warm-up do not yield values for time rates of heat production during warm-up. Similarly, the mean rates of oxygen consumption (total O₂ ÷ duration of warm-up) cannot be used as a satisfactory estimate for an instantaneous rate of energy metabolism during warm-up because heat production changes continuously as a function of the continuously increasing thoracic temperature. Simultaneous measurements of thoracic temperature and oxygen consumption throughout warm-up of the sort made by Kammer & Heinrich (1974) on bumble-bees are obviously needed over a large size range. Until more data are available, we must accept the paradox that in heterothermic birds and mammals rate of warm-up is strongly and inversely correlated with mass (Heinrich & Bartholomew, 1971), but in insects the rate of warm-up is either independent of body mass or weakly but positively correlated with body mass.

In a flying moth the energy expended by the flight motor, and consequently the oxygen consumed, is a function of the wing beat frequency and the work per wing beat. Our data on sphingids indicate that during flight varies with mass^0·74 (Table 4), whereas their wing beat frequency varies with mass^−0·27. Therefore, the oxygen consumed per wing beat scales almost exactly with mass^1·0. These allometric relationships are the same as those proposed on general theoretical grounds by Calder (1974) who proposed (1) that frequency terms such as heart rate and breathing rate should scale with mass^−0·25, (2) that capacity terms such as stroke volume and tidal volume should scale with mass^1·0, and therefore, that flow rates such as cardiac output and minute volume scale with mass^0·75. Scaling arguments require geometric similarity. Sphingids are geometrically uniform, whereas saturniids show a wider variety of wing sizes and shapes (Table 1). In addition, our measurements of are based on hovering flight, and sphingids hover more readily than do saturniids. Because of these considerations, neither the lack of correlation between wing beat frequency and body mass in saturniids (Table 6) nor the departure of the slope of their from mass^0·75 is surprising.

This study was supported by a grant (DEB 75-14045) from the National Science Foundation. We are indebted to Kathleen K. Casey for technical assistance.