INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

MANY BIRDS SHOW MARKED SEASONAL and diurnal variations in body mass (8, 15, 23, 27, 29, 30, 43, 54, 56). For example, small passerines may have masses up to 29% greater in winter than in summer (29) and experience mass changes of 7-15% each day (15, 23, 30, 43, 56). Also, many migratory passerines accumulate over 50% additional mass before migration (8). The consequences of these dramatic mass changes for power requirements and behavior during flight are not yet clear (21). Accordingly, this study explores the trade offs involved in load carrying, in the context of short, energy-demanding flights by birds.

Aerodynamic theory suggests that any rise in body mass should increase the mechanical power required for flight, if wing dimensions and flight muscle mass remain constant (34, 39). Gessaman and Nagy (12) found mass loading to increase the metabolic cost of flight in homing pigeons, Columba livia, and the red knot, Calidris canutus, showed an increase in the metabolic cost of flight with increasing fuel load (21). Both the pigeon and red knot studies, however, were concerned with sustained cruising flight, which did not involve frequent takeoffs and landings. If mechanical power does not increase with larger loads or if the efficiency of conversion of metabolic power to mechanical power decreases, then a higher body mass may result in a decrease in flight performance. Such a decrease in performance should be evident in takeoff velocities and angles among birds performing short flights (37). In line with this, a mass-induced compromise in takeoff performance has been found in birds performing both routine flights (i.e., nonalarm) and alarm flights (22, 26, 30, 55). On the other hand, some studies have found no effects of mass on takeoff performance during either alarm or routine flights (19, 20, 49, 50) and this warrants an explanation.

Behavioral effects of load carrying on takeoff performance are usually interpreted as costs in terms of an increased predation risk (4), but there may be other disadvantages. Although birds are assumed to be using maximum power (both metabolic and mechanical) during alarm flights, none of the earlier studies of takeoff performance looked at the effects of mass change on energy expenditure. Possibly, the absence of an effect of mass loading on takeoff performance (19, 49) is due to an increasing metabolic power input.

During steady-state flight at minimum power speed, a 25% increase in body mass is predicted to increase the mechanical power (P_mech) required for flight by 45-50% (39). Although the effect of mass loading on pigeons was found to be less than predicted by aerodynamic theory (3, 21), nonetheless a substantial increase in P_mech with increasing mass loading was recorded. Assuming that the efficiency for the conversion of metabolic power to P_mech is not altered by the additional body mass, then metabolic power (P_met) and hence flight energy expenditure should be affected in a similar way.

Short flights differ from steady-state flights in that they are performed at air speeds below minimum power speed. At low air speeds, induced power is dominant, increases as the square of body mass, and is the component of flight power most affected by additional mass loading (34, 35, 39). Short flights should therefore be more affected by mass loading than steady-state flights. Any increase in the power demands of a short flight, therefore, which is already known to be an extremely demanding activity (36, 37), will have implications, particularly for small passerines that perform repeated short flights.

This study used the doubly-labeled water (DLW) technique to measure the effects of mass loading on short-flight energetics. Video was used in tandem with the DLW technique to determine any effect of mass loading on takeoff velocity and angle. Dorsal packs ("backpacks") were used to increase the mass of male zebra finches by 26-27%. The hypothesis under test was that the addition of mass would impact negatively on short flight performance, specifically by increasing the energetic cost of flight, decreasing takeoff performance, or both.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Training. Thirty male zebra finches were trained to fly back and forth between two automated perches positioned at either end of a 5.46-m-long plastic-walled flight aviary. Flights from one end of the aviary to the other were induced by the simultaneous withdrawal of one perch and presentation of the other (37). After 3 wk of training, all the finches were fitted with weighted backpacks (Figs. 1 and 2).

View larger version (119K): [in this window] [in a new window]   Fig. 1.   Photograph illustrating the position of the backpacks when mounted on the finches.

View larger version (94K): [in this window] [in a new window]   Fig. 2.   Photograph showing the construction of the backpacks.

Experimental protocol. On the morning of the experiment, six finches were randomly assigned to one of three treatments: control, empty backpack (manipulated controls), and loaded backpack (manipulated). Control finches had their backpacks removed. Manipulated controls had the weights removed from their packs and cotton wool inserted tightly into each straw. The cotton wool ensured that the drag caused by empty-packs (only 2.5 ± 0.1% of body mass) was equal to that caused by full packs. Manipulated finches had their loads (backpack + weights) adjusted so that they equaled 26% of their body mass; due to body mass loss during the experiment, this increased to a mean of 27.4 ± 0.4% (Table 1).

Data analysis. All data were normally distributed and consequently all statistical tests were parametric and were performed using MINITAB version 11.2. Means are displayed as ±SE unless otherwise stated. Due to the loss of some DLW samples during analysis and withdrawal of some finches midexperiment for welfare reasons (because apparently fatigued), final group sizes were controls, 8; manipulated-controls, 7; and manipulated, 9. To standardize with respect to body mass, energy expenditure has been expressed in multiples of basal metabolic rate (BMR) (37). To allow for wider comparisons, however, it has also been expressed in kilojoules.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Energy expenditure. An analysis of variance showed that energy expenditure did not significantly differ between treatments after controlling for proportion of time spent in flight (T_f; treatment, F_2,20 = 2.0, P = 0.161; T_f, F_1,20 = 4.05, P = 0.058). Means (×BMR) adjusted for variation due to T_f were 2.51 ± 0.11 (control), 2.72 ± 0.11 (manipulated control), and 2.44 ± 0.10 (manipulated). Similarly, total energy expenditure, when not adjusted for either body mass or T_f, did not differ between treatment groups (F_2,21 = 1.13, P = 0.341). The corresponding means (kJ) were 61.74 ± 3.0 (control), 64.53 ± 3.2 (manipulated control), and 58.13 ± 2.8 (manipulated).

Flight performance. A repeated-measures general linear model (GLM) showed takeoff velocity to vary between finches but observation time and flight direction to have no detectable effect on takeoff velocity (GLM: finch, F_23,78 = 2.57, P = 0.003; observation, F_1,78 = 0.59, P = 0.445; direction, F_1,78 = 3.44, P = 0.070). A similar analysis showed that takeoff angle did not differ between finches or with direction and did not vary with observation time (GLM: finch, F_23,78 = 0.93, P = 0.563; observation, F_1,78 = 0.23, P = 0.637; direction, F_1,78 = 0.01, P = 0.912). Because observation time and direction proved nonsignificant, they were removed from the statistical models and the resulting ANOVAs (with finch as the single factor) were used to calculate the repeatability of takeoff velocity (F_23,47 = 2.67, P = 0.002) and angle (F_23,47 = 0.8, P = 0.719). The repeatability of takeoff angle was extremely low (negative as MS_A < MS_W, where MS_A and MS_W refer to mean square among and within individuals, respectively). In contrast, the analysis suggested that within-finch takeoff velocity was highly repeatable (0.88). Flight duration also did not differ with observation time or flight direction, but was dependent on the finch performing the flight (GLM: finch, F_23,78 = 1.92, P = 0.018; observation, F_1,78 = 2.12, P = 0.127; direction, F_1,78 = 1.26, P = 0.264). Flight durations of individual finches also appeared to be very repeatable (0.84).

Body mass. The initial body masses of the three treatment groups did not differ significantly (F_2,21 = 0.80, P = 0.461). Means (g) were 13.69 ± 0.35 (control), 13.33 ± 0.45 (manipulated control), and 14.0 ± 0.32 (manipulated). In contrast, manipulated finches lost more body mass over the experimental period than either control or manipulated control finches. Again the P value obtained from an ANOVA using all three treatment groups was not quite significant at the 5% level (F_2,21 = 3.29, P = 0.057). Means (% of body mass) were 5.77 ± 0.56 (control), 5.87 ± 0.61 (manipulated control), and 8.00 ± 0.66 (manipulated). The power of the statistical test, however, was again very low (1   = 0.22), allowing that the effect may have been real. There was no difference in percent body mass lost by control and manipulated control finches (F_2,13 = 0.02, P = 0.903). Amalgamating the two control groups and comparing the joint mass loss among controls (5.82 ± 0.51) against that of manipulated finches (8.00 ± 0.66) suggested that manipulated finches did experience a significantly increased mass loss during the experiment (F_2,21 = 6.88, P = 0.016).

Mechanical power. Bird flight aerodynamic models have not considered an acceleration component (35, 39). Therefore predictions derived from them and applied to flight with an acceleration component (for example, takeoff) are at best an indication of relative P_mech values and may bear little relation to reality. An overall effect of the increased mass loss and reduced takeoff velocity, however, was assessed by estimating the instantaneous mechanical power output between the eighth and ninth video frames after takeoff (Table 4). Instantaneous energy gain (42) in joules was calculated from m · 1/2 (V²) + g z, where, m is mass, V is velocity, g is acceleration due to gravity, and z is height gained and the result multiplied by 25 (camera speed 25 Hz) to obtain a mechanical power estimate in watts (W). Height changes between the eighth and ninth video frames were very small and did not differ between treatment groups (F_2,21 = 2.35, P = 0.120). Means (m) were 0.002 ± 0.002 (control), 0.006 ± 0.002 (manipulated control), and 0.000 ± 0.002 (manipulated). Similarly, mechanical power was almost identical between treatment groups (F_2,21 = 0.02, P = 0.977). Means (W) were 2.19 ± 0.19 (control), 2.23 ± 0.19 (manipulated control), and 2.17 ± 0.18 (manipulated).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

No increase in energy expenditure of finches in response to carrying additional mass equal to 27% of body mass was found. Instead the effect of the mass burden was manifest as a 12% reduction in takeoff velocity and an apparent downward regulation of body mass during the experiment. Furthermore, mechanical power outputs did not differ between treatment groups. If mass-loaded finches had not experienced a reduction in flight velocity or an increased rate of mass loss, then the estimate of mechanical power would be 2.80 W (29% higher than that calculated with trade offs). The estimate of 2.80 W assumes that velocity and mass loss with no trade offs would be equal to that of control finches (where velocity = 3.6 m/s instead of 3.18 m/s and mass loss = 5.77% instead of 8.0%, giving a mean body mass of 13.60 g instead of 13.43 g). Hence, the additional loss of body mass and decrease in velocity during takeoff appears to offset the increased energy demand imposed by the mass loading. Of course, this assumes that the conversion of metabolic (energy expenditure) to mechanical power was similar in each treatment group. There is no obvious reason why finches carrying external loads are likely to suffer a reduced efficiency of conversion. More likely, an increase in efficiency would be appropriate to allow for loading (21), in which case metabolic power could be expected to remain low while relative mechanical power was increased. The lack of a measurable difference in energy expenditure is therefore likely to be more robust than the variation in the data would suggest, particularly as no trend toward greater energy expenditure in mass-loaded finches was apparent and feeding rate (an indirect indication of energy use) was not elevated. Because a compensatory mechanism for reducing energy demand (or maintaining energy expenditure at a given level) was identified, the inference of additional energy-saving mechanisms such as a reduction in RMR (5, 38) is not necessary. Even so, no difference in either the day- or nighttime RMR of control and mass-loaded finches was detected.

The results of this study appear to contradict previous studies with pigeons (12) and red knot (21), which found that mass loading increased the metabolic cost of flight. A direct comparison, however, is inappropriate. The pigeons and knot performed long-distance sustained cruising flights as opposed to the short flights performed by the finches in the present study. An upward takeoff is likely to be the most energetically costly component of a flight and therefore the range of metabolic power levels required to perform such a maneuver is narrow. In contrast, steady-state cruising flight can be achieved over a wider range of power levels usually corresponding to a change in flight velocity (33, 40). Thus the physiological restraints and behavioral options are different in each situation. Hence, the lack of an increase in energy expenditure during the short flights found here does not necessarily, and most likely does not, extrapolate to steady-state flight. Similarly, whether the reduction in flight velocity would have occurred had the finches reached cruising flight speed, as found in previous mass loading studies (12, 17, 40, 51), is not clear. A curious finding was that despite the mass-loaded finches having slower takeoffs, their flight durations were not longer than the control group. An increase in mass, however, increases wing loading and subsequently increases the rate of altitude loss during gliding flight (39). Therefore, when coming into land, a loaded finch will have to maintain flapping flight for longer. Perhaps the landing approach was made at higher speed, which offset the time lost during takeoff.

The observed decrease in velocity and the observed mass loss may be both a behavioral and physiological response. Presumably, full compensatory mass loss (i.e., equivalent to the mass of the backpacks) was not possible over the relatively short period of the experiment or perhaps not possible physiologically while maintaining flying ability. Nevertheless, although the mechanism was not apparent, body mass reduction may be an adaptive response to an increase in energy demand. In the absence of an increase in energy expenditure, the most plausible mechanisms of mass loss are likely to be a reduced food intake or dehydration. Finch-mediated anorexia (i.e., eating less) did not appear to occur here; feeding rates were similar among treatment groups (see above). Whether dehydration can explain the increase in mass loss is not clear. Uncertainties concerning final body water volumes in the calculations of water flux rates means that dehydration cannot be ruled out as the mechanism for body mass reduction. Similarly, feeding rates and digestive efficiency may vary independently, and therefore a reduction in caloric intake cannot be wholly disregarded as a cause of mass loss.

When given a weight burden, zebra finches during routine short flights either lack the power surplus required to accommodate it or choose not to expend more power. The fact that previous studies have found birds capable of increasing takeoff velocity when alarmed (49, 50) suggests that during routine flights birds choose to use the same power level. The power level may be determined by the upper limit of aerobic power available. Perhaps only when alarmed are finches prepared to use anaerobic power, which would fatigue the flight muscles and possibly result in a reduced flight performance if a predator attack occurred soon after the routine flight. Alternatively, the upper limit for power used during routine flights may be set by an increased risk of cell or tissue damage caused by either a muscle strain resulting from increased force production or perhaps a requirement for more physically demanding wing kinematics (14). To accommodate mass loading during takeoff, pigeon wing kinematics have been found to change (3) and birds are known to vary wing beat gait in response to changes in flight demands (46). Although the upper limit of aerobic power is set physiologically, the maintenance of metabolic power at an unloaded level when loaded is likely to represent a behavioral decision.

In agreement with aerodynamic theory (34, 39), pigeons have been shown to increase mechanical power output in response to mass loading (3, 6). The increases in power measured, however, were lower than predicted and unlike the present study were not proportional to mass loading. Metabolic power was not measured and again measurements were taken during steady-state level flight. Therefore the results are not necessarily applicable to short flights. Nevertheless, consistent with the present study, the pigeon data suggest a behavioral component in responses to mass loading. Mechanical force production by the pectoralis muscles increased with mass loading, suggesting that at lower mass loading the pigeons were using a submaximal power level. Moreover, large variations in the mechanical forces recorded for individual pigeons (3) indicate that response strategies to mass loading may differ markedly between individuals. Together, studies of mass loading during flight suggest that caution should be exercised when developing predictions of the effects of mass loading from aerodynamic models. Their application to many ecological situations is limited because they do not incorporate an acceleration component, a ceiling for available power, or knowledge of metabolic-mechanical conversion efficiency, nor do they allow for behavioral and physiological trade offs. The response to manipulations during flight will likely involve multi-level trade offs, and therefore outcomes will be considerably more complex than simple changes in mechanical flight performance. Indeed, the failure of at least two studies to find an effect of mass on mechanical takeoff performance (19, 49) may be because costs were experienced at a different level to that being investigated.

Our findings may have implications for foraging economics and foraging tactics of birds when short flights are common (for example, during the chick-rearing period). The mass loading used in this study was both artificial and substantial at 27% of body mass, and perhaps an "other-things-being-equal" scenario (i.e., increase in nonflight motor mass, without a concomitant increase in organ or flight apparatus mass) is likely to be rare in a wild situation. Some studies, however, both observational and experimental, have shown birds to reduce their body masses in response to increased predation pressure and have suggested it to be indicative of birds adaptively adjusting their body masses in relation to the perceived level of predation risk (11, 13, 25). Although the finches did not encounter a predator during the present study, they were captured at regular intervals throughout the training period and experiment. The capture event may have been perceived as a predation event and conceivably the finches' reduction in body mass was a response to an increased perception of predation risk. The reason why the body mass loss was greater in loaded finches than unloaded finches may have been due to the optimum body mass/physiological state being different for each group. Predation risk is thought to be traded off against future body reserves (11, 28, 30). Hence, the loaded finches may have responded more to the perceived risk and lost more mass, whereas controls remained at their optimum.

Under the conditions in this study the finches apparently regulated their body masses, and as a consequence of the weight burden suffered a reduction in takeoff velocity. The loss in body mass is likely to be peculiar to the methodology of adding mass artificially, but suggests that body mass can be regulated by the finches either behaviorally or physiologically. In contrast, although the level of metabolic power may be selected, the reduction in velocity is a consequence of physical constraints and is likely to be universal. In terms of costs, a reduction in velocity may increase "potential" predation risk.

Perspectives