Relationships Between Body Mass, Motor Output and Flight Variables During Free Flight of Juvenile and Mature Adult Locusts, Schistocerca Gregaria

During flapping flight of insects such as locusts, the functional task of the flight motor system is the generation of a motor pattern providing an aerodynamic output sufficient to support the body weight during flight. The pattern for alternating wing depression and elevation during flapping flight is generated by a neuronal oscillator (e.g. Wilson and Weis-Fogh, 1962; Delcomyn, 1980; Robertson and Pearson, 1985; Stevenson and Kutsch, 1987) that is controlled by sensory information from wing-related or head-located sense organs (e.g. Wendler, 1974; Gewecke, 1975; Möhl, 1985; Pearson and Wolf, 1987; Reye and Pearson, 1988; Wolf and Pearson, 1988; Fischer and Ebert, 1999).

In locusts, the hemi-metabolous development to the imago involves five non-flying larval stages and is characterised not only by an increase in body length but also by a large increase in body mass (Phipps, 1950; Cheu, 1952). After final ecdysis, the adult juvenile locust continues accumulating body mass up to maturation; however, there is no further increase in body size. Furthermore, the body mass of adult mature locusts exhibits considerable fluctuations, in particular for females after oviposition. However, as reported from field studies, locusts are capable of free flight from the second day of adult life and maintain this flight capability throughout the complete adult life span, apparently independent of age and body mass (Kennedy, 1952). This suggests that the motor output of the adult locust flight system is somehow adjusted to the developmental stage during adult ontogeny, i.e. to the required increase in lift production during flight that is demanded by an increasing body mass. In previous studies involving animals flying either tethered or suspended on a flight balance, changes in particular flight variables, such as wingbeat frequency and flight speed, were observed to change coincidentally with adult age (Kutsch, 1971, 1973; Gewecke and Kutsch, 1979; Kutsch and Gewecke, 1979).

There is, however, little information about (i) how the motor output of the locust flight system depends on body mass, (ii) which free-flight variables are influenced by the body mass and change accordingly as body mass increases, and (iii) how, in terms of behaviour, the locust flight system might ‘balance’ a large increase in body mass such as that observed during adult maturation.

To understand more about the functional implications of such morphological changes in relation to motor output and flight behaviour in maturing adult locusts, we investigated changes in flight variables in relation to body mass during adult maturation. We manipulated, reproducibly, the body mass of individuals and investigated the resulting changes in motor output and flight variables under free-flight conditions. Functional relationships between particular flight variables that changed during maturation were also investigated after partial wing reductions.

All experiments were carried out using male and female desert locusts Schistocerca gregaria Forskål. The animals were bred at 42 °C and 70 % relative humidity in crowded colonies of both sexes. Maturation occurred on average 18 days after final ecdysis (determined by observations of copulating pairs). Individual flight sequences were investigated in a room measuring 10 m×5.5 m×2.2 m at a temperature of 34 °C. The average illumination of the room was 4500 lx. After an initial warming to 40 °C, each animal was placed on a platform. For the majority of animals, flight activity started either with the animal jumping spontaneously or with a jump in response to hand-clapping or to a short puff of wind to the cerci. When jumping could not be elicited, flight was initiated by launching the animal vertically into the air (speed at release from the hands 2–3 m s⁻¹). To avoid any residual effects of the jump and the launch during the observations, the platform and the point of launch were 5 m in front of the observation area (see also Fischer and Kutsch, 1999).

For acquisition of electromyographs (EMGs) during free locust flight, one-channel radiotelemetry was used. The mass of the device was 0.3 g including the power supply. For a detailed description of the transmitter, attachment of the device to the animal and the receiver system used, see Fischer (1998) and Fischer and Ebert (1999). The metathoracic subalar wing depressor muscle (M129, muscle nomenclature according to Snodgrass, 1929) was recorded using low-resistance electrode wires (Manganin, Dillenburg, Germany; 50 µm in diameter).

The individual body length (in mm) was measured and the individual body mass m (in g) was determined before or after flight by weighing. The term m* indicates the body mass including any additional loading (see following section) and/or the mass of the transmitter. For each individual investigated, the wing area A_w (in cm²) was measured by weighing the clipped wings after the experiments. The remaining wing area, after partial reduction of the fore- and the hindwing as well as of both wings, was determined in a similar manner after the flight experiments. After determining the body mass of freshly killed individuals, the proportion of the flight muscle mass (as a percentage of the body mass) was determined by the mass difference between the isolated pterothorax, including the flight musculature, and the mass of the same thorax after maceration in KOH. For additional loading of the animals, small lead balls (Fischerei Zubehör, Germany) were flattened and fixed onto the pronotum of the experimental animal using beeswax.

The free-flight sequences of individual male and female locusts were recorded using a stationary video camera (Panasonic NV-MS4EG; frame frequency 50 Hz, shuttering time 1 ms). If telemetric EMG recording was performed synchronously, the EMGs were displayed on an oscilloscope (Tektronix D11) and stored on tape (Racal Store 4DS). The oscilloscope screen was filmed by a second video camera (Revue Video 8, frame frequency 50 Hz, frame interval 20 ms). The video outputs of both cameras were mixed online by a Panasonic digital mixer (WJ-AVE5) and recorded on a video master tape.

In a different approach, the flights of 12-day-old adult females were recorded by digital high-speed video (maximum frame frequency 500 Hz, shuttering time 750 µs: Weinberger Systems, Switzerland) for resolution of the instantaneous forewing angle of attack α (in degrees) in relation to the longitudinal body axis (in degrees). In accordance with Wolf (1990), the values obtained were transformed with reference to the parallax (for a description of the measurement errors, see Fischer, 1998).

The video recordings were screened on a 27 inch Sony Trinitron colour video monitor. Flight variables were acquired by frame-by-frame analysis (Panasonic digital video recorder AG7355) of the continuous displacement of the body silhouettes transferred to overhead transparencies (for methods, see Baker and Cooter, 1979; Baker et al., 1981). For a detailed evaluation of the measurement errors with respect to silhouette length and flight variables, see van der Wall (1996). All variables are considered as ‘instantaneous’ flight variables since they were constant within the given frame interval.

Data analysis and statistical procedures were computer-aided (software KaleidaGraph, StatView and Origin) and follow the criteria described by Sachs (1978). Correlation and linear regressions were performed at a significance level of P<0.05, with r indicating the linear coefficient of regression. The statistical significance of non-linear regressions of data is given by the coefficient of determination, r², at a significance level of P<0.05. Unless stated otherwise, results are presented as means ± S.D.

Ten male and ten female locusts from a colony cage with approximately 150 individuals of both sexes were individually labelled and kept isolated by a mesh inset cage. The body mass of each labelled individual was determined daily starting with the first day after the final ecdysis. Maturation was determined by the observation of copulating couples. In the adult females investigated, relative body mass increased on average by 208±11 % during maturation (N=10). For adult males, an average increase of 149±8 % was observed before maturity. In the animals tested, maturation occurred, on average, 18±1 days postmoult (N=10). In a different group of 25 males and 25 females, postmoult wing area did not change significantly in either sex (P>0.05, data not shown), indicating that there is a significant increase in wing loading L_w during the adult life span: for males, L_w increased from 5.1±1 to 7.4±0.6 N m⁻² (N=10); wing loading increased from 4.9±0.8 to 11.3±0.7 N m⁻² (P<0.05, N=10) in females. In general, L_w increased by 0.01 N m⁻² day⁻¹ in males and by 0.24 N m⁻² day⁻¹ in females (data not shown).

In a further set of experiments, 10 adult females and 10 adult males were separated from the colony on the first day after final ecdysis and kept isolated by a mesh cage inset. Starting 5 days postmoult, free-flight experiments were carried out at 5-day intervals, and the mean flight speed was determined. Before the experiments began, the body mass of each individual tested was measured, and the animals were returned to the cage after flight. The results are shown in Fig. 1A. In both sexes, flight speed was significantly correlated with body mass (P<0.05), both increasing over the adult life span. The mean flight speeds of males and females were not significantly different for the first 15 days postmoult (P>0.05); however, the mean flight speed of females was significantly higher after maturation than that of the males (P<0.05). The slopes of the regression lines in Fig. 1A indicate the mean increase in flight speed per gram body mass accumulated. There was no significant difference between the slopes (P>0.05): the increase in flight speed was 2.04 m s⁻¹ g⁻¹ body mass in males and 2.06 m s⁻¹ g⁻¹ in females. Is an increase in flight speed functionally required by an increase in body mass? To answer this question, we set up a different experiment, in which 15 adult females (16–18 days postmoult, mean body mass 2.8±0.6 g) were additionally loaded with flattened lead attached to the pronotum. For each animal, 5–7 flight sequences from two experimental flights were evaluated. In the first flight, seven animals started flying with the additional mass attached, and this was then removed for the second flight. The remaining eight animals underwent the same procedure in reverse. The results are shown in Fig. 1B. In 11 of the 15 (i.e. in 73 %) locusts tested, flight speed was significantly increased when flying under loaded conditions. The average increase in flight speed as a function of body mass was 0.0236 m s⁻¹ %⁻¹ mass increase (r=0.57, N=31, P<0.05), indicating that loading the females by 1 g would result in an increase in flight speed of 0.83 m s⁻¹.

In these experiments, 40 females and 40 males were separated from the colony immediately after the final ecdysis. At intervals of 5 days and starting on the fifth day after final ecdysis, the average wingbeat frequency f_WB was measured in eight males and eight females by radiotelemetric EMG recordings at 34 °C. For a representative 5-day-old and 20-day-old animal of either sex, the activity of the subalar depressor M129 during free flight is shown in Fig. 1C; the increase in wingbeat frequency observed over maturation is highlighted by a grey shaded area. The body mass m* (including the transmitter mass, which averaged 0.3 g) of each individual tested was determined before the experiment.

In both sexes, f_WB was significantly correlated with m* during maturation (P<0.05, Fig. 1D). f_WB in males was significantly higher than in females throughout maturation (P>0.05). The mean increase in f_WB was 4 Hz g⁻¹ body mass in females and 6 Hz g⁻¹ in males; however, the slopes of the regression lines were not significantly different (P>0.05). To investigate whether an increase in f_WB is functionally related to an increase in body mass, 12 adult females (16–18 days postmoult, mean body mass 2.9±0.7 g) were loaded at the pronotum in addition to the transmitter. For each animal, f_WB in 15–20 consecutive wingbeat cycles of 5–7 flight sequences from two flight attempts was evaluated. In the first attempt, six animals started flying with the additional mass attached; this was then removed for the second attempt. The remaining six animals were tested without additional load and were then loaded in the second set of experiments. The results are shown in Fig. 1E. In 25 %, i.e. in three of the 12 individuals tested, f_WB increased significantly (P<0.05); in another 25 %, f_WB decreased significantly after loading. However, in the remaining six animals, i.e. in 50 % of the locusts investigated, there was no significant change in f_WB (P>0.05). This indicates that, unlike flight speed, wingbeat frequency does not appear to be strongly affected by acute changes in body mass.

The relationships between the free-flight variables investigated are summarised in Table 1. In general, a wide range of ascent angles was observed for the individuals tested (0 °<χ<60 °). The distribution of χ was not significantly different either between the sexes or between the maturation stages (P<0.05). In the majority of animals investigated,wingbeat frequency and flight speed did not change significantly as a function of ascent angle (Table 1).

The non-linear characteristics of the maximum effective lifting power, P_EL,max, measured during adult maturation as a function of the ascent angle χ in 10 males and 10 females are shown in Fig. 2A,B. The maximum effective lifting power, P_EL,max, is the effective additional lifting power generated during the transition from horizontal (χ=0 °) to climbing (χ>0 °) flight. The increase in P_EL,max that occurs during maturation is not only a result of the increase in body mass but is also achieved by the maturational increases in flight speed.

The proportion of the flight muscle mass as a function of body mass during adult maturation is shown in Fig. 2C. For males, flight muscle proportion was maximal 5 days postmoult, while in females it was maximal at 10 days postmoult. During the first 15 days postmoult, flight muscle mass proportion did not differ significantly between males and females (P>0.05); however, after maturation the proportion of the flight muscle mass was significantly higher in males than in females (P<0.05, N=10).

The body angle φ as a function of the ascent angle χ was investigated in 10 males and 10 females during adult maturation. In both sexes, sampling was first carried out 5 days postmoult, and measurements were then performed at 5 day intervals at 34 °C to provide the results shown in Fig. 3. Fig. 3A illustrates locust body silhouettes at three different body angles to the horizontal (the drawing was reconstructed from video recordings). In general, body angle was correlated with ascent angle during climbing flight in each of the individuals investigated, irrespective of sex and stage of maturation (P<0.05; for mean relationships for the gender-specific groups see Table 1). However, in both males and females, mean φ decreased significantly with an increase in body mass during adult maturation (P<0.05, Fig. 3B), and the slopes of the regression lines obtained for the two sexes were not significantly different (P>0.05). A comparison between the body angles of 5-day and 20-day postmoult females as a function of ascent angle is given in Fig. 3C, which shows that mature females exhibit significantly lower body angles during flight than do immature females (P<0.05). A similar result was found for males (P<0.05, data not shown). This result led us to consider whether a decrease in body angle is functionally related to an increase in body mass.

To investigate this, the instantaneous body angle φ was determined as a function of the ascent angle in 10 female locusts (15 days postmoult). In five animals, loads that averaged 10 % of body mass (and represented the average increase in body mass occurring between 15 and 20 days postmoult) were attached on the pronotum after control flights. For the remaining five animals, the control flights were performed after flights under loaded conditions. The results showed that, in loaded animals, φ decreased significantly compared with that observed in control flights (Fig. 3D, systematic difference: Δφ=11.8±8 °, N=10, P<0.05) and that the decrease in φ was comparable with that observed in relation to the natural increase in body mass in females (see Fig. 3C).

Preliminary experiments revealed that, 18 days postmoult, females lost their free-flight capability (i.e. the ability to support their body weight during flight, as indicated by ascent angles below 0 °) after a reduction (i) of the forewing area by more than 20 % (N=15), (ii) of the hindwing area by more than 25 % (N=12), or (iii) of both forewing and hindwing area by more than 27 % of the gross area (N=10, data not shown). After control flights, in three groups each containing 10 female locusts (18 days postmoult), the gross area of the wing pairs was reduced by 13 % by reduction of (i) the forewing pair, (ii) the hindwing pair and (iii) both wing pairs in equal proportion (Fig. 4).

The body angle φ decreased slightly after forewing reduction (the intercepts of the regression lines were significantly different, P<0.05; Fig. 4A), but increased strongly after hindwing reduction (P<0.05; Fig. 4B). Compared with intact flights, φ increased during shallow ascending flight (0 °:s;χ:s;10 °) but decreased during steep ascending flight (χ>10 °) following a 13 % reduction of both wing pairs in equal proportions (Fig. 4C). Trimming of the wing pairs also influenced the maximum ascent angles observed: during intact flight, ascent angles of 30–40 ° were observed in all three groups; however, these did not occur after reduction of the forewings or both wings (as indicated by the grey areas in Fig. 4A,C). Under both experimental conditions, the distribution of the observed ascent angle was shifted significantly towards lower values compared with the intact condition (P<0.05, N=10). A reduction of the hindwing area, however, had no significant effect on the maximum ascent angle achieved (Fig. 4B, observed distributions of χ are not significantly different, P>0.05, N=10).

The effects of wing reduction on the motor output of the adult locust flight system was investigated in three groups of animals each containing 12 adult females (18 days postmoult) flying at 34 °C. Any reduction in the wing area increases the wing loading L_w. Since the animals investigated in this particular set of experiments performed flight sequences with the additional mass of the transmitter for telemetric acquisition of f_WB, reductions in the gross wing area were limited to, on average, 10 % (L_w was increased on average by 8 % after transmitter attachment). All animals tested maintained the capability for free flight. As shown in Fig. 5Ai, f_WB was not significantly changed in the majority of animals following a reduction in forewing area (for intact locusts, mean f_WB was 24.8±0.8 Hz, N=12). In the remaining locusts, mean f_WB was either significantly increased or decreased (in general, these changes were 5–12 % of the intact value) compared with the intact situation (P<0.05). The effects of forewing reduction on f_WB resembled those observed after additional loading of the animals at the pronotum (see Fig. 1D). Hindwing trimming (Fig. 5Aii) did not affect f_WB in 40 % of the animals tested (mean f_WB was 25.2±1.0 Hz, N=12 in intact females). However, f_WB was significantly changed in 60 % of the locusts investigated (relative changes of 6–11 % of the intact value). Wing area reduction spread over both wings changed the mean f_WB in 75 % of the animals (Fig. 5Aiii, intact situation, mean f_WB=24.2±0.88 Hz, N=12; changes observed in the range 5–14 %).

In a different set of experiments, the effects of wing reduction on flight speed were investigated. The wing area was reduced in three groups of 10 adult females (18 days postmoult, without transmitters) to the limits required for support of body weight during free flight (for details, see above). All females tested maintained their free-flight capability throughout the experiments. As shown in Fig. 5B, a reduction in wing area decreases the flight speed in all groups on average from 4.34±0.38 m s⁻¹ (N=30) to 3.09±0.47 m s⁻¹ (N=30), representing 71.2 % of the intact value, irrespective of which wing pair was trimmed (values from the three groups were not significantly different from one another, P<0.05).

Free-flight sequences of juvenile adult females (12 days postmoult) were recorded using high-speed video. The stroke plane inclination (shown schematically in Fig. 6A,B) as a function of longitudinal body axis (Θ_BL) did not significantly change (P>0.05, data not shown) when the animals performed climbing flight, i.e. increased χ (P>0.05; forewing, Θ_BL=21±9 °, N=27; hindwing, Θ_BL=12±11 °, N=23; data pooled from six animals). However, stroke plane inclination to the vertical (Θ_v) changed because the body angle increased during ascending flight (Fig. 6C).

For the forewing, the angle of attack of the wing profile (α, with the longitudinal body axis used as reference) during the wingbeat cycle was determined in four animals performing horizontal flight sequences (i.e. χ≈0 °) and in one particular flight with a strong ascending flight path (39 °≥:χ≥:48 °; shown schematically in Fig. 7A,B). A value of α<0 ° indicates wing pronation (i.e. depression of the wind leading edge), whereas α>0 ° indicates supination of the wing. Determination of a ‘common’ angle of attack in the hindwing was not possible since the hindwing plane, in particular the anal part, undergoes considerable deformation during the wing stroke. The results from a representative animal are shown in Fig. 7C. During the last two-thirds of the downstroke, values of α were almost steady for both horizontal and climbing flight. However, values of α during the last two-thirds of the downstroke during climbing were significantly lower (P<0.05; mean α=−54±12 °, N=8) than during horizontal flight (mean α=−32±14 °, N=5, data pooled from one of the animals tested), indicating a higher degree of pronation during climbing (these results were confirmed in three of four animals).

An understanding of the relationships between wing kinematic variables and the aerodynamic forces produced by wing movements has been provided by the application of unsteady aerodynamic calculations to locust flight (Send, 1992, 1994; Zarnack and Send, 1994; for an overview, see Zarnack, 1997). Investigations using locusts flying on flight balances revealed (i) that the functional task of the complex three-dimensional wing movement, i.e. alternating up and down movement, the time course of the wing angle of attack during the cycle and the pro- and remotion of the wing along the longitudinal body axis, is the generation of thrust not lift, (ii) that the generation of lift during the wing beat relies mainly on the steady angle of attack and on the speed of the air stream faced by the wings, and (iii) that stroke amplitude and wingbeat frequency have only a weak, if any, influence on the basic lift production required for stable level flight.

Emerging from these results, the locust flight system might theoretically control lift generation during flight in two main ways: (i) by changing the flight speed, and (ii) by changing the wing angle of attack during the wingbeat cycle.

In both sexes of locust, we observed that the increase in body mass during postmoult maturation was correlated with an increase in free-flight speed (Fig. 1A). In addition, mean flight speed of mature females increased after an acute addition of body mass (Fig. 1B). Even though we cannot provide evidence that this addition of small mass units adequately addresses the gradual increase in mass associated with maturation, our data support the hypothesis that the increased lift production required by maturational increases in body mass is achieved functionally by the observed increase in mean flight speed. Since both the accumulation of body mass in both sexes during maturation and the acute addition of mass led to an increase in wing loading, flight speed also increased in proportion to relative wing loading (data not shown). This matches the finding that wing loading is positively correlated with flight speed in various other species (e.g. Dudley and Srygley, 1994).

As reported from tethered flight studies in mature locusts, the minimum flight speed required for continuous support of the body weight was approximately 3 m s⁻¹ (Weis-Fogh, 1956; Zarnack and Wortmann, 1989). In accordance with this result, freely flying mature locusts that were artificially ‘braked’ by clipping their wings to reduce thrust production lost their capability for flight when flight speed was less than 3 m s⁻¹ irrespective of which wing pair was clipped (Fig. 5).

In addition to supporting the increasing body mass during maturation, additional lift (i.e. additional effective lifting power, Fig. 2; see also Fischer and Ebert, 1999) is generated when the locust changes from horizontal to climbing flight. Consistently, 37 % of the animals tested in this study increased their instantaneous flight speed when changing to climbing flight. In 47 % of the animals investigated, however, instantaneous flight speed did not change during climbing (data from Table 1). This indicates that the control of the instantaneous flight speed might be only one mechanism used by the locust flight system to regulate additional lift production during climbing, since stroke amplitude and/or downstroke duration in particular wings were also observed to change during ascending flight (Fischer and Kutsch, 1999).

In the field, flight activity changes throughout postmoult maturation (Kennedy, 1952): while freshly moulted animals did not fly, flight activity starts 2–3 days postmoult and is maximal between 5 and 12 days postmoult when the animals become ‘fully migrant’ (Weis-Fogh, 1956). This is coincident with the maximal postmoult mass proportion of flight muscles in both sexes (Fig. 2). Flight speed did not differ between the sexes in the fully migrant phase, i.e. when the animals were between 5 days and 15 days postmoult (Fig. 1).

During free flight in maturing adult locusts, the mean wingbeat frequency f_WB increased as a function of adult age and, thus, coincided with the observed increase in average flight speed and body mass (Fig. 1). However, unlike flight speed, instantaneous f_WB did not change significantly as a function of an acute application of mass (Fig. 1E) or as a function of climbing (in 60 % of the locusts tested, f_WB was not significantly correlated with the ascent angle, irrespective of sex and maturation stage; data from Table 1). This indicates that additional lift generation either in response to an acute addition of mass or for climbing flight is, apparently, not achieved by an increase in instantaneous f_WB. Furthermore, in 59 % of the locusts investigated, irrespective of sex and maturational stage, instantaneous free-flight speed was not correlated with instantaneous f_WB (data from Table 1). This constant f_WB, which has been observed during a wide range of flight speeds in insects as well as in birds, was previously explained as the characteristic f_WB of particular flight systems, which were assumed to be driven damped oscillators in which the characteristic f_WB (or ‘resonance frequency’) depended both on the driving frequency and on the mechanical damping properties of the system (Soltavolta, 1952; Greenwald, 1960). Recently, it has been shown that detached wings of the tobacco hawkmoth Manduca sexta caused to ‘beat’ by a piezoelectric device produced maximal aerodynamic output when driven in resonance (Scharstein, 1998a,b). In the mechanical, highly condensed thoraces of more recent ‘myogenic’ flyers (e.g. dipterans), which exhibit much higher values of f_WB than locusts, wing clipping increased f_WB (Soltavolta, 1952). This indicates that, in a myogenic flight oscillator, the characteristic f_WB may depend mainly on the damping properties of the system, i.e. on the wing mass and thoracic internal and external frictional forces that determine the ‘resonance conditions’ and, thus, f_WB.

In evolutionarily earlier ‘neurogenic’ flyers, however, wing clipping had little effect on the characteristic f_WB (e.g. in cockroaches, Roeder, 1951; Soltavolta, 1952) or led to increased variability (for locusts, see Fig. 5). These effects of wing reduction experiments may indicate that the characteristic f_WB in neurogenic flyers such as locusts depends mainly on the driving frequency of the flight oscillator, i.e. on the instantaneous frequency output generated by the neuronal flight network (see Robertson and Pearson, 1985). During postmoult maturation of locusts, specific sensorimotor pathways that participate in the generation of a functional flight motor pattern are reorganised following the first 2 weeks of imaginal ecdysis (Altman et al., 1978; Gray and Robertson, 1996) with respect to the flight interneurons that grow during maturation (Gee and Robertson, 1994). These network-related changes, together with apparent changes in the driving capacity of the flight muscles during maturation (Brösemer et al., 1963), support the hypothesis that the ‘proximate’ reason for the increase in characteristic f_WB during maturation in locusts is a change in driving frequency, i.e. in the resonance conditions of the maturing flight system.

Although f_WB seems not to be directly involved in the instantaneous control of the flight speed responsible for additional lift production (e.g. for climbing or in response to acute addition of mass), we cannot conclude from our experiments that f_WB, in general, plays no part in the propulsive mechanism of the flight system that provides basic aerodynamic output enabling flight capability throughout maturation. Temperature experiments in mature locusts have shown that, in general, a minimum wingbeat frequency is required for the support of the body mass during free flight (Fischer and Ebert, 1999).

The fact that body angle and ascent angle are strongly correlated suggests a significant role for the body angle during lift production (see also Baker et al., 1981). Studies in locusts flying in a flight balance revealed that changes in body angle reproducibly manipulate lift production and that, during ‘steady’ flight, maximum lift is obtained at a mean body angle of 20 ° (Zarnack and Wortmann, 1989), which matches body angles observed during horizontal free flight (Fig. 3). In the majority of locusts investigated, the body angle was correlated neither with flight speed nor with wingbeat frequency (Table 1).

The mean body angle changes in relation to the increase in body mass that occurs during adult maturation in both sexes (Fig. 3). These changes are more obvious in females, in which a greater postmoult increase in body mass was observed than in males; the relative change in the average body angle was comparable in both sexes. A preliminary study has shown that the location of the centre of gravity of female locusts changed in relation to the longitudinal body axis during the adult life span (probably as the ovaries, which are located in the abdomen, matured). A slight caudal shift in the centre of gravity by, on average, 4 % of relative body length was found when comparing 5-day and 20-day postmoult females (Fischer, 1994). It is not clear whether this slight shift in the centre of gravity could be co-responsible for the considerable decrease in body angle during maturation.

A large decrease in body angle was also observed after acute addition of mass at the pronotum (Fig. 3D). We do not believe that this decrease in body angle is simply the result of placing the mass anterior to the centre of gravity and causing the locust to pitch forward. In these particular experiments, the total mass applied was limited to 10 % of the body mass, an amount that did not significantly influence the location of the centre of gravity (Fischer, 1994).

Our experiments suggest that the instantaneous body angle during flight is largely controlled by the aerodynamic force production of the wings rather than by changes in leg position or in abdominal posture. Partial clipping of the forewings decreased the body angle slightly during free flight, while the instantaneous body angle was strongly increased after hindwing reduction (Fig. 4). This also indicates that both wing pairs may contribute to the regulation of body angle during the adult life span in an antagonistic manner. Forewing clipping not only decreased the instantaneous body angle but also limited the range of ascent angles to lower values. This effect was not, however, observed after hindwing reduction. This further supports the idea that increases in body angle play a significant role during climbing flight. When wing area was reduced in both wing pairs, body angle adjustment seemed to be impaired; and this had a general effect on the ability of the insect to perform climbing flight.

Changes in body angle during flight strongly influenced the stroke plane inclination of both wings (Fig. 6); however, in general, stroke plane inclination did not change in relation to longitudinal body axis (see also Gewecke and Kutsch, 1979). The main difference between the time course of the wing angle of attack during horizontal and steep ascending flight was an increase in wing pronation during the downstroke in climbing flight (Fig. 7). It is not clear whether these changes in the angle of attack result in increased thrust production (i.e. flight speed) and, thus, in increased lift production during flight.

The relationship between stroke plane and body axis and, to some extent, the basic time course of the wing angle of attack during the wingbeat cycle are mainly determined (and limited) by wing joint mechanics (Pfau, 1977, 1978, 1982). This suggests that body angle adjustment during locust flight plays an important part in the directional control of the resultant flight force vector, which is directed towards the vertical during climbing. Whether the steeper body angle exhibited by very young adults during flight enables the juvenile immature flight system to provide a resultant flight force vector that is directed more vertically (e.g. for additional support of body mass) is a question that needs to be addressed by further experimentation.

We are grateful to Drs H. Scharstein, A. Büschges and G. Wendler (Universität zu Köln, Germany) and Mr S. Merrywest (University of St Andrews, Scotland) for their valuable comments, helpful advice and discussion. We wish to thank two anonymous referees for their constructive criticism and valuable suggestions. Thanks are especially due to Ms B. Janisch and Mr E. Ebert (Universität Konstanz, Germany) for their assistance. This project was supported by a DFG grant to W.K. (Ku 240/17-1, 2). H.F. is currently supported by The Wellcome Trust.