The locust tegula: kinematic parameters and activity pattern during the wing stroke

In most animals, the generation of functional motor patterns for the repetitive movement of the body, limbs or wings during locomotion relies both on pattern generators in the central nervous system and on peripheral sensors,which provide feedback to the locomotor control circuitry (for a review, see Büschges and El Manira,1998). In locust flight, for example, the proprioceptive sense organs associated with the wings, such as stretch receptors, campaniform sensilla and tegulae, have been shown to contribute to the generation and modulation of the flight motor command(Gettrup, 1966; Wendler, 1974; Möhl, 1985; Wolf and Pearson, 1987b, 1988; Reye and Pearson, 1988). The locust tegulae, knob-shaped sense organs at the anterior base of the wings(Kniazeva, 1970), have been investigated in particular detail, mostly with regard to the sensorimotor pathways involved in flight pattern generation. During flight, tegula input contributes mainly to the timing of wing elevation(Wolf and Pearson, 1988; Pearson and Wolf, 1989; Wolf, 1993), while nerve recordings indicate that the tegula organs are excited during the (preceding)downstroke of the wing (Neumann,1985; Wolf and Pearson,1988; Fischer and Ebert,1999).

Contrasting with the rather detailed information available about tegula sensorimotor pathways, little is known about the functional morphology of the tegula organs or their mode of activation. The tegula is composed of two types of mechanosensor, an external hair plate located on the cupola, which contains approximately 40 sensory hairs, and a chordotonal organ inside the cupola,which consists of approximately 30 scolopidial sensilla(Kutsch et al., 1980). While single (filiform) hairs are typical exteroreceptors responsive to the degree and direction of hair bending (for reviews, see Thurm, 1982, 1984), hair plates (e.g. Kent and Griffin, 1990; Mücke, 1991; Newland et al., 1995) or hair rows often subserve a proprioceptive function, for example in monitoring joint position and movement (e.g. Wong and Pearson, 1976; Pflüger et al., 1981; Dean and Wendler,1983; Bässler,1983). This is achieved through the successive deflection of adjacent hairs by skeletal elements such as neighbouring limb segments. Chordotonal organs (for a review, see Matheson, 1990) are typical proprioceptors responsive to the (relative) position and movement of skeletal elements, including acceleration and vibration (e.g. Zill, 1985; Hofmann and Koch, 1985; Kittmann and Schmitz, 1992;also in tympanal organs, Yack and Fullard,1993). In addition, chordotonal organs are employed to monitor the position and movement of an appendage in the context of motor control (e.g. Burns, 1974; Field and Pflüger, 1989; Matheson and Field, 1995; Büschges, 1994).

The tegula would be equipped to encode almost every parameter of the locust wing stroke important for aerodynamic force production, flight control and steering (amplitude and angular velocity of the wing stroke, e.g. Lehmann and Dickinson, 1998; timing of the stroke reversals, e.g. Dickinson et al., 1999). The tegula is also involved in phase-tuning muscle activity during the wingbeat cycle, particularly regarding the wing elevators(Wolf and Pearson, 1988; Pearson and Wolf, 1989; Wolf, 1993, Fischer and Ebert, 1999). Since the activation phase is one of the key features controlling the mechanical output of synchronous oscillatory insect muscles (e.g. Josephson, 1985; Stevenson and Josephson,1990), this would provide a direct functional context for wingbeat-synchronous mechanosensory pathways, such as that of the tegula, in flight pattern generation.

In the present study, we examined the functional morphology of the tegula organs, their timing and activity pattern during the wing stroke and possible stroke parameters encoded by the tegula. In a videographic analysis, the relationship between wing movement and tegula kinematics (including the kinematics of selected wing hinge elements) was examined, and electrophysiological and wing movement recordings were combined to analyse the relationship between tegula discharge and wing stroke parameters. The data suggest that the tegula does not just signal the downstroke movement, but rather monitors details of stroke timing and the angular velocity of the wing.

Adult female locusts (Locusta migratoria L.) were used for all experiments 6-12 days after the imaginal moult. Experiments were performed at 25-30 °C. The animals were attached to a holder by the ventral meso- and metathoracic sterna using beeswax resin. The tarsi of all legs were removed to avoid termination of flight episodes by tarsal contact. Flight was induced by short wind puffs onto the frons or cerci.

For high-speed video analysis, the tegula organs (encircled in Fig. 1A,B, which shows electron microscographs of the location of the tegula organs, the main structural components of the wing base and the pterothorax) were marked with circular dots of black ink (Texpen, USA, marked with open arrows in Fig. 1D) under a dissection microscope as a reference point for the examination of kinematic parameters. Ink dots were placed centrally on the organs, without covering the posterior hair fields of the tegulae or touching other structures of the wing base. To allow video recording of the forewing tegula, the posterior edge of the pronotum had to be clipped without damaging the ligament between the pro- and mesothorax.

For electrophysiological experiments, a flight preparation was used(Wolf and Pearson, 1987a) in which the animal was glued to the holder in an inverted position. A flap of the ventral cuticle was removed to provide access to either the mesothoracic or the metathoracic ganglion and to the proximal segments of nerve 1 (N1;nomenclature after Campbell,1961).

Aspects of functional morphology were studied either in freshly killed animals or in isolated pterothoraces, macerated in concentrated KOH, to investigate cuticular anatomy (see Pfau and Koch, 1994). Anatomical descriptions are based on Albrecht(1953).

A commercially available digital high-speed video system (HSVS; hardware:Weinberger Systems, Switzerland; software; Speedcam, Fraunhofer Institute,Erlangen, Germany) was used which allows synchronous recording by two separate cameras (frame frequency adjustable between 1 and 1000 frames s^-1). Tegula movement was recorded from the dorsal side with one camera (Sigma macrophoto lens, f=90 mm, Fig. 1D); the other camera was equipped with a zoom telephoto lens(Cosimar, f=1.4-50 mm) and recorded the stroke movements of the fore-and hindwings from a lateral view (not shown). The frames of both cameras were system-internally synchronised during recording. Between the two cameras,frames corresponding in time were identifiable by the displayed frame numbers. Recordings were stored on-line on computer disc. For analysis, the digitally recorded episodes of both cameras were transferred onto VHS videotape. In each individual, the lengths of the fore- and the hindwings (base to tip) were measured, and the dimensions of wing hinge components and tegula organs were determined with an ocular micrometer after the experiments.

Flight motor activity was monitored by bipolar electromyographic (EMG)electrodes (30 μm stainless-steel pins) from the first basalar depressor(forewing, M97; hindwing, M127) and a tergosternal elevator (M83/84 and M113,respectively; nomenclature according to Snodgrass, 1929)(Fig. 2Ai,Bi). To record tegula activity, bipolar hook electrodes were placed either on nerve N1 or on nerve branch N1C (nomenclature after Campbell,1961), which contains the afferent axons from the wing base. The recording site was isolated with silicone grease. In this experimental arrangement, wing position during flight was recorded by an optical position detector (von Helversen and Elsner,1977) (Fig. 2Ai,Bi)in parallel with the extracellular nerve recordings. For each animal, the detector was calibrated by positioning the wing passively at given stroke angles after the experiment. In addition, after the completion of electrophysiological recordings, the recorded tegula organ was severed as a control (Fig. 2Aii,Bii). The data were stored on compact disc (CD recorder, Pioneer PDR 04) and transferred onto a computer hard disc using an analog/digital converter (Biologic DRA-800).

The VHS recordings were monitored on a 27 inch Sony Trinitron colour video monitor. To measure wing stroke parameters, recordings by the lateral camera were screened frame by frame, and the instantaneous stroke position and stroke deviation (see Fig. 1C for explanation) were transferred to overhead transparencies for further analysis(see also Baker and Cooter,1979). The distance between the upper and lower reversal points of the wing was measured, and the total (peak-to-peak) stroke amplitude (Φ,see also Sane and Dickinson,2001) was calculated from these data as a cosine function of wing length (base to tip; see Fischer and Kutsch, 1999).

The pterothorax (i.e. the fused meso- and metathorax with fused sterna and pleura, which is further stabilised by several sternal and pleural apostemata;e.g. Albrecht, 1953) was studied after maceration by applying mechanical stress, which revealed rigidity along the longitudinal axis. Experiments in which the thorax was filmed during flight showed that, by using a ventral attachment of the locust by both meso- and metathoracic sterna (e.g. Zarnack and Wortmann, 1989),flight activity did not result in any longitudinal (i.e. foreward and backward) movements or lateral displacements of the pterothorax relative to the tether. The tether or the margin of the video frames was therefore used as a reference for measuring movements of the tegula and other skeletal elements of the wing hinge. In contrast, suspension of the animals by the pronotum(e.g. Dugard, 1967; Baker, 1979) resulted in strong oscillatory displacement of the body relative to the tether during flight and,thus, prevented accurate focusing on the tegulae and other wing structures.

The degree of tegula rotation in the wingbeat cycle was estimated from the transparencies, according to Fig. 1Diii, with the orientation of the ink dot at the upper reversal point of the wing serving as a reference. Initially, this procedure was tested using a Styrofoam sphere marked with a circular ink dot and rotated through known angles while being filmed from the same view as the tegula during experiments. The degree of tegula inclination (i.e. the inclination of its longitudinal axis) was estimated by measuring the relative changes in the visible area of the ink dot using the area of the dot at the upper reversal point of the wing as a reference. For individual calibration, the wing was positioned at given angles in quiescent locusts before flight experiments.

The evaluated parameters of wing stroke and tegula activity are explained in the legend to Fig. 2. The mean amplitude of the tegula discharge was calculated as the integral of the rectified tegula burst divided by burst duration (e.g. Chau et al., 1998). The data were analysed using the Spike 2 data software package (Cambridge Electronics,UK) and the Data View signal-analysis program (W. J. Heitler, University of St Andrews, UK).

Statistical analyses were computer-aided (KaleidaGraph, MS Excel, StatView)and followed the criteria described by Sachs(1978). Correlation and linear regression analyses were tested for significance levels of P<0.05,with r indicating the linear correlation coefficient. Partial correlation coefficients (r^*) were determined according to Sachs (1978). The statistical significance of non-linear regressions of data is given by the coefficient of determination, r². Mean phase values are given as ϕ± mean angular deviation, with r describing the mean vector. Circular two-sample comparison was performed using the Watson—Williams test (Batschelet, 1981). Unless stated otherwise, data are given as mean ± S.D.

The results described in the following two sections were obtained by high-speed video recordings (500 frames s^-1) from 10 different animals, five investigating the forewing and five the hindwing structures. Three to five flight episodes were evaluated per animal.

The cupola of the tegula organ is integrated into a common ligament attached to the scutum, basalar sclerite, pleura and leading edge of the wing. This is illustrated for the forewing in Fig. 3A. The schematic forewing diagram in Fig. 3B shows that,during the downstroke, the scuta of the wing segments moved dorsally but were also displaced posteriorly along the body axis. During this posteriorly directed movement of the scutum, the wing was promoted, i.e. shifted in the anterior direction (anterior stroke deviation, see Fig. 1C). The changes in stroke position and stroke deviation of the wing were strictly phase-coupled during the wingbeat cycle (Fig. 3C; at the upper reversal point, the wing tip has reached its posterior extreme position; the anterior extreme position is reached when the wing passes through the lower reversal point). This appears to be due to the tight mechanical coupling between most elements of the wing hinge(Pfau, 1982). The pterothoracic scuta are connected by an elastic ligament. Thus, during the downstroke, the two scuta moved posteriorly at slightly different times;during the upstroke, their anterior-directed movements were almost synchronous(and in phase with the hindwing upstroke, shaded area in Fig. 3D). Furthermore, both scuta also underwent a vertical displacement during the wingbeat cycle,roughly in anti-phase to the wing movement, because of their location on the inner side of the wing hinge. The scuta were displaced dorsally during the downstroke and returned ventrally during the upstroke. We were, however,unable to quantify this vertical movement because both scuta showed considerable dorso-ventral deformation superimposed on their vertical movements, which appeared to be caused mainly by the contraction of the dorsal longitudinal muscles.

In the video recordings taken from the dorsal side, the mesothoracic first basalar sclerite performed rotational, rather than horizontal, movements during the wingbeat cycle (determined via its horn-shaped anterior process, 1ba in Fig. 1; shown schematically in Fig. 3B using the orientation of the process at the upper reversal point of the wing as a reference). This is probably (i) because the first basalar depressor muscle attaches to the posterior part of the sclerite(Albrecht, 1953) and (ii)because the sclerite itself is attached to the scutum by the medial and anterior edges of the common ligament. The posteriorly and upward-directed components of the scutum movement, together with the contraction of the first basalar muscle during the downstroke, are thus transformed into an `inward rotation' of the first basalar sclerite. This change in orientation was slightly phase-shifted with respect to the stroke position of the forewing[Fig. 3C, advanced by a mean phase (ϕ) of 0.18±0.03, r=0.967, N=45, data pooled from five animals] but occurred almost in synchrony with the horizontal movement component of the scutum (ϕ=0.02±0.02, r=0.945, N=45). The first basalar sclerite of the hindwing could not be investigated because it is located below the plane of the hindwing and was not visible in the video recordings.

The tegula organ followed a complex three-dimensional trajectory during the wingbeat cycle in both the fore- and hindwings. During the downstroke, the longitudinal axis of the oval-shaped tegula was inclined horizontally. This is illustrated schematically for the forewing in Fig. 4A (the longitudinal axis of the organ is indicated by a dashed line). At the same time, the organ was rotated around its longitudinal axis (anti-clockwise on the animal's right-hand side, i.e. anterior margin upwards; indicated by the elliptical red arrow in Fig. 4A). During the upstroke, this movement was reversed; the longitudinal axis of the organ moved vertically, with a synchronous downward rotation of the anterior tegula margin.

The temporal pattern of tegula inclination and rotation, with respect to stroke position and deviation movements, is shown for four consecutive wingbeat cycles in Fig. 4B(forewing parameters in Fig. 4Bi and hindwing parameters in Fig. 4Bii; both panels represent typical experimental animals, data confirmed in all animals studied). Movements of the wing and tegula organ exhibited a stable phase relationship, with the tegula reaching maximum rotation (and minimum inclination) near the lower reversal point of the wing beat (right in Fig. 4A) and vice versa near the upper reversal point (left in Fig. 4A).

The mean values of inclination and rotation were lower in the forewing than in the hindwing organs (P<0.05, data not shown). For both tegulae,total inclination and rotation movements during a wingbeat cycle (determined as peak-to-peak values, Fig. 4B) were significantly correlated with wing stroke amplitude in all animals investigated (0.66<r<0.73, P<0.05, N=10, data not shown). Furthermore, the angular velocity of tegula inclination and tegula rotation during a wingbeat cycle was significantly correlated with cycle period (0.68<r<0.83, P<0.05, N=10, not shown) and with the angular velocity of the wing itself(0.71<r<0.89, P<0.05, N=10, data not shown). These findings indicate that tegula movement reliably reflects wing movement, albeit slightly differently in the fore- and hindwings.

In addition to the rotational movements described above, the tegula organs of both pairs of wings shifted in the horizontal plane during the wingbeat cycle (Fig. 4C). During the downstroke, the tegulae were displaced posteriorly in synchrony with the posteriorly directed movement of the adjacent segmental scutum(Fig. 3) and the anterior stroke deviation. During the upstroke, these movements were reversed. The tegulae also showed a proximo-distal component of movement during the wingbeat cycle since the distance between the right and left thoracic pleurae decreased during the downstroke and increased during the upstroke (ordinate in Fig. 4C), this effect being more pronounced in the metathorax (open circles in Fig. 4C).

The pattern of tegula activity was investigated with respect to specific wing stroke parameters in 20 animals (10 for the forewing organs, 10 for the hindwing organs). For both pairs of wings, the relationships between stroke parameters and tegula burst duration are given in Fig. 5.

In the hindwing, mean tegula burst duration was 27.1±3.2 ms(N=10), which was slightly, although significantly, higher than the mean burst duration observed in the forewing organs (24.6±3.7 ms, P<0.05, N=10). In the majority of animals, and for both pairs of wings, the duration of tegula discharge was significantly related to the duration of the wing downstroke (P<0.05, forewing, 7/10 animals; hindwing, 8/10 animals; shown for four individuals in Fig. 5Ai,ii). A summary of the individual data is provided in Table 1. In the hindwing, burst duration increased, on average, by 0.36±0.18 ms per millisecond of increase in the downstroke interval(N=8), while in the forewing, the increase was 0.37±0.18 ms per millisecond (N=7). These values were not significantly different from one another (P>0.05). For the majority of these animals, the downstroke interval was correlated with cycle period (P<0.05, see Table 1). Thus, the duration of the tegula burst was also correlated with cycle period (P<0.05,data not shown). In the remaining (two and three, respectively) animals, burst duration was not significantly related to the downstroke interval(P>0.05).

In contrast, in the majority of animals, tegula burst properties were not correlated with stroke amplitude in either pair of wings (P>0.05;results shown for four individuals in Fig. 5Bi,ii). In the remaining animals, burst duration was either slightly negatively (one hindwing) or slightly positively (three forewings,one hindwing) correlated with stroke amplitude (see Table 1).

The duration of tegula bursts was significantly related to the angular velocity during the downstroke (ω, rad s^-1) in the majority of individuals (P<0.05; hindwing: 8/10 animals; forewing: 6/10 animals; data from four individuals each are shown in Fig. 5Ci,ii). Tegula burst duration decreased by 0.28±0.13 ms rad^-1 s^-1(N=6), on average, in the forewing, and by 0.22±0.11 ms rad^-1 s^-1 (N=8) in the hindwing organ. The remaining (two and four, respectively) locusts exhibited no significant relationship between burst duration and angular velocity (P>0.05, Table 1).

The tegula organs are activated with some delay after the beginning of the downstroke movement. In the hindwing, this latency was, on average,15.5±2.6 ms (N=10); it was 11.6±4.1 ms (N=10)in the forewing. These two values are significantly different(P<0.05, N=10). In both sets of wings, the latency between the start of the downstroke and the onset of tegula activity was related to the stroke parameters examined above. The results are shown in Fig. 6. Latency was related to downstroke interval in all 20 animals examined (P<0.05; Fig. 6A,Aii shows data from four individuals; see Table 1). In the hindwing, the latency increased by an average of 0.67±1.5 ms per millisecond increase in downstroke interval (N=10). Comparable values were observed in the forewing organ (0.72±1.9 ms ms^-1, mean values not significantly different, P>0.05, N=10). In contrast, latency was not significantly related to stroke amplitude for the forewing in seven out of 10 individuals and in the hindwing in eight out of 10 individuals (P>0.05; Fig. 6Bi,Bii illustrates data from four individuals; see Table 1). In the remaining animals, the relationship between latency and stroke amplitude was not consistent: two locusts showed a positive relationship, the remaining three a negative relationship (P<0.05).

In all 20 animals investigated, the latency of tegula discharge was dependent on the angular velocity of the wing during the downstroke in a non-linear manner. For both the fore- and hindwing organs, the typical characteristics of latency, as dependent on angular velocity, are shown in Fig. 6Ci,ii (data from four animals; r² significantly different from zero in all animals investigated, P<0.01, Table 1). In seven of the 10 animals, the latency of hindwing tegula activation reached a minimum at approximately 66.5±9.2 rad s^-1 (N=7). The corresponding minimum value was 60.7±5.9 rad s^-1(N=6) in the forewing organ, with six of the 10 animals reaching such a minimum (minimum values were calculated from the equations used to fit the data points). In the remaining seven animals, the graph did not reach a consistent minimum value within the angular velocities recorded.

In 12 of the 20 animals, the phase of the onset of tegula activity in the wingbeat cycle (insets Fig. 6)was not significantly correlated with the cycle period (P>0.05). The results were inconsistent among the remaining animals (positively correlated in five and negatively correlated in three individuals, see Table 1). Similarly, there was no clear relationship between the phase of tegula discharge and stroke amplitude in the majority of animals (15/20, Table 1; P>0.05,insets Fig. 6B). The findings that latency was inversely related to angular velocity and that this relationship, shown in Fig. 6C,was hyperbolic, suggest that the tegula is activated at a nearly constant phase irrespective of the wing's angular velocity. Indeed, in 10 of 20 animals, phase was not significantly correlated with angular velocity(P>0.05, Table 1). In the majority of the remaining animals, the phase of tegula activation varied little over a wide range of angular velocities (insets in Fig. 6C). Consistent with these observations, the mean coefficient of determination r²between phase and angular velocity(⁠

In 12 animals (six forewings, six hindwings), the `mean amplitude' of the rectified and integrated tegula burst was calculated and related to instantaneous wing stroke parameters. In all 12 animals investigated, mean burst amplitude was correlated with the angular velocity of the wing(P<0.05, r ranging from 0.41 to 0.77 in the forewings and from 0.40 to 0.68 in the hindwings). For each wing, data from four animals are shown in Fig. 7Ai,ii. In eight of the 12 animals, mean burst amplitude was not significantly related to the stroke amplitude of the wing (P>0.05, Fig. 7Bi,ii), while in the remaining four animals, such a correlation was observed(0.45<r<0.56). To examine whether this dependency of tegula burst amplitude on angular velocity was based on a common influence related to the correlation between stroke amplitude and angular velocity reported above,the partial correlation coefficients r^* were calculated(to remove the interfering variable). In three of the four animals, r^* was significantly different from zero(P<0.05, r^* ranging from 0.41 to 0.49),indicating a stronger influence of wing angular velocity on mean burst amplitude than on stroke amplitude.

In five out of nine animals examined, the stroke amplitude (Φ) of the hindwings did not change significantly after removal of the hindwing tegulae(control Φ_c=121.5±21.2°, deafferentedΦ _d=121.6±22.6°; N=5, P>0.01). In these animals, however, tegula removal significantly delayed the start of wing elevation with respect to the preceding downstroke (determined as the phase of elevation onset in the wingbeat cycle defined by the start of the downstroke, ϕ; ϕ_c=0.531±0.033, r=0.924;ϕ _d=0.587±0.026, r=0.963; P<0.01). This indicates that tegula removal in the hindwings prolongs the downstroke interval (cf. Büschges and Pearson,1991; Wolf, 1993; Fischer and Ebert, 1999). In the remaining animals, the effects of hindwing tegula removal on stroke amplitude were inconsistent: in three of the nine animals, Φ_ddecreased (on average by 13%, P<0.01), and Φ_dincreased by 9% in one animal (P<0.01). ϕ remained unchanged in three of these animals (P>0.05) and decreased in one individual(P<0.01). The ablation of the forewing organs had a small and inconsistent effects on forewing stroke amplitude: Φ increased, on average, by 4% in three out of seven animals(Φ_c=108±9.4°), remained unchanged in two and decreased in two. In six out of seven animals, ϕ did not change after ablation of the forewing organs (ϕ_c=0.462±0.022, r=0.924; ϕ_d=0.464±0.021, r=0.963, P>0.05; see, for example, Büschges and Pearson,1991).

In both the fore- and hindwings, a failure of tegula discharge was usually observed when the downstroke movement was terminated prematurely (examples are shown for the hindwing in Fig. 8Ai,ii). In both pairs of wings, stroke amplitude (Φ) and angular velocity (ω) were determined for such wingbeat cycles and in a number of cycles where premature termination was suspected. Histograms of these data are given in Fig. 8B. A failure of the hindwing tegula was observed if the amplitude of the wing beat remained within 50° of the upper stroke reversal(Fig. 8Bi, indicated by the grey shaded area). This is less than 40% of the mean wingbeat amplitude(Φ_hw=119.9±21.8°, N=8). Similarly,excitation of the forewing tegula failed at stroke amplitudes below 44°,or 40% of the mean stroke amplitude of the forewing(Φ_fw=110.9±18.6°, N=8). Occasional failures were also observed at higher stroke amplitudes for reasons as yet undetermined. In both sets of wings, tegula failure was apparently unrelated to a minimum angular velocity of the downstroke movement(Fig. 8Bii).

An understanding of the sensorimotor integration processes that participate in the selection and production of locomotor behaviour requires not only the identification of the response properties of the primary mechanoreceptors (see references in the Introduction), including the complex transduction `cascade'that underlies the excitation of a particular sensory cell (e.g. Moran et al., 1976), but also a consideration of the possibility that a large population of, possibly different, afferent input fibres might encode particular signals (for an overview, see Sparks et al.,1997).

The present study focuses on an insect mechanosensory organ, the tegula,the knob-shaped cupola of which is integrated into a common ligament attached to the scutum, basalar sclerite, thoracic pleura and leading edge of the wing(Fig. 3). This wing-associated sense organ plays an important role in the generation and modulation of the flight motor pattern. However, in contrast to other wing-related sense organs,which usually consist of one morphological type of mechanoreceptor, the tegula houses two morphologically distinct sensory systems. Each consists of a relatively large number of primary mechanosensory axons, approximately 40 from mechanosensory hairs located on the posterior cupola and approximately 30 scolopidial sensilla from a chordotonal organ attached to the inner surface of the posterior cupola. These sensory cells each project into the central nervous system in a single afferent axon(Kutsch et al., 1980), which makes (excitatory) monosynaptic connections with motoneurons driving the wing elevator muscles (Pearson and Wolf,1988) and also supplies all known interneurons of the flight oscillator in parallel (see Pearson and Wolf, 1988, 1989).

At present, however, little is known about what wingbeat parameters might be encoded by the tegula organs or how the organ might be activated during flight. To address these questions, the present study employed electrophysiology and high-speed video recordings to monitor the collective activity patterns of tegula afferents, the kinematic movements of the wing and of the tegula organs themselves as well as of the cuticular structures of the wing hinge attached to the tegula organs.

It has been hypothesised that the tegula is excited during flight by the organ touching a membranous fold during the downstroke, probably resulting in the bending of the mechanosensory hairs on the posterior cupola(Kutsch et al., 1980). The high-speed video recordings confirmed that, during the downstroke, the posterior region of the tegula on which the hair plate is located touches a membrane fold located just ventral to the subcosta. This contact is intensified by the anterior deviation of the wing(Fig. 1C) during the downstroke and by the synchronous, posteriorly directed shift of the scutum(Fig. 3). Since the hair plate region was covered partly by the membrane fold itself and partly by the ligament (and, thus, was not visible in the video recordings) when the wing approached its lower reversal position, we were unable to quantify accurately the total time of contact between the hair plate region and the membrane fold from the high-speed recordings. We conclude from our recordings, however, that at least part of the hair plate region touches the membrane fold during approximately half of the cycle period, including one-third of the upstroke interval. The nerve recordings (Figs 2, 6) show that tegula activity starts with a brief delay after the upper stroke reversal, thus roughly matching the time in the cycle when part of the hair plate makes first contact with the membrane fold. Nevertheless, stimulation of the hair plate may not play a key role in tegula excitation because (i) a tegula discharge during the wingbeat cycle cannot be prevented by covering the hair plate with wax(Neumann, 1985) and (ii)tegula bursts reliably terminated at the lower reversal point of the wing(Fig. 2), i.e. at a time when the hair plate was still in contact with the membrane fold.

The mechanisms that, apparently, limit the tegula discharge to the downstroke interval are not clear at present. The obvious coincidence between the posterior stroke deviation, which starts when the wing passes its lower stroke reversal (Fig. 4B), and tegula burst termination might indicate that the tegula is also sensitive to wing motion in a plane perpendicular to the stroke plane and, thus, that the anterior stroke deviation during the downstroke (see Fig. 4B) might limit tegula activity to the downstroke interval. However, this hypothesis now needs to be addressed by further experimentation.

For both wings, a failure in tegula excitation was almost exclusively observed during stroke cycles in which the wing did not reach its normal lower reversal point (Fig. 8B). However, tegula failure was not restricted to cycles in which the angular velocity was particularly low (Fig. 8B). In addition, moving the wing passively at angular velocities far below those observed during active flight, but at comparable amplitudes,activated the tegula organs (see Fig. 2 in Fischer and Ebert,1999). This indicates that excitation of the tegula relies on the wing passing a `critical' position during the downstroke rather than the wing reaching a certain angular velocity (Fig. 8Bii). Together with the observations of Neumann(1985), this suggests that the response of position-sensitive afferents in the chordotonal organ of the tegula might play a role in the activation of the organ during the wing stroke.

The pterothoracic scuta are attached to each other by a flexible ligament. The two scuta move posteriorly at different phases during the downstroke, but they shift anteriorly almost in synchrony during the upstroke(Fig. 3). The hindwing tegula is integrated into this ligament, which inserts in a mesothoracic fold located posteriorly between the scutum and the dorsal border of the epimeron. One might expect, therefore, that hindwing tegula activity would be affected by the kinematics of the forewing. However, we found no indication of a mechanical influence of the mesothoracic scutum on hindwing tegula activity and vice versa. Apparently, the two tegula organs are functionally quite separate.

The mechanical elements of the locust thorax and wing hinge are tightly coupled and move in strict phase relationships during flight (e.g. Fig. 3). The movement of these elements, to which the tegula organ is attached by a common ligament, results in the rotational and tilting movements of this organ, which occur at stable phase values with respect to wing movement(Fig. 4). Both the latency of tegula activity with respect to the onset of the wing downstroke and the burst duration decrease when the downstroke interval is reduced with increasing wingbeat frequency (Figs 5, 6). This implies that the tegula organs are activated at an almost constant phase during the wingbeat cycle (Fig. 6; Table 1). These observations are in accord with the phase-locked kinematics of the organ's movements.

The present results thus seemingly disagree with previous findings suggesting a relatively constant latency of tegula discharge, at least at lower wingbeat frequencies (Wolf and Pearson, 1988). However, latency was determined with respect to the activity of single wing depressor muscles (first basalar or subalar) in previous studies, and the relationship between the activity of a particular muscle and wing movement can be variable(Wilson and Weis-Fogh, 1962;Pfau, 1978, 1982; Möhl, 1985, 1988). Evaluation of the present data set with regard to first basalar muscle activity indeed demonstrated considerable variability between the individuals tested,including, almost equally, no, positive or negative relationships between latency and cycle period (data not shown). A variable discharge pattern was observed in particular in the first basalar muscle, which is involved in a number of tasks (e.g. flight steering, Zarnack and Möhl, 1977; Baker, 1979; climbing flight, Fischer, 1998; Fischer and Kutsch, 1999;adjustment of the angular setting of the wing, Pfau, 1978, 1982). The quantitative data presented by Wolf and Pearson(1988) were from just one animal, and the slope of the relationship between latency and cycle period,although small (approximately 0.1; H. Wolf and K. G. Pearson, unpublished data), is well within the range of slopes observed in the present study (data not shown, Table 1). The effects of the tegula on the flight motor pattern reported previously(Wolf and Pearson, 1988; Fischer and Ebert, 1999) thus appear to result to a large extent from the constant delay required to elicit elevator activity in response to a tegula discharge, particularly at lower wingbeat frequencies (Wolf,1993), rather than from a constant latency of tegula activation(see also the effects of tegula ablation reported above).

Key parameters of tegula discharge, such as latency and phase(Fig. 6), duration(Fig. 5) and the mean amplitude of the tegula burst (Fig. 7),were related to the stroke amplitude, downstroke interval and angular velocity of the wing, which are important for aerodynamic force production during flight (e.g. Ellington, 1984;Lehmann and Dickinson, 1997, 1998; Thüring, 1986).

Apart from the fact that a minimum amplitude seems to be required for tegula activation (Fig. 8),stroke amplitude does not appear to play an important role in determining the pattern of tegula discharge during flight since the duration, the latency and the amplitude of the tegula bursts were not significantly related to the amplitude of the wing stroke (Figs 5B, 6B, 7B). Furthermore, removal of the tegula organs had no consistent, if any, effect on stroke amplitude in either pair of wings.

In contrast, the latency and duration of the tegula discharge were significantly related to the downstroke interval and, thus, to the cycle period, since these two parameters are correlated during flight(Table 1; see also Wolf, 1993). The dependency of the duration and latency of tegula discharge on the downstroke interval is evident when considering the fact that the tegula is excited by (e.g. Wolf and Pearson, 1988)(Fig. 8Ai) and is active during the downstroke, and it suggests that the tegula encodes parameters of the downstroke such as timing (Wolf,1993) and velocity (see also Figs 5C, 6C).

In both sets of wings, latency (Fig. 6C), burst duration (Fig. 5C) and burst amplitude (Fig. 7A) depend on the angular velocity of the wing. The relationship between latency and angular velocity is non-linear. Towards higher angular velocities, latency approaches or reaches a minimum value and often stays near this minimum if angular velocity increases further. The minimum is reached between approximately 45 and 75 rad s^-1. The corresponding wingbeat frequencies are between approximately 15 and 19 Hz, i.e. they mark the lower limit of frequencies observed during free flight. The latency of the tegula discharge thus appears to be kept within a narrow range during normal flight,indicating that a feedback loop is functioning. Furthermore, the tegula organs are sensitive to a very wide range of angular velocities(Fig. 8), including very low values (e.g. 2-5 rad s^-1) that are not observed during flight (data shown in and extracted from Fig. 2 in Fischer and Ebert,1999). In principle, angular velocity could be adjusted by controlling stroke amplitude or cycle period or both. However, in 85 % of the animals examined, angular velocity was (negatively) related to the cycle period (Table 1), while in only 60 % of the locusts was it (positively) related to stroke amplitude (data not shown). It appears, therefore, that angular velocity is primarily related to wingbeat frequency.

The above conclusions are based on the changes in the collective activity of the tegula afferents with respect to variation in particular wingbeat parameters. It has previously been suggested that the tegula might work as a functional unit since the two sensory systems in the tegula are in close vicinity to each other (Kutsch et al.,1980). However, since the tegula consists of a large number of afferents (in contrast to other wing-associated sensory organs, e.g. the single-cell stretch receptor), there is the distinct possibility of different response properties and range fractionation (for reviews, see Field and Matheson, 1998; Newland et al., 1995; Neumann, 1985) among the 70-80 sensory cells of the tegula. This might allow population-coding of wingbeat parameters, although at present this possibility must remain speculative because of the absence of data concerning tegula receptor physiology or the the specificity of the central connections of different receptor cell types.

In the present study, a meaningful distinction of different spike amplitudes, or discharge properties of different axon populations, was not possible because of the rather homogeneous distribution of both spike amplitudes and discharge characteristics in the compact tegula bursts (data not shown, but see Fig. 2). Even in the few cases where large, small and sometimes intermediate spike amplitudes could be differentiated in the tegula discharge, these groups of sensory axons had very similar discharge patterns (data not shown).

The locust pterothorax is a compact mechanical structure composed of a large number of coupled cuticular elements that move in strict phase relationships during flight (e.g. Fig. 3). The mechanical composition and properties of the pterothorax(including flight musculature) both determine, to a major extent, the complex three-dimensional wing movements (Pfau, 1978, 1982) and limit the power output of the flight oscillator through their mechanical damping properties(Roeder, 1951; Soltavalta, 1952). A more global assessment of flight performance should consider at least two points. First, the activation phase of the flight muscles, which controls their mechanical power output (e.g. Stevenson and Josephson, 1990), and in turn is translated into basic flight parameters, such as wingbeat frequency, stroke amplitude, the angular velocity of the wing and finally into aerodynamic force production, should be examined. Second, an appropriate tuning of these flight parameters may be important not only for flight performance per se but also in the context of a possible resonance stabilisation of the flight oscillator(Greenwald, 1960; Scharstein, 1998a,b).

Insects need to activate their flight muscles at appropriate times and phases to generate a functional flight pattern under a variety of conditions,e.g. during changes in stroke frequency. This provides a direct functional context for wingbeat-synchronous (i.e. phase-locked, Fig. 6) sensorimotor pathways,such as that involving the tegula, which are important in flight motor control(e.g. Wendler, 1974; Möhl, 1985; Wolf and Pearson, 1988; Wolf, 1993; Fischer and Ebert, 1999). Thus, the respective sense organs may serve to monitor, and control, the appropriate tuning of critical flight variables, e.g. the angular velocity of the wing, which plays an important role in aerodynamic performance (e.g. Ellington et al., 1996;Lehmann and Dickinson, 1998; Dickinson et al., 1999). If there is not just a correlation between particular parameters of wing movement and tegula discharge, but if this correlation is non-linear and a preferred parameter combination is maintained during normal flight (e.g. the phase of tegula activation during the cycle or the relationship between latency and angular velocity, Fig. 6), this may indicate that a feedback loop is involved. Tegula afferent pathways might thus be part of a feedback loop controlling not only the basic flight motor pattern(Wolf and Pearson, 1988) but also the angular velocity of the downstroke, with the variations in tegula activity serving as error signals for these control loops. However,experiments demonstrating conclusively such feedback loops can be obtained only by examining the control circuit in cybernetic experiments (e.g. Wendler, 1974).

We gratefully acknowledge the Fachbereich Biologie, Universität Konstanz (Germany), for providing the digital high-speed video system (funded by an HBFG grant to the University). We thank W. Kutsch (Universität Konstanz) for providing laboratory space and basic equipment during the initial parts of the study. T. Breithaupt (Universität Konstanz)generously provided a variety of special equipment. We are grateful to C. Graef (Universität Köln, Germany) for technical assistance and to H. Scharstein and G. Wendler (Universität Köln) for valuable discussions. Special thanks are due to W. J. Heitler (University of St Andrews, Scotland) for a custommade upgrade of Data View. We thank three anonymous referees for valuable comments and suggestions regarding the analysis and interpretation of our data, and S. D. Merrywest, D. L. McLean and R. J. Chapman (University of St Andrews) for critically reading the manuscript.