The parasitoid wasps of the genus Trichogramma use the surface curvature of their insect egg hosts to set an upper limit to the number of progeny allocated to the host, as well as the duration of their host examination. In addition, host recognition and host acceptance are in part mediated by surface curvature. In this paper, the relationships between the positions of body parts of the wasp and surface curvature are examined in order to determine a possible mechanism for curvature detection by ‘the wasp.
Wasps of different sizes were photographed in profile while examining glass bead models of different diameters. The positions of selected body parts were analysed using a digitizer and microcomputer. The height of the wasp above the model surface did not change with surface curvature. Furthermore, the angle of the head relative to the thorax was also constant over the range of models used. Only the scapal-head angle and flagellar-head angle changed significantly with surface curvature.
A curvature detecting mechanism is proposed in which the wasp uses the scapal-head angle to measure the curvature of the surface. The body of the wasp is maintained at a preferred height and angle to the substrate, serving as a fixed platform from which curvature measurements are made. Additional features of this mechanism, as well as its correlation with morphological and behavioural findings, are discussed.
The minute parasitoid wasps of the genus Trichogramma use the eggs of a wide variety of insect species as hosts for their larval stages. An important component in host recognition and clutch size determination by these wasps is their ability to detect and quantify differences in substrate curvature. Salt (1935) and de Jong & Pak (1984) showed that only objects within a particular range of curvature are accepted for ovipositing, but the wasps will accept unsuitable objects such as mercury globules, glass beads or seeds, provided they are of suitable size and shape.
J. M. Schmidt & J. J. B. Smith (in preparation) demonstrated that surface curvature alone is sufficient to elicit host examination and ovipositing behaviour.
The amount of time taken by the wasps to examine the host surface also depends upon the radius of the host (Klomp & Teerink, 1962; Schmidt & Smith, 1985a,b, and in preparation). This time dependence is unaffected by the extent to which the surface area of the host is exposed above the substrate, and is the constant for hosts and glass bead models of the same diameter (Schmidt & Smith, 1985a,b, and in preparation).
Trichogramma are gregarious parasitoids, and the female deposits a variable number of eggs into each host depending upon several factors, including host size, shape, distribution and chemical content (Klomp & Teerink, 1962, 1967; Marston & Ertle, 1969; Nettles et al. 1982, 1985; Schmidt & Smith, 1985a,b). The adjustment of clutch size to host volume results in improved reproductive success for the ovipositing female by reducing the deleterious effects of larval competition for limited host nutrient (Klomp & Teerink, 1967; Charnov & Skinner, 1985). Both curvature and the length of the wasp’s initial transit over the surface of the host are used by the wasp to determine clutch size (Schmidt & Smith, 1985a, and in preparation). In the absence of other cues, host curvature sets an upper limit to the number of progeny allocated to the host.
In this paper, correlations between body angles and substrate curvature are examined to determine the possible mechanisms’ of host curvature measurement by Trichogramma.
MATERIALS AND METHODS
Culture of Trichogramma minutum
Trichogramma minutum Riley (Hymenoptera: Chalcidoidea : Trichogrammati-dae) was cultured as described by Schmidt & Smith (1985a). Hosts used for rearing were frozen (killed) eggs of the tobacco hornworm, Manduca sexta (L.) (Lepidop-tera: Sphingidae), obtained from Carolina Biological Supply Co. (Burlington, NC, USA 27215).
The wasps used in the experiments ranged in body length from 0 ·35 to 0 ·85 mm. These differences are expected phenotypic variations resulting from different numbers of larvae developing in a single host (Flanders, 1935; Salt, 1934; Klomp & Teerink, 1967). Smaller emergents are produced under conditions of larval crowding. Wasps with obvious physical deformities were not used.
Glass bead models
The glass bead models used in the experiment were spherical, with diameters ranging between 0 ·30 and 1 ·5 mm. The beads were cleaned by washing in acetone, ethanol and distilled water, and mounted individually on 2 ×2 cm white cardboard squares with a minimal amount of gum arabic. Clear plastic Petri dishes (5 cm diameter) with tightly fitting lids were used as experimental arenas.
All photographs were taken on Ektachrome colour slide film (Daylight, ASA 64), with a Nikon F camera attached with an adapter tube to a Bausch & Lomb binocular dissecting microscope. Illumination was provided with an electronic flash unit with a 30 W Xenon flash tube (Carl Zeiss Ukatron UN 60) mounted in a Zeiss microscope lamp housing. The lens of the flash tube was located 9 cm from the specimens at a 45 °angle from the substrate. Additional lighting for locating the specimens and focusing was provided by a heat-filtered tungsten lamp. All photographs were taken from directly above the centre of the host.
Wasps of various body lengths were combined with glass bead models of different diameters in order to obtain the greatest possible range of relative substrate curvatures. Each of 320 wasps was photographed once in lateral profile during the examination of a glass bead model. 78 exposures were selected as the final data set, based on the following criteria: (1) the wasp was strictly in lateral profile, with only one antennal base and one lateral ocellus visible and the crest of the scutum clearly outlined; (2) the wasp was completely in focus, including the details of antennal structure and leg position; (3) the antenna on the side of the wasp facing the camera was in contact with the host, with the flagellar portion not bent by compression against the host. In addition, wasps standing on a level substrate were photographed, both from directly above and from the side.
Experimental design and data processing
Data from the photographs were entered into a microcomputer by means of a digitizer (Summagraphics Bit Pad One). The selected slides were projected onto the digitizer and the data points entered manually using a stylus. Eight points were entered for each wasp (Fig. 1). These were used to calculate head length, body length, height of the wasp above the surface, and the angles between the various body parts of the wasp and their relationship to the host surface. Three points on the circumference of the model were entered, from which the radius and the coordinates of the centre of the model were determined. All linear measurements, including the diameters of the models used, were calculated in terms of head length (DE) of the wasp used with that host. Accordingly, the relative radii of the models ranged from 0 ·75 to 5 ·0 head units. In addition, the distance between the pronotum of the wasp and the model centre (FO), the height of the crest of the pronotum above the substrate (FO— R), and the height of the pronotum above the mesothoracic tarsus (FT) were determined. Three angles were calculated for each wasp: the flagellar-scapal angle (ABC), the scapal-head angle (BCD) and the neck angle (CFG) (Fig. 1). All processing was done using a Corona Model PC-HD microcomputer using programs written in the STSC-APL PLUS language.
The proportional lengths of body parts of the wasps were analysed in order to determine if the size relationships between parts change over the range of wasp sizes used. The flagellar segments and scape of the antennae were measured, as well as head length and the distance between the pronotal crest and the abdominal tip
(Fig. 1). By comparing the correlation between two perpendicular measures (head and abdominal length) with the correlation found between nearly parallel measures (head and flagellar length), the possible distortion due to deviation from a completely lateral profile was evaluated. The leg span, height of the pronotum and antennal angles were also measured from the photographs of wasps standing on a level substrate. In addition, the lengths of the head and the antennal segments were measured directly from anaesthetized specimens using a binocular microscope and ocular micrometer (Table 1).
RESULTS AND DISCUSSION
Allometry of body parts
Over the range of wasp sizes used in this experiment, linear proportional relationships were found between scapal length and head length (r=+0 ·867, N = 78), between flagellar length and head length (r= +0 ·752), and between head length and abdominal length (r= +0 ·898) (Fig. 2). For all three correlations, the power (log-log) regressions showed linear relationships over the range of wasp sizes used (exponent of regression between 0 ·96 and 1 ·01). Thus, the correlations measured between body angles and surface curvature are not significantly confounded by changes in the relative proportions of body parts, despite differences in body length between wasps. Since the relative proportions of the wasps do not differ significantly, the radii of the models can be compared to changes in body angle and height in terms of head length units. Since the correlations of scapal or flagellar length with head length did not differ significantly from the correlation found between head length and abdominal length, any deviations of the wasps’ positions from exact lateral profiles were presumed to be not sufficient to introduce a significant bias in the angle measurements.
No significant change was found in the height of the pronotum above the glass bead surface (FO-R) (1 ·4 ±0 ·1 mm) over the range of bead radii (r =+0 ·036). The relationship between the distance of the pronotum from the centre of the bead (FO) and bead radius (R) was strictly linear (r = +0 ·994) (Fig. 3). In addition, no significant correlation was found between the height of the pronotum above the mesothoracic pretarsus and the radius of the bead (r -+0 ·11). These results indicate that the wasp keeps its body at a preferred height above the host surface.
From these data, it is not evident how the wasp controls its height above the substrate, or to what degree the wasp must actively adjust its position relative to the host. For example, the wasp could change the span between contralateral legs in order to raise or lower its body. To investigate the basis of the wasp’s constant height above the substrate, the change in the height of the wasp above the centre of the glass bead with changing surface curvature was modelled (Fig. 4A). If the leg span remains constant, then the distance between the pronotum and the bead centre is given by :
Using observed values for the height of the pronotum when the wasp is standing on a flat surface (1 ·4 ±0 ·1 head units) and the span of the mesothoracic legs perpendicular to the long axis of the wasp (1 ·6 ±0 ·1 head units) (Table 1), the pronotal-bead centre distance can be calculated and plotted against the observed data for wasps on models of differing relative radius (Fig. 3). From these, it is apparent that over most of the range of host radii the calculated values correspond closely with the observed data. This suggests that the wasps do not need actively to adjust either leg span or body height above the pretarsi in order to maintain the preferred height above the substrate. Only when host diameter is relatively small, close to the leg span distance itself, is there an apparent departure from the predicted line (Fig. 3), indicating some adjustment of wasp height to maintain the distance between the wasp and the surface. This adjustment probably results from a reduction of leg span in order to maintain contact with the host surface. However, the change in leg position must be very small, since no significant change in the height of the pronotum above the mesothoracic pretarsi was observed. The maintenance of the body at a preferred distance from the substrate has been shown for other insects (Cruse, 1976; Kemmerling & Varjú, 1982), in which the placement and relative position of the tarsi on the substrate depends upon contact of the tarsi with a supporting surface (Cruse, 1979).
Clearly, for very small hosts the adjustment of leg position to maintain body height could provide the wasps with some measure of host curvature. However, it would not appear to provide significant measures over the range of host sizes accepted by the wasp (radius > leg span).
During the short period in which all legs contact the surface of the host, some relationships of relative position do arise between the legs and the body of the wasp which are dependent upon surface curvature. However, these relationships change rapidly during this period because of the forward movement of the wasp’s body. Furthermore, it is unlikely that the wasp can use relative leg position to detect differences in curvature because of the small distances between the wasp’s legs. For ipsilateral legs, the distance between the mesothoracic pretarsus and the pro- or metapretarsus is only 0 ·75 –0 ·85 head units. For such small distances there is little proportional change (<5 %) in the height of the mesothoracic pretarsus relative to the pro- and metathoracic pretarsi for most of the range of host radii tested. A larger change in leg position is required to maintain body height only when the host radius is near or below one head unit. Since such hosts are too small for larval development (Salt, 1935), this change of relative leg position could serve as a cue for host rejection, but would not provide a useful measure of size for larger hosts.
The alternate triangle gait is well suited to maintaining the body of the wasp at a fixed height and angle above the substrate. Furthermore, by keeping the span between contra- and ipsilateral legs relatively small, the wasps reduce the amount of change in height above the substrate without actively adjusting their posture. This suggests that the effect of the legs may be to maintain the body as a platform at fixed height and attitude with respect to the surface. Using the antennae, and possibly the angle between the body and the head, the wasp could measure the slope of the surface away from the centre of this platform to estimate curvature. To explore this possibility, changes in the neck and antennal angle with surface curvature were determined.
Correlation of model radius with body angles
Of the three body angles measured, only neck angle (Fig. 1) showed no significant correlation with model radius (r = +0 –09). The neck angle is fixed at between 70 ° and 80 °, maintaining the head at a constant angle and height relative to the legs and thorax. The same angle was observed on wasps walking or standing on a level substrate. During oviposition, the head is raised further from the substrate and retracted against the thorax so the neck angle is decreased to 60 –65 °.
In contrast, both the flagellar-scapal angle and the scapal-head angle (Figs 5, 6) change significantly as model radius is varied. The greatest linear correlation coefficient was found for the scapal-head angle (r = − 0 ·81), the angle decreasing as radius is increased (Fig. 5). The relationship is non-linear and concave, the magnitude of the slope decreasing with increasing radius. The scapal-head angle ranges between 135 ° and 70 ° over a range of model radii from 0 ·75 to 5 ·0 head units.
The linear correlation coefficient between the flagellar-scapal angle and model radius was also significant and negative (r = −0 ·57) (Fig. 6). The flagellar-scapal angle ranges between 100 ° and 70 ° decreasing with greater model radius. Flagellar-scapal angle and scapal-head angle with host radius also correlate significantly with each other (r = +0 ·83) ; the relationship is linear and has a positive slope (Fig. 7).
These observations suggest that Trichogramma could use the scapal-head angle to measure surface curvature. By maintaining its body at a fixed height and attitude with respect to the substrate, and the head at a fixed angle to the body, the wasp can determine the tangential slope of the substrate using its extended antennae. This method clearly has the advantage of utilizing the greatest linear span available to the wasp. The changes in scapal-head angle (Fig. 5) could also provide information about surface curvature over a wide range of host sizes. Since the relationship is concave, the sensitivity of the measure apparently decreases for comparatively large hosts. However, it is unlikely that the wasps require an as accurate measure of curvature for large hosts (radius>2 ·5 head units), since small variations in the number of progeny allocated to larger hosts will be less detrimental than for small hosts (Klomp & Teerink, 1967). Very large objects (radius > 6 head units) are rarely accepted as hosts (Salt, 1935; de Jong & Pak, 1984), probably because the wasp cannot distinguish the host object from the substrate on the basis of such small curvatures and scapal angles.
In deriving this model, it was assumed for simplicity that the points/), C and O lie on a straight line (Fig. 4B). As measured in the experiment, the angle DCO was found to be relatively small (5 ·7 ± 2 ·3 °), indicating that the estimates obtained from equations 2, 3 and 4 for the scapal-head angle may be slightly lower than the actual values. However, the resulting small increase in the predicted values of the scapal-head angle does not appreciably change the fit of the model to the observed values (Fig. 5A), nor does it invalidate the model as a first approximation.
The proposal that the wasp determines surface curvature by measuring changes in the maximum scapal angle is also supported by morphological studies of the antennae (J. M. Schmidt & J. J. B. Smith, in preparation). In most insects, groups of hairs in the joint regions of the legs, neck and antennae act as proprioceptors providing information about the relative displacement of body parts (McIver, 1985). In Trichogramma, hairplates found at the antennal joints (J. M. Schmidt & J. J. B. Smith, in preparation) could provide the wasps with information about the angles described previously. The position of the flagellum is monitored by a dorsal and a lateral hairplate. The arrangement of these hairplates is such that they can provide the wasp with information about antennal position only when the flagellum is near the limits of its movement, either retracted against the scape or fully extended. It is unlikely that these hairplates can contribute much to the wasp’s curvature measures. The scapal-head joint is monitored by four distinct hairplates arranged around the ventral surface of the scapal radicular base. As the scape moves, the individual hair sensilla make contact with the socket of the antenna and are displaced to differing degrees from their resting orientations. Clearly, these sensilla could provide the wasp with detailed information about the scapal-head angle and hence surface curvature.
Although the correlations observed in these experiments do not demonstrate conclusively the mechanism by which Trichogramma measures surface curvature, they do indicate a set of body postures and relative angles which could be used to mediate the wasp’s response. Since the wasps can respond to host curvature in the absence of other external cues, including visual information (Schmidt & Smith,1985a, and in preparation), some mechanism involving proprioceptive cues appears most likely. In the absence of other significant changes in the relative position of body parts with surface radius, measuring the antennal angles is an obvious method of curvature detection.
Similar use of the antennae has been implicated for the egg parasitoid Telenomus heliothidis Ashmead (Hymenoptera: Scelionidae) (Strand & Vinson, 1983a). These wasps require both chemical and physical cues to mediate host acceptance (Strand & Vinson, 1982) and investigate the host surface by drumming with their elbowed antennae in a manner similar to that used by Trichogramma (Strand & Vinson, 1983b). By observing the response of Telenomus to models of various shapes, Strand & Vinson (1983a) have suggested that host shape is sensed with the antennae. Antennal measurement of host size and shape has also been proposed for Encarsia formosa Gahan (Hymenoptera: Chalcidoidea: Aphelinidae) (van Lenteren, Nell & Svenster-van der Lelie, 1980).
That insects use their antennae to determine the topography of their local surroundings suggests itself intuitively, as indicated by the common use of the term ‘feelers’ (Fühlern) for these structures. Despite this obvious function, relatively little work has been done to show how the mechanosensory structures associated with the antennae could be involved-in the detection of form or texture. Trichogramma, with its specialized dependence upon information about the geometry of hosts, is a good candidate for the detailed investigation of these sensory and integrative systems.
We thank Rosemary Tanner for assistance and Dr J. Machin for helpful discussion. Support was provided by a grant from the Natural Sciences and Engineering Research Council of Canada.