External Design and Field of View of the Compound Eyes in a Raptorial Neuropteran Insect, Mantispa Styriaca

A first glance at Mantispa styriaca, a predatory insect of the neuropteran family Mantispidae, suggests a small praying mantis. There are large bulbous eyes and a stalked prothorax bearing spiny forelegs which are powerful prehensile claws. The two distal members of the foreleg can be snapped against the next member, like a jack-knife. We have seen that Mantispa prefers to spend its time in trees or bushes, where it lies motionless in wait of prey. If a small insect comes within range, Mantispa moves in its direction and fixes it with its eyes. Then the two claws are extended with a sudden forward jerk of the body and the insect is caught. When the capture effort fails, it can happen that Mantispa ‘follows up’ its prey. We have also seen that even while Mantispa is eating its prey, held in the mandibles, it may reach out with its two free extremities after another insect scurrying along at its side, but this effort usually fails.

Though Mantispa and the praying mantis show similarities in capture behaviour, they are different in life cycle. The praying mantis shows hemimetabolous development and even the very small first-instar larvae show the same prey-capturing behaviour with success rates similar to those of the large adult animals (Mittelstaedt, 1957). Mantispa, in contrast, is polymetabolous with two very different larval stages (Brauer, 1852, 1869). The first larval stage is slender and motile, and has suckers which it uses to penetrate the egg cocoons of the wolf spider. It then eats the eggs and young spiders and later moults into the second maggot-like larval stage which pupates in a cocoon it spins for itself within the spider cocoon (Hungerford, 1936; Schremmer, 1959, 1983). Only the adult insect that emerges from the metamorphosis has the powerful forelegs. Field observations show that the praying mantis hunts in the daytime and that the males mate at twilight and at night (Rossel, 1979). Similar behaviour is probably shown by Mantispa (New and Haddow, 1973).

In this paper we investigate whether the capture of prey by Mantispa is optically directed in a way similar to that in the praying mantis (Roeder, 1959; Levin and Maldonado, 1970; Maldonado and Barros-Pita, 1970; Maldonado et al. 1970; Rossel, 1979, 1980, 1983a,b, 1986), i.e. with a forward-looking acute zone, an adaptation for enhanced acuity of fixation, and fields of view of the two compound eyes with a large overlap. We approached this problem by measuring the total visual and binocular field size of both compound eyes and by surveying the optical parameters of single ommatidia with simple optical methods, such as the pseudopupil method.

Adult females and males of Mantispa styriaca Poda (Planipennia, Neuroptera) (Fig. 1) were obtained from the southernmost tip of the Istrian peninsula near Premantura, Yugoslavia. The insects were collected early in July in the late afternoon and early evening in a pine wood. All the females came from a community of more than 80 individuals in one tree; males were found only occasionally on other nearby pines. All animals were about the same size. They were kept under controlled laboratory conditions (15h:9h light:dark cycle) in individual cages and were fed fruit flies in the evening.

Whole eyes were fixed in a standard mixture of 2.5 % glutaraldehyde and 2.5 % paraformaldehyde in 0.1 mol I⁻¹ sodium cacodylate buffer (pH 7.2) at 4 °C for some hours. The tissue was then washed for 1h in four changes of 0.1 mol l⁻¹ cacodylate buffer. Postfixation was in 2 % osmium tetroxide buffered with veronal acetate buffer (pH 7.2) at 4°C for 2h. Dehydration was through an ascending series of ethanol to propylene oxide; embedding followed in soft Epon 812. Semithin sections 1–2 μm thick were stained with Toluidine Blue or were left unstained. For scanning electron microscopy, (SEM), air-dried specimens were sputter coated with a 20–30 nm thick layer of gold and examined in a Cambridge scanning electron microscope.

Before the optical measurements began, a Wild M400 photomicroscope equipped with an ocular micrometer was used to record eye size and the smallest distance between the eyes. Mapping and counting of facets was on the basis of SEM photographs.

The visual fields were measured by using the luminous (corneal) pseudopupil caused by axially incident light (orthodromic illumination). The insect was mounted in the centre of a three-dimensional goniometer. The eucentric point corresponded to the middle point of the spherical coordinate system. The angle of each axis of the goniometer was measured to one decimal place by a linear recording potentiometer with digital display. The high-intensity, bluish-yellow, round pseudopupil (see Fig. 4) was allowed to wander around the marginal ommatidia, starting from a reference point on the equator (defined by the smallest distance between both eyes, see Fig. 2A; fused rhabdoms are unrotated so there is no obvious equator; cf. open rhabdoms of flies). Meanwhile, the spherical coordinates of the goniometer were read at 10° intervals, as described extensively by Beersma et al. (1977). The spherical form of the eye ensured that measured optical axes were in accordance with real optical axes.

The effective acceptance angles of ommatidia (Δ ρ), which determine the size of the ommatidial visual field, were estimated in monocular and binocular eye regions by measuring the angular span in which the luminous pseudopupil remains in the ommatidium (similar to Nilsson and Odselius, 1982, who, however, used a true pseudopupil). This optical measuring method defines the angle at which the distal rhabdom on the lens surface remains fully visible. In contrast, with electrophysiological determination, the acceptance angle is defined as the halfwidth of the ommatidia’s or receptor’s acceptance function. The general agreement of Ap (light-adapted) and interommatidial angle (Δ ϕ) confirms the suitability of the optical measuring method.

Interommatidial angles (Δ ϕ) were determined from R (the distance between the midpoint of the corneal surface of an ommatidium and the imaginary intersection of its optical axis with the optical axis of a neighbouring ommatidium) and the diameter D of the regular hexagonal facet, determined in semithin sections (Fig. 3) in the horizontal plane (x,z) and in the sagittal plane (y,z). Eight different sections of the eye were measured (see Table 3). The interommatidial angles were calculated according to Snyder (1977) using the D/R formula.

The emerald-green eyes of adult Mantispa are almost hemispheres, covering a large part of the rather simple head capsule (there are no ocelli) (Figs 1, 2A). The eye surface measures about 1.2 mm² and contains some 3500 ommatidia with a regular hexagonal shape, though those at the edges can be irregular. The surface of the ommatidia and the interommatidial area are covered with numerous hemispherical corneal nipples that are considerably lower on the cupola of the lens than on the surrounding mantle (Fig. 2B).

Figs 5 and 6 and Table 1 show the extent of the areas of binocular, monoculae and blind visual field as a percentage and in mm² (all values are mean±S.D.). The extent of the binocular visual field was 6.6% in females and 8.2% in males, the 20 % greater area in males producing a reduction in the blind spot. In contrast, the monocular field was about the same size in both sexes. The maximum extent of the binocular visual field in the dorsoventral direction was 135.2° in females and 142° in males (Figs 5, 6). The maximum horizontal overlap averaged 42.7° in females and 52.4° in males; in males there was a suggestion of two maxima in the dorsoventral direction at about 0° and 30°.

Trigonometry, using the angular data and the interocular separation, showed that the projected optical axes of marginal ommatidia intersected in the area of maximal overlap in the sagittal plane of the animal at 1.017 mm in males and 1.305 mm in females. This ‘critical’ distance amounts to some 10–20% of the length of the extended forelegs. The convergence angle with an interocular separation of 1.07 mm in males and 1.09 mm in females corresponded to the overlap area in degrees and was 32.4° on average in males and 38.3° in females. The blind spot in the neck region had a dorsoventral extent of 224.8° in females and 218.3° in males on average.

Horizontal measurements in the binocular region near the equator produced average values of Δ ρ of 1.9° in the light-adapted eye (Δ ρLA) and Δ ρ in the dark-adapted eye (Δ ρDA) of 3.6°, and Δ ρLA of 2.0° and Δ ρDA of 3.6° in the monocular tegion of the lateral part of the eye. Table 2 summarizes the calculated values and Fig. 7 shows the dependence of the effective acceptance angle of a single ommatidium in the binocular and monocular areas on the duration of illumination. The curves are quite similar; there is no significant difference between the monocular and binocular areas (t-test).

In the frontal equatorial region, the diameter D of the facets averaged 21.3 μm. In the anterior dorsal area it was 20.9 μm and on the outer margin of the eye, 20.1 μm; thus it was quite constant over the entire eye. There were slight variations in the values of R, between 450 and 550 μm.

Values of Δ ϕ measured in the horizontal sections of the eye from median to lateral near the equatorial plane, and from ventral to dorsal in the sagittal plane, were generally constant and ranged from 1.8° to 2.3°. Table 3 demonstrates the uniformity of the angular values and the significant decrease in the angular values in the mediofrontal part of the eye, where the binocular overlap was greatest. Table 3 also shows the eye parameter p (D²/R).

Our results lead to the conclusion that the ommatidia of the eyes of Mantispa have a high light-gathering capacity, owing to an anti-reflection system made up of corneal nipples and a reflector consisting of tracheal tapetal layers on the base of the ommatidia. The corneal nipples increase in height from the periphery to the centre of the lens to compensate for the increase in the angle of incident light and the related degree of reflection. Furthermore, it is possible that the higher nipples operate on longer wavelengths, while the lower ones would be most effective for the shorter wavelengths received by the photoreceptors (see Bernhard et al. 1970). The striking tracheolar tapetum is similar to that of the closely related Chrysopa (green lacewing) (Ast, 1920; Horridge and Henderson, 1976) and is formed from thick tracheal ridges at the base of each rhabdom. As in the eyes of butterflies (Horridge et al. 1972; Ribi, 1979), it probably serves to return light that has already passed the rhabdom, thus effectively doubling the rhabdom’s length.

In Mantispa, the visual field and the binocular overlap are relatively large. Mantispa is similar in appearance and mode of capture of prey to the praying mantis, and it is therefore of interest to compare the present results with those obtained in mantids (Ciulfina, Mantis, Tenodera-. Horridge and Duelli, 1979; see also Horridge, 1980; Rossel, 1979, 1986). The horizontal overlap of the visual fields in Mantispa is of a magnitude similar to that of mantids, i.e. 48–70°. The praying mantis has a vertical binocular field of 240° compared with only 135.2° and 142° in Mantispa, which results in a reduction of the posterior blind spot (the area that the animal cannot see) in the praying mantis.

While the size and characteristics of the binocular field in Mantispa and the praying mantis are similar with regard to horizontal overlap, they differ in vertical overlap. As far as the size of ommatidial angles and their relationship to the part of the eye are concerned, there is a striking difference between the two species. In Mantispa, both the effective acceptance angles and the interommatidial angles seem to be rather constant over the entire eye. It should be noted that Ap of light-adapted ommatidia and A</> correlate very well, indicating a 1:1 ratio. Eye mapping in the equatorial plane also fails to show significant decreases in Δ ρ and Δ ϕ in the inner frontal eye region. Therefore, the situation in Mantispa is different from the situation in the praying mantis. In Tenodera, Rossel (1979) recorded a Δ ρLA of around 2° near the edge of the eye and in the inner frontal regions. The same is true of the interommatidial angles, which are of about the same order of magnitude as the acceptance angles. Similar values were found for Ciulfina (Horridge and Duelli, 1979). Thus, the praying mantis, unlike Mantispa, has a distinct acute zone in the frontal eye area.

This difference is especially apparent when one compares longitudinal sections through the eye in the horizontal plane in the equatorial region in Mantispa (Fig. 3) and the praying mantis (Horridge and Duelli, 1979; Horridge, 1980; Kirmse and Kirmse, 1985). In Mantispa, R, ommatidial diameter and cone length are constant throughout the entire section (Fig. 3), while in the praying mantis there is an enormous difference in R and cone length between the inner frontal region and the rest of the eye. The mantispid eye seems to be homogeneous from the optical and anatomical points of view, while the mantis eye shows extreme regional specialization. It is likely that mantispid eyes are too small and spherical to have a fovea.

Numerous reports on behaviour and optical measurements in praying mantis (e.g. Mittelstaedt, 1957; Maldonado and Barros-Pita, 1970; Rossel, 1979, 1980, 1983a, b, 1986) suggest that, in this animal, the acute zone is an important prerequisite for the binocular triangulation needed to estimate distances to prey.

Behavioural experiments under physiological conditions would be necessary to provide evidence of binocular vision in Mantispa. The prey-capturing behaviour of Mantispa under natural conditions and in the laboratory shows fixation and following movements (i.e. target following of the prey) which give the impression that prey capture is mainly binocularly controlled. It is interesting to note that Mantispa can catch not only prey located on a projection of its longitudinal body axis, which requires binocular distance estimation, but also prey considerably to the side of that axis. In this situation, Mantispa also uses both forelegs to strike, while the mantis often uses only one. Movement parallax may play a role in this behaviour, as it does in the praying mantis (see Rossel, 1983a, b, 1986).

Both Mantispa and the praying mantis show a distinct dependence of acceptance angles on adaptional state. In Mantispa, the effective acceptance angles of single ommatidia in both monocular and binocular regions decrease by a factor of 1.9 with illumination of the eye. In Tenodera (Rossel, 1979), mean Δ ρ in the light-adaptated eye is 0.7° in the acute zone in the frontal region, and 2.0° under dark adaptation at night. In the dorsal eye region, Ap increases from 2.4° to 6° with dark adaptation. In Mantispa and praying mantis (Stavenga, 1979) massive pigment shifts in the ommatidia are probably the reason for the illuminationdependent change in angle. Rossel (1979) tried to explain these pronounced adaptional changes on the basis that Tenodera uses its eyes in the daytime for hunting and at night for mating. In Mantispa this also seems to be true (McKeown and Mincham, 1948; New and Haddow, 1973), as we saw in Istria and in the laboratory, as well as in insects captured with light traps. When Mantispa is flushed from its roost into broad daylight, it flies immediately to the protection of another bush or tree. At night, the males seek mates and the females gather at a laying site; the latter often have to fly some distance to get there. Mantispa thus needs eyes that can perform specialized tasks around the clock. In this context, it is interesting that in Mantispa there is a distinct difference in binocularity between male and female. As our data reveal, Mantispa males have a degree of overlap in the dorsal direction that is greater by a factor 1.2 in the horizontal and by 1.1 in the vertical direction than in females. In the male, there is a second, more pronounced, overlap maximum above the equator. No such gender-specific differences are known from praying mantis, but Kirmse and Kirmse (1985) briefly mention that, in Mantis, males have a greater percentage area of frontal acute zone than do females. The reason for sexual difference in Mantispa could be that the male uses binocular vision not only to capture prey but also to find a mate.

We thank Ms Brunegger and Ms Pongratz (Forschungsinstitut für Elektronen-mikroskopie und Feinstrukturforschung, Graz) and Mr Jurschitsch for collaboration. Special thanks to Karin Herbst for histological material and to Duian Devetak of the University of Maribor for help with collecting animals. We are also appreciative of Dr Labhart’s valuable comments. The study was supported by grant P6766 (to KK) from the Austrian Science Foundation.