The high degree of diversity among insects can be expected to be reflected in a great variety of sensory modes and capabilities. Extensive knowledge of the different types of sensory processing used by insects would provide insights into their perceptual worlds, and allow us to understand how the sensory systems function and how they relate to insect evolution and ecology. Not surprisingly, however, our knowledge is still limited because most investigations are restricted to insects that are easy to obtain, rear and maintain. Scientists interested in sensory research should be encouraged to focus more on insects with striking biology and behaviour, whether or not they are easy to handle. The mantispids, also known as mantidflies, mantis flies or mantid lacewings, are an example of such insects.
The family Mantispidae is exceptional among the 17 families of the order Neuroptera (Insecta). Five mantispid species are native to Europe, especially the Mediterranean area (Ohl, 2004). Worldwide, approximately 400 mantispid species have been described, most of them in tropical and subtropical regions (Aspöck et al, 1980; Ohl, 2004). They are predatory insects that feed on other living insects. Most species are relatively small, with a forewing length < 15 mm; however, in the largest species, the forewing length is greater than 35 mm (Ohl, 2004). Their close resemblance to the praying mantis (order Mantodea) is particularly noteworthy. Although not closely related phylogenetically to mantids (Svenson & Whiting, 2004; Wedmann & Makarkin, 2007), mantispids have evolved powerful raptorial forelegs as a result of similar selective pressures. The forelegs have an elongated thin coxa, a femur armed with spines and a tibia that can fold closely against it, resembling a jack knife. The distinctive habitus of mantispids is exemplified by Mantispa scabricollis (MacLachlan, 1875), a representative of the Mediterranean area (Fig. 1).
The hypermetamorphosis during development, with diverse larval stages, is another fascinating feature of mantispids. This complex development was first described by Brauer (1852, 1869) in a study of Mantispa styriaca (Poda, 1761) (= pagana Fabr.). Depending upon the species and body size, a female can lay a few hundred to several thousand eggs. The eggs (< 0.3 mm long) are on short stalks, and are attached, for example, to the underside of tree limbs and similar structures that provide protection from rain and direct sunlight. It is remarkable that many females may meet at the same location to lay their eggs (Schremmer, 1959). For example, in Istria (Croatia), a large number of M. styriaca females are observed on free-standing Pinus halpensis and Pinus pinea in an open pine wood, and on a single pubescent oak Quercus pubescens (Eggenreich, 1989; K. Kral, unpublished data). However, the aggregation site and oviposition location must not be the same, as could be found in the wasp mantidfly Climaciella brunnea (Say, 1824) (Redborg & Redborg, 2000).
First-instar larvae of species of the subfamily Mantispinae actively seek spiders as hosts or wait for them and then parasitize their eggs. For example, in extensive studies of the life cycle of mantispids, Redborg & MacLeod (1984, 1985) observe that larvae of Mantispa uhleri Banks climb onto female spiders, settling preferentially on the pedicel (between the cephalothorax and abdomen), and then wait for the production of eggs, or cling to the membrane between the edge of the spider carapace and the base of the legs. Mantispa uhleri larvae that remain on spiders over the winter, until eggs are produced in the spring, are observed feeding on spider haemolymph. However, larvae of M. viridis Walker, for example, do not feed on spider haemolymph. The mantispid larvae move to a spider egg sac when the egg sac is being formed, or move to and penetrate an almost completed egg sac. Pointed suckers formed from modified mandibles and maxillae are used in both haemolymph feeding and egg sac penetration (Fig. 2; Kral, 1989). Almost every family of hunting spider is parasitized, including Lycosoidea and Clubionoidea; however web-spinning spiders can also be potential hosts (Gilbert & Rayor, 1983; Schremmer, 1983; Redborg & MacLeod, 1985; Hoffman & Brushwein, 1989; Hirata & Ishii, 1995; Guarisco, 1998). Mantispid larvae of species of the subfamily Symphrasinae can feed on the eggs, larvae or pupae of wasps and other hymenopterans; it should be noted that some mantispids, such as Anchieta fumosella (Westwood, 1867) and Trichoscelia santareni (Navás), benefit from morphological wasp mimicry or chemical camouflage in the adult stage (Parker & Stange, 1965; Dejean & Canard, 1990; Beck, 2005; Buys, 2008).
In rearing experiments with M. viridis, Davidson (1969) reports that the habitus of the larva changes dramatically as it grows in a unique environment, inside the spider egg sac. The well-developed legs and other extremities of the first-instar larvae disappear and the size of the head increases. The second-instar larvae resemble a fat, immobile maggot with a plump body form. The third and last larval stage pupates within the larval skin, thus being enveloped in three layers of protection (the egg sac, jelly coat and cocoon). A Mantispa cocoon may entirely fill the egg sac, which means that no eggs or spiderlings survive (Killebrew, 1981). However, in the case of M. decorate Erichson, a number of spiderlings may survive in an infested egg sac (Capocasale, 1971). The body size of M. uhleri, for example, increases rapidly during the approximately 18 days of development before the pupal stage is reached; under laboratory conditions, unfed first-instar larvae measure approximately 0.9 mm, whereas mature third-instar larvae exceed 10 mm. The pupal stage then lasts approximately 10 days. Under laboratory conditions, adults of M. uhleri (forewing length 8–16 mm) are found to live a mean of 114 days (Redborg & MacLeod, 1985).
The first-instar larvae of mantispids have simple, single-lens eyes, referred to as stemmata. On each side of the head, there is a single posterior eye, and a pair of eyes located anterolaterally (Fig. 2) (Kral, 1989). The number of stemmata differs in the larvae of species of other neuropteran families. For example, the Ascalaphidae (owlflies), Chrysopidae (green lacewings), Hemerobiidae (brown lacewings), Osmylidae (osmylids), Sisyridae (spongillaflies, with aquatic larvae), Myrmeleontidae (antlions), Nemopteridae (spoonwings or spoon-winged laceflies), Nevrorthidae (with aquatic larvae) and Nymphidae (split-footed lacewings) have six stemmata in a group on each side of the head, located dorsally or dorsolaterally, or six dorsal or dorsolateral stemmata, with a seventh located ventrally. The latter is usually reduced or rudimentary (Jokusch, 1967; Henry, 1976; Popov, 2002; Beutel et al, 2010a; Devetak et al, 2010). In the Myrmeleontidae, Myrmecaelurus trigrammus larvae have eight pairs of stemmata (Mirmoayedi, 2008). However, in the larvae of species of the families Dilaridae (pleasing lacewings), Ithonidae (moth lacewings) and Polystoechotidae (giant lacewings) there is only one pair of stemmata, or none at all (Grebennikov, 2004). Larvae of species of the family Coniopterygidae (dustywings) have five stemmata located on each side. A general overview of the stemmata of holometabolous insects, with reference to neuropterans, is provided by Gilbert (1994).
As shown by longitudinal sections through the stemmata of mantispid larvae (Kral, 1989; Fig. 3), each eye possesses a corneal lens and a bell-shaped crystalline body, composed of two Semper cells. Directly below the dioptric apparatus is the retina, which is two- or three-tiered. Distal to proximal consecutive serial sections reveal 12 elongated photoreceptor cells. The central fused rhabdom, which is photosensitive, appears to be always composed of four photoreceptor cells, arranged radially around the visual axis of the eye. The rhabdom in the first cell tier almost completely fills the space under the dioptric apparatus. The orientation pattern of the rhabdomeric microvilli remains constant in all tiers. Each photoreceptor cell narrows proximally to an axon, and the bundle of axons forms the optic nerve. The pathway of the optic nerve has not yet been determined. In the case of larval scorpionflies, Melzer (1994) reports that the nerve of the stemmata enters the brain anteroventrally and proceeds dorsally along the lateral surface of the brain before reaching the lamina and medulla neuropils. (A lobula neuropil is lacking; see discussion of the optic lobe below.) It would be interesting to determine whether the photoreceptor cells of the different layers of the retina terminate in different areas. Such an arrangement could indicate photoreceptor cells with differing properties and functions, as is found in the case of stemmata of butterfly larvae (Ichikawa, 1991) and water beetle larvae (Mandapaka et al, 2006; Stecher et al, 2010).
It is particularly interesting that stemmata are absent in the second-instar larvae. As observed by Davidson (1969): ‘the eyes and sclerotized areas were lost and the legs much reduced’. Thus, beginning with the second instar, vision can no longer play a role in larval activity.
What role is played by the stemmata of mantispid larvae? With only three stemmata on each side compared with six, seven or eight in the larvae of other neuropteran species, the visual field might appear to be limited. However, the reduced number of stemmata may be compensated for by their positioning, and differing orientations. On each side, the two anterolateral stemmata are positioned so that they face mainly forward, whereas the posterior eye, clearly separated from the other two by a bulge, definitely faces backward. The visual angle of 110–120° for each eye is relatively large for stemmata. The visual fields of the anterolateral eyes overlap but do not appear to overlap with the visual field of the posterior eye. A consideration of the two anterolateral eyes raises the question of why they are arranged vertically. Does this have any functional significance? The vertical arrangement might play a role in distance determination via the use of retinal elevation, with objects in the visual field of the lower eye perceived as being closer than those in the visual field of the upper eye. A similar role for vertically overlapping visual fields is proposed by Zeil & Al-Mutairi (1996) for fiddler crabs living on a flat substrate on the seashore. It is likely that behavioural experiments with controlled visual inputs (e.g. achieved by painting single eyes with a lightproof coating), where the larvae approach and strike at an artificial host, could help to address this interesting question.
This hypothesis raises the further question of whether the axial tiers of the photoreceptor cells may play a role here. In the larvae of the water beetle Thermonectus marmoratus, where the eyes have a two-tiered retina, the presumably bifocal lens may produce two best images that are vertically displaced from one another (Stowasser et al, 2010). It is hypothesized that the two images tend to move toward each other as the item of prey in the field of view comes closer. This system sounds plausible and could make accurate distance estimation possible, despite the simple structure of the single-lens eyes.
However, in the case of neuropteran stemmata, nothing is known about the morphological composition and physical features of the lenses, or about how images are projected onto the retina. Investigations in this area would be interesting and could lead to discoveries, and perhaps new insights, into visual distance estimation. Although it is reasonable to assume that the stemmata of mantispid first-instar larvae are optimized and specialized for the extremely efficient host seeking and attacking behaviour, it should be noted that olfactory cues and mechanoreceptors could also play a decisive role in locating a host. In M. viridis and M. uhleri, Redborg & MacLeod (1985) observe that first-instar larvae are able to direct their movements toward spiders and egg sacs from a distance.
The compound eyes
Adult mantispids have two large hemispheric compound eyes (Fig. 1). Additional dorsal ocelli are absent; this is also the case in other neuropteran insects, with the exception of Dilaridae and Osmylidae species, which have three ocelli on the vertex (New, 1986; Beutel et al, 2010b). In M. styriaca adults, each compound eye is composed of approximately 3500 ommatidia, occupying almost 60% of the surface of a sphere with a radius of approximately 500 µm. The facets have a mean diameter of 20 µm; when viewed frontally, they appear as regular hexagons, similar to a honeycomb. Short bristles are located at intervals between the facets (Eggenreich & Kral, 1990) (Fig. 4).
Electron microscope investigations of the compound eyes (Kral, 1990; Kral et al, 1990) reveal that each ommatidium has a dioptric apparatus, which consists of a convex corneal lens completely covered with nipples (Fig. 4). Corneal nipples are also described in other neuropteran species (Schneider et al, 1978; Yang et al, 1998) and insect groups (Peisker & Gorb, 2010). The crystalline cone under the corneal lens is elongated and its core consists of four segments, each belonging to a Semper cell. The central part is optically dense and the peripheral mantle is optically thin. Around the crystalline cone, there are 12 accessory pigment cells. In the dark-adapted state, the pigment is completely proximal; with light adaptation, it extends to the base of the corneal lens. Two main pigment cells surround the proximal part of the crystalline cone, and may function as a pupil around the tip of the crystalline cone. The dioptric apparatus and the retina are separated by a distinct clear zone that is approximately 170 µm thick in M. styriaca. The structural elements of the clear zone are the distal processes of the photoreceptor cells and the pigment-free part of the accessory pigment cells (Kral, 1990; Kral et al, 1990) (Fig. 5).
Each ommatidium consists of eight photoreceptor cells. Distally to proximally, the central fused rhabdom is initially formed only by the rhabdomere of the seventh photoreceptor cell, and then by the eighth photoreceptor cell, which completely displaces the seventh cell. From then on, cell 7 has no rhabdomere. The mean diameter of the rhabdom of these two small cells is 2 µm. At the base of the clear zone, the six large photoreceptor cells then form the rhabdom, 10–12 µm in diameter; the eighth cell is displaced more to the edge, although its rhabdomere extends almost to the middle of the rhabdom (Fig. 5) (Kral et al, 1990). With the aid of serial sections, it can clearly be shown that, initially, there is a rectangular rhabdom formed by cell 7, which then loses its shape as a result of the centrally located rhabdomere of cell 8 and the emergence of rhabdomeres of cells 1–6, which finally become a flower-shaped, six-lobed rhabdom. This type of retina, arranged in three tiers, is also found in the Chrysopidae (Horridge & Henderson, 1976; Yang et al, 1998) and Ascalaphidae (Schneider et al, 1978). At the base of the compound eye, there is a reflective tapetal layer composed of six flattened air-filled tracheoles in each ommatidium, isolated from neighbouring ommatidia, as is also found in the Chrysopidae and Ascalaphidae families.
Mantispid compound eyes are of the refracting superposition type. However, they differ considerably from the superposition eyes of insects that are exclusively crepuscular and/or nocturnal (Nilsson, 1989). For example, the relatively small interommatidial angles, or angular separations of the visual axes of adjacent ommatidia, of 1.8–2.2°, are attributable to small facet diameters and the relatively large radius of the eye. Furthermore, the clear zone is bridged by the distal processes of the photoreceptor cells and the rhabomeric part of cell 7. These structures may function as light guides as a result of their high refractive index (Kral et al, 1990) (Fig. 5). However, in addition to their optical importance, they may also play a role in mechanically fixing the rhabdoms, counteracting the elasticity of the tracheoles (Schneider et al, 1978). Similar features are also found in the diurnal superposition eyes of the Ascalaphidae (Schneider et al, 1978), Chrysopidae (Yang et al, 1998), Lepidoptera and Ephemeroptera (Horridge & Giddings, 1971; Horridge & Stange, 1972; Horridge et al, 1977; Wolburg-Buchholz, 1976), as well as the Mysidacea (Crustacea) (Hallberg, 1977). The resolution of such diurnal superposition eyes is thus comparable to that of some photopic apposition eyes. It should also be noted that the light sensitivity of diurnal superposition eyes is higher than that of apposition eyes with a comparable resolution. This not only is a result of the superposition optics, but also the effects of special structures, such as the corneal nipples (Fig. 4), which emerge from the epicuticular layer of the cornea. The corneal nipples presumably fulfil the physical conditions necessary to serve as an anti-reflection system for stimulus light (Dey, 2007). Another characteristic feature of the compound eyes of mantispids and other neuropterans is the reflective tapetal layer immediately below the retina. It consists of a network of fine, flattened tracheoles, which serves as a shining layer (tapetum lucidum) that doubles the light rays passing through the photoreceptors, thus improving their photon catch. The glow of the eyes, when dark-adapted, is caused by the tapetum lucidum (Eggenreich & Kral, 1990); in comparison, in the eyes of reptiles, birds and mammals, there are reflective tapeta defined by crystals or collagen fibres (Ollivier et al, 2004). The function of the short bristles found between the facets of mantispid compound eyes is unclear (Fig. 4). Interfacetal hairs, which are quite common in the compound eyes of insects, have diverse functions (e.g. mechanoreception: Neese, 1967; Chi & Carlson, 1976; Herbstsommer & Schneider, 1979; reception of infrared light: Neville, 1975; Gingl & Tichy, 2001; Dey, 2007).
The forward-facing part of the large compound eyes of mantispids has a large binocular visual field (Eggenreich & Kral, 1990). With the aid of a goniometer, by using the luminous pseudopupil arising from axially incident light (orthodromic illumination), a horizontal overlap between the visual fields of the two compound eyes, measuring more than 40° in females and more than 50° in males, is reported in M. styriaca (Eggenreich & Kral, 1990) (Fig. 6). In males, two overlap maxima are found dorsoventrally, at 0° (the eye equator), and at 30° above the eye equator; whereas, in females only one maximum is found, at approximately 20° above the eye equator. The first overlap maximum of males and that of females have a similar horizontal extension. However, the second maximum of males is approximately 10° larger. The binocular overlap has a dorsoventral extension of approximately 135° in females and more than 142° in males. In males, the total area of the binocular overlap is therefore 9.5% larger than that of females. In both males and females, the binocular overlap begins at approximately 45° below the eye equator. By contrast, the monocular visual field is approximately the same in both sexes. The backward-facing blind area of the eyes, which is out of the field of view, is quite small. It is approximately 12% smaller in males than in females. The extensive visual field demonstrates the almost panoramic view of the compound eyes.
What is currently known about the spectral sensitivity of the compound eyes of mantispids? Spectral sensitivity curves derived from electroretinogram measurements on M. styriaca show a prominent maximum in the green region, with a peak at the 546 nm wavelength of stimulus light (Mayer & Kral, 1993) (Fig. 7). The stimulus threshold is the lowest for wavelengths in this range, significantly lower than for short wavelengths. This type of dominant green sensitivity, with a peak at a similar wavelength of stimulus light, is also found in the compound eyes of the neuropteran insect Chrysoperla carnea (Kral & Stelzl, 1998). In the mantispid, the green receptor is most probably the rhabdom of the large cells 1–6 (Fig. 5). The size, structure and position of cells 7 and 8 suggest that they may form a system for short wavelengths. Because the morphological features are uniform over the entire hemispherical eye, all eye regions can be expected to have the same spectral sensitivity, with no difference between the dorsal and ventral parts of the eye, as is found in diurnal owlflies (Gogala, 1967; Schneider et al, 1978; Fischer et al, 2006).
The optic lobes and brain
In the compound eyes of insects, the axons of the six large photoreceptor cells (presumably green-sensitive in the mantispid) generally project into the neuropil of the first optic ganglion, referred to as the lamina, and form synaptic contacts with the monopolar cells (first-order interneurones). There are also other interneurones, local interneurones and terminals of centrifugal cells (Strausfeld & Nässel, 1981; Kral, 1987). In mantispids, however, it is not yet known how far the dendrites of the monopolar cells extend across the many terminals of the photoreceptor axons. In addition, the spatial arrangement of the compound eye should be maintained. For insects (Strausfeld & Blest, 1970; Strausfeld & Nässel, 1981), the monopolar cells project to the neuropil of the second optic ganglion, referred to as the medulla. The axons of the two small (ultraviolet-sensitive) photoreceptor cells also project to the medulla. In the neuropil of the medulla, as well as complex intrinsic neuronal circuits, there are incoming and outgoing neurones that project to the neuropil of the lamina, or to the neuropil of the third optic ganglion, referred to as the lobula, or to the neuropils of the optic lobe of the other compound eye. Thus, the medulla, the largest and most complex component of the optic lobe, appears to process all of the visual information received by the eye. The main function of the lobula is the neural coding of spatial and motion information (via the giant tangential neurones). This is also the case in the lobula of superposition eyes (Horseman et al, 2011). From the lobula, fibres project to the dorsal neuropil of the lateral and medial protocerebrum, from which descending neurones proceed to the neuropils of the neck, leg and flight motor systems. Although these aspects have been intensively investigated in flies (Borst et al, 2010), this is far from the case for mantispids and other neuropteran insects. Insights into the neural basis of the remarkable visual behaviour of mantispids and other neuropterans could provide new perspectives.
Daily pattern of visual activity
Mantispids are active during the day, with orientation flights, prey capture and mating (Eggenreich, 1989; Kral et al, 2000; Beck, 2005; D. Devetak, personal communication). In Mediterranean mantispids, such activities can be observed near bushes, conifers and deciduous trees, often free-standing. Nevertheless, artificial night lighting, as is used in light traps, can attract M. styriaca and other mantispids, if air temperatures are sufficiently high (New & Haddow, 1973; Ábrahám & Papp, 1994). This indicates that mantispids are also active at night under good weather conditions. Observations of non-European mantispids at light sheets show that they are active predators, opportunistically capturing a wide range of prey of dimensions suited to their individual body size (J. Oswald, personal communication). Based on 24-h experiments on M. styriaca concerning diurnal activity patterns under controlled laboratory conditions, Ábrahám & Mészáros (2006) argue that positive phototaxis at night is an abnormal, light-dependent behaviour. This could mean that artificial light triggers a positive phototactic response and interferes with the natural behaviour patterns of mantispids, such as resting in nearby vegetation. The same also applies to diurnal mantids (K. Kral, unpublished data). From this perspective, it might be presumed that an insect is exclusively nocturnal only when activity can never be observed during the daytime.
Mantispids are sophisticated ambush and stalking predators. Their predatory behaviour is highly dependent upon vision, as is also the case in the praying mantis. In investigations of M. styriaca, it is possible to learn something about their prey-capture strategies (D. Devetak, personal communication; K. Kral, unpublished data; Kral et al, 2000). On bright, hot sunny days, mantispids wait in ambush for insect prey (e.g. on a pine or deciduous tree). Mantispids, such as C. brunnea, also hunt in lower vegetation and on flowers (Redborg & Redborg, 2000). In the morning and afternoon, they prefer a warm, sunny place, whereas, around mid-day, they are usually found in more shaded locations, presumably to regulate their body temperature. When waiting in ambush, mantispids either stand or hang upside down, with the head raised, the forelegs drawn up close to the thorax, and the antennae continually oscillating back and forth. From time to time, mantispids may move to another strategic location, where they again assume the ‘ambush’ posture. Insects that are suitable as prey generally have a body size of 2–12 mm; adult antlions, even as large as mantispids, are also appropriate prey. As soon as an insect appears, walking, running, hopping, fluttering or flying around, the mantispid turns its head and long prothorax toward the moving object. With moving flies as targets, reaction times of 60–160 ms have been measured between the initial lateral stimulus and the beginning of the turning movement, at air temperatures of 27–31 °C. Such fast reaction times mean that the visual system must be able to process the visual stimuli arising from passing flies very quickly. Accordingly, the prominent green sensitivity of the compound eyes is significant (Mayer & Kral, 1993). From bees and other insects, it is known that it is primarily the green-sensitive photoreceptor cells that are responsible for fast movement reactions; indeed, they are the fastest photoreceptors in insects. For example, in bees, green-sensitive photoreceptor cells react to sufficient light stimuli in less than 8 ms (Skorupski & Chittka, 2010).
It is evident that the closer the insect prey, the higher the probability it will be detected by the mantispid. This relationship is demonstrated in Fig. 8 (Kral et al, 2000). With flies as prey, calculations have also shown that, for an item of prey to be detected, its image must cover a critical visual angle of the eye (Kral et al, 2000). In accordance with the critical visual angle, approaching flies are detected later if they are smaller than if they are larger. Furthermore, flies of equal size are detected later if they approach from the side than if they approach from the front. Thus, the critical visual angle is larger in the lateral region of the eye (4–10°, corresponding to two to five ommatidia) than in the frontal region of the eye (4–6°). The situation for insect prey that exhibit less contrast than flies and that also have shapes and movement patterns differing from those of flies and, hence, are perceived differently by the eyes of mantispids, is not yet known. In mantids, such parameters can have a significant influence on the recognition and detection of prey (Kral & Prete, 2004). One reason for the difference between the frontal and lateral field of view could be a difference in the spatial and/or temporal resolving power of the frontal and lateral regions of the eye. Even though the structure of the hemispheric eye is quite uniform, there could be differences in the neuronal organization of the frontal and lateral parts of the optic lobe with respect to spatial and/or temporal vision. As noted above, light also plays an important role in prey-capture behaviour. However, the effect of light on mantispid reaction distances has not yet been systematically studied. Based on the structure of the eye and the retina, it can be expected that spatial acuity will increase significantly with increasing light.
Behavioural studies have shown that if the insect prey does not approach, the mantispid may exhibit stalking behaviour but only as long as the insect is moving past or moving away (Kral et al, 2000). The probability that stalking behaviour will occur increases exponentially as the predator–prey distance decreases (Fig. 8). As soon as the insect prey stops moving, the mantispid also stops, possibly to avoid eliciting an escape response or to keep the prey in a constant position in the visual field. During such stops, the mantispid sometimes executes small rocking movements. It is unclear whether these movements are related to range estimation, as is the case in the praying mantis (Poteser & Kral, 1995; Kral, 2003). Stalking behaviour ceases as soon as the prey moves too far away (Fig. 8). However, if the prey remains within the detection range, the mantispid continues to execute tracking movements with its head and thorax. If the prey moves beyond the critical visual angle and can no longer be detected by the mantispid, the mantispid alternately stretches its raptorial legs forward several times and may then exhibit grooming and pitching movements, with its forelegs reaching behind its back. In the praying mantis, such movements have been interpreted as an expression of displacement behaviour (Zack, 1978).
When the item of prey enters the capture range, the mantispid immediately stops stalking and assumes a position in which its forelegs, head and thorax are directed exactly toward the centre of the item of prey. Thus, immediately before the lunge and strike, the prey is precisely centred within the binocular visual field, in the corresponding region of the left and right compound eye. To achieve this, if necessary the mantispid makes small corrections of body alignment, moving a little forward and backward, as well as laterally. Because different corresponding ommatidia are stimulated at different object distances, it may be presumed that the information provided to the nervous system can be used for range estimation. Video analyses indicate that the lunge and strike are elicited primarily when the item of prey is located at a distance of 10–15 mm from the mantispid (Fig. 8), and that this is the case regardless of the size of the prey (Kral et al, 2000). It is remarkable that if one compound eye is completely painted with a lightproof material, the mantispid appears to modify its strategy by adding translatory scanning movements with the head and thorax, provided that the prey is still (K. Kral, unpublished data). Thus, it appears that if information from corresponding ommatidia of the left and right eye is insufficient to determine the striking range, the mantispid can compensate for this with other cues such as self-induced retinal image motion. This is supported by the fact that prey capture can also be successful in such situations, although, for fast-moving prey, the success rate is much lower than it is under binocular conditions. The question arises as to whether and how these visual cues are interrelated (Kral, 2003).
To capture the prey, the mantispid lunges forward, often so far that the standing legs are stretched back to a great extent. The strike is executed with both forelegs moving forward from behind the head (Kral et al, 2000) (Fig. 9). Video analyses have shown that the prey-capture movement is extremely fast, and can take less than 60 ms. However, the strike, unlike that of the praying mantis, is a stereotyped, ballistic action (Kral, 1999; Prete & Hamilton, 1999; Kral et al, 2000). This means that, in mantispids, the strike is always performed exactly in the medio-sagittal plane, with the same sequence of movements, and at the same speed, regardless of the size and mode of movement of the prey. Unlike the praying mantis, mantispids are not capable of executing lateral strikes. This may be a result of the different morphological structures that are responsible for the strike. By contrast to mantispids, the subcoxal muscles of the praying mantis include abductors, redactors, rotators and subinators, which allow strikes to be executed in different planes (Ulrich, 1965; Frantsevich, 1998; Bullaro & Prete, 1999; Frantsevich & Wang, 2009).
Not all strikes are successful. In the case of relatively slow-moving prey, such as adult antlions, moths and butterflies, the prey-capture success rate is high. However, in the case of fast-moving prey such as flies, which can recognize the beginning of a strike movement within approximately 15 ms, and require only 30–50 ms to execute an escape manoeuvre (N. Franceschini, personal communication; Holmqvist & Srinivasan, 1991), only one in four or five strikes succeeds in reaching the target. Because of the stereotyped nature of their strikes, unlike the praying mantis, mantispids are not capable of outmanoeuvring a fly by striking into the escape direction.
Because mantispids usually hunt in areas with bright green vegetation, they may benefit from the fact that their compound eyes are highly sensitive to green light (Fig. 7). This makes the green background appear especially bright, such that the brightness enhances the contrast and visibility of items of prey, particularly if they are dark in colour (Mayer & Kral, 1993). After a mantispid has captured an item of prey with its forelegs, it tries to turn the body of the prey and starts eating at the head in the case of moths, butterflies or antlions, or at the ventral or lateral thorax (flight musculature) in the case of flies. During this procedure, items of prey, such as fleshflies, which make a strong defence with vehement flaps of the wings, can be lost. However, once this critical first step has been successfully completed, the item of prey is firmly fixed between the femur and tibia, and the forelegs remain closed, and further defence by the prey is of no avail.
Vision is an important cue that triggers mating behaviour in mantispids (Eltringham, 1932; Redborg & MacLeod, 1985; Redborg, 1998). Facing one another plays a significant role in mate location and courtship; before moving and coming together, mates face each other for a few minutes to several hours, apparently trying to determine whether there is any risk of danger. However, it is the male that takes the initiative in approaching the female. For example, for nine pairs of the tiny neotropical mantispid T. santareni (forewing length < 10 mm), Dejean & Canard (1990) report that the male approaches the stationary female face-to-face until reaching a distance of 4–5 cm, which is sufficiently far away to eliminate the possibility of a physical attack. The male then stops and begins to wave its raptorial forelegs slowly, and the female responds by also waving the forelegs. Similar courtship behaviour, with complex motion pattern of the forelegs, is observed in M. uhleri (Redborg & MacLeod, 1985). This behaviour lasts for several minutes and appears to be a clear signal for mating because the male then comes to the side of the stationary female.
With regard to the role of vision in this behaviour, the question arises as to how the male is guided so precisely toward the female. How does the male determine when to stop or, in other words, how can the male estimate the critical distance to the female? It may be that when moving continuously forward, the male uses the increasing size of the retinal image on its compound eyes as a cue. Because the body size of the females is quite constant, the males could make use of the fact that a particular image size corresponds to a certain distance. Collett & Land (1975) demonstrate the use of image size as a visual cue for absolute distance estimation in males of the hoverfly Syritta pipiens L. (body length 7–10 mm) during the pursuit of a female. They also show that the male controls its body position relative to the female with the aid of the image position in both compound eyes. Because of the binocular overlap of the compound eyes of the male, at a distance of 10 cm, an image of the pursued female appears on both eyes, and, at a distance of 7 cm, the images on the two eyes begin to merge. In mantispids, the male could thus benefit from having a large binocular visual field (Fig. 6) with extensive binocular vision in the horizontal plane (Eggenreich & Kral, 1990).
Dejean & Canard (1990) report that, after the mantispid male reaches the side of the female, both sexes then rotate back-to-back. During copulation, which can last from under an hour to several hours, the male and female are facing away from each other, and it may happen from time to time that one moves forward, pulling the other, or vice versa. This copulatory position is similar to that in the lovebug, Plecia nearctica, a dipteran (Thornhill, 1976). After copulation, the male usually goes away, whereas the female remains motionless or exhibits grooming behaviour.
The fact that mantispids visit particular locations such as free-standing trees (D. Devetak, personal communication; K. Kral, unpublished data; Eggenreich, 1989) for the purpose of feeding or mating, arriving at these locations from a distance, demonstrates their spatial orientation capability in a larger environment. However, it is not known how such aimed travelling is controlled. A prominent tree, perhaps, standing in a clear, exposed area, serves as a landmark that guides mantispids over longer distances. From a biological point of view, the question arises as to how mantispids can recognize a suitable tree, in other words a location with sufficient insects to serve as prey and spiders to act as hosts for the larvae. Perhaps mantispids and other insects are simply attracted by trees and other plants that are in bright, sunny locations and are fragrantly scented. However, local travel also requires an overview of the spatial surroundings. For example, if mantispids are disturbed, they fly up and around, often for several metres, to find new shelter or to return to the original location (K. Kral, unpublished data). In this case, vision clearly plays an important role in determining the flight path.