Zool. J . Linn. SOC.,58: 1-37. With 2 plates and 8 figures January 1 9 7 6 The defensive behaviour of Ghanaian praying mantids with a discussion of territoriality MALCOLM EDMUNDS* Department of Zoology, University of Ghana Accepred for publication March 1975 The defensive behaviour of 18 species of mantids is described and further details are given of the defences of 12 species whose behaviour was described in an earlier paper. As a result the defences of 3 7 genera of Ghanaian mantids are now known. Most species of Mantinae have a general resemblance to their background and in Sphodromanris lineola and Miomantis paykullii there is a correlation between colour of background and colour of insect. Most other species of mantid have a highly specific resemblance t o bark, grass, sticks or leaves, and the evolution of these insects is discussed. Fire melanism occurs in Galepsus foganus and Pyrgomanfis pallida. Evidence is presented which suggests that the selective factor leading to reduction of wings in females of many species is predation by birds. Ten species of mantid have first instars which mimic ants, and in S. lineola first instars are positively associated with Oecophylla, the ant they most closely resemble. Tamchodes afzelii not only has nymphs which resemble ants but it preys extensively on ants as well. The functions of startle displays and the evolution of ocelli in rnantids are discussed, and four types of territorial display which result in spacing out of nymphs are described. CONTENTS Introduction . . . . . . . . . Descriptions of defensive behaviour . . Summary of defensive adaptations . . Discussionof primary defence . . . . Body colour polymorphism . . Leafmimicry . . . . . . Barkmimicry . . . . . . Grassandstickmimicry . . . Reductionof wings . . . . . Defence against ants and ant mimicry Discussion of secondary defence . . . Function of the startle display . Variation in the startle display . Evolution of ocelli . . . . . Territoriality in mantids . . . . . Summary . . . . . . . . . . Acknowledgements . . . . . . . Note added in proof . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Present address: Department of Biology, Preston Polytechnic, Preston PRl 2TQ. 1 . . . . . . . 1 2 4 17 17 17 21 22 23 2 30 30 31 31 32 35 35 35 35 M. EDMUNDS 2 INTRODUCTION The anti-predator defensive behaviour of 25 species of Ghanaian praying mantids has been described recently (Edmunds, 1972). In this paper I report on the defensive behaviour of an additional 18 species of Ghanaian mantids, and further observations are also presented on several of the species described in my earlier paper. As a result of these two papers, the defensive behaviour of 37 out of the 49 genera of praying mantids recorded from Ghana has now been described. The classification adopted is that of Ragge & Roy (1967) except for the transfer of Oxypilus from the Mantidae to the Hymenopodidae (Edmunds, 1972). In addition the species which I describe here under the name Catasigerpes occidentalis is listed by Ragge & Roy (1967) as C.toganus. The reasons for this change of name are given under the section on this species. Ghanaian mantids whose anti-predator behaviour has now been described are thc following: Family Subfamily Species Amorphoscelidae Amorphoscelis maculata * Amorphoscelis lagrecai Paramorphoscelis gondokorensis Thespidae Hoplocorypha nigerica* Mantidae Sibyllidae Tarachodinae Tarachodes afzelii * Galepsus toganus * Pyrgomantis pallida* Liturgusinae Theopompella westwoodi Theopompella chopardi Iridopteryginae Negroman tis modesta * Nilomantis edmundsi Oxyothespinae Oxyothespis longipennis Angelinae Leptocola phthisica* Mantinae Plistospilo ta guineensis * Cataspilota misana Proh ierodula orna tipenn is Polyspilota aeruginosa * Tenodera superstitiosa * Sphodromantis lineola * Sphodromantis aurea Paraman tis prasina * Paramantis togana Mantis religiosa* Statilia apicalis Miomantis paykullii* Miomantis aurea * Sibylla limbata DEFENSIVE BEHAVIOUR O F PRAYING MANTIDS Hymenopodidae Epaphroditinae Phyllocrania paradoxa * Hy menopodinae Pseudoharpax virescens * Chloroharpax modesta Panurgica compressicollis Chlidonoptera lestoni Pseudocreobotra ocellata * Acromantinae Oxypilus hamatus* A nasigerpes b ifasciata * Catasigerpes occidentalis * Chrysoman tis speciosa Chrysomantis cachani Vatidae Stenovates strachani" Danuria buchholzi* Popa undata Empusidae Idolomorpha lateralis * Hemiempusa capensis* 3 In this list, species whose behaviour has been described in my earlier paper are marked with an asterisk, though for some of these further information is presented here. Mention is also made in the text to the following Ghanaian mantids, but their defensive behaviour has not been studied in detail: Amophoscelis laxeretis *, Compsothespis occidentalis, Miomantis lam toensis and Chlidonoptera chopardi? As in my earlier paper, two types of anti-predator behaviour are recognised: primary defensive adaptations, which reduce the probability of a predator initiating a prey-capture attempt, and secondary defensive adaptations, which operate only after a prey-capture attempt has been initiated and which reduce the chances of its being successful. There are two principal types of primary defence: a general resemblance to the background so that the insect is camouflaged, and a highly specific resemblance to part of the environment such as a stick or a ,leaf. These specific resemblances are here considered to be examples of mimicry and are named in accordance with the object resembled, e.g. stick mimicry, grass mimicry, leaf mimicry and bark mimicry. A study of the different species of stick and grass mimics reveals that there are two very different ways in which the resemblance can be attained, so it is useful to distinguish short stick and grass mimics from long stick and grass mimics. The evolution of these different forms of mimics is discussed in this paper. In the terminology of Cott (1940) all of the mantids described here are cryptic, not mimetic. Resemblance to a plant, such as a stick, implies crypsis, whilst resemblance to an animal implies mimicry. However, the resemblance to a stick or a leaf can be very precise and hence comparable in morphological specialization to mimicry of another animal. Wickler (1968) and Edmunds (1974) define batesian mimicry in terms of its effects on a signal receiver, which in this context is the predator. Following Wickler, if an animal produces a counterfeit signal which deceives a signal receiver, this is mimicry. Cryptic animals rely for protection on producing few or no signals, hence they are not 4 M. EDMUNDS easily found by predators. Mimetic animals, by contrast, are detected by predators, but the signals which they produce cause the predators to refrain from attacking them. This distinction may be valid in theory, but since we know so little about how predators recognise prey it breaks down in practice. Thus if one accepts this distinction, then stick mimics and leaf mimics are probably truly mimetic since they rest fully exposed to a predator, but are not apparently recognised as being insects. Bark mimics, however, merge into their background, and hence they are simply extreme examples of crypsis. But it is very difficult to decide if grass mimics such as Leptocola and Pyrgomantis are mimetic or cryptic: we just do not know what predators perceive when they encounter such a mantis in a natural situation. In this paper therefore I have called all forms of extreme resemblance “mimicry”, although it should be understood that the principles by which the animal escapes being eaten may not be the same in all cases. Secondary defensive adaptations include active escape by running, jumping or flying, death feigning (thanatosis), startling or frightening displays, and overt attack to which all species resort when seized by a predator. In the discussion at the end of this paper consideration is given to the function and evolution of displays: whilst many displays function to intimidate potential predators, displays may have other functions as well, for example, spacing out individuals. Several species of mantid exhibit some form of territorial behaviour which results in spacing out of individuals and hence lessens the chances of cannibalism occurring, and this form of behaviour is also discussed. Observations are also reported on the defences of nymphs and these include a survey of ant mimicry amongst first-instar mantids. DESCRIPTIONS OF DEFENSIVE BEHAVIOUR 1. Amophoscelis lagrecai Roy, 1964 Two females examined. A forest species. Arboreal. Primary defence. A greenish brown, mottled, dorsoventrally flattened mantis which lives on tree trunks as do other species of the genus (see Edmunds, 1972). Secondary defence. When poked both insects ran quickly or flew, as did A . maculata Roy (Edmunds, 1972). No startle display could be elicited. 2. Paramorphoscelis gondokorensis Werner, 1907 Two males examined. A savanna species. Judging from its colour and resting posture it is probably arboreal. Primary defence. This species is grey (becoming browner when pinned), elongated, and is a short-bodied stick mimic with similar defences to the short grass mimic Pyrgomantis (Edmunds, 1972). When placed on a stick it rests with the body closely apposed to the stick and the legs held close to the body. The head is held in the prognathous position with the mouthparts directed forwards (Fig. 1). Thus the head and body taper gradually to the stick and there is no conspicuous recognition mark by which a predator lateral to the mantid might DEFENSIVE BEHAVIOUR OF PRAYING MANTIDS 5 Figure 1 . Primary defence posture of male Paramorphoscelis gondokorensis: a short-bodied stick mimic. Drawn from a colour transparency. recognise that a mantid is there. The head also has two posterior projections which conceal the neck from dorsal view. Paramorphoscelis rests either with the head upwards or downwards. Secondary defence. When disturbed Paramorphoscelis usually runs up the stick or dodges round to the other side and then becomes motionless, just as does Pyrgomuntis. It may also fly, especially in the unnatural situation of being disturbed when on a flat surface rather than on a stick. There is no startle display. 3. Hoplocorypha nigerica Beier, 1930 One female examined. The defences of the male have already been described (Edmunds, 1972). Primary defence. The female is a grey-brown stick mimic with long legs and two projections on the back of the head to conceal the neck from dorsal view, just as in the male, but with a much fatter abdomen. There is a mid-dorsal black line on the abdomen. When disturbed the male protracted the forelegs so as to increase the resemblance to a stick, but this occurred only very rarely in the female. Secondary defence. When poked the female ran or hopped, just as did the male, but since she is apterous, she cannot fly. The foreleg femora are black on 6 M. EDMUNDS the inner surface, and the ventral surface of the thorax is crimson, suggesting that these colours may possibly have significance in some display, but no display was ever seen. They are unlikely to be flash colours since they are not easily visible when the mantis moves. 4.Theopompella westwoodi (Kirby, 1904) One male examined. A forest species. Arboreal. Primary defence. This mantis is dorso-ventrally flattened with the forewings having the coxal area angled from the rest of the wing and normally lying lateral to the abdomen when at rest. In this way the contour of the insect is smoothed from either anterior or posterior view, and any lateral shadow is obscured (Fig. 5 , Plate 1B). The insect rests on tree trunks, most usually with the head facing downwards. As in Taruchodes, the forelegs are held slightly abducted so that they too smooth the contour from head to substrate, but the head is in the prognathous position with the mouthparts directed forwards, unlike Turachodes. The body and wings are mottled with dark brown, pale brown and cream, and there are disruptive bands on the forelegs, so the insect is a bark mimic. Secondary defence. When disturbed Theopompella runs or tries to fly, or it may slash with the forelegs at the object attacking it. No startle display was seen, but the forelegs are brightly coloured on their inner (ventral) surface: the coxa has a dark brown spot proximally, and is then cream merging to mauve distally; the femur is mostly dark brown with a cream band followed by brown and then white distally; the tibia is cream with a mauve band half-way along its length. This colour must have some function, though it need not be in a startle display. 5. Theopompella chopardi Roy, 1963 One male examined. A forest species. Arboreal. Primary defence. The resting posture of this species is exactly as in T. westwoodi with prognathous head, slightly abducted forelegs, and wings concealing lateral shadow. If disturbed it flattens itself still more closely t o the bark. Secondary defence. When poked it opened its wings as if about to fly, but it did not display. The wings are transparent with purple-brown and white veins, except that the bases of the hind wings are grey. The abdomen is dark brown dorsally. This lack of bright colours further suggests that no startle display occurs in this species. 6. Nilomantis edmundsi Roy, 1975 Numerous males examined. A savanna species. Probably arboreal. This species is very close toNilomuntis floweri, Werner 1907, from East Africa, Arabia and Sudan (Roy & Leston, 1975). Primary defence. This is a very small species with bright green body (18 mm long, but with wings protruding a further 2-3 mm) and legs. (The legs become DEFENSIVE BEHAVIOUR O F PRAYING MANTIDS 7 yellow in pinned insects.) There is a yellow stripe mid-dorsally on the thorax and also on the abdomen, but this latter is normally hidden by the pale green wings. Thus the mantis has a general resemblance to green vegetation. Secondary defence. When disturbed this species almost invariably flies away. One male, however, was observed to give an alternate left and right “boxing” display, similar to that of Cutasigerpes (Edmunds, 1972), with both legs extended synchronously first to the left, then to the right. No startle display was observed. I t is not known if this boxing occurs regularly in Nilomantis, and if so whether it should be regarded as a defensive response or if it has some intraspecific function as in Oxypilus and Catasigerpes. 7 . Oxyothespis longipennis Chopard, 1941 One male examined. A savanna species which probably lives amongst grass. Primary defence. An elongated grass mimic with long legs and no unusual resting posture (Fig. 2). It is buff in colour and highly cryptic on dead grass. The forelegs are often held slightly protracted, but not in line with the head as occurs in Hoplocorypha. Secondary defence. When disturbed Oxyothespis either remains almost motionless (i.e. moving minimally to one side or downwards to avoid being poked) or it flies. The wings are transparent and there are no bright colours on the feeble forelegs, so probably there is no startle display in this species. The related Compsothespis occidentalis Sjostedt, 1930, is another elongated brown mantid with even more feeble forelegs than Oxyothespis (see fig. 30 of Gillon & Roy, 1968). I t has the head and thorax shaped so as to conceal the neck from dorsal view (as in Danuria and Hoplocorypha described by Edmunds, 1972), and it is probably an arboreal stick mimic. The hind wings, h Figure 2. Primary defence posture of male Oxyothespis longipennis: a long-bodied grass mimic with no unusual cryptic posture. Drawn from a colour transparency. 8 M. EDMUNDS however, are orange basally and black distally suggesting their use either in a display or (more likely) as flash colours. Unfortunately the only specimen available to me was moribund when found. 8. Leptocola phthisica (Saussure, 1869) One female examined. The defences of the male of this common grassland species have already been described (Edmunds, 1972). Primary defence. The female rests amongst long grass, as does the male, with no particular cryptic posture (Fig. 3). Secondary defence. Unfortunately the insect was unhealthy and had a broken leg, so it was not possible to examine its responses at all adequately to being attacked. When seized with fingers it retaliated by scratching with the forelegs, but no startle display could be elicited. However, the wings are reduced to very small erectile organs, quite useless for flying, but of possible use during a startle display. The forewing is mauve with a green costal area, whilst the hind wing is greenish-white proximally with a black spot posterodistally. This striking colour suggests that the wings may be erected in display, either to a conspecific or to a potential predator, and the resemblance Figure 3. Primary defence posture of female Leptocolu phthisicu: an extremely elongated grass mimic with no unusual cryptic posture. Drawn from life. Inset: details of colour pattern on wings which may perhaps be used for secondary defensive display. DEFENSIVE BEHAVIOUR OF PRAYING MANTIDS 9 of the black spot to an eyespot suggests that such a display is directed against predators (see Fig. 3, inset). Therefore it is probable that, as with Turachodes and Dununu, the wings have been reduced in the female because they are an easily recognisable visual cue to predators, but they have been retained as small brightly coloured structures because they are of protective value during a startle display. 9. Plistospilota guineensis Roy, 1965 The behaviour of a single male of this species has already been described (Edmunds, 1972 p. 8 & pl. l ) , but the caption to the plate illustrating its display incorrectly refers to it as a female. A second male has now been examined: when poked it either gave a startle display or it flew. 10. Cutuspilotu misunu (Giglio-Tos, 191 1) Two males and two females examined. A forest species. Arboreal. Primary defence. The adult insect is mottled grey and dark brown with a green costal area to the forewing. I t is thus cryptic on branches and trunks of forest trees. When disturbed the adult occasionally lowers the body towards the substrate and protracts the forelegs in front of the head, just as do the stick mimics Dununa and Hoplocoryphu. Secondary defence. When poked, males usually run or fly, females run. Only occasionally did the two females give a startle display with the forelegs abducted, the hind wings raised, and sometimes with stridulation. There are six prominent cream spots on the forecoxa which is orange proximally, merging into brown distally. The hind wings are mostly black with three yellow bars anteriorly. Hence although the display was not often given it is quite dramatic. 11. Prohierodula omatipennis (Bolivar, 1893) One male and one female examined. A forest species. Arboreal. Primary defence. The female is bright green, very similar in appearance to Sphodromantis lineola, whilst the male has green head, thorax, legs and coxal area of the forewing, but the rest of the forewing is brown mottled with black. Both sexes are thus cryptic with a general resemblance to vegetation. Secondary defence. When poked the male ran, flew, or gave a startle display, and the female ran or gave a display. The forefemur is yellow with two black spots whilst the coxa has 6 or 7 cream spots. The hind wings are mostly black with a crimson or bright magenta coxal strip, large bright orange spot at the wing tip, and two clear spots just behind the crimson region. Hence the startle display is one of the most dramatic of any Ghanaian mantis. In addition the abdomen is pink dorsally with a middorsal black line. 12. Sphodromantis aurea Giglio-Tos, 1917 One male and one female examined. A forest species. Arboreal. Primary defence. Both animals were bright green like Sphodromantis lineola 10 M. EDMUNDS (see Edmunds, 1972), and hence they have a general resemblance t o green vegetation. Secondary defence. When poked the male always flew, the female gave a startle display. The three or four creamy white spots on each forecoxa are very large and conspicuous, and there is also a large blue-black or greenish black spot on the forefemur, so that with the legs abducted and held close together the spots resemble eyes (Plate 1A). During display the red jaws are also visible and yellow marks intersegmentally on the ventral surface of the abdomen can sometimes be seen. The female is slightly larger than is the female of S. lineola, and the display, with brighter marks on the forelegs, is more conspicuous. 13. Paramantis prasina (Serville, 1839) One male examined. The defences of two animals of this species have been described in Edmunds (1972), but these were females, not males as stated in that paper. Primary defence. The male is green with purplish brown stripes mid-dorsally and at the edges of the pronotum, as well as on the posterior margin of the head. The posterior half of the pronotum is also purplish brown, and so also are the wings except for the green costal areas. Thus the male is much browner than the female which is entirely bright green in dorsal view. Both sexes have a general resemblance to vegetation. Secondary defence. When poked the male ran, flew or gave a weak display involving the forelegs but not the wings. Since the forecoxa has bright red spots this display is quite dramatic. The hind wings are suffused purplish grey and the abdomen is crimson dorsally suggesting that the wings may sometimes be raised during display, or that these are flash colours. 14. Paramantis togana (Giglio-Tos, 1912) One male and one female examined. A forest species which occurs occasionally in the savanna (e.g. Kwabenya, near Accra). Probably arboreal. Primary defence. The male is brown with green eyes, the female is green. Both are therefore camouflaged on vegetation. Secondary defence. When poked the insects either ran, flew or gave a startle display. During display the female abducted the forelegs but did not raise the wings whilst the male displayed both legs and wings. This is probably individual rather than sexual variation since normally females of mantids display far more readily than males. The forelegs each have three large yellow spots on the coxa, the jaws are black, and the wings are brightly coloured as follows: hindwing mostly black with yellow veins, but with the costal area crimson; undersurface of forewing mottled red and yellow (hence appearing orange), but crimson and grey proximally, transparent posteriorly, and with conspicuous yellow and crimson bars anteriorly in the costal region. Hence the display is very dramatic. 15. Statilia apicalis (Saussure, 187 1) One male examined. Mainly a forest species but also occasionally found in the savanna at Legon. Probably arboreal. DEFENSIVE BEHAVIOUR O F PRAYING MANTIDS 11 Primary defence. A dark brown mantid with a general resemblance to twigs or sticks. Secondary defence. When poked the insect ran, flew, or displayed. The display always involved erection of the hind wings, usually with stridulation. Sometimes the forelegs were held in the abducted position close to the body, thus displaying their bright colours to the full, whilst at other times they were held t o one side as in Hemiempusa (Edmunds, 1972). The wings are suffused brown but are transparent, as in Polyspilota, whilst the forelegs have two cream spots distally on each femur, one cream spot proximally on each coxa, the rest of these regions being bluish black with small yellow spots (see Plate 6 of Edmunds, 1974). The ventral surface of the thorax is red anteriorly, bluish grey posteriorly, so the displaying mantis is very impressive. 16. Miomantis paykullii Stil, 1871 One brown and two green females were examined in addition t o those whose behaviour has already been described (Edmunds, 1972). These insects did give a slight display when poked, with erected hind wings, but without abduction of the forelegs. The hind wings are pale yellow, and the posterior surfaces of the forewings are also pale yellow. The abdomen is orange-yellow dorsally, possibly as a flash colour. None of these or of the earlier insects could be induced t o fly, either when poked or when thrown into the air. 17. Miomantis aurea (Giglio-Tos, 1917) A second female of this species was examined. Unlike the first individual (see Edmunds, 1972), it did not display, but ran or jumped when poked. 18. Sibylla limbata Giglio-Tos, 191 5 Two males examined. A forest species. Arboreal. Primary defence. The normal resting posture of this species is upside down so the head and pronotum are reverse countershaded, dark brown ventrally, more yellow-brown dorsally. The legs and abdomen are dark brown, but there are cream spots at the bases of legs two and three. The wings are pale green with brown veins. There are foliations on legs two and three, the head has an irregularly shaped vertex, and the pronotum is long and narrow. It is not known if this species normally rests on branches or on leaves, but obviously it is cryptic. Secondary defence. When poked the insects ran or flew. No display was observed, and since the forelegs have no conspicuous markings it is probable that they do not display. 19. Phyllocrania paradoxa Burmeister, 1838 An additional male and female were examined which enable me to add to my previous description. The female is pale straw in colour in contrast to the previous female which was dark brown (Edmunds, 1972). This female is illustrated in colour in Edmunds (1974: plate 3). Neither insect gave a display 12 M. EDMUNDS although the basal region of the hind wing is blackish, suggesting that a display may possibly occur under some circumstances. 20. Chloroharpax modesta (Gerstaecker, 1883) One female examined. A forest species. Presumed to be arboreal though the present individual, collected as a first or second instar nymph, was found on a shrub in secondary forest. Primary defence. The female is green with a black-rimmed yellow spot on the forewing, and black marks dorsally on the head and thorax (Fig. 7, Plate 1C). It probably has a general resemblance to vegetation, but it is not known if this spot is only of use in display or if it is disruptive. Secondary defence. When poked it attempted to run, jump or fly, but on two occasions it gave a startle display. The display posture is similar to that of Miomantis aurea and of Pseudocreobotra (Edmunds, 1972) with the wings raised and the forelegs held extended and partially abducted. The small spots on the wings are then visible and could possibly be mistaken for small eyes, though they are nothing like as dramatic as the eyespots of Pseudocreobotra or Chlidonoptera The hind wings are white, but there are no conspicuous colour marks on the forelegs. 2 1. Panurgica compressicollis (Saussure, 1898) Two females examined. A forest species. Probably arboreal. Primary defence. The body and wings are mottled with various shades of brown and purplish brown. The normal resting posture of this short and broad mantid is upside down with the legs held close to the body and the wingtips practically touching the substrate posteriorly. The result is that the insect closely resembles a folded dead leaf (Plate 2A). If the twig is moved the insect pivots on its legs and sways laterally, just as if the leaf is about to break off from the twig. Secondary defence. When disturbed both insects either attempted to run, jump or fly, or they gave a startle display or struck at the stimulus source with the forelegs. The startle display posture is similar to that of Chloroharpax and Pseudocreobotra, with the forelegs extended laterally, and the wings opened. It may be directed either frontally or laterally. The inner surface of the forelegs is black with pale yellow spots. The hind wings are crimson-red basally and along the anterior edge, but this merges into black posterodistally with yellow veins. The under surface of the forewings is also tinged with crimson, and the abdomen is orange mid-dorsally. Hence the display is very dramatic. 22. Chlidonoptera lestoni Roy, 1975 One male examined. A forest species, but whether from the canopy, shrubs or herbs is not known. This species is closely related to C. vexillum from the Cameroons and is fully described by Roy & Leston (1975). Primary defence. The insect is disruptively marked with green and yellow with a large eyespot on the forewing. Although this eyespot can be conspicuous DEFENSIVE BEHAVIOUR OF PRAYING MANTIDS 13 during display I consider that it is not conspicuous during normal resting when the insect is on vegetation or flowers (Fig. 7). Secondary defence. When the mantid was poked it tried to run, fly or give a startle display. The display is very similar to that of Pseudocreobotra ocellata (Edmunds, 1972), but the sub-circular eyespots are much less perfect imitations of a vertebrate eye. The basal half of the otherwise transparent hind wing is orange. A single female was also examined, but unfortunately it was moribund and soon died. The female has larger, more circular eyespots than the male (Fig. 7, Plate 2D), as do other species of Chlidonoptera (Roy, 1964, 1965). The hind wings were yellow with red spots, very like those of C.vexillurn, so its display must be very dramatic. 23. Pseudocreobotra ocellata (Beauvois, 1805) Two further females were examined which enable me to add to my previous description (Edmunds, 1972). During display both sexes erect the hind wings, but whilst in the male these are entirely transparent with white veins, in the female the proximal half is yellow, the distal half transparent. The forelegs may be extended anterolaterally in display (as shown in Edmunds, 1972: plate l), or they may be held in the normal resting position. 24. Oxypilus hamatus Roy, 1966 Two females examined. The behaviour of the more easily obtained males has been described earlier (Edmunds, 1972). Primary defence. The female is entirely brown and completely apterous. She has a general resemblance to twigs but has no specialized resting posture such as occurs in Catasigerpes or Danuna. Secondary defence. When disturbed the insects either ran or jumped, but they never gave a startle display. 25. Anasigerpes bifasciata Giglio-Tos, 1915 One male and one female were examined which enable me t o add to the description given earlier (Edmunds, 1972). The wings of the male are truncated posteriorly, just as they are in male Catasigerpes, Phyllocrania and Panurgica, so that when at rest the wing tips touch the substrate. In this way the insect resembles a short twig arising from the branch on which it is resting. The dorsal surface of the abdomen is coloured in both sexes, orange in the male but mauve in the female. Probably these are flash colours since no startle display was ever seen. 26. Catasigerpes occidentalis (Wood-Mason, 1879) One additional female was examined. The females described and illustrated by Edmunds, 1972, as Gztasigerpes toganus are actually C. occidentalis (identified by R. Roy). The species of Catasigerpes from the C6te 14 M. EDMUNDS d'Ivoire-Ghana-Nigeria region are in a state of confusion, as discussed by Gillon & Roy (1968). Females with a long vertex correspond with the type of C. occidentalis (Wood-Mason, 1879), whilst those with a short vertex correspond with the type of C. nigericus (Giglio-Tos, 1915). All males appear to correspond with the type of C. toganus (Giglio-Tos, 1915), which has a short vertex. Numerous males have now been examined but so far it has not proved possible t o distinguish those of C. occidentalis from those of C. nigericus, hence, if the female is unknown, males are conveniently referred t o as C. toganus. Since the female Catasigerpes from Legon are all clearly C. occidentalis, the males which were reared from them (Kumar, 1973) as well as those caught at Legon must also belong to this species. Hence the correct name for the Catasigerpes described by Edmunds (1972) and listed in Ragge & Roy (1967) as C. toganus should be C. occidentalis. Secondary defence. Two points need t o be added to my earlier account: the hind wings of the female are black, which may possibly add to the flash effect of the red abdomen when the insect flies; and one of the females exhibited thanatosis for a few seconds after repeated poking. 27, Chrysomantis speciosa Giglio-Tos, 1915 One female examined. A forest species. Probably arboreal. Primary defence. The face, pronotum and outer surface of the forelegs are greenish white; the wings are mottled with shades of grey and are truncated posteriorly as in Anasigerpes and Catasigerpes. The resting place of this species is not known but its very pale colour indicates that it must live either on bark or twigs. The second and third legs have minute foliations which presumably break up their outline. Secondary defence. The insect was unfortunately sluggish and unreactive when first examined, and it died soon afterwards. The inner surface of the foreleg is black with cream spots distally on the femur. It is likely that these marks are exposed during a boxing display similar to that of Chrysornantis cachani and of Catasigerpes. 28. Chrysomantis cachani Roy, 1964 One female examined. A forest species. Probably arboreal. Primary defence. The dorsal surface is greyish mottled with darker brown and with a transverse brown bar on the wings. There is also some green on the eyes and wings. The outer surface of the foreleg is grey-brown and greenish grey. The wings are truncated posteriorly as in Anasigerpes and Catasigerpes. Legs two and three are minutely foliaceous. The insect is therefore cryptic on greyish-brown twigs. Secondary defence. When poked the insect usually tried to fly. The hind wings are transparent and the dorsal surface of the abdomen is blackish brown, so there do not appear to be any flash colours. No startle display was seen. When resting or if slightly disturbed the mantis gave a boxing display similar to that of Catasigerpes. The forelegs were extended forwards at an angle of about 45" to the midline, first to one side, then to the other. Sometimes both legs DEFENSIVE BEHAVIOUR OF PRAYING MANTIDS 15 were extended simultaneously, first to the right then to the left; at other times the legs were extended alternately. The inner face of the foreleg is black with white distally on the coxa and orange proximally on the femur, so these marks are conspicuous during the display. By analogy with Oxypilus and Cutasigerpes, this boxing is probably a territorial, not a defensive display (Edmunds, 1972). 29. Dunuria buchholzi Gerstaecker, 1883 Numerous males and three more females enable me t o add to my earlier description of the defences of this species (Edmunds, 1972). Secondary defence. When the insect was poked at with forceps, it often dodged to one side and so avoided being hit. The movement is simply one of flexing the legs on one side so that the body is moved quickly sideways, but the position of the tarsi on the substrate remains unchanged. One male also exhibited thanatosis when repeatedly poked and roughly handled. The startle display is very variable: sometimes the forelegs remain extended as in the fully cryptic posture, sometimes they are flexed as in Polyspilotu or Sphodromantis. The tip of the abdomen is often repeatedly reflexed over the back during display, possibly resembling a scorpion. In the female the wings are very short, useless for flying, but still used during startle display since the lower surface of the forewing is red and the hind wing is black. However, none of the females examined gave a sustained display with the short wings held open such as occurs in the similarly brachypterous Turuchodes female which often sustains a startle display for 10 or more seconds. In Danuriu, after much poking, the wings may be opened and closed several times in rapid succession, or if a leg is seized they may be held open for perhaps two or three seconds. Hence it appears that the display is designed to startle a predator into quickly releasing it by sudden appearance and disappearance of bright colours whereupon the mantis once again resembles an innocuous stick. In Turuchodes, on the other hand, the display may actually be intimidating to the predator. 30. Popu undatu (Fabricius, 1793) Two males examined. A savanna species. Arboreal. Primary defence. Popu is of typical shape (i.e. similar t o Mantis) but resembles Dunuriu in its defensive adaptations. The insect is greyish brown. The head is prognathous with two projections posterolaterally which conceal the neck (as in Danuriu). The forecoxa is notched so that when the forelegs are protracted anteriorly the head fits into the notch, and the profile of body, head and forelegs resembles that of an attenuated stick (also as in Dunuriu) (Plate 2C). The wings are truncated posteriorly, as in Cutasigerpes. The legs are much shorter than are those of Dunuriu so that when resting the tips of the wings often touch the substrate so that the body resembles a branch arising from the substrate. finally the second legs have a small frill, presumably to break up the outline. Thus this mantis is a stick mimic which has many of the adaptations of the extremely elongated and long-legged Dunuriu, but it also has similarities to the short-legged stick mimics such as Catusigerpes and Chryso- M. EDMUNDS 16 mantis. I t is related closely to Danuriu and therefore gives an indication of how the elongated stick mimics may have evolved. Secondary defence. When the mantis was slightly disturbed the forelegs were protracted into the fully cryptic posture, as described above. On more sudden disturbance the mantis flew. Occasionally it gave a brief startle display exposing pink on the lower surface of the forewings and black on the hind wings. The forelegs were not flexed during display, but since the forecoxa is pink with black and brown spots, it is probable that in some circumstances the forelegs are exposed during a startle display. Table 1. Summary of defensive adaptations of Ghanaian mantids + indicates a particular behaviour or characteristic occurs; - indicates i t does not occur; ? indicates that there is some uncertainty-see text for detail. A blank indicates no observations were made Genus General Special Runs Flies resemblance resemblance d P dQ (mimicry) Amorphoscelis bark Poromorphoscelis short stick Hoplocorypho long stick Torochodes bark Golepsus short grass Pyrgomon tis short grass Theopompella bark Negromantis green Nilomantis green Oxy o thespis longgrass Leptocola long grass Plistospilota grey-green Catospilo to grey-bro wn Prohierodulo greenbrown Polyspilota greenlbrown Tenodero greenlbrown long grass Sphodromantis greenbrown Poramantis green or brown Mantis greenlbrown Statilia brown Miomantis greenhrown Sibyllo green &brown Phyllocrania leaf Pseudoharpax green Chlorohorpax green Panurgica leaf Chlidonoptera disruptive Pseudocreobotro disruptive Oxypilus brown Anosigerpes short stick Ca tasigerpes short stick Chrysomantis short stick Sten o vo tes long stick Danuria long stick Pop0 stick Idolomorpho long grass Herniempuso long grass ++ + ++ ++ + ++ + + --? + - ++ ++ ++ ++ ++ ++ ++ + ++ + ++ ++ + + + ++ ++ ++ ++ + -- ++ - ++ ++ Thanatosis Flash Startle Wings d Q reduced in Q ++ + +++ +- + + + + ++ +- +++ ++ ++ ++ ++ + ++ ++ +- + + + ++ +- ++ ++ + -- +- + ++ ++ + -- - Boxing display DEFENSIVE BEHAVIOUR OF PRAYING MANTIDS 17 SUMMARY OF DEFENSIVE ADAPTATIONS The defensive adaptations of the 37 genera of mantid described in this and in my earlier paper (Edmunds, 1972) are summarized in Table 1. No attempt has been made to indicate the probability of a particular response occurring, but in general small species run or fly and rarely or never give a startle display, whereas large species are more likely to display than to attempt to escape actively. Question marks indicate that there is some uncertainty about a particular behaviour; for example there are several mantids which have brightly coloured dorsal surfaces to the abdomen which could be flash colours. Since it is not easy to prove their function I have put a question mark in the appropriate space. DISCUSSION OF PRIMARY DEFENCE Body colour polymorphism In my earlier paper (Edmunds, 1972) I attempted to show that there is a correlation between the frequencies of green and brown morphs of Sphodromantis lineola and Miomantis paykullii and the colour of the surrounding vegetation. Further data confirm the conclusions reached at that time. For S. lineola I now have a sample of 684 males examined between November 1967 and August 1973 at Legon. Brown insects only occur in the dry season (October to March), apart from a single insect on 18 April 1972, and if the data are summed by month the browns reach a peak in January which is normally the driest month of the year (Table 2). It is possible that the Table 2. Numbers of green and brown Sphodromantis lineola caught at Legon per month, 19671973 Month Green Brown %brown January 34 64 39 45 79 117 38 13 18 60 99 65 3 1 8.1 1.5 February March April May June July August September October November December 0 1 0 0 0 0 0 1 4 2 0 2.2 0 0 0 0 0 1.6 3.9 3 .O single brown insect in April had survived since February, or (more likely) it had spent its last few days before the final moult in a building where light conditions tend to produce a high frequency of browns. Barnor (1972) has shown that the most important factor determining whether a nymph of S. lineola becomes green or brown at the next moult is light intensity. Nymphs 2 M. EDMUNDS 18 commonly occur on small trees and shrubs which are always green in the wet seasons but which normally lose their leaves and then flush with new growth in the dry season. A green nymph on a tree that loses its leaves will be very conspicuous, and selection would naturally favour any insect that could become brown at this time and then green again once the new leaves had opened. Loss of leaves means that the tree will be exposed to greatly increased incident light and this induces insects to become brown at the next moult. As the leaves grow again the light reaching the insects is reduced so they are induced to become green at the next moult. This is speculation, but it does accord with experimental evidence and also with the observable fact that brown insects are never very numerous: this is because leaf fall and flushing are not synchronous on all trees but are spread over a long period of the dry season. In Miomantis paykullii the situation is rather different. In my earlier paper, on the basis of six dry and one wet season samples, I concluded that colour is correlated directly with rainfall: 70% of the population were green in the rainy season whilst from 70% to 90% were brown in the dry season. I now have samples of Miomantis collected monthly from 1970 till 1973. Figure 4 shows 5 ?! 40 20 - 0 0- 2 3-4 5- 6 7- 9 10 - 17 Rain days per month Figure 4. Relationship between rainfall (as number of days on which rain fell per month) and body colour (as % of population which were green) of male Miomantis paykullii at Legon, Ghana. The marks to the right of each block are the monthly figures on which the means are based. that there is a very strong correlation between percentage of green insects and number of days on which rain fell in that month: in months with 0-2 days of rain, only 24% of mantids were green whilst in months with 10 or more days rain, 69% of insects were green. Using the Spearman rank correlation test (Siegel, 1956), the correlation between number of rainy days and percentage of green insects in that month is significant at the 0.01% level (rs = 0.65 for N = 30). If the percentage of green Miomantis is compared with the number of rainy days in the preceding month, the correlation is even more significant (r, = 0.74 for N = 30), presumably because the mantids can only respond to DEFENSIVE BEHAVIOUR OF PRAYING MANTIDS 19 change in humidity at the next moult which may be a week after the rain, and because once adult they cannot change colour during the week or more that they live. Miomantis lives amongst grass which changes colour according to rainfall: if there is a drought the grass becomes brown, and if there is rain it becomes green. Barnor (1972) found that in this species, unlike in Sphodromantis, relative humidity is of more importance than light in determining the colour of the insect at its next moult. High humidity induces an insect to become green, low humidity induces it to become brown. Since humidity of the grass is likely to correlate with its colour it is obviously biologically advantageous for humidity to determine the colour of the insect at its next moult. Conversely with Sphodromantis, humidity does not determine times of leaf fall and flushing of trees (at least not in a simple way), so again it is obviously more advantageous to use some factor other than humidity to determine the colour of the insect. For both species, the colour is likely to coincide with that of its surroundings, so both are likely to be camouflaged. In addition Barnor (1972) found that when given a choice between brown and green backgrounds, green Miomantis showed a significant preference for resting on green whilst brown Miomantis showed a significant preference for brown (summarized by Edmunds, 1974). In Sphodromantis, green insects choose to rest on green leaves rather than on brown ones, but brown insects appear to have no such preference: indeed they may even prefer green (the figures do not differ significantly either from random or from the choice of green insects). Since brown Sphodromantis occur rarely, and since they are likely to occur only on entirely leafless trees (see above), they are not likely to have much choice of resting place, so selection has not favoured the evolution of a matching response. Table 3. Numbers of green and brown Mantis religiosa and Tenodera superstitiosa caught at Legon in wet and dry seasons Species Mantis religiosa Tenodera superstitiosa Colour Wet season Dry season brown green %brown 10 26 14 42 10 15 82 brown green %brown 41) P 4.40 <0.05 72 2 5 88 94 not significant Mantis religiosa also has a green-brown polymorphism, 60%of insects caught over a four year period being brown. If the year is divided into a wet six month period (April to September) and a dry period (October to March), there are significantly more greens in the wet season than in the dry (Table 3). This wet six months includes the major rains in June and the small rains of September. August is normally dry, but no insects were caught in this month. The dry six months includes October and March which can be fairly wet in some years. If the same is done with the grass-living Tenodera superstitiosa the figures are not significant, possibly because of the low frequency of green insects at all M. EDMUNDS 20 seasons. In this species the brown form always has a green costal area to the forewing so it is disruptively cryptic, and since even in the rainy season there is always some dead, brown grass, selection may not have favoured such a marked seasonal change in frequency of the two morphs. Seasonal changes in the frequencies of green and brown forms of forest species such as Polyspilota aeruginosa and Prohierodula ornatipennis have not been studied but would be of considerable interest since there can be very much less change in colour of vegetation there than in the savanna. In Polyspilota at Legon brown males are rare; only 2 out of 31 males seen were brown (the rest being “bicoloured” green and brown), but 5 out of 7 females were brown. All but 4 of these insects ( 1 green female, 2 bicoloured males and 1 brown male) were caught in the dry six month period. Another type of colour change which is probably environmentally induced is fire melanism. Many species of grasshoppers moult into black forms shortly after bush fires, and these are obviously better camouflaged than green or brown insects on the blackened ground (Hocking, 1964). Fire melanism also occurs in the mantids Galepsus toganus and Pyrgomantis pallida (Table 4). In Pyrgomantis melanics have only been found in January and February, the end of the dry season when bush fires are most numerous, whilst in Galepsus a few black insects have also been taken as late as June. I t is not known if these were long-lived individuals, or if they were induced to turn black by late burning, or if they arose for some other reason. Table 4. Frequencies of melanics amongst male Galepsus toganus and Pyrgomantis pallida attracted to light at Legon between November 1970 and August 1973 Month January February March April MaY June July Aug-Oct November December Galepsus toganus Pyrgomantis pallida melanics total % melanic melanics total % melanic caught caught 10 14 9 13 8 16 10 36 33 15 13 13 0 7 0 4 8 5 0 0 1 5 3 2 1 2 0 0 1 2 0 0 0 0 0 0 0 0 0 0 15 7 10 4 20 0 10 0 15 0 0 0 0 0 0 16 6 12 34 13 A further polymorphism has also been found to occur in Tarachodes afzelii, the common bark mantis. In this species occasional individuals are found with a longitudinal black stripe running mid-dorsally from the pronotum to the tip of the abdomen, and a second morph also occurs with two transverse black bars, one on the pronotum, the other on the abdomen (Fig. 5). Between November 1970 and August 1973, 205 male Tarachodes were observed at Legon, but only three were striped and one had a black bar (frequencies of 1.5 and 0.5% respectively). Insects with a black stripe are better concealed to the human eye DEFENSIVE BEHAVIOUR OF PRAYING MANTIDS 21 Figure 5 . Bark mimicry in mantids. Left: two female Turuchodes ufzelii one with black bars, the other with a black stripe, on the body. Right: male Theopompella wesrwoodi with broad, downwardly angled costal areas to the forewings. Based o n colour transparencies and mounted specimens. on deeply furrowed bark, but not on smooth uniformly grey bark. This is particularly important in the last instars and in the adult female, but the adult male has the abdomen covered by wings so that the black stripe is far less conspicuous. One adult female with a black stripe was captured. She had already been mated, presumably by a typical male. Young from this female were reared to the 3rd or 4th instar when stripes (if present) can be seen: 5 were striped, and 3 unstriped. One of the striped young was reared to adulthood and was a female. She was mated with a typical unstriped male and of her young 6 were striped, 5 unstriped, and 4 had a line of black dots mid-dorsally. On the basis of such meagre results it would be rash to suggest the genetic basis of the striped condition, but I assume that it is controlled genetically and is not an environmentally induced polymorphism. Leaf mimicry The defensive behaviour of two species of leaf mimics from Ghana have now been described, Phyllocrania paradoxa and Panurgica compressicollis (Fig. 6 ) . Both species resemble dead brown leaves. Phyllocrania is the more “perfect” leaf mimic with the pronotum extended laterally and with frills on vertex, abdomen and legs, but the legs are held away from the body so may be a visual recognition feature for predators. Phyllocrania probably evolved from a typical mantid (similar to Mantis or Sphodromantis in general shape) by developing 22 M . EDMUNDS 10 rnrn Figure 6 . Leaf mimicry in mantids. Left: female Panurgica cumpresricullis; right: female Phyllocrania paradoxa. Drawn from mounted specimens. these frills and body extensions whilst retaining long legs. Sibylla represents an intermediate stage in this sequence although its pronotum and body are too elongated for it to be ancestral to Phyllocraniu. Panurgica on the other hand has rather shorter legs and these are normally held very close to the body so that they do not form a visual cue to predators. I t also has a very short body compared with the enormous size of its head and forelegs. I t has evidently evolved from a typical mantis by shortening and broadening the abdomen, developing disruptive marks on the wings, and by adopting a posture with the legs held close to the body. It is very closely related to the target mantids Pseudocreobotra and Chlidonoptera which have a similar body shape and resting position with the legs held close to the body (see below). Bark mimicry The defences of three genera that live on tree trunks in Ghana have now been described, A morphoscelis, Tarachodes and Theopompella. All are dorsoventrally flattened, but whereas Amorphoscelis is an active, running mantid with only a general resemblance to moss or lichen covered bark, the other two are less active species with highly cryptic resting postures. The middle and hind legs are fairly long to give a good grip on vertical or overhanging branches, but because of the low, flattened body they are held close above the substrate and so are not conspicuous. The forelegs of both genera are held slightly abducted in typical resting posture so as to smooth the contour of the insect DEFENSIVE BEHAVIOUR O F PRAYING MANTIDS 23 when viewed from in front and to obscure any lateral shadow. In addition the wings of female Tarachodes are greatly reduced in size whilst those of Theopompella are disruptively coloured and are angled laterally (Fig. 5 ) . Thus in neither of these insects do the wings form a conspicuous feature by which predators might recognise the insect. One important difference between the two genera is that whilst in Tarachodes the head is reflexed beneath the body in the opisthognathous position, in Theopompella it is directed forwards in the prognathous position. Neither position is suitable for actual feeding, but both result in the line from the head to the substrate when viewed from the side being smoothed and hence becoming less conspicuous. Grass and stick mimicry First of all it is necessary to distinguish two types of stick and grass mimics: long legged insects with elongated bodies, and short legged insects with (usually) short bodies. These two types have different morphological adaptations and different resting positions. For convenience I shall refer to them simply as long and short stick mimics and long and short grass mimics. I shall consider long grass and stick mimics first. The two habitats of long grass and the canopy of a tree or shrub both contain many elongated structures, but whereas in long grass the stems are very dense, in a tree the terminal branches are comparatively sparse. Hence in long grass, provided an insect is elongated and of comparable body diameter to the grass, it will be cryptic, but in a tree or shrub a more precise resemblance to a stick is required to give a comparable level of protection. Thus it is that long-legged grass mimics such as Oxyothespis and Leptocola have extremely attenuated bodies, but they lack either a special resting posture or morphological adaptations such as narrow head or notched forecoxae. Compsothespis by contrast has a narrow head with the neck concealed in dorsal view, so perhaps this rare, greyish species lives on shrubs rather than amongst grass. Idolomorpha and Herniempusa have more cryptic resting postures with narrow heads, pointed vertex and antennae, and forelegs held fully flexed close beneath the pronotum, but they have broader and hence more conspicuous wings and abdomens. Long stick mimics such as Danuria, Hoplocorypha and Stenovutes have distinctive resting postures with numerous morphological adaptations that increase their resemblance to sticks (details given in Edmunds, 1972). The grass-living Tenodera is of interest since it demonstrates one way in which grass or stick mimicry may have evolved. Like typical Mantinae which lack morphological resemblance to sticks or grass, Tenodera has a green-brown polymorphism. But it also has a longer body and legs than do typical mantids, and it rests in a stick mimicking posture with forelegs protracted in line with the body. It has no other adaptations to stick mimicry (such as narrow head or notched forecoxa), so it is thus intermediate in structure between typical mantids and long stick mimics. A study of the defensive behaviour of nymphs of certain mantids suggests another way in which stick mimicry may have evolved. Balderrama & Maldonado (1973) have found that late instars and adults of the large South 24 M. EDMUNDS American mantid Stagmatoptera biocellata have a dramatic startle display similar to that found in all large species of the Mantinae, but they found that the early instars respond to a bird by protracting the forelegs and thus increasing their resemblance to a stick. They found that as the insects approached maturity so they responded less often by becoming stick-like but more often by a display, and also that the further off the stimulus the more likely they were to respond by becoming stick-like. I have found that the forest mantid Cataspilota misana behaves in a similar way: from about the third instar onwards (younger insects were not available) the brown mantids often adopted a stick-like posture with the body lowered towards the substrate and the forelegs extended in front of the head. This occurred occasionally in the adults as well, though these also displayed. It should be stressed that this stick mimicry is very imperfect since neither Cataspilota nor Stagmatoptera has any of the adaptations to perfect the resemblance to a stick such as long body, narrow head, notched forecoxae and projections posteriorly on the head. Nymphs of Sphodromantis, Miomantis and Prohierodula have also been studied but were never observed to adopt this stick-like posture. I t is possible that the initial stage in the evolution of some stick mimics was simply adoption of a stick-like posture; in other words that the behavioural adaptation preceded elongation of the body and other morphological modifications. The defensive adaptations of several short grass and stick mimics from Ghana are now known, for example Paramorphoscelis, Galepsus, Pyrgomantis, Catasigerpes, Anasigerpes and Chrysomantis. Pyrgomantis is obviously an extremely specialized grass mimic. Since Galepsus is golden brown and has what appear to be fire-induced melanics, this genus is probably also a grass mimic. I have already pointed out that Galepsus has the structure and behaviour that could evolve into either a bark mimic (e.g. Tarachodes) or a grass mimic (e.g. Pyrgomantis) (Edmunds, 1972). The short stick mimics are of two types, the amorphoscelid Paramorphoscelis, and the hymenopodid genera Catasigerpes, Anasigerpes and Chrysomantis. These last three genera all conceal the head by having it in the opisthognathous position; they have short and hence inconspicuous legs with frills to break up their outline; and the wings (in the male) are truncated posteriorly so that the body appears to arise directly from the branch on which it is resting (Plate 1D). Paramorphoscelis is much more similar in its structure and behaviour to Galepsus and Pyrgomantis. Its legs are short and it rests close and parallel to the substrate, not angled away from it, so that the insect appears to be a part of the twig on which it rests rather than a branch arising from it. The head is held in the prognathous position as in Theopompella and Danuria, but unlike either Pyrogomantis or Catasigerpes. The prognathous position in a prostrate resting form probably gives better concealment from lateral view than does the opisthognathous position. Finally Popa is an insect of very great interest since it has some of the adaptations found in long stick mimics such as the closely related Danuria, but it also has some adaptations of short stick mimics such as Catasigerpes (these have been summarized above at the end of the section “Descriptions of defensive behaviour”). It is thus in a morphological condition such that it could evolve either into a long or into a short stick mimic. DEFENSIVE BEHAVIOUR O F PRAYING MANTIDS 25 Reduction o f wings In 1972 I pointed out that size of wings in the female is a compromise, selection on the one hand favouring large wings for flight, and on the other favouring reduced wings for improved camouflage. I t is important for the male to be mobile since he must find a female, but it may not matter if he is short-lived and dies soon after mating. Hence he has long wings. For the female, however, it may be more important to be highly cryptic so that she can live long and produce many oothecae. Some data which support this hypothesis are given by Gillon & Roy (1968). At Lamto, Cbte d’Ivoire, eight species of mantid in which females have reduced wings have sex ratios ranging from 16%to 33% males, whilst two species in which both sexes are fully winged have sex ratios of 43% (Mantis religiosa) and 39% (Statilia apicalis). The low percentage of males in all species supports the hypothesis that females live longer than males, but the very low percentage of males in species where the female is flightless requires further consideration. Gillon and Roy also report on the stomach contents of 11 black kites (Milvus migrans) and 1 grasshopper buzzard (Butastur rufipennis). These birds had eaten a total of 44 adult mantids: 37 were of species fully winged in both sexes, 20 male and 17 female; whilst of species in which the female has reduced wings 7 were male and none female. These two sex ratios are significantly different (using Fisher’s exact test, P = 0.032). The figures show that these birds were capturing a higher proportion of females in species with fully winged females than in species with flightless females. Since these two species of birds capture most of their food on the ground, the explanation of the difference is probably that flightless females are more cryptic and so less easy to find than are fully winged mantids. This implies that the very low percentage of males in mantids with flightless females is partly due to selective predation by birds. I have now been able to examine females of eight species which have reduced wings. Wings may be reduced in size so that they are useless for flight but still important for display, or they may be completely lost. In Hoplocorypha and Oxypilus the female is completely apterous and a startle display h.is never been seen in either sex of these two genera. Consequently once the ancestors of these genera had reduced the wings in the female till they could no longer be used for flight, selection favoured their complete loss. In the remaining genera, however, the wings have been reduced in size but not completely lost. In Tarachodes and Danuria they are small but used during display. (They are probably also used in display in Galepsus, since the male displays readily, but no females were available for study.) In Pyrgomantis and Leptocola no display was observed in the females but it probably does occur under some circumstances. The single female Leptocola was damaged so could not be properly studied, but each hind wing has what appears to be a large eyespot. Possibly the wings are opened and closed rapidly when the insect is disturbed thus giving a flash of two large eyes which may startle a predator into withdrawing. In Miomantis paykullii and M . aurea the wings are only slightly reduced from the size one finds in typical flying mantids of comparable size, yet none of the females I examined was capable of flying. In both species the females use the wings for display occasionally. Comparison with other species 26 M. EDMUNDS of Miomantis indicates how wing reduction has evolved. In eight female M. aurea from various localities in Ghana the wing length as a percentage of abdomen length varies from 85% to loo%, average 91%. In thirteen M. paykullii from Ghana it varies from 79% to 90% average 85%.Two females of M. lamtoensis Gillon & Roy, 1968, have wings 44% and 48% of abdomen length (one from the Mole Game Reserve, Ghana, the other measured from Gillon and Roy’s plate of a female from Chte d’hoire). Finally M . bitumanensis Roy, 1971, has wings of 6.5-8.3 mm length which I estimate to be about 35-40% of the abdomen length. In addition, wing length can vary in different populations of the same species: thus M . misana (Giglio-Tos, 1911) from Loma has wings that are 14-17 mm long whilst females from Nimba have wings 9-12 mm long (Roy, 1971). Presumably once female mantids ceased to fly selection then favoured a progressive reduction in size of the wings, partly for reasons of economy and partly because the insect can be better camouflaged with smaller wings. Eventually a balance was reached in each species such that the advantages of small wings for crypsis are balanced by the advantage of long or conspicuous wings for startle display. I t is of interest that the species with the smallest wings, M . bitumanensis, is the only one t o have evolved conspicuous colouration: the anal region of the hind wing of the female is tinged with red (Roy, 1971). Large wings in a startle display may intimidate because they give the illusion of increased size, but small wings can also intimidate if they are brightly coloured. Defence against ants and ant mimicry The commonest insect on every shrub and tree in tropical Africa is almost certain to be an ant. Usually each individual plant is dominated by one or (rarely) two species of ants, and these are typically very aggressive towards any insect they encounter which does not belong to their own colony or is not one of their associates (e.g. a scale insect or aphid) (Leston, 1972). If an ant tries to climb on a mantid’s leg the mantid is most likely to fly or run away, or it may kick the ant away so that it falls to the ground. Tarachodes lives on tree trunks which are particularly densely covered with ants since the trunk provides the only route from ground to canopy. It is not surprising therefore that Tarachodes is adapted to living amongst ants. First, ants are probably its principal food under natural conditions. From the third instar until adult, if a Tarachodes sees an ant it immediately chases, seizes, and then eats it. The hunting behaviour of Tarachodes with other insects such as grasshoppers is quite different: it normally watches these and waits until they come near before the final short rush and strike. This difference is probably adaptive since ants have comparatively poor eyesight and are rarely aware of the running mantis when chased, but larger insects such as grasshoppers or flies would see the mantid coming and take evasive action. Second, Tarachodes has a special defensive behaviour against ants. If a mantid is satiated, or if an ant manages to crawl onto a leg without the mantid noticing, then Tarachodes immediately strikes at the ant with the forelegs in such a way that the ant is knocked off the substrate and falls to the ground. Since Tarachodes nymphs are occasionally killed by ants in laboratory cages, DEFENSIVE BEHAVIOUR O F PRAYING MANTIDS 27 and so presumably also in nature, this behaviour must be of considerable importance. Finally, Tarachodes females normally guard their oothecae until the young hatch. Ene (1962) showed that this behaviour reduces the incidence of parasitization of the eggs, but it must also prevent ants from finding and destroying the ootheca. The young mantids are themselves ant mimics and they scatter from the ootheca about an hour after emerging. In one of the four oothecae which I observed at emergence the mother also moved away after an hour. Three of the four mothers did not attack their own young, either ignoring them even when they climbed over her head, or gently knocking them away with her forelegs. One of these mothers struck at the ant Camponotus acvapimensis several times, sometimes knocking it violently away, sometimes capturing and eating it. A second Tarachodes behaved in a similar way towards the ant Oecophylla longinoda, knocking all ants violently away, but not harming her own young. One of the four Tarachodes, however, struck at and ate two Camponotus acvapimensis, and she also struck at but then released her own young. Next day her young were still nearby in her cage and she killed and ate several of them. Evidently the parent can and does distinguish her own young from ants, and for at least a few hours she has a strike inhibition against her nymphs but not against ants. Gillon & Roy (1968) show photographs of Galepsus toganus and Pyrgomantis pallida resting on their oothecae, and they state that Pyrgomantis feeds on Camponotus acvapimensis, one of the commonest grassland ants in West Africa. It is likely, therefore, that these species have similar relationships with ants as has Tarachodes afzelii. The first three instars of many species of mantid are ant mimics. Since ants are so abundant and are not often eaten by insectivorous birds, mimicry of ants probably gives good protection to small mantids. In my earlier paper I described the resemblance between first instars of Tarachodes afzelii to Camponotus acvapimensis, Sphodromantis lineola to Oecophylla longinoda, and Miomantis paykullii to Pheidole spp. Kumar (1973) has also described the resemblance between Miomantis paykullii and Tetramorium sp., Polyspilota aeruginosa and Pheidole spp., Phyllocrania paradoxa and Camponotus acvapimensis, and Catasigerpes occidentalis and C. acvapimensis. To these can now be added the following: Miomantis aurea. The first three instars mimic Oecophylla longinoda. The first instar is orange-red with black eyes and a black spot posteriorly on the thorax. The second and third instars are similar but lack the black spot on the thorax. The fourth instar is more reddish brown and less like Oecophylla, and the fifth instar is mottled brown and is cryptic. Later instars may be either green or brown. B-ohierodula ornatipennis. Four small mantids, probably in the first instar, were examined. They were similar in colour to the dark forest form of Oecophylla longinoda with the head brownish cream; eyes pale grey; thorax and forelegs black, but with a cream spot posteriorly on the thorax; abdomen mottled red-brown and black above, black below; legs pale greyish yellow. Later instars were green and cryptic. Pseudocreobotra ocellata and Panurgica compressicollis. The first instars of 28 M. EDMUNDS both of these species are shiny black and resemble Camponotus or Crematogaster spp. The closeness of the similarity between mantid and ant varies; thus mantids resembling Oecophylla are very similar indeed. Some black mantids could be said to resemble a variety of ants, for example Kumar (1973) and I give different model species for Miomantis paykullii (see above) and the similarity of Phyllocrania to an ant is not close since it has a pronounced vertex. The following mantids do not have ant mimicking first instars: Stenovates strachani. The young are elongated, like miniature editions of the adult, but they rest in typical praying posture until the 4th instar when they first start to rest with forelegs extended laterally as does the adult (Edmunds, 1972). Tenodera superstitiosa. The young examined by myself and by Kumar (1973) are elongate and do not resemble ants. This conflicts with the report of Shelford (1903) to the effect that they are ant mimics. Pseudoharpax virescens. The first instar is mottled brown, but becomes greener in the second and third instars. Chloroharpax modesta. The first instar is brown and is not an ant mimic. Oxypilus hamatus. The first three instars have a dark brown head and abdomen, buff pronotum and forelegs. They are contrastingly marked and do not resemble any of the common ants at Legon, contrary to my 1972 statement. The young mantids referred to in that paper which were shiny black must belong to another species, but unfortunately they were not reared to maturity. Thus in ten out of the fifteen species of mantid examined, the first instar nymphs are ant mimics (summarized in Table 5 ) . These ten all belong to the Mantidae or the Hymenopodidae. The five whose first instar nymphs do not mimic ants (or at least not closely) belong to the Mantidae, Hymenopodidae and Vatidae. In the case of Stenovates it is probable that the extreme specialization of the later instars and adults as stick mimics renders ant mimicry of the first instar impossible. There is a limit to the amount the body can increase in length at each moult, and it is likely that the first instar must of necessity be elongated in such exceptionally long mantids as Stenovates. Thus it is unlikely that ant mimicry occurs in any very long species of mantid. The record (above) that Tenodera nymphs do not resemble ants supports this hypothesis since this mantid belongs to a family whose nymphs normally do resemble ants, but Tenodera is a very elongate species. Since so many mantids have ant-mimicking young one must suppose that there is protective value in the resemblance, but I know of no evidence in the literature to support this view. Further, one might expect selection to favour any behavioural mechanism leading to a positive association between an ant-mimicking mantid and the particular species of ant which it resembles. In the case of Sphodromantis lineola at Legon I have searched carefully 3 3 1 shrubs and large herbs and scored each for presence or absence of Oecophylla longinoda and for presence or absence of first instar Sphodromantis lineola which mimic Oecophylla. The results, given in Table 6, indicate a positive association between the mantids and Oecophylla. This association could arise either through the female ovipositing amongst Oecophylla colonies rather than elsewhere, or through the young dispersing mantids resting preferentially near DEFENSIVE BEHAVIOUR OF PRAYING MANTIDS 29 Table 5 . Characteristics of nymphs of Ghanaian mantids Family Species Subfamily Amorphoscelidae A morphoscelis lagrecai Mantidae Tarachodinae Tarachodes afzelii Mantinae Cataspilota misana Prohierodula ornatipennis Polyspilo ta aeniginosa Tenodera superstitiosa Sphodroman tis lineola Mioman tis aurea Mioman tis pay kullii Ant mimic foreleg display + + - + + + + Hymenopodinae Pseudoharpax virescens Chloroharpax modes ta Panurgica c o mpressicollis Pseudocreobotra ocellata Vatidae boxing display + Hymenopodidae Epaphroditinae Ph yllocrania paradoxa Acromantinae tail display target display - + + Oxypilus hama tus Ca tasigerpes occiden talis - + Stenova tes strachani - Table 6. 3 3 1 plants at Legon scored for presence or absence of Sphodrornantis lineola nymphs and Oecophylla longinoda workers Oecophylla Sphodroman tis present absent Total present 5 97 102 absent 1 228 229 Total 6 325 331 Fisher’s exact test (including the more extreme cases of 6:O and 96:229) gives P = 0.01 17 to Oecophylla, or through young mantids elsewhere suffering higher mortality than do mantids close to Oecophylla. Since all six young which I found during the survey were single individuals, they had evidently already dispersed; hence their association with Oecophylla must be maintained either by the second or 30 M. EDMUNDS the third of these mechanisms. I have no data on whether or not oothecae are laid close t o Oecophylla colonies. DISCUSSION OF SECONDARY DEFENCE Function o f the startle display Several points need to be added to my 1972 discussion of secondary defence in mantids. First there are four possible functions to the display which I have called a startle display: 1. For recognition o f a sex partner. I have observed mating in Miomantis paykullii, Sphodromantis lineola, Tarachodes afzelii, and Oxypilus hamatus, but I have never observed mantids to display prior to a copulation attempt (Edmunds, 1972, 1976). Kumar (1973) also reports no display prior to mating in Miomantis paykullii, Sphodromantis lineola, Tenodera superstitiosa and Catasigerpes occidentalis. On one occasion a female Sphodromantis did display to a male, but she was not in a condition to mate: although the male mounted her, copulation did not occur for at least three days during which time she laid an ootheca. Hence whilst it is possible that the display signifies to the male that the female is not available for mating, there is no evidence that it forms a normal part of courtship. 2. For demarcation o f territory. I t is possible that the startle display is often given by conspecifics and that it results in one or both insects retreating. This would result in spacing out and hence reduce the likelihood of cannibalism. I have never observed a display occurring in adult mantids kept in the same or in adjacent containers, but it does occur in nymphs of some species such as Tarachodes afzelii. This is further discussed in the section on territoriality. 3. For luring prey to the mantid. Wickler (1968) illustrates the dramatic display of the African Idolium diabolicum. This species has swollen, brightly coloured forelegs which are reputed to act as a lure to insects which mistake the mantis for a flower. However, Carpenter (1921) describes how an Idolium reared up and displayed when it was approached by a monkey which backed away in response t o the display. The related genera Idolomorpha and Hemiempusa also have bright colours on the forelegs which are only exposed when they are attacked, so it is clear that in Idolium too the posture is an anti-predator display. In addition, I have found that it is almost impossible to induce a displaying Sphodromantis to strike at a prey insect, and this indicates that it is unlikely that the display of Idolium could ever evolve into a lure to attract Prey * 4.As an anti-predator display. The display can usually be elicited by attacking the mantid in a manner simulating attack by a predator. It can also be elicited by presenting the mantid with a bird or with a model of a bird (Maldonado, 1970). Maldonado has also shown that the display of Stagmatoptera biocellata does actually intimidate several species of birds, although the more persistent troupials often succeeded in killing the mantid. In natural situations I have observed a single Sphodromantis display when approached by a lizard (Edmunds, 1972), and I have several times seen male Tarachodes afzelii trapped in a spider's web display and slash at the approaching spider (Nephilengys DEFENSIVE BEHAVIOUR OF PRAYING MANTIDS 31 cruentata). Hence it is reasonable to conclude that the display is directed against predators, especially birds. Following Maldonado (1970) the startle display of mantids, which appears to be designed to frighten predators, can be called a deimatic response. Two types of deimatic display can be distinguished: in Sphodromantis and Polyspilota the display increases the apparent size of the insect, and in these and other species the display is sustained and there may also be exposure of bright colour marks, stridulation and periodic slashing at the stimulus source. In female Danuria, however, the insect is so elongated and comparatively feeble in appearance that raising the wings does not increase its apparent size to any marked extent. In this case, the display simply involves sudden exposure of bright colours which then vanish quickly so that the display is more in the nature of a flash. Although I have not been able to examine a healthy Leptocola, its wing pattern is such that I consider it probable that it too has a startle display involving sudden opening and closing of the wings. Obviously the importance of such a startle display involving a quick flash of colour or of eyespots can only succeed if the insect then becomes cryptic again, and this sort of display only occurs in species with elaborate primary defence. Variation in the startle display In my 1972 paper I drew attention to the variation in colour of spots on the forelegs of Sphodromantis lineola .Further data are now available. Between November 1967 and August 1973, 684 male mantids were examined at Legon. 18.5% have a black spot on the femur; 6.3%have red or blackish red spots on the coxa instead of the usual yellow spots; and 15.2%have either more or less than the usual eight spots on the two forecoxae. Since November 1970 I have collected data on rainfall and also on population size, but there is no obvious correlation between variation in the foreleg spots and either of these parameters, nor could I detect any significant seasonal changes in frequency of any of the morphs. Colour of foreleg markings also varies in Mantis and Paramantis. Thus although all four Paramantis prusina that I have examined from Ghana have red spots, elsewhere in its range the spots may be black or white (Roy, 1965). Intrapopulation variation in spotting may prevent predators from learning that a particular display pattern is a bluff, whilst interpopulation variation may represent adaptations towards intimidating different species of predators. Evolution of ocelli There are a pair of large ocelli on the forewings of Pseudocreobotra and Chlidonoptera. These two genera are closely related to Panurgica on morphological grounds. They all belong to the subfamily Hymenopodinae; they have similar nymphs (ant-mimicking first instars, and with a “target” mark in the later instars); and they have similar startle displays with the forelegs held extended rather than flexed. Species of Chlidonoptera have rather less precise ocelli than do Pseudocreobotra, and males have less circular markings than females (Fig. 7) (Roy, 1964). But the problem remains of how the ocellus first 32 M. EDMUNDS Figure 7. Ocelli in hymenopodid mantids. Left: female Chloroharpax modesra; centre: male and female Chlidonoptera lestoni; right: male Pseudocreobotra ocellara. Green areas are stippled; yellow and whitish areas are left white. In Chloroharpax the ocellus on the left forewing is visible through the transparent greenish right forewing which lies above it. In the other three mantids the ocellus on the lower wing is almost exactly overlaid by the ocellus of the upper wing so it cannot be seen in the resting insect. Based on colour transparencies and mounted specimens. started to evolve. Many mantids have a small spot on the forewing. In Chloroharpax this takes the form of a small black circle with a yellow centre which perhaps resembles a leaf blemish or is a disruptive mark. The two spots are also conspicuous during the startle display so selection could lead to a gradual enlargement of the spots until a pair of ocelli formed. At the same time, since one forewing is always visible in the resting insect, such a mark must not be too conspicuous or else it might attract predators to the otherwise cryptic insect. In the males of Chlidonoptera it appears that selection has led to crypsis and active escape by flight (since they are lighter than the females) rather than to perfection of the ocelli. But in the females, which cannot fly off so quickly when disturbed, selection has led to more effective ocelli for intimidation of predators even if this renders the primary defence less effective. In Pseudocreobotra both sexes have equally perfect ocelli. I assume that these ocelli intimidate predators during display, and that they are directed against birds, but the only evidence to support this view is the work of Blest (1957) on models with eyespots and on butterflies, using birds as predators. Similar large ocelli on the forewings also occurred in the Carboniferous protorthopteran insect Protodiamphipnoa (Carpenter, 1971). This insect also had raptorial forelegs, but it apparently evolved quite independently from modern mantids. Presumably the ocelli of this insect were directed against amphibians or reptiles since there were no birds or mammals at that time. TERRITORIALITY IN MANTIDS Following MacKinnon’s description of territorial behaviour in the target mantis Pseudocreobotra wahlbergi, I have investigated several other species of DEFENSIVE BEHAVIOUR OF PRAYING MANTIDS 33 mantid to see if spacing out behaviour occurs in these as well. Four types of territorial behaviour have been observed: 1. Target display. When nymphs of Pseudocreobotra ocellata and Panurgica compressicollis see another nymph of the same or different species but of similar size to themselves, they pump blood into the abdomen so that it swells up, thus rendering visible a large circular “target” mark on the abdomen (Plate 2B). This mark in Panurgica is dark blue and yellow, and during display the forelegs are either held extended laterally (as during a startle display) or boxed in a way not unlike the boxing display of Oxypilus. This similarity in behaviour between the two genera Panurgica and Pseudocreobotra suggests that they are closely related (see section on evolution of ocelli). Evidence that this display serves to space out individuals is given by MacKinnon (1970). Figure 8. Tail display of nymph of Amorphoscelis lagrecai. The tip of the abdomen and the cerci are moved alternately left and right as indicated by the arrows. Based on a colour transparency and sketches from life. 2. Tail display. Nymphs of Amorphoscelis lagrecai have enormously elongated anal cerci compared with adults or with other mantids. If two nymphs are placed together they display by moving the tip of the abdomen in a circle so that the cerci are conspicuously waved, rather like a semaphore signal, first to one side then to the other (Fig. 8 ) . This is done particularly when about to move and when coming to rest after having moved. Unfortunately these insects are very difficult to rear away from their normal habitat (tree trunks in humid 3 34 M. EDMUNDS forest) and it was not possible to study the behaviour further. Nevertheless it seems likely that this display serves to inform other nymphs of the presence of a mantis, and if this reduces cannibalism by spacing individuals out, it may benefit the species. 3. Boxing display. Oxypilus, Catasigerpes, Anasigerpes and Chrysomantis all give a boxing display, particularly in the presence of another insect of the same species, and I have already (1972) given evidence suggesting that the display serves to space out conspecific males. A possible boxing movement has also been observed in one male Nilomantis, but further observations are required to determine if it occurs regularly in this genus. In Oxypilus hamatus and Catasigerpes occidentalis young nymphs also display to one another and usually one of the two eventually retreats, thus indicating that the display results in spacing out of individuals. Spacing out by means of boxing displays probably occurs in all genera of the Acromantinae. In Oxypilus the colour of the forelegs in the first three instars is quite different from that of later instars or of adults. If there is a marked difference in size between two nymphs it is probably more advantageous to the smaller one to retreat quickly rather than stay put, display, and perhaps be captured and eaten. Hence if late instars have a different pattern to early instars, this enables the small nymphs to recognise quickly that an approaching mantis is not of comparable size to themselves, and to take evasive action. In Anasigerpes bifasciata there is widespread geographical variation in the foreleg pattern (Roy, 1966), but it is not known if each population has a single pattern, or if populations are polymorphic for this character. 4.Foreleg display, If two fourth instar nymphs of Tarachodes afzelii are placed together in the same cage, they often display at one another. The display is exactly like the adult startle display apart from the absence of wings: the head and abdomen are raised from the substrate, and the forelegs are abducted, fully flexed, thus displaying black and pale marks on their inner surfaces. Occasionally the forelegs may be extended laterally for a second or two and then flexed again. One or both insects may maintain a display for up t o five minutes. Usually one of the nymphs then moves away and the remaining insect eventually returns to a normal resting posture. Since the display is given towards conspecifics, and since one usually retreats, I consider it probable that the function of the display is to space out individuals, and so to reduce cannibalism. The colour marks on the forelegs of all instars of this species are illustrated by Gillon & Roy (1968). I have not observed males or females display to conspecifics, so it appears that the display is confined to nymphs. For several other species of mantid whose nymphs I have reared I have not observed any behaviour which might possibly function t o space out individuals (summarized in Table S), but obviously careful study involving several individuals would be required to be certain that it does not occur. In both Tarachodes and Pseudocreobotra the movements associated with territorial display are similar to those used by adults giving an anti-predator startle display. On the available evidence it is impossible to say which display evolved first, but in both displays the mantis shows a reduced tendency to make a predatory strike. It is perhaps surprising that there is so little evidence of similar movements being used in courtship displays. If the displays I have DEFENSIVE BEHAVIOUR OF PRAYING MANTIDS 35 described above do indeed space out individuals then presumably the colour marks can be perceived by conspecifics. Yet although Oxypilus male and female do box during courtship (Edmunds, 1976) I have never observed Sphodromantis or any other large mantid displaying prior t o mating. SUMMARY The defensive behaviour of 18 species of mantids is described in this paper and further details are given of the defences of 12 species whose behaviour was described earlier (Edmunds, 1972). As a result the defences of 37 genera of Ghanaian mantids are now known. Most species of the Mantinae lack a specific resemblance to particular objects but have a greedbrown polymorphism, and evidence is presented showing that there is a correlation between background colour and colour of the mantid in Sphodromantis lineola and Miomantis paykullii. Most other species of mantid have a highly specific resemblance to bark, grass, sticks or leaves, and the evolution of these insects is discussed. Fire melanism occurs in Galepsus toganus and Pyrgomantis pallida. Evidence is presented which suggests that wing reduction in females results in decreased predation by birds. Wing reduction gives better primary defence but the wings cannot be completely lost if they are also used for a startle display in secondary defence. Many species of mantid have first instars which mimic ants and the fact that S. lineola first instars have a positive association with the ant Oecophylla indicates that ant mimicry is of defensive importance. Species which have relatively unspecialized adults normally have ant-mimicking nymphs (e.g. most of the Mantinae), but some species which have highly modified adults do not have ant-mimicking nymphs, probably because already at the first instar the nymph is “preadapted” t o becoming an adult and so cannot resemble an ant. Tarachodes afzelii not only has ant-mimicking nymphs but also lives amongst and feeds on ants. I t has a special technique for knocking off ants which come near it and which it cannot eat. The functions of startle displays and the evolution of ocelli in Pseudocreobotra are discussed. Territorial displays which result in spacing out of nymphs are described. These result in reduced cannibalism and they may also serve to partition out the available food resources. Not all mantids have territorial displays, but four different displays have evolved in different species which appear to have this function. ACKNOWLEDGEMENTS I t is a pleasure to thank Dr P. Grubb, Dr R. Kumar, Dr D. Leston, Mr E. A. Aryeetey, Mr J . H. Hynes, Mr D. Louis, Mr S. B. V. Mkhize and Mr S. Mensah for providing me with specimens of mantids, especially forest ones, and for their interest in my work. I am also grateful t o my wife, Janet, for discussion at various stages of the work and for critically reading the manuscript, and to Dr D. Leston, Dr D. R. Ragge and Dr R. Roy for help with identification of some of the species. This work was prepared for publication during the tenure of an Inter University Council Resettlement Fellowship at Exeter University. 36 M. EDMUNDS REFERENCES BALDERRAMA, N. & MALDONADO, H., 1973. Ontogeny of the behaviour in the praying mantis, J . Insect Physiol.. 19: 319-36. BARNOR, J. L., 1972. Studies on colour dimorphism in praying mantids. 85 pp. M.Sc. thesis, University of Ghana. BLEST, A. D., 1957. The function of eyespot patterns in the Lepidoptera. Behaviour, J I : 209-56. CARPENTER, F. M., 1971. Adaptations among Paleozoic insects. Proc. N. A m . Paleontological Convention 1969, p t . I: 1236-51. CARPENTER, G . D. H., 1921. Experiments o n the relative edibility of insects, with special reference to their colouration. Trans. R. ent. SOC.Lond., 54: 1-105. COTT, H. B., 1940. Adaptive coloration in animals, 508 pp. London: Methuen. EDMUNDS, M., 1972. Defensive behaviour in Ghanaian praying mantids. Zool. J. Linn. SOC.,51: 1-32. EDMUNDS, M., 1974. Defence in animals. A survey of anti-predator defences, 357 pp. Harlow: Longman. EDMUNDS, M., 1976. Courtship, mating and possible sex pheromones in three species of praying mantids. Entomologist’s rnon. Mag., in press. ENE, J. C., 1962. Parasitisation of mantid oothecae in West Africa. Int. Congr. En?., XI, W e n , Band 2: 725-7. GILLON, Y. & ROY, R., 1968. Les mantes d e Lamto et des savanes de CGte d’Ivoire. Bull. Inst. fond. Afr. noire (Sir. A ) , 30: 1038-1151. HOCKING, B., 1964. Fire melanism in some African grasshoppers. Evolution, Lancaster, Pa., 18: 332 3 5 . KUMAR, R., 1973. The biology of some Ghanaian mantids (Insecta: Dictyoptera), Bull. Inst. f o n d . Afr. noire (Str. A ) , 35: 551-78. LESTON, D., 1972. The ant mosaic: a basic property of the tropical forest zone. Abstr. Int. Congr. Ent., XIV, Canberra: 328. MACKINNON, J., 1970. Indications of territoriality in mantids. Z. Tierpsychol., 27: 150-5. MALDONADO, H., 1970. The deimatic reaction in the praying mantis Stagmatoptera biocellata. Z . vergl. PhySiOl., 68: 60-71. RAGGE, D. R. & ROY, R., 1967. A review of the praying mantises of Ghana (Dictyoptera Mantodea). Bull. Inst. fond. Afr. noire (Sir. A ) , 29: 586-644. ROY, R., 1964. Les mantes de la CGte d’lvoire forestikre. Bull. Inst. fr. Afr. noire (Shr. A ) . 26: 735-93. ROY, R., 1965. Les mantes d e la GuinCe forestihre Bull. Inst, fr. Afr. noire (Shr. A ) , 27: 577-612. ROY, R., 1966. Notes sur le genre Anasigerpes (Mantodea Hymenopodidae). Bull. Inst fr. Afr. noire (Sir. A ) , 28: 12845. ROY, R., 1971. Dictyoptera Mantodea. In: Le Massif des Monts Loma (Sierra Leone), fascicule 1 . Mem. Inst. f o n d . Afr. noire, 86: 241-248. ROY, R. & LESTON, D., 1975. Mantodea of Ghana: new species further records and habitats. Bull. Inst. fond. Afr. noire (Sir. A ) , 37, in press. SHELFORD, R., 1903. Bionomical notes o n some Bornean Mantidae. Zoologist, 7: 293-304. SIEGEL, S., 1956. Nonparametric statistics f o r the behavioral sciences: 312 pp. New York: McGraw-Hill; and Tokyo: Kogakusha. WICKLER, W., 1968. Mimicry in plants and animals, 255 pp. London: Weidenfeld & Nicholson. NOTE ADDED IN PROOF Roy & Leston’s 1975 paper (referred to above) makes several changes of name and of allocation to families, but it has appeared too late to modify this paper. The main name changes relevant to this paper are as follows: Amorphoscelis maculata becomes Maculatoscelis maculata; A. lagrecai becomes Caudatoscelis lagrecai ; Galepsus toganus becomes Paragalepsus toganus ; and Stenovates strachani becomes Heterochaeta strachani. EXPLANATION O F PLATES PLATE 1 A. Deimatic (startle) display of female Sphodromanth aurea. B. Primary defence of t h e bark mimic, male Theopompella wesrwoodi. C. Primary defence of female Chloroharpax modesta. Note the small eyespots which may perhaps be disruptive. Zool. r. Linri. SOC.,58 (1976) M. EDMUNDS Plate 1 (Facing p . 36) Zoo]. J . Linn. Soc., 58 (1976) M. EDMUNDS Plate 2 DEFENSIVE BEHAVIOUR O F PRAYING MANTIDS D. Primary defence of the short stick mimic, male Cutusigerpes occidentulis. Note the position of the body with the abruptly truncated wings touching the substrate. PLATE 2 A. Primary defence of t h e leaf mimic, female Panurgicu compressicollis. B. Final instar nymph of Panurgicu cornpressicollis showing the ‘target’ mark on the abdomen. C.Primary defence of the stick mimic, male Popu undutu. D. Resting position of female Chlidonopteru lestoni. 37
© Copyright 2026 Paperzz