Descriptions are based mainly on the final feeding instars. Examination of larvae by GG, and his photographs of reared larvae at various stages, showed that the mature larvae differ very little in morphology or colouration from earlier instars. However, the colour pattern, particularly the black markings, becomes more strongly developed as the larva matures.

Xenapates Kirby, 1882

Figs 3–10, 11–12

[based on the larvae of X. braunsi and X. gaullei]

Description

Head. Slightly wider than high (to epistomal suture). Without surface sculpture. Antenna with 5 antennomeres on large subconic antacorium. 4 basal antennomeres ring-like, apical antennomere peg-like. Clypeus with 4 setae. Postclypeus about half as long as preclypeus. Preclypeus with medial, laterally sclerotised (brown), transverse division (ca. medial third unsclerotised). Labrum slightly asymmetric (right lobe larger than left), undivided by longitudinal or transverse depressions, with 6 setae. Stipes and palpifer of maxilla each with a prominent lobe directed towards mandibles. Left mandible (Figs 3, 4) with straight-edged blade on outer (dorsal) surface, ending with a tooth on left side; a long tooth medially, contiguous with a well-developed mesal ridge, its edge produced in two rounded lobes, separated by a wide, rounded emargination (Fig. 3); inner surface with three well-defined teeth. Right mandible (Fig. 5) outer surface similar to left mandible, but straight-edged blade not ending in tooth; medial tooth very broad; inner surface with one wide, asymmetrical tooth closely adjoined by a pair of short, narrow teeth on a shared, rectangular flange; to left of this a straight, unsclerotised edge, about as long as three adjacent teeth together, and much lower than these; no mesal ridge present.

Figures 3–10. 

Xenapates larvae: 3. X. braunsi left mandible, outer face; green line along edge of mesal ridge. 4. X. braunsi left mandible, ventral; (a) medial tooth, (b) inner face. 5. X. braunsi right mandible, inner face. 6. X. gaullei right pro- and mesothoracic legs; (c) expanded apex of femur. 7. X. gaullei thorax, lateral. 8. X. braunsi proleg and ventral part of abdominal segment 8, external surface; (d) reticulate-spiculate surface structure on surpedal lobe; (e) setae on proleg (some missing). 9. X. braunsi mature larva; dorsal, lateral and ventral (from left). 10. X. gaullei mature larva.

Cuticle of thorax and abdomen. Without dorsal macrostructures such as tubercles, spines, or longer setae, but with reticulate-spiculate surface structure on dorsum reaching to just above spiracles (Fig. 7), sometimes also on subspiracular and surpedal lobes (Fig. 8). To the naked eye, these parts appear “velvety”, but at magnifications of 40–90× under a stereo microscope they might almost be described as “scaly”. The cuticle thus strongly resembles that of the easy bleeder species Rhadinoceraea micans (Klug, 1816) (Blennocampinae) as described by Boevé (2009).

Thorax. Meso- and metathoracic substigmal lobes strongly protruding laterally. Prothorax with more or less strongly developed dorso-lateral lobes. Legs with 5 articles, bearing numerous long setae. Coxa longer than basal width. Trochanter wider than long; about half as long as coxa. Femur apically expanded on inner surface (Fig. 6); femur thus much wider apically than apical width of trochanter, and about twice as wide as base of tibia.

Abdomen. Segments 1–8 with 6 dorsal annulets. Prolegs on segments 2–8 and 10; 1.5–1.7× as long as basal width, with 16–20 setae on outer surface (Fig. 8) and 12–20 on inner surface. Apex of suranal lobe and antero-dorsal surface of subanal lobe with numerous short setae.

Xenapates braunsi (Konow, 1896)

Figs 3–5, 8–9, 11

Description

Colour (Fig. 9). Head including posterior vertex yellow, changing to white on genae. A black patch, not more than 4× as long as ocularium, surrounds stemmatum and does not reach an imagined horizontal line extended from dorsal apex of frons. Thoracic legs pale brown, except for black coxal suture. Trunk pale green, appearing dorsally darker because contents of gut show through; subspiracular lobes and anal area including anal prolegs pale orange.

Head (Fig. 11). With a pair of large, pale depressions, one each side of coronal suture dorsal of junction with frontal suture. Setae on vertex and parietals very short (less than third diameter of antennomere 1) and pale. Some setae on genae and mouthparts as long as antenna. Right lobe of labrum about twice as long as visible part of left lobe.

Figures 11–18. 

Xenapates immature stages: Larva; head frontal. 11. X. braunsi 12. X. gaullei. Prepupa; ventral, lateral and dorsal (from left). 13. X. braunsi 14. X. gaullei. Pupa; ventral, lateral and dorsal (from left). 15. X. braunsi 16. X. gaullei. 17. X. braunsi: cocoon containing pupa. 18. Vacated egg pocket of X. gaullei, exuvia of first instar larva visible in perforation, with feeding hole (at left) made by second instar. Scale bars = 1 mm.

Trunk. Prothorax dorsal annulets 1 and 2 laterally fused. Annulet 2 medially divided; lateral lobes very much higher than lateral parts of 1 or 3. Dorsal surface of lobes on annulet 2 with partly pigmented spicules that are more strongly developed than on other parts of thoracic dorsum. Surface of integument of abdomen above spiracles almost smooth, with very minute, unpigmented spicules. Some partly dark spicules on hypopleurite, surpedal and substigmal lobes (Fig. 8).

Length of fully grown larvae: 19–20 mm (n = 16).

Figs 13–16

The colour pattern of the larva of Xenapates gaullei is largely lost at the moult to the prepupal stage (Fig. 14), whereas the prepupae of X. braunsi (Fig. 13) retain, for example, extensive black markings on the head, and the orange anal area. The pale depressions on the faces of larvae of X. braunsi (not present in X. gaullei) at the base of the coronal suture adjoining the frons are retained in the prepupa (Fig. 13), thus further facilitating their separation. The pupae are at first only weakly pigmented. At this stage they can nevertheless be distinguished by colour: X. braunsi (Fig. 15) is predominantly bright green, whereas X. gaullei (Fig. 16) has a yellow ground colour with only the abdomen tinged slightly green. As pigmentation develops, the much more extensively dark colour pattern of X. braunsi easily distinguishes it from X. gaullei.

X. braunsi: Poaceae: Digitaria horizontalis Willdenow (Jamaican crabgrass), Pennisetum purpureum Schumacher (elephant grass), Setaria barbata (Lamarck) Kunth (bristly foxtail grass), Zea mays Linnaeus (maize).

All four plant species were observed to be hosts under field conditions. Though adults were regularly found on Z. mays during the wet seasons, larvae were rarely observed feeding on this species, compared to the other host plants. A few larvae collected in the field on maize were successfully reared to the cocoon stage on potted maize plants.

X. gaullei: Commelinaceae: Commelina benghalensis Linnaeus (Bengal dayflower) and C. communis Linnaeus (Asiatic dayflower). Both species were found to be hosts under field conditions.

Adults of Xenapates braunsi and X. gaullei were frequently observed courting and mating on leaves of Alternanthera brasiliana (Linnaeus) Kuntze (Amaranthaceae), which offer a relatively large surface area on which the sawflies can engage with each other. These activities took place particularly during sunny periods following rainfall. Neither oviposition nor occurrence of larvae was observed on A. brasiliana. Despite the provision of honey solution, adults never lived longer than three days in captivity. Compared to other sawflies, the incubation period of X. braunsi and X. gaullei eggs is extremely short (lasting a maximum of five days: see Table 1 and below). Larvae of both species are easy bleeders (Boevé 2009). Rupture of the cuticle and emission of haemolymph was induced even by light contact with a 000 size paint brush. No parasitoids were observed or reared, but attacks on cocoon stages by nematodes were frequent when the soil was too wet.

X. braunsi: Oviposition was not observed during this study, but 3 mm long larvae were detected on young leaves of Setaria barbata five days after the introduction of adults to the experimental cages. Larval feeding followed a similar pattern to X. gaullei, except that early instars first grazed on the leaf surface without perforating it. Later, the still young larvae moved to the leaf edge and fed on its margin. They were then also observed to feed on older leaves. When mature, larvae stopped feeding and entered the ground, without moulting. The cocoon (Fig. 17), made of sand grains bound together by a secretion, was ellipsoidal with largest dimensions approximately 7 by 11 mm. The external surface of the cocoon was rough, the inner surface smooth. It was not possible to remove the sand grains from the fabric of the cocoon without destroying it. The larva moulted slowly to the prepupal stage, shedding its exuvia caudally. The pupa became fully developed only a few days before emergence of the adult. The intensely green-coloured haemolymph of the cocoon stages of X. braunsi was striking, and remained highly visible through the integument of freshly emerged adults (Fig. 1).

X. gaullei: Eggs were found mostly singly on sub-apical leaves, but not on the youngest, apical leaves. Oviposition was into a slit cut in the mesophyll from the upper side of a leaf, at no specific locus: eggs were found close to the leaf edge or at some distance from it, but never in the mid-rib. On average, larvae left their egg cavities (Fig. 18) to feed externally 5 days after oviposition (n = 11). The first moult, however, takes place just before the larva exits the leaf. It seems likely that larvae of the first instar, which must be of very short duration, feed within the leaf. Although such feeding was not observed directly, the cavities surrounding hatched eggs appeared too large to have been caused by mere disruption during oviposition, or eclosion of the larva. Whereas young larvae made feeding holes in the leaf blade, close to where they eclosed (Fig. 18), older instars fed directly on the leaf margin. One to two days before they burrowed into the soil, mature larvae stopped feeding, evacuated their gut and thereby became slightly smaller. Each larva then made a cocoon in the earth, close to the bottom of the plant pot. Cocoons were never recovered from within the root system of the host plant. The cocoon resembled that of X. braunsi.

Although too few adults were reared to allow definite conclusions to be drawn on voltinism, field observations of adults in 2012 through 2014 indicated that Xenapates braunsi and X. gaullei are multivoltine, with two or three generations per year. Possibly the number of generations depends on prevailing weather conditions. The first generation emerges at the beginning of the main wet season, which lasts from April to late July. In years with early and sustained rainfall such as 2013, a second generation of X. braunsi can emerge as early as 11–14 days after the first larvae have burrowed into the ground, i.e. 34–36 days after the egg was laid (Table 1). In X. gaullei the equivalent period spent in the ground was 19 days, and complete development from the egg took 40–42 days (Table 1). However, in years with normal onset of rain (e.g. 2012 and 2014), the immature stages take longer to develop, and adults of the second generation emerge during the short, less intense wet season that starts between late September and early November. Offspring that have developed during the latter period undergo a prolonged diapause and adults emerge in April of the following year, after having spent about 193 days in the soil in X. braunsi and 202 days in X. gaullei (Table 1). Under field conditions in southern Benin, three generations were less often observed in X. gaullei than in X. braunsi. Whereas X. braunsi appeared to have an approximately 1:1 sex ratio during the short wet season, females of X. gaullei seemed far less frequent than males during the same period. During the long wet season, females were apparently equally frequent in both species. Slight protandry was observed to be usual, in both species.

The phylogenetic position of Xenapates within the Tenthredinidae, as in several other lineages traditionally placed in the Allantinae, remains unclear. Xenapates larvae differ from previously described exophytic tenthredinid larvae (Lorenz and Kraus 1957), in lacking on the dorsal annulets of the trunk any cuticular appendages such as tubercles or setae. Other sawfly taxa whose larvae have an extremely reduced complement of macroscopic cuticular structures belong notably to the tribes Waldheimini (e.g. Waldheimia) and Blennocampini (e.g. Tomostethus) of the Blennocampinae (Smith 1969, Smith and Williams 2014). The complete absence of these structures might be an apomorphy of Xenapates. However, as long as this character state remains unknown in most other Afrotropical and Oriental genera placed in the Allantinae and Blennocampinae, its taxonomic significance cannot be evaluated. Of the taxa that were included in the Xenapateini by Koch (1996), only the larva of Takeuchiella pentagona Malaise, 1935 has been described (Okutani 1959): annulets 2 and 4 of the abdominal segments carry conical tubercles and setae. In some characters, the larvae of Xenapates agree with those of Allantini as characterised by Lorenz and Kraus (1957): antenna with 5 members, clypeus with 4 setae, abdominal segments with 6 annulets, and abdominal prolegs on abdominal segments 2–8 and 10. However, the same combination of character states is also found in some Blennocampinae (Smith 1969). Noteworthy is that the lobed stipes and palpifer of the maxilla of Xenapates are found together only in the allantine genera Allantus and Taxonus, and Tenthredininae (Lorenz and Kraus 1957). On the other hand, larvae of the Allantinae and Blennocampinae taxa described by Lorenz and Kraus (1957) and Smith (1969, 1979) are without setae on the external surface of the abdominal prolegs, whereas Xenapates has at least 16 setae. According to Lorenz and Kraus (1957), in the Tenthredinidae some Athaliinae, Nematinae, Selandriinae and Tenthredininae share this character. The abdominal prolegs of the larva of T. pentagona are setose only on the inner surface (Okutani 1959). The mandibles of Xenapates larvae resemble those of Blennocampinae in that only one mandible has a mesal ridge, rather than the Allantinae, in which both mandibles have a mesal ridge (Smith 1969, 1979). However, in all the Blennocampinae studied by Smith (1969), the right mandible has a mesal ridge, whereas in Xenapates it is only developed on the left mandible. Xenapates also differs from Allantini in its use of monocots as larval hosts. The easy bleeding trait exhibited by X. braunsi and X. gaullei larvae is shared by a number of species in various lineages of Tenthredinidae: Athaliinae, Selandriinae, Tenthredininae and particularly the Blennocampinae, but is not known in the Allantinae (Boevé 2009; note subsequent alteration to family group taxonomy by Malm and Nyman 2014). Although we called the structure made by the last larval instar of Xenapates a cocoon, one could alternatively regard it as a cell formed in the soil, which is held together by a secretion. Smith (1969) notes that Blennocampinae which enter the ground form a cocoon. On the other hand, Smith (1979) states that Allantinae rarely, if ever, spin cocoons and that those which enter the soil form an earthen cell. However, it has seldom been recorded whether or not any secretion was used to hold the walls together, and perhaps in some cases this does occur.

The spectrum of host plants utilised by Tenthredinoidea is very large, with larvae of most taxa feeding on angiosperms, but a large minority on gymnosperms, and rather fewer on ferns, horsetails and mosses (Liston 1995, Vikberg and Nuorteva 1997). However, most available data on host plant associations relate to the temperate zones of the Palaearctic and Nearctic. Very little is known about host plant associations of Afrotropical sawflies, and the pattern in this region might therefore ultimately prove to be quite different. All hosts of the two Xenapates species in this study are placed in the monocots, within a clade called the commelinids, respectively in the Commelinaceae of the Commelinales (X. gaullei) and Poaceae of the Poales (X. braunsi) (APG 2009). Worldwide, the number of tenthredinoid lineages so far known to be attached to monocots in general, and grasses (Poales) in particular, is relatively low. The grass-feeders comprise only some tenthredinid species in the Holarctic and partly Oriental genera Selandria, Dolerus (Selandriinae), Eutomostethus (Blennocampinae), Tenthredo, Tenthredopsis (Tenthredininae) and Euura (as Pachynematus in Lorenz and Kraus 1957; see Prous et al. 2014 on generic placement) (Nematinae) (Lorenz and Kraus 1957). As far as we are aware, X. braunsi is so far the only sawfly whose larvae have been found feeding on Zea mays.

Until now, only one record existed of an Afrotropical sawfly species using a member of the Poales as a larval host. Larvae of a Distega species (Blennocampinae), identified as Distega aff. nigeriae Fors.[ius, 1927], were stated to feed on “Mil” (French for millet; cultivated Pennisetum spp.) in Senegal by Risbec (1950). He noted that during his stay in Senegal, feeding damage to crops by this sawfly did not seem very significant. It is difficult, because of the brevity of the published information and lack of voucher material, to assess the accuracy of identification of the sawfly involved in this record. In particular, it is not stated whether adults were reared from the larvae. The drawing of the larva in Risbec (1950) lacks any indication of the presence of cuticular processes on the dorsum, and its very strongly developed prothoracic lobes also closely resemble those of X. braunsi, which in this study was found to use Pennisetum as a host. Furthermore, Risbec referred only to the presence of setae on the front of the head and on the abdominal prolegs, not mentioning any setae on other parts of the body. This resembles the distribution of setae that we observed in Xenapates larvae. The velvety appearance of the cuticle described by Risbec suggests that it, too, was an easy bleeder. On the other hand, his larvae exhibited a conspicuous row of lateral black spots, just above the spiracles, which are not found in X. braunsi or X. gaullei. We conclude that Risbec’s identification of Pennisetum as a host of Distega requires reinvestigation, because the sawfly larva involved may have been misidentified.

In the Holarctic, some species of Selandriinae and Nematinae whose larvae feed on Poaceae reach levels of abundance sufficient to make them ecologically important in grasslands (for example, the larvae are a major source of nutrition for some bird species: Potts 1986), and occasionally they may damage crops of grass fodder (Mühle and Wetzel 1965), or grain crops (Miller and Pike 2002). It remains to be seen, if grass-feeding sawflies perform similar roles in the Afrotropical Region.