First records of the spider wasps Ctenocerus Dahlbom and Paraclavelia Haupt from Asia, with discussions on the systematics of Ctenocerinae (Hymenoptera: Pompilidae)

. We describe two new species of Ctenocerinae (Pompilidae) from Asia, i.e., Ctenocerus srilankae Shimizu sp. nov. from Sri Lanka and Paraclavelia arabiae Shimizu sp. nov. from Oman. These represent the first records of Ctenocerus Dahlbom, 1845 and Paraclavelia Haupt, 1930 in Asia. Two new combinations and a new synonymy are proposed: Ctenocerus fasciatus (Smith, 1851) (= Micropteryx fasciata Smith, 1851); Paraclavelia decipiens (Arnold, 1932) (= Clavelia decipiens Arnold, 1932); and Ctenocerus Dahlbom, 1845 (= Euclavelia Arnold, 1932). We discuss the difficulty of systematics of Ctenocerinae at generic level. The head and pronotal structures of Ctenocerus and Paraclavelia are presumably specialized for preying on trapdoor spiders. These structures are considered to have evolved independently of other unrelated parasitoid Pompilidae that have been confirmed or proposed to prey on trapdoor spiders. We also discuss the biogeographical distribution of these genera.


Introduction
We dedicate this paper to the recently deceased Pompilidae taxonomists Michael C. Day, Raymond Wahis, and Marius Wasbauer. The pompilid subfamily Claveliinae was first proposed by Haupt (1929) (type genus: Clavelia Lucas, 1851) for pompilids that had seemingly odd modifications of the head and pronotum: the vertex rises behind the ocelli before producing a sharp angle, and the pronotum is very long. Later, Haupt (1930) found further taxa with similar modifications, describing the South American genus Lepidocnemis Haupt, 1930 and the African genus Paraclavelia Haupt, 1930 and included these genera in Claveliinae. In his monograph of the African Claveliinae, Arnold (1932) recognized 11 genera, including Clavelia and Paraclavelia. Afterward, Arnold (1934) synonymized Clavelia with Ctenocerus Dahlbom, 1845. As a result, the subfamily name was changed from Claveliinae to Ctenocerinae Arnold, 1934 (see also Day 1981: 6). Waichert et al. (2015) constructed a molecular phylogeny of Pompilidae Latreille, 1804 with global coverage, from Bayesian and maximum-likelihood analyses of four nuclear DNA markers. The resulting consensus tree showed that the sampled Ctenocerinae, including Trichosalius Arnold, 1934, Paraclavelia, and Ctenocerus, constituted the earliest diverging clade in the Pompilidae, sister to all other subfamilies.
When the first author, AS, visited the Natural History Museum, London in 2004, the late Mr M.C. Day, who had retired from the museum, introduced AS to the Pompilidae collection in the museum that he had been managing. At that time, he showed AS a female and three males of a species from Sri Lanka collected by K.V. Krombein (National Museum of Natural History, Washington D.C., USA) et al. and a female of another species from Oman, and stated that these were true Ctenocerinae from Asia (although Oman is a biogeographically complex country in the Middle East, straddling the Afrotropical, Indo-Malayan and Palaearctic realms). This surprised AS because Ctenocerus and Paraclavelia had been considered to be endemic to the African continent. In fact, Arnold (1932) stated that Ctenocerus was distributed in North Africa and the Ethiopian (now Afrotropical) Region, and Paraclavelia in the Ethiopian (= Afrotropical) Region. In 2019, AS visited the National Museum of Natural History, Washington D.C., USA and inspected the Pompilidae collection there. At that time, he found three female and five male specimens conspecific with the Sri Lankan species mentioned above. AS decided to study the Asian Ctenocerinae based on the above specimens borrowed from the two museums. The results led him the conclusion that both were undescribed species, with the Sri Lankan species belonging to Ctenocerus and the Arabian species belonging to Paraclavelia.
In this paper, we revise the generic characters of Ctenocerus and Paraclavelia, and describe two new species of both genera, which are recorded from Asia for the first time. We discuss the difficulty of systematics of Ctenocerinae at generic level. The head and pronotal structures of these wasps are presumably specialized for preying on trapdoor spiders (Araneae: Ctenizidae). These structures seem to have evolved independently of other unrelated parasitoid Pompilidae that have been confirmed or proposed to prey on trapdoor spiders. We also discuss the biogeographical distribution of these genera.

Morphological analysis
Specimens were photographed with a digital camera (Nikon Coolpix 4500 and MDC lens equipped with a stereo microscope Leitz TS and a transmitted light microscope Leitz Dialux). Photographs were stacked by using combineZM (Hadley 2008) and the final synthesized photographs were post-processed for contrast and brightness using Adobe Photoshop software.
The terminology of general morphology, including the wing veins and cells, follows Day (1988). In addition, the following term is used: antennocular line, the anterior margin of the frons from the antennal base to the eye in dorsal view.
The label information on each specimen of these species is listed in Supp. file 1.

Diagnosis
Ctenocerus is distinguished from other genera of Pompilidae by the combination of the following features.
Redescription (as a complement to Arnold's (1932) description, based on the type species)

Female
MeasureMents. Small to fairly large wasps, 9 to 22 mm in length.

Male
MeasureMents. Much smaller and slenderer than female, 5.5 to 18 mm in length.
Head. Broader than long ( Legs. Apical margin of fore tibia with short, stout, decurved spine or lacking such spine mesally. Fore femur slender, thinner than mid femur (Figs 4C, I, 5G). Fore tarsomeres 2-4 combined as long as or shorter than fore tarsomere 1. Orbiculae of all legs similar to those of female or orbicula of hind leg remarkably small, its pecten absent. Fore and mid tarsal claws bifid; hind tarsal claw bifid, or edentate, rectangularly bent subapically and both claws parallel to each other or slightly divergent.

Distribution
Africa (Palaearctic and Afrotropical Regions) and South Asia (Sri Lanka).

Distribution
South Asia (Sri Lanka).

Diagnosis
This genus has most of the features of Ctenocerus, but differs from the latter in the following.

Female
Lower frons lateral to and ventral to antennal sockets deeply depressed, median area between antennal sockets steeply receding into lowermost transverse depression, hence frontal bridge absent (Figs 6A, D, G, J, 7D). Clypeus flat and polished with deep depression across whole width of its base, depression being broader laterally than medially. Pronotal collar usually situated slightly below level of dorsum (Figs 6C, F, I, L, 7H), but rather deeply depressed in Parac. somalica (Fig. 6N). Fore femur not swollen (Figs 6F, I, 7G).

Male
Supra-antennal area produced anteriorly into frontal ledge overhanging antennal radicle (Fig. 8C, F, I) Lowermost frons across its whole width depressed much below level of supra-antennal area, slightly below level of clypeus, lacking any trace of frontal bridge (Fig. 8A, D, G).
Description (as a complement to Haupt's (1930) description, based on the type species)

Female
MeasureMents. Small to fairly large wasps, 12 to 26 mm in length.
Wings. FW with three SMCs (Fig. 7C). Pterostigma as long as or longer than cross-vein 2r-rs at bottom. Marginal cell lanceolate, acute at apex. Second abscissa of vein M (basal vein) curved. Last abscissa of vein M not attaining outer wing margin. Discal cell 1 usually with indistinct membranous irregularity (fenestra) basally. Cross-vein cu-a originating distal to separation of vein M+CuA, oblique to vein A. HW cross-vein cu-a originating basal to, at (Fig. 7C), or distal to fork of vein M+CuA, confluent with vein A, forming smooth arc.
Legs. Apical margin of fore tibia usually with short, stout, decurved spine mesally (Figs 6M, 7F, arrow) (apical spines short, stout, but not decurved in Parac. caffer). Fore tarsomeres 2-4 combined much shorter than fore tarsomere 1. Mid and hind femora with distinct basal ring. Mid tibia with short spines dorsally. Hind tibia with spines short or rudimentary dorsally and latero-and dorsoapically (Fig. 7J). Tarsomere 5 with several short, irregularly arranged spines, one or two spines, or lacking spines beneath. Orbicula small, narrower than 0.6 × width of tarsomere 5, with orbicular pecten consisting of a few rather strong, straight setulae, some of them being as long as, or longer than orbicula itself. All tarsal claws bifid.
Head. Broader than long. Vertex strongly convex above level of eye tops, chevron-shaped (Fig. 8A, D, G); juncture of anterior and posterior faces broadly rounded (Fig. 8B, E, H). Frons with numerous long pubescence and setae. Clypeus distinctly narrower than LID (Fig. 8A, G), trapezoid or rectangular, its surface convex, covered with long pubescence and setae. Malar space longer than in female. Scape short (Fig. 8B, F, H-I), not compressed laterally, with numerous setae, those on ventral side longer and denser than elsewhere. Flagellum uni-or biramous (Fig. 8F, K), catenulate (Fig. 8J), or in a few species, basal and apical ends of each flagellomere contiguous all round (Arnold 1932: 67). Mandible short (Fig. 8A) with small tooth subapically on inner margin; anterior face flattened and polished. Occipital suture complete (Fig. 8E), its uppermost portion situated immediately or rather deeply below vertex crest.
Wings. HW cross-vein cu-a originating usually at or distal to separation of vein M+CuA, confluent with vein A, forming long smooth arc.
Legs. Apical margin of fore tibia without short, stout, decurved spine mesally. Fore femur slender (Fig. 8F, I), thinner than mid femur. Fore tarsomeres 2-4 combined as long as or shorter than fore tarsomere 1. Fore, mid and hind orbiculae similar to those of female, or hind orbicula remarkably small, its pecten indistinct. Fore and mid tarsal claws bifid; hind tarsal claw bifid or edentate, rectangularly bent subapically, and both claws parallel to each other or slightly divergent .

Etymology
The species name is derived from its locality, the Arabian Peninsula.
Wings (Fig. 7C). FW marginal cell from wing tip by 0.65 × its own length. SMC2:SMC3 = 1:1.1 on vein M, 1:1.5 on vein Rs. SMC2 narrowed on vein Rs by 0.57 × its length on vein M, receiving cross-vein 1m-cu at basal 0.45. SMC3 narrowed on vein Rs by 0.75 × its length on vein M, receiving cross-vein 2m-cu at basal 0.53. Cross-veins 2rs-m and 3rs-m gently curved outwardly. Cross-vein cu-a slightly apical to point of separation of vein M+CuA, oblique but abruptly recurved near its terminal. Cross-vein 2m-cu strongly curved near its origin. HW cross-vein cu-a originating at point of separation of vein M+CuA.

Male
Unknown.

Distribution
The southeastern Arabian Peninsula (Oman). Haupt, 1930 (The key is not specific to the Asian species but applies to the African species.) Females 1. Lower frons lateral to and ventral to antennal sockets deeply depressed, leaving frontal bridge (Figs 1A,F,J,2E,G,I,3D,arrow) this being continuous to clypeus on same plane; clypeus deeply and broadly depressed only basilaterally; fore femur swollen, thicker than mid femur ( frontal bridge absent (Fig. 8A, D, G), i.e., lowermost frons across its whole width depressed much below level of supra-antennal area and slightly below level of clypeus ... Paraclavelia Haupt, 1930 Based on these key characters, at least Ct. decipiens, which is known only from a male from South Africa, should be transferred to Paraclavelia. Examination of the males of this species has revealed that they bear the characteristics of Paraclavelia, i.e., the frons protruded over the antennal sockets ( Fig. 8F) and lacking the frontal bridge (Fig. 8D), although the flagellum is biramous (Figs 8D-F, K). Moreover, the male of Ct. srilankae sp. nov. has the crenulate flagellum (Fig. 5H). Thus, the males of both Ctenocerus and Paraclavelia may have the flagellum uni-or biramous, or crenulate, as discussed below, invalidating this character as diagnostic for the genera as proposed by Arnold (1932).
At the Natural History Museum, London, AS compared female specimens identified as Ct. klugi (Fig. 1F-I) with one of the female syntype of Euclavelia fasciatus Arnold, 1932 (type species of Euclavelia) (Figs 1J-M, 2A-D) and concluded that Euclavelia should be regarded as a synonym of Ctenocerus, because both species share almost all diagnostic and generic characters of Ctenocerus listed above, although in Ct. klugi, T1 is not petiolate nor parallel-sided anteriorly. Arnold (1932) stated that the characters of Pseudopedinaspis Brauns, 1906 applied to Euclavelia, except that the wings of Pseudopedinaspis were remarkably reduced to small adnate flaps. As such, it may be better to treat Pseudopedinaspis also as a junior synonym for Ctenocerus. This genus, however, bears great modification in the mesosomal structure in accordance with the reduction of the wings. Thus, we leave the problem of the taxonomic position of Pseudopedinaspis to future studies.
According to Arnold (1932), females of Paraclavelia are distinguishable from Ctenocerus by the clypeus being deeply depressed across the whole width of its base; hence, the lack of a frontal bridge. However, his distinction between the genera seems to be wrong in the males from the following situation: he succeeded in associating conspecific females and males in one species of Ctenocerus (Ct. ramosus) and two species of Paraclavelia (Parac. crudelis and Parac. caffer) (Arnold 1932); based on these associations, he considered that the male antenna of Ctenocerus was uni-or biramous, while that of Paraclavelia was crenulate, and used these different conditions to separate the males into the two genera. As stated above, however, in this study we conclude that these antennal conditions occur in both genera, which invalidates the characters Arnold (1932) used for genus-level recognition.
A further problem in the genera placed in Ctenocerini lies in the range and definition of Parapompilus.
The taxonomic history of this genus is complicated. Arnold (1932) (1855) proposed the name Parapompilus as a substitute for Micropteryx (see Arnold 1932: 89;Grünwaldt 1933), because Micropteryx was a pre-occupied name (see also Pate 1946). Pate (1946) proposed the genus Marimba Pate, 1946 as a new name for the African Cryptosalius in the sense of Arnold (1932), not of Turner (1917). According to Schultz (1905), Grünwaldt (1933), Day (pers. com.), and Wahis (pers. com.), Cl. pompiliformis is conspecific with M. brevipennis based only on the holotype male from Barbaria, Algeria. There is, thus, a great possibility that Clavelia becomes a senior synonym for Parapompilus. We, however, leave the name Parapompilus here because we have not examined types of the above taxa.
To revise this group, further studies, including the examination of the types (mostly deposited in the collections of South African Museum, Cape Town, South Africa and the Natural History Museum, London, UK) and molecular analyses, are needed.

Convergent evolution of morphological modifications in trap-door spider-preying Pompilidae
Ctenocerus and Paraclavelia share the following morphological features that seem to be associated with preying on trap-door spiders, as discussed below: (1) the vertex is raised far beyond the postocellar line (Figs 1F, J, 2G, I, 3D, 6A, D, G, J, 7D); (2) the frons is broad (half of MID much broader than eye); (3) the lower frons is deeply depressed lateral to and ventral to the antennal sockets, the depressions accommodating the antennal scape; (4) the clypeus is narrower than LID, lamelliform, usually truncate apically, its surface flat or slightly convex with a basal or basilateral depressions that are continuous to the lower frontal depression; (5) the scape is curved outward and concave on its lateral face (Figs 1L, 2B, 3E, 6B, E, K, P, 7E), which may fit in the frontal depression; (6) the pronotum is elongate (usually longer than the mesoscutum at the midline) (Figs 1K-L, 3A, 7B), flattened above and in a low position, i.e., the dorsum lying only slightly above the level of the collar (Figs 1C, H, 2A, F, H, J, 3H, 6C, F, I, L, 7H); and (7) the apical margin of the fore tibia has a stout curved spine mesally (Figs 2C, 3F, 6M, 7F).
The modifications in head and pronotal structures of Ctenocerus and Paraclavelia are very similar to those of the New World genus Psorthaspis (Pompilinae) (especially to Ctenocerus in having the frontal bridge). They share the above features 1-6 ( Fig. 10A-D), although Psorthaspis does not have a stout curved spine on the fore tibia apicomesally (Fig. 10E). It is thus not surprising that Bradley (1944) placed Psorthaspis in Ctenocerini, Pompilinae [!].
Nothing is known about the biology of Ctenocerinae, except that Paraclavelia caffer was taken from the nest of a trapdoor spider, Stasimopus robertsi Hewitt, 1910 (Ctenizidae) (Arnold 1932). In contrast, the North American species Psorthaspis planata is well known to behave as an ectoparasitoid of the trapdoor spider Bothriocyrtum californicum (O. Pickard-Cambridge, 1874), Ctenizidae (Jenks 1938). Behavioral observations by Jenks (1938) are summarized here. After detection of the burrow, the female wasp pounced on the trapdoor, raised it with her strong claws and supported it on her dorsum (Jenks 1938: 812, above figure;Shimizu et al. 2021: fig. 11) as she slipped through into the burrow to attack the spider. When unable to lift the flap sealed tightly with silk and mud, the wasp tore a hole in it with her powerful mandibles. In a 'furious rough-and-tumble fight', the wasp stung the spider into paralysis, laid an egg on the dorsum of the spider's abdomen posteromedially in a longitudinal position, and left the burrow without closing it.
Based on similarities in morphology with Psorthaspis, it is presumed that species of both Paraclavelia and Ctenocerus are also ectoparasitoids of trapdoor spiders living in subterranean burrows with lids. Arnold (1932: 48) thought similarly. The anteriorly flattened head of these genera, including the lamelliform clypeus and mandible, likely play an important role when the wasp locates and pries open the host-burrow lid as a lever and supports it. The elongate and dorsally flattened pronotum also must be employed in lifting and supporting the lid (Jenks 1938: 812, above figure;Shimizu et al. 2021: fig. 11). The stout curved spine of the fore tibia is suspected to function as a hook when the wasp lifts the trapdoor.