Review of world genera of Ceinae , with the description of two new Palaearctic species of Spalangiopelta Masi ( Hymenoptera , Chalcidoidea , Pteromalidae )

The paper provides the fi rst illustrated key to all described genera of Ceinae, i.e., Bohpa Darling, 1991, Cea Walker, 1837, and Spalangiopelta Masi, 1922. Based on the study of the original material, the genus Diparisca Hedqvist, 1964 stat. nov. is removed from the synonymy with Spalangiopelta and its higher classifi cation is discussed. Spalangiopelta rameli sp. nov. and S. viridis sp. nov. are described from Greece and the Canary Islands, respectively.

Of all species of Ceinae hosts are known only for Cea pulicaris and Spalangiopelta alata and include small Diptera from the families Agromyzidae and Drosophilidae, respectively (Noyes 2015).
In this study, two new West Palaearctic species of Spalangiopelta are described, and S. ferrierei is transferred back to Diparisca Hedqvist, 1964 which is removed from the synonymy with Spalangiopelta.The taxonomic position of Diparisca is also briefly discussed.

Material and methods
As pointed out by Darling (1991b) specimens of Spalangiopelta are usually extremely rare in collections.All specimens of the new species have been collected using Malaise traps (MT) and / or yellow pan traps (YPT), which are usually the only methods that potentially yield series of specimens.Prior to mounting on rectangular cards, specimens were dried using hexamethyldisilazane (HMDS).
The terminology used in this paper follows Darling (1991b) and Gibson (1997), except for the terms mesoscutellum and metascutellum, which are used for scutellum and dorsellum, respectively.A character used to separate some species groups of Spalangiopelta is the presence or absence of a "hyaline break" on the fore wing parastigma (Darling 1991b: 15, fig. 35).
For the new species, a description of the holotype and allotype is given, followed by a discussion of variation found among paratypes.The description format follows Darling (1991b) in order to allow easy comparisons among the new and previously described species.The differential diagnoses for the new species encompass the characters used by Darling (1991b) to separate the world species of Spalangiopelta and are intended to allow the separation of the new species from the previously described ones as well as from each other.Colour of the eyes and ocelli should be used with care, since it seems to vary according to the methods used for the preservation and preparation of specimens.

Male
Unknown.

Male
Unknown.

Females
Body length: 1.6-1.8mm.F1 length 1.60-1.85×width, distally slightly to distinctly wider than A3.Each of the following funicular segments subequal to or longer than F1.Clava length about 5.0-5.5×width.Fore wing length about 2.8-2.9×width, with 4-8 admarginal setae arranged in 1-2 rows.M about 1.6-1.9×P and 3.4-4.0×S. Gaster length about 1.8-2.1×width.The specimens collected near Vironia differ from those collected near Promahonas in being larger and having the F1 subequal in length to any of the following funicular segments; the latter specimens are smaller, and F1 is distinctly shorter than any of the following segments; these differences are regarded as intraspecific variability.

Distribution
Greece.

Hosts
Unknown.

Remarks
This species appears to be intermediate between S. procera and S. dudichi, due to its moderately robust mesosoma.From both species it differs mainly in having distinctly shorter ovipositor sheaths and a different general body colour.Additionally, from S. dudichi it differs in having a longer F1.
Male (Fig. 4E-F) Petiole conical; fore wing linear, front and hind margins parallel from middle of marginal vein to end of postmarginal vein; fore wing with 2-4 admarginal setae below marginal vein, without hyaline break.

Male allotype
Differs from the female holotype as follows.Body length: 1.25 mm.Head colour lighter, violet reflections less strong (Fig. 4E).Propodeum with golden reflections less strong; both fore and mid coxae brown, rest of leg segments brownish yellow (Fig. 4E).Petiole brown.Antenna (Fig. 4F): A1 visible, although strongly transverse; A2 and A3 transverse, subequal in length; each funicular segment with one basal whorl of erect setae that are longer than the segments; F1 length about 3× width, F2 slightly longer than F1, F3 subequal in length to F1, F4 and F5 slightly shorter than F3; clava length about 5× width, longer than F4-F5 combined.Petiole longer, conical, almost smooth.Infumation on fore wing less conspicuous; fore wing linear, front and hind margins parallel from middle of marginal vein to end of postmarginal vein (Fig. 4E); fore wing length about 4× width.SM : M : S : P = 50 : 33 : 7 : 18. Gaster length about 2.3× width, slightly and gradually widening toward apex in dorsal view.

Distribution
Canary Islands.

Hosts
Unknown.

Remarks
This is one of the few species of Spalangiopelta with bright metallic colours.Following Darling's key (1991b), S. viridis sp.nov.most closely resembles S. brachyptera, from which it differs mainly in the well developed wings, and different colour of antennae and body.
The highly modified morphology of Bohpa (Fig. 1A-B) makes its classification difficult.Its taxonomic position is discussed in detail by Desjardins (2007), who concluded that its inclusion in the analysis made the Ceinae paraphyletic.However, taking into consideration that it cannot be included in Diparinae either (it lacks most of the synapomorphies that define Diparinae), until further evidence, its present classification in Ceinae is maintained.
The monotypic genus Diparisca Hedqvist, 1964 stat.nov.(Fig. 5A-D), described from Brazil, is removed from synonymy with Spalangiopelta, based on the examination of the holotype, one paratype and several other specimens of D. ferrierei Hedqvist, 1964 from BMNH.The synonymy was proposed by Desjardins (2007: 82), who stated that the "holotype specimen has been examined and conclusively identified as Spalangiopelta".It is assumed that the above assertion was based on the raised mesopleuron of Diparisca, which is similar to the mesopleuron structure in Spalangiopelta (Fig. 5C).However, the vertex and mesosoma of Diparisca are covered with strong paired setae (Fig. 5A-D), the propodeal spiracles are more advanced, the metascutellum is not smooth and convex (Fig. 5C), the mesoscutellum is much higher, and the cuticle is rather strongly sclerotized.At the same time, it is also clear that Hedqvist's description of Diparisca is partly inadequate: the female antenna (Fig. 5D) has three anelli not two, and the clava consists of three segments and not one (although their limits are very difficult to observe because of the pale colour).Thus, the correct antennal formula in female Diparisca is 11353.Hedqvist (1964: 56)  differs by having antennae 10-joined [not true, see above], median carina on propodeum, notaulices not meeting in the middle of mesoscutum [Fig.5C] and mandibles with 2 teeth".I compared Diparisca with Dipara Walker, 1833 and agreed that is a distinct genus, based mainly on the structure of the antenna, the propodeum and the gastral tergites (especially GT1, which is only slightly expanded in Diparisca, as compared with Dipara).
In conclusion, Diparisca appears to be distinct from both Spalangiopelta and Dipara, and is considered here a valid genus.The characteristic shape of the mesosoma in Spalangiopelta (Fig. 3E) and Diparisca (Fig. 5C) is probably a convergence due to the same adaptation to subterranean environments, as suggested by other morphological characteristics, such as the reduction of wings, which is encountered in both genera.
However, the taxonomic placement of Diparisca is problematic.Although the presence of strong bristles is not unique to Diparinae, their configuration on the vertex of diparines apparently is (Desjardins 2007)."In Diparinae, the vertex bristles are arranged along the occipital margin, ocellar triangle, and along the dorso-frontal margins of the eyes" (Desjardins 2007: 20).This is exactly the case in Diparisca (Fig. 5D).The other characters, i.e., the position of propodeal spiracles, the shape of metascutellum, the structure of the cuticle, also point out that the genus is closer to Diparinae than to Ceinae.At the same time, other characters found in Diparisca apparently are closer to Ceinae than to Diparinae, in particular the shape of GT1, which is greatly enlarged in Diparinae (except Pyramidophoriella Hedqvist, 1969), and only moderately so in Diparisca (Fig. 5A-B), and the absence of transverse striation on hind coxae (there are, however, a few diparines that lack striations on hind coxae).The presence or absence of a cercal brush which, according to Desjardins (2007), is always present in Diparinae, could not be determined in the examined specimens of Diparisca.In conclusion, I prefer to keep Diparisca in Diparinae until further evidence.Darling (1995) stated that the subfamily Ceinae may be a sister group of Diparinae, based on the presence of papilliform sensilla on the antenna.Since then, papilliform sensilla have been discovered also in Pireninae (Mitroiu 2010;Fig. 6), suggesting a common ancestry of Ceinae + Diparinae + Pireninae.Ceinae and Pireninae also share several other morphological features such as the absence of or generally faint, metallic coloration (except some species, including the newly described ones and especially S. viridis sp.nov.), low insertion of antennae, and mostly weakly sclerotized, almost smooth cuticule.
In Desjardins' analysis, (Cea + Spalangiopelta) was placed either as a sister-group of (Bohpa + Diparinae) when Bohpa and the bristle positional characters were included in the analysis, or as a sister-group of Coelocybinae, when Bohpa was excluded (Desjardins 2007).
In the most recent phylogenetic analysis of Chalcidoidea (Heraty et al. 2013), Spalangiopelta, the only analyzed genus of the subfamily, appeared to be more related to Tetracampidae and Eulophidae than to other Pteromalidae, but the support for this hypothesis is rather weak.
In conclusion, both the taxonomic placement of Ceinae in relation to other pteromalid subfamilies (especially Diparinae), and their monophyly remain uncertain.Finding additional characters (including molecular ones) could perhaps elucidate these aspects.