The skeletomuscular system of male terminalia in Evaniomorpha (Hymenoptera) is described and the functional morphology of male genitalia is discussed. Confocal laser scanning microscopy is the primary method used for illustrating anatomical phenotypes, and a domain-specific anatomy ontology is employed to more explicitly describe anatomical structures. A comprehensive data set of ceraphronoid male genitalia is analyzed, yielding the first phylogeny of the superfamily. One hundred and one taxa, including three outgroups, are scored for 48 characters. Ceraphronoidea are recovered as sister to the remaining Evaniomorpha in the implied weighting analyses. Numerous character states suggest that Ceraphronoidea is a relatively basal apocritan lineage. Ceraphronoidea, Ceraphronidae, and Megaspilinae are each retrieved as monophyletic in all analyses. Megaspilidae is not recovered as monophyletic. Lagynodinae is monophyletic in the implied weighting analyses with strong support and is a polytomy in the equal weighting analysis. Lagynodinae shares numerous plesiomorphies with both Megaspilinae and Ceraphronidae. Relationships among genera are weakly corroborated. Masner is sister of Ceraphronidae. Trassedia is nested within Ceraphronidae based on the present analysis. Because of this and numerous features shared between it and Ceraphron we transfer Trassedia from Megaspilidae to Ceraphronidae. Dendrocerus forms a single monophyletic clade, with modest support, together with some Conostigmus species. This result challenges the utility of such traditional diagnostic characters as ocellar arrangement and shape of the male flagellomeres. Aphanogmus is monophyletic in the implied weighting, but remains a polytomy with Ceraphron in the equal weighting analysis. Gnathoceraphron is always nested within a well-supported Aphanogmus clade. Cyoceraphron and Elysoceraphron are nested within Ceraphron and Aphanogmus, respectively. The male genitalia prove to be a substantial source of phylogenetically relevant information. Our results indicate that a reclassification of Ceraphronoidea both at the family and generic level is necessary but that more data are required.
Hymenoptera, which includes sawflies, wasps, bees, and ants, is one of the four most species-rich insect orders, with more than 145 000 known species (Huber 2009) and perhaps more than 1 million species remaining to be described (Sharkey 2007). The evolutionary history of Hymenoptera has yet to be resolved, though recent efforts have edged closer towards a robust estimate (Vilhelmsen et al. 2010; Heraty et al. 2011). The emerging evolutionary topology reveals a highly supported basal grade of herbivorous hymenopterans (sawflies and woodwasps, known as “Symphyta”), leading to an extraordinarily diverse and notorious rapid radiation of Apocrita (Whitfield & Kjer 2008). Most apocritan superfamilies are robustly monophyletic, but the relationships between them are weakly resolved. Perhaps the most uncertain is the position of the small hyperparasitoid lineage, Ceraphronoidea (Sharanowski et al. 2010, Sharkey et al. 2012), a putatively basal apocritan (Vilhelmsen et al. 2010).
Ceraphronoidea was recently cataloged by Johnson & Musetti (2004) and currently includes four families: Ceraphronidae, Megaspilidae, and the fossil groups Stigmaphronidae and Radiophronidae. These four families include 27 valid genera (plus 28 generic concepts now considered junior synonyms) and 613 valid species (plus 235 species-level concepts now considered junior synonyms). Ceraphronoidea is one of the smallest of the major apocritan clades, yet they are the fourth most commonly collected hymenopterans (Martínez de Murgía et al. 2001; Schmitt 2004). Most ceraphronoids are parasitoids of entomophagous insects that develop in weakly concealed environments, inside cocoons or puparia or hosts that are prepupae (Haviland 1920; Withycombe 1924; Kamal 1939), a lifestyle that is probably the ground plan biology for the clade. Many of their hosts are primary parasitoids or predators of economically important insects (e.g., predators of the coffee berry borer (Evans et al. 2005), spider mite predators (Diptera: Cecidomyiidae) (Oatman 1985), and parasitoids of lepidopteran pests on oil palm (Polaszek & Dessart 1996). Despite their abundance and economic importance, only one systematist, Paul Dessart (active 1962–2001, deceased 2001), has worked on the group in modern times, and his core revisionary hypotheses have never been tested phylogenetically.
It is widely accepted that Apocrita is monophyletic, with Orussoidea as its sister lineage (Rasnitsyn 1988, 2002; Ronquist et al. 1999; Vilhelmsen 2003; Schulmeister 2003a; Sharkey 2007; Heraty et al. 2011). However, our knowledge of the phylogeny of the suborder remains incomplete, and the relationships between putatively basal apocritans (Ceraphronoidea, Evanioidea, Trigonalidae and Megalyroidea, together referred to as Evaniomorpha) remain elusive (Heraty et al. 2011). The importance of Evaniomorpha for resolving the phylogeny of Hymenoptera is broadly recognized, and exemplars are usually involved in analyses attempting to resolve higher-level phylogenies (Shcherbakov 1981; Gibson 1985, 1999; Heraty et al. 1994; Vilhelmsen 1996; 2000a,b, 2003; Schulmeister 2003b). Unfortunately, Ceraphronoidea has been excluded from most of these analyses and in some morphology-based analyses that did include ceraphronoid exemplars, character states were misinterpreted, probably due to their minute size. Numerous observations, however, indicate, that unlike other apocritans some Ceraphronoidea share putatively plesiomorphic character states with “Symphyta”, including: (1) presence of a propleural arm-postoccipital muscle, shared with some Tenthredinoidea (Vilhelmsen et al. 2010), (2) presence of a metanoto-metacoxal muscle, shared with most non-apocritan Hymenoptera (Vilhelmsen et al. 2010), (3) presence of a posterior apical spur on the protibia, shared with most non-apocritan Hymenoptera (Rasnitsyn 1988), (4) presence of a median mesoscutal sulcus that corresponds to an internal ridge, shared with most non-apocritan Hymenoptera and with basal Apocritan lineages (Megalyroidea, Stephanoidea) (Gibson 1985), (5) presence of a mesonotal and mesofurcal depressor of the mesotrochanter with this muscle inserting distinctly ventrally of the site of insertion of the mesonotal depressor on the mesotrochanter (pers. obs.). This latter state of the mesotrochanteral extracoxal depressor complex was considered by Gibson (1999) to represent the hypothetical ancestor of Apocrita, though he did not observe the ceraphronoid furcal muscle in his studies.
The collective phenome of ceraphronoid wasps is not only relatively monotonous (i.e., lacking distinct apomorphies) compared to other microhymenoptera (e.g., Chalcidoidea, Platygastroidea), but the few possibly suitable morphological variables (e.g., body shape, sculpture) are often affected by allometry (Fig. 1). Male genitalia, however, provide a source of discrete and size-independent characters. The utility of male genitalia in species delimitation was recognized relatively early and from the mid 20th century became the key element in species diagnoses (majority of Paul Dessart’s publications since 1963; Teodorescu 1967; Takada 1973).
Male genitalia characters are certainly informative for higher level classification of basal Hymenoptera (Schulmeister 2003b) and have been successfully applied in generic level phylogenetic studies within Apocrita (Rozen 1951; Andena et al. 2007; Owen et al. 2007; Brajković et al. 2010; Žikić et al. 2011). While ceraphronoid male genitalia serve as an important source of species diagnostic characters their phylogenetic signal has never been tested. Here we provide the first phylogenetic analysis of Ceraphronoidea, based exclusively on 48 morphological characters related to the skeletomuscular system of the male terminalia.
Materials and Methods
Depositories and locality data of specimens examined in the present study are documented in Appendix C.
Resulting anatomical phenotype descriptions were based on observations made during dissections under stereo (Olympus SZX16 with SDFPL APO 26PF objective, 2306) and compound (Olympus BX51 with LMPLFLN506 objective, 5006) microscopes. Wet specimens in a glycerin droplet and critical point dried specimens were both dissected on Blu-Tack (Bostik Findley, Wauwatosa, WI, USA) using Super Personna razor blades (American Safety Razor, Cedar Knolls, NJ, USA) and #2 insect pins. For the better visualization of skeletal structures some specimens were macerated in 20% KOH for one week (Figs 10, 11, 25, 57).
Confocal laser scanning microscopy (CLSM), scanning electron microscopy (SEM) and bright field digital imaging were used to visualize anatomical structures. Bright field images were made using an Olympus CX41 compound microscope and DP71 digital camera. SEM micrographs were made using a Hitachi S-3200 Scanning Electron Microscope (wd = 23.5, av = 5kV). Specimens were critical point dried and coated with palladium prior to examination. CLSM images were made on glycerin-stored specimens between 1.5 mm thick, 24 × 50 mm cover glasses with a Leica LSM 710 laser scanning confocal microscope using 488 nm laser for excitation of the sample. We collected the autofluorescence of insect anatomical structures between 500 and 700 nm with two channels (500–580; pseudocolor green and 580–700 pseudocolor red) using 106 and 206 Plan Achromat objectives. Volume rendered images and media files were generated by Imaris Bitplane (Bitplane, Zürich, Switzerland) software. Media files, SEM micrographs and bright field images are available from figshare.com (Appendix A).
We applied a dissection-based morphological approach for the description of complex anatomical systems because section based techniques (micro-CT or histological sections) would have limited the breadth of our taxon sampling. The appropriate visualization of very small anatomical structures (the male genitalia are 100–400 μm in most Ceraphronoidea) and the textualization of the highly complex and diverse skeletal systems were the two most challenging components of our study. Imaging dissected specimens with bright field digital imaging techniques results in the loss of the ability to see anatomical structures in three dimensions and a very low tissue contrast (i.e., the ability to differentiate muscles and skeletal structures). By applying CLSM we were able to eliminate these complications. The 3D reconstructions of CLSM data have very high tissue specific contrast due to differences in the wavelength of emitted fluorescence light by soft and skeletal structures (Deans et al. 2012) and CLSM volume rendered media files let us to share our observations in 3D with a reader.
Cladistic analyses were carried out with TNT 1.1 (Goloboff et al. 2008). Space for 1000 trees was reserved in memory. Traditional searches in equal-weighting analyses and implied-weighting analyses (Goloboff 1993) with the concavity constant k set in turn to 3, 5, 10 and 25 were run to test the stability of clades under different weighting conditions. Analyses were run with collapsing rules set to maximum length = 0. One thousand replications with 1000 trees saved per replication were run, followed by a round of branch breaking on the optimal trees. Jackknife support values were calculated with 10000 pseudoreplications.
Morphological terminology follows Schulmeister (2001, 2003b), Snodgrass (1941) and Bohart and Menke (1976). The integumentary part of the external male genitalia, similar to other areas of the insect cuticle, is composed of sclerites (more rigid, usually well tanned areas with thick exocuticle) and conjunctivae (less flexible, usually less tanned areas with thin exocuticle) whose number and spatial distribution change during the course of evolution. The statements “sclerites fused” and “sclerite subdivided” are widely used to refer to processes during which the usually narrow and elongate conjunctivae separating sclerites disappear or appear during the course of evolution. Unfortunately this phrasing presents two major problems:
The phenotypic concepts “fused” and “separated” are relational, based on comparisons with the phenotypes of other taxa and therefore can be decoded only by comparisons with the phenotypes of other taxa. This predicament limits the accessibility of character descriptions to the non-expert community.
If the above mentioned phenotypic concepts are applied in descriptions supported by anatomical ontologies, i.e., semantic phenotype annotations (Deans et al. 2012; Mullins et al. 2012), where concepts are explicitly defined by structural relationships (e.g., by their bordering sclerites), “fused” and “separated” actually refer to anatomical structures that do not exist.
For example, the descriptors that “the parossiculus is delimited medially from the other parossiculus and laterally from the gonostipes by conjunctivae” are necessary components of its differentia (what distinguishes this sclerite from other sclerites). In the statement “the parossiculi are fused laterally with the gonostipites in Ceraphronidae and fused medially with each other and laterally with the gonostipes in most Dendrocerus species” qualities are provided for anatomical structures that lack the necessary conditions and therefore do not exist in these wasps. We therefore chose a more objective approach to describe what was observed rather than what is hypothesized—“medioventral conjunctiva of the male genitalia (separating parossiculi) present or absent”. We have also mapped all anatomical terms used to anatomical concepts in the Hymenoptera Anatomy Ontology (Appendix B).
A cercal plate is present in Xyela, Orthogonalys, Pristaulacus, Megalyroidea and Megischus. The cercus is located on a posterior lobe of T9 in Gasteruptidae (crc: Fig. 18) and Ceraphronoidea (Fig. 2A) and no cercus is present in Evaniidae.
This character is modified after character 353 of Schulmeister (2003b).
T10 is surrounded laterally by the posteriorly extended lateral part of the T9, which bears the cerci and is connected to it via a pair of rodlike T9-T10 muscles (Figs 2C, D). T10 is present in Xyela among the outgroup taxa. The relative position of T10 in Ceraphronoidea is similar to that of Orussus (T10: fig. 15D in Schulmeister 2003b; XT: plate IV in Snodgrass 1941). Unfortunately we were not able to study the musculature of the male terminalia of Orussus and hence can only suggest that T10 of Ceraphronoidea is structurally equivalent to the abdominal tergum 10 of Orussus.
This character is modified after character 352 of Schulmeister (2003b).
The anterior margin of S9 is concave in Ceraphronidae and Trassedia and convex in all other taxa examined. The shape of the proximal margin of S9 corresponds to the site of origin of the mediolateral S9-cupulal muscle. The anterior margin is convex if the muscle arises medially from the sternite and concave if it arises from the anterolateral edge. In most taxa, the site of origin of the muscle corresponds with a distinct spiculum (Figs 2B and 3A), but the spiculum is absent from Masner (Fig. 4A).
This character is modified after character 347 from Schulmeister (2003b).
In Megaspilidae the cercus is uniformly setose whereas in Ceraphronidae a row of shorter setae delimits a posterior area from which 1 or 2 longer setae arise.
A proximal lobe emptying into the lumen of the vas deferens is present in most Ceraphronoidea and numerous outgroup taxa. Based on its shape and location, the lobe might be an accessory gland. In Megalyra, Pseudofoenus, Megaspilus and Conostigmus sp. 23 the lobe is absent. The distal part of the vas deferens, just proximally of the vesicula seminalis, is enlarged relative to the more distal and more proximal areas in Megaspilus and Conostigmus sp. 23 (Fig. 5B). This enlarged area of the duct might also have an accessory gland function. Male accessory gland structures are great sources of species diagnostic and phylogenetic characters in some Hymenoptera (Schulmeister 2003b; Ferreira et al. 2004; Mikheyev 2004). The shape of the vesicula seminalis and the putative accessory glands is variable in different ceraphronoid taxa, and might be useful for species delimitation or even for phylogenetic analyses. Unfortunately, however, the internal male genitalia is often damaged during the dehydration and dissection of the metasoma, and characters other than the presence of a proximal lobe, cannot be accurately scored in most taxa.
The cupula is a continuous ring in most taxa examined but is dorsally incomplete in Pseudofoenus and Gasteruption. In Pseudofoenus, the cupula ventrally is also interrupted by a conjunctiva and thus is composed of two crescent-shaped sclerites located ventrolaterally on the male genitalia. The presence of an incomplete cupula might be related to the opening of the male genitalia (see character 48), because unlike in other taxa, where the cupula is relatively rigid, in Gasteruptiidae it is bent medially when the male genitalia is opened.
This character is modified after character 299 of Schulmeister (2003b).
Shape of ventral part of cupula: 0, Cupula ventromedially extended more proximally than dorsomedially (c: Figs 3C,D, 4D, 7B and 8B); 1, Cupula ventromedially not extended more proximally than dorsomedially (c: Fig. 6A,C)
The ventral part of the cupula is strongly extended proximomedially in Ceraphron, Cyoceraphron, Trassedia and Masner, which results in the most proximal point of the ventral part of the cupula being distinctly more proximal than the most proximal point of the dorsal part of the cupula.
The proximodorsal notch separating the site of origins of the dorsomedial cupulo-gonostyle/volsella complex muscles is present in numerous Conostigmus species and in Trichosteresis. Although the proximodorsal margin is concave medially in numerous other taxa (e.g., Megalyra and Orthogonalys), the medial concavity never separates the sites of origin of the dorsomedial cupulo-gonostyle/volsella complex muscles, which extend medially along the proximal margin of the concave area.
The dorsal submedian impression is present in most Ceraphronidae. The impression might correspond to the presence of the dorsolateral cupulo-gonostyle/volsella complex muscle.
Gonostyle/volsella complex dorsal continuity: 0, continuous, dorsomedian conjunctiva of gonostyle/volsella complex incomplete, not reaching proximal and distal margins of the complex (gvc: Fig. 7A); 1, discontinuous, dorsomedian conjunctiva of gonostyle/volsella complex complete, extending between proximal and distal margins of the complex (gvc: Fig. 9A)
The gonostyle is discontinuous along the dorsomedian conjunctiva in most outgroup taxa and is continuous proximodorsally in Ceraphronoidea except in Dendrocerus wollastoni where it is continuous distodorsally (Fig. 16A). The character is not applicable in Evanioidea where the aedeagus is continuous proximodorsally with the gonostyle (Fig. 5C).
This character is modified after character 310 from Schulmeister (2003b).
The proximoventral margin of the gonostyle/volsella complex is pointed proximally in Lagynodinae and Ceraphronidae and in numerous outgroup taxa, whereas in other ceraphronoid taxa and in Megischus and Pseudofoenus it is straight, not pointed proximally.
This character is modified after character 303 from Schulmeister (2003b).
The apex gonostipitis is present in outgroup taxa and in Lagynodinae. The apex gonostipitis is present only in those taxa where the parossiculus is present (character 13) and the gonostyle/volsella complex is continuous medially (character 17).
The parossiculus is absent (fused with the gonostipes) from Megischus, Pseudofoenus and numerous Ceraphronoidea. The site of “fusion” between the parossiculus and the gonostipes might be marked by the submedian notch on the distoventral margin of gonostyle/volsella complex (see character 21).
This character is modified after character 328 of Schulmeister (2003b).
This character is not applicable if the parossiculus is evenly covered with dense setae (Megalyroidea). The number of setae and the presence of corresponding distal projections of the parossiculus vary within in Ceraphronoidea. Only one parossiculal seta is present in Ceraphronidae, whereas in Megaspilidae the number of setae is usually 2.
Submedian conjunctiva on the distoventral margin of gonostyle/volsella complex: 0, absent (Figs 4D, 7B and 8B); 1, present (sdv: Figs: 6C, 13A and 5D)
The submedian conjunctiva on the distoventral margin of the gonostyle/volsella complex separates a median area from the ventral part of the complex, which serves as the site of origin of the medial gonostyle/volsella complex-volsella muscle. The median area delimited laterally by the conjunctiva might be homologous with the parossiculus based of the site of origin of the muscle and the conjunctiva might mark the “site of fusion” of the parossiculus with the gonostipes. The conjunctiva is present in Pseudofoenus, Megischus and in most Megaspilidae. There is a submedian notch that laterally separates the area that contains the parossiculal setae and the gonostyle/volsella complex-gonossiculus articulation in Ceraphronidae (pn: Fig. 4D). There is no conjunctiva in the notch and it does not separate the site of origin of the medial gonostyle/volsella complex-volsella muscle, and thus might not mark the “site of fusion” of the parossiculus and volsella, but rather evolved for some other reason. In Pseudofoenus, a second, proximoventral conjunctiva also delimits the median area from the proximal region of the gonostyle/volsella complex (Fig. 5D).
Medioventral conjunctiva of the gonostyle/volsella complex (fusion of parossiculi): 0, absent (parossiculi fused; Figs 6C and 13A): 1, present (parossiculi independent or fused proximally; mvc: Figs 8D and 9D)
The medioventral conjunctiva of the gonostyle/volsella complex is present in outgroup taxa, in Ceraphronidae, and in numerous Megaspilidae. The conjunctiva is absent from most Dendrocerus and three Conostigmus species.
This character is modified after character 332 of Schulmeister (2003b).
Gonostyle/volsella complex continuity proximoventrally: 0, continuous proximoventrally, medioventral conjunctiva extending to proximal margin of gonostyle/volsella complex (Figs 6C, 13A and 16B,C); 1, discontinuous, medioventral conjunctiva not extending to proximal margin of gonostyle/volsella complex (Figs 4D, 6B, 8D, 10B, 11B,C and 12B)
The gonostyle/volsella complex is continuous proximoventrally in all Megaspilinae and discontinuous in other taxa examined. In those megaspilid taxa where the parossiculus is present, the complex is continuous along a narrow area ventrally of the complex. This area might be homologous with the “fused” apices gonostipitis based on the site of origin of the distoventral gonostyle/volsella complex-penisvalva muscle.
The medioventral area of the gonostyle/volsella complex is inflected vertically (oriented distodorsally) along the medioventral conjunctiva in some Aphanogmus species and in Gnathoceraphron. The vertical orientation of the medioventral area of the gonoforceps is most probably related to the movement of the ventrolaterally oriented gonossiculus (see character 21).
This character is modified after character 329 of Schulmeister (2003b).
The volsella is generally known to have a clasper function in Hymenoptera (Schulmeister 2003b; Snodgrass 1941; Allard et al. 2002). The volsellal “forceps” is composed of the cuspis, the distomedial projection of the parossiculus, and the usually elongated gonossiculus. In Ceraphronoidea (and numerous other prototrupomorph taxa; personal observation), the cuspis is absent and thus the volsella has seemingly lost its clasper function. A distolateral projection that corresponds with the parossiculal setae is present in some Ceraphronoidea, but it seemingly never acts together with the gonossiculus as forceps. This character is not applicable in Megalyroidea, where the gonossiculus is absent. A lateral projection located distolaterally on the volsella in Dinapsis (Figs 12A,B,D) might be homologous with the cuspis; however, because the gonossiculus is absent, this projection certainly does not have a “forceps” function.
Distivolsellar apodeme: 0, absent; 1, present (dva: Fig. 14D)
The distivolsellar apodeme serves as the site of insertion of the lateral and median gonostyle/volsella complex-volsellal muscles. The apodeme is present in those taxa where the muscle inserts on the parossiculus, e.g., in Orthogonalys, Xyela and Megischus.
Position of gonossiculus vs. distal region of aedeagus: 0, gonossiculus parallel with and adjacent to distal aedeagus (gs: Figs 3A,B and 6A,C); 1, gonossiculus oriented differently and not adjacent to distal aedeagus (gs: Figs 13D and 14B)
Because the cuspis is absent from Ceraphronoidea (see character 26), the volsella has lost the forceps function present in most basal Hymenoptera and the majority of basal Apocritan taxa (Snodgrass 1941Schulmeister 2003b; see also character 19). The ceraphronoid gonossiculus is usually flattened, and lays parallel with and adjacent to the distal aedeagus in the majority of ceraphronoid taxa. In these taxa, the gonossiculo-gonostyle articulation is located more ventrally than the site of origin of the gonostyle/volsella complex-volsella muscles, which inserts on the dorsalmost digital spine (ddgtt: Fig. 15C). The sclerite presumably pivots dorsomedially when the muscles are contracted. Because the gonossiculus is parallel and adjacent to the distal end of the aedeagus, it might press the latter to the distodorsal margin of the gonostyle/volsella complex. In numerous taxa, the apical part of the gonossiculus hooks on the distodorsal margin of the gonostyle/volsella complex (gss: Fig. 6A) or on the median margin of the harpe (Fig. 8A), which possibly stabilizes the dorsally slanted position of the aedeagus. Notches accommodating the gonossiculus on these structures are quite often present in different ceraphronoid taxa (ddn: Fig. 11D, pnh: Fig. 43). In numerous taxa, the gonossiculi surround the distal end of the aedeagus in the dorsally slanted position and might also be involved in an opening/closure mechanism of the phallotrema (gss: Figs 3B, 4D, 6A,C, 7A and 8A). The penisvalvo-gonossiculal muscle inserts ventromedially on the gonossiculus, and hence most probably is antagonistic of the gonostyle/volsella complex-volsella muscles and pivots the sclerite ventrolaterally.
In those Ceraphronoidea species where the medioventral area of the gonoforceps is vertical (numerous Aphanogmus species and Gnathoceraphron; see character 18), the gonossiculus is sickle shaped, does not lay parallel and is not adjacent to the distal aedeagus. The gonostyle/volsella complex-gonossiculus articulation is more dorsal than the site of origin of the gonostyle/volsella complex-volsella muscles and the sclerite therefore most likely pivots lateroventrally when the muscles are contracted. Beside its importance for the attachment of the male specimen to the female during copulation, the volsella might be involved also in the infliction of copulatory wounds in Hymenoptera (Kamimura 2008). The above described, sickle-shaped (only one digital spine present) and ventrally oriented gonossiculus might be involved in damaging the female copulatory tract and preventing her from remating with subsequent males. Although numerous proctotrupomorph taxa share the presence of digital spines (dgt, ddgt: 4D, 7A, 15C) on the flattened gonossiculus with Ceraphronoidea, the teeth are structurally different. In Platygastroidea, Chalcidoidea, Cynipoidea and Proctotrupoidea the digital spines are spurs that usually are more heavily sclerotised than the rest of the gonossiculus, whereas in Ceraphronoidea they are spines that are similar in sclerotization as the proximal gonossiculal areas.
The proximal part of the penisvalva is curved ventrally in most Ceraphronoidea, but curved dorsally in Ceraphron, Cyoceraphron, Masner and Trassedia, as well as in most outgroup taxa. The proximal end of the penisvalva serves as the site of origin of the penisvalvo-gonossiculal muscle in Ceraphronoidea, and consequently the direction of the curvature might be related to the movement of the gonossiculus.
The apodeme is present in Megalyroidea, Evanioidea, Megischus and Orthogonalys, and is absent from Xyela and Ceraphronoidea.
Continuity of dorsomedian apodeme of the aedeagus and gonostyle: 0, continuous (dae in Fig. 5C); 1, not continuous
The dorsomedian apodeme of the aedeagus is continuous with the proximodorsal margin part of the gonostyle in Evanioidea and in Orthogonalys while independent in Megischus and Megalyroidea.
The dorsal apodeme of the penisvalva is present in some Megaspilidae and numerous Aphanogmus species. The distodorsal gonostyle/volsella complex-penisvalval muscle inserts on the apex of the apodeme. Because the site of origin of the penisvalvo-gonossiculal muscle extends dorsally along the apodeme, the presence of the apodeme might be correlated to the moving mechanism of the gonossiculus.
The ventromedian apodeme of the aedeagus is present in some Aphanogmus species and in Gasteruption. The cupulo-gonostyle/volsella complex muscles might be involved in male genitalia opening (see character 48). However, these muscles are absent from Gasteruptiidae and opening of the male genitalia might be facilitated by the ventrally pivoting aedeagus (Fig. 18) during the contraction of the ventral gonostyle/volsella complex-penisvalval muscles. The ventromedian apodeme is seemingly involved in this procedure in Gasteruption. The function of the apodeme is questionable in Aphanogmus, but certainly has evolved for a different reason because Ceraphronoidea are not able to open/close the male genitalia. Another smaller rod like apodeme is present in numerous Ceraphronoidea, which arises distoventrally from the aedeagus and is continuous with the ventromedian conjunctiva of the gonostyle/volsella complex (vra: Fig. 14C). This rod like apodeme is variously sclerotized and seemingly often broken off from the aedeagus during dissection; therefore, the presence of this apodeme was not included in the present analysis. This rod-like apodeme might be homologous with the median sclerotized style present in Cephidae and Siricidae (Schulmeister 2003b).
The phallotrema is located ventrally in all outgroup taxa and distodorsally in Ceraphronoidea. The distal end of the aedeagus surrounds the phallotrema in Megaspilidae and in numerous Aphanogmus species, whereas the phallotrema is more distal than the dorsal end of the sclerotized part of the aedeagus in Ceraphron, Masner and Trassedia. The gonossiculi therefore are pressed to the aedeagus in the former taxa and to the phallotrema in the latter group (see character 20).
This character is modified after character 324 of Schulmeister (2003b).
A harpe is present in a majority of Ceraphronoidea and in Trigonaloidea among the apocritan taxa examined. Schulmeister (2003b) considered the absence of a harpe as one of the synapomorphies of Vespina and concluded that the distally delimited sclerite on the gonostyle of some taxa, including Trigonaloidea, is most probably not homologous with the harpe of basal Hymenoptera due to the lack of muscle attachment sites. The distal sclerite of the gonostyle/volsella complex is, however, connected to the complex via muscles in both Ceraphronoidea and Trigonaloidea (Fig. 9A,B). The harpe is absent from Trichosteresis glabra (Fig. 9C,D), Dendrocerus wollastoni (Fig. 16A) and Aetholagynodes stupendus (Fig. 10A,B) among the examined ceraphronoid taxa.
This character is modified after character 287 from Schulmeister (2003b).
The proximomedial apodeme is present in Megaspilidae and Trassedia and absent from Ceraphronidae (see character 47).
The distolateral projection is present on the dorsoventrally flattened harpe of Aphanogmus sp. 10–12 and in A. dyctinna.
T9-T10 muscle: 0, present (T9-T10: Fig. 2C,D); 1, absent
The T9-T10 muscle connects abdominal tergum 9 with abdominal tergum 10 in Ceraphronoidea and in Xyela.
The mediolateral S9-cupulal muscle arises medially from the anterior margin of S9 in all outgroup taxa, Megaspilidae and Masner, and from the anterolateral corner of S9 in other Ceraphronidae except Masner. Protraction and retraction of the male genitalia occurs very often in Hymenoptera and almost certainly is the reason for the protection of the anatomical system. Based on their site of attachment, the mediolateral S9-cupulal muscles protract, whereas the lateral S9-cupulal muscles retract the male genitalia. In Megaspilidae, similarly to numerous other Hymenoptera (Snodgrass 1941; Schulmeister 2001, 2003b), the mediolateral S9-cupulal muscles arise medially from the anterior margin of S9, whereas the lateral S9-cupulal muscles arise laterally from the posterior region of S9. The posterior margin of S9 is usually strongly projected anteriorly into a median projection, the spiculum, increasing the distance between the sites of origin of the S9-cupulal muscles and allowing a longer anteroposterior movement of the male genitalia. The anterior margin of S9 is concave in all Ceraphronidae except Masner and the mediolateral muscle arises from the anterolateral edge of the sternum, which projects slightly anteriorly. Often in this group, the proximoventral margin of the cupula is projected proximomedially. Projections that correspond to sites of origins of muscles on two sclerites that move antagonistically relative to each other because of action of the muscles, increase the distance between muscle attachment points and, therefore, might play a crucial role in complete movement. The key element to protract and retract the male genitalia is the anteromedially projected proximal margin in Megaspilidae (and in numerous other Hymenoptera), and the anteromedially projected cupula in numerous Ceraphronidae. Although the two S9-cupulal muscles govern the movement of the male genitalia relative to the metasoma, it is most probable that other mechanisms are also involved in at least the retraction of the male genitalia in Ceraphronoidea. In numerous museum specimens the male genitalia is distinctly more retracted than what might be possible if only the retractor muscle is involved in the retraction (Fig. 3C,D). Another possibility is that change of the hydrostatic pressure in the metasoma affects male genitalia retracting forces. Austin (1983) hypothesized that the change of hydrostatic pressure plays a crucial role in movement of the ovipositor system in and out of the metasoma in Scelionidae (Hymenoptera: Platygastroidea). The Ceraphronoidea metasoma has the main requirement for the hydrostatic pressure hypothesis of Austin because, similarly to Platygastroidea, it lacks spiracles (Vera & Kumar 1972, pers. obs.).
In Evanioidea, Orthogonalys and Megalyra the muscle inserts submedially on the ventral part of the cupula, whereas in Dinapsis, Xyela, Megischus and Ceraphronoidea it inserts medially on the proximal margin of the cupula. The cupulal site of attachment of the muscle sometimes corresponds to the gonocondyle.
This character is modified after character 267 of Schulmeister (2003b).
Two cupulo-gonostyle/volsella complex muscles exist in Megaspilidae, whereas three exist in Ceraphronidae. The more dorsally located muscle is structurally equivalent with the dorsomedian cupulo-gonostyle/volsella complex muscle of non-apocritan Hymenoptera (muscle g of Schulmeister 2001) based on the dorsomedian position of the insertion site of the muscle on the gonostyle/volsella complex. The more laterally located muscle of Megaspilidae is structurally equivalent with the ventrolateral cupulo-gonostyle/volsella complex muscle of basal Hymenoptera (muscle e Schulmeister 2001) based on its site of insertion medially on the proximoventral margin of the gonostyle/volsella complex (on apex gonostipitis if the structure is present). In Ceraphronidae, three muscles originate from the cupula and insert on the proximal margin of the gonostyle/volsella complex. The dorsal muscle is equivalent with the dorsomedial cupulo-gonostyle/volsella complex, whereas the lateroventral muscle is equivalent with the ventrolateral cupulo-gonostyle/volsella complex muscle of basal Hymenoptera. The third muscle arises just laterally of the dorsomedian cupulo-gonostyle/volsella complex muscle, and therefore might be homologous with the dorsolateral cupulo-gonostyle/volsella complex muscle of basal Hymenoptera (muscle f in Schulmeister 2001). This muscle is absent from Pseudofoenus and Gasteruption where the cupula is discontinuous dorsally.
This character is modified after characters 273 of Schulmeister (2003b).
This muscle is present in Megischus and Xyela and absent from other taxa examined.
This character is modified after character 271 of Schulmeister (2003b).
The lateral gonostyle-penisvalval muscle is absent from Ceraphronoidea and present in most outgroup taxa.
This character is modified after character 279 of Schulmeister (2003b).
Site of origin of gonostyle/volsella complex-volsellal muscles: 0, not extending distodorsally on gonostyle/volsella complex (Figs 10A, 11A and 15C, gs-vl: 8C, 15C and 17B); 1, extending distodorsally on gonostyle/volsella complex (gs-vl: Figs 6A, 7C, 8A and 16A)
This muscle arises proximally from the gonostyle/volsella complex in most Conostigmus, Megaspilus and Lagynodinae, and distodorsally on the complex in Ceraphronidae, Trassedia and Dendrocerus.
Penisvalvo-gonossiculal muscle: 0, absent; 1, present (pv-gss: 6A, 7C, 8C, 13D, 14C and 15A,C,D)
This muscle originates proximally from the penisvalva and inserts medially on the gonossiculus, and shares the site of origin with the gonostyle/volsella complex-gonossiculus muscle if the latter is present. Based on its site of origin, the muscle could be homologous with the penisvalvo-phallotremal muscles (muscles nb, nd in Schulmeister 2003b), which are involved in the opening/closure of the ejaculatory duct. In Ceraphronoidea, the phallotrema is located distodorsally on the aedeagus and is possibly opened/closed by the gonossiculus, at least in taxa with the gonossiculus adjacent with the distal aedeagus (see character 29). The penisvalvo-phallotremal muscles are present in Orthogonalys and Xyela, absent from Megalyroidea, and was not observed in other outgroup taxa. The presence of the penisvalvo-gonossiculal muscle is a possible synapomorphy for Ceraphronoidea.
This character is modified after character 281 of Schulmeister (2003b).
Gonostyle/volsella complex-parossiculal muscle: 0, present (gs-ps: Figs 9A, 11B and 12A–D); 1, absent
This muscle is present in those outgroup taxa where the parossiculus is present (i.e., Xyela, Evaniidae, Pristaulacus, Megalyroidea and Orthogonalys), but absent from other taxa examined.
This character is modified after character 283 of Schulmeister (2003b).
Site of insertion of gonostyle/volsella complex-volsellal muscles: 0, distivolsellar apodeme (proximal gonostyle/volsella complex-parossiculal muscle present; dva: Fig. 14D); 1, gonossiculus (gonostyle/volsella complex-gonossiculal muscle; gss: Figs 8C and 15C)
This muscle inserts exclusively on the gonossiculus and shares a common tendon in Ceraphronoidea and Evanioidea. In Xyela, Orthogonalys and Megischus, the muscle inserts on the distivolsellar apodeme (dva: Fig. 14D). This character is not applicable for Megalyroidea, in which the gonossiculus is absent and the muscle inserts on the single sclerite (Fig. 12B) representing the volsella.
This character is modified after character 284 of Schulmeister (2003b).
Two muscle bands, the lateral and median gonostyle/volsella complex-volsella muscles, are present in most taxa examined. One band arises ventrolaterally from the complex and one ventromedially from the complex (from the parossiculus if it is separated from the gonostipes or from the area that is separated by the submedian conjunctiva on the distoventral margin of gonostyle/volsella complex (see character 15). Only one band is present in Ceraphronidae and Trassedia.
This character is not applicable in Megalyroidea, in which the gonossiculus is absent.
This muscle is present in Xyela, Evanioidea, Orthogonalys and in some Dendrocerus species, though it was absent or not observed (questionable) from other taxa examined. The muscle arises medially from along the proximal margin of the gonostyle/volsella complex in Dendrocerus and shares a common site of insertion on the gonossiculus with the penisvalvo-gonossiculal muscle. This character is not applicable if the gonossiculus is absent (Megalyroidea).
This character is modified after character 285 of Schulmeister (2003b).
Parossiculo-penisvalval muscle: 0, absent (Figs 6A,C, 7B, 8A,C, 9A, 10A, 10B,D, 11A, 15C,D, 16A–C and 17B); 1, present (pss-pv: Figs 9B and 11B)
This muscle is present in Xyela and Orthogonalys but absent from other taxa examined.
Proximal gonostipes/volsella complex-harpal muscle: 0, absent (Figs 5C,D and 12B–D); 1, present (gs-hrp: Figs 6A,C, 7B, 8A,C, 9A, 10A,B,D, 11A, 15C,D, 16A–C and 17B)
Two gonostyle/volsella complex muscles insert on the proximomedial margin of the harpe in Ceraphronoidea similar to some basal Hymenoptera. The proximal muscle always arises from the gonostipes/volsella complex, whereas the distal muscle is usually divided and partly or exclusively arises from the lateral wall of the harpe. Three muscles insert on the harpe in Xyela—the apical gonostipes/volsella complex-harpal muscle arises from along the distal margin of the gonostyle and inserts on the lateral wall of the harpe (gs-hra: Fig. 11A,B), whereas two other muscles, the distal and proximal gonostipes/volsella complex-harpal muscles (gs-hrp, gs-hrd: Fig. 11A,B) arise proximally of the distal margin of the gonostyle and insert on the proximomedial margin of the harpe. Based on their relative location on the gonostyle/volsella complex and their site of insertion on the harpe, the distal muscle of Ceraphronoidea is structurally equivalent with the distal gonostipes/volsella complex-harpal muscle and the proximal muscle to the proximal gonostipes/volsella complex-harpal muscle of Xyela. The apical gonostipes/volsella complex-harpal muscle is absent from Ceraphronoidea, whereas the proximal gonostipes/volsella complex-harpal muscle is present in all Ceraphronoidea having a harpe, except for Trichosteresis, where no gonostipes/volsella complex-harpal muscles were observed. Among those taxa where the harpe is missing, one muscle, which is possibly homologous with the proximal gonostipes/volsella complex-harpal muscle, is present in Aetholagynodes and in Dendrocerus wollastoni (Figs 10A,B and 16A (gs-hrp). The muscle inserts on the medial wall of the distal region of the gonoforceps in Aetholagynodes and might move it laterally when contracted (Fig. 10A,B). Although the primary function of the muscle is to move the harpe in different directions, it might also be important for “anchoring” it in those Ceraphronoidea taxa where at least some bands of the distal gonostipes/volsella complex-harpal muscle extends between the proximomedian and proximolateral margins of the harpe. Two muscles are inserted on the harpe in Orthogonalys. The smaller proximal gonostipes/volsella complex-harpal muscle arises from the dorsomedian margin of the gonostyle/volsella complex and inserts on the median wall of the harpe. The larger muscle (gs-hr?: Fig. 9A,B) arises proximolaterally from the gonostyle/volsella complex and inserts partly on the proximedian edge of the ventral part of the complex and partly on the conjunctiva between the harpe and the gonostyle.
This character is modified after character 287 of Schulmeister (2003b).
Proximal gonostipes/volsella complex-harpal muscle site of origin: 0, proximally from the gonostyle/volsella complex (gs-hrp: Figs 8C, 11A, 15C,D and 17B–D); 1, distodorsally from the gonostyle/volsella complex (gs-hrp: Figs 6A,C, 7C, 8A and 16A–C)
This muscle always arises distodorsally in Ceraphronidae and in most Dendrocerus, whereas in Conostigmus, Megaspilus and in Lagynodinae it arises proximally from the gonostyle/volsella complex. The site of origin of the proximal gonostipes/volsella complex-harpal muscle possibly functions in movement of the harpe or stabilizes it when the distal gonostyle/volsella complex muscle is contracted. It moves the harpe laterally if the muscle arises from the distal part of the gonostyle/volsella complex, whereas it moves the harpe ventrally if it arises proximodorsally.
Distal gonostipes/volsella complex-harpal muscle: 0, present (gs-hrd: Figs 6C, 7B, 8A,C,D, 10A,B,D, 11A,B, 15C,D, 16A–C and 17B; gs-hr?: 9A, 9B); 1, absent (Figs 5C,D, 9D, 10A,B and 12B–12D)
The distal gonostipes/volsella complex-harpal muscle is present in all Ceraphronoidea except for taxa lacking a harpe. A small muscle band is present in Orthogonalys and this might be homologous with the distal gonostipes/volsella complex-harpal muscle based on the site of origin (see character 44). The muscle is absent from the rest of outgroup taxa examined except Xyela.
This character is modified after character 288 of Schulmeister (2003b).
Site of origin of distal gonostipes/volsella complex-harpal muscle: 0, from gonostyle/volsella complex (gs-hrd: Fig. 11A,B); 1, at least partly from proximolateral margin part of harpe (gs-hrd: Figs 8D and 17A,B); 2, exclusively from harpe (gs-hrd: Figs 7C, 8A, 10D and 13D)
The distal gonostipes/volsella complex-harpal muscle arises exclusively from the gonostyle/volsella complex in Xyela, partly from the complex and from the lateral wall of the harpe in Megaspilidae, and in Ceraphronidae attaches exclusively on the harpe, extending between its median and lateral walls. In Megaspilidae and Trassedia the proximomedial part of the harpe is extended proximally to form the proximomedial apodeme of the harpe (Figs 85, 86, see character 29), which serves as the site of insertion of the muscle. When the muscle bands extending between the median and lateral margins of the harpe contract, the harpe is possibly bent medially, which moves its distal part towards the midline, i.e. they “adduct” the gonoforceps (the proximal gonostipes/volsella complex-harpal muscles might play an important role in anchoring the proximal end of the apodeme when the distal muscle is contracted; see character 44). The contraction of the bands connecting the harpe with the gonostyle might be involved in the adduction or dorso-ventral movement of the harpe. We do not know what the function of the muscle attached exclusively on the harpe might be in Ceraphronidae. The most striking modification of the site of origin is found in Conostigmus triangularis, in which the muscle extends along almost the entire proximal margin of the gonostyle/volsella complex (Fig. 17A,B). This modification might be related to the extremely well developed, bilobed harpe. This character is not applicable if the muscle does not attach to the gonostyle/volsella complex.
Ability to open and close the male genitalia: 0, present (Fig. 18); 1, absent
Schulmeister (2003b) reported that the two halves of the male genitalia can be folded towards each other in Orussus, Stephanoidea and Ichneumonoidea. The ability to open and fold the male genitalia is present in Evanioidea and Megischus among the examined taxa. The ability to open and close the male genitalia might be related to the absence of a harpe. Adjusting the apical regions of the male genitalia might be important during copulation. In those taxa having a harpe, this function is facilitated by the gonostipes/volsella complex-harpal muscle. In those taxa lacking a harpe, the apical geometry of the male genitalia might be regulated by the opening and closure of the entire gonostyle/volsella complex.
Results and Discussion
The relationships among putatively basal apocritans (Evaniomorpha: Ceraphronoidea, Evanioidea, Trigonalidae and Megalyroidea) remain uncertain (Vilhelmsen et al. 2010, Heraty et al. 2011). Compared to other microhymenoptera (e.g., Platygastroidea, Chalcidoidea), the male genitalia of Ceraphronoidea is a complicated structure, composed of four to ten sclerites (cupula, two gonostipites, two parossiculi, two gonossiculi, two harpes, and the penisvalvae) with 12–14 muscles. The diversity in shape and pilosity of different male genitalia components (especially the apical harpe), as well as the orientation and thickness of muscles, provide characters for species delimitation (e.g., for Conostigmus compare Fig. 15C,D with Fig. 17A, B, and for Aphanogmus compare Fig. 10D with Fig. 14A). The presence or absence of muscles, conjunctivae or sclerites are also informative for higher level classification within Ceraphronoidea and even for the placement of the superfamily within Hymenoptera.
Based on our analyses, Ceraphronoidea is monophyletic with strong support and its position is unresolved (Fig. 19), or it is the sister to other evaniomorph taxa (Fig. 20). Ceraphronoidea share numerous plesiomorphic character states with Xyela (Xyeloidea: Xyelidae) that are either absent from other Evaniomorpha or are present only Trigonaloidea (sister to other non-Ceraphronoidea Evaniomorpha in the implied weighting analyses), including presence of a harpe and its muscles (ch. 28:1, 44:1 and 46:0), presence of the penisvalvo-phallotremal muscle (ch. 38:1), absence of the dorsomedian apodeme of the aedeagus (ch. 23:0), and the presence of T10 (ch. 2:1). Evaniomorph outgroup taxa were retrieved as monophyletic only in the implied weighting analyses with moderate jackknife support.
Schulmeister (2003b) concluded that the absence of a harpe is a synapomorphy for Vespina (Orussoidea + Apocrita) because even if there is a harpe-like apical sclerite there is no associated musculature. Our study reveals that a musculated harpe is present in Ceraphronoidea and Trigonaloidea. However, in Trigonaloidea, the arrangement of the harpal muscles is significantly different from that of Ceraphronoidea and lower Hymenoptera, thus hindering our ability to deduce primary statements of homology. As was mentioned in the introduction, numerous other, sometimes relatively complex character systems (i.e. mesotrochanteral depressor muscles; Gibson 1999), are shared by Ceraphronoidea and basal Hymenoptera.
Although the clades of outgroup taxa are not well supported in the present analysis, our study reveals some putative synapomorphies that can be tested in future phylogenetic studies. Perhaps the most important of these is the absence of the gonossiculus from the megalyroid volsella, which is composed of a single sclerite. The absence of the gonossiculus is preceded by the loss of numerous muscles that are connected to this sclerite in other taxa. There is no uncontradicted synapomorphy in our analysis for the weakly supported Evanioidea, but the absence of the cercal plate (ch. 1:1) is a putative synapomorphy for Gasteruptiidae + Evaniidae and may be an important character for future study in this group.
A longstanding hypothesis states that Ceraphronoidea is composed of two easily distinguishable and possibly monophyletic extant families—Megaspilidae and Ceraphronidae (Masner & Dessart 1967; Dessart 1995a,b; Mikó & Deans 2009). We retrieved Megaspilidae as monophyletic only in the implied weighting analyses, and with very weak support. Putative synapomorphies for the family are the ventrally curved penisvalva (ch. 22:1; with reversals in Aphanogmus, Elysoceraphron and Gnathoceraphron) and the presence of the proximomedial apodeme of the harpe (ch. 29:1; reversal in Trassedia).
Although the monophyly of Lagynodinae is strongly supported in the implied weighting analyses, it is resolved as a basal polytomy in the equal weighting analysis. Lagynodinae exhibits the largest number of character states shared with Xyela (Xyeloidea: Xyelidae). Lagynodinae and Ceraphronidae share two plesiomorphies with the outgroup taxa—the proximally pointed proximoventral margin of the gonostyle/volsella complex (ch. 11:1) and presence of the dorsolateral cupulo-gonostyle/volsella complex muscle (ch. 34:1). These findings challenge the current classification of the superfamily and suggest that Lagynodinae should probably be elevated from subfamilial status in Megaspilidae to family status.
Although our results do not resolve family level relationships in Ceraphronoidea with strong support, Megaspilinae and Ceraphronidae were monophyletic in both analyses. A monophyletic Megaspilinae had greater support in the implied weighting analyses than in the equal weighting analysis, with the following synapomorphies: straight proximoventral margin of the gonostyle/volsella complex (ch. 11:0), absence of apex gonostipitis (ch. 12:1), dorsoventrally continuous gonostyle/volsella complex (ch. 17:0), and absence of the dorsolateral cupulo-gonostyle/volsella complex muscle (ch. 34:0). The monophyly of the subfamily is also supported by the presence of a pterostigma (Mikó and Deans 2009), which is also present in two ceraphronid taxa, Masner and Trassedia.
Delimitation of taxa within Megaspilinae has been a longstanding problem. The work of previous authors (Dessart & Cancemi 1987; Fergusson 1980) suggested that the species of Megaspilinae are found in two assemblages surrounding Conostigmus and Dendrocerus, respectively. These lineages have been previously differentiated on the basis of the shape of the interocellar triangle (equilateral in Conostigmus) and the shape of the male flagellomeres (trapezoidal in Dendrocerus). However, these characters are not fixed for each lineage and the genera are defined by the morphology of their constituent species and not by clear generic concepts. Our analysis retrieves a Dendrocerus clade that contains three species of Conostigmus, a clade of “Conostigmus sensu stricto” species group (sensu Dessart and Cancemi 1987) and Trichosteresis, and a basal polytomy of Conostigmus and Megaspilus species.
Dendrocerus clade is supported by three synapomorphies—absence of medioventral conjunctiva of the gonostyle/volsella complex (ch. 16:0), the distodorsally extended site of origin of the gonostyle/volsella complex-volsellal muscles (ch. 37:1), and the distodorsal site of origin of the proximal gonostipes/volsella complex-harpal muscle (ch. 45:1). Three Conostigmus species with symmetric, elongate flagellomeres and an equilateral ocellar triangle are nested in this group. Dessart and Cancemi (1987) used two male genitalia characters in the delimitation of Conostigmus sensu strict—the presence of a parossiculus (ch. 13:0), and a proximodorsal notch of cupula (ch. 8:0). These two characters are the only synapomorphies for this clade in our analysis. The non-equilateral arrangement of the ocelli and non-trapezoidal flagellomeres found in Trichosteresis give this genus a dubious position within the Megaspilinae. This, and three Conostigmus species with typical, Conostigmus type structures of the ocelli and flagellomeres nesting within the Dendrocerus clade, provide additional evidence for the plasticity of ocellar and flagellar features previously used for classification of the subfamily.
Masner lubomirus was retrieved as sister of Ceraphronidae in the implied weighting analyses. The monophyly of the rest of Ceraphronidae is supported by only two character states—the concave anterior margin of S9 (ch. 3:1), and the lateral site of insertion of the mediolateral S9-cupulal muscle (ch. 32:1). These character states are related with the protraction and retraction of the male genitalia, which is cardinally different in Ceraphronidae, excluding Masner, and the rest of the Hymenoptera studied in our analysis. Reversal of this character is hence less likely, and given the presence of two possible megaspilid synapomorphies (presence of the occipital depression and the pterostigma; Mikó & Deans 2009) in Masner, it seems to be reasonable to hypothesize that Masner lubomirus is the most basal ceraphronid taxon.
Dessart and Cancemi (1987) classified Ceraphronidae into two “satellite” species groups, the Ceraphron-group and Aphanogmus-group based on the degree of lateral compression of the body and the shape of the male flagellomeres. Our implied weighting analyses retrieved a weakly supported clade of an Aphanogmus-group (Aphanogmus, Gnathoceraphron and Elysoceraphron) and a basal polytomy of Ceraphron, Cyoceraphron and Trassedia. The monophyly of Aphanogmus is not supported in the equal weighting analysis.
Ernst et al. (in press) suggest that Trassedia belongs to Ceraphronidae, contradicting its original classification within Megaspilidae (Cancemi 1996). Cancemi (1996) classified Trassedia in Megaspilidae based on the presence of a pterostigma and the presence of nine female flagellomeres. The position of Trassedia within the Ceraphronidae is supported by numerous other body characters—presence of Waterston’s evaporatorium, absence of an occipital depression, absence of a narrow sclerite anterior to the synsternum, presence of axillar setae, and a single mesotibial apical spur. Based on the present analysis and these additional characters, we transfer Trassedia from Megaspilidae to Ceraphronidae.
Members of the Aphanogmus-group share two character states—ventrally curved penisvalva (ch. 22:1) and cupula not extended proximoventrally (ch. 7:1). Although these two features appear symplesiomorphic, they more likely represent reversals because they are shared with Megaspilidae and with outgroup taxa. Our analysis retrieved two monophyletic lineages in the Aphanogmus-group clade in both analyses. Although these lineages are well supported, they are defined only by a few synapomorphies. Members of the Aphanogmus fasciipennis group share the vertical medioventral area of the gonostyle/volsella complex (ch. 18:0) and the gonossiculus not being associated with the distal aedeagus (ch. 21:1). The gonossiculus might have a different function in these taxa than in other Ceraphronoidea (see character 21). Gnathoceraphron is nested within this lineage, suggesting that it might best be treated as a junior synonym of Aphanogmus. The Aphanogmus nigripes-group is supported by only one synapomorphy, the presence of the distolateral projection of the harpe (ch. 30:0).
The importance of internal characters in systematics has been demonstrated by numerous studies (Deans et al. 2012). Besides providing an extra set of informative characters in a phylogenetic analysis, internal structures are crucial for the understanding and accurate interpretation of the external skeletal phenome. Song and Bucheli (2010) demonstrated that the phylogenetic signal of male genitalia is statistically similar to that of other body characters in many insects. The current phylogenetic analysis, the first for Ceraphronoidea, is based exclusively on male terminalia characters. A more comprehensive analysis of the greater somatic phenome is a prerequisite for clarifying unresolved parts of the Ceraphronoidea phylogeny and could reveal relevant information about the phylogenetic signal of external body vs. male genitalia characters. The specimen pool of our study is seemingly broad enough to use as a template for further molecular and morphological studies, which will be crucial for establishing a more robust and pragmatic Ceraphronoidea classification.
We thank for the following curators for loaning specimens for the present research: David Smith (USNM), Dave Karlsson (Sweden), George Melika (Hungary), Andrew Bennett (CNC), Bob Copeland (ICIPE), Brian Fisher (CALS), Tony van Harten (Portugal), Bob Blinn (NCSU) and Simon van Noort (SAMC). Bob Blinn, Andrew Ernst, Gary Gibson, Lars Vilhelmsen and Elijah Talamas are also acknowledged for their help in specimen data management and valuable discussions, Scott Shaw for the identification of Dinapsis specimens, and Chuck Mooney (NCSU) for his assistance with Scanning Electron Microscopy. This research was funded in part by the US National Science Foundation (NSF; grant DBI-0850223) and benefitted from meetings supported by the Biodiversity Synthesis Center of the Encyclopedia of Life the Phenotype RCN (http://www.phenotypercn.org; NSF DEB-0956049).
Imaris Bitplane. Available online at http://www.bitplane.com/ (accessed 27 August 2012).
PolaszekA. & DessartP. (1996) Taxonomic problems in the Aphanogmus hakonensis species complex; (Hymenoptera: Ceraphronidae) common hyperparasitoids in biocontrol programmes against lepidopterous pests in the tropics. Bulletin of Entomological Research 86: 419–422.
VilhelmsenL.IsidoroN.RomaniR.BasibuyukH.H. & QuickeD.L.J. (2001) Host location and oviposition in a basal group of parasitic wasps: the subgenual organ, ovipositor apparatus, and associated structures in the Orussidae (Hymenoptera, Insecta). Zoomorphology 121: 63–84.