Phylogenetic relationships among superfamilies of Hymenoptera

The first comprehensive analysis of higher-level phylogeny of the order Hymenoptera is presented. The analysis includes representatives of all extant superfamilies, scored for 392 morphological characters, and sequence data for four loci (18S, 28S, COI and EF-1a). Including three outgroup taxa, 111 terminals were analyzed. Relationships within symphytans (sawflies) and Apocrita are mostly resolved. Well supported relationships include: Xyeloidea is monophyletic, Cephoidea is the sister group of Siricoidea + [Xiphydrioidea + (Orussoidea + Apocrita)]; Anaxyelidae is included in the Siricoidea, and together they are the sister group of Xiphydrioidea + (Orussoidea + Apocrita); Orussoidea is the sister group of Apocrita, Apocrita is monophyletic; Evanioidea is monophyletic; Aculeata is the sister group of Evanioidea; Proctotrupomorpha is monophyletic; Ichneumonoidea is the sister group of Proctotrupomorpha; Platygastroidea is sister group to Cynipoidea, and together they are sister group to the remaining Proctotrupomorpha; Proctotrupoidea s. str. is monophyletic; Mymarommatoidea is the sister group of Chalcidoidea; Mymarommatoidea + Chalcidoidea + Diaprioidea is monophyletic. Weakly supported relationships include: Stephanoidea is the sister group of the remaining Apocrita; Diaprioidea is monophyletic; Ceraphronoidea is the sister group of Megalyroidea, which together form the sister group of [Trigonaloidea (Aculeata + Evanioidea)]. Aside from paraphyly of Vespoidea within Aculeata, all currently recognized superfamilies are supported as monophyletic. The diapriid subfamily Ismarinae is raised to family status, Ismaridae stat. nov. !The Will Henning Society 2011. Hymenoptera have traditionally been divided into two suborders, ‘‘Symphyta’’ (sawflies) and Apocrita, with the latter subdivided into ‘‘Parasitica’’ (parasitoids) and Aculeata (stinging wasps, bees, and ants). Even before the advent of cladistic principles, ‘‘Symphyta’’ were not generally considered a monophyletic group. The division between ‘‘Parasitica’’ and Aculeata was usually based on behavior, that is parasitic versus predatory behaviour, but the biology of many Aculeata is ectoparasitoid. Whereas Aculeata have a diagnostic synapomorphy (the sting), none has been forthcoming for Parasitica, and the likelihood that this group is paraphyletic has been widely accepted. Phylogenetic research on higher level relationships within the order as a whole has been based primarily on morphological data. Intuitive approaches that included character diagnoses began with a series of papers by Rasnitsyn (1969, 1980, 1988) and Königsmann (1976, 1977, 1978a,b). Rasnitsyn!s research was not cladistic, but incorporated much fossil data and was original. Königsmann!s research mostly comprised characters taken from the literature, and although ostensibly cladistic, the data were analyzed intuitively, with little *Corresponding author: E-mail address: [email protected] ! The Willi Hennig Society 2011 Cladistics 10.1111/j.1096-0031.2011.00366.x Cladistics 27 (2011) 1–33 resolution obtained at the superfamily level within the ‘‘Parasitica’’ grade. Rasnitsyn!s (1988) summary paper came to be used as the starting point for quantitative phylogenetic studies of higher level relationships across the Hymenoptera. Ronquist et al. (1999) revisited the 1988 character list, and coded it into a numerical cladistic matrix. However, their analyses of the matrix resulted in little resolution, especially within the Apocrita, and thus they were unable to corroborate many of the relationships suggested by Rasnitsyn. Nevertheless, the Ronquist et al. (1999) data set was used in the combined molecular ⁄morphological analyses of the higher-level relationships of the Hymenoptera as a whole (Carpenter and Wheeler, 1999) and Apocrita (Dowton and Austin, 2001). The matrix of Ronquist et al. (1999) represents Rasnitsyn!s (1988) interpretations of many character systems, but in many cases these are not cladistic in nature. Rasnitsyn!s interpretations owed much to fossils and many character definitions are not ideal for recent taxa. Moreover, the idiosyncratic manner in which Rasnitsyn (1988) analyzed his data allowed him to ignore homoplasy and down-weight characters on an ad hoc basis. The characters were coded for hypothesized ground-plans instead of representing observations, and many characters, obviously including those published after 1988, were not included. Sharkey and Roy (2002) critically examined the characters of the matrix that pertain to wings (37 of the 169 characters). They edited these but did not deal with other morphological characters, which in many cases need to be reconsidered. Detailed analyses of the Symphyta by Vilhelmsen (2001) and Schulmeister et al. (2002) produced reasonably well resolved relationships of the basal lineages of Hymenoptera. In contrast, Sharkey and Roy!s (2002) reanalysis of the Ronquist et al. (1999) data set (recoded wing characters) is representative of what little is known of apocritan superfamilial phylogenetic relationships, being largely unresolved. Several molecular analyses have been published and in general these have focused on either the Apocrita (Dowton and Austin, 1994, 2001; Dowton et al., 1997; Castro and Dowton, 2006) or Symphyta (Schulmeister et al., 2002; Schulmeister, 2003). Davis et al. (2010) recently compiled a hymenopteran supertree at the family level from the studies published up until then. Besides the usual deficiencies of supertree analyses (see Goloboff and Pol, 2002), Davis et al. were unable to include the first HymAToL outputs relevant to higherlevel phylogeny of the Hymenoptera, i.e., Vilhelmsen et al. (2010) on mesosomal characters, including a multitude of muscular and other internal characters, and Heraty et al. (2011), presenting a large molecular analysis employing 28S, 18S, COI and EF-1a. Most recently, Sharanowski et al. (2010) proposed a novel hypothesis, based on EST analyses, in which Chalcidoidea were excluded from Proctotrupomorpha. However, although gene sampling was extensive, the taxon sampling was minimal (10 Hymenoptera) and the results varied depending on the method of analysis, thus we do not address their results further in this study. It is clear from previous research that both morphology and sequence data are informative. Combining these data sources into a comprehensive dataset including all hymenopteran superfamilies was an obvious path to take, and the focus of this paper. Materials and methods