Systematic Entomology (1987) 12, 413-431 Phylogenetic relationships and classification of the Vespinae (Hymenoptera: Vespidae)· JAMES M. CARPENTER Museum of Comparative Zoology, Harvard University,'! Cambridge, Massachusetts, U.S.A. ABSTRACT. The phylogenetic relationships of the genera, subgenera and species-groups ofthe Vespinae are analysed using cladistic techniques. The results are used as the basis for a natural classification of these wasps. The cladogram for the four genera recognized is: Vespa + .cProvespa + (Dolichovespula + Vespula)). No subgenera are recognized; all those previously described are synonymized with the appropriate genus. The . synonymies of Nyctovespa with Vespa and Rugovespula with Vespula are new. Introduction Nomenclature. More serious is the lack of a comprehensive phylogenetic system. As a The yellowjack6ts and hornets are the· most result, paraphyletic taxa are in use, and much­ familiar of wasps. Almost ubiquitous through­ discussion of the evolutionary development of out the north temperate regions and oriental behaviour is misplaced. Without differentiation tropics, they are generally recognized - and of features into derived and primitive states, feared - by layrrmh. The history of human disputes ongroupings and evolutionary develop­ knowledge of these wasps is ancient (cf. Sprad­ ment are unresolvable (cf. Yamane, 1976; bery, 1973; Edwards, 1980). The fascination Greene, 1979; Matsuura & Yamane, 1984; Mac­ engendered by their eusocial behaviour is also Donald & Matthews, 1975, 1984). The present ancient, and so they are the subject of an ever­ work is intended to redress this situation, by burgeoning number of behavioural studies. In providing the first comprehensive cladistic treat­ addition to social organization, considerable ment of supraspecific taxa in the Vespinae. current interest is focused on aspects of venom chemistry, economic injury and biological con­ trol potential (e.g. Akre & MacDonald, 1986). Previous treatments With the attention paid to these wasps, it might be expected that their classification would be The taxonomic history of the vesplne genera has stable, and their evolutionary relationships been reviewed by Bequaert (1930, 1932), grasped at least in outline. But this is not the Guiglia (1971) and Edwards (1980). I review case. Great controversy currently exists; both here only recent studies dealing with evolution­ nomenclature and classification vary from ary relationships, independent of taxonomy. author to author, and evolutionary relationships MacDonald & Matthews (1975, 1984) ques­ are in dispute. Some current classifications do tioned the placement ofVespula squamosa in the not follow the International Code of Zoological ruta species group, or Vespula s.str. (Bequaert, 1932). Correspondence: Dr J. M. Carpenter, Museum This stance has been followed by general of Comparative Zoqlogy, Harvard University, authors, who have regarded this species and its Cambridge, Mass. 02138, U.S.A. sister-species sulphurea as of uncertain place- 413 414 lames M. Carpenter ment (Edwards, 1980; Alae et aI., 1981). Mac­ characters (tyloides) is also incorrectly inferred, Donald & Matthews (1975: 1004) stated: and the states of the other two characters 'Comparative nest architectural, behavioral, associating Boreovespula and Dolichovespula morphological, and electrophoretic data cur­ are also true ofsylvestris (Metavespula). Finally, rently under study strongly indicate that the autapomorphic state for Boreovespula is not V.squamosa is wrongly aligned with the V.rufa true of all the included species, and this taxon is group'. No morphological or electrophoretic paraphyletic. data pertinent to this question have ever been Greene (1979) presented a discussion of published, but MacDonald (1977) and Mac­ behavioural characters which he considered to Donald & Matthews (1984) listed a number of show a 'phyletic sequence' with Dolichovespula nest architecture and behavioural characters as the most primitive yellowjacket genus and the which, they suggested, showed greater similarity vulgaris group the most recently derived group. to the V. vulgaris species group (Paravespula). This sequence, he argued, was the opposite It will be shown below that these features are of that proposed by Yamane (1976). In fact, either plesiomorphic in Vespinae, or uninforma­ Greene's discussion conflated character state tive. MacDonald (1977) stated that the male change and branching sequence. Although genitalia of squamosa are distinctive (also men­ Greene (1979: 614) stated that Dolichovespula tioned by Yamane et aI., 1980), and MacDonald 'probably has a relatively closer ancestral rela­ & Matthews (1984) cited a study by Kugler et al. tionship with Vespa', he later stated (p. 618): 'I (1976) showing seven ovarioles per ovary in have assumed the yellowjackets constitute a squamosa (versus six in most other vespines) as monophyletic group in which the genus Vespula also indicating wrong placement ofsquamosa in evolved from Dolichovespula stock, which in the rufa group. These traits, of course, are turn arose from the Vespa lineage'. This is the autapomorphies, and such unique characters are same sequence as Yamane proposed. Thus, irrelevant to the placement of the species - they whereas Greene is probably correct in the . only show that it is monophyletic. Synapomor­ polarities inferred for the behaviours discussed, phies uniting the squamosa and rufa groups are they are irrelevant to the question ofphylogene­ presented below. tic relationships among yellowjackets, as either Yamane (1976) presented a cladogram for polarity is in accord with the same scheme. vespine subgenera; however, it did not include Yamane et al. (1980) argued that the division Provespa or the parasitic subgenera (viz Vespula of Vespula s.l. into the rufa and vulgaris groups (only with austriaca), and Pseudovespula). This was inappropriate, and subdivided Vespula s.l. latter omission is unimportant, as recognition of into seven groups. Their character states were a subgenera for the parasitic species renders other mixture of primitive and derived, and they subgenera paraphyletic, as discussed below. stated (p. 34): 'These groups are, however, Yamane (1976: Table 3) listed fifteen characters rather phenetic'. They did not investigate the used in the construction of his tree, and differen­ interrelationships of these groups (nor establish tiated these into primitive and derived states. their monophyly), but did adopt Guiglia's (1972) He upheld the monophyly of Vespula + division of Vespula into the subgenera Vespula Dolichovespula, and Dolichovespula, but could and Paravespula! They considered the assign­ not determine whether Allovespula was the sis­ ment of the koreensis group to Paravespula and ter-group of Paravespula or Dolichovespula. As the squamosa group to Vespula to throw doubt developed below, the polarity of two of on this division. They stated (p. 34): 'They have Yamane's characters was incorrectly inferred, characters both peculiar to them and transitional and there are no grounds for grouping between the two subgenera'. The interrelation­ Allovespula with Dolichovespula (done on the ships of these groups are discussed below, aI).d it basis of grey, pliable nest paper). Yamane's is shown that the peculiar characters are cladogram also showed the interrelationships of autapomorphies, whereas the 'transitional' are the subgenera of Dolichovespula as: Meta­ symplesiomorphies. vespula + (Boreovespula + Dolichovespula). Archer (1982) described the subgenus The latter two taxa were grouped by three char­ Rugovespula for the koreensis group, primarily acters, and single autapomorphies were noted because-the male genitalia ofthis group 'are very for each of them. The polarity of one of these different from those of other species of Para- Phylogenetic relationships and classification ofVespinae 415 vespula'. It is shown below that the differences are vespine autapomorphies not mentioned by discussed all show the pnmitive state in Carpenter (1981). The autapomorphies are Rugovespula, and this taxon is the sister-group included in the matrix to establish the mono­ of Paravespula. phyly of each terminal taxon in the 'Cladistic Varvio-Aho et al. (1984) presented an Diagnoses' section. allozyme data set for the eight species of Euro­ pean yellowjackets, and analyses which they Character matrix (Table 1) suggested showed that Dolichovespula was not a 1 Prestigma length. Prestigma shorter than monophyletic group, and that Paravespula + pterostigma, O. Prestigma ~ pterostigma, 1. Pre­ Vespula was not monophyletic. Carpenter stigma 3x pterostigma, 2. (1987) showed that these authors incorrectly 2 lRI cell length. Length of first submarginal (lRI) analysed their data, and that their data are in fact cell ::; distance from ap<;x of cell to apex of wing, largely uninformative· on the relationships of O. Length of cell> this distance, 1. these yellowjacket species. 3 Base of lRS cell. M vein obliquely oriented with Finally, Matsuura & Yamane (1984) respect to m-euI at base of second submarginal (IRS) cell, O. M vein vertically oriented (apex of published two matrices totalling forty-two char­ discal cell truncate), 1. acters for the Vespinae (including Vespula, 4 lRS cell M: RS. Second submarginal cell with Paravespula, Rugovespula and the squamosa section ofM vein distal to m-eu2 crossvein shorter group as terminal taxa). The matrix is than section of RS vein basal to r2 crossvein, O. M reproduced here as Table 2, with correction of section