Systematic Entomology (2003) 28, 433–450 Phylogeny of Syrphidae (Diptera) inferred from combined analysis of molecular and morphological characters GUNILLA STA˚HLS1 , HEIKKI HIPPA2 , GRAHAM ROTHERAY3 , JYRKI MUONA1 andFRANCIS GILBERT4 1Finnish Museum of Natural History, University of Helsinki, Finland, 2Swedish Museum of Natural History, Stockholm, Sweden, 3National Museums of Scotland, Edinburgh, U.K. and 4School of Biological Sciences, Nottingham University, Nottingham, U.K. Abstract. Syrphidae (Diptera) commonly called hoverflies, includes more than 5000 species world-wide. The aim of this study was to address the systematic position of the disputed elements in the intrafamilial classification of Syrphidae, namely the monophyly of Eristalinae and the placement of Microdontini and Pipizini, as well as the position of particular genera (Nausigaster, Alipumilio, Spheginobaccha). Sequence data from nuclear 28S rRNA and mitochondrial COI genes in conjunction with larval and adult morphological characters of fifty-one syrphid taxa were analysed using optimization alignment to explore phylogenetic relationships among included taxa. A species of Platypezidae, Agathomyia unicolor, was used as outgroup, and also including one representative (Jassidophaga villosa) of the sister-group of Syrphidae, Pipunculidae. Sensitivity of the data was assessed under six different parameter values. A stability tree sum- marized the results. Microdontini, including Spheginobaccha, was placed basally, and Pipizini appeared as the sister-group to subfamily Syrphinae. The monophyly of subfamily Eristalinae was supported. The results support at least two independ- ent origins of entomophagy in syrphids, and frequent shifts between larval feeding habitats within the saprophagous eristalines. Introduction At the beginning of the last century, Syrphidae was divided into 2–20 subfamilies by different authors. A system Syrphidae (Diptera: Lower Cyclorrhapha) commonly called of three subfamilies (subfamilies Microdontinae, Eristalinae flower- or hoverflies, comprise more than 5000 described and Syrphinae) was adopted for Syrphidae more than species, one of the most speciose of dipteran families 25 years ago, largely for the sake of convenience (Thompson (Thompson & Rotheray, 1998). In contrast to the fairly & Rotheray, 1998). The traditional classification of Syrphi- uniform flower-feeding habits of adult syrphids, larvae are dae is based largely on adult characters. In their cladistic found in a very diverse array of habitats. Those of subfam- study of larval characters, Rotheray & Gilbert (1999) ily Eristalinae are saprophagous in dead wood, copropha- considered all previous estimates of syrphid classification, gous, phytophagous, aquatic filterfeeders or inquilines in dealing with a reasonably large section of the family social insect nests, whereas larvae of Microdontinae are and addressing hypotheses about syrphid phylogenetic inquilines in ants’ nests, and larvae of Syrphinae are mostly relationships based on non-traditional characters (e.g. predaceous on soft-bodied Homoptera. chromosomes). Two recent studies address the monophyly of Syrphidae and the systematic position of various clades within the Correspondence: Dr G. Stahls, Finnish Museum of Natural family. Skevington & Yeates (2000) investigated the use History, PO Box 17, FIN-00014 University of Helsinki, Finland. of the mitochondrial genes 12S and 16S for developing E-mail: [email protected] a phylogeny of superfamily Syrphoidea (Pipunculidae þ # 2003 The Royal Entomological Society 433 434 Gunilla Sta˚hls et al. Syrphidae). Their combined analysis presented convincing unique attributes of both the larvae and the adults of the support for this sister-group relationship (bootstrap sup- Microdontinae (containing the single tribe, Microdontini) port of 87%), but support for the monophyly of Syrphidae and lack of clear synapomorphies from adult characters to was weak, and the resolution within Syrphidae was limited. combine it with other Syrphidae have led to many Rotheray & Gilbert (1999) presented a hypothesis of intra- hypotheses of its relationships (Skevington & Yeates, familial relationships of Syrphidae based on the first com- 2000). Thompson (1969, 1972) considered Microdontinae prehensive study of larval morphological characters, to form a basal, monophyletic group with respect to the including many several outgroup taxa, and concluded rest of Syrphidae, and this view was partly supported by Syrphidae was monophyletic. Within the family, their Shatalkin (1975a). Contrary to the traditional monotypic results differ fundamentally from the traditional classifica- classification of Microdontinae, Shatalkin classified Sphe- tion, with few traditional suprageneric taxa recovered. A ginobaccha, Nausigaster and Microdon into Microdontinae. study by Cheng et al. (2000) employed both molecular and Thompson (1969) and Speight (1987) discussed raising morphological data to address the question of the place- Microdontinae to familial rank, but were not followed ment of Pipizini, but their small dataset (especially by subsequent authors. In sharp contrast, larval characters the molecular component) was too limited to elucidate placed the Microdontinae within a monophyletic Syrphidae relationships. as the sister-group to the Syrphinae þ Pipizini, a placement The studies of Rotheray & Gilbert (1999) and Skevington & never suggested from studies of adult characters (Rotheray & Yeates (2000) both questioned the monophyly of sub- Gilbert, 1999; this study Fig. 1). The molecular phylogeny of family Eristalinae (previously called Milesiinae). The Skevington & Yeates (2000) showed a basal position for Platypezidae Pipunculidae Alipumilio Eumerus Merodon Cheilosia morio Cheilosia-borer Portevinia Cheilosia-fungivore Ferdinandea Rhingia Volucella inflata Volucella pellucens Volucella inanis Microdon mutabilis Microdon analis Pipizella Pipiza Trichopsomyia Melanostoma Toxomerus Platycheirus Baccha Sphaerophoria Paragus Chrysotoxum Syrphus Hammerschmidtia Chrysogaster Sphegina Neoascia Syritta Xylota Ceriana Blera Sericomya Anasimyia Eristalis Helophilus Chalcosyrphus Brachypalpoides Nausigaster Spilomyia Milesia Criorhina Temnostoma Callicera Fig. 1. Strict consensus of sixteen most parsimonious trees inferred from analysis of 187 larval characters (length ¼ 578). # 2003 The Royal Entomological Society, Systematic Entomology, 28, 433–450 Phylogeny of Syrphidae – combined analysis 435 Microdontini, thus supporting the hypothesis proposed by position of particular genera (Nausigaster, Alipumilio, Thompson (1972) and Shatalkin (1975a). Spheginobaccha). In addition, we examined the evolution Like the Microdontinae, the classification of Pipizini of the larval feeding modes in the light of the phylogeny (geographical distribution world-wide) has been a long- recovered. We used three different independent datasets: standing problem, the reason being that its members have molecular (this study), and larval (Rotheray & Gilbert, a syrphine-like larva, but a milesiine-like adult (Thompson, 1999) and adult (H. Hippa & G. Sta˚hls, unpublished data) 1972). Vockeroth (1969), Thompson (1972) and Vockeroth morphology. We discovered considerable conflict between & Thompson (1987) referred Pipizini to Eristalinae, phylogenetic hypotheses based on any one of these datat- whereas Shatalkin (1975a, b) classified them within Cheilo- sets, and hence we employ combined analysis (Total Evi- siinae. Rotheray & Gilbert’s (1989, 1999) work suggested dence, Kluge, 1989) to evaluate syrphid relationships in that Pipizini are the sister-group of Syrphinae. Kuznetzov light of the diverse and comprehensive dataset we gath- (1987, 1992) proposed elevation of Pipizini to subfamilial ered for this purpose. We explored which groupings rank, based on scanning electron microscope studies of consistently appeared in the phylogenetic trees under dif- first-instar larvae of syrphines, pipizines and eristalines. ferent weighting schemes. Such a sensitivity analysis (sensu Like the Syrphinae, the pipizines have separated posterior Wheeler, 1995) gives information on node stability and respiratory tubes in L1 larvae, but Kuznetzov claimed that congruence. We restrict our analyses to this kind of the long sclerotized tubes of the Pipizini were very different taxonomic congruence within a combined dataset. The from the small non-sclerotized protuberances of the Syrphi- result is summarized in a stability tree (sensu Schulmeister nae: such an autapomorphy is a weak basis for a change et al., 2002). rank. Parsimony analysis of molecular characters of Skeving- ton & Yeates (2000) suggested a weakly supported sister- group relationship between Pipizini (Triglyphus) and Materials and methods Milesiini (Orthoprosopa). DNA sequences collected by Cheng et al. (2000) to test specifically the placement of Taxa and characters Pipizini, included four representatives in a study comprising only seven taxa in total. Their parsimony analysis showed The taxon sampling covered as much taxonomic diversity that the relationship between Pipizini and Syrphinae (Meta- as possible, representing thirteen of the fourteen recognized syrphus corollae) was closer than the relationship between syrphid tribes (Thompson & Rotheray, 1998), totalling Pipizini and Eristalinae (Eumerus strigatus and Eristalis fifty-one syrphid taxa (Table 1). Most included species tenax), and hence Pipizini should be transferred to Syrphi- have Palaearctic or Holarctic distributions, but a few