1999. The Journal of Arachnology 27:53±63 aTOWARDS A PHYLOGENY OF ENTELEGYNE SPIDERS (ARANEAE, ARANEOMORPHAE, ENTELEGYNAE) Charles E. Griswold1, Jonathan A. Coddington2, Norman I. Platnick3 and Raymond R. Forster4: 1Department of Entomology, California Academy of Sciences, Golden Gate Park, San Francisco, California 94118 USA; 2Department of Entomology, National Museum of Natural History, NHB-105, Smithsonian Institution, Washington, D.C. 20560, USA; 3Department of Entomology, American Museum of Natural History, Central Park West at 79th Street, New York, New York 10024 USA; 4McMasters Road, R.D. 1, Saddle Hill, Dunedin, New Zealand ABSTRACT. We propose a phylogeny for all entelegyne families with cribellate members based on a matrix of 137 characters scored for 43 exemplar taxa and analyzed under parsimony. The cladogram con®rms the monophyly of Neocribellatae, Araneoclada, Entelegynae, and Orbiculariae. Lycosoidea, Amaurobiidae and some included subfamilies, Dictynoidea, and Amaurobioidea (sensu Forster & Wilton 1973) are polyphyletic. Phyxelidinae Lehtinen is raised to family level (Phyxelididae, NEW RANK). The family Zorocratidae Dahl 1913 is revalidated. A group including all entelegynes other than Eresoidea is weakly supported as the sister group of Orbiculariae. The true spiders or Araneomorphae (ara- Raven (1985) and Goloboff (1993a) for My- neae verae of Simon 1892) comprise more galomorphae (15 families); Coddington (1986, than 30,000 described species. The classi®- 1990a, b) for Orbiculariae (13 families) and cation of this group has undergone a revolu- Entelegynae; Platnick et al. (1991) on haplo- tion in the last 30 years, sparked by Lehtinen's gynes (17 families) and Araneomorphae; Gris- (1967) comprehensive reassessment of ara- wold et al. (1994, 1998) for Araneoidea (12 neomorph relationships and steered by Hen- families), and Griswold (1993) for Lycosoidea nig's phylogenetic systematics (Hennig 1966; and related families (11 families). Platnick & Gertsch 1976). Spider classi®ca- The latter studies targeted large but rela- tion, portrayed by some authors as chaotic tively well-de®ned lineages. It is now feasible (Head 1995; Elgar et al. 1990; Vollrath & to probe how these large lineages are related. Parker 1997: Prenter et al. 1997) is actually We chose exemplars from all cribellate fami- one of the better-understood megadiverse or- lies, reasoning that taxa retaining this plesiom- ders (Coddington & Levi 1991): including the orphic feature are more likely to straddle the results reported here, 100 of the 108 currently basal nodes of the phylogeny of higher groups recognized families (93%) have been placed than are their relatives that have lost the cri- cladistically, that is, in a higher taxon based bellum: therefore they are most likely to re- on evidence assessed phylogenetically. New ¯ect phylogenetic groundplans. Although character systems compared across worldwide phylogenetically ancient, the cribellum is a samples of taxa have led to many new and complex feature unlikely to have evolved thought-provoking hypotheses in araneo- more than once. Most major araneomorph morph phylogeny. The strongest test of such clades have cribellate members (exceptions hypotheses is how simply they can account for are Palpimanoidea and Dionycha). A phylog- the available data, i.e., most parsimonious eny of these basal taxa should mirror the re- cladograms based on matrices of taxa by char- lationships of the large clades they exemplify. acters. Tests speci®cally designed at the fam- As Lehtinen (1967: 202) declared, ``because ily level and above are increasingly common: of the central position of the Cribellate groups in Araneomorphae, a detailed revision of them aPresented as part of a symposium on ``Higher is a short cut to a rough classi®cation of the Level Phylogenetics of Spiders.'' whole suborder.'' 53 54 THE JOURNAL OF ARACHNOLOGY A parsimonious cladogram based on an ex- We omitted Gradungulidae because the cri- plicit taxon-character matrix is concise, logi- bellate genera are extremely rare in collections cal, and testable. Our analysis tests many su- and its placement in Austrochiloidea seems prafamilial hypotheses of the last 30 years and ®rm (Forster, Platnick & Gray 1987; Platnick is the ®rst attempt to relate them using quan- et al. 1991). Voucher specimens for exemplars titative phylogenetic techniques: Amaurobioi- are deposited in the California Academy of dea (sensu Forster & Wilton 1973), Amauro- Sciences (CAS) with the exception of Vytfutia biidae and included subfamilies (sensu (Deeleman coll.) and male Raecius (NHMV). Lehtinen 1967), Dictynoidea and Desidae Character data taken from the literature in- (sensu Forster 1970), Entelegynae (sensu clude the suite of classical characters from Coddington 1990b; Coddington & Levi 1991), spider internal anatomy (characters 43±49: Lycosoidea (Homann 1971; sensu Griswold Platnick 1977; ex Millot 1931, 1933, 1936; 1993); Orbiculariae (sensu Coddington 1986, Marples 1968 [these have been recorded for 1990a, b); and the 'RTA clade' (sensu Cod- hypochiloids, austrochiloids, and such a wide dington & Levi 1991). variety of haplogyne and entelegyne Araneo- clada that we are con®dent that the assumed TAXA AND CHARACTERS states for entelegyne exemplars in Table 1 are Table 1 comprises 43 exemplars from 21 of justi®able]) and character 114, presence/ab- the 22 araneomorph families with cribellate sence of the muscle M29 in the male palp (as- members. As outgroups we included HYPO- sumed for all taxa in Table 1 following Huber CHILIDAE (Hypochilus), AUSTROCHILI- 1994). Silk ultrastructure data are taken from DAE (Hickmania and Thaida), and FILIS- Eberhard & Pereira (1993) and from unpub- TATIDAE (Filistata and Kukulcania, lished observations (R. Carlson in lit.). Filistatinae). From eresoids, we included Characters, character states, and codings are OECOBIIDAE (Oecobius and Uroctea) and listed in Table 1. Some features are most suc- ERESIDAE (Eresus and Stegodyphus). From cinctly described by reference to a taxon for Orbiculariae we included DEINOPIDAE which they are typical, e.g, 'dictynid conduc- (Deinopis and Menneus), ULOBORIDAE tor.' For ®gures of entelegyne genitalia see es- (Octonoba and Uloborus) and an araneoid pecially Lehtinen (1967), Coddington (1990a) groundplan. Recent phylogenetic study of this and Griswold (1993); for features of spinner- superfamily (Griswold et al. 1998) gives us ets see especially Platnick et al. (1991) and con®dence that the reconstructed groundplan Griswold et al. (1998). Character evolution is accurately re¯ects the primitive conditions for summarized on the cladogram (Fig. 1) opti- Araneoidea. From ``dictynoids'' we included mized via Clados (Nixon 1992) and MacClade DICTYNIDAE (Dictyna and Nigma, Lathys, (Maddison & Maddison 1992). and Tricholathys representing Dictyninae, Ci- curinae, and Tricholathysinae, respectively), METHODS AND ANALYSIS DESIDAE (Badumna candida, B. longinquua, Spigot classi®cation follows Coddington and Matachia, formerly Matachiinae), and NI- (1989); all specimens were critical point dried CODAMIDAE (Megadictyna). From ``amau- before scanning electron microscope (SEM) robioids'' we included AMAUROBIIDAE examination of spinning organs. Behavioral (Amaurobius and Callobius (Amaurobiinae), observations were made on living animals in Metaltella (Metaltellinae), Retiro and Pimus the ®eld or lab. (Macrobuninae), Phyxelida, Vytfutia, and Xe- The matrix (all characters unordered and vioso (Phyxelidinae)), AMPHINECTIDAE equally weighted) was analyzed with three (Maniho), NEOLANIDAE (Neolana), AGE- phylogenetic packages: Nona 1.6 (Goloboff LENIDAE (Neoramia), and TITANOECI- 1993b), Hennig86 1.5 (Farris 1988), and Pee- DAE (Goeldia and Titanoeca). From lyco- Wee 2.6 (Goloboff 1997), using a wide variety soids and related groups we included of randomized and directed search strategies. CTENIDAE (Acanthoctenus), MITURGIDAE Nona (using both `amb 5' and `amb-' options (Raecius and Uduba, Uliodoninae), PSE- for clade support) and Hennig86 found the CHRIDAE (Psechrus), STIPHIDIIDAE same three topologies, including Fig. 1 (length (Baiami and Stiphidion); TENGELLIDAE 376, ci 0.43, ri 0.69). The two alternate to- (Tengella), and ZOROPSIDAE (Zoropsis). pologies involved local rearrangements of Ni- GRISWOLD ET AL.ÐENTELEGYNE SPIDER PHYLOGENY 55 codamidae and Eresoidea. The strict consen- ber of additional steps required in the shortest sus has one 4-tomy at the entelegyne node, trees for which that node collapses. The fol- otherwise identical to Fig. 1. lowing Bremer Support values were found We used successive and implied weighting for the clades on Fig. 1: Austrochiloidea (5), (Carpenter 1988; Goloboff 1993c) to further Araneoclada (1), Entelegynae (1), Haplogy- evaluate the data. Successive weighting in nae (8), Eresoidea (1), Stegodyphus-Eresus Nona (length 16,346, ci 0.63, ri 0.80) pre- (8), Uroctea-Oecobius (4), Canoe Tapetum ferred Fig. 1. Successive weighting in Hen- Clade (0), Orbiculariae (2), Deinopis-Octo- nig86 (length 1127, ci 0.79, ri 0.88) found Fig. noba (3), Deinopis-Menneus (4), Uloborus- 1 as well as two other trees one step longer. Octonoba (5), Megadictyna-Zoropsis (0), Di- Pee-Wee at concavity functions of 3 and 4 (®ts vided Cribellum Clade (1), Titanoecoids (1), 962.6 and 1009.0, length 378) found one tree Titanoeca-Goeldia (2), Vytfutia-Phyxelida in which Retiro and Pimus swapped places, (2), Xevioso-Phyxelida (2), RTA Clade (1), otherwise identical to Fig. 1. Concavity 5 (®t Dictynidae (2), Tricholathys-Nigma