THE PHYLOGENETIC SYSTEM OF THE GAYELLINI ( HYMENOPTERA : VESPIDAE ; MASARINAE ) * BY

The Gayellini is one of the two. tribes of Masarinae (Carpenter, 1981). Endemic to the Neotropics, the majority of the species are Patagonian, but one ranges as far north as Mexico. With ten described species, the group is far less speciose than its sister-tribe Masarini, which has over 200 described species (cf. Richards, 1962), and most species are rarely collected. These wasps have a very distinctive appearance among Vespidae (Fig. 1), and their taxonomic history has been more turbulent than any other higher vespid taxon. Although the phylogenetic placement of the group as a whole has now evidently been settled (Carpenter, 1981), no study has been made of the species. The current generic classification is fragmented, and there have been no keys to all of the taxa. In this paper, I investigate the phylogenetic relationships of the species, and present a revised generic classification along with keys to all taxa.

tres" of the "Eum6niens" because the forewing recurrent veins (m-cu_2) are received in separate cells (Fig. 6), as in the other genera placed in this section (Raphiglossa, and Stenoglossa-Psili- glossa).In other vespids he studied both veins were received by the second submarginal cell.Ashmead (1902a) described the subfamily Raphiglossinae (in his Eumenidae) for this group, but by that time other genera had been described which had the diagnostic character of the recurrent veins.These were Euparagia and Paramasaris, both considered probable masarines by their authors (Cresson, 1879, and  Cameron, 1901, respectively).Ashmead (1902b) proposed the tribe Euparagiini in his Masaridae for these two genera.So the recurrent *Manuscript received by the editor September 28, 1988   211 Psyche [Vol. 95  veins were no longer uniquely diagnostic.Bequaert (1918) ques- tioned whether Gayella belonged in the Raphiglossinae, since the longitudinal plaiting of the forewings "is very obsolete" in the genus, and Bradley (1922) placed it in its own subfamily in Vespidae s./.Bequaert (1928) transferred Paramasaris to this subfamily, based on its possession of the characteristic hindwing venation of Gayella (Fig. 3).Richards (1962) included the Gayellinae in his Masaridae, but his dendrogram showed Euparagiinae as more closely related to the subfamily Masarinae.I (Carpenter, 1981)demonstrated that Richards' Masaridae was a paraphyletic group, since Euparagiinae is the sister-group of Vespidae as a whole.Four synapomorphies were adduced which showed a sister-group relationship between gayellines and masarines: presence of hypostomal apodemes, loss of the midfemur basal ring, loss of the scutal lamella and provisioning with nectar and pollen.Gayellines and masarines were treated as tribes in an expanded Masarinae, the system followed here.
Gayella was originally described as monotypic for G. eumenoides by Spinola (1851).Saussure (1855 in Vol. 3 of ttudes), Willink   (1956, 1963) and Willink and Ajmat de Toledo (1979) added five species.The latter paper provided a key to the six species, however I believe that the key given here will be easier to use.Paramasaris was also originally described as monotypic, for P. fuscipennis Cameron.
Cameron (1904) later described a new genus Zethoides (non Zethoides Fox, 1899; Plesiozethus Cameron, 1905, and Metaze- thoides Schulz, 1906, are replacement names) for Z. flavolineatus, which differed from Paramasaris in having only two (Fig. 5), as opposed to three (Fig. 6), submarginal cells.Zavattari (1912) ques- tioned whether Cameron had described this character correctly, and Bradley (1922) suspected that Plesiozethus was a synonym of Para- masaris.This was confirmed by Bequaert (1928), who showed that the number of submarginal cells was variable, and synonymized flavolineatus withfuscipennis.Giordani Soika (1974) described two new species in the genus, but provided no key.He also described a new genus, Paragayella, monotypic for the new species Paragayella richardsi.I consider this genus a synonym of Paramasaris, as dis- cussed below.I also give the first key to species of the group.

MATERIALS AND METHODS
All of the species have been examined.Types of Paramasaris have been seen; treatment of Gayella follows Willink's concepts.Com-Psyche [Vol. 95  plete label data for all material of Paramasaris are listed under taxonomic notes for each species; for the relatively better known Gayella only provinces are noted.Acronyms for collections are listed below, along with the name of the individuals who provided the material where this was borrowed.
AMNH American Museum of Natural History, New York (M. S. Favreau)  Carpenter (1981), except that I have adopted Snelling's (1986) more descriptive terms "preoccipi- tal" and "postocular" for the carinae previously termed "dorsal occipital keel" and "ventral occipital keel" (Richards, 1962).Detailed examination of the labiomaxillary complex and male geni- talia was made by dissection of these structures, clearing slightly in cold lactophenol, and exaoaination in glycerin.Measurements were made with an ocular micrometer.Illustrations were made with a Wild M-400 photomacroscope employing Kadak TMAX 400 film.Cladistic analysis (Hennig, 1966) was performed for all the features discussed in this paper.Outgroup taxa include Masarini and Euparagiinae, with reference to other Vespidae also occasionally made.

Subfamily characters
First discuss some morphological features important in higher- level vespid relationships, before turning to consideration of the phylogenetic relationships among the species.Autapomorphies of  the Gayellini listed by Carpenter (1981) include the hindwing with Cu diverging from M+Cu far distad of the insertion of cu-a (Fig. 3), the clypeus with the dorsal margin bisinuate (Fig. 12), the first metasomal tergum and sternum fused and metasomal segments after II retractile (the latter two convergent with other vespids).Some other autapomorphies are mentioned below.
Forewing discal cell.Carpenter (1981:14) noted that the discal cell is shorter than the submedian in Paramasaris.This is also the case in Gayella (Fig. 1), and this should be considered a reversion from the state of an elongate discal cell in other Vespidae (Fig. 4; cf Carpenter, 1981), and thus an autapomorphy of Gayellini.
Forewing radial region.The variation in the number of sub- tnarginal cells in Paramasaris was alluded to above.Besides fuscipennis, I have seen loss of r-m2 producing two submarginal cells in several specimens of brasiliensis (including the allotype and para- type, Fig. 5).The placement of m-cu varies as well, sometimes meeting M at the fork where RS diverges.But this is not correlated with number of cells, and most specimens have the usual condition of RS diverging first (Figs.5-6).In addition, the specimen of Para- gayella richardsi from Formoso, Brazil, has a very small adventi- tious cell at the junction of r-m3 and RS on one wing (Fig. 6), and both Goias specimens have an adventitious vein spur arising from the middle of r-m3 (Fig. 6).
Clypeus.The clypeus is narrower than its height in all species, particularly in males .Richards (1962:46) stated that the reverse is true in Paramasaris, perhaps a lapsus.This is not the usual state in Vespidae, and is perhaps apomorphic, although the degree of narrowing varies in the tribe.Occipital carinae.Gayellini have both the postocular and preoccipital carinae in the groundplan, contrary to Richards   (1962:12).The postocular carina is reduced in length, and may be present only as a trace just ventral to the eye in Gayella, but is typically obvious in the eumenoides group.The carinae are almost confluent in many specimens of eumenoides and araucana, separated by only a slight gap (Fig. 18).The "complete" carina in Para- masaris (Richards, 1962:46; Fig.  19) is evidently produced by confluence of the postocular with the preoccipital carina, as occurs in some Masarini (Snelling, 1986) and probably other Vespidae (Carpenter, 1988).The postocular carina is effaced in Paragayella (Fig. 20).
Hypostomal apodemes.These are present in all species, which supports the interpretation of synapomorphy with Masarini.They are always very narrow (Fig. 23).
Pronotal carina.Paramasaris and Paragayella are notable for having two parallel carinae on the pronotum.One is present at the anteroventral margin of the pronotum; the other is posterior to this and runs towards the humeri and dorsum of the pronotum (Figs. 29-31,40).The second carina shows a transformation series in development, ranging from short lateral sections only (Paragayella, Figs. 24, 29), to extending to the dorsum (Paramasaris brasiliensis, Figs. 25, 30), to complete across the dorsum (P.cupreus and fuscipennis, Fig. 26).This series apparently corresponds to the phylogenetic relationships among these species (Fig. 2).Gayella has only the anterior carina (Fig. 7).Euparagiinae also has only the anterior carina (Fig. 8), although the humeri are somewhat raised in scutellaris.In Masarini the anterior carina is usually blunt, and a lateral carina on the humeri may be present (Fig. 37).In all these groups, the anterior carina precedes a groove which is frequently crenate (secondarily reduced in various Masarini).
19. P. brasiliensis, 2IX.cc: apical clypeal carinae; g: gap between mandibular teeth; po: postocular carina; pr: preoccipital carina."prominence" (Fig. 39).Richards (1973) confused the dorsal carina in Polistes, behind the fovea, with the anterior carina in other polistines, in front of the fovea.Eumeninae, which also have a fovea, also have a carina in front of the fovea (Fig. 9), which continues across the dorsum in the groundplan (Carpenter and  Cumming, 1985).The single carina may be a composite structure, derived from a state resembling certain polistines with two closely approximated carinae (Fig. 38).This is also the case in the ground- plan of Vespinae (Carpenter, 1987), where there is a single carina, preceding the pronotal fovea and running across the dorsum (Fig. 10).Stenogastrinae have a highly modified pronotum and lack a posterior carina and fovea (Carpenter, 1988), but have a blunt ridge anteriorly that may correspond to the anterior carina (Fig. 32).
Claws.Richards (1962:44) erroneously characterized the claws of Gayella as simple, and Carpenter (1981:26) initially followed this (corrected in Carpenter and Cumming, 1985:907).In fact, the claws are toothed in all species of Gayellini (variable in G. mutilloides).This is the plesiomorphic condition in Vespidae.
Male genitalia.I have examined the genitalia of all species except Paramasaris cupreus and Paragayella richardsg where the males are still unknown.In the groundplan of the tribe, the aedea- gus is broadly rounded apically, the digitus is a prominent triangular lobe that is desclerotized ventrally, the cuspis is a small lobe completely fused to the lamina volsellaris, and the parameral spines are long and sharply pointed (Figs.[56][57][58][59][60][61][62][63]. Figure 39 of Richards  (1962), showing a large, triangular cuspis and rounded digitus in Gayella araucana, is incorrectly labelled.What is there termed cus- pis is actually the digitus, and the structure labelled as digitus must be the aedeagus (cf. Fi. 58).This figure was the reason I previously was unable to characterize the groundplan of the volsella in the tribe (Carpenter, 1981:26), as I had not seen that species at the time.
Within genera, the genitalia are relatively uniform, with species dif- fering in details (especially of the volsella); however, there are some consistent differences letween the genera.These are discussed below.

Paramasaris
Giordani Soika (1974) characterized Paragayella as related to Gayella, and stated (my translation): "This genus appears at first sight a Gayella by the general aspect and dimensions."The type material I have seen he even labelled as "Gayella richardsi.In fact, Paragayella is not really even superficially similar to Gayella.Paragayella lacks some of the apomorphies shared by the species of Paramasaris, and for some other derived traits which Paramasaris and Paragayella share the latter has a less developed state.Thus it shares some primitive similarity with Gayella, which of course indi- cates nothing about phylogenetic relationship (Hennig, 1966).On the other hand, Paragayella shares several clear synapomorphies with Paramasaris.These include the forewing with r-m3 more or less straight (Figs.5-6; sinuous in Gayella and other Vespidae, Figs. 1, 4), the pronotum with two carinae (Figs. 29-31, 40; one in Gayella Fig. 7), the metanotum with a longitudinal median carina (Figs. 33, 46; none in other Masarinae), and the metasoma petiolate (tergum I in dorsal view at least twice as long as wide and half the width of tergum II, Figs. 35,40-41; it is differently shaped in Euparagiinae, Masarini and Gayella, Figs. 1, 43-45).Paragayella is the sister- group of Paramasaris.Autapomorphies of Paragayella include the reduced postocular carina (Fig. 20) and the transversely carinate metanotum (Fig. 46).
The three species of Paramasaris share numerous synapomorphies.The postocular and preoccipital carinae are apparently con- fluent (Fig. 19).These carinae are separated in other Gayellini, and the postocular carina reduced in several species (Paragayella, the Gayella mutilloides group).The mandibles are tridentate with the proximal teeth separated from the apical one by a gap (Fig. 13).The mandibles are quadridentate in females of Paragayella and Gayella (Figs. 11, 14-15), and tridentate in males of the latter (Fig. 17; Richards, 1962:44, erroneously characterized the mandibles of Paramasaris as quadridentate and those of Gayella as simple); there is no gap.Quadridentate mandibles is the groundplan state of most of the Vespidae (Carpenter, 1981), although Euparagiinae has bidentate mandibles.The glossa is shortened and lacks acroglossal buttons, the paraglossae are broadened, and the posterior lingual plate is cordate in shape (Fig. 21).The posterior lingual plate is slightly broadened in other gayellines, but the length of the structure still exceeds its width (Fig. 22).The clypeus is broadly truncate (Fig. 12), which is here treated as derived, convergent with the ground- plan of Masarini.Paragayella has the clypeus narrowly truncate (Fig. 11), as in Euparagiinae, which is considered the primitive state.A broad truncation seems most simply interpreted as derived from a narrow emargination, as does the pointed clypeus of Gayella (Figs. 14-17).The posterior carina of the pronotum extends further dor- sad in Paramasaris (Figs. 25-26) than Paragayella (Fig. 24), a further apomorphic development.The propodeum has oblique carinae more or less developed (Figs.31, 33), a unique trait in Psyche [Vol. 95  Figs.29-36.29-32.Lateral view of pronotum and mesepisternum.29.Paramasaris richardsi, 13X.30.P. brasiliensis, 14X.31.P. fuscipennis, 17X.
Lateral view of metasoma.36.Gayella reedi , 3. Lateral view.ap: anterior pronotal carina; at: anterior truncation of metasomal tergum I; cg: carina delimiting Masarinae.The first metasomal tergum has a blunt posterior ridge that is continued anterolaterally and drawn out into projections posterolaterally (Figs. 35,40), a feature unique in Vespidae.The tergum is also strongly truncate anteriorly (Fig. 35).The second tergum has a median longitudinal ridge (Figs.40, 42), which how- ever is variably developed in brasiliensis (strong in the male and not developed in the female, Fig. 35).A longitudinal ridge is found elsewhere in Vespidae only within Eumeninae (Cyphomenes, where it is anterior).Finally, the parameral spines of the male genitalia are extremely elongate in brasiliensis andfuscipennis, being longer than the parameres and extending far beyond the apex of the aedeagus (Fig. 56).This is apparently a derived condition; in Gayella, Euparagiinae and Masarini the spines are shorter than the parameres and extend little beyond the aedeagus (Figs.[57][58][59][60][61][62][63].Males of cupreus are predicted to share this synapomorphy, and possibly also Paragayella. Within Paramasaris, cupreus and fuscipennis are sister-groups.42).Autapomorphies of the species are: for cupreus the propodeal median groove delimited by more lamellate carinae (Fig. 34); forfuscipennis the oblique propodeal carina better defined (Fig. 33), and the dorsal groove and scrobal furrow of the mesepisternum broader and deeper (Fig. 31, cf.with 29-30, 40).I have not discovered any autapomorphies of brasiliensis.
Since Paragayella is the sister-group of Paramasaris, recognition of both genera is consistent with monophyly.However, it serves little useful purpose.Paragayella itself has few apomorphiesmit mostly lacks those of Paramasaris.Recognition of Paragayella thus contributes little to the process of efficient diagnosis.Since Paragayella is monotypic, and Paramasaris consists of but three propodeal median groove; dg: dorsal groove; lm: longitudinal metanotal carina; oc: oblique propodeal carinae; pp: posterior pronotal carina; sf: scrobal furrow; TI: metasomal tergum I; TII: metasomal tergum II; tc: posterolateral tergal projection.

Gayella
The monophyly of the genus is shown by the pointed clypeus (Figs.14-17; not similar to that of Stenogastrinae and Polistinae), the temples projecting somewhat and the emarginate and bispinose last metasomal tergum (Figs.[50][51][52].In other Masarinae and Euparagiinae the clypeus is truncate or emarginate, the temples do not project and the last visible metasomal tergum is neither emarginate nor spined.Several characters of the male genitalia are also synapomorphies.The digitus is enlarged relative to Paramasaris (cf.Figs. 5 and 57-63), the cuspis is tuberculate basally (Figs.57-61, 63), and the paramere has an enlarged lobe (dorsal to the spine, Fig. 62).The combination of features of the male genitalia is unique in Vespidae.Finally, the globose shape of the first metasomal tergum may be apomorphic, but this is variable within the genus (Figs.43-45).
Within the genus, two monophyletic species groups may be recognized, which allows a classification that is phyletieally se- quenced (Wiley, 1979).These are the eumenoides group, for eumenoides and araucana, and the mutilloides group, inlcuding reedi, patagonica, luispenai and mutilloides.
48. Gayella araucana, 21X.49.G. reedi,.15. 50--52.Dorsal view of metasomal in the mutilloides group (cf Figs. 57-58and 59-61, 63), and is here inferred as an elaboration.A sister-group relationship therefore obtains between eumenoides and araucana.The first has the autapomorphy of the humeri projecting above the anterior pronotal carina (Fig. 27).A very weak angle is also found in reedi, and Paramasaris has an angle of a different form (Fig. 24), but the projection is much stronger in eumenoides.Willink and Ajmat de Toledo (1979: fig. 3) depict eumenoides as having an apically bilobed aedeagus; however the shape varies among specimens in my dissections, and most have a broadly rounded apex as in other Gayella (Fig. 57).The sister-species of eumenoides, araucana, also has some autapomorphies.The acroglossal buttons are very reduced in size in the male, whereas they are elongate in the female and other Gayella.The pronotal punctation in araucana is relatively coarser than in the rest of the tribe, so this may also be a derived feature.and 28), but the difference from eumenoides is slight.The first metasomal tergum is narrower than in other species of the genus (cf Figs. 43and 44-45),   but this is approached in some specimens of eumenoides.The male genitalia has the cuspis with the basal tubercle sharply pointed (Fig. 58).The tubercle is usually less pointed in eumenoides (Fig. 57), but some specimens approach araucana.Willink and Ajmat de Toledo (1979: fig.5) depict a rather different digitus in araucana.However, their figures were evidently drawn from specimens flat- tened on slides, and do not accurately portray the relative uniform- ity in this structure (or the aedeagus) among the species (Figs.[57][58][59][60][61][62][63].

Mutilloides group
The most obvious feature supporting the monophyly of the mutil- loides group is the coat of elongate black hairs (Figs. 36, 45,53).

Psyche
[Vol. 95  derived features, however.The preoccipital carina is evanescent and separated from the postocular carina by more than an ocellar diameter in species of this group, whereas both are well developed and closely approximated in the eumenoides group and Euparagii- nae.The femur is punctate in the mutilloides group; it is smooth in other gayellines.The spines defining the emargination of the last visible metasomal tergum are narrow and elongate in the mutil- loides group (Fig. 50), whereas they are broader and shorter in the eumenoides group (Figs.51-52).Since both states uniquely charac- terize monophyletic groups, the polarity cannot be clearly inferred.
However, the state of the metasomal spines in the mutilloides group is a more extreme development, and is here suggested as relatively apomorphic.
Within the group, reedi is the sister-group to the remaining three species.I have not discovered any clear autapomorphies of this species.Synapomorphies uniting patagonica, mutilloides and luis- penai include greater development of the long black hairs on the metasoma (extending over the disc of tergum II, Fig. 53), and an elongate malar space (length ! 2 to greater than the width of the interantennal distance, Fig. 14; shorter than this in other gayellines, Figs.15-17).The postocular carina tends to be more effaced (as in Fig. 20), but traces appear to be present in some specimens.Meta- somal sternum II in the male and to some extent also the female is bordered posterolaterally with blunt ridges (Figs.53-54), however these are variably developed in patagonica, and reedi approaches this condition.Among these three species, the features I have polar- ized are autapomorphies.Hence, the relationships are at present unresolved (Fig. 2).Autapomorphies of the species are: for mutil- loides the very long malar space (Fig. 14), the acroglossal buttons more elongate and the glossa more deeply bifid than other Gayella, and the cuspis tubercle quite blunt (Fig. 61); for luispenai the male metasomal sternum II projections elongate (Fig. 54); and for pata- gonica the female propodeal median groove narrowed before broadening dorsally (Fig. 55; smoothly narrowed in other Gayella, Fig. 45).
In addition to the material in the British Museum, I have seen female specimens from the following localities in Goias in Brazil: "24 kil.E. Formoso, June 6, 1956 (F. S. Truxal)" UCD; "S.Isabel do Morro, Ilha do Bananal, June 1961 (M.Alvarenga)" MCZ.
Giordani Soika (1974) alluded to various differences in sculpture between cupreus and fuscipennis in his description, but several of these do not hold up in the additional material I have seen.The pronotal carina and tergal punctation are similar in most specimens, and the clypeus is not more narrowly emarginate in cupreus.The finer and sparser punctation on the dorsum of the mesosoma in cupreus is usually consistent, particularly the pronotum, but one of the Magdalena specimens has the punctures on the scutum and scutellum about as in fuscipennis.
Bequaert and Ruiz (1942) summarized early literature on this species.Reed (1893) pointed out that Spinola had confused the sexes in the original description, and Willink (1956) observed that Spinola and Saussure confused eumenoides and araucana.This is the only gayelline for which any behavioral information has been published, by Claude-Joseph (1930).It provisions clusters of free mud cells with nectar.
Willink and Ajmat de Toledo (1979) recognized the same species groups as the present paper, but stated that araucana is the most distinct and should possibly be treated as a different genus.They even stated that morphologically it approaches eumenines of the genera Ancistrocerus and Stenodynerus, but it does not possess any of the synapomorphies of that subfamily (Carpenter, 1981).The characters cited as distinguishing araucana are only trivially different from other Gayella.These characters comprise the mesosoma with lateral margins subparallel, the malar space nearly obliterated, the form of the emargination of male metasomal tergum VII and the form of the male genitalia.The condition of the malar space is primitive, and similar to eumenoides (cf.Figs. 15 and 16); the emar- gination of male tergum VII is little different from that of eume- noides (cf.Figs. 51and 52).As noted above, their figures of the male genitalia are misleading; araucana is no more different in the digitus or aedeagus than the other species are from each other (Figs.57-61,  63).The cuspis is autapomorphic in having a sharp tubercle, but this again is not properly illustrated in the other species, all of which have some projection.The subparallel mesosoma is also autapo- morphic, but this is a minor difference compared to the outstanding similarities shared by all species of Gayella.Placement of araucana in a separate genus would render Gayella paraphyletic, which is reason enough to reject doing so.
This species is very similar to patagonica, as noted by Willink and   Ajmat de Toledo (1979:430).Most of the characters they cite will not distinguish females.The series of patagonica I have seen from Rio Negro (MF) overlaps in size, length of the malar space and development of protuberances on female metasomal sternum II.The punctation tends to be less coarse on the scutellum of pata- gonica, but this varies among the specimens.The propodeal median groove will separate them (Figs.45, 55), but as I have seen only one female of luispenaL I cannot be certain that this feature does not vary.Males are readily distinguished by the sternal projections in luispenai (Fig. 54).

BIOGEOGRAPHY
It is clear from the few records for some species that their distribu- tions are very poorly known, and further collecting, particularly of Paramasaris, will doubtless extend the ranges of some of these.Nevertheless, a few remarks about biogeography may be made.
Paramasaris and Gayella occupy completely different regions, Tropical American versus Patagonia, which corresponds to a well-known vicariant break.Within Gayella, most of the species overlap broadly in distribution.The clade mutilloides + patagonica + luispenai is the only group which occurs on the eastern side of the Cordillera, but the first two species are also found in Chile.By contrast, Para- masaris shows a pattern of endemism.Within the genus, the distri- bution of the sister-species fuscipennis and cupreus is basically trans-Andean: Central America versus western Amazonia.There is a record of cupreus from the western side of the Sierra de Perijh, but this was elevated in the late Oligocene (Kellogg, 1984).In turn, the sister-group of this clade, brasiliensis, is southeastern Brazil, and the sister-group of all three species, richardsL is southern Amazon basin.This pattern does not correspond to that shown by the avi- fauna, for example Cracraft and Prum (1988), where southeastern Brazil is not closely related to a western Amazon/trans-Andean clade.However, that study showed southeastern Brazil as a compos- ite area, implying either dispersal or differing ages for components of the regional biota.The latter factor may well explain the incon- gruence; Gayellini is an ancient group, since the Masarinae as a whole is gondwanian (Carpenter, 1981).

(
BMNH British Museum of Natural History, London (M. C.