Back to the Trigonalidae Homepage
Morphological terminology follows that of Gauld and Bolton (1988), except that the terms mesosoma, metasoma, sternum, and tergum are used here. The term metasomal armature is used for the various modifications (projections, stout spines, or raised posterior edges) found on sterna II and/or III of some females. The awl is the sclerotised tip of the terminal sternum of the female that is shaped and functions like a hypodermic needle to pierce foliage for the oviposition of eggs (Yamane & Yamane, 1975, Fig. 16). In many cases it is held in the preceding sterna which are flattened into a capsule (Figs 1-3). The term supra-antennal elevation (SAE) is used for any elevation or modification above and mesad of the torulus (Tsuneki, 1991). The number of antennal 'segments' includes the scape, pedicel, and flagellomeres. Malar space is "short" if it is shorter than the width of the first flagellomere, "long" if it is longer. Only the metasomal segments are used for counting sterna and terga. Forewing (FW) length is measured from the break between the costal vein and the humeral plate to the wing tip. Body length is measured from the antennal insertion to the furthest point of the metasoma, and is not standardised due to changes in alignment of the head and metasomal segments. To see the angle of the gena in the horizontal plane, the termination of the genal carina, and whether the genal carina is mesad of or on the genal angle the head must be oriented so that it is viewed ventrally (Figs 22-23).
Clausen (1940), Cooper (1954), Carlson (1979), and Gelhaus (1987) used the term secondary host for the folivore while Gauld and Bolton (1988) and Weinstein & Austin (1991) used the term primary host. Because of this ambiguity, we use the terms intermediate host when a trigonalid needs to develop to adulthood in a second host, and ultimate host for the host that the larva uses to complete its development.
While trigonalids from key geographical regions, including Southeast Asia, Africa, and South and Central America remain undercollected, we have had about 2300 specimens from over 50 collections available for study. Most genera were represented by series of 40 or more specimens. Still, many species and some genera are represented by only one to a few specimens, often of one sex, and many other species originally described from one to a few specimens remain unknown to us. We have seen the types (holotypes, syntypes, or lectotypes) of 37 species, including eleven types previously listed as missing or with their deposition unknown (Weinstein & Austin, 1991). We have seen paratypes of an additional nine species. The types of many species are apparently lost (Weinstein & Austin, 1991), and the species descriptions are often inadequate to determine their identity.
The characters in the data matrix (Table 1) are identified by numbers assigned in the section on character analysis; these numbers serve to identify the characters in the cladograms (Figs 24-30). The terminal taxa used in the analyses are the hypothetical ancestor and the genera of Trigonalidae. So that results from the phylogenetic analysis would challenge our generic definitions and synonomisations, we also included species representing the previously recognised (and here synonymised) genera Labidogonalos, Nanogonalos, and Poecilogonalos Schulz, as well as taxa whose placement is tentative, including Taeniogonalos flavocincta (Teranishi) and Taeniogonalos maga (Teranishi). Due to the small number of specimens and in some cases their poor condition, it may not be possible to evaluate the relationships of Tsuneki's new genera and species until more specimens are collected (Tsuneki, 1991).
Parsimony analyses were carried out using MacClade 3.01 (Maddison & Maddison, 1992) and PAUP 3.0s (Swofford, 1991). MacClade was used for entering data and comparing different phylogenetic hypotheses (cladograms) and character evolution while PAUP was used for finding the most parsimonious trees and their statistics. PAUP's default heuristic search settings were used with more than fourteen taxa, and 100 random stepwise-addition replicates were used to search for additional parsimonious trees. With fewer than fourteen taxa, we used the branch and bound search, which will find all the most parsimonious trees.
Hennig86 ver. 1.5 (Farris, 1988) was also used for parsimony analyses. Polymorphic characters in the data matrix (Table 1) were converted to monomorphic characters using the ACCTRAN option in MacClade on the tree in Fig. 28. In the case of the propodeal scutellum of Xanthogonalos the polymorphism was not resolved, so the character state was changed to unknown, resulting in a tree one step shorter. This file was exported to Hennig86, and then analysed using the implicit enumeration option, which will find all the most parsimonious trees, and successive approximations character weighting (Farris, 1969). The Hennig86 file was imported back to PAUP, and reanalysed, giving the same results as the original PAUP input file with polymorphic characters.
In the species lists, no attempt is made to duplicate the catalogues of Bischoff (1938) and Weinstein & Austin (1991). Synonymies and bibliographic details listed by them are not repeated except for clarification. Type repository information is based on Weinstein & Austin (1991) and our correspondence with the collections. Repository and label information is given for specimens examined of less common species in Carmean (1993). In a few cases we transferred or synonymised species without seeing the types because the type species of the genus they were placed in (Poecilogonalos, Nanogonalos, and Discenea) was also transferred, or because, based on the description, they were obviously misplaced.
Collection Abbreviations
Specimens were obtained from the following collections and individuals, using standard abbreviations from Arnett et al. (1993) for institutions and the first four letters of the last name for personal collections: AEIC- American Entomological Institute, Gainesville, D. Wahl; AMNH- American Museum of Natural History, New York, E.L. Quinter; ANIC- Australian National Insect Collection, Canberra, I. Naumann; ANSP- The Academy of Natural Sciences of Philadelphia, D. Azuma; BMNH- Natural History Museum, London, L. Ficken, T. Huddleston, I. Gauld, and M.C. Day; BPBM- Bishop Museum, Honolulu, K. Arakaki; CARM- D. Carmean personal collection; CASC- California Academy of Sciences, San Francisco, W.J. Pulawski; CDAE- California Department of Food and Agriculture, M. Wasbauer; CMNH- The Carnegie Museum of Natural History, J. Rawlins; CNCI- Canadian National Collection, Ottawa, L. Masner and J. Huber; CUIC- Cornell University, Ithaca, J.K. Liebherr; DENH- University of New Hampshire, D.S. Chandler; EMUS- Utah State University, T. Griswold and F.D. Parker; FSAG- Collections Zoologiques, Gembloux, J. Leclercq; FSCA- Florida State Collection of Arthropods, Gainesville, J. Wiley; HNHM- Hungarian National Museum of History, Budapest, J. Papp; IMLA- Fundacion Miquel Lillo, San Miguel de Tucuman, A. Willink; INBIO- Instituto Nacional de Biodiversidad, Costa Rica, D. Janzen; INHS- Illinois Natural History Survey, Champaign, K.C. McGiffen; ISNB- Institut Royal des Sciences Naturelles de Belgique, Brussels, P. Dessart; IZAV- Universidad Central de Venezuela, Maracay, Venezuela, J.L. Garcia R.; KIMS- L. Kimsey personal collection, Davis; LACM- Natural History Museum of Los Angeles County, Los Angeles, R.R. Snelling; LEMQ- Lyman Entomological Museum, McGill University, Quebec, P.M. Sanborne; LSUC- Louisiana State University, V. Mosely and C.B. Barr; MACN- Museo Argentino de Ciencias Naturales 'Bernardino Rivadavia', Buenos Aires, A. Bachmann; MAMU- Macleay Museum, Sydney, Australia, D.S. Horning, Jr.; MCZC- Harvard Museum of Comparative Zoology, Cambridge, Massachusetts, D. Furth; MEMU- Mississippi Entomological Museum, Mississippi State University, T.L. Schiefer; MLPA- Universidad Nacional de La Plata, Ricardo A. Ronderos; MNHN- Museum National d'Histoire Naturelle, Paris, J.C. Weulersse; MRAC- Musee Royal de l'Afrique Centrale, Tervuren, E. De Coninck; MRSN- Spinola Collection, Museo Regionale di Scienze Naturali, Torino; MZSP- Museu de Zoologia da Universidade de São Paulo, C.R.F. Brandão; NCSU- North Carolina State University, Raleigh, R.L. Blinn; NHMW- Naturhistorisches Museum, Vienna, M. Fisher; NHRS- Naturhistoriska Riksmuseet, Stockholm, B. Gustafsson; OMNH- Osaka Museum of Natural History; OSUO- Oregon State University, Corvallis, G. Ferguson and J.A. DiGiulio; OXUM- Oxford Museum (Hope Entomological Collections), Oxford, C. O'Toole; PAGL- G. Pagliano personal collection, Torino; PORT- C.C. Porter personal collection, Gainesville; PSUC- Frost Entomological Museum, Pennsylvania State University, D.W. Love; RMNH- Rijksmuseum van Naatuurlijke Historie, Leiden, C. van Achterberg; ROME- Royal Ontario Museum, Toronto, C. Darling; SCAR- L. Scaramozzino personal collection, Torino; TAMU- Texas A&M University, E.G. Riley; TARI, Taiwan Agricultural Research Institute, L.Y. Chou; TMSA- Transvaal Museum, Pretoria, K.N. Dower; UCDC- Bohart Museum of Entomology, University of California, Davis, S. Heydon; UCRC- University of California, Riverside, S. Frommer; UOPJ- University of Osaka, Hirowatari; UMMZ- University of Michigan, Ann Arbor, M. O'Brien and B.M. OConnor; USNM- United States National Museum, Washington D.C., D.R. Smith and G.F. Hevel; WSUC- Washington State University, Pullman, R.S. Zack; YAMA- Sk. Yamane personal collection, Kagoshima; ZMHB- Zoological Museum of Humboldt University, Berlin, F. Koch; ZMUC- Universitetets Zoologiske Museum, Copenhagen, B. Petersen; ZSMC- Zoologische Staatssammlung, Munich, E. Diller. Abbreviations for other institutions mentioned in the text are: EIHU- Hokkaido University, Sapporo; MCSN- Museo Civico di Storia Naturale, Genoa; MLUH- Universität Halle, Halle.
Outgroup analysis
Trigonalidae have been placed in or near most other major lineages of apocritan hymenopterans at one time or another. Recently, Whitfield (1992), Rasnitsyn (1988), and Johnson (1988) placed the Trigonalidae in the Evaniomorpha, along with the Evanioidea, Ceraphronoidea, and the Megalyridae. Rasnitsyn (1988) considered the Stephanidae also to be evaniomorphs but Whitfield (1992) placed the Stephanidae basal to all other Apocrita. Whitfield (1992) also considered a second possibility, with the Trigonalidae part of an unresolved trichotomy between Trigonalidae, Evaniomorpha, and the 'Microhymenoptera'. Dowton and Austin (1994), using DNA sequence from the mitochondrial 16S rRNA gene, found the Trigonalidae close to the base of the Apocrita along with the Evanioidea, but otherwise unresolved. Basibuyuk & Quicke (1995) found the structure of the basitarsal comb in the Trigonalidae and Evaniidae, but not other proposed Evaniomorpha, to be almost identical. Like the other families in the Evaniomorpha, trigonalids have the derived character of an elongated mesal lobe on the surface of the mesocoxa (Johnson, 1988).
The Trigonalidae are so divergent morphologically from all the other Hymenoptera that outgroup analysis for most characters is speculative. Of the 30 characters used in this study, 23 can be found in the outgroups and contribute to outgroup analysis, including 2 of the 20 morphological characters that Shaw (1990) used in his analysis of the Megalyridae. The ancestral states of these later two characters are areolate-rugose propodeal sculpturing and elongate flagellomeres. Elongate body-form is the ancestral state for megalyrids (Shaw, 1990) and appears ancestral in trigonalids, but because of intermediate states was not used in our analysis. Comparing trigonalids with other Evaniomorpha, the Ceraphronoidea have metasomal tergum I longer than all the following terga, the antennae close to the mandibles, reduced wing venation, a long malar space, and the genal carina meeting the hypostomal carina. The Evaniidae have a long malar space, with the genal carina just mesad of the relatively sharp genal angle, and ending at the mandibular base. Aulacidae have a long malar space, genal carina usually meeting the hypostomal carina (in some taxa ending at the mandibular base instead), and a propodeal foramen that is evenly curved or 'U'-shaped dorsally. Stephanidae have a long malar space, genal carina ending near the hypostomal carina, and a 'U'-shaped propodeal foramen (which is otherwise very different from the form in Trigonalidae). No trigonalids have reduced wing venation or an elongate first metasomal segment. In addition, trigonalids have the antenna far removed from the mandible, except the derived Pseudonomadina and Bakeronymus. We conclude that the ancestral trigonalid, as in most evaniomorphs, had the genal carina ending near the hypostomal carina, a 'U'-shaped propodeal foramen, and a long malar space, although this feature is too ambiguous in most taxa to be included in the analysis. The ancestor probably lacked such advanced trigonalid characters as male tyloids, female armature, and sclerotisation in the last female abdominal sternum.
We used two outgroups separately: one is based on the presumed primitive Evaniomorpha and is here called the Evaniomorph while the other is based on the hypothetical ancestral trigonalid groundplan and is called the Ancestor. The Evaniomorph is based on the families Stephanidae, Aulacidae, Evaniidae, Gasteruptiidae, and Megalyridae. Characters for Megalyridae were taken from the literature (Naumann, 1991; Shaw, 1990). The Ancestor is similar but is influenced by Orthogonalys and has the family synapomorphies of asymmetrical mandibles and trochantellus divided; the character state for the SAE is unknown in the ancestor while in the evaniomorph outgroup it is flattened. The number of antennal segments is greater than 17 in the ancestral trigonalid and less than 17 in the evaniomorph. The shape of the propodeal foramen is a low 'U' shape in the ancestor and unknown in the evaniomorph. The remaining characters have the same state in the ancestor and the evaniomorph. The ancestor and the outgroup were used separately in the parsimony analyses, and the exact same results were obtained from each, with Orthogonalys basal to the rest of the Trigonalidae (Fig. 24).
Character Analysis
Characteristics used in this analysis are discussed below. Their
inferred primitive and derived states are discussed below and
in the previous section on outgroup analysis. When the evaniomorph
or ancestral outgroup state could not be deduced, it has been
coded as unknown (?) because it contributes no information about
polarity of character state change. Characters were treated as
unordered and reversible, and not weighted except in one analysis
using a posteriori successive approximation
character weighting (Farris, 1969). Inferences about polarity,
based upon outgroup comparison, are given below; but ultimately
polarity was determined after the cladistic analysis as a consequence
of outgroup rooting of the trees. The data matrix is presented
in Table 1.
1. Head shape. (0) Rounded, normal (Figs 14-17; Tsuneki, 1991 Figs 49, 156); (1) subrectangular and wide (in anterior view) (Figs 18, 20; Yamane & Yamane, 1975 Fig. 5; Yamane & Kojima 1982 Figs 1a, 2a; Tsuneki, 1991 Figs 2, 29). Heads of evaniomorphs are generally rounded and tall. Bareogonalos, Nomadina, Pseudonomadina, and Bakeronymus have the derived state.
2. Vertex shape. (0) Normal, convex, or flat (Figs 14-18, 20); (1) concave. Only two Asian genera, Bakeronymus and Pseudonomadina, have their head deeply indented along the dorsal sagittal plane, and this is apparently the derived condition.
3. Supra-antennal elevation (SAE). (1) Prominent, meeting or nearly meeting at midline (Fig. 14); (2) prominent, not meeting at midline; (3) reduced to small triangular protuberance and extending toward midline (Fig. 17); (4) flattened, not extending toward midline, flat between toruli (Figs 15, 18). This character was used extensively by Schulz (1907a) and Tsuneki (1991). In all other evaniomorphs there is no SAE: the area above and mesad of the torulus is flat, or the torulus is on a shelf. The ancestral condition is unknown, and the evaniomorph outgroup is coded as SAE flattened with the area between the toruli flat. In Orthogonalys the SAE is generally prominent, projecting forward, and slightly separated at the midline. In Taeniogonalos the SAE is reduced so that the dorsal edge of the torulus forms a small triangular lip, and the intertorulus area is relatively flattened. In the Nomadinini the SAE is reduced and the intertorulus area is generally very flattened.
4. Intertorulus Distance. (0) Short, distance between toruli less than 0.9 times the shortest distance between the inner eye margin and the torulus; (1) medium, the two distances about equal; (2) long, toruli set far apart, shortest distance between inner margins of the toruli greater than the shortest distance between the inner eye margin and the outer edge of the torulus. In most evaniomorphs, and Orthogonalys, the intertorulus distance is short; this is assumed to be the ancestral condition. The distance is long in the Stephanidae.
5. Toruli placement. (0) Distant from mandibular base; (1) next to mandibular base. The toruli are above the clypeus, far removed from the mandibular base in most evaniomorphs and in most Trigonalidae. The derived condition is only found in Bakeronymus and Pseudonomadina.
6. Number of antennal segments. (0) Greater than 17 segments (including scape and pedicel in count); (1) 13-16 segments. There is a great amount of variability in antennal segment number but no overlap between these states. Most Trigonalidae have greater than 17 segments and this is considered the ancestral condition. Only Nomadina, Bakeronymus, and Pseudonomadina have 13-16 antennal segments and, as they are otherwise relatively derived genera, this is assumed to be the derived condition. However, the evaniomorph outgroup taxon is coded as having 13-16 segments since most evaniomorphs, except the Stephanidae, have their antennae with 13-14 segments or less.
7. Antennal shape. (0) Filiform; (1) thickened or spindleform. The evaniomorphs and most Trigonalidae have filiform antennae of even thickness, though the Stephanidae have much thinner flagellomeres. The derived state occurs in Seminota, Lycogaster, Nomadina, and related genera. A single undescribed male from Costa Rica with tyloids has spindleform antennae (AEIC), but in other respects it is close to Taeniogonalos, and its antennal shape is considered convergent.
8. Tyloids. (0) Absent; (1) present. Tyloids (Figs 10-12) are not present in the evaniomorphs or Orthogonalys and their presence is considered derived.
9. Tyloid shape. (1) Short, oval-round (Fig. 11); (2) elongate-broadly oval (Fig. 12); (3) elongate-narrow (Fig. 10). The shape of the tyloids has been generally ignored by previous authors but is valuable phylogenetically. The states are not ordered. To avoid giving additional weight to the absence of tyloids, this character is coded as unknown for taxa that lack tyloids, including the ancestor. We separated the character of tyloid presence from tyloid shape because we believe that change between different shapes is a different process than the gain or loss of tyloids. However, the topology of the strict and majority rule consensus trees was unchanged when these characters were combined (at the same time characters 27 and 28 were similarly combined) and when the ancestral state was 'tyloids absent', and the three tyloid shapes were treated as unordered and derived.
10. Genal carina. (0) Meets (or ends near) hypostomal carina (Fig. 9; Tsuneki, 1991, Fig. 64); (1) meets (or ends near) lateral edge of mandibular base (Figs 22, 23). The primitive state occurs in Orthogonalys and most evaniomorphs.
11. Occiput excavation. (0) Occiput not excavated (Fig. 9; Tsuneki, 1991, Fig. 64); (1) occiput slightly excavated (Fig. 23); (2) occiput deeply excavated but not near mandible; (3) occiput deeply excavated all the way to mandible (Fig. 22). In the ancestral condition, shared by Orthogonalys and the evaniomorphs, the genal carina is on a flat plane, and the occiput is not excavated. In Pseudogonalos the occiput is deeply excavated, but this ends in a flat plane before the mandible. The occiput of Trigonalys is deeply excavated all the way to the mandible. In Taeniogonalos the occiput is only slightly excavated.
12. Genal angle. (0) Located laterad of genal carina; (1) at genal carina. In most Trigonalidae and evaniomorphs the genal angle is located laterad of the genal carina (Figs 9, 23). The genal angle and genal carina only overlap in Trigonalys (Fig. 22); this is the derived state.
13. Clypeal width. (0) Wider than base of antennae; (1) as wide as base of antennae. The derived state is only found in the most specialised of Trigonalidae: Nomadina, Bakeronymus, and Pseudonomadina. In the ancestral condition and the normal apocritan condition, the clypeus is wider than the distance between the base of the antennae.
14. Mandible symmetry. (0) Asymmetrical; (1) symmetrical. In most Trigonalidae there are three teeth on the left mandible and four on the right mandible (Fig. 19), rarely, there are four on the left and five on the right. The normal apocritan condition is symmetrical, and in several taxa within the Nomadinini the mandibular teeth are symmetrical. No other hymenopteran families are known to have asymmetrical mandibles as their groundplan. The groundplan synapomorphy for the family is believed to be asymmetrical and the evaniomorph outgroup condition is symmetrical.
15. Maxillary palps. (0) Normal, 6-segmented, as long or longer than mandibles; (1) 4-segmented, shortened or rudimentary. In most taxa the palps are much longer than the mandibles and this is considered the ancestral condition for Trigonalidae. In Bakeronymus, Nomadina and Pseudonomadina the palps are usually 4-segmented, but in the latter two genera they are rudimentary and may be indistinctly segmented.
16. Notauli of mesoscutum. (0) Straight; (1) parallel at base and then diverging strongly. Only Bakeronymus and Pseudonomadina have curved notauli, making the median mesoscutal area nearly heart-shaped (Yamane & Kojima, 1982, Fig. 9). The evaniomorphs and most trigonalids have relatively straight notauli.
17. Submarginal cell II. (1) Petiolate; (2) not petiolate. This character has been overemphasised in the past, and is somewhat variable within genera, but it is still phylogenetically informative. In most evaniomorphs venation is not comparable, except in the Aulacidae where the submarginal cell II is petiolate. Thus the ancestral and outgroup states are unknown. In Trigonalys and Taeniogonalos this character is variable, and was coded as unknown.
18. Hind trochantellus. (0) Divided; (1) undivided. Most Trigonalidae have the trochantellus diagonally divided into two apparent segments and this is believed to be the ancestral condition for the Trigonalidae. The evaniomorph outgroup was coded as undivided because they have the trochantellus undivided as do the derived genera Bareogonalos, Nomadina, Bakeronymus, and Pseudonomadina.
19. Propodeal sculpturing. (0) Areolate-rugose; (1) rugose; (2) punctate; (3) smooth (4) areolate. Shaw (1990), using Ceraphronoidea, Evanioidea and Trigonalidae as outgroups, found that the primitive state for megalyrids is areolate-rugose, as is found in Orthogonalys. Most trigonalids are punctate but some are rugose or very smooth. Bareogonalos is strongly areolate.
20. Propodeal foramen. (0) Low 'U' shape (wider than high); (1) high 'U' shape (at least as high as wide); (2) 'V' shape (acute angle at apex). In most evaniomorphs the foramen is 'U' shaped, but it is also closed ventrally, while in all the Trigonalidae it is open ventrally, so comparison with these taxa may not be useful. However, within the Trigonalidae the propodeal foramen of Orthogonalys is a low 'U' shape, and there appears to be a transition from a low 'U'-shaped to 'V'-shaped foramen. The evaniomorph outgroup was coded as unknown and the ancestral trigonalid outgroup was coded as a low 'U' shape.
21. Propodeal foramen carina. (1) Thick and double-walled; (2) narrow and single walled. Several taxa, including Orthogonalys, Pseudogonalos, Bareogonalos, and an undescribed genus from Papua New Guinea have a thick double-walled carina while most taxa have a thin carina. The carina in the evaniomorphs does not appear analogous and the ancestral state is considered unknown.
22. Tergal plate thickness. (0) Thin, with transparent to translucent edges folding over sterna; (1) not thin, and meeting sterna laterally with little overlap. Several evaniomorphs and Orthogonalys have very thin metasomal plates. Their terga overlap the sterna ventrally, and are distinctly transparent not just at the very margin but over a wide area. Bareogonalos, Nomadina, and Pseudonomadina also have the ancestral condition of thin terga, apparently secondarily.
23. Metasomal sternum II (male). (0) Rounded medially; (1) flattened or concave medially. Most trigonalids and evaniomorphs have the metasoma rounded ventrally. This character has been used in the past to separate genera which are synonymised herein under Taeniogonalos, but are included separately in the cladistic analysis.
24. Metasomal segment lengths. (0) Segments II & III subequal in length; (1) segment II slightly longer; (2) segment II approximately long as all following combined. This character was determined using male specimens when possible but generally applies to both sexes. It is difficult to determine in some specimens of taxa (especially Orthogonalys) with thin terga that distort during drying. Many evaniomorphs have an elongated first segment, which is unknown among the Trigonalidae, but most have segments II and III the same length, so the evaniomorph outgroup and ancestor were coded as segments II & III subequal in length.
25. Female awl. (0) No awl; (1) awl present. The awl (Yamane & Yamane, 1975, Fig. 16) is apparently unique to the Trigonalidae, and seems to have evolved after tyloids originated.
26. Female capsule. (0) No capsule present; (1) capsule present. In the derived state the terminal and penultimate sterna are flattened and often lyre-shaped, forming a capsule that positions the apical sternum to point anteriorly. No evaniomorphs have a capsule and the ancestral condition is assumed to be without a capsule. As this character may be linked to armature presence it was experimentally deleted without affecting the outcome of the analysis.
27. Female armature. (0) Absent; (1) female armature present in some members. Because of the complexity of this character it is most parsimonious to assume that armature only originated once, and that the groundplan for taxa with more than one state is presence of armature. Armature is variable, i.e. present or absent, in two genera, Taeniogonalos and Trigonalys. In the data matrix, Taeniogonalos is divided into several representative species, for each of which armature is not variable. Experimentally coding Trigonalys as "armature absent" did not affect the topology of the strict or majority rule consensus trees. Armature is absent in all evaniomorphs. Those trigonalid taxa assumed to have secondarily lost their armature have their metasoma more strongly sclerotised ventrally, and often have the second sternum ventrally swollen or expanded.
28. Sternal armature location. (1) Present on sternum II; (2) present on sternum III. If the armature is present on sternum III there are usually traces of armature on sternum II, except in Trigonalys. To avoid giving the absence of armature additional weight, this character is coded as unknown for taxa that always lack armature, including the ancestor and outgroup taxon. In experiments where character 27 was combined with this character (similarly and simultaneously with characters 8 and 9 being combined) and the ancestral state was 'armature absent' and the unordered derived states were 'armature present on sternum II' and 'armature present on sternum III' the topology of the strict and majority rule consensus trees was unchanged.
29. Paramere. (1) Rounded (about as long as wide); (2) elongate (longer than wide); (3) angulate. The parameres are either relatively short and wide, as is the case in Orthogonalys and several Nomadinini; narrow and elongate, as in Pseudogonalos, Trigonalys, and Taeniogonalos; or sharply angled near the base, as in many Nomadinini. Male genitalia are figured by Tsuneki (1991). The parameres are fused to the basiparamere in Aulacidae, Evaniidae, Gasteruptiidae, and Stephanidae, and though in these taxa the parameres appear about as long as wide, they are not considered comparable. Thus the ancestral and outgroup states are unknown.
30. Aedeagus. (1) Cylindrical, not bilaterally flattened, apex capitate; (2) elongate- thin; (3) strongly bilaterally flattened with the tip expanded, 'T' or plough-shaped. In Orthogonalys, the aedeagus is rod-shaped and not bilaterally flattened, and the tip is capitate or slightly expanded into a bulb. In Trigonalys and Taeniogonalos, the aedeagus is elongate-thin, and slightly bilaterally flattened with the tip variously shaped but not capitate. In most Nomadinini it is strongly bilaterally flattened and the tip shaped like a plough or length-wise 'T' shape. In the evaniomorphs, the aedeagus of Gasteruptiidae, Aulacidae, and Evaniidae is cylindrical but not capitate; in Stephanidae it is apically slightly flattened but the shaft is cylindrical. The aedeagus of the ancestor may be assumed to be a simple cylindrical shaft, closest to the condition in Orthogonalys, but without further evidence we are considering the ancestral and outgroup states unknown.
Phylogenetic Analysis
The amount of homoplasy in the Trigonalidae is a challenge to any method of analysis. Some characters that initially appear monomorphic within a taxon, such as the length of the malar space or the presence of a petiolate second submarginal cell, become increasingly variable as more specimens and taxa are studied.
There are at least two phylogenetically significant, structurally complex features that have arisen within the Trigonalidae: tyloids and metasomal armature. Since some species have armature but do not have tyloids and some species have tyloids but do not have armature, and other species have both, it is necessary to resolve which taxa have secondarily lost or convergently gained these characters. Despite the variety of forms of armature and tyloids, we believe that they are too complex to have arisen more than once. Most taxa that fall within the Trigonalini have either armature, traces of armature, or a swollen sternum II, which may be a remnant of armature. Based on this evidence we assume that when both states are present within a taxon, the groundplan for that taxon is presence of armature. In the data matrix this assumption only applies to Trigonalys, and experimentally changing the coding for Trigonalys does not change the results. It is also more parsimonious to assume that the female capsule arose only once and in tandem with the armature. The capsule serves as a guide to point the ovipositor anteriorly, and the armature serves as a brace for oviposition into a leaf (Carmean, 1988, 1991). Experimentally deleting this character also did not change the results.
The results of the phylogenetic analyses are shown in Figs 24-30. Exactly the same 32 trees resulted from using the trigonalid groundplan (trigonalid ancestor), the evaniomorph outgroup, or only the trigonalid taxa. Using the hypothetical ancestor in the analysis resulted in a tree 72 steps long, C. I. 0.597; excluding the single uninformative character, C. I. 0.592. Using the hypothetical outgroup taxon resulted in 4 additional steps (76 steps), reflecting four synapomorphies for the family (Characters 3, 6, 14, and 18), with a consistency index of 0.566 (excluding uninformative characters, 0.560). The large number of trees are partially a result of including several species of Taeniogonalos that were previously separated into different genera and are not strongly differentiated. Including only T. gundlachii from Taeniogonalos resulted in only 17 trees, 71 steps (using ancestral taxon for rooting).
Table 2 compares the classification from Weinstein & Austin (1991), which is primarily based on Schulz (1907a), Bischoff (1938), and Benoit (1952), with the classification proposed in this study, which is based on the results of a cladistic analysis (Figs 24-28). Several taxa within the Trigonalinae remain with their status uncertain. They do not share the defining characters of the two tribes nor do they have any characters unambiguously uniting them or clarifying their relationships with other Trigonalinae. Use of scanning-electron micrographs of the tyloids or DNA sequence data may help establish clades including these taxa.
In the consensus trees (Figs 24, 25), the Trigonalini are paraphyletic with the Nomadinini. This may be in part because several species of Taeniogonalos secondarily lack armature. Further study is required to ascertain if Taeniogonalos and Trigonalys together form a monophyletic group, and what their relationships are to other taxa. With a reduced number of taxa, the bootstrap consensus tree (Fig. 26) generally agrees with the proposed phylogeny, and the Trigonalini are not paraphyletic.
Successive approximations character weighting (Farris, 1969) provides an objective method of a posteriori character weighting when confronted with several equally parsimonious cladograms. Applied to the characters in this study, the successive approximations character weighting reduces the number of equally most parsimonious cladograms from 32 to 12 and increases the consistency index from 0.59 to 0.71 (Fig. 25). Table 3 shows the final weights assigned each character by this procedure. One major difference between the unweighted and weighted consensus trees is at the base: Teranishia joins with Pseudogonalos in the unweighted tree (Fig. 24) but with an undescribed genus from Japan in the weighted tree (Fig. 25).
One alternative to Orthogonalys being the most primitive of the trigonalids is that Bareogonalos, Nomadina, Pseudonomadina and Bakeronymus are the most primitive Trigonalidae. Bareogonalos, Nomadina, Pseudonomadina and Bakeronymus share three generalised apocritan traits, flat SAE, symmetrical mandibles, and undivided hind trochantellus, which are absent from most trigonalids. These four genera are, however, highly derived in other respects, and their 'primitive' character states may have arisen secondarily. If a genus in the Bareogonalos-Nomadina group is assigned to a basal position in the Trigonalidae (Figs 29, 30), then unique trigonalid characters including the awl, sclerotisation of the capsule, and female armature would have to be interpreted as primitive familial characters lost in various lineages. Rerooting the most parsimonious tree so that Bareogonalos is basal gives the unlikely result of the evaniomorph outgroup nesting well within the ingroup (Fig. 29). Alternatively, experimentally constraining the Nomadinini to be basal (= the sister group) to all other Trigonalidae required three more steps than the most parsimonious tree (Fig. 30).
The tribe Nomadinini now includes several taxa previously treated
as separate subfamilies. While the Nomadininae under Schulz's
(1907a) and Weinstein & Austin's (1991) classification are
a monophyletic group, the other taxa in their classification are
either polyphyletic (Lycogastrinae), too narrowly defined (Bareogonaloinae),
or at best, paraphyletic (Seminotinae). Removing the unrelated
taxa previously placed in the Lycogastrinae and treating these
taxa together as a single tribe eliminates these problems.
This page is maintained by Dave Carmean with an eye towards speed and clarity, and last modified 15 April 1997. Comments or suggestions are welcomed!