INTRODUCTION

The hover wasps, so called for their remarkable hovering flight (Carpenter, 1988), form the subfamily Stenogastrinae of the family Vespidae. These wasps are endemic to the Oriental Region, ranging from India to New Guinea. The group is taxonomically not well understood: 45 species in six genera are formally recognized (Carpenter and Kojima, 1997), but numerous undescribed species were provided with manuscript names by the late J. van der Vecht. Van der Vecht was revising the entire subfamily before his death, and described five genera (van der Vecht, 1969, 1972, 1977), but published only four generic revisions (van der Vecht, 1972, 1975, 1977). One of Van der Vecht's genera has since been synonymized (Carpenter, 1988), and although a few species descriptions have been published by other workers (Reyes, 1988; Turillazzi, 1988; Turillazzi and Carfì, 1996), the largest genera (viz., Eustenogaster, Liostenogaster, Parischogaster) lack published revision, but work is in progress on Liostenogaster, by Turillazzi, and Parischnogaster, by CKS.

The species of Stenogastrinae that have been studied are social wasps, but are primitively social, and present some notable differences in their biology from other social wasps (reviewed by Turillazzi, 1991). Cladistic analysis of the relationships among subfamilies of the Vespidae supported a sister-group relationship between Stenogastrinae and the other social wasps, Polistinae and Vespinae (Carpenter, 1981), a conclusion that was upheld in consideration of basal vespid fossils (Carpenter and Rasnitsyn, 1990) and more detailed consideration of behavior (Carpenter, 1988). That conclusion has recently been challenged by Schmitz and Moritz (1998), who published a molecular data set and analysis that supported closer relationship of their eumenine sample to the polistines + vespines than the stenogastrines, but remarkably showed Vespidae as paraphyletic in terms of Apidae. This latter result is certainly not credible: Vespidae and Apidae are in different superfamilies, and a large body of evidence indicates that they are well separated (see Brothers and Carpenter, 1993). The morphological evidence supporting a sister group relationship between Stenogastrinae and Polistinae + Vespinae is not strong, as pointed out by Carpenter (1988), but the evidence that Vespidae are monophyletic is substantial (Carpenter, 1981; Brothers and Carpenter, 1993), as is the behavioral evidence grouping stenogastrines with other social wasps (Carpenter, 1988). Lacking two of the six extant subfamilies of Vespidae, Schmitz and Moritz's conclusion differs from Carpenter (1981) only in the placement of the root (see Nixon and Carpenter, 1993, for basic concepts), and their placement is clearly spurious, presumably an artifact of low taxonomic sampling, as is all too common in current molecular analyses. Combination of Schmitz and Moritz's molecular data with morphological characters from Carpenter (1981, 1987, 1988, 1991) and Brothers and Carpenter (1993) supports vespid monophyly, and combination with the behavioral characters from Carpenter (1988) and others supports closer relationship of Stenogastrinae to Polistinae + Vespinae than Eumeninae. We will not present details of that exercise here; instead JMC is pursuing a more comprehensive analysis of molecular data within Vespidae, which will re-examine subfamily relationships with a better taxon sample.

In 1997, the Center for Biodiversity and Conservation at the American Museum of Natural History initiated a collaborative project with the Institute of Ecology and Biological Resources, Ha Noi, Viet Nam. The project aims to develop scientific knowledge and expertise to protect biotically rich and endangered habitats in Viet Nam. As part of this effort, biotic inventories are being conducted in several areas, and in April, 1998, JMC participated in surveys of sites in the northern Truong Son Mountain Range. One of these sites is a mountain pass known as Cha Lo, in the Minh Hoa District of Quang Binh Province. The pass is close to the Lao frontier, and the road through it was part of the old Ho Chi Minh Trail. In 1998, the road was being improved, and JMC spent four days in a preliminary survey of the forest systems near the pass. A riparian area near the road, at the Cha Lo Frontier Post, was undergoing rapid degradation due to the construction. The vegetation along the streams was disturbed, secondary growth, but further away from the road old secondary and some primary forest was intact. It consisted of evergreen seasonal broadleaf hill forest, dominated by Lithocarpus and Mangletia. Along one of the streams of the watershed, collecting with an aerial net produced two species of hover wasps, patrolling along the stream. One of these species presents some remarkable features, justifying recognition of a new genus. This species was hitherto undescribed, but more than a decade ago CKS recognized it as new based on two specimens from northern Thailand. In this paper, we describe this new genus, and analyze its phylogenetic relationships to the other hover wasps.

The new genus is clearly distinguished from other stenogastrines, as the following key will show. A cladistic analysis is presented after the description, to address the placement of this genus within Stenogastrinae.

CLADISTIC ANALYSIS

The phylogenetic relationships among the genera of Stenogastrinae proposed by Carpenter (1988) are shown in figure 9. That study considered 24 characters of adult morphology, two of larvae and 14 of behavior. The cladogram of figure 9 resulted from analysis of 10 morphological characters that varied among the genera. Those ten characters are scored for Chalogaster in table 1. In addition, three other characters are included in table 1. These were characters discussed by Carpenter (1988) but were considered autapomorphies, thus uninformative on relationships among genera. The discovery of Chalogaster changes that interpretation. The outgroup is the subfamily Eumeninae, relatively basal to the Stenogastrinae in the cladogram of Carpenter (1981).

Analysis of the data in table 1 with the exact routine of the program Nona (Goloboff, 1998a) results in two cladograms (figs. 10–11). The length is 27 steps, the consistency index is 0.81, and the retention index is 0.66. These same cladograms result from implied weighting using the program Piwe (Goloboff, 1998b; default concavity).

The consensus tree for the two cladograms resulting from the analysis is given in figure 12. The length for the consensus is 28 steps, just one more step than either cladogram. The clade concordance index (Nixon and Carpenter, 1996) is zero, indicating complete disagreement between the characters supporting the differing resolutions of figures 10–11. Those characters are #10, metasomal segment II petiolate or not, and #9, propodeal sculpture. In the cladogram of figure 10, a petiolate metasomal segment II is interpreted as a synapomorphy for the clade composed of Anischogaster + (Parischnogaster + (Metischnogaster + Chalogaster)) - but with a subsequent reversal in Chalogaster. In the cladogram of figure 11, the propodeal sculpture is interpreted as transforming from punctate to striate in the clade composed of (Stenogaster + Eustenogaster) + (Parischnogaster + (Metischnogaster + Chalogaster))—but then changing to smooth separately in Eustenogaster and Chalogaster, and moreover reversing to punctate within Parischnogaster, which is polymorphic for this character (as for that matter are Liostenogaster and the outgroup Eumeninae). Both of the characters are homoplastic, but in particular we regard the propodeal sculpture as unreliable, because of the ambiguity due to polymorphism. If the polymorphisms in table 1 are replaced with ambiguity codes, as is necessary for analysis with programs such as Hennig86 (Farris, 1988), only the cladogram of figure 10 results. But although the character supporting this cladogram was perfectly consistent in the earlier analysis by Carpenter (1988), that is not the case now: the placement of Chalogaster as sister-group to Metischnogaster is supported by several characters, and thus the petiole of segment II is homoplastic. In this light, the variation in length of the petiole within Parischnogaster may indicate that it is not a reliable character. We therefore regard the consensus tree of figure 12 as a better representation of phylogenetic relationships among stenogastrine genera given present knowledge, with the placement of Anischnogaster unresolved.

In any case, the analysis indicates that Metischnogaster and Chalogaster are sister-groups. This is supported by the bidentate male mandibles (#3), the strongly impressed notauli (#6), and the laterobasal processes of the aedeagus (#13), none of which show homoplasy. In addition, the posteriorly attenuate valvulae (#8) is shared by Parischnogaster, Chalogaster, and Metischnogaster (further reduced in this genus), as is the apically dilated aedeagus (#12). These three genera form a clade. The flash-bands of the metasomal terga are an additional character uniting these three genera: males of all studied species have Terga IV–VII with broad, basal white or pale yellow bands, which can be exposed or hidden as the segments are more or less telescoped together and which function in sexual display (reviewed in Turillazzi, 1991). The phylogenetic relationships of Chalogaster thus appear clear.

At the same time, Chalogaster is distinct from its sister-group Metischnogaster by a number of apomorphies: the male clypeus rounded apically (#2, as scored in table 1 convergent in Eustenogaster, but of a different form in that genus), the scutellum clearly carinate in females (#6, as scored in table 1 convergent in Eustenogaster, but in that genus the males have the scutellum carinate as well, and weak traces of a carina are present in other stenogastrine genera), propodeal sculpture smooth (#9; it is striate in Metischnogaster), the parameral spines with a median process (#11; it is spiniform in Metischnogaster). Among the characters scored in Table 1, Metischnogaster is distinguished from Chalogaster by the apomorphy of the propodeal valvulae further reduced (#8), but also has a large number of other derived features, discussed by Carpenter (1988), viz., vertex with median longitudinal groove; spiracles of metasomal Segment II are visible in dorsal view; male with antennae having flagellomeres flattened, pleural areas sunken and densely pubescent, foretibiae with spatulate setae, hindwings with posterior fringe of hairs greatly elongated, and the aedeagus with a pair of small subapical projections. Given this distribution of features, recognition of a new genus for Chalogaster seems amply justified, thereby allowing the generic classification to convey information on the distribution of a number of remarkable characters, which is the point of classification (Farris, 1979).

All recorded localities of Chalogaster spatulata and the two species of its sister-genus Metischnogaster are shown in figure 8. Extensive collecting by CKS in various parts of the Philippines proper (i.e., the Philippines except Palawan and smaller associated islands on the continental shelf) persuades us these genera are in fact absent from that area. It is less certain that they are absent from Sulawesi or the Lesser Sunda Islands, but each has been subjected to a moderate amount of collecting. The present distributional information is consistent with the hypothesis that the speciation separating Chalogaster and Metischnogaster occurred as a single vicariance event around the Isthmus of Kra.

The recognition of a new genus underscores the point made earlier that the hover wasps are not well understood. The threat to the biota brought about by rapid deforestation and population growth in nations such as Viet Nam and Thailand underscores the need for a committed effort to protect and manage natural heritage, or the elements of that biota, such as the fascinating hover wasps, will never be understood.