Introduction

The Aculeata or stinging Hymenoptera is a monophyletic taxon whose the common evolutionary origin is supported, among other synapomorphies, by the unique presence of the sting apparatus (Sharkey et al 2012), which is a modification of the ovipositor. It includes a glandular portion where the venom and others substances are produced, and a motor portion composed of chitinous and muscular structures that act together in the injection of venom and sting protrusion and extrusion (Manzoli-Palma & Gobbi 1997).

Considering the chitinous parts of the sting, one can recognize a pair of spiracular plates that represent the eighth tergum and bear the last pair of abdominal spiracles; a pair of quadrate plates (ninth hemitergites) that are below the spiracular plate and are connected to the arch and anal plate; one pair of oblong plates homologous to the second valvifers (basal plates of ovipositor) that articulate posteriorly with the gonostylus (third valvulae); one pair of triangular plates (first valvifers), articulated with the quadrate and oblong plates; the sting, which represents the fusion of the second valvulae and is divided into bulb and stem and is connected to the anterior portion of oblong plates; and articulated with the anterior dorsal portion of stem is the furcula (Silveira & Silveira 1994).

The primitive role of the sting apparatus is associated with prey capture. However, it can also serve as a powerful deterrent to large enemies, so that in several lineages, it has taken on an important defensive function, especially in social insects (Macalintal & Starr 1996). As noted by Starr (1985, 1988), defensive use of the sting in social insects might be due to the pressure exerted by predators attracted by the increased size of the colonies, particularly in the tropics. Because of its defensive potential, the sting apparatus is considered a key factor that favored the multiple origins of social behavior within Aculeata. Added to this the fact that only females act as workers in wasps, ants, and bees (Silveira & Silveira 1994).

The family Vespide includes more than 5,000 species in 259 genera, distributed in six extant subfamilies: Euparaginae, Masarinae, Eumeninae, Stenogastrinae, Polistinae, and Vespinae (Pickett & Carpenter 2010). The tribe Epiponini belongs to Polistinae and comprises of 19 genera in the New World, all represented in Brazil. The wasps of this tribe show a varied nest architecture, swarm-founding, polygyny (multiple functional queens) (Richards & Richards 1951), and complex caste differentiation (Richards & Richards 1951, Richards 1971, Noll & Wenzel 2008).

A robust phylogenetic tree is of paramount importance as a beginning point for many kinds of evolutionary studies. Understanding the evolutionary history of a character can allow us to tease apart important factors for adaptation to a specific environmental condition from those due to history (Arévalo et al 2004).

Studies in other Aculeata, such as social bees and ants, showed variation among the structures of the sting apparatus, demonstrating its potential for systematic purposes. Kugler (1979) used the sting apparatus in a study of myrmicine ants and proposed a phylogeny for this group. Cardinal & Packer (2007) used characters derived from chitinous structures of sting apparatus for a systematic study that corroborated the monophyly of advanced eusociality in corbiculate bees.

Silveira & Silveira (1994) and Macalintal & Starr (1996) asserted that the morphological variation of the sting apparatus in polistine wasps and all the family Vespidae is small compared with other aculeates such as bees and ants. Silveira & Silveira (1994) also stated that the contribution of the morphology of this apparatus for the phylogeny of Polistinae is proportional to the low differentiation found. However, the conclusions were based only on observations of the morphologies of the sting apparatus, without conducting any study of phylogenetic relationships.

In the present study, the use of the chitinous structures of the sting apparatus was evaluated for phylogenetic analysis and we point out that the characters collected from these structures were helpful in resolving relationships within Epiponini and to recover relationships among the subfamilies of Vespidae.

Material and Methods

Species from each genus of Epiponini were analyzed, totaling 22 species. Representatives of tribes Polistini, Mischocyttarini, and Ropalidiini (Polistinae) and subfamilies Eumeninae, Stenogastrinae, and Vespinae composed the outgroup (Table 1). The sting apparatus was removed with the aid of tweezers and boiled in a 10% KOH solution during 10 min, until all the remaining muscles were separated from chitinous structures. Subsequently, it was dissected into pieces of interest and mounted on slides with Hoyer’s solution (30–40 mL of distilled water, 30 g of Arabic gum, 200 g of chloral hydrate, and 20 mL of glycerin) for temporary storage (adapted from Silveira & Silveira 1994, Diniz 1997). The structures of interest were observed and photographed with a stereomicroscope (Leica MZ16 and light box Leica CLS 150×) with a digital camera attached (Leica DFC320). Later, structures were outlined in the software CorelDRAW.

Table 1 List of species used in the analysis.
Full size table

The characters, obtained mainly from the spiracular and quadrate plates according to Silveira & Silveira (1994), gonostylus and stylet (bulb end stem) were scored in a data matrix (Table 2). For the phylogenetic analysis, the software TNT 1.1 (Tree Analysis Using New Technology—Goloboff et al 2008a, b) and Winclada (Nixon 1999) were used to perform a parsimony analysis and edit the obtained cladograms, respectively. Characters 3, 5, and 11 were treated as ordered because a linear transformation series could be noted in each. The parameters used were traditional search with 500 replicates and tree bisection reconnection branch swapping, with 10 trees saved per replication. The analyses were performed with equal weighting (EW) and implied weighting (IW) with the default concavity. Based on previous studies of relationship of Vespidae, the trees are rooted on the subfamily Eumeninae. The robustness of branches was tested using symmetric resampling. Also, sting characters (char.) were optimized on the most widely accepted tree for the Epiponini (Wenzel & Carpenter 1994).

Table 2 Data matrix. Symbols (− inapplicable).
Full size table

Results and Discussion

Seventeen characters from the sting apparatus were delimited (Table 3, Figs 1–4). The analyses under equal and implied weighting resulted in 95 and 3 most parsimonious trees, respectively.

Table 3 Characters of sting apparatus.
Full size table
Fig 1
figure 1

Chitinous structure of the sting apparatus used for the analysis of characters. a Spiracular plate (Chartergellus communis): 1 lamina, 2 spiracle, 3 dorsal bar, 4 oblique ridge, 5 basis of plate, 6 anterior margin, 7 apodeme. b Quadrate plate (Angiopolybia pallens): 1 lamina, 2 apodeme, 3 ventral margin of apodeme, 4 dorsal margin of plate, 5 ventral margin of plate, 6 anal arc. c Sting (Parachartergus fulgidipennis): 1 bulb, 2 stem. d Gnostylus (Pachymenes ghilianii): 1 segment, 2 bristles.

Full size image
Fig 2
figure 2

Characters of the sting apparatus (stem and gonostylus, ventral view). Segmentation of gonostylus: gonostylus with 2 segments = 0, gosnostylus monossegmented = 1; arrangement of bristles in gonostylus: predominantly at the tip = 0, throughout the gonostylus = 1, higher concentration at the tip = 2; stylet: bulb and stem undifferentiated in width = 0, bulb wider than stem = 1. a Angiopolybia pallens, b Parachartergus fulgidipennis, c, f Pachymenes ghilianii, d Agelaia vicina, and e Mischocyttarus leicointei.

Full size image
Fig 3
figure 3

Characters of the sting apparatus (spiracular plate, dorsal view). Oblique ridge: present = 0, absent = 1; extension of oblique ridge: surpassing half the height of plate = 0, until half of the plate height = 1, around the plate = 2; margin crossing the apodeme of spiracular plate: absent = 0, present = 1; lateral margin of spiracular plate on the edge with dorsal bar: continuous = 0, not continuous = 1; if continuous: continuous since the anterior portion of margin = 0, continuous since the posterior portion of margin = 1, emarginated on posterior portion, before becoming parallel to the bar = 2; if not continuous: straight = 0, emarginated on the edge with bar = 1; shape of posterior margin of spiracular plate: forming obtuse angle = 0, rounded = 1, forming acute angle = 2, straight = 3; thickness of dorsal bar (between spiracular plates): thick and well sclerotized = 0, moderately thick = 1, very thin and little sclerotized = 2. a Nectarinella xavantinensis, b Polybia jurinei, c Pachymenes ghilianii, d Synoeca surinama, e Dolichovespula maculata, f Metapolybia unilineata, g Mischocyttarus leicontei, h Agelaia vicina.

Full size image
Fig 4
figure 4

Characters of the sting apparatus (quadrate plate, lateral view). Central margin of quadrate plate on the edge of anal arc: straight = 0, emarginated = 1, with a “step” = 2; dorsal margin of apodeme: parallel to ventral margin = 0, rounded = 1, posteriorly angled = 2; dorsal margin of quadrate plate: slightly curved before anal arc = 0, strongly curved before anal arc = 1, straight = 2; ventral margin of quadrate plate: rounded = 0, straight = 1; posterior margin of apodeme (quadrate plate): not covering the posterior margin of plate = 0, under posterior margin of plate = 1; position of anal arc regarding the posterior margin of quadrate plate: anteriorly = 0, centralized = 1. a Angiopolybia pallens, b Chartergellus communis, c Synoeca surinama, d Dolichovespula maculate, e Ropalidia latebalteata, f Charterginus fulvus, g Chartergus metanotalis.

Full size image

The strict consensus tree of the EW analysis shows a low level of resolution (Figs 5 and 6), with Eumeninae as the first subfamily to diverge. Stenogastrinae appears as sister group of a polytomy that includes Vespinae and Polistinae, supported by five synapomorphies as follows: bulb wider than stem (char. 1, state 1), bristles throughout the gonostylus (char. 3, state 1), lateral margin of spiracular plate emarginated on posterior portion before becoming parallel to the dorsal bar (char. 8, state 3), dorsal bar moderately thick (char. 11, state 1), and dorsal margin of quadrate plate straight (char. 14, state 2). Analysis of IW (Figs 7 and 8) recovered a similar relationship to the EW analysis, but with a better resolution for Polistinae and within Epiponini. However, Vespinae and Epiponini were still together in a clade. Even though lacking resolution, these findings reinforce a single origin for eusociality of Vespidae (Pickett & Carpenter 2010) instead of a dual origin (Hines et al 2007).

Fig 5
figure 5

Strict consensus cladogram obtained with equal weighting (length 53, consistency index 49, retention index 69). Black circles indicate unreversed transformations and white circles homoplastic transformations. Numbers above and below the circles indicate number and the transformation series of characters.

Full size image
Fig 6
figure 6

Result of symmetric resampling analysis under equal weights.

Full size image
Fig 7
figure 7

Strict consensus cladogram obtained with implied weighting (weighted score 4.263, consistency index 46, retention index 65). Black circles indicate unreversed transformations and white circles homoplastic transformations. Numbers above and below the circles indicate number and the transformation series of characters.

Full size image
Fig 8
figure 8

Result of symmetric resampling analysis under implied weights.

Full size image

Regarding the evolution of characters, the distribution of some of them showed a clear evolutionary trend. The segmentation of the gonostylus (char. 2, Fig 9), for example, shows the shortening and reduction of gonostylus segments to one segment that occurs in the most apomorphic groups within Apocrita (Smith 1970). Another example is the thickness of the dorsal bar (char. 11, Fig 9). This structure is thinner in more apical taxa; whereas in the plesiomorphic condition, it is thick and well sclerotized. Given the homology among the spiracular plates and the eighth tergum, it is reasonable to assume that evolution progressed towards simplifying the connection among the plates, the remainder of the tergum (Silveira & Silveira 1994), especially in species with high degree of sociality.

Fig 9
figure 9

Optimization of sting apparatus characters on the phylogeny proposed by Wenzel & Carpenter (1994). Black circles indicate unreversed transformations and white circles indicate homoplastic transformations. Numbers above and below the circles indicate number and the transformation series of characters.

Full size image

This fact can be corroborated by observing the subfamily Vespinae (highly social) which also shows the apomorphic condition. But Polistes (Polistinae), which have less complex sociality, retain the plesiomorphic condition. The loss of the continuity of the dorsal bar with the spiracular plate (char. 7, Fig 9) shared by almost all Epiponini also indicates a trend towards the separation of the spiracular plates. In Apis, a highly social genus, the spiracular plates are completely separated, and this fact is associated with changes in the movement of the sting (Cardinal & Packer 2007).

The oblique ridge was lost in Epiponini (char. 4, Fig 9), but some reversals happened independently. Nevertheless, in this group, it appears reduced especially in the most apical genera. Also it was observed an increase in the amount of bristles on the gonostylus (char. 3, Fig 9) and an enlargement of the bulb of the sting through the phylogeny (char. 1, Fig 9).

When the characters were optimized on the phylogeny proposed by Wenzel & Carpenter (1994) (Fig 9), it was possible to identify several characters that were highly homoplastic, with several reversions or independent origins. On the other hand, some of them offered important information. Characters 1 (state 1), 3 (state 1), and 10 (state 2) were synapomorphic for the clade Vespinae + Polistinae, adding more support to the close relationship between these two subfamilies. Characters 4 (state 1) and 17 (sate 0) were synapomorphic for the Epiponini, supporting the monophyly of the tribe. Others characters still were synapomorphic for small clades within this tribe.

The overall uniformity of the sting structures, as previously noted for the Polistinae (Silveira & Silveira 1994) and Ropalidiini (Macalintal & Starr 1996) was confirmed here, resulting in a weakly resolved tree under equal weights. However, the results under implied weighting were much better resolved, consistent with the findings of Goloboff et al (2008a, b) that weighting against homoplasy improves resolution (and support) for morphological data. Our data also establish that the sting apparatus is apparently informative for higher level taxonomic studies because it supported the relationships between the solitary and social subfamilies of Vespidae, reinforcing a single origin for the eusociality, found in the ancestor of Vespinae and Polistinae. Also, some sting characters showed support for the tribe Epiponini.