The phylogeny of the Hymenoptera, and particularly the Aculeata, has recently been investigated critically by several authors (Brothers 1999; Ronquist 1999; Ronquist et al. 1999 ; Sharkey 2007; Davis, Baldauf and Mayhew 2010; Heraty et al. 2011). Generally, recent authors agree that the Aculeata is monophyletic and comprises two monophyletic lineages, the Chrysidoidea and the (Apoidea + Vespoidea), the latter group sometimes called the “Aculeata sensu stricto (s.str.)”. All also agree that the Plumariidae is the sister group of the remaining Chrysidoidea (see also Carpenter 1999). This paper concentrates on the Chrysidoidea, and the relevant relationships so far established are shown in Figure 1.

As chrysidoids, plumariids are very unusual morphologically, the males having broad wings with a relatively rich wing venation including well developed accessory veins in the apical membrane similar to those of many of the very distantly related Mutillidae (Vespoidea) and Heterogynaidae (Apoidea) (the latter consequently mistakenly assigned to Plumariidae by Brothers 1974), and the females being wingless with the thorax highly modified (in particular with the propleura fused into a tube and a deep ventral constriction at the base of the laterally expanded metathorax-propodeum). No females have been directly associated with males and the correspondence is putative although very strongly supported on distributional and morphological grounds (Evans 1967; Brothers 1984). (The supposed female of Plumaroides tiphlus Diez, 2008 is actually a member of the genus Pseudisobrachium Kieffer, Bethylidae, according to Quintero and Cambra 2010.) Plumariids occur in the more arid regions of the southern hemisphere, the males often being attracted to lights and the females having been collected under stones or in pitfall traps, but nothing further is known of their biology. There are seven modern genera: Plumarius Philippi, 1873, Plumaroides Brothers, 1974, Maplurius Roig-Alsina, 1994, Mapluroides Diez, Fidalgo and Roig-Alsina, 2007 and Pluroides Diez, Roig-Alsina and Fidalgo, 2010 from South America, and Myrmecopterina Bischoff, 1914 and Myrmecopterinella Day, 1977 from southern Africa. Phylogenetic analyses of generic relationships within the family (Roig-Alsina 1994; Carpenter 1999; Diez, Roig-Alsina and Fidalgo 2010) have shown that each southern African genus is most closely related to one or more of the South American genera rather than their being most closely related to each other (Figure 2). The distribution is unusual and putatively Gondwanan, with each of the two primary lineages occurring on both continents and therefore probably having arisen before the breakup of Gondwana. There are no records from the other major Gondwanan continent, Australia, nor from India or the Middle East (other Gondwanan derivatives with arid environments) however.

Within the Chrysidoidea, the family which apparently arose next (after the Plumariidae had diverged) is the Scolebythidae (see Figure 1). This also has a putatively Gondwanan distribution, based on modern representatives, with Pristapenesia Brues, 1933 in the neotropics, Clystopsenella Kieffer, 1911 in the neotropics and Australia, Ycaploca Nagy, 1975 in South Africa, Australia, Fiji, New Zealand and New Caledonia, and Scolebythus Evans, 1963 in Madagascar and South Africa (Engel and Grimaldi 2007; CSIRO 2011; Brothers pers. obs.); the recent description of a species of Pristapenesia from Thailand and China (Oriental and Palaearctic regions) (Azevedo, Xu and Beaver 2011) has cast doubt on the validity of the previous statement, though. Again, the closest relationships are between genera separated by long distances, with Clystopsenella and Scolebythus being sister groups, as are Ycaploca and Pristapenesia, as shown by Engel and Grimaldi (2007). However, fossil species of Scolebythidae (including two Pristapenesia) have been found in Early Cretaceous Lebanese amber, Late Cretaceous New Jersey amber, Eocene French amber, Eocene Baltic amber and Miocene Dominican amber (Engel and Grimaldi 2007), demonstrating that the family was much more widespread in the past. Strikingly, the fossil members generally are more derived with more reduced wing venation than most modern members. The other (“higher”) families of Chrysidoidea are all also known from fossils, in each case, as for Scolebythidae, the earliest being from the Early Cretaceous (see Engel and Grimaldi 2006; Rasnitsyn 2010; Ortega-Blanco, Delclòs and Engel 2011).

No fossil Plumariidae have yet been described, but two conspecific male specimens from Late Cretaceous New Jersey (USA) amber were recently stated to be members of the family (Rasnitsyn in Grimaldi, Shedrinsky and Wampler 2000; Brothers and Rasnitsyn 2000a, b) and this interpretation has been incorporated in some general accounts (e.g. Rasnitsyn 2002: 244, 2010), although some others (e.g. Grimaldi and Engel 2005: 430; Engel and Grimaldi 2006; Quintero and Cambra 2010) have reflected my subsequent view (Brothers 2003, 2004) that they may represent a new family. The long setae on the flagellomeres, reduced pronotum, large propleura and broad wings with a large pterostigma are particularly reminiscent of Plumariidae, although none of these features is unique to that family. The specimens differ from all modern male plumariids in having much simpler wing venation, lacking the accessory veins and with the second submarginal cell broadly sessile anteriorly, a much smaller anal (vannal/plical) lobe on the hind wing, the clypeus simple, the mandible truncate and four-toothed, the metanotum short and the mesopleuron less swollen. Preliminary analyses using the characters and taxa from Brothers and Carpenter (1993) had indicated their placement in Chrysidoidea (Brothers pers. obs.), but a more focused study in the context of the Chrysidoidea as a whole is required to establish their true relationship, and coincidentally confirm or refute existing estimates of the relationships of the chrysidoid families. As reported and discussed below, it is concluded that they represent not only a new genus and species, but are indeed most appropriately allocated to a new family.

The two amber pieces (Figures 3–4), each embedded in epoxy as described by Nascimbene and Silverstein (2000), were studied using standard methods and illustrated using stacked photographs taken with a Canon Powershot G10 digital camera adapted to Wild M7 and Wild M11 microscopes using a Clearshot 600 adapter kit (Alexis Scientific) and combined with CombineZP software (Hadley 2010). Drawings were done using a drawing tube and a Wild M8 microscope and subsequently digitised and corrected with reference to photographs, using CorelDraw X4. As with all fossils, some character states were not absolutely clear, but the most-probable states inferred (as explained in the species description) are those used in the analyses; if such states could not be inferred then they are coded as unknown. The specimens used for the analyses are listed in Appendix A. Terminology has been adapted from previous relevant studies.

Previous cladistic analyses of the Chrysidoidea and Plumariidae (e.g. Brothers, 1999; Carpenter 1999; Diez, Roig Alsina and Hidalgo 2010) have all rooted the trees using hypothetical ancestors with all-primitive states, based on comparisons with either Aculeata s.str. or other chrysidoids, and have utilised groundplans. For this analysis I used exemplars of the various taxa, both outgroup and ingroup, and thus made no a priori assumptions about probable direction of state changes, and thereby also included estimates of polymorphisms. Exemplars rather than groundplans were utilised, as advocated by Prendini (2001), but scorings for individual specimens were often combined to produce “summary” terminals with specified polymorphisms (see Table 1) rather than maintaining them as separate terminals. Since Chrysidoidea is the sister group of Aculeata s.str. (a clade with more-derived states for many characters, as shown by previous analyses), using only members of Aculeata s.str. as outgroups may have been misleading. In addition to other aculeates, I therefore included specimens representing various non-aculeate taxa which have previously been suspected as close relatives of Aculeata (Ichneumonidae, Trigonalidae and Evanioidea). Separate analyses were done using only Aculeata s.str. representatives as outgroup (similar to the approach of Carpenter 1999) and using the expanded outgroup to investigate the influence of using more or less distant taxa as outgroups. For the ingroup, in addition to the two fossil specimens, specimens of all genera of Plumariidae and most subfamilies or (for the smaller families) genera of the other families of Chrysidoidea were examined (Appendix A). In all cases, only males were used since we have no idea what the females of the fossil taxon were like, and there is often considerable sexual dimorphism in chrysidoids and aculeates in general. In a few cases states were derived from or checked in the literature (e.g., Olmi 1984, 1995, 2005; Gauld and Bolton 1988; Kimsey and Bohart 1990; Brothers and Carpenter 1993; Finnamore and Brothers 1993; Huber and Sharkey 1993; Prentice, Poinar and Milki 1996; Brothers and Janzen 1999; Terayama 2003b) specially where the condition of the specimens caused uncertainty, or states were scored as unknown (“?”) for taxa where I had only one or two specimens which could therefore not be dissected. The 90 characters used (see Appendix B) were chosen from those used in previous analyses for plumariid genera, chrysidoid families and aculeates in general, as well as a few newly discovered. The range of characters was thus greater than used in previous analyses.

Parsimony analyses by TNT (Goloboff, Farris and Nixon 2008a, b) were performed using the default settings unless otherwise noted (traditional search, 10 000 replications, tree memory 100 000 trees); implied weighting was implemented using various values of k but only those for k = 2.5 are reported (this seems to be a reasonable value for the size of the matrix and level of homoplasy found, see Goloboff et al. 2008). Where several most-parsimonious cladograms (MPCs) were found, only the strict consensus is reported. WinClada (Nixon 2002) was used for tree analysis and drawing; branch lengths reflect optimisation of unambiguous states only, with branches unsupported by such states collapsed. In addition to analyses where most characters were considered to be additive (as shown in Appendix B), analyses were also done considering all characters non-additive to investigate the effects of removing all hypotheses of evolutionary direction. Relative group support for all analyses, using GC values which are frequency differences (Goloboff et al. 2003), was estimated by symmetric resampling using TNT (new technology search using ratchet, drift and tree fusing, 10 000 replications, tree memory 100 000 trees).

Plumalexius rasnitsyni Brothers, sp. n.

urn:lsid:zoobank.org:act:20DEDD72-2F53-43D6-B455-E987604C0265

http://species-id.net/wiki/Plumalexius_rasnitsyni

Figs 3 –13

Type material:

Holotype male (Figures 3, 5–9), in heavily fractured block of yellowish amber embedded in a trapezoidal epoxy matrix about 22 × 10 × 7 mm, with labels as follows: “NEW JERSEY Amber: / Late Cretaceous / NEW JERSEY: Middlesex Co / Sayreville, White Oaks Pit / 1995, coll.Paul Nascimbene / AMNH no. NJ-695”, “NEW JERSEY Amber: / Late Cretaceous / AMNH no. NJ-695 / HYMENOPTERA:”, “Plumariidae” [Rasnitsyn’s handwriting], “HOLOTYPE / Plumalexius / rasnitsyni ♂ / D.J. Brothers, 2011” [red label, printed].

Paratype male (Figures 4, 10–13), in heavily fractured block of yellowish amber embedded in a rectangular epoxy matrix about 18.5 × 13.5 × 9 mm, with labels as follows: “NEW JERSEY Amber: / Late Cretaceous / NEW JERSEY: Middlesex Co / Sayreville, White Oaks Pit / 1995, coll.Paul Nascimbene / AMNH no. NJ-175”, “NEW JERSEY Amber: / Late Cretaceous / AMNH no. NJ-175 / HYMENOPTERA: / Family? (PN-2a) / Plumariidae” [Rasnitsyn’s handwriting], “?Family / Det. L. Masner 1996”, “PARATYPE / Plumalexius / rasnitsyni ♂ / D.J. Brothers, 2011” [yellow label, printed]. (This specimen is presumed to be a male because of its similarity to the holotype even though the metasoma is mostly not visible.)

Etymology:

The species name, a noun in the genitive case, honours Professor Dr Alexandr Rasnitsyn, who first recognised the significance of the specimens.

Description

(based on holotype, paratype data in parentheses where different or feature not visible in holotype): Male. Entirely pale yellowish (reddish) brown with venation slightly darker. Head and body length as preserved 2.03 (2.37) mm; estimated head length 0.24 (0.29) mm; estimated mesosoma length 0.80 (0.77) mm; estimated metasoma length 0.90 (0.89) mm; approximate forewing length 1.32 (1.46) mm; approximate hindwing length 1.03 (1.17) mm. Head and metasoma with scattered fine short erect setae, mesosoma almost glabrous, antennal pedicel and flagellomeres with fine long erect setae; legs with dense recumbent setae and scattered semi-erect setae.

Head: Hypognathous; about as wide as high; vertex evenly rounded. Eye ovate with convex inner margin, moderately protuberant, apparently glabrous, ommatidia distinct. Ocelli ovate, large. Occipital carina distinct. Frons and clypeus weakly convex; clypeus transverse with convex anterior/apical margin. Gena simple. Antennal sockets simple, apparently about as close to eyes as to each other, apparently close to posterior/dorsal margin of clypeus. Antennal scape about as long as wide (distinctly flattened posterolaterally and broadened towards apex), with several erect setae; pedicel (about half length of scape and of first flagellomere), with many fine long erect setae; (flagellomeres 1–4 becoming slightly longer sequentially, with many fine long erect setae). Mandible long, evenly broad and curved; two prominent short curved setae on lateral surface; apex truncate with four similar sharp teeth, apical tooth the longest. Maxillary palp at least 5-segmented; labial palp at least 3-segmented [palp bases concealed by foam but segmentation inferred from assumed points of origin].

Mesosoma ovate, about twice as long as wide/high. Pronotum forming a curved oblique ribbon anterolateral to mesoscutum, broader medially than posterolaterally; posterodorsal margin evenly concave; posterolateral margin strongly emarginate and approaching tegula dorsally; posteroventral angle broadly rounded; anteroventrally with slight collar (flange) but leaving propleura exposed anteriorly. Propleura closely associated or fused; anteriorly produced as a short neck; swollen posteriorly; posterior margin apparently almost straight but exposing small part of prosternum medially; forecoxae approximated. Mesoscutum shorter than wide, moderately convex; notaulus distinct and complete, weakly diverging anteriorly; tegula small and convex. (Mesoscutellum apparently almost as long as scutum, weakly convex.) Mesopleuron large and convex; meso-metapleural suture distinct, almost straight. Mesosternum with posteromedial margin almost straight and apparently slightly overhanging mesocoxal base; mesocoxae slightly separated. (Metanotum short and transverse, flattened, not constricted medially. Metapostnotum apparently as long as metanotum, slightly depressed.) Metasternum apparently somewhat depressed. Metacoxae slightly separated. Propodeum slightly longer than mesoscutum, weakly convex but more strongly so posteriorly although without any defined posterior declivity; incision between mesosoma and metasoma weak.

Forewing broad, about 2.2 × as long as wide, about 1.7 (1.8) × as long as mesosoma, with seven closed cells, veins approaching but not reaching margin. Costal cell well developed, broad. Pterostigma large, about 0.19 × as long and 0.17 (0.20) × as wide as wing, entirely sclerotised. Marginal cell about 2.38 (2.17) × as long as wide, 1.41 (1.52) × as long as pterostigma, apex acute. First submarginal cell about 2.08 (2.61) × as long as wide, 0.81 (0.80) × as long as marginal cell. Second submarginal cell almost as large as pterostigma, broadly sessile anteriorly, about 0.64 (0.57) × as long as first submarginal cell. Veins tubular except for nebulous free apical sections of M, Cu and A. No trace of any accessory vein(s) in apical membrane. Prestigmal vein (Sc+R) scarcely swollen, about 1.33 (1.37) × as long as vein 1Rs. Crossvein cu-a distinctly postfurcal, about 1.33 (1.75) × as long as 1Cu. Vein Cu2 absent, first subdiscal cell broadly open apically.

Hind wing about 0.8 × as long as forewing. (Basal and subbasal cells closed by tubular veins; costal cell open anteriorly, vein C present only basally. Veins tubular except for nebulous free apical sections of Rs, M, Cu and A. A few basal hamuli present in a cluster; about five apical hamuli. Crossvein rs-m long, about 2.67 × as long as 1Rs. Vein 1M very short, about 0.13 × as long as 2M+Cu. Crossvein cu-a distantly antefurcal, 2M+Cu about 0.57 × as long as 1M+Cu. Anal (vannal/plical) lobe apically delimited by moderate incision, lobe about 1.19 × as long as submedian cell, about 0.4 × as long as wing.) Jugal lobe absent.

Legs well developed, moderate in size; trochanters well developed and cylindrical; no trochantelli; tibiae without any spines or strong setae; basitarsomeres long, about as long as next three tarsomeres combined; all arolia large and flattened; claws simple ventrally. Foreleg with coxa subglobose, trochanter inserted apically; femur slightly swollen, with inner/anterior surface flattened; tibia with simple, weakly curved, bladelike calcar subapically. Mid- and hind legs with coxae globose, hind coxa somewhat larger than mid-coxa; tibiae each with two straight simple apical spurs, inner spur somewhat longer than outer.

Metasoma elongate oval, about 2.6 (indeterminable in paratype) × as long as wide/high; terga subequal in length. First tergum broad, weakly contracted toward base, profile evenly merging with second. First sternum apparently simple and evenly overlapping second. Seventh tergum apparently simple with apical margin convex. Seventh sternum apparently reduced and mostly concealed. Hypopygium simple, weakly convex, with narrowly rounded apical margin. Cercus apparently present, cylindrical. Genitalia with paramere apparently broadly rounded apically.

Female. Unknown.

Figures 3–7.

New Jersey amber containing specimens of Plumalexius rasnitsyni sp. nov. 3 Specimen NJ-695, holotype (circled) 4 Specimen NJ-175, paratype (circled) 5–7 Specimen NJ-695, holotype 5  Ventrolateral view 6 Dorsolateral view 7 Detail, ventrolateral view.

Figures 8–9.

Plumalexius rasnitsyni sp. nov., holotype. 8 Ventrolateral view 9 Dorsolateral view.

Figures 10–11.

Plumalexius rasnitsyni sp. nov., specimen NJ-175, paratype. 10 Dorsolateral view 11 Details, dorsolateral view.

Figures 12–13.

Plumalexius rasnitsyni sp. nov. 12 Wings, based on both specimens 13 Paratype, dorsolateral view, right wings in grey. Abbreviations. Wing veins: A = anal, C = costa, Cu = cubitus, M = media, R = radius, RS = radial sector, Sc = subcosta (numerals indicate abscissae, all lower-case indicates crossveins); cells: BC = basal cell (cell R), CC = costal cell (cell C), DC = discal cell (cells 1M, 2M), MC = marginal cell (cell 2R1), SBC = subbasal cell (cell 1Cu), SDC = subdiscal cell (cell 2Cu), SMC = submarginal cell (cells 1R1, 1Rs, 2Rs).

Table 1 shows the distribution of character states across the taxa.

The cladograms resulting from the analyses using an aculeate outgroup (illustrated using Anthoboscinae, but the relationships within the Chrysidoidea were not affected by changing this to any of the other three aculeates) are shown in Figures 14–17. The consensus tree from the “equally weighted additive” analysis (Figure 14) shows Chrysidoidea as monophyletic, all chrysidoid families also as monophyletic (although with their relationships often not convincingly resolved, as shown by several apparent clades having no or very low relative resampling support), Plumalexius sister to Plumariidae (this clade sister to the remaining chrysidoids), and the plumariid genera with similar relationships to those found earlier (see Figure 2). In contrast, although the consensus tree from the “equally weighted non-additive” analysis (Figure 15) also shows Chrysidoidea as monophyletic, all chrysidoid families also as monophyletic (with their relationships even less resolved), and the same relationships for the plumariid genera, (Plumalexius + Plumariidae) now groups with Sclerogibbidae and Dryinidae, although without relative support. The consensus tree from the “implicitly weighted additive” analysis (Figure 16) shows similar relationships as the equally weighted version (Figure 14), except that there is slightly greater resolution for the families of Chrysidoidea and Scolebythidae is no longer sister to Chrysididae which is now monophyletic with Bethylidae, although some branches lack positive relative support values; there is also greater resolution for the plumariid genera. Similarly, the single MPC from the “implicitly weighted non-additive” analysis (Figure 17) produced improved resolution, showing (Bethylidae + Chrysididae) as monophyletic, but Plumalexius is grouped with the chrysidoids other than Plumariidae (although without positive relative branch support). The differences from previous analyses, most strikingly involving Scolebythidae and Sclerogibbidae (see Figure 1), are probably due to two factors: the use of exemplars instead of groundplans (introducing polymorphisms), and the position of the Aculeata s.str. outgroup taxa as relatively more derived than the Chrysidoidea.

To try to address the second of the above concerns, analyses were done using four non-Aculeata as outgroup. The resulting cladograms (Figures 18–21) are presented with Ichneumonidae as outgroup, but using any of the other three non-aculeates made no difference to the relationships shown for the Chrysidoidea. The consensus tree from the “equally weighted additive” analysis (Figure 18) shows Chrysidoidea as monophyletic, all chrysidoid families also as monophyletic (but their relationships still sometimes unconventional), Plumalexius basal to Plumariidae (with high support), and the plumariid genera with similar relationships to those found earlier. Although the study was not intended to reflect the relationships amongst the outgroup taxa, it is interesting that Aculeata s.str. (Vespoidea and Apoidea) appears as paraphyletic in this analysis. The consensus tree from the “equally weighted non-additive” analysis (Figure 19) also shows Chrysidoidea as monophyletic (but with Embolemidae as basal), all chrysidoid families also as monophyletic but with most relationships very different from previous findings (except that Bethylidae and Chrysididae are well supported as monophyletic, and many of the family relationships have no positive relative branch support), with Plumariidae (and Plumalexius sister to it) appearing as most closely related to Sclerogibbidae and Dryinidae; the relationships for the plumariid genera remain consistent. Aculeata s.str. now appears as polyphyletic, with Evanioidea interpolated between Vespoidea and Apoidea. The single MPC resulting from the “implicitly weighted additive” analysis (Figure 20) is fully resolved, shows Chrysidoidea as monophyletic, all chrysidoid families as monophyletic, Plumariidae (and Plumalexius sister to it) as sister to the remaining chrysidoids, and the relationships of those families as found by previous analyses (see Figure 1); the relationships of the plumariid genera also agree with previous analyses (see Figure 2), except that Plumaroides appears as sister to Pluroides rather than Mapluroides. Aculeata s.str. is paraphyletic but with the apparent sister-group relationship of Apoidea to Chrysidoidea not supported (the resampling analysis instead showed a monophyletic Aculeata s.str. as supported with a value of 12). The “implicitly weighted non-additive” analysis also produced a single MPC (Figure 21) with a monophyletic Chrysidoidea, monophyletic chrysidoid families, Plumalexius sister to Plumariidae, and Bethylidae and Chrysididae forming a monophyletic group but apparently sister to Scolebythidae (but without relative branch support); relationships for the plumariid genera are the same as for the “additive” analysis. Aculeata s.str. is now monophyletic (with good support) and sister to Chrysidoidea (with some positive support). It is notable that the chrysidoid relationships shown are more similar to those for the “additive” analyses than those seen in the “unweighted non-additive” analysis.

At first sight, consideration of all of the above results, involving not only the placement of Plumalexius but even more the relationships amongst the other chrysidoids, has produced a slightly confused picture, perhaps not unexpected for a set of analyses using exemplars and considerable polymorphism, and also based on characters which have previously been used at very different levels. The limitation of having to exclude all characters restricted to females (many of which have proved extremely informative in previous analyses, and one of which, the presence of an articulation within gonocoxite IX, is probably the most significant unique synapomorphy for Chrysidoidea) has also had an effect. Nevertheless, it is gratifying that the results of most previous studies have been confirmed, or at least not convincingly contradicted. Accordingly, I consider that the cladogram which agrees best with those results, one using an expanded outgroup and additive characters, and derived using implied weighting (an approach advocated by Goloboff et al. 2008), should be considered the preferred current estimate of the relationships of the families of Chrysidoidea and the genera of Plumariidae. This is shown in Figure 22, elaborated and adjusted from Figure 20, with Aculeata s.str. shown as monophyletic (which increased the length of the tree by a single step) and the relationships of the tribes of Chrysidinae resolved to reflect that found (and supported) most often in all analyses (which did not alter the tree length). It must be noted that, although for each family some subfamilies, tribes and genera are also shown, and their apparent relationships often (but not always) agree with other recent studies (such as Carpenter 1999; Terayama 2003a; Engel and Grimaldi 2007; Carr, Young and Mayhew 2010), not all subfamilies or tribes are represented by exemplars, nor are all genera included (except for the Plumariidae), and the characters used did not necessarily include all those which have been found useful within all of the families, so these results are incomplete in that respect; the aim of including the exemplars used was to reflect the variation found within the families rather than to discover intra-family relationships (except for Plumariidae).

Plumalexius seems convincingly indicated as sister to the Plumariidae, although one analysis was ambiguous about this; trees with it placed as sister to the remaining chrysidoids or as sister to the Chrysidoidea as a whole differ in length from that shown in Figure 22 by only 4 and 5 steps respectively (lengths 478 and 479 compared with 474), emphasising its relatively basal position. It does not share any unique synapomorphies with Plumariidae, however (the five unambiguous states supporting the sister relationship to Plumariidae are 11-2: flagellomere setae conspicuous and erect; 21-1: pronotal posteroventral margin strongly concave; 45-1: metasternum weakly depressed anteromedially; 50-0: pterostigma large and prominent; and 87-0: hypopygium completely exposed or almost so, all states found elsewhere in relatively distantly related taxa). The long erect flagellar setae of Plumalexius and some Plumariidae have been indicated as a putative synapomorphy for the family (Rasnitsyn 2002: Fig. 331). The arrangement of the setae in Plumalexius is most similar to that in Myrmecopterina, although the setae are less dense and considerably longer in Plumalexius, but the present analysis has shown that other plumariid genera lack such setae and, conversely, they are also present in some Scolebythidae and Bethylidae at least; prominent flagellomere setae are actually found widely in the Chrysidoidea. The other most obvious similarity, more extensive venation in both wings than in other chrysidoids, is a symplesiomorphy. It is thus evident that there is no key apomorphy associating Plumalexius with the Plumariidae sufficient to assign it to that family. Were that to be done, the expanded family would lose its present defining features, such as the presence of apical accessory veins in the wing membrane, the reduced second submarginal cell and the tapered mandibles with few apical teeth. In view of this, I conclude that the best solution is to propose a new family for it, as has been done above, something which also emphasises its distinctiveness. In contrast to the specialised morphology of Scolebythidae, showing several adaptations enabling the effective parasitisation of wood-boring beetle larvae, the morphology of Plumalexius provides little clue as to its biology, specially since the female is unknown. The male looks like a very generalised wasp, probably very similar to the form ancestral to Chrysidoidea as a whole.

Apart from the above results, the variety of analyses performed has shown that the use only of an outgroup which is sister to the ingroup, and which may have many characters with relatively more-derived states than the ingroup, may produce misleading or ambiguous results (Figs 14–17 all show different relationships from the preferred result). Instead, the outgroup should be expanded to include taxa similarly related to both the ingroup and its sister group. Furthermore, the use of additive characters where reasonable inferences of additivity can be made is likely to produce better-resolved cladograms than if all characters are considered non-additive, and it seems that using implied weighting not only improves the results obtained under both scenarios, but also reduces the uncertainty induced by considering all characters non-additive. The results obtained here, therefore, indicate that wherever possible additive characters and a method (such as implied weighting) which gives greater weight to the more reliable characters should be used.

Whether Plumalexius is sister to Plumariidae or not affects the estimated minimum age of Plumariidae: if it is, then Rasnitsyn’s (2002, 2010) estimate remains reasonable (after all, the common ancestor of two lineages must be at least as old as either lineage), but if it is sister to the remaining chrysidoids or to Chrysidoidea as a whole, then that estimate for Plumariidae is poorly founded. Since all other chrysidoid lineages date from the Early Cretaceous (Engel and Grimaldi 2006; Rasnitsyn 2010), that would also be the estimated minimum age for Plumariidae itself if it is considered to be sister to the other chrysidoids rather than to Plumalexius. In any case, the presence of a group apparently closely related to Plumariidae in North America in the Cretaceous requires reassessment of ideas on the geographic origin of Plumariidae, making it unlikely that the group arose on Gondwanaland. Like the Scolebythidae, it is probable that the modern members are scattered relicts of a group with a previously much more extensive distribution. The discovery of fossils clearly attributable to Plumariidae will be critical in solving the puzzle, but such fossils are not very likely to be found if members of the family have always been adapted to arid environments, probably being parasitoids of subterranean hosts (Evans 1967), most likely beetle larvae. In contrast, Scolebythidae tend to be found in wooded or forest habitats as parasitoids of wood-boring beetle larvae, and thus have often been entombed in exuded resin and become inclusions in amber, facilitating their later discovery. Both Plumalexius and Boreobythus Engel and Grimaldi 2007 (Scolebythidae) were found in the same amber deposits, putatively derived from temperate coastal or deltaic swamps of coniferous trees (Grimaldi, Shedrinsky and Wampler 2000), an environment very different from those where modern Plumariidae exist.