Sphecomyrmini Wilson and Brown, 1967

Borysenko, Leonid H., 2017, Description of a new genus of primitive ants from Canadian amber, with the study of relationships between stem- and crown-group ants (Hymenoptera: Formicidae), Insecta Mundi 2017 (570), pp. 1-57 : 15-17

Tribe Sphecomyrmini Wilson and Brown, 1967

Diagnosis (workers, gynes). (1) head capsule unspecialized; (2) mandibles unspecialized; (3) anterolateral clypeal margins not produced over mandibular bases in rounded lobes; (4) peg-like setae on anterior clypeal margin present or not; (5) ocelli present; (6) F1L>PL<F2L, F1L>F2L; F1 often longest flagellomere; (7) neck short or long; (8) petiole subsessile or pedunculate; (9) gastral constriction present or not.

Comment. Under the PhyloCode, the taxon should be named “ Sphecomyrmini P, converted clade name” and defined as “the clade consisting of Sphecomyrma freyi Wilson and Brown, 1967 and all species that share a more recent common ancestor with Sphecomyrma freyi Wilson and Brown, 1967 than with Haidomyrmex cerberus Dlussky, 1996 or Zigrasimecia Barden and Grimaldi, 2013 ”.

Composition. Genera Sphecomyrma Wilson and Brown (type genus), 1967, Cretomyrma Dlussky, 1975 , Armania Dlussky, 1983 , Pseudarmania Dlussky, 1983 , Orapia Dlussky, Brothers and Rasnitsyn, 2004 , Gerontoformica Nel and Perrault, 2004 .

Discussion. The species of this tribe seem to be less specialized than those of the other two tribes of the Sphecomyrminae . Also, the tribe is heterogeneous due to the presence of possibly non-monophyletic genera Sphecomyrma and Gerontoformica .

In the genus Sphecomyrma , the SL/HL index has an unusually large range. Even in a single species, S. freyi , values range from 0.28 to 0.62 ( Table S2), i.e., the difference is 120 %. Data on 11 modern genera ( Radchenko 1991, 1994; Seifert 1992, 2000, 2003; MacKay 1993; Radchenko and Elmes 1998, 2003; Ward 1999; Radchenko et al. 2002; Baroni Urbani and de Andrade 2003; Wild 2004; Bolton 2007; Bolton and Fisher 2011) show that the range of the SL/HL among workers, gynes, and workers + gynes within a given species rarely exceeds 10 %, with a single maximum of 20 % in Myrmica ( Radchenko 1994) , 31 % in Linepithema Mayr ( Wild 2004) , 31 % in Dolichoderus Lund ( MacKay 1993) . Such results raise the possibility that S. freyi #3, which has the SL/HL value of 0.28, does not belong to Sphecomyrma .

Regarding the variation of the SL/HL index within a genus, the findings are the following. In the Cretaceous ant males as exemplified by Baikuris , the range is 77 %; in the workers of Sphecomyrma , it is 160 % (or 106 % if the clypeal lobe of S. mesaki is excluded). In modern genera, males are characterized by a large range: 100 % in Proceratium Roger ( Baroni Urbani and de Andrade 2003) , 204 % in Dolichoderus ( MacKay 1993) , 175% ( Seifert 1988) and 260 % ( Radchenko 1994) in Myrmica . The workers and gynes of modern genera range from 30 % in Pseudomyrmex Lund to 55 % in Proceratium ( Seifert 1992, 2000, 2003; Radchenko 1994; Ward 1999; Baroni Urbani and de Andrade 2003; Bolton and Fisher 2011); only in the Dolichoderinae are there single large values, causing larger ranges up to 128 % in Dolichoderus ( MacKay 1993) , and 121 % in Technomyrmex Mayr ( Bolton 2007) . Therefore, whereas the range in Baikuris is consistent with the data on recent genera, the range in Sphecomyrma is not. Although the limits of within-genus variation are not known for Cretaceous representatives, taking into account the unique morphology of S. mesaki (short scape, antennal scrobes), its taxonomic position should also be re-examined.

The genus Gerontoformica , which includes 13 species, is even more problematic. Indeed, Gerontoformica is variable in generic- and higher-level characters, such as the scape length, palp formula, the presence of a petiolar peduncle, ocelli, gastral constriction, and trochantellus. In some species, the head seems to be specialized as in the Zigrasimeciini (frontal part thickened, anterolateral clypeal lobes present). All the species have peg-like setae on the anterior clypeal margin, again as in the Zigrasimeciini . The SL/HL index ranges from 0.19 (the minimum value obtained in this study) to 0.67 (one of the highest values of all stem-group ants) ( Table S2), i.e., the difference is 250 %, which is larger than any within-genus range obtained here.

Gerontoformica and Sphecomyrmodes Engel and Grimaldi have recently been synonymized ( Barden and Grimaldi 2016), which made the genus even more heterogeneous. Of special interest is the species G. cretacica , transferred from incertae sedis. In terms of the results of the statistical analysis, G. cretacica occupies an intermediate position between the crown- and stem-group ants. For SL/HL, G. cretacica is similar to the crown-group ants; its scape is the longest of all stem-group ants known, but its flagellum is also elongated proportionally to the scape, and thus the indices FL/HL and AL/HL are among the largest reported. The same applies to the pedicel, F1 and F2. Other interesting features are: the terminal flagellomere is not elongated (a rare character also found in two species of the Haidomyrmecini ), antenna without the club, flagellomeres diminish in length towards the apex ( Nel et al. 2004).

It is possible that Gerontoformica can be split into several genera, and these genera can be placed in different tribes within the Sphecomyrminae . But G. cretacica may fall out of Sphecomyrminae P; its unique combination of characters (if it is not an artifact of preservation as suggested by the latest report ( Barden and Grimaldi 2016)) may indicate that it is a representative of a subclade of Pan-Formicidae P branched close to the crown clade.

The problem of large within-genus variation of the scape length seen in Sphecomyrma and Gerontoformica may have, however, another angle. There is a long-lasting debate on whether stem-group ants were eusocial or not, and, as it was already pointed out, the answer to this question is probably linked to the problem of scape elongation ( Dlussky 1983; Dlussky and Fedoseeva 1988). It is reasonable to assume that the transition to eusociality did not occur instantly but gradually, via the stage of facultative sociality, i.e., social behavior in stem-group ants depended on, for example, abiotic environmental conditions and varied even among closely related species. A similar pattern of social organization is found in the modern halictine bees ( Yanega 1997). If this is the case, then the scape length was not stable in stem lineages.

Lastly, I will review the taxonomic status of the Armaniinae , an enigmatic group known mainly from winged forms preserved as imprints in rock. Dlussky initially assigned all his new Cretaceous species to the Sphecomyrminae ( Dlussky 1975) but then transferred to the new family Armaniidae ( Dlussky 1983) . Wilson returned them to the Sphecomyrminae , and synonimized almost all genera of the Armaniinae (except Cretomyrma ) with the genus Sphecomyrma ( Wilson 1987) . Bolton changed the status of Dlussky’s family to the subfamily Armaniinae ( Bolton 1994, 2003); Dlussky first accepted Bolton’s approach ( Dlussky 1996) but then again called the group Armaniidae ( Dlussky 1999b) , and then again mentioned the subfamily Armaniinae ( Dlussky et al. 2004) . In 2005, Wilson mentioned this group as the family Armaniidae ( Wilson and Hölldobler 2005) . Such confusion is caused by the absence of reliable characters distinguishing the Armaniinae from the other groups, especially from the Sphecomyrminae .

The gastral constriction is one of the most confusing characters. For example, Dolichomyrma was described without ( Dlussky 1975) or with ( Wilson 1987) the gastral constriction; Armania - with ( Dlussky 1983) or without ( Wilson 1987); Arhaeopone - with ( Dlussky 1975) or without ( Dlussky 1983); Petropone most probably has the gastral constriction ( Dlussky 1975). It is now clear, however, that the gastral constriction is not a stable character in Sphecomyrminae , being present in some genera and absent in others. The trochantellus may be an important character distinguishing the Armaniinae and Sphecomyrminae ( Dlussky 1983) , but it not stable either, being variable even within a genus (e.g., in Gerontoformica ) ( Barden and Grimaldi 2014).

Also, the Armaniinae are characterized by a “very short scape”, in contrast to just a “short scape” of the Sphecomyrminae ( Bolton 2003) . This is viewed as one of the most important diagnostic characters of the Armaniinae ( Dlussky 1983) . However, as can be seen from the statistical analysis, the scape indices as well as other antennal indices of the Armaniinae are not at all different from those of the Sphecomyrminae or Sphecomyrmini ( Tables S2, S 4–S 8; Fig. 4 ). To give more support to these findings, I compared within-subfamily ranges of the scape index SL/HL in workers. The difference between the extreme values was the following: 3.5 times in the Sphecomyrminae (between Gerontoformica orientalis and G. rugosus ); 2.2 times in the Ponerinae (from 0.53 in Feroponera Bolton and Fisher to 1.16 in Diacamma Mayr ) ( Shattuck and Barnett 2006; Bolton and Fisher 2008); 3.4 times in the Dolichoderinae (from 0.65 in Anillidris Santschi to 2.3 in Leptomyrmex ) ( Shattuck 1992; Lucky and Ward 2010; Schmidt et al. 2013); 4.6 times in the Myrmicinae (from 0.28 in Metapone Forel to 1.3 in Aphaenogaster Mayr ) ( Alpert 2007; Shattuck 2008); 6.2 times in the Formicinae (from 0.34 in Cladomyrma Wheeler to 2.13 in Euprenolepis Emery ) ( Agosti et al. 1999; LaPolla 2009). The difference observed between the Armaniinae and Sphecomyrminae is much lower than these values ( Table S4). Thus, even in comparison with diverse modern groups, the Sphecomyrmini and Armaniinae can be viewed as one homogeneous entity.

Another important feature listed in the diagnosis of the Armaniinae is the shape of the petiole. In the Armaniinae , the petiole is broadly attached posteriorly to the gaster, which resembles the petiole of non-ant Vespoidea (e.g., Sierolomorphidae ), while in the Sphecomyrminae it is nodiform ( Dlussky 1999b). This may be the only remaining argument against changing the taxonomic status of the Armaniidae / Armaniinae , because such an attachment is unique and obviously plesiomorphic.

The drawings and photos of the Armaniinae , however, do not speak in favor of this conclusion. For example, in Orapia, Pseudoarmania , Armania curiosa the petiole looks rounded above and on the sides, not different from the petiole of the Sphecomyrminae . In the putative males of the Armaniinae (genera Archaeopone and Poneropterus ), the petiole is massive but again similar to that of the putative males of the Sphecomyrminae ( Sphecomyrma and Dlusskyridris). From the poorly preserved imprints of Khetania , Armania pristina , and A. capitata , one may conclude that the petiole does not appear to be differentiated from the gaster, which raises the question about the assignation of these specimens to the ants.

The only exception is a well-preserved imprint of Armania robusta with a relatively massive petiole, which seems, indeed, to be broadly attached to the gaster (although partly hidden by a coxa). Interestingly, Wilson (1987) considered this particular specimen as a gyne of Sphecomyrma , and its massive petiole as a sexually dimorphic feature, i.e., not as a subfamilial or even species-level character. Taking into account that there is no definition of a “broad attachment”, it is necessary to fill this gap by calculating the index PG/ PH in the A. robusta holotype as well as in several species of stem- and crown-group ants. The values of this index in the ants with a nodiform petiole are the following: 0.5 ( Gerontoformica subcuspis ), 0.6 ( Sphecomyrma freyi ), 0.7 ( Boltonimecia and gynes of Leptanilla ). The larger value 0.8 (indicating that the petiole is more broadly attached to the gaster) is found in A. robusta ; but in the workers and gynes of Stigmatomma (Amblyoponinae) values vary from 0.7 to 0.9. It is therefore impossible to reach the unambiguous conclusion that the petiole of A. robusta is a unique feature in terms of the width of its attachment to the gaster.

In summary, it is hard to disagree with Wilson (1987) that the Armaniinae do not have any subfamilial-level feature, to say nothing of a familial one. So my interpretation of the taxonomic status of the Armaniinae is similar to that of Grimaldi et al. (1997). Relatively well-preserved genera used in the statistical analysis ( Armania, Pseudoarmania , Orapia ) were transferred to Sphecomyrmini P. All the other genera were considered to be groups of uncertain taxonomic position within Formicidae P or Aculeata.

Figures

Figure 4. Bivariate plots (females). A) Scape length versus head length. B) Flagellum length versus head length. C) Scape length versus flagellum length. D) Pedicel length versus head length. E) Flagellomere 1 length versus head length. F) Flagellomere 2 length versus head length. (Note: Filled circles are crown-group ants; open circles are Sphecomyrmini; triangles are Haidomyrmecini; A - Armaniinae (sensu Bolton 2003); C - Cananeuretus occidentalis; G - Gerontoformica cretacica; M - Myanmyrma gracilis. Regression lines for crown-group ants (solid line), Sphecomyrmini (dashed line), and Haidomyrmecini (dash-dotted line) are shown in cases where statistical difference between at least two groups was found).