Introduction

The subfamily Dolichoderinae represents one of the oldest evolutionary stocks of the modern ant subfamilies: the oldest palaeontological records for a member of this subfamily belong to the Late Cretaceous (Canadian amber, Campanian, ca. 80 Ma) (Dlussky 1999), though the modern genera are known first from the middle Eocene deposits (ca. 45 Ma) (Wilson 1985a; Dlussky and Rasnitsyn 2002). The dolichoderine ants dominated numerically in the Eocene taphocenosis of Europe, but from the start of the Oligocene their proportion strongly decreased compared to the members of other modern subfamilies—Myrmicinae and Formicinae (Dlussky 2002, 2008; Dlussky and Rasnitsyn 2002, 2007). Now Dolichoderinae has worldwide distribution, but the highest generic and species diversity occurs in the Neotropical, Austrolasian, and Oriental Regions.

The genus Leptomyrmex Mayr, 1862 is highly specialized morphologically and is one of the most enigmatic genera among Dolichoderinae ants. Most probably, it forms a sister group for the rest of dolichoderines (Shattuck 1995; see also discussion below). Discovery of the fossil Usomyrma gen. nov. not only strongly support the uniqueness of the Leptomyrmex genus-group, but helps to throw light on questions concerning the evolution and zoogeography of the tribe Leptomyrmecini as a whole, that are not yet fully resolved.

Institutional abbreviations.—ZMUC, Zoological Museum of the University of Copenhagen, Denmark.

Other abbreviations.—AH, height of mesosoma seen in profile, measured from the upper level of scutum perpendicularly to the level of lower margin of mesopleuron; AL, diagonal length of mesosoma seen in profile, measured from the anterior-upper margin of pronotum to the posterior margin of propodeal lobes; HL, length of head in full-face view, measured in a straight line from the anterior point of median clypeal margin to mid-point of the posterior margin; HW, maximum width of head in full-face view behind the eyes; HTL, length of tibia of hind leg; OL₁, maximum diameter of eye; OL₂, minimal diameter of eye, PH, maximum height of petiole in profile; PL, maximum length of petiole seen from above from the posterodorsal margin of petiole to the articulation with propodeum; SCL, maximum length of scutum + scutellum seen from above; SCW, maximum width of scutum seen from above; SL, maximum straight-line length of antennal scape seen in profile.

Indices.—AI, AL/AH; CI, HL/HW; HTI, HTL/HL; OI₁, OL₁/HL; OI₂, OL₁/HW; OI₃, OL₁/OL₂; PI, PL/PH; SCI, SCL/SCW; SI₁, SL/HL; SI₂ , SL/HW.

Material and methods

We investigated two pieces of amber, each containing one male; both preserved in ZMUC (Danish [= Scandinavian] Amber). The figures are based on the original drawings of the specimens and photographs made using a Canon A640 digital camera fitted to an Olympus SZX9 stereomicroscope in conjunction with the computer software Corel Draw 8 program.

The specimens were measured (accurate to 0.01 mm) using a stereomicroscope Leica S6E, and the measurements were used to calculate the various indices (see Other abbreviations section). We measured as many features as possible on each specimen, given that not all were properly visible and measurable.

Systematic palaeontology

Fig. 1.

Dolichoderine ant Usomyrma mirabilis gen. et sp. nov., the holotype male, ZMUC, Bórge Mortensen coll., 28.2.1965, no. 380, from the Danish Amber, late Eocene; body in lateral view. Photograph (A), explanatory drawing (B).

Fig. 2.

Dolichoderine ant Usomyrma mirabilis gen. et sp. nov., the paratype male, ZMUC, Bórge Mortensen coll., 1.11.1964, no. 398, from the Danish Amber, late Eocene; explanatory drawings based on the original photographs, body in lateral view (A), head in lateral view (B), head in dorsal view (C), antenna (D).

Discussion and conclusions

The majority of Leptomyrmex workers are quite large, up to 10 mm, and slender, with very long antennae and legs; their bodies are often a quite bright orange colour or markedly bicoloured.

Nowadays, the distribution of Leptomyrmex is restricted to the eastern Australia, New Guinea (and in the adjacent Aru Islands and Ceram Island), and New Caledonia (with Loyalty Islands) (Wheeler 1934; Shattuck 1992; Smith and Shattuck 2009; Lucky and Ward 2010). They inhabit sclerophyll or rainforests in mountains or sea coasts, nesting mainly in soil or in wood remnants, and foraging predominately on the ground surface. Interestingly, when moving, the workers, which forage solitarily, fold their gaster vertically or somewhat forward, over the mesosomal dorsum (similar to many specialized members of Cataglyphis Foerster, 1850 or Oecophylla Smith, 1860 species). Leptomyrmex has sub-caste of plerergates, which make honey-storing role, like in American honey-ants Myrmecocystus Wesmael, 1838 or Palaearctic Proformica Ruzsky, 1902 (Formicinae), but such lifestyle is unknown among all other Dolichoderinae genera. Lastly, Leptomyrmex queens are ergatoid (except for a single known exception, see below) and differ from their workers mainly by the somewhat larger size, by the presence of ocelli and by the somewhat enlarged mesosoma (Wheeler 1934; Shattuck 1992; Lucky and Ward 2010); it should be noted that queens have been described only for seven “macro” Leptomyrmex species (Lucky and Ward 2010).

The preceding taxonomic revisions of Leptomyrmex were made by Wheeler (1915, 1934) and it was need of a modern revision. Then Smith and Shattuck (2009) described six socalled “micro” Leptomyrmex species from Australia, that at first were identified as members of the genus Iridomyrmex Mayr, 1862. Superficially, these species quite well differed from the “typical”, previously known Leptomyrmex species by their much smaller size, by their brownish body colour and by their less slender body and shorter appendages, etc. Shattuck (1992: 112) noted that “…several species (as yet undescribed) …have fully winged queens…”. Surprisingly, when they (Smith and Shattuck 2009) described the new “micro” Leptomyrmex species, they recorded a gyne (called by them a queen) and male only for the newly described L. burwelli. Except for the measurements and indices for one gyne and two males they provided no formal description of them, noting only: “Queen description. As described in Shattuck (1992). Male description. As described in Shattuck (1992), except the antenna lacks the distinctive bend between funicular segments 3 and 4” (Shattuck 1992: 61). Unfortunately, such a “description” tells us very little. Finally, a modern comprehensive revision of the “macro” Leptomyrmex species were made by Lucky and Ward (2010) who recognized 27 species in this genus.

As we describe a new genus based on males and obviously the most important information for our discussion are the characteristics of Leptomyrmex males. The male of this genus was described first by Emery (1891), and males of 18 “macro” Leptomyrmex and one “micro” Leptomyrmex species are known now (Lucky and Ward 2010). Emery (1913: 16) stressed that in Leptomyrmex males “La nervulation de l'aile antérieure présente une disposition qui ne se retrouve chez aucune fourmi”, and Wheeler (1934: 76) emphasized that “Unquestionably, the most singular and intriguing feature of the male is its wing venation”. The most peculiar in the fore wing venation are absence of the pterostigma and, in many species, the presence of a pterostigmal appendage—a short, often curved or twisted vein, arising from the costal vein in the area where there should normally be the pterostigma. Secondly, the closed cell 3r is very long, and most importantly, it is very narrow; all other closed cells are absent. Although Shattuck (1992) noted that rarely two “cubital” and one “discoidal” cells may be present (according to the nomenclature used by us, these are cells 1+2r and mcu, respectively), Lucky and Ward (2010) did not mentioned such features. Additionally, as discussed by Wheeler (1934), the frenal apparatus is vestigial and quite functionless. Such degeneration of the frenal apparatus combining with the reduction of the wing venation in the “macro-”Leptomyrmex males may be caused by the absence of true nuptial flight, because gynes are ergatoid and mating, probably, takes place inside the maternal nests or at most close to them on a ground.

Fig. 3.

Dolichoderine ant Usomyrma mirabilis gen. et sp. nov., the paratype male, ZMUC, Bórge Mortensen coll., 1.11.1964, no. 398, from the Danish Amber, late Eocene; explanatory drawings based on the original photographs, fore wing (A), mesosoma and petiole in dorsal view (B), hind tibia (C), genitalia in lateral view (D).

The antennae in Leptomyrmex males (except for two New Guinean species, see Lucky and Ward 2010) are 13-segmented with a rather short scape, but they are very long, subequal to or sometimes longer than body. The funicular segments, starting from the third one, are very long and thin, the third and fourth funicular segments are curved and forming an angle between them (except for L. burwelli from the “micro” Leptomyrmex group), the second funicular segment is straight (not curved) and varying in the length, but always it is shorter than the third one (Wheeler 1934; Shattuck 1992; Lucky and Ward 2010). The mandibles of Leptomyrmex males are well developed, elongate-triangular, with a distinct masticatory margin that has no teeth (except in some species an apical one) but in some cases has minute denticles. Generally, the body is long and slender, with a nodiform peduncle, and the legs are very long.

Based on the set of plesiomorphies and autapomorphies (see Remarks, above), we conclude that Usomyrma has in general, more archaic features and Leptomyrmex has the evolutionary derived morphology.

Our next question was: what is the limit of the tribe Leptomyrmecini? Emery (1913) included two genera—Leptomyrmex and Leptomyrmula Emery, 1913—when he first established the tribe (Leptomyrmula was established to accommodate the a fossil species from the Sicilian Amber, previously described as Leptomyrmex maravignae Emery, 1891). Until the early 1990's all subsequent authors considered Leptomyrmecini as the tribe sensu Emery, 1913. Then Shattuck (1992) reduced all tribes in the Dolichoderinae but his reasons for doing this were not clear and do not seem to be well supported. Ward et al. (2010) re-established tribe Leptomyrmecini, although their treatment of its generic composition cardinally differed from the earlier one (see below).

Baroni Urbani (1980) described Leptomyrmex neotropicus based on workers found in the Dominican Amber (Central America, Miocene) and in the same paper he recorded another fossil specimen (male) found in the same deposit, ascribing this very interesting specimen to the genus Leptomyrmex. Wilson (1985b) cast doubt on the taxonomic position of L. neotropicus, but in a subsequent paper Baroni Urbani and Wilson (1987) studied a second worker specimen and confirmed the attribution of this species to Leptomyrmex. In the same paper they also discussed the morphological peculiarities and possible taxonomic position of the “Leptomyrmex” male mentioned above; they wrote that L. neotropicus by the general appearance most resembles the extant L. fragilis (Smith, 1859), which is distributed in New Guinea, Ceram Island, and Aru Island, but differs from the latter by its smaller body size and some other morphological features (for details see Baroni Urbani and Wilson 1987: 2, 4, 5).

The characteristics of the “Leptomyrmex” male from the Dominican Amber, as given by Baroni Urbani and Wilson (1987), are most relevant to this discussion. The most important feature is again the fore wing venation: the pterostigma, pterostigmal appendages, and closed cells, including the cell 3r, are completely absent! In our opinion, this character of fore wing venation is probably sufficient to consider that this male belongs to a separate new genus. A similar suggestion was also made by Baroni Urbani and Wilson (1987: 5): “The leptomyrmecine male, whether Leptomyrmex neotropicus or not, may be sufficiently different from the living Leptomyrmex and Sicilian amber Leptomyrmula to warrant separation as a distinct genus”; but they went on to say “However, we do not believe it prudent to take this step until enough material has accumulated to make the worker-male association more probable, and to better estimate the extent of variation in both cases”, an opinion with which we agree.

So, how do we see the modern generic composition of the tribe Leptomyrmecini? Ward et al. (2010) in their very interesting and useful paper included 16 genera to the tribe Leptomyrmecini; six genera are distributed exclusively or chiefly in the Neotropics, and the other 10—in the Austrolasian and Oriental Regions. Concurrently, they excluded from this tribe the genus Leptomyrmula, even though relatedness of this genus to Leptomyrmex was never doubted by any of the preceding authors. Ward et al. (2010: 352) also stated that the earlier (e.g., Shattuck 1995) “morphological studies suggested that Leptomyrmex is sister to the rest of the Dolichoderinae, an arrangement not supported by our results”. It is quite obvious that Ward et al. (2010: 361) reached this conclusion based almost exclusively on molecular-genetic analysis and that morphological features were not taken into account in the phylograms. Moreover, somewhat inconsistently with their own molecular results, they have written in the diagnosis of the tribe Leptomyrmecini: “Morphologically heterogeneous assemblage, recognized primarily by disagreement with the 3 preceding diagnoses (i.e., diagnosis of the tribes Dolichoderini, Tapinomini, and Bothriomyrmecini—our remark)”. In our opinion, the Ward and co-authors' treatment of the tribe Leptomyrmecini is too wide and should be reassessed based on morphological analysis or perhaps combined morphological and molecular-genetic analysis.

The complete revision of the subfamily Dolchoderinae at the suprageneric level lies far outside the main task of this paper and at the moment we do not have sufficient data to reassess formally the generic composition of the Leptomyrmecini sensu Ward et al. (2010). Nevertheless, based on morphological evidence (obviously there is no molecular genetic data for fossil amber specimens) we propose as a first step towards such a revision, to group together the extant genus Leptomyrmex and the fossil genera Usomyrma, Leptomyrmula, and “Leptomyrmex” male specimen from the Dominican Amber, to establish informal Leptomyrmex genus-group within the tribe Leptomyrmecini. The main synapomorphies for this genus-group (based on the males morphology) are the peculiarities of the fore wing venation, particularly loss of the vein section Rs+M (which led to the fusion of the cells [1+2r]+mcu in Usomyrma) with the further reduction of the central vein sections and cells in the other genera. Such wing venation is unique not only for the subfamily Dolichoderinae, but for all known ants at all.

Fig. 4.

Dolichoderine ant Leptomyrmula maravignae (Emery, 1891), the holotype male, from the Sicilian Amber, body in lateral view (after Emery 1891).

We suggest the following evolutionary trend in the fore wing venation within this group. The most plesiomorphic, i.e., putatively ancestral to other genera of this genus-group, is the venation of Usomyrma that has the well developed pterostigma, the closed central cell and the relatively wide cell 3r (Figs. 1A, B, 3A). We suggest that the next evolutionary step was the complete reduction of the closed central cell in Leptomyrmula while conserving the pterostigma (though it is somewhat smaller than that in Usomyrma) and a wide closed cell 3r (Fig. 4). Then, in the modern Leptomyrmex males the pterostigma was reduced and the cell 3r became very narrow (see Lucky and Ward 2010). Finally, the most apomorphic condition is illustrated by the venation of “Leptomyrmex” male from the Dominican Amber, which has no pterostigma and pterostigmal appendages combined with the absence of the closed cell 3r (see Baroni Urbani and Wilson 1987; Fig. 2).

Ward et al. (2010) proposed the following scenario for the origin and further spread and diversification of the tribe Leptomyrmecini (“s.l.”): the tribe arose in the Neotropics about 50 Ma, and two subgroups of leptomyrmecines (including Leptomyrmex) later colonized Australia. The older of these, Leptomyrmex, is a sister to the New World clade of (Forelius Emery, 1888 + Dorymyrmex Mayr, 1866). Based on the latter, they supposed that “Leptomyrmex” from the Dominican Amber (15–20 Ma) represents an extinct stem lineage of (Forelius + Dorymyrmex) in America. Their data also suggested the average time of the origin of the Leptomyrmex stock ca. 37 Ma, with the oldest possible age being somewhat more than 40 Ma. Our finding of Usomyrma in the late Eocene amber (ca. 37 Ma), and the modern dating of the Sicilian Amber (where Leptomyrmula was found) as not the Miocene but the early Oligocene (Skalski and Veggiani 1990; 29–34 Ma), does not contradict the timing proposed by Ward et al. (2010).

Fig. 5.

Map of distribution of fossil species Usomyrma gen. nov. (triangle), Leptomyrmula Emery, 1913 (circle), “Leptomyrmex” male (square) and of extant species Leptomyrmex Mayr, 1862 (dots).

On the other hand, if one accepts our treatment of the Leptomyrmex genus-group (above), then this group probably does not have an American origin (see also Fig. 5). The main reasons are: the representative of the oldest stem and most primitive morphologically, is the genus Usomyrma (Danish Amber, late Eocene, ca. 37 Ma), the representative of the next evolutionary step, Leptomyrmula, was found in Sicilian Amber (early Oligocene, 29–34 Ma) while modern Leptomyrmex (Austrolasian Region) is more derived. The only known “Leptomyrmex” male from the America (Dominican Amber, Miocene, 15–20 Ma) is, in our opinion, the most derived. Under this scenario, Ward et al.'s (2010) supposition that South America was a possible place of the origin of Leptomymecini (and modern Leptomyrmex) with their later dispersion to Australia seems unlikely. We think it more probable that the ancestor of Leptomyrmex penetrated from the Europe during the Eocene (or even earlier) to Australia.

If so, what was the way of dispersion of this ancestor? As there are no strict evidences for a such way, one may propose two alternative scenarios.

The first one claims that the ancestor of Leptomyrmex might penetrate to Australia via South-East Asia. We are well aware that there were no direct land connections between South-East Asia and Australia (e.g., see Sinitsyn 1962, 1965), but species might cross the relatively narrow straits between Soenda Isles, New Guinea, and Australia. The argument that the ergatoid queens observed in modern Leptomyrmex species might indicate poor dispersion among ancestral species is very doubtful: ergatoid queens might evolve much later after Leptomyrmex reached Australia. Furthermore, the indirect evidence of Aenictinae species distribution supports our suggestion: all species of Aenictus Shuckard, 1840 have wingless gynes and highly specialized physogastric queens, but there is no doubt the genus has an Asian origin but occurs also in Australia and New Guinea. Then Leptomyrmex penetrated to South America, where the most advanced morphologically form (i.e., “Leptomyrmex” male from the Dominican Amber) arose, following the mode suggested by Ward et al. (2010), but in the reverse direction: i.e., Leptomyrmex species may have migrated from Australia to South America via Antarctica, which was connected by land masses until the late Oligocene (Sanmartín and Ronquist 2004).

On the other hand, there are several facts that at least partly contradict such scenario. Perhaps the most important is that there are no any known either modern or fossil representatives of the Leptomyrmex genus-group in South and South-East Asia, despite the habitat conditions in these regions both in a past and present seem quite suitable for them. Moreover, despite we have no evidence to positively answer the question where genus Leptomyrmex arose: in Asia, New Guinea or in Australia, most probably the place of its origin was Australia. This may be supported by the fact that in New Guinea lives more derived morphologically species than in Australia (see also Lucky and Ward 2010). Hence, the alternative way of dispersion of the ancestor of Leptomyrmex might be from Europe to America via so-called De Geer's Land Bridge in the late Eocene (see McKenna 1983; Tiffney 1985), and then—to Australia (by the way from America via Antarctica as proposed by Ward et al. 2010). In such a case, the Dominican Amber “Leptomyrmex” might evolve later, but extinct after Miocene.

Whichever, the species radiation and diversification of the genus took place in Australia.

In conclusion, one can find many examples when the modern distribution of extant groups of animals does not correspond with the paleontological evidence of distribution. One example among ants may be the “bulldog ants”, subfamily Myrmeciinae. Their modern distribution is restricted to Australia and New Caledonia but fossils suggest a much wider distribution in the past: a couple of fossil genera were found in the middle and late Eocene deposits of the North (USA and Canada) and South America (Argentina), and Europe (Denmark, Germany) (Dlussky and Perfilieva 2003; Archibald et al. 2006; Dlussky 2012).