Newly discovered sister lineage sheds light on early ant evolution

Abstract

Ants are the world's most conspicuous and important eusocial insects and their diversity, abundance, and extreme behavioral specializations make them a model system for several disciplines within the biological sciences. Here, we report the discovery of a new ant that appears to represent the sister lineage to all extant ants (Hymenoptera: Formicidae). The phylogenetic position of this cryptic predator from the soils of the Amazon rainforest was inferred from several nuclear genes, sequenced from a single leg. Martialis heureka (gen. et sp. nov.) also constitutes the sole representative of a new, morphologically distinct subfamily of ants, the Martialinae (subfam. nov.). Our analyses have reduced the likelihood of long-branch attraction artifacts that have troubled previous phylogenetic studies of early-diverging ants and therefore solidify the emerging view that the most basal extant ant lineages are cryptic, hypogaeic foragers. On the basis of morphological and phylogenetic evidence we suggest that these specialized subterranean predators are the sole surviving representatives of a highly divergent lineage that arose near the dawn of ant diversification and have persisted in ecologically stable environments like tropical soils over great spans of time.

Since the Cretaceous, ants have evolved to become one of the most diverse and abundant animal families to ever appear on Earth (1–5). A robust phylogeny is indispensable for elucidating the evolutionary origin of ants and for exploring the selective forces that have produced their extraordinary specializations. Previously published studies, however, led to contradicting views of early ant evolution, in part because of high levels of morphological convergence, the secondary loss of characters, and a lack of informative paleontological data (5–14). As a result, numerous taxa have been proposed as the most basal lineage (15–23). Two recent, comprehensive molecular studies have reconstructed the formicid phylogeny, resulting in largely congruent topologies that agree that ants as a family and most of the subfamilies are monophyletic (20, 21). In contrast with previous morphological studies (5–14), the blind, subterranean subfamily Leptanillinae was recovered as sister to all extant ants, as suggested by two earlier partial molecular studies (22, 23), implying an early origin and diversification of cryptic, blind species with hypogaeic foraging habits.

Results and Discussion

Here we describe a new ant species, Martialis heureka (Figs. 1 and 2), comprising a new subfamily, the Martialinae, and discuss the implications of this discovery for the early evolution of ants. Morphological results confirm that M. heureka is clearly a member of the family Formicidae because of the presence of the metapleural gland orifice, geniculate antennae, and a morphologically differentiated petiole (Figs. 1 and 2). However, it cannot be placed within any extant or extinct subfamily (20–22, 24). M. heureka is close to, but not within, the subfamily Leptanillinae, because it exhibits several autapomorphies (see diagnosis below) and retains the plesiomorphic condition for other morphological characters, like the flexible promesonotal suture and the exposed antennal sockets. Hence, we place M. heureka in its own subfamily, the Martialinae.

• Taxonomic Treatment

• Family Formicidae Latreille, 1809 (25)

• Subfamily Martialinae Rabeling & Verhaagh, subfam. nov.

• Martialis heureka Rabeling & Verhaagh, gen. et sp. nov.

• Martialinae Rabeling & Verhaagh, subfam. nov.

Fig. 1.

Holotype worker of Martialis heureka gen. et sp. nov. The single specimen has been collected in the leaf litter of a terra firme rainforest near Manaus, Amazonas, Brazil. Martialis heureka is inferred to be the sister lineage to all extant ants. (A) Lateral and (B) dorsal view of the worker. (Scale bar: 1 mm.) Photographs courtesy of C. Rabeling and M. Verhaagh.

Fig. 2.

Worker of Martialis heureka foraging in a subterranean tunnel. This drawing depicts M. heureka in its assumed natural habitat as inferred from its external morphology. Please see the Inferred Biology section for a more detailed discussion of the species' biology. (Scale bar: 1 mm.) Color pencil drawing by Barrett A. Klein.

Worker diagnosis: small, blind, pale, presumably subterranean ants with the following combination of characters. Putative apomorphic conditions are marked by an asterisk. (1) Prementum partly visible with closed mouthparts, not entirely concealed behind the broad labrum and stipites; (2*) clypeus highly reduced; (3*) eyes absent; (4) frontal lobes absent; (5*) antennal sockets fully exposed and distant from each other, positioned at anterior margin of clypeus and projecting anteriorly beyond it; (6*) toruli raised vertically, forming cups to hold condylar bulbs; (7) antennae 12 segmented; (8) promesonotal suture present and flexible; (9) propodeal lobes absent; (10) metapleural gland orifice slit shaped, dorsal orifice margin projecting slightly forward, but not overhanging or concealing opening; (11*) metacoxal cavity closed, with a complete cuticular annulus surrounding the cavity; (12) metasternal process absent; (13*) petiole tergosternally fused, without a trace of a suture between tergite and sternite; (14*) abdominal segment III broadly attached to segment IV but distinctly smaller than the latter, tergosternally fused; (15) helcial tergite well developed, posterior part with a girdling constriction; (16) abdominal segment IV not tergosternally fused, posttergite broadly overlapping abdominal segment V; (17) stridulitrum on presclerite of abdominal segment IV absent; (18) spiracles of abdominal segments I–IV exposed, of V–VII concealed by preceding postsclerites; (19) pygidium (tergite of abdominal segment VII) well developed, simple, neither armed with teeth or setae nor heavily sclerotized nor otherwise modified; (20) sting present; (21) metatibial gland absent; (22) tarsal claws simple, without preapical teeth.

• Queen, male, and larva are unknown.

• Martialis Rabeling & Verhaagh, gen. nov.

• Type and only known species: Martialis heureka Rabeling & Verhaagh, sp. nov. (described below).

Worker diagnosis: including the characteristics of the subfamily description given above. (1*) Mandibles elongate, slender, forceps-like, not crossing at rest (although doing so in dried condition), without differentiated basal and masticatory margins; (2) labrum broad with a U-like insertion at apical margin; (3*) clypeus reduced to a narrow strip at the sides, with a small trapezoidal median portion between antennal sockets, densely covered with a brush of setae; (4) metanotal groove distinct; (5) propodeum evenly rounded at the transition from basal (= dorsal) to declivitous face; (6) propodeal spiracle orifice not enlarged and nearly equally distant from basal and declivitous face of propodeum and the anterior part of metapleural opening, with circular orifice directed posteroventrally; (7) petiole with short peduncle, tergosternally fused, without a trace of a suture between tergite and sternite; (8*) presclerite of abdominal segment IV well developed, posterior part with a girdling constriction, forming the posterior constriction of a weakly defined postpetiole; (9) sting present, but small; (10) mesotibial spur present, simple; (11) metatibial spur present, shortly pectinate.

Martialis heureka Rabeling & Verhaagh, sp. nov.

Worker description (holotype and only known specimen). Measurements: HW 0.65 mm, HL 0.62 mm, SL 0.46 mm, FL 1.03 mm, ML 0.90 mm, WL 1.02 mm, PW 0.40 mm, PEW 0.19 mm, PEL 0.27 mm, PPW 0.34 mm, PPL 0.30 mm, HFL 0.60 mm, HTL 0.57 mm, CI 105, MI 145, SI 72, DI 45 (See Table S1 for definitions and abbreviations of morphometrics). Including the characters of the subfamily and genus description given above: small (HW 0.65, WL 1.02), pale yellow, and blind, integument in dried condition partly translucent. Very long, slender, forceps-like mandibles (ML 0.90, MI 145) inserted on outer anterior margin of head capsule, projecting straight forward to ≈5/7 of its length, then curving slightly mesally. Inner margin bearing a tiny sudmedian tooth at ≈2/5 of its length, measured from the base and a group of 3 subapical teeth at ≈5/7 of its length: a median larger tooth and a somewhat smaller tooth to each side of it. The subapical teeth and the sharply pointed mandibular tip frame an oval space. Mandibles crossed in dried condition (Fig. 1), not crossing in live individual (Fig. 2) or while stored in 80% EtOH. Head capsule bulbous in lateral view, narrowing toward posterior margin. Clypeus reduced, narrow strip between lateral head margin and wider between antennal sockets, covered with at least 20 setae that project forward, beyond the margin of antennal sockets, resembling a brush. Antennal scape relatively short (SI 72, DI 45), 1st and 2nd funicular segment ¾ and ¼ longer than 3rd funicular segment. Funiculus more than twice as long as scape. Mesosoma long and slender. Promesonotal suture present, pronotum and mesonotum presumably capable of movement relative to each other. Pronotum forming a slender cervix. Front legs enlarged: procoxae twice as long and wide as meso- and metacoxae; profemura and tibiae also enlarged. Petiole subsessile with a short anterior peduncle; petiolar node rounded, smooth, with a short sloping posterior face; no teeth or projection ventrally. Metasoma (abdominal segment IV–VIII visible) laterally compressed, drop shaped in lateral view. Head, pronotum, and legs densely covered with erect to suberect hairs and sparsely with longer erect setae; mandibular pubescence dense, consisting of short suberect hairs. Inner margin with two rows of at least 18 long straight setae, which are arranged pair wise. Propodeum without any hairs; petiole with few suberect setae on dorsal surface of node and abdominal segment III–VIII with irregularly spaced long erect setae. Dense appressed pubescence absent from entire body. Only few body parts bear distinct sculpturing: neck, mesonotum, propodeum, and ventral surface of petiole punctate; lateral surface of propodeum faintly striate.

Holotype and Locality.

Brazil: Amazonas, Manaus. Headquarters of Empresa Brasileira de Pesquisa Agropecuária (EMBRAPA)-Amazônia Ocidental, located at kilometer 28 of highway AM 010; 2°53′S, 59°59′W; elev. 40–50 m; 09 May 2003; col. C. Rabeling; ex leaf litter at dusk, primary tropical lowland rainforest. The holotype is deposited in Museu de Zoologia da Universidade de São Paulo, Brazil (MZSP).

Etymology.

The genus name refers to the unknown combination of aberrant morphological characters, which led Stefan P. Cover and Edward O. Wilson to the conclusion that this ant has to be from the planet Mars; hence, the genus name Martialis (gr.: “of Mars” or “belonging to Mars”). The species epithet heureka (gr.: I found it!) epitomizes the troubled story of the species' rediscovery. Five years after two workers were discovered by M. Verhaagh in a soil sample and subsequently lost, a single worker, the present holotype, was recollected in a nearby patch of primary rainforest.

Discussion

We assume that the present specimen of M. heureka is a worker, because it lacks ocelli, the enlarged mesosoma and the extra sclerites associated with wings (Figs. 1 and 2). In addition, the specimen was collected in the leaf litter suggesting foraging activity usually performed by the worker caste. However, queens with worker-like morphology have been reported from several poneroids and socially parasitic Formicinae and Myrmicinae (26–29). Because we did not perform a dissection of the single specimen no statement can be made about palpal segmentation or internal anatomy.

Inferred Biology.

On the basis of the specimen's external morphology, we are able to infer some aspects of the species' biology. The pale integument and the absence of eyes suggest that M. heureka lives hypogaeically or in covered low-light environments, like leaf litter or rotting wood. The fact that the first two M. heureka individuals were collected in soil core samples during the day, and the present specimen in leaf litter at dusk, supports this hypothesis. Possibly, M. heureka surfaces during the night to forage. The unusually enlarged procoxae and long front legs could potentially be an adaptation to prey capture. Presumably, they are used less for digging activities, because the legs are relatively thin and lack the characteristic erect setae of actively digging species. We speculate that M. heureka might take advantage of preexisting underground cavities, like hollow rotten roots (Fig. 2). The forceps-like mandible type is not seen in any other ant species. These long, filigree instruments could be used to drag soft prey items out of cavities. Annelids, termites, insect larvae, and other soft-bodied arthropods are possible prey. We do not expect M. heureka to prey on heavily sclerotized invertebrates.

Phylogenetic Study.

To infer the phylogenetic position of M. heureka and to evaluate competing phylogenetic hypotheses, we sequenced 4.2 kb of Martialis heureka's nuclear ribosomal and single-copy nuclear DNA. DNA sequence data were added to a previously published molecular data set of 151 ant species from 20 subfamilies and 11 aculeate outgroup taxa (20), except for four single-copy nuclear genes, which we could not amplify from the limited amount of DNA extract. The phylogeny was inferred using maximum likelihood (ML, 30) and Bayesian inference (BI, 31). Both approaches place Martialis as sister to the remaining extant ants and support the monophyly of all ants, including Martialis (Fig. 3). Martialis did not associate with any extant subfamily in the molecular phylogeny, supporting the designation of the Martialinae as a distinct subfamily on morphological grounds. The basal position of Martialis had a Bayesian posterior probability (BPP) of 0.912, a maximum likelihood bootstrap proportion (ML BP) of 0.76 (Fig. 3), and was strongly supported by Bayes factors (BF, 32–34) over the next most strongly supported rooting (Leptanillinae as basal to all extant ants; 2ln(BF) = −16.4; Fig. 4). However, the more conservative Shimodaira-Hasegawa (SH, 35, 36) test did not reject the alternative rootings (Fig. 4). Even though the BPP and ML BP were only moderately high, the basal positions of Martialis and the Leptanillinae were robust to taxon sampling (Figs. S1 and S2). When Martialis was removed from analyses, the leptanillines were recovered as sister to all extant ants. The converse was true when the leptanillines were removed. Because the basal position of neither Martialis nor the leptanillines changed in the absence of the other, we suggest that their basal position is not because of a particular taxon-sampling scheme.

Fig. 3.

Maximum likelihood (ML) tree with Bayesian posterior probabilities (BPP) and ML bootstrap proportions (ML BP) support values. Tree is rooted on the branch leading to the outgroup, Pristocera. The formicoid clade has been collapsed to increase resolution of relationships among basal ant groups (Martialis, Leptanillinae, and poneroids). Bipartitions with strong Bayesian support are indicated by blue triangles (BPP = 1.0), green circles (0.95 < BPP < 1.0) or orange rectangles (0.9 < BPP < 0.95).

Fig. 4.

Alternative rootings evaluated with Bayes factors (BF) and the Shimodaira–Hasegawa (SH) test. Constraint trees used to evaluate alternative rootings are depicted on the left. Rootings 1–3 correspond to the first rooting considered by Brady et al. (20), but with alternative arrangements of Martialis and Leptanillinae. Rootings 4–7 correspond to Brady et al.'s rootings 2–5. For each alternative rooting, the corresponding value of 2ln(BF) is given for the comparison of that rooting against the maximum posterior probability (MPP) rooting 1. Values lower than −10 can be interpreted as strong evidence against the alternative. P values resulting from the comparison of alternative rootings to the maximum likelihood (ML) rooting 1 using the SH test are also given. Rootings significantly worse than the ML rooting would have P values < 0.05.

The basal position of Martialis is not likely caused by long-branch attraction (LBA), as was previously suggested of the leptanillines (20), because Martialis has a substantially shorter branch than all outgroup taxa and the Leptanillinae (Fig. 3). Additionally, the ribosomal DNA sequences in this study (87% of our data set) are the most slowly evolving of the genes used by Brady et al. (20) (58% of their data set) and are least likely to contribute to LBA artifacts (37). Taxon sampling was also increased by one important lineage over previous studies, breaking the long branch that subtended the extant ants in previous studies (20, 21). However, the reconstruction of the molecular ant phylogeny still faces at least two sources of uncertainty that could introduce analytical artifacts. First, even though Martialis breaks the long branch leading to the remaining extant ants, the branch connecting the ingroup to the aculeate outgroup taxa remains long (Fig. 3). Second, the basal ant lineages seem to have originated in a relatively short period (20, 21), potentially making the unambiguous resolution of their relationships quite difficult and sensitive to methodological error. Increased gene and taxon sampling for both ants and outgroup aculeates should reduce the potential for statistical artifacts in future analyses.

Our phylogeny supports paraphyly of the poneroids (Fig. 3). The Ponerinae were strongly supported as sister to the formicoids by Bayesian analysis (BPP = 1), although less strongly by ML bootstrap analysis (ML BP = 68). Relationships among the remaining poneroid subfamilies were more ambiguous. The best estimate from both ML and Bayesian analyses (Fig. 3) has an Amblyoponinae/Agroecomyrmecinae clade basal to all other poneroids, followed by a Paraponerinae/Proceratiinae clade, although support is weak (BPP = 58, ML BP <50). Paraphyly of the poneroids was recovered, even if the outgroups, Martialis, the leptanillines, both the outgroups and Martialis, or both the outgroups and the leptanillines were removed from the analysis (Figs. S1 and S2).

Our phylogenetic analyses, combined with the inferred biology of M. heureka, suggest that the most basal extant ant lineages are cryptic, hypogaeic foragers, rather than wasp-like, epigaeic foragers (Fig. 3). This finding is congruent with recent molecular studies (20–23), which previously suggested the Leptanillinae, another subfamily of subterranean predators, to be sister lineage to all extant ants. This result has puzzled ant systematists for two reasons. First, Wilson et al.'s (16, 17) classic study of the Mesozoic amber ant Sphecomyrma postulated that the ancestral ant was a large-eyed, wasp-like, ground forager, creating a strong expectation that the most basal extant ant lineages would also be epigaeic foragers, presumably similar to Sphecomyrma. Second, the Leptanillinae share common morphological (10, 38) and behavioral (39, 40) characteristics with the Amblyoponinae, implying the monophyly of this group (20). In contrast, our results and recent molecular systematic studies (20–23) suggest that blind, subterranean, specialized predators, like Martialis, the Leptanillinae, and some poneroids, evolved early during ant diversification. We hypothesize, that once these hypogaeic predators adapted to their specialized subterranean environment, their morphology and biology changed little over evolutionary time because their hypogaeic habitat has likely been ecologically stable and provided a refuge from competition with other, more recently evolved, ants. It is important to note that no definitive statement about the morphology and life history of the ancestral, Mesozoic ant can be derived from our current knowledge about the surviving basalmost ant lineages, because the relative probabilities of evolutionary transitions between epigaeic and hypogaeic habits are uncertain. On the basis of the combined evidence of morphological, phylogenetic, and Mesozoic fossil data, the possibility remains that the ancestral ant was a large-eyed, wasp-like, epigaeic forager (16–19), rather than a blind, specialized, hypogaeic predator.

The Mesozoic fossil Sphecomyrma freyi has been interpreted as a representative of the stem-group ants and ant ground-plan characteristics have been derived from its morphology, suggesting that the most recent common ancestor (MRCA) of ants could be similar to tiphioid wasps (16–18). By considering both the topology of our inferred phylogeny (Fig. 3), and the timing of ant diversification presented in previous studies (20, 21), the Sphecomyrminae could alternatively be placed within the subfamilies grouped in the formicoid clade or as stem-group members of the formicoid subfamilies rather than the Formicidae as a whole (16–19). However, the morphology-inferred sister group relationship of Sphecomyrma to all extant ants (14) and the absence of hypogaeic ant fossils in Cretaceous deposits (11, 24, 41–50) do not support this hypothesis.

Conclusion

Martialis heureka exhibits a mosaic of plesiotypic ant characters and derived morphological adaptations to its hypogaeic predaceous lifestyle. The discovery of Martialis and our phylogenetic analyses support the hypothesis that several of the most basal extant ant lineages are hypogaeic foragers, suggesting that hypogaeic foraging evolved at least thrice independently during the early evolution of ants or that the ancestral ant was a hypogaeic forager. These evolutionary scenarios are congruent with the dynastic succession hypothesis for ant evolution (5, 20, 21), proposing a ground-associated ant ancestor, derived from a wasp-like, aculeate predator that radiated into specialized soil, leaf litter, and arboreal habitats potentially coinciding with the diversification of the angiosperms. Martialis and other soil-dwelling basal ants might therefore be relicts of early ant diversification, which survived in an environmentally stable refuge habitat, permitting them to retain plesiotypic characteristics because of reduced selective pressures. The exact nature of the ancestral ant remains uncertain, given that the propensity for repeated evolution of a hypogaeic lifestyle may be higher than for reevolution of an epigaeic lifestyle. Interestingly, M. heureka is a unique Neotropical representative of an extant, basal ant lineage. This discovery hints at a wealth of species, possibly of great evolutionary importance, still hidden in the soils of the remaining rainforests. Nocturnal leaf litter sampling and subterranean traps would be promising additions to future biodiversity surveys.

Materials and Methods

Photomicrographs were taken using a JVC KY-F70 digital camera mounted on a Leica Z6 APO dissecting scope. Composite images were assembled from 20 sequential photographs with Syncroscopy AutoMontage (v 5.0) software. DNA was extracted from the right front leg using a QIAGEN DNA Micro Kit. Two nuclear ribosomal genes and one single-copy nuclear gene were amplified using conventional PCR methods and sequenced (20). We obtained a total of 4926 aligned bp: 1904 bp for 18S, 2505 bp for 28S, and 517 bp for elongation factor 1 F2. We failed to sequence additional genes from the limited amount of DNA extract. The obtained DNA sequences were aligned with the data matrix of Brady et al. (20).

The phylogenetic position of Martialis was estimated using both ML and BI. ML trees were inferred using Garli v0.951 (30) with a single model of evolution. Bayesian phylogenies were inferred using MrBayes v3.2 (31) with a four-partition model (1st and 2nd codon positions of EF1αF2, 3rd codon positions of EF1αF2, 18S, and 28S). Convergence of Bayesian analyses was assessed with the program MrConverge (34). Choices for sequence evolution models (all GTR + I + Γ) were taken from Brady et al. (20) for accuracy of comparison with their results and because our taxon-sampling schemes are nearly identical. Adequacy of the model of sequence evolution was evaluated with Bayesian posterior predictive simulation (51) using PuMA v0.903 (see Fig. S3) (52). Phylogenies were also estimated, using identical methods, when excluding Martialis, the Leptanillinae, the outgroups, both Martialis and outgroups, or both the Leptanillinae and outgroups. Alternative rootings were evaluated using Bayesian bipartition posterior probabilities, Bayes factors (32–34), and the Shimodaira-Hasegawa test (35, 36). For more detail on phylogenetic analyses and results, see SI Materials and Methods.