A new genus of hell ants from the Cretaceous (Hymenoptera: Formicidae: Haidomyrmecini) with a novel head structure
Abstract
An unusual Cretaceous trap jaw ant is described from Burmese amber dated to the Late Cretaceous. Linguamyrmex vladi gen.n. sp.n. is distinguished by an unusual suite of morphological characters indicating specialized predatory behaviour and an adaptive strategy no longer found among modern ant lineages. The clypeus, highly modified as in other closely related haidomyrmecine hell ants, is equipped with a paddle-like projection similar to Ceratomyrmex. X-ray imaging reveals that this clypeal paddle is reinforced, most probably with sequestered metals. Presumably this fortified clypeal structure was utilized in tandem with scythe-like mandibles to pin and potentially puncture soft-bodied prey. This unique taxon, which stresses the diversity of stem-group ants, is discussed in the context of modern and other Cretaceous trap jaw ant species.
This published work has been registered in ZooBank, http://zoobank.org/urn:lsid:zoobank.org:pub:40D636A3-4D88-470A-BC5B-85ABFD1A49E2.
Introduction
The earliest definitive ants range in age from approximately 100 to 78 Ma and are described from six fossil deposits throughout Laurasia (Wilson et al., 1967a; Dlussky, 1975, 1996; Wilson, 1985; Nel et al., 2004) and a single locality in present-day Botswana (Dlussky et al., 2004). Only two of the 45 named ant species from the Cretaceous are unambiguously assignable to extant lineages (belonging to crown-group Formicidae; Barden, 2017); the majority of early species are distinct from modern lineages – they are composites of modern features and plesiomorphic traits. All definitive Cretaceous ants possess a constricted waist segment, or petiole, and a distinct gland (the metapleural gland) visible as an opening on the metapleuron (LaPolla et al., 2013). At the same time, many early ants exhibit shortened antennal scapes and often a very demarcate metanotal sclerite not found in modern species; these features are probably plesiomorphic. The phylogenetic position of noncrown group Cretaceous ants has been tested three times; however, only recently has there been an assessment of monophyly (Grimaldi et al., 1997; Ward & Brady, 2003; Barden & Grimaldi, 2016). There is now evidence that nearly all described species from the Cretaceous are stem-group taxa – they are paraphyletic with respect to crown ants and so belong to lineages that arose prior to the common ancestor of all extant ants. Ultimately stem ants became extinct, sometime between the Late Cretaceous and Paleocene, and so the only window into these lineages, their morphology and behaviour, remains the study of fossil specimens (Barden & Grimaldi, 2016).
- Haidomyrmex cerberus Dlussky – the first hell ant was named 20 years ago by the prolific paleomyrmecologist Gennady Dlussky (Dlussky, 1996). Haidomyrmex cerberus was described from a single partial worker specimen, providing the basis for what are now tribal synapomorphies for the Haidomyrmecini, established by Bolton (2003) and revised by Perrichot et al. (2016): unusual L-shaped mandibles and a bulging clypeus possessing a clypeal brush comprising dense, stout setae. Locality: Burmese amber, ∼98.8 Ma, Kachin State, Myanmar (Shi et al., 2012)
- Haidomyrmodes mammuthus Perrichot, et al. – described from both partial gyne and worker specimens, H. mammathus is the only alate specimen yet described (Perrichot et al., 2008). In addition, the type material is the only reported conspecific worker and putative queen caste for any Cretaceous taxon and provides important information regarding sociality in early ants. The description of multiple specimens from an additional locality confirmed beyond any doubt that the highly unusual features observed in H. cerberus were not preservational. Largely similar to H. cerberus, H. mammathus possesses ocelli, which have long been presumed to be an ancestral ant trait (Wilson et al., 1967b). Locality: Charentese amber, ∼100 Ma, Charentes, France (Perrichot et al., 2010)
- Haidomyrmex scimitarus, Haidomyrmex zigrasi Barden & Grimaldi – reconstructions based on two entirely complete dealate (H. scimitarus) and worker (H. zigrasi) specimens suggested that the enigmatic L-shaped mandibles probably articulated, at least partially, in a vertical plane. In particular, the placement of trigger hairs on the clypeus as well as the hypognathous-like orientation of the head (all other non-haidomyrmecine ant heads are prognathous; Keller, 2011) suggested that hell ants were Cretaceous analogues to modern trap-jaw predators (Barden & Grimaldi, 2012). H. scimitarus is also the largest haidomyrmecine, measuring over twice the total length of most other species. Locality: Burmese amber
- Haidoterminus cippus McKellar, et al. – the youngest known haidomyrmecine, H. cippus is described from a single worker specimen. This species extends the temoporal and geographic range of hell ants considerably, indicating that the specialized mouthparts and probable trap-jaw behaviour were evidently successful for some time and over a wide area of Laurasia. Haidoterminus cippus is also unique among haidomyrmecines, possessing an elongate antennal scape somewhat similar to modern ant taxa. Locality: Medicine Hat amber, ∼78.5 Ma, Alberta, Canada (McKellar et al., 2013)
- Ceratomyrmex ellenbergeri Perrichot, et al. – an exceedingly unusual taxon described from four worker specimens of varying completeness. Retaining the same overall Cretaceous trap-jaw bauplan as other haidomyrmecines, C. ellenbergeri possesses a unique clypeal horn coated ventrally in elongate, tapered setae. The mandibles are correspondingly very long, apparently extended to pin prey between the mandibular points and lengthy clypeal projection. This configuration is a dramatic variant on the trap-jaw syndrome and would appear to confirm vertical mandibular articulation. Locality: Burmese amber.
Of the six named haidomyrmecine species, four are from Burmese amber. As with other insect groups [Grimaldi et al., 2009; e.g. camouflage in Neuroptera and Reduviidae (Wang et al., 2016); social parasitism in staphylinid beetles (Yamamoto et al., 2016); apomorphic symphytan wasps (Engel et al., 2016); brood care in scale insects (Wang et al., 2015)] and even vertebrate groups [tropical lizards (Daza et al., 2016); nonavian theropods (Xing et al., 2016)], this deposit has made a substantial contribution towards the reconstruction of the evolutionary history of ants. Here, we describe an additional unusual haidomyrmecine ant morphotype from Burmese amber, further expanding the complex history of stem-group ants.
Materials and methods
All studied material was mined from the Hukawng Valley in Kachin State, Republic of the Union of Myanmar. The amber or ‘burmite’ recovered from this deposit is dated to the Cenomanian (∼98 Ma) based on radiometric uranium–lead dating (Shi et al., 2012).
Light microscopy equipment included a Leitz Wetzlar stereoscope with 48–144× magnifications and a Nikon SMZ1500 stereoscope with Nikon nis software (Melville, NY, U.S.A.) allowing for measurements and digital photography with z-stacking. X-ray micro-computed tomography (micro-CT) scanning of specimens BuPH-01 and BuPH-02 was performed at the American Museum of Natural History Microscopy and Imaging Facility. Scans were taken with a GE Phoenix v|tome|x s240 (Germany), equipped with either a 240 or 180 kV X-ray source. Specimen BuPH-01 was scanned at 60 kV and 250 μA with a voxel size of approximately 3 μm. A total of 1800 images were taken with an exposure time of 1000 ms each. Specimen BuPH-02 was scanned at 80 kV and 200 μA with a voxel size of approximately 8 μm. In total, 1800 images were taken with an exposure time of 750 ms each. Specimen BuPH-02 is infiltrated with a large calcite inclusion, which obscured most of the clypeal horn. Volume reconstruction of raw X-ray images was achieved using GE phoenix datos|x v2.3.2 with automatic geometry correction and default settings. Volumes were postprocessed and rendered using volume graphics vg studio max v3.0 and 3d slicer 4.7 nightly builds (Fedorov et al., 2012). To explore the relative grey values of amber matrix, void space and fossil inclusion tissues for holotype specimen BuPH-01, histogram values (grey values and relative frequency of grey value pixel occupancy) were exported and plotted in r v3.3.2 with ggplot2 v2.1. While it is not possible to definitively infer density values or the elemental composition of materials present in the scan, grey values do provide relative information relating to the degree of X-ray attenuation for each material.
Results
Description
Order Hymenoptera Linnaeus, 1758
Family Formicidae Latreille, 1809
Tribe Haidomyrmecini Bolton, 2003
Genus Linguamyrmex Barden & Grimaldi, new genus
http://zoobank.org/urn:lsid:zoobank.org:act:9F04FAEE-59DA-4B3C-AB8B-EE0D234CC118





Diagnosis, worker
As in other haidomyrmecines (Haidomyrmex, Haidoterminus, Haidomyrmodes, Ceratomyrmex), head hypognathous-like with mandibles projecting primarily downward; mandible scythe-like, with flattened basal margin leading to a curved apical tooth that is expanded perpendicular to axial plane of head. Cephalic clypeal ‘horn’ present but abbreviated, differs from Ceratomyrmex by horn stalk of Linguamyrmex being glabrous, that of Ceratomyrmex with fine, stiff setae of various lengths; clypeal horn much shorter in Linguamyrmex, less than head length/depth, stalk short, with the expanded, flat, paddle-shaped setose pad comprising >50% total horn length; clypeal pad slightly trough-shaped ventrally, covered with very short, dense velcro-like vestiture; trigger hairs originate not at base of stalk as in Ceratomyrmex but near basal margin of setose pad; ocelli present. In addition, Linguamyrmex with first and second gastral segments with deep constriction between them (a gastral constriction is figured in description of Haidomyrmodes mammuthus but is less developed).
Type species
L. vladi sp.n.
Etymology
From Latin ‘lingua’, meaning tongue – in reference to the tongue-like clypeal projection – and the Greek ‘myrmex’, meaning ant.
Linguamyrmex vladi Barden & Grimaldi sp.n.
http://zoobank.org/urn:lsid:zoobank.org:act:2C55A44F-974D-4A33-BDC2-26269687577A
Diagnosis
As for genus.
Description
Head: measuring 0.90 mm postero-anteriorly along dorsal margin, 0.96 mm in length/depth from vertex of head to anterior margin of clypeus. Occipital foramen positioned highly dorsad, just under vertex of head. Postgena broadly depressed; postgenal suture visible, deeply furrowed. Vertex broadly rounded and glabrous with gena gradually tapered ventrally towards mandibular socket and oral opening with fine, sparse setae. Ocelli faintly visible on vertex, positioned dorsally. Eye situated high on head capsule and bulging in frontal view, ovoid, measuring 0.38 mm in length and 0.25 mm in width when viewed laterally. Three antennal segments fully preserved (scape 0.94 mm in length; pedicel 0.12 mm; flagellomere I 0.55 mm). Antennal socket approximately in line with ventral margin of eye; socket exposed and immediately flanking a medial frontal triangle (sensu Perrichot et al., 2016). Frontal triangle extends the vertex, contrasted with antennal sockets, which are present within cuticular depressions. Clypeal horn originating at both frontal triangle and clypeal stalk, both structures heavily sclerotized with cleared, membrane-like cuticle connecting from frontal triangle to stalk. Horn paddle-shaped, total length 0.64 mm, diameter 0.49 at greatest; narrow (0.04 mm) glabrous stalk 0.22 mm in length leading to setose pad; setose pad (0.42 in length) with long trigger hairs originating at pad base; dorsal margin of setose pad glabrous, underside coated in a large number of stout setae in centre and longer, more tapered setae along edges. Anterior margin of clypeus medially triangulate; distinct medial ridge extending to clypeal horn; lateral margins, beginning just above mandibular insertion, extending diagonally toward clypeal horn. Cheek-like lobes projecting anteroventrally above mandible insertion. Mandible scythe-like, comprise a linear basal margin (0.55 mm in length) and curved apical tooth (0.92 mm measured as a straight line from base to tip, ignoring curvature) meeting nearly at right angle; preserved with apical teeth in parallel, nearly touching. Basal portion of mandible with anterior flange-like expansion, concave inner margin coated with pointed setae; leading edge of anterior flange expansion smooth; apical tooth rounded broadly with slight point. Maxillary and labial palps not visible.
Mesosoma
Pronotum broad, coated in fine setae; propleuron reduced, not visible in lateral view except where abutting head capsule anteriorly. Pronotal length 0.72 mm, measured along dorsal margin. Mesonotum 0.34 mm; metanotum 0.24 mm; propodeum 0.49 mm. Weber's length 1.77 mm, mesosomal height 0.56 mm measured perpendicular to Weber's length line at pronotum. Procoxal length 0.72 mm, max width 0.22; mesocoxal length 0.31, max width 0.21; metacoxal length 0.49, max width 0.26; protrochantal length 0.14, max width 0.11; mesotrochantal length 0.19, max width 0.11; metatrochantal length 0.19, max width 0.12; profemoral length 1.01, max width 0.13; mesofemoral length 1.09, max width 0.12; metafemoral length 1.48, max width 0.18; protibial length 1.09, max width 0.11; mesotibial length 1.30, max width 0.08; metatibial length 1.59, max width 0.08. Trochantellus present, length included in femur measurements. Three protibial spurs present, the largest approximately 2× length of other two; mesotibia with two spurs, the larger 2× the length of the smaller; two conspicuous setae of equal length positioned along anterior margin of mesotibial apex; metatibia with two spurs, the largest pectinate and ∼3× the size of the smaller. Pretarsal claw with subapical tooth positioned closer to apex than to claw origin. Dorsal margin of propodeum gradually rounded, posterior margin with sheer face. Propodeal spiracle a dorsoventrally elongate slit situated medially. Metapleural gland opening gaping, present within horizontal invagination of cuticle, opening visible posteroventrally following well-developed bulla.
Metasoma
Petiole 0.49 mm in length, 0.49 mm in height at greatest, pedunculate; not tergosternally fused, deep suture visible; anterior margin gradually sloping; with gradually rounded dorsal margin; broadly attached to gastral segment I (abd seg III) with striated helcium clearly visible. Lateral sulcus visible running anteroposteriorly along petiole, signifying incomplete fusion; ventral to sulcus lies thin, membraneous cuticular expansion with distinct anteriorly positioned keel. Sternite III with pointed projection on anterior margin exhibiting striations running laterally. Circumgastral constriction present between gastral segments I and II with dorsal, posteriorly extended notch. Gastral cuticle translucent, revealing heavily sclerotized and expanded apodemes of gastral sternites IV and V withdrawn into the gaster.
Type material
Holotype, AMNH BuPH-01, in the Department of Invertebrate Zoology, American Museum of Natural History.
Additional material examined
Although not named, three specimens, BuPH-02 (Figs 2D, 3C, 7, Video S1), BuPH-03 (Figs 2B, 3A, B, 4) and BuPH-04 (2C), were studied and are illustrated here. These three specimens probably represent additional species, particularly with respect to overall size and clypeal paddle composition. Moreover, these specimens possess mandibles that appear to interlock basally, joining the distal apical tooth portions of the mouthparts (Figs 3D, E, 7, Video S1). As the inner surface of each mandible is concave, their joining forms a channel that is open dorsally near the pointed mandibular apex (Fig. 7). However, there remains some ambiguity regarding the reliability of measurements and identification of other discrete features due to incomplete preservation. We therefore delay description until more complete material is obtained. Measurements are provided here (in mm): (BuPH-02) [BuPH-03] {BuPH-04} Head length measured postero-anteriorly along dorsal margin (1.14) [0.87] {0.69}, head depth (1.48) [1.22] {1.2}, scape length (1.56) [1.36] {1.28}, paddle length (1.5) [1.52] {1.26}, paddle diameter (−) [∼0.57] {−}, eye length (−) [−] {0.29}, eye width (−) [−] {0.19}, basal margin of mandible (0.71) [0.53] {0.39}, apical tooth (1.97) [1.57] {1.48}, pronotal length (1.08) [0.97] {0.8}, mesosomal height (∼0.81) [∼0.75] {0.66}, Weber's length (2.64) [2.49] {1.99}, petiole length (0.82) [0.76] {0.60}. Unfortunately, no specimen retained all antennal segments, preventing a reliable comparison of scape length relative to all other segments. In addition, all specimens, including the type, are missing terminal abdominal segments.
Etymology
In reference to Vlad III, or Vlad Dracula (c. 1429–1476), prince of a region of Romania then called Wallachia. His moniker, Vlad the Impaler, refers to his favoured and frequent method of execution, which inspired the vampirous character Count Dracula fictionalized by Bram Stoker in 1897. The patronym is in reference to the presumed impalement of prey by Linguamyrmex and its liquid diet (see later).
Systematic comments
Haidomyrmecines have been recovered as monophyletic (Barden & Grimaldi, 2016) and have traditionally been placed as tribe within Sphecomyrminae (Bolton, 2003). The subfamily is defined largely by plesiomorphic features and is now suspected to be paraphyletic (Barden & Grimaldi, 2016). We tentatively follow this subfamilial assignment here but note that, following additional phylogenetic analyses, haidomyrmecines may be best served by the erection of a separate subfamily.
Discussion
In crown-group ants, trap-jaw behaviour has evolved at least four times in three subfamilies: Formicinae, Myrmicinae and Ponerinae (Larabee & Suarez, 2014). Species employing trap jaws are capable of ‘setting’ their mandibles in the open position before rapidly striking them closed, usually in the service of prey capture but occasionally in a defensive effort (Patek et al., 2006; Larabee & Suarez, 2015). While mechanisms for storing and releasing the potential energy required for swift prey capture vary, nearly all modern trap-jaw ants possess trigger hairs that initiate mandible closure. All described haidomyrmecines possess elongate, paired setae present on the clypeus. These setae have been interpreted as trigger hairs (Barden & Grimaldi, 2012; McKellar et al., 2013; Perrichot et al., 2016) and probably function as sensors for rapid mandible closure, as in modern trap-jaw ants. In all cases, these elongate clypeal setae rest directly in the most probable path of mandibular motion. Additionally, all hell ants exhibit an unusually modified clypeal sclerite, which stretches from the anteroventral margin of the head, near the oral opening, to the apex of the head – a syndrome unknown in other ants. The elongated nature of the clypeus may be a modification that would allow for the capture of prey positioned anteriorly – a strategy that would otherwise be untenable given the hypognathous head. The clypeal cuticle of ‘hell ants’ is characteristically raised along the upper margin, always with a clypeal brush composed of shortened, stout setae. Mandibular apices, in turn, are dorsally expanded to meet this clypeal process when fully closed, and would therefore be effective in prey capture. In some cases, raised clypeal cuticle may appear as a slight node whereas in the most extreme sense, as in Ceratomyrmex and, to a lesser extent, Linguamyrmex, the clypeus is distended into a conspicuous cephalic horn. Here, the function of the clypeal process acting as a pinning or stopping point is underscored. Almost certainly, the modifications of the mandibles and clypeus occur in a complementary fashion.
Remarkably, during the course of X-ray imaging, we observed significant heterogeneity in the X-ray penetration of the clypeal paddle (Figs 5B–D, 6). Density and elemental composition are rendered solely into ‘grey values’ that indicate the degree of X-ray attenuation, a solid black pixel with a grey value of 0 indicating a material that is entirely X-ray-transparent. Conversely, material with higher X-ray attenuation due to either high density or absorption coefficient (material-based value impacted by compositional factors including atomic number) will register as higher grey values (e.g. 1 μm of carbon will appear as a lower grey value than 1 μm of iron in the same scan). Scan results indicate that the underside of the clypeal paddle appears to be reinforced, either by greater cuticular density or, more likely (based on the great contrast in X-ray attenuation shown in Fig. 6), through the incorporation of metals into cuticle. This reinforcement occurs primarily along the centre of the paddle and, as the specimen is preserved with the mandibles largely ‘closed’ and positioned near this spot, suggests that the reinforcement is intended to accommodate mandibular impact. Insects are known to sequester metals – in particular, calcium, manganese, zinc, and iron – in ovipositors and mandibles, to increase strength and reduce wear [ants (Edwards et al., 1993); Orthoptera and Coleoptera (Vincent & Wegst, 2004); Isoptera (Cribb et al., 2008); Hymenoptera (Quicke et al., 1998)]. By bracing the contact region of the clypeal paddle, Linguamyrmex may have been able to withstand repeated misfires (i.e. missing prey) or perhaps complete penetration of soft-bodied prey such as larvae. While possible, it is unlikely that the reinforcement area is the result of mineral infilling or partial contamination, due to its distance from amber fractures elsewhere in the specimen and its significantly higher density (Video S2). With resepect to density, the grey values corresponding to the paddle centre are approximately 30% greater than the next closest found within the head capsule.

The mandibles and paddle of Linguamyrmex may have functioned to puncture soft-bodied prey and feed on the haemolymph. Although this would be a novel strategy, this morphological arrangement is itself entirely novel and it is not at all apparent how these ants could otherwise masticate food items. Although it is obtained from their own larvae, some living ant species do utilize haemolymph as a food source (Masuko, 1989; Ward & Fisher, 2016). As with other haidomyrmecines, Linguamyrmex foragers probably hunted with their mandibles cocked wide open and gaping, quickly snapping shut vertically when the tips of the trigger hairs touched prey, evidently with sufficient force that, were the clypeal paddle not reinforced, it might have been damaged. Teeth at the base of the mandibles overlap each other, apparently locking the two mandibles together to move in unison. With prey pinned against the furrowed paddle, the mandibles would be embedded into soft cuticle, while haemolymph from the prey could be channelled into the gutter between the mandibular tusks (Fig. 7, Video S1); a large droplet would then accumulate in the elbowed section, held in place by the surface tension from the minute, dense spicules. The haemolymph droplet could then be guided through the opening at the base of the mandibles by the spicules (which point backwards) and by the negative pressure created by absorption/suction from the adjacent glossa. Specimen PH-03 is preserved adjacent to a large beetle larva. Its mandibles are not embedded into the larva, but the proximity is consistent with this being prey. The presence of the tube-like channel located between the mandibles along with the distinct mesal teeth indicates that these animals probably did not masticate, but rather fed on liquid. Nevertheless, these features do not preclude the possibility that they also functioned to grasp prey items while being immobilized by the sting.

There are multiple examples of predatory specialization in ants, particularly within the subfamily Ponerinae (Hölldobler & Wilson, 1990) and the cryptic lineages that diversified early within crown-group Formicidae (Rabeling et al., 2008; Brandão et al., 2010). In perhaps the most dramatic case, the unique mandibles of the Neotropical ponerine genus Thaumatomyrmex Mayr possess elongate mesial and apical teeth modified into sparsely arranged spikes (Kempf, 1975). Incredibly, these mouthparts effectively function as shears utilized solely for the removal of defensive setae from polyxenid millipedes (Brandão et al., 1991).
Although it may not be possible to definitively ascertain the behaviour of Linguamyrmex, this new taxon highlights the adaptive diversity of a highly specialized, extinct lineage of Cretaceous, stem-group ants. Just as the diversity and adaptive spectrum of carnivorous theropod dinosaurs could not have been predicted by the study of modern vertebrates alone, the bizarre adaptations of haidomyrmecines (indeed, their existence) would remain unknown if not for preservation in amber.
Acknowledgements
We thank Steve Thurston for contributing to figure preparation as well as Andrew Smith and Morgan Hill of the AMNH Microscopy and Imaging facility for assistance with CT imaging. This work funded in part by National Science Foundation no. 1523788 to PB. We also thank Thomas Simonsen and three anonymous reviewers for taking the time to provide comments and suggestions, which improved this contribution. The authors declare no conflicts of interest.




