A remarkable troglomorphic ant, Yavnella laventa sp. nov. (Hymenoptera: Formicidae: Leptanillinae), identified as the first known worker of Yavnella Kugler by phylogenomic inference

Keywords: Insecta, Iran, Milieu Souterrain Superficiel, molecular systematics, Palaearctic, phylogenomics, subterranean biology, troglomorphism.

The Leptanillinae (Hymenoptera: Formicidae) are a group of miniscule, cryptic ants, found in many tropical and warm temperate areas of the Old World. Workers are completely blind and exclusively soil dwelling. The biology of only a few leptanillines has been studied – these are specialised predators of geophilomorph centipedes (Masuko 1990; Hsu et al. 2017) or of forcepstails (Diplura: Japygidae) (Ito et al. 2022). The subfamily is divided into two monophyletic tribes, Leptanillini and Anomalomyrmini (Bolton 1990; Borowiec et al. 2019), with the monotypic genus Opamyrma unplaced to tribe (Ward and Fisher 2016). Of the tribe Leptanillini, in only Leptanilla Emery, 1870 are the worker and queen known, with the remaining three genera – Scyphodon Brues, 1925; Noonilla Petersen, 1968; and Yavnella Kugler, 1987 – known only from male specimens. So far as is known, queens of Leptanilla are wingless and blind, resembling miniature versions of the dichthadiigynes observed in army ants of the subfamily Dorylinae (Hölldobler and Wilson 1990; Ito and Yamane 2020), whereas queens in the Anomalomyrmini are alate (Bolton 1990; Baroni Urbani and de Andrade 2006) or ergatoid (Billen et al. 2013).

Conversely, males of Leptanillinae are always fully winged. Owing to collecting bias towards males, three genera in the tribe Leptanillini have been described solely from male specimens, as have some species of Leptanilla. Further, there is a large diversity of undescribed male morphospecies in the subfamily (Griebenow 2021). The Leptanillinae are therefore afflicted by parallel taxonomy. Out of more than 60 described species, the sexes have been associated only in Leptanilla japonica Baroni Urbani, 1977 (Ogata et al. 1995), Opamyrma hungvuong Yamane, Bui & Eguchi, 2008 (Yamada et al. 2020) and Protanilla lini Terayama, 2009 (Griebenow 2020). Males were collected from the same nests as corresponding workers only in L. japonica and O. hungvuong, with the male of P. lini being indirectly identified using phylogenomic inference.

Yavnella is the most recently described of the male-based leptanilline genera, established by Kugler (1986) for two species from Israel and Kerala (India), by original designation. The phylogenetic analyses of Griebenow (2020, 2021) included a variety of undescribed male Yavnella morphospecies along with Yavnella argamani Kugler, 1987 and recovered the genus as monophyletic with high support, irrespective of data or inferential framework. The genus is most diverse in mainland South-east Asia (this diversity remaining entirely undescribed), with additional representatives in the Indian subcontinent and the Arabian subcontinent as far south as Yemen (Collingwood and Agosti 1996). Borowiec et al. (2019) also robustly recovered this clade, under a sampling regime overlapping with Griebenow (2020, 2021), although Borowiec et al. (2019) did not explicitly identify this clade as Yavnella.

Here, we describe Yavnella laventa sp. nov. from Fārs Province, Iran, based on seven worker specimens, collected in the Milieu Souterrain Superficiel (MSS) at depths of 0.6–1 m within a debris flow adjacent to a salt cave entrance. According to Uéno (1980) and Culver and Pipan (2014), the MSS is a subterranean network of empty air-filled cracks and voids. Yavnella laventa is here identified as belonging to Yavnella based upon phylogenomic inference from ultra-conserved elements (UCEs) and constitutes the first known representatives of the worker caste in that genus. The phenotype of Y. laventa is strikingly different from that of all other known leptanilline workers, with the mandibles, antennae and legs being elongated in what are apparently examples of strong adaptation to subterranean habitats, i.e. troglomorphism. With the discovery of the worker caste of Yavnella we provide a revised worker-based key to the genera of the Leptanillinae, with figures.

A natural classification of the Leptanillinae is made difficult by dissociation of the dissimilar worker and male forms of leptanilline ants, which results in parallel taxonomy. Therefore, the identification of the worker caste of Yavnella, a major leptanilline clade heretofore known only from male specimens, begins to correct this parallel taxonomy. The phylogenomic approach by which Y. laventa was identified as Yavnella, despite the lack of known conspecific male specimens with which to determine the generic identity of this species, affirms the utility of molecular data in connecting unassociated and drastically divergent forms in polymorphic organisms.

From an ecological perspective, the troglomorphism exhibited by Y. laventa is remarkable in the context of the Formicidae as a whole. After Leptogenys khammouanensis Roncin & Deharveng, 2003 (Ponerinae: Ponerini) and Aphaenogaster gamagumayaa Naka & Maruyama, 2018 (Myrmicinae: Stenammini), Y. laventa is only the third arguably troglomorphic ant species described (Roncin and Deharveng 2003; Naka and Maruyama 2018) out of >15 000 described species. In contrast to L. khammouanensis and A. gamagumayaa, Y. laventa was not collected in an underground space between rocks accessible to humans (i.e. a ‘cave’), but in the network of subterranean fissures and voids that constitutes the MSS, which is known to harbour species with troglomorphic traits (see Mammola et al. 2016 for a comprehensive review on the topic).

In total, 63 specimens belonging to the Leptanillinae were included in this study, in addition to Martialis heureka Rabeling & Verhaagh, 2008 (Martialinae) as an outgroup. Ultra-conserved element data are included in this study for 51 taxa of those 63, including M. heureka, a representative of Y. laventa (CASENT0842745), along with representatives of all major subclades of that subfamily. Thirteen of these specimens are newly sequenced in this study. Twelve specimens of Protanilla Taylor in Bolton (1990) and Leptanilla along with one Anomalomyrma Taylor in Bolton (1990) were morphologically examined, but not sequenced (Table 1). Generic assignment of sequenced material follows Griebenow (2020) rather than Borowiec et al. (2019) where sampling overlaps with those studies. Additional collection data for these specimens are provided in a Supplementary Table S1.

Table 1.  Specimens used in this study, with summary statistics pertaining to the 313 498-bp alignment in the cases of those specimens for which ultra-conserved elements (UCEs) were enriched using the hym-v2 probe set of Branstetter et al. (2017).

Specimens are deposited at the following institutions (abbreviations follow http://hbs.bishopmuseum.org/codens/, where applicable): the California Academy of Sciences, San Francisco, USA (CAS); the California State Collection of Arthropods, Sacramento, CA, USA (CSCA); the Biodiversity Museum, University of Hong Kong, Hong Kong, PR China (HKUBM); the Jalal Afshar Zoological Museum, Department of Plant Protection, College of Agriculture and Natural Resources, University of Tehran, Karaj, Iran (JAZM); the Museum of Comparative Zoology, Harvard University, Cambridge, MA, USA (MCZC); Lund University, Lund, Sweden (MZLU); National Changhua University of Education, Changhua, Taiwan (NCUE); the Okinawa Institute of Science & Technology, Onna-son, Japan (OIST); the Royal Ontario Museum, Toronto, ON, Canada (ROME); the R. M. Bohart Museum of Entomology, University of California, Davis, CA, USA (UCDC); the Museum für Naturkunde der Humboldt-Universität, Berlin, Germany (ZMHB); and the Zoological Museum, University of Isfahan, Isfahan, Iran (ZMUI).

Specimens of Yavnella laventa sp. nov. were collected using subterranean sampling devices, i.e. two buried pitfall traps, placed in the MSS. The buried pitfall traps (hereinafter ‘MSS traps’) consisted of a rigid plastic cup, with holes bored in them midway from top to bottom to prevent flooding, and a horizontal stone placed on top. These were set in a clast on the bank of a wadi, opposite to a salt diapir (Fig. 1) (the Khoorab Salt Dome; Abbassi et al. 2015). These MSS traps were placed by slope boring at depths of 60–100 cm, baited with sardines and dates jointly contained in small vials, and half filled with brine. We attempted to measure relative humidity (RH) within the MSS using a Lascar EL-USB-2 data logger buried adjacent to the MSS traps. Brine is not an ideal preservative for purposes of acquiring DNA, but was used in a broad survey of salt karst fauna in the vicinity of Khoorab because of low evaporative rate. Although extraction of a genome-scale molecular dataset was successful when attempted with a single specimen (see ‘Sequencing & data processing’), we recommend that future targeted efforts to collect leptanilline ants in the MSS or in salt caves use ethanol as a preservative, with traps being set for much briefer periods than outlined below.

Fig. 1.  Layout of MSS traps, and position relative to the adjacent salt diapir (the Khoorab Salt Dome) in which Last Cave is located.

Traps were left for 15 months, from 14 February 2019 to 26 June 2020, at the end of which specimens were transferred to 80 or 95% ethanol, the latter if intended for non-destructive DNA extraction. A few specimens remained in brine and were used for dissection and imaging. In addition, four pitfall traps baited in the same way and containing the same liquid were placed at ground level inside an adjacent salt cave (‘Last Cave’) within the Khoorab Salt Dome, for the same duration as the MSS traps, but no Y. laventa were collected in these traps. We measured RH within Last Cave using the same data loggers as listed above.

For the specimens newly sequenced for this study, DNA was extracted non-destructively using a Dneasy Blood & Tissue Kit (Qiagen Inc., Valencia, CA, USA) with H₂O at room temperature to elute DNA, or, in the case of CASENT0842745 and several other samples, 56°C buffer AE (Cruaud et al. 2019) in order to increase DNA yield. Genomic concentrations were quantified for each sample with a Qubit fluorometer (ver. 2.0, Life Technologies Inc., Carlsbad, CA, USA). Input DNA was sheared using a Diagenode Bioruptor (Diagenode, Denville, NJ, USA) or Qsonica Q800R3-110 (Qsonica Inc., Newtown, CT, USA). Sheared product was used as input for the modified library preparation protocol of Branstetter et al. (2017), with the ant-specific version of the UCE probe set hym-v2 (Branstetter et al. 2017). Enrichment success and size-adjusted DNA concentrations of pooled libraries were assessed using the SYBR FAST qPCR kit (Kapa Biosystems, Wilmington, MA, USA), and all pools were combined into an equimolar final pool. Final pools were sequenced on an Illumina HiSeq X at Novogene (Sacramento, CA, USA) or prepared, enriched and sequenced using similar protocols at RAPiD Genomics (Gainesville, FL, USA). Refer to Ward and Blaimer (2022) for further details on library preparation and enrichment. For sequencing protocols implemented for the phylogenomic data used in this study that have been previously published, refer to Griebenow (2020).

The FASTQ output was demultiplexed and cleansed of adapter contamination and low-quality reads using illumiprocessor (ver. 2.1, B. C. Faircloth, see https://github.com/faircloth-lab/illumiprocessor) in the PHYLUCE bioinformatic software package (ver. 1.7.1, see https://phyluce.readthedocs.io/en/latest/; Faircloth 2016). Raw reads were assembled with SPAdes (ver. 3.12.0, see https://github.com/ablab/spades; Bankevich et al. 2012). Species-specific contig assemblies were obtained with the ant-specific hym-v2 probe set (Branstetter et al. 2017), aligned with MAFFT L-INS-I (ver. 7.741, see https://github.com/GSLBiotech/mafft; Katoh and Toh 2010), and trimmed with Gblocks (ver. 0.91, see https://home.cc.umanitoba.ca/~psgendb/doc/Castresana/Gblocks_documentation.html; Castresana 2000) within a PHYLUCE workflow modified from Faircloth (2016) with min_identity = 80 within phyluce_assembly_match_contigs_to_probes.py, resulting in an alignment 313 498 bp long. This alignment was 76.79% complete, composed of 39.1% parsimony-informative sites; AT content was 57.4%. Summary statistics for this alignment were computed with the summary command in AMAS (see https://github.com/marekborowiec/AMAS; Borowiec 2016) (Table 1).

Partitioning to generate subsets of each UCE locus was performed using PartitionUCE (see https://github.com/Tagliacollo/PartitionUCE; Tagliacollo and Lanfear 2018). Using IQ-Tree (ver. 2.1.2, see http://www.iqtree.org/; Minh et al. 2020) on the CIPRES Science Gateway (ver. 3.3, see http://www.phylo.org/; Miller et al. 2010), partition schemes were inferred with ModelFinder in IQ-Tree (ver. 2.1.2; Kalyaanamoorthy et al. 2017) using subsets generated by PartitionUCE for the complete alignment, with the Bayesian Information Criterion (BIC) deciding among available partitioning schemes, followed by maximum-likelihood (ML) phylogenetic inference under these partition schemes (Chernomor et al. 2016) for 1000 ultrafast bootstraps (UFBoot) (Hoang et al. 2018) and SH-like approximate likelihood ratio test (SH-aLRT) (Guindon et al. 2010) replicates. The relaxed hierarchical clustering algorithm was implemented in these analyses (Lanfear et al. 2014), with ModelFinder considering only the most likely 20% of partition schemes. Substitution models with I+G extensions to accommodate among-site rate heterogeneity were permitted, as version 1.4.3 onward of IQ-Tree implements an optimisation heuristic that effectively compensates for the non-identifiability of these models (Nguyen et al. 2018). Bayesian inference was performed in ExaBayes (ver. 1.5.1, see https://github.com/aberer/exabayes; Aberer et al. 2014) on the CIPRES Science Gateway under the partitioning scheme produced by ModelFinder in IQ-Tree (as described above) with GTR+G imposed across all partitions, all parameters unlinked, with the single Markov Chain Monte Carlo (MCMC) running until the average standard deviation of split frequencies (ASDSF) of topologies <0.05, for 100 000 generations. Convergence of the MCMC with respect to continuous parameters was visually assessed in Tracer (ver. 1.7, see https://github.com/beast-dev/tracer/; Rambaut et al. 2018).

Nomenclature for sculpturation follows Harris (1979); setation, Wilson (1955) and Boudinot et al. (2020). Notation of palp and tibial spur formulae follows Bolton (2003). Cephalic nomenclature follows Richter et al. (2021) and Boudinot et al. (2021). Mesosomal nomenclature follows Liu et al. (2019); metasomal, Lieberman et al. (2022). Male genital nomenclature follows Boudinot (2018).

Sorting and initial examination of the material was done using an Echo-Lab SM203H stereomicroscope (DEVCO, Milan, Italy). Morphometric data for four specimens of Y. laventa are included in Table 2. Detailed morphological study of these specimens was performed with a Leica MZ75 compound microscope (Leica Microsystems, Oak Grove, IL, USA) at magnifications of up to 50×. Photographs were obtained as image stacks using a Leica DMC2900 camera attached to a Leica MZ16A stereomicroscope or using the Visionary Digital Imaging System (Visionary Digital, Richmond, VA, USA), with z-stepping in the Leica Application Suite (LAS) software (ver. 4.13.0, see https://www.leica-microsystems.com/products/microscope-software/p/leica-las-x-ls/) and montaged with Helicon Focus Pro (ver. 7.7.4, Helicon Software Ltd, Kharkiv, Ukraine). Scanning electron microscopy was performed with a Hitachi TM4000 (Hitachi Global, Tokyo, Japan). Measurement and index definitions are provided below.

Table 2.  Measurements and indices for the type series of Yavnella laventa.

HW, maximum width of cranium in full-face view.

HL, head length, maximum length of head in full-face view from anterior margin of head to cranial vertex.

SL, scape length, maximum length of scape in medial view, excluding bulbus.

MaL, mandible length, maximum length of mandible from view orthogonal to lateral mandibular margin, measured from ventral mandibular articulation to mandibular apex.

WL, Weber’s length, maximum diagonal length of mesosoma in profile view, measured from most anterior extent of pronotum excluding cervical shield to most posterior extent of propodeal lobes, when present.

PrW, pronotal width, maximum width of pronotum, measured in dorsal view.

MW, mesonotal width, maximum width of mesonotum in dorsal view, measured immediately anterior to mesocoxal foramen.

PTL, petiolar length, maximum length of petiole in dorsal view, not including presclerites.

PTH, petiolar height, maximum height of petiole in profile view, including sternal process and dorsal node, if distinct.

PTW, petiolar width, maximum width of petiole in dorsal view.

PPL, postpetiolar length, maximum length of postpetiole in dorsal view, not including presclerites.

PPW, postpetiolar width, maximum width of postpetiole in dorsal view.

PPH, postpetiolar height, maximum height of postpetiole in profile view, including sternal process and dorsal node, if distinct.

1. CI = (HW ÷ HL) × 100

2. SI = (SL ÷ HW) × 100

3. MI = (MaL ÷ HW) × 100

4. PI = (PTW ÷ PTL) × 100

5. PPI = (PPW ÷ PPL) × 100

6. PPHI = (PPH ÷ PPL) × 100

Maximum-likelihood phylogenomic inference from a 313 498-bp alignment consisting of 473 UCEs, partitioned within-locus, corroborates the phylogeny of Leptanillinae as recovered by Griebenow (2020). All nodes along the backbone of the tree are recovered with high support under ML, with sub-maximal UFBoot/SH-aLRT values being restricted to the sister-group relationships of two terminals within Yavnella. Phylogenomic inference under a Bayesian framework, partitioned using a scheme derived according to an information–theoretic criterion (BIC) using ModelFinder in IQ-Tree (ver. 2.1.2), recovers all internal nodes of the phylogeny with maximal Bayesian posterior probability (BPP), with nearly all estimated parameters having an effective sample size (ESS) of >200 (the exceptions with ESS = 190–191). Yavnella laventa is likewise robustly recovered within Yavnella with maximal support under ML and Bayesian frameworks (UFBoot, SH-aLRT = 100; BPP = 1) and sister to Yavnella argamani (UFBoot, SH-aLRT = 100; BPP = 1) (Fig. 2).

Fig. 2.  Maximum-likelihood phylogeny of the Leptanillinae, inferred from an alignment of 473 ultra-conserved elements (UCEs) partitioned within-locus (Kalyaanamoorthy et al. 2017; Tagliacollo and Lanfear 2018) and rooted a posteriori on Martialis heureka as an outgroup. The Anomalomyrmini (9 terminals) and Leptanilla s. l. (26 terminals) are collapsed. Node support values are UFBoot/SH-aLRT and are only noted when <100. The phylogeny of the Leptanillinae inferred from the same data using a Bayesian approach was identical to the ML phylogeny shown here and is provided on Zenodo (doi:10.5281/zenodo.5595290). Branch length is expressed in number of expected substitutions per site.

Palp formula 2,1 or 1,1. Mandible without differentiated basal and masticatory margins. Medial mandibular margin without regularly spaced serration (Fig. 3c, d). Peg-like chaetae absent from mandible and labrum. Clypeus without median demarcation from frons by posterior carina, anteroposteriorly compressed anterior to antennal torul; antennal torulus adjacent to, or abutting, anterior margin of cranium (Fig. 3c, d); antennal socket fully exposed. Compound eye absent. Frontal carina absent. Antenna 12-merous. Promesonotal articulation highly flexible. Mesotibia with 0–2 apical spurs. Propodeal lobe absent; propodeal spiracle situated low on propodeum. Abdominal segments II–III with tergosternal fusion. Spiracle of abdominal segment III very large and placed far forward. Abdominal segment III posteriorly constricted, forming postpetiole (Fig. 4). Spiracles of abdominal segments IV–VII concealed by posterior margins of preceding tergites. Abdominal segment IV without tergosternal fusion; stridulitrum absent from abdominal presclerite IV. Abdominal tergite VII large, with simple posterior margin. Sting present. Pretarsal claw without apical tooth on inner margin.

Fig. 3.  Full-face view of mandibular armature across the Leptanillinae. (a) Protanilla beijingensis (CASENT0842639). Regular serration at mandibular apex outlined in red. (b) Anomalomyrma indet. (CASENT0178553). (c) Leptanilla thai (CASENT0842784). (d) Yavnella laventa (CASENT0842745). PLC, peg-like chaetae. Scale bars: a, b, 0.2 mm; c, 0.04 mm; d, 0.1 mm.

Fig. 4.  Propodeum and abdominal segments II–IV of Yavnella laventa (CASENT0842746), dorsal view. Scale bar: 0.1 mm.

Yavnella Kugler, 1987 [published in a 1986 volume in Kugler 1986], p. 52. Type species: Yavnella argamani, original designation.

Three mandibular teeth present (Fig. 5a). Palp formula 2,1 (Fig. 5c). Mesotibia with 2 apical spurs. Petiole much longer than wide in dorsal view (PI ≤ 31) (Fig. 4), without distinct anterior peduncle. Abdominal segment IV constricted anteriorly in dorsal view; total length of abdominal segment IV greater than that of abdominal segments V–VII combined.

Fig. 5.  Mouthparts of Yavnella laventa. (a) Mandibles, dorsal oblique view (CASENT0842789). (b) Mandibles, profile view (CASENT0842748). Putative ‘trigger hairs’ outlined in black. (c) Mouthparts of Yavnella laventa, ventral view (CASENT0842789). Maxillary palp (MXP) outlined in black. Scale bars: a, b, 0.1 mm; c, 0.05 mm.

Palp formula 1,1. Ocelli present and set on a distinct tubercle (Griebenow 2020, fig. 5A), rarely absent (Griebenow 2020, fig. 6A); if present, anteromedian ocellus orthogonally dorsal to compound eye in profile view (Griebenow 2020, fig. 12Bi). Procoxa without distal transverse carina (cf. Petersen 1968, p. 583, fig. 8). Protrochanter not elongated relative to meso- and metatrochanter. Profemur without sinuate medial carina or ventral hook. Medioventral carina (Griebenow 2021, fig. 1) and comb (Griebenow 2021, fig. 3) absent from protibia. Notauli absent. Pronotum and mesoscutum not anteroposteriorly elongated. Pterostigma absent (Griebenow 2020, fig. 4B). Recurved posteroventral process absent from mesoscutellum (Griebenow 2021, fig. 16A). Lower metapleuron indistinct. Propodeal declivity concave in profile view (Griebenow 2021, fig. 19B); propodeum without dorsolateral carina. Petiole reduced, without distinct dorsal node. Abdominal tergite VIII broader than long in posterodorsal view. Volsellae present, not dorsoventrally compressed and lamellate; fully articulated medially; parossiculus and lateropenite not distinguishable. Phallotreme apical, not surrounded with dense vestiture of setae.

(Fig. 4–11)

ZooBank: urn:lsid:zoobank.org:act:1BCBAA0B-753E-4DD4-8CFA-43CC25BCE68E

Holotype. Iran, Fārs: 1.3 km E Khoorab [in Milieu Souterrain Superficiel], 60 cm [below surface], 28.59843°N, 52.32863°E [±10 m], alt. 620 m, MSS2, 14.II.2019-26.VI.2020, M. Isaia & M. Moradmand leg. (ZMHB CASENT0842746), ♀.

Paratypes. Iran, Fārs: 1.3 km E Khoorab [in Milieu Souterrain Superficiel], 60 cm [below surface], 28.59843°N, 52.32863°E [±10 m], alt. 620 m, MSS2, 14.II.2019-26.VI.2020, M. Isaia & M. Moradmand leg., 1 ♀ (ZMUI CASENT0842745); ibid., 3 ♀ (ZMUI CASENT0842747, ZMUI CASENT0842795, ZMUI CASENT0842796); ibid., 1 ♀ (JAZM CASENT0842797); 1.3 km E Khoorab [in Milieu Souterrain Superficiel], 100 cm [below surface], 28.59841°N, 52.32856°E [±10 m], alt. 618 m, MSS4, 14.II.2019-26.VI.2020, M. Isaia & M. Moradmand leg., 1 ♀ (ZMHB CASENT0842748).

Other material examined. Iran, Fārs: 1.3 km E Khoorab [in Milieu Souterrain Superficiel], 60 cm [below surface], 28.59843°N, 52.32863°E [±10 m], alt. 620 m, MSS2, 14.II.2019-26.VI.2020, M. Isaia & M. Moradmand leg., 1 ♀ (head and hind leg) (ZMHB CASENT0842789); ibid., 1 ♀ (mesothorax and metapectal-propodeal complex with hind leg) (ZMHB CASENT0842790).

As for genus (see above).

Fig. 6.  Yavnella laventa, holotype (CASENT0842746). (a) Profile view. (b) Dorsal view. (c) Full-face view. Scale bars: a, b, 0.5 mm; c, 0.1 mm.

Fig. 7.  Cranium of Yavnella laventa (CASENT0842789), posterior view. OCC, occipital carina; OCF, occipital foramen. Scale bar: 0.1 mm.

Fig. 8.  Frontoclypeal process of Yavnella laventa (CASENT0842789), full-face view. CAR, lateral carina; FCP, frontoclypeal process. Scale bar: 0.1 mm.

Fig. 9.  Propodeum and anterior petiole of Yavnella laventa (CASENT0842745), profile view. MLF, metapleural longitudinal flange. Scale bar: 0.05 mm.

Fig. 10.  Tibial spurs of Yavnella laventa. (a) Antennal strigil, anterior view (CASENT0842745). (b) Protarsus, posterior view; calcar outlined (CASENT0842745). (c) Mesotibia and mesobasitarsus, posterior view (CASENT0842746). (d) Metatibial spur (CASENT0842789), posterior view. CLC, calcar; PTB, protibia; PBS, probasitarsus; MSP, anterior mesotibial spur; LSP, posterior mesotibial spur; MTP, metatibial spur. Scale bars: a, 0.1 mm; b, 0.2 mm; c, 0.05 mm; d, 0.1 mm.

Fig. 11.  Ventral view of abdominal sternite II of Yavnella laventa (CASENT0842745). Scale bar: 0.1 mm.

Male unknown.

Named after La Venta Esplorazioni Geografiche, the organisation that facilitated the 2019 faunal survey of south-western Iranian salt caves and their vicinity, during which the type series of this species was collected. The specific epithet is a noun in apposition and is therefore invariant.

 Head

 Cranium longer than wide in full-face view (CI = 68–72). In full-face view, vertex of cranium emarginate; occiput anteroposteriorly narrow, occipital carina completely encircling occipital foramen (Fig. 7). Lateral margins of cranium slightly convex. Frontal carina absent. Antennal insertion exposed. Frontoclypeal process present, delimited from cranium by lateral carinae (Fig. 8), without posteromedian delimitation from cranium, projecting well anterior of labrum in full-face view; frontoclypeal process laminate, broad in outline, with apex emarginate, and anterolateral corners lobate (Fig. 8). Clypeus anteroposteriorly compressed anterior to antennal toruli; epistomal sulcus absent.

Anterior tentorial pit not visible. Antennal torulus circular. Hypostomal carina present. Postgenal ridge extending from hypostoma to occipital carina. Mandible projecting anteriorly at rest (Fig. 3d). Mandalus small and bean-shaped in outline. Lateral mandibular groove extending along 1/3 of mandible surface, with a smaller groove laterad the longitudinal line, beginning at the basal tooth; both grooves merging proximad subapical tooth. Medial mandibular margin not divided into basal and masticatory portions. Three teeth present on mandible, apical tooth acute; basal tooth larger than subapical tooth, tip recurved; margin distal to subapical tooth irregularly serrate (Fig. 5a). Large, tapering basal and subapical setae present on mandible (Fig. 5b). Peg-like chaetae absent from mandible. Labrum concealed by frontoclypeal process in full-face view; peg-like chaetae absent from labrum. Palp formula 2,1 (Fig. 5c). Ventral premental face elliptical. Antennae 12-merous. Scape elongated, extending well beyond cranial vertex at rest (SI = 160–163); margins subparallel, slightly expanded towards apex. Pedicel longer than broad; constriction separating pedicel from flagellum not pronounced. Flagellum filiform; all flagellomeres longer than broad, with antennomere 3 longer than length of any of the distal antennomeres (Fig. 6a, b); apex of antennomere 12 slightly tapered.

 Mesosoma

 In dorsal view, pronotal outline anteroposteriorly oblate, maximum width (PrW = 0.199–0.228) greater than that of mesonotum, or of the propodeum (Fig. 6b). Pronotal dorsum convex, elevated above dorsal mesonotal vertex. Promesonotal suture present, highly flexible. Mesonotum constricted anteriorly in dorsal view, with maximum width <PrW (Fig. 6b); indistinct from mesopleural region. Mesothorax dorsoventrally constricted and anteroposteriorly elongated posterad the promesonotal suture in profile view (Fig. 6a). Meso-metapleural suture absent; in profile view, fusion of mesonotum with propodeum marked by excavation. Propodeum not constricted anteriorly in dorsal view, with outline subrectangular. Metapleural gland bulla large, anterior margin extending slightly anterior to anterior margin of propodeal spiracle. Metapleural gland orifice longitudinally elongated, curving posteriorly towards dorsum, overhung by longitudinal flange (Fig. 9).

Propodeal declivity convex in profile view. Coxae robust, pro- and mesocoxae well separated; distal leg articles elongated (Fig. 10b). Metacoxal dorsum unarmed. Tibial spur formula 2b,1p. Calcar large, anterior margin densely pectinated (Fig. 10a), posterior surface bare, velum large; apex of posterior margin with two subapical spines (Fig. 10b); posterior stout seta absent from protibia. Anterior mesotibial spur reduced, barbulate with slight splintering; posterior mesotibial spur with pronounced barbulation. Metatibial spur pectinate (Fig. 10d). Meso- and metabasitarsus less than one half the length of meso- and metatibia, respectively. Anterior surface of probasitarsus with single row of acute scale-like cuticular processes (Fig. 10a); posterior surface bare of such processes. Tarsomeres with traction chaetae small and restricted to distal margins. Pretarsal claws unarmed, length less than that of tarsomere 5. Arolium present.

 Metasoma

 Anterior margin of petiole linear in dorsal view. Abdominal spiracle II very large, situated well forward on petiole. Petiole much longer than wide (PI = 29–32) (Fig. 4), without distinct dorsal node or ventral process; sessile; tergosternal fusion complete, with anterior ½ of abdominal sternite II delimited by longitudinal carinae, converging anteriorly in ventral view (Fig. 11); lateral margins subparallel in dorsal view. Dorsal and ventral surfaces of petiole shallowly convex in profile view. Abdominal spiracle III very large, situated well forward on postpetiole. Abdominal segment III posteriorly constricted, forming postpetiole; somewhat longer than wide (PPI = 59–65); tergosternal fusion complete, with longitudinal sutures not converging anteriorly in ventral view; lateral margins convex in dorsal view (Fig. 4). Prora distinct. Abdominal segment IV longer than length of posterior abdominal segments combined, constricted into ‘neck’ immediately posterior to abdominal segment III. Abdominal segments IV–VIII without tergosternal fusion. Abdominal sternite VII entire and unarmed. Sting well developed.

 Integument

 Somal surface smooth to scabrous; mostly scabriculous. Anterior margins of pronotum, meso- and metapleuron, and abdominal sternite II areolate (Fig. 11) to rugose. Occiput substrigulose (Fig. 7). Appendages mostly unsculptured. Colouration orangish-yellow, extremities paler. Cuticle covered with short setae, subdecumbent to appressed; sparse on cranium, mesosoma, and abdominal sternite II. Setae longest on abdominal segments III and V–VII.

Yavnella laventa is known only from the type locality, inhabiting the MSS within a debris flow on the bank of an ephemeral stream adjacent to a salt diapir (Fig. 12). The Khoorab Salt Dome is one of ~130 salt diapirs occurring in southern Iran (Talbot and Alavi 1996). It is therefore possible that Y. laventa occurs across this area, at least in microhabitats resembling those present at the type locality.

Fig. 12.  Map of the type locality within Iran, and the spatial relationship of the MSS traps in which Yavnella laventa was collected relative to each other and to Last Cave. Yellow marks indicate approximate positions of MSS traps, which were located below the surface of the ground. (a) Location of MSS2. (b) Location of MSS4.

Mean annual precipitation around the type locality is ~400 mm (Zarei 2010), meaning that moisture is a limiting abiotic factor. Microclimatic conditions in the MSS at the type locality were not directly measured becuase of data logger malfunction. Indeed, RH in the MSS is rarely measured for this reason (Mammola et al. 2016). Contrary to cave habitats, in which RH is generally constant, studies of this parameter in the MSS show seasonal variation, with a drop in spring through summer (Barranco et al. 2013). Using a data logger and measuring RH in the nearby salt cave (the ‘Last Cave’), we found that RH varied seasonally from 50 to 80%, contrasting with the humidity and climatic constancy commonly associated with subterranean habitats (Cigna 2002; Badino 2010). We hypothesise that the hygroscopic property of salt causes a strong drop in humidity inside Last Cave, which may account for the absence of Y. laventa.

Fig. 13.  Profile view of abdominal segments II–III across the Leptanillinae. Profile of abdominal tergite II outlined in magenta; profile of abdominal segment III outlined in blue. (a) Opamyrma hungvuong (AKY05vii17-06) (Yamada et al. 2020, fig. 1C). (b) Protanilla bicolor (CASENT0235341), Estella Ortega (AntWeb, ver. 8.68.7, California Academy of Sciences, see https://www.antweb.org). (c) Anomalomyrma helenae (CASENT0220220) (Borowiec et al. 2011, fig. 6). Scale bars: a, 0.5 mm; c, 0.2 mm; c, 0.5 mm.

Fig. 14.  Full-face view of the frontoclypeal margin across the tribe Leptanillini. (a) Leptanilla KE01 (CASENT0842721). (b) Leptanilla boltoni (CASENT0260440). (c) Leptanilla thai (CASENT0842784). (d) Yavnella laventa (CASENT0842789). Scale bars: a, b, 0.02 mm; c, 0.035 mm; d, 0.05  mm.

The habitus of Y. laventa is exceptional among worker Leptanillinae in the elongation of the appendages, including the scape (Table 3), flagellomeres, and tarsomeres (Fig. 6c, 15b). The anterior constriction of abdominal segment IV is also unique among the Leptanillinae, exceeding the constriction observed in Leptanilla tanakai Baroni Urbani, 1977 (Baroni Urbani 1977, fig. 33). Protanilla spp. have long scapes (SI > 100) by comparison to Leptanilla (Richter et al. 2021), but these are less elongated than in Y. laventa (Table 3). Elongation of the scapes in Protanilla was hypothesised to be a secondary reversal from the ancestral condition in the Leptanillini (Richter et al. 2021). This implies that the elongation of the scape in Y. laventa is also a secondary reversal.

Table 3.  Comparison of scape index (SI) and petiole index (PI) in Yavnella laventa to that observed in a selection of other leptanillomorph species.

Fig. 15.  Protarsi of selected Leptanilla spp. and Yavnella laventa, lateral view. (a) Leptanilla boltoni (CASENT0842753). (b) Yavnella laventa (CASENT0842745). (c) Leptanilla thai (CASENT0842752). (d) Leptanilla theryi (CASENT0842751). TRC, traction chaetae. Scale bars: a, 0.05 mm; b, 0.2 mm; c, 0.1 mm; 0.05 mm.

The elongation of appendicular articles in Y. laventa is unparalleled in the ‘leptanillomorph clade’, i.e. Martialis + Leptanillinae (Borowiec et al. 2019; Richter et al. 2021), as is the elongation of the petiole (PI = 29–32) (Tables 2, 3). Leptanillomorph workers generally have robust, short limbs, with a submoniliform antennal funiculus; this tendency is most pronounced in the Leptanillini. Along with positioning of the antennal toruli anterior to their ancestral position for the Formicidae, shortening of extremities is associated with motion in confined subterranean conditions (Eisenbeis and Wichard 1987; Richter et al. 2021). By contrast, the extremities of Y. laventa are attenuated and fragile, with the antennal funiculus filiform. This convincingly restricts this species to subterranean voids, as predicted for M. heureka (Rabeling et al. 2008, fig. 2). The sparseness of traction chaetae on the ventral tarsal surface (Fig. 10b, 15b) also implies limited digging capability in Y. laventa compared to examined Leptanilla spp. (Fig. 15).

The emarginate frontoclypeal process of Y. laventa resembles that observed in many Leptanilla spp., mostly distributed in the Indo-Malayan ecoregion. Although regarded as clypeal in origin by previous authors (e.g. Leong et al. 2018), the homology of the frontoclypeal process is unclear, since it is difficult to delimit the clypeus in the absence of the epistomal sulcus.

The mandibular surface of Y. laventa bears sparse, tapering suberect setae of mostly uniform length and diameter. Two pairs of more robust, longer suberect setae are present on the medial mandibular surface, with the distal pair positionally homologous with the putative ‘trigger hairs’ present in Protanilla lini and Protanilla rafflesi Taylor in Bolton (1990) (Richter et al. 2021). This is the first purported example of trigger hairs in the Leptanillini. The definition of trigger hairs has always been functional rather than anatomical, relying upon confirmation of ‘trap-jaw’ behaviour, or assertion by analogy to other ants for which behavioural observations exist (e.g. Creighton 1930; Barden and Grimaldi 2012; Richter et al. 2021). Observations of living Y. laventa, or three-dimensional modelling of mandibular movement in this species based upon micro-CT data, would test the hypothesised function of these mandibular setae (cf. Richter et al. 2021). Consultation of the primary literature (e.g. Man et al. 2017, fig. 5; Aswaj et al. 2020, fig. 2C; Baidya and Bagchi 2020, fig. 1C), photographs on AntWeb (ver. 8.68.7, California Academy of Sciences, see https://www.antweb.org), and available specimens showed that the subapical mandibular seta is present and robust in all described species of Anomalomyrmini for which this information is available. The presence of a robust subapical mandibular seta was also confirmed in all available undescribed specimens of that tribe (Table 4). There were few available worker specimens belonging to the Leptanillini in which mandibular setation could be assessed. A subapical mandibular seta is present in those that were examined and in O. hungvuong (Yamada et al. 2020, fig. 2E) (Table 4) but is less produced than in the Anomalomyrmini or Y. laventa, leaving its function as a trigger hair doubtful.

Table 4.  Presence or absence of a subapical mandibular seta in all described species belonging to the tribe Anomalomyrmini, and in 8 specimens belonging to undescribed morphospecies of this tribe.

In Y. laventa the mandibles are elongated such that, when closed, these rest in a position subparallel to the anteroposterior axis of the cranium (Fig. 3d, 5b, 6c), resembling the Anomalomyrmini. This is not a condition previously observed in the Leptanillini. Save for the posture of the mandibles at rest, Y. laventa has little morphological commonality with the Anomalomyrmini to the exclusion of other Leptanillini. Anomalomyrmine workers are uniformly distinguished from the Leptanillini, including Yavnella, by the presence of four maxillary palpomeres; the presence of regular serration on the medial mandibular margin, and absence of large teeth from that margin; the presence of at least one peg-like chaeta on the labrum; and the median demarcation of the clypeus from the frons by a carina.

Yavnella laventa was collected ≥60 cm. below the surface, setting the workers of this species apart from other Leptanillinae for which soil depth of origin is recorded: these workers approach the depths of Y. laventa only in Leptanilla taiwanensis Ogata, Terayama & Masuko, 1995 and Protanilla beijingensis Man, Ran, Chen & Zhu, 2017 have been collected with unbaited pitfall traps at depths of up to 55 cm (Man et al. 2017).

The biotope of Y. laventa appears to be the MSS, rather than soil as is the case in other leptanilline ants, which is consistent with the strikingly gracile phenotype of this species. The elongated, delicate limbs preclude the endogean (i.e. soil dwelling) biology otherwise observed in this subfamily, instead indicating hypogean habits. This elongation of extremities is consistent with troglomorphism. Worker Leptanillinae lack compound eyes, and so that condition in Y. laventa does not constitute troglomorphism per se, although it corroborates our supposition that this species is exclusively subterranean, as are all other leptanilline ants. Rather, the argument for troglomorphism in Y. laventa rests upon overall elongation of the extremities in conjunction with subterranean biology. This is analogous to troglomorphic Japygidae and Campodeidae (Hexapoda: Diplura), which likewise belong to an ancestrally eyeless, endogean clade, and differ from endogean relatives by larger size and elongation or multiplication of appendicular articles (Sendra et al. 2021). A similar pattern is also recovered in subterranean spiders (Araneae), e.g. Troglohyphantes spp. (Linyphiidae), in which leg length appears to correlate with habitat (pore) size (Mammola and Isaia 2016).

We here follow the definition of Bichuette et al. (2015) for troglomorphism, regarding it as constituting phenotypic traits selectively favoured by subterranean biology and apomorphic relative to non-subterranean relatives of the putatively troglomorphic lineage. By this definition, troglomorphism does not necessarily coincide with habitation in subterranean voids that are considered caves by dint of being ‘commensurable to the human scale’ (Mammola et al. 2016, p. 3). Therefore, that Y. laventa was collected in the MSS does not exclude its being troglomorphic. Indeed, this condition is unsurprising, since troglomorphic organisms are frequently encountered in the MSS (Christiansen 2005; Juberthie and Decu 2006; see Mammola et al. 2016 for further evidence from the literature).

Few ants are known to reside permanently in caves (Pape 2016) or exhibit troglomorphism (Christiansen 2005). Whereas Nylanderia pearsei Wheeler, 1938 (Formicinae: Lasiini) from the Yucatán Peninsula and an undescribed Leptogenys sp. (Ponerinae: Ponerini) from central Texas are subterranean so far as is known (Wheeler 1938; Reddell 1977; Cokendolpher et al. 2009), neither is unambiguously troglomorphic in phenotype. Compelling arguments for troglomorphism among described ants have heretofore only been made for Leptogenys khammouanensis and Aphaenogaster gamagumayaa. When compared to their respective closest relatives, both display a gracile habitus, pale colouration and reduced compound eyes (Roncin and Deharveng 2003; Naka and Maruyama 2018). Additionally, L. khammouanensis was collected in two large calcareous caves in Laos, ranging from 500 m to several kilometres from the cave entrance (Roncin and Deharveng 2003), whereas A. gamagumayaa was collected ~20 m within a calcareous cave on Okinawa, apparently nesting in the floor of an aphotic guano hall (Naka and Maruyama 2018, pp. 138–139).

Since Y. laventa is the sole species of Yavnella for which the worker has been identified, the range of morphological variation in the worker caste of Yavnella is unknown, as is the prevalence of troglomorphism in Yavnella. If all that is required for the evolution of troglomorphism in the Leptanillinae is the presence of the MSS, troglomorphism could be prevalent across the worker caste in Yavnella, since the geographical extent of the MSS is unknown (Juberthie and Decu 2006; Mammola et al. 2016).

It must be cautioned that without troglomorphic elongation of the soma and extremities, and putative trigger hairs, workers of Y. laventa cannot be discriminated from those of Leptanilla. Disregarding these apomorphies, the phenotype of Y. laventa shows close affinity to Leptanilla escheri (Kutter, 1948) and Leptanilla judaica Kugler, 1987. The 2,1 palp formula of Y. laventa (Fig. 5c) resembles that in Leptanilla havilandi Forel, 1901, L. escheri, L. judaica, and Leptanilla ujjalai Saroj et al., 2022 (this study; Kugler 1986; Saroj et al. 2022), whereas the anteromedian frontoclypeal process of Y. laventa resembles that observed in these and other Leptanilla spp. (Kugler 1986; Wong and Guénard 2016, fig. 1A, B; Leong et al. 2018, fig. 14B, C, 15A; Saroj et al. 2022, fig. 3B).

It is possible that L. escheri and L. judaica are non-troglomorphic representatives of Yavnella. In the absence of molecular data for L. escheri, L. judaica or their close relatives Leptanilla lamellata Bharti & Kumar, 2012 and L. ujjalai, we refrain from transferring any Leptanilla spp. to Yavnella. The hypothesis that these Leptanilla spp. represent Yavnella has biogeographical plausibility, because these species are known from the Indian subcontinent and Israel, with Iran intervening between these regions (Fig. 16). Under this hypothesis, it is plausible that Yavnella indica Kugler, 1987 and Y. argamani respectively represent the males of L. escheri and L. judaica, a prediction that could be tested with molecular data, as in this study and others (e.g. Ward and Brady 2009; Griebenow 2020).

Fig. 16.  Distribution of known specimens attributed to Yavnella, according to Collingwood and Agosti (1996) and AntWeb (see https://www.antweb.org), with generic attribution of undescribed morphospecies following Griebenow (2021). Generated using SimpleMappr (https://www.simplemappr.net).

Identification of L. escheri or any of its relatives as representatives of Yavnella would erase the distinction between that genus and Leptanilla in the worker caste. The subapical mandibular setae have not been comprehensively surveyed across the known diversity of the Leptanillini and therefore are not of monothetic use. Yavnella and Leptanilla s. l. are uniformly discriminated based upon male morphology (Griebenow 2021) and robustly recovered as reciprocally monophyletic by ML and Bayesian phylogenomic inference (Griebenow 2020, 2021; this study). Resolving the taxonomic status of the major subclades of the Leptanillini, including Yavnella, will require sequencing of further worker material across this tribe, including the as-yet-unknown worker caste of Scyphodon, Noonilla, and the undescribed Bornean morphospecies-group (Griebenow 2020, 2021).

Supplementary material is available online.

Configuration files, DNA alignment and output for all phylogenetic analyses employed in this study are available on Zenodo (doi:10.5281/zenodo.5595290). Accession numbers for raw UCE reads, uploaded to the Sequence Read Archive (SRA) (https://www.ncbi.nlm.nih.gov/sra), are provided in Table 1.

The authors declare that they have no conflicts of interest.

This research was supported by the UC Davis Department of Entomology, UC Davis Jastro-Shields, and by NSF grant DEB-1932405 to P. S. Ward.