The primary field research took place in the lowland dipterocarp rainforest at the Kuala Belalong Field Studies Centre (KBFSC), Temburong District, Brunei Darussalam (4°32'48.2"N, 115°09'27.9"E), where Colobopsis explodens sp. n. was sampled during five collecting trips (each of 30 days duration) encompassing different seasons from 2014 to 2016.

The behaviour of C. explodens sp. n. was observed at multiple nesting sites on several height-levels, starting from the forest floor and understory up to the canopy and emergent layer. The activity of ants was recorded in situ and in vitro using a CANON 70D Digital SLR Camera with a CANON EF 100 mm macro lens and a Tamron AF 28–200 mm F/3.8–5.6 XR Di aspherical (IF) macro zoom lens (Suppl. material 2: S2a). For macro and close-up filming the Neewer adjustable LED light with LCD display was used. When necessary, the camera was mounted with the use of a Manfrotto Gorillapod 494RC2 tripod. The movie (Suppl. material 7) was annotated and cut using Corel VideoStudio X10 Software.

Sampling of Colobopsis badia in southern Thailand was conducted by H. Zettel and W. Jaitrong in June 2016. The sampling site was located in the Khao Chong Botanical Garden, near the Ton Pliw Waterfall (07°32'34"N, 99°47'33"E); a single male specimen was caught at a light at the Botanical Garden headquarters.

Nesting habits of C. explodens sp. n. were observed based on the model colony occupying several trees and an artificial nest (nest #38, Fig. 9) in direct vicinity to the kitchen facility at KBFSC. The artificial nest consisted of a 100 cm tall and 6 cm wide square wooden stake, with a cavity of approximately 15 mm in diameter drilled into the centre and a 4 mm wide entrance hole in the top third of the stake. The nest was painted with green acrylic paint and fastened to a small tree with rope (for detailed method of construction, see Davidson et al. 2009 and Laciny et al. 2017). The host trees were identified by comparison with type samples preserved in the herbarium of Universiti Brunei Darussalam, Brunei. The main host tree was DNA barcoded (see Suppl. material 6 “accession numbers”).

The activity of C. explodens sp. n. occupying artificial nest #38 was observed from 14^th to 30^th November of 2015 at different times during the day, for 30 minutes each by counting the ants entering and leaving the nest. Temperature, barometric pressure, and weather conditions were recorded, as well as any observed noteworthy behaviour (see Fig. 9; Suppl. material 6 “activity”).

DNA extraction, gene fragment amplification, and sequencing were performed for minor worker ants of five different taxa (C. explodens sp. n., C. badia, C. nr. saundersi, C. aruensis Karawajew, 1933, and C. cylindrica (Fabricius, 1798)), as well as for mandibular gland reservoir content of C. explodens sp. n., one symbiotic cricket (Camponophilus sp.) from artificial nest #38, and the host plant of C. explodens sp. n. (Shorea johorensis).

For DNA barcoding of ant specimens, DNA was extracted from ant legs using Qiagen’s tissue QIAamp DNA Micro kit following the manufacturer’s protocol (Qiagen, Venlo, Netherlands). To obtain sufficient DNA quantity for further processing, the amount of legs used per sample varied. For the DNA extraction of queens, a minimum of three legs of one individual were transferred into one 1.5 ml microcentrifuge tube and frozen with liquid nitrogen. Three legs of one individual was also the minimum amount for males and major workers. For minor workers, all legs from two to four individuals were pooled (see Suppl. material 6 “accession numbers”). The frozen legs were ground into small pieces with disposable pestles attached to a pestle motor (Kimble, Vineland, NJ, USA). Subsequent steps were performed according to manufacturer's instructions with the following exceptions: sample lysis for 20 hours, final elution step with 25–50 µl elution buffer. To assess the purity of the extraction, DNA concentration and 260/280 nm ratio were measured with a NanoDrop ND-1000 Spectrophotometer (Software Version ND-1000 v.3.8.1, Thermo Fisher Scientific, MA, USA).

For DNA barcoding of symbiotic crickets, DNA was extracted from one whole specimen applying the same procedure as used for ants legs, but with a pretreatment with an enzymatic lysis buffer (Tris·Cl 20 mM, pH 8.0, sodium EDTA 2 mM, Triton X-100 1.2%, add lysozyme to 20 mg/ml for 60 min).

For DNA barcoding of the host plant, 100 mg of a leaf were ground with mortar and pestle under the use of liquid nitrogen and DNA was extracted using Qiagen’s DNeasy Plant Mini Kit according to manufacturer’s instructions.

For the ants, the gene fragments cytochrome C oxidase subunit I and II (COI, COII), cytochrome B (cytB), and carbamoyl-phosphate synthase II (cad) were amplified, for the cricket only COI. Additionally, a fragment of 16S rRNA was amplified from the DNA extracted from the mandibular gland content of C. explodens sp. n. minor workers, to assess the presence of bacteria. For the plant, the gene fragment maturase K (matK) was amplified. Primer sequences and specific annealing temperatures are given in Tab. 1. Final concentrations for PCR were 1× GoTaq Flexi Buffer (Promega, Madison, Wisconsin, USA), 0.16 mM dNTP’s, 3 mM MgCl₂ (Promega), 0.4 µM forward and reverse primer (Microsynth, Balgach, Switzerland), 0.8 Units GoTaqG2 Flexi polymerase (Promega) and 2–50 ng sample (diluted with HPLC water, ROTH), in a final volume of 50 µl. PCR was performed with a Biometra T3 Thermocycler (Biometra, Göttingen, Germany) with the following conditions: 2 min at 94 °C, 35 cycles of 1 min at 94 °C, 1 min at primer specific annealing temperature, 90 sec at 72 °C and finally 7 min at 72 °C. PCR products were separated by 1.5% agarose gel electrophoresis. PCR products were purified using mi-PCR Purification Kit (Metabion, Planegg, Germany) and one direction sequencing was performed at Microsynth (Austria).

Sequences are deposited in NCBI GenBank. Accession numbers for ant specimens are given in Table 2; see Suppl. material 6 “accession numbers” for additional details. Sequences of non-ant material are deposited under MG582639 for COI of myrmecophilous crickets (Camponophilus sp.), MF993320 for matK of Shorea johorensis and MF996752 for 16S rRNA of cf. Blochmannia (Enterobacteriales).

GapStreeze v. 2.1.0 (https://www.hiv.lanl.gov/content/sequence/GAPSTREEZE/gap.html) was used for COI gene alignment with 95 % gap tolerance in order to retain only the conserved region. The individual gene alignments were subjected to best substitution model selection using the BIC criterion in SMS (Lefort et al. 2017). Consecutively, HKY85, HKY85+I, HKY85+G, and GTR+G were chosen as best substitution models for genes cad, cytB, COI, and COII respectively. The concatenated alignment was partitioned for each locus using MrBayes v. 3.2.5 (Ronquist et al. 2012), and the respective substitution models were assigned to each partition. The substitution and branch length estimates were allowed to vary independently between each partition. Priors for an exponential distribution with mean 1 to all branch lengths and to all shape parameters were assigned for all four partitions. Metropolis-coupled Markov chain Monte Carlo (MCMCMC) sampling was performed using MrBayes v. 3.0B4 (Ronquist et al. 2012) with two simultaneous runs of four incrementally heated chains that performed 1 million generations. Bayesian posterior probabilities (PP) were obtained from the 50 % majority rule consensus of trees sampled every 100 generations after removing the first 25 % of trees using the “burnin” command. According to the protocol of Leache and Reeder (2002), PP values higher than 0.94 were considered significant. The phylogenetic trees were visualized in FigTree v. 1.4.3 (Rambaut 2016) and then annotated using vector graphic software.

All specimens used for morphometry were card-mounted, individually numbered, and measured at magnifications from 25.6× up to 256× with a Nikon SMZ1500 binocular microscope. Genital structures of two male specimens were dissected and mounted separately. Results represent minimum and maximum values for each morph; in cases where a character could not be measured in all individuals, the number of measured specimens is given in parentheses. The complete dataset of measurements is provided in Suppl. material 6 “measurements”.

Measurements and indices (* = only gynes and males)

EL Eye length. Maximum diameter of compound eye, measured in lateral view.

FeL Femur length. Maximum length of metafemur, measured from base to apex.

FWL* Forewing length. Length of forewing, measured from tegula to distal tip.

HaL Hair length. Length of the longest standing hair on first gastral tergite, measured from base to apex.

HL Head length. Maximum length of head in full-face view, excluding mandibles, measured from anteriormost point of clypeus to posterior-most point of head vertex, parallel to midline.

HS Head size. (HW + HL) / 2.

HW Head width. Maximum width of head in full-face view (including eyes if protruding; only in gynes).

ML Mesosoma length. Measured laterally from anterior surface of pronotum proper (excluding collar) to posterior extension of propodeal lobes.

MSW* Mesoscutum width. Maximum diameter of mesoscutum, measured dorsally.

NH Node height. Height of petiolar node, measured laterally, from the intersection point of the axes of maximum height and length to dorsal apex

OcD* Ocellar distance. Minimum distance between lateral ocelli, measured between median borders.

OcW* Ocellus width. Maximum diameter of median ocellus.

OED* Ocellar eye distance. Minimum distance between lateral ocellus and outer border of compound eye.

PH Petiole height. Maximum height of petiole in lateral view, measured from ventral-most point of petiolar sternum to dorsal apex.

PL Petiole length. Maximum length of petiole in lateral view, measured from inflexion point of anterior constriction to posterior margin, perpendicular to axis of maximum height.

PS5 Length of maxillary palp segment V, measured from base to apex.

PS6 Length of maxillary palp segment VI, measured from base to apex.

SL Scape length. Maximum length of antennal scape in dorsal view excluding basal neck and condyle.

SW Scape width. Maximum width of antennal scape, measured dorsally.

TL Total length. The added lengths of head (excluding mandibles), mesosoma, petiole, and gaster.

2r* Maximum length of 2^nd radial crossvein (see Figs 5e, 6b).

4Rs+M* Length of 4^th radial sector fused with median (see Figs 5e, 6b).

CI Cephalic index. HW / HL × 100

EI Eye Index. EL / HW × 100

FeI Femur Index. FeL / HW × 100

OI* Ocellar Index: OED / OcD × 100

PI Petiole Index. PH / PL × 100

PSI Palp Segment Index. (PS5 + PS6) / HS × 100

SI Scape index. SL / HW × 100

SWI Scape width index. SW / SL × 100

WVI* Wing Vein Index. 4RsM / 2r × 100

Digital stacked images of most specimens (Figs 2–6) were acquired with a Leica DFC camera attached to a Leica MZ16 binocular microscope with Leica Application Suite v3 and stacked with Zerene-Stacker 64-bit. Images of labels were taken with a Nikon D60 camera with an AF-S Micro Nikkor 105 mm objective and an EM-140 DG macro ring flash. Photographs of genital structures of males (Figs 7, 10c–f) as well as of the male C. badia specimen (Fig. 10 a, b) were created with the help of Leica Application Suite v3.8, using a Leica DFC450 camera attached to a Leica Z16APO optics carrier. All images were processed with Adobe Photoshop 7.0.

Type material of C. explodens: Holotype (minor worker): Brunei, Temburong, Kuala Belalong Field Studies Centre, 04°33'N, 115°09'E, 60 m a.s.l., 10.XI.–5.XII.2015, leg. A. Laciny & A. Kopchinskiy (“YG Vienna Colony”, specimen number COCY 01565).

Paratypes (59 minor workers, 8 major workers, 16 gynes, and 6 males dry mounted; > 500 imagines stored in 96 % ethanol): 19 minor workers, 2 major workers, 12 alate gynes, 4 dealate gynes, 6 males (including allotype) (all dry mounted), as well as 8 males, 2 alate gynes, ca. 500 minor workers (in alcohol) from the same nest sample as holotype; 1 major worker, same locality and collector as holotype, 17.IV.2015, “YG 373 main natural nest”; 1 major worker, same data as holotype (“YG doorkeeper #19”); 2 major workers, same data as holotype except 20.IV.2015, leg. A. Kopchinskiy (“cf. YG 39 (351) artificial nest”); 8 minor workers, 2 major workers, same locality as holotype, 2002, leg. D.W. Davidson (“YG KB02-108”); 4 minor workers, same locality and collector as previous, no collection date, “YG 2025”; 5 minor workers, same locality and collector as previous, I.2012, “YG T-trail (202)”; 5 minor workers, same locality and collector as previous, 15.V.2014, “YG-2 (73)”; 7 minor workers, same locality and collector as previous, 15.V.2014, “YG-2 (49)”; 2 minor workers, same data as previous except Batu Apoi Forest Reserve, N04°32', E115°10', 200 m a.s.l., 25.XI.2004, (“CAYG A-370”); 15 minor workers (on 5 pins), same data as previous, except N04°55', E115°19', 60 m, 3.VII.2002, (“YG KB02-108 voucher”); 4 minor workers, Thailand, Chumphon Province, Krom Luang Chumphon W.S, 3.II.2002, leg. W. Jaitrong (“WJT02-TH-0116”); 5 minor workers, West Malaysia, Kelantan, 60 km NE Tanah Rata, Tanah Kerajaan, 1000 m a.s.l., 12.–30.IV2007, leg. P. Cechovský.

Additional material: 3 pupae (Suppl. material 5) and 6 myrmecophilous crickets (Camponophilus sp., det. S. Ingrisch), from the same nest sample as the holotype.

For unique identification numbers of all 90 dry mounted specimens (60 minor workers, 8 major workers, 16 gynes, and 6 males), as well as information on caste and colony affiliation, see Suppl. material 6 “measurements”.

The holotype and a portion of the paratypes will be deposited at the Brunei Museum; additional paratypes will be housed in the Universiti Brunei Darussalam, the Natural History Museum Vienna, the University of California (Davis, USA), the Natural History Museum of Los Angeles County (Los Angeles, USA), the Thailand Natural History Museum (Technopolis, Thailand), and the collection of H. Zettel (Vienna, Austria).

In this study, three mitochondrial DNA loci and one nuclear DNA fragment were applied for the construction of a molecular phylogenetic tree (Fig. 1). The evolutionary analysis based on four loci showed that C. explodens sp. n. is clearly genetically distinct from morphologically similar species. A minor level of infraspecific polymorphism within the specimens from Brunei was observed for COI, COII, and in particular for cytB marker. While tree topologies based on single mitochondrial loci were concordant, the cad tree (nuclear locus) was not resolved for the entire COCY group. The comparison of COI sequences with several hundred COCY sequences available in our local database and 13 sequences deposited in NCBI (Nov. 2017) suggests that this marker can be used for the reliable molecular identification (DNA barcoding) of C. explodens sp. n., as COI sequences of the nearest COCY taxa share only 91% similarity (C. badia), and the similarity to the selected non-COCY species C. aruensis is 83%.

The DNA extraction from the gastral parts of the mandibular gland reservoirs of C. explodens sp. n. minor workers resulted in drastically low yields indicating no abundant microbial symbionts present there. The successful 16S rRNA PCR amplification gave a sharp band that was sequenced with high reproducibility. The 16S rRNA fragment corresponding to the whole genome sequenced Blochmannia endosymbiont of North American Colobopsis obliquus strain 757 (NCBI GenBank accession number CP010049, Williams and Wernegreen 2015) was 92 % similar to mOTU revealed in this study resulting in 56 polymorphic sites. This confirms that C. explodens sp. n. also harbours these bacteria that usually colonize the midgut of Camponotini workers (Sauer et al. 2002) and are considered to be beneficial for N-nutrition of these ants; they may also contribute to the general health of the workers and gynes. Thus, the detection of cf. Blochmannia bacteria rather indicated the contamination of the MG sample by fragments of the digestive system. In this respect, it is interesting to note that no Wolbachia (Alphaproteobacteria) mOTUs were recovered, but neither digestive tract nor ovaries were specifically investigated.

The treatment of Colobopsis as a genus separate from Camponotus is supported by molecular, morphological, and biological data (Blaimer et al. 2015, Ward et al. 2016). Naked pupae (Wheeler 1904; see Suppl. material 5) and presence of phragmotic soldiers and gynes are important features of Colobopsis, although unknown in many of the 94 valid species assigned to this genus by Ward et al. (2016). The morphological separation of minor workers of Colobopsis and Camponotus is chiefly based on head morphology, but complicated by extensive evolutionary changes within each group (Ward et al. 2016); however, the phylogeny of Colobopsis species has not been studied to date. The molecular data published by Blaimer et al. (2015), obtained from only four species, do not allow an interpretation of the relationships of intrageneric clades. Attempts to classify the species by morphological characters (Emery 1925, McArthur 2012), although useful for a rough sorting of species, probably hardly reflect their evolutionary relationships.

A first attempt of a comprehensive classification of the species of Colobopsis (as a subgenus of Camponotus) was done by Carlo Emery. In his outstanding treatment of Formicinae (Emery 1925) he treated 58 species and established six groups to hold 49 of them (nine remained unclassified). He defined the [Camponotus (Colobopsis)] cylindricus group by a gradual variation between worker and soldier, interspecific variation of head in soldiers and females (from concave and marginate to oblique and obtuse), and generally large size. Emery included eight species presently classified as Colobopsis (Ward et al. 2016), of which Colobopsis calva Emery, 1920, C. quadriceps (Smith, 1859), and C. smithiana (Wheeler, 1919) are not presently assigned to this group (see below), whereas C. badia and C. corallina were not included (listed under incertae sedis). Although Emery (1925) correctly recognized the size variation of workers, he failed to recognize the unique characteristics of the soldier caste (see Laciny et al. 2017).

More species of the COCY group were subsequently described by Stitz (1925), Menozzi (1926) and Karawajew (1929, 1935). A second attempt at classification was made by McArthur (2012): His Camponotus (Colobopsis) cylindricus group consists of species with “neck attached to head well below vertex” and is broader than Emery’s (1925)cylindricus group. It includes the following species (according to current classification) that do not fit the characteristics of the COCY group in the present sense: Colobopsis anderseni (McArthur & Shattuck, 2001), C. brachycephala Santschi, 1920, C. cotesii (Forel, 1893), C. desecta (Smith, 1860), C. excavata (Donisthorpe, 1948), C. hosei (Forel, 1911a), C. mutilata (Smith, 1859), and C. quadriceps, as well as Camponotus dedalus Forel, 1911b, and Camponotus kutteri Forel, 1915.

According to our morphological studies the COCY group can be defined as such: polymorphic Colobopsis with at least three distinct female castes: (i) winged, phragmotic gynes, (ii) phragmotic soldiers (doorkeepers), and (iii) minor workers with a considerable size variation; intermorphic workers may occur in addition (Laciny et al. 2017). Minor workers: Vertex highly raised above foramen. Eyes of minor worker small and flat, not breaking head sides in full-face view. Entire trunk with dense, reticulated microstructures; punctures of integument often reduced. Head with moderate, mesosoma with dense pubescence. Mesosoma (at least the pronotum, except in C. clerodendri Emery, 1887) dorsally with long undulated setae, never arranged in distinct rows. Gaster with appressed pubescence and two or three types of setae of different lengths (not arranged in rows, except at hind margin). Soldiers (not known of all taxa): differing from minor workers by enlarged heads and short appendages (antennae, palpi, legs); in most species with a clearly circumscribed head shield for phragmosis. Microsculpture and pilosity similar to minor worker.

Following this definition, the COCY group presently comprises 17 names in the rank of species, subspecies or variations, which are partly in synonymy to each other. The strong intraspecific variation of minor workers, the frequently lacking knowledge on soldiers (or gynes), and the historical descriptions of taxa from different morphs (either workers or gynes) make the species taxonomy a true challenge. A preliminary analysis of morphological and molecular data (unpublished) supports the division of the group into four species complexes (molecular data of one complex not available). We restrict the following analysis to the species of the C. saundersi complex, which includes C. explodens sp. n. and can be defined by the following combination of characters observable in minor workers and soldiers: head always red or brown (not black); mesosoma moderately elongated and dorsally with some long undulated setae, at least on pronotum; node of petiole without long setae; gastral tergites with dense (in most species strongly transverse) micro-reticulation and with small hair pits (without large grooves). Soldiers and gynes (not known of all species) have a strongly truncated head, in most species with a well-defined (crested) head shield. The following names are available in this group (listed chronologically): Colobopsis badia (Smith, 1857), C. corallina Roger, 1863, C. saundersi Emery, 1889, C. badia var. krama Forel, 1912, C. badia saginata Stitz, 1925, C. solenobia (Menozzi, 1926), and Colobopsis trieterica (Menozzi, 1926), comb. n.

Colobopsis corallina (=C. solenobia syn. n., =C. trieterica syn. n.) is an endemic species from the Philippines. Soldiers and gynes differ strongly from C. explodens sp. n. and other taxa of the complex (as far as such morphs are known) by a very obtusely margined, not crested head shield. Minors have a bright orange colour on head, mesosoma, and petiole, often extending to gastral tergite I. Morphometrically, the examined minor workers of C. corallina (n = 31) mainly differ from those of C. explodens sp. n. by a greater average length of appendages (SI 92–109 vs. 87–104; FeI 136–159 vs. 123–151; PSI 30–39 vs. 28–35).

The greatest similarity is observed between C. explodens sp. n. and C. saginata (stat. n.), a taxon only known from a single alate gyne from Northern Borneo. The important structures of the head shield are almost identical. Although strongly different from C. explodens sp. n. by pale orange brown colour, this gyne differs only by subtle morphometric characters of which the long and distally wide scape seems to be the most reliable (SI 83 vs. 78–80). The length of vein 4Rs+M is considerably longer in C. saginata than in C. explodens sp. n. (WVI 65 vs. 26–58).

Colobopsis badia var. krama, described from a soldier from Java (Forel 1912) is a very poorly known taxon. We have not been able to study any further material from Java yet. The type (illustrated by AntWeb.org under CASENT0910610) differs from C. explodens sp. n. by a pale red head that strongly contrasts with the dark brown mesosoma, by a well-developed median carina of the head shield that reaches the foremargin of the clypeus, and by a stronger punctation of the preocellar area.

Colobopsis badia was described by Smith (1857) from Singapore and Sarawak (Borneo). However, the original description is too brief to draw any meaningful taxonomic conclusions. Viehmeyer (1916) describes workers of this species in more detail, also noting the secretion of a sticky liquid upon capture. He mentions a strong variability in colouration (from red to almost black with reddish head) and propodeal shape. This raises the question whether all examined specimens were truly members of the same species or perhaps belonged to one of the other, similar species of the C. saundersi complex. We examined the three syntype minor workers of C. badia in the Oxford University Museum of Natural History. To fix the identity of this taxon, we have chosen the syntype from Singapore (imaged by AntWeb.org under CASENT0901897) as the lectotype of Formica badia. The two syntypes from Sarawak are in a relatively poor condition, which does not allow a complete morphometric analysis, and therefore the conspecificity with the C. badia type remains doubtful. We were not successful in obtaining fresh material of C. badia from Singapore, but a nest sample (minors only) from southern Thailand (Trang Province) which agrees well with the lectotype in morphology, especially morphometry, was available for a molecular analysis. It shows that C. badia and C. explodens sp. n. are closely related, but distinct (Fig. 1). Although very similar to C. explodens sp. n. in overall habitus and colouration, the examined C. badia minor workers are on average somewhat larger with less size variation (HW 1.22–1.57 vs. 1.46–1.59) and possess longer appendages (e.g., FeI 123–151 vs. 141–168; see Fig. 8a).

We examined two syntype minor workers of Colobopsis saundersi from Myanmar (“Tenasserim, Thagata”, one illustrated by AntWeb.org under CASENT0905463). Morphometric analysis revealed no differences between the types of C. saundersi and C. badia, suggesting that the two species should be synonymized. Colobopsis saundersiwas considered a junior synonym of C. badia by Carlo Emery himself (Emery 1896) but revived from synonymy by Bingham (1903) without providing a reason. The large geographic distance of the type locality of C. saundersi and some minor differences in morphology led us to the decision to refrain from a formal synonymization at this time. A comparative molecular analysis of specimens from the type localities (Myanmar, Singapore) would most likely be necessary to corroborate this synonymy.

The morphology of males of the tribe Camponotini is insufficiently studied, so that a complete comparison at generic level is not possible. The modified (enlarged) first funicular segment is presumably characteristic for males of Colobopsis. This characteristic has been described in the type species, Colobopsis truncata (Spinola, 1808), by Kutter (1977) and has been equally observed in several species of the C. cylindrica group.

Males of the COCY group have previously been described for three species (see below). However, these descriptions largely lack the necessary details to meaningfully compare taxa. No previous accounts of genital morphology or illustrations of male specimens have been found in the literature.

Colobopsis badia: Viehmeyer (1916) gives a brief description of a male from Singapore. Colouration, size, proportions of head and ocelli, as well as the enlarged first funicular segment correspond well to the examined male from Thailand.

Colobopsis severini (Forel, 1909): The extremely brief description of a male from the island Labuan (near Borneo) is not sufficient to draw any meaningful taxonomic conclusions.

Colobopsis leonardi (Emery, 1889): Karawajew (1929) gives a rather detailed description of males collected within a nest-series on Sumatra. The correct species identification by Karawajew is uncertain; the series may belong to another species of the C. leonardi complex as well. The pattern of pilosity on the gaster, with standing setae only present on the posterior half, also corresponds to our observations in males of the C. saundersi complex.

According to our knowledge, males of the C. cylindrica group can be distinguished from other Southeast Asian Colobopsis species by the relatively rich subdecumbent pilosity and the dense microreticulation of gastral tergites.

The behavioural observations on C. explodens sp. n. conducted at KBFSC revealed multiple modes of behaviour that are either poorly studied or new to science.

The diurnal activity pattern, as well as the positive influence of high temperatures correspond to the results of previous studies in related taxa (see Hamdan et al. 2013). Similarly large colonies containing several thousand individuals and extending to multiple trees and / or artificial nesting structures have also been described for other members of the genus (Federle et al. 1998, Laciny et al. 2017). However, it is still unclear whether individual workers are linked to certain parts of the colony or whether all foragers can move freely through the entire territory of the colony. An interesting and hitherto undescribed phenomenon in this regard is the presence of one or multiple “guards” at the artificial nest’s entrance: These minor workers were frequently observed to touch any incoming or leaving workers with their antennae. In some instances, returning foragers were delayed or altogether denied entrance by the guarding ants. One reason for this may be that some workers are linked to different parts of the colony. Alternatively, the observed guarding behaviour may be related to the limited capacity of the artificial nest, which is also suggested by the conspicuously balanced numbers of workers entering and leaving the nest during times of foraging activity. These behavioural patterns are hitherto undescribed and must be investigated in future studies.

A further noteworthy activity observed in foraging workers was so-called “grazing” behaviour, in which minor workers were frequently seen using their mandibles to pluck and chew at various mosses, lichens, and other epiphytes on the bark of trees or other surfaces. While this activity can last for up to one hour at a time, its exact purpose remains unclear. It is possible that minor workers cut and consume parts of the plants and microorganisms or merely ingest fluids. As previous analyses of nitrogen isotopes (Davidson et al. 2016) suggest a largely plant-based diet for COCY ants, it seems likely that “grazing” contributes to their nutrition. However, other previously hypothesized modes of nutrition, such as tending of scale insects (Davidson et al. 2016) were not observed, and recent investigations on Colobopsis leonardi (Emery, 1889) (Zettel et al., ms submitted to Asian Myrmecology) even suggest a higher prevalence of carnivory in COCY ants than previously suspected.