Surviving the winter: Tetramorium sibiricum n. sp., a new Central Siberian ant species (Hymenoptera: Formicidae)

In 2019, the author received eleven nest samples of Tetramorium workers collected by Igor ANTONOV in two sites in the surroundings of Irkutsk and two sites in stony steppe habitats southwest of Ulan Ude. With mean January temperatures of – 21 °C (Irkutsk) and – 24 °C (Ulan Ude), these Central Siberian sites represent the most winter-cold places worldwide from which Tetramorium ants are known so far. All collected ants belonged to the Tetramorium caespitum complex – an ant group in which some species appear almost inseparable based upon worker morphology. Subjective visual inspection of the samples surprisingly suggested that the ants from near Ulan Ude should represent another species than those from Irkutsk.

In 2019, the author received eleven nest samples of Tetramorium workers collected by Igor ANTONOV in two sites in the surroundings of Irkutsk and two sites in stony steppe habitats southwest of Ulan Ude. With mean January temperatures of -21 °C (Irkutsk) and -24 °C (Ulan Ude), these Central Siberian sites represent the most winter-cold places worldwide from which Tetramorium ants are known so far. All collected ants belonged to the Tetramorium caespitum complex -an ant group in which some species appear almost inseparable based upon worker morphology.
Subjective visual inspection of the samples surprisingly suggested that the ants from near Ulan Ude should represent another species than those from Irkutsk.
The recent revision of the Westpalaearctic members of the Tetramorium caespitum complex published by WAGNER et al. (2017) strongly improved the taxonomic knowledge on this difficult ant group and distinguished ten species: Tetramorium caespitum (LINNAEUS, 1758), T. impurum (FOERSTER, 1850), T. indocile SANTSCHI, 1927, T. immigrans SANTSCHI, 1927, T. hungaricum RÖSZLER, 1935, T. staerckei KRATOCHVIL, 1944, T. fusciclava CONSANI & ZANGHERI, 1952, T. alpestre STEINER et al. 2010, T. breviscapus WAGNER et al., 2017and T. caucasicum WAGNER et al., 2017. Based on the known climate niche, only two of these species would be capable of postglacial invasion of the Baikal region. This is firstly T. caespitum, as the only species of the group known from Fennoscandia and found there north to 63 °N (SEIFERT, 2018) and secondly T. indocile which is known from near Chelyabinsk in Western Siberia and the montane zone of the Tian Shan (WAGNER et al., 2017). Using exploratory data analyses of a very complex morphological character system, this paper provides evidence for the existence of an undescribed Central Siberian species Tetramorium sibiricum n.sp. and of two further species occurring in Siberia west of the zoogeographic divide known as the REINIG Line (DE LATTIN, 1967).

MATERIAL AND METHODS
A total of 111 nest samples with 359 worker individuals were investigated. These were 7 samples with 21 workers of Tetramorium sibiricum n. sp., 20 samples with 72 workers of T. indocile and 84 samples with 266 workers of T. caespitum. The material originated from the whole known Palaearctic range of these species with exception of unclear samples near to T. caespitum from Iberia. As a rule, worker ants were taken directly from the nest populations -hence representing nest samples of close kinship. Detailed data on geographic origin and morphology of these samples are given in the digital supplementary information SI1 and SI2. Equipment and methodology of stereomicroscopic evaluation is described elsewhere (e.g. SEIFERT, 2020). Numeric description was performed in a total of 35 characters which are more thoroughly described in STEINER et al. (2010) andWAGNER et al. (2017). I repeat here the character descriptions with the wording of some characters partially changed (box A + figures 1-3).  Box A. Phenotypic characters used in this study (after STEINER et al., 2010 andWAGNER et al., 2017, partially changed).
 ALPH. Angle between the imaginary line from spine tip to center of propodeal stigma and the imaginary line from spine tip to caudodorsal corner of the propodeal lobe (figure 1); calculated as cos α = (SPST² + PLSP² -PLST²) / (2*SPST*PLSP).  BETA. Angle between the imaginary line from anterioventral corner of the metapleuron to center of propodeal stigma and the imaginary line from anterioventral corner of the metapleuron to the caudodorsal corner of propodeal lobe; calculated as cos β = (MPST² + MPPL² -PLST²) / (2*MPST*MPPL).  CL. Maximum cephalic length in median line; the head must be carefully tilted to the position with the true maximum; excavations of occiput and/or clypeus reduce CL. Surface irregularities due to sculpture, carinae in particular, are considered by averaging between peaks and valleys of sculpture.
 CS. cephalic size; the arithmetic mean of CL and CW, used as a less variable indicator of body size.  CW. maximum cephalic width across eyes.  dCV. Mean distance of longitudinal carinae /carinulae of sculpture on central vertex in dorsal view. The transversal counting line is placed at level of anterior eye margin and is demarcated by the imagined prolongations of posterior parts of the frontal carinae; the distance between the frontal carinae is divided by the number of longitudinal sculpture elements crossing it; sculpture elements just touching the measuring line and those exactly at its endpoints are counted as 0.5.  DELT. Angle between the imaginary line from caudodorsal corner of propodeal lobe to spine tip and the imaginary line from caudodordal corner of propodeal lobe to anterioventral corner of metapleuron; calculated as cos δ = (PLSP² + MPPL² -MPSP²) / (2*PLSP*MPPL).

MC1TG = 18
 EL. Maximum large diameter of the elliptic eye. All structurally defined ommatidiae, pigmented or not, are included.  EPSI. Angle between the imaginary line from the center of the propodeal stigma to the caudodorsal corner of the propodeal lobe and the imaginary line from the center of the propodeal stigma to the anterioventral corner of the ventral margin of the metapleuron; calculated as cos ε = (PLST² + MPST² -MPPL²) / (2*PLST*MPST).  EW. Maximum small diameter of the elliptic eye. All structurally defined ommatidiae, pigmented or not, are included.  FL. Maximum anterior distance of frontal carinae; when FL is not defined because frontal carinae converge frontad anterior of the FRS level, FL is taken equal to FRS.  FRS. Distance of the frontal carinae immediately caudal of the posterior intersection points between frontal carinae and the lamellae dorsal of the torulus. If these dorsal lamellae do not laterally surpass the frontal carinae, the deepest point of scape corner pits may be taken as reference line. These pits take up the inner corner of scape base when the scape is fully switched caudad and produce a dark triangular shadow in the lateral frontal lobes immediately posterior of the dorsal lamellae of scape joint capsule. Only measured, if FL is not defined.  GAMM. Angle between the imaginary line from the center of the propodeal stigma to the tip of the spine and the imaginary line from the center of the propodeal stigma to the caudodorsal corner of the propodeal lobe; calculated as cos γ = (SPST² + PLST² -PLSP²) / (2*SPST*PLST).  ML. Mesosoma length measured in dorsal view from the caudalmost portion of propodeum to the anteriormost margin of the pronotal slope (i. e. the breaking point where the coarser sculpture of the pronotal slope changes into the finer sculpture of anterior pronotal shield); equivalent measuring also possible in lateral view.  MPPL. Distance between anterioventral corner of the ventrolateral margin of the metapleuron and caudodorsal corner of propodeal lobe (figure 1). When there is no corner, measure to the most distant point of caudodorsal propodeal lobe. This does not necessarily mean a coincidence of this point with PL but the effective error in measuring MPPL is low.  MPSP. Maximum distance between the anterioventral corner of the metapleuron and the tip of the spine (figure 1).  MC1TG. Quantification of stickman-like or reticulate microsculpture units on anterior surface of 1st gastral tergite (use > 150 × magnification): number of connected lines building units and being separated by line intersections and by angular direction changes > 10° is counted. Also very short lines are full counts. Arithmetic means of at least three units per worker are taken. In example of figure 2, MC1TG is 18.  MPST. Maximum distance from the anterioventral corner of the ventrolateral margin of the metapleuron to the center of the propodeal stigma (figure 1).  MW. Maximum mesosoma width.  PeH. Maximum petiole height; the more or less straight section of ventral petiolar profile at node level is the reference line perpendicular to which the maximum height of petiole node is measured at node level. Note that maximum PEH is not necessarily found at the level of node top; it is the longest section line achieved during shifting of the petiole from frontal to caudal.  PeNL. Maximum length of petiolar node from the dorsocaudal corner of petiole to anterior petiolar slope; the measuring line is directed orthogonal to a reference line formed by the dorsalmost linear part of anterior petiolar slope.  PeW. Maximum width of petiole.  PLSP. Distance between the summit of the caudodorsal corner of the propodeal lobe to the tip of the spine (figure 1).  PLST. Distance between the summit of the caudodorsal corner of the propodeal lobe and the center of the propodeal spine (figure 1).  PoOc. Postocular distance; using the cross-scaled ocular micrometer the head is adjusted to the measuring position of CL; caudal measuring point: median occipital margin as average between peaks and valleys of microsculpture; frontal measuring point: median head at the level of the posterior eye margin; because of head asymmetries, the average of left and right postocular distance is calculated.  PosSPl. Orthogonal distance of the uppermost point of spine tip to a ventral reference line of the mesosoma. The ventral reference line is the line from the ventralmost point of the pronotum to the ventralmost point of the metapleuron (figure 3). With the graduated scale of the ocular micrometer directed perpendicular in the visual field, this reference line is brought to coincidence with the horizontal line of the cross-scale at magnifications of about × 100. Then, at magnifications of ≥ × 250, the mesosoma is carefully tilted to a position in which the ventralmost point of the metapleuron and the tip of the spine are at the same focal level. Note that this character is not necessarily measured in lateral view, but frequently in a dorsolateral view.  PosSPu. Measured in the same adjustment as for PosSPl and orthogonal to the ventral reference line. The distance of the uppermost point of spine tip to the dorsalmost point of propodeum. Note that this point is in the given adjustment usually behind the median line of propodeum and can only be found after focussing movements. With the graduated scale of the ocular micrometer kept perpendicular in the visual field, focussing cannot induce a parallax error.  POTCos. Number of post-oculo temporal costae and costulae as mean of both head sides. With head in lateral view and the longitudinal axis of head adjusted horizontally, counted by focussing along a perpendicular line from the caudalmost point of eye down to underside of head. Costae/ costulae just touching the measuring line are counted as 0.5, those positioned just at ventral margin of head silhouette are not counted. A costa beginning exactly at the upper reference point is counted 0.5.  PpH. Maximum height of postpetiole measured perpendicular to the suture between dorsal and ventral sclerites. Note that maximum PpH is not necessarily found at the level of node top; it is the longest section line achieved during shifting of the petiole from frontal to caudal.  PpW. Maximum width of postpetiole.  PreOc. Preocular distance in lateral view; measured as the shortest distance between the anterior eye margin and the sharp frontal margin of the gena. Do not confuse this margin with the edge of the basal mandibular wulst.  PnHL. Length of the longest hair near to the corner of antolateral pronotum; arithmetic mean of both sides.  SLd. Maximum straight line scape length in dorsal view from the most proximal point of basal scape lobe to most distal point of scape end.  SPBA. Smallest distance of the outer margins of the spines at their base, measured in dorsofrontal view. If the lateral margins of spines diverge continuously from the tip to the base, a smallest distance at base is not defined. In this case, SPBA is measured at the level of the bottom of the interspinal meniscus.  SPST. Distance between the tip of the spine and the center of the propodeal stigma (figure 1).  SPWI. Maximum distance between outer margins of spines; measured in same position as SPBA.  SWd. Maximum scape width which is given in the distal third of scape; transmitted light is used to measure the real cuticular surface and not the pubescence surface. In order to show in comparative tables interspecific differences independent of body size and to increase resolution of principle component analyses, a removal of allometric variance (RAV) was performed for worker ants following the basic procedure described by SEIFERT (2008). Evaluation of scatter plots suggested the use of linear monophasic allometry functions. RAV was calculated assuming all individuals to have a cephalic size of CS = 0.75 mm. RAV functions were calculated as the arithmetic mean of the species-specific functions of eight Palaearctic species with more than 50 workers measured per species -i. e. Tetramorium alpestre, T. caespitum, T. fusciclava, T. hungaricum, T. impurum, T. immigrans, T. indocile and T. staerckei (for primary data see supplementary information SI2). The RAV functions are given in the box B.
Using these 34 RAV-corrected characters and cephalic size CS as input, three different forms of exploratory data analyses (EDA) using nest centroids as input data were run (NC clustering, SEIFERT et al., 2013). These were hierarchical NC-Ward clustering, the hierarchical method NC-part.hclust and the iterative vector-quantization method NCpart.kmeans -the latter two methods are implemented in partitioning algorithms based on recursive thresholding (for details see CSŐSZ & FISHER 2015). A controlling linear discriminant analysis (LDA) was run with species hypotheses being formed for samples in which the EDA's arrived at identical classifications whereas samples with controversial classification were run as wild-cards. The final classification ("final species hypothesis") was established by the LDA in an iterative procedure and there remained no undecided cases even if their posterior probabilities were close to 0. Only the first seven components extracted by this PCA were then used as input in a LDA which solved the overfitting problem in T. sibiricum n. sp. that was confirmed as a clearly separate cluster with posterior probabilities of p > 0.998 for each sample (figure 6). As a result, all three exploratory data analyses fully agreed in classifications of T. sibiricum n.sp. whereas the few disagreements (a mean of 1.8 % of the total) occurred in the T. indocile and T. caespitum samples.
These data show that three species of the T. caespitum complex occur in Siberia west and north of the Reinig Line. This faunal divide separates East Siberian, Inner Mongolian, Chinese and Tibetan species from those of Central Siberia, West Siberia and the Turanian region (DE LATTIN, 1967    The methodology of species identification applied here is extremely complex and data recording is by far the most time-consuming of any ant genus investigated so far by the author -a challenge even for experienced and specialized investigators having access to adequate equipment. An attempt to simplify the separation of the three West and Central Siberian species was done in writing a key that uses absolute primary data (i. e. without RAV-correction) and a strongly reduced character set.

Etymology
The name of the species refers to the region where the species is occurring.

Type material
Depository of all specimens: SMN Görlitz (Germany

Geographic range
Only known so far from the two type localities. Apparently, a species restricted to Siberia.

Description
Worker (table I,

Consideration of synonymies
T. sibiricum n. sp. shows a rather unique character combination making its identification comparatively easy. It has been stated above that taxa described from east and southeast of the Reinig Line -from China, Korea and Japan in particular -are very unlikely to have invaded the extremely winter-cold Central Siberia. Apart from climatic reasons, there are also morphological arguments to exclude synonymies. The images of a type of Tetramorium jacoti WHEELER, 1927 (specimen CASENT0901253 in www.antweb.org), described from Tartar City / Peking and of a type of T. tsushimae EMERY, 1925 (CASENT0904809 in www.antweb.org), described from Tsushima /Japan probably refer to the same species. This assumption is based on similarities in sculpture and mesosomal shape and zoogeographic proximity. These types differ from the T. sibiricum n. sp. condition in having shorter mesosomal setae and the spine tips reaching more dorsad relative to the level of the dorsal mesosomal profile. Tetramorium annectens PISARSKI,1969 from Pei-Hai / Peking was described by WHEELER (1927) to have the petiole and postpetiole "subopaque and sculptured throughout, the former irregularly rugulose, the latter regularly longitudinally rugulose". This strongly contrasts the situation in T. sibiricum n. sp. and is confirmed by the image of a T. annectens type (CASENT0916601 in www.antweb.org). The image additionally shows a remarkably strong longitudinal sculpture at the base of the first gaster tergite -a character not observed in any species of the TCC. Notably, these two characters strongly contradict BOLTON's proposal (BOLTON, 1995) of T. annectens being a junior synonym of T. tsushimae. Four taxa described on species level from China by WANG et al. (1988) are fully excluded as synonyms of T. sibiricum n. sp. due to most deviating morphology and climate niches.
It was stated above that there is no taxon other than T. caespitum and T. indocile from west and southwest of the Reinig line approaching T. sibiricum n. sp. in the combination of phenotype and climate niche. Yet, Tetramorium caespitum var. barabensis RUZSKY, 1925, described from Karachinskoe Ozero (probably 55.35 °N, 76.96 °E) near Tomsk / West Siberia requires consideration. The original description gives the following data: "Head weakly shiny; dorsum of promesonotum very smooth ('so znachitel'no sglazhennoi skul'pturoj'); sides of both waist segments fully smooth, furthermore postpetiole narrowed, having the width of petiole, from above flatter ('sverkhu bolee ploskij'). Color light, head brownish-cinnamon, mesosoma brownishyellowish; mandibles, scapes and tarsae yellow. Size 2.1-3 mm. Nesting in soil under a stone in rout sprouts of Stipa pennata, sunny tschernozem northeast of the health resort." This description makes a synonymy with T. sibiricum n. sp. unlikely. The weak sculpture of T. barabense rather points to T. caespitum or T. indocile as well as the mean ratio PpW/PeW which is 1.349 in T. sibiricum n. sp. but 1.286 in T. caespitum and 1.284 in T. indocile. According to RADCHENKO (1992), types of T. barabense are unknown but, if they should be discovered, the taxon could turn out as senior synonym of T. indocile. The former taxon is considered here as incertae sedis as long as no conclusive species identification is possible.

Biology
All seven nests were collected under stones. The sparse information currently available suggests that Tetramorium sibiricum n. sp. is distributed in more natural steppe habitats whereas sympatric T. caespitum seems to occur in habitats with anthropogenous influence.