Taxonomy of the hyper-diverse ant genus Tetramorium Mayr in the Malagasy region (Hymenoptera, Formicidae, Myrmicinae) – first record of the T. setigerum species group and additions to the Malagasy species groups with an updated illustrated identification key

Abstract In this study we provide an update to the taxonomy of the ant genus Tetramorium Mayr in Madagascar. We report the first record of the Tetramorium setigerum species group in Madagascar and describe the only Malagasy representative as Tetramorium cavernicola sp. n., which is known only from a cave in Ankarana. In addition, we provide an overview of the 19 proposed Malagasy species groups, and discuss their zoogeography and relationships to other groups and larger lineages within the hyper-diverse genus Tetramorium. At present, we recognise a highly unique Malagasy Tetramorium fauna with 113 species endemic to the island of Madagascar out of a total of 125 translating into an endemism rate of 93%. We hypothesise that this fauna is based on one or a few colonisation events from the Afrotropical region, with subsequent adaptive radiation in Madagascar. Furthermore, we present an updated and illustrated identification key to the Tetramorium species groups in the Malagasy region.


Introduction
The genus Tetramorium Mayr, widely distributed throughout all zoogeographical regions, is among the most species-rich ant genera in the world. Currently, we recognise around 600 valid species, but because the authors are aware of a larger number of undescribed species, we expect the total count to be closer to 700 or more species. Most Tetramorium species are found in the tropics and subtropics of the Old World, where the genus can be considered hyper-diverse by the definition of Wilson (2003), and are key elements of most local ant communities, especially in the Afrotropical and Malagasy regions. Recent studies in the latter region have revealed an astonishingly diverse and highly endemic Tetramorium fauna consisting of 107 valid species plus approximately 17 undescribed species (Hita Garcia and Fisher 2011, 2012a, 2012b, 2014b. Based on these figures, Tetramorium is by far the most species-rich ant genus in Madagascar, where it seems to have undergone a radiation that was particularly successful in the forested eastern and northern areas of the island. On a global scale, Bolton (1976Bolton ( , 1977Bolton ( , 1979Bolton ( , 1980 revised the taxonomy of most regional faunas with the exception of the Palaearctic region. These works provided an excellent foundation for the many later revisions or treatments of species groups/complexes or regions/subregions (e.g. Csösz et al. 2007;Csösz and Schulz 2010;Hita Garcia et al. 2010; Hita Garcia and Fisher 2011, 2012a, 2012b, 2013, 2014a, 2014bBharti and Kumar 2012;Sharaf et al. 2012;Vásquez Bolaños et al. 2011). The Malagasy Tetramorium fauna was first monographed by Bolton (1979), who treated eight species groups with 36 species (29 of these endemic to Madagascar). The later synonymisation of Triglyphothrix Forel (Bolton 1985) under Tetramorium added an additional species group with one tramp species; two additional tramp species have been recorded since then (Blard et al. 2003;Roberts and McGlynn 2004). This means that 39 Malagasy Tetramorium species were known prior to 2011. In that year, we began a large-scale taxonomic revision of the genus for the Malagasy region based initially on more than 160 morphospecies with more than 40,000 mounted specimens. As a foundation for a series of monographs, we proposed 14 species groups for the Malagasy region and provided a preliminary identification key to these groups (Hita Garcia and Fisher 2011). In addition, we revised the T. bicarinatum, T. obesum, T. sericeiventre, and T. tosii species groups. In that study we described one species and sank another to the rank of junior synonym, which did not change the total species richness for the region. Based on that work, we revised the T. bessonii, T. bonibony, T. dysalum, T. kelleri, T. marginatum, T. tortuosum, T. tsingy, and T. weitzeckeri species groups shortly afterwards (Hita Fisher 2012a, 2012b). These studies treated 58 species, of which 45 were described as new, and raised the species count for the region to 84. We also proposed additional species groups leading to a total of 18 for the region. In the most recent study we revised four additional groups: T. naganum, T. plesiarum, T. schaufussii, and T. severini (Hita Garcia and Fisher 2014b). We treated 31 species, of which 22 were newly described, and raised one junior synonym to species status. This increased the current species count for Malagasy Tetramorium to 107 (not 106 as mentioned in the introduction of Hita Garcia and Fisher 2014b).
In this study we report the first record of the presence of the Afrotropical T. setigerum species group on Madagascar and describe the single representative in the region as a new species, T. cavernicola sp. n.. With the T. setigerum group, there are now 17 Malagasy species groups that have undergone a current taxonomic revision. Nevertheless, the last two groups, the T. ranarum and the T. simillimum groups, have not been revised since Bolton (1979). The revisions of these two groups are currently in preparation. In addition to the revision of the T. setigerum group, in this study we also present an updated discussion on the currently recognised Malagasy species groups. We give an overview, discuss their biographical affinities, and try to assess their relationships to other key lineages within this hyper-diverse genus. Also, as a consequence of the recent revisions (Hita Garcia and Fisher 2012a, 2012b, 2014b this study) that proposed additional species groups and added species not known during the preparation of the species group key published in Hita Garcia and Fisher (2011), we present an updated illustrated key to the 19 proposed species groups.

Abbreviations of depositories
The collection abbreviations follow Evenhuis (2014). The material upon which this study is based is located and/or was examined at the following institutions:

BMNH
The Natural History Museum (British Museum, Natural History), London, U.K.

CASC
California Academy of Sciences, San Francisco, California, U.S.A.

Material and methods
The material examined for this study and the previous Malagasy Tetramorium revisions (Hita Garcia and Fisher 2011, 2012a, 2012b, 2014b was collected during ant inventories carried out in the Malagasy region from 1992 to 2013. These inventories included material from more than 6,000 leaf litter samples, 4,000 pitfall traps, and 9,000 additional hand collecting events (see Fisher 2005 for additional details). All new type material and all imaged specimens can be uniquely identified by specimenlevel codes affixed to each pin (e.g. CASENT0247028). Digital colour montage images were created using a JVC KY-F75 digital camera and Syncroscopy Auto-Montage software (version 5.0), or a Leica DFC 425 camera in combination with the Leica Application Suite software (version 3.8). All images used for the colour plates illustrat-ing the identification key or for the presentation of species are available online and can be seen on AntWeb (http://www.antweb.org). We predominantly have used images of valid species, but in a few cases used images of undescribed species. The latter have morphospecies codes (e.g. Tetramorium fhg-forc) and can be seen under their respective codes on AntWeb. The distribution map provided below was generated with R (R Core Team 2014). Morphometric measurements were performed with a Leica MZ 12.5 equipped with an orthogonal pair of micrometres at a magnification of 100×. Measurements and indices are presented as minimum and maximum values with arithmetic means in parentheses. In addition, all measurements are expressed in mm to two decimal places. The measurements and indices used in this study follow Hita Garcia and Fisher (2011, 2012a, 2012b, 2013, 2014a, 2014b:

HL
Head length: maximum distance from the midpoint of the anterior clypeal margin to the midpoint of the posterior margin of head, measured in full-face view. Impressions on the anterior clypeal margin and the posterior head margin reduce head length. HW Head width: width of the head directly behind the eyes measured in full-face view.

SL
Scape length: maximum scape length excluding basal condyle and neck.

EL
Eye length: maximum diameter of compound eye measured in oblique lateral view.

PH
Pronotal height: maximum height of the pronotum measured in lateral view.

PW
Pronotal width: maximum width of the pronotum measured in dorsal view.

WL
Weber's length: diagonal length of the mesosoma in lateral view from the posteroventral margin of propodeal lobe to the anteriormost point of pronotal slope, excluding the neck. PSL Propodeal spine length: in dorsofrontal view the tip of the measured spine, its base, and the centre of the propodeal concavity between the spines must all be in focus. Using a dual-axis micrometre the spine length is measured from the tip of the spine to a virtual point at its base where the spine axis meets orthogonally with a line leading to the median point of the concavity. PTH Petiolar node height: maximum height of the petiolar node measured in lateral view from the highest (median) point of the node to the ventral outline. The measuring line is placed at an orthogonal angle to the ventral outline of the node. PTL Petiolar node length: maximum length of the dorsal face of the petiolar node from the anterodorsal to the posterodorsal angle, measured in dorsal view excluding the peduncle. PTW Petiolar node width: maximum width of the dorsal face of the petiolar node measured in dorsal view. PPH Postpetiole height: maximum height of the postpetiole measured in lateral view from the highest (median) point of the node to the ventral outline. The measuring line is placed at an orthogonal angle to the ventral outline of the node. PPL Postpetiole length: maximum length of the postpetiole measured in dorsal view. PPW Postpetiole width: maximum width of the postpetiole measured in dorsal view.

Overview Malagasy species groups
The Tetramorium ant fauna of the Malagasy region can be divided into 19 species groups that represent different major lineages within this hyper-diverse genus. However, not all groups are native to the region. A proper assessment of the biogeographical affinities of a region, such as Madagascar and its surrounding South West Indian Ocean island systems, is only possible if comprehensive knowledge on the origin and distribution of each species group is available. This is of special importance when dealing with hyper-diverse genera that possess hundreds of species and dozens of evolutionary lineages throughout most or all zoogeographical regions. Fortunately, in the case of Malagasy Tetramorium we are able to assess the whole fauna and classify the groups into native, exotic, or shared with the Afrotropical region. Six of the 19 groups are either completely exotic, contain partly global tramps, or species of African origin. Two of these, the T. bicarinatum and T. obesum groups, are only present in the region with a few very successful global tramp species that very likely originated in the Oriental and Indo-Australian regions (Bolton 1977(Bolton , 1979McGlynn 1999;Wetterer 2010;Hita Garcia and Fisher 2011). Four other groups are clearly Afrotropical in origin: the T. sericeiventre, T. setigerum, T. simillimum, and T. weitzeckeri groups. The Malagasy representatives of these groups are either species that have likely been recently transferred from eastern and/or southern Africa by humans, or species that are Malagasy endemics that have evolved from much older colonisation events from Africa, or combinations of both. The T. simillimum group in the Malagasy region is a good example of the latter. It contains two global tramp species of African origin (T. caldarium (Roger) and T. simillimum (Smith), one very widespread non-tramp species of African origin (T. delagoense Forel), as well as some species endemic to the Malagasy region (T. anodontion Bolton, T. scytalum Bolton, and a few undescribed species). The T. sericeiventre group has one species that is widespread in the southern Palaearctic, Afrotropical, and Malagasy regions (T. sericeiventre Emery) and one species endemic to Madagascar (T. mahafaly Hita Garcia & Fisher). The T. weitzeckeri group is represented by T. humbloti Forel, a widespread species distributed in eastern and southern Africa that is probably a more recent introduction to the Malagasy region. The T. setigerum group, here recorded for the first time from the Malagasy region, contains one recently discovered species endemic to Madagascar that we describe below.
Most of the abovementioned, mostly non-native, groups possess twelve-segmented antennae and a triangular to dentiform sting appendage (except the T. weitzeckeri group, which has eleven-segmented antennae and a spatulate sting appendage). The only other Malagasy species group with twelve-segmented antennae is the T. tosii group, which seems to be endemic to Madagascar (Bolton 1979;Hita Garcia and Fisher 2011). In Hita Garcia and Fisher (2011) we discussed the very strong morphological similarities between the T. tosii group and some Afrotropical members of the T. setigerum group, which was not known from Madagascar at the time. At present, we propose to keep both groups separate until more data becomes available, but provide a more thorough discussion in the species group treatment of the T. setigerum group below.
One intriguing finding of the recent revisions (Hita Garcia and Fisher 2011, 2012a, 2012b, 2014b is that the vast majority of native species groups (T. bessonii, T. bonibony, T. dysalum, T. kelleri, T. marginatum, T. naganum, T. plesiarum, T. ranarum, T. schaufussi, T. severini, T. tsingy, and T. tortuosum) described from Madagascar share the three following key characters: eleven-segmented antennae, anterior clypeal margin notched, and sting appendage spatulate. These twelve groups together contain more than 110 species that are endemic to Madagascar, plus three species found only on Mayotte or the Comoros. The only other group with eleven-segmented antennae, notched anterior clypeal margin, and spatulate sting appendage is the T. weitzeckeri group. However, as noted above, T. humbloti represents a recent arrival from the Afrotropical region and did not evolve independently in Madagascar (Hita Garcia and Fisher 2011). Not considering the latter group, most native species groups could have originated from one or a few ancient colonisation events, with subsequent adaptive radiation that is most pronounced in the humid forests of eastern and northern Madagascar. Assessing the geographic origins of the first colonists with eleven-segmented antennae appears challenging at first glance. A number of species groups have elevensegmented antennae and a spatulate sting appendage in the Afrotropical, Oriental, and Indo-Australian regions. However, the species morphologically closest to the Malagasy groups are mostly found in Africa.
The most closely related ants seem to belong to the comparatively species-rich T. weitzeckeri species group (Bolton 1980;Hita Garcia et al. 2010). The group is very widespread and ecologically successful in sub-Saharan Africa, and has both elevensegmented antennae and a spatulate sting appendage. Many of its species resemble a number of members of Malagasy groups, such as the T. bessonii group, parts of the T. bonibony group, parts of the T. dysalum group, and parts of the T. marginatum group. The older species of these groups were initially even placed in the T. weitzeckeri group (Bolton 1979) until recent rearrangements of the Malagasy species group system (Hita Garcia and Fisher 2011, 2012a, 2012b, 2014b. In Hita Garcia and Fisher (2011) we proposed to treat the Malagasy members (with the exception of T. humbloti that stayed in the group) as independent developments from the T. weitzeckeri group and created the new T. bessonii, T. bonibony, T. dysalum, and T. marginatum groups. These groups contain a high degree of morphological diversity often differing from the Afrotropical T. weitzeckeri group. One reason for the distinctiveness of the Malagasy groups was the shape of the petiolar node. We stated that in the T. weitzeckeri group the node is often squamiform with anterior and posterior faces approximately parallel, whereas in Madagascar this form of node is only present in few species of the T. dysalum group. Most other species of the groups in question have petiolar nodes that are anteroposteriorly compressed, often very strongly so, but with a more triangular to cuneiform shape, the node narrowing towards the dorsum, or the anterodorsal margin much more angled than the posterodorsal margin causing the dorsum to strongly taper backward posteriorly. The dorsum of the node is greatly reduced in a number of species, especially from the T. bonibony and T. marginatum groups. Despite being absent in the T. weitzeckeri group, this more triangular node shape is also found in the Afrotropical T. squaminode group, which is also predominantly East and South African. This group also has a spatulate sting appendage but twelve-segmented antennae, and despite this difference seems to be closely related to the T. weitzeckeri group (Bolton 1980;Hita Garcia et al. 2010). Nevertheless, the argument about the difference in node shape is less valid now after the recent description of T. mpala Hita Garcia & Fischer (Hita Garcia and Fischer 2014). This interesting species from Kenya belongs to the T. weitzeckeri group but has a more triangular squamiform node shape like in the Malagasy groups mentioned above. However, instead of reuniting several species groups with more than 70 species, we prefer to wait until more data is available. Morphological similarities in as diverse a genus as Tetramorium can be misleading, and better taxonomic resolution on a supraspecific level can only be achieved with a large-scale analysis that combines morphology with informative molecular phylogenetic or phylogenomic data. Regardless, the Malagasy T. bessonii, T. bonibony, T. dysalum, and T. marginatum groups are very likely part of a larger Tetramorium lineage that also includes the Afrotropical T. weitzeckeri and T. squaminode groups, even though the relationships remain unclear at present.
The Malagasy T. naganum, T. schaufussi, and T. severini groups (and parts of the T. dysalum group) also appear to have a strong African influence since they share a spatulate sting appendage, high nodiform petiolar node shape, and a lack of any sculpture on the waist segments with the South African T. grassii group, even though the latter group has twelve-segmented antennae. Bolton (1980) stated that the T. grassii group is related and possibly ancestral to the T. weitzeckeri and T. squaminode groups. We concur that these three groups are very likely closely related. Based on their unique morphology, the T. plesiarum, T. ranarum, and T. tsingy groups seem to be independent Malagasy developments since there are no species groups with similar morphology in any region. However, they are very probably also part of the same larger lineage as the other species groups with eleven-segmented antennae and a spatulate sting appendage. This leads us to hypothesise that there is a larger Afrotropical and Malagasy Tetramorium clade/lineage that contains the following groups: T. bessonii, T. bonibony, T. dysalum, T. grassii, T. marginatum, T. naganum, T. plesiarum, T. ranarum, T. schaufussi, T. severini, T. squaminode, T. tsingy, and T. weitzeckeri species groups. The situation for the T. kelleri and T. tortuosum groups is less clear. These two groups are very close and were separated recently on the basis of their distinctiveness in Madagascar (Hita Garcia and Fisher 2012b). However, the T. tortuosum group is present in the Neotropical, Afrotropical, Malagasy, Oriental, and Indo-Australian regions, and displays great differences in morphological diversity and species richness from region to region (Bolton 1977(Bolton , 1979(Bolton , 1980Fisher 2012b, 2013). It is difficult to assess if all the species now listed as T. tortuosum group indeed belong to one very widespread, possibly very old, monophyletic clade, or some have evolved independently to share several key morphological characters. Consequently, we cannot assess with any certainty whether the Malagasy T. kelleri and T. tortuosum groups are more closely related to the African or Asian members of the T. tortuosum group, or represent a more independent lineage. In summary, we were able to identify a highly unique Malagasy Tetramorium fauna. We recognise 12 of the 19 species groups and an astonishing 113 of the 125 species as Malagasy endemics. This results in an endemism rate of 93%, which is more or less in agreement with the published value for the whole Malagasy ant fauna (ca. 96% in Fisher 2003). In Table 1 we provide an overview of the Malagasy species groups with data on their biogeography, key characters, taxonomic revisions, and preferred habitats.

Identification key to Tetramorium species groups in the Malagasy region (workers)
The species group key presented here is based on the one published in Hita Garcia and Fisher (2011). Although the key in that publication still works for most species in Madagascar, it does not accommodate them all. The recent revisions of most species groups, with the establishment of some new groups (Hita Garcia and Fisher 2012a, 2012b, 2014b, require an updated and improved key. The following key applies to the 19 groups we currently recognise, which contains around 125 species (the T. ranarum and T. simillimum groups will be revised in a future publication and the species count for these two groups is temporary).  (Fig. 2C), or with reduced pilosity but short appressed pubescence (Fig. 2D)  Lateral portion of clypeus prominent, raised to a tooth or denticle in full-face view (Fig. 4B); propodeal spines medium-sized and spinose, approximately of same length as metapleural lobes (Fig. 5C) ........... T. sericeiventre group -Lateral portion of clypeus never modified as above (Fig. 4C); propodeum either unarmed (Fig. 5B) or armed with small triangular teeth or denticles that are shorter than metapleural lobes (Fig. 6E, F)    6 Head in full-face view relatively thin (CI < 80) and antennal scapes very long (SI > 120) (Fig. 6A); in general appearance head, antennae, and legs elongate and slender (Fig. 6E)    Pronotum anterodorsally with distinct protuberance or bulge (Fig. 8A, B)  Pronotum anterodorsally without any protuberance or bulge (Fig. 8C, D) ...9 9 First gastral tergite with strongly appressed pubescence of varying length and without any standing hairs (Fig. 9A, B), or with short appressed to erect pilosity without any long, erect to suberect hairs (Fig. 9C)

Synopsis of the T. setigerum species group in Madagascar
Tetramorium cavernicola Hita Garcia & Fisher, sp. n.

Taxonomic and biogeographic notes on the group
Prior to this study, the T. setigerum species group appeared endemic to the Afrotropical region where it is widely distributed. Of the 13 species recognised by Bolton (1980), most are found in more arid areas of eastern and southern Africa, a few are distributed in the rainforests of Central Africa, while two species are also found in Ethiopia and the southwestern Arabian Peninsula. The recent finding of T. cavernicola in Madagascar was unexpected since there was no previous indication of the presence of the group on Madagascar or any of the surrounding islands of the South West Indian Ocean. However, as outlined above, considering the strong biogeographical affinities of the Tetramorium ant fauna of Madagascar with the Afrotropical region, this is less of a surprise. Indeed, the T. setigerum group has its highest abundance and diversity in South and Southeast Africa, which is geographically comparatively close to Madagascar. As outlined above, other species or species groups that made it from Africa to Madagascar are often of predominantly eastern and southern African origin; examples include T. humbloti from the T. weitzeckeri group and T. delagoense from the T. simillimum group. The T. setigerum group cannot be mistaken for any other Malagasy species group. Its possession of twelve-segmented antennae, an entire and convex clypeal margin, and simple pilosity distinguish it from most other groups, except the T. sericeiventre, T. simillimum, and T. tosii groups. In the T. sericeiventre group the clypeus is distinctly modified, with the lateral portion being very prominent and raised into a tooth/denticle in full-face view while the clypeus of the T. setigerum group lacks such a tooth/ denticle. Also, the species of the T. simillimum group possess much shorter antennal scapes (SI always much shorter than 100) than the T. setigerum group (SI over 120). The differentiation of the latter from the T. tosii group is more problematic. Despite the fact that the only representative of the T. setigerum group in Madagascar and the two species of the T. tosii group are easily separable (see key couplets 4 to 6), only a few morphological characters separate both groups if one also considers all members of the T. setigerum group from the Afrotropical region. Nevertheless, we prefer to keep both groups separate for the following reasons. First, the shape of the petiolar node is low, elongate, clublike, and always longer than high in the T. tosii group (Fig. 29A), whereas it is variably nodiform in the T. setigerum group, but usually higher than long and never low and elongate (Fig. 29B, C, D, E, F). Second, the standing pilosity in the T. tosii group consists of long, fine, acute hairs (Fig. 29A), whereas the pilosity in most members of the T. setigerum group is thick, short to moderately long, and usually blunt apically (Fig. 29B, E, F). Nevertheless, this is not the case in T. metactum Bolton and T. youngi Bolton since they have long and fine pilosity (Fig. 29C, D). This may seem contradictory, but leads to our next argument. Third, we strongly suspect that the T. setigerum group is not a monophyletic group, but might be composed of different lineages that share a number of morphological characters that have evolved convergently. For example, the morphology of T. cavernicola from Madagascar is certainly closer to the species complex around T. setigerum Mayr and allies (Bolton 1980) than to T. metactum or T. youngi. In sum, the relationships between the T. tosii and the T. setigerum groups, as well as within the latter group, remain unclear, and we prefer to keep the groups as they are until additional data can provide better resolution of the groupings.

Diagnosis.
Tetramorium cavernicola differs from all other Malagasy congeners by the following combination of characters: 12-segmented antennae; anterior clypeal margin entire and convex; lateral clypeus not modified into tooth or denticle; antennal scape very long (SI 120-123); mesosoma in profile relatively low and slender ; and propodeum armed with very short teeth/spines (PSLI 7-11).
Worker measurements ( Worker description. Head much longer than wide (CI 77-79); posterior head margin weakly to moderately concave. Anterior clypeal margin entire and convex. Frontal carinae strongly developed, moderately raised, usually becoming weaker after posterior eye level, approaching or ending at posterior head margin; antennal scrobes very weak to absent. Antennal scapes very long, weakly surpassing posterior head margin (SI 120-123). Eyes moderately large (OI 23-26). Mesosomal outline in profile relatively flat, elongate and low (LMI 35-36), weakly marginate from lateral to dorsal mesosoma; promesonotal suture and metanotal groove absent. Propodeum armed with short, triangular teeth (PSLI 7-11), propodeal lobes moderately developed, triangular to elongate-triangular, slightly longer and broader than propodeal teeth. Petiolar node nodiform with moderately rounded antero-and posterodorsal margins, in profile between 1.2 and 1.4 times higher than long (LPeI 73-79), anterior and posterior faces not parallel, node weakly narrowing towards dorsum, anterodorsal and posterodorsal margins situated at about same height and both weakly to moderately angled, petiolar dorsum flat to very weakly convex; node in dorsal view around 1.2 to 1.3 times wider than long (DPeI 121-127), in dorsal view pronotum around 2.0 to 2.1 times wider than petiolar node (PeNI 47-49). Postpetiole in profile approximately globular, around 1.0 to 1.1 times higher than long (LPpI 90-98); in dorsal view around 1.1 and 1.2 times wider than long (DPpI 113-123), pronotum around 1.7 to 1.8 times wider than postpetiole (PpNI 56-60). Postpetiole in profile appearing distinctly more voluminous than petiolar node, postpetiole in dorsal view around 1.2 to 1.3 times wider than petiolar node (PPI 120-127). Mandibles striate; clypeus longitudinally rugose/ rugulose with well-developed median ruga and usually one or two weaker, sometimes irregular, lateral rugae/rugulae on each side; cephalic dorsum between frontal carinae anteriorly towards posterior clypeal margin with three or four distinct but irregularly shaped longitudinal rugae with numerous cross-meshes, halfway between eye level and posterior head margin fluent transition to well-developed rugoreticulum ranging to posterior head margin; scrobal area only weakly sculptured, remainder of lateral head clearly reticulate-rugose. Mesosoma laterally and dorsally conspicuously reticulate-ru-gose; forecoxae unsculptured, smooth, and shining. Petiole and postpetiole irregularly rugulose, better developed on dorsum than sides. First gastral tergite unsculptured, smooth, and shiny. Ground sculpture on cephalic dorsum between frontal carinae weak, distinctly reticulate-punctate on lateral head, mesosoma, and waist segments, absent from gaster. All dorsal surfaces of body with short to moderately long, thick, and apically blunt pilosity; appressed pubescence on first gastral tergite strongly reduced to absent. Anterior edges of antennal scapes and dorsal (outer) surfaces of hind tibiae with decumbent to suberect hairs. Head and mesosoma reddish brown; waist segments lighter in colour, usually orange brown; mandibles, antennae, and legs yellowish brown.
Etymology. The name of the new species is a Latin noun and means "cave dweller" or "cave inhabitant". It refers to the microhabitat where the type series was collected. The species epithet is a nominative noun in apposition.
Distribution and biology. Currently, T. cavernicola is only known from Ankarana (Fig. 30D), where it was collected from a cave. The collection locality and the fact that the species is not known from outside the cave imply that T. cavernicola might be a specialised, cave-adapted ant. The generally very slender body and very long antennae and legs also support cave specialisation. Nevertheless, we do not consider the new species an obligate cave inhabitant. Arthropods that have evolved a cave-obligate lifestyle usually have a distinct set of morphological adaptations: reduction or loss of eyes, pigments, and wings; thinning of the cuticle; elongate antennae and legs; and slender body (Christiansen 1962;Culver 1982;Barr 1985). Yet the eyes, pigment, and wings in T. cavernicola are clearly not reduced since its eyes are always well developed, as are the wings in the queen and male castes, and the body colouration is brownish. In addition, we cannot detect any thinning of the cuticle. The slender gestalt and long antennae and legs could argue for cave adaptation, but are actually very typical of most species in the T. setigerum group. The antennae and legs of T. dolichosum Bolton and T. perlongum Santschi (Fig. 29E, F) are much longer than in T. cavernicola, even though these species do not live in caves. Tetramorium cavernicola appears to nest in the ground since most of the type series was collected from a ground nest, but no additional natural history data exists for this species.
Discussion. Tetramorium cavernicola is a very distinctive element of the Malagasy Tetramorium fauna and cannot be mistaken for any other congener based on the diagnosis provided above. There are some morphological similarities to the two species of the T. tosii group, as outlined earlier, but the distinction between these is easily found by comparing the shape of the head, the length of the antennal scapes, and the propodeal spines. In T. cavernicola the head is noticeably thinner (CI 77-79) and the antennal scapes are much longer (SI 120-123) while the propodeal spines are reduced to short teeth (PSLI 7-11). By contrast, the species in the T. tosii group have a thicker head (CI 85-91), much shorter scapes (SI 79-104), and much longer propodeal spines .
Variation. Since T. cavernicola is only known from the type locality, there is no observable intraspecific variation.