A revision of the purse-web spider genus Calommata Lucas, 1837 (Araneae, Atypidae) in the Afrotropical Region

Abstract The purse-web spider genus Calommata Lucas, 1837 is revised in the Afrotropical Region. Following examination of the female type material, Calommata transvaalica Hewitt, 1916 is removed from synonymy with Calommata simoni Pocock, 1903 and revalidated. The females of both species are redescribed and their males described for the first time. While Calommata simoni is very widespread across tropical Africa, Calommata transvaalica is endemic to northern South Africa. Four new species are described, all known only from males: Calommata megae sp. n. (Zimbabwe), Calommata meridionalis sp. n. (South Africa), Calommata namibica sp. n. (Namibia) and Calommata tibialis sp. n. (Ivory Coast and Togo). Notes are presented on the biology of each species.

In this paper the Calommata of the Afrotropical Region are revised, C. transvaalica is removed from synonymy with C. simoni, and four new species are described. While the biology of C. simoni and C. transvaalica has been studied (Van Dam and Roberts 1917, Blandin 1971, Charpentier 1995, little is known about the habits of the new species, which is largely due to the general scarcity of these spiders. We suspect that the biological studies related to C. simoni do, in fact, refer to C. tibialis sp. n., based on available material from one study and contrasting habitat preferences of the two species. By resolving the taxonomy of the Afrotropical fauna, the representative species can be included in future phylogenetic analyses and conservation assessments.

Material and methods
All spiders were studied under an Olympus SZX10 stereomicroscope under 10× magnification. Measurements (body and legs) were taken with a measuring eyepiece and are given in millimetres (mm). The proportional indices used follow Levy (2007) and are the carapace index (length divided by width), and patella-tibia index (combined length of the patella and tibia segments of the first leg divided by the length of the carapace). Total length is given as the length from the front of the chelicerae to the tip of the abdomen, excluding the spinnerets. A range of total length measurements is provided for each species from the material examined. Where geographical locality data was not provided on specimen labels or was not available in collection databases, geographical co-ordinates were searched for in the Global Gazetteer Version 2.2 (http://www.fallingrain.com) or using Google Earth Version 5.0 (http://www.google. com/earth), and are indicated in square parenthesis.
Material of male specimens of Calommata meridionalis sp. n., C. namibica sp. n., C. simoni and C. tibialis sp. n. was prepared for Scanning Electron Microscopy (SEM). Specimens were transferred to absolute (100 %) ethanol and left overnight. After drying in hexamethyldisilazane (HMDS), the specimens were glued to rounded aluminium rivets using two-sided copper strips and then coated with gold for examination using a JEOL JSM-6480 scanning electron microscope at the MRAC. Digitised micrographs were taken.
Digital photographs of the dorsal habitus of a C. simoni female were taken with a Leica MZ16 and stacked using the LAS automontage software at the MRAC. Digital photographs of the dorsal habitus of the remaining species and a C. simoni male were taken using a Nikon Coolpix 8400 mounted on a Nikon SMZ800 stereomicroscope at the UFS. The extended focal range images were stacked using CombineZM software to increase depth of field (www.hadleyweb.pwp.blueyonder.co.uk.). Genus Calommata Lucas, 1837 http://species-id.net/wiki/Calommata Type species. Pachyloscelis fulvipes Lucas, 1835 from Java and Sumatra Diagnosis. Calommata can be distinguished from Atypus and Sphodros by three main characteristics: 1) Spermathecal structure. In Atypus, there are two broad plates each bearing two or more small receptacula (Schwendinger 1990: figs 14-25), in Sphodros the four spermathecae are each highly coiled and without distinct receptacula (e.g. Gertsch and Platnick 1980: fig. 29), whereas in Calommata there are four spermathecae (Fig. 62), each bearing several closely packed terminal receptacula positioned in a cauliflower-like arrangement,). However, the latter definition was based only on the spermathecal structure of three of the seven Calommata species (Gertsch and Platnick 1980), and variation does indeed occur within the spermathecal structure of the genus. The female genitalia of C. transvaalica only bear one pair of oval spermathecae (Fig. 69). 2) Male palpal cymbium structure. In Calommata the palpal cymbium is short and truncate (Fig. 48), while in Atypus and Sphodros it is short and acuminate (Gertsch and Platnick 1980: figs 15, 25;Raven 1985). 3) Labiosternal suture. In Calommata the labiosternal suture is positioned anteriorly on the sternum (Levy 2007: fig. 8), but is considered by Gertsch and Platnick (1980) to be absent and by Raven (1985) to have migrated posteriorly in Atypus and Sphodros (see Gertsch and Platnick 1980: figs 14, 22). Further morphological characters unique to Calommata (Gertsch and Platnick 1980) include the bipartite, longitudinal thoracic groove (fovea) (Figs 1-9); basal ledge on the outer surface of the fangs of both sexes, the posteriorly positioned ocular tubercle, enormously elongated endites and dorsally expanded chelicerae (Figs 10, 11); short leg I, particularly the femur, and flattened palpal tibia and tarsus of females (Figs 4, 7); and the greatly elongate tarsi of legs III and IV of males, which are clearly pseudosegmented (e.g. Figs 3, 5, 30).

Key to the Afrotropical Calommata
[females of C. megae sp. n., C. meridionalis sp. n., C. namibica sp. n. and C. Chelicerae with a single row of teeth along promargin running from fang base close to cheliceral base (Fig. 15); epigyne with two pairs of spermathecae ( Chelicerae with one or two very large teeth near fang base, remaining teeth distinctly smaller; several small denticles retrolateral of teeth row near base of chelicerae (  Diagnosis. The male of this species is recognised by the conductor that narrows and makes a half twist before broadening distally, and the obliquely orientated embolus and conductor . This species shares with C. tibialis sp. n. the carapace that is subequal in length and width (longer than wide in the other four species).
Etymology. The specific epithet is a patronym in honour of the collector of the holotype, Meg Cumming, in recognition of her contributions to African arachnology, particularly in Zimbabwe.
Carapace and chelicerae brown in colour ( Fig. 1). Carapace flat and robust. Median ocular tubercle raised, darker in colour. Chelicerae with a single row of small teeth, increasing in size from fang base to base of chelicerae, with a few denticles near base of chelicerae (Fig. 12). Sternum and coxae light brown, remainder of legs dark brown, fading to light yellow-brown at tarsi. Legs weakly covered with bristles; prolateral side of patellae, tibiae and metatarsi of legs II-IV covered with spinules (thicker and shorter than bristles). Abdomen dark brown, with an irregular brown scutum present anteriorly (Fig. 1). Palpal cymbium short with rounded distal margin; embolus and conductor orientated obliquely, pointing retrolaterally towards base of chelicerae; conductor narrow, making a half twist before broadening distally, with single small curved tooth on its dorsal surface distally; embolus short and straight, slightly curved near tip (Figs 53-55).
Female. Unknown. Distribution. Known only from the type locality (Fig. 73).
Biology. Poorly known. The holotype male was collected in mid-summer on the soil surface. The second instar juvenile specimen was captured while ballooning and landing on the porch of a house (M. Cumming, pers. comm.). Amongst atypids, ballooning has previously been recorded in both Atypus and Sphodros (Wiehle 1953 cited in Pedersen and Loeschcke 2001, Coyle 1983, Coyle et al. 1985. Diagnosis. The male of this species can be easily recognised from African congeners by the presence of one or two very large teeth at the fang base (Fig. 13), and the transversely orientated curved embolus with distally broadened conductor bearing a single tooth on its dorsal surface (Figs 42,43,57). The raised median ocular tubercle is broader than in the other species.

Calommata meridionalis
Etymology. This specific epithet is Latin for southern, referring to the distribution of the species, southernmost in the genus.
Description. Male holotype. Carapace and chelicerae dark brown in colour (Fig. 2). Median ocular tubercle raised, darker in colour; median ocular tubercle broader than in other species. Chelicerae with single prolateral row of teeth, one very large tooth close to fang base, sometimes accompanied by second large tooth, remaining teeth distinctly smaller and subequal in size, with several denticles retrolateral of teeth row close to cheliceral base (Fig. 13). Sternum and coxae pale brown, remainder of legs brown, gradually fading to yellow at tarsi. Legs weakly covered with bristles; prolateral side of patellae, tibiae and metatarsi of legs II-IV covered with spinules. Abdomen grey-brown, with a round brown scutum anteriorly (Fig. 2). Palpal cymbium short with rounded distal margin; embolus and conductor orientated transversely across palpal axis, pointing retrolaterally; conductor broadened distally, with a single small tooth on its dorsal surface distally; embolus long, slightly curved in a S-form along its length (Figs 42,43,(56)(57)(58).
Female. Unknown. Distribution. Endemic to central and northern Free State Province, South Africa (Fig. 73).
Biology. The species was collected exclusively by pitfalls in spring and early summer (September to early December) in the Grassland Biome of South Africa. Specimens were only collected in dark vertic clay and loamy-clay soils and not from sites with sandy soils. Most of the specimens were collected from sites near to freshwater streams and dams. Despite exhaustive attempts to locate burrows in the vicinity of pitfall sites (Erfenis Dam Nature Reserve and Botanical Gardens) none could be found. Other material examined. None. Diagnosis. The male of this species can be recognised by the tiny cheliceral teeth in a single row, without retrolateral denticles (Fig. 14), and the obliquely orientated conductor and embolus projecting far beyond the retrolateral cymbial margin (Fig. 60).
Etymology. The specific epithet refers to the country of the type locality. Carapace and chelicerae dark brown (Fig. 3). Median ocular tubercle raised, narrow, darker in colour. Chelicerae with single prolateral row of tiny teeth, without denticles near cheliceral base (Fig. 14). Sternum and coxae light brown, femora dark brown; subsequent segments fading to light yellow at tarsi. Legs weakly covered with bristles; prolateral side of patellae, tibiae and metatarsi of legs II-IV covered with spinules. Abdomen dark grey, nearly black, with brown scutum present in the anterior half (Fig. 3). Palp with elongate cymbium, with tapering pointed distal margin; embolus and conductor orientated obliquely, pointing retrolaterally and distally, projecting far beyond retrolateral cymbial margin; conductor short, slightly broadened distally, with a very prominent, very long and slender tooth distally on its dorsal surface; embolus long, with slight bend in distal half (Figs 45,46,(59)(60)(61).
Robustly built with short legs, carapace faded to creamy brown (Fig. 4). Median ocular tubercle raised, narrow, sloping sharply at fovea (Fig. 10). Single median line running from anterior of eye area to approximately middle of chilum. Chelicerae pale orange brown, darker laterally; chelicerae with a single row of small and medium sized teeth along promargin running from fang base close to cheliceral base, with extensive denticle field retrolateral of teeth row near cheliceral base (Fig. 15). Endites strongly elongated prolaterally, strongly curved upwards (Fig. 10). Sternum and legs light yellowish brown. Legs short and stout, leg formula 4231; legs III and IV more robust than legs I and II; leg I without bristles or spinules; leg II with few spinules distally on tibiae, and dorsal and lateral spinules on metatarsi and tarsi; legs III and IV with spinules from patellae to tarsi (mainly dorsal and prolateral) and covered in bristles. Abdomen globose and pale grey, with indistinct median heart marking in anterior half (Fig. 4). Epigyne forming a broad, weakly sclerotised plate ventrally, in dorsal view with two pairs of spermathecae; median pair subrectangular, rounded anteriorly, lateral pair subtriangular (Fig. 62) Carapace and chelicerae brown (Fig. 5). Median ocular tubercle raised, narrow, darker in colour (Fig. 11). Chelicerae with single prolateral row of teeth, largest teeth in midsection of teeth row interspersed with smaller teeth anteriorly and posteriorly, with several denticles retrolateral of teeth row close to cheliceral base (Fig. 16). Endites elongated prolaterally, curving upwards (Fig. 11). Sternum and coxae light yellowish brown, rest of leg segments brown, fading to light yellow at tarsi. Legs weakly covered with bristles; prolateral side of patellae, tibiae and metatarsi of legs II-IV covered with spinules. Abdomen grey brown, with elongate brown scutum present in anterior half (Fig. 5). Palp with short cymbium, with rounded distal margin; embolus and conductor orientated obliquely, pointing retrolaterally and distally, not projecting beyond retrolateral cymbial margin; conductor short, broadened distally, with a prominent elongate tooth and sharp edge opposite the tooth, appearing as a second tooth; embolus short and straight (Figs 48, 49, 63-65).
Remarks. Benoit's (1967: 286, figs 1-4) drawings of a male "allotype" of C. simoni correspond with the males we have studied. However, Pocock (1903: 259) never described the male of C. simoni nor listed any males in his material studied, and thus the specimen examined by Benoit could not possibly be an allotype. The loan request to BMNH also only yielded the female lectotype of C. simoni, and no allotype or paratypes as indicated by Benoit. Benoit indeed wrongly considered the specimen used to describe the unknown sex for the first time to be the allotype, even when that occurred separately from the original description. Comments on the revalidation of C. transvaalica are provided under remarks for that species below. Charpentier (1995)  Distribution. Widespread across tropical Africa in forests and savanna woodlands (Fig. 73).
Biology. The biology of "C. simoni" was studied by Blandin (1971) and Charpentier (1995) in Côte d'Ivoire and Benin, respectively (localities listed above). However, examination of the specimens reported on by Blandin indicates that they are, in fact, C. tibialis sp. n.. In considering the habitats of the available material of C. simoni and C. tibialis sp. n., it is evident that the two species are ecologically separated, the former occurring in forests and the latter in woodland savannah. As the material collected by Charpentier (1995) could not be traced, it is impossible to determine whether he studied the biology of C. simoni or C. tibialis sp. n.. However, his indication of the habitat types at the four localities he sampled (grassland, patches of subsistence agriculture, near rivers and open ground near palm forests) suggests that the material he studied is C. tibialis sp. n. and not C. simoni. Thus, we have included biological information from their two studies under C. tibialis sp. n..
Most of the specimens studied here from the MRAC collected in Guinée, Côte d'Ivoire, Kenya, Tanzania and Congo D.R. were collected in contrasting forest types across tropical Africa, indicating that C. simoni is tolerant and adaptable to a wide range of soil, vegetation and climatic variables. Diagnosis. Th e male of the species can be recognised by the carapace that is subequal in length and width (Fig. 6), the short, swollen palpal tibia, and the narrow conductor ending in a thick prominent tooth (Figs 51, 52).
Etymology. Th e specifi c epithet refers to the palpal tibia of the male, which is distinctly shorter and more swollen compared to that of other African congeners.
Female. Unknown. Remark. The specimens may possibly have faded over time in 70% ethanol, which can only be confirmed should fresh material become available. Although a subadult female is available it will not be described as the genitalic structure cannot be studied.
Distribution. Central Côte d'Ivoire and northern Togo (Fig. 73). Biology. Present data indicates that C. tibialis sp. n. occurs in woodland savannah habitats and avoids forests, where C. simoni has been collected. This may indicate some degree of ecological separation between the species but requires further study. Taking this into account we consider the studies of Blandin (1971) and Charpentier (1995) to relate to C. tibialis sp. n. and not C. simoni, as indicated by them. Charpentier (1995) located more than 50 nests of C. tibialis sp. n. at four localities in southern Benin with quite contrasting habitat structures, including grassland, patches of subsistence agriculture, in close proximity to rivers, and open ground near palm forests. He did not indicate the occurrence of the species in forests. In one of the habitats that he found Calommata, the soil was described as sandy, of poor quality and relatively acidic, and covered in 'grassland' vegetation, similar to the habitat characteristics described by Blandin (1971) for the Lamto area, from where C. tibialis sp. n. specimens are available.
The burrow of C. tibialis sp. n. slants obliquely downwards into the soil, and was estimated to be 25-30cm deep by Blandin (1971), while Charpentier (1995) indicated a maximum depth of 21cm in a female specimen, although generally shallower in other specimens (12-19cm). The top 1-2cm of the burrow is expanded to form a chamber covered by silk webbing that is camouflaged with soil, and the spider lies in wait hanging upside-down from the web for wandering prey (Charpentier 1995). Egg sacs are suspected to hatch in May; during incubation the female spins a silk veil at the base of the chamber that is suspected to firstly allow the spider access to the chamber to capture potential prey, and secondly hide the spider and its eggs from potential parasites once they have entered the chamber (Charpentier 1995). Hewitt, 1916species-id.net/wiki/Calommata_transvaalica Figs 7, 8, 18, 19, 69-72 C. transvaalicus Hewitt, 1916 fig. 3, pl. 26, fig. 11 revalidated C. simoni Benoit, 1967: 283  Diagnosis. The female of this species has an additional row of two to four large prolateral teeth close to the fang base in addition to the main row of teeth, which are larger and more strongly curved than in C. simoni. The epigyne comprises a single pair of transversely oval spermathecae, while C. simoni possesses two pairs of smaller spermathecae. The male of the species shares with C. meridionalis the transversely orientated embolus and conductor, but the conductor of C. transvaalica is clearly narrower at the tip and the embolus is straight and not slightly curved as in C. meridionalis. The male chelicerae of C. transvaalica also lack the one or two large teeth found near the fang base in C. meridionalis.

Calommata transvaalica
Redescription. Female from Blouberg Nature Reserve. Measurements: CL 6.72,CW 5.80,AL 13.70,AW 10.50, Robustly built with short legs (Fig. 7), carapace pale creamy brown. Median ocular tubercle raised, narrow, sloping sharply at fovea. Single median line running from front of median ocular tubercle to middle of chilum. Chelicerae orange, darker laterally; chelicerae with a row of two to four large teeth close to fang base, prolateral of promarginal teeth row; teeth row comprising very large teeth curved at tips, interspersed with small teeth, with extensive denticle field retrolateral of teeth row near cheliceral base (Fig. 18). Endites strongly elongated and slender prolaterally, strongly curved upwards. Sternum and legs pale yellow-brown. Legs short and stout, leg formula 4231; legs III and IV more robust than legs I and II; leg I with three to five spines on patellae and two on tibiae; leg II with few spinules on patellae and several spinules on tibiae and metatarsi; legs III and IV with spinules from patellae to tarsi (mainly dorsal and prolateral); legs II to IV covered in bristles. Abdomen globose and pale grey, with indistinct median heart marking anteriorly (Fig. 7). Epigyne forming a broad, weakly sclerotised plate ventrally, in dorsal view with single pair of large, transversely oval spermathecae (Fig. 69). Female palp short, tibiae and tarsi flattened.
Carapace and chelicerae dark brown in colour (Fig. 8). Chelicerae with single row of large teeth, gradually decreasing in size from fang base to cheliceral base, without denticles near cheliceral base (Fig. 19). Carapace oval in shape. Median ocular tubercle raised, narrow, darker in colour. Sternum and coxae yellow-brown, femora, patellae and tibiae brown, metatarsi yellow-brown, tarsi yellow. Legs weakly covered with bristles; prolateral side of patellae, tibiae and metatarsi of legs II-IV covered with spinules. Abdomen dark grey, nearly black, with dark orange-brown scutum anteriorly (Fig. 8). Palp with short cymbium, cymbium tip in ventral view tapering to rounded point; embolus and conductor orientated transversely to palpal axis, pointing retrolaterally, distal ends projecting beyond retrolateral cymbial margin; conductor short, slightly broadened distally, with single sharp, curved tooth on its dorsal surface; embolus long and straight (Figs 70-72).
Remarks. It is clear from the redescriptions of both sexes of C. simoni, and the redescription of the female and first description of the male of C. transvaalica provided here, that the two species have distinct differences in their somatic and genitalic morphology, most notably regarding their cheliceral dentition, number of spermathecae in the female epigyne and orientation and length of the male embolus and conductor. Consequently, we reject Benoit's (1967) synonymy of the two species and propose the revalidation of C. transvaalica. The specimens collected by Van Dam and Roberts (1917) between Villeria and Derdepoort near Pretoria could not be traced.
The abdomen of the female holotype of C. transvaalica is damaged and therefore the specimen was not redescribed. The holotype is the smallest of the known females of this species (18.60mm long). The colour of all available female specimens has likely faded over time in 70 % ethanol.
Distribution. Limpopo and Gauteng Provinces, South Africa (Fig. 73). Biology. The biology of C. transvaalica was studied by Van Dam and Roberts (1917) at Roodeplaat near Pretoria, South Africa, in the days following heavy rainfall. They first detected a female (the holotype) by kicking up a tuft of grass that disclosed white webbing, which was followed downwards into the ground to locate the spider. They subsequently discovered additional nests on bare ground. They described the nests as slightly raised above the ground at the top, and then from the inner rim they were neatly rounded off, gradually sloping outwards and downwards to the level of the ground with the outer surface covered with earth that resembled the surroundings. The interior of the tube was lined with loose, highly adhesive silky webbing. They suggested that the adhesive webbing may afford the spider some protection against the intrusion of enemies. The nests were deep (22-25cm) and vertical for the greater part of their depth (Van Dam and Roberts 1917). Hewitt (1916) commented that C. transvaalica specimens had a very pronounced and objectionable odour and compared it to decomposing stable manure.
The recently collected series of males from Zwartkoppies Farm (TMSA 23875) was collected in open pitfalls without preservative from a site in open Acacia karroo woodland on red structured clay soils (30-45% clay in A horizon, Shortlands form). The site has a gentle slope and had recently been burned. The activity of the males appears to be related to heavy rainfall that had fallen two days prior to the collection of the males. It thus seems that males do not emerge on the night immediately following a heavy shower, but instead on the night thereafter (Ian Engelbrecht, pers. comm.).

Discussion
Calommata is a small but widespread genus that is known from Africa, Israel and South East Asia. Two species are present in West Africa, one species in Central Africa and four species in southern Africa. The type of habitat seems to play a role in separating the species biogeographically. Calommata simoni is perhaps the most flexible in terms of habitat requirements, and occurs in forests, savannahs and grasslands in tropical Africa. In contrast, C. tibialis was only found in wooded savannahs in West Africa. In southern Africa, C. megae, C. namibica and C. transvaalica occur in the Savannah biome, but with contrasting vegetation structures and climatic variables. Lastly, C. meridionalis occurs only in the Grassland biome of central South Africa (Fig. 73).
Atypid spiders are widely regarded as being of conservation importance due to their generally specific environmental requirements, low rates of dispersal and general scarcity (Pedersen and Loeschcke 2001, Řezáč et al. 2007, Řezáč 2009). Several species are considered critically threatened or endangered and are included in various country or regional Red Data lists (e.g. Platen et al. 1996, Komposch and Steinberger 1999, Blick and Scheidler 2003, Farkač et al. 2005, Sato et al. 2007. Calommata transvaalica (previously as C. simoni in South Africa) was until recently presumed nationally extinct in South Africa as the species had last been reported in the 1920's (Dippenaar-Schoeman 2002). It was subsequently rediscovered in 2003 (Groenkloof NR) and 2005 (Blouberg NR). This species was submitted for an initial Red List assessment in April 2008, but during the process of assessment, questions arose regarding the taxonomic status of this species and it was included in the Data Deficient category for taxonomic reasons (Engelbrecht 2008). As the genus has now been revised, the South African species can be resubmitted for Red Data listing to promote conservation of these spiders. In comparing the two South African species, populations of C. transvaalica are under severe threat in the south of its range due to rapid urbanisation and habitat loss in the Gauteng Province. The recent collection of 13 males from Zwartkoppies Farm near Pretoria suggests that healthy populations do still exist in natural habitats in the province. In contrast, C. meridionalis is experiencing considerably lower threat levels (the three known localities are from undisturbed grassland) and potential agricultural expansion, especially from cultivation agriculture, perhaps represents its greatest threat. If more attention can be paid to the apparent soil preferences of South African Calommata in future when conducting pitfall surveys it is likely that further populations could be located in suitable habitat. As such, South African Calommata may serve as ideal candidates of predictive modelling due to their restricted distributions, limited knowledge of their biology, and increasing threats to their survival (see Lobo 2006, Jiménez-Valverde et al. 2007).