Cyphoderus (Cyphoderidae) as a major component of collembolan cave fauna in Thailand, with description of two new species

Abstract Distinguishing features of Cyphoderus Collembola of the bidenticulati group are described. Taxonomic problems in the bidenticulati group of Cyphoderus are emphasized, and new characters of taxonomic value are introduced and discussed. Two new species are described from caves of Thailand, differing mainly in claw morphology.


Introduction
The species richness of Thai cave faunal communities are poorly known. Most studies in Thailand have focused on low-energy cave habitats, and large regions of the country have seldom been sampled. Consequently, the taxonomy, evolution, and biogeography of Thai cave Collembola are insufficiently known. Surveys of the Thai cave inverte-brates revealed that Collembola in the family Cyphoderidae were the dominant arthropods in non-oligotrophic habitats of the dark zone. All examined specimens belong to the bidenticulati-group of the genus Cyphoderus (sensu Delamare-Deboutteville 1948), which previously included 16 species worldwide. Cyphoderidae are typically myrmecophilous or termitophilous, with few records outside of ant and termite nests (Imms 1912, Folsom 1927, Delamare-Deboutteville 1948, Yoshii 1987. The abundance of Cyphoderus in caves in the absence or rarity of ants, and the striking morphological similarity of cave forms with myrmecophilous species raises several evolutionary and ecological questions. Börner (1906) created Cyphoderini as a tribe of Entomobryinae to include Cyphodeirus albinos Nicolet, 1842 and three other species that he described in the same paper. In 1913, he upgraded Cyphoderini to subfamily rank, which he placed in Entomobryidae, a concept followed by Delamare-Deboutteville (1948). Subsequently, the taxon was upgraded yet again and was considered a family by most authors (Absolon and Kseneman 1942, Szeptycki 1979, Yoshii 1980, 1987, 1992, Deharveng 2004, Fanciulli et al. 2006. Soto et al. (2008) considered the group to be a subfamily in the Paronellidae on the basis of their non-annulated dens. However, the dens of cyphoderids is clearly reduced in length compared to that of all other Paronellidae sensu stricto and always bears characteristic feathered scales (more accurately termed feathered chaetae) consisting of a strong rachis with two symmetrical vanes made of long parallel barbs, a unique structure unknown from other Collembola. On this basis alone, we believe that Cyphoderidae deserve family rather than subfamily status. Twelve genera have been described in Cyphoderidae (Bellinger et al. 2013): Calobatinus Silvestri, 1918 (4 species), Cephalophilus Delamare-Deboutteville, 1948 (3 species), Cyphoda Delamare-Deboutteville, 1948 (10 species), Cyphoderinus Denis, 1942 (1 species), Cyphoderodes Silvestri, 1910 (7 species), Cyphoderus Nicolet, 1842 (64 species), Delamareus Mitra, 1976 (2 species), Megacyphoderus Delamare-Deboutteville, 1948 (4 species), Mimoderus Yoshii, 1980 (5 species), Paracyphoderus Delamare- Deboutteville, 1948 (1 species), Pseudocyphoderus Imms, 1912 (4 species) and Serroderus Delamare-Deboutteville, 1948 (26 species). The genus Cyphoderus is the largest in the family and has a worldwide distribution. Like most cyphoderid species, most Cyphoderus species are termitophilous or myrmecophilous (Delamare-Deboutteville 1948, Christiansen 1957, Yoshii 1980, 1987, 1990, 1992. In his extensive revision of Cyphoderidae, Delamare-Deboutteville (1948) divided Cyphoderus into 5 groups according to the shape of the mucro (tridenticulati, bidenticulati, inermes, quadridenticulati and multidentati), to accommodate the 42 species known at that time.
Cyphoderus "bidenticulati-group" created by Delamare-Deboutteville (1948) and studied in this paper are easily recognized by their long, thin, and yellow mucro ending in two subequal small teeth. This group includes a large number of forms described as species, several only known from a single location, and a few species given as widespread on the account of numerous literature records. However, most of these records are doubtful because most species in this complex lack conspicuous morphological features, and are therefore difficult to distinguish.
Not only the so-called species are difficult to separate, but the description of the taxon's widespread type species, Cyphoderus albinus Nicolet, 1842, is poor by modern standards. In fact, the original description of Nicolet (1842) is so vague that it could apply to almost any species in the bidenticulati group. The most reliable, recent information comes from three sources: Delamare-Deboutteville (1948), whose description is probably based on French material; Yoshii (1990) based on material of Macaronesia; and Fjellberg (2007) describing material from Scandinavia. However, these contradictory accounts add further confusion, as there are disagreements about major diagnostic characters. According to Delamare-Deboutteville (1948), the species has no unpaired inner tooth on claw; the other two descriptions mention one unpaired tooth, but not at the same level. Fjellberg (2007) stated that there is no sublobal hair on outer maxillary lobe; Yoshii figured one. These contradictions may represent variability among populations, different species placed under the same name, or inaccurate observations. The only certainty is that the bidenticulati group of Cyphoderus is a complex of extremely similar forms after Delamare-Deboutteville (1948), where morphological examination reaches its limit for delimiting species. In this paper we describe new morphological characters, beyond those introduced by Yoshii, and provide detailed descriptions that could serve as references for future taxonomic works. The redescription of type material or topotypes will be necessary to extend the present work. In parallel, the use of molecular taxonomy might be the easiest way to assess the status of populations.

Materials and methods
Collembola were extracted from cave substrate samples using Berlese funnels and pitfall traps and stored in 90% ethanol at 5°C. Caves were sampled throughout Thailand (Fig. 1). The two described species come from two caves that yielded abundant populations, one from eastern Thailand and the other from the peninsula. Specimens were cleared in lactic acid and mounted on slides in Marc Andre II gum. The morphological analyses used a Leica DMLB light microscope. Images taken on a Cambridge 600 scanning electron microscope (SEM) were used for intepretating fine morphology of some chaetae. Figures were improved with Photoshop CS5 (Adobe Inc.).

Material deposition PSU
Prince of Songkla University, Hat Yai, Songkhla, Thailand MNHN Muséum national d'Histoire naturelle, Paris, France Abbreviations used in the descriptions: Abd.= abdominal segment; Th.= thoracic segment; Ant.= antennal segment; AIIIO = Ant.III organ; M (in figures) or mac (in text) = macrochaeta(e); mes = mesochaeta(e); mic = microchaeta(e); sens = S-chaeta; T (in figures) = trichobothria; Man = manubrium (in tables). Chaetae notation: frontal chaetae of head and ventral tube chaetae after Yoshii (1980), tergite chaetotaxy after Szeptycki (1979), labial palp after Fjellberg (1999), AIIIO and ventral cephalic chaetae after Chen and Christiansen (1993 Character assessment. Several characters of taxonomic importance were discovered or re-appraised in the course of this study. 1) All antennal segments were examined on both dorsal and ventral sides, revealing 10 types of chaetae (Fig. 3A). Their distribution pattern on the antennae is complex, but similar in the two species. Similarities are also obvious with the few Entomobryoidea where antennal chaetotaxy has been described. For instance, sens 1 to 5 and 8 of AIIIO as figured in Sinella by Chen and Christiansen (1993) were easily retrieved in our Cyphoderus (Fig. 3G). Several of the chaetal types recognized here are also found in other genera of Entomobryoidea. However, patterns are very complex and their comparisons would require detailed analyses beyond the scope of this paper.
3) Pseudopores on tergites are arranged as in the Entomobryoidea species where they have been recorded (Jordana 2012 for instance): 1,1/1,1,1,1,0,0 from Th.II to Abd.VI. The presence of dorso-distal pseudopores on manubrium (2+2 in the studied Cyphoderus, Fig. 7H, 8D) is also characteristic of Entomobryoidea. Special to Cyphoderus described here are the 2+2 pseudopores behind the posterior row of chaetae of Abd.IV, found also in other unidentified Cyphoderus of the bidenticulati group (Fig.  4B). This pseudopore location is only known in Troglopedetinae, i.e., in Troglopedetes (Deharveng 1988), in Cyphoderopsis (Jantarit et al. 2013) and in Trogolaphysa (Soto-Adames and Taylor 2013), with a number of pseudopores different for each genus. A ventral pseudopore is present on antennal area, in the same location as in Isotomidae (Deharveng 1979), Neanuridae (Deharveng 1983) and Onychiuridae (Pomorski 1998, Martinez et al. 2004. At least, the presence of 1+1 or 2+2 pseudopores on head anteriorly to the antenno-basal line (Fig. 2H) is a new pseudopore location for Collembola, unnoticed as far as we know in other genera of the class. 4) Important features of dorsal head chaetotaxy have been discovered by Yoshii (1980Yoshii ( , 1987Yoshii ( , 1992, useful for characterizing the family Cyphoderidae and several taxa of lower rank. The number and arrangement of post-labial chaetae as well as the presence of one mic among them are the same in the two species described here. However, they differ when compared with other species and might provide another promising set of taxonomic characters. 5) Body chaetae of various types were detected and tentatively grouped in categories. The mes of type-5 are the most numerous chaetae dorsally. They are seen as smooth under microscope examination, but serrated under SEM, Fig. 4A5; distinguishing them from S4 sens is especially difficult on Abd.IV where both are present, and the same confusion may arise for many other Entomobryoidea. As patterns of these mes as well as those of S4 sens seem to be stable inside population and different between species, further investigations will have to re-examine this character for its use in taxonomy.
6) The chaetotaxy of dorsal side (Figs 4-6) matches in most cases that given by Szeptycki (1979) for Cyphoderus albinus, and is very similar to that of Entomobryoidea (see Zhang et al. 2009). Main differences include the relative position of S1 and S2 on Th.II (see above), and chaeta "as" of Th.III as a mes in our material versus a short S-chaetae in Szeptycki (1979). 7) One of the most important characters for differentiating species of the bidenticulati group is claw morphology, and it is the most diagnostic feature of the species described here. Although some variability in size and position of the various dental teeth has been noticed by other authors, it has not been taken into account in previous descriptions, leaving doubts about the validity of several species. Type material. Holotype, male adult (#PSUZC2011.SONG-001H) and 44 paratypes (6 males, 3 females, 35 of unknown gender) mounted on slides. Holotype and 29 paratypes at PSU (25 slides, 4 males, 2 females and 23 subadults, collection #PSUZC2011.SONG-001P-030P) and 15 paratypes at MNHN (12 slides, 2 males, 1 female and 12 subadults).
Description. Habitus thick ( Fig. 2A), not troglomorphic, body length about 1.2 mm excluding antennae and furca. Furca well developed, about 2.5 times shorter than body. Body color white. Eyes absent, no ocular patch. Dense cover of scales on head, body and furca (ventrally on manubrium, both sides on dens); scales present on Ant.I-II dorsally, absent on legs and ventral tube. Four categories of chaetae: ordinary chaetae (mac, mes and mic), scales, trichobothria and S-chaetae (= sens), described below separately for antennae and body.
Measurement. in µm (from type specimen #PSUZC2011.SONG-001H, male). Etymology. From the name of the province "Songkhla" where this species was discovered.
Distribution. Only known from the type locality. Ecology. Collected on guano in the dark zone of a cave developed in a karst covered with rainforest.
Discussion. The new species is similar to Cyphoderus javanus Börner, 1906 andto C. sumatranus Yoshii, 1987. The only detailed description of C. javanus is that of C. borneensis by Yoshii (1980Yoshii ( , 1987, which was synonymized with C. javanus by the same author in 1992. C. songkhlaensis sp. n. differs from C. borneensis as described by Yoshii in the following combination of characters: the posterior face of its ventral tube with chaetae L1 and L2 ciliated but M smooth (given however as ciliated in Yoshii 1980) (versus L1, L2, M all ciliated chaetae), anterior mac of ventral tube serrated versus smooth, spatulate versus blunt tenent hairs, no versus a few smooth basal chaetae on manubrium and claw with two inner teeth versus one inner tooth on claw.
Description. Habitus thick, not troglomorphic, body length about 1.3 mm excluding antennae and furca. Furca well developed, about 2.4 times shorter than body. Body color white. Eyes absent, no ocular patch. Dense cover of scales on head, body and furca (ventrally on manubrium, both sides on dens); scales present on Ant.I-II dorsally, absent on legs and ventral tube. Types of chaetae as in C. songkhlaensis sp. n.
Mouthparts. Outer maxillary lobe with one basal chaeta, a simple palp and one sublobal hair. Maxilla with 3-toothed capitulum and a complex of 5-6 pad-shaped lamellae not analyzed in detail. Mandible head stocky, asymmetrical with 5 (left) and 4 (right) teeth. Labial palp with 5 papillae (A-E) and 13 guards, as in C. songkhlaensis sp. n.; hypostomal chaetae (H, h1, h2) present; 4 proximal chaetae. Labial basis formula m,e,l1,l2, with all chaetae smooth or indistinctly serrated, and l2 reduced to a minute but thick mic. Labral formula 4/5, 5, 4 with all chaetae smooth; two chaetae of the mid-row stronger and longer than others (similar to C. songkhlaensis sp. n.); dorso-distal limit of primary granules with a deep central incision; labral edge without structure; ventro-distally, two asymmetrical combs with many teeth variously developed and two central tubules.
Ventral chaetotaxy of head. 4+4 post-labial mes smooth or very finely serrated along linea ventralis, and one mic between G3 and H3 probably homologous with X ( Fig. 2 in Chen and Christiansen 1993). Legs. Without scales. Trochanteral organ with 18 to 30 simple, straight, smooth chaetae arranged in V-form (Fig. 8A). Tibiotarsus chaetotaxy mostly composed of strong mes, with one thick smooth ventro-subapical chaeta on hind tibiotarsus. Each tibiotarsus with one tenent hair rather stout, apically spatulated, 3/4 to 4/5 as long as inner edge of claw; distal row of 9-10 serrated chaetae irregularly arranged on all tibiotarsi (Fig. 8B). Claw broad, not slender, without tunica; with 2 small teeth at 12% and 25% of inner edge from the tip of the claw, a small dorsal tooth basally and a pair of inner basal teeth of unequal size, the outer one much larger than the inner one (Fig.  8B). Unguiculus pointed and broad, more than a half as long as claw, lanceolate, with a strong outer tooth (Fig. 8B).
or subequal to mucrodens (versus manubrium longer than mucrodens). The number of teeth on the claw has been confirmed on 5 specimens of C. khaochakanus sp. n. and 8 specimens of C. songkhlaensis sp. n. Characters 2 and 3 are more difficult to observe, and their variability need to be assessed more firmly. In any case, these very slight differences are those usually reported in the literature between the species of the albinus group of Cyphoderus. Whether they indicate species-status would require re-examination of many species of the genus, especially for testing the variability of inner teeth on claw. We surmise that there are too few consistently different morphological characters in this group to further describe new species based only on morphology. We believe that molecular data will be helpful in providing additional information relevant to alpha taxonomy.