The dorsal chaetotaxy of Trogolaphysa (Collembola, Paronellidae), with descriptions of two new species from caves in Belize

Abstract Species diagnosis in Trogolaphysa has been based, until now, almost exclusively on number of eyes and shape of claws and mucro. Chaetotaxy, a character system important to diagnose species in other genera of scaled Entomobryoidea, has been described only for a few Trogolaphysa species. Here the complete dorsal chaetotaxy of six species of Trogolaphysa is described using the AMS and Szeptycki’s systems for head and body, respectively. A morphology-based parsimony analysis was performed to evaluate whether chaetotaxic characters overcome the influence of putatively cave adaptive convergent characters to resolve species level relationships, and to evaluate the evolution of the dorsal macrochaetae of the head. Phylogenetic analysis using only putative cave-adaptive characters support clades of unrelated taxa, but the addition of chaetotaxy overcomes the influence of convergent characters. A phylogeny based on all characters supports a trend towards reduced head macrochaetae number. Head macrochaetae are lost beginning with A3 and followed, in order, by S5, S3 and M3. In addition, a checklist of New World Trogolaphysa is provided and two new species, Trogolaphysa giordanoae sp. n. and Trogolaphysa jacobyi sp. n., are described on the basis of material collected in six caves in southern Belize.


introduction
The collembolan fauna of Belize is among the least known of any Central American country. The Catalogue of Neotropical Collembola (Mari Mutt and Bellinger 1990) and subsequent updates (Mari Bellinger 1996, Mari Mutt et al. 2009) list one species, the troglomorphic Trogolaphysa belizeana Palacios-Vargas & Thibaud, 1997, for Belize. A recent biospeleological expedition to the Toledo District of Belize yielded several new springtail species, including two new species in the genus Trogolaphysa.
What is understood about the evolution of morphological adaptations to cave habitats in entomobryoid springtails is derived from northern temperate members of the genera Pseudosinella and Sinella (Christiansen 1961, Gama 1984. The evolution of troglobiont species in tropical Trogolaphysa (Palacios Vargas et al. 1985[1986) and Troglopedetes (Deharveng 1987, Deharveng andGers 1993), could provide important independent information to test hypotheses about the direction of character evolution in Entomobryoidea. The characters utilized in the descriptions of most of the 33 species of Trogolaphysa reported from the New World (Mari Bellinger 1990, Mari Mutt et al. 2009; Table 1) are limited to claw complex and mucronal shape (e.g., Palacios-Vargas et al. 1985[1986), two character systems prone to convergent evolution in cave habitats (Christiansen 1961, Christiansen andCulver 1987). Chaetotaxy is known for few species (Gruia 1987, Mari Mutt 1987[1988, Najt 1988[1989]), and is limited to the number of macrochaetae. While convergence itself is of interest in understanding evolution in caves (Derkarabetian et al. 2010, Hedin andThomas 2010), distinguishing convergent characters adaptive for subterranean life from characters that better reflect phylogenetic history has proven to be important in a variety of groups of animals (e.g., Wiens et al. 2003).
The relationships between the genera Paronella, Troglopedetes, Trogolaphysa, and Dicranocentruga has been a source of confusion. Najt (1988[1989]) evaluated morphological characters of these genera and provided clear diagnoses for all of them: Paronella was retained for species with 1+1 rows of external spines on the manubrium; Troglopedetes was restricted to species with a single subdivision of the fourth antennal segment; Trogolaphysa was circumscribed to include Paronella-like species with Ant. 4 not subdivided, manubrium without spines and a short mucro (in relation to dens) with 3-5 teeth; whereas Dicranocentruga was placed as a junior synonym of Trogolaphysa. Najt (1988 [1989]) did not consider the presence of EOS (extra ocular structure) as a diagnostic character. Mitra (1992Mitra ( , 1993Mitra ( , 2002 argued that species without manubrial spines but sharing the presence of an EOS and 8+8 (or apparently 6+6) eyes should be placed in the genus Dicranocentruga, whereas species with fewer than 6+6 eyes and without EOS should be allocated to Trogolaphysa or Troglopedetes. Mitra (1993) suggested that further observations of the chaetotaxy would furnish characters to support this separation, but until now, the complete dorsal chaetotaxy of these taxa remained undescribed.
Here we present complete descriptions of the dorsal chaetotaxy of the head and trunk for the two new species of Trogolaphysa and for T. belizeana, and compare their chaetotaxy to that of T. jataca (Wray, 1953), T. geminata (Mari Mutt, 1987[1988) and T. riopedrensis (Mari Mutt, 1987[1988), three surface species from Puerto Rico. Finally, we present a morphology-based phylogenetic analysis to assess the value of chaetotaxy in elucidating species relationships in this genus, and to evaluate the evolution of some elements of the dorsal chaetotaxy of the head.  Najt (1988[1989]) of the New World, with distribution by country (given as ISO 3166-1 alpha-3 code).

Materials and methods
Springtails were collected with aspirators and preserved in 70% ethanol. Samples were associated with substrate characterizations and field-collected measurements of temperature, light intensity and humidity. Selected specimens were cleared in Nesbitt's solution, mounted in Mark André II (Mari Mutt 1979) on glass slides, and examined under a compound microscope with phase contrast. The extra-ocular structure (EOS) was examined under polarized light. Drawings were made using a drawing tube, with final illustrations completed using Adobe Illustrator CS5, version 15.0. To protect vulnerable sites, latitude and longitude are not provided for the Belize material. These locations are controlled by, and may be requested from, the Institute of Archaeology, Belmopan, Belize (see Acknowledgements). Holotypes and paratypes of the new species are deposited in the Illinois Natural History Survey Insect Collection (INHS).
Here we describe only elements of the chaetotaxy that are modified into microchaetae, macrochaetae or sensilla (i.e., idiochaetotaxy, Szeptycki 1979). We follow the nomenclature of Szeptycki (1979) for the body and the AMS system (Jordana and Baquero 2005, Soto-Adames 2008, 2010 for the head. Mitra (1993) proposed a system for the chaetotaxy of the head in Paronellidae, but it has not been widely embraced, whereas the AMS system has been applied to entomobryoids since the 1970's (Szeptycki 1973, Mari Mutt 1979 and allows evaluation of homologies between families of Entomobryoidea. The idiochaetotaxy of Trogolaphysa is reduced, and in naming body setae we assume it represents the remnant of primary chaetotaxy. The setae closest to the mesothoraxic pseudopore (Figs 12,32,53) are identified as m1 and m2, even though they occupy positions that in entomobryoids with more abundant idiochaetotaxy might be assigned to setae m2i and m2e, respectively. The nomenclature of setae on the fourth abdominal segment follows Szeptycki's system (Soto-Adames 2010): setae in columns A and B are named sequentially from posterior to anterior, irrespective of their relative insertion. Columns A and B have a maximum of four setae, and when all are present they are always setae 3-6 (e.g., A3, A4, A5 and A6). In the species described below, setae A3, A6, B3 and B6 are always present, and it is assumed that setae A4 and B4 are always suppressed before A5 and B5.
For the labial chaetotaxy, upper case letters represent macro-or mesosetae and lower case represent microsetae, an underscore in the formula identifies ciliate setae.
The eye patch of a generalized springtail comprises a group of 5 anterior and 3 posterior simple eyes, we refer to the space between these two groups of eyes as the 'eye patch well' to distinguish it from the inter-ocular space, which is the gap between the eye patches on either side of the head.
The formula of the dorsal macrochaetae of head and trunk is based on Gisin's (1967) model, but we consider all macrochaetae associated with the bothriotricha on abdominal segments 2-4, instead of only those found between the bothriotrichal complexes. The number of macrochaetae on the head is presented as two digits; the first digit refers to macrochaetae anterior to the head sulcus (series A, M and S), the second to the posterior macrochaetae (series Ps, Pa and Pm). The macrochaetae on abdominal segment 4 are represented by three digits separated by plus (+) symbol, where the first, second and third numbers refer to the inner (series A and B), medial (assumed, in Szeptycki's system, to be series C) and outer macrochaetae (series T, D, E, F and Fe). The last number in the macrochaeta formula may be represented by a range because the number of outer macrochaetae may be variable, as some macrochaetae external to series F appear to be added as individuals grow older. The formula is based on the relative size of the sockets and includes all macrochaetae, irrespective of whether they are large (i.e., short, thick and blunt) or small (long, slender and acuminate).
The habitat parameters substrate temperature, air temperature, light, and relative humidity were measured with hand held meters. Differences in abiotic parameters between habitats occupied by the two new species were tested using a Wilcoxon rank sum test in R 2.15.2 (R Developmental Core Team 2012), with continuity correction. Najt (1988[1989]) http://species-id.net/wiki/Trogolaphysa Diagnosis. Paronellidae with finely denticulate scales covering dorsum of head and body, and ventral face of furcula; Ant. 4 sometimes annulated, never subdivided in two; labial seta L2 normal, not reduced; eyes 0-8; EOS present; Abd. 2-4 with 2, 3, 3 bothriothricha; manubrium without spines, dens with 1-2 rows of spines; mucro short, with 3-5 more or less evenly spaced teeth.

Genus Trogolaphysa Mills, 1938 sensu
Remarks. As currently circumscribed Najt 1988 [1989]), the absence of a subdivision on Ant. 4 in Trogolpahysa is the only character that distinguishes this genus from Troglopedetes.
It is not known if the type species of the genus, T. maya Mills, 1938, has EOS, but the presence of this structure in all species discussed below, including the two troglomorphic forms, suggests it is likely also present in that species. Diagnosis. Trogolaphysa giordanoae sp. n. is unique among species with 6-8 eyes in having 5 dorsal head macrochaetae, 3 metathoracic macrochaetae and 4 inner macrochaetae on Abd. 4. Among species with known dorsal chaetotaxy, the new species is most similar to T. riopedrensis, but the two species are easily distinguished by the combination of characters given above and by the presence of a relatively shorter mucro in the new species (Table 2). Additional diagnostic characters distinguishing the new species from all other New World Trogolaphysa with 6-8 eyes and capitate/spatulate tenent hair are presented in Table 2. Among the species described before the introduc-  tion of chaetotaxy, the new species is most similar to T. distinguenda (Denis, 1931), but the two species can be separated by the presence of a relatively long mucro with 5 teeth in distinguenda, and a 4-toothed short mucro in T. giordanoae sp. n. Trogolaphysa belizeana is the only other New World Trogolaphysa with 3 metathoracic macrochaetae. However, T. belizeana is a troglobiont (sensu Sket 2008, Culver and Pipan 2009)-blind, with long antennae and modified ungues. Description. Size. Body length up to 2.1 mm.
Color. Pattern, if any, obscured by green dye present in the alcohol in which specimens were preserved (Fig. 1).
Scale distribution. Scales dark brown, present on Ant. 1-2 and base of Ant. 3, more abundant on dorsal face than on ventral face of segment. Scales absent from ventral tube, legs and dorsal face of manubrium.
Legs. Trochanteral organ with up to 36 setae. Metathoracic claw complex as in Fig. 18. Tenent hair weakly spatulate. Smooth posterior setae on metathoracic legs 0.76× as long as unguiculus. Unguis with 4 inner teeth: 1 basal tooth sometimes ap- Figure 22. Trogolaphysa giordanoae sp. n. paratype habitat Okebal Ha entrance/twilight zone. Specimens were collected from a small pile of fruit bat guano near the researchers in the foreground, below a bat roost site. Sample site was much darker than it appears in this enhanced image. Photo courtesy of MES. pearing slightly larger than other, both paired teeth ending near middle of inner edge; proximal unpaired tooth as large as basal paired teeth, ending on distal half of inner edge; distal unpaired tooth smallest of all inner teeth and ending on distal fourth of inner edge. Outer tooth ending on basal quarter of outer ungual edge. Unguiculus lanceolate, with outer margin serrate.
Etymology. This species is dedicated to Rosanna Giordano, the senior author's wife, for her years of support and contributions to science.
Distribution. The species is known only from Belize Habitat. Trogolaphysa giordanoae sp. n. is a guanophile, recorded from entrance, twilight (Fig. 22) and dark zones of caves (6.7, 53.3 & 40.0 % of 15 collections, respectively), often in association with fruit bat or other guano (Fig. 23) (noted for 40% of 15 collections). It was commonly found on the floor of caves (76.9% of 13 collections where position was noted), but also on cave walls (23.1% of 13 collections where position was noted). Diagnosis. Trogolaphysa jacobyi sp. n. is the only member of the genus that is blind, has 3-toothed mucro and unguis, and has a single macrochaeta on the metathorax. Trogolaphysa belizeana is the only other New World Trogolaphysa lacking eyes and having a 3-toothed mucro, but it differs from T. jacobyi sp. n. in having 3 metathoracic macrochaetae (1 in T. jacobyi sp. n.), in the arrangement and identity of inner macrochaetae on Abd. 4 (cf. Figs 38, 49 see discussion below), in having few postlabial scales (absent in T. jacobyi sp. n.) and setae (many in T. jacobyi sp. n., cf. Figs 30, 46), in the presence of sensillum d2 on Abd. 3 (absent in T. jacobyi sp. n.), in the absence of unpaired ungual teeth (1 tooth in T. jacobyi sp. n.) and in having a typical lanceolate unguiculus (basally swollen in T. jacobyi sp. n.). Table 3 provides a list of characters that distinguish T. jacobyi sp. n. from all other New World Trogolaphysa lacking eyes and having paired basal ungual teeth inserted near the basal fourth of the inner edge.
Description. Size. Body length up to 2.0 mm. Color. Living specimens yellowish, with pigment only on a small eyepatch and mesothorax (Fig. 24). Specimens in alcohol white, without trace of pigment. Scale distribution. Scales transparent, present on Ant. 1-2. Scales absent from postlabial region of head, ventral tube, legs and dorsal face of manubrium.
Legs. Trochanteral organ with up to 25 setae. Claw complex as in Figs 41-42. Tenent hair acuminate, longer on L1 than L3. Smooth posterior setae on metathoraxic legs as long as unguiculus. Unguis with 3 inner teeth: basal teeth small, subequal and ending on basal fourth of inner edge; unpaired tooth distinctly larger than basal teeth, ending near middle of inner ungual edge. Outer tooth absent on all claws; lateral teeth present only on pro-and mesothoracic legs, and ending on basal quarter of outer edge of unguis (Fig. 41). Unguiculus basally swollen, with basal fifth of outer margin weakly serrate.

Distribution. The species is known only from caves in southern Belize
Remarks. Trogolaphysa jacobyi sp. n. is a troglobiont (sensu Sket 2008, Culver and Pipan 2009). Living specimens seem to have eye pigment (Fig. 24), but we were unable to identify corneas on specimens examined. The only structure resembling a cornea corresponds to the EOS.
It could be argued that the constriction of the fourth antennomere places this species in Troglopedetes. However, the presence of a well-developed ciliate labial seta L2, the incomplete nature of the constriction on Ant. 4, and the similarity with T. belizeana (presumably with complete, unconstricted Ant. 4, and therefore an uncontested Trogolaphysa) suggest that T. jacobyi sp. n. should be retained in Trogolaphysa. Etymology. This species is dedicated to JoAnn Jacoby, the junior author's wife, in gratitude for her enthusiasm and assistance in the planning and execution of field-work in the caves of Belize and in many earlier excursions. Habitat. This species is a troglobiont, and all 5 collections (11 individuals) were taken in the dark zone (0 lux) on the floor (Fig. 44), often (80% of collections) in wet conditions associated with flowstone or calcite and drip pools, sometimes with scattered cricket droppings. Additions to the original description. Head. Dorsal chaetotaxy of the head identical to that of T. jacobyi sp. n., with macrochaetae A0, A2, A3, M3, S3, S5, Pa5 and Pm3. Labral margin smooth. Sublobular plate of outer maxillary lobe without setae-like appendages. Labial papilla E with lateral appendage reaching tip of papilla; 5 proximal smooth labial setae present, seta z (Soto-Adames 2010) longest. Labial triangle formula as M1M2rEL1L2A1-5 (Fig. 45): M1 ciliate, shorter but thicker than M2; r short, stout, apically acuminate; A2 close to r. Postlabium with few scales; columns ICEL with 7732 setae (Fig. 46); seta L2 shortest; ventral cervical setae 6+6.
Legs. Claw complex of pro-and metathoracic legs as in Figs 50-51. Tenent hair acuminate. Outer and lateral unguis teeth small, inconspicuous; inner paired teeth with one tooth slightly, but clearly larger, unpaired teeth absent. Unguiculus lanceolate.
Ventral tube. With 2+2 distal macrochaetae on anterior face. Remarks. The paratypes examined differ from the original description of the species in having labial seta L2 smooth instead of ciliate, in having only 2 posterior mesothoraxic macrochaetae, in the claws having lateral teeth and in the number of bothriotricha on Abd. 2 and Abd. 4.

Remarks:
The individual from St. Thomas lacks head macrochaetae A2. In the individual from Aguadilla the dorsal and outer teeth of the unguis end on the basal fourth of the claw instead of the distal half.

Dorsal chaetotaxy
The dorsal chaetotaxy of Trogolaphysa has not been fully described in the context of the AMS (Soto-Adames 2010) and Szeptycki (1979) systems of nomenclature. The notes presented below are based on the study of T. jataca, T. geminata and T. riopedrensis, three surface species from Puerto Rico, in addition to the three species of cave Trogolaphysa from Belize.
Head. The dorsal chaetotaxy of the head is reduced when compared to other genera of scaled Entomobryidae (e.g., Seira, Pseudosinella; cf. Fig. 26 here to fig. 1 in Soto-Adames [2008] and fig. 4 in Soto-Adames [2010]). In the species studied, series A includes setae A0-3. Some species have additional microchaetae that can be construed as belonging to this series (e.g., T. jataca, Fig. 52) but only A0-3 are present in all species examined. Seta A1 is always a normal, coarsely ciliate microchaeta, all other members of the series can develop into macrochaetae.
Series M includes 2 setae, probably homologous to M3-M4. In most species the lateral seta in series M is internal to S5, but in troglomorphs T. jacobyi sp. n. and T. belizeana the seta is inserted external to S5 and just internal to the dorsal cephalic suture. M0 is absent (seen only in one individual of T. giordanoae sp. n.), whereas M3 is often developed into a macrochaetae. Series S includes setae S1-5, S0 is absent (seen only in one individual of T. geminata). Among the species examined only setae S3 and S5 are modified into macrochaetae. Most setae in series S are inserted along the dorsal cephalic sulcus; the exceptions are S1, which is anterior to all others, and seta S3 when it is modified into a macrochaeta (cf., T. giordanoae sp. n. [Fig. 5] versus T. jacobyi sp. n. [Fig. 26]).
There is a pattern in the addition of macrochaetae on the interocular region of the head for species with 3-4 macrochaetae, but the pattern in not retained for species with five macrochaetae: whenever three macrochaetae are present they are always A0, A2 and M3; the species with four macrochaetae carries A0, A2 and M3 plus S3; the species with five macrochaetae have A0, A2, S3, S5, and either A3 or M3.
Series Ps includes only two setae (Ps2 and Ps5) whereas series Pa has four setae (Pa2, 3, 5 and bothriotrix Pa6), and series Pm and Pp has one seta each (Pm3 and Pp3). Posterior setae Pa5 and Pm3 are often modified into macrochaetae.
Mesothorax. The chaetotaxy of the mesothorax is reduced, as in scaled Entomobryidae (e.g., Seira, Pseudosinella [Soto-Adames 2008, 2010), the closest group of Entomobryoidea for which detailed information about chaetotaxy is available. All Trogolaphysa species share the presence of macrochaetae a5 and p3, and microchaetae m2, m4, p4, p5, and what we provisionally call p6. Setae p1 and p2 are present in the three species from Belize but either absent or integrated in the p3 macrochaetae complex in the three surface species from Puerto Rico (Fig. 53) The homologies of the posterior macrochaetae across the species examined are unclear. The presence of setae p1 and p2 in T. giordanoae sp. n. suggests that the cluster of six posterior macrochaetae represent a multiplication of seta p3; whereas the transfor-mation of p1 and p2 into macrochaetae in T. jacobyi sp. n. and T. belizeana, and their absence in the surface species T. jataca, T. geminata and T. riopedrensis suggest that the three setae have been integrated (and duplicated) into the macrochaetal complex. We propose three hypotheses to explain the evolution of posterior macrochaeta: the macrochaetae evolved independently more than once in the genus, either as 1) a duplication of p1-3 or as 2) multiplication of p3 alone; 3) the cluster evolved only once, a duplication of p1-3, and the setae we have identified as p1 and p2 in T. giordanoae sp. n. are secondary and not homologous to those present in T. jacobyi sp. n. and T. belizeana. A study of the postembryonic development of these setae or molecular phylogenetic analysis may provide evidence in support one of the hypotheses proposed above.
Metathorax. The chaetotaxy of this segment is reduced to five setae (e.g., T. geminata, Fig. 58). The homologies of these setae are uncertain, and names provided in Fig. 58 are based on comparison with the general organization of the chaetotaxy in first instar Seira dowlingi (Wray, 1953), Heteromurus nitidus (Templeton, 1835) and Willowsia buskii (Lubbock, 1870) (Soto-Adames 2008, Szeptycki 1979. The single macrochaeta present in T. jacobyi sp. n. appears to be homologous to p3, whereas the three macrochaetae present in T. giordanoae sp. n. and T. belizeana appear to be homologous to a displaced a2, p2 and p3.
Abdomen 1. This segment also has a reduced chaetotaxy, carrying not more than six setae (Figs 14, 59). The homologies proposed are based on comparisons with first instar S. dowlingi, H. nitidus and W. buskii (Soto-Adames 2008, Szeptycki 1979. Seta a6 is present in T. giordanoae sp. n., T. geminata and T. riopedrensis and absent in T. jacobyi sp. n. and T. jataca. Abdomen 2-3. The chaetotaxy of these segments was previously described by Mari Mutt (1987Mutt ( [1988) and the species examined here, including T. belizeana, conform to that description. These two segments do not carry inner microchaetae beyond those associated with the bothriotichal complexes. The macrochaetae on Abd. 2 are homologous to m3 and m5. Lateral setae a6, m6 and p5 appear to be present in all species, although a6 and m6 are often difficult to see. The socket of p5 is enlarged, macrochaeta-like, but this seta falls off in most slide-mounted individuals, it was observed in T. jacobyi sp. n, where it is a ciliate mesochaeta and in T. jataca, where it is enlarged and fusiform (Fig. 54).
Abdomen 4. The chaetotaxy of Abd. 4 is similar to that in scaled Entomobryidae and setae modified in, for example, Seira or Lepidocyrtus, can also be modified in Trogolaphysa. The chaetotaxy displays some unique peculiarities. For example, what appears to be seta B6 is, in most species, a meso-or small macrochaeta inserted just posterior to B5 (Fig. 17). In addition, the posterior bothriotrix corresponds to D4 (D3 in Seira, Soto-Adames 2008). The number, identity and relative insertion of inner macrochaetae varies between Trogolaphysa species. Trogolaphysa giordanoae sp. n. and T. riopedrensis share the same inner macrochaetae (A4, A5, B4, B5), but the insertion of B4 in relation to the pseudopore and seta C2 differ between these two species (cf.,Figs 17,60). Trogolaphysa geminata, T. jataca and T. belizeana have three inner macrochaetae and share macrochaetae A5 and B5, but whereas in T. jacobyi sp. n. the third macrochaetae is B4, in the other two species it is A4. Trogolaphysa jacobyi sp. n. is also unusual in having macrochaeta B5 displaced towards A6 instead of C4 (Fig. 38).
The external macrochaetae in the first three rows of columns D, E and F are stable in the species of examined. All species have macrochaetae D3, E2, E3, F1 and F2. Macrochaeta F3 is present in all species except T. jacobyi sp. n. The number of macrochaetae external to column F and posterior to row 3 varies intra-and interspecifically. However, the lateral and posterior fields are often difficult to see in regular preparations and it is possible that some of the apparent differences are simply incomplete observations.
Phylogenetic analysis based on all characters supports two equally parsimonious trees  in which the two troglobiontic species from Belize form a monophyletic group and T. giordanoae sp. n. is placed at the base of the species from Puerto Rico. The parsimony trees support the sister species relationship between T. subterranea and T. luquillensis, but relationships between the other three species from Puerto Rico are unresolved, as T. riopedrensis is placed as sister to either T. jataca or to a clade that includes all other island species.
The apparently rare occurrence of metathoracic macrochaetae in the three Belizean species suggests a close relationship between them, but the parsimony trees show the troglobiontic species diverging before the separation of T. giordanoae sp. n. from the ancestor of the island species. The lack of support for the monophyly of Belizean species may be an artifact of a disproportionate contribution of characters under strong cave habitat selection to the final topology of the tree. However, parsimony analysis based only on chatotactic characters results in a single shortest tree (Fig. 63), which also supports the monophyly of troglobiontic species while retaining T. giordanoae sp. n. at the base of the island species clade.
To assess whether putative adaptive characters provide support for alternative relationships, we conducted a phylogenetic analysis using only eye number, ornamentation of labral papilla, and claw and mucro morphology. These characters support a single tree (Fig. 64) that places most surface forms at the base of the tree while supporting a clade comprising the cave species (T. jacobyi sp. n., T. belizeana, T. subterranea) and T. luquillensis. Trogolaphysa luquillensis is endemic to the tropical rainforest and is unique among surface species examined here in having an acuminate tenent hair and three inner ungual teeth close to each other and inserted in the basal half of the claw. These characters of the claw have been identified as adaptations to walking on water surface or other, permanently wet, surfaces such as those found in rainforest leaf litter and caves (Christiansen 1961(Christiansen , 1965. Evaluation of the direction of evolution of head chaetotaxy using trees in Fig. 61 and Fig. 62 supports a trend towards a reduction in number of macrochaetae. However, the pattern is equivocal because some macrochaetae may be lost independently through out the tree, depending on tree topology. For example, S5 might have been lost once and regained or it might have been lost twice independently. What is clear from this analysis is that A3 is the first macrochaetae to be lost, followed by S5, S3 and M3 (Table 4). Trogolaphysa riopedrensis is the only species in which this pattern seems to be disrupted: under either tree this species is hypothesized to have lost M3 and gain A3 independently.

Taxonomic status of Dicranocentruga and Trogolaphysa
The character used by Mitra (1993Mitra ( , 2002 to diagnose genera Trogolaphysa and Dicranocentruga can be difficult to apply. The presence of EOS is difficult to ascertain using phase contrast or DIC light microscopy. The retention of Dicranocentruga as a valid genus hinges on whether Trogolaphysa maya, the type species of Trogolaphysa, carries the EOS. As pointed out above, the presence of EOS in the two troglomorphic species considered here suggests that this structure is also present in T. maya. We examined the single alcohol preserved syntype of T. maya deposited at the Illinois Natural History, but the condition of the specimen is such that confirmation of the presence of the EOS is impractical. It is possible, as proposed by Mitra (1993), that a more extensive analysis of idiochaetotaxy may provide diagnostic characters for these two genera that are easier to see and interpret. The present study does not support this idea. The organization of the idiochaetotaxy is the same in all the species studied. Changes in the distribution of setae, as in the case of the metathorax, are related to the morphology of the elements (whether macro-or microsetae), and not to the presence of EOS, number of eyes, or other cave adaptive characters. Until such time as the presence of EOS can be reliably determined, or other diagnostic characters are found, we retain all New World Dicranocentruga in the genusTrogolaphysa, as proposed by Najt (1988[1989]).
pecially cave forms, have been diagnosed almost exclusively based on characters of the claw complex, mucronal shape and development of the antennae, characters identified as malleable under selective pressures (Christiansen 1961). A new set of characters or character systems, would be needed to perform more reliable phylogenetic analyses. Most other studies addressing the evolution of morphological convergence in caveadapted arthropods have used molecular data (e.g., Trontelj et al. 2012) to generate phylogenies for hypothesis testing. However, many troglobiontic springtail species are known only from a few individuals from few, seldom visited localities (as is evident from the small number of records reported in Mari Mutt and Bellinger 1990, and Mari Mutt et al. 2009), which are not suitable or available for molecular analysis.
For these species only morphological information can be used to evaluate the evolution of other morphological characters. Ever since the publication of Gisin's (1967) "systématique ideal," collembolan systematists have assumed that ideochaetotaxic characters are non-adaptive characters that evolve neutrally, are less prone to convergence and, therefore, more valuable for phylogenetic analysis. However, this assumption has never been tested in a phylogenetic context. The simple test performed here supports the traditional view of chaetotaxy as less vulnerable to directional convergence than characters related to claw structure. Analysis based exclusively on putative cave-adaptive characters support a clade comprising cave species from Puerto Rico and Belize, whereas analysis of chaetotaxy alone supports the placement of cave species from Puerto Rico and Belize in independent clades. Despite the clear difference in signal in the character partitions it should be noted that analysis of the complete character set results in higher bootstrap values for what is basically the chaetotaxy-only tree, than when only chaetotactic characters are analyzed. It is clear that some putative adaptive characters retain phylogenetic information concordant with chaetotaxy characters, an observation which argues in favor of the retention of all characters in the analysis. The simple test preformed here has to be expanded to include many more species, to determine if the result obtained are consistent or just an artifact of the sparse taxon sampling. It is unclear if chaetotaxy will provide sufficient characters to resolve relationships in an analysis that includes all species. In any case, there are problems related to the evolution and homology of some chaetotactic characters (e.g., posterior macrochaetae on the meso-and metathorax, and the inner macrochaetae on the fourth abdominal segment) that may be intractable on morphology-based datasets, and will require the use of putatively independent molecular characters.