Two New Cave-Dwelling Species of the Short-Tailed Whipscorpion Genus Rowlandius (Arachnida: Schizomida: Hubbardiidae) from Northeastern Brazil, with Comments on Male Dimorphism

Two new species of the arachnid order Schizomida, Rowlandius ubajara sp.nov. and Rowlandius potiguar sp.nov., are described based on both male and female specimens collected in caves from northeastern Brazil. Rowlandius ubajara is known only from the Ubajara Cave, in the state of Ceará; R. potiguar is recorded from 20 caves of the Apodi Limestone Group, in the state of Rio Grande do Norte. A remarkable dimorphism in male pedipalp length is described and analyzed in R. potiguar. The distribution of male pedipalp length is clearly bimodal in the species, but the two male morphs (homeomorphic and heteromorphic) present some overlap in the sizes of this structure. Moreover, males show a steeper allometry in pedipalp length than females, indicating that this trait is under a different selective regime in males and in females.


Introduction
The arachnid order Schizomida, the short-tailed whipscorpions, is represented in South America mainly in northern forested areas. Besides its great diversity and broad distribution in Central America and the West Indies, in the continental South America it can be considered a poorly-known Amazonian group. Until very recently, the only species known from southern parts of the subcontinent was Stenochrus portoricensis Chamberlin, 1922 [1,2], a widespread and synanthropic species. However, the scarcity of schizomid species outside the Amazonian forest can be attributed, at least in part, to collecting bias. This happens because schizomids occur mostly in poorly sampled habitats, like the soil leaf litter, termite and ant nests, and caves [2,3]. Subterranean habitats in particular can harbor large schizomid populations [3], and are a promising source of undescribed species.
Among South American countries, Brazil is known for its abundance of caves, which are scattered in several geological formations [4]. Despite the richness of cave habitats, the Brazilian subterranean fauna is mostly unknown [5], though it is currently attracting the interest of many biologists.
The genus Rowlandius Reddell and Cokendolpher, 1995, the focus of this study, is a good example of how little is known about the schizomids and the cave fauna of Brazil. The majority of the 54 species of the genus [2,6,7,8,9,10] occur in Caribbean islands, mainly Cuba, Jamaica and Hispaniola (see for instance [11,12,13,14,15]). The genus is represented in continental South America by one species in Venezuela (R. arduus Armas, Villareal & Colmenares, 2009), one in the eastern Brazilian Amazonia (R. sul Cokendolpher & Reddell, 2000) and another, recently described species from the northeastern Brazilian Atlantic Forest (R. linsduarteae Santos, Dias, Brescovit & Santos, 2008) [8]. In this study, two new species are described based on specimens collected in cave habitats in northeastern Brazil. Most caves from which specimens were retrieved are imbedded in the caatinga, a northeastern Brazilian, semi-arid biome [16], an unexpected habitat for a group mostly considered as a humid forest dweller.
Several species of Rowlandius and other genera of schizomids are known by the remarkable variation in the length of male pedipalp. Depending on the species, the length of the male pedipalp can vary from approximately the same size as those of the female to up to three times as long (e.g. [11,15]). Authors describing species of these genera usually discriminate males in two categories, namely homeomorphic males and heteromorphic males, for specimens with short and long pedipalps, respectively [11]. This classification is based on the assumption that two discrete male morphs can be recognized based on the pedipalp size, relative to total body size. Male dimorphism has been described for several groups of arthropods, like insects [17], crustaceans [18] and, among arachnids, spiders [19], mites [20], pseudoscorpions [21] and harvestmen [22,23]. This phenomenon has been attributed to intrasexual selection pressures favoring a conditional strategy with different tactics for small and large males [24]. However, the detection of male dimorphism can be statistically challenging, especially when the differences between the morphs are subtle [25]. For schizomids, no criteria have been proposed to discriminate between homeomorphic and heteromorphic males, and this distinction is generally based on small sample sizes, usually no more than five or six specimens. In this study, we take advantage of a large sample of specimens to evaluate whether or not it is possible to recognize discrete male morphs in one of the newly described species.

Specimens and Laboratory Procedures
The specimens examined for this study were collected in several caves in two states in northeastern Brazil (Fig. 1) [26]) to remove soft tissues, air-dried and mounted on the stub. Description format and abbreviations follow Santos et al. [2] and Santos & Pinto-da-Rocha [27]. Morphological nomenclature follows [28] for chelicerae setation, [29] and [30] for flagellum setation and [3] for female internal genitalia. All measurements are in millimeters.

Male Dimorphism Analysis
Male dimorphism was at first visually evaluated in scatterplots of prosoma length (as a proxy for general body size) and pedipalpal patella length (indicative of pedipalp size). To check whether a non-linear or discontinuous allometric pattern was apparent, natural values were used without logarithm transformation (caveats of such transformations explained in [31]), and the scatterplot axes were isometric (such that DX = DY; see [32] for why this is important). Both variables were checked for bimodality through adjustment of non-parametric kernel density estimates to frequency distributions. Pedipalp size distribution seemed strongly bimodal, and therefore was parameterized as a mixture of two 'facing gamma distributions' using finite mixture models, according to model ,d. in [33]. The 'facing gamma distributions' proposed by these authors accommodate skews in the distributions of the trait for each male morph, modeling a skew to the right (towards higher values) among homeomorphic males and a skew to the left (towards lower values) among heteromorphic males.
This parameterization was performed in SAS version 9.2 (SAS Institute 2004) with code kindly provided by J.M. Rowland and C.R. Qualls. Finally, the likelihood of belonging to each male morph was computed from the mixture models for each male based on their pedipalp length.
Simple linear models were fitted to the allometric relationship between pedipalp length and prosoma length for females and for each male morph separately, through standard major axis regression using package 'lmodel2' [34] in R version 2.14.2 [35]. For this last step of the analysis, males were classified as homeomorphic or heteromorphic when their likelihood of belonging to one particular male morph (computed from the mixture models) was higher than 95%. Males with less than 95% chances of belonging to either male morph were left out when adjusting the allometric relationships between pedipalp length and prosoma length for homeomorphic and heteromorphic males.

Nomenclatural Acts
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature [36], and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http://zoobank.org/". The LSID for this publication is: (urn:lsid:zoobank.org:pub:4882594A-5176-4668-A8B3-20DFFDD4DA73). The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central, LOCKSS.

Etymology
The specific epithet is a noun in apposition from the type locality.

Diagnosis
Males of R. ubajara can be distinguished from the remaining species of the genus by the flagellum ovoid, wider at the basal third, with the anterior surface between setae dl3 wide and convex and with the dorsal projections wide, separated by a distance equivalent to their width and not touching or surpassing the lateral border of the flagellum (Fig. 3A, B). Females are recognized by the chitinized arch of the internal genitalia u-shaped, closed and pointed laterally, lateral lobes of the spermathecae with curved stalks and small bulbs and the median lobes cylindrical, wrinkled and with half the length of the lateral lobes (Fig. 3F).

Description
Male (holotype). Propeltidium pale brown, chelicerae and pedipalp reddish-brown. Metapeldium and legs pale-brown. Opisthosoma greenish-brown. Anterior process with one apical seta. Propeltidium with an anterior and a posterior pair of setae.
Female (Paratype, UFMG 3896). As in male, except by the following. Prosoma, pedipalp and leg I reddish-brown (Fig. 2). Opisthosoma, chelicerae and legs II-IV greenish brown (Fig. 2). Anterior process with 1+1 setae. Flagellum with four segments, setation as in the male, except by three pairs of lateral microsetae: one pair in segment II (adjacent to dm1); one in III and one in IV (between dm4 and dl3).

Natural History
This species was recorded from a single limestone cave located in a small patch of Brazilian Atlantic forest in northeastern Brazil (in Ubajara National Park). Part of the cave is open for visitors and is illuminated by electric lamps. Specimens of R. ubajara were found in the inner part of the cave, which consists of a single passage through which a small stream flows. Potential food resources observed in the cave are mainly bat guano from insectivorous and hematophagous bats and potential prey for R. ubajara could be mainly springtails (Collembola) and booklice (Psocoptera: Psyllipsocidae). The population is apparently small, and only about 10 specimens were observed in the cave during the collection. The average temperature of the cave was 23.5uC and the humidity was 99% in the inner parts, where the species occurs. Although the population only occurs in the inner part of the cave and the species was observed in a single cave, the absence of eye spots is the only potentially troglomorphic trait exhibited by the species. This characteristic, as well as the absence of pigmentation and elongation of appendages, is frequently suggested as evidence of troglobiotism in schizomid species [3,37]. However, it is not uncommon to find schizomids with those traits associated with epigean habitats, as well as species without any apparent troglomorphism living in caves ( [37] and references therein). It is reasonable to consider schizomids as pre-adapted for living in caves, mostly due to their low reliance on visual orientation. Except for four species with convex eye lenses [38,39], the members of the order have either no eyes or only vestigial, pale eye spots [3]. It is uncertain whether the eye spots, or even the well formed eyes, are truly functional, since no species have been evaluated through histological or behavioral methods. Thus, the presence or degree of development of the eyes does not seem to be a good indicator of adaptation to caves in this group. Since not all troglobiotic species necessarily possess troglomorphic traits [40], further studies should be conducted in the area (especially in the external environments) to actually confirm the degree of association of this species with hypogean habitats.

Distribution
Known only from the type locality in the northeastern Brazilian state of Ceará (Fig. 1).

Etymology
The specific epithet is an adjective used in Brazil to designate the natives from the state of Rio Grande do Norte.

Diagnosis
Males of Rowlandius potiguar resemble those of R. linsduarteae Santos et al., 2008 in the subquadrate flagellum, but differ in its rounded sides, dorsal projections narrower in dorsal view and the setae dm4 closer to the posterior border ( Fig. 4A-C, 5A). The flagellum of this species is also similar to that of R. monticola Armas, 2002 in general shape ( fig. 3b in [41]), but differs in the wider stalk and the dorsal projections closer to each other. Females are similar to those of R. sul Cokendolpher & Reddell, 2000 and R. linsduarteae in the conical shape of the median lobes of the spermathecae ( fig. 6 in [42]) and specifically to R. linsduarteae in the lateral lobes of the spermathecae with curved stalks and large bulbs ( figure 6 in [2]). These species can be distinguished by the lateral lobes with thinner stalks and larger bulbs and by the presence of a gonopod in R. potiguar (Figs. 4G, 6D-F). As in several other species of Rowlandius, male pedipalp are variably longer than in the female (Figs. 2B, C; 4D, E; 7A, B -see discussion below). However, unlike some species, such as R. gladiger (Dumitresco, 1977), R. biconourus (Rowland & Reddell, 1979) R. longipalpus (Rowland & Reddell, 1979) and R. falcifemur Teruel, 2003 (figs. 51, 55, 57 in [14], fig. 7 in [15]), neither the trochanter nor the femur has anterior projections (Fig. 4D, E).

Remarks
A peculiar lobed aperture was described from the appendages of Schizomida and Thelyphonida by Santos & Pinto-da-Rocha [27] and was proposed as a potential synapomorphy of Uropygi. This peculiar aperture has been reported from the tarsus of leg I and the flagellum, in both males and females of three schizomid and two thelyphonid genera. In the current study, we also report this structure in the male pedipalp of R. potiguar (Fig. 6A). Although Santos & Pinto-da-Rocha [27] suggest this aperture could be a glandular opening, its function remains unclear. Thus, we here propose to refer to this structure simply as the uropygid pore.

Natural History
This species is distributed in 20 caves along the Apodi limestone group, in the western part of the state of Rio Grande do Norte. The area is located in the caatinga, a seasonally-dry forest formation [16]. This limestone group is composed of huge outcrops, located in various municipalities, including Jandaíra, Felipe Guerra, Apodi, Martins and Baraúnas. Populations of R. potiguar were found in many caves (Fig. 1), although the population sizes were variable, possibly due to different conditions in each cave. The species is apparently associated with moist caves, since no specimens have been observed in completely dry caves. In the dry Casa de Pedra cave, the southernmost record of the species, the single specimen observed was in the only wet area of the cave, in sediments humidified by dripping water. Field observations suggest that this species occurs in large populations, with hundreds of individuals, particularly close to bat guano or seed deposits transported by bats. These organic piles harbor several scavenger invertebrates, like springtails (Collembola), small flies (Diptera) and booklice (Psocoptera); which could serve as prey for schizomids.

Distribution
Known from 20 caves in the northeastern Brazilian state of Rio Grande do Norte (Fig. 1).

Male Dimorphism
Pedipalp length in R. potiguar males is remarkably variable, presenting 20.11 times more variance than prosoma length (variances standardized by the means of the traits), and its distribution is clearly bimodal ( Figure 7A). Meanwhile, the same trait in conspecific females presents a unimodal distribution, with only 0.50 of the variance presented by prosoma length (variances standardized by the means of the traits). Moreover, a mixture of two 'facing gamma distributions' (h = 0.163, l = 1.91, lower bound = 0.48 mm for homeomorphic males; h = 0.201, l = 2.88, upper bound = 2.25 mm for heteromorphic males; Figure 7A) fits the distribution of male pedipalp length very well. This model estimates that at least 95% of males with pedipalpal patella shorter than 1.02 mm represent the homeomorphic morph (empty circles, Figure 7B) and at least 95% of the males with pedipalpal patella longer than 1.23 mm represent the heteromorphic morph (full black circles, Figure 7B).  The only known cases of elongated male-dimorphic traits in arachnids were described for the order Opiliones [22,23], where the elongation of the second or fourth pair of legs is bimodal and extremely variable among males. In these cases, the male dimorphic traits are sexually selected weapons used in male-male fights. As far as we know, there is no evidence of male-male fights in Schizomida, and the only studies that describe the courtship and copulation in the order [43,44] do not indicate any participation of the male pedipalp in mating. However, there is evidence that males of the schizomid Hubbardia pentapeltis use their  [33]). The bars represent the frequency (scale on the left) of males in each pedipalpal patella length bin, and the curves represent the distributions of predicted probabilities (scale on the right, in %) of each pedipalpal patella length for both male morphs, as estimated by the model. (B) The relationship between the length of pedipalpal patella and the length of prosoma. Filled black dots and black line indicate males with a probability of being heteromorphic higher than 95%, empty dots and dashed line indicate males with a probability of being homeomorphic higher than 95%, filled gray dots indicate males with probabilities lower than 95% of being either morph, and crosses and dotted line indicate females. Axes are isometric to show male morphs in the most objective fashion. doi:10.1371/journal.pone.0063616.g007 pedipalps during courtship. In this species, the male stretches out his pedipalps, picks up small twigs, and manipulates them in various ways while the female seemingly examines the male's behavior, repeatedly touching his anterior body parts with her first legs (JM Rowland, personal communication). In conclusion, it is also possible that the elongation of pedipalps in R. potiguar evolved in the context of courtship.
The fact that the static allometry of pedipalps is significantly steeper in males than in females ( Figure 7, Table 1) cannot be readily taken as evidence that such pedipalps are under sexual selection [45], as it has been interpreted in the past for other taxa [46]. However, it does indicate that pedipalp length is under a different selective regime in males than in females. We urge for studies on the sexual behavior of R. potiguar and other schizomids, which would be promising approaches to shedding light on the evolutionary causes of pedipalp elongation and male dimorphism in the group.
In conclusion, an objective and robust discrimination of male morphs might be crucial for in depth behavioral and ecological studies of particular male dimorphic species, but it does not seem to be so important for taxonomic descriptions. Nevertheless, it is still essential that taxonomists realize the existence of male dimorphism, and hence avoid describing male morphs as different species. This does not seem to be a problem in schizomid systematics, since species distinction is based mainly on male flagellum. Yet, it is relevant to report variation in male pedipalp length within species and, whenever a reasonably big sample is available, discriminate morphs properly. The detection of male dimorphism by taxonomists can in itself stimulate behavioral ecology studies on these animals, which could elucidate their mating systems and unveil the alternative mating tactics of males that are potentially behind the evolution of male dimorphism in Schizomida. Table 1. Coefficients (and their respective 95% confidence intervals) from the simple linear models fitted to the allometric relationship between pedipalp length and prosoma length for each male morph and for females.