Further study of two Chinese cave spiders (Araneae, Mysmenidae), with description of a new genus

Abstract The current paper expands knowledge of two Chinese cave spider species originally described in the genus Maymena Gertsch, 1960: M. paquini Miller, Griswold & Yin, 2009 and M. kehen Miller, Griswold & Yin, 2009. With the exception of these two species, the genus Maymena is endemic to the western hemisphere, and new evidence presented here supports the creation of a new genus for the Chinese species, which we name Yamanetagen. nov. The male of Y. kehen is described for the first time. Detailed illustrations of the habitus, male palps and epigyne are provided for these two species, as well as descriptions of their webs. DNA sequences are provided for both Yamaneta species. We build on a previously published phylogenetic analysis of Mysmenidae to assess the phylogenetic position of Yamaneta and its relationship to true Maymena.


Introduction
The genus Maymena Gertsch, 1960 was established in the context of a taxonomic paper describing several American spiders of the family Symphytognathidae Hickman, 1931. At that time, the concept of Symphytognathidae was broader than it is today; the taxa described therein are currently distributed among four families (Symphytognathidae, Mysmenidae Petrunkevitch, 1928, Anapidae Simon, 1895and Theridiosomatidae Simon, 1881. The world's Symphytognathidae were reviewed and redefined by Forster and Platnick (1977), and several symphytognathid genera were transferred to other families, including Maymena to the Mysmenidae. Gertsch's (1960) original description included observations of several characteristics of Maymena but did not provide diagnostic characters for separating the genus from its close relatives. It was not until the recent publication of Lopardo and Hormiga (2015) that a rigorous and convincing diagnosis of Maymena was finally published. Miller et al. (2009) described several symphytognathoid spiders from the Gaoligong Mountains, Southwest China. Two species from caves were placed in the genus Maymena: M. paquini Yin, 2009 andM. kehen Miller, Griswold &Yin, 2009, the latter species being known only from females. The genus Maymena currently contains 13 described species (World Spider Catalog 2019). Except for the two Chinese species, all are known from the western hemisphere, from the USA south through Mexico, Central America, the Caribbean, and Peru. In addition, two undescribed taxa in Lopardo et al. (2011) considered to belong to Maymena (Lopardo and Hormiga 2015) were from Argentina. Most Maymena species are clearly associated with caves, although a few species are occasionally or typically found in surface habitats (Gertsch 1960, 1971, Brignoli 1974, Baert 1990, Eberhard et al. 1993. In August 2008, students and professors of Sichuan University carried out a collecting survey in the Gaoligong Mountains. Both males and females of Miller et al.'s Chinese Maymena species were collected from their type localities and their web structures were discovered and photographed. In addition to new detailed morphological data and the description of the previously unknown male of M. kehen, multiple individuals of both species were sequenced for five loci. To test the relationships of Chinese Maymena to western Maymena and other Mysmenidae, we added this DNA sequence data to the molecular phylogenetic dataset of Lopardo et al. (2011).

Material and methods
Specimens were acquired by hand from the dark zone of caves and preserved in 95% ethanol. They were examined using a Leica M205 C stereomicroscope. Further details were studied under an Olympus BX43 compound microscope. Male palps and epigynes were examined and photographed after dissection. Epigynes were treated in lactic acid before being embedded in Arabic gum to take the photos of the vulva. To reveal the course of the spermatic ducts, male palps were also clarified using lactic acid and subsequently mounted in Hoyer's Solution. The left palp was photographed and described. Photos were taken with a Canon EOS 60D wide zoom digital camera (8.5 megapixels) mounted on an Olympus BX 43 compound microscope. The images were montaged using Helicon Focus 3.10 (Khmelik et al. 2006) image stacking software. All measurements are in millimeters. Leg measurements are given as follows: total length (femur, patella, tibia, metatarsus and tarsus).
Tissue samples were taken from eight individual specimens of Chinese Maymena representing both known species. Whole genomic DNA was extracted from tissue samples with TIANamp Micro DNA Kit (TIANGEN) following the manufacturer's protocol for animal tissues. Five gene fragments were amplified in 25μL reactions: mitochondrial large-subunit ribosomal RNA (16S), nuclear small-subunit ribosomal RNA (18S), nuclear large-subunit ribosomal RNA (28S), cytochrome c oxidase subunit I (COI), and histone H3 (H3). Primer pairs and PCR protocols are given in Table 1. Raw sequences were edited and assembled using BioEdit v.7.2.5 (Hall 1999). New sequences generated for this study were deposited in GenBank; accession numbers are reported in Table 2. All molecular vouchers and examined materials are deposited in the Natural History Museum of Sichuan University in Chengdu (NHMSU), China.
The most recent molecular phylogeny of Mysmenidae was Lopardo et al. (2011). Lopardo supplied alignments of the six genes used in their analysis (the five above plus the mitochondrial small-subunit ribosomal RNA 12S). Taxonomic determinations were updated according to notes in Lopardo and Hormiga (2015). The COI sequence "Mysmena-MYSM-018-MAD" (GU456888) was omitted because it was flagged on GenBank as removed at the submitter's request because of possible contamination. We used the MAFFT version 7 online service (https://mafft.cbrc.jp/alignment/server/add_ sequences.html) with the following settings to add the Chinese Maymena sequences to the existing alignments of the five shared loci (Strategy: Auto, scoring matrix for nucleotide sequences: 200PAM/k=2, Gap opening penalty: 1.53, offset value: 0.0; Katoh et al. 2017). Alignments of all six loci were concatenated in Geneious version 8.1.8 (https://www.geneious.com). The final alignment consisted of 6038 positions (Suppl. material 1). Uncorrected pairwise distances between terminals in the expanded alignment were calculated using MEGA X (Kumar et al. 2018) and are included as Suppl. material 2. We did not build on the morphological data matrix of Lopardo et al. (2011) or the expansion of this dataset in Lopardo and Hormiga (2015), and these data were not used in our analysis. This is because DNA sequence datasets are relatively simple to expand with additional data; it would be difficult for us to do the same for the large and complex morphological dataset without risking the introduction of errors and artifacts.
The most parsimonious tree was found using 1000 replicates of random taxon addition and TBR (Tree-Bisection-Reconnection) branch swapping using MEGA X (Kumar et al. 2018). To assess support, bootstrap values were calculated using MEGA X (Kumar et al. 2018) with 1000 bootstrap replicates, each consisting of 1000 replicates of random taxon addition and TBR branch swapping. Departing from the approach of Lopardo et al. (2011), gaps/missing data were treated as ambiguities, not as a 5 th character state.
The Bayesian phylogenetic inference was performed using MrBayes version 3.2.6 (Ronquist and Huelsenbeck 2003) through the Cipres Science Gateway (Miller et al. 2010). Partitions and models followed Lopardo et al. (2011;table 4). A total of 10 partitions were defined (three independently modeled regions each for the nuclear ribosomal genes 18S and 28S, single model for each of the remaining four loci). The Bayesian search consisted of 50,000,000 generations using four chains, with the chain sampled every 1000 generations (see Suppl. material 1). Tracer version 1.7.1 (Rambaut et al. 2018) was used to establish the appropriate level of burn-in, which was set at 5000.

Results
Parsimony analysis of the expanded sequence alignment recovered a single most parsimonious tree (Fig. 1). This tree features a monophyletic, but weakly supported, Mysmenidae. Western hemisphere and Chinese Maymena are reciprocally monophyletic, with moderate bootstrap support. The two Chinese species are coherent. Few clades, especially along the backbone of the phylogeny, have high bootstrap support, and relationships among outgroup taxa are complicated. The low support values seem in part to be attributable to a number of unstable taxa. Maymena (western and Chinese) and Trogloneta Simon, 1922 together form a paraphyletic complex, with one branch of Trogloneta sister to Maymena (western and Chinese) and the other Trogloneta branch sister to the remaining Mysmenidae. After 50,000,000 generations of Bayesian analysis, the average deviation of split frequencies fell below 0.05. The combined effective sample sizes of the two MCMC chains were 7425.9 and 7654.5 (12,520.9 combined), comfortably above the recommended minimum of 200 (Lanfear et al. 2016). The Bayesian topology (Fig. 2) features a monophyletic Mysmenidae, which in contrast to the parsimony analysis enjoys high support from posterior probability. As in the parsimony analysis, western and Chinese Maymena are reciprocally monophyletic with high support, and the two Chinese species are coherent. However, branch lengths suggest a long separation between the Chinese and western lineages. The Bayesian analysis also indicates complicated relationships among the outgroup taxa. Maymena (western and Chinese) and Trogloneta together form a paraphyletic complex similar to that found in the parsimony analysis; one branch of Trogloneta is sister to all other Mysmenidae, and the other branch of Trogloneta is sister to the remaining Mysmenidae, except for Maymena (western and Chinese).

Discussion
Monophyly of and relationships between the so-called symphytognathoid families (including Mysmenidae, Anapidae, Theridiosomatidae and Symphytognathidae) are complicated and inconsistent across various analyses. Early attempts based on morphological data (e.g., Griswold et al. 1998) proposed a "symphytognathoid" clade composed of monophyletic families, but Schütt (2003) warned that some of this could be based on a syndrome of parallel reductions and simplifications related to the evolution of small size. DNA sequences initially seemed promising as a source of phylogenetic data independent of morphology, where homology assessment could be confused with parallel evolution. The analysis of Rix et al. (2008) tested the relationships of Anapidae and Micropholcommatidae (currently considered part of Anapidae; Schütt 2003, Lopardo et al. 2011, World Spider Catalog 2019, with representatives of several other relevant families, based on DNA sequence data from two nuclear ribosomal loci. Results concerning the primary focal group of the study, the Micropholcommatidae, were fairly robust and consistent.    However, results regarding the Anapidae and key outgroup taxa were generally of poor resolution and inconsistent across tree-building methods. This despite being built upon what was, for the time at least, a rigorous and sophisticated analytical approach. The results of the analyses reported in Lopardo et al. (2011) were similarly sensitive to changes in analytical parameters. This analysis was based on an expanded set of loci compared to Rix et al. (2008), plus a set of morphological characters, and many permutations of data partitions and phylogenetic optimization methods were employed. The monophyly of Mysmenidae was relatively robust to permutations of the analysis, but the inclusion of morphological data had a tendency to support the monophyly of outgroup families, which sometimes collapsed in analyses based on molecular sequences alone. A series of studies by Dimitrov et al. (2012Dimitrov et al. ( , 2016 used roughly the same set of loci with a progressively expanded sample of taxa to explore deep questions of spider relationships. These consistently recovered two clades of anapids and never found support for the monophyly of symphytognathoid families. The phylogenetic analysis of Wheeler et al. (2017) further expanded the taxon sample, but not the selection of loci. Anapidae (including a monophyletic Micropholcommatinae), Symphytognathidae, and Mysmenidae were each monophyletic and moderately well supported; Theridiosomatidae was monophyletic only after the pruning of one problematic taxon, and still presented with low support. The four symphytognathoid families together were found to be closely related but not monophyletic. The parsimony and Bayesian phylogenies presented here disagree about outgroup relationships in several important ways, including the monophyly of Anapidae, its relationship to Micropholcommatinae, and the sister clade to Mysmenidae. Such results are not surprising, because previous studies relying on the same limited set of reliable loci have seen similar results for nearly a decade, and also because taxon sampling outside Mysmenidae in this study and its predecessors (Lopardo et al. 2011, Lopardo andHormiga 2015) is very limited.
Recent phylogenomic approaches have finally expanded the volume of DNA sequence data used to investigate spider phylogeny (Bond et al. 2014, Fernández et al. 2014, Garrison et al. 2016), but only Fernández et al. (2018) has achieved the taxon sampling necessary to address some of the longstanding symphytognathoid questions. Their study, based on ca 2500 genes, found monophyletic Theridiosomatidae (4 terminals), Mysmenidae (3 terminals) and Anapidae (2 terminals; Symphytognathidae was not represented); none of the symphytognathoid families present were found to be sister taxa. We look forward to further studies with greatly expanded DNA sequence coverage and the taxon sample necessary to address longstanding symphytognathoid questions. Lopardo and Hormiga (2015) noted that the placement of the several mysmenid species and genera described by Miller et al. (2009) had yet to be tested phylogenetically; here we have begun to rectify this. Both parsimony and Bayesian analyses found the Chinese Maymena species formed a clade sister to the remaining Maymena. This suggests that placing the Chinese species in Maymena was defensible. However, multi- ple lines of available evidence seem sufficient to justify the creation of a new genus to accommodate the Chinese species. The two Chinese Maymena are the only members of the genus known from beyond the western hemisphere. Although the Chinese Maymena resemble and share several characters with those from the west (e.g., aspects of the genital morphology, trichobothria on the male palpal tibia, troglophily, web architecture), they also express distinguishing features (e.g., elongate male palpal tibia and patella, relatively long and setose epigynal scape). Bayesian branch lengths (Fig. 2) and uncorrected pairwise distances based on our alignment (Fig. 3)  Etymology. Formed from Yama, the figure in Chinese mythology who oversees the realm of the dead, and -neta (-νήτης), an element in several spider names conventionally taken to mean 'spinner' (Cameron 2005). The gender is masculine.
Diagnosis. Distinguished from other mysmenid genera except Maymena by the presence of a modified spatulate seta on the PLS (Miller et al. 2009: fig. 57D, F;Lo-pardo and Hormiga 2015: fig. 11G, H), the proximal position of the male metatarsus I clasping spur (more proximal in Maymena than Yamaneta; Miller et al. 2009: fig. 53A; Lopardo and Hormiga 2015: fig. 16G), the shape of the apical part of the cymbium, which appears to form a functional conductor (Miller et al. 2009: fig. 55A;Lopardo and Hormiga 2015: fig. 10D, G) that interacts with the embolus. The presence of trichobothria on the male palpal tibia is a rare character in Mysmenidae, occurring in such genera as Maymena, Yamaneta, Trogloneta, Mysmenopsis Simon 1898 and Isela Griswold, 1985. Distinguished from Maymena by the elongate male palpal tibia and patella, long and setose epigynal scape, by the absence of a modified seta with a long row of branches near the major ampullate gland spigot on the anterior lateral spinnerets (Miller et al. 2009: fig. 57B), and by the clustered arrangement of male epiandrous fusules (Miller et al. 2009: fig. 56D; dispersed in Maymena: Lopardo and Hormiga 2015: figs 12B, 16A). The type species Maymena mayana (Chamberlin & Ivie, 1938) has been described as having a small rounded scape (Gertsch 1960), although this is a glabrous structure (setose in Yamaneta), and M. mayana is coded as absent for a scape in phylogenetic data matrices (Lopardo et al. 2011, Lopardo andHormiga 2015: character 60). There are also similarities in the female reproductive path shared between M. mayana and Yamaneta, such as the fertilization ducts arising from the copulatory ducts rather than the spermathecae (Lopardo and Hormiga 2015: fig. 128B); internal female reproductive structures and spinneret spigot morphology have been documented for only a few Maymena species.
Distribution. Gaoligong Mountains, Yunnan, China. Affinity with Maymena. Lopardo and Hormiga (2015) highlighted several key morphological characteristics of Maymena and discussed their status as putative synapomorphies and utility as diagnostic characters. These observations were based on a selection of western species, but many of the characteristics discussed are consistent with Yamaneta. The modified spatulate seta on the PLS (Lopardo and Hormiga 2015: fig. 11G, H) is present in Yamaneta paquini (Miller et al. 2009: fig. 57D, F [indicated by arrow]). The variable shape of the aciniform gland spigots on both pairs of posterior spinnerets and in both sexes (Lopardo and Hormiga 2015: fig. 11F-H, 13F, G) is visible in Yamaneta paquini (Miller et al. 2009: fig. 57C-F). However, the modified seta with a long row of branches near the major ampullate gland spigot on the anterior lateral spinnerets (Lopardo and Hormiga 2015: fig.  11E, 13C, 16B) is not visible in Yamaneta paquini (Miller et al. 2009: fig. 57B). The presence of macrosetae on the femora, tibiae, and metatarsi, especially on the anterior legs (Lopardo and Hormiga 2015: figs 140M, 141C), is shared by Maymena, Yamaneta (Miller et al. 2009: fig. 53A, B, see also text), and the kleptoparasitic clade Mysmenopsinae. A roughly cylindrical palpal tibia (i.e., distal width less than two times proximal width; Lopardo and Hormiga 2015: fig. 10A) is difficult to discern in Yamaneta, which have the palpal tibia elongated and modified in shape compared to Maymena species (Figs 5, 8). Like Maymena, males of Y. paquini and Y. kehen have a femoral spot on femur I, a clasping spur in a proximal position on male metatarsus I (Miller et al. 2009: fig. 53A;Lopardo and Hormiga 2015: fig. 16G), and lack a tegular conductor. Also consistent across Maymena and Yamaneta is the presence of macrosetae on the female palpal tarsus (Miller et al. 2009: fig. 53B;Lopardo and Hormiga 2015: figs 13A, 15A). Unlike the Maymena species studied by Lopardo and Hormiga (2015: fig. 10H), Yamaneta species do not appear to have a deeply grooved embolic rim. As in the Maymena species studied by Lopardo and Hormiga (2015), described as having the primary cymbial conductor apically bent over the ventral side (Lopardo and Hormiga 2015: figs 10D, G, 14D), the cymbium of Yamaneta species has a complex, almost helical shape, with the embolus and cymbium interacting distally (Figs 5, 8;Miller et al. 2009: figs 54, 55A, B). Unlike Maymena (Lopardo and Hormiga 2015: figs 12B, 16A), where the epiandrous fusules are arranged in a dispersed row, those of Y. paquini are arranged in a few rough clusters (Miller et al. 2009: fig. 56D). Lopardo and Hormiga (2015: 778) report that the respiratory system of Maymena distinguishes it from other mysmenids, but this has not been investigated for Yamaneta. Diagnosis. Yamaneta kehen can be distinguished from its congener Y. paquini by having only a single proximal-dorsal trichobothrium (TTr) and a single long distalventral setae (TiS) on the male palpal tibia, but 2 of each in Y. paquini (Fig. 5A, B vs. Fig. 8A, B); and by the form of the epigyne in the female, which features a basally wider and shorter scape (Fig. 6C vs. Fig. 9C), a distinctly notched lateral margin of the scape (Fig. 6D vs. Fig. 9D), and a more strongly curved profile of the dorsal surface of the scape (Fig. 6B vs. Fig. 9B). Lateral margins of scape proximal to notches are nearly parallel in Y. paquini (Fig. 9C, D), converging in Y. kehen (Fig. 6C, D).
Distribution. Known from a single cave in Yunnan, China. Natural history and web architecture. This species lives in the dark zone of the cave. They build a web typical of Maymena (e.g., Eberhard 1986, Lopardo andHor-miga 2015: fig. 147D, E). Aerial lines extend upwards from web radii and hub and are attached to frame lines or the substrate above. The web is under tension and the hub is lifted. The catching spiral is dense and nearly horizontal. Above the catching spiral is an irregular network of horizontal and angled lines under tension (Fig. 10A). The spider usually hangs in the irregular area above the catching spiral.
Female. Somatic characters see Fig. 7D Vulva (Fig. 9A-D): Scape relatively narrow basally (Fig. 9A, C), with gently notched lateral margins (Fig. 9D) and moderately curved dorsal profile (Fig. 9B). Lateral margins of scape proximal to notches are nearly parallel (Fig. 9C, D). Distribution. Known from a single cave in Yunnan, China. Natural history and web architecture. This species lives in the dark zone of the cave. The web documented in Fig. 10B resembles that of Y. kehen, except that it is under less tension, causing lines in the superstructure to bow. The occupant of this web was observed below the catching spiral.    (Miller, Griswold & Yin, 2009) comb. nov. from Fugong Co., Lishadi, "a nameless cave", female B Yamaneta paquini (Miller, Griswold & Yin, 2009)