Fifty-seven specimens were included in this study (Suppl. material 1), including 40 Travunioidea, 14 non-travunioid Laniatores (eight Triaenonychidae, two Synthetonychiidae, and four Grassatores), and single representatives from each of the other three harvestmen suborders. Forty-nine samples were newly sequenced for this study; raw reads for other taxa are from Starrett et al. (2017). We included two samples for most travunioid genera (in most cases from two different species), except for Trojanella and Travunia Absolon, 1920, both of which were represented by a single specimen. We were unable to include three European travuniid genera (Arbasus Roewer, 1935, Buemarinoa Roewer, 1956, and Dinaria Roewer, 1935) as these are extremely rare and difficult to obtain cave-dwelling taxa.

Genomic DNA was extracted from whole bodies using the Qiagen DNeasy Blood and Tissue Kit (Qiagen, Valencia, CA). For several larger specimens (body size greater than 3–4 mm) only legs, pedipalps, and chelicerae were used in extractions. Extractions were quantified using a Qubit Fluorometer (Life Technologies, Inc.) Broad Range kit, and quality was assessed via gel electrophoresis on a 0.8% agarose gel. Up to 500 ng of genomic DNA was used in sonication procedures, using a Bioruptor for 7 cycles at 30 seconds on and 90 seconds off, or a Covaris M220 Ultrasonicator for 60 seconds with a Peak Incidence Power of 50, Duty Factor of 10%, and 200 cycles per burst. Samples were run out on a gel to verify sonication success.

Library preparation followed the general protocol of Starrett et al. (2017) and the UCE website (ultraconserved.org), with some modifications. Briefly, libraries were prepared using the KAPA Hyper Prep Kit (Kapa Biosystems), using up to 250 ng DNA (i.e., half reaction of manufacturer’s protocol) as starting material. Ampure XP beads (Beckman Coulter) were used for all cleanup steps. For samples containing <250 ng total, all DNA was used in library preparation. After end-repair and A-tailing, 5 μM universal adapter stubs (University of Georgia, EHS DNA Lab) were ligated onto libraries. Libraries were then amplified in a 50 μl reaction, which consisted of 15 μl adapter-ligated DNA, 1X HiFi HotStart ReadyMix, and 0.5 μM of each Illumina TruSeq dual-indexed primer (i5 and i7) with modified 8-bp indexes (Glenn et al. 2016). Amplification conditions were 98 °C for 45 s, then 16 cycles of 98 °C for 15 s, 60 °C for 30 s, and 72 °C for 60 s, followed by a final extension of 72 °C for 60 s. Samples were quantified to ensure amplification success. Equimolar amounts of libraries were combined into 1000 ng pools consisting of eight samples each (125 ng per sample).

Target enrichment was performed on pooled libraries using the MYbaits Arachnida 1.1K version 1 kit (Arbor Biosciences) following the Target Enrichment of Illumina Libraries v. 1.5 protocol (http://ultraconserved.org/#protocols). Hybridization was conducted at 65 °C for 24 hours, then libraries were bound to streptavidin beads (Dynabeads MyOne C1, Invitrogen) and washed. Following hybridization, pools were amplified in a 50 μl reaction consisting of 15 μl of hybridized pools, 1X Kapa HiFi HotStart ReadyMix, 0.25 μM of each of TruSeq forward and reverse primers, and 5 μl dH20. Amplification conditions consisted of 98 °C for 45 s, then 16 cycles of 98 °C for 15 s, 60 °C for 30 s, and 72 °C for 60 s, followed by a final extension of 72 °C for 5 minutes. Following an additional cleanup, libraries were quantified using a Qubit fluorometer. Molarity was determined with an Agilent 2100 Bioanalyzer and equimolar mixes were prepared for sequencing on an Illumina NextSeq (University of California, Riverside Institute for Integrative Genome Biology) with 150 bp PE reads.

Raw demultiplexed reads were processed entirely in the Phyluce pipeline (Faircloth 2015). Quality control and adapter removal were conducted with the Illumiprocessor wrapper (Faircloth 2013). Assemblies were created with Trinity r2013-02-25 (Grabherr et al. 2011) and Velvet 1.21 at default settings. For each sample, the fasta files from both assembly methods were combined into a single file. The combined assembly contigs were matched to probes using minimum coverage and minimum identity values of 65 with a modified version of the “phyluce_assembly_match_contigs_to_probes” script to allow multiple hits to a single probe. UCE loci were aligned with MAFFT (Katoh and Standley 2013) and trimmed with Gblocks (Castresana 2000, Talavera and Castresana 2007) with custom blocks settings (b1 = 0.5, b2 = 0.5, b3 = 6, b4 = 6) implemented in the Phyluce pipeline. Individual UCE alignments were imported into Geneious 11.0.4 (http://www.geneious.com, Kearse et al. 2012) and manually inspected for obvious alignment errors and to remove any potential non-homologous sequences. In this case, ingroup sequences more divergent than outgroup taxa based on pairwise genetic distance calculated in Geneious were flagged and removed as potential non-homologs. Two datasets were produced including loci at two different taxon coverage thresholds (50% and 70%). All bioinformatic analyses were performed on a late 2015 iMac, except for contig assembly, which was run on the University of California, Riverside Institute for Integrative Genome Biology Linux cluster.

Analyses of both 50% and 70% datasets included concatenated maximum likelihood, concatenated Bayesian, and coalescent-based analyses, while partitioned maximum likelihood analyses were run only on the 70% dataset with partitions and models determined by PartitionFinder v1.1.1 (Lanfear et al. 2012). Maximum likelihood trees were estimated with RAxML v8.2 (Stamatakis 2014) using the rapid bootstrap algorithm, 500 bootstrap replicates, and the GTRGAMMA model. Bayesian analyses were conducted using the BEAST v2.4 package (Bouckaert et al. 2014), run for 100 million generations, logging every 1000 generations, with 10% burnin. To assess convergence, Tracer (Rambaut et al. 2007) was used to check for ESS values >200 and examine stationarity of parameters. Two separate analyses were run to check for convergence between runs. All RAxML and BEAST analyses were run through the CIPRES PORTAL (Miller et al. 2010). To incorporate coalescent approaches, ASTRAL-II (Mirarab et al. 2014, Mirarab and Warnow 2015) was used with individual gene trees estimated in RAxML, and SVDQuartets (Chifman and Kubatko 2014, 2015) was run through PAUP* 4.0a159 (Swofford 2003), with 100 bootstrap replicates.

Sequencing results and data matrix statistics are presented in Suppl. material 1. Raw sequence reads are available in the NCBI Short Read Archive Accession no. SRP142540 (BioProject ID PRJNA451420). Untrimmed contigs, all trimmed individual locus alignments (pre- and post-manual editing), and trimmed concatenated matrices are available from the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.tj48997. Combining assemblies resulted in higher numbers of resulting UCE loci relative to using only a single assembly method, suggesting assembly-specific contigs as found in Hedin et al. (2018). The average number of UCE loci sequenced was 748 across all travunioid samples and 675 across all samples included in this study. The 70% taxon coverage matrix included 317 loci (272 average loci per sample, total length of 83,990 bp, 253.86 bp average locus length) and the 50% taxon coverage matrix included 677 loci (488 average loci per sample, total length of 165,096 bp, 264.95 average locus length).

All analyses, with the exception of the 70% concatenated BEAST analysis, recover Travunioidea as sister group to all other Laniatores lineages (Figure 3 and Suppl. material 2: Figure 1), a relationship not recovered in previous molecular phylogenetic studies. However, based on morphological data, this hypothesis has been put forth by Kury (2015) who created the name Tricospilata Kury, 2015 for Triaenonychoidea + Grassatores (the sister group to Travunioidea). The 70% concatenated BEAST analysis recovers Travunioidea + Triaenonychoidea Sørensen, 1886 (= Synthetonychiidae + Triaenonychidae), previously called Insidiatores Loman, 1900, sister group to the Grassatores. The only other phylogenetic analysis resulting in Insidiatores as sister group to Grassatores is the transcriptome-based study of Fernández et al. (2016). Synthetonychiidae is recovered as sister group to Triaenonychidae with full support in all RAxML and SVDQuartets analyses and the concatenated BEAST analysis, but is sister group to all non-travunioid Laniatores (Triaenonychidae + Grassatores) in the ASTRAL analyses. It is apparent that denser taxonomic sampling and further phylogenomic datasets will be required to resolve the base of Laniatores.

Figure 3. 

Phylogenetic relationships among major laniatorean lineages. Lower phylogenies correspond to results presented in this study. Nodes are fully supported (100% bootstrap or 1.0 posterior probability), unless indicated otherwise. Numbers in lower left phylogeny correspond to support values from 50% RAxML concatenated (top), and from 70% RAxML partitioned (bottom) analyses. Numbers in ASTRAL phylogeny based on 50% (top) and 70% (bottom) matrices.

Travunioidea is monophyletic and fully supported across all analyses (Figure 3). Within Travunioidea, no families (and all but one subfamily) as currently defined in Kury et al. (2014) are monophyletic in any analyses (Figs 4, 5). A highly supported sister relationship between Travunia and Trojanella is recovered, and this group is sister to all remaining travunioids in all analyses. The western North American genera Cryptomaster Briggs, 1969 and Speleomaster Briggs, 1969 are recovered as sister taxa, and although the placement of Cryptomaster + Speleomaster is inconsistent across analyses, they are never sister group to or included within the Travuniidae or CladonychiinaesensuKury et al. (2014). In all analyses, the traditional Briggsinae (Briggsus Özdikmen & Demir, 2008 + Isolachus Briggs, 1971) are recovered within a largely travuniid clade, and always the sister group to Speleonychia Briggs, 1974. Eastern North American Erebomaster Briggs, 1969 + Theromaster Briggs, 1969 (traditional Cladonychiinae) is the sister group to the Briggsinae + Peltonychia clavigera (Simon, 1879) in all analyses. The two samples of the European genus Peltonychia Roewer, 1935 included in this study are never sister, P. leprieurii (Lucas, 1861) is found as the sister group to Holoscotolemon Roewer, 1915 while P. clavigera is the sister group to Briggsinae + Speleonychia. The relationships among these lineages are generally weakly supported at multiple nodes. All analyses recover a clade comprised of the former “northern triaenonychids”, currently in the families Paranonychidae and Nippononychidae, although neither family as currently defined is monophyletic. The Californian endemic genus Zuma Goodnight & Goodnight, 1942 is recovered within a clade comprised of Japanese taxa except Yuria Suzuki, 1964 and Paranonychus fuscus (Suzuki, 1976). The relationships within this lineage are all highly supported and identical across all analyses. The placement of the Japanese genus Yuria differs considerably across analyses and is recovered as the sister group to either the traditional travuniids or to a clade comprising Cryptomaster + Speleomaster and the Paranonychidae + Nippononychidae. Most importantly, Yuria is never recovered with the other Japanese nippononychids.

Figure 4. 

Phylogenomic relationships among travunioid genera. Left: RAxML and 50% BEAST concatenated topologies, with bootstrap support from the partitioned analysis. All nodes in the BEAST topology have posterior probability of 1.0. Abbreviations indicate placement in classification at the time of Kury et al. (2014): Tt = Travuniidae, Travuniinae; Tc = Travuniidae, Cladonychiinae; Tb = Travuniidae, Briggsinae; Pp = Paranonychidae, Paranonychinae; Ps = Paranonychidae, Sclerobuninae; N = Nippononychidae. Right: 70% BEAST concatenated. Nodes are fully supported (100% bootstrap or 1.0 posterior probability), unless indicated.

Figure 5. 

Phylogenomic relationships among travunioid genera. Left: 70% SVDQuartets. Right: 70% ASTRAL. Nodes are fully supported (100% bootstrap), unless indicated.

Our approach to establish a stable classification involved identifying the largest group of terminal taxa that are always monophyletic and always highly supported across all analyses. We discovered four multi-genus clades consistent across all analyses (Figs 4, 5), treated here as families:

1) A clade containing Travunia + Trojanella. Because of the inclusion of Travunia, this clade retains the name Travuniidae.

2) A clade containing the majority of travuniid genera sensuKury et al. (2014): Peltonychia, Holoscotolemon, Erebomaster, Theromaster, Speleonychia, Briggsus, and Isolachus. This clade will use the re-elevated and re-circumscribed familial name Cladonychiidae (see below).

3) A clade containing all genera currently included in the Paranonychidae and Nippononychidae of Kury et al. (2014). This clade retains the familial name Paranonychidae.

4) A clade consisting of the two former Cladonychiinae genera Cryptomaster and Speleomaster, endemic to the Pacific Northwest of North America, described below as the new family Cryptomastridae fam. n. (Figure 6).

Figure 6. 

Morphology of Cryptomastridae. A) male Cryptomaster, arrow denotes sexually dimorphic swelling diagnostic of Cryptomastridae; B) Speleomaster, habitus; C) Scanning electron micrograph of tibial swelling, from Starrett et al. (2016); D) Cryptomastrid distribution.

The genus Yuria is considered incertae sedis given its uncertain phylogenetic placement (see Discussion). The new phylogenomics-based classification, used hereafter, is summarized in Table 2.

D1 diverticulum 1;

OD2 opisthosomal diverticula 2;

OD3 opisthosomal diverticula 3.

Terminology and homology for penis/glans structure follows Martens (1986).

Included genera and species. Yuria (Figure 1J). A monotypic genus endemic to Japan, Yuria pulcra Suzuki, 1964 includes two subspecies distributed throughout southern Honshu, Shikoku, and Kyushu: Y. p. pulcra and Y. p. briggsi Suzuki, 1975.

Remarks. When Yuria pulcra was first described it was placed in Travuniidae because the tarsal claw is a peltonychium (Suzuki 1964, 1975a). Phylogenetic analysis of morphological data placed Yuria as the sister group to Nippononychus (Paranonychidae) with two synapomorphies (Mendes 2009). Kury et al. (2014) later transferred this genus to the Nippononychidae, which contained most of the other Japanese travunioids. Placement and support for Yuria varies depending on analysis (Figs 4, 5). Morphology complicates matters further, as Yuria possesses a free ninth tergite and lateral sclerites, plesiomorphic characters that are potentially neotenic and shared with the Briggsinae. The penis morphology of Yuria is also relatively unique within Travunioidea, possessing a dorsal plate with fused stylus (Figure 7 and Suppl. material 2: Figure 4).

Travunioidea includes 80 nominal taxa (species/subspecies), four families, and one unplaced genus. Traditionally within Travunioidea the subfamilial rank has been used to further subdivide taxa, and the composition of subfamilies has changed across classification schemes (Table 1). Here, we refrain from using the subfamilial rank for three reasons: 1) the confusing taxonomic history of these lineages, specifically with regards to relative rank and composition; 2) the poorly supported nodes in Cladonychiidae and non-monophyly of Peltonychia; 3) and the relatively sparse composition each subfamily would have (i.e., 4/6 subfamilies would contain only 1–2 genera). Based on phylogenomic analyses, the composition of all subfamilies would have changed again (Table 2). Relationships among several traditional Travuniidae genera are still uncertain, and the absolute stability of the familial rank will be dependent upon future incorporation of the unsampled European genera Arbasus and Buemarinoa.

The traditional Travuniidae have had an incredibly long and complex taxonomic history beginning with the description of the first travunioid in 1861. Kury and Mendes (2007) focus entirely on nomenclatural issues, resolving them at the familial and generic level, and in doing so note that “the [traditional] family Travuniidae constitutes one of the worst problems of the laniatorid taxonomy of the 20^th century”. Several others have discussed the diverse array of problems plaguing travunioid taxonomy (Novak 2005, Novak and Gruber 2000, Karaman 2005). These issues include, but are not limited to, description of two species in two different genera based on the same material, autosynonymy of a genus name, proposal of unavailable family and genus names, genus description without designation of type species after 1930, species descriptions based on juveniles, mistranslation of foreign languages, disregard for correct taxonomic changes, mismatched type localities, type localities accidentally and intentionally incorrectly named, inability to find further specimens from type localities, and destruction of type localities. Even though the European taxa have received significant attention from a taxonomic standpoint, a great deal of focused and devoted research including fieldwork, and modern morphological and phylogenetic analyses will be needed to fully resolve the taxonomic issues of this notoriously difficult clade.

The goal of this research was to provide a stable classification of Travunioidea at the familial level. This stability relies on incorporating potential future changes if unsampled taxa are included. We believe our familial level classification, disuse of subfamily rank, and leaving Yuria unplaced, minimizes future taxonomic changes. All familial name-holding genera are included. Only Arbasus and Buemarinoa are missing but given morphological similarities and the geographic distribution of these taxa, their inclusion in Cladonychiidae given future sampling seems likely. The stability of the familial level provided by this phylogenomics-based reclassification and the recovered distinction between Travuniidae and European Cladonychiidae can guide future efforts. The morphologically enigmatic Yuria remains phylogenetically elusive. A potential solution to the unreliable placement of Yuria is to create a monotypic family. However, we refrain from this until all genera can be included in phylogenomic analyses.

The tarsal claw – A type of modified tarsal claw termed a peltonychium united the “traditional Travuniidae”, a structure now known to be convergent in several unrelated cave-dwelling taxa (e.g., Peltonychia, Speleonychia, Trojanella). The morphological distinction between the typical trident-shaped tarsal claw (with variable number of side-branches) and a peltonychium is not entirely clear in some travunioids (e.g., Izunonychus, Metanippononychus), and the transition between forms is best documented in the triaenonychid genus Lomanella Pocock, 1902 (Hunt and Hickman 1993). However, all of the 18 travunioid species (not including subspecies) with a clear peltonychium are found in caves, 10 of which are either described to be troglobitic (cave-obligate) and/or show high levels of troglomorphy. An additional six species are described to have reduced pigmentation (relative to surface-only species), five of which are only reported from caves. Two remaining species are inconsistent with this pattern. First, Peltonychia leprieurii, from the Alps of northern Italy and Switzerland is recorded from both cave and surface habitats (considered a troglophile) and retains black pigment. Second, Y. pulcra pulcra and Y. p. briggsi from southern Japan show some reduced pigmentation, though they differ in their habitat: Y. p. pulcra is recorded from both cave and surface habitats, while Y. p. briggsi is known only from surface habitats under woody debris. It has been suggested that the peltonychium, and other plesiomorphic characters, are convergent through neoteny (Rambla 1980, Hunt and Hickman 1993, Karaman 2005). The hind tarsal claws of juvenile triaenonychids and travunioids have more side branches than adults (e.g., Hunt and Hickman 1993, Suzuki 1975b), and some juveniles additionally possess a pseudonychium (median tarsal claw) (Shultz and Pinto-da-Rocha 2007, Gnaspini 2007). As such, the peltonychium may be a retained adult form of this juvenile structure (Hunt and Hickman 1993).

The ninth tergite and lateral sclerites – The traditional Briggsinae (Briggsus + Isolachus) were hypothesized to be a relatively early diverging lineage within Travunioidea (Briggs 1971, Giribet and Kury 2007). This was due to the presence of a free ninth tergite, and lateral sclerites, a plesiomorphic condition found in the other suborders of harvestmen. Conversely, penis morphology suggested that this group is derived given the relatively simple penis structure (Martens 1986). Speleonychia also possess a distinct ninth tergite and lateral sclerites, now shown to be a shared condition with Briggsus + Isolachus. Rambla (1980) argued that these characters are neotenic, retained in adults from nymphal stages, and as such are derived. All phylogenomic analyses here support the derived nature of these characters, as they are recovered well within the Cladonychiidae in all analyses. However, not all travunioid taxa with a free ninth tergite and free lateral sclerites are restricted to this clade, as Yuria also possess both characters (Suzuki 1964, 1975a). While the placement of Yuria is uncertain, it is never recovered with or as the sister group to the traditional Briggsinae + Speleonychia. Outside of Travunioidea, the only other laniatorean taxa known to possess free lateral sclerites are in the genus Hickmanoxyomma Hunt, 1990, a largely cave-dwelling triaenonychid genus endemic to Tasmania (Hunt 1990). In this genus the presence of five pairs of free lateral sclerites (2–3 in travunioids) are a diagnostic character for the H. cavaticum species group containing four species, all recorded only from caves and some showing troglomorphy. Hunt (1990) suggested it might be correlated with the overall reduced sclerotization associated with troglomorphy, but also reiterated Rambla’s (1980) view of neoteny. No clear phylogenetic, evolutionary, or ecological pattern exists for these characters and their presence in multiple unrelated lineages suggests their plesiomorphic nature.

The midgut – Studies focusing on the digestive tract began in the 1920s, but the work of Dumitrescu (e.g., 1974, 1975, 1976) contributed most significantly to the phylogenetic utility of midgut morphology in Opiliones. Through examination of intestinal morphology Dumitrescu (1975, 1976) noticed the “northern triaenonychids” are more similar to the Travuniidae as they share a bipartate OD3, instead of the southern hemisphere triaenonychids (3-branched). As such, he placed the “northern triaenonychid” taxa into the family Paranonychidae Briggs, 1971. The name Paranonychidae was used by a few subsequent authors (e.g., Shear 1982, 1986, Ubick and Dunlop 2005), but from a classification standpoint, was not incorporated into the taxonomy as the taxa were left in the Triaenonychidae (e.g., Kury 2003a, Pinto-da-Rocha and Giribet 2007; Kury 2007b). Later, molecular phylogenetic studies also showed the “northern triaenonychids” being grouped with the travunioid families (Giribet et al. 2010, Hedin and Thomas 2010, Derkarabetian et al. 2010, Sharma and Giribet 2011), and the familial name Paranonychidae was finally included in Opiliones taxonomy by Kury (2013).

Our phylogenomic analyses allow for a reexamination of the intestinal morphology research of Dumitrescu (1975, 1976) (Suppl. material 2: Figure 2). First, the recovery of Travunioidea as the earliest diverging Laniatores lineage is reflected in the branching pattern of OD3. All Travunioidea possess an OD3 with two branches, a characteristic shared with the other harvestmen suborders, while the Synthetonychiidae, Triaenonychidae, and Grassatores possess three branches. This phylogenetic utility of OD3 was also noted in the morphological analyses of Mendes (2009), who recovered Travunioidea as the earliest diverging lineage of Laniatores. The intestinal complex also shows some phylogenetic value within Travunioidea; three of four families can be differentiated entirely based on intestinal morphology. Of the taxa examined by Dumitrescu (1975, 1976), the paranonychids possess a simple and circular D1 with no branches, while all cladonychiid genera possess a relatively complex D1 with 2–3 branches or is distinctly triangular in shape. Karaman (2005) also reports a D1 with two branches for Trojanella. Cryptomastridae possess a triangular D1, but the shorter OD2 and OD3 are more similar to those of the paranonychids. Samples of Briggsus and Speleonychia were included and look very similar to each other, particularly the D1, which is recorded to be 3-branched in Briggsus and triangular in Speleonychia (Dumitrescu 1976). Dumitrescu did not note the similarity, instead stating Speleonychia was most similar to Nippononychus japonicus (then Peltonychia japonica, placed in Travuniidae), perhaps subjectively limited by the classification system of the time. It is clear that there is phylogenetic utility in midgut morphology at higher taxonomic levels, and further descriptions are needed to confirm the consistency of diagnostic characteristics noted here.

The penis – Based on descriptions and drawings, the musculature and glans complexity can be used to diagnose and differentiate travunioid lineages recovered here (Figure 7) (Briggs 1971b, Suzuki 1975a, 1975b, Martens 1986, Karaman 2005, Novak 2005, Pinto-da-Rocha and Giribet 2007, Derkarabetian and Hedin 2014, Starrett et al. 2016). In Travunioidea, the glans is relatively simple and plate-like without dorsal, dorsolateral, or ventral plates in Trojanella, Travunia, Cryptomastridae, and Cladonychiidae, while in Yuria and Paranonychidae (except Paranonychus) the glans is relatively complex, and some possess a dorsolateral plate. The dorsal plate is absent from all travunioids, except Yuria, which possesses a dorsal plate that is fused to the stylus.

Given the consistent and highly supported relationships within Paranonychidae, diagnostic differences of paranonychid lineages can be seen in penis morphology (Figure 7; Suppl. material 2: Figure 4). Within Paranonychidae, the earliest-diverging genus Paranonychus possesses the simplest glans. Dorsolateral plates are present but reduced in two pairs of sister genera Sclerobunus + Metanonychus and Metanippononychus + Nippononychus (Briggs 1971b, Suzuki 1975b, Derkarabetian and Hedin 2014). Additionally, Sclerobunus + Metanonychus show a distinct setae-bearing process similar to a ventral plate (“sensillenträger” of Martens) that is fused to the base of the stylus in the Zuma + Japanese taxa. The Zuma + Izunonychus + Kainonychus clade possesses a modified stylus that is expanded distally (Briggs 1971b, Suzuki 1975b). As opposed to other Japanese paranonychids which have a stylus that is fused to the ventral plate, the stylus of both Izunonychus + Kainonychus is separated from the ventral plate (Suzuki 1975b).

The overall trend across Opiliones suborders is one of apparently increasing genitalic complexity. The earliest-diverging suborder Cyphophthalmi has a spermatopositor, the Dyspnoi and Eupnoi have a simple penis with relatively little modifications, and the derived Laniatores possess the most complex penes (Macías-Ordóñez et al. 2010). Within Laniatores the trend of increasing complexity is maintained, as Travunioidea with relatively simple glans morphology is possibly the most early-diverging Laniatores. Travunioidea and Triaenonychoidea use muscles for glans expansion (Shultz and Pinto-da-Rocha 2007, Pérez-González and Werneck 2018), a condition shared with Eupnoi and Dyspnoi. Relative to travunioids, the synthetonychiids and triaenonychids have slightly more complex glans structures with dorsolateral plates. Finally, in the most diverse laniatorean lineage Grassatores, muscles are absent, and the glans requires hydraulic pressure to expand (Shultz and Pinto-da-Rocha 2007). In Travunioidea, Martens (1986) noted a tendency towards simplification of the glans, and Karaman (2005) additionally noted a correlation where simplification of the glans structure is associated with reduction of penis musculature to basal portion of the truncus. Both Martens and Karaman argued that the simple glans of travunioids (i.e., traditional travuniids) is a derived condition, as all Triaenonychoidea have relatively complex glans with dorsolateral plates and a distinct process bearing setae, two characters lost in many travunioids. Most phylogenomic analyses conducted here recover Travunioidea as the most early-diverging Laniatores lineage suggesting a trend of simple to complex penis structure. It is interesting to note that the simplest glans and some taxa with relatively reduced musculature are found in cave-inhabiting taxa (e.g., Trojanella, Dinaria, Travunia), which may be a potential confounding factor in establishing a clear evolutionary trend. More detailed morphological examinations of the male and female genitalia using modern approaches (e.g., Pérez-González and Werneck 2018) will help elucidate evolutionary trends and provide more diagnostic characters for the families established here.

This phylogenomic study provides a more stable taxonomy for Travunioidea, which serves as a starting point for species-level phylogenomics and provides the phylogenetic context to explore evolutionary questions relating to character evolution, alpha taxonomy, and biogeography.

Morphological and chemical evolution – Many Travunioidea are cave-obligate taxa with species from 14 genera showing some degree of troglomorphy, possessing homoplastic morphological features that evolve as a response to cave life. Travunioidea can be an excellent system to study the repeated evolution of troglomorphy, as it has evolved at multiple taxonomic levels (e.g., within families, genera, species) with multiple independently derived taxa showing varying degrees of troglomorphy. For example, within the genus Sclerobunus, troglomorphy has evolved at least five times independently across multiple species and within single species and is time-correlated (Derkarabetian et al. 2010, Derkarabetian and Hedin 2016). A species-level phylogeny for all travunioids would allow for an accurate estimate of the number of independent evolutions of troglomorphy and allow for in-depth morphological analyses exploring the rate and timing of this potentially adaptive morphology, as well as providing the phylogenetic framework for comparative studies (e.g., gene expression).

Similarly, chemical evolution can be explored in this phylogenomic context. Harvestmen possess repugnatorial glands, which are used to store chemical cocktails that are secreted in defensive behavior. Chemical composition across lineages has been shown to have some phylogenetic value, particularly in Laniatores (Raspotnig 2012, Raspotnig et al. 2015). Similarly, a study focusing on Travunioidea and Triaenonychoidea has shown high levels of divergence in chemical composition between taxa formerly united under the traditional taxonomy (i.e., Cladonychiidae) (Shear et al. 2014). Given the phylogenetic context, detailed chemical analyses can be used to discover novel chemicals, compare pathways of identical chemicals found in independent lineages, and explore the evolution of biochemical pathways across taxa (e.g., Rodriguez et al. 2018).

Biogeography and alpha taxonomy – Cryophilic harvestmen are useful in biogeographic analyses because of restricted ecological constraints and extremely low vagility. Previous molecular phylogenetic studies on harvestmen with these biological characteristics have shown compelling biogeographic patterns (e.g., Boyer et al. 2007, Thomas and Hedin 2008, Giribet et al. 2011, Schönhofer et al. 2015, Boyer et al. 2015). The Travunioidea have a temperate Laurasian distribution with species found in eastern Asia, eastern and western North America, and central Europe, with notable absences from central Asia (Figure 2). Recent divergence dating analyses show an ancient origin of Travunioidea dating to >200 million years (Giribet et al. 2010, Sharma and Giribet 2011), although these time estimates should be revisited. Several travunioid lineages show trans-continental distributions. For example, members of the Cladonychiidae are distributed in Europe and eastern and western North America, while the Paranonychidae show a Beringian distribution with genera in western North America, Japan and South Korea. Many other harvestmen show similar broad distributions within the Holarctic including the Sironidae, Caddo, Leiobunum, and many Dyspnoi (Suzuki 1972b, Suzuki et al. 1977, Shultz and Regier 2009, Hedin et al. 2012a, Hedin et al. 2012b, Schönhofer et al. 2013).

The biological characteristics of essentially all travunioids are quite similar (e.g., dispersal-limited, restricted to cryophilic microhabitats in north temperate latitudes). Regional clades are relatively ancient, allowing ample time for the accumulation of species diversity. In addition, the presence of rare, completely blind, troglobitic species in several different geographic areas speaks to the ancient origin of the Travunioidea. It is likely that troglobitic species, especially those in central Europe, have an unknown diversity concealed by troglomorphy. Similarly, many ancient lineages can be found in the moist, coastal forests of the Pacific Northwest of North America (Briggsus, Metanonychus) that have likely persisted in refugia through climatic cycles. Additionally, all of these taxa consist of species and subspecies that are short-range endemics (Harvey 2002). Congeneric species syntopy is rare, probably because of ecological niche conservatism that prevents resource partitioning. This ecological niche conservatism likely plays an important role in speciation (see model of Wiens 2004). As such, there is a high potential for species discovery, and recent species-level studies focusing on the travunioid genera Sclerobunus and Cryptomaster have resulted in the description of new species (Derkarabetian and Hedin 2014, Starrett et al. 2016). Our ongoing studies of Travunioidea are continuing this trend: every genus currently being revised (Briggsus, Erebomaster, Theromaster, Metanonychus) shows evidence for new species.