Three new species in the harvestmen genus Acuclavella (Opiliones, Dyspnoi, Ischyropsalidoidea), including description of male Acuclavella quattuor Shear, 1986

Abstract In Shear’s (1986) cladistic analysis of the Ischyropsalidoidea, he described the new genus Acuclavella including four new species from the Pacific Northwest states of Washington and Idaho. Several of these species descriptions were based on very limited sample sizes. Our recent field work has increased by more than an order of magnitude both the number of specimens and known localities for Acuclavella. We use this new material to interpret species limits in Acuclavella using morphometric analyses and DNA sequence data from four gene regions. We sequence for the first time the protein-coding homolog of the Wnt2 gene for phylogenetic reconstruction in Opiliones. Our multi-locus phylogeny corroborates a sister relationship between Acuclavella and Ceratolasma, as hypothesized using morphology by Shear (1986). Within Acuclavella, morphometric clusters and reciprocal allelic monophyly allows recognition of three additional species: Acuclavella leonardi sp. n., Acuclavella sheari sp. n., and Acuclavella makah sp. n. This work also describes the previously unknown male of Acuclavella quattuor, from specimens collected at the type locality. Our research identifies a number of novel morphologies for Acuclavella, including females with four pairs of spines, individuals with three pairs of spines on scute areas I-III, and a population with two pairs of spines disjunct from Acuclavella quattuor, which was diagnosed with this spination character. We were unable to assign these populations to existing species, and conservatively do not yet recognize them as new. Intrageneric morphometrics and phylogenetic inference in Acuclavella were often concordant. However, we demonstrate that species delimitation signal would not be detected if only a single line of evidence were utilized.


Three new species in the harvestmen genus Acuclavella
(Opiliones, Dyspnoi, Ischyropsalidoidea), including description of male Acuclavella quattuor Shear, 1986 Introduction The genus Acuclavella was described in 1986 by Shear in his revision of the superfamily Ischyropsalidoidea (Opiliones, Dyspnoi).In this work, four new species in the genus were described (Shear 1986).These descriptions were based in large part on very small sample sizes: Acuclavella cosmetoides Shear, 1986, the generic type species, was described from a single male and female; A. shoshone Shear, 1986 was described from two males and one female; and A. quattuor Shear, 1986 was described from a single female.Each of these three species was known only from their respective type localities in northern Idaho, north of the Salmon River.The fourth species, A. merickeli Shear, 1986, was represented by thirteen individuals.
Twelve of these (four males and eight females) were from the type locality also in Idaho north of the Salmon River.The remaining individual was collected from the Cascade Mountains of western Washington, and due to morphological similarity, was considered by Shear to be conspecific with A. merickeli despite a geographic separation of over 500 kilometers.The biogeographic situation in Acuclavella, with several short-range endemic species from a small geographic area, coupled with an apparently widespread (but disjunct) species, clearly invites further investigation.The biogeographic barriers separating the Cascade Mountains from the Rocky Mountains of northern Idaho have promoted speciation in a variety of taxa (e.g., amphibians: Nielson et al. 2001, Steele et al. 2005; arachnids: Derkarabetian et al. 2010;insects: Barr 2011).In addition, recent surveys have uncovered populations of Acuclavella from the Olympic Peninsula of northwestern Washington State, and from south of the Salmon River in Idaho.Cascade Mountain and Olympic Peninsula ecoregions of western Washington have been shown to house distinct species of non-vagile organisms (e.g., Good and Wake 1992).In Idaho, a habitat corridor of Abies grandis, north-facing and high above the south side of the Salmon River, may house unique populations of organisms that survived in an unrecognized compartment of a structured Pleistocene glacial refugium (Brunsfeld and Sullivan 2006).
The goal of this paper is to use morphometrics and molecular phylogenetics to investigate the validity of the four species hypothesized by Shear (1986), to delimit potentially new species within Acuclavella, and to revise the genus as necessary.We hypothesize that molecular phylogenies will reveal consistent reciprocal monophyly conforming to Idaho versus Washington taxa.Furthermore, we expect that Acuclavella populations from the two ecoregions in western Washington will show evidence of molecular divergence.In Idaho, expectations are that a morphologically distinct new population discovered south of the Salmon River will show molecular phylogenetic divergence from the four northern species described by Shear (1986).

Taxon sampling
Most fieldwork was conducted in the summer of 2008, with additional adult specimens collected from May to September in 2006, 2007, and 2009 (Appendix I -Collection Locality Information).Acuclavella are crenophilic denizens of small, perennial water features such as headwater streams and seeps in the Tsuga heterophylla Zone and the coastal Picea sitchensis Zone of the Pacific Northwest (Franklin and Dyrness 1988).Appropriate habitat and microhabitat was targeted for specimen collection throughout northern Idaho and western Washington, and collections include all type localities.When encountered, at least one specimen from each locality was preserved for molecular analyses by placing specimens into 100% EtOH, and subsequently transferring specimens to a -80° C freezer (following Vink et al. 2005).

Molecular data collection
Genomic DNA was extracted from two legs per specimen using the Qiagen DNeasy kit, per manufacturer's protocol.Currently, few genes are available for resolving shallow phylogenetic relationships in Opiliones (reviewed in Hedin et al. 2010).In this study, PCR amplification targeted four gene regions, including mitochondrial protein-coding cytochrome oxidase I (COI), the 28S large subunit ribosomal RNA (28S), nuclear protein-coding Elongation Factor-1 alpha (EF-1α), and the nuclear protein-coding wingless (Wnt2).COI is a standard gene used for species-level delimitation in animals, and has been used extensively in opilionid phylogenetic studies (e.g.Boyer et al. 2005Boyer et al. , 2007;;Thomas and Hedin 2008;Hedin and Thomas 2010;Derkarabetian et al. 2010).28S rRNA data has been used successfully to elucidate deeper relationships in opilionid phylogenies (e.g.Boyer et al. 2007, Derkarabetian et al. 2010, Giribet et al. 2010, Schönhofer and Martens 2010, Sharma and Giribet 2011), but generally evolves too slowly to resolve shallow relationships (e.g.Hedin and Thomas 2010).EF-1α has been shown to have phylogenetic utility at two levels (Hedin et al. 2010); in the region amplified for Acuclavella, more slowly evolving exons bracket a relatively rapidly evolving intron.The Wnt2 gene region has been used to reconstruct spider phylogenies (e.g., Blackledge et al. 2009, Satler et al. 2011), but is used here for the first time in Opiliones.PCR primer and amplification conditions are available as a supplementary file (Appendix II -PCR Primer Information).Amplicons were visualized on agarose gels, purified via polyethylene glycol (PEG) precipitation or on Millipore plates, and Sanger sequenced at the San Diego State University Microchemical Core Facility (http://www.sci.sdsu.edu/dnacore/sdsu_dnacore.html) or at Macrogen USA.

Phylogenetic analyses
Bi-directional Sanger reads were assembled into contiguous sequences using Sequencher v4.5 (Gene Codes Corporation, MI).EF-1α haplotypes were reconstructed in PHASE (Stephens and Donnelly 2003) using an ingroup-only matrix with the following parameter settings: 100 iterations, thinning interval of 1, burn-in of 100, probability cut-off set to 0.70 (Harrigan et al. 2008).Haplotypes with nucleotides inferred with probability values < 0.70 were left ambiguous.The PHASE input file was converted from a FASTA alignment using SeqPHASE (Flot 2010).COI, EF-1α exon, and Wnt2 gene regions were unambiguously aligned using amino acid translations in MacClade v4.06 (Maddison and Maddison 2003).EF-1α intron and 28S rRNA data were aligned using MAFFT (Katoh et al. 2005), utilizing the Q-INS-i alignment algorithm strategy with parameters: BLOSUM62 alignment scoring matrix, 200PAM / K=2 scoring matrix value, an opening gap penalty of 1.53, and an offset value of 0.1.Regions of alignment uncertainty were removed from the 28S matrix with Gblocks (Castresana 2000) with the following parameter settings: minimum length of a block = 3, allowed gap position with half, minimum number of sequences used to identify a conserved position and a flanking position 11 and 17 respectively, and 8 as the maximum number of contiguous non-conserved positions.
Individual gene trees were reconstructed using maximum likelihood and Bayesian inference.Bayesian analyses were implemented using MrBayes v3.1.2(Ronquist and Huelsenbeck 2003) run on an XSEDE utilizing the CIPRES portal (Miller et al. 2010).Models of DNA sequence evolution used in Bayesian analyses were determined using jModelTest v0.1.1 (Posada 2008, Guindon andGascuel 2003).Likelihood scores were computed with three substitution schemes (24 models), unequal base frequencies, proportion of invariable sites, rate variation among sites, and used the BIONJ algorithm of Gascuel (1997).The goodness of fit of alternative models was determined using AIC (Akaike 1974) with CI set to 100%.Bayesian gene tree analyses were run for 5 X 10 6 generations sampling every 1000 trees, using best-fit models of molecular evolution.At this many generations, the average standard deviation of split frequencies between runs was < 0.01.The first 40% of the sampled trees were discarded as burn-in, and stationarity was confirmed visually using Tracer v2.4 (Rambaut and Drummond 2007).Maximum likelihood gene tree analyses were conducted using RAxML v7.2.8 (Stamatakis 2006), also on the CIPRES portal using the GTR + GAMMA model for tree inference and bootstrapping (Stamatakis et al. 2008).The EF-1α data set was partitioned by intron and exon, and the COI data was partitioned by codon position.The Wnt2 and 28S matrices were not partitioned.Outgroup sequences were used to root all gene trees.
All available DNA sequences, with the exception of apparently nuclearized copies of COI (see Results), were concatenated for phylogeny reconstruction.The concatenated matrix was analyzed using both maximum likelihood (RAxML) and Bayesian approaches (MrBayes v3.1.2) using a seven-partition strategy (EF-1α intron + exon, Wnt2, 28S, individual COI codon positions).The Bayesian analysis was run for 1 X

Morphometric analyses
Specimen measurements were taken using an Olympus SZX12 dissecting microscope with an ocular micrometer.Individuals with missing data (i.e., no Leg II) were excluded from analyses.Standard measurements (Acosta et al. 2007) of 16 characters were determined for 261 individuals (131 males, 133 females); the measurement data, including a figure showing our character measurements scheme, are available as a supplementary file (Appendix III).Total body length was not measured due to the variation in length allowed by free tergites and sternites in individuals that are gravid or have recently eaten; this measurement is known to vary with nutritional condition in Opiliones (Acosta and Machado 2007).
Principle components analyses (PCA) and discriminant function analyses (DFA) were carried out using SYSTAT 12 (Systat Software, Inc.).Though both PCA and DFA are multivariate analyses, DFA is a validation approach that deals specifically with the problem of separating predetermined groups.In taxonomy, DFAs are almost exclusively used to differentiate between morphologically similar species that are difficult to identify from single characteristics (see Klimov et al. 2004, Seifert 2003, Fisher 1936, Lubishew 1962).Multispecies delimitation via DFA, as done here, is becoming more common (see Schlick-Steiner et al. 2006, Schönhofer and Martens 2008, Gebiola et al. 2012).
In taxonomy, PCAs are regularly employed for species delimitation.This utilization has ranged from comparing the vector angles in a scatter plot of predefined species groups (Hamm 2010), to mathematical inference of groups without a priori assignment of samples to species (Ezard et al. 2010).Because a PCA has no intrinsic measure of group exclusivity, the parameters of group recovery in morphospace need to be defined.For purposes of this research we defined three PCA categories: a "recovered" group clustered together and did not overlap with samples from other groups; a "nearly recovered" group clustered together with minimal overlap with other groups (one or two individuals at the periphery of their respective group that are not enclosed within another); a "not recovered" group failed to cluster exclusively.For these analyses, three methods were employed to explore this morphospace: (1) hypothesized species were compared pairwise; (2) all combinations of principle components explaining a significant part of the variation in the data were explored; and (3) a nested approach where if clusters of hypothesized species were recovered by initial analyses, further analyses were conducted on those clusters.An in-depth discussion of morphometric methods is available as a supplementary file (Appendix IV).
Due to sexual dimorphism, male and female specimens were analyzed separately for all morphometric analyses.Since the data used to conduct PCAs were measured in the same units (mm), analyses were conducted on both correlation and covariance matrices.PCAs using covariance matrices tend to be dominated by characters showing the most variability (Jolliffe 2002).In order to reduce possible excessive influence on the principle components, variables were coded to have a mean of zero and variance of one; analyses on this data are based on a correlation matrix.Analyses run on covariance matrices assume a multivariate normal distribution.This is often violated in taxonomic studies where data consists of multiple species of varying morphological distinctiveness; outliers could represent members of an under-sampled group.All unusual measurements were confirmed by repeated measurement, then included in analyses under the assumption they do not excessively affect interpretation of the data.

Species delimitation
We initiated analyses with seven a priori hypothesized species, with species-level distinctions based on geographic criteria or on preliminary specimen sorting (i.e., not morphometrics).Geographic criteria were based on the results of work on amphibians endemic to the disjunct mesic forests of the Pacific Northwest (Nielson et al. 2001), or the physiogeographic regions of western Washington (Good and Wake 1992).Areas of expected endemism included western Washington versus northern Idaho, as well as between Cascade Mountain and Olympic Peninsula regions in Washington, areas separated by the low valley of the Puget Trough.In Idaho, a morphologically diminutive population was discovered south of the Salmon River; this group was also tested as a species.
North of the Salmon River in Idaho, A. merickeli and A. quattuor were distinguished using the diagnostic features outlined by Shear (1986).Due to a gradation of intermediate morphologies spanning the diagnostic characters (Shear 1986) for A. cosmetoides and A. shoshone (Appendix V: males Figure 1, females Figure 2), coupled with the observation that the described morphologies of these species more or less occur only at the type localities (Appendix V: males, Figure 3; females, Figure 4), an initial attempt to distinguish these two species was abandoned.Since A. cosmetoides is the type species for the genus, all morphotypes north of the Middle Fork Clearwater were preliminarily identified as A. cosmetoides.A newly discovered population in northern Idaho matched the diagnostic morphology of A. quattuor of having pairs of spines on scute areas I and II.However, the new population is distributed between the Selway and Lochsa rivers, whereas A. quattuor occurs between the South Fork Clearwater and Salmon Rivers.This population was tested as a new species with the working name Acuclavella cf.quattuor.

Cybertaxonomy
A cybertaxonomic approach was undertaken for enhanced dissemination of this work (e.g., Miller et al. 2010).This work is available as an open access PDF (doi: 10.3897/ zookeys.311.2920).Molecular sequence data has been uploaded to GenBank (http:// www.ncbi.nlm.nih.gov/Entrez;see Table 1 for accession numbers).Some species of Acuclavella are difficult to diagnose using morphological characters alone.An effort has been made to facilitate DNA barcoding by uploading COI sequences to the Barcode of Life Data Systems (BOLD) v. 2.5 (http://boldsystems.org).Aligned matrices and all phylogenetic trees have been deposited in the Dryad Digital Depository (doi: 10.5061/dryad.16737).Anatomical images included in this publication and many additional images have been deposited in MorphBank (http://www.morphbank.net/?id=822371).A Keyhole Markup Language file (KML) for interactive viewing of species distributions and collection information in Google Earth (http:// earth.google.com) is available as a supplementary file (Appendix VI).All nomenclatural acts have been registered with ZooBank (http://www.zoobank.org).New species described in this study are provided to Encyclopedia of Life (http://www.eol.org) where interconnection of information is linked back to this open access, peerreviewed work.

Data resources
The data underpinning the analyses reported in this paper are deposited in the Dryad Data Repository at doi: 10.5061/dryad.16737.

Taxon sampling
Fieldwork resulted in the collection of 272 Acuclavella specimens from 61 localities.Populations were sampled from throughout northern Idaho, as well as in the Cascade Mountain and Olympic Peninsula physiogeographic regions of western Washington (Figure 1).The first state records for Montana were secured from the Bitterroot Mountains.Specimens were collected from the four type localities (Shear 1986), including male specimens from the type locality of A. quattuor, which Shear originally described from a single female.The vast majority of specimens were collected from beneath moist woody debris near small perennial water features.
Although the overall geographic distribution of Acuclavella is now reasonably wellknown, additional collecting efforts in the following areas may prove fruitful: south of known localities in Idaho, east into more inland areas of Montana, the Coast Range of Oregon, and the Willapa Hills of Washington.For example, most animal taxa that include populations in the Olympic Peninsula and Cascade Mountains are also found in the Willapa Hills (e.g.Rhyacotriton, Good and Wake 1992; Octoglena anura, Shelley et al. 2010).

DNA sequence characteristics
Mitochondrial COI DNA sequences for a geographical subset of Acuclavella specimens were inconsistent with expectations of protein-coding gene evolution.These COI se-quence reads showed ambiguous nucleotides at individual sites, a large number of mutations resulting in replacement amino acid substitutions, as well as insertions and deletions.Some deletions resulted in frame-shift mutations resulting in stop codons.These patterns of variation suggest that nuclearized copies of the mitochondrial COI gene (NUMTs; White et al. 2008, Baldo et al. 2011) exist in Acuclavella.
Sequences showing evidence for nuclearization were discarded, resulting in the removal of all sequences for individuals collected north of the Middle Fork Clearwater River, including all A. cosmetoides and A. shoshone (sensu stricto).In addition to apparent authentic COI sequences, we have multiple nuclear markers providing a similar history for three new species, so our taxonomic conclusions should not be compromised.The final COI matrix included 18 ingroup sequences (L = 1224 basepairs; Table 1).The first, second, and third codon positions included 34, 9, and 282 parsimony-informative characters in the ingroup.GenBank numbers for all mitochondrial gene sequences can be found in Table 1.
Gblocks removal of ambiguous sites in the 28S matrix resulted in a reduction of alignment length from 1211 to 1173 positions.The aligned matrix included 26 parsimony informative characters for 14 ingroup sequences.For the EF-1α gene, eight of 22 Acuclavella sequences showed signs of heterozygosity.Three of these individuals were ambiguous at a single base pair, two were ambiguous at two sites, and remaining individuals were ambiguous at 3, 4, or 5 sites respectively.All haplotypes were reconstructed using PHASE.The resulting aligned matrix consisted of 690 bp of exon data, and an 80 bp intron with gaps.The EF-1α exon contained 30 parsimony informative characters in the ingroup.The intron sequences did not have outgroup representatives; there were 18 parsimony informative characters in the intron.The Wnt2 gene matrix (L = 370) included 7 ingroup sequences, with only 4 parsimony informative characters.GenBank numbers for all nuclear gene sequences can be found in Table 1.

Phylogenetic analyses
Models of evolution inferred from jModelTest are summarized in Table 2.These partition-specific models were used in a Bayesian analysis of the concatenated matrix.Figure 2 shows results from the Bayesian analysis and associated posterior probabilities; the phylogeny resulting from the RAxML analysis is available in a supplementary file (Appendix VII).The concatenated Bayesian and ML phylograms are congruent in their strong support for a monophyletic Acuclavella.Ceratolasma is recovered as sister to Acuclavella with strong support, confirming the hypothesis of Shear (1986).Deeper phylogenetic relationships within Ischyropsalidoidea are beyond the scope of this research, and no emendations to the current taxonomy are proposed, pending collection of additional data.
Within Acuclavella, samples from Washington and Idaho are reciprocally monophyletic in both ML and Bayesian analyses, with regional clades strongly-supported and subtended by relatively long branches (Figure 2, Appendix VIII).Within Washington, both analyses recover samples from the Olympic Peninsula and southern Cascade Mountains as monophyletic lineages separated by long branches with high support.Both ML and Bayesian analyses also strongly support A. sheari as monophyletic and sister to all other Idaho samples.Acuclavella cf.quattuor was recovered by both analy-   Idaho samples.The COI gene tree shows significant support for a monophyletic A. sheari, which is sister to all remaining Idaho sequences.This species is recovered as monophyletic but without support by EF-1α and 28S gene trees.The Wnt2 gene tree (Figure 3B) supports reciprocal monophyly of samples from Washington and Idaho, but lacks phylogenetic signal within Idaho.The lack of structure for shallow evolutionary events within Acuclavella, coupled with high support for most nodes within Ischyropsalidoidea above the generic level, points to a deeper phylogenetic utility for Wnt2 in Opiliones.1).RAxML concatenated phylogeny (Appendix VII, Figure 2) and gene trees (Appendix VII, Figures 3-5) are also available.

Morphometric analyses
DFA and PCA morphometric analyses of male and female data sets most frequently recovered Acuclavella sheari, followed by A. leonardi and A. makah.Undescribed morphologies from northern Idaho clustered with samples of A. cosmetoides and A. shoshone (sensu stricto) of Shear, 1986.Acuclavella merickeli was also frequently recovered in these analyses.The two populations with two pairs of scutal spines on scute areas I and II (A.quattuor and A. cf.quattuor) were frequently recovered as distinct from other hypothesized species, but were not recovered as morphometrically distinct from each other.
Since the group frequencies of hypothesized species varied greatly (Table 3), jackknifed classification of some data sets reduced the number of individuals used to define a group to just two or three individuals.An assumption is made here that variation seen within hypothesized species is normal in terms of representing the cluster.Male Acuclavella are robustly discriminated by DFA analysis, with nearly all species correctly classified with 100% accuracy.An exception is A. cf.quattuor, with two specimens classified as an A. quattuor (classification error rate of 0.09).Results of DFAs are available Results of DFAs are available in Appendix IV.The analysis was repeated with a jackknife classification resulting in successful discrimination of 94% of male individuals.Considering the high variation in group frequencies of hypothesized species (Table 3), analyses were rerun allowing prior probabilities of group membership.These analyses resulted in very similar groupings (not shown).
Groupings based on female specimens were not recovered as frequently as malebased groups.Females of A. leonardi, A. merickeli, and A. sheari were correctly discriminated in the classification matrix.There was support for female samples of A. makah and A. cosmetoides, with correct classification 93% and 97% of the time respectively.About 25% of A. quattuor and A. cf.quattuor females were misclassified as the other species.Results from the jackknifed classification matrix show similar, if slightly lower percentages of correctly classified individuals.In this matrix, one A. sheari is misclassified as A. cosmetoides, and three A. cosmetoides are misclassified as A. sheari.This is likely the result of convergent similarities -females of A. sheari and some A. cosmetoides (A.shoshone sensu stricto) lack scutal spines.
Principle components analyses regularly recovered all hypothesized species of Acuclavella with the exception of A. cf.quattuor from A. quattuor.Figure 4 shows a graph recovering or nearly recovering these species.Acuclavella sheari was most frequently recovered in morphospace, with strong evidence for A. leonardi and A. makah as morphometrically distinct.Also well-supported, A. merickeli and A. cosmetoides were recovered in PCA analyses.Acuclavella cf.quattuor and A. quattuor were regularly recovered as distinct from other hypothesized species, but were not distinguishable from each other.Generally, analyses run on covariance matrices had more success of clustering hypothesized species than analyses run on a correlation matrix.Similar to DFAs, males tended to be recovered more frequently than females.Graphs plotting principle components from all PCA analyses conducted (n=143), as well as a description of strategies employed to explore morphospace, are available as a supplementary file (Appendix IV).The number of individuals from each hypothesized species used in morphometric analyses.
Figure 4. PCA bi-dimensional plot.PCA results of male data using a covariance matrix; plot of principle components 2 and 3.

Difficulties with previously described species
Both phylogenetic and morphometric analyses strongly support three new species: Acuclavella leonardi, Acuclavella makah, and Acuclavella sheari, but do not clearly support already recognized taxa within Idaho samples north of the Salmon River.Molecular and morphological data sets agree that phylogeographic structure associated with the Salmon, Selway, and Lochsa Rivers exists, and morphological data suggests differentiation across the South Fork Clearwater River.Although the number of individuals and known localities was greatly improved by this research, further specimen and genetic sampling is needed to adequately test species limits in northern Idaho.In particular, more rapidly-evolving genetic markers are needed for resolution of what is likely a relatively recent evolutionary radiation in this region.A lack of phylogenetic structure within most of northern Idaho, coupled with the finding that some of the undescribed morphologies cluster together with A. shoshone and A. cosmetoides, make it possible that A. cosmetoides is a widespread and morphologically highly variable species that encompasses A. shoshone.The same hypothesis may apply to the A. merickeli plus A. quattuor lineage.
Though they lack molecular support, A. merickeli and A. quattuor are readily discriminated by morphometric analyses (Figure 4).Since male A. quattuor were collected from the type locality and agree with the diagnostics for females (Shear 1986), the male for this species is here described and figured for the first time.

Biogeography
The Western Hemlock Zone (WHZ) (Franklin and Dyrness 1988) of northern Idaho is disjunct from its coastal counterpart.This disjunction is explained by the orogeny of the Cascade Mountains 8-5 MYA (Mitchell and Montgomery 2006), and subsequent xerification of the Columbia Basin due to the rain shadow formed from this uplift (Leopold and Denton 1987).Since this isolation, the WHZ in Idaho has been subject to the numerous glacial oscillations including alpine glaciers and continental ice sheets (Gates 1983).During peak glaciation it is believed that organisms of the WHZ were forced into compartmented refugia, either in valleys or south of glaciated regions (Brunsfeld andSullivan 2006, Nielson et al. 2006).
Reconstructed molecular phylogenies show that Acuclavella biogeography is largely congruent with that seen in other dispersal-limited taxa of the Pacific Northwest.The deep phylogenetic split between Washington and Idaho (Figures 2, 3) conforms to the ancient vicariance hypothesis (Brunsfeld et al. 2001) seen in amphibians including Ascaphus (Nielson et al. 2001), Dicamptodon (Steele et al. 2005), and Plethodon (Carstens et al. 2004).Relatively long branches separate Acuclavella species from the Cascade Mountain and Olympic Peninsula ecoregions of western Washington, which is congruent with patterns seen in Rhyacotriton salamanders (Good and Wake 1992).

Natural history
Acuclavella are riparian obligate forest-dwellers.The vast majority of specimens were collected within the channel wall of perennial headwater streams or in seep-like features.Most specimens were found underneath large woody debris, though excavation of woody sediment wedges (May and Gresswell 2003) or mats of moss adjacent to these streams also resulted in securing individuals.Coniferous canopy cover is typically consistent across these water features at northern Idaho localities.The dominant vegetation at Idaho sites included a canopy of Tsuga heterophylla and Thuja plicata, or Abies grandis in southern areas.Washington localities typically had a riparian vegetation corridor along small streams dominated by Alnus rubra and Rubus spectabilis (Appendix I).
The primary defense of Acuclavella appears to be one of crypsis.Their dark bodies are cryptic against moist woody debris, onto which specimens adhere with outstretched legs.When detected, specimens are easily collected by grabbing onto an outstretched leg; a hurried scramble is soon followed by limb rigidity.Autospasy was not encountered.A strategy to overcome the crypsis of Acuclavella is to expose appropriate areas and wait a few minutes for animals to come out of thanatosis to scurry for cover and appropriate microclimatic conditions.Acuclavella likely employ mechanical defense via spination.When turning a cover object the first author placed the pad of his finger directly on an individual; it felt like being pricked by a rose.It is likely that heavy sclerotization of the integument and the ubiquitous hemispherical warts add to the structural integrity of these spines.Though circumstantial, this proposal of heavy sclerotization for mechanical defense is consistent with recent findings in other harvestmen (Souza and Willemart 2011).
Acuclavella appear to have an annual life cycle, with seasonal adults.Penultimate stage subadult and adult animals have been collected in April and May; additionally, a young instar was collected in mid October (CHR3435; same locality as CHR 1409; Appendix I).Adults have been found from May to September.Mating was not encountered, nor was feeding.A single individual (CHR2403) (Appendix I) was found dead in the web of a Pimoa spider (CHR2404).

Poor-person's rainforest
Recent surveys of litter dwelling animals in mesic forests of the Pacific Northwest have led to the discovery of many new species.Taxa include terrestrial gastropods (Leonard et al. 2011, Leonard et al. 2003), millipeds (Shear and Leonard 2003, Shelley and Shear 2006, Shear and Leonard 2007, Shelley et al. 2010), pseudoscorpions (Mark Harvey personal communication), and Opiliones (this work, additional undescribed species).An appreciation of these creatures benefits society in numerous ways.Taxonomic knowledge provides a better understanding of forest ecosystems, by identifying potential nutrient recyclers and vertebrate prey items.Taxonomic knowledge also provides future researchers with a baseline from which to compare trendsin the words of Joseph Grinnell (1910) "this value will not, however, be realized until the lapse of many years, possibly a century…".There is promise of novel chemical discovery in the repugnant secretions of harvestmen and millipeds.To some they embellish the aesthetic beauty of our forests.Though many of these newly described organisms are widespread, and without conservation concern, this is likely not true for all taxa (e.g.Shear and Leonard 2004).It is important that surveys continue in this geographic region and for litter faunas in particular.Litter dwellers are ideal organisms for establishing or adding to a metric of biodiversity, and it is important that true microendemics are identified; this information greatly informs conservation efforts (Harvey et al. 2011).

Systematics
The genus Acuclavella, and species A. merickeli, A. cosmetoides, A. shoshone, and female A. quattuor are described by Shear (1986); where descriptive information follows those entries, it should be considered an addendum to his descriptions.Depository abbreviations: American Museum of Natural History (AMNH), California Academy of Sciences (CAS), University of Washington Burke Museum (UWBM), personal collection of primary author (CHR).
Key to the species of Acuclavella.This dichotomous key should allow users to identify the new species described herein.However, discovery of new morphologies in northern Idaho not described by Shear (1986), and not ascribed to species herein, are currently of uncertain placement at the species level.These morphologies include females with pairs of spines on scute areas I-IV, individuals with three pairs of spines on scute areas I-III, and individuals with pairs of spines on scute areas I-II which are geographically and genetically distinct, but morphologically similar to Acuclavella quattuor.For the purposes of this work, the four species named by Shear (1986) 5 and 6, Appendix VIII: Figure 1, Figure 2 Acuclavella merickeli Shear, 1986 (in part: specimen from Pg 21-22) Type material.Male holotype (AMNH), and male (CAS, CASENT9039218) and female (AMNH) paratypes from a tributary of Iron Creek, Forest Service Rd 25 4.6 miles south of FS Rd 300, Gifford Pinchot National Forest, Lewis County, Washington; male paratype (UWBM, WA2392/6319) and female paratype (CAS, CASENT9039224) from upper Iron Creek, FS Rd 28 0.1 miles E of FS 25, Gifford Pinchot National Forest, Skamania County, Washington; female paratype (UWBM, WA2391/6027) from a tributary of Goble Creek, Cowlitz County, Washington.Further information on type localities can be found in Appendix I with the exception of female paratype from Goble Creek.This specimen was collected by the first author 3 August 2005 and deposited in a research collection; this specimen was not used in morphometric analyses, but is used to characterize the species in the description below.This specimen was collected at a location accessed via S Goble Creek Rd; 2.0 miles east of Rose Valley Rd turn right, 1.6 miles turn right, 0.6 miles park; N46.0963°, W122.7607°, elevation 227 meters.
Etymology.The specific epithet is a patronym in honor of the naturalist and careful observer William P. Leonard for his work on litter-dwelling organisms in the poorperson's rainforest of the Pacific Northwest.
Diagnosis.Distinguished from all Acuclavella except A. makah by the combination of having paramedian tubercles as enlarged spines on area II only, and having light, strongly contrasting ends to sclerotized leg segments, giving the appearance of banding.Also distinguished from these taxa in that false leg articulations on the metatarsi of legs II are present, or single dark prolateral tubercles on the palpal patellae are present, but these features are not consistently found in A. leonardi.Scutes posterior to spines containing many raised mounds bearing warty tubercles, more distinct than in A. makah.Though the height of scutal spines is similar, the base of the spines in A. leonardi appears broader than in A. makah.Diagnostic COI sequences have been uploaded to the Barcode of Life Data Systems (BOLD: ACUOP005-13).
All scutal tergites with pairs of median tubercles, these prolonged into large spines on area II only; lateral tubercles distinct.Tergite I with paired median tubercles as raised mounds adorned with warts standing 0.09 mm (n=7, 0.08-0.13mm) above the scutal surface; two additional pairs of warty mounds reduce in size laterally.Median tubercles on area II tergite greatly enlarged into erect spines standing 1.04 mm (n=7, 0.85-1.16mm) above the surface of the scutum; lateral to these tubercles are raised mounds adorned with warts.Area III tergite with three pairs of relatively large tubercles in the form of raised mounds adorned with warts; these not decreasing in size laterally; apical setae on each mound; median pair 0.06 mm above scutum (n=7, 0.03-0.08mm).Tergite areas IV and V with three or four (UWBM) pairs of tubercles in the form of raised mounds adorned with warts, these not decreasing in size laterally or posteriorly; setae as previous; area IV median tubercle height 0.06 mm (n=7, 0.05-0.08mm).First free tergite (VI) with relatively large and numerous tubercles in the form of raised warty mounds.Second free tergite (VII) as previous in the holotype and paratype (CAS); paratype (UWBM) without tubercles.Free tergite VIII without distinguishable tubercles, or with a median pair only (CAS paratype).
Leg measurements given in Table 4. Microsculpture of femora, patellae, and tibiae scattered, distally elevated scales, bilobed scales not observed; scales subtend setae, occasionally housing seta apically.Leg trochanters, femora, patellae, tibiae light brown, dark brown, or black, lighter at joints; metatarsi of leg III with proximal one-third to one-half light brown, brown, or black; leg IV with proximal three-quarters light brown, brown, or black; proximal end of metatarsi of legs I and II pale brown, brown, or black; remaining metatarsal areas pale brown; tarsi pale brown, darkening distally.Scaled microsculpture subequal to darkened areas, remainder with setae and microtrichia.Metatarsi of leg II with false leg articulations (n=7).Penis length 2.48 mm (n=1), glans plate 0.37 mm, stylus 0.09 mm, stylus slightly twisted, not decurved.
Leg measurements given in Table 4. Leg trochanters, femora, patellae, and tibiae brown to dark brown, lighter at joints; metatarsi of legs III with proximal one-half brown, of leg IV with proximal three-quarters light brown to brown, proximal ends of legs I and II light brown to brown, remaining metatarsal areas pale brown; tarsi pale brown, darkening distally.
Etymology.The specific epithet refers to the Makah Nation, which historically occupied much of the known distribution of the species.The name Makah was given to these people by their neighbors; it means "generous with food".These people have shared with many people access to their beautiful land, next to the rocks and gulls.For more information on the Makah Nation see: http://www.makah.com.
Diagnosis.Distinguished from all Acuclavella except A. leonardi by the combination of having paramedian tubercles on area II only, having light, and strongly contrasting ends to sclerotized leg segments, giving the appearance of banding at joints.Though not always present, false leg articulations on the metatarsi of legs II, and single dark prolateral tubercles on the palpal patellae also diagnose it from these species.Scutal tubercles lateral to paramedian tubercles tend to be on more distinctive raised mounds in A. leonardi.Area II spines more narrow at base than in A. leonardi.Best diagnosed from A. leonardi using molecular data.Diagnostic COI sequences have been uploaded to the Barcode of Life Data Systems (BOLD: ACUOP007-13).
All scutal areas with pairs of paramedian tubercles.Area I paramedian tubercles cluster of cuticular warts standing 0.04 mm above the surrounding scute (n=10; 0.025-0.075mm), two additional pairs of warty tubercles reduce in size laterally.Paramedian tubercles of area II rise to form large acute spines standing 1.37 mm above the scutum surface (n=10; 1.13-1.75mm); lateral to spines a pair of tubercles typically small cluster of warts, though one individual (CHR1536) with lateral pair of short spines.Scute areas III, IV, and V with three pairs of wart-clustered tubercles each; paramedian pair largest, diminishing in size laterally along scute, and posteriorly across scute areas.Paramedian tubercles of area III stand 0.03 mm (n=10, 0.03-0.05mm) above surface of the tergite; area IV tubercle height 0.04 mm (n=10, 0.03-0.08mm).Holotype and paratype (CAS) with first free tergite (area VI) with barely distinguishable tubercle as median pair of enlarged warts, tergites VII and VIII without tubercles; paratype (UWBM) tergite VI with median and lateral pair of enlarged warty tubercles, tergite VII with barely distinguishable tubercles, these lacking on tergite VIII.Tergite IX divided, triangular, bracketing tergite X, which forms the anal operculum.
Leg measurements given in Table 5. Leg trochanters, femora, patellae, and tibiae black, lighter at joints; metatarsi of leg III with proximal half black or darkened, leg IV with proximal three-quarters black or darkened, proximal end of legs I and II black or darkened, remaining metatarsal areas yellow-brown or pale brown; tarsi yellow-brown or pale brown, darkening distally.
Etymology.The specific epithet honors Dr. William A. Shear, eminent milliped and opilionid taxonomist.His influence has been important to the authors' aspirations to be systematic biologists, and we thank him sincerely, and with pleasure.
Diagnosis.Generally reduced dimensions overall.Males with one pair of scutal spines on area II only; scutal spines ≤ 0.50 mm, distance from ventral edge of eye to tip of ocularium ≤ 0.60 mm distinguishes it from other males with single pair of spines.Females lack scutal spines.Diagnosed from spine-less A. shoshone (Shear, 1986 sensu stricto) females by having palpal femora ≤ 0.88 mm, leg II tarsi ≤ 3.92 mm, leg II femora ≤ 2.75 mm.
Eye tubercle at anterior edge of carapace, hemispherical warts cover entirety of ocular spine, prolonged into an acute spine standing 0.67 mm above the surface of the carapace (n=3, 0.64-0.70mm), 0.53 mm (n=4, 0.48-0.60mm) from the ventral edge of the eye to the tip of the tubercle, eye spine less acute than in other species of Acuclavella.Eyes dark brown to brown, located basally on tubercle.Surface of carapace evenly curved, posterior margin arcuate.Metapeltidial paramedian sensory cones raised into sharp, acute spines standing 0.21 mm (n=4, 0.17-0.24mm) above surface of the carapace, curving slightly towards the midline, lacking warty microsculpture, shiny; lateral to spines, clusters of warts form tubercles, missing in paratype (UWBM).
All scutal tergites with pairs of paramedian tubercles; relatively short, pointed spines on area II only.Fused tergite I with paramedian tubercles as raised mounds, relatively tall but not spike-like, standing 0.125 mm above the surrounding scute (n=4, all 0.125), these adorned with warts; lateral to these, two additional raised mounds become smaller away from the midline, these lacking in holotype.Paramedian tubercles of area II form acute spine 0.39 mm above the scutum surface (n=4, 0.34-0.43mm), curved slightly posteriad; lateral to these are raised tubercles adorned with warts.Fused tergites III, IV, and V with paramedian tubercles in the form of raised mound adorned with warts; lateral to these, area III with 2 or 3 additional pairs of tubercles, area IV with 1 or 2 additional tubercles, area V with 0 or 2 additional tubercles; these tend to diminish in size away from the midline and posteriorly across tergites.Area III tubercle height above tergite 0.081 mm (n=4, 0.075-0.10mm); area IV tubercle height 0.075 mm (all 0.075 mm).Free tergites without discernable tubercles, or tubercles occur in single pairs as small warty mounds.
Leg measurements given in Table 6.Microsculpture of femora, patellae, and tibiae scattered, distally elevated scales, bilobed scales not observed; scales subtend setae, occasionally housing seta apically.Trochanters, femora, patellae, tibiae black, or in holotype, trochanters light brown, femora, patellae, and tibiae brown; metatarsi of leg III proximal one-third to one-quarter brown or black; leg IV with proximal half black or brown; proximal end of metatarsi on legs I and II black or brown, remaining metatarsal areas pale brown or brown; all tarsi pale brown or brown, darkening distally.Scaled microsculpture subequal to darkened areas, remainder with setae and microtrichia.False leg articulations not observed.Leg claws single, black, not toothed, evenly curved.
Eye tubercle height above surface of carapace 0.67 mm (n=3, 0.56-0.73mm), distance from ventral surface of eye to tip of spine 0.54 mm (n=4, 0.46-0.60mm).Eye color brown or gray.Metapeltidial spine 0.11 mm (n=4, 0.08-0.15mm), curving slightly posteriad and towards the midline, lateral to these spines clusters of warts form tubercles.Dorsal armature lacking.Dorsal adornment generally with median tubercles as raised mounds adorned with warts, decreasing in size away from midline and posteriorly across tergites.Two lateral pairs of warts on areas III and IV, area V with median pair of tubercles barely discernable, two pairs of lateral tubercles, or shallowly raised mounds adorned with warts.Free tergites with tubercles in the form of clusters of warts, or without discernable tubercles.Paramedian tubercles heights above the surface of the tergite (n=4): of area I 0.07 mm (0.05-0.08 mm); area II 0.10 mm (0.08-0.15 mm); area III 0.06 mm (0.05-0.08 mm); area IV 0.05 mm (0.03-0.08 mm).white, with two seta-bearing tubercles; trochanters light brown, dark brown, or white, with three (AMNH) or four (CAS, UWBM) seta-bearing tubercles; femora and patella pale brown, brown, or white, patella without both distal tubercles and diffuse bands; tibiae white or pale-brown; tarsus white or pale-brown, fading to brown distally.Claw rudiment very small (AMNH, UWBM), or absent.
Leg measurements given in Table 6.Leg trochanters brown, black, or pale-brown; femora, patellae, and tibiae dark brown, back, or pale brown; metatarsi of legs III with proximal third brown, black, or pale-brown; leg IV metatarsi with proximal half brown, black, or pale-brown; proximal end of legs I and II metatarsi brown, black, or pale-brown; remaining metatarsal areas pale-brown, brown, or white; tarsi pale brown or white, darkening distally.
For female description see Shear (1986).Distribution and habitat.Acuclavella quattuor populations are bracketed by the Salmon River to the south and the South Fork Clearwater River to the north, whereas A. cf.quattuor is found north of the Selway River and south of the Lolo Trail Ridge.Acuclavella quattuor habitats are dominated by Abies grandis and Picea englemannii; microhabitat in litter, moss, and moist woody debris adjacent to headwater streams.

Figure 1 .
Figure 1.Distribution of Acuclavella.Black circles are localities where Acuclavella were specifically targeted, but not encountered.Black crosses indicate localities where general surveys of litter invertebrates (e.g., Opiliones, Diplopoda, terrestrial Gastropoda) were conducted and Acuclavella were not encountered.Blue lines point to labeled rivers.Type localities (T) from Shear (1986) are indicated with paddle icons; green paddle = A. shoshone.Type localities of species described in this work are indicated with inverted triangles.Image saved from Google Earth.
ses, but is not supported by either (PP = 0.52, BS = 13).No other molecular clades are recovered conforming to other described species of northern Idaho Acuclavella.A KML file has been created to more easily visualize where sequences used to construct the Bayesian phylogeny occur in geographic space (Appendix VI).All Bayesian individual gene trees had their longest branch separating Washington and Idaho genetic groups (Figure3).The COI, EF-1α, and 28S gene trees (Figure3A, 3C, and 3D respectively) also show a deep split separating monophyletic A. leonardi and A. makah samples.Acuclavella sheari was also recovered in each of the individual trees, but with varied support and topological placement within a clade of

Figure 2 .
Figure 2. Bayesian phylogram resulting from analysis of concatenated dataset.Numbers at nodes correspond to Bayesian posterior probabilities.The outgroup topology is shown at bottom.The inset picture is a female A. shoshone (sensu stricto) collected from the type locality (Shear 1986).

Table 3 .
Sample sizes for PCA and DFA analyses.
length 0.82 mm, width 0.46 mm (n=1); otherwise as A. makah.Distribution and habitat.Known from north-facing, horizontal band of Abies grandis forest above south shore of Salmon River.This habitat is mainly roadless, though intersected by FS 592 and Burgdorf Road in Payette National Forest, Idaho County, Idaho.Adults collected in June and September.Abies grandis, Pseudotsuga menziesii, and Picea engelmannii associated canopy cover; collected under woody debris adjacent to headwater streams.Material examined.Males (AMNH; CAS, CASENT9039220; UWBM, ID0013/5661) from the type locality: Slate Creek Road 10.2 miles east of US 95, Nez Perce National Forest, Idaho County, Idaho.Diagnosis.Diagnosed from all Acuclavella species except for A. cf.quattuor by having two pairs of erect spines on tergites I and II; similar morphologically to A. cf.quattuor, and is best diagnosed using molecular data: basepair 23 of the EF-1α intron cytosine in A. cf.quattuor, adenine in A. quattuor; basepair 44 guanine in A. quattuor, thymine or adenine in A. cf.quattuor (thymine if sequences aligned due to a 3 basepair deletion in 40% of A. cf.quattuor at basepair 39-41).

Table 7 .
Acuclavella quattuor male palpus and leg measurements.All measurements in millimeters, n=sample size.
Shear, 1986 http://species-id.net/wiki/Acuclavella_merickeli MorphBank images of specimens considered this species include: CHR2100.1,MorphBank Specimen Id: 822457, 2 images; type locality CHR2100.2,MorphBank Specimen Id: 822597, 1 image CHR2121.2,MorphBank Specimen Id: 822601, 3 images CHR2121.4,MorphBank Specimen Id: 822602, 2 images CHR2121.5, MorphBank Specimen Id: 822459, 3 images CHR2140.0,MorphBank Specimen Id: 822460, 3 images CHR2140.2,MorphBank Specimen Id: 822603, 1 image Figures: Appendix IX Diagnosis.For species descriptions see Shear 1986.Discriminated from other Acuclavella having a single pair of paramedian spines on tergite II by having relatively homogenously colored leg segments, not giving the appearance of banding at leg joints (from A. makah and A. leonardi) or by having scutal spines > 0.50 mm and ocularium height from ventral edge of eye to tip of spine > 0.60 mm (from A. sheari).Distribution and habitat.Acuclavella merickeli populations are bracketed to the south by the South Fork Clearwater River and to the north by the Selway River; all localities in Nez Perce National Forest, Idaho County, Idaho.Coniferous habitat dominated by Picea englemannii or Thuja plicata and Pseudotsuga menziesii; microhabitats include moist woody debris and moss adjacent to small perennial water features such as side-slope seeps and headwater streams.collected with help from Daniel Richart, William Leonard, Shahan Derkarabetian, Jeffery Underwood, Adrienne Richart, Paul Marek, and Dean Levitt.Hours of morphological measurements were expedited with help from Kristen Emata, David Carlson, Nicasia O'Neal, Adrienne Richart, and Kyle Wilson.Nicasia O'Neal, Roy Larimer, and Joseph Warfel helped with images.Maureen McCormack, Shahan Derkarabetian, and Robin Keith Hedin helped with laboratory training.Stas Vidyakin helped with accessioning MorphBank images.This research was enriched from discussions of concepts and methodologies with Ricardo Carretero, Rod Crawford, Shahan Derkarabetian, Andrew Gottscho, William Leonard, Dean Levitt, Maureen McCormack, Tod Reeder, Jordan Satler, Axel Schönhofer, Peter Scott, William Shear, Jack Sullivan, and others.The Wnt2 Ischyropsalis sequence came from the collection of Jochen Martens via Axel Schönhofer, which was collected by S. Huber.The manuscript was improved via suggestions and edits by Angela DiDomenico, Kristen Emata, Ryan Fitch, Erika Garcia, Marc Hayes, Jordan Satler, Axel Schönhofer, Prashant Sharma, William Shear, and an anonymous reviewer.