Two new species of Fluminicola (Caenogastropoda, Lithoglyphidae) from southwest Oregon, USA, and a range extension for F. multifarius

Abstract We describe two new species of pebblesnails (Lithoglyphidae: Fluminicola) from southwest Oregon based on morphologic and mitochondrial DNA (COI, cytB) evidence. Fluminicola umpquaensis sp. n., which had been traditionally identified as F. virens prior to the recent restriction of the latter to the lower Columbia River drainage, lives in lotic habitats in the Umpqua River basin. This species is readily distinguished from closely related F. gustafsoni and F. virens by shell and anatomical characters, and by its mtDNA sequences (divergence >3.6% for both genes). Fluminicola fresti sp. n. ranges among lotic habitats in the North Umpqua River basin, and in the upper Rogue River drainage north of Little Butte Creek. This species differs from other congeners by >9.1% for both genes and is distinguished from closely similar and geographically proximal F. multifarius by several anatomical characters. Additionally, new records are provided for F. multifarius from the upper Rogue River basin south of Little Butte Creek, which extend the geographic range of this species about 80 km northward from the Sacramento River headwater region. This continues a recent series of taxonomic papers on the poorly known and little studied pebblesnail fauna of the vast Pacific watershed from northern California to southern British Columbia.


Introduction
Fluminicola (Truncatelloidea, Lithoglyphidae) is a western North American genus of small freshwater gastropods with globose to conical shells, commonly known as pebblesnails, which is distributed in diverse habitats ranging from small seeps to springinfluenced lacustrine reaches to large rivers. Both morphological (Hershler and Frest 1996) and molecular (Hershler and Liu 2012) evidence suggest that the 25 currently recognized Fluminicola species belong to two evolutionarily distinct lineages. One of the lineages contains two species in the Columbia River basin -F. gustafsoni Hershler & Liu, 2012;F. virens (Lea, 1838) -while the other (containing the remaining congeners) is distributed in this drainage and also the Great Basin, upper Colorado River basin, and Sacramento River basin. Fluminicola continues to be recognized as a non-monophyletic genus pending clarification of the phylogenetic relationships of its poorly known and possibly extinct type species, F. nuttallianus (Lea, 1838).
The broad geographic range of Fluminicola includes most of the Pacific Coastal watershed from northern California to southern British Columbia. Pebblesnail populations are scattered throughout much of this huge area, yet are currently undescribed aside from six species in the Columbia River basin (Hershler and Frest 1996, Hershler and Liu 2012, Liu et al. 2013) and 14 species in the Sacramento River basin (Hershler et al. 2007). This taxonomic knowledge gap is hampering efforts by the conservation community to obtain legal protection for pebblesnails (e.g., USFWS 2012), which are groundwater-dependent and threatened by various anthropogenic activities.
The Fluminicola fauna of southwestern Oregon includes a relatively large pebblesnail in the Umpqua River basin that was traditionally identified as F. virens prior to the restriction of the latter to the lower Columbia River drainage (Hershler and Frest 1996), and numerous unstudied populations of smaller pebblesnails in the upper reaches of both the Rogue and Umpqua River basins that were recently reported in grey literature (Frest and Johannes 1999, 2000, 2004, 2005. Herein we utilize DNA sequences from two mitochondrial genes in delineating the Fluminicola species in the Rogue-Umpqua basins. This continues our recent series of integrative taxonomic studies of the pebblesnails of the Pacific Coastal drainages (Hershler et al. 2007, Hershler and Liu 2012, Liu et al. 2013).

Methods
We sequenced specimens from 35 sites in the Rogue-Umpqua basins, including eight localities containing the large pebblesnail resembling F. virens and 27 localities containing smaller pebblesnails. Thirty-two of these localities were sampled by RH and HPL during September 2015 specifically for this project. Specimens were collected by hand or with a small sieve and preserved in 90% (non-denatured) ethanol in the field. Vouchers were deposited in the Smithsonian Institution's National Museum of Natural History (USNM) collection. Other relevant material from the USNM and the Bell Museum of Natural History (BellMNH) was also examined during the course of this study.
A total of 155 and 146 sequences of cytochrome c oxidase subunit I (COI) and cytochrome B (cytB), respectively were obtained from 161 analyzed Rogue-Umpqua pebblesnails. Genomic DNA was extracted from entire snails using a CTAB protocol (Bucklin 1992); each specimen was analyzed for mtDNA individually. LCO1490 and HCO2198 (Folmer et al. 1994) were used to amplify a 708 base pair (bp) fragment of COI; cytB427F (5'TGA GGK GCN ACT GTT ATT ACT AA3') and cytB1049R (5'GTG AAA ACT TGS CCR ATT TGC TC3') were used to amplify a 644 bp fragment of the cytB gene. The cytB427F and cytB1049R primers were designed based on conserved regions of cytB in an alignment using previously published sequences from Oncomelania hupensis (Gredler) (NC13073) and Potamopyrgus antipodarum (Gray) (GQ996433). Amplification conditions and sequencing of amplified polymerase chain reaction product methods were those of Liu et al. (2013). Sequences were determined for both strands and then edited and aligned using SEQUENCHER™ version 5.4.1 (Gene Codes Corporation, Ann Arbor, MI). Our analysis of the COI dataset also included 35 previously published sequences from 23 congeners, two taxonomically indeterminate Fluminicola lineages from the Sacramento River basin (F. sp. A, F. sp. B, Hershler et al. 2007), and representatives of two other North American lithogyphid genera (Somatogyrus, Taylorconcha). Trees were rooted with Pristinicola hemphilli (Pilsbry) (Hydrobiidae). The cytB dataset also included 34 previously published sequences from 22 Fluminicola species (a cytB sequence is not available for F. gustafsoni) and the two taxonomically indeterminate Fluminicola lineages. Given that cytB sequences are not available for other North American lithoglyphid genera, we used basally positioned F. virens to root the trees (Hershler et al. 2007, Hershler andLiu 2012). In order to generate easily readable topologies, one example of each detected haplotype was used in the phylogenetic analyses, which were performed separately for the COI and cytB datasets. Sample codes, locality and voucher details, and GenBank accession numbers for the sequences used in the molecular phylogenetic analyses are in Suppl. material 1.
Genetic distances were calculated using MEGA7 (Kumar et al. 2016), with standard errors estimated by 1,000 bootstrap replications with pairwise deletion of missing data. MRMODELTEST v. 2.3 (Nylander 2004) selected the GTR + I + G model parameters as the best fit for both the COI and cytB datasets (using the Akaike Information Criterion). Phylogenetic analyses were performed using three different methodologies -maximum parsimony (MP), maximum likelihood (ML), and Bayesian inference. The MP and ML analyses were performed using PAUP* v. 4.0b10 (Swofford 2003) and the Bayesian analyses were conducted using MRBAYES v. 3.2.6 (Ronquist and Huelsenbeck 2003). The MP analysis was conducted with equal weighting, using the heuristic search option with tree bisection reconnection branch-swapping and 100 random additions. Nodal support was evaluated by 10,000 bootstrap replicates. The ML analysis was performed using the GTR + I + G model. The optimized parameter values for COI were base frequencies of A = 0.3097, T = 0.3952, C = 0.1612, G = 0.1339; shape of gamma distribution = 1.3430; proportion of invariant sites = 0.5912. The optimized parameter values for cytB were base frequencies of A = 0.3274, T = 0.3671, C = 0.1919, G = 0.1145; shape of gamma distribution = 0.7958; proportion of invariant sites = 0.54237. A GTR distance based neighbor-joining (NJ) tree was used as the initial topology for branch-swapping. Nodal support was evaluated by 1,000 bootstrap pseudoreplicates. For the Bayesian analyses Metropolis-coupled Markov chain Monte Carlo simulations were run with four chains (using the model selected by MRMODELTEST) for 5,000,000 generations. Markov chains were sampled at intervals of 100 generations to obtain 50,000 sample points. We used the default settings for the priors on topologies and the GTR + I + G model parameters. At the end of the analyses, the average standard deviations of split frequencies were 0.005692 (COI dataset) and 0.004193 (cytB dataset) and the potential scale reduction factor (PSRF) was 1, indicating that the runs had reached convergence. The sampled trees with branch lengths were used to generate 50% majority rule consensus trees, with the first 25% of the samples removed to ensure that the chain sampled a stationary portion.
Large adult females were used for shell measurements. The total number of shell whorls (WH) was counted for each specimen; and the height and width of the entire shell (SH, SW), body whorl (HBW, WBW), and aperture (AH, AW) were measured from camera lucida outline drawings (Hershler 1989). Descriptive statistics were generated using SYSTAT FOR WINDOWS 11.01 (SSI 2004). Other methods of morphological study were routine (Hershler et al. 2007).

Results
Forty-one COI and 55 cytB haplotypes were detected in the analyzed specimens from the Rogue-Umpqua basins (Suppl. material 2-3, respectively). The molecular phylogenetic analyses of both the COI and cytB datasets consistently resolved the Rogue-Umpqua haplotypes into three distinct clades. The trees generated by the three methods of phylogenetic analysis were closely similar; the Bayesian topology based on the COI dataset is shown in Figure 1.
Two of the clades (clades A, B, Fig. 1) contained the smaller Rogue-Umpqua pebblesnails. Clade A, which was weakly supported generally, but well supported in the cytB Bayesian topology (100% posterior probability), was composed of the haplotypes from populations in the upper Rogue River basin to the south of Little Butte Creek (Fig. 2) and the two representative sequences of F. multifarius Hershler, Liu, Frest & Johannes, 2012, a morphologically variable species that is distributed in the upper Sacramento River basin in northern California (Hershler et al. 2007). The divergence between the Rogue pebblesnails in clade A and all published sequences of F. multifarius (2.6% for COI and 3.6% for cytB, Hershler et al. 2007) was slightly greater than the variation within these two groups (2.2 and 1.4% for COI; and 3.5 and 2.4% for cytB, respectively) and falls into the range of pairwise differences among currently recognized Fluminicola species: 1.4-18.7% for COI and 1.6-25.7% for    cytB (Hershler et al. 2007, Hershler andLiu 2012). However, we could not differentiate these geographically disjunct yet phylogenetically intermixed groups of populations based on morphologic criteria and thus are treating them as conspecific.
Clade B, which was well supported in all but the cytB Bayesian analysis, is composed of the small pebblesnails in the North Umpqua basin, and the upper Rogue River basin north of Little Butte Creek (Fig. 2). The members of this clade differ from all currently recognized Fluminicola species by >9.1% for both genes (the sequence divergence within the clade is c. 2% for both genes) and although having the generalized morphology shared by most of the smaller Fluminicola species, can also be distinguished from closely similar and geographically proximal F. multifarius by several anatomical characters. Based on the sum of this evidence we recognize clade B as a somewhat variable new species which is described below.
The third clade (C), which was moderately supported in most of the analyses and well supported in the COI ML tree (100%), contained the haplotypes detected in the large pebblesnails from the Umpqua River basin (Fig. 2). The sequence divergence within this clade was slight -0.3% for COI and 0.6% for cytB, respectively. Clade C, in turn, formed a well-supported monophyletic group with F. gustafsoni and F. virens, and is most similar genetically to the latter, from which it differs by 3.6% for COI and 3.9% for cytB. The large pebblesnail in the Umpqua River basin is also readily differentiated morphologically from both F. gustafsoni and F. virens, and thus we recognize it as a new species which is described below.   The new records detailed herein extend the range of F. multifarius about 80 km northward from the Sacramento River headwaters (Fig. 2). It is not known whether F. multifarius is also distributed in the intervening Klamath River basin; the pebblesnail fauna of this large watershed is currently undescribed. Populations of F. multifarius in the Rogue basin were referred to as the Chinquapin pebblesnail, Emigrant pebblesnail, Keene Creek pebblesnail, Little Butte pebblesnail, and Pilot Rock pebblesnail by Frest and Johannes (2000, 2004, 2005.  Diagnosis. A small to medium-sized Fluminicola (2.3-5.5 mm shell height) having a trochoidal to ovate-conic shell and small, gently tapered penis. Differs from closely similar and geographically proximal F. multifarius in the hooked shape of the anterior end of the osphradium, larger number of ctenidial (gill) filaments, smaller seminal receptacle, and in its mtDNA sequences.

Fluminicola fresti
Description. Shell (Fig. 5A-D) trochoidal to narrow-conic, whorls 3.5-4.0. Teleoconch whorls medium convex, sometimes weakly shouldered. Aperture ovate, slightly angled above; inner lip complete, variably thickened and reflected, sometimes forming a rather wide parietal-columellar shelf that sometimes covers the umbilical region. Outer lip thin, prosocline. Umbilicus very small or absent, umbilical region sometimes excavated. Shell white, periostracum brown, sometimes covered with thick black deposits. Shell measurements and whorl count data are summarized in Table 1. Operculum (Fig. 5E-F) as for genus; muscle attachment margin thickened on inner side. Radula (Fig. 5G-I) as for genus; dorsal edge of central teeth concave, lateral cusps two-five, basal cusp one-two. Lateral teeth having two-three cusps on inner side and three-four cusps on outer side; length of outer wing 175-185% length of cutting edge. Inner marginal teeth with 23-31 cusps, outer marginal teeth with 27-40 cusps. Radula data are from USNM 1422223, USNM 1144426.
Snout, cephalic tentacles, pallial roof, visceral coil usually medium pigmented (brown); foot varying from near pale to medium pigmented along anterior edges. Distal section of penis having dense core of internal black pigment. Ctenidial filaments 21-24 (N = 5), lateral surfaces smooth. Anterior end of osphradium distinctly hooked (not illustrated). Glandular oviduct and associated structures shown in Figure 4E-F. Coiled oviduct circular, proximal arm kinked, posterior arm sometimes having small pouch containing sperm. Bursa copulatrix large, reniform, partly overlapped by albumen gland. Bursal duct slightly shorter than bursa copulatrix, narrow. Seminal receptacle small, sac-like, almost completely overlapped by albumen gland. Albumen gland having short pallial component. Capsule gland longer than albumen gland, composed of two glandular zones. Genital aperture a small, sub-terminal pore. Penis (Fig. 4G) small, slightly curved, gently tapering, distal section abruptly narrowing to small pointed tip. Medial section having a few weak folds along inner edge. Penial duct near centrally positioned, straight, narrow.
Etymology. This species name is a patronym (in the genitive singular) honoring recently deceased malacologist Terrence Frest for his many contributions to the documentation of molluscan biodiversity in the northwestern United States.
Distribution. Fluminicola fresti is distributed in spring-fed habitats in the North Umpqua River drainage and in the Rogue River basin north of Little Butte Creek.
Remarks. As noted above, the shells of F. fresti vary in overall shape and in the width of the inner apertural lip. Although this variation is generally continuous in the material that we examined, two rather distinct forms -ovate-conic with a narrow inner apertural lip (Fig. 5B) and trochoidal with a wide parietal-columellar shelf (Fig. 5C) -can be identified in one of the springs in Joseph Stewart State Park. We sequenced specimens from this locality and found that the two shell forms (samples RU18 and RU17, respectively) differed by 0.7% for COI and 1.3% for cytB, which was less than the variation within these two groups (0.2 and 1.2% for COI; and 0.2 and 1.5% for cytB, respectively). The scant genetic differentiation of the two shell forms when in sympatry provides additional evidence that they are conspecific. Additional studies incorporating rapidly evolving nuclear markers (e.g., microsatellites) may help sift through the possible explanations for the interesting variation in the shells of this species (e.g., ecophenotypic plasticity, incipient speciation). Twenty-three COI hap-  Frest and Johannes (1999, 2000, 2004, 2005. In order to avoid confusion, we suggest that "Frest's pebblesnail" be used as the common name for F. fresti.  Genital aperture a small, sub-terminal pore. Penis (Fig. 4J) large, straight, broad, little tapered, distally rounded, deeply folded along most of length. Penial duct near centrally positioned, rather wide, undulating along entire length, opening through small terminal papilla.

Fluminicola umpquaensis
Etymology. The species name is an adjectival geographic epithet referring to the distribution of this pebblesnail in the Umpqua River basin.
Distribution. Fluminicola umpquaensis is widely ranging in the Umpqua River basin, and is distributed in riverine habitats as well as springs and streams.
Remarks. As mentioned above, the smaller of the two divergent Fluminicola clades (containing F. gustafsoni and F. virens) was previously confined to the Columbia River basin (Hershler and Liu 2012). Fluminicola umpquaensis extends the geographic range of this lineage >200 km southward from the lower Columbia River. Eight COI haplotypes and 11 cytB haplotypes were detected in the sequenced specimens of F. umpquaensis (Suppl. material 2-3, respectively).
Populations identified herein as F. umpquaensis were referred to as the Jade pebblesnail by Frest and Johannes (2000:182). We propose that this common name continue to be applied to this species.