A molecular phylogeny of Pseudocrangonyx from Japan , including a new subterranean species ( Crustacea , Amphipoda , Pseudocrangonyctidae )

A subterranean species of pseudocrangonyctid amphipod, Pseudocrangonyx gudariensis Tomikawa & Sato, sp. n., is described from the spring-fed stream Gudari-numa in Hakkoda Mountains, Aomori Prefecture, northern Japan. Pseudocrangonyx gudariensis is morphologically similar to P. coreanus Uéno, 1966 and P. febras Sidorov, 2009 based on its relatively small body size, small number of articles of rami of pleopods, and urosomite 1 without basal setae. However, P. gudariensis is distinguished from those species based on the following characteristics: from P. coreanus, antenna 2 of female without calceoli, palmar margins of gnathopods 1 and 2 with distally notched robust setae, inner margin of inner ramus of uropod 2 with 4 robust setae, and basal part of inner ramus of uropod 2 without slender seta; and from P. febras, carpus of gnathopod 2 without serrate robust setae on posterodistal corners, peduncle of pleopods 1 and 2 with setae, and longer article 2 of uropod 3. Phylogenetic analyses using nuclear 28S rRNA and histone H3, and mitochondrial cytochrome c oxidase subunit I and 16S rRNA markers showed that P. gudariensis is placed among known Pseudocrangonyx Akatsuka and Komai, 1922 species. However, its exact phylogenetic position within the genus could not be determined. The polyphyly of the Japanese Pseudocrangonyx species indicates that multiple colonization events of Pseudocrangonyx ancestors to the Japanese Archipelago could have occurred. The reliability of the past Pseudocrangonyx records from Japan is briefly discussed.


Introduction
Amphipods that belong to the genus Pseudocrangonyx Akatsuka & Komai, 1922 inhabit subterranean waters of Japan, the Korean Peninsula, eastern China, and the Far East of Russia; this genus currently includes 20 species (Sidorov and Gontcharov 2013).
Over recent decades, additional species have been described from the Far East of Russia (e.g., Sidorov and Gontcharov 2013).These results indicate that Pseudocrangonyx is highly diversified, and several species await description (Sidorov and Gontcharov 2013).Therefore, it is highly possible that there are also many undescribed species in the Japanese Archipelago.
During field surveys of the benthic invertebrate fauna in the spring-fed stream Gudari-numa in Hakkoda Mountains, Aomori Prefecture in the northern Honshu, two of the authors (AS and AO) and their colleagues collected several Pseudocrangonyx specimens.After careful examination of the materials, it was revealed that the collected Pseudocrangonyx amphipods represent an undescribed species.Thus, this new species is described herein.In addition, the phylogenetic position of the new species within Pseudocrangonyx was estimated using nuclear 28S rRNA and histone H3, and mitochondrial cytochrome c oxidase subunit I and 16S rRNA sequence data.
The taxonomic description was prepared by the first and third authors (KT and AS).The second author (TN) conducted the molecular analyses, the fourth author (YO) assisted in manuscript preparation, and the last author (AO) provided the material of the new species and conducted this study.

Sample
Specimens of Pseudocrangonyx species were collected from 14 localities in Hokkaido, Honshu, and Shikoku, Japan (Fig. 1).Most of specimens were collected by scooping various groundwater environments in caves with a fine-mesh hand-net.The Gudari-numa specimens were pumped up with 10-55 L of interstitial water at 21-57 cm  1.
beneath the surface of gravelly bottom using a handmade core sampler, and then collected, or directly collected by scooping together with bottom gravels of the stream.Most of the specimens were fixed in approximately 10% formaldehyde solution but a few were in 99% ethanol.

Morphological observation
All appendages of the examined specimens of the undescribed species were dissected in 70% ethanol and mounted in gum-chloral medium on glass slides under a stereomicroscope (Olympus SZX7).Specimens were examined using a light microscope (Nikon Eclipse Ni) and illustrated with the aid of a camera lucida.The body length from the tip of the rostrum to the base of the telson was measured along the dorsal curvature to the nearest 0.1 mm.The nomenclature of the setal patterns on the mandibular palp follows Stock (1974).The specimens are deposited in the Tsukuba Collection Center of the National Museum of Nature and Science, Tokyo (NSMT) and the Zoological Collection of Kyoto University (KUZ).
The PCR reaction and DNA sequencing for a part of COI and 16S sequences followed Tomikawa et al. (2014); the PCR and CS reactions were performed using a PC-320 Thermal Cycler (ASTEC).Those for the other sequences were performed using the modified method outlined by Nakano (2012) and Tomikawa et al. (2016); reactions were performed using a T-100 Thermal Cycler (Bio-Rad).When using a PC-320 Thermal Cycler, the PCR mixtures were heated to 94°C for 7 min, followed by 35 cycles at 94°C (45 s), 42°C (1 min) and 72°C (1 min), and a final extension at 72°C (7 min).In the other reactions using a T-100, the PCR mixtures were heated to 94°C for 6 min, followed by 35 cycles at 94°C (10 s), 50°C for 28S and H3 or 48°C for COI and 16S (20 s each), and 72°C (1 min 24 s for 28S, 24 s for H3, and 42 s for COI and 16S), and a final extension at 72°C for 6 min.When using a PC-320, the CS conditions were 25 cycles at 96°C (10 s), 50°C (5 s) and 60°C (4 min).The sequencing mixtures for the other reactions were heated to 96°C for 2 min, followed by 40 cycles at 96°C (10 s), 50°C (5 s) and 60°C (36 s).The obtained sequences of a portion of COI and 16S were edited using MEGA6.03(Tamura et al. 2013), and the reminders were assembled using DNA BASER (Heracle Biosoft S.R.L.).These DNA sequences were deposited with the International Nucleotide Sequence Database Collaboration (INSDC) through the DNA Data Bank of Japan (DDBJ) (Table 1).
The phylogenetic position of the Pseudocrangonyx amphipod from the Gudari-numa Stream within the genus was estimated based on the gene fragments of 28S, H3, COI, and 16S sequences.The alignments of H3 and COI was trivial, as no indels were observed.The 28S, and 16S sequences were aligned using MAFFT v. 7.299b L-INS-i (Katoh and Standley 2013).The lengths of 28S, H3, COI and 16S sequence lengths were 1,480, 328, 658, and 431 bp, respectively.
Prior to construction a phylogenetic tree based on the concatenated sequences, maximum likelihood (ML) trees were constructed based on each of the 28S, COI, and 16S markers using RAxML v. 8.2.8 (Stamatakis 2014) with the substitution model set as GTRCAT, immediately after nonparametric bootstrapping (Felsenstein 1985) conducted with 1,000 replicates.Based on the three obtained ML trees (not shown), a 28S (HQ286019) and a COI (HQ286032) sequences of C. thingvallensis were removed from the dataset to prevent long branch attraction.Then, 28S sequences were re-aligned using MAFFT L-INS-i and refined with Gblocks Server v. 0.91b (Castresana 2000) with a default setting, and thus their final length was 980 bp.The concatenated sequences yielded 2,397 bp of alignment positions.One of the completely identical sequences (G1297 and G1298) was removed from the dataset using the "pgelimdupseq" command implemented in Phylogears v. 2.0.2014.03.08 (Tanabe 2008).
ML phylogenies were conducted using RAxML v. 8.2.8 with GTRCAT, immediately after nonparametric bootstrapping (BS) conducted with 1,000 replicates.The best fit-partitioning scheme for the ML analysis was identified with the Akaike information criterion (Akaike 1974) using PartitionFinder v. 1.1.1(Lanfear et al. 2012) with the "all" algorithm: 28S/1st and 2nd positions of H3/3rd position of H3/1st position of COI/2nd position of COI/3rd position of COI/16S.BI and Bayesian poste-Table 1.Samples used for the phylogenetic analyses.The information on the vouchers is accompanied by the collection localities and the INSDC accession numbers.Sequences marked with an asterisk were obtained for the first time in the present study.Acronym: NSMT, the Tsukuba Collection Center of the National Museum of Nature and Science, Tokyo.Identification sources: a, by the first author KT; b, Narahara et al. (2009);c, Nunomura (1975);d, Uéno (1927); e, Uéno (1971a).rior probabilities (PPs) were estimated suing MrBayes v.
3.2.6 (Ronquist et al. 2012).The best-fit partition scheme as well as models for each partition were selected based on the Bayesian information criterion (Schwarz 1978) using PartitionFinder with the "all" algorithm: for 28S, GTR+G; for the 1st and 2nd positions of H3, JC69; for the 3rd position of H3, K80+G; for the 1st position of COI, SYM+I; for the 2nd position of COI, HKY85+I+G; for the 3rd position of COI, GTR+I+G; and for 16S, GTR+I+G.Two independent runs of four Markov chains were conducted for 20 million generations, and the tree was sampled every 100 generations.The parameter estimates and convergence were checked using Tracer v. 1.6.0(Rambaut and Drummond 2013) and the first 50,001 trees were discarded based on these results.Type locality.Japan, Aomori Prefecture: Aomori, Gudari-numa Stream (northern Honshu).
Uropod 1 (Fig. 7E) with basofacial slender seta on peduncle; inner ramus 0.70 times as long as peduncle, inner margin of former with 2 robust setae, outer margin bare, basal part with 3 slender setae; outer ramus 0.76 times as long as inner, its inner and outer margins with 0 and 2 robust setae, respectively.Uropod 2 (Fig. 7F, G) with inner and outer rami; inner ramus 0.90 times as long as peduncle, its inner margin with 4 robust setae, outer margin bare, distal part with 2 serrate and 4 simple robust setae and 1 slender seta (Fig. 7G); outer ramus 0.89 times as long as inner ramus, its outer margin with 1 robust seta.Uropod 3 (Fig. 7H) with peduncle 0.33 times as long as outer ramus, with 1 dorsal and 3 ventral robust setae; inner ramus absent; outer ramus 2-articulate, proximal article with robust setae, terminal article 0.36 times as long as proximal article, with 3 distal setae.Telson (Fig. 7O) length 1.2 times as long as wide, cleft for 0.08 times of length, each telson lobe with 2 lateral long penicillate setae, 2 apical robust and 1 apical short penicillate setae.
Uropod 1 (Fig. 9A) with 3 robust setae on inner margin of inner ramus, basal part with 2 slender setae; outer ramus 0.80 times as long as inner.Uropod 2 (Fig. 9B) with 6 simple robust setae and 1 slender seta on distal part of inner ramus.Uropod 3 (Fig. 9C) with peduncle 0.32 times as long as outer ramus; terminal article of outer ramus 0.35 times as long as proximal article.
Etymology.The specific name is an adjective derived from Gudari-numa, the type locality of the new species.
Distribution and habitat.This species is known only from the type locality.The specimens were collected from interstitial water in the gravelly bottom.Water temperature of the habitat was stable and around 7°C throughout the year (Baba and Ohtaka unpublished).Remarks.Pseudocrangonyx gudariensis is morphologically similar to P. coreanus described from the Korean Peninsula.The deposited female paratypes of the latter species have calceoli on antenna 2 and pleopods without bifid setae on inner basal margin of inner ramus, which are features that were not mentioned in the original description (NSMT-Cr 13521-13522; Tomikawa and Onodera, personal observation).These two species share the following features: 1) relatively small body size (smaller than 6 mm), 2) eye completely absent, 3) carpus of gnathopod 2 without serrate robust setae on posterodistal corners, 4) outer margin or outer distal corner of pleopods 1 and 2 with setae, 5) inner basal margin of inner ramus of pleopods without bifid setae, and 6) small number of articles (less than 5) of rami of pleopods.However, P. gudariensis is distinguished from P. coreanus by the fol-lowing features (features of P. coreanus in parentheses): 1) antenna 2 of female without calceoli (present); 2) palmar margins of gnathopods 1 and 2 with distally notched robust setae (absent); 3) inner margin of inner ramus of uropod 2 with 4 (0 or 1) robust setae; and 4) basal part of inner ramus of uropod 2 without slender seta (present).
Pseudocrangonyx gudariensis is also similar to P. febras from river basin of Primorye, Russia in having 1) relatively small body size (smaller than 6.5 mm), 2) eye completely absent, 3) palmar margins of gnathopods 1 and 2 with distally notched robust setae, 4) small number of articles (less than 6) of rami of pleopods, and 5) urosomite 1 without basal setae.However, P. gudariensis is distinguished from the latter by the following features (features of P. febras in parentheses): 1) carpus of gnathopod 2 without serrate robust setae on posterodistal cor-ners (present), 2) peduncle of pleopods 1 and 2 with setae (absent), and 3) article 2 of uropod 3 longer (shorter) than setae on distal part of article 1.

Discussion
As mentioned in the Remarks, P. gudariensis is morphologically similar to P. coreanus and P. febras.These three species share the following characteristics: relatively small body size, absence of basal setae on urosomite 1, and small number of articles of rami of pleopods.However, our phylogenetic analyses failed to recover monophyly of P. gudariensis + P. febras.In addition, P. coreanus sensu Narahara et al. (2009) is genetically distant from P. gudariensis and P. febras.Therefore, the aforementioned shared characteristics do not reflect phylogenetic relationships of Pseudocrangonyx species.Whatever the case, the results of our morphological examination and molecular phylogenetic analyses fully support the distinct taxonomic status of the present new species in this genus.Sidorov and Holsinger (2007) suggested that the colonization events of the ancestral Pseudocrangonyx species to the Japanese Archipelago could have taken place twice: in the Middle-Late Miocene and the Early Pleistocene through land bridges between the continental China and the Japanese Archipelago.The obtained phylogeny also indicated that multiple colonization events of Pseudocrangonyx amphipods to the Japanese Archipelago occurred.The present Japanese specimens were split into two clades, and three distinctive lineages, of which phylogenetic positions still remain uncertain.Moreover, the obtained tree showed that the phylogenetic positions of the several Japanese individuals did not reflect their geographical distributions.The specimens distributed in the Chugoku region (#11, 14, 15) did not form a monophyletic group with geographically close samples (e.g., #12, 13).These three specimens formed a clade with the northern Japanese P. yezonis and two Russian species.These results indicate that Japanese Pseudocrangonyx species experienced a complicated biogeographical history.To clarify the origin and dispersal routes of Pseudocrangonyx amphipods, comprehensive taxa sampling and more detailed genetic data are needed.
Our phylogenetic results also revealed that the species diversity of Japanese Pseudocrangonyx is quite high, and they should be classified into the known three species and additional undescribed species.Therefore, taxonomic studies should be conducted to determine the systematic accounts of these undescribed amphipods.First, however, the taxonomic status of the three known species described by Akatsuka and Komai (1922) should be revisited.In the obtained tree, P. kyotonis sensu Nunomura (1975) (#6) and sensu Uéno (1971a) (#12) were not monophyletic.The latter was genetically more closely related to the Pseudocrangonyx species distributed in Shikoku (e.g.,#16,17), where the type locality of P. shikokunis is located.Alternatively, P. shikokunis sensu Uéno (1927) (#14) was distantly related to the Pseudocrangonyx specimens collected in Shikoku.These results highlighted that the previous studies of Pseudocrangonyx in Japan contained misidentified or taxonomically uncertain records.
In addition to P. gudariensis, P. kyotonis sensu Uéno (1971a), P. kyotonis sensu Nunomura (1975), and P. coreanus sensu Narahara et al. (2009), five unidentified Japanese Pseudocrangonyx species were identified based on molecular phylogenetic analyses (Fig. 10).Among these five species, it is highly possible that Pseudocrangonyx sp. 5 and Pseudocrangonyx sp. 4 represent the true P. shikokunis and P. kyotonis, respectively, because their collection localities are close to each of the type localities of these species.Accordingly, P. kyotonis sensu Nunomura (1975) and P. shikokunis sensu Uéno (1927) could be considered unidentified Pseudocrangonyx sp.6 and Pseudocrangonyx sp. 2, respectively.To confirm whether these taxonomic treatments are adequate, molecular phylogenetic analyses including topotypic sequences of described species and detailed morphological analyses are needed.
Our phylogenetic analyses also shed light onto the taxonomic account of Eocrangonyx Schellenberg, 1936.This genus have been placed under the family Pseudocrangonyctidae along with Pseudocrangonyx (Holsinger 1989).These two genera bear a close resemblance to each other in general morphology.Eocrangonyx is distinguished from Pseudocrangonyx by the absence of article 2 of uropod 3 (Holsinger 1989).However, Sidorov and Holsinger (2007) revealed the presence of extremely reduced article 2 of uropod 3 in the Russian E. stygoedincus (Sidorov and Holsinger, 2007).Subsequently, Tomikawa and Shinoda (2016) also revealed the same characteristics in E. japonicus (Uéno, 1930), which is the type species of Eocrangonyx.Our phylogenetic tree showed that Pseudocrangonyx and Eocrangonyx are phylogenetically closely related (Fig. 10).However, the phylogenetic position of E. primoryensis remains unresolved.In addition, E. japonicus genetic data have been never assessed.To evaluate the independence of Pseudocrangonyx and Eocrangonyx, and the validity of article 2 of uropod 3 as a generic diagnostic character, it is necessary to clarify the phylogenetic relationships between Pseudocrangonyx and Eocrangonyx by including additional species and genetic markers.Consequently, there will be an enhanced understanding of species diversity and evolutionary history of the Far Eastern subterranean Crangonyctoidea species.
Museum), Naoyuki Nakahama (Kyoto University), Ryosuke Okano (Ehime University), Dr Tomislav Karanovic (Sungkyunkwan University), and Naoshi Sato (Ofunato City) for providing specimens of Pseudocrangonyx.KT thanks Satoko Tashiro (Hiroshima University), Yukiko Narahara-Nakano (Hiroshima University), and Daisuke Saiga (Hiroshima University) for supporting field work.Thanks are also due to Dr Ronald Vonk (Naturalis Biodiversity Center), Professor Boris Sket (University of Ljubljana), and Dr Michael Ohl (Museum für Naturkunde) for their critical reading and valuable comments on this manuscript.This work was partly supported by JSPS KAKENHI Grant Numbers JP25242015, JP25840140, JP15J00720.The open access publication of this manuscript was supported by the Museum für Naturkunde.

Figure 1 .
Figure 1.Map showing the collection localities of the specimens examined in this study and type localities of the known Japanese Pseudocrangonyx species.The closed circles indicate the localities of the referred materials used for the phylogenetic analyses.The star in red denotes the type locality of the new species; in purple, P. shikokunis; in blue, P. kyotonis; and in green, P. yezonis.Names of localities are shown in Table1.

Figure 10 .
Figure 10.Bayesian inference tree for 2,397 bp of nuclear 28S rRNA plus histone H3 and mitochondrial COI and 16S rRNA markers, with the map modified from Fig. 1.Numbers on nodes represent bootstrap values for maximum likelihood and Bayesian posterior probabilities.Specimen numbers are also shown in Fig. 1 and Table1.