A new genus and two new species of cavernicolous amphipods ( Crustacea : Typhlogammaridae ) from the Western Caucasus

This study deals with the biodiversity and distribution of cavernicolous Amphipoda in caves of the Arabika massif (Western Caucasus). The Sarma, Trojka and Orlinoe Gnezdo caves were explored during speleological expeditions over the years 2011–12. Two new species of Amphipoda were found: a sub-surface dweller Zenkevitchia sandroruffoi sp. nov. is reported from the Sarma, Trojka and Orlinoe Gnezdo caves at depths from -30 m to -350 m; the second one, a deep dweller Adaugammarus pilosus gen. et sp. nov. is reported from the Sarma Cave at depths of -1270 to -1700 m. Adaugammarus gen. nov. shares similarities with Typhlogammarus Schäferna, 1907 and Zenkevitchia Birstein, 1940. The species Anopogammarus birsteini Derzhavin, 1945 is also re-described herein based on new samples that suggest close affinity of this species with the family Gammaridae. The original taxonomic combination is resurrected for Zenkevitchia revazi Birstein & Ljovuschkin, 1970, comb. resurr. (from Anopogammarus Derzhavin, 1945). To accommodate morphologically different species in the genus Zenkevitchia, two new groups are proposed. These are the admirabilis-group (Z. admirabilis Birstein, 1940 and Z. yakovi Sidorov, 2015) and the sandroruffoi-group (Z. sandroruffoi sp. nov. and Z. revazi). An updated molecular (mt-cox1) phylogeny, an identification key to the genera and a distribution map for the typhlogammarid amphipod species of Transcaucasia are provided.


Introduction
The Arabika karst massif is one of the largest and highest in the limestone band of the Western Caucasus.It is bounded by the canyons of the rivers Kutu-Sara, Gega and Bzyb on the North and East, by the Black Sea coast on the Southwest and by the valleys of the Khashupse and Tsandrypsh rivers in the West.The following separation and nomenclature for the speleological areas of the Arabika massif is adopted: the Ortobalagan trough, Gel'geluk trough, Treugol'nik caving district (Zont-Utug-Khyrka), Minskaja Valley trough and the Dzou tract.
Since Birstein and co-authors (Birstein 1940;Birstein & Lopaschov 1940) presented their first studies on the biodiversity of subterranean animals in Transcaucasia, both significant speleological discoveries were made and taxonomic views changed considerably.Dozens of new caves have been explored (Dublyansky et al. 1987) and several new taxa have been discovered (Birstein & Ljovuschkin 1967).The contemporary period is marked by several publications on the cavernicolous fauna of the region (Marin & Sokolova 2014;Vinarski et al. 2014;Sidorov et al. 2015), including biospeleological studies on caves located in the Arabika massif (Jordana et al. 2012;Sendra & Reboleira 2012;Sidorov et al. 2014).However, the invertebrate troglofauna of the Western Caucasus and surrounding areas is still poorly known (Barjadze et al. 2015).
In this paper, we present results of a biospeleological survey from the most studied caving area, Treugol'nik, with about 30 known caves including one of the world's deepest caves, the Sarma Cave.The Sarma, Trojka and Orlinoe Gnezdo caves were studied and among other things (Sidorov et al. 2014), amphipods of the Typhlogammaridae family were collected there.The DNA barcode region of the mitochondrial cytochrome c oxidase subunit 1 (COI) gene sequence was used to verify the distinction of described species, along with comparative morphological analyses.

Specimen collection
The samples of blind, unpigmented aquatic amphipods were collected in three limestone caves; the meters refer to the explored cave depth: Sarma (-1830 m), Trojka (-257 m), and Orlinoe Gnezdo (-75 m) in the Arabika massif of Abkhazia (Fig. 1, Table 1).Additionally, samples containing specimens of Anopogammarus birsteini Derzhavin, 1945 were collected at the type locality near Goluboe Lake in the Gagra District of Abkhazia.Specimens were preserved in 96% ethanol at each field site.The geographical map showing the sampling sites and distribution pattern of the Typhlogammaridae species in the Western Caucasus was constructed with GMT 4.5.6.-1.GIS software (Fig. 1).

Morphology and taxonomic terms
The body length of the amphipods was recorded by holding the specimen straight and measuring the distance along the dorsal side of the body from the base of the first antenna to the base of the telson, using a micrometer eye piece in a Lomo MBS-9 dissecting microscope.Appendages were drawn with a Carl Zeiss NU-2 compound microscope equipped with a drawing device as described in Gorodkov (1961).Heavily calcified specimens of crustaceans were placed in 4% lactic acid and then washed and boiled in clean water to remove air bubbles within the segments.Permanent preparations were made using a methylene blue staining solution, and polyvinyl lactophenol (PVL) was used as the mounting medium.The descriptive terminology follows the classification system in the original conventional sense and does not agree with the homology concepts proposed by Watling (1989).To make the nomenclature more stable, we use the term "spine" for robust setae and the term "seta" for slender, usually flexible structures.The term "palmar angle" of the gnathopod propodi refers to the angle formed at the end of the palm and beginning of the posterior margin or at the point where the tip of the dactylus closes on the propodus (Birstein 1941).The fore-gut lateralia comprise a potentially useful morphological character in the phylogenetic analysis (Coleman 1991).The nomenclature for setal patterns on article 3 of the mandibular palp follows the practice of Karaman (1970) and Stock (1974).The descriptions are based on the type series and all material examined is deposited in the Zoological Museum of the Far East Federal University, Vladivostok (FEFU; the holotypes are kept there) or in the research collection of D.A. Sidorov at the Institute of Biology and Soil Science, Vladivostok (IBSS).

DNA extraction, amplification, sequencing and analysis
Total DNA was extracted from the muscle tissue with a DNeasy Blood & Tissue kit (QIAGEN GmbH, Hilden, Germany) according to the manufacturer's protocol.The COI gene fragment was amplified using the universal primers HCO2198 and LCO1490 (Folmer et al. 1994).The annealing temperature was set at 40°C for 20 s.The PCR products were sequenced directly using the same primers and a BigDye terminator v. 3.1 sequencing kit (Applied Biosystems, USA).Sequences were analyzed on an ABI 3130 genetic analyzer (Applied Biosystems, USA) and assembled with the Staden Package v. 1.4   (Bonfield et al. 1995).The Akaike information criterion (AIC2) in ModelGenerator 0.85 (Keane et al. 2006) was used to select the model of sequence evolution best fitting our data set (HKY+G+I).The data set was analyzed using the maximum-likelihood (ML) algorithm in Mega 6.0 (Tamura et al. 2013), and pairwise sequence divergence was estimated using the same environment.A standard BioNJ initial tree was obtained automatically by applying the Nearest-Neighbor-Interchange (NNI) for tree inference.To assess support for clades 1,000 bootstrap replicates (Felsenstein 1985) were performed.
Fig. 3. Unrooted ML-tree with bootstrap support values based on the mt-cox1 sequences (values less than 50% not shown).Specimen labels refer to information given in Table S1.Scale bar indicates the number of substitutions per site.

Remark
Although A. birsteini was described by Derzhavin (1945) based on an 8.0 mm female and was subsequently redescribed in detail (Birstein & Ljovuschkin 1970) based on both sexes, we further introduce an amended diagnostic description to emphasize some important characters.
GnathopodS 1-2 (Fig. 5C-D).Gnathopod 1: propodus almond-shaped, palm convex, with cutting margin acanthaceous and 2× longer than posterior margin; along posterior margin two sets of simple setae; antero-distal group of anterior margin with 10 setae; palmar margin with short, notched setae along outer and inner faces, palmar angle undefined, a group of 10 distally-notched strong spines on both faces (with 2 strong mid-palmar spines in the place where tip of nail close); nail long, 0.25× total length of dactylus, 1 seta along anterior margin, with 4 setules at hinge.Gnathopod 2: propodus small (compared to the body) and slightly larger than propodus of gnathopod 1; propodus almond-shaped, palm convex, with cutting margin acanthaceous and as long as posterior margin; posterior margin with 6 sets of stiff setae; antero-distal group of anterior margin with 10 setae; palmar margin with short, notched setae along outer and inner faces, palmar angle undefined, a group of 8 distally-notched strong spines on both faces (with 2 strong mid-palmar spines in the place where tip of nail close); dactylus similar to that of gnathopod 1.

Discussion of affinities
Describing the monotypic genus Anopogammarus Derzhavin (1945) noted a lack of eyes in A. birsteini as the only difference from the genus Gammarus and considered this feature characteristic.Later, Birstein & Ljovuschkin (1970) re-described Anopogammarus birsteini in detail and considered this species, along with Metohia carinata Absolon, 1927, as derived from Gammarus, implying a subgeneric status for the genera Anopogammarus and Metohia.However, according to their view, Zenkevitchia revazi occupies an intermediate position between the specialized Zenkevitchia admirabilis and Gammarus (Stock 1973).Subsequently, Karaman & Barnard (1979) transferred Z. revazi to Anopogammarus, based on the non-moplike structure of maxilla 1 and reduced palps in the former species.Later, the same authors (Barnard & Karaman 1980) again confirmed that Anopogammarus, along with the rest of the taxa placed in the Family group 2 (Typhlogammarus group, hypogean large gammarids) sensu Bousfield (1977), has no strong distinction from the Gammarus-Echinogammarus group (see Barnard & Karaman 1980: 7-9).Ruffo (1995) and Karaman & Ruffo (1995), describing two genera of cavernicolous amphipods (Albanogammarus and Sinogammarus from Albania and Southwest China, respectively), discussed the position of the new taxa in depth and hypothesized an obvious affinity with Anopogammarus.
The revision of the group cannot be considered as completed, because the genus Anopogammarus is heterogeneous and needs to be split.We propose transferring Anopogammarus revazi to the genus Zenkevitchia (Zenkevitchia revazi Birstein & Ljovuschkin, 1970, comb. resurr.)

Remarks
Although, as described below, Zenkevitchia sandroruffoi sp.nov.differs from other species of Zenkevitchia (Z.admirabilis, Z. yakovi) in the non-filtrative maxilla 1, we decided that it, along with Z. revazi, differs sufficiently to warrant recognition as a new species group within the genus Zenkevitchia, which we here designate as the sandroruffoi-group.The sandroruffoi-group (Z.sandroruffoi sp.nov.and Z. revazi Birstein & Ljovuschkin, 1970) is phenotypically more distant from the type species of the genus Zenkevitchia, bearing only up to 22 multi-toothed spines (non-falcate in shape).In some sense, the group occupies an intermediate position between Zenkevitchia (admirabilis-group) and other species of the Typhlogammaridae (cf.Accubogammarus), although, in our opinion, it is closer to the former.The indicators for affinity to Zenkevitchia are: interantennal cephalic lobes sub-acute, urosomal segments weakly armed, shortened antenna 1, antennal gland cone of antenna 2 long, reduced and subsymmetrical palps of maxilla 1, structure and armament of gnathopods (both appendages with welldefined palmar angle).

Etymology
This species is named in honor of Professor Sandro Ruffo (Museo Civico di Storia Naturale, Verona) for his outstanding contribution in the field of Amphipoda systematics.
antennae (Figs 2,.Antenna 1 0.38× of body length; main flagellum with up to 23 articles; each article with 5-9 short setae; peduncular articles in ratio 1 : 0.7 : 0.4; proximal article of peduncle distally with 1 medial set of long setae; accessory flagellum 3-articulated.Length ratio of antenna 1 to antenna 2 is 1 : 0.6; flagellum of antenna 2 with 10 articles, each article densely setose; length ratio of peduncle articles 4 and 5 is 1 : 0.8; flagellum 0.4× longer than peduncle (articles 4+5); peduncular articles 4 and 5 with sets of long, stiff setae on lateral and medial faces; gland cone long, reaching half of peduncle of segment 4. mouth partS.Typical gammarid, except for unusual maxilla 1 (Figs 4H, 8A-E, G-L).Labrum rounded, clypeus unfused, longer than broad.Inner lobes of labium absent, outer lobes broad with stiff curved setae marginally, mandibular process distinct (narrow).Left mandible: incisor with 5 teeth, lacinia mobilis with 4 teeth; between lacinia and molar a row of 14 serrate spines.Right mandible: incisor process with 4 teeth, lacinia mobilis bifurcate, with several small denticles, between lacinia and molar a row of 12 serrate spines; triturative molar process with long lanose seta.Mandibular palp article 2 slightly longer than article 3 (distal); proximal palp article without setae; second article with 12 setae; distal article narrowed, with 1 A group of 2 setae, 2 B setae, 13 D setae and 4 E setae.Maxilla 1 palp reduced, distal article with 4 or 5 apical setae (both palps sub-symmetrical); outer plate with 22 multi-toothed spines; inner plate triangular, with 16 plumose setae.Maxilla 2 inner plate smaller than outer one with, oblique row of 11 plumose setae; both of them apically with numerous setae in two rows.Maxilliped palp article 2 narrow, with about 55 setae along inner margin; article 3 narrow, with 2 dense sets of setae on inner face; article 4 (distal) with dorsal seta, bearing 5 setae at the nail base, nail shorter than pedestal; outer plate with 15 flattened naked spines and 7 long plumose setae on apex (3 of them flattened); inner plate with 3 strong spines (1 supplemental spine on medial face) and 8 stiff, naked setae on apex, 27 plumose setae on ventral face, 10 stiff denticulate setae in 2 rows on dorsal face.Foregut lateralia with 15 strong pectinate spines, with densely setose row of stiff setae.

Variability
Flagellum of antenna 1 with 21-23 articles.Uropod 3 endopodite with 2-3 setae apically.One individual had an abnormal 2-articulated accessory flagellum.The population of Z. sandroruffoi sp.nov.inhabiting streams in the Orlinoe Gnezdo cave differs slightly by having a smaller body size and further shortened antenna 1, but are otherwise indistinguishable.

Ontogenetic variation
Almost all adults presented with significantly threadbare (Fig. 8F) or broken spines on the outer plate of maxilla 1.Thus, the first three of the most powerful spines turned out to be broken off in almost of all of the specimens.

Distribution and ecology
Zenkevitchia sandroruffoi sp.nov.inhabits aquatic biotopes in several caves (Fig. 1) in the Eastern Arabika massif and it apparently avoids deep cave habitats.This species was observed in the Sarma Cave not deeper than -350 m (Fig. 2), in the nearby Trojka cave at a depth of -30 m and in the notfar distant Orlinoe Gnezdo cave at -75 m.The setae on the antennal appendages of all specimens were densely covered (Fig. 7C) with attached filamentous algae (cyanobacteria?).The nature of the morphological variability and divergence of COI gene sequences (pairwise distances <2.2%) suggests that, in the Orlinoe Gnezdo cave, the other population of the same species is present.

Etymology
The generic epithet (Adaugammarus) is derived by the combination of Adau, the name of fearsome malicious giants in Abkhazian mythology, with Gammarus, a closely related genus.Gender masculine.

Discussion of affinities
Adaugammarus gen.nov. is an interesting taxon possessing features typical for a number of groups, but tending more towards Typhlogammarus because of the following characters: lateral cephalic lobe with recess, propodi of both gnathopods with palmar angles undefined, uropod 3 exopodite distinctly broad in males.Adaugammarus gen.nov. is related to the sandroruffoi-group of Zenkevitchia and Typhlogammarus in the similar structure of the outer plate, which has 14 multi-toothed spines.The urosomal segments are armed in a similar manner to those of Zenkevitchia.Adaugammarus gen.nov.differs from Metohia in the absence of dorsal cuticular elements and from Accubogammarus in having a short antennal gland cone, in variramus uropod 3, and differs in having 14 multi-toothed spines on the outer plate of maxilla 1, in contrast to the 38 very densely toothed spines present in Accubogammarus.Superficial comparison with the other "deep dweller", cf.Zenkevitchia sp.sensu Jaume in Sendra & Reboleira (2012) from the Krubera-Voronja cave, revealed that the two taxa differ significantly from each other by the body shape and by the extremely large body size (~19.0-28.0mm in females from Krubera) (pers.obs.).As for the genus.A. pilosus gen.et sp.nov. is readily distinguished from the other cavernicolous typhlogammarids by the sexually dimorphic uropods.See the section 'Discussion of affinities' and the key below.

Etymology
The species epithet (pilosus, Latin), meaning "covered with hairs", refers to the extremely setose uropods in males.
mouth partS (typical gammarid,Figs 4K,13).Labrum rounded, clypeus unfused, longer than broad.Inner lobes of labium absent, outer lobes broad, with stiff curved setae marginally, mandibular process distinct (narrow).Left mandible: incisor with 5 teeth, lacinia mobilis with 4 teeth; between lacinia and molar a row of 11 serrate spines.Right mandible: incisor process with 4 teeth, lacinia mobilis bifurcate, with several small denticles, between lacinia and molar a row of 9 serrate spines; triturative molar process with long plumose seta.Mandibular palp article 2 longer than article 3 (distal); proximal palp article without setae; second article with 21 setae; distal article narrowed, with 2 A groups of 2 and 7  setae, respectively, 2 B groups of 3 and 4 setae, respectively, 13 D setae and 4 E setae.Maxilla 1 palp long, distal article with 1 or 3 apical spines accompanied with setae (both palps asymmetric, right palp broader, with 1 seta on outer margin); outer plate with 14 multi-toothed spines; inner plate trapezoidal, with 10 plumose setae.Maxilla 2 with both plates narrowed, inner plate with oblique row of 8 plumose setae; both of them apically with numerous setae in two rows.Maxilliped palp article 2 broad, with  about 75 setae (some of them in 8 oblique clusters) along inner margin; article 3 narrow, densely setose, with long setae on inner face; article 4 (distal) with dorsal seta, bearing 5 setae at nail base, nail half as long as pedestal; outer plate with 18 flattened, naked spines, 2 long serrate spines and 3 long plumose setae on apex; inner plate with 3 strong spines (1 supplemental spine on medial face) and 13 stiff, naked setae apically and sub-apically, 24 plumose setae on ventral face, 5 stiff denticulate setae on dorsal face.Foregut lateralia with 17 strong pectinate spines, with densely setose row of stiff setae.

Variability
Not observed.

Distribution and ecology
Adaugammarus pilosus gen.et sp.nov.inhabits aquatic biotopes in the Sarma Cave (Figs 1-2) in the eastern Arabika massif (species was observed at depths of -1270 m and -1700 m) and revealed an ability to move in a hygropetric zone.

Phylogenetic analysis
To access the affinities of the newly described taxa and confirm their genetic distinctness, we sequenced a partial mt-cox1 gene from Z. sandroruffoi sp.nov.(6 accessions, Table 1) and Adaugammarus gen.nov.(4 accessions) and assembled a data set including their putative relatives, Accubogammarus, Metohia, Typhlogammarus, and Zenkevitchia spp.(Table 1).Specimens of Zenkevitchia sandroruffoi sp.nov.from the Trojka (-30 m depth) and Sarma caves (-350 m) produced identical sequences (KT427522 and KT427521, respectively), as did two further specimens from the Orlinoe Gnezdo cave (-75 m; KT427523 and KT427524); the intraspecific distance is 0.34 ± 0.17 SE (Table 2).All four specimens of Adaugammarus pilosus gen.et sp.nov.from the deepest part of the Sarma Cave (-1270 to -1700 m; KT427516 -KT427519) were also almost identical in sequence; the intraspecific distance is 0.09 ± 0.09 SE (Table 2).Redundant sequences were excluded from the analyses.The resulting topology (Fig. 3) was generally in agreement with that previously obtained for a smaller data set (Sidorov et al. 2015).Species of Zenkevitchia were members of a clade (81% BP) that also included Accubogammarus and Adaugammarus gen.nov.Topologically, Z. revazi was a basal divergence in this lineage.Adaugammarus gen.nov.was resolved as a sister (96%) to a cluster composed of Z. sandroruffoi sp.nov., Z. admirabilis, and Z. yakovi.The branching pattern among these species remained unresolved.Four highly similar Numbers in brackets correspond to geographic locality on Fig. 1; n/a = not applicable.
(p-distances 0.002-0.01)Z. sandroruffoi formed a robust (100%) clade characterized by a relatively long branch.Accessions from the Sarma and Trojka caves shared two synapomorphic substitutions and were clustered together to the exclusion of specimens from Orlinoe Gnezdo.

Discussion
Brief observation on the vertical distribution of Amphipoda in the Sarma cave An integrated comparative morphological and genetic analysis of the new amphipod taxa in the Sarma Cave revealed their obvious dissimilarity, which in turn allows us to pay attention to the structure of this cave.The system of the Sarma Cave is heterogeneous in its physical structure.Following Burmak (unpubl.)there are three parts: upper, middle and lower.In the upper part, down to -350 m, Z. sandroruffoi sp.nov.was discovered; in the lower part, down to -1700 m, A. pilosus was present.In the middle part amphipods were not detected, but we can't exclude that they were overlooked.In the lower level (from -900 m to -1100 m) in the zone of major faulting leading to the deepest part, the cave abruptly changes direction (see Fig. 2).In this part of the cave thick deposits of red clay are observed.There is also a strong flowing stream in this part of the cave.
Zenkevitchia sandroruffoi sp.nov. is not endemic to the Sarma Cave and was noted in two other caves.
The most remote of the investigated caves, Orlinoe Gnezdo, is located about 3.5 km to the south of Sarma.This suggests that Z. sandroruffoi sp.nov. is vagile, settling in suitable biotopes in the eastern Arabika massif.The affinity of Z. sandroruffoi sp.nov. to the Zenkevitchia lineage and its limited distribution suggest a relatively young phylogenetic age of this species.In contrast, A. pilosus is, in our view, morphologically closer to Typhlogammarus and hence its cohabitation with Z. sandroruffoi sp.nov. in the same cave at different depths is apparently not a result of parapatric speciation.Although there are no reliable data about the geophysical structure and history of the Sarma Cave, it seems that the lower part of the cave belongs to another speleological system, probably older than the overlying cave system.

Table 1 .
List of the specimens, sampling sites and accession numbers of the sequences included in this study.References are given for sequences obtained from GenBank.

Table 2 .
Estimates of pairwise sequence divergence (uncorrected p-distances) of partial mitochondrial COI gene among and within species (bold type) of the Balkan and Caucasian Typhlogammaridae.