AGRAFIA SZAROWSKA ET FALNIOWSKI, 2011 (CAENOGASTROPODA: HYDROBIIDAE) IN THE CAUCASUS

a bStract : Freshwater gastropods of the Caucasus are poorly known. A few minute Belgrandiella -like gastropods were found in three springs in Georgia. Molecular markers: mitochondrial cytochrome oxidase subunit I (COI) and nuclear histone (H3) were used to infer their phylogenetic relationships. The phylogenetic trees placed them most closely to Agrafia from continental Greece. The p-distances indicated that two species occurred in the three localities. Two specimens from Andros Island (Greece) were also assigned to the genus Agrafia . The p-distances between the four taxa, most probably each representing a distinct species, were within the range of 0.026–0.043 for H3, and 0.089–0.118 for COI.


INTRODUCTION
The freshwater gastropods of Georgia, as well as those of all the Caucasus, are still poorly studied; this concerns especially the minute representatives of the Truncatelloidea. About ten species inhabiting karst springs and caves have been found so far (SHadin 1932, 1952, tzvetkov 1940, StaroboGatov 1962, kantor et al. 2010, vinarSki et al. 2014, barJadze et al. 2015, cHertoprud et al. 2016, vinarSki & kantor 2016, Sitnikova et al. 2017. Morphology and anatomy of the soft parts are known only for some of them, not to mention DNA sequences used in phylogeny reconstruction. Belgrandiella A. J. Wagner, 1927 is a group of minute, dioecious, caenogastropod snails, stygobionts, known to inhabit caves and springs of southern Europe, from Spain across Austria and the southern Balkans to the Caucasus and Asia Minor. However, 'Belgrandiella' determined or described in this way include nearly all minute conical or turriform-shelled caenogastropods with a moderately high spire. Anatomically, several genera were recognised within the European 'Belgrandiella ': Litthabitella Boeters, 1970, between closely related congeners are not always to be expected (e.g. radoman 1983). The 'lock-andkey' mechanism needs not to work, especially in molluscs, without any sclerotised structures, and the variation of the reproductive/copulatory organs is striking, although not always noted (faLniowSki 1987, SzarowSka & faLniowSki 2008, faLniowSki & SzarowSka 2011, faLniowSki et al. 2012a). On the other hand, miniaturisation coupled with adaptations to freshwater habitats (osmoregulation, internal fertilisation, embryonic development inside a capsule) has resulted in a simple and uniform anatomy. Thus, molecular data are crucial for reconstruction of phylogenetic relationships within the Truncatelloidea.
Recently a few shells of Belgrandiella-like gastropods were collected in springs in Georgia (Table 1, Fig. 1). In each locality only 1-2 live specimens (including juveniles) were found. Thus, no anatomical examination was possible. The aim of this study was to infer phylogenetic relationships of those snails by means of molecular markers. Additionally, we sequenced one of the two Belgrandiella-like gastropods collected in 2013 on Andros Island, Greece.

MATERIAL AND METHODS
The snails were collected in three localities in Georgia and one in Greece (Andros) ( Table 1, Fig. 1). The Caucasian localities (1-3 in Table 1) were not typical karst springs: 1 -an outflow from a metal pipe well-paved with Lower and Middle Eocene marls and quartzy sandstones and conglomerates, situated in a forest; 2 -small spring at the contact of clastic-limestone with argillites at the bottom and clay shales on top, close to the river bank in a shaded forested place; 3 -two small springs on Upper Cretaceous clastic bedded limestone, artificial wall with metal pipe and a small sedimentary basin, inside a forested area below a walkway. Empty shells and live specimens were obtained from the springs by visual search and washing of vegetation and substratum with the use of either hand-net or metal sieve. The samples were preserved in 70-80 % ethanol and then transferred to 80% ethanol. The shells were photographed with DNA was extracted from foot tissue using a Sherlock extraction kit (A&A Biotechnology) and dissolved in 20 ml of tris-EDTA buffer. Details of PCR conditions, primers used and sequencing were given in SzarowSka et al. (2016). Sequences were initially aligned in MUSCLE (edGar 2004) pro-gramme in MEGA 6 (tamura et al. 2013) and then checked in Bioedit 7.1.3.0 (HaLL 1999). The saturation test (Xia 2000, Xia et al. 2003) was performed using DAMBE (Xia 2013). In the phylogenetic analysis additional sequences from GenBank were used as reference ( Table 2). The data were analysed using approaches based on Bayesian inference and maximum likelihood (ML). We applied the GTR mod- Two simultaneous analyses were performed, each with 10,000,000 generations, with one cold chain and three heated chains, starting from random trees and sampling the trees every 1,000 generations. The first 25% of the trees were discarded as burn-in. The analyses were summarised as a 50% majority-rule tree. The ML approach was applied with RAxML v. 8.0.24 (StamatakiS 2014). One thousand searches were started with initial trees obtained using the randomised stepwise addition maximum parsimony method. The tree with the highest likelihood score was considered as the best representation of phylogeny. Bootstrap support was calculated with 1,000 replicates and summarised on the best ML tree. RAxML analyses were performed using the free computational resource CIPRES Science Gateway (miLLer et al. 2010).

RESULTS
All shells from locality 1 (Figs 2-5), with still incomplete peristomes, were juvenile. Two (Figs 6-7) of the three (Figs 6-8) shells in locality 2 were adult. The single specimen from locality 3 ( Fig. 9) was also juvenile, but the shell habitus differed from one of the other presented shells (narrow spire, broad body whorl) more than could be explained by ontogeny. The shells from Andros Island (Figs 10-11) resembled the ones from locality 2 in Georgia (Figs 6-7), as well as the ones of Agrafia wiktori from northern Greece (Figs 12-13).
We sequenced histone H3 in six specimens (314 bp, GB Accession Numbers MG543153-MG543158) and cytochrome oxidase subunit I (COI; 442 bp: GB Accession Numbers MG543150-MG543152) in three specimens. Both H3 and COI sequences were identical in localities 1 and 2 ( Table 3). The p-distances between the four taxa were within the range of 0.026-0.043 for H3, and 0.089-0.118 for COI (Table  3). For coding of COI, Xia's (2013) saturation test revealed a significant degree of saturation in the third position of the sequences. In the truncatelloids, COI approaches saturation of ca. 18.6% or 120 nucleotide differences (daviS et al. 1998(daviS et al. , wiLke et al. 2000, which seems to happen after approximately 10 million years. However, to avoid a substantial loss of information in the case of closely related species, this position was not excluded from the dataset and was used for the analysis. The maximum likelihood (ML) tree for COI (Fig.  14) placed all the Caucasian and Greek populations within the subfamily Sadlerianinae in one, monoplyletic and well supported clade (bootstrap value 77%), whose sister clade grouped Pontobelgrandiella, Fissuria Boeters, 1981, Avenionia Nicolas, 1882, Alzoniella and Islamia Radoman, 1973, but neither Belgrandiella nor Graziana. The ML tree for H3 (Fig. 15) clearly showed distinctness of the two Caucasian and two Greek (Agrafa and Andros Island) taxa, the mean p-distance between these two subclades being 0.099 for COI and 0.035 for H3. The divergence level between the Caucasian/Greek populations and Belgrandiella was higher, the p-distance was 0.124 for COI and 0.054 for H3. In both COI and H3 trees the snails from the five studied populations formed a monophyletic group, and in both the population from Andros Island clustered between the ones from Georgia and continental Greece.

DISCUSSION
Considering the inferred H3 tree, localities 1 and 2 hold the same taxon, while the taxa at each of the other localities (3-5) are distinct. Most probably, the two Georgian taxa belong to the genus Agrafia, like the snails from Andros Island, and most probably each taxon represents a distinct species. Anyway, they form a distinct clade, not close to Belgrandiella.