Revision of the freshwater genus Atyaephyra (Crustacea, Decapoda, Atyidae) based on morphological and molecular data

Abstract Atyaephyra de Brito Capello, 1867 was described from the Mediterranean region almost 200 years ago. Since then, the genus has been recorded from various freshwater habitats in Europe, North Africa and the Middle East. Despite its long history, the taxonomic status of Atyaephyra species remains confusing and uncertain. Consequently numerous specimens from the known range of Atyaephyra were analysed using morphological characters and mitochondrial COI sequences in an attempt to clarify the taxonomy of this genus. The present study recognises seven Atyaephyra species, more than twice as many as previously recorded (three), four of which are considered as new. The new species are described, additional information to the original descriptions are provided for the remaining three taxa, while neotypes of Atyaephyra desmarestii Millet, 1831 and Atyaephyra stankoi Karaman, 1972 are designated to stabilize their taxonomy. Non-overlapping distinguishing morphological characters are used to discriminate the examined material into five species, e.g., Atyaephyra desmarestii, Atyaephyra stankoi, Atyaephyra orientalis Bouvier, 1913, Atyaephyra thyamisensis sp. n., Atyaephyra strymonensis sp. n. In addition, the genetic analysis supports the existence of multiple phylogenetic clades in the broader Mediterranean area and distinguishes two new cryptic species, namely Atyaephyra tuerkayi sp. n. and Atyaephyra acheronensis sp. n. The geographic distribution of these species is confirmed and their phylogenetic relationships are described.

In the beginning of the 20th century, Bouvier (1913) described two varieties of A. desmarestii: (a) a western variety named A. desmarestii var. occidentalis Bouvier, 1913, distributed in North Africa up to Tunisia, and the entire area of Southern Europe, up to and including Macedonia; (b) an eastern one, A. desmarestii var. orientalis Bouvier, 1913, found in Syria. Fifty years later, these two forms were elevated to subspecies level by Holthuis (1961) and since A. d. var. occidentalis contained the name-bearing type of the species it was re-named to A. d. desmarestii. A third subspecies, A. d. stankoi, was described by Karaman (1972) from Doirani Lake which is situated at the borders between Greece and Former Yugoslav Republic of Macedonia (F.Y.R.O.M.). Finally, Al-Adhub (1987) described A. d. mesopotamica from Shatt Al-Arab River and Hammar Lake (Iraq) thus increasing the number of subspecies to four.
Subsequent studies (Gorgin 1996, Anastasiadou et al. 2004) questioned the validity of these four subspecies based on the observed overlapping in the key characters used to separate them. However, Anastasiadou et al. (2004) stated that given the wide distribution of this species and the degree of isolation of its populations it is likely that a detailed examination of other morphological features could reveal real differences among the various populations of this species.
After examining two mitochondrial genes (COI, 16S) from specimens collected mainly from the western Mediterranean area, Garcia Muñoz et al. (2009) proposed the existence of two species: A. desmarestii, distributed in West Europe and North Africa and A. stankoi Karaman, 1972 distributed in Greek freshwaters which was elevated from the subspecies to the species level. Furthermore, the authors argued about the existence of a third genetically distinguished group, A. mesopotamica Al-Adhub, 1987(or A. orientalis Bouvier, 1913, without confirming its status as a distinct species. In addition, they synonomised A. rosiana, as described by Anastasiadou et al. (2008), with A. desmarestii. The species A. stankoi was characterized as cryptic since previous studies failed to detect any distinguishing morphological characters (Anastasiadou et al. 2004) that would enable its discrimination from the A. desmarestii complex (Garcia Muñoz et al. 2009).
A comprehensive revision of synonyms of the Atyaephyra, at species level, has been provided by De Grave and Fransen (2011) while a list of synonyms at genus level is given by Holthuis (1993).
This eventful taxonomic history, and the high intra-and inter-specific morphological variability observed among the Atyaephyra taxa make the recognition of discrete species intricate. Also, the wide distribution of the genus and the apparent isolation between populations may support the existence of new non-described species. Therefore the lack of any study including material covering all the known distribution of the genus provoked the present current multidisciplinary study.
In an attempt to recognize and delimit species within Atyaephyra, samples covering the known distribution of the genus were analysed, using morphological and molecular methods to evaluate the consensus of groupings as inferred by both datasets. In the last decade molecular data have been widely used in conjunction with decapod morphology, and have been instrumental in discriminating cryptic or sibling species (e.g. Macpherson and Machordom 2005, Jesse et al. 2010. This study specifically aims to: (a) test the status of the species already recognized based on morphological and molecular data; (b) describe new species based on morphological and molecular data; (c) provide knowledge on the current geographic distribution of the Atyaephyra species; (d) describe the phylogenetic relationships of new and previously described species based on COI gene.

Morphological analyses
Specimens were collected with a hand dredge over the period 2000-2012 from numerous river catchments in Greece, while additional material from the rest of the Mediterranean region was either offered or loaned by researchers and Museum collections. Samples were loaned or offered from the following museums: NHM, NMW, MNHN, MMNH, ZMAUTH, OUMNH and SMF. In total 1,082 adult individuals (A. acheronensis sp. n.: 4, A. desmarestii: 431, A. thyamisensis sp. n.: 194, A. orientalis: 111, A. stankoi: 106, A. strymonensis sp. n.: 92, A. tuerkayi sp. n.: 2; furthermore 112 and 30 additional individuals were examined pending their assignment to A. acheronensis and A. tuerkayi respectively) were examined from 122 different stations (49 river basins, 20 countries) spanning throughout the known distribution of the genus Atyaephyra from Middle East to North Africa and Europe (Fig. 1). Part of this examined material has been included in the studies of Kinzelbach and Koster (1987) and Anastasiadou et al. (2004Anastasiadou et al. ( , 2006Anastasiadou et al. ( , 2008. A total of 135 morphological characters including 68 somatometric distances were analysed (see Appendix: Table 1). Morphometric measurements were taken using a Carl Zeiss standard trinocular microscope or an Olympus VM stereoscope both with ocular micrometer. Only adult individuals were taken into account in order to exclude deviations in the features which appear in the juvenile individuals. A threshold of CL ≥ 5 mm was set for all the specimens examined except for those belonging to A. orientalis for which the threshold was set to CL ≥ 3.8 mm. The threshold corresponds to the smaller ovigerous individual found. Atyaephyra orientalis is of smaller size and thus the threshold must be lower than in the other species. Drawings were made based on photos taken which were subsequently digitized and processed with CorelDRAW® Graphics Suite X5.

Electronic publication
All data (e.g. taxon descriptions, figures, characters measured) underlying this publication can also be accessed on Atyaephyra Scratchpad (http://atyaephyra.myspecies. info/). Scratchpads (http://scratchpads.eu) is a Virtual Research Environment, that enable taxonomists to collaborate in the production of websites documenting the diversity of life (Blagoderov et al. 2010).

Molecular analyses DNA extraction, amplification and sequencing
Genomic DNA was extracted exclusively from abdominal tissue using ammonium acetate protocol (provided by Poulakakis N, NHMC, University of Crete, Greece). Abdominal tissue was dissolved in 600μl extraction buffer (0.05M Tris-HCl pH 7.5, 1mM EDTA pH 8.0, 0.15M NaCl, 0.3% sodium dodecyl sulfate, and 0.6μg/μl proteinase K) and incubated in a shaking waterbath at 56°C overnight. Following the incubation, 340μl of 4M ammonium acetate were added to each sample and incubated at room temperature for 60 min. Samples were mixed several times during this period by inversion. The solution was centrifuged at 18,000g for 20 min and supernatant was transferred to 2.0ml centrifuge tubes and 1ml of absolute ethanol was added to each sample. The tubes were inverted several times and centrifuged at 18,000g for 30 min. Following the removal of ethanol samples were dried overnight. DNA pellet was diluted by adding 50μl ddH 2 O and incubated at 4°C overnight. A fragment of the 5' region of mitochondrial (mtDNA) cytochrome c oxidase subunit I (COI) gene was amplified using the polymerase chain reaction (PCR). Two pairs of primers were used for each DNA extract, following the technique of nested PCR. Different combinations of primers were used as first pair: (a) LCO-1490 (5'-GGTCAACAAATCATAAAGATATTGG-3';Folmer et al. 1994) and HCO-2198 (5'-TAAACTTCAGGGTGACCAAAAAAT-CA-3';Folmer et al. 1994); (b) LCO-1490 and C1-N-2191 (5'-CCCGGTAAAAT-TAAAATATAAACTTC-3'; Simon et al. 1994); (c) Pals-COI-F1 (5'-GAGCTGAAC-TAGGTCAACC-3', designed on Palaemoninae sequences) and HCO-2198 specifying a ~700 bp to ~600 bp fragment of the COI gene. Thermocycling was performed with an initial denaturation step of two min at 94°C; followed by 35 cycles of one min at 94°C, one min at 42-52°C (depending on the primer pair used), and one min at 72°C, with a final extension of 72°C for 10 min. Then, the primary PCR product was directly used for another amplification reaction, without further purification, using two different combinations of primers as second pair: (a) the newly designed Pals-COI-F1 and Pals-COI-R1 (5'-AGTATAGTAATAGCTCCAGC-3', designed on Palaemoninae sequences) and (b) C1-J-1718, (5'-GGAGGATTTGGAAATTGATTAGTTCC-3'; Simon et al. 1994) and Pals-COI-R1 which amplified a ~450 bp and ~330 bp fragment respectively. The thermal profile for the secondary amplification reaction was the same as that of the primary amplification reaction. All amplification reactions were carried out in a final volume of 20μl. Each reaction contained 1.0μl template DNA, 0.15μM of each primer, 0.15mM dNTPs, 1.5mM or 3mM MgCl2 (depending on the primer pair used), 1X PCR reaction buffer, and 0.5U Taq (Gennaxon).
In some cases after the nested PCR a re-amplification was made using a modified Band-stab PCR protocol (Bjourson and Cooper 1992). The re-amplification reaction was carried out in a final volume of 50μl containing: 0.1μM of each primer, 0.08mM dNTPs, 1mM MgCl2, 1X PCR reaction buffer, and 1.25U Taq (Gennaxon). After an initial denaturation step of two min at 94°C, 25 cycles of one min at 94°C, one min at 45°C, and one min at 72°C were performed, followed by a final extension of five min at 72°C. The amplified fragments were then purified using ethanol and sodium acetate precipitation method and sequenced using Big Dye Terminator Cycle Sequencing 3.1 (Applied Biosystems) standard protocol on an ABI 3730 Genetic Analyzer (Applied Biosystems). All individuals were sequenced either with the forward or the reverse COI primer or with both (Pals-COI-F1, Pals-COI-R1).

Alignment and genetic divergence
Thirty-seven new COI sequences were generated (GenBank accession numbers JX289898-JX289919, JX289921-JX289933, JX289935-JX289936; Table 1). Our dataset was supplemented with eight COI sequences of Atyaephyra from the study of Garcia Muñoz et al. (2009), one from Franjević et al. (2010), one from Zakšek et al. (2007) and four from Page et al. (unpublished data). Furthermore, three COI sequences (Page et al. 2005a, Zakšek et al. 2007, Garcia Muñoz et al. 2009) from another two atyid genera, were included as outgroups (i.e. Dugastella valentina (Ferrer Galdiano, 1924) from Spain, Dugastella marocana Bouvier, 1912 from Morocco, and Paratya curvirostris (Heller, 1862) from New Zealand, accession numbers provided in Table 1). The choice of the taxa used as outgroup was based on their close relationship with the genus under study since they all belong to the same atyid group (Paratya group) (Von Rintelen et al. 2012).
COI sequences were aligned using FSA (Fast Statistical Alignment) (Bradley et al. 2009) and translated into amino acids prior to analysis, to ensure that no gaps or stop codons were present in the alignment. The number of distinct haplotypes was estimated with the software Arlequin version 3.5.1.3 (Excoffier and Lischer 2010). jMod-elTest (Posada 2008) was used to determine the model of DNA sequence evolution that best fit the data using AIC and BIC criteria. Sequence divergences were estimated with the software MEGA version 5.1 (Tamura et al. 2011).

Phylogenetic analyses
Phylogenetic inference analyses were conducted using Neighbor Joining (NJ), Maximum Likelihood (ML), and Bayesian Inference (BI) methods. The nucleotide substitution model selected by jModeltest [Tamura-Nei, 1993 (TrN) + gamma (G)] was applied to the data matrix in all analyses. A NJ tree was produced with the software MEGA where branch support was assessed with 1,000 bootstrap replicates. ML estimates were made using PhyML online web server (Guindon et al. 2010; http://www.atgc-montpellier.fr/phyml/). Nearest neighbor interchanges (NNIs) and subtree pruning and regrafting (SPR) topological moves were used to explore the space of tree topologies. Approximate likelihood-ratio test (aLRT) based on a non-parametric Shimodaira-Hasegawalike (SH-like) procedure was employed to estimate branch support (Guindon et al.  Atyaephyra specimens and COI sequences accession numbers listed by area and species. The sex and the CL are given for each specimen sequenced in parenthesis (first column). Museum accession numbers are given in parentheses (second column). GenBank accession numbers of published sequences, used in this study, are provided with their corresponding studies indicated by the letters a-e [a: Garcia Muñoz et al. 2009, b: Page et al. (unpub sequences), c: Franjević et al. 2010, d: Zakšek et al. 2007, e: Page et al. 2005a].

Specimen
Sampling site 2010). BI analysis was performed in BEAST version 1.7.2. (Drummond et al. 2012) assuming an uncorrelated lognormal relaxed-clock model, setting the tree prior to Yule process, run for 100,000,000 generations (10% was discarded as burn-in period). Finally, TreeAnnotator was used to find the Maximum Clade Credibility tree. In order to show the geographic distribution of the distinct haplotypes, in all the analyses, not only the unique haplotypes were used, but all the sequences acquired.

Phylogenetic analyses
Out of the 51 Atyaephyra COI sequences 35 distinct haplotypes were distinguished. Shared haplotypes were observed among individuals in close geographical proximity. Of the 600 nucleotide sites examined, 237 were variable of which 197 were parsimony informative (14% in the first, 2% in second, and 84% in third codon position). The nucleotide substitution model that best fits our data according to both AIC and BIC criteria is Tamura and Nei (1993) + gamma (G) based on which Atyaephyra sequence divergence ranged from 0% to 25.7%. All employed methods yielded consistent tree topologies (Fig. 2). The monophyly of the genus is highly supported in all methodologies (BI posterior probability: 1.0, ML SH-like value: 96, NJ bootstrap value: 95).
In all phylogenetic analyses four main and well-supported phylogroups were identified, corresponding to different groups of species designated by morphology (presented in the next section) and/or well defined geographic regions throughout the Mediterranean region (Fig. 2). The first phylogroup comprises specimens from the Middle East which were classified to the nominal species, A. orientalis by morphology. Specimens from the topotypical populations of the subspecies A. d. orientalis (Orontes River, Syria) and A. d. mesopotamica (Shatt Al-Arab River, Iraq) were also included. However, present data do not allow for within clade fine scale resolution. The mean genetic distances between the Middle East phylogroup (A. orientalis) and the other groups/subgroups were very high ranging from 18.7% to 24.5% while the average intraspecific distance was 5.8% (Table 2).
The second phylogroup which is strongly supported by both BI and ML methodologies while in NJ yielded lower bootstrap values (BI posterior probability: 0.99, ML SH-like value: 94, NJ bootstrap value: 65) includes sequences exclusively from Greek table 2. Nucleotide mean distances (% Tamura-Nei 1993 + G model) of cytochrome c oxidase I (COI) within (first column) and among the Atyaephyra species.
The range of pairwise distances is given in parenthesis. populations. The Greek phylogroup is further subdivided into three well supported groups. The first subgroup corresponds to the nominal species, A. stankoi, found in West-central Greece. It is worth noticing that specimens from the type locality (Doirani Lake) of A. d. stankoi are also included. The remaining Greek specimens are grouped in two well defined subgroups, one distributed in North-east Greece while the other is located in West Greece (Fig. 1). The mean genetic divergence among the three subgroups ranges from 11.9% to 18.2%, while the mean genetic distances within subgroups varied from 0% to 2.4% ( Table 2). The third phylogroup contains specimens from the Syrian River Nahr Al-Kabir and it is strongly supported in all methodologies (BI posterior probability: 0.99, ML SH-like value: 100, NJ bootstrap value: 100). The mean genetic distances between the Syrian subgroup and the other groups/subgroups were very high ranging from 19.7% to 25.7% ( Table 2).

Within species
The fourth phylogroup which is well supported by BI, ML and NJ (BI posterior probability: 0.97, ML SH-like value: 99, NJ bootstrap value: 100) includes specimens originating from West-central Europe, North Africa and the Balkans. Within this phylogroup, specimens from Croatia, Slovenia and Greece form a distinct highly supported subgroup (BI posterior probability: 0.99, ML SH-like value: 100, NJ bootstrap value: 99). The remaining specimens within the phylogroup i.e. specimens from West-central Europe and North Africa, although classified as A. desmarestii (nominal species) by morphology (discussed in the next section) do not constitute a well supported subgroup except in NJ analysis where it is relatively well supported (NJ bootstrap value: 89). Sequences from the topotypical populations of the A. desmarestii (Mayenne and Sarthe River), and A. rosiana described by de Brito Capello (Ceira River, tributary of Mondego River) were included in this subgroup as well as a sequence acquired from river Bordeira (Portugal) which is near to São Barnabé River from where A. rosiana was re-described by Anastasiadou et al. (2008). The genetic distances between these two subgroups are quite large, ranging from 5.9% to 11.6% ( Table 2). The lowest values (5.9-6.8%) were observed between the specimens of the Balkan subgroup and those of South Iberian Peninsula and North Africa (Morocco), located in the distant end of A. desmarestii distribution. On the contrary higher values (7.5-10.2%) were observed between the nearest to the Balkan subgroup populations (e.g. Danube River) as well as between the topotypical populations of A. desmarestii (Mayenne and Sarthe River) and the Balkan populations. Furthermore, no haplotypes were shared between these two subgroupings.
Distribution. Atyaephyra desmarestii is found in freshwater habitats of North Africa and West-central Europe (see material examined and Fig. 1).

Remarks.
A. desmarestii has been exhaustively described and illustrated by Anastasiadou et al. (2006). Anastasiadou et al. (2008) also re-established and redescribed in detail A. rosiana, a species currently considered as a synonym of A. desmarestii. In the present paper the same material used for the redescriptions of A. desmarestii and of A. rosiana (Anastasiadou et al. 2006(Anastasiadou et al. , 2008 was examined. Although Anastasiadou et al. (2006) stated that the "holotype" of A. desmarestii could not be traced in French institutions, Bouvier (1913) clearly stated that he examined material from "Maineet-Loire (H. Milne Edwards, probablement des cotypes de Millet)". As Millet and H. Milne Edward were contemporary, and it seemed possible that H. Edwards may have asked for some specimens from the MNHN, this material was recently looked for in the MNHN collection, where the material listed in Bouvier (1913) is indeed still present (registration number Na480). However, there appears to be a discrepancy (and thus possible clarification) on the actual specimen label to this information. The specimen label (see Appendix: Fig. 3) provides the following information: (1) "Maine et Loire", (2) "Caridina Desmarestii Millet", (3) "A. Milne Edwards det.", (4) "E.L. Bouvier ver. 1899" and (5) "A. Milne Edwards, 1900". It is difficult to definitively interpret the label information in view of what Bouvier (1913), a contemporary of A. Milne-Edwards, wrote, as he may have had access to direct, personal information. However, the sample is herein interpreted as having belonged to the A. Milne- Edwards collection, who died in 1900Edwards collection, who died in (1835Edwards collection, who died in -1900 and was then accessioned in the museum collection (label item 5), with the material being examined and verified, i.e "ver." in 1899, by Bouvier (label item 4), but that the material originally was identified by A. Milne Edwards (label item 3), and that the material may not have been seen by H. Milne Edwards (although it may have passed from father to son without being recorded as such on the museum labels). It seems, therefore, impossible to certify that these are indeed syntypic specimens of Hippolyte Desmarestii Millet, 1831, as indicated by Bouvier (1913). However, in deference to Bouvier's potential knowledge on the matter and in line with Recommendation 75A (ICZN, 1999), a neotype for A. desmarestii is herein selected from this lot, the largest ovigerous female. The designation of a neotype is deemed justified under Art. 75 (ICZN, 1999), as (1) the taxon is involved in a complex nomenclatorial problem which cannot be solved without fixing the identity of the oldest name; (2) the taxon is differentiated from the other taxa in this complex by having 0-8 mesial spines on terminal segment of third maxilliped, the basal endite of first maxilliped clearly reaching beyond distal end of exopod, having 1-5 post orbital rostral teeth, having a not protruding, rounded pterygostomial angle and by the slightly curved endopod of first male pleopod with its distal part elongated and tapering; (3) the selected specimen is the largest (of only two) ovigerous females in lot MNHN-Na480; (4) the reasons the name-bearing types are considered lost (or the contrary cannot be conclusively proven) are given above (see also Anastasiadou et al. 2006); (5) the neotype is from the general locality (Maine et Loire) of the type locality of A. desmarestii from which no other species is known and thus it corresponds morphologically and genetically with data presented herein and in Anastasiadou et al. (2006); (6) the neotype is selected from the "Maine et Loire" sample in Bouvier (1913), corresponding to the area mentioned in Millet (1831); and (7) the neotype has been selected from a sample already belonging to MNHN (Na480). Therefore, all conditions of Art. 75 are considered to be met and the selection of neotype is justified.

Size.
A. orientalis is a small-medium sized species of Atyaephyra, with maximum carapace length to be 4.8 mm in ♂♂, 6.8 mm in ♀♀ and 5.5 mm in ovig. ♀♀. Molecular characters. A. orientalis can be differentiated from all other species of Atyaephyra by molecular characters, as demonstrated by the phylogenetic analysis of mtDNA COI sequences. Additionally, 5 haplotypes, each from a different location, found in A.orientalis were not shared by any other species of the genus. It also differs from all the other species in the following nucleotide positions in the COI gene of A. desmarestii specimen Dour1, position 273: guanine (G), position 276: guanine (G) and position 369: cytosine (C).
Distribution. Atyaephyra orientalis is found in freshwater habitats of Middle East, from Turkey to Iraq (see material examined and Fig. 1). Remarks. Bouvier (1913) after examining the Atyaephyra material deposited in the MNHN collections he assigned it into two varieties (A. d. var. orientalis and A. d. occidentalis) based mainly on differences observed in the endopod of first male pleopod. A. d. var. orientalis was originally described from Syria (from Orontes River, near the Lake Qattinah (Lake Homs), from a stream in Kousseir (probably Qoussair) near Damascus and from Barada River, Ataibe, East of Damascus) and was elevated to subspecies level by Holthuis (1961). Apart from A. d orientalis, a second subspecies, A. d. mesopotamica, was found to exist in the Middle East and was described by Al-Adhub (1987). Al-Adhub (1987) described the new subspecies based on the presence of a distinct subterminal process (vs. absent from A. d. orientalis and A. d. desmarestii) and the presence of 50 spines on dactylus of fifth pereiopod (vs. 40 in A. d. orientalis and A. d. desmarestii). Furthermore he noticed that the rostrum of A. d. mesopotamica resembles that of A. d. desmarestii from Greece but differs in having the distal ventral part always devoid of teeth. Indeed the individuals from Shatt Al-Arab River had the highest number of spines on dactylus of fifth pereiopod ranging from 41-55 but specimens from the River Orontes were also found with up to 47 spines (33-47). Additionally, male individuals having endopod with a distinct subterminal process were found again in River Orontes as well as in other Middle East Rivers. Gorgin (1996), after studying 150 males from two different localities in Iran found individuals with a distinct subterminal process and without inside the same population. Finally, specimens from Greece belonging to A. stankoi (as the sample of Holthuis to which Al-Adhub refers to) were found to be also devoid of teeth in the distal part of the rostrum. Even in the illustration included in Holthuis (1961) work, the Greek specimen is devoid of teeth in the distal part of the ventral margin. Although the genetic distances within the A. orientalis phylogroup were high (0.9%-10.2%) no firm conclusion could be drawn whether the hypothesis of multiple species is valid or not. Sequences from Orontes River (topotypical location of A. d. orientalis) and from Shatt Al-Arab River (topotypical location of A. d. mesopotamica) presented a noticeable mean genetic divergence (5.0%) but still not strong enough to support the hypothesis of different species. Detailed future studies on the morphological and genetic variability within the Atyaephyra distributed throughout the Middle East will help clarify the relationships between the populations in this region. However, only one species is currently considered to exist, A. orientalis. Therefore, A. d. mesopotamica is here proposed as a synonym.
Remarks. Bouvier (1913) assigned the material of MNHN originating from Portugal, France, Corsica, Macedonia, Tunisia, Algeria and Morocco to var. occidentalis while the material from Syria he assigned to var. orientalis. The material from Macedonia was collected from the region of Vardar (Axios) north of Thessaloniki, from the Lake of Amatovo (drained in the early twentieth century) near Kirdzalar (today called Adendron). The two varieties described by Bouvier were elevated in subspecies level by Holthuis (1961) and the var occidentalis was re-named to A. desmaresii desmarestii since it contained the name-bearing type of the species. Few years later, Karaman (1972) described a new subspecies from Doirani Lake which is part of the Vardar (Axios) basin and named it A. desmarestii stankoi ignoring the available name of Bouvier's (A. d. var. occidentalis). However, after designating a neotype of A. desmarestii from Bouvier's material the nomen A. d. var. occidentalis becomes unavailable since it becomes a junior synonym of A. desmarestii (see A. desmarestii remarks) and thus the nomen A. stankoi can be used for the Macedonian taxon (as used herein).
Efforts made to trace Karaman's type material in the MMNH were unsuccessful. According to the director of the Museum, Dr Petkovski S. (pers. comm.), Karaman's material is considered lost after a fire that took place in the Museum.
A neotype for A. stankoi is proposed for reasons of taxonomic clarification and stability, as foreseen by Art. 75 (ICZN, 1999). The neotype will contribute to the stability of the taxonomic status of the species and avoid further confusion due to nomenclature (see also A. desmarestii remarks). Furthermore, it incorporates novel characteristics that distinguish it from the remaining Atyaephyra species such as: having 11-35 mesial spines on terminal segment of third maxilliped, basal endite of first maxilliped failing or reaching to distal end of exopod, distal boarder of telson with spines arranged in a fork-like pattern, a rounded antennular lobe, a pterygostomial angle not protruding, and a slightly curved and distally more or less elongated but always tapering endopod of male first pleopod. The name-bearing types are considered lost while the neotype has been collected from Doirani Lake, the same locality from where Karaman (1972) collected A. d. stankoi type material and it will replace the lost type material.
Size. Atyaephyra thyamisensis sp. n. is a large sized species with a maximum carapace length of 6.4 mm in ♂♂, 8.0 mm in ♀♀ and 8.1 mm in ovig. ♀♀. Molecular characters. A. thyamisensis sp. n. is different from all the other species of Atyaephyra by molecular characters, as shown by the phylogenetic analysis of mtD-NA COI sequences. The one haplotype found was unique in the genus. Furthermore, Etymology: Atyaephyra thyamisensis sp. n. is named after the Thyamis River, Greece, the type locality.
Distribution. Atyaephyra thyamisensis sp. n. is found in fresh water habitats of Northwest Greece as well as in the islands Corfu and Lefkada (see material examined and Fig. 1).
Remarks: A. thyamisensis can be discriminated from A. stankoi by the presence of a sharply protruding pterygostomial angle (Fig. 7B). It should be noted that this character has been observed to be missing from one side (either the left or the right) in some very large sized individuals (Fig. 7C). This character is shared by A. orientalis  inner margin and 9-15 setae arranged on outer margin (Fig. 10I). 210-250 eggs of 0.50-0.70 × 0.40-0.50 mm in size.
Size. Atyaephyra strymonensis sp. n. is a large sized species with maximum carapace length to be 5.6 mm in ♂♂, 7.9 mm in ♀♀ and 7.5 mm in ovig. ♀♀. Molecular characters. Atyaephyra strymonensis sp. n. is unique in the genus in having 2 haplotypes not found in any of the other species. Also, it differs from all the other species in the following nucleotide positions in the COI gene of A. desmarestii specimen Dour1 Etymology: Atyaephyra strymonensis sp. n. is named after the Strymon (Strymonas) River, Greece, the type locality.
Distribution. Atyaephyra strymonensis sp. n. is found in North-western Greece in the Rivers Strymon and Nestos (see material examined and Fig. 1).
Size. Atyaephyra acheronensis sp. n. is a large sized species with maximum carapace length to be 5.1 mm in ♂♂, 7.6 mm in ♀♀ and 7.0 mm in ovig. ♀♀. Molecular characters. Molecular information based on the COI sequences provides compelling evidence that is a well defined species. Atyaephyra acheronensis sp. n. is unique in Atyaephyra in having 2 haplotypes not shared by any other species. Furthermore, it differs from all its congeners in the following nucleotide positions in the COI gene of A. desmarestii specimen Dour1, position 255: adenine (A) and position 318: cytosine (C). Finally, the mean genetic distances between A. acheronensis and the remaining Atyaephyra species range from 8.3% to 23.8% (Table 2).
Distribution. Atyaephyra acheronensis sp. n. is found in freshwater habitats of Croatia (Krka River), Slovenia (Dragonja River) and Greece (Acherontas River and Louros River) (see material examined and Fig. 1). Although this study was based on a limited  19-23 pre orbital teeth on dorsal margin of rostrum arranged up to tip. With two post orbital teeth and 4-7 teeth on ventral margin of rostrum (Fig. 13A). Carapace smooth with pterygostomial angle not protruding, rounded (Fig. 13B). Pleuron of fifth abdominal segment pointed with an acute posterior angle (Fig. 13C). Telson with four pairs of dorsal spines arranged in curved fashion (Fig. 13D). Distal border of telson with 9 spines (5 pairs) arranged in fan-like pattern. Outermost pair of spines shortest, similar to dorsal spines, adjacent pair stronger terminating before the finely setulose, inner pairs (Fig. 13E). Antennulary stylocerite with its tip failing to reach or reaching distal margin of basal peduncle segment. Anterolateral lobe of basal segment short and round (Fig. 13G). Distal segment of antennular peduncle with 1-2 spines (Fig.  13F). Basal lower endite of maxilla densely covered with long simple setae arranged in 18-20 oblique parallel rows. Endite of maxilla 1.58-1.59 × as long as basal lower endite (Fig. 14G). Basal endite of first maxilliped reaching clearly beyond distal end of exopod (Fig. 14F). Distal one-third of terminal segment of third maxilliped bearing 1-6 mesial spines and one subdistal lateral spine near the base of larger terminal spine ( Fig 14H). Armature along flexor margin of dactylus of third and fourth pereiopod consisting of 6-7 and 6-7 spines respectively (Figs 14B, 14D). Merus of third and fourth pereiopod with 4 and 3 spines respectively (Figs 14A, 14D). Armature along flexor margin of dactylus of fifth pereiopod consisting of 28 spines (Fig. 14E).
Size. Atyaephyra tuerkayi is a large sized species with maximum carapace length to be 7.1 mm for ♀♀  (Table 2).

Molecular characters. A haplotype found in
Etymology. Atyaephyra tuerkayi sp. n. is named after Professor Michael Türkay, in appreciation of his contribution to the study of Decapoda.
Distribution. Atyaephyra tuerkayi sp. n. is found in the Nahr Al-Kabir River situated between Syria and Lebanon (see material examined and Fig. 1).
Remarks. In addition to the type-material we investigated the morphology of the 23 female individuals (6 ovig.) and 7 males originating from Nahr Al-Kabir River (Fig. 1, stn 122; SMF 12189, SMF 12191, SMF 12192). All the individuals examined (including the sequenced ones) were morphologically identical. However, their placement to A. tuerkayi, sp. n. has still to await sequencing. Since no male or ovigerous individual was sequenced observation regarding the form of the endopod of first male pleopod and number of eggs carried by the female were not included in the description. But observations were made in other individuals of the same sample and population and thus given here: endopod of first male pleopod expanded proximally and with a distal portion elongated and tapering, endopod with 9-16 spines arranged on a slightly curved inner margin and 9-11 setae arranged on outer margin. 430-450 eggs of 0.45-0.50 × 0.30-0.35 mm in size. Maximum carapace length to be 5.7 mm for ♂♂, 7.9 mm for ♀♀ and 7.6 mm for ovig. ♀♀.

Discussion
Given the highly structured nature of freshwater habitats and the limited potential for dispersal of the freshwater species (mainly due to natural barriers) in combination with the wide distribution of Atyaephyra in the Mediterranean region, a hypothesis under which several species are expected to be harbored in the genus seemed highly possible.
However, until recently, Atyaephyra was considered as a monotypic genus. Over the last 100 years many authors (Bouvier 1913, Holthuis 1961, Karaman 1972, Al-Adhub 1987 have attempted to challenge this perception. However, the high intra-and inter-population variability, which made even the previously proposed subspecies questionable (Gorgin 1996, Anastasiadou et al. 2004) along with the lack of a complete series of samples covering all the known distribution of Atyaephyra, proved to be far more challenging than many taxonomists would ever anticipate.
In the latest revision of the Atyaephyra (Garcia Muñoz et al. 2009), which was based on the genetic information deriving from two mitochondrial genes (COI, 16S), two species were recognized while a third was proposed but without confirming it. In the current study seven species are defined, based both on morphological and molecular data. This difference in numbers is attributed to the limited geographical focus of the former study, which was primarily carried out on material collected from the Western Mediterranean area.
After an exhaustive study of a large number of specimens from 20 different countries and a thorough examination of more than 135 morphological characters, including somatometric distances, new characters were found which could differentiate species or groups of species within the Atyaephyra. One of these characters is the number of mesial spines on the terminal segment of the third maxilliped according to which two main groups can be distinguished. The first group is characterized by 10-38 mesial spines and comprises three species, A. thyamisensis sp. n., A. stankoi, A. orientalis whereas the second by 1-8 mesial spines including the remaining four, namely A. desmarestii, A. acheronensis sp. n., A. strymonensis sp. n. and A. tuerkayi sp. n.
The species included in the first group can subsequently be distinguished by a series of features, e.g. presence-absence of a protruding pterygostomial angle, shape of antennular lobe and shape of endopod of first male pleopod. Atyaephyra thyamisensis sp. n., A. stankoi and A. orientalis are morphologically and phylogenetically well defined. In the phylogenetic tree they represent three well supported clades (16.7%-22.6% divergent from each other). In the second group, A. strymonensis sp. n. is also a well defined species morphologically and can be distinguished from the remaining members by a combination of characters such as the lack of post orbital teeth, presence of a short unarmed proximal gap on rostrum and ratio of basal lower endite of maxilla in relation to the whole maxilla endite. The genetic divergence observed between A. strymonensis sp. n. and its closest congeners by morphology is quite high (21.9%-25.4%). Thus, both morphological and molecular data show congruent patterns and jointly support its recognition as a distinct species within the genus. In addition, although A. strymonensis sp. n. seems to be morphologically closer to the members of the second group e.g. A. desmarestii, A. acheronensis sp. n., A. tuerkayi sp. n., genetically it is more closely related to the other two species of the first group from Greece (e.g. A. thyamisensis sp. n. and A. stankoi) with which it forms a strongly supported phylogroup (genetic divergence range: 11.9%-18.2%).
No diagnostic morphological characters were found to distinguish the species A. desmarestii, A. acheronensis sp. n. and A. tuerkayi sp. n. from each other, a fact which is mainly caused by the high morphological variability observed in A. desmarestii. However, their genetic distinctiveness coupled with their discrete geographical distribution provides enough evidence to distinguish the three species as distinct taxa.
The range of genetic divergence observed between the specimens of A. desmarestii and of A. acheronensis sp. n. (TrN distances: 5.9%-11.6%, Uncorrected p-distances: 5.3%-8.7%) is comparable to those found for other cryptic or sibling species of freshwater shrimps (e.g. Page et al. 2005a, Uncorrected p-distances: Caridina sp. A vs Caridina sp. B or C: 8.4-10.9%; Caridina sp. B vs Caridina sp. C: 6.7-8.8%), freshwater crabs (e.g. Jesse et al. 2011, Uncorected p-distances: interspecific variability between 14 Potamon species range: 3.1%-11.2%) as well as for other decapod sibling or well defined species (e.g. Jones and Macpherson 2007, TrN distances: interspecific variability between 14 Munidopsis species range: 1.5%-19.6%). The mean genetic divergence observed between A. desmarestii and A. acheronensis (8.3 %) was the smallest among the Atyaephyra species (remaining genetic distances ranging from 11.9 to 25.7%). This level of divergence was also evident in morphology, indicating a more recent speciation event within the genus (compared to the ones that gave rise to the other species of Atyaephyra) and thus less time for these two species to diverge both morphologically and genetically.
Furthermore, the fact that no haplotypes were shared between A. desmarestii and A. acheronensis sp. n. would suggest that the populations of shrimps from both species, although recently evolved, had independent evolutionary histories for a relatively long period of time. Additional support, although further research is still needed, comes from their geographical distribution since A. desmarestii and A. acheronensis sp. n. seems to be allopatric. Atyaephyra acheronensis is found in the western Balkan Peninsula, ranging from Croatia to Greece. In Greece, this species is found only on the west side of the mainland reaching most probably as far as South Peloponnese although with a remarkable fragmented distribution. In comparison A. desmarestii is distributed in West-central Europe and North Africa. It should be noted here that the native distribution of A. desmarestii is limited to Southern Europe and its presence in North-Central Europe up to the Danube River is believed to have been caused by its dispersal through the canals that were opened to connect the main rivers of Europe (Dhur and Massard 1995, Moog et al. 1999, Grabowski et al. 2005 Špaček 2009). Geographical barriers like the Alps and the Balkan mountains that isolated the Balkan drainages preventing faunal exchanges with the rest of Europe ) could also account for this secluded population. Although, the current evidence deriving from mitochondrial data along with the geographic distribution supports the discrimination of A. acheronensis as a distinct species, further support could come from additional mitochondrial sequence data (especially from the Balkan peninsula) as well as by combining information provided by nuclear sequence data.
The monophyly of the species A. desmarestii, although supported by NJ, was poorly or not supported at all by BI and ML analyses, respectively. In the study of Garcia Muñoz et al. (2009) the monophyly of this species, based on the COI sequences, was strongly supported. This difference should be attributed to the larger number of sequences used in this study. A. desmarestii (Millet, 1831) does not comprise a strongly supported genetically distinct group and appears as a not well resolved part of the phylogeny. However, the genetic distances observed within this group are quite small in comparison with the other Atyaephyra species and this in combination with the morphological data supports the consideration of all the populations inside this group as one taxonomic entity. More sequence and morphological data, especially from the area of South Portugal and Morocco (the monophyly of the species is strongly supported once the sequences originating from Morocco and South Portugal material are removed), as well as other molecular markers are needed in order for the relationships within A. desmarestii to be clarified.
In the southwestern part of the Mediterranean area, only two species of Atyaephyra have been described to date: A. desmarestii and A. rosiana. These two species had been considered synonyms until Anastasiadou et al. (2008) resurrected A. rosiana after studying material from São Barnabé River (Odelouca River) in South Portugal. In their study Garcia Muñoz et al. (2009), stated that the hypothesis of the two distinct species could not be supported although they did note some genetic variability in the specimens originating from South Iberian Peninsula. Similar results are obtained in the current study. Sequences from North African and South Iberian individuals presented a noticeable mean genetic divergence (3.1% and 2.3% respectively) from the rest of west European and Tunisian sequences, but although noticeable is still weak to sup-port the hypothesis of different species. A high variability in morphological characters, especially in the individuals from the South Iberia was also observed. Characters such as the length and height of the rostrum (the tendency is for rostra to be longer and narrower) and the number of rows in maxilla basal lower endite (usually 15-18) varied greatly from the typical form present in North Iberia and the rest of Europe as well as Tunisia (shorter and broader rostra, 17-21 rows on maxilla basal lower endite). Genetic diversity among the South and North-central Iberia populations is observed in many other freshwater species whereas only in a few of them is it robust enough to justify distinct species , Durand et al. 2003, Sanjur et al. 2003). An explanation for this should be sought in the eventful geological history as many basins of the Iberian Peninsula almost dried up and the southwestern part of the Peninsula became completely isolated during the Messinian period (Sanjur et al. 2003). In addition, the genetic diversity observed mainly between the Moroccan and remaining populations should be sought again to the geological history and the isolation of the North-west Africa from Europe and where dispersal between these land mass, across the Gibraltar strait ceased to be an option since Pliocene (Sanjur et al. 2003).
The Tunisian populations, on the other hand, are more closely related to the western European ones, probably due to the past connections through the Sicily Strait with European populations (Butler et al. 1999).
The second cryptic species A. tuerkayi sp. n. has been found only in the River Nahr Al-Kabir which is located along the borders of Lebanon with Syria. A. tuerkayi sp. n. is completely isolated geographically from the other two morphologically closest to it species, A. desmarestii and A.acheronensis sp. n. In fact A. tuerkayi sp. n. is surrounded by A. orientalis populations which show a wide distribution from Turkey to Iraq. Atyaephyra tuerkayi sp. n. is genetically well discriminated from A. desmarestii and A. acheronensis (genetic distances are 23.0% and 22.2% respectively) as well as from A. orientalis that is found in the adjacent areas (genetic distance is 19.7%). The genetic distances are among the highest observed between Atyaephyra species and by far exceed currently published records of intra-population variability of other fresh water decapods (e.g. Jesse et al. 2011). Furthermore, they are comparable with genetic distances of COI sequences described elsewhere for taxa recognized at the generic level (Avise 2000, Lefébure et al. 2006, Matzen da Silva et al. 2011). Therefore such an extensive differentiation should be attributed to speciation.
In the area of the Middle East, two subspecies were previously described, A. desmarestii orientalis Bouvier, 1913 andA. desmarestii mesopotamica Al-Adhub, 1987. However, no observable morphological characters where found that could differentiate them (see remarks of A. orientalis). Furthermore, although the genetic distances within the A. orientalis phylogroup were high (0.9%-10.2%) no firm conclusion could be drawn whether the hypothesis of multiple species is valid or not. Sequences from Orontes River (topotypical location of A. d. orientalis) and from Shatt Al-Arab River (topotypical location of A. d. mesopotamica) presented a noticeable mean genetic divergence (5.0%) but still not strong enough to support the hypothesis of different species. Detailed future studies on the morphological and genetic variability within the samples of Atyaephyra distributed throughout the Middle East will help clarify the relationships between the populations in this region, however given the present data, only one species is considered to exist, A. orientalis.
Four species (A. acheronensis sp. n., A. thyamisensis sp. n., A. stankoi, and A. strymonensis sp. n.) were found to co-exist in Greece with well defined and clearly separated distributions. Only two species (A. acheronensis sp. n. and A. thyamisensis sp. n.) were found to co-exist in the same river (River Louros, Epirus). Multiple individuals collected from the Louros estuary and further upstream, dating back to 1977 until 2001 were examined. These specimens were all identified as A. thyamisensis sp. n. However, in a recent sample (2012) both species were found. Probably, this could be attributed to fish transfers or translocation where shrimps could have accidentally been introduced. Additionally, the distance between the estuaries of the Rivers Louros and Acherontas is less than 30 km making human mediated dispersal, between the two watersheds, highly possible. Furthermore, numerous translocations of fish were made within Greece over the last 70 years (Economidis et al. 2000) making this scenario even more justified. However, the natural co-existence of the two species cannot be entirely excluded.
It is surprising that four out of the seven Atyaephyra species examined for the present study are recorded from Greece and three of these are endemic. Greece is considered to be a faunal and floral biodiversity hot spot within the Mediterranean region where freshwater fauna is not an exception (Reyjol et al. 2007, Jesse et al. 2011). Jesse et al. (2011 after studying the diversity of the freshwater Potamon crabs, revealed the existence of 14 species within the greater Mediterranean region. Eight of these species (three endemic and five with limited distribution in adjacent countries) were found in Greece. High diversity and endemism is recorded in other freshwater groups too, such as fishes. Greece harbours the largest number of fish species of any region in the Mediterranean basin where the number of endemic species exceeds 45% of the total number of native species (130) recorded , Blondel et al. 2010. Freshwater endemism in Greece is considered as one of the highest in the Mediterranean region and has been ascribed to its eventful geological history combined with complex climatic events (Bobori et al. 2001. The importance of morphology versus molecular data in order to resolve the phylogeny of a taxon still provides a forum for scientific debate (Tautz et al. 2003, Blaxter 2004, Page et al. 2005b. Although additional work is needed towards the exhibited morphological variability within the genus, the data provided by the present study demonstrate a case in which conventional and molecular taxonomy do not provide different patterns but, rather, complimentary. Finally, an additional step was taken by considering the molecular validation of the two cryptic species which couldn't be supported by morphological data alone. It seems, therefore, that when both molecular and morphological effort is combined towards a "total evidence" approach a whole greater than the sum of its parts emerges which is instrumental in our understanding the diversity of life (Page et al. 2005b).