Elasmopus yucalpeten sp. n. (Crustacea, Amphipoda, Maeridae) from the northern Yucatan coast, with a key for the genus in the Gulf of Mexico and biogeographic comments

A new amphipod species of the genus Elasmopus Costa, 1853 is described and illustrated based on material collected in a harbor on the northern Yucatan coast, southern Gulf of Mexico. Elasmopus yucalpeten sp. n. is recognized from its congeners by a two-articulate accessory flagellum, a group of long robust setae on the anterodistal margin of the gnathopod 2 basis, a distomedial concave portion on palm of gnathopod 2 propodus, long setae on basis posterior margin of pereopods 5–7, and an entire telson. The differences among closely related species are pointed out and they are compared with the new species. An identification key to species of the genus Elasmopus in the Gulf of Mexico and biogeographic comments at the regional and global scales are also provided.


Language and style
• The language of publication is English. There is no general limitation of the length of manuscripts, but please contact the editor before submitting papers exceeding 30 printed pages (approximately 60 manuscript pages including figures). • Manuscripts should be written in a clear, straightforward style and must not have been published or submitted elsewhere. • The text should be 12 pt, double-spaced, one-sided, left justified and with a margin of at least 3 cm. • Use a standard typeface, e.g. Times New Roman as little formatted as possible (without tabulators, several blank spaces, etc.). Avoid footnotes. • Divide the text into sections using headlines and subheadlines. Do not number the headlines. Inline headers should be set in italics and followed by a full stop. • The names of genera and species must be in italics.
• Taxonomic descriptions must comply with the rules of the 4th edition of the ICZN (see http://www.iczn.org/). • Enter the page number on every page. • Submit figures with a minimum resolution of 300 dpi. • The preferred file formats are PSD (Photoshop) and TIFF for colour and grayscale illustrations, and EPS for vector graphics. • JPG files are only accepted in high resolution.

General manuscript structure
If appropriate, the manuscript should be structured using headlines and sub-headlines, but without numbering, according to the following sections:  The genus Elasmopus Costa, 1853 is the most diverse genus in the family Maeridae Krapp-Schickel, 2008, roughly with 101 of the 328 species in the family distributed worldwide in temperate and tropical seas (Ahyong et al. 2011, Vader andKrapp-Schickel 2012). The species are mostly found on the continental shelf (≤ 200 m depth), mainly associated with macrophytobenthos (e.g. algae, marine angiosperm) and secondarily with epifauna (e.g. anemone-hermit crab symbiosis, sponges, zoanthids) (Souza-Filho andSenna 2009, Vader andKrapp-Schickel 2012).
Morphologically, two groups of species inside Elasmopus have been recognized by Vader and Krapp-Schickel (2012): the pectenicrus-group and the rapax-group. The

Biogeographic comments
Up to now, nine species of Elasmopus (including the new species) have been recorded in the Gulf of Mexico (Table 1). At the horizontal axis, according to the regionalization of Felder et al. (2009) for the Gulf of Mexico basin, three of those species (E. levis, E. pocillimanus and E. rapax) have been widely reported in the Gulf basin regions, two species (E. balkomanus and E. pectenicrus) have been mostly reported in the northern regions, and the remaining four species are so far confined (endemic) to the northeast region (E. cf. magnispinatus) or southeast region (E. lemaitrei, E. thomasi and E. yucalpeten sp. n.). At the vertical axis, according to the zonation by depth of Yáñez-Arancibia and Day (2004) for the Gulf of Mexico basin, all species have been reported on the coastal shallow (0-20 m) and only four species have been reported on the continental shelf (21-200 m). Furthermore, those species more widely distributed in the Gulf of Mexico also displayed a broad range of depth; in contrast, species with a constricted distribution displayed a narrow range of depth. Nevertheless, as LeCroy (2000) has pointed out for Elasmopus species narrowly distributed, further samplings may reveal that those species are actually more widespread than previously expected, for example, E. cf. magnispinatus has been reported only from the northeast region, but with a great number of records and a broad range of depth. Regionally, the amphipod fauna in the Gulf of Mexico shows an affinity to the biogeographic provinces from the tropical western Atlantic (Carolinian and Caribbean) quoted by Neigel (2009) and Briggs and Bowen (2012). The Carolinian province corresponds to northern regions (NW and NE) representing a warm-temperate condition; whereas the Caribbean province corresponds to southern regions (SW and SE) representing a tropical condition. According to those provinces, the genus Elasmopus has species with tropical and temperate affinities. The tropical component is dominant with eight species, three of which are endemic so far for the Gulf of Mexico (Table 1). Globally, E. yu-  Ortiz and Lalana (1994), LeCroy (2000), LeCroy et al. (2009), Vader and Krapp-Schickel (2012), and Paz-Ríos et al. (2013).
• ≤1 • calpeten sp. n. is geographically related to four species which in turn are morphologically similar by a unique trait in the genus, an entire telson. Moreover, from those species, E. yucalpeten sp. n. is similar to three species of the rapax-group (E. integer, E. pseudinteger and E. visakhapatnamensis) by an anterodistal margin of gnathopod 2 basis with group of long robust setae and P5-7 basis having long setae, revealing possibly a separate complex of widely distributed species. The distribution of those species resembles the trans-Indo-Pacific-Caribbean tracks described by Myers and Lowry (2009) (Figure 5), which is explained by plate tectonic/sea-level changes during the Cretaceous and is represented by distribution of a number of amphipod taxa at the family category (e.g. Neomegamphopidae) and genus (e.g. Mallacoota, Shoemakerella). Therefore, with the similarity among species closely related to Elasmopus it was possible to recognize that biogeographic track, which proposes according to Myers (1991) and Myers and Lowry (2009) an ancient connection among seas and a current isolation by means of disjunct distributions of related taxa.

Introduction
Natural history museums provide society with a number of indispensable, although often underappreciated, services in the fields of homeland security, public health and safety, agriculture, monitoring of environmental change, traditional taxonomy and systematics (Suarez and Tsutsui 2004). They are also indispensable for the study of the state and trends of biodiversity, providing baseline information useful to assess change at the genetic, species, community and landscape level. They are sometimes the only repositories of specimens of taxa which have gone extinct in historical times, and allow studies on endangered taxa avoiding new captures. The 'Museo di Zoologia dell'Università di Bologna' is no exception. In its present form, it dates back to the 1930s, but it actually contains specimens dating back to the 16 th century (collected by Ulisse Aldrovandi, 1522-1605). It was then enriched by the collections of F. Cospi (1609-1686) and L.F. Marsigli (1658Marsigli ( -1730, and further developed during the 19 th century, mainly due to the work of its directors C. Ranzani (1775-1841) and G.G. Bianconi (1809Bianconi ( -1878, and 20 th century, due to the commitment of A. Ghigi (1875A. Ghigi ( -1970. The value of the museum as repository of specimens of extinct or endangered species has been already recognized for vertebrates; for example it contains a head of a great auk, Pinguinus impennis (Linnaeus, 1758), which became extinct in the mid-19 th century. However, no recognition was ever given to the value of its invertebrate collections, of which here we present the North-American Unionida.
The Unionida is a diverse order of bivalves with ca. 840 species worldwide. The Neartic (especially the SE United States) has the highest concentration of Unionida diversity in the world, comprising ca. 300 species alone (Graf andCummings 2007, Bogan 2008). However, this richness has been threatened by the construction of dams, pollution and sediment toxicity, wetland drainage and channelization, sedimentation and siltation resulting from poor agricultural and silvicultural practices, highway and bridge construction, interbasin transfer schemes, habitat loss through dredging, and other land-use activities (Lydeard et al. 2004). At present, 235 species have been assessed, of which 27 (11%) are considered extinct, 50 (21%) critically endangered, 31 (13%) endangered and 11 (5%) vulnerable (IUCN Red List, last accessed July 2013).
Our work recovered the collection of Unionida in the Museo di Zoologia dell'Università di Bologna, Italy, a collection dating back to the 19 th century. This collection is likely to be only the tip of the iceberg among many other collections, hidden in the Italian fragmented natural history museum system. Our aims are therefore to highlight the value of historic natural history collections as repositories of specimens of extinct or endangered species. In a time of increased awareness of global biological changes, museums are valuable reservoirs of baseline information. Moreover, we wish to bring to the attention of the international scientific community the hidden treasures in the Italian museums, and foster research on their material.

Methods
We recovered the collection of North American Unionida, cleaned all specimens, labels and original boxes, and transferred them into zip-lock plastic bags along with all original labels. Also the original boxes, likely to belong to early 20 th century were preserved. The most interesting lots were photographed. Identification was checked, nomenclature updated following the MUSSELp database (Graf and Cummings 2013), which proved particularly useful in tackling old names, and the conservation status of each species was recorded on the basis of published assessments (IUCN Red List, U.S. Fish and Wildlife Service). Samples were databased, and research into the biography of the main contributors to the collection performed. The collection also contains several lots from Europe, as well as samples from South America, Africa and Asia, which were cleaned, but not analysed in detail.

Results
Eighty-six specimens of North-American Unionida are preserved in the Museum (Table 1 gives a list) in 76 lots, representing two families (Unionidae and Margaritiferidae) and 57 species. The condition of specimens is generally very good, with most valves being paired, and only a few having cracks or other defects. The collection comprises further 34 lots (104 specimens and 22 valves) from Europe, 21 lots (31 specimens and two valves) from the Central and South America, three lots (three specimens) from sub-Saharan Africa, and two lots (two specimens) from New Zealand, which will not be further commented upon here.
Most of the lots were received from M.E. Moricand (1779-1854) (Fig. 2 A-B), a Swiss collector, in the mid-1800s. His collection was particularly rich, especially of land shells: his son M.J. Moricand catalogued his collection in 1859 censing 5,950 species and about 25,000 specimens (Cailliez, 1983). A few more lots from the Mississippi River are dated 1863 and belonged to the Capellini collection ( Fig. 2 C). Giovanni Capellini (1833Capellini ( -1922 was a professor of geology at the University of Bologna, and travelled in 1863 to the United States (Vai, 2002).

Discussion
Museum collections are increasingly becoming the only source of information on extinction rates and range contractions, or the only way to access species extinct in historial times (Allmon 1994). The case study here reported shows that the effort to recover an historical 19 th century collection allowed the recognition of an extinct species and of several endangered species, belonging to one of the most imperiled groups of molluscs worldwide. The specimens themselves, and the data accompagnying them, provide researchers with a historic record of where extinct and endangered species once lived. Notwithstanding natural history collections are then a vital resource for conservation research, the recognition of their importance is often lacking.
Even when the value of natural history collections as sources of long-term or past datasets is recognized (Lister et al. 2011), locating material of interest is often difficult due to the fragmentation of the museum system in some countries, the lack of computerization of collection data, or the lack of availability to the public (e.g. via internet) (O'Connell et al. 2004). The Museo di Zoologia dell'Università di Bologna does not have permanent personnel, and all curatorial work is carried out by university researchers and students in the framework of their research assignments or on voluntary basis. The malacological collection is partly databased, but the database is not readily available to the public. Publishing the results of this work will hopefully contribute to arise consciousness on the value of natural history collections, and start spreading information to the scientific community on the Italian invertebrate historical collections. Indeed, the Italian museum system had its golden years in the second half of the 19 th century and early 20 th century (see for a narrative on vertebrate collections Gippoliti (2005)). Therefore, further collections of interest can be expected when digging into the old cabinets of the Italian museums. Cyprogenia stegaria (Rafinesque, 1820) ( Fig. 1 A,  Unidentified -2 lots, 1 complete specimen and 1 valve --
Since its description the species has rarely been mentioned in the literature. Based on osteological characters (Martin 1972) and acoustic criteria (Tandy and Keith 1972) it was placed in a monotypic group within the genus Bufo. Mills Tandy studied the three toads A. perreti, A. maculatus and A. regularis in 1964 and 1965, occurring in sympatry at the A. perreti type locality, and observed that they differ in size, advertisement call parameters, reproductive season and biology, as well as in microhabitat selection (Tandy and Keith 1972). Based on karyology (2n= 20; Bogart 1968), but without genetic data, Frost et al. (2006) transferred B. perreti into their new genus Amietophrynus, and Channing et al. (2012) redescribed the tadpoles. Schiøtz and Tandy (2004) and Stuart et al. (2008) give the conservation status of the species as 'Vulnerable'.
However, only Mills Tandy seems to have collected new data from the field (Tandy and Keith 1972) and until now, the species is still only known from its type locality. On the webpage of the Amphibian Survival Alliance (ASA 2014) Amietophrynus perreti is listed as being lost, the last observation dating to 1970. We do not know from which data the later date stems. In 2013, 50 years after its original description, we decided to search for the species. Our aims were fourfold: i) document if the species still exists at the type locality, ii) collect data on the population status, iii) search in neighbouring areas for further populations, and iv) investigate the systematic position of this species.

Material and methods
All field work was carried out by two persons (ABO and a field assistant) in the Idanre Hills, Ondo Province, south-western Nigeria (approximately 7°06'N, 5°06'E; see Appendix). The region was visited three times; the first visit from 10-12 October 2013 was only used to get familiar with the topography and to select a suitable field assistant. The second visit took place at the beginning of the dry season (26-28 October 2013). During three days and two nights we applied visual and acoustic encounter surveys (compare  at three sites (see below). Day surveys were conducted at all sites while night surveys were conducted at sites A and C only. The third visit comprised one day and one night survey at sites A and B, respectively, and took place from 13-14 March 2014, the ending of the dry season (Table 1).
Site A (7°06'37.3"N, 5°06'24.1"E) comprised approximately 2 ha situated between two huge rocks, only about 50 m from the town of Idanre (Fig. 2a, c). Water flowed in two valleys between these rocks. The vegetation consisted mainly of shrubs and small trees. Site B (7°06'14.2"N, 5°04'41.3"E), covering about 1.5 ha, was approximately 250 m height on the rocks. This area was relatively flat with patches of vegetation, shrubs and trees, growing on the rocks. Water was flowing down from higher locations, forming rock-pools in some depressions (Fig. 2b). Site C (7°06'32.9"N, 5°04'28.0"E) was part of a cocoa farm adjacent to a large inselberg, including a creek flowing through the farm (Fig. 2e). The creek had a rocky bottom and was 6-11 cm deep. Along the water course, there were few grasses and shrubs.
In addition, we surveyed two further sites with similar habitats (rocky inselbergs). Ikere was about 28 km from Idanre. Two areas were surveyed here on 3-4 November 2013 and 16 March 2014. Site 1 comprised giant rocks and was surveyed for 6 and 4 hours per day and night, respectively (3 day and 2 night surveys, each 2 hours length). With the same sampling effort we investigated site 2, a cocoa plantation near rocks. The second area was situated about 6 km from Akure and was searched on 6 November 2013 for 3.5 hours during the day and night, respectively. A map of all surveyed sites is provided in Figure 3.
All amphibians encountered were recorded. A few vouchers were collected to ascertain species determination. They were euthanized in a chlorobutanol solution and preserved in 75% ethanol. Vouchers are deposited at the zoological collection of A.B. Onadeko at the University of Lagos (no accession numbers) and at the Museum für Naturkunde Berlin (see Appendix for ZMB and GenBank   Onadeko, A.B. et al.: The rediscovery of Perret's toad, Amietophrynus perreti (Schiøtz, 1963) 116 were compared to GenBank data as well as sequences for additional Amietophrynus species of interest (see results). Sequences were aligned using MAFFT v7.017 (Katoh and Standley 2013) and a Bayesian phylogenetic inference was carried out using MrBayes version 3.2.2 (Ronquist et al. 2012). Four independent runs were executed to assess convergence, each for 3 million generations, sampling every 1000 generations. Four chains per run were allowed to sample across the entire general time reversible substitution model space and the first 10% of each run were discarded as burn-in before generating a 50% majority-rule consensus tree from the posterior tree samples. Uncorrected 16S p-distances of A. perreti and included Amietophrynus spp. were calculated using PAUP * 4b10 (Swofford 2003).

Results
Occurrence, abundance and threats. We rediscovered Amietophrynus perreti and observed them in sympatry with A. regularis, Phrynobatrachus spp. and Arthroleptis spp. at the study sites of the Idanre Hills inselbergs. We did not observe any A. maculatus, known to occur here as well (Tandy and Keith 1972). Occasionally we spotted A. regularis in close vicinity to A. perreti, but then usually at the rock bases. A. regularis were particularly dominant around the town of Idanre, where A. perreti was lacking. Puddle frogs, Phrynobatrachus, were observed on the rocks, in vegetation around water-filled rock-pools. Arthroleptis were abundant in the cocoa farm and in tertiary vegetation but absent from the rocks. In total we observed 111 A. regularis, 49 Phrynobatrachus spp. and 98 Arthroleptis spp. during 39 person-hours of searching.
Amietophrynus perreti (Fig. 4) were observed during day and night, mainly on higher parts of the rocks. They were encountered on the hard rock surfaces and among low herbal vegetation on sandy soil between the rocks. During October 2013 we observed a total of 31 A. perreti, 16 during the day (12 person-hours search time, Table 1) and 15 during the night (6 p-h). In March 2014 we recorded a total of 12 A. perreti, 5 during day (11 p-h) and 7 (10 p-h) during night time (Table 2). Compared to the other anuran taxa, A. perreti had the lowest abundances.
We observed two potential threats to the survival of the toads. Most importantly some areas around the sites are being cleared for cocoa plantations, the major cash crop in the region. At higher places on the rocks some crops such as pepper, corn and banana are planted. Another cash making venture is the intense cutting of trees that grow among the rocks.
We could not observe A. perreti in two other localities, namely Akure, 6 km and Ikere, about 28 km from Idanre. Both had similar habitats and topography compared to the A. perreti type locality.
Morphology. We examined three male and four female A. perreti vouchers (see Appendix). The males' snoutvent-lengths (SVL: 39.4-40.2 mm) were within the range Schiøtz (1963) and Tandy and Keith (1972) provided for their specimens. None of the presumed male toads exhibited nuptial pads (dry season and hence no reproductive season). The females (55.3-64.4 mm SVL) slightly exceeded the sizes of the known females (Schiøtz 1963, Tandy andKeith 1972). Small differences to the original description concern the size of the metatarsal tubercles, being quite big and clearly distinct (larger in females than in males), and the presence of small, very flat and indistinct warts on the top of the head (no warts according to original description). All toads had comparatively flat heads, slightly pointed in lateral view (rounded in A. regularis and A. maculatus), almost no webbing between toes (distinct webbing between toes in A. regularis and A. maculatus), no tarsal fold (present in A. regularis and A. maculatus), and a white throat in males (dark in A. regularis and A. maculatus). In the preserved vouchers (some of them very flattened after transportation) the parotid glands ranged from less distinct to slightly more pronounced than in A. maculatus, but they were never as huge and smooth as in A. regularis. All our vouchers had more or less the same dorsal pattern as the holotype (compare Fig. 4). In life they have the olive (to yellowish) tinge Schiøtz (1963) mentioned in his description. All   preserved animals had a much clearer and more contrasting pattern than A. regularis and A. maculatus.
Generic assignment. Tandy and Keith (1972) placed A. perreti in its own, monotypic group, although having the same number of chromosomes as the A. regularis-maculatus-latifrons complex (2n= 20, Bogart 1968). A crossing of a female A. maculatus with a male A. perreti produced some embryos to the larval stage, however none of them reached metamorphosis (Blair 1972) [such crossing worked to gastrula or even larval stage even with American species nowadays placed in different genera, namely Anaxyrus boreas, see Blair 1972].

Discussion
We succeeded in rediscovering Amietophrynus perreti at its type locality, 50 years after its original zse.pensoft.net Onadeko, A.B. et al.: The rediscovery of Perret's toad, Amietophrynus perreti (Schiøtz, 1963) 118 description, and 43 years after the last observation, if the date of 1970 is correct (ASA 2014). A welcome byproduct of our investigations was that we could confirm the morphological description of the species by Schiøtz (1963), and more importantly, confirm that it is indeed a member of the genus Amietophrynus. It was however, surprising, and in contrast to the osteological results of Martin (1972) and the biological and acoustical analyses of Tandy and Keith (1972), that A. perreti has no isolated phylogenetic position within the genus, but seems to be most closely related to the savannah dwelling A. maculatus (see Rödel 2000 for details of the biology of this species). The 16S sequence data did not sufficiently resolve a number of nodes in the phylogeny, however, the Amietophrynus crown group was well supported, as was the A. perreti-maculatusregularis-latifrons-togoensis clade.
The long time gap since the previous observations and the rediscovery of the toad was not due to particular rarity of the species but can be explained by the fact that nobody else searched for the species. We found the toad being locally quite abundant, although our search time at the beginning and towards the end of the dry season certainly was suboptimal. According to Tandy and Keith (1972) A. perreti reproduces during the core rainy season.
Due to their age and peculiar edaphic and climatic conditions, inselbergs are very special habitats with often unique plant and animal communities (Porembski and Barthlott 2000). Unfortunately, so far they have attracted almost exclusively the attention of geologists and botanists. We only know a few studies either focusing entirely (e.g. McLachlan and Cantrell 1980, McLachlan 1981, Osborne and McLachlan 1985, Köhler and Böhme 1996, Schorr 2003 or partly on amphibians on inselbergs (Rödel 1998). As far as we know, A. perreti is the only African anuran species depending exclusively on this habitat type. Schiøtz and Tandy (2004) speculate that A. perreti may occur in other areas of western Nigeria, offering similar habitats, but we did not record the species in two similar areas in closer vicinity of the type locality, nor was the species recorded by Onadeko and Rödel (2009) from other sites in south-western Nigeria. It thus seems possible that A. perreti is indeed an endemic species of the Idanre Hills.
During the current survey we observed no tadpoles of A. perreti. However, the tadpoles require a constant water film on the rocky surfaces where they graze on algae ( Fig.  1), most likely in a situation like figured in Figure 2b. During the dry season, although occasional rains occur, such habitats do not exist. A reason for the absence of A. perreti on other inselbergs might be that the existence of this species, with its very special reproductive biology, requires an environment with high humidity persisting for prolonged periods and some water storing vegetation. Only then it seems possible that a constant water film on the rocks is maintained and provides the tadpoles of A. perreti with enough time to finish metamorphosis. The tadpoles seem to actively avoid deeper water and in-stantly, when forced into water, climb out again (Schiøtz 1963), showing a behavior comparable to some Petropedetes tadpoles (Barej et al. 2010). It thus seems that they will not make use of deeper rock-pools (Fig. 2e) and ultimately depend on the wet rock surfaces (Fig. 1).
We know from other inselbergs in the forest zone of Ivory Coast, that these provide a much hotter and dryer environment compared to the surroundings and thus are home to true savannah species, even within a rainforest matrix (Schorr 2003). The fact that true forest in south-western Nigeria disappeared almost completely (see Onadeko and Rödel 2009), may explain that A. perreti simply cannot survive on other nearby inselbergs because the overall climate is too hot and dry. Already Schiøtz (1963) had failed to record the species in nearby inselbergs, surrounded by savannah vegetation. It is also possible that the mere dimensions of the inselberg landscape near Idanre are simply unequaled elsewhere.
According to the original description (Schiøtz 1963) A. perreti is abundant within its habitat. Arne Schiøtz collected 16 adults in July and August, 10 semi-adults and juveniles in January, July and August and 74 tadpoles in July and August 1961. On our surveys we likewise encountered the species frequently, although probably less frequent than Schiøtz (1963). We guess that the toads are not only reproducing during the rainy season (Tandy and Keith 1972), but that this is also the peek activity period and that the Idanre Hills population is still in good shape. The population thus seems viable, but we do not entirely agree with the judgment of Schiøtz and Tandy (2004) that the species does not face "any serious threats at present" because the "rocky habitat is largely inaccessible and useless to humans". Schiøtz (1963) reported that he found all adults at night on bare rocks. He heard males calling in July 1961 during daytime, the voices coming from "very small patches of grass and low vegetation on the rocks, never from the wooded area" (Schiøtz 1963). Tandy and Keith (1972) report similar observations. It thus seems that the toads, at least during the rainy season, do not need forested areas. However, we believe that the current clearing of larger trees from the area, and the use of smaller patches to grow various crops on the rocks may result in a changing micro-climate and altered hydrology, both potentially affecting the toad's survival, especially during the dry season when humid areas are rarer. In addition, the close proximity to a human settlement and to cash crop plantations may expose the toads to agrochemicals and other potentially harmful pollutants.
As the species could not yet be recorded from any other site, the only known population deserves particular attention. Schiøtz and Tandy (2004) already wrote that the species "is intrinsically at risk because of its restricted range". We thus urge, and agree with Schiøtz and Tandy (2004), for setting up a continuous monitoring program of this population and suggest aiming for a protection status for the Idanre Hills in order to assure the long-term survival of Perret's toad. B (2008)

Introduction
Chemosymbiotic bivalves typically host their autotrophic bacteria in tissues of the ctenidium (gill). As a consequence gill morphology is modified and from a gross perspective these gills are thick and fleshy in comparison with those adapted for filter feeding (Taylor and Glover 2010). The fleshy nature of the gills of thyasirids and lucinids was recognised long before their function was discovered and was shown to be a result of elongation of the abfrontal zone of the filaments (Allen 1958). Bacterial symbiosis associated with this modification was confirmed by a number of studies in the 1980s ( see review by Taylor and Glover 2010). In the heterodont families Lucinoidea and Vesicomyidae the association is apparently obligate but the Thyasiridae display both symbiotic and asymbiotic taxa. Variation in thyasirid gill structure was first demonstrated by Southward (1986), and later in a wider ranging study by Dufour (2005). Dufour (2005) classified thyasirid gills into three categories based on the number of demibranchs, the degree of abfrontal extension and the extent of bacterial symbiosis. In all three types the filaments were either rod-like or lamellar, the lamellar form having greatest abfrontal extension and strongest association with chemosymbiotic bacteria. This lamellar gill (type 3 of Dufour 2005) has recently been described from two other heterodont families the Montacutidae (Oliver, Southward and Dando 2012) and the Basterotiidae (Oliver 2013).
A more complex gill, akin to the tubular form of many lucinids (Distel and Felbeck 1987) has been suggested to occur in Conchocele (Taylor and Glover 2010; Dufour 2005) and as such was partially described by Nakazima (1958) in C. disjuncta Gabb (= C. bisecta Conrad). In thyasirids the bacteria are extracellular with the exception of "Maorithyas" hadalis where an intra-cellular arrangement was described and where the frontal cilia are absent (Fujiwara et al. 2001). Zoosyst. Evol. 90 (2) 2014, 121-132 | DOI 10.3897/zse.90.8323 Recently I was sent two thyasirids from the Regab pockmark (Gulf of Guinea) to identify and name following their inclusion in the bacterial study of Rodrigues and Duperron (2011). From a cursory examination it is immediately apparent the gill filaments are not laminar but that the gill is a three dimensional network of tubules.
Using scanning electron microscopy this paper describes the gross structure of these "tubular" gills and compares them with those described by Dufour (2005), especially the lamellar (Type 3) gill associated with chemosymbiotic taxa. Both Conchocele and the Regab thyasirid have been shown to be chemosymbiotic (Imhoff et al 2003;Rodrigues and Duperron 2011). Comparisons with the lucinoid gill are made and finally the Regab thyasirid is described as Ochetoctena tomasi gen. et sp. nov.

Methods
All specimens had been previously fixed in ethanol or formaldehyde of unknown concentrations. For scanning electron microscopy, gill tissues were excised and cut transversely and longitudinally using a thin razor blade. Tissues were dehydrated in 100% ethanol overnight and critically point dried with liquid CO 2 as the intermediate fluid in a Quorum K850 critical point dryer. Dried samples were mounted and gold coated before examination using a Jeol Neoscope SEM.

Institutional Abbreviations
AMS -Australian Museum, Sydney NMW.Z -National Museum of Wales RBCM -Royal British Columbia Museum SBMNH -Santa Barbara Museum of Natural History

Results
The laminar filament. (Type 3L) in Thyasira flexuosa, T. sarsi, Axinus cascadiensis, "Conchocele" excavata and the undescribed genus from Quatsino Sound Both demibranchs are present, the outer extending over approximately half of the inner (Figs 1A-C). The filaments are fully reflected forming a descending and ascending lamella. The filaments are extended abfrontally giving each a laminar form. The supra-branchial chamber is represented by the large dorsal space (sbc) between the descending and ascending laminae. Ventrally the abfrontal regions extend and fuse to form continuous inter lamellar septae of varying patterns (ils) (Figs 1A-C, 2A, B). Using Thyasira sarsi as typical of this group the micro-structure is as follows. All parts of the abfrontal surfaces and the septae (ils) are lined with polygonal bacteriocytes (Fig.  2B). The frontal face of the filaments are fully ciliated (Figs 2C, D) with a median band of frontal cilia (fc) bordered on both sides by a band of lateral frontal cirri (lfc) and behind these a band of lateral cilia (lc). If removed the ciliated surfaces can be recognised by a remaining intricate pattern of scars on the epithelial surfaces ( Fig. 2E, arrowed). Immediately behind the ciliated bands there is a network of inter-filament junctions (ifj) (Fig. 2D).
In "Conchocele" excavata only alternate filaments fuse to form inter lamellar septae and the middle portions of these septae do not bear bacteriocytes (Fig. 2F).

The tubular/laminar filament (Type 4T/L) in Conchocele bisecta
Both demibranchs are present, the outer and inner of approximate equal size (Fig. 1D). The filaments are fully reflected forming descending and ascending lamellae and appear to be narrowly separate along their entire length. At low magnification the filament is seen to be extended abfrontally but rather than appearing to be laminate the frontal region appears as a series of small blocks but these are not apparent abfrontally. Under the SEM the frontal surface appears as a series of vertical bands with rows of openings between the bands (Fig. 3A, openings white arrows) The vertical bands represent the frontal face of each filament and are strongly angulate with no trace of ciliation (Figs 3E, F). Nakazima (1958) assumed that the cilia had been lost through poor preservation but if this was the case the intricate pattern of ciliary 'roots' would remain and be visible under the SEM (see Fig. 2E for comparison). The lateral margins of the frontal bands are smooth ( Fig. 3E) but between each band a small projection can be seen sitting on the inter filament junction (Fig.  3F, arrowed).
With the frontal bands ripped off a regular tubular structure is seen ( Fig. 3B) with inter filament junctions (arrowed). Viewed from the abfrontal face of a lamella the regular tubular structure is not apparent and replaced by laminar filaments (Fig. 3C abf-lam). Widely separated but torn inter lamellar junctions (ilj) are visible in Fig.  3C, and best seen in cross-section in Fig. 3D (ilj). In cross-section (Figs 3D, 4) it can be seen that the frontal zone (fz) is narrow consisting primarily of the angular bands. The tubular zone (tbz) is a little over half the thickness of the lamella with the remainder being laminar (lz). Inter-lamellar junctions maintain a series of inter lamellar spaces (Fig. 3D, ils). Behind the frontal zone and between the laminar and tubular zones are inter filament junctions (ifj).
The tubes are lined with densely packed bacteriocytes as is the surface of the laminar zone, the inter lamellar junctions do not bear bacteriocytes. Preservation was not sufficient to acquire detailed images of the bacteria.

The tubular filament (Type 5T) in Ochetoctena tomasi
Both demibranchs are present but the outer is about half the depth of the inner (Fig. 1E). The normal filament structure is not apparent even under low magnification but appears as a series of small blocks (Fig. 8F). In trans-verse view (Figs 5, 7C) the ascending and descending arm of each filament is seen to be made up of a series of tubules (lt), abfrontally these are fused to form a median tube (mt). There is a small dorsal supra branchial chamber (Fig. 5 sbc). In lateral section the tubes can be seen to open between the frontal faces of the filaments

Discussion
The results presented here reveal further modification of the thyasirid gill towards a complex three-dimensional structure. The gill types described by Dufour (2005) can now be added to with two additional types; "tubular/laminar"  Figure 2F "Conchocele" excavata, longitudinal section of ventral portion of inner demibranch. abs abfrontal surface; fc frontal cilia; lc lateral cilia; lfc lateral frontal cirri; ifj inter filamental junction; ils inter lamellar septum; sbc supra branchial chamber.   and "tubular". The known gill types can be summarized as follows (Table 1) with a revised nomenclature based on that of Dufour 2005. The revised nomenclature incorporates the shape of the filaments and the number of demibranchs. Filaments that have no or little abfrontal extension are termed rods and these equate to Dufour Types 1 and 2. For Dufour Type 2 those with single demibranchs and non-reflexed filaments are termed "Type 2b (R)" while those with both demibranchs are termed "Type 2a (R)". The letters "L" and "T" refer to the laminar and tubular structures respectively. For each gill type the corresponding genera are listed, but some remain unallocated due to lack of data.
The extent of frontal ciliation is noted but the present observations give rise to difficulties in interpretation of the functioning of the tubular/laminar and tubular gills. All Type 3 (L) gills have complete ciliation of frontal and lateral cilia with eulateral cirri as also seen in the Lucinoidea (Taylor and Glover 2010). Here, it has not been possible to show any similar ciliation in Conchocele (Type 4T/L). Without ciliation it is difficult to comprehend how the currents to drive the water flow through the gill tubules are generated. Bernard (1972) records ciliary currents over the gill but gives no details of the ciliation. The lack of the basal structure of the cilia and cirri usually seen when the cilia are removed suggests that the Conchocele gill is atypical but its functioning remains enigmatic and requires observation on live material. It seems improbable that the small projections sitting on the frontal interfilamentar junctions, even if motile, could drive water currents into the tubules and could not produce the currents indicated by Bernard (1972).
Although there is an indication of frontal ciliation in Ochetoctena this is reduced and in both this genus and Conchocele they may be unable to create sorting and feeding currents on the gill.
Thyasirids typically hold their bacteria extra-cellularly (Dufour 2005) but the taxon Maorithyas hadalis is reported to hold the bacteria intra-cellularly (Fujiwara et al 2001). This unique condition suggests a further gill type   (2005) and can be seen to be similar to Thyasira sensu stricto in both shell and anatomy. This tubular structure increases the surface area of the bacteriocyte zone and creates a more rigid network further facilitating the movement of water from the infra-branchial chamber to the supra-branchial chamber. Similar structures are present within the Lucinoidea; Dando et al (1985) noted the inter-lamellar bridges in Myrtea spinifera and Distel and Felbeck (1987) illustrat-ed a complex tubular structure in Lucinoma aequizonata. Such complex structures are found throughout the Lucinoidea but this gill differs in consisting of a only a single demibranch (Taylor and Glover 2010). The tubular structure is yet another convergent morphological feature shared by chemosymbiotic thyasirids and lucinids.
The gill structures described here can potentially impact on the systematics of the Thyasiroidea. The observations on the gills of Conchocele bisecta and "C" excavata indicate that these taxa are not congeneric and this is part of the subject of another paper that also describes a new genus containing species previously assigned to Conchocele (Oliver and Frey, in press).   Definition. Equivalve, Equilateral. Outline subcircular, lunule margin depressed, posterior margin with a single weak sinus. Posterior sulcus shallow but prominent. auricle absent. Escutcheon narrow, shallow. Ligament deeply sunken. Hinge edentulous. External sculpture of commarginal lines and growth stops, overall with microscopic conical, calcareous spines these randomly distributed with some linked by low ridges . Ctenidia of two demibranchs, filaments tubular, "Type 5T".
Distribution. Known only from the Regab pockmark off west Africa.  Etymology. Named for my son Tomas.
Discussion. The shell of Ochetoctena has a weak posterior sulcus and the escutcheon is excavated but lacks any auricle to support the sunken ligament. In this it differs from Thyasira sensu stricto where there is a well developed auricle and from Conchocele which is oblique with a very strong posterior sulcus. Species lacking an auricle are often placed in the genus Parathyasira (Oliver and Killeen 2002; Payne and Allen 1991). The type species of Parathyasira is P. resupina Iredale, 1930 (Fig. 9F) and has a shell microsculpture of radial rows of calcareous spines. This character is also seen in the Atlantic species Ochetoctena is the only thyasirid known to have ctenidia of the Type 5T structure; Conchocele has a partly tubular gills of the Type 4TL morphology. The ctenidia of these Parathyasira species show poorly developed abfrontal extension resulting in a flimsy open structure. Dufour (2005) reported that the gills of P. granulosa are of her type 2 (type 2aR above) suggesting that the genus Parathyasira is characterised by a weak symbiont partnership.
There are, therefore, shell and anatomical synapomorphies that separate Ochetoctena from all other known thyasirids.

Introduction
Driftwood specialist hoppers are a rare, difficult to find, ecological group of Talitridae which are virtually confined to rotting driftwood where they live in galleries, consuming rotting driftwood and reproducing with relatively small broods (1 to 19 ova per brood in the Macarorchestia so far described). The ova are incubated in the brood pouch formed by the oostegites on peraeon segments 2 through 5 and hatch as juvenile forms. Because talitrids have no larval stage their dispersal, particularly to distant oceanic islands, was seen as problematic. The general view was that some form of passive rafting dispersal was involved (Wildish 1988). In the lineal "island" theory (Wildish 2012) this view was focussed and the hypothesis proposed that driftwood hoppers, with near-permanent residence in driftwood, were important agents of long distance dispersal for talitrids, particularly to distant oceanic islands. Circumstantial evidence Academic editor: Matthias Glaubrecht supporting the lineal "island" theory was presented in Wildish (2012), showing that driftwood talitrids reached the northeast Atlantic islands on driftwood transports. After reaching the remote, recently-formed, volcanic islands they evolved further there into terrestrial or cavernicolous forms if sub-tropical rain forest or seashore caves were contiguous habitats with the supralittoral and also lacked a talitrid fauna. Alternatively driftwood specialists remained in place in their primary ecotope as supralittoral, driftwood hoppers if the supply of driftwood habitat was plentifully available.
Currently known driftwood taxa (Wildish et al. 2012) comprise 5 species presently grouped in 2 genera from the Mediterranean and northeast Atlantic (Macarorchestia and Orchestia), with another genus (Platorchestia) reported from the west coast of North America by Bousfield (1982). The preponderance of driftwood hoppers in the northeast Atlantic/Mediterranean coastal areas is suggested to be the result of more intensive sampling there. Other geographic areas and particularly the southern hemisphere, where no driftwood hoppers have yet been found, have not been sampled intensively enough to conclude that they are without driftwood talitrid specialists.
As a result of molecular studies of Macarorchestia remyi it was suggested that two genetically distinguishable forms: one centred in the Tyrrhenian and the other in the Adriatic Sea were present (Pavesi et al. 2011). This result was confirmed with further relative growth studies (Wildish et al. 2012) and a new species referable to Macarorchestia is described herein. Molecular and relative growth studies (Wildish et al. 2012;Pavesi et al. 2014) determined the taxonomic identity of an unknown talitrid taxon found in a driftwood log which stranded in the Swale, U.K., and found morphological evidence supporting a generic level change in nomenclature for Orchestia microphtalma Amanieu & Salvat, 1963. A formal presentation of the taxonomic changes outlined above is provided with a key for the currently known driftwood species from the Mediterranean and northeast Atlantic coastal region.

Material and methods
Slides were prepared as temporary mounts without staining and after dissecting mouthparts and limbs. Some were prepared as permanent mounts by Sara LeCroy, after staining with lignin pink and permanently mounted in CMCP-10 (Master's Company, Inc.).
Photographs of limb parts were made with a Carl Zeiss photomicroscope and digital Canon 990 camera. Adobe Photoshop (version 7.0), Illustrator (version 11.0) and a Wacom tablet were used to draw limb parts and prepare plates, essentially as outlined in Coleman (2003Coleman ( , 2006. Antenna, mouth and limb part abbreviations used throughout are: A1 = first antenna, A2 = second antenna, LL = lower lip (labium), UL = upper lip (labrum), RMnd = right mandible, LMnd = left mandible, Mx1 = first maxilla, Mx2 = second maxilla, Mxpd = maxilliped, Gn1 = first gnathopod , Gn2 = second gnathopod, P3 to P7 = peraeopods 3 through 7, Pl1 to Pl3 = pleopods 1 through 3, Up1 to Up3 = uropods 1 through 3, and T = telson. Body length was measured accurately from digital photographs of each individual pinned in a dissecting dish. The total body length (TBL) was measured from the most anterior part of the cephalon to the telson tip, on digital photographs with the aid of Image Pro Plus software. Limb ratios were calculated as Pl3/TBL, in units of mm.

Systematics
Genus Macarorchestia Stock, 1989Stock 1989: 1109Ruffo 1993: 738. Type species. M. martini Stock, 1989 Component species. Currently includes five species: M. martini, M. roffensis, M. remyi, M. pavesiae sp. n., M. microphtalma (Amanieu & Salvat, 1963) new comb. Stock (1989) except that the lacinia mobilis of the left mandible is 4 -5 -dentate. The propodus of the seventh peraeopod carries distinctive tufts of long, slender setae, which are sexually dimorphic. The first antenna with up to 5 articles (versus 3 in Stock). The pleopods are biramous with a basis which is not reduced, but both rami are variously reduced dependant on the total body length of the largest adults of the species. Thus in the smallest species, martini, there is no segmentation in the rami and 1-3 plumose setae and in the largest, microphtalma, there up to 5 segments and 11 plumose setae. The pleopod rami may be sexually dimorphic, as in male M. remyi, and the second antennal flagellum articles are sexually dimorphic in female M. pavesiae sp. n. In addition all species are small, that is to say < 15 mm in total body length and lack epidermal pigment patterns.
Material examined. Male holotype and 8 females (allotype and paratypes) on loan from Zoological Museum of Amsterdam, Amsterdam, the Netherlands (AMPH. 108.57). Collected by J. Stock on 2 August, 1987 from Gruta das Agulhas in Porto Judeus on the island of Terceira, Azores archipelago.

Distribution.
Known only from the type locality on the island of Terceira.
Remarks. This is the type species and smallest Macarorchestia. Stock (1989) was equivocal about whether this was a recent troglobiont specialist or trogloxenous form. I concur with the latter view and believe that the adaptations (small body length, body length greater in the female, small eyes, reduced pleopod rami and shortened length of the first 5 peraeopods) indicate that M. martini is a driftwood specialist. Distribution. Besides the locations found by L. Pavesi for this species, published records include that of Ruffo (1960) in Sardinia and Ruffo (1993) in France, Greece and Italy. Possibly some of these records refer to the species described below.

Epidermal pigment patterns. Absent.
Remarks. This is the second largest Macarorchestia.
In recent times the type locality has been destroyed by cleaning beaches for the benefit of tourists (L. Pavesi, pers. comm.). Diagnosis. M. pavesiae sp. n. is distinguished from its close relative, M. remyi, by its smaller size, sexually dimorphic second antennal flagellum articles in adult females and absence of sexual dimorphism in pleopod rami.

Macarorchestia
Description. Based on male paratypes: total body lengths in the range of 8.0 to 6.7 mm. Figs 1 and 2. The male holotype has similar morphology to male paratypes dissected in preparing the figures. Head deeper than long (1: 0.7); eyes small, round, less than half the head length. Antenna 1 flagellum 5-articulate with tip just exceeding the junction of peduncle segments 4 and 5 of antenna 2. Antenna 2 short, flagellum 12-articulate, peduncle not incrassate.
Mouthparts. Lower lip with lateral lobes, minute setae on the inner clefts. Left mandible with a 4-dentate lacinia mobilis and large molar process. Right mandible with the tip of the dentate incisor bilobed, 6-dentate lacinia mobilis. Maxilla 1 inner plate narrow with 2 terminal plumose setae, inner margin with long, fine setae; outer plate with a palp, apical robust setae curved and serrated on the inner edge. Maxilla 2 both plates equal in size, inner with a single, plumose seta, with shorter robust setae on the distal edge; outer plate with simple robust setae. Maxilliped inner plates with 3 stout teeth apically, inner, outer and palp distal edges covered with robust setae; palp large and 3-articulated.
Peraeon. Gnathopod 1 weakly subchelate with palmate lobes on propodus and carpus; dactylus as long as propodus lobe; the largest robust setae are present on the posterior edge of the merus. Gnathopod 2 subchelate with propodus and dactylus massively enlarged, dactylus drawn out in a short, blunt tip, its inner surface lacking fine setae. Ventral edge of each coxal plate rounded and with fine robust setae. Peraeopods 3 to 5 short, peraeopods 6 and 7 longer, the latter just longer than uropod 1. Peraeopods 3 and 4 lack a dactylus notch ("pinched unguis"). Peraeopods 6 and 7 not sexually dimorphic in merus and carpus. Six distinctive tufts of long, slender setae originating from the propodus of peraeopod 7 as follows: anterodistal (2 setae), distal (6 setae), anterior side of peraeopod near the first insertion of robust, bifid-tipped setae (4 setae), then on the posterior side of the peraeopod: first insertion of robust, bifid-tipped setae (2 setae), second (4 setae) and third (3 setae). The maximum length of the longest seta from the distal tuft was 144 µm (Fig. 3).
Pleosome. Pleopod basis not reduced and with a pair of hooked coupling spines, robust setae and fine marginal setae absent. All rami are are shorter than the basis. The second pleopod rami have 3 articles each bearing a pair of long, plumose setae: exopod -7, endopod -5. Urosome. In uropods 1 and 2 the inner and outer rami are of similar length, with 2-4 apical robust setae and 1 or 2 inter-ramal robust setae, basis with 1 or 2 dorsoventral, robust setae distally. Uropod 3 basis longer than the ramus and with 2 large, robust setae dorso-laterally. Smaller robust setae at the ramus tip but lacks inter-ramal robust setae. Telson with a mid-dorsal groove and 6-7 dorso-lateral robust setae on each lobe. Sexually dimorphic differences. Based on non-breeding adult female paratype of 7.3 mm total body length.
Gnathopod 1 without palmate lobes on propodus and carpus. Gnathopod 2 basis slender and with weak robust setae. Palmate lobes present on merus, carpus and propodus, dactylus small (described as "mitten-shaped"gnathopod 2). Pairs of non-ovigerous oostegites on peraeopods 2 to 5. Adult females greater than 7.5 mm body length with no more than 11 antennal flagellum articles (versus up to 13 in males). The 6 tufts of long, simple setae on the propodus of peraeopod 7 in males compares with 2 tufts, of sparser and shorter length setae in females.

Epidermal pigment patterns. Absent.
Etymology. The name honours Dr. Laura Pavesi who originally discovered and collected the new species during graduate studies at the University of Rome, Italy.

Distribution.
There are three known locations for this species on the shores of the Adriatic Sea and one on Corfú Island, Ionian Sea.

Component species.
Since the erection of Orchestia Leach, 1814 the genus has been uncritically used to include many new species from around the World. In more recent times genera have been split off from Orchestia including: Platorchestia by Bousfield (1982), Palmorchestia by Stock and Martin (1988) and Macarorchestia by Stock (1989). Bousfield (1982) re-defined the range of Orchestia, limiting species to those found in the Atlantic/ Mediterranean region. I have further limited the geographic range of this taxon to the northeast Atlantic, the Mediterranean and Black Seas, but excluding the western Atlantic coastline of North America. The northeast Atlantic islands including: Canary, Madeira and Azore archipelagos are also included in the region. The northerly limit is arbitrarily set at the Arctic Circle (thus including Iceland) and the southern one at the Tropic of Cancer. Circumstantial evidence (Henzler and Ingolffson 2008) supports the presence of Orchestia gammarellus on northwest Atlantic coastlines (as far south as Maine), as a result of recent, post glacial, synanthropic, dispersal from the northeastern shores of the Atlantic.
Taking only Orchestia species which occur within this newly defined geographic range and excluding those outside it, synonyms, and where the taxonomic or ecological status is unclear (inclusive of O. kosswigi Ruffo, 1949-which is figured and described in Ruffo (1993) but its ecological status remains unclear); O. guerni Chevreux, 1889 andO. gambierensis Chevreux, 1908), leaving a total of 13 species (Table 1). The placement of these 13 species in five clearly separate habitats is consistent with a polyphyletic origin for them and that we can expect further generic splitting of Orchestia. In fact Lowry and Fanini (2013) have recently proposed a revision of the genus Orchestia in which all the species belonging to freshwater and terrestrial rain forest leaf litter of the northeast Atlantic islands (columns 3 and 4 in Table (2014) suggest the polyphyletic status of Orchestia and a generic level re-alignment like that shown in Table 2. Further genetic and taxonomic work is needed to include all the species listed in Tables 1 and 2. Diagnosis. An interim diagnosis is provided based on the type species, O. gammarellus from the Medway estuary, U.K., as listed in Table 2. This is because of the demonstration of polyphyly (Pavesi et al. 2014) within the older view of the genus Orchestia and because of the resultant taxonomic uncertainty regarding which of the taxa in Table 1 should be included within Orchestia. A diagnosis of the 5 genera listed by letter in Table 2 is delayed because the current COI phylogeny (Pavesi et al. 2014) does not include 6 species of "Orchestia" (indicated by brackets in  Table 2). This omission might change the final phylogentic tree obtained with all species listed in Table 2 included.
Adult total body length up to 22 mm; dorsal pigment patterns present; eyes medium in size, approximately one quarter of head length; antenna 1 flagellum just reaching antenna 2 peduncle of article 4; antenna 2 sexually dimorphic, peduncle slightly incrassate in adult males and without ventral plate on peduncle article 3; upper lip without robust setae; mandible left lacinia mobilis 4 dentate; maxilliped palp 3 articulate, article 2 with well developed medial lobe; gnathopod 1 of male subchelate with palm equal to dactyl, carpus and propodus free and with rounded lobes covered with palmate setae; gnathopod 1 of female parachelate, without lobes on carpus and propodus; gnathopod 2 of male strongly subchelate, merus and carpus free, dactylus with blunted tip and is half the length of the enlarged propodus; gnathopod 2 of female, ovigerous oostegite long and wide with many, long, simple, marginal setae, basis expanded anteriorly; peraeopods 3-7 cuspidactylate; peraeopods 5-7 lack slender setae lining the anterior margin of the dactyl; peraeopod 7 sexually dimorphic, adult males with merus and carpus enlarged; distinctive tufts of long simple setae on propodus of peraeopod 7 absent in both sexes; pleon segments 1-3 lacking vertical slits; pleopod rami slightly, or not, reduced; uropods without apical, spade-like robust setae, uropod 1 not sexually dimorphic, peduncle lacking well developed dorsolateral robust setae distally, outer ramus with marginal robust setae, uropod 2 rami equal in length, uropod 3 ramus shorter than peduncle; telson apically notched with 6-8 robust setae per lobe and shorter than uropod 3. Type material. Holotype-immature male of 7.2 mm total body length (NHMUK 2014. 397) and slide preparation from this individual (NHMUK 2014. 397). Nine juvenile paratypes (NHMUK 2014. 398 -406) and 2 immature females (destructively sampled for temporary slide mounts and mtDNA analysis) removed from a cast-up driftwood log resting at the base of the seawall in the Enteromorpha zone by K.J. Wildish. The driftwood log had been tethered to an old cattle fence on the shore, so it could not float away. It was sampled on 14 th June 2011, 23 rd July 2011 and 13 th August, 2011 from the Swale, near Kingsferry Bridge, U.K.  Maximum body length. Unknown, the largest found was an 8.9 mm TBL immature female. Mature adults predicted by relative growth methods to be 12 to16 mm total body length (Pavesi et al. 2014).
Diagnosis. N. kenwildishi sp. n., can readily be distinguished from other driftwood hoppers of the genus Macarorchestia by its medium size eyes (versus small) and unreduced pleopods (rami sub-equal to basis, versus rami shorter than basis).
Description. Based on immature male holotype of 7.2 mm total body length. Figs 4 and 5.
Mouthparts. Upper lip with minute setae on the apical margin. Lower lip deeply cleft and with minute setae on the inner face. Maxilla 1 inner plate slim and with two terminal, plumose setae; inner margin with long fine setae; outer plate with a vestigial palp, apical robust setae curved inwards, some simple and others serrated on the inner edge. Maxilla 2 with inner plate subequal to the outer, inner with a single, plumose seta and fine marginal setae below it; both inner and outer plates with long, simple robust setae which curve inwards. Left mandible with a 4-dentate lacinia mobilis, 6-dentate incisor, strong molar process and setose accessory blades. Maxilliped with 3 strong api-cal teeth on the inner plate; inner, outer and palp edged with simple, robust setae; palp large and 3-articulated.
Pleosome. Pleopods large and well developed, with 6-7 ramal segments in the endopod, 8 segments in the exopod; rami subequal to basis with paired coupling spines on the inner, distal margin of the basis; 6 simple setae present on the basis of pleopod 1.
Urosome. Uropod 1 rami subequal to peduncle. Peduncle with 2 rows of 2 robust setae. Terminal setae on each ramus consists of 1 large and 1 or 2 smaller robust setae. 3 interamal robust setae on inner and 2 on outer ramus.

Sexually dimorphic differences.
Based on immature female of 8.9 mm total body length.
In the absence of sexually mature males and females the only unique female characters found were: absence of palmate lobes on propodus and carpus of gnathopod 1, presence of pinched unguis on peraeopod 3, and presence of small, rudimentary oostegites on coxae of peraeopods 2-5.
Etymology. The name honours Kenneth J. Wildish who discovered and collected the new taxon in the Swale during the summer of 2011.

Epidermal pigment patterns. Absent.
Distribution. Known only from the type locality.

Discussion
The following arguments were considered in deciding how to name the unknown taxon. Relative growth data was available to show that the unknown taxon fundamentally differed from juvenile Orchestia mediterranea (Wildish et al. 2012;Pavesi et al. 2014). The relative growth of the unknown taxon differed from that in O. mediterranea by: -being slower growing with a reduced terminal moult size, -sexualization beginning at an earlier moult stage and with fewer moult stages per life history, and -dorsal pigment patterns being absent.
The first two of these three phenotypic characters are described as neotenous dwarfism and are the basic adaptations possessed by all driftwood specialist talitrids of the genus Macarorchestia Stock. Because the unknown taxon clearly does not belong to Macarorchestia: by possession of fully developed eyes and pleopods, by COI divergence differences (Wildish et al. 2012) and because it is clearly a driftwood specialist, both in habitat and adaptive morphology, the unknown taxon should be placed in a new genus.
On the other hand molecular evidence suggests that the differences between the unknown taxon and O. mediterranea are small (Wildish et al. 2012;Pavesi et al. 2014). Thus for the mitochondrial gene, COI, the difference for K2P = 2% with a divergence time of 0.83 ± 1.5 MYA (Pavesi et al. 2014). The low K2P% would not reach species level difference (assumed to be K2P = 3% in Amphipoda, Radulovici et al. 2009;4% in Wildish et al. 2012). It is hypothesized that the unknown taxon was formed from an O. mediterranea population which found itself in a secondary ecotope (driftwood) where a few, or single pleiotropic, nuclear gene mutations occurred which resulted in slower growth and sexualisation occurring at an earlier moult. Thereafter, the unknown taxon would follow life within driftwood and be unable to breed with O. mediterranea because of size and habitat differences. If this mutant form arose recently (in geological time), as the COI divergence time suggests, the magnitude of K2P differences may be an inapplicable measure of species or genus level difference. Clearly further genetic evidence to test this hypothesis is required: such as a search for the hypothesized nuclear gene(s) controlling neotenous dwarfism.
Pragmatically, and in the absence of conclusive molecular data, it is considered prudent to remove the unknown taxon from Orchestia, a supralittoral wrack generalist genus and create a new driftwood specialist genus: Neotenorchestia for the unknown taxon. Finding adult males and females of the new genus is needed to complete the description and diagnosis of the new taxon.
Inclusive of the taxonomic actions taken above brings the total genera of driftwood talitrids to three: Macarorchestia Stock 1989, "Platorchestia" chathamensis Bousfield 1982 and Neotenorchestia gen. n. For the latter genus only one species is known and with the transfer of microphtalma Amanieu & Salvat, 1963 from Orchestia brings the species belonging to Macarorchestia to a total of 5. Thus the known driftwood specialist talitrids total to date is 7 species.
The scarce locality records for each species documented here suggest either rareness and/or that they are difficult to find on shores of the northeast Atlantic and Mediterranean seas. Further evidence for this is that 2 of the 6 driftwood taxa dealt with here are known only from the type locality and the rest from only a few locations. A problem for future discoveries of driftwood talitrids is that the habitats are fast being destroyed by human activities. Documented examples include the destruction of the type locations for two species as mentioned above.
Both Macarorchestia and Neotenorchestia gen. n. probably originated from ancestors that were larger and faster growing. The evolutionary process in these genera involves reductions in metabolic and growth rates as well as sexualization occurring at an earlier moult number (neotenous dwarfism). A recent, common ancestor gave rise to modern O. mediterranea and N. kenwildishi gen. n., sp. n., which is consistent with morphological (relative growth) and molecular genetic studies (Wildish et al. 2012;Pavesi et al. 2014). The common ancestor of Macarorchestia is unknown, but within the genus of two lineages defined genetically (Wildish et al. 2012): evolution involves further neotenous dwarfism. Thus taxa to the left are larger and plesiomorphic, whereas those to the right are smaller and apomorphic. Neotenous dwarfism of this kind in driftwood talitrids poses a special challenge to taxonomy because many of the slope values between pairs of species are isometric. In these cases only regression constants, or plots, can be used to separate two species populations. Isometric relative growth is rare within the Amphipoda and where it does occur ratios cannot be used to express the relative growth differences and recourse to regression predictions appears to be the only way to handle the differences due to neotenous dwarfism. Permanent slides of limb and mouth parts prepared by Sara LeCroy (Gulf Coast Research Lab, University of Southern Mississippi) of four species of Macarorchestia identified possible, taxonomically important, morphological differences between groups 1 and 2 as defined in the preceeding paragraph. Thus the left mandible lacinia mobilis in Macarorchestia appeared to be predominantly 4-dentate (Table 3). In an immature male M. roffensis the teeth were lateral to the viewing plane and consequently easy to count. In all other preparations, including temporary ones, the teeth were dorsal to the viewing plane ("end on") making it necessary to focus up and down to see the teeth. An adult male of M. remyi appeared to be 5-dentate, whereas an immature female of this species was 4-dentate. Intraspecific variation of left mandible lacinia mobilis dentition has been recognized in other talitrids (Wildish and LeCroy 2014) and this may be the case in Macarorchestia, which renders this character of dubious value in taxonomic discrimination. Further studies are needed to resolve this point. The presence of long, simple, fine setae on the propodus of peraeopod 7 ("comb" setae) proved to be useful in distinguishing species of Macarorchestia (Fig. 3). Thus males of M. microphtalma had groups of 6 tufts: one near the first insertion of bifid-tipped, robust spines on the anterior side of peraeopod 7 (5 setae), distal (6 setae of 192 µm), anterodistal (0 setae), then 4 setal tufts on the posterior side of the peraopod at the first insertion of bifid-tipped, robust setae (5 setae), second (6 setae), third (5 setae) and fourth (4 setae). Setal length was approximately the same at each insertion, varying from 156 to 168 µm, except for the most proximal where it was 120 µm. Males of M. remyi and M. pavesiae sp. n., also had groups of 6 tufts on the propodus of peraeopod 7, but differed from M. microphtalma in having a single tuft at anterodistal position (2 setae) and only 3 tufts on the posterior propodus. A subadult male M. roffensis had a single tuft (2 setae, length 70 µm). This character is sexually dimorphic in all species of Macarorchestia and females have only two tufts at distal and anterodistal positions on the propodus of peraeopod 7 (each of 3-6 setae), which are of smaller setal length than in males. The number of setae in each tuft is growth dependant, so the number of setae per tuft cannot be used as a definitive taxonomic character.
In considering current data of Table 3 and assuming that the left mandible lacinia mobilis dentition is not taxonomically useful, it is possible to propose a subgeneric split based on Atlantic versus Mediterranean coastal ranges of Macarorchestia. This is clearly premature because: -of missing data as indicated by question marks in Table 3.
-The left mandible lacinia mobilis is 5 dentate according to Stock (1989), rather than 4 dentate, as would be the case if it were close to M. roffensis as the molecular data indicates (see Table 5 in Wildish et al. 2012). If intraspecific variation in dentition (4 or 5 dentate) is common this might explain this apparent anomaly.
Further molecular and morphological studies are needed to resolve the subgeneric status of Macarorchestia.

Introduction
Temnocephalida (Platyhelminthes, Rhabditophora) is the most diverse group of symbiotic turbellarians typically associated with crustaceans, with 122 valid species and 24 genera described in the world (Tyler et al. 2006(Tyler et al. -2012. Recently, Temnocephalida was confirmed as a monophyletic group included in Lymnotyphloplanida, which in turn makes up part of the Dalytyphloplanida clade, a major group of Rhabdocoela (Van Steenkiste et al. 2013). Within the Temnocephalida, the family Temnocephalidae Monticelli, 1899, is the most diverse, distributed in the Australian region with high species richness, but low host diversity, and in the Neotropics with an apparently lower number of temnocephalan species, but a greater diversity of host taxa (Damborenea andBrusa 2009, Sewell 2013). In fact, in the Neotropics, 32 species belonging to the genus Temnocephala and four taxa belonging to Didymorchis, endemic to this region and associated with crustaceans, mollusks, insects and chelonians have been described (Damborenea andCannon 2001b, Garcés et al. 2013 and cited therein).
The inventory work of the temnocephalan fauna in the Neotropics began in the 18 th century, when the first species of Temnocephalida was described, Temnocephala chilensis (Moquin-Tandon 1846), associated with anomuran crabs, Aegla laevis (Latreille), from Chile (Damborenea and Cannon 2001a). Since then, more than 50 studies have been published regarding aspects of the temnocephalan fauna in the Neotropics, including descriptions of new species, analyses of temnocephalan symbiotic community structure of particular host species, and studies with phylogenetic and biogeographic inferences (e.g. Damborenea 1998, Volonterio 2007a, Garcés et al. 2013). However, in many cases, information about the reported biodiversity in particular geographical locations of these rabdocoel turbellarians is scattered among myriad bibliographic sources and difficult to access. Therefore, attempts to generate inventories and compile information are highly valuable for understanding the global diversity of freshwater flatworms (Schockaert et al. 2008). The main objectives of this paper are to compile all the available published accounts on the symbiotic freshwater temnocephalans from the Neotropics and to incorporate new data derived from our own work of the last few years to construct a checklist of symbiont-host associations.

Bibliographic search
All the published records on Neotropical temnocephalan species reported from Malacostraca (Decapoda), Gastropoda (Caenogastropoda), Insecta (Hemiptera, Megaloptera, Plecoptera and Trichoptera) and Reptilia (Testudines) strictly in freshwater systems were compiled. Databases such as Biological Abstracts, Biological and Agricultural Index Plus and Scopus, Google Scholar, Helminthological Abstracts, ISI Web of Knowledge, Turbellarian Taxonomic Database and Zoological Record were used to ensure that we retrieved all available information; the bibliographic search was undertaken up to June, 2014. We considered all the studies whose datasets provide taxonomic information regarding the Neotropical temnocephalan taxa, even those found in a single individual host. Papers containing compiled records of Neotropical temnocephalans that require taxonomic revisions due to problems were indicated (e.g. Vianna and Melo 2002). The host species names were used according with IUCN (2014)

Survey work
Original data from our own studies of the last few years were included. A total of 11 taxa of decapod crustaceans of five families was examined for Argentine and Mexican temnocephalans. Furthermore, two species of Chelonia and one of Gastropoda from Argentina also were examined. Decapod crustaceans were collected with seine nets in one locality of central Mexico (Table 1). The collected decapod crustaceans were kept alive and examined for temnocephalans no more than 4 h after their capture. Decapod crustaceans were sacrificed and immediately examined for temnocephalans; external (e.g. carapace and claw surface) and internal structures (e.g. branchial cavity) were analyzed separately in Petri dishes with 0.65% saline solution, under a stereomicroscope. Gills from each decapod were also obtained and placed in tap water to search for temnocephalans. In the case of mollusk hosts, their mantle cavity was opened after sacrificing. Temnocephalan collections from live turtles were carried out by the catch-andrelease method (e.g. FAO 2012); therefore, the live turtles were identified directly in the field (L. Alcalde, personal communication). Temnocephalans were fixed with hot (steaming) 4% formalin or hot (steaming) distilled water. In some cases, specimens from the same host and with the same external aspect were fixed in 100% ethanol in the field for future molecular studies. All temnocephalans were processed following standard procedures (Sewell 2013). Species identification was achieved using specialized literature, and voucher specimens of some temnocephalans were deposited at the Colección Helmintológica of Museo de La Plata, Argentina (MLP-He) and the Colección Na

Results
In total, 60 papers have been published establishing host and locality records of the freshwater temnocephalan fauna in the Neotropics. The analysis of all available information (bibliographic and new original data) allowed us to establish a list of 38 symbiotic temnocephalan taxa in invertebrates and vertebrates in the Neotropical region, which are contained in four groups of hosts. Malacostraca (Decapoda): 4 taxa of Didymorchis associated with 3 taxa of crabs, 17 species of Temnocephala associated with 32 taxa of decapod crustaceans and only one species of Diceratocephala associated with one species of decapod crustacean; Gastropoda (Caenogastropoda): 5 species of Temnocephala associated with 5 taxa of freshwater snail hosts; Insecta: 1, 2, 2 and 1 taxa of Temnocephala associated with 1, 5, 3 and 1 taxa hosts of Trichoptera, Hemiptera, Megaloptera and Plecoptera, respectively; Chelonia (Testudines): 4 taxa of Temnocephala associated with 7 species of freshwater turtle hosts.
The results of this study are presented in the Table 1 which shows the symbiont-host list, where temnocephalans are organized by taxonomic groups and ordered alphabetically by family name. Then species within each family are listed alphabetically followed by authority name and date. The next category is the host species in which the temnocephalids were found, followed by the locality, and the bibliographic reference from which the information was obtained, except for those records established in the present work. In the temnocephalan species found in more than one host species, the latter are listed alphabetically, and host species for which more than one locality was recorded, are listed together. Furthermore, a host-symbiont list (See Appendix 1) is taxonomically and alphabetically organized.
The decapods are the most species-rich host group with temnocephalans (27), followed by the insects (5 taxa) and snails (5 species). Of the 38 taxa of Temnocephalidae listed in this work, all appear to be specific to particular host groups, while at least only one species of the family Diceratocephalidae have successfully associated with hosts after their anthropogenic introduction, i.e. Diceratocephala boschmai. The most widely distributed species are T. axenos, T. chilensis and T. iheringi, which are present in 9 and 10 crab host species and 5 snail host species, along 20, 25 and 49 localities, respectively.
In terms of hosts, Hydromedusa tectifera (a turtle) is the host with the highest temnocephalan species richness with 4 taxa, followed by Aegla neuquensis, A. platensis, Dilocarcinus pagei (decapod crabs) and Pomacea canaliculata (snail), all with 3 species; meanwhile, 49 host taxa show only one record of temnocephalid taxa for one locality.
The species accumulation curve for Neotropic temnocephalans plotted against the total number of species (Figure 1) shows irregular growth over 15 decades of studies in Temnocephalida (each decade divided into two periods of five years). This graph shows that the asymptote has not been reached yet and, if the systematic studies of the group are continued, a significant increase in the number of species in the Neotropical region can be expected. This graphic also reflects two important periods of research. The first shows the initial prospecting for temnocephalid species in the Neotropical region, between 1890 and the beginnings of the 20th century. The second period, beginning around 1970, shows an increase in the research on temnocephalans from different host species, with some stationary periods.

Discussion
The genus Temnocephala is an endemic component of the Neotropical region (Damborenea and Cannon 2001a). At the moment, it includes 35 taxa, of which 14 (40%) are considered microendemic (only one record for locality) ( Table 1). In total, 57 host taxa are associated with one or more temnocephalan taxa, which belong to seven orders and 14 families within four classes. It is worth pointing out that each major group of hosts is characterized by a particular assemblage of temnocephalan species, with host specificity at family level. For example, 17 taxa of Temnocephala are associated with three families of freshwater crab hosts (Aeglidae, Pseudothelphusidae and Trichodactylidae), while five Temnocephala species are associated with 8 taxa of freshwater shrimps included in three families (Cambariade, Palaemonidae and Parastacidae). Information about the natural history of this endemic genus is key to understanding the role of different factors that shaped its diversification patterns across several hydrological basins in the Neotropics and the possible implications of codivergence with host groups (see below) (e.g. Thompson 2005, Martínez-Aquino et al. 2014b. In this inventory, only Diceratocephala boschmai was detected as an introduced species because of translocation together with their crustacean hosts, the invasive redclaw Cherax quadricarinatus in Uruguay (Volonterio 2009a), due to human activities such as aquaculture and breeding of ornamental species (Lodge et al. 2012, Saoud andGhanawi 2013). According to several authors, D. boschmai causes a detrimental economic impact because of an aesthetic effect of the eggs on the body surface of the C. quadricarinatus (Herbert 1987, Volonterio 2009a. However, it is more important to mention the detrimental biological and ecological impact of these introduced  (Gelder 1999, Sicard et al. 2006, Witte et al. 2008, Tsuchida et al. 2011, Ohtaka et al. 2012. In this context, the data generated in this checklist can be used to support conservation strategies for freshwater biodiversity (Cardoso et al. 2011a, b, Stendera et al. 2012, Collen et al. 2013).
One hundred sixty eight years have passed since the first description and record of a temnocephalan from the Neotropics (Damborenea and Cannon 2001a), and, currently, ±236 records of temnocephalans have been published. However, considering the number of described species and the time passed, it can be stated that most of the diversity of Temnocephala remains yet to be described. There is also a significant number of potential hosts that have not been studied with regards to symbiotic temnocephalans. On the other hand, Schoackaert et al. (2008) mentioned that the few species recorded in South America were mostly recorded up to about 1970. Based on the species accumulation curve (Figure 1), this study shows clearly the increase in knowledge about the biodiversity of the temnocephalan fauna in recent times, but based on all of the information compiled for Neotropic temnocephalans, we show the necessity to continue inventory work. The Neotropic temnocephalan fauna contains 31% of Temnocephalida taxa described at the moment, representing 37 taxa allocated to two genera. This checklist presents data on almost all the extant species of temnocephalans along their distributional ranges in 11 Neotropical countries, which represents 35% of the total political territories (i.e. countries) in the Neotropics ( Figure 2). Argentina, Brazil and Uruguay are the countries with the most records of temnocephalans and with the most endemic species of Temnocephala, which are represented by 6, 9 and 4 species, respectively, while Colombia, Costa Rica, Mexico and Peru hold 1, 2, 1 and 1 endemic species, respectively. The relatively high number of records in Argentina, Brazil and Uruguay can imply that in these countries there are more research groups working with turbellarians compared to other Neotropical countries (e.g. Damborenea and Brusa 2008, Volonterio 2010, Amato et al. 2011. Therefore, the values of endemism for these particular countries are subjective -a function of the research effort -and it is probable that the endemism may be increased/decreased in future studies from different Neotropical countries. With regards to its exclusively Neotropical distribution, morphological evidence (mosaic syncytial plates) (e.g. Cannon and Joffe 2001, Damborena and Cannon 2001b), plus the recorded host specificities shown in this study (Appendix 1), allow for the inference that the biological radiation of Temnocephala may be the result of a complex combination of ancestral allopatric speciation processes (as a result of the separation of South America and Australia), plus the diversification of their host groups (e.g. Parastacidae) in South (and subsequent radiation in Central) America. For example, the species of Temnocephala associated with mollusks appear to be a morphologically homogeneous group with a phylogenetic structure (Volonterio 2007a, Damborenea andBrusa 2008). On the other hand, the almost exclusive distribution in the Southern Hemisphere of the family to which Temnocephala belongs (Temnocephalidae) is noteworthy and alludes to a Gondwanian origin (Gelder 1999, Cannon andJoffe 2001). However, a reliable molecular clock of the Temnocephalida is required to support or reject this hypothesis. Future studies combining research programs in integrative taxonomy (Schlick-Steiner et al. 2010, Ceccarelli et al. 2012, Fujita et al. 2012) with approaches of historical association (e.g. genes, organism and areas; see Page and Charleston 1998) will decipher the evolutionary history of Temnocephala.
At least 60 papers have been published dealing with the records of Neotropic symbiotic temnocephalans; however, the scarcity of studies in many countries is clear, and needs to be rectified. For example, some countries comprising complex geographic areas (i.e. Mexican Transition Zone, South American Transition Zone) only have one record of these turbellarians, and the diversity of the four major hosts groups is also unknown (Martínez-Aquino et al. 2014a). Therefore, we contend that future survey work should be strategic, aimed at enhancing the biodiversity inventory, combining identification of the host spectrum with choice of appropriate drainages based on biogeographic, faunistic, and hydro-  terstitial, some are pelagic and some are commensals or ectoparasites of other invertebrates. Pycnogonids are normally small animals; littoral species have a leg span of at most a few centimetres, while polar and deep-water species can achieve a leg span of 70 cm (Ruppert et al. 2004). The phylogenetic position of the Pycnogonida has long been controversial and is still under debate. Today pycnogonids are placed either within the Chelicerata as sister taxon of the Euchelicerata or as sister taxon of all other Euarthropoda (Dunlop and Arango 2005, Regier et al. 2010, Giribet and Edgecombe 2012. The phylogenetic relationships within the group are discussed as well. Traditionally the Pycnogonida were divided into eight families (Hedgpeth 1947) but with uncertain relationships. Recently studies on this point, using morphological and / or molecular characters were realized (Arango 2002, Arango 2003, Arango and Wheeler 2007, Nakamura et al. 2007). On the basis of these studies Bamber (2007) and Bamber and El Nagar (2013) suggested 11 families.
The Mediterranean Sea as we know it today developed about 5 million years ago. After the Messinian salinity crisis (5.6 to 5.3 Mya) the Mediterranean Sea was filled in a major flood approximately 5.3 million years ago, in which water poured in from the Atlantic Ocean and through the Strait of Gibraltar (Tichy et al. 2001). Consequently, the Mediterranean marine biota including the pycnogonid fauna is derived primarily from the Atlantic Ocean. The opening of the Suez Canal in 1869 created the first salt-water passage between the Mediterranean and Red Sea allowing some Red Sea species to invade the Mediterranean (lessepsian migration) . Anoplodactylus californicus, A. digitatus, and Pigrogromitus timsanus are regarded as being lessepsian migrants (Chimenz-Gusso and Lattanzi 2003).
In the Mediterranean Sea 56 pycnogonid species are known until now. A current species list of the Mediterranean pycnogonid fauna is found in Chimenz-Gusso and Lattanzi (2003), with the completion of Anoplodactylus nanus, described by Krapp, Kocak, andKatagan in 2008 (Krapp et al. 2008). The Mediterranean pycnogonid fauna consists mostly of littoral species, and deep-sea species are an exception. It is assumed that, because of the geological barrier caused by the Strait of Gibraltar (max. depth 286 m) and the temperature barrier in the deep-sea (Atlantic 1-4°C, Mediterranean 12°C), the deep water fauna in the Mediterranean is relatively species-poor . Deep-sea pycnogonid species adapted to cold water such as representatives of the Colossendeidae and Pallenopsidae are either absent or rarely occur in the Mediterranean. A similar barrier also exists between the western and eastern basin of the Mediterranean and between the Mediterranean Sea and the Black Sea.
There is a rapidly growing body of molecular studies especially on the Antarctic and Subantractic fauna (e.g. Mahon et al. 2008, Nielsen et al. 2009, Krabbe et al. 2010, Dietz et al. 2011, Weis and Melzer 2012b, Dietz et al. 2013, Carapelli et al. 2013. In contrast, there is a lack of publications with molecular studies for the Mediterranean. However, so far -as at January 2014 -11 Mediterranean pycnogonid species are listed in BOLD (Ratnasingham and Hebert 2007) and 10 species in GenBank (Benson et al. 2010). Detailed morphological analyses of Mediterranean Sea species are required to support future molecular studies, and specimens stored in natural history collections can provide a useful basis for these studies (see also Dunlop et al. 2007, Weis et al. 2011, Weis and Melzer 2012a. In the present study representatives of some of the major genera of Mediterranean pycnogonids were sourced from the Bavarian State Collection of Zoology. Additional material was collected during field trips. A significant objective of this study is to remove any ambiguity in species identifications by providing a pictorial atlas principally based on high resolution Scanning Electron Microscope (SEM) images. Classification follows Bamber (2007) and PycnoBase (Bamber and El Nagar 2013).

Material and methods
Material: The majority of specimens were sourced from the Bavarian State Collection of Zoology. Additional material was collected during field trips to Banyuls-sur-Mer, France (June/July 2006) and Rovinj, Croatia (September 2006 andMay 2007). Collecting sites are summarized in Figure 1. Details are given under material section of each species. All material was conserved and stored in 75% ethanol. Preparations were made according to methods described in Bolte (1996). Only adult animals were used for light microscopic and SEM imaging. Species determinations are based on the original descriptions and a variety of literature suitable for Mediterranean pycnogonids (e.g. Dohrn 1881, Bouvier 1923, Stock 1968, Bamber 2010. Synonyms followed PycnoBase (Bamber and El Nagar 2013) and Müller (1993). All specimens (SEM and alcohol material) used for this study are deposited at the Bavarian State Collection of Zoology.
Light microscopy: Light microscopic pictures were taken using an Olympus SZX stereo microscope and a Jenoptic Prog-Res C12 digital camera (2580 × 1944 px; 96 dpi; colour depth 24 bit). Larger specimens (Endeis charybdaea, E. spinosa) were photographed using a Wild-Heerbrugg M5A stereo microscope and a Canon Digital IXUS 850 IS digital camera (3072 × 2304 px; 180 dpi; colour depth 24 bit). Up to 12 pictures with different focus steps along the z-axis were combined to a single respective image with a greater field of depth using the computer software Auto Montage (Syncroscopy) or Helicon Focus (HeliconSoft).
Scanning electron microscopy: For SEM preparation, specimens were dehydrated in a graded acetone series (70%, 80%, 90%, 10 min. each, plus 3 × 100%, 20 min. each) and critical-point-dried in a Baltec CPD 030. Dried specimens were mounted on SEM stubs with self-adhesive carbon stickers and coated with gold on a Polaron Sputter Coater. SEM pictures (2048 × 1536 px; 72 dpi; colour depth 8 bit) were made with a LEO 1430VP at 10-20 kV. Scales were inserted using the measurement utility of the SEM.
Nomenclature: The present study follows the nomenclature in Arnaud and Bamber (1987) and Bamber (2010). Hence, the trunk of a pycnogonid is divided into 4 segments, the cephalon (=segment 1) and the segments 2, 3, and 4. The cephalon carries the proboscis anteroventrally, the ocular tubercle dorsally, and four pairs of extremities: the chelifores above the proboscis, the palps laterally, the ovigers ventrally, and one pair of legs laterally. Segments 2, 3, and 4 each carry one pair of legs laterally. The fourth segment carries the abdomen posterodorsally. Some species do not possess palps or chelifores. Ovigers are present in males (with few exceptions, e.g. Pycnogonum subgen. Nulloviger). In some genera ovigers are reduced or absent in females. The number of articles of the chelifores, palps, and ovigers varies within the systematic groups. Each leg is composed of nine articles: first, second, and third coxa, femur, first and second tibiae, tarsus, propodus, and claw. The first coxa articulates with the lateral process of the trunk while the second coxa carries the genital opening ventrally. In males, the cement glands are on the femur and rarely on other articles. The claw is often flanked by two smaller auxiliary claws. All abbreviations used in the figures are provided in Table 1.

General remarks
This study is based on 21 Mediterranean species. In all cases, the major morphological characteristics correspond with published descriptions; exceptions see remarks section of each species.

Achelia vulgaris (Costa, 1861)
Here, only coxa 2 of the first leg has three protuberances with spine on the one and two on the other side; the coxa 2 of the other legs has two protuberances with spine on each side (Fig. 11C, D). The other characters of the male, like lateral process not touching each other (Fig.  11B) correspond with the descriptions of Costa (1861), Dohrn (1881) and Bouvier (1923).
Here all 6 specimens from 4 different locations have more or less rudimentary auxiliary claws ( Fig. 31F; Fig. 33F, G). Remarks. According to Dohrn (1881) and Bamber (2010) each lateral process and coxa 1 is armed with one protuberance. Here, in the males the lateral process 1 and the coxa 1 of the first leg has two protuberances (Fig.  42D, F), the remaining legs correspond with previous observations (e.g. Fig. 42E, G). The females are without such protuberances (Fig. 44C, D).

Remarks.
In literature there is diverse information about the articulation of the oviger, 7 articles in Dohrn (1881) and 6 articles in Bouvier (1923) and King (1986) are reported. Here the ovigers have 6 articles.

Remarks.
In the literature there is diverse information about the articulation of the oviger, 7 articles in Dohrn (1881) and 6 articles in Bouvier (1923) and King (1986) are reported. Here the ovigers have 6 articles (Fig. 50E, F Ledoyer (1986) described ?Oedicerina megalopoda from the Mozambique Channel, western Indian Ocean, also based on a single anterior fragment. This species was defined by the characteristic shape of coxa 4 and by the strongly developed carpal articles of gnathopods 1 and 2. The description of a third species, O. denticulata Hendrycks & Conlan, 2003 from the northeast Pacific Ocean, was the first to record complete specimens and was accompanied by a more detailed appraisal of the genus.
The new material from the Bergen Museum has made possible a re-description of O. ingolfi. Specimens belonging to Oedicerina from the northeast Atlantic Ocean and from New Zealand waters demonstrate further diversity within the genus and require the recognition of two new species which are described below.
This paper is the third to utilize material sorted at Skibotn in 2009, following d'Udekem d' Acoz (2010) and Krapp-Schickel and Vader (2013).

Material and methods
Norwegian Sea material assigned to O. ingolfi from the Natural History Collection of the University Museum of Bergen (ZMBN) was collected with an RP sledge (Rothlisberg and Pearcy 1977) by Torleiv Brattegard (Brattegard and Fosså 1991) in the period 1981-1986. Additional material from the Norwegian and Greenland Seas from the Museum of Zoology, Lund University was collected during the NORBI expedition (Dahl et al. 1976) using an epibenthic sledge (drague Sanders) (Guennegan and Martin 1985).
Specimens from the Discovery Collections at the National Oceanography Centre, Southampton were obtained in the East Iceland Basin on an RRS Discovery cruise that contributed to the Institute of Oceanographic Sciences investigations of mid-water and benthic faunas in the eastern North Atlantic Ocean (1965)(1966)(1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975)(1976)(1977). The material, from an epibenthid sledge, was fixed in 4% formaldehyde and later transferred to 70% Industrial Methylated Spirits. These specimens have been deposited in the Amphipoda collections at The Natural History Museum, London.
The New Zealand material was collected during the Ocean Survey 2020 expeditions with RV Tangaroa to the Chatham Rise and the Challenger Plateau (Knox et al. 2012) by means of a "Brenke" epibenthic sledge (Brenke 2005). The material was sorted on board, fixed in 96% ethanol and later transferred into 70% ethanol. It has been deposited in the National Institute of Water and Atmosphere Research (NIWA) Marine Invertebrate Collection in Wellington, New Zealand.
For habitus drawings the specimens were transferred into glycerol on a cavity slide. Specimens were then dissected under a stereomicroscope (Leica M205 or Wild M5) using dissecting needles. Mouthparts and appendages were mounted temporarily in glycerol on slides for microscopic examination and drawing. Appendages were later mounted as permanent slides with glycerol jelly, or transferred into small glass microvials. Microvials were stoppered with a cotton ball wrapped in Japan paper to avoid the appendages being entangled in the cotton fibres.
After dissection, mouthparts and appendages of Discovery Collection material were made directly into permanent mounts using Polyvinyl-lactophenol stained with lignin-pink. Drawings of habitus and appendages were made using a camera lucida attached to a compound microscope (Leica DMLB or Wild M20). Pencil drawings were scanned, inked digitally and arranged to plates using the methods described in Coleman (2003Coleman ( , 2009.
Body lengths were measured along the dorsal outline from the tip of the rostrum to the end of the telson. Lengths of individual articles of gnathopods and pereopods measured along anterior or posterior margins can vary depending on the degree of flexure of the appendage. All articulations except those between coxae and tergites and between merus and carpus of gnathopods are bicondylar. Measurements made between condyles gives a length that is not affected by limb flexure. Length ratios herein have been derived using this principle.
Urosome. Urosomite 1 (Fig. 1a) longest, with an inconspicuous boss close to the posterior margin; urosomite 3 longer than 2, with short, acute mid-dorsal projection. Uropod 1 (Fig. 5e): peduncle about as long as outer ramus, margins with short setae; inner ramus 1.3 × length of outer ramus, with small setae on both margins; outer ramus with setae on lateral margin only. Uropod 2 (Fig.  5f): peduncle slightly tapering, with short setae on both margins; inner ramus 1.7 × length of outer ramus, with short setae on both margins; outer ramus with setae on lateral margin only. Uropod 3 (Fig. 5g) peduncle short, about as long as telson, with ventral subacute projection; rami subequal, plumose setae on lateral margins. Telson (Fig. 1h) tapered, notched 30%. Sexual dimorphism. Male antenna 1 with shorter peduncle articles in the ratio 1:0.7:0.3 and more numerous flagellum articles compared to female. Article 1 of the flagellum is elongate, about as long as peduncle article 3. Subsequent proximal articles are shorter than wide. The 1-articulate slender accessory flagellum is about 1/3 as long as article 1 of the primary flagellum.
Remarks. Stephensen's (1931) specimen was damaged and incomplete. Only the head and pereonites 1-4, pereopods 3-4 and coxae and bases of pereonites 1-2 on one side were available for study. The material from the Bergen Museum used for this description consists of numerous specimens of all sizes, both female and male, and was col-lected relatively close to the type locality of O. ingolfi, but nevertheless we cannot be absolutely sure that our material represents Stephensen's species (see discussion below).
Etymology. The specific name vaderi recognises the important contributions to amphipod studies made by Professor Wim Vader.
Variability. The paratypes bear a small posteriorly directed tooth on pleonite 3. It may be that this process has been present and is worn down in the holotype. Antenna 1 of the female (Fig. 6b) has a longer and more slender peduncle and fewer and more elongate flagellum articles compared to the male.

Distribution. Chatham Rise, east of New Zealand.
Remarks. The female specimen has the same antenna 1 morphology as the male: short peduncle articles and numerous flagellum articles. The proximal articles of the flagellum are shorter than wide. Remarks. Only the head and pereonites 1-2 are present. Coxae 1-2 bear long setae along the distal margins. The animal appears similar to O. ingolfi, but as a result of incompleteness it is impossible to attribute it to any species.

Discussion
The three species described herein are morphologically very similar. Mouthparts and appendages show only minor and subtle differences and the species are best discriminated by habitus characters. Two of the species, O. ingolfi and O. loerzae sp. n. have mid-dorsal carinae on pleonites 1-2. Oedicerina ingolfi differs from O. loerzae sp. n. in having a small, slender, acute, upright tooth on pleonite 3 and a small pointed process on the posterior margin of urosomite 3 both of which are absent in O. loerzae sp. n. Oedicerina vaderi sp. n. has a small pointed process on the posterior margin of urosomite 3, as found in O. ingolfi, but pleonites 1-2 are evenly vaulted lacking any trace of a carina. Pleonite 3 of the holotype of O. vaderi sp. n. appears to be dorsally unarmed, but the paratypes have a small acute process (see female paratype, 6.3 mm in Fig. 10e). Coxa 5 of O. ingolfi is longer than wide, that of O. vaderi sp. n. is about as wide as long, and that of O. loerzae sp. n. much wider than long.
Oedicerina sp. indet. was collected from the warm-temperate east Atlantic Ocean off Western Sahara. The unique specimen is incomplete, only the head and pereonites 1 and 2 are present, preventing a full identification.
Apart from the type species O. ingolfi, two other species had been described in the genus prior to this study: Oedicerina megalopoda Ledoyer, 1986, collected close to Mayotte, Mozambique Channel, western Indian Ocean (200-500 m) and Oedicerina denticulata Hendrycks & Conlan, 2003 from the northeast Pacific Ocean off California (4050 m). Knowledge of O. megalopoda is limited as the unique specimen is incomplete, but the massive rostrum of this species differs markedly from all other species of the genus. The palm of gnathopod 2 of O. megalopoda is straight, similar to that of O. loerzae sp. n., and thus different from the convex pattern seen in both North Atlantic species. In Oedicerina denticulata pleonite 1 is smooth and pleonites 2 and 3 have a posteromarginal process. The process on pleonite 3 is directed posteriorly and is reminiscent of that seen in paratype material of O. vaderi sp. n. thus contrasting with the upright condition found in O. ingolfi. The posterior margin of urosomite 3 bears a small process (as do all species except for O. loerzae sp. n.) and urosomite 1 has a small upright process in the male but not the female, a character unique within the genus. Coxa 5 appears to be longer than wide, as in O. ingolfi.
The mouthparts and appendages of all species of this genus are remarkably similar to each other. Examination of the extensive Norwegian Sea material which we attribute to O. ingolfi indicates that intraspecific variability is minimal, except for sexual dimorphism in antenna 1. In females of both Atlantic species peduncle articles of antenna 1 are longer and more slender than in males. Flagellum articles in females are uniformly much longer than wide and relatively few in number, whereas in males they are more numerous and proximally wider than long, forming an incipient callynophore. This sexual dimorphism is not apparent in O. loerzae sp. n. where the structure of antenna 1 is very similar in males and the one ovigerous female paratype.
Because of minimal differences among appendages and mouthparts in Oedicerina species, differentiation within the genus relies significantly on patterns of ornamentation of pleonites and urosomites. As the posterior segments of the type material of O. ingolfi are missing, the question remains as to which of the two Atlantic species represents the species that Stephensen described. We allocate the material from the museums in Bergen and Lund studied herein to O. ingolfi on geographical grounds in that it was collected much closer to the type locality of that species, and on the morphological grounds of the dense fringe of setae on the distal margins of coxae 1 and 2 and the shape of the rostrum that our material shares with Stephensen's original description.

Introduction
The far north of Madagascar comprises a mosaic of heterogeneous landscapes ranging from rainforests on volcanic basement to deciduous dry forests in karstic massifs and littoral habitats on sandy ground (e. g. Lavranos et al. 2001, Vences et al. 2009, Crottini et al. 2012. The geological and climatic diversity of this area is reflected by a high species diversity and a high degree of microendemism (e. g. Andreone 2004, Wilmé et al. 2006, Ranaivoarisoa et al. Montagne des Français and adjacent littoral habitats (Ramanamanjato et al. 1999, Raselimanana et al. 2000, Glaw et al. 2001, Köhler et al. 2010a, 2010b, Miralles et al. 2011. These major habitat types are separated from each other by rather steep ecotones in northern Madagascar and thus in part constitute "habitat islands" for several species, possibly allowing allopatric speciation. Several taxa including dwarf frogs (Stumpffia), dwarf chameleons (Brookesia), burrowing skinks (Paracontias), leaf-tail geckos (Uroplatus), and the nocturnal geckos of the genus Paroedura have undergone remarkable diversification in northern Madagascar (Jackman et al. 2008, Köhler et al. 2010a, 2010b.
The genus Paroedura is widely distributed throughout Madagascar's biomes, including eastern rainforest, western dry forest, extremely arid thornbush savanna and high mountain habitats (Angel 1942, Guibé 1956, Dixon and Kroll 1974, Rösler and Krüger 1998. Five of the 15 described species from Madagascar occur in the far north, and two additional species from the Comoro islands have close relationships to the northern P. lohatsara and P. stumpffi suggesting that they originated by two colonization events from the northern species . Recent surveys indicate that the karstic limestone massifs in this region still harbour further undescribed reptile species (e. g. Jackman et al. 2008, Recknagel et al. 2013. Some of them might be microendemic and threatened by substantial habitat destruction. In the following we describe a new Paroedura species from Montagne des Français to contribute to the taxonomic inventory of this massif, and to highlight the threats affecting this microendemic species and other biota in the region.
To obtain molecular comparisons of the new species with previously unstudied nominal Paroedura species, we sequenced a fragment of the mitochondrial gene for cytochrome oxidase subunit I (cox1) with primers and protocols defined by Nagy et al. (2012) for several Paroedura samples from Tsingy de Namoroka (corresponding voucher specimens to be catalogued in MNHN). These samples have become available through a recent herpetological survey by one of us (II) and include topotypical P. karstophila, previously unstudied from a molecular perspective. The resulting sequences were combined with those of Nagy et al. (2012) and Koubová et al. (2014) in MEGA version 5 (Tamura et al. 2011) to yield an alignment of 664 bp. We performed phylogenetic inference in MEGA under the Maximum Likelihood optimality criterion, with a general time-reversible substitution model with gamma-distributed rates and invariant sites, NNI branch swapping, and assessing robustness of nodes with 500 bootstrap replicates. Newly obtained sequences were submitted to GenBank (accession numbers: KM978078-KM978080).

Molecular differentiation in the genus Paroedura
The multigene phylogenies of Jackman et al. (2008) and Hawlitschek and Glaw (2013) as well as the combined cox1 sequences of Nagy et al. (2012), Koubová et al. (2014) and those newly obtained (Fig. 1) provide evidence of strong genetic divergence of P. hordiesi to all other described Paroedura species except P. vahiny for which DNA sequence data is not yet available.
The cox1 data place the new species sister to an undescribed candidate species from Nosy Hara (79% bootstrap support), and this clade forms part of a more inclusive clade with P. oviceps and the undescribed Ankarafantsika species (bootstrap support 69%; no cox1 sequences were available for P. homalorhina from Ankarana which in a previous study was the sister taxon of P. hordiesi, see Jackman et al. 2008). All studied species and candidate species of Paroedura included in the cox1 data set were differentiated by very high pairwise uncorrected distances. P. hordiesi differed from its Nosy Hara sister lineage by a p-distance of 16.3-16.5%, and from all other Paroedura including the relatively distantly related P. karstophila by >20%. Two samples sequenced from the Tsingy de Namoroka, the type locality of P. karstophila, were sister to each other but showed a substantial divergence of 15.9% cox1 p-distance, suggesting that possibly two cryptic species may be hiding under the name P. karstophila in this karstic massif. The sample from Ankarafantsika (P. sp. Ankarafantsika in Fig. 1), identified in Jackman et al. (2008) as P. karstophila, in fact belongs to yet another, undescribed species according to our subsequent comparisons and is only distantly related to P. karstophila (Fig. 1). The DNA barcoding voucher sequence for the karyotype of P. karstophila (Koubová et al. 2014) is identical to our sequences of P. hordiesi, suggesting that the karyotype description in this paper refers to our new species rather than to P. karstophila.

Description of a new species
Paroedura hordiesi sp. n.
The new species can be easily attributed to the genus Paroedura based on its nested phylogenetic position within the genus (Jackman et al. 2008) and its morphological similarity to other Paroedura species, especially concerning the ventral structure of their fingers and toes which comprise a pair of squarish terminal adhesive pads. Among Malagasy geckos this terminal toe structure is only found in Paroedura and the related genus Ebenavia . The latter genus can be easily distinguished from Paroedura by its much narrower and strongly pointed head, its elongated body, and smaller size.

Comparisons.
The new species can be distinguished from the 17 other currently recognized Paroedura species (including the three available junior synonyms in the genus) as follows: From P. androyensis, P. bastardi, P. ibityensis, P. lohatsara, P. maingoka, P. picta, and P. vahiny by having the nostril in contact with the rostral scale; from P. gracilis by absence of a raised vertebral ridge on the body and shorter forelimbs which are not extending forward beyond tip of snout; from P. masobe by much smaller size (SVL up to 58 mm versus 107 mm), much smaller eyes with a pigmented iris (versus black iris) and absence of a dorsal row of paired spines on the tail; from the two Comoroan species P. sanctijohannis and P. stellata by slightly smaller size (SVL up to 58 mm versus 68 mm and 62 mm, respectively) and absence of whorls with distinct spiny tubercles of the original tail; from the syntopically distributed P. stumpffi by smaller size (SVL up to 58 mm versus 70 mm) and absence of whorls with distinct spiny tubercles of the original tail; from P. tanjaka by much smaller size (SVL up to 58 mm versus 102 mm) and absence of whorls with distinct spiny tubercles of the original tail; from P. vazimba by larger size (SVL up to 58 mm versus 49 mm) and absence of whorls with distinct spines of the original tail; from P. oviceps from its type locality Nosy Be by smaller size (SVL up to 58 mm versus 69 mm) and rather regularly arranged tubercle rows on the back (versus rather irregular rows of tubercles); from P. karstophila by the absence of whorls with distinct spiny tubercles of the original tail (and by a smoother regenerated tail, see Nussbaum and Raxworthy 2000) and by colouration (see Fig. 2 versus Fig. 5); and from its close relative P. homalorhina (Jackman et al. 2008) by shorter limbs (finger tips reach the anterior margin of eye versus snout tip when forelimbs are adpressed along the body), slightly smaller size (SVL up to 58 mm versus 65 mm, see Table 2), distinct and generally regularly arranged tubercles rows on the back (versus less distinct and less regular rows), and a more slender habitus (see Fig. 2 versus Fig. 5). In addition, Paroedura hordiesi can be easily distinguished from most other Paroedura species (P. androyensis, P. bastardi, P. gracilis, P. ibityensis, P. lohatsara, P. maingoka, P. masobe, P. oviceps, P. picta, P. sanctijohannis, P. stellata, P. stumpffi, P. tanjaka, P. vahiny and P. vazimba) by adult colouration in life (see colour photographs in , Schönecker 2008) and from P. androyensis, P. bastardi, P. gracilis, P. homalorhina, P. ibityensis, P. lohatsara, P. maingoka, P. masobe, P. picta, P. sanctijohannis, P. stellata, P. stumpffi, P. tanjaka, and P. vazimba by juvenile colouration , Schönecker 2008 juvenile colouration of the other species still unknown). Genetically, P. hordiesi can be distinguished from all other species in the genus by its molecular differentiation in mitochondrial and nuclear genes (Jackman et al. 2008, Nagy et al. 2012itschek and Glaw 2013, Fig. 1) except for P. vahiny for which DNA sequences are not yet available.
Description of the holotype. SVL 52.6 mm, further measurements and counts are given in Table 1. Holotype in good condition, with complete but broken original tail and everted hemipenes. Head distinctly wider than neck and wider than body. Snout angled downward to tip, slight depression between poorly developed canthal ridges. Ear opening is a vertical slit. Original tail slightly shorter than snout-vent length, nearly round in cross section in its proximal part, laterally compressed in its distal half, with sharply pointed tip; ventral pygal section with pair of postcloacal sacs. Digits moderately expanded at tips. Rostral scale rectangular, much wider than tall and wider than mental. Nostril in contact with Table 1. Morphometric and meristic variation of several type specimens of Paroedura hordiesi from the type locality Montagne des Français. Abbreviations for measurements and counts (see Materials and Methods for other abbreviations): ZSM = Zoologische Staatssammlung München; SVL = snout-vent length; TL = tail length; HL = maximum head length (from tip of snout to posterior margin of ear); HW = maximum head width, at widest point; HH = maximum head height; AGL = axilla-groin distance; ED = maximum eye diameter; EO = maximum ear opening diameter; FOL = forelimb length, from axilla to tip of longest finger; HIL = hindlimb length, from groin to tip of longest toe; SPL = number of supralabial scales; IFL = number of infralabial scales; NAS = number of nasals in direction from rostral to labial including nasorostrals, supranasals, postnasals; IN = number of internasals; IO = number of interorbitals; PM = number of postmentals; SLM4 = number of subdigital lamellae on fourth digit of manus; SLP4 = number of subdigital lamellae on fourth toe of pes; PCT = number of postcloacal tubercles; TLT = number of tubercle rows on tail. Counts are listed left-right. All measurements in Tables 1 and 2  rostral, first supralabial, and five further scales. First supralabial largest, labials smooth. Snout and interorbital scales juxtaposed, some raised, some scales in front of orbits tuberculate, as are some larger lateral occipital scales. Dorsolateral neck and body scales very heterogeneous with about eight partly poorly recognizable longitudinal rows at midbody of enlarged tubercles; enlarged tubercles separated partly by small flat scales and smaller tubercles. Dorsal scales of forelimbs mostly flat. Dorsal scales of hindlimbs large and tuberculate, much smaller above knee. Ventral scales of forelimbs slightly smaller than surrounding ventral scales of the body. Dorsal pygal scales like dorsal body scales; lateroventral pygals tuberculate. Tail scales flat, tail segments without any transverse row of spiny tubercles. Mental triangular, bordered posteriorly by a pair of elongate, irregular hexagonal postmentals. Postmentals contact mental, first and second infralabial, one enlarged lateral gular, one smaller posterolateral gular, and one larger central gular. First three infralabials not significantly larger than others. Gulars small, granular. Ventrals of chest and abdomen flat. Proximal subdigitals in rows of 2-3. Pair of squarish, terminal pads. Claws curving downwards between terminal pads of digits.
Colour after 10 years in alcohol (Fig 3): head dorsally beige with almost no recognizable pattern except a whitish dorsolateral spot above each ear opening and a well-defined blackish area above each eye which represents, however, no pigmentation of the skin, but is due to the blackish eyeball shining through the skin. Dorsum greyish with four poorly defined beige spots on the back which are the remains of dorsal crossbands: one distinct spot between the forelimbs, two poorly recognizable spots on the back between forelimbs and hindlimbs, and a fourth poorly recognizable spot between the hindlimbs. Dorsal surfaces of forelimbs and hindlimbs marbled with beige and grey. Flanks with similar colour as dorsum, but without any recognizable traces of light crossbands. Tail dorsally with whitish-grey and brown alternating transverse bands which are poorly delimited in the distal portion of the tail. Throat, chest, venter, ventral parts of forelimbs and hindlimbs and ventral side of tail whitish. Iris dark grey, pupil white. Colour of holotype in life unknown (no colour photographs available).  Variation. Morphometric and meristic variation of ten specimens of the type series is summarized in Table 1, but there is also a remarkable individual variation in dorsal colouration and pattern (Fig. 2). Paratype ZSM 2106/2007, the smallest known juvenile (SVL 28.1 mm, tail length 25.2 mm), shows a moderately distinct juvenile colouration both in life (Fig. 2) and in preservative consisting of four light crossbands on the back between the insertion of forelimbs and the insertion of hindlimbs. A similarly distinct banding on body and original tail is still visible in the larger juvenile paratype ZSM 2107/2007 (Fig. 2, SVL 35.2 mm, tail length 34.7 mm). Although the juvenile colouration is distinct compared to most adults, it is less colourful and less contrasting in comparison with many other Paroedura species. The adult paratypes usually have a less distinct dorsal colour pattern than the juveniles, ranging from poorly recognizable (e. g. ZSM 532/2000) to distinct and well delimited (e. g. ZSM 531/2000) after 14 years in alcohol. Additional photographs of living individuals are provided in Henkel and Schmidt (1995), , and Schönecker (2008). Most of the adult type specimens have a regenerated tail (without distinct transversal banding of dark and white), indicating a high pressure by predation or intraspecific aggression.

Hemipenis.
The following description is based on the everted right organ of ZSM 343/2004 (Fig. 4). Hemipenis medium-sized (total length 6.4 mm, width at apex 4.5 mm), apex divided in two lobes. Sulcus spermaticus forms an S-shaped, narrow and deep groove, bordered by moderately distinct lips reaching the apex. On the apex, the sulcus is becoming broader and divides in two branches. A horn-like apical cone is present in the center of each lobe. Calyces on the apex are only present at the border of the lobes. The hemipenes of the holotype (ZSM 342/2004) are virtually identical in every respect.
Habitat and habits. Paroedura hordiesi was observed multiple times at night in karstic dry forest in the rainy season, mainly climbing on karstic rocks and the ruins of an old fort. It was found in close syntopy with P. lohatsara in the karstic limestone areas, whereas P. stumpffi was only encountered on the slope between the massif and the sea, in areas without karstic formations.
Etymology. The specific name is dedicated to Freddy Hordies, in recognition of his support for biodiversity research and conservation through the BIOPAT initiative.
Distribution and conservation status. Paroedura hordiesi is reliably known only from the recently established nature reserve of Montagne des Français. The species possibly also occurs at Ampombofofo, ca. 25 km north of this massif (Megson et al. 2009: ZSM 1531, but the identity of this population needs further study as molecular data are not available thus far. Surveys in other limestone areas of northern Madagascar revealed a superficially similar Paroedura at Nosy Hara (Metcalf et al. 2007) which is, however, strongly differentiated by colouration (see photo in  and mitochondrial DNA sequences ( Fig. 1 and Nagy et al. 2012), and is therefore considered a further undescribed candi-  Raxworthy 2000, Raselimanana 2008). Our own surveys in Ankarana revealed only P. homalorhina and P. stumpffi and we consider the occurrence of P. hordiesi unlikely in this massif. Rakotondravony (2006a) found Paroedura sp. at Binara in the Loky-Manambato region which potentially could refer to P. hordiesi, but the identity of this record remains to be studied. No unidentified Paroedura species was found at Analamera (Rakotondravony 2006b). Thus current evidence suggests that P. hordiesi is microendemic of Montagne des Français and perhaps the adjacent Ampombofofo region. For consistency with the IUCN Red List Assessment for Paroedura lohatsara (Raxworthy et al. 2011) and other potential microendemic species of the Montagne des Français region, we suggest a classification as "Critically Endangered" on the basis that P. hordiesi has an extent of occurrence of at most 50 km², it is known from a single location, and there is a continuing decline in the extent and quality of its habitat.

Discussion
With the description of Paroedura hordiesi we add a further, probably microendemic new species to the herpetofauna of Montagne des Français. Although this population is already known for approximately 20 years, several factors have hampered the clarification of its identity, including the variability in colouration and in distinctiveness of longitudinal rows of dorsal tubercles of P. hordiesi, P. homalorhina and P. karstophila, the existence of several undescribed species with similar key characters, the rarity of individuals with an original tail, and the absence of genetic data and colour photographs reliably referable to P. karstophila. This situation has led to substantial uncertainty about the correct name for the species from Montagne des Français, as is reflected by the different preliminary names that have been used for this species in the literature (see above). DNA sequences are now available from two P. karstophila-like specimens from the type locality Namoroka (Fig. 1). These strongly differ from each other, suggesting that P. karstophila is possibly a composite taxon including two cryptic species which occur in syntopy at Namoroka. Although the identity of P. karstophila remains to be further studied, we have little doubts that one of the two Namoroka lineages will turn out to correspond to this taxon.
Due to PCR failure with universal reptile primers (Nagy et al. 2012), no cox1 sequences are available for several species of Paroedura, including P. homalorhina which was sister to P. hordiesi in the study of Jackman et al. (2008;as P. sp. n.). On the other hand, the new species from Nosy Hara and P. karstophila have not yet been sequenced for the genes used by Jackman et al. (2008). This unfortunate lack of overlap of the two molecular data sets, to be remedied by future sequencing efforts, hampers our ability to determine whether P. hordiesi is more closely related to P. homalorhina, or to the undescribed species from Nosy Hara. It does not, however, compromise our taxonomic conclusion of P. hordiesi being a valid species differentiated from both these close relatives by a substantial genetic divergence and morphological characters.
Paroedura hordiesi is a typical example of a microendemic karst specialist of an isolated habitat island that apparently has lost its ability to survive outside its special habitat and thus has lost the genetic exchange with populations from neighboured karstic massifs. Although the surrounding massifs are only separated by several kilometers of grassland or other non-karstic habitats, they are often populated by different species which might have evolved in isolation for millions of years. The cox1 data presented in Fig. 1 suggest that the undescribed Paroedura species from the island Nosy Hara is a close relative of P. hordiesi. The two species differ by a substantial genetic distance, and distinctly by their colouration (see photographs in , although the distance between this island and Montagne des Français is just about 35 km. The distance between Montagne des Français and Ankarana is approximately 50 km, yet the latter massif is populated by P. homalorhina, the second close relative of P. hordiesi (see discussion in previous paragraph). Similar patterns of microendemism and comparable phylogenetic relationships have been found in dwarf chameleons of the genus Brookesia , in a clade of dwarf frogs of the genus Stumpffia (Köhler et al. 2010a), and in lizards of the genus Zonosaurus (Raselimanana et al. 2009, Recknagel et al. 2013, suggesting that similar vicariant processes among karstic habitat islands might have affected these groups of organisms in the far north of Madagascar. The karstic massifs might have provided sufficient resources and protection from desiccation during periods of drier climate to allow the long-term survival of these populations, finally leading to completed speciation. The evolution and long-term survival of these microendemic species suggests rather stable conditions without catastrophic events over very long time periods. The apparent lack of gene flow among the karstic habitat islands also suggests a limited dispersal capacity of karst specialists across the non-karstic interspersed matrix. by a cooperation accord between the UADBA and the ZSM. Malagasy authorities are acknowledged for collection and export permits. The research of F. Glaw was supported by the "Deutsche Forschungsgemeinschaft" (grant no. GL 314/1) and the European Association of Zoos and Aquaria (EAZA). Fieldwork of I. Ineich was made possible with the help of Ewe Madagascar for logistical support (Marc Gansuana), La Colas and the Société Aquamas for providing boats, thanks to them. He also wants to thank Lucile Allorge and Thomas Haevermans (MNHN) for scientific organisation, MNHN Labex for providing founds, and the Parc naturel du Namoroka, its guides and the inhabitants of the village of Vilanandro for their useful help.

Introduction
Most West African countries have lost the majority of their natural forests. Ivory Coast has been experiencing a great loss of rainforest cover in the west (e.g. Chatelain et al. 1996), and this has been worsening during the political crisis in the first decade of the 21 st century (Bible 2013, Hansen et al. 2013). This situation is also severe in other parts of this country, although forest loss in these other areas has received much less attention. The rainforests in south-eastern Ivory Coast are actually among the most highly threatened African forests (Norris et al.  Mayaux et al. 2013). In particular they face logging, shifting agriculture and poaching (Lauginie 2007).
Although, since 1926, the Ivorian State created protected areas throughout the country (Lauginie 2007), the eastern forested areas were neglected (Bakarr et al. 2004). To reinforce its conservation policy, the Ivorian state is now encouraging compensatory measures, i.e. the creation of volunteer nature reserves (VNR), in addition to protected areas. One of these VNRs is the Tanoé-Ehy Swamp Forests, which has been designated as a "Very High" priority area for primates. To further enhance protection it has been recommended to collect zse.pensoft.net Kpan, T. F. et al. : The anuran fauna of the Tanoé-Ehy Swamp Forests 262 and update the scientific information for this forest area (Koné et al. 2008).
Recently it has been shown that the amphibian faunas of the western (western Ivory Coast and westwards) and eastern (eastern Ivory Coast to the Dahomey Gap in Benin and Togo) Upper Guinea forests are distinctly different (Penner et al. 2011). The discovery and descriptions of several new species from eastern Ivory Coast and western Ghana, further underline the faunistic uniqueness of this region (Assemian et al. 2006, Rödel et al. 2009a, 2009b, Kouamé et al. 2014a.
So far the scientific knowledge of the Tanoé-Ehy Swamp Forests is fragmentary (Koné and Akpatou 2004, Ahon 2010, Zadou et al. 2011) and completely excludes amphibians. However, amphibians are important parts of tropical ecosystems and provide numerous ecosystem services (Mohneke andRödel 2009, Hocking andBabbitt 2014). Furthermore representatives of this taxonomic group are known to react sensitively to habitat alteration. The composition of amphibian assemblages thus may reflect the degree of habitat degradation and destruction (for Ivorian examples compare e.g. Ernst and Rödel 2005, Ernst et al. 2006, Hillers et al. 2008. Hence, conservation recommendations could be based on the presence or absence of particular amphibian species and their continued monitoring.
The Tanoé-Ehy Swamp Forests have recently been the focus for exploitation by an agro-industrial company, making a comprehensive survey of its biological richness more pressing. We therefore participated in a survey of the Tanoé-Ehy Swamp Forests organized by the "Centre Suisse de Recherche Scientifique." The amphibian data are presented in this paper.

Material and methods
The Tanoé-Ehy Swamp Forests (TESF; 5°05′-5°15′ N; 2°45′-2°53′ W) constitute 12,000 ha of remaining rainforest in the department of Tiapoum, south-eastern Ivory Coast. The mean annual temperature is 26 °C, the mean annual precipitation is about 2000 mm. A longer dry season lasts from December to March, and is followed by the period with highest precipitation in March to July. A minor rainy season extends from October to November (Eldin 1971). The River Tanoé crosses the southern and eastern parts of the TESF while the western part of these forests is marked by the Ehy lagoon (Fig. 1). The TESF mainly consist of moist, partly primary forests on predominantly sandy soil with vegetation typical for south-eastern Ivory Coast (Béligné 1994).
The survey was carried out in the northern part of TESF from 17 June to 29 July 2010 (long rainy season) and from 19 September (minor dry season) to 5 October 2010 (minor rainy season). We searched for frogs along eight sites during day and night. Searching techniques included visual search for frogs, the investigation of potential hiding places or very specific habitats (e.g. exceptional breeding sites such as water-filled tree holes), and the acoustic monitoring of frog calls (see Heyer et al. 1994, Zimmerman 1994. All available habitats were examined by two people. We searched for frogs on 26 days, each day for seven hours (07:00-11:00 h & 19:00-22:00 h GMT). The sampling effort therefore was always 14 person-hours per day and thus comparable throughout the survey. A GPS receiver (Etrex venture HC Garmin) was used to record geographic positions. Coordinates and short site descriptions are given in the Appendix 1. As our sampling design provides only qualitative and semi-quantitative data we calculated the estimated species richness, and thus the sampling efficiency, with the Chao 2 and Jack-knife 1 estimators (software: EstimateS, Colwell 2006). These estimators are incidence based, calculating using the presence/ absence data of the daily species lists (26 days of survey work) for 33 species. To avoid order effects we accomplished 500 random runs of the daily species lists.
Encountered individuals were usually determined to species level and the nomenclature used herein follows Frost (2014). Arthroleptis species in the Upper Guinean forests are currently difficult to identify because of overlapping intraspecific and interspecific variation in morphology . Judging from their advertisement calls our Arthroleptis records comprise more than one species, but are treated herein as one taxon Arthroleptis spp. Snout-vent-lengths (SVL) of the living frogs were taken with a dial caliper (accuracy ± 0.5 mm). Voucher specimens of all frog species were euthanized in a chlorobutanol solution and preserved in 70% ethanol. Voucher specimens are deposited at the "Laboratoire de Zoologie et de Biologie Animale" at the Félix Houphouët Boigny University, Abidjan. Some specimens of particular interest (see below) have been deposited in the Museum für Naturkunde, Berlin (ZMB).

Results
Species richness and community composition. Overall, we recorded 33 anuran species in nine families and 13 genera. A total species list with sites records (compare Appendix 1), known habitat preferences, distribution and IUCN Red List Category (IUCN 2013) is given in Table 1. The comparison of the species accumulation curve to the two incidence-based species richness estimators revealed that more amphibian species could probably be encountered within the TESF (Fig. 2). The Jack-knife 1 estimator calculated 39 (sd: ± 2.3) species for the area, the Chao 2 estimator estimated 40 (sd: ± 8.6) species. We hence probably recorded about 77.5-79.5% of the local species pool.
We recorded the highest species numbers at KA (21 species), followed by RE (20), MN, EB, YA (each 10). Species richness was lowest at MK (8), KW (7) and PN (5). While the site RE consisted of a swampy forest with Raphia palms and partly open canopy, EB, KA and KW comprised dense vegetation characterized by large canopy gaps and thick shrubby undergrowth along a river. At KA we recorded creeks, numerous puddles and ponds, and more leaf-litter compared to other sites. The site PN, which was relatively dry, mainly consisted of a more uniform forest with dense canopy. The four other sites (MK, MN and YA) were partly degraded and comprised also rice fields, small farms or oil palm plantations.
The majority of the encountered species are closely associated with forest habitats (13 species, 39.4%). Eight species (24.2%) predominantly occur in forest, but tolerate degraded habitats such as farmbush (secondary growth or degraded forest) or even savanna (Table 1). Twelve species (36.4%) usually prefer savanna and farmbush habitats and normally do not occur in pristine forests. At most survey sites, the amphibian assemblages were dominated by forest species (Table 1). Nevertheless, we observed species with preferences for farmbush and savanna habitats at all sites, suggesting that the area has already suffered from habitat degradation.
Most recorded species (54.5%) do not occur outside West Africa (defined as the area west of the Cross River in Nigeria), and are often restricted to smaller parts of West Africa. More than one quarter (27.3%) of all recorded species only occur in the Upper Guinea forest zone (forests west of the Dahomey Gap), while three records (Morerella cyanophthalma, Phrynobatrachus ghanensis and P. cf. intermedius) are potentially endemic to the forests in south-eastern Ivory Coast and adjacent Ghana.
According to the IUCN Red List, almost a quarter (24.2%) of all recorded species are threatened or near threatened: four species are Near Threatened (Afrixalus nigeriensis, Leptopelis occidentalis, Phrynobatrachus alleni and P. liberiensis), two are Vulnerable (Kassina arboricola and Morerella cyanophthalma), and two Hylarana occidentalis and Phrynobatrachus ghanensis, are Endangered (IUCN 2013; Table 1). If the Phrynobatrachus sp. males are P. intermedius (see below), the list would also comprise a Critically Endangered species.
Notes on selected species. We only comment on four species of particular interest.
Afrixalus fulvovittatus fulvovittatus (Cope, 1861) was described from Liberia and is mainly distributed in the flanks and back, not known to us to occur in P. liberiensis and not reported from P. intermedius either. The pair of white dorsal roundish dots occurs as well in P. liberiensis, however, there usually only in juveniles.
Phrynobatrachus sp. was associated with small creeks in swampy forest areas, dominated by Raphia palms. The males could be the unknown males of P. intermedius, although sex dependent differences in webbing would be unusual for the genus. In order to clarify the taxonomic situation molecular data are needed. Future work in south-western Ghana and south-eastern Ivory Coast should carefully examine P. liberiensis and P. plicatus to assess if P. intermedius (or other cryptic species) occur in the few remaining rainforests in these regions, and in particular to collect tissue samples and call recordings.

Discussion
During our survey we recorded 33 anuran species. The overall species richness of TESF was lower compared to species richness recorded in western Ivorian forest areas, for instance the Taï National Park (Ernst et al. 2006), Mont Sangbé National Park (Rödel 2003), or the Haute Dodo and Cavally Classified Forests (Rödel and Branch 2002), but similar to Mont Péko National Park (Rödel and Ernst 2003). Compared to these and other West African forest areas with known amphibian assemblages such as south-western Ghana , south-eastern Guinea  or north-western Liberia (Hillers and Rödel 2007), the TESF ranks among the areas of median to low amphibian species richness. Amphibian surveys in two other eastern Ivorian forests, Yakassé-Mé village forest and Banco National Park, documented 24 and 28 species, respectively (Assemian et al. 2006, Kouamé et al. 2014b. It thus could be speculated that this region is comparatively species poor concerning amphibians. However, our estimations revealed that we probably do not yet know the entire amphibian fauna of the TESF. More intensive surveys, especially in areas and microhabitats not yet investigated, may result in an increasing number of species. Further species likely to be recorded in TESF are Acanthixalus sonjae, Hyperolius laurenti and H. viridigulosus. The occurrence of Cardioglossa occidentalis, Astylosternus laticephalus, Leptopelis macrotis, Phlyctimantis boulengeri, Ptychadena aequiplicata and P. superciliaris, seems possible because these species have been found in the Ankasa Conservation Area , a rainforest in south-western Ghana and thus almost adjacent to TESF. Although we spent always the same number of hours per visit searching for frogs at the eight sites, the number of visits per site varied largely from 2-8 between sites (compare in Appendix 1). Therefore, the probability to record all species at a given site, and thus across different habitats, likely varied. Moreover, as the survey did not cover the entirety of the swamp forests, it is possible that further species occur in forest parts which were not studied. In contrast it is also possible that the documented forest alteration is responsible that some species, including the above mentioned ones, all being forest specialists, declined or even went extinct in our study site.
We recorded different subsets of the 33 anuran species in TESF. The highest species richness at site KA (21 spp.) was most probably due to the fact that this area was less altered and included more breeding sites, puddles, ponds, creeks, as well as thicker layer of leaf-litter than the remaining sites. More than half of the recorded species are restricted to West Africa, or to smaller parts of this region; the majority of these including all threatened species, are forest specialists. However, the records of many TESF species with wide distributional ranges, and broad habitat tolerance, clearly reflects altered forest conditions, due to deforestation and conversion of forests into palm plantations or rubber monocultures. The latter are steadily increasing in the eastern forest zone (Kouamé et al. 2014a). From western Ivorian forests it has been shown that logging has a serious effect on the composition of frog assemblages. Many forest specialists seem to be unable to prevail in degraded forests, most probably due to an altered microclimate with which they cannot cope , Ernst and Rödel 2006, Hillers et al. 2008. Hence, exploring more of the potentially pristine areas in the southern parts of TESF is needed to get a complete impression concerning the presumed original anuran composition in TESF.
South-eastern Ivory Coast suffered from intensive deforestation and only a few forest remnants still prevail (Parren and de Graaf 1995, Bakarr et al. 2004, Zadou et al. 2011, Mayaux et al. 2013. This is worrying because the rainforests of this region and neighbouring western Ghana are supposed to have acted as a Pleistocene forest refugium (e.g. de Graaf 1995, Maley 1996), and thus may comprise a unique fauna and flora. This is especially true for amphibians (Penner et al. 2011), which however, are still far from being completely known. Therefore, further surveys are highly recommended for all remaining forests, with a particular focus on primary forest. Such intensified research would lead to a better knowledge of the regional amphibian fauna, population sizes and distribution pattern of particular species. This would be especially important for species with high conservation concern, such as Phrynobatrachus intermedius, Morerella cyanophthalma and other rare or threatened endemics. Finally, further investigations could lead to a better understanding of the diversity in VNRs which would help to provide further conservation recommendations.  and Kutschera 2011;Elliott and Dobson 2014). Utevsky and Trontelj (2005) provided a key to all known European species in the genus Hirudo, and Kutschera (2012a, 2012 b) summarized their geographical distribution.
In the present article, we describe the morphology of juvenile and adult H. medicinalis-individuals, add information on its evolutionary distance to its sister taxon H. verbana, and summarize observations on the behaviour, ecology and distribution of this endangered species.

Materials and methods
Adult and juvenile European medicinal leeches (H. medicinalis) (plus cocoons) were obtained from undisturbed habitats of eastern Germany (Elliott and Kutschera 2011), and specimens of Mediterranean medicinal leeches (H. verbana) were purchased from a commercial supplier (Sudak, Tr-59560 Murefte Tekirdag, Turkey) (Kutschera and Roth 2005). The leeches were kept in aqua-terraria (90 x 40 x 60 cm, depth of the pond water ca. 10 cm; temperature 22 to 26 °C), and observed/photographed alive. Specimens of H. medicinalis were killed by adding 80 % ethanol to the water, so that the animals were preserved in their non-contracted, natural shape, and photographed. Extraction of DNA from part of the posterior sucker, sequencing of a fragment of the mitochondrial gene cytochrome c oxidase subunit I (CO-I), and phylogenetic analyses, based on newly acquired (and deposited) Gen-Bank-data were performed as described (Kutschera et al. , 2013Wirchansky and Shain 2010).

Morphology of adult H. medicinalis-individuals
Leeches are animals with an organization akin to that of earthworms, but having certain modifications associated with a predatory or parasitic mode of life. The limitation of the number of body segments facilitates a greater degree of agility than would be the case if the body was as long as that of most earthworms. The segments are each subdivided into a number of annuli, five in the Hirudinidae. There is some disagreement about the relationship between annulation and segmentation (Mann 1962). Externally, the annuli look much alike, and there is little indication of segmentation. Perhaps the best guide is the pattern of colouring, which often repeats itself once per segment. For example, a distinctive pattern separates H. medicinalis from its sister taxon H. verbana (Figs 1, 2). On the middle annulus of each segment are sensory papillae (Pap in Fig. 3A). These may be prominent, and are often marked by spots of light pigment. Papillae may also be present on other annuli of a segment. On the first few segments of the body, some of the sensory papillae are replaced by black-pigmented eye spots, five pairs of eyes arranged in a crescent in Hirudo medicinalis (four eyes are marked in Fig. 3B).
The size of the suckers relative to the body varies according to the mode of life of the leech species and, in H. medicinalis, the anterior sucker is quite small. The buccal cavity is lined by muscular ridges surmounted by cuticular teeth, and the mouth is a wide aperture occupying the whole of the anterior sucker (Fig. 3C). Following the pharynx is a region of the alimentary canal, the crop, which is dilated for the storage of food. In the sanguivorous H. medicinalis, it is drawn out into lateral arms referred to as diverticula.
The clitellum is situated towards the anterior of the body (Fig. 4A). The male reproductive aperture is median and unpaired. There are two internal ducts leading to it but these unite to form a single genital atrium with one external gonopore and a 'tube-like' male copulatory organ (Fig. 4B). The female pore is likewise median and unpaired, and is posterior to the male pore. chiefly in July and August. Over one to 12 days, each mature leech will lay 1 to 8 cocoons with usually 12 to 16 eggs per cocoon; sometimes more, but with some infer-

Cocoons and juvenile H. medicinalis
Mature medicinal leeches leave the water to deposit their cocoons in a moist place just above the water line on the shore or bank. The spongy cocoons (Fig. 5A) are laid tile eggs. In the laboratory, each adult laid 1 to 7 cocoons with 3 to 30 eggs per cocoon, and produced 2 broods per year under optimum conditions. Hatching time varied from 4 to 10 weeks, depending upon the temperature, and the live mass of each newly-hatched leeches (length: 8-12 mm) varied from 12 to 60 mg.
The markings of the juveniles are very similar to those of the adults except there is less pigment on the ventral surface (Figs 5,6). Hatchlings can survive for up to 100 days without feeding, but fed leeches in the laboratory attained a live mass of 0.5 to 0.6 g at the end of their first year, about 1.4 g in their second, and about 2.4 g in their third year. Similar results were obtained for H. verbana (Kutschera and Roth 2006). Although there is a paucity of field information, it is general agreed that H. medicinalis and H. verbana take at least two years to reach the breeding stage in the wild, and slow-growing leeches may not breed until they are three or four years old.

Behaviour of H. medicinalis vs. H. verbana and hyperparasitism
Living, adult individuals of H. medicinalis and its sister species H. verbana were maintained in aqua-terraria. Despite the fact that the species were clearly distinguishable based on their pigment patterns on both the dorsal and ventral sides of their body (Fig. 1), qualitative observation of their behavioural patterns revealed no differences. For a large part of the year when water temperatures are low, medicinal leeches are quiescent and remain buried in the mud or under submerged objects at the edge of the pond. As water temperature increases, the leeches become very responsive to water disturbance caused by a potential host, and swim towards the source of blood. Laboratory experiments showed that 86 % and 95 % of unfed leeches responded to low-amplitude surface waves (about 1 mm high) by swimming, whilst only ca. 60 % of fed leeches displayed a reaction. The neurophysiology of this detection of water motion was described in detail by Friesen (1981).
Laboratory experiments have also shown that when a medicinal leech is near a mammalian host, such as the skin of a human, it uses heat detection, the optimum response occurring at 33 to 40 °C (Dickinson and Lent 1984), and also chemosensory stimuli (Elliott 1986), both receptors being located in the anterior end of the leech (Fig. 3B). The leech explores the outer cell layer of the host for a suitable feeding site, then pierces the skin with its three jaws armed with numerous sharp teeth, and finally sucks the blood of its host. We also observed that, in the wild and in the laboratory, H. medicinalis suck blood from amphibians, such as the edible frog (Fig. 7).
However, other leech species will sometimes feed on H. medicinalis. Young Glossiphonia complanata that were co-cultivated with medicinal leeches frequently obtain their first meal by feeding on the body of H. medicinalis. In a quantitative study in a tarn (= pond) in Northwest England, H. medicinalis were found to be carrying all sizes of Helobdella stagnalis that were feeding on the host. The proboscis was inserted deep into the body wall of the host and the anterior portion of the body contracted regularly as fluid was extracted from the host, i.e., hyperparasitism was documented unequivocally. H. stagnalis did not kill its host or produce any obvious reactions. Similar observations were reported for H. verbana ).  The disk-shaped sucker is largely composed of muscle tissue containing numerous mitochondria. DNA-extractions for mt-sequence analysis (fragments of the gene CO-I) were performed from this part of the body that is not contaminated with the gut content of the blood-sucking annelid.