A new species of Typhlatya (Crustacea: Decapoda: Atyidae) from anchialine caves on the French Mediterranean coast

A new species of the thermophylic Tethyan relict prawn Typhlatya is described from two anchialine caves near Per-pignan (southern France). The new species is closely related to a congener known only from a freshwater cave at Castellón (eastern Spain), about 400 km to the south-west, differing apparently only in the size and shape of the rostrum and the armature of the dactylus of the ﬁfth pereiopod. Based on palaeogeographical evidence and assuming a sister-group relationship between both species, we suggest that their common ancestor could not be older than early Pliocene in age, and that it was already a stygobiont taxon adapted to live in shallow-water marine crevicular habitats. This ancestor would have vanished from the western Mediterranean after the cooling associated with the onset of northern Hemisphere glaciation, about 3 Mya, as documented for other Mediterranean marine taxa. Indeed, the genus is completely stygobiont and does not occur in ﬂuvial environments. The Pyrenees represent a watershed boundary that eliminates the possibility of the derivation of one species from the other by active dispersal after establishment in continental waters. © 2005 The Linnean Society of London, Zoological Journal of the Linnean Society , 2005, 144 , 387–414. spines, each with row of tightly set denticles on one side, and row of long pinnules on other side and dorsally; spine series ending distally with two longer and stouter spines with ornamentation reduced to few marginal denticles along one side only. Exopod of ﬁfth pereiopod well developed, attaining ischiomeral articulation of endopod. Telson 2.0–2.3 times longer than wide, with 12 terminal spines.


INTRODUCTION
In 1973 the French biologist J. Y. Bertrand collected a single specimen of a blind atyid prawn in the cave of Font Estramar, a small cave opening just above a spring of the same name in Massif des Corbières (Pyrénées-orientales, France). He identified it as Troglocaris sp., presumably with reference to the presence of Troglocaris inermis Fage, 1937 in the nearby karsts of Hérault, where the stygofauna is similar to that of Corbières (Bertrand, 1974). Hitherto Troglocaris Dormitzer, 1853, together with the monotypic Spelaeocaris pretneri Matjasic, 1956 from caves in Herzegovina, were the only stygobiont atyid genera known from the peri-Mediterranean region. Troglocaris , with nine recognized species (see d 'Udekem d'Acoz, 1999) is limited to coastal regions on the northern rim of the Mediterranean, from the Gulf of Lyon in MATERIAL AND METHODS Most of the prawns were caught in traps baited with strips of raw beef and set in the submerged passages of the caves using specialized cave diving techniques. Other specimens were captured by hand, the animals often entering the open water column and being illuminated by torchlight. Salinity and temperature profiles of the water column in Grotte des Fées de Leucate were produced with a Seabird SBE19 CTD oceanographic profiler programmed to take readings every 0.5 s. The descent of the profiler was controlled to a rate of c. 10 cm s -1 ; the device was held out in front by the diver to avoid disturbing the water column. Specimens were fixed in the field in ethanol and treated in the laboratory with hot lactic acid to remove internal tissue. Black Chlorazol B cuticular staining was used following the procedure described in Wagner (1994). Drawings were prepared using a camera lucida on an Olympus BH-2 microscope equipped with Nomarski differential interference contrast. Body measurements were derived from the sum of the maximum dorsal dimensions of the carapace (rostrum included), free somites, and telson. Material is deposited in the Crustacea collection of the Departement des Milieux et Peuplements Marins, Muséum national d'Histoire naturelle, Paris (MNHN).

T HE CAVES
Font Estramar is an important artesian spring that opens in the limestone of the Massif des Corbières (Early Cretaceous; see Aunay et al ., 2002) about 150 m inland from the inner shore of the Mediterranean coastal lagoon of Salses-Leucate (Salses-lechâteau; Pyrénées-orientales, south-west France; Fig. 1). The lagoon, with a surface area of 54 km 2 and a maximum depth of 3.70 m, is separated from the sea by a sand bar. Water salinity is slightly lower (22-35‰, depending on the local conditions) than in the sea nearby (37.1‰).
Underwater exploration of the spring started in 1949 and the extent of surveyed passages has increased progressively since then, in line with the improvements in cave diving techniques. Considering the complexity of the cave, the Association de Recherches de Font Estramar (ARFE) was created in 1991 to coordinate its exploration, leading to the survey of more than 2800 m of underwater passages. The current explored limit in the cave was reached in 1996 by ARFE's cave diver Cyrille Brandt, who reached a depth of -164 m at 850 m from the entrance (Brandt, 1997). It has been estimated that the karstification at the spring area extends 350 m, with an even greater potential as the thickness of the limestone is around 1000 m (Ladouche et al ., 2004).
Font Estramar and Font Dame are the main water sources within the large and complex karstic system of Corbières d'Opoul (Salvayre, 1978(Salvayre, , 2002. The large associated aquifer derives around 30% of its water from rainfall, the remainder coming from sinks in the Agly and Verdouble riverbeds (Ladouche et al ., 2004). The outflow of Font Estramar has, over the last 30 years, averaged 2.1 m 3 s -1 , fluctuating from 0.8 m 3 s -1 to more than 30 m 3 s -1 during floods. The water is characterized by a slightly elevated temperature and oligohaline state (salinity range 0.2-2.2‰, depending on the spring's outflow volume; Kiener & Petit, 1968;Ladouche et al ., 2004). Recent studies (Ladouche et al ., 2004) indicate that the salinity derives from the water of the coastal lagoon. The deep intrusion and circulation of this water would explain its elevated temperature, ranging from 15.6 to 18.6 ∞ C at the spring's outlet (Cyrille Brandt, pers. comm.).
Typhlatya arfeae sp. nov . has been collected in Font Estramar from the entrance to the so-called Puits du Silence (see map in Brandt, 1997), although it has also been observed by divers further within the cave. The species seems to be more abundant in the Galerie Chaude , where the temperature and salinity are higher, and the conditions (turbidity, flow) more stable. It has been observed together with a cirolanid isopod ( Faucheria sp.) and the amphipod Niphargus delamarei Ruffo, 1954(Bertrand, 1973. Grotte des Fées de Leucate ( Fig. 1), excavated in Early Cretaceous limestone (see Aunay et al ., 2002), opens only 40 m from the northern shore of the coastal lagoon of Salses-Leucate. It has five entrances interconnected by underwater passages. The main entrance leads to a submerged room followed by a gallery that reaches a depth of -34 m at its furthest point. The total length of the cave exceeds 140 m. The water is known to be brackish and recent analyses have demonstrated that the cave is currently disconnected from any karstic system (Ladouche et al ., 2000).
Typhlatya arfeae sp. nov. has been observed in Grotte des Fées de Leucate throughout the entire water column, frequently swimming in open water. The accompanying fauna consisted of the stygobiotic amphipods Pseudoniphargus sp., Niphargus delamarei , and Niphargus angelieri Ruffo, 1954, the copepods Diacyclops bicuspidatus odessanus (Schmankevitch, 1875) and Calanipeda aquaedulcis Kritschagin, 1873, and the eel Anguilla anguilla (L.). Anguilla , Calanipeda , and Diacyclops are common in the Salses lagoon, and their presence in the cave suggests the existence of a connection to the lagoon that is inaccessible to divers.

O N THE ANCHIALINE CONDITION OF THE CAVES
During our first visit to Grotte des Fées de Leucate (17 April 2003), we noticed that a fresh-to-oligohaline water lens of 2 m overlaid a brackish water layer, with a sharp pycnocline between. No water currents were observed. Using an oceanographic probe, the CTD profile of the water column was produced on 7 December 2003 after several days of strong rainfall. The water level in the cave was then nearly 3 m higher than during our earlier visit. Aside from a surface layer (3 m) of mixed water, three additional pycnoclines were discernible at depths of 5-12, 18-20, and 32-34 m (Fig. 2). This superimposed series of pycnoclines could have been generated by sequential influxes of fresh water into the cave, which eventually mixed by diffusion with the saline layer immediately beneath. At the cave bottom, at a depth of 37 m, salinity reached 22‰, which was the same as in the neighbouring lagoon at that time. Water temperature in the cave remained comparatively warm and stable, between 15.5 and 16.5 ∞ C along the entire column.
These hydrographic features, plus the presence of stygobiont organisms, suggest a typically anchialine Figure 1. Location of the two caves harbouring Typhlatya arfeae sp. nov. in south-eastern France plus the single locality of the closely related T. miravetensis Sanz & Platvoet, 1995 in eastern Spain. condition for the cave (Stock, Illife & Williams, 1986;Sket, 1996). Grotte des Fées de Leucate is therefore the first anchialine cave recorded in metropolitan France. Similar anchialine habitats on the inner shore of coastal lagoons have been reported from the Dominican Republic (Jaume & Wagner, 1998) and Mallorca (Encinas, 1983).
Font Estramar differs dramatically from Grotte des Fées de Leucate by its low salinity, its important outflow, and its lack of obvious stratification in the water column. Nevertheless, it can also be considered an anchialine habitat according to Sket (1996), who contended that any subterranean habitat where continental and marine waters and/or faunas may influence each other should be considered anchialine.

Diagnosis
Eyes lacking pigment. Rostrum small, pointed, not extending beyond eyestalk and hardly beyond insertion of first peduncle segment of antennule. Sixth pleonite about 1.6 times longer than fifth pleonite. Antennulary stylocerite not reaching distal margin of first peduncle segment. Flagellar lobule of exopod of first maxilliped well developed and delimited from remainder of exopod. Armature along medial margin of dactylus of fifth pereiopod consisting of 19-28 spines, each with row of tightly set denticles on one side, and row of long pinnules on other side and dorsally; spine series ending distally with two longer and stouter spines with ornamentation reduced to few marginal denticles along one side only. Exopod of fifth pereiopod well developed, attaining ischiomeral articulation of endopod. Telson 2.0-2.3 times longer than wide, with 12 terminal spines.

Description
Body up to 17.85 mm long, unpigmented (white when alive), with completely regressed eyes. Rostrum small and variably pointed (compare Figs 3A, 4B and 5), subtriangular in cross-section, triangular in dorsal aspect (Fig. 4A), lacking teeth and rostral ridge; sometimes with 1-2 tiny setae on ventral rim (see Fig. 3A). Rostrum hardly extending beyond insertion of first peduncle segment of antennule, in general shorter than eyestalks, but exceptionally reaching tip of eyestalk in some smaller specimens (Fig. 3A).
Lateral body wall of fourth to eighth thoracomeres (corresponding to thoracomeres bearing first to fifth pereiopods) with one phyllobranchiate pleurobranch on each side, those on eighth thoracomere reduced (see Figs 3C, 13F, 15A). Sternite of eighth thoracomere with posterior margin weakly produced anteriorly into triangular process with subterminal tuft of setules (Fig. 4C); process not extending beyond insertion of fifth pereiopod. Pleon ( Fig. 3A) with intercalary dorsal sclerite between carapace and first pleonite; Pleurae of first five pleonites rounded. Sixth pleonite about 1.62 times longer than fifth, with posterolateral lobe on each side; lobe with tiny denticle as in Figure 3B. Posterior margin of sternite of sixth pleonite projecting mediodistally into narrow triangular process bearing subdistal setule on rounded tip (Fig. 18A); process probably homologous with the preanal plate of T. galapagensis, Stygiocaris stylifera Holthuis, 1960 and Antecaridina lauensis (Edmondson, 1935) (see Monod & Cals, 1970: 97, figs 60-62), although in the new species, in T. miravetensis and in T. rogersi (and apparently also in T. mitchelli and T. campechae; see Hobbs & Hobbs, 1976: 2, 6) this process is incorporated into the sternite and does not appear as a sep-arate sclerotized plate. Anus positioned on posterior margin of sixth pleonite, just below insertion of telson ( Fig. 18A).
Eyestalks ellipsoid, about 1.6 times longer than wide, with reduced area of thin cuticle at distolateral corner ( Fig. 4B). Conical frontal process present between insertion of eyestalks and insertion of antennules (Fig. 4A). Paired soft finger-like processes inserted laterodorsally, adjacent to insertion of antennules, and apparently functioning to lock eyestalks laterally (Fig. 4A); process homologous with two processes, beneath the eyes, placed above the dorsal part of the antennulary praecoxa of T. galapagensis (see Monod & Cals, 1970: 73, fig. 23). At least in the new species this process is an integumental outgrowth of the body wall and does not form part of the antennule, as supposed by Monod & Cals (1970).
Antennule with 3-segmented peduncle reaching transverse suture line (= diaeresis) on antennal scaphocerite (see Fig. 6A). Length ratio of peduncular segments 1-3 measured along medial margin approximately 0.41 : 0.29: 0.30. Dorsodistal margin of first and second peduncular segments each with setae inserted on biarticulate socle, remaining setation as in Figure 6A; first segment with stylocerite extending 82% length of segment; third segment with rounded process dorsodistally, apparently articulating with segment; process homologous with calcified plate mentioned in same position in T. galapagensis (see Monod & Cals, 1970 Antenna with 5-segmented peduncle; first segment (coxa) with recurved, finger-like soft process on ventrolateral surface (Fig. 7C); process tentatively homologous with coxal orifice placed on a membranous outgrowth of T. galapagensis (see Monod & Cals, 1970: 75, fig. 24). Second segment (basis) with two soft finger-like processes dorsolaterally (Fig. 7A, C), and acute tooth on ventromedial corner extending slightly beyond distal margin of third peduncular segment; outlet of antennary gland visible as soft conical process with terminal pore, located behind tooth. Third segment with single seta on distolateral corner. Fourth peduncular segment inserted on dorsomedial surface of preceding segment, with row of stout plumose setae, each on biarticulate socle, along distal and distolateral margins (Fig. 7B, C). Fifth segment elongate, about 2.2 times longer than wide, subrectangular, inserted dorsodistally on preceding segment, with cluster of smooth setae at distolateral corner.  Scaphocerite inserted ventrolaterally on second peduncular segment, about 2.2 times longer than wide, with straight medial margin ending in pointed process; transverse integumental suture line (= diaeresis) extending from insertion of pointed process to about midway along scaphocerite. Outer margin of scaphocerite convex, distal margin evenly rounded; both margins fringed with plumose setae with transverse endocuticular interruptions giving a segmented appearance (Fig. 4A). Flagellum multiannulate, with proximal annuli incompletely separated (Fig. 4B).
Mandibles asymmetrical, lacking palp and lacinia mobilis, with sparsely set short plumose setae on surface of proximal part of coxal gnathobase (Fig. 8A). Left mandible (Fig. 8A) with incisor bearing three superimposed triangular teeth at anterodistal corner and blunt, truncate tooth at posterodistal corner. Setal row (= 'ventral row of lifting spines' of Fryer, 1977) consisting of four spinulose elements as in Figure 6B. Proximal portion of mandibular sinus with two superimposed marginal rows of slender plumose setae (= 'dorsal row of lifting spines' of Fryer, 1977), anteriormost row composed of shorter setae. Molar process large, with grinding surface strongly curved transversely and composed of parallel striated plates with tuberculate rim, each plate with short spinule distally (Fig. 8C); spinules progressively shorter, absent from plates on molar crown. Right mandible (Fig. 8D) with incisor narrower than on left counterpart, composed of six sharp triangular teeth (Fig. 8E). Setal row of five spinulose elements, more slender and less spinulose    than on left mandible. Two superimposed rows of setae on mandibular sinus composed of fewer, and stiffer, setae than on left mandible (Fig. 8F). Molar process similar to left counterpart, but proximal margin of grinding plates with short dendriform spinules progressively shorter, absent from plates on molar crown (Fig. 8G); presence of such spinules on left mandible unconfirmed.
Maxillule (Fig. 9A-D) lacking expressed segmentation; tentatively interpreted as comprising coxa with broad, foliaceous endite (= 'proximal lobe' of Fryer, 1977), and two lobes representing the basis with basal endite (= 'distal lobe' of Fryer, 1977), and the endopod. Coxal endite pointed distally, setulose, with armature consisting of two superimposed rows of setae along medial margin; setae disposed in middle portion of anterior row stiff and with truncate tip (Fig. 9C), corresponding to the 'fine funnelling setae of proximal endite' of Fryer (1977). Two roughly parallel clusters of short setae on posterior surface of lobe; those comprising middle cluster with brush-like tips ( Fig. 9B; setae probably homologous with the 'lifting spines of the proximal endite' of Fryer, 1977: fig. 78). Basal endite spatulate; distal half of medial margin with two parallel rows of stout conical spines with rounded tubercles (Fig. 9D); remaining armature on margin and on lateral surface of endite as figured. Endopodal lobe finger-like, with four subdistal setae.
Maxilla (Fig. 10A) phyllopodial, biramous, with basic structure corresponding tentatively to coxa with endite, basis with bilobed endite, reduced finger-like endopod and unsegmented exopod (= scaphognathite). Intersegmental articulations not expressed except partially between coxa and basis. Coxal endite (= 'proximal endite' of Fryer, 1977) with marginal comb row of long setae each swollen proximally; proximal and distal setae of comb row plumose, others rigid and apparently smooth; posterior surface of endite with row of about nine rigid setae each bipinnate distally; anterior surface of endite with cluster of heterogeneous setae (some plumose proximally but smooth distally, some fully plumose, some bipinnate distally, and others apparently completely smooth) arranged as in Figure 10B. Basis indistinct, with three plumose setae, one on medial margin, two on posteromedial surface of segment.
Proximal endite lobe (= 'teaseling lobe' of Fryer, 1977) with proximal margin slightly convex, medial margin straight, and distal margin oblique, not reaching distal margin of distal endite lobe. Posterior surface of lobe unarmed, anterior surface densely covered with rigid setae each unipectinate distally (Fig. 10E), arranged in parallel oblique rows (Fig. 10C); medial margin of lobe with sparsely set, smooth setae with rounded tips (Fig. 10F) mixed with typical unipecti-nate setae; row of slender setae with bipinnate tips (Fig. 10D) lying close and parallel to insertion of endite on anterior surface of segment. Distal endite lobe (= 'distal endite' of Fryer, 1977) spatulate, not extending beyond distal margin of scaphognathite, with heterogeneous array of armature elements consisting of plumose setae with smooth distal part, plumose setae with unipectinate distal part, smooth setae with rounded tip, and denticulate spines (as Fig. 10G); several additional plumose setae located distomarginally on anterior surface of lobe.
Scaphognathite (= exopod) inserted on lateral margin of basis, distal lobe truncate, proximal lobe (= 'caridean lobe') well developed; margins fringed with plumose setae, those on distal margin of distal lobe longer and with several transverse endocuticular interruptions giving a segmented appearance. Endopod (= palp sensu Monod & Cals, 1970) extending to middle of distal lobe of scaphognathite, smooth except for three minute plumose setae inserted proximally on outer margin.
First maxilliped (Fig. 11A) phyllopodial, biramous, with coxa, basis, exopod and endopod indistinctly separated. Coxa with rounded endite bearing complex armature array comprising two marginal rows of long setae (Fig. 11F), one row of smooth geniculate setae with denticulate tips (Fig. 11G), and one of plumose setae with smooth tips; numerous short plumose setae each with smooth distal part set sparsely on anterior surface of endite, plus three setae with denticulate tips on anterior surface near distomedial margin. Coxal epipodite short, conical, with short smooth seta on tip instead of lamelliform podobranch apparently present in homologous position in T. iliffei (see Hart & Manning, 1981: fig. 38).
Basis indistinct, bearing row of about six long plumose setae, each with smooth distal part, on anterior surface, and row of about 14 shorter setae on posterior surface. Basal endite elongate, subrectangular, with complex armature comprising plumose setae with smooth distal portion and pointed tips, same type of setae but with rounded tips, short smooth setae with rounded tips (Fig. 11D), and plumose setae with denticulate tips (Fig. 11E); medial margin of endite with row of tightly set, rigid plumose setae displaying transverse lamellar expansions distally (Fig. 11B, C).
Exopod subrectangular, with distomedial corner produced into elongate process ('flagellar lobule' of Hobbs & Hobbs, 1976) extending about 46% of total length of segment. Lateral and distal margins of segment fringed with plumose setae; inner margin and margins of flagellar lobule with plumose setae each smooth distally. Anterior surface of segment with sparsely set plumose setae with smooth distal parts and several short smooth setae distributed as figured.   Endopod reaching distal margin of basal endite, roughly triangular; inner (= medial) margin with row of plumose setae with smooth distal part, single similar seta on tip, and two tiny smooth setae on proximal part of outer (= lateral) margin.
Second maxilliped (Fig. 12A) stenopodial with expanded dactylus. Coxa unarmed, with podobranch reaching distal margin of ischium, and no trace of epipod. Basis and ischium almost completely fused, former slightly shorter, with sparsely set plumose setae with smooth distal portion along both margins and on anterior surface of segment, proximalmost seta on medial margin enlarged. Exopod elongate and slender, bowed, with plumose setae with smooth distal part along both margins of proximal part of segment; middle part of segment naked; distal part with plumose setae along both margins, setae with transverse endocuticular interruptions giving a segmented appearance; two short smooth setae on anterior surface of segment as figured.
Ischium with row of heterogeneous setae along medial margin (some plumose, others with denticulate tips as in Fig. 12B). Merus about as long as ischium, but more slender, with row of plumose setae with smooth distal parts along medial margin. Carpus half length of merus, with digitiform process with microtuberculate ornamentation distally, on anteromedial surface; armature of segment reduced to single plumose seta with smooth distal part positioned subdistally on lateral margin. Propodus with expanded distal margin covered with smooth rigid setae, denticulate setae, and long plumose setae with denticulate distal parts and rounded tips (Fig. 12C); latter setae with subdistal pinnules stouter than rest.
Dactylus arising on medial margin of preceding segment and completely incorporated into it except for expression of vestige of intersegmental articulation on posterior surface; dactylus locked into flexed position by rounded process on carpus. Segment falcate, with smooth convex medial (but anatomically lateral) margin and slightly concave lateral (but anatomically medial) margin. Slit connected with duct present proximally on both anterior and posterior sides of segment. Anteromedial surface of segment ( Fig. 12D) with smooth rigid setae, long plumose setae with smooth distal part, and single plumose seta distally. Medial margin of segment with row of tightly set, pectinate spines as in Figure 12E, F.
Third maxilliped (Fig. 13A) stenopodial, with small phyllobranchiate arthrobranch (i.e. associated with arthrodial membrane between coxa and body). Coxa with transverse row of plumose setae on inner margin and cluster of setae on distomedial corner; epipod bifurcate, posterior branch slender, not extending beyond coxa distally, with subdistal notch; anterior branch short and roughly conical, unarmed. Basis partially fused to ischiomerus, short, with row of unequal setae along medial margin. Exopod unsegmented, long and slender, bowed, extending beyond distal margin of ischiomerus; armature as described for preceding limb. Relative length of ischiomerus, carpus and propodus as shown in Figure 14. Ischiomerus somewhat bowed, about six times longer than wide, with sparsely set short setae along both margins.
Carpus with 6-7 transverse rows of short bipectinate spines on medial surface of segment (Fig. 13B); row of smooth setae along distal margin. Propodus with heterogeneous array of transverse rows of spines on medial surface of segment, as follows: six rows of short bipinnate spines with expanded hyaline tips ( Fig. 13C; cochlear setae of Monod & Cals, 1970; 'distally spatulate setae' of Hobbs & Hobbs, 1976) on proximal half of segment; four rows of spines with serrate triangular denticles along one margin only and with spinulose tips (Fig. 13D) located adjacent to preceding rows of spines; and about 12 smooth, conical spines ( Fig. 13E) along distal part of segment; latter spines indistinguishable from larger terminal spine, interpretable as dactylus.
First pereiopod (Fig. 13F) chelate, with tip of propodus and dactylus covered with tuft of heterogeneous spines and setae (similar to those on second pereiopod, described in detail below and Fig. 15B-G). Coxa, basis and exopod as in third maxilliped except for uniramous coxal epipodite (nevertheless similar to posterior branch of third maxilliped), presence of 2-4 long pinnate coxal setae ('setobranchs') on small outgrowth near origin of epipodite, and basis completely separated from ischium. Ischium completely separate from and slightly shorter than merus, both segments bearing few short smooth setae on both margins. Carpus with protruding conical outgrowth located about two-thirds distance along lateral margin; several smooth setae on outgrowth and margins of segment as figured. Chela (propodus + dactylus) slender, about 2.7 times longer than wide, with dactylar (= outer) process clearly extending beyond tip of propodal (= inner) process. Relative length of ischium + merus, carpus and propodus as in Figure 14.
Second pereiopod (Fig. 15A) chelate, similar to preceding limb but more elongate due to ischium, merus and carpus proportionally longer (see Fig. 14). Exopod proportionally shorter, reaching distal margin of merus only (exopod reaching almost to conical lateral process on propodus in preceding limb; nevertheless, both exopods about equal in absolute length). Three 'setobranch' setae present on coxa. Chela similar to preceding limb both in size and armature, with terminal tuft of setae on dactylar process composed of heterogeneous array of spines and setae, as follows:    Four stout sigmoid spines mid-ventrally, each with expanded tip and with transverse lamellar processes along posterior margin (Fig. 15B, C); spines apparently lacking internal canal described by Monod & Cals (1970: 67, fig. 6) for Typhlatya galapagensis. Sigmoid spines partially hidden by several rows of setae evolving progressively in form from innermost to outermost rows from coarsely serrate (Fig. 15D) to finely serrate scrapers (Fig. 15E), flexible finely unipinnate setae (Fig. 15F), and finally flexible setae with terminal brush (Fig. 15G). Terminal tuft on propodal process similar to that on dactylus but lacking sigmoid spines.
Third pereiopod (Fig. 16A) long and slender, stenopodial, with coxa, basis and exopod as in preceding limb; tip of exopod reaching about two-thirds of merus. Ischium incompletely separated from merus, about 4.7 times longer than wide, with one spine on medial margin. Merus with two spines on medial margin. Carpus lacking spines. Propodus with row of about 15 small spines along medial margin. Relative lengths of ischiomerus, carpus and propodus as in Figure 14. Dactylus slender, with eight short, stout smooth spines on medial margin (Fig. 16B); spines progressively longer towards distal part of dactylus. Unguis smooth, with faint transverse endocuticular interruption proximally.
Fourth pereiopod (Fig. 16C) as for preceding limb except for relatively shorter ischiomerus and carpus (see Fig. 14), and for slight differences in number of armature elements, as follows: one additional medial spine present subdistally on merus, and 7-9 spines along medial margin of dactylus (Fig. 16D).
Fifth pereiopod (Fig. 9E, F) as for two preceding limbs except for relative length of ischiomerus, carpus and propodus (see Fig. 14), absence of coxal epipodite, presence of only two coxal 'setobranch' setae instead of three, only one spine on inner margin of merus, shorter exopod (as long as ischium only), and armature of medial margin of dactylus, consisting of series (grooming comb?) of about 19-28 similar spines, denticulate along one side, pinnate on other, and with additional row of pinnules dorsally; two additional, longer and stouter spines positioned distal to series, with ornamentation reduced to few marginal denticles on one side only (see Fig. 9F).
Pleopods 1-5 biramous. Pleopod 1 somewhat shorter than pleopod 2, others decreasing progressively in length from 2 to 5. Protopods with evidence of subdivision (into coxal segment?) proximally, partially fused to basis anteromedially (Fig. 17C); protopod of pleopod 5 with row of smooth setae along lateral margin (Fig. 17F), protopods of pleopods 1-4 with setation reduced to two short setae on distolateral margin (Fig. 17A, D). Rami of pleopods unequal, exopod longer than endopod, both with marginal setation consisting of plumose setae with transverse endocuticular interruptions giving a segmented appearance. Endopod of all pleopods except first with slender appendix interna with cluster of denticulate retinacles on oblique tip (Fig. 17E); no trace of appendix masculina in any of the specimens studied. Exopod of first pleopod modified, expanded proximally and with distal portion elongate, but variable from hardly developed to well developed, as in Figure 17B; short setae on segment as figured.
Uropods clearly longer than telson (Fig. 18A), with undivided protopod; exopod slightly longer than endopod. Most plumose setae along distal and medial margins of exopod and on margins of endopod, with transverse endocuticular interruptions giving a segmented appearance. Surface of both rami armed with mixed array of short setae, as in Figure 18A-C. Incomplete, sinuous transverse suture line (= diaeresis) on exopod, not reaching inner margin of segment, with short plumose seta and thorn-like spine inserted submarginally, and with row of smooth setules between insertion of spine and outer margin of segment (Fig. 18B). Endopod inserted dorsal to exopod on protopod.
Telson (Figs 4D, E, 18A) elongate, trapezoid, 2.0-2.3 times longer than wide, with two short smooth spines dorsolaterally on each side. Dorsal surface with four pairs of short smooth setae present along midline, plus several setules as in Figure 4D. Distal margin with six pairs of spines, outer two pairs smooth, rest spinulose. Outermost pair shortest, adjacent pair longer and stronger than rest; four inner pairs decreasing progressively in length towards midline (Fig. 4E).

Sexual dimorphism
None of the specimens analysed displayed evidence for appendix masculina on the second pleopods or carried eggs, thus preventing sex determination. Not one displayed the extra-setulation developed on the pleopods and on the margins of the epimeral plates of ovigerous females of some species, such as T. campechae (Hobbs & Hobbs, 1976: fig. 3a, b;see also Juberthie-Jupeau, 1974). Likewise, none displayed an anteriorly vaulted, pointed extension (thelycum?) of the posterior margin of the sternite of the eighth thoracomere (= thoracomere carrying the fifth pereiopods; see Fig. 4C); this structure seems to develop on (adult?) females in T. galapagensis (see Monod & Cals, 1970: fig. 58).

Etymology
The species name is based on the acronym ARFE, Association de Recherches de Font Estramar, and is dedicated to the members of this Association.

DISCUSSION
The genus Typhlatya currently comprises 15 species (see Table 1), although it is in need of thorough revision. In addition, the validity of genera within the Antecaridina-Stygiocaris-Typhlatya-Typhlopatsa complex has been questioned (d'Udekem d'Acoz, 1999; see also Monod & Cals, 1970). In fact, inspection of the original descriptions of several Typhlatya species brings their status into question. For example, T. garciadebrasi, with the third maxilliped lacking spatulate spines on the propodus (an armature displayed by all the species of the genus thus far investigated, and also by Antecaridina Edmondson, 1954 andStygiocaris Holthuis, 1956; see Monod & Cals, 1970), and with an extremely elongate sixth pleonite (3.6 times longer than the fifth pleonite, compared to 2.2 times longer, at most, in other species of the genus), should probably be removed from Typhlatya. Similarly, T. jusbaschjani, although a Typhlatya, should be regarded as species inquirenda since its original description contains almost no information of diagnostic value, making the species practically unidentifiable. In addition, the quality of the original species descriptions is uneven and in many cases inadequate. The characters used to diagnose the species are frequently ambiguous or even wrong. For example, the expression/lack of expression of the ischiomeral intersegmental articulation of pereiopods has been used by Sanz & Platvoet (1995) as a taxonomically valid character. However, it has been shown above (and by Sanz & Platvoet, 1995: fig. 3a, d) that this articulation is only partially expressed in both T. arfeae and in T. miravetensis -on one side of the corresponding limbs only -and consequently could easily have been overlooked, or interpreted as completely expressed. The same holds true for the expression of the articulation between the propodus and dactylus in the second maxilliped, considered by Hobbs & Hobbs (1976) to be of taxonomic value. Sanz & Platvoet (1995) used Table 1. Species and distribution of the genus Typhlatya Creaser, 1936. In addition to the formally described species, there are records of unidentified species from Curaçao, Bonaire, and Guadalupe islands (Lesser Antilles), and from Isla Juventud (Cuba; see Silva, 1974;Stock, 1993;Sanz & Platvoet, 1995;Hobbs, 1998)
Again, examination of topotypic material has revealed that the species displays the typical number of spines (see Fig. 19A, B). Other inconsistencies in the original description can be explained by the confusion between left and right mandibles (see Sanz & Platvoet, 1995: fig. 1c, d), and by the swapping of the labels between the figures of the maxilla and first maxilliped (their fig. 2a, b). Finally, the apparent absence of a caridean lobe on the scaphognathite of the maxilla of T. miravetensis could not be confirmed since dissection of the specimens was not permitted.
Typhlatya arfeae sp. nov. is almost identical to T. miravetensis, differing only in details of the armature on the medial margin of the dactylus of the fifth pereiopod, and in the shape and degree of development of the rostrum. Differences in armature of the fifth dactylus have been observed to be constant in all specimens analized. T. miravetensis displays a marginal series of about 36 slender spines ornamented with large, sparsely set denticles along one side, and a second row of tiny pinnules along either the other side and dorsally (Fig. 19C, D); this series of slender spines terminates distally in an additional, stouter spine with ornamentation reduced to one row of short denticles along one side, and another row of tiny denticles    dorsally. In contrast, the new species displays 19-28 stouter spines along the medial margin of dactylus, and the ornamentation of these spines differs, consisting of a row of tightly set denticles on one side, and a row of long pinnules on the other side and dorsally. In addition, this series of spines ends distally in two longer and stouter spines with ornamentation reduced to few marginal denticles along one side only (see Fig. 9F). A remarkable trait of T. miravetensis, viz. its virtual lack of rostrum (it is reduced to a scarcely developed, rounded vestige in dorsal view, whereas in other species it is variably elongate, triangular), could also be considered as diagnostic for the new species, but one of the three specimens studied by us displayed a faint, weakly developed, pointed rostrum.
Typhlatya arfeae sp. nov. could be sympatric with Troglocaris inermis Fage, 1937, an atyid species reported from the groundwaters of Hérault, about 90 km to the north of the known range of Typhlatya arfeae. However, Troglocaris can be easily differentiated by the presence of supraorbitary and suborbitary spines on the carapace, the longer rostrum, extending anteriorly beyond the eyestalks, the vestigial exopod on the fourth and fifth pereiopods, and the more elongate sixth pleonite (2.2 times longer than the fifth pleonite, compared to only 1.6 times longer in T. arfeae).
As remarked above, Typhlatya arfeae sp. nov. seems to be very closely related to T. miravetensis, recorded from a freshwater cave located 4 km inland from the coast of Castellón (SE Spain), about 400 km southwest of the range of T. arfeae and with the Pyrenees in between. As regards the possible origin and age of the new species, we adopt some of Sanz & Platvoet's (1995) suggestions for the genus: (1) ancient age, earlier than late Cretaceous; (2) thalassoid condition; (3) occupation of and speciation in continental waters due to stranding after marine regressions; and (4) that dispersal episodes could have occurred in ancestors of the current species only prior to the development of troglobiosis. We also agree with Sanz & Platvoet (1995) in considering that the new species, like T. miravetensis, descended from a (common) ancestor that inhabited the shallow coastal waters in a region corresponding to the current Roussillon and Spanish Levant.
Assuming a sister-group relationship between the two species and since the coastline of Roussillon was drastically affected by the Messinian regression (Aunay et al., 2002), we suggest that their common ancestor could not be older than early Pliocene in age, and that it was already a stygobiont taxon adapted to live in shallow-water marine crevicular habitats. Assuming a tropical-temperate origin for the genus, this ancestor could have vanished from the western Mediterranean after the cooling associated with the onset of northern hemisphere glaciation, about 3.0 Mya, as documented for other Mediterranean marine taxa (Raffi, Stanley & Marasti, 1986;Sorbini, 1988). Indeed, as the genus is completely stygobiont and does not occur in fluvial environments, and as the Pyrenees represent a watershed boundary, we can eliminate any chance of the derivation of one species population from the other by active dispersal, once established in continental waters.