A new species and new records of deep‐water Calappidae (Crustacea: Decapoda) from the Indian Ocean with a key to the Mursia Desmarest, 1823 species of the region

Deep‐water Calappidae collected by German and Russian expeditions to the western Indian Ocean and additional material from several European and American museums were studied. Calappa japonica Ortmann, 1892 was recorded from the Red Sea for the first time, where it occurs in unusually deep waters and C. pustulosa Alcock, 1896 was recorded for the first time off Madagascar. Eight species of Mursia Desmarest, 1823 are known from the Indian Ocean (M. africana Galil, 1993, M. aspera Alcock, 1899, M. bicristimana Alcock and Anderson, 1894, M. coseli Crosnier, 1997, M. cristiata H. Milne Edwards, 1837, M. flamma Galil, 1993, M. curtispina Miers, 1886, and M. minuta n. sp.). For M. bicristimana the status was clarified and a new species described. This description is partly based on a specimen which prior to this study had been erroneously considered a syntype of M. bicristimana. New characters and a key are provided to facilitate identification of the Indian Ocean species of Mursia. The contrasting geographic distribution patterns of the two genera are discussed with regard to their vertical distribution.


Introduction
Two genera of box crabs (Calappidae), Calappa Weber, 1795 andMursia Desmarest, 1823, are known to occur in deep-water habitats of the Indian Ocean. There are, however, considerable differences in vertical and geographical distribution between the two genera. Even though the recent revisions by Galil (1993Galil ( , 1997 facilitate the identification and interpretation of the available material, the distribution in particular of deep-water species is still insufficiently known. In Calappa most of the 15 species currently known from the Indian Ocean (Galil 1997;Ng et al. 2002) inhabit shallow water. Only two species, C. depressa Miers, 1886 andC. japonica Ortmann, 1892, were previously reported below 200 m (Galil 1997;Bellwood 1998). Both species, however, are not restricted to deep-water habitats but

Size (CL6CW)
Juvenile specimens from the Red Sea (SMF 13592): 13.6615.9, 13.3615.4, and 6.967.4 mm; male from the Gulf of Aden: 46.4661.6 mm; female from the Gulf of Oman: 56.3677.9 mm; specimens from Japan: 53.5670.2 and 96.66133.7 mm.

Habitat
Outside the Red Sea the species is known from 30 to 380 m depth. In the Red Sea it occurs between 363 and 785 m. The habitat in the Red Sea where the IORAS specimen was collected, was observed from the submersible Pisces and characterized as ''small plateau in front of the steep rift zone covered with pteropodous mud''. In the Gulf of Aden the species was collected together with the swimming crab Charybdis smithii MacLeay, 1838, known to inhabit muddy substrates in a depth range from approximately 200 to 300 m (Tü rkay and Spiridonov 2006).

Distribution
Southern to central Red Sea, Gulf of Aden, Gulf of Oman, Arabian Sea, East to South Africa, Gulf of Bengal, East China Sea, Japan. Sakai (1937) was the first author to synonymize C. exanthematosa Anderson, 1894 with C. japonica. Galil (1997) followed this view. Our specimens from the Gulf of Aden and the Gulf of Oman show no considerable differences to those from Japan with regard to the carapace and cheliped sculpture, the shape of the basal antennal segment, and the morphology of the Go/2. However, the carapace of the juvenile specimens from the Red Sea have a subcircular rather than an elliptical shape and, in general, more distinct tubercles and stronger granulation than those from Japan. The smaller the crabs are, the more these characters are expressed and might lead to confusion with C. pustulosa. The tubercle pattern, however, is the same as in adult specimens from the Gulf of Aden and from Japan, and the basal antennal segment is very similar to other specimens of C. japonica examined ( Figure 1A-C). Thus the specimens from the Red Sea are considered as belonging to C. japonica and the differences in carapace shape and granulation are considered to be size-dependent variation.

Type locality
Off Ganjam and Orissa coasts, Gulf of Bengal.

Type material
Probably in the ZSI. Number of specimens (syntypes) not known.

Deep-water Calappidae
Size (CL6CW) Specimen from the Strait of Mozambique: 54.6660.3 mm; specimen from Japan: 33.7637.1 mm.

Habitat
On sandy or muddy bottom. The present record extends the known depth range which is now from 40 to 228 m.

Distribution
The present record extends the known range to the south-western Indian Ocean, i.e. the Strait of Mozambique. Further distribution range: Maldives, Bay of Bengal, Andaman Sea, Philippines, south China Sea, Japan.

Remarks
Both examined specimens agree well with existing descriptions and illustrations with regard to the carapace and cheliped morphology and the shape of the basal antennal segment ( Figure 1D). The tubercles in the large specimen from the Strait of Mozambique are somewhat more smoothed than in most C. pustulosa, including our specimen from Japan.

Distribution
Along East African coast from Kenya to Mozambique. A record from Thailand by Ng et al. (2002) is questionable, since especially the relative length of the lateral spine is a difficult character to distinguish between M. africana and similar species, such as M. flamma, and use of the key provided by Galil (1993) might lead to misidentifications (see ''Remarks'').

Remarks
The present specimens agree well with the description of Galil (1993) and the holotype deposited in the RMNH, which has been examined for comparison. Even though M. africana is similar to M. flamma Galil, 1993, in particular regarding the shape of the Go/2 [compare Figures 3a, b and 10c, d in Galil (1993) and our Figure 3A, B], the two species are distinguished by the following characters: carapace relatively wider and less convex in M. africana, lacking the additional tubercle between median and mesial branchial row of tubercles which is present in M. flamma; posterior margin of carapace with flattened, nearly effaced lobes in M. africana and three sharp teeth in M. flamma; the merus of the cheliped bispinose (but with a minute granular tubercle in addition to the two spines) in M. africana and trispinose in M. flamma. Contrary to the description provided by Galil, the median frontal lobe is more or less triangular in M. africana, while it is more rounded and bulbous in M. flamma.
The relative length of the lateral spines, a character often used for distinction of the species within the genus, however, is subject to considerable size-related variation. In M. africana the LS/CW ratio ranges from 0.35 in small specimens to 0.20 in larger size specimens, while in M. flamma it varies from 0.20 to 0.07. Thus, if this character is used for distinction of the two species the size of the actual specimen needs to be taken into account. Otherwise the use of this character might result in misidentifications.  Lloyd 1907, p 6;Kemp andSewell 1912, p 29. ? Mursia bicristimana: Galil 1993, p 356-357 (part: Alcock and Anderson (1894) and Anonymous (1914)].

Type material
In the original description by Alcock and Anderson (1894) there is no mention of the number or sex of the specimens obtained at Investigator St. 151. However, since the size (CL6MCW521641 mm) is given for only one specimen without indication of further material, this might indicate that the description and original record is based on a single specimen. In this case, that specimen has to be considered a holotype. Its size indicates that it probably is the same as that illustrated by Alcock and Anderson (1896, Plate 24, Figure 5). If not lost or destroyed, this specimen should still be deposited in the ZSI in Calcutta, but all efforts to get any information or the specimen on loan failed. In case that there is more than one specimen in the original sample from Investigator St. 151 a lectotype should be selected.
Subsequently a large male (CL6MCW547.8688 mm) was reported from Investigator St. 204 off Colombo (06u509200N, 79u369200E, 180-271 fathoms, 18 April 1895) by Anderson (1896). This specimen was mentioned in Alcock's (1896) account of the Calappidae and described and figured in great detail by Alcock (1899, p 23-24, Plate III, Figures 3, 3a, b). This specimen, however, has no type status and most probably it is deposited in the ZSI.
A specimen in the NHM (1898.8.26.3) was considered as a syntype by Galil (1993). Examination of this specimen revealed that it is accompanied by a label carrying the following information ''Syntype? or Alcock det. 11u149300N, 74u579150E, 68-140 fathoms, coll. Indian Museum''. Thus, the locality data on the label points to the Laccadive Islands, not to the Gulf of Manaar. Most probably the specimen was collected at Investigator St. 246, 15 October 1898, since the coordinates as well as the depth information given on the label agree exactly with the data for that station. It is therefore clear that the specimen was collected at a different locality and four years after the description of M. bicristimana was published, consequently this is not a syntype of that species. As will be discussed in detail below, a close examination of the specimen reveals that it belongs to a distinct and yet undescribed species.

Size
The measurements of the presumed type given by Alcock and Anderson (1894) are CL 21 mm, MCW 41 mm; those of the large male mentioned by Alcock (1896Alcock ( , 1899 are CL 47 mm, CW 67 mm. Adult males from the Gulf of Aden measure CL 28.1-48.2, CW 35.5-65.8, LS 4.6-6.2 mm; the females from the same station measure CL 36.1-43.7 mm, CW 45.9-56.5 mm, LS 5.5-6.3 mm (broken in smallest specimen). The male specimen from Error Seamount (ZMMU Ma 5353) measures CL 50.0 mm, CW 65.5 mm, LS 8.0 mm. The female from the Nicobar Islands measures CL 39.9 mm, CW 51.1, LS 9.8; the largest male from the Andaman Sea measures CL 44.2 mm, CW 59.8, LS 8.0.

Habitat
Records range from 250 to 420 m depth. Information on the sediments is scarce. In the case of Meteor 5, St. 267 it consisted of fine sand with numerous gastropod and bivalve shells (A. Allspach, personal communication). At Investigator St. 360 the sediment was probably fine sand (Lloyd 1907) and at Investigator St. 391 it was recorded that the trawl contained a remarkable number of the gastropod Xenophora pallidula Reeve, 1842 (Kemp and Sewell 1912).

Distribution
Gulf of Aden, Arabian Sea (off Socotra Is., Error Seamount), Laccadive Sea, Gulf of Manaar. Specimens tentatively assigned to M. bicristimana, but probably belonging to very similar but distinct species, have been recorded from the Andaman Sea and the Philippines.

Extended diagnosis (based on Alcock and Anderson and specimens from the Gulf of Aden and the Arabian Sea)
Carapace transversally oval, convex, 1.6-1.7 times broader than long (including lateral spines) and 1.25-1.35 times broader than long not counting lateral spines. Dorsal surface finely granular with one median and six radial rows of tubercles (protuberances according to Crosnier 1997aCrosnier , 1997b and one additional tubercle in each anterolateral region. Front wider than orbit, trilobate, median lobe broadly triangular, projecting well beyond rounded lateral lobes. Supraorbital margin unifissured. Inner suborbital tooth subtriangular with external edge straight or slightly convex, separated from outer orbital margin by U-shaped hiatus opening into oblique subhepatic canal. Anterolateral margin arched, with 11 acuminate granular denticles decreasing in size posteriorly. Lateral spines of moderate length, measuring from about one-tenth of CW in adult specimens from the Gulf of Aden to about one-third of CW in small juveniles, directed horizontally and slightly anterolaterally. Posterior margin with lateral lobes sharply triangular, median lobe small to indistinct.
Merus of cheliped bispinose, distal spine shorter than lateral spine of carapace. Upper margin of palm bearing a row of seven triangular teeth increasing in size up to sixth, last tooth smaller than the previous one. Outer surface bearing a continuous generally smooth ridge above lower margin with low sharp tooth proximally and finely granulated depression in distal half. Above this ridge seven granular tubercles forming three indistinct longitudinal rows. Area between the ridge and the rows of tubercles almost smooth. Lower margin serrated, but without well-developed teeth.
Ambulatory legs laterally compressed. Length to width ratio in meri varies from 3.4 in P2 to 3.1 in P5. Propodi with length to width ratio varying from 3.1 in P2 to 2.5 in P5, sharply keeled dorsally.
Crest on tergite 2 of abdomen trilobate, lateral lobes rounded, median subrectangular with almost straight or slightly convex posterior border, not projecting considerably beyond laterals. Penultimate segment of male abdomen subquadrate with sinuous margins. Ultimate segment (telson) forming equilateral triangle with slightly concave lateral margins.
Go/1 elongate, curved and tapering distally towards spinulose tip. Spinules forming terminal field around opening and extending in a thickening row along mesial face. Go/2 much longer than Go/1 with hook-shaped terminal tube, tip not reaching half distance to junction with basal part; junction mesially with pair of stiff bristles.
Female genital sternite and two previous sterna partly included in sterno-abdominal cavity densely beset with minute spinules similar to those on Go/1. Genital opening covered with spheroid cap, shifted close to anterior edge of sternite, posteriorly bordered by smooth elevated thickened cuticle, mesially bordered by subtriangular smooth area with median eminence.

Size-related variation
Within the material from the Gulf of Aden and the Arabian Sea the relative length of the lateral carapace spines decreases with size. Small specimens have lateral spines of about one-third of CW, while those of adult specimens do not exceed about one-tenth of CW. The expression of the median lobe on the posterior margin of the carapace is also variable. While small, but usually well discernible in adult males, it is very indistinct in females, and in juveniles this lobe cannot be recognized at all.

Coloration
Alcock and Anderson (1894) describe the coloration of M. bicristimana as ''salmon pink''. When obtaining the second specimen, Anderson (1896, p 103) made the following notes regarding the colour in life: ''upper surface of leg and carapace pale bluish white studded with orange red granules, lower surface white, inner surface of merus of cheliped deep orange''. All specimens studied by the authors had been stored in alcohol for extended periods and their colour was pale cream with a slightly darker carapace. In some specimens an orange spot on the inner surface of the palm near the junction with the movable finger was still visible.

Affinities
Mursia bicristimana resembles M. africana, M. flamma, M. coseli, and M. danigoi in carapace morphology and tubercle pattern on the carapace, median lobe of front projecting beyond laterals, unifissured supraorbital margin, and Go/2 morphology. It is, however, distinguished from all above species by the outer surface of the palm bearing a continuous, indistinctly trilobed, mostly smooth ridge composed of three fused lobes above the lower margin. In M. africana and M. flamma all three of these lobes are widely separated. In M. coseli the lobes are separated from each other by granular depressions forming a wave-like ridge. In M. danigoi a subsidiary tubercle partly fills the gap between the proximal and the median lobe while the distal lobe is widely separated. Further distinguishing characters and their respective character states in the different species are given in Table I.

Remarks
The identity of M. bicristimana is difficult to clarify because of the inaccessibility of the type specimen (or specimens) in the ZSI. As already discussed above, the specimen from NHM certainly is not a type of M. bicristimana and, due to some considerable morphological differences such as the number of spines on the cheliped merus (four spines versus two in bicristimana), it cannot even be considered conspecific with the specimen described and figured by Anderson (1894, 1896). The diagnosis given by Galil (1993) and in particular the characteristic shape of the Go/2 of the specimen thus do not refer to M. bicristimana sensu Alcock and Anderson, but to a yet undescribed taxon. Lloyd (1907) reported M. bicristimana from the western Gulf of Aden (RIMSS Investigator St. 360, 13u369N, 47u329E, 130 fathoms, probably fine sand) and Kemp and Sewell (1912) mention a record from the Laccadive Sea (RIMSS Investigator St. 391, 9u149100N, 75u459E, 237 fathoms). In both cases the material is probably housed in the collections of the ZSI and was not available for examination. It is, however, probable that the specimens are conspecific with the other material from the northwestern Indian Ocean and is therefore assigned to M. bicristimana.
The specimens from the Andaman Sea listed by Galil (1993) under the same species, however, do not have the characteristic Go/2 and apparently are not conspecific with the new taxon mentioned above. These specimens appear to be conspecific with a specimen from the Nicobar Islands collected by the Deutsche Tiefsee-Expedition and identified as Mursia armata bicristimana by Doflein (1904). The specimens from the Andaman Sea and the Nicobars as well as those from the Gulf of Aden and the Arabian Sea agree with M. bicristimana sensu Alcock and Anderson in most characters including the characteristic ridge above the lower margin of the palm, the morphology of the front and the posterior border. There are, however, important differences between the material from the northwestern and the eastern Indian Ocean. In specimens from the Nicobars and the Andaman Sea the carapace bears a well-developed additional tubercle between median and first radial row of tubercles ( Figure 6A). In specimens from the Gulf of Aden, this additional tubercle is hardly visible ( Figure 5A). The lateral spines of the carapace in adult specimens from the Andaman Sea and the Nicobars reach about 0.15-0.18 times CW, while the Gulf of Aden specimens have somewhat shorter spines reaching only about 0.10 times CW in adult specimens. Similarly, the distal spine on the cheliped merus in specimens from the Eastern Indian Ocean is between one-quarter and one-third of the length of the merus excluding the spine, while it is somewhat shorter in specimens from the Gulf of Aden reaching only about one-fifth of the merus length.
Comparing both forms with the figure in Alcock (1899), it appears that the specimens from the Gulf of Aden are more similar to the one depicted there, since the carapace is quite smooth and the merus of the cheliped is clearly bispinose. Thus, the specimens from the Gulf of Aden would be true M. bicristimana (providing, that the specimen depicted in Alcock 1899 is the same species as that described by Alcock and Anderson 1894), while those from the Andaman Sea are similar, but differ in several characters and might belong to a distinct species. Since, however, direct comparison with the type material was not possible the identity of true M. bicristimana is still not completely clarified, therefore no further taxonomic and nomenclatural actions were taken. The specimens from the Nicobars and the Andaman Sea are therefore preliminarily assigned to M. bicristimana.
Furthermore, five specimens from the USNM collected by the RV Albatross in the Philippines and identified as M. bicristimana have been examined. The specimens, ranging in CW from 20.2 to 40.6 mm, agree with M. bicristimana by possessing the characteristic  ridge above the lower margin of the palm and by having a bispinose cheliped merus. They do, however, differ from M. bicristimana as described above, by having a smoother and less granular carapace, relatively longer lateral spines (0.21-0.24 times CW) and longer distal spines on the cheliped merus (one-third to almost half the length of the merus excluding the spine). Furthermore, the meri of the last pereiopods are more slender and have a less granular surface and upper margin in specimens from the Philippines than in specimens of similar size from the Gulf of Aden. For example, the largest specimen from the Philippines (CW 40.6 mm) has a length/width ratio of 3.5 for the merus of the last pereiopod, while the same ratio in a specimen from the Gulf of Aden (CW 34.4 mm) is only 3.2, even though the specimen is smaller and the legs of smaller specimens tend to be more slender in both forms. As for the specimens from the Andaman Sea, they are tentatively assigned to M. bicristimana for the time being although they may represent a distinct taxon.   Zarenkov).

Type locality
Gulf of Manaar, Sri Lanka.

Etymology
The epithet ''minuta'' (Latin for ''small'') refers to the body size of the species which is much smaller than most other species within Mursia.

Habitat and vertical range
For the type specimens from the Gulf of Manaar no ecological data are available, but due to the fact that the material originates from a study of the pearl oyster banks it can be assumed that they probably originate from relatively shallow waters. The specimen from the Laccadives (NHM 1898.8.26.3) is from a station with sandy/stony bottom and a depth range of 124-271 m. The specimens from the Gulf of Aden have been collected on muddy sand at depths of 83 and 87 m.

Distribution
Gulf of Aden, Laccadive Islands, Gulf of Manaar.

Description (based on holotype)
Carapace transversally oval, convex, ca 1.65 times broader than long (counting lateral spines) and ca 1.25 times broader than long not counting lateral spines. Dorsal surface coarsely granular excepting frontal and orbital region, which is finely granulated. A longitudinal row of four blunt tubercles (protuberances according to Crosnier 1997aCrosnier , 1997b along the midline, two closely set tubercles anterior to it on mesogastric region. Branchial regions with two converging radial rows of four blunt tubercles each, the external one sinuous and meeting base of lateral spine. A mesial additional row of three less-marked tubercles present between median row and mesial branchial row, with anterior tubercle being rather well-developed and others hardly discernible. Second additional row consisting of two tubercles present between two branchial rows. Each hepatic region with a row of four tubercles, increasing in size anteriorly. Two additional tubercles on each side of gastric region (anterior being smaller than posterior) and one in each anterolateral region.
Front of about the same width as orbit, depressed, trilobate, median lobe broadly triangular, pointed; lateral lobes rounded, somewhat elevated, less produced than median. Supraorbital margin parabolic, with a single fissure but some grooving in place of second fissure. Inner suborbital tooth sub-triangular, with outer margin straight and inner margin convex, separated from outer orbital margin by V-shaped hiatus opening into oblique subhepatic canal. Subhepatic region markedly granular. Anterolateral margin arched, carrying nine acuminate granular denticles with largest in middle part of arch and two to three additional granules anteriorly.
Lateral spines moderately long and stout, comprising ca 0.15 of CW, nearly straight, directed horizontally with markedly granular surface. Posterolateral margin sinuous, granular, with some granules in anterior half larger than others. Posterior margin moderately granular, trilobate: both median and lateral lobes low, triangular, similar in size. Thoracic sternum and visible parts of sterna with moderate development of even granulation, those parts of sterna within sterno-abdominal cavity smooth.
Third maxillipeds elongate; exopod bearing longitudinal fringe of hair along complete length. Endopod ischia not closing in midline, leaving a narrow triangular gap; inner margins bearing rows of rounded teeth decreasing in size distally; merus with distolateral corner forming rounded process.
Chelipeds equal in size, evenly covered with granules similar to those on carapace excepting part of lower surface of merus and inner surface of palm. Distal margin of merus bearing three laterally directed rather stout spines; distal spine largest reaching distinctly more than half length of lateral spine of carapace and ca 0.4 of maximal merus length (excluding spines); following spine reaching about half length of distal spine, and third spine being smallest. Proximally to third spine is a small spinule or sharpened tubercle. In holotype, this tubercle is more developed in left cheliped, almost indistinguishable in right one. Carpus with three tubercles on outer surface, hardly distinguishable from background granulation, anterodistal corner triangular. Upper margin of palm bearing a row of two rather blunt proximal teeth and seven subtriangular teeth. Outer surface in lower part bearing a granular ridge with three distinct elevations, proximal triangular and sharp, median and distal rounded, subequal in size. Depressions between elevations are markedly granular. Above this ridge eight granular tubercles form three indistinct longitudinal rows: lower and median ones with three, and upper with two tubercles each. Lower surface granular, lower margin with coarse granules and six to eight denticles in anterior half. Inner surface nearly smooth in upper part, sparsely granular in lower part, with angled fringe of hair along lower margin and base of fixed finger. Fixed finger slightly deflexed, cutting edge bearing a row of irregular rounded teeth in left cheliped, and a large crushing tooth in right one. Movable finger with granular outer face and typical molariform proximal process in major cheliped; inner surface with milled ridge extending for most of length and becoming less clear distally.
Ambulatory legs laterally compressed, P3 being longest (but not much longer than P2). Length to width ratio in meri varies from 3.35 in P2 to 3.08 in P3 and 3.06 in P4. Upper surface and lower margin of meri granular, distal margin unarmed; in last pair of legs granulation extends to most of posterior face excepting lower distal corner. Upper surface of carpi with three granular carinae, median one ending in a terminal spine; distal margin with well-developed spines decreasing in size from P2 to P4. Propodi with length/width ratio comprising ca 3.7 in P2-P4, sharply keeled dorsally, in P2 and P3 keels serrated and preceded by a sparse row of granules at anterior surface of article; anterior and posterior surfaces with usual grooves. Dactyli longer than propodi, slender and slightly curved.
Female characters and variation. Female sterna sparsely and finely granulated. Female abdomen smooth, terga separate: third tergite being broader than second and slightly broader than fourth and fifth. Genital opening covered with spheroid cap, shifted close to anterior edge of sternite, posteriorly bordered by smooth somewhat elevated cuticle, mesially bordered by subtriangular smooth area with median eminence.
Egg mass extends laterally somewhat beyond coxa of anterior ambulatory legs. Eggs small, measuring ca 0.3-0.4 mm in diameter.
The male from the Laccadive Islands is similar to the holotype, with somewhat less coarse granulation on the carapace. Only the anterior tubercle in the mesial additional row between the median and the mesial branchial row of tubercles is recognizable. The meri of P5 (missing in the holotype and the paratype) lack a distinct spine distally and are sparsely granular over most of posterior surface; carpi lack anterior carina. Length to width ratio in meri varies from 3.40 in P2 to 3.26 in P3, 3.22 in P5, and 3.16 in P4. Propodi length/width ratio varies from 3.0 in P2, 3.72 in P3, 4 in P4 to 2.9 in P5.

Affinities
The present species is closest to Mursia australiensis Campbell, 1971, known from Australia, New Caledonia and, until recently with some reservation, from Japan (Campbell 1971;Galil 1993). A male specimen from Sagami Bay, Japan (Haberer coll., ZMB, unregistered), examined by us clearly belongs to that species, thus confirming the occurrence of M. australiensis in Japan.
Similarity between the two species includes general outline, carapace and cheliped granulation, and the Go/2 having a rather short and relatively stout curved terminal part. Within the genus this type of gonopod morphology is shared only by the above two species and Mursia trispinosa Parisi, 1914 (see Galil 1993, Figure 10h, i).
Mursia minuta n. sp. can be distinguished from M. australiensis by the lobes near the lower margin of the cheliped manus forming a granular ridge with three elevations, while in M. australiensis the three lobes are widely separated and do not form a ridge-like structure. The median lobe of the crest on the second abdominal tergite is broadly subrectangular with an almost straight margin in M. minuta, while it is rounded and much narrower in M. australiensis. The Go/2 in M. minuta is somewhat spiralled while it is sinuous in M. australiensis.
Mursia trispinosa is distinguished from M. minuta n. sp. by the three isolated large and pointed teeth near the lower margin of the cheliped palm and the strongly denticulated lower margin of the palm. Furthermore, in M. trispinosa the posterior margin of the carapace is bilobed, and the Go/2 is crochet-shaped.

Coloration
Even though the specimen has been stored in alcohol for an extended period and has lost most of its coloration, there are orange markings present in the pterygostomial region anterior to the base of the chelipeds, on the upper face of the cheliped merus, and on the inner surface of the cheliped palm near the base of the movable finger. In the specimen from Taiwan assigned to M. danigoi by Tan et al. (2000) there is a vertical red patch on the base of the cheliped movable finger as well as on the inner surface of the palm and there are four minute dots on each side of the distal margin of the buccal cavity. According to the colour photograph of the above authors, the background colour of the carapace is orange. A pattern of dark red patches extends posteriorly up to the epibranchial regions while carapace and cheliped tubercles are whitish (Tan et al. 2000).

Habitat and vertical range
The specimen from Saya de Malha was collected at a depth of 230-235 m. Mursia danigoi from the Western Pacific was collected at a depth of 143-204 m.

Distribution
The examined specimen originates from Saya de Malha Bank in the western Indian ocean, while M. danigoi is known from the Philippines and Taiwan.

Remarks
The present specimen is close to M. danigoi and M. coseli in having a quadrispinose cheliped merus and a trilobate posterior margin of the carapace. According to Crosnier (1997b), the distinguishing characters between the two species include a more uneven granulation of the carapace in M. coseli and the length/width ratio of the P5 merus being 2.6-2.8 in M. danigoi versus 3.3 in M. coseli. The specimen from Saya de Malha shows the character state as described for M. danigoi in having a length/width ratio of the merus of the fifth pereiopod of 2.6.
It differs, however, from both species in having the anterior part of the anterolateral margin of the carapace unevenly curved, the lobes near the lower margin of the cheliped merus being incompletely fused forming a ridge-like structure and the Go/2 having a shorter recurved terminal part, not reaching the junction with proximal part. Since, however, only a single specimen is available and the extent of individual variation of these characters is unknown we refrain from describing a new taxon.

Habitat and vertical range
Known from 10 to more than 300 m depth.

Distribution
South Africa from Natal to Namibia. St Helena Island in the South Atlantic. The latter record by Doflein (1901) was questioned by Manning and Holthuis (1981) but confirmed by our examination of the original material in the ZSM.

Remarks
Full synonymy and a more detailed discussion of the nomenclatorial problems are given by Holthuis (1993, p 603ff).

Type locality
Madagascar.

Type material
Male holotype (MNHN B24371) and numerous paratypes deposited in the MNHN; one paratype in the South African Museum (Galil 1993).

Habitat
The overwhelming majority of records are from a depth range of 300-510 m. The shallowest station where we recorded the species had a depth of 210-230 m. Galil (1993), however, indicates it from St. 3 of Mascareignes III Cruise (22u27.39S, 43u079E) at a depth of 35 m.

Distribution
Off Madagascar and along east African coast from Natal to Tanzania; West of Sumatra, Indonesia.

Remarks
An important character not mentioned in the original description of the species, but seen both in the photographs of type specimens (Galil 1993, Figure 7; Crosnier 1997b, Figure 5B) and in our material, is an extra pair of tubercles located between the inner branchial and median rows of tubercles on the carapace. They may be indistinct in juvenile specimens. Contrary to the description given by Galil, the frontal median lobes of M. flamma are rounded, not triangular. The female genital opening and surrounding cuticular field are similar to those in M. bicristimana, but are located a greater distance from the sternite edge and the mesial eminence is much less pronounced and more regularly shaped than in M. bicristimana. Examination of the specimens collected by the Deutsche Tiefsee-Expedition off Nias Island, west of Sumatra and published by Doflein (1904) as Mursia armata curtispina Miers, 1886, revealed that these specimens correspond better with the description of M. flamma Galil, 1993 than with that of M. curtispina. In particular, the surface and anterolateral border of the carapace are more coarsely granulated than in M. curtispina, and the patterns are identical to those of M. flamma from East Africa. The median tooth on the posterior margin is not as small as in M. curtispina and the teeth forming the lowest row of tubercles on the outer face of the palm are more elevated and sharper than in M. curtispina. Furthermore, the Go/1 of the male from Nias agrees well with that of M. flamma from East Africa. Since, however, M. curtispina had been described on the basis of a single female specimen, comparison of that character is not possible.
Poore (2004) reported M. curtispina to occur in Australia (Western Australia, Northern Territories, Queensland, and New South Wales). However, the lateral spine of the figured specimen (Poore 2004, Figure 97b) is about a quarter of the maximal carapace width. In that character this specimen is different from M. flamma and thus it is possible that M. curtispina or a similar taxon occurs in the easternmost part of the Indian Ocean off the Western Australian coast.
Key to the Indian Ocean species of Mursia *M. danigoi has not been recorded from the Indian Ocean, but is included in the key due to its similarity to the specimen from Saya de Malha identified as M. sp. aff. danigoi.

Discussion
Traditional views of deep-water biodiversity and zoogeography assume that species diversity is low and that most species are widely distributed, this being attributed to the lack of distributional barriers and uniformity of environmental conditions (Ekman 1935(Ekman , 1953. This concept of a species-poor, more or less homogeneous deep-water fauna within broad latitudinal ranges is to some extent based upon a scarcity of available material and fragmental faunistic knowledge. Limited understanding of species diversity in deep water has limited incorporation of deep-sea organisms into zoogeographic analyses. With regard to the Calappidae and the Indian Ocean region, prior to the revisionary works by Galil (1993Galil ( , 1997, six species from Calappa and Mursia were known to occur below 200 m depth. In Calappa, two species, C. japonica and C. depressa Miers, 1886, were known from deep-water localities within the region. New collections and intensive taxonomic research that included Galil's (1997) revision have not led to increased numbers of deep-water species being reported from the region, except for a new record of C. pustulosa from Madagascar in the present paper. All three of the species recorded from the shelf break in the Indian Ocean are widespread with distributional ranges including the western Indian Ocean as well as the western Pacific.
The presence of C. japonica in the Red Sea, where it inhabits unusually deep water of up to 785 m depth, is of particular interest. There are no clear signs of morphological differentiation between the Red Sea and Indian Ocean populations. In many other decapod taxa the populations in the deep Red Sea are considered to be distinct subspecies, vicariant species, or species without a clear affinity to the Indian Ocean congeners (Tü rkay 1986(Tü rkay , 1996Tü rkay 2001, 2007;Tü rkay and Spiridonov 2006). These varied levels of evolutionary differentiation are probably the result of multiple isolation events caused by sea-level changes and current reversals in the Straits of Bab al-Mandeb since the Miocene (Klausewitz 1983a(Klausewitz , 1983b(Klausewitz , 1989Tü rkay 1986). Thus it can be concluded that C. japonica either has invaded the Red Sea very recently, probably during the holocene transgression following the last Ice Age, or the observed morphological homogeneity of C. japonica indicates a slow evolutionary change of morphological characters within that species, or even within the genus Calappa.
Mursia in contrast shows a very different pattern of diversity and distribution in deep waters of the Indian Ocean. Four species (M. aspera Alcock, 1899, M. bicristimana, M. cristiata, andM. curtispina Miers, 1886) were known from deep waters throughout the Indian Ocean prior to the work by Galil (1993), which described two additional species, M. africana and M. flamma. Subsequently, Crosnier (1997b) described M. coseli. Moreover, in the present paper one species M. minuta n. sp. is described from more shallow waters, and a second (M. sp. aff. danigoi) is reported from the central Indian Ocean. In M. bicristimana regional variation in several important characters was observed and could indicate more than one taxon is included under the current concept of that species. Thus, within the last decade, the number of species of Mursia known to occur in the Indian Ocean has doubled and probably will further increase in the future. Several of these species have more or less restricted geographical ranges within the Indian Ocean ( Figure 15). Contrary to the situation in the genus Calappa, there is a comparatively high diversity of species with more restricted distributional ranges in Mursia. This agrees well with data from other recent taxonomic revisions and biogeographical analyses of tropical deep-water decapod fauna, which indicate that species numbers are considerably higher and distribution ranges often much more restricted than previously thought (Manning and Holthuis 1981;Macpherson and Baba 1993;Guinot and Richer de Forges 1995;Lemaitre 1996Lemaitre , 1999Spiridonov and Zhadan 1999;Spiridonov and Tü rkay 2001;Tü rkay and Spiridonov 2006).
Ecologically, both genera are similar in inhabiting soft sediments (Sakai 1976), but there are considerable differences regarding their bathymetric ranges. While the species of Mursia are almost completely restricted to the shelf break and upper continental slope with a bathymetric distribution centred around 300 m, most species of Calappa inhabit shallow shelf regions (Bellwood 1998) and even those occurring deeper usually range into relatively shallow waters.
Mursia and Calappa belong to different clades of calappid crabs, and palaeontological data indicate that they diverged not later than the Oligocene and by that time might have already had different depth preferences (Bellwood 1998). Apparently Mursia is adapted to living in relatively cold deep waters, while Calappa is a mainly shelf-inhabiting group adapted to warm shallow waters (Galil 1997). The unusual occurrence of C. japonica at a depth of more than 700 m in the Red Sea can be easily explained by the extraordinary temperature characteristics of the Red Sea with a homoiothermic water column, measuring 21.5-22.0uC from 200 m down to the greatest depths (Siedler 1969;Morcos 1970).
We suppose that the observed differences in geographical distribution patterns are closely related to the differences in vertical distribution between the two genera. Other studies of decapod distribution also indicate that shallow-water shelf taxa are generally more widely distributed in the Indo-Pacific than upper slope taxa (Spiridonov and Zhadan 1999).
A possible explanation for this somewhat surprising pattern might be the more patchy distribution of suitable habitats along the continental slope and shelf break. The mosaic distribution and variety of habitats in that zone stand in sharp contrast to the environmental conditions on the shelf and in the abyssal plains, which are usually areas of sediment accumulation. The slope, however, is mainly an area of denudation, but where possible mosaic accumulation of sediments takes place and generates a high habitat heterogeneity (Zenkevich 1977). Zenkevich's generalizations derived from the observations of the great deep-sea expeditions of the mid-20th century and were recently confirmed by direct observations using submersibles (Roux 1994).
In the case of Mursia, being restricted to soft sediment substrates and a relatively narrow vertical range, the mosaic set of habitats results in a fragmental distribution along the slope. As in other brachyuran decapods the larvae of Mursia are probably planktonic with a high ability for dispersal, but the fragmentary distribution of suitable habitats and the strong vertical gradient leads to elimination of a high proportion of larvae if there are no mechanisms of retention.
Continental slopes and seamounts usually form a complicated pattern of meso-and macroscale eddies. It is generally accepted that local circulation systems developing over seamounts provide retention of the larvae of seamount-dwelling demersal organisms, in particular fishes, even though these eddies are not stationary. To achieve retention it is sufficient if these eddies are stable over periods as long as the duration of larval pelagic life (Parin et al. 1985). This was also demonstrated using a simulation model incorporating turbulent diffusion (Rudyakov and Tseitlin 1985). The eddies usually associated with boundary currents over continental slopes may similarly contribute to the restriction of larval dispersal of continental slope species, but case studies do not exist. In the northern Indian Ocean there may be other oceanographic mechanisms restricting larval dispersal, in particular the seasonal current reversal related to the monsoon (Wyrtki 1973;Neiman et al. 1997). Larvae released in the transitional season probably have better chances to settle close to the parental habitats and not to be transported to areas where settling is impossible. Moreover, seasonal reproduction appears to be a rather common phenomenon in the north-western Indian Ocean in several animal taxa including brachyuran crabs (Zarenkov 1971;Daniel and Chakrapany 1983;Couwelaar et al. 1998;Tü rkay and Spiridonov 2006) and might be an adaptation to the above-mentioned oceanographic phenomena.
Both mosaic distribution of habitats and oceanographic phenomena might result in a considerable restriction of gene flow and consequently in increased species diversity and restricted species distribution in shelf break and upper slope taxa.