BULLETIN OF MARINE SCIENCE, 41(2): 309-321, 1987

ETHOLOGY AND DISTRIBUTION OF PYLOCHELIDAE (CRUSTACEA COENOBITOIDEA)

Jacques Forest

ABSTRACT The Pylochelidae differ from the other hermit crabs by the complete segmentation of the abdomen and the presence of paired appendages on each of its segments. They do not usually inhabit gastropod shells, but dwell in decayed pieces of wood, stones, tusk-shells, or living . A recent revision, founded on most of the previously recorded specimens and on a large unidentified collection, increased the number of known species from 16 to 39, and the genera from 5 to 7. Two new subgenera have been established, and the family divided into six subfamilies. This paper deals first with the eco-ethological characteristics of the different taxa. According to their dwelling, genera and subgenera can be classified, as a whole, as xylicolous, petricolous, tusk-dwellers, spongicolous, with a few specifical or individual ex- ceptions. In connection with the habitat, adaptive features have been described: opercular structures, boring "rasp," stridulating apparatus ... The Pylochelidae are present in the Indo- West Pacific (36 species or subspecies in 6 genera), and in the NW Atlantic (4 species in 3 genera). Two genera only, belonging to the sole non monotypic subfamily, provide a biogeo- graphical link between the two areas. In I-W.P., the family is known from the SW Indian Ocean to Japan, Kermadec Islands and New Zealand. Indonesia, with 14 species and 5 genera appears as a center of dispersion and diversification. Japanese endemism is noteworthy: one genus and six of the seven species have not been reported elsewhere. The probable relation between the availability of dwelling material and the geographical distribution is also dis- cussed. The vertical distribution extends from 30 to 1,570 m, but the group is mostly rep- resented between 200 and 500 m, where 28 species are living.

The Pylochelidae differ from all other hermit crabs in having a well developed abdomen, all segments of which are articulated and provided with a pair of appendages, similar to the normal abdomen of other reptant decapods. Pylochelids are commonly called "symmetrical" hermit crabs, but this is not entirely correct, because in one genus the abdomen, telson and pleopods are noticeably asym- metrical. The group has been considered as rare, only 16 species being recorded from a small number of rather deep water stations in Indo-West Pacific (I-W.P.) and western Atlantic, and the most being known only from their type localities. The abundance of a new material, originating mainly from ALBATROSSdredgings and from recent French explorations in the I-W.P. has led to a systematic revision (Forest, 1987). As a result, 24 new species-group taxa have been added to the 16 previously established valid species, the five known genera have been redefined, the genus Pylocheles has been divided into three subgenera, and two new genera have been proposed. Until now the Pylochelidae has been considered as a relatively restricted family of infrequently encountered species; aside from three forms captured on several occasions in Japanese waters, the total number of specimens recorded in literature did not exceed 60, caught at about 30 stations. The present revision includes more than 400 specimens, collected at some 200 stations! The importance of pylochelid fauna in tropical and subtropical waters has been, therefore, much underestimated and, most probably, new taxa and localities will be added in the future. This research, however, has not been restricted to the description of new forms.

309 310 BULLETIN OF MARINE SCIENCE, VOL. 41, NO.2, 1987

Investigations on relationships between the various genera have shown that the whole group is made up of several distinct phyletic lines whose respective affinities are not entirely clear, and is likely polyphyletic. Therefore, the family has been divided, provisionally, into six subfamilies. The classification and content of the family Pylochelidae is given in Appendix 1. The Pylochelidae, within the section Paguridea of the infraorder , are classified in the superfamily Coenobitoidea. A comparative study of their main characters allows us to suggest that they are close to the Diogenidae. They cannot however be regarded as primitive representatives of that family; both Diogenidae and Pylochelidae (if monophyletic) probably have a common ancestor, but they apparently evolved independently along various phyletic lines. The richness of the new material upon which the systematic revision of the family has been based, has also provided considerable information on the ecology or the habitat of many forms, and allows one to interpret the adaptive value of some morphological structures. The first part of the present communication is devoted to this aspect of the pylochelid biology. The second and third parts give an account of the biogeographical and bathymetrical distributions of these most interesting pagurids. Etho-ecology of Pylochelidae. - The abdominal tergites in the Pylochelidae are better calcified and accordingly less vulnerable than in most other marine pagurids. However, these seek protection by dwelling in hollow objects. From lit- erature it was known that gastropod shells are infrequently used. Rather, pylo- chelids use wood or stone fragments, tusk-shells or sponges. Examination of numerous specimens still in their dwellings, sometimes observed at the time of their capture, provides the basis for these statements, and allows in many cases the accurate determination of the exclusive or preferential habitat of species be- longing to at least eight of the nine genera or subgenera. These dwellings are discussed below: WOOD DWELLINGS. These come from the shore or rivers and consist of decayed fragments oftrees excavated by wood-boring organisms, or hollow stems such as . All species of the subgenera Xylocheles and Bathycheles. in the genus Pylocheles. are wood-dwellers, or xylicolous. The size of the piece of wood is usually in relation with that of the inhabitant; small animals are found in stems of about I cm in diameter, middle-sized or large ones in much larger fragments weighing sometimes more than I kg. The cavity is subcylindrical, with a diameter of more than 2 cm for large animals, and always longer than the inhabitant. The walls of the lodging are extremely regular, owing to the activity of the , using rasps of tubercles on the carpus of the chelipeds. Presumably, the hermit first selects a piece of wood already hollowed by wood-borers, smoothens the walls with its rasps, and widens the cavity as it grows. The back of the lodge is rounded but remains often open to the exterior through the small primitive gallery. On the schematic illustration of Pylocheles (Xylocheles) macrops in its dwelling one can see the cylindrical main chamber, and the small primitive gallery (Fig. lb). All species of Xylocheles and Bathycheles have rasps and, as a rule, are xylic- olous. However, of 20 specimens of B. incisus collected at one station, about half dwelled in large tusk-shells of Fissidentalium magnificum, Parapylocheles scorpio, a monotypic genus, is also xylicolous and is usually found in . However, we were surprised to discover in the Philippines FOREST: SYMMETRICAL HERMIT CRABS 311

(; ,\ , ~~ .. I ,I~!!' I I, !. .:r- I' ), '~~'

,~: I ": I

k,': \ , ~' J ,; \

, ; ,r,", 'I > \' ',,\, " :', ¥, I 1 ',J ft a }, III b , ~ l

Figure I, a. Pylocheles (pylocheles) mortensenii Boas in a pumice stone; b, pylocheles (Xylocheles) macrops Forest in a piece of wood. one specimen settled in a corn cob! Apparently, there is no rasping apparatus in the genus Parapylocheles. STONE DWELLINGS. These are found mainly in broken pieces of soft stones such as sand conglomerates, limestone and volcanic rock, especially pumice. The two species of the nominotypic subgenus Pylocheles, P. (P.) agassizii and P. (P.) mortenseni, live most frequently in that last type of habitat; many specimens of the latter have been found in pumices weighing from a few to 80 g. Here again, the animal is located inside a more or less deeply hollowed subcylindrical cavity 312 BULLETIN OF MARINE SCIENCE, VOL. 41, NO.2, 1987

(Fig. la). Species of the genus Cheiroplatea also live in stone dwellings, and have similar opercular chelipeds. In these petricolous pylochelids, the chelipeds are narrowly coadapted, the hands perpendicular to the axis of the cavity, closing it perfectly. This is not the case in xylicolous forms where the appendages remain more or less parallel to the axis of the cavity and where both hands are not perfectly coadapted. On the other hand, there is no rasping apparatus on the carpus of the chelipeds in Pylocheles (pylocheles) and Cheiroplatea. Strong teeth at the edge ofthe carpus and propodus may be instrumental in fitting out the habitat. Some specimens have been collected in dead bryozoans or in living sponges. Presumably these latter organisms were epibiont on the primitive mineral dwelling of the hermit. The new genus, Cancellocheles, is probably also petricolous. The four known individuals of C. sculptipes have been described without any information on their habitat, but, from a strong convergent similarity with Cancellus (the known species of which live in stones, dead corals or calcareous algae), it can be presumed that they have a similar way oflife. In both genera, chelipeds and second thoracic legs are coadapted in the formation of a perfect operculum. SHELLDwELLINGS.Apart from a small number of Trizocheles, and from Mix- topagurus paradoxus A. Milne Edwards, from the Western Atlantic, pylochelids do not use gastropod shells. There are 12 known specimens of Mixtopagurus, only one of which has been collected with its dwelling, a shell of Xenophora sp. All specimens however have a slightly twisted, asymmetrical abdomen, and it can be presumed that the species is adapted to life in the same sort of wide, low-spiralled shell. The tendency of pylochelids to seek shelter in a more or less straight tubular cavity explains why they often inhabit tusk-shells, since they are elongated, conical and slightly arched. These shells are the exclusive dwelling of Pomatocheles jef- freysii from Japan, the most frequently recorded species of the family. Its chelipeds are modified as an operculum, and the following legs are coadapted for a perfect fit into the tube. SPONGEDWELLINGS.In the true pylochelid- association, the latter or- ganisms are not epibionts that secondarily settled on a mineral or shell habitat, but rather apply themselves directly and exclusively to forming a shelter for the crab. This type of association appears predominant amongst the members of the genus Trizocheles, though the habit is known for only half of the 18 species or subspecies recorded. Two sorts of sponges may shelter Trizocheles: T. boasi, T.loquax, T. caledonicus and T. brevicaulis live in dictyonin hexactinellids, thin walled sponges with more or less internal tubular structures. On the other hand T. pulcher, T. spinosus spinosus and T. spinosus bathamae (New Caledonian specimens) may be found in lithistid or haplosclerid demosponges, which are much more massive. In all of these examples, the association appears as a permanent mutualistic one, the hermit being sheltered by the sponge, which in tum may benefit from alimentary remains of the host. Some Trizocheles have another type of dwelling: T. sakaii, from Japan, lives in tusk or gastropod shells or in serpulid tubes. The type specimen of T. balssi, from South Africa, was lodged in a coral skeleton, and some New Zealander T. spinosus bathamae in gastropod shells. There is a good general correspondence between type of habitat and taxonomic divisions in the family. Genera and subgenera are in part established from shared adaptations to a same way of life, boring rasps in the xylicolous, operculiform FOREST: SYMMETRICAL HERMIT CRABS 313 chelipeds in petricolous forms, and coadaptation of thoracic appendages in tu- bicolous, for example. Exceptions to this rule may be explained by scarcity or lack of the usual material, and can be occasional and opportunistic, as for the normally xylicolous Bathycheles found in tusk-shell, or, rarely, specific, for T. sakaii, already mentioned. With the exception perhaps of some forms living in tusk shells, pylochelid pagurids have a permanent dwelling that they enlarge progressively while growing (xylicolous and petricolous forms) or that grows with them (spongicolous forms). They are most probably not able to move on the bottom with their shelter like other pagurids. Because of these heavy, clumsy dwellings, they presumably leave them only temporarily for alimentary andlor reproductive purposes. This sed- entary life is a limiting factor in the dispersion of species, in possible relation with an abbreviated development, similar to that of petricolous diogenids such as Cancellus or of some Paguristes. The development of a stridulating apparatus, allowing sexual recognition, may also be related to low population densities in addition to sedentary habits.

Biogeographical Distribution. -Members of the Pylochelidae are known from two disjuncted geographical areas, the Indo-West Pacific and the western Atlantic. Their representation in these two areas is very unequal; 35 species belonging to 6 genera and 5 subfamilies in I-W.P., against 4 species belonging to 3 genera and 2 subfamilies in W. Atlantic. They are primarily inhabitants of tropical waters, but some species extend to temperate regions. In the I-W.P., their distribution is extremely wide, extending from South Africa to the Kermadec Islands and from southern New Zealand (ca. 46°S) to Japan (ca. 38°N). Table 1illustrates the geographical distribution of genera, subgenera and species. Indonesia, with 14 species in 5 genera appears as the center of the present dis- tribution of pylochelids. Nine species in three genera are now recorded from the Philippines. Japan is in third place for species richness (7), but its generic diversity (5 genera) is higher. The high degree of Japanese endemism (6 species and 1 genus) must be emphasized. Although only two genera and three species are recorded from the north-eastern part of the Indian Ocean, diversity and endemism are again noticeable in the south-western part of this ocean, with four species belonging to four genera, all of which are present in Indonesian waters. . In the south-west Pacific, the wide area extending from 20° to 46°S between E. Australia and the Kermadec includes seven species in three widely distributed genera; only one of these seven species extends outside this sector to Indonesia and Japan. In the western Atlantic, as previously indicated, pylochelid representation is poor. There are four species recorded, of which one belongs to the endemic, phylogenetically isolated genus Mixtopagurus (as here restricted). The three re- maining species belong to genera or subgenera much better represented in I-W.P. and provide a link with the fauna of the latter region. If now, instead of global biogeographical relations, the range of each of the subfamilies or genera is considered, the situation appears rather variable. Their distribution is illustrated in three maps (Figs. 2-4), two for the I-W.P. (Figs. 2, 3), the third for W. Atlantic (Fig. 4). The five monotypic subfamilies are (I) the Cancellochelinae, known from four 314 BULLETIN OF MARINE SCIENCE, VOL. 41, NO.2, 1987

Table 1, Geographical distribution of Pylochelidae. For genera and subgenera, the figure after the black dot indicates the number of species in the considered region

AllanlU: Indian Ocean Jndlan Ocean Indonesia PhIlippines Japan Auslraha N. Calednnia N. Zealand Kermadec W S.W N.E E & SE

genus PVLOCHEL(~ .2 .' .2 .5 .5 ., ., - ., - subgenus PYLOCIHlES ., - - ., ., .' - ., - P. (p.)

Number 01 tipeCieS 4 4 3 14 9 7 2 3 4 1

specimens of the unique species, Cancellocheles sculptipes from Tosa Bay (Japan); (2) the Parapylochelinae, with also a monotypic genus, Parapylocheles, have a wider range, since P, scorpio, is recorded from the North-eastern Indian Ocean and the Philippines; (3) the Pomatochelinae include the rather common Poma- tocheles jeffreysii, distributed in Japanese waters to about 38°N, and two other FOREST: SYMMETRICAL HERMIT CRABS 315

fA IlOCH(l(S

D Illnglc8 •• 11. iEl .10".11 fJbO •• 1 m brevlctull. EJ 11lqU •••. IE b.-I.ll

D cal"donle .••• EEl pu I cher

II "lb.-Cro'il Em Ipinolul 'PlnOI ••• CJ g,.c ill .• m ap,nol'" b.themae o moo •• 1 ~ perplal(ul o I'-Ufllnl •• m mannlngl m bllchYQPI POM,. TOCHE L [5

b. leftr.}'111 .A atrldullnl .£. galll.rdl

CANC[LlOCHEL[S

)( II:U I pt I pel

o'

~""

30'

50'

Figure 2. Distribution of species of Trizocheles, Pomaloche/es and Cancelloche/es. species, P. gaillardi from Banda Sea, and P. stridulans from Comoro Islands. The extension of the genus is thus wide, but disjuncted; (4) the Trizochelinae, with 18 species or subspecies, has an exclusive Indo-West Pacific range, but within this area considerably more extended than the preceding ones. The genus Trizocheles is by far the more diversified, though the larger part of its species have a restricted range, the knowledge of a number of them being in fact limited to their type localities. There are five species (T. longicaulis, T. boasi, T. moosai, T. brevicaulis and T. mutus) in Indonesia, three (T. gracilis, T. laurentae, and T. manningi) in the Philippines, and three (T. loquax, T. albatrossi and T. sakaii) in Japan. In the south-western Pacific, Trizocheles is present in eastern Australia (T. spinosus spinosus), New Caledonia (T. spinosus bathamae, T. caledonicus and T. pulcher), and in New Zealand (T. spinosus bathamae, T. brachyops and T. perplexus). Only one species, T. balssi, has been recorded from the south-western Indian Ocean; and (5) the isolated subfamily Mixtopagurinae is exclusively western Atlantic, with a rather wide range; Mixtopagurus paradoxus is distributed from Guiana to North Carolina and from Barbados to the north western part of the Gulf of Mexico. The subfamily Pylochelinae includes two genera, pylocheles and Cheiroplatea. Both have in the I-W.P. a distribution pattern more or less similar to that of Trizocheles, but they are also represented in the western Atlantic. The 10 species of Pylocheles are classified in 3 subgenera, Pylocheles (2 sp.), Xylocheles (2 sp.) and Bathycheles (6 sp.). 316 BULLETIN OF MARINE SCIENCE, VOL. 41, NO.2, 1987

PVLOCH[LES CH[ I ROPLAT("

o morlon'llnll Q) 'Illc.ud. emltrt' ® Itanu'''1 Om.crop, @c.nobll. 0'""'·' @)mllol 4) prolundus ® pumlclcola o Inl ••g •• , P"RAPYLOCH(l (5 G CIOlnl.rl .corplo o m.cgllchrlltl *

0'

".: ~ ....

30'

30· sri'

Figure 3. Distribution of Parapyloche/es and ofI-W.P. species of Pylocheles and Cheiroplatea.

Pylocheles (P.) mortenseni is rather common in eastern Japan, from Kyushu to Sagami Bay, and some specimens are known from the Philippines, Banda Sea, eastern Australia and around New Zealand. The second, near by species P. (P.) agassizii has a Caribbean distribution. The two species of the subgenus pylocheles (Xylocheles), P. (X) macrops and P. (X) miersii, live in the central part of the Indonesian Archipelago and in the Philippines; one, P. (X) miersi, is also present in Andaman Sea. Of the six recorded species of the subgenus pylocheles (Bathycheles), three [P. (B.) incisus, P. (B.) integer and P. (B.) profundus] are restricted to the same Indonesian-Philippines area. One [P. (B.) macgi/christi] is known from the Gulf of Bengal only, another [P. (B.) crosnieri] from the western coast of Madagascar. The last species, P. (B.) chacei, belongs to the western Atlantic fauna. Some peculiar features confer to that last form an isolated systematic position. The genus Cheiroplatea is widely distributed, but every one of the six species appears limited to a narrow range. Four are known from their types only, and all specimens of each of the two others have been collected in adjacent stations. Two, C. laticauda and C. cenobita, come from Banda Sea, one, C. mitoi, from Japan; two other I-W.P. forms originate, one from the Kermadec lsi. (c. pumicicola), the other from the Comoro (c. stenurus). The last species is western Atlantic, and has been described from the Gulf of Mexico (c. scutata). FOREST: SYMMETRICAL HERMIT CRABS 317

• 0' ... TO' .0'

3d'

pvt.aCH[I.[S

• O·t;I·IIIEII 01 8 ..11 •••• ,

CHE IROPLlH[A O_CUI.'.

Ml.llTOPACURUS

• p ••••••••

'D'

10'

Figure 4. Distribution of species of Pylochelidae in the western Atlantic.

Bathymetric Distribution. -Our present knowledge on the vertical distribution of the members of the Pylochelidae is illustrated in Table 2. They occur between 30-1,570 m, but representation is extremely variable. Five of the 8 species col- lected between 30 and 200 m are only occasional at these depths and live usually deeper. Diversity is highest between 200 to 500 m, with 28 species. From 500 to 750 m, only nine species are found; beyond 750 m the number of species decreases to five, one only having been collected twice under 1,000 m. Vertical distribution of genera and subgenera is unequal: Pomatocheles appears restricted to rather shallow waters; the Japanese P. jeffreysii has been recorded from 30 to 200 m, more frequently between 100 and 150 m. The type and unique specimen of P. gaillardi was captured at 90 m. The third species of the genus, P. stridulans, is somewhat deeper, at 250 and 300 m. Cancellocheles, endemic from Japan, has a narrow bathymetric known range, 250-300 m. This is also the case of the western Atlantic Mixtopagurus paradoxus, captured from 200 to 365 m. Relative stenobathy occurs also in Cheiroplatea; five species (laticauda, cen- abita, mitoi, pumicicola and stenurus) range from 300 to 565 m. The depth of capture of the sixth species, scutata, is not known. The monotypic genus Para- pylocheles is deeper, with a wider range; the nine stations of collections of P. scorpio increase regularly from 400 to 900 m. The two large genera Pylocheles and Trizocheles have, together with a wide geographical distribution, the largest vertical range. Subdivision of Pylocheles in three subgenera corresponds to eco- logical and bathymetrical preferential requirements: (I) The petricolous subgenus Pylocheles lives beween 100 and 400 m, P. (P.) mortenseni mostly from 250 to 318 BULLETIN OF MARINE SCIENCE, VOL. 41, NO.2, 1987

Table 2. Vertical distribution of Pylochelidae. The broken line indicates the possible distribution

Depth In meters 0 100 200 300 400 500 600 700 800 900 1000

genus PYlOCHElES

subgenus PYlOCHEl(S

P (P ) lliI ••,.1 % Ii

P (P ) mo len •• nll

subgenv s XYlOCH(LE 5

P (X ) mlar ,I

p (X ) moc-rops _1510 subgenus OATHYCHEL(S

p [8 >. chico I •••••••••••_•••••••••••~••••••••••••••••••••••••••••••••••••••••••••••••••••••••~•••••••••••••_& ••••••••••••••••••_ ••••_••_ ••••_ ••••••••••••••••••- ••-.

P (0 ) Inc is •• , _ 1344 et 1570 P (D ) profundus

P (8 ) In t ger

P (0 ) c rOn i a I I P (8 ) mlegi Ich Istl :::::::::: ....::.~:::::::::::::::.:::::::::::: ..::.::::::::::::::···················I··--···········_····_~.~:.:::::::::::::.::~=:::::::::=:=.:::=:=:::=::::::::.::::::.:: genus CHE IROPlATEA C \.t I cauda ':::::::::::::'::.: ..=::'..::::::~ .~.~.~.~:.~.~ ~, ~.=::=::::==::=:::::::=:=:=::::.::::::.:::::::::::::=:::===:'.:=:= C Jeu tel. C s tllnu I u .....- -...... ------. .- - -...... •..•...... - .

C "Ii to I ~=~~~=.~~~~..=~~=~~.~~~=~~~..~~~~.~....~~:~~:~~~~~~~~~~~~~~~=~~.~~:~~..~~~~~~~~~~~=~~~:.~~~~~~:=:~:~::~:=:=~_..~~=~:::=~~'.'~'.'.'.~~~'.:::::~ C pumlcleol. - _ -..•.....•...... -.-.....•...... • -- -- .....•...... -.- -...... •..•.•..... genus POMATOCHElES .....•.....•...... - - -...... •...... •. -- -..-.- -.-- .•..... P J. i r t.y I i I

P gllliudi

P .t,ldvl.".

genus PARAPYlOCHElES

P leal p I 0

genus C~NC[tlOCHEL[S

C Iculptlpea

genul TRIZOCHELES

T langle.ulll

T bOI. I

T I aqui. . -•...... _-.- - - -....•. - -...... •. -..- - - -" ,.-- -..--- - . 1 cllodonlcu.

T .Ibat 0 I

T 9' Ie I II • T .mooslld

T'lurent.e

T brlchyops ,. I.kell

T b •••• I

T pulcher

T .plnolu •• plnolu.

T splnolu, bathama.

T p. rp IlllCUI T mannlngl .-.- -...... •.....•. - - - - -. T mutu a

genus MIXTOPAGURUS ---.- ..-..- ,

400 m, P. (P.) agassizii from 200 to 300 m. (2) Xylicolous Xylocheles have about the same depth requirements and are found from 170 to 450 m. P. (X.) miersi is, however, more frequent between 300 and 400 m and P. (X.) macrops between 180 and 220 m. (3) The subgenus Bathycheles, known from 400 to 1,570 m, is the deepest. This last subgenus differs mainly from Pylocheles (Xylocheles) by reduced eyes. The vertical distribution of the six species is rather extended and FOREST: SYMMETRICAL HERMIT CRABS 319

sometimes overlaps widely. All are present in the 500-l,000-m range, but three [P. (B.) incisus, integer, crosnieri] may be shallower (400 m), whereas one [Po (B.) profundus] can exceed 1,000 m, with captures at 1,344 and 1,570 m. The overlap between the ranges of the subgenera Xylocheles and Bathycheles is not large, but clear enough since both of them may be found from 400 to 450 m. One must notice that in the three species of Bathycheles eventually collected at less than 500 m there are no morphological features intermediate with those of Xylocheles. One can therefore presume that Bathycheles differentiated in rather deep water, around 600-800 m, and that a few species were able to migrate up to shallower bottom where Xylocheles species still lived. The genus Trizocheles is encountered from about 100 to 750 m. The average depth of capture of 12 of the 17 recorded species is between 200 and 500 m. One species only has been collected around lOOm, and three have a restricted range between 580 and 680 m. The deepest comes from 750-913 m. As a general rule, the vertical range for each species of Trizocheles is relatively narrow (120 m at most). The genus thus appears as the most stenobath amongst Pylochelids. Ten species however have been collected only once, so that their vertical extensions remain unknown. These species will possibly be caught at different levels, but, on the other hand, they may be restricted to a very narrow depth range. That could explain why they have not been collected at other depths in nearby stations.

SUMMARY AND CONCLUSIONS

My first comment concerns the localization and extension of the various subfam- ilies in the distribution area of the family. As already mentioned the pylochelid group of pagurids appears as an assemblage of distinct ancient lineages. The differences observed in the present geographical pattern of distribution of these various lines are undoubtedly related to differences in their respective capacities for adaptation to ecological variations during the course of geological times. The fact that exclusively Indo-West Pacific subfamilies are best represented in Indo- nesian and Japanese waters may be explained alternatively: either Indonesia or Japan were the past centers of diversification and dispersion of an ancestral group, or the persistence of adequate conditions has permitted in these two areas, the maintenance of various members ofa formerly widely distributed Tethyan fauna. One can note that the two subfamilies Pylochelinae and Trizochelinae are together the richest in species number and the more vastly distributed in I-W.P. The other subfamilies, mono- or paucispecific, may represent relicts of lineages with poor adaptive potentialities. Concerning the relations between I-W.P. and W. Atlantic faunas, similar hy- potheses may be formulated. Setting apart the monotypic, W. Atlantic endemic Mixtopagurinaes the Atlantic fauna includes three species belonging to the subfamily Pylochelinae, which has also a broad I-W.P. distribution. Moreover, the Caribbean form, P. (P.) agassizii, is so close to the I-W.P. P. (P.) mortenseni that, if not for their distinct range, they could be considerecd as subpspecies. Disjuncted distribution of Pylochelinae thus appears to represent the survival of Tethyan relicts. The other subfamilies, Mixtopagurinae in W. Atlantic, Parapylochelinae, Po- matochelinae, Trizochelinae, Cancellochelinae in I-W.P., may have shared in the past a wide common distribution, and ecological changes may have induced their disappearance from numerous areas. However the possibility of a separate di- versification since the fragmentation of the Tethyan Sea, may also be hypothesized. 320 BULLETIN OF MARINE SCIENCE, VOL. 41, NO.2, 1987

There is apparently no correlation between geographical and vertical distri- butions. That is, shallower species are either strictly endemic or widely distributed. Pomatocheles jeffreysii and Pylocheles mortenseni, for instance, are both living above 300 m; the first is restricted to Japanese waters, while the second known to range from Japan to Indonesia and New Zealand. Similarly, of the two deeper species, Parapylocheles scorpio and pylocheles (Bathycheles) profundus, the first is recorded from the North-Eastern Indian Ocean and the Philippines, while the second is known only from the Mindanao Sea. A last comment concerns possible relations between habitat and geographical distribution. Definite and usually strict habitat requirements condition the set- tlement of most of these Pagurids. There is apparently a correlation between the distribution of genera, subgenera or species and the presence of appropriate sub- strata. Xylicolous pylochelids need wood fragments to settle and are numerous in places wherever these fragments are available. In fact, nearly all xylicolous Pylocheles (Xylocheles), P. (Bathycheles) and Parapylocheles have been collected in deep sea channels between the wooded islands ofIndonesia and the Philippines where there is an accumulation of tree debris. Similarly, petricolous pylochelids live preferentially in soft stones like pumice pebbles, mostly found in relatively recent volanic areas. And all known Pylocheles (pylocheles) and Cheiroplatea originate from such areas: Japan, Kermadec, New Zealand, Indonesia, Comoro, West Indies, and from nowhere else. The hypothesis that the distribution of spongicolous pylochelids, that is of the genus Trizocheles, is dependent upon that of particular sponges could in the same way be advanced. However, available data are still too fragmentary and no de- finitive conclusions can be drawn. The lack of appropriate hosts and the possession of the necessary adaptive abilities may have induced the Japanese species Tri- zocheles sakaii to settle in various shells, mainly Dentalium but also gastropod shells or serpulid tubes.

LITERATURE CITED Forest, J. 1987. Les Pylochelidae ou "Pagures symetriques" (Crustacea Decapoda Coenobitoidea). Res. Camp. MUSORSTOM. I & II. Philippines, Yol. 3. Mem. Mus. natn. Hist. nat" Paris, ser. A, Zoo!., 137: 1-254, figs. 1-82, pis. I-IX.

DATE ACCEPTED: April 7, 1987.

ADDRESS: Laboratoire de Zoologie (Arthropodes). Museum national d'Histoire naturelle et Laboratoire de Carcinologie et Oceanographie biologique. Ecole Pratique des Hautes Etudes, 61, rue de Buffon. 75005 Paris. France.

ApPENDIX I

Family PYLOCHELIDAE Bate, 1888

Subfamily Pylochelinae Bate, 1888 incisus Forest, 1987 Genus Pylocheles A. Milne Edwards, 1880 profundus Forest, 1987 Subgenus Pylocheles A. Milne Edwards, 1880 integer Forest, 1987 agassizii A. Milne Edwards, 1880 crosnieri Forest, 1987 mortensenii Boas, 1926 macgilchristi (Alcock, 1905) Subgenus Xylocheles Forest, 1987 Genus Cheiroplatea Bate, 1888 miersi Alcock et Anderson, 1898 laticauda Boas, 1926 macrops Forest, 1987 scutata Ortmann, 1892 Subgenus Bathycheles Forest, 1987 stenurus Forest, 1987 chacei Forest, 1987 cenobita Bate, 1888 FOREST: SYMMETRICAL HERMIT CRABS 321

mitoi Miyake, 1978 albatrossi Forest, 1987 pumicicola Forest, 1987 gracilis Forest, 1987 Subfamily Pomatochelinae Stebbing, 1914 moosai Forest, 1987 Genus Pomatocheles Miers, 1876 stenolepis Forest, 1987 jeffreysii Miers, 1876 brachyops Forest et de Saint Laurent, gaillardi Forest, 1987 1987 stridulans Forest, 1987 sakaii Forest, 1987 Subfamily Parapylochelinae Forest, 1987 brevicaulis (Boas, 1926) Genus Paraphylocheles Alcock, 1901 balssi (Stebbing, 1914) scorpio (Alcock, 1894) pulcher Forest, 1987 Subfamily Cancellochelinae Forest, 1987 spinosus spinosus (Henderson, 1888) Genus Cancellocheles Forest, 1987 spinosus balhamae Forest et de Saint sculptipes (Miyake, 1978) Laurent, 1987 Subfamily Trizoche1inae Forest, 1987 perplexus Forest, 1987 Genus Trizocheles Forest, 1987 manningi Forest, 1987 longicaulis (Boas, 1926) mutus Forest, 1987 boasi Forest, 1987 Subfamily Mixtopagurinae Bouvier, 1895 loquax Forest, 1987 Genus Mixtopagurus A. Milne Edwards, 1880 caledonicus Forest, 1987 paradoxus A. Milne Edwards, 1880