Norntates PUBLISHED by the AMERICAN MUSEUM of NATURAL HISTORY CENTRAL PARK WEST at 79TH STREET, NEW YORK, N.Y

Home , Axelrodichthys, Coelacanthidae, Diplocercides, Latimeria, Macropoma, Mawsonia (fish)

AMERICAN MUSEUM Norntates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 2866, pp. 1-30, figs. 1-26, 1 table December 18, 1986

Coelacanths from the Lower Cretaceous of Brazil

JOHN G. MAISEY1

ABSTRACT Specimens of Mawsonia sp. (possibly repre- dermal bones, the dermosphenotic morphology, senting the type species, M. gigas), plus a new and the number of paired posterior elements in genus and species of fossil coelacanth, are de- the skull roof suggest that the new coelacanth is scribed from the Romualdo Member of the San- related to Mawsonia. A phylogenetic hypothesis tana Formation, Chapada do Araripe, Ceara (Bra- is offered in which Mawsonia plus the new taxon zil). The new form, Axelrodichthys araripensis, is represent the sister group ofMacropoma and Lati- distinguished from other coelacanths on the basis meria. The new coelacanth is represented by sev- ofits cranial anatomy. The posterior moiety ofthe eral complete, articulated skeletons, among the first skull roofhas a median dermal element plus three to be discovered in South America. paired ossifications. Heavy rugose ornament ofthe

INTRODUCTION Fossils of coelacanth fishes from South 40 km north-south across the Ceara-Pernam- America are extremely rare and until now buco border and into Piaui (Lima, 1978). The have consisted mostly of incomplete and principal collecting areas in Ceara are around fragmentary specimens. The discovery of Santana, Jardim, and Missao Velhao. Access complete examples in the Lower Cretaceous is generally difficult and hazardous. The Cha- Santana Formation of Ceara (northeastern pada do Araripe is capped by the Exu For- Brazil) is therefore ofconsiderable interest to mation, some 200 m of current-bedded red paleontologists. sandstones and siltstones above the Santana The Santana Formation is exposed along Formation. The latter has much greater lith- the flanks of the Chapada do Araripe, a pla- ological variation, with interbedded lime- teau extending some 180 km east-west and stones, evaporites, marls, and siltstones 1 Associate Curator, Department of Vertebrate Paleontology, American Museum of Natural History. Copyright © American Museum of Natural History 1986 ISSN 0003-0082 / Price $3.60 2 AMERICAN MUSEUM NOVITATES NO. 2866

(Mabesoone and Tinoco, 1973). Fossil fishes tion. One of these is referable to Mawsonia, occur in concretionary limestones at various but according to Campos and Wenz (1982, levels in the Santana Formation, but partic- p. 1152), "Les proportions relatives de ularly toward the base of the upper (Romu- l'avant et de l'arriere-crane, notamment la aldo) Member. Slight differences of lithology presence d'un museu etroit et allonge, indi- allow recognition of concretions from differ- quent que le Mawsonia de Ceara est speci- ent localities and horizons. Santana Forma- fiquement distinct de l'espece de Bahia et des tion coelacanths in the AMNH collection formes Africaines, toutes datees du Neocom- come mostly from Santana, some 50 km west ien a l'Albien." Unfortunately in the other of Juazeiro do Norte, but also from Jardim, coelacanth taxon from Ceara (here regarded slightly under 50 km to the southeast. Some as a new genus), the anterior skull roofis even specimens (e.g., AMNH 11759, from San- more elongate and narrow than in Santana tana) are associated with ostracods resem- Formation specimens referred to Mawsonia, bling Pattersoncypris; the low diversity ofos- and so it is impossible to determine which of tracods within the concretionary limestones the taxa Campos and Wenz (1982) were re- is said to indicate a fresh or brackish envi- ferring to Mawsonia. As discussed below, the ronment (Bate, 1972), although this may only two coelacanth taxa from the Santana For- be true of the lower part of the Romualdo mation share several peculiar features, in- Member (Mabesoone and Tinoco, 1973). It cluding postparietals (also found in other is uncertain whether the coelacanths from Mawsonia spp.), suggesting that they are Araripe were fully marine or else tolerant of closely related. They are certainly separate brackish to fresh water. species, but it is my view that the two are Despite the great abundance and diversity sufficiently distinct to merit generic separa- of the fossil fishes from the Santana For- tion. mation (Silva Santos and Valenca, 1968), the There is some disagreement concerning fauna has still not been adequately surveyed. coelacanth bone terminology. In the present This is surprising when one considers the ex- work I have mostly followed Forey's (1981) cellent state ofpreservation and the ease with terminology, except for features not dis- which these fossils may be prepared in acid cussed by him. (e.g., Toombs and Rixon, 1950; Rixon, 1976, The age of the Santana fishes was set as pl. 8). Although the Santana fishes have been Aptian by Silva Santos and Valenca (1968), known since the early 19th century (Spix and on the basis of comparison with other ich- Martius, 1828; Agassiz, 1841, 1844) several thyofaunas. This determination was rein- rare taxa have only recently been discovered. forced by the discovery of supposedly iden- Among the latest additions to the Santana tical species of fossil fishes in the Riachuelo paleofauna are coelacanths, which were first Formation of Sergipe-Alagoas, below am- reported on the basis of incomplete cranial monite zones representing the Albian and and postcranial remains by Campos and Wenz Upper Aptian (Silva Santos, 1981). While this (1982), who reported two taxa, Mawsonia sp. may place an upper limit to their age, the and "Forme B." Mawsonia was recognized Santana fish fossils could nevertheless pre- on the basis of two large skulls showing sup- date the Upper Aptian by a significant amount posedly characteristic "posterior parietals" (i.e., Neocomian or Upper Jurassic), accord- (termed postparietals here). "Forme B" was ing to available palynological and ostracod founded on a fragment ofthe body, preserved data (Maisey, in prep.). in part and counterpart. Without complete specimens it was not possible for Campos and Wenz (1982) to make direct comparisons ABBREVIATIONS between these supposedly different coelacanths. INSTITUTIONAL The AMNH collection confirms the pres- AMNH, American Museum of Natural History, ence of two distinct coelacanth taxa in the New York Romualdo Member of the Santana Forma- DGM, Divisao de Geologia e Mineralogia, De- 1986 MAISEY: COELACANTHS 3

partamento Nacional da Producao Mineral, Rio SYSTEMATICS de Janeiro CLASS OSTEICHTHYES SUBCLASS SARCOPTERYGII ANATOMICAL ORDER ACTINISTIA afs, anterior fossa of parietal FAMILY COELACANTHIDAE AGASSIZ 1844 ang, angular Genus Mawsonia Woodward antart, antotic articulation (in Mawson and antpr, antotic process of basisphenoid Woodward, 1907) apa, anterior apophysis of parietal EMENDED DIAGNOSIS: Coelacanths of large apr, ascending process of prootic size (estimated body length up to 3 + m); skull art, articular roofing bones and angular ornamented by au, autopalatine heavy rugosities; bsp, basisphenoid operculum and gular or- cc, cranial cavity namented by numerous radiating striae; an- col, supraorbital sensory canal terior moiety of skull roof 1.5 to 2 times as coron, coronoid long as posterior part and from 2 to 2.5 times den, dentary as long as broad; posterior skull roof with di, internal depression of anterior apophysis three paired bones (parietals, supratempor- dpr, dorsal process of ectethmoid als, and postparietals); dermosphenotic with dsp, dermosphenotic splintlike anterior projection, and with in- faeb1, facet for articulation of first epibranchial fraorbital sensory canal located away from foc, fossa for otic canal the anterior margin; lachrymojugal elongate, fr, frontal the posterior g, gular two-thirds almost straight, an- hm, hyomandibular facet teriorly extending to the tectal series; anterior in, intemasal margin of metapterygoid slopes obliquely j, jugular canal forward as it rises from the pterygoid. la, lachrymal process of lateral rostral TYPE SPECIES: Mawsonia gigas Woodward lj, lachrymojugal (1907); Lower Cretaceous (Neocomian), Ba- mppa, median postparietal hia Basin, Brazil. mpt, metapterygoid my, myodome Mawsonia cf. gigas Woodward na, nasal np, notochordal pit REFERRED MATERIAL op, operculum AMNH 11758 Nearly complete skull roof and pa, parietal associated elements oforobran- padl, descending lamina of parietal chial skeleton. par, parasphenoid AMNH 12216 Partial skull roof with sensory pls, "pleurosphenoid" suture canals exposed. pop, preopercular AMNH 12217 Large basisphenoid with part of postcor, "posterior coronoid" parasphenoid attached. ppa, postparietal (paired) AMNH 12218 Large badly damaged basisphe- prcon, processus connectens noid and prootic. preart, prearticular ?DGM 11 07-P, 1109-P (mentioned in Campos pro, prootic and Wenz, 1982). prof, canal for superficial ophthalmic pt, pterygoid All the above specimens are from the Ro- qu, quadrate mualdo Member, Santana Formation, Cha- so, supraorbitals pada do Araripe, Ceara, Brazil. spl, splenial sq, squamosal SYSTEMATIC NOTE st, supratemporal tec, tectal AMNH 11758 closely resembles the ho- VII, facial nerve lotype of Mawsonia gigas, BM(NH)P 10355 la

I in

5cm i

-antpr atb' ; dpr

bsp

hm, j ' pro B

Fig. 1. Mawsonia cf. gigas, AMNH 11758. Acid-prepared skull in (A) ventral; (B) lateral; and (C) dorsal aspects. Prior erosion has resulted in loss of parasphenoid and parts of prootics. 1 986 MAISEY: COELACANTHS 5

- foc

my6:

2 cm B Fig. 2. Mawsonia cf. gigas, AMNH 11758. (A) Anterior view ofotico-occipital portion. (B) Posterior aspect of ethmosphenoid moiety.

(Mawson and Woodward, 1907, pl. VII) in and close agreement is found with the new its palatoquadrate and lower jaw morphol- material (see below). On the basis of these ogy, and in dermal bone ornamentation. Cer- descriptions, in my view the Mawsonia from tain other features distinguish the new spec- Santana cannot be distinguished from the type imens from other Mawsonia spp. (e.g., species (cf. Campos and Wenz, 1982). proportions of skull roof, morphology of basisphenoid, parasphenoid, and otic moiety, DESCRIPTION arrangement and shape of cheek bones), but are unknown in the type species. A few ad- Elements preserved in AMNH 11758 in- ditional features of the parietal in M. gigas clude both the anterior and posterior moi- have been described in detail by Casier (196 1), eties of a comparatively small skull (dimen-

Fig. 3. Mawsonia cf.ggs AMNH 12216. Partial skull in (A) dorsal and (B) ventral aspects. Dermal bone surfaces are wor away, revealing intenal sensory canals. 6 AMERICAN MUSEUM NOVITATES NO. 2866

5 cm ' -4

Fig. 4. Mawsonia cf. gigas, AMNH 12217. Large basisphenoid and associated dermal elements, in (A) ventral and (B) dorsal views.

sions are given below), except for the tapers to approx. 25 mm at the level of the parasphenoid and ventral part of the opis- nasals. The snout is consequently much nar- thotic region (fig. 1). The right palatoquadrate rower than in M. tegamensis and slightly more and articular, both dermosphenotics, squa- tapered than in M. lavocati (Wenz, 1975, fig. mosals and lachrymojugals, and fragments of 1, pl. 1; 1981, fig. 3, pl. 2A), but the propor- the operculars, ceratohyals, and other ele- tions in the holotype of M. gigas are un- ments are also preserved. Another skull roof, known. together with part of the basisphenoid and The antotic process of the basisphenoid is prootic, is represented by AMNH 12216 (fig. broad and winglike (figs. 1, 2). Its anterior 3). Much ofthe superficial ornamentation has face, consisting ofunfinished cancellous bone, been lost by erosion, exposing some parts of is shaped like a gull wing in anterior view, the sensory canal system (discussed further with the antotic process forming the "wing- below). In addition, two very large but frag- tip" where it forms a broad suture. The ba- mentary basisphenoids are represented by sisphenoid is not ossified as far anteriorly as AMNH 12217 (fig. 4) and 12218. Basisphe- the optic nerve. Sphenoid condyles are well noid morphology seems to be a reliable cri- developed, and there is a strong sliding joint terion for distinguishing between the Santana between the processus connectens of the ba- coelacanths (see descriptive sections below), sisphenoid and a groove on the mesial surface and the two large basisphenoids may be fairly of the prootic, but a basipterygoid process is confidently referred to Mawsonia. Only an absent. The basisphenoid is poorly preserved outline description of the cranial anatomy in M. tegamensis. The basisphenoid of M. will be given here. lavocati (Wenz, 1981, fig. 1, pl. 2D-F) is The preserved length ofthe skull in AMNH shorter than that of AMNH 11758, but also 11758 is almost 170 mm, ofwhich the otico- has a broad, winglike antotic process. As in occipital region takes up some 54 mm. The M. tegamensis (Wenz, 1975, p. 180, fig. 2) posterior skull roof is arched like an arc of a the "pleurosphenoid" merges with the ventral circle, and is approx. 80 mm wide at the pos- surface of the posterior frontal, rather than terior margin. The anterior apophyses of the forming a separate ossification as in Lati- parietals (see fig. 5) are approx. 25 mm apart. meria (Millot and Anthony, 1958). The intracranial joint marks the widest part Three other incomplete Mawsonia skulls (40 mm) ofthe ethmosphenoid region, which from Santana (AMNH 12216, 12217, 12218) 1986 MAISEY: COELACANTHS 7

TABLE 1 Relative Dimensions-Santana Mawsonia (in mm) AMNH Catalog no. 11758 12216 12217 12218 Basisphenoid width (max.) 24.5 36 67.5 60.5 Gap between antotic processes 26 42.3 77 Length condylar surface of basisphenoid 23 - 82 Prootic length 63 78 - 110 Basisphenoid "wing" length 14 23 54 40+ Width between parietal articular processes 17.5 27 46 - Length epipterygoid 2cm articulation 27 41.5 82 - Cranial opening in eth- Fig. 5. Mawsonia cf. gigas, AMNH 11758. mosphenoid Posterior skull roof in ventral view, after removal width 12 17.5 26 20 of prootics. height 17 - 60 46

gion ofAMNH 11758 are arranged as in coel- include two extremely large basisphenoids, acanths generally (Schaeffer, 1952; Forey, dimensions of which are given in table 1. In 1981). The frontonasal series consists of six all these specimens the basisphenoid antotic paired bones, although only five pairs are well process is shaped as in AMNH 11758, and developed. The left supraorbitotectal series there is no basipterygoid process. consists ofseven bones, but only six are pres- The otico-occipital region has a pro- ent on the right (the anterior two apparently nounced prootic process jutting anteriorly fused together). The anteriormost tectal is su- over each processus connectens of the basi- tured to the lateral rostral as in Latimeria sphenoid. Dorsolateral to this is a deep myo- (Millot and Anthony, 1958, fig. 3c). The lach- dome, mesial to which is a large foramen, rymal processjuts laterally from the skull roof presumably for the jugular vein and facial and defines parts ofthe anterior and posterior nerve (fig. 2A). This foramen opens into a nares. A small intemasal is present, sutured canal which emerges posteriorly in the mid- anteriorly between the foremost nasals. Pre- dle of an area of unfinished cancellous bone maxillaries and most of the rostral elements (the hyomandibular facet), as in Latimeria. have not been found, probably for preser- Another canal, presumably for the hyoman- vational reasons. dibular nerve, meets this canal mesially. The The otico-occipital region is roofed by absence of perichondral bone over the hyo- paired parietals (intertemporals; parietoder- mandibular facet suggests that the unossified mopterotics), supratemporals, and postpari- epihyal either lacked a mobile articulation etals (seen in AMNH 11758, 12216; figs. 1, with the cranium, or else was separated by a 2, 5). Postparietals were thought to be pe- layer of cartilage. The prootic ascending pro- culiar to Mawsonia tegamensis according to cess is extensively sutured to a descending Wenz (1975), but they are evidently wide- lamina of the parietal, below a large orifice spread among other (perhaps all) Mawsonia for the otic canal. A comparable, but less ex- spp., and they also occur in the other coel- tensive, descending lamina is present in M. acanth taxon from the Santana Formation aff. tegamensis (Wenz, 1975, fig. 4). A su- (see below). Below these postparietals in pratemporal descending process is absent in AMNH 11758 and 12218, separate supra- AMNH 11758. occipital elements (fig. 6) and arcual bones Roofing bones of the ethmosphenoid re- were found; these are apparently better os- 8 AMERICAN MUSEUM NOVITATES NO. 2866

postparietals. However in AMNH 11758, a much smaller skull, the commissure lies behind the supratemporals and postparietals, within a shallow groove (as in the restoration, fig. 7A). AMNH 12216 represents a much larger individual than AMNH 11758, and the observed variation between them could be accounted for by secondary enclosure of the commissure by continued postparietal A growth. In neither specimen is there evidence of anteriorly directed branches of the com- 2cm missure as in Latimeria or Macropoma (For- ey, 1981). Unlike Latimeria (Millot and An- thony, 1958, fig. 10; Jarvik, 1980, figs. 222, 223) and a number offossil coelacanths (e.g., Whiteia, Rhabdoderma; Lehman, 1952; For- ey, 1981), in these specimens the transverse branch of the otic canal forms a continuous medial commissure across the parietals, rather than blind-ending pit lines. This commis- C D sural canal is exposed in AMNH 12216, ex- Fig. 6. Supraoccipital of Mawsonia cf. gigas, tending across the interparietal suture (fig. 3). AMNH 1 1758, in (A) ventral, (B) dorsal, (C) an- Casier (1961) has compared the arrange- terior, and (D) posterior views. ment of sensory canals within the parietal in several coelacanths including Mawsonia gigas, M. ubangiana, Latimeria, Whiteia, and sified in Mawsonia than in Recent Latimeria Laugia. Comparison of AMNH 12216 with (Millot and Anthony, 1958; Jarvik, 1980). his figures reveals closest similarity with M. The parietal morphology of M. gigas has ubangiana in the posterolaterally directed ar- been described in detail by Casier (1961, figs. rangement of the parietal branch of the otic 4A, 5A, 7, 8A, 9A) in comparison with that canal. In the type specimen of M. gigas, the of M. ubangiana from Zaire. In AMNH parietal branch seems somewhat straighter 11758 and 12216, the parietal architecture than in either M. ubangiana or AMNH matches that of M. gigas in virtually every 12216. However that canal was not com- respect, particularly the arrangement of the pletely traced in M. gigas by Casier (1961, articular process (fig. 5). This close agreement fig. 9A), and it could therefore have curved perhaps offers the strongest evidence that the laterally. One notable difference from both Santana specimens belong to the type species M. gigas and M. ubangiana figured by Casier of Mawsonia. (op. cit.) is the continuous transverse branch In many coelacanths an extrascapular se- ofthe otic canal; in AMNH 12216 the trans- ries either fringes (or is attached to) the skull verse and parietal branches of the otic canal table posteriorly, and contains the supratem- were both well developed. poral commissure (Schaeffer, 1952). Forey The circumorbital and cheek bones of (1981) regarded the presence of seven ex- AMNH 11758 do not overlap (fig. 7A). The trascapulars as a synapomorphy of Macro- dermosphenotic (postorbital) is peculiar in poma and Latimeria. It is possible that the having an elongate, narrow splint that pro- postparietals in Mawsonia are the homologs jects anteriorly (figs. 8, 9). The dermosphen- of lateral extrascapulars, which otherwise otic of M. tegamensis is similarly attenuated seem to be absent. In AMNH 12215 the sur- but the process is deeper (Wenz, 1975, fig. 1, faces of the dermal bones have been partly pl. 1). In both cases the process ofthe dermo- stripped away, revealing the internal arrange- sphenotic probably layjust dorsal to the lach- ment of sensory canals (fig. 3). The supratem- rymojugal. The infraorbital canal in M. te- poral commissure extends through the su- gamensis and AMNH 11758 is located toward pratemporals and continues across the the posterior end of the dermosphenotic, 9 1 986 MAISEY: COELACANTHS

,A--

op I. \ sq

% I

Fig. 7. Restorations of Mawsonia head in lateral view; (A) showing dermal bones and sensory canals (light shading); palate, basisphenoid, parasphenoid (dashed), and prootic shaded darker. (B) Cranium and upper jaw.

rather than at its anterior margin as in many of the AMNH specimens referred to Maw- coelacanths. Below the dermosphenotic sonia. Wenz (1975, fig. 1) identified a "pre- other dor- is a small squamosal. A preopercular and opercule inferieur" and "preopercule subopercular have not been identified in any sale" (here = preopercular and squamosal, 10 AMERICAN MUSEUM NOVITATES NO. 2866

.dsp

li sq A 2 cm i i

dsp

B q Fig. 8. Cheek elements of AMNH 1 1758, lateral view; (A) right dermosphenotic, squamosal, and lachrymojugal; (B) left dermosphenotic, squamosal, and incomplete operculum. respectively) in M. tegamensis, and suggested AMNH 11758 and M. tegamensis, the eye that a subopercular was absent, as in some of Mawsonia probably lay farther forward Triassic coelacanths. The lachrymojugal of than in most fossil coelacanths, where the AMNH 11758 is long and slender, reaching posterior border of the eye is situated more the tectal series anteriorly. A similar elongate or less level with the dermal intracranial joint lachrymojugal occurs in M. tegamensis, and (e.g., Whitea, Chinlea, Rhabdoderma; Leh- may be a characteristic of Mawsonia species man, 1952; Schaeffer, 1967; Forey, 1981). In in general. Latimeria the eye is also somewhat anterior From the overall configuration ofthe cheek to the dermal intracranial joint (Millot and series, and the general form of the skull in Anthony, 1958, fig. 1; Jarvik, 1980, figs. 205, 11 1 986 MAISEY: COELACANTHS

dsp

A 2cm i

B Fig. 9. Cheek elements of AMNH 1 1758, mesial view; same bones as in figure 8.

223c). This peculiarity is shared by the other ticular surface for the antotic process; and a coelacanth taxon from the Santana Forma- small autopalatine anteriorly. There is no tion (see below). evidence for an ectopterygoid apart from The palatoquadrate includes a pterygoid, some suturelike rugosities along the thin ven- with a single strengthening ridge on the lateral tral margin of the pterygoid. surface, and a fine covering of teeth over its Fragments of the lower jaw (angular, artic- visceral surface (fig. 10). To the pterygoid is ular, coronoid) and hyoid arch are also pres- sutured the quadrate, with a prominent ar- ent in AMNH 1 1758 (fig. 1 1). As in M. gigas, ticular facet for the lower jaw; the metapter- the articular is not fused to the angular (Maw- ygoid (the "hyomandibula" of Mawson and son and Woodward, 1907). The right artic- Woodward, 1907, pl. VII) with a broad ar- ular of AMNH 11758 is still attached to part 12 AMERICAN MUSEUM NOVITATES NO. 2866

A i 2cm

mpt antart

qu B bw Fig. 10. Right pterygoid, metapterygoid, quadrate, and autopalatine ofAMNH 11758, in (A) mesial and (B) lateral views. 1 986 MAISEY: COELACANTHS 13

A B

2 cm

C D Fig. 1 1. Right coronoid of AMNH 11758 in (A) lateral and (B) mesial views; right articular and fragmentary angular, in (C) lateral and (D) mesial views.

of the angular, including the coronoid em- parietals and supratemporals, plus a much inence. The right coronoid is also preserved smaller pair of postparietals separated by a and can be directly compared with that ofM. median element; dermosphenotic rectangu- gigas (Mawson and Woodward, 1907, pls. lar, extends as far in front ofintracranialjoint VII, VIII), where its "triangular" shape may as behind it; lachrymojugal semicircular, not be an artifact of incomplete preservation. In elongated anteriorly and not reaching lateral AMNH 1 1758 its anterodorsal edge is thick- rostral; supraorbital sensory foramina con- ened and indented, with a posterolaterally fined to snout; antotic process of basisphe- directed process which would have contacted noid short and robust; foramina for VII and the coronoid eminence of the angular. There jugular vein extremely small, jugular canal are small tubercular teeth on only a small area reduced or closed; scales ornamented with of the coronoid visceral surface. narrow, sometimes interrupted ridges; swim bladder wall strongly ossified. Type species: AXELRODICHTHYS, NEW GENUS A. araripensis (see below). ETYMOLOGY: Named in recognition of Dr. DIAGNOSIS: Coelacanths reaching estimat- Herbert R. Axelrod, for his generous and ed lengths of 1-2 m; D1 10-11; D2 9-10; C valuable support of paleoichthyology at the 17-18 in upper lobe, 15-16 in lower lobe; P American Museum of Natural History. 12-15; V 17-18; roofing bones of skull and angular rugose, particularly supraorbitals and tectals; operculum and gular ornamented with Axelrodichthys araripensis, new species numerous fine radiating striae; ethmosphenoid moiety of skull roof 2.5 times longer DiAGNosIs: As for genus. than otico-occipital part, and at least 3 times HOLOTYPE: AMNH 11759; complete fish as long as broad; posterior moiety with large 710 mm long (figs. 12, 13), Lower Romualdo 14 AMERICAN MUSEUM NOVITATES NO. 2866

Member, Santana Formation, Santana, Cha- pada do Araripe, Ceara. REFERRED MATERIAL (from Santana except where noted): AMNH 11760 Large incomplete skull, acid prepared. AMNH 11761 Badly damaged skull prepared in acid. AMNH 12207 Large incomplete skull roof, approx. 300 mm long. AMNH 12208 Skull, approx. 220 mm long. AMNH 12209 Complete fish, cranium not exposed (fig. 1SA); from Jardim. AMNH 12210 Nearly complete fish lacking snout (fig. 15B). AMNH 12211 Incomplete composite fish (fig. 15C). AMNH 12212 Complete fish (fig. 15D). AMNH 12213 Complete small fish with flattened skull (fig. 15E). AMNH 12214 Flattened head with rostral bones. AMNH 12215 Partial skull prepared in acid (fig. 25). AMNH 12219 Large incomplete fish. AMNH 12220 Small (370 mm long) complete fish in part and counterpart (fig. 16). DESCRIPTION Although several specimens have the body intact, reliable estimates ofoverall length are obtainable only from AMNH 11759, 12209, 12212, 12213, and 12220 (in 12209 and 12212, however, overall skull length is un- available). Comparison of ethmosphenoid length with overall body length provides an approximate estimate of the probable max- imum size. The longest ethmosphenoids (AMNH 11760, 12207, 12211) are approximately 185-190 mm long, giving an extrap- olated overall length of 1.25 m (4.1 ft). Al- though it is possible that Axelrodichthys grew to even greater size, the available material suggests that it was generally smaller than Mawsonia (e.g., AMNH 12217, 12218) from the same formation. An outline restoration of the fish is given in figure 14. Although the dermal skeleton can be examined in most of the specimens, the brain-

Fig. 12. Axelrodichthys araripensis, new genus and species. Holotype, AMNH 11759, entire fish. 5cm

in (A) Fig. 13. Axelrodichthys.eara4ipensisrfiewtgenus and species. Holotype, AMNH 11759, head right; (B) left; and (C) dorsal aspects. 16 AMERICAN MUSEUM NOVITATES NO. 2866

0 0

C13

4- 0

,~'0 .~ - ,

. o~ o ;j

C3 A

E Fig. 15. Specimens referred to Axelrodichthzys araripensis; (A) AMNH 12209; (B) AMNH 12210; (C) AMNH 12211; (D) AMNH 12212; (E) AMNH 12213. Scale bars = 10 cm. 18 AMERICAN MUSEUM NOVITATES NO. 2866

5cm

Fig. 16. Small, complete specimen referred to Axelrodichthys araripensis, in part and counterpart; AMNH 12220. case can be studied only after preparation in angle is only 1100. For comparison, in Moen- acid. Most of the information below con- kopia wellesi this angle is approximately 1250 cerning the braincase was obtained from (measured from Schaeffer and Gregory, 1961, AMNH 11760, 11761, and 12215 after acid fig. 1 B); in Rhabdoderma elegans and Nesides preparation (figs. 17-21, 23-26). schmidti it is approximately 1400 (from Jar- The basisphenoid ofAxelrodichthys differs vik, 1980, fig. 218A; Forey, 1981, fig. 1); and in several respects from that of Mawsonia in Latimeria chalumnae is it approximately from Santana (e.g., AMNH 11758, 12217, 1450 (estimated from X-ray of anterior cra- 12218). In lateral profile, the angle between nium; Millot and Anthony, 1958, pl. 22). its ventral and posterior surfaces (indicated In anterior view the unfinished cancellous in fig. 18B) is approximately 1300, whereas bone of the basisphenoid in Axelrodichthys in the Santana Mawsonia basisphenoids (and forms a deep V, rather than a gull-wing shape. in the Moroccan Mawsonia sp. basisphenoid The upper part of the exposed anterior mar- figured by Wenz, 1981, fig. 4, pl. 2D-F) this gin is tilted back, forming the "floor" of the 1 986 MAISEY: COELACANTHS 19

5 cm

r, ; A

Fig. 17. Ethmosphenoid moiety of Axelrodichthys araripensis, AMNH 11760, in (A) dorsal; (B) lateral; and (C) ventral aspects. triangular gap (presumably cartilage-filled in somewhat deeper notch in Axelrodichthys life) between the skull roofand basisphenoid. (e.g., AMNH 11760) than in Mawsonia (e.g., Sphenoid condyles are well developed, as is AMNH 11758, 12218). In posterior view, the slidingjoint formed by the processus con- differences are also found in the shape ofthe nectens (seen in AMNH 11760, 11761, cranial cavity and in the sweep of the "pleu- 12215). A basipterygoid process is absent. rosphenoid" suture (compare figs. 2B, 19B). The sphenoid condyles are separated by a As in the Santana Mawsonia (e.g., AMNH 20 AMERICAN MUSEUM NOVITATES NO. 2866

tec

col col' A 5cm

- bsp dpr

par\

prcon B

C Fig. 18. Annotated drawing of ethmosphenoid moiety in AMNH 1 1760.

11758, 11217, 11218) and M. tegamensis The parasphenoid and ectethmoids are long (Wenz, 1975), the "pleurosphenoid" ofAxel- and slender (figs. 17, 18, 22B). A fine shagreen rodichthys is continuous with the posterior of teeth covers the anterior three-quarters of frontal. the parasphenoid. The dorsal process of the 1 986 MAISEY: COELACANTHS 21

pis padI ant

2cm B

Fig. 19. Annotated drawing of AMNH 1 1760; (A) anterior view of otico-occipital moiety; (B) posterior view of ethmosphenoid moiety. ectethmoid is straighter and less steeply in- rostral organ, and larger openings, fore and clined than in either M. lavocati or the San- aft of the lateral rostral, for the anterior and tana Mawsonia (AMNH 11758), with a pro- posterior nares. At the ventral margin of the nounced lateral protruberance anteriorly. The snout is a pair of slender bones, possibly the lateral margin of the tectal series is curved premaxillae, but there is no sign of teeth on outward, unlike that in M. lavocati and Maw- these elements. The lateral rostral forms a sonia from Santana (e.g., AMNH 11758), broad platform extending for some distance which curves down to flank the dorsal process anteroposteriorly, lateral to the ectethmoid laterally. dorsal process (e.g., AMNH 11759, 12214; The ethmosphenoid roofing bones com- figs. 13, 14, 22A, 26). prise a frontonasal series offive paired bones, In the otico-occipital region (figs. 19-21), flanked on either side by a supraorbitotectal the prootic process ofAMNH 11760 is large. series of seven bones (e.g., AMNH 11759, Dorsolateral to the process is a shallow myo- 12207, 12208, 12211, 12212, 12213, 12214, dome. Unlike the foramen in Mawsonia sp. 12219, 12220). The superficial ophthalmic from Santana (e.g., AMNH 11758), the fo- canal is enclosed for much of its length, al- ramen for the facial nerve and jugular canal though in the acid-prepared skull of AMNH is not separated from the myodome but in- 11760 the bone overlying the right canal has stead lies laterally within its floor (fig. 19A). been stripped off anteriorly (fig. 18). Foram- The ascending process of the prootic region ina for nerve branches supplying the supraor- is much broader than in AMNH 11758, and bital sensory canal are numerous in the snout has a correspondingly broad suture with the region, but farther posteriorly (i.e., above the parietal descending lamina. The supratem- orbit back as far as the intracranialjoint) there poral also has a low descending lamina, but are virtually no sensory foramina (e.g., this lacks a suture. AMNH 11760, 11759, 11220), suggesting that The otico-occipital region includes paired the laterosensory system of Axelrodichthys parietals, supratemporals, and postparietals was somewhat specialized, with its main in- (e.g., AMNH 11759, 11760, 12208), an ar- nervation confined to the snout. rangement known otherwise only in Maw- A complex pattern of small rostral bones sonia spp. (Wenz, 1975; see above). How- covers the snout (e.g., AMNH 12214; fig. ever, there is an additional median element, 22A). Elements corresponding to the medial located at the posterior skull margin between and lateral rostrals, postrostral, and nasals the postparietals (fig. 14, 20A, 21A). In all 1-5 ofLatimeria are identified. Between some specimens examined the median and paired ofthese are small round openings, presumed postparietals are anamestic, and the supra- to be for anterior and posterior tubes of the temporal commissure presumably would 22 AMERICAN MUSEUM NOVITATES NO. 2866

have lain farther posteriorly. An extrascapular series like that found in most other coelacanths seems to be absent even in complete specimens of Axelrodichthys. In Axelrodichthys the pterygoid posterior margin is almost vertical (figs. 23, 24), with a single strengthening ridge. In lateral view, the ossified metapterygoid is shaped much as in Mawsonia, but the antotic articular surface 1u is relatively narrower (e.g., AMNH 12215; fig. 25). Anteriorly the pterygoid is tapered into a thin splint, with a more downturned anterior extremity than in Mawsonia from Santana (fig. 14). There is no evidence of an ectopterygoid. The visceral surface of the pterygoid in AMNH 1 1760 is overlain by an extensive shagreen of fine teeth. A triangular A 2 cm autopalatine is present (AMNH 1 1759; figs. 1 3B, 26B), with an elongate, almost horizon- tal, sutural contact with the splintlike pterygoid. As in Mawsonia, the circumorbital and cheek bones do not overlap (figs. 13-16). The dermosphenotic dorsal margin follows the anterolateral margin of the parietal and the lateral margin of the first two supraorbitals. Although the dermosphenotic extends far an- teriad to the intracranial joint, as in Maw- sonia spp., it is not splintlike. Again as in Mawsonia, the infraorbital canal lies poste- -' t riorly within the dermosphenotic. The squamosal and preopercular are small. A sub- -CY opercular has not been found. The operculum is rounded, with fine striae radiating from its articular area over the entire lateral surface except at its very margin (figs. 1 3B, 16, 22A). Instead of forming an elongate bar, as in some Mawsonia spp., the suborbital moiety of the lachrymojugal is semicircular, as in ppa many other fossil coelacanths (figs. 13, 14). The eye would have been located dorsal to Ct this curved part. In Mawsonia a corresponding curved region occurs at the anterior extremity (see above and Wenz, 1975, fig. 1). In both Mawsonia and Axelrodichthys the eye > pro probably occupied a more anterior position than is usual among fossil coelacanths. In nei- is there evidence of a Fig. 20. Otico-occipital moiety of AMNH ther genus preorbital as occurs in some coel- 11760, in (A) dorsal; (B) ventral; an(d (C) right bone, primitive lateral views. Central part of skull roof is missing, acanths (e.g., Rhabdoderma; Forey, 1981). partly obscured by matrix in (B). The lachrymojugal does not extend anteriorly far enough to meet the lateral rostral process 1 986 MAISEY: COELACANTHS 23

in Axelrodichthys, although it does so in Mawsonia. The lowerjaw ofAxelrodichthys resembles that of Macropoma, except that there (according to Woodward, 1909, p. 176) the articular is apparently fused to the angular (not so according to Forey, personal commun.). In this respect Axelrodichthys is like Maw- sonia (see Mawson and Woodward, 1907, p. 136). A bilobed coronoid is present (figs. 23, 24, 26), as in Mawsonia. The postcranial skeleton is best preserved in AMNH 11759 and 12220, the two smallest specimens in the sample (figs. 12, 16). Fin- ray counts (given above in generic diagnosis) A are based mainly on these specimens, and 2 cmii seem fairly consistent where observable in others (fig. 15). Generally the tip of the tail is not preserved, but there are traces of a median caudal lobe in AMNH 12211 and 12220 (figs. 15C, 16). Basal elements and/or girdles are ossified (see fig. 14). The left shoulder girdle has been partially prepared in AMNH 11759, but it is damaged and incomplete. A supracleithrum (ano- cleithrum) has not been identified. The cleithrum is virtually unomamented, but the clavicles are ornamented by ridges ventrally. These elements appear to meet ventrally. Only a small fragment of the extracleithrum is B present. The scapulocoracoid has not been observed. In none of the specimens is it possible to see whether the pectoral fin was lobed or else pedunculate, as in Latimeria (cf. Berg, 14 1940; Forey, 1981). From the positions offin rays it is fairly certain that the pelvic, anal, and second dorsal fins were lobed as in La- timeria. The swim bladder is well ossified, as in Macropoma and many other fossil coelacanths, and is not usually compressed, unlike the remainder ofthe postcranial skeleton. The swim bladder evidently resisted sedi- C Wi mentary compaction, which suggests that it was filled with fatty tissue in life (further sug- Fig. 21. Otico-occipital moiety of AMNH gested by the lack of sedimentary filling; the 1 1760, in (A) dorsal; (B) ventral; and (C) left lat- swim bladder is usually preserved hollow but eral views. with a geodelike lining of calcite). In Axel- rodichthys the swim bladder extends posteriorly past the level of the second dorsal fin, unlike those in Undina and Macropoma (Woodward, 1891, fig. 53; 1909, fig. 49). Fur- 24 AMERICAN MUSEUM NOVITATES NO. 2866

5cm

_-r-.1t<

A:!-?--f*;?,, "7';A

B <,< 1 986 MAISEY: COELACANTHS 25

5 cm i -i A B

C D Fig. 23. Axelrodichthys araripensis, AMNH 11760; (A, B) right pterygoid, metapterygoid, and quadrate in (A), lateral and (B), mesial views; (C, D) incomplete angular and articular in (C), lateral and (D), mesial views. thermore, in Axelrodichthys the swim blad- lower part of the Romualdo Member of the der seems to consist of two or perhaps even Santana Formation. One of these taxa is re- three chambers, separated by constrictions ferable to Mawsonia and may even represent (e.g., AMNH 11759, 12210, 12211, 12212, the type species, M. gigas. Mawsonia is rel- 12213, 12219, 12220). In most ofthese spec- atively rarer than the other taxon, being rep- imens(AMNH 11759, 12210, 12211, 12212? resented in the AMNH collection only by a and 12213) the anterior chamber lies some- partial skull and cranial fragments. The other what lower in the body cavity than the re- form (Axelrodichthys, n. gen.) differs from mainder (see outline, fig. 14). Mawsonia in several aspects of its cranial anatomy, including the proportions and gen- DISCUSSION: THE RELATIONSHIP eral shape of the skull roof, the cheek bone OF MAWSONIA AND arrangement, the presence of a median element in the posterior margin of the skull ta- AXELRODICHTHYS ble, basisphenoid morphology, and associ- As reported by Campos and Wenz (1982), ated differences in metapterygoid morphology there are two distinct coelacanth taxa in the and its articulation with the skull. This new

Fig. 22. Axelrodichthys araripensis, flattened head, AMNH 12214; (A) skull roof with rostral bones in place, plus right operculum, cheek bones, and coronoids; (B) parasphenoid, right pterygoid, prootics, and ?ceratohyals. 26 AMERICAN MUSEUM NOVITATES NO. 2866

14-IL thl-antart

-1\ mpt

> pt

qu ,,

5cm A>< p.tI postcor ang

C D Fig. 24. Annotated drawings ofelements shown in figure 23; (A, B) pterygoid (lacking anterior ramus), metapterygoid, and quadrate; (C, D) articular and angular.

form is considered sufficiently distinct to merit generic separation from other coelacanths. Axelrodichthys resembles Mawsonia in possessing three rather than two pairs of dermal bones in the posterior skull roof, in the anterior extent of the dermosphenotic, and in the very shallow anterior limb of the pterygoid. Furthermore, in both taxa the ectopterygoid may have been unossified. Maw- sonia (and Axelrodichthys) resembles Macropoma in its rugose dermal bone ornamentation, and on this basis it has long been suggested that Macropoma and Maw- sonia are closely allied (Mawson and Wood- Fig. 25. Axelrodichthys, AMNH 12215, in- ward, 1907, p. 134; Woodward, 1909, p. 176). complete skull with visceral arches, showing me- In a cladistic analysis of coelacanth phy- tapterygoid (projecting, top) attached to pterygoid, logeny, Forey (1981, fig. 14) proposed that and right ceratohyal beneath quadrate. Macropoma and Latimeria are sister taxa, 1 986 MAISEY: COELACANTHS 27

ppa

op la

Ii,

/ _ 5 < ~~~~~~~~~~~~~~den

A~~~

s 0~~~~~an sp

Fig. 26. Outline restorations of the head in Axelrodichthys, based mainly on AMNH 11759, 11760, and 12215. (A) with cheek bones, operculum, and lower jaw in place; (B) cranium and upper jaw. 28 AMERICAN MUSEUM NOVITATES NO. 2866 united by possession of a prootic ascending discovery of Axelrodichthys, a genus sharing process, seven extrascapulars (subsequently the same pattern of dermal bone ornamen- omitted as a character; Forey, 1984, fig. 2), tation as Mawsonia, means that records of and anterior branches developed from the the latter genus require careful reinvestiga- supratemporal commissure. Mawsonia was tion in order to determine whether some excluded from that analysis, but may now be should be referred to Axelrodichthys. Distin- included (along with Axelrodichthys). guishing criteria have been noted in the Investigation of the present material from dermosphenotic, lachrymojugal and parietal- Brazil generally corroborates Forey's (1981) postparietal complex, but basisphenoid and hypothesis. In particular, there is an ascend- metapterygoid morphology may also be di- ing process on the prootic in both Mawsonia agnostic. At present, Axelrodichthys is known and Axelrodichthys. However, the descend- only from the Santana Formation of Brazil, ing process ofthe supratemporal (which meets where it accompanies Mawsonia. The close the hyomandibular articulation in many similarity of Mawsonia from Ceara and Ba- coelacanths) is no more than a weak ridge in hia suggests that the Lower part of the Ro- Axelrodichthys (AMNH 11760, 11761, mualdo Member (Santana Formation) may 12215) and is absent in Mawsonia (AMNH be of Neocomian rather than Aptian age. 11758). According to Forey's (1981) char- acterization, this is probably a homoplasious ACKNOWLEDGMENTS condition in these taxa, since a supratemporal descending process is supposedly ab- I am indebted to Dr. Herbert R. Axelrod sent in the most primitive coelacanths, such (T.F.H. Publications, Inc.) for making his im- as Diplocercides. portant collection of Santana vertebrate fos- Neither Mawsonia nor Axelrodichthys pos- sils available for study, and for his outstand- sess a series of seven extrascapulars or an- ing generosity in donating this collection to terior branches from the transverse commis- the AMNH. sure. Thus of the three characters which I also thank the following for their help in (according to Forey, 1981) unite Macropoma completing this description: Drs. Peter For- and Latimeria, only one (the ascending pro- ey, Richard Lund, Bobb Schaeffer, and Sylvie cess of the prootic) is present in Mawsonia Wenz, for useful and informative discus- and Axelrodichthys. I conclude that Maw- sions; Ms. Ellen Garvens and Mr. Peter Gold- sonia and Axelrodichthys together form a sis- berg, for compiling the illustrations; Ms. Ale- ter group to Macropoma and Latimeria. jandra Lora, for typing the various versions These taxa also primitively share with Hol- of this manuscript. The paper was reviewed ophagus the presence of dorsal laminae on by Dr. Peter Forey (BMNH), and by Profs. the parasphenoid and the ventral position of Donald Baird (Princeton) and Keith S. the infraorbital canal within the squamosal. Thomson (Yale), and was edited by Ms. Adult Axelrodichthys and Mawsonia at- Brenda Jones. tained large dimensions, comparable with Finally, a special word of thanks to Mr. modem Latimeria. Most other Paleozoic and Walter Sorensen, who prepared the speci- Mesozoic coelacanths were considerably mens for this investigation. By the time this smaller, although the Triassic genus Moen- article appears, Walter will have retired from kopia, known only by the basisphenoid, must a career spanning 36 years as a preparator at also have attained a similar size to Latimeria the American Museum of Natural History. (Schaeffer and Gregory, 1961). In the Santana Coelacanth buffs are doubtless aware ofWal- ichthyofauna, only some Cladocyclus, Note- ter's long-standing contributions, particular- lops, and Enneles specimens rival the coel- ly his discovery ofthe first specimen of Chin- acanths in size (Silva Santos and Valenca, lea sorenseni from the Chinle and Dockum 1968). Formations (Schaeffer, 1967). It therefore The distribution of Mawsonia is thought seems appropriate that Walter's career should to be of paleobiogeographical significance culminate in the investigation of this new (Patterson, 1975; Wenz, 1980). However the coelacanth discovery. 1986 MAISEY: COELACANTHS 29

ADDENDUM ner in the Province ofCeara, in the north of Brazil. Edinburgh New Phil. Jour., The postcranial skeleton of Mawsonia is vol. 30, pp. 82-84. virtually unknown. According to Forey (per- 1844. Sur quelques poissons fossiles du Bresil. sonal commun.), the holotype of M. minor Comptes Rendus Acad. Sci. Paris, vol. has closely segmented lepidotrichia in the fins, 18, pp. 1007-1015. as in some Jurassic coelacanths, but the fin Bate, R. H. rays of the first dorsal and caudal fins lack 1972. Phosphatized ostracods with append- denticles. These are present in many Jurassic ages from the lower Cretaceous of Bra- and all other hitherto described Cretaceous zil. Palaeontology, vol. 15, no. 3, pp. in 379A-393. coelacanths, and also occur adult Lati- Berg, L. S. meria. Evidence from baby Latimeria sug- 1940. Classification offishes, both Recent and gests that these denticles appear late in on- Fossil. Trudy Zool. Inst. Leningrad, vol. togeny. Denticles are present on the fins of 5, pp. 87-517. the "Forme B" coelacanth described by Cam- Campos, D. A., and S. Wenz pos and Wenz (1982). 1982. Premiere decouverte de Coelacanthes The fin rays of all the AMNH Axelrod- dans le Cretace inf'erieur de la Chapada ichthys specimen are closely segmented dis- do Araripe (Bresil). C. R. Acad. Sci., II, tally, although this segmentation becomes in- vol. 294, pp. 1151-1154. distinct proximally. In AMNH 11759 the Casier, E. three or four of D1 and 1961. Materiaux pour la faune ichthyologique posteriormost rays eocretacique du Congo. Ann. Mus. Roy. the ventralmost rays of the anal fin all bear Afr. Cent., Tervuren, Sci. Geol., vol. 62, a few sparse denticles. In the small specimen, xii + 96 pp. AMNH 12220, the last three rays of D1 and Forey, P. L. the third or fourth hypocaudal fin ray also 1981. The Coelacanth Rhabdoderma in the bear a few denticles. Additionally, one or two Carboniferous of the British Isles. Pa- denticles have been found on the fringe of a leontology, vol. 24, no. 1, pp. 203-229. large uncataloged caudal fin associated with 1984. The coelacanth as a living fossil. In N. an Axelrodichthys head. I have been unable Eldredge and S. M. Stanley (eds.), Liv- to find denticles on the fins ofany ofthe other ing fossils. New York, Springer, pp. 166- specimens listed earlier. 169. Jarvik, E. I conclude from this that Axelrodichthys 1980. Basic structure and evolution of verte- possessed few denticles on its first dorsal or brates, vol. 1. New York, Academic caudal fins, even in large individuals (in fact Press, 575 pp. more denticles are present on two of the Lehman, J.-P. smallest specimens, which seems to rule out 1952. Etude complementaire des poissons de ontogenetic variation). The apparent absence l'Eotrias de Madagascar. K. svenska of denticles in Mawsonia minor may conse- VetenskAkad. Handl., vol. 4, no. 2, pp. quently be of phylogenetic importance, per- 1-201. haps providing a further synapomorphy with Lima, M. R. De 1978. Palinologia da Formacao Santana (Cre- Axelrodichthys. t'aceo do Nordeste do Brasil). Introdu- It should be noted that in figures 2B and cao geologica e desceicao sistematica dos 1 9B the opening labeled "prof' probably esporos da Subturma Azonotriles. housed the superficial ophthalmic branches Ameghiniana, vol. 15, pp. 333-365. of the facial and trigeminal nerve, and that Mabesoone, J. M., and I. M. Tinoco the canal designated "col" is for the supraor- 1973. Palaeoecology of the Aptian Santana bital sensory canal. I thank Dr. Peter Forey Formation (Northeastern Brazil). Pa- for pointing this out to me. laeogeogr., Palaeoclimat., Palaeoecol., vol. 14, pp. 97-118. Mawson, J., and A. S. Woodward LITERATURE CITED 1907. Cretaceous formation of Bahia and its Agassiz, L. vertebrate fossils. Quart. Jour. Geol. 1841. On the fossil fishes found by Mr. Gard- Soc., vol. 63, pp. 128-139. 30 AMERICAN MUSEUM NOVITATES NO. 2866

Millot, J., and J. Anthony Spix, J. B., and C. F. P. Martius 1958. Anatomie de Latimeria chalumnae 1. 1828. Reise in Brasilien, vol. 2. Munich, pp. Squelette, muscles et formations de sou- 770-779. tien. Paris, Cent. Natl. Rech. Sci., 118 Toombs, H. A., and A. E. Rixon pp- 1950. The use ofplastics in the transfer meth- Patterson, C. od of preparing fossils. Mus. Jour., vol. 1975. The distribution ofMesozoic freshwater 50, pp. 105-107. fishes. Mem. Mus. Nat. Hist. Natl., ser. Wenz, S. A, Zool., vol. 88, pp. 156-174. 1975. Un nouveau Coelacanthide du Cretace Rixon, A. E. inf'erieur du Niger. Remarque sur la fu- 1976. Fossil animal remains. London, Ath- sion des os dermiques. Colloq. Intn. lone Press, vi + 304 pp. Cent. Natl. Rech. Sci., Paris, no. 218, Schaeffer, B. pp. 175-190. 1952. The Triassic coelacanth fish Diplurus 1980. A propos du genre Mawsonia, Coel- with observations on the evolution of acanthe geant du Cretace inferieur d'Af- the Coelacanthini. Bull. Amer. Mus. rique et du Bresil. Mem. Soc. Geol. Fr., Nat. Hist., vol. 99, art. 2, pp. 25-78. n. s., no. 139, pp. 187-190. 1967. Late Triassic fishes from the western 1981. Un Coelacanthe geant, Mawsonia lav- United States. Bull. Amer. Mus. Nat. ocati Tabaste, de l'Albien-Base du Hist., vol. 135, art. 6, pp. 287-432. Cenomanien du Sud Marocain. Ann. Schaeffer, B., and J. T. Gregory Paleont. (Vertebres), vol. 67, no. 1, pp. 1961. Coelacanth fishes from the continental 1-20. Triassic of the western United States. Woodward, A. S. Amer. Mus. Novitates, no. 2036, pp. 1- 1891. Catalogue ofthe fossil fishes in the Brit- 18. ish Museum (Natural History), pt. II. Silva Santos, R. Da London, Brit. Mus. (Nat. Hist.), xliv + 1981. Sobre a occurencia de uma faunula de 567 pp. peixes na formacao Riachuelo, estado 1909. The fossil fishes of the English Chalk, de Sergipe. An. Acad. Brasil Cienc., vol. pt. V. Monogr. Palaeont. Soc. London, 53, no. 1, p. 203 (summary). pp. 153-184. Silva Santos, R. Da, and J. G. Valenca 1968. A Formacao Santana e sua Paleoictio- fauna. An. Acad. Brasil. Cienc., vol. 40, no. 3, (3), pp. 339-360.

Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates, Bulletin, and Anthropological Papers published during the last five years are available free of charge. Address orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, New York 10024.