AMERICAN MUSEUM Novitatesx T PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 3132, 20 pp., 7 figures May 24, 1995

First Records of Fossil Sergestid Decapods and Fossil Brachyuran Crab Larvae (Arthropoda, Crustacea), with Remarks on Some Supposed Palaemonid Fossils, from the Santana Formation (Aptian-Albian, NE Brazil)

JOHN G. MAISEY1 AND MARIA DA GLORIA P. DE CARVALHO2

ABSTRACT A new genus and species of shrimp is described Araripe Basin. It also indicates that Tharrhias was from limestone concretions of the Romualdo a plankton feeder, and probably did not compete Member. These shrimps occur within the matrix with predators such as Rhacolepis, which preferred of concretions and in the stomach contents of the to eat benthic decapods and small pelagic fishes. teleost fish Rhacolepis. Based on rostral and car­ The fact that Rhacolepis and Tharrhias rarely oc­ apace characters, it is suggested that these deca­ cur in the same horizon suggests that their pre­ pods are representatives of the family Sergestidae. ferred food sources usually did not occur together. Other fossil decapods recovered from the stomach New morphological information is also provided contents of a different fish {Tharrhias) are iden­ for the supposed palaemonid Beurlenia from tified as "protozoea" larvae of a brachyuran crab. freshwater strata of the Crato Member. Although This new discovery suggests that some brachyu- this taxon is a caridean decapod, its palaemonid rans completed part of their life cycle within a affinity is questionable. restricted marine or brackish environment in the

1 Curator, Department of Vertebrate Paleontology, American Museum of Natural History. 2 Professora, Instituto de Geocidncias, Departamento de Geologia, Universidade Federal do Rio de Janeiro, Brazil, CEP 21.910.000.

Copyright © American Museum of Natural History 1995 ISSN 0003-0082 / Price $2.10 2 AMERICAN MUSEUM NOVITATES NO. 3132

RESUMO E descrito um novo genero e especie de camarao salobro, na Bacia do Araripe. Isto indica tambem encontrado em concregoes calcarias provenientes que Tharrhias possuia habito alimentar plancto- do Membro Romualdo. Estes ocorrem tambem nico e provalmente nao competia com Rhacolepis, no conteudo estomacal de peixes teleosteos - Rha- um predador que preferia comer decapodos ben- colepis. Com base nos caracteres do rostro e da tonicos e pequenos peixes pelagicos. O fato de carapa^a sugerimos que estes decapodos sejam re- Rhacolepis e Tharrhias raramente ocorrem no presentantes da Familia Sergestidae. Outros de­ mesmo horizonte sugere que suas fontes de ali- capodos fosseis foram obtidos do conteudo esto­ mentos preferidos normalmente nao ocorriam macal de diferente genero de peixe - Tharrhias, juntas. Sao dadas novas informa9oes morfologicas identificados como larva de caranguejo Braquiuro, sobre Beurlenia, um suposto palaemonideo ocor- no estagio "protozoea." Esta nova descoberta su- rente nos depositos de agua doce do Membro Cra- gere que braquiuros poderiam completar seu ciclo to. Embora este taxon seja um decapodo carideo, vital dentro de um ambiente marinho restrito ou sua afinidade com palaemonideo e questionavel.

INTRODUCTION The Santana Formation of northeast Brazil (Beurlen, 1963), but were not described or is famous for its abundant and well-preserved named. Fragmentary decapod remains have Lower Cretaceous fossil assemblages includ­ also been noted in stomach contents of fossil ing vertebrates, insects, and plants (for more fishes from the Romualdo Member (Maisey, details of the fauna and flora, see Maisey, 1991,1994; Wilby and Martill, 1992;Kellner 1991, 1993). It is particularly well known for et al., 1994). Until now, however, no attempt fossil fishes, especially from limestone con­ has been made to provide a morphological cretions in its upper (Romualdo) Member. description of this material. From the Crato These fishes are abundant, well preserved, Member, much lower in the Santana For­ and represent a wide range of taxa including mation, Martins-Neto and Mezzalira (1991) chondrichthyans, coelacanths, and many ac- described a complete decapod which they tinopterygians. named Beurlenia araripensis and identified The most informative method of studying as a palaemonid. Little has been published these fishes is to dissolve the limestone ma­ about Santana Formation decapods, reflect­ trix with dilute formic or acetic acid, usually ing either their rarity (in the case of Beurlenia after embedding the fossil in clear epoxy. This from the Crato Member, for example) or the preparation technique reveals the finest de­ difficulty in finding specimens that are mor­ tails of skeletal anatomy, and may even pro­ phologically informative (as with the Ro­ vide data on associated soft tissues (e.g., mus­ mualdo Member shrimps). cle, gill lamellae). In some predaceous fishes, Here we provide new data about fossil dec­ the stomach contents are also revealed, pro­ apods from both the Crato and Romualdo viding an opportunity to study predator-prey members of the Santana Formation. We de­ relationships and reconstruct parts of the tro­ scribe decapod crustaceans from the Ro­ phic network (Maisey, 1994). Phosphatic prey mualdo Member, using specimens obtained items (e.g., bones, crustacean carapaces) are during acid preparation of fossil fishes at the resistant to the acid, but calcareous items (e.g., American Museum of Natural History. These molluscan shells) do not survive the prepa­ specimens were recovered both from the ma­ ration method. Besides small fishes and iso­ trix surrounding some fish fossils, and from lated bones, many small crustaceans have the stomach contents of others. Although been discovered in the stomachs of acid-pre­ fragmentary, this material provides an op­ pared fish fossils from the Santana Formation portunity to make some morphological ob­ (% 1). servations, identify the decapods, and pres­ Decapods have been reported at two dis­ ent some ecological interpretations. New tinct levels in the Santana Formation (see specimens of the supposed palaemonid Beur­ next section). In the Romualdo Member, lenia from the Crato Member also are figured shrimps were found in limestone concretions and described (figs. 6, 7), supplementing the 1995 MAISEY AND DE CARVALHO: BRACHYURAN CRAB LARVAE 3

1 cm

Fig. 1. (A) Specimen of the teleost fish Rhacolepis buccalis, AMNH DVP 19380, revealing the stomach (parts of which are preserved) completely full of shrimps which are referred here to the new genus Paleomattea. One individual with appendages intact (illustrated in fig. 4Q lies within the pylorus and was about to enter the intestine. (B) Detail of specimen showing compacted mass of shrimps in the stomach. data provided by Martins-Neto and Mezza- ern part of the State of Ceara (fig. 2). The lira(1991). topography is dominated by the Chapada (plateau), which consists mainly of fluvial STRATIGRAPHIC FRAMEWORK sandstones belonging to the Exu Formation The Chapada do Araripe lies in the interior (Middle Cretaceous). Below these sand­ of northeast Brazil, mostly within the south­ stones, the Santana Formation crops out at 4 AMERICAN MUSEUM NOVITATES NO. 3132

in the Santana Formation. Bituminous shales near the base of the Romualdo Member are of Middle Albian age (Pons et al., 1990). The concretions are from higher in the succession, and their precise age has yet to be deter­ mined. It is unlikely that they are appreciably younger than the bituminous shales, because similar fossil fishes occur in both. Through­ out the Romualdo Member there is evidence from ostracods and dinoflagellates of pulsa­ tory marine ingressions (Arai and Coimbra, 1990). Well-preserved decapod fossils de­ scribed here are associated with fossil fishes in concretions from the Romualdo Member.

MATERIALS AND METHODS Fig. 2. Outline map to show the location of the Chapada do Araripe. All decapod remains from the Romualdo Member were obtained following acid prep­ aration of limestone concretions containing various sites around the foot of the plateau. fossil fishes. In some cases it was possible to The most famous and productive fossil lo­ remove decapods from the matrix, especially calities lie along its northern border, but fos­ where they were originally separate from the sils are found in many other places around fish. Specimens removed from matrix were the Chapada. mounted on standard glass microscope slides. The Santana Formation is customarily di­ Where decapods formed part of the stomach vided into three members. The lowest (Crato) contents in a fish, however, they were usually Member is generally regarded as a lacustrine left in place to avoid damage or destruction sequence, representing a quiescent phase of during removal. Also, the original trophic re­ basin infilling after a rifting episode (Maisey, lationship was considered of sufficient value 1990). Palynological data (Pons et al., 1990) to be retained. suggest a Middle to Late Aptian age for the The decapod specimens described here Crato Member. Laminated limestones in have been cataloged and deposited in the De­ these strata have yielded a rich fauna, in­ partment of Invertebrates at the American cluding several hundred species of insects, as Museum of Natural History. Reference is also well as fossil spiders, scorpions, fishes, frogs, made to fossil fish specimens from which feathers, and terrestrial plants (Grimaldi and some of the decapods were recovered. These Maisey, 1990; Maisey, 1993). It is from the fishes were originally cataloged by the De­ Crato Member that specimens of Beurlenia partment of Vertebrate Paleontology at the have been found. same institution; to distinguish between de­ The second unit is the Ipubi Member, which partmental collections, the original catalog consists mainly of evaporites (gypsum) with numbers of the fossil fishes are here prefixed some intercalated shales. These shales have by DVP. Most of the fishes containing deca­ yielded conchostracans, small gastropods, and pod prey have been deposited in the De­ poorly preserved fishes, but no decapods. partment of Invertebrates, except for AMNH The upper part of the Santana Formation DVP 19380, which reveals important fea­ is formed by the Romualdo Member. This tures of stomach anatomy in Rhacolepis (fig. consists of a heterogeneous sequence of bi­ 1). tuminous shales, marls, sandstones, and car­ New specimens of Beurlenia (figs. 6,7) from bonate sediments. Layers of calcareous con­ the Crato Member were prepared mechani­ cretions occur within oil shales in the middle cally. The fragile condition of their exoskel- part of the Romualdo Member. These con­ eton precluded more than local preparation cretions are the best known source of fossils with a needle. 1995 MAISEY AND DE CARVALHO: BRACHYURAN CRAB LARVAE 5

ABBREVIATIONS ORDER DECAPODA LATREILLE, 1803 Institutional SUBORDER DENDROBRANCHIATA BATE, 1888 AMNH American Museum of Natural History DVP Department of Vertebrate Paleontology FAMILY SERGESTIDAE DANA, 1852 Morphological Paleomattea, new genus aped antennal peduncle DIAGNOSIS: Small sergestid with mostly asc antennal scaphocerite smooth carapace and abdominal segments, ds dorsal spine lacking keels and grooves; carapace with short hs hepatic spine rostrum bearing three small spines; small su­ per pereiopod pie pleopod praorbital and hepatic spines present on car­ rs rostral spine apace; length of sixth abdominal segment al­ ss supraorbital spine most three times its depth; none of pereio- pods enlarged. A NEW GENUS OF SERGESTIDLIKE ETYMOLOGY: Greek, palaios, ancient; Lat­ DECAPOD FROM THE in, mattea, a delicacy. ROMUALDO MEMBER Paleomattea deliciosa, new species Small shrimplike decapods were first noted in Romualdo Member concretions by Beur- DIAGNOSIS: AS for species. len (1963). Some indeterminable fragments ETYMOLOGY: Latin, deliciosa, delicious were subsequently described by Martins-Neto (self-evident in Rhacolepis). and Mezzalira (1991). Other fragments were HOLOTYPE: AMNH 44985, Department of mentioned by Maisey (1991) in matrix as­ Invertebrates: carapace and abdominal seg­ sociated with fossil fishes and from stomach ments from stomach contents of Rhacolepis contents (fig. 1). This occurrence has proven buccalis AMNH DVP 13696, recovered fol­ to be an excellent source of fragile decapod lowing acid preparation; Lower Cretaceous, fossils, especially following acid preparation Albian, Romualdo Member, Santana For­ of the concretions (Maisey, 1991, 1994; Wil- mation, Chapada do Araripe, Ceara, Brazil by and Martill, 1992; Kellner et al., 1994). (fig. 3B). Until now, however, no attempt has been DESCRIPTION: The carapace and abdominal made to describe their morphology. Martins- pleurae are laterally compressed and smooth Neto and Mezzalira (1991) regarded them as and, apart from a small hepatic spine on the indeterminate carideans, but concluded that lateral surface of the carapace, there is little they did not belong to the family Palaemon- ornamentation; keels and grooves are absent. idae. The material described below is cer­ The carapace is almost three times longer than tainly distinct from Beurlenia, and in our view its maximum depth. Its dorsal margin is al­ does not represent any caridean family. most straight, except anteriorly where there No complete specimens of these shrimps is a short rostrum which is slightly upturned. have been recovered. Among the material In some specimens a small supraorbital spine available, however, several partially articu­ was observed near the anterior margin of the lated and incomplete individuals have been carapace lateral to the rostrum and just be­ studied (fig. 3). These were collected after acid hind the eye (fig. 3B, C, F). preparation of fishes belonging to the tele- The rostrum of Paleomattea is very differ­ ostean genus Rhacolepis (fig. 5A). One ad­ ent from that in Beurlenia from the Crato vantage of the acid-prepared decapod ma­ Member, consisting of little more than a raised terial is its preservation in three dimensions, keel with three or four small spines dorsally, unlike the compressed fossils from the Crato and hardly projecting in front of the carapace Member. Despite the small size of our sam­ (fig. 3B, C, F). A large, spiny rostrum is pres­ ple, we can determine some morphological ent in many living and fossil penaeid and features with confidence. At the present time, candid decapods, where it probably repre­ we do not have reason to believe more than sents a primitive character. A small rostrum one taxon is represented in this sample. like that of the fossils described here has a AMERICAN MUSEUM NOVITATES NO. 3132

Fig. 3. Fossil and Recent sergestids. Specimen numbers prefixed by DVP refer to the Department of Vertebrate Paleontology (fossil fishes); other numbers refer to Department of Invertebrates. (A-F), examples of the sergestid Paleomattea deliciosa, new genus and species from the Romualdo Member of the Santana Formation. Scale bar = 1 mm. (A) The largest example, AMNH 44988, recovered from matrix of a concretion during acid preparation; (B) the holotype, AMNH 44985, from stomach contents of Rhacolepis buccalis, AMNH DVP 13914; (C) specimen with parts of appendages intact, AMNH 44989, from stomach contents of Rhacolepis buccalis, AMNH DVP 19380 (illustrated in fig. 1); (D, E) two abdomens (AMNH 44986 and 44987 respectively) associated with the holotype of Paleomattea 1995 MAISEY AND DE CARVALHO: BRACHYURAN CRAB LARVAE 7 more limited occurrence, for example among first is regarded by Burkenroad (1981) as a Recent genera included in the dendrobran- derived character of all dendrobranchiates. chiate family Sergestidae (e.g., Acetes, Serg- Very little can be said about the append­ estes). There is a particularly close similarity ages of Paleomattea from the material avail­ with the extant Acetes erythraeus from South able. One specimen (in the stomach of Rha­ Africa (Kensley, 1972: fig. 10F; see fig. 3G colepis buccalis, AMNH DVP 19380) has here). The eye stalks have not been observed proximal parts of the antennules, antennae, in any of our material, and were presumably and pereiopods (fig. 3C), and in another ex­ very fragile, but probably were longer than ample (recovered from a concretion after acid the rostrum. That character would also sup­ preparation) the third, fourth, and fifth ple­ port inclusion in the family Sergestidae and opods are preserved (fig. 3A). The antennal therefore ought to be verified. scaphocerite (asc, fig. 3C) is bladelike, ap­ The first five abdominal pleurae in Paleo- proximately four times longer than its max­ mattea are of approximately equal size. Their imum width, and is expanded distally. Both depth is slightly greater than their length, al­ the antennules and antennae are supported though the third pleura is deeper than the by broad, cylindrical peduncles. others. The sixth is more elongated, with a Coxae and proximal parts of the pereio­ length approximately three times its depth. pods are known, but their distal extremities All the pleurae have a slight overlap with the have not been observed and it is presently one behind; there is no "reversed" overlap unknown which of them were chelate (in Re­ of the first by the second pleura. Midlateral cent sergestids the first three pereiopods are locking hinge joints are exposed between chelate). From the relatively equal sizes of pleurae 4 and 5, and 5 and 6, but are not the ischia it is possible to conclude that none observed farther anteriorly. The articulated of the pereiopods was enlarged relative to the nature of the Romualdo Member material others, although such an extrapolation from suggests that concealed hinge joints were ischium size may be unreliable. Fourth and present between the anterior pleurae. fifth pereiopods are present (these are re­ According to Burkenroad (1981), exposed duced in Recent sergestids). The pleopods hinges between all abdominal pleurae prob­ (pie, fig. 3 A) of abdominal segments 3-5 are ably represents a primitive pattern among laterally flattened and leaf-shaped, with very decapods which became modified in certain abbreviated distal segments; they appear to lineages. There is no well-developed hinge have maintained their swimming function. joint between the third and fourth segments Pleopods of the more anterior abdominal in stenopodids, eukyphids (including cari- segments have not been observed and it is deans), and dendrobranchiates (penaeids and unknown whether any were modified toward sergestids). The anterior hinge joints usually a reproductive function. Pleopods of the sixth are exposed in members of these groups, but abdominal segment form a characteristic tail in the Upper Cretaceous penaeid Sicyonia fan along with the telson. from Germany the anterior two are hidden The absence of "reversed" overlap be­ under the pleurae (Marck, 1863). Absence of tween the firstan d second abdominal pleurae a hinge joint from all three anterior abdom­ in Paleomattea rules out affinity with palae- inal pleurae in stenopodids probably repre­ monids or a more general caridean relation­ sents an autapomorphy of that group. Over­ ship. Overlap of the second abdominal pleura lap of the second abdominal segment by the by the first is (according to Burkenroad, 1981)

deliciosa, from the stomach contents of Rhacolepis buccalis, AMNH DVP 13914; (F) anterior detail of carapace in stomach contents of Rhacolepis buccalis, AMNH DVP 19380, showing short rostrum, supraorbital spine, and hepatic spine (anterior margin restored); (G) carapace of Recent Acetes erythraeus from South Africa, showing great similarity with Paleomattea deliciosa in rostral morphology and spine arrangement (G after Kensley). Note that the eye stalks are longer than the rostrum (not known in Paleomattea). 8 AMERICAN MUSEUM NOVITATES NO. 3132

a dendrobranchiate synapomorphy. Con­ a smooth carapace and elongate sixth abdom­ cealed hinge joints between anterior abdom­ inal pleura. We suspect, however, that pen­ inal pleurae apparently represent a shared de­ aeids are not monophyletic when all the fos­ rived character with some penaeids such as sils are included. It is even possible that the Sicyonia. A different derived condition in accepted division of dendrobranchiates into which the anterior pleurae lack a hinge joint penaeids and sergestids is artificial. (seen in stenopodids, a different suborder) is No modern dendrobranchiates occur in not present in Paleomattea. fresh water, although some are found in We conclude that Paleomattea resembles brackish environments. Some Triassic pen­ sergestid dendrobranchiates in several fea­ aeids are known from brackish or lagoonal tures. Unfortunately, many of these similar­ facies, but no brackish-water decapods are ities are probably primitive (dendrobran­ known from the Jurassic (Glaessner, 1969: chiate) characters, but we are more confident 426). that rostral morphology of the fossils repre­ sents an apomorphic character shared with FOSSIL BRACHYURAN PROTOZOEA some modern sergestids. Among Recent ser­ LARVAE gestid taxa, rostral morphology in Acetes is close to that seen in the fossils, with only It is well known that the majority of deca­ three dorsal spines on the abbreviated ros­ pod crustaceans undergo a series of meta­ trum (fig. 3G). In other Recent sergestids the morphoses during ontogeny, passing through rostrum lacks a spine or else has only one at distinctive larval stages (e.g., "nauplius, cy- its tip (e.g., Sergestes, Sergio). The rostrum pris, protozoea, zoea, mysis, megalops," etc.; in most penaeids is much longer and has many Gurney, 1942). Some exceptions are noted, dorsal spines, but even in penaeids where the particularly among some freshwater brach- rostrum is short, several dorsal spines are yurans and astacidean crayfish (which have present. direct development), and also among certain Additional similarities between Paleomat­ shrimps and crabs that have various kinds of tea and Acetes include the smooth carapace abbreviated development different from the lacking keels or grooves, and the presence of norm usually seen in a taxon (Gore, 1985). only the supraorbital and hepatic spine. One Terminology for various larval stages within important difference from Acetes is the pres­ the older literature is cumbersome, and has ence of a full set of pereiopods in the fossils been simplified by more recent workers in (the last two pairs are absent in Acetes; they the field of decapod development (e.g., Wil­ are present but reduced in other sergestids). liamson, 1969). The larval term "protozoea" This suggests that Paleomattea represents an used here is also sometimes referred to as extinct sergestid related to Acetes, which has "Stage 1 zoea." plesiomorphically retained the posterior per­ Two larval decapod carapaces were dis­ eiopods. It would be premature to suggest covered among the stomach contents of a that Acetes and Paleomattea are sister taxa small (approx. 170 mm standard length) on the basis of available data. specimen of the gonorynchiform fish Thar- If we have correctly identified these deca­ rhias araripis, AMNH DVP 13680 (fig. 5B). pods as sergestids, then they are the first to One of these larval carapaces is almost com­ be recognized in the fossil record. The oldest plete and is shown in figure 4. The other spec­ dendrobranchiates are of Permo-Triassic age. imen is slightly damaged. They were first no­ These are customarily included in the Pe- ticed because of their relatively large com­ naeidae although the relationships of these pound eyes, which are well preserved, and early forms is not well established (Glaessner, were referred to elsewhere as a "large-eyed 1969). If penaeids and sergestids are sister decapod crustacean" (Maisey, 1994). The groups, they should have an equally long an­ preserved length of the carapaces is slightly cestry, and the presence of Lower Cretaceous less than 2.5 mm including the rostral and sergestids would not be unexpected. Bombur dorsal spines. If the spines are excluded, the (from the Triassic and Jurassic of Europe) carapaces are only a little more than 1 mm resembles modern sergestids in its short ros­ in length. No appendages are attached, al­ trum, and is similar to Paleomattea in having though fragments are littered around in the 1995 MAISEY AND DE CARVALHO: BRACHYURAN CRAB LARVAE

Fig. 4. Carapace of brachyuran "protozoea" larva, AMNH 44990, from stomach contents of Thar- rhias araripis, AMNH DVP 13680: (A) lateral and slightly dorsal view; (B) posterolateral view; (C) dorsal and slightly oblique view. Scale bar = 1 mm.

stomach region of the fish from which they cervical, postcervical, and branchiocardiac were recovered. grooves in adult decapods. The eyes are bul­ The most distinctive feature that led to their bous and have a diameter close to 0.5 mm. identification as brachyuran larvae is the dor­ Their compound lenses are clearly preserved. sal spine (ds, fig. 4). This feature is charac­ The fact that the eyes remained attached to teristically absent in adult decapods, and is the carapace suggests that they were not yet well developed only in brachyuran "proto­ movable. zoea" and later zoea larvae; in nonbrachyu- We conclude that these carapaces represent ran decapod larvae a median dorsal spine is brachyuran larvae because they possess a sin­ not developed, although in anomurans there gle dorsal spine. In modern brachyuran lar­ are short paired spines (Gurney, 1942; Gore, vae prior to the zoea, the dorsal spine is ab­ 1985). The distal extremity of the dorsal spine sent (Gore, 1985: fig. 1), and it is lost again in both our specimens is broken and its orig­ in postzoeal (e.g., megalopa) stages (Gore, inal length is unknown, but it was probably 1985: figs. 2, 6, 7). The larva is thought to at least 1 mm long when complete. The dorsal represent an early zoea ("protozoea") stage spine arises close to the posterior margin of because the dorsal spine was not much longer the carapace and is directed posterodorsally. than the rostral spine, and because the eye It is cylindrical and tapered, but is otherwise stalks remained attached to the carapace after featureless. ingestion by the fish and even after acid prep­ The rostral spine (rs, fig. 4) is complete, aration. In the early zoeal stages the eyes are and is approximately 1 mm long. It has a fixed, and the eye stalks become movable only round cross section distally, but is broader in later zoeal molts ("zoea" of Gurney, 1942). and has paired longitudinal furrows proxi- The dorsal spine frequently becomes much mally, between the eyes. The spine is curved longer in successive molts of brachyuran zoea downward gently as it leaves the carapace, larvae. but is slightly upturned again distally. Rostral The brachyuran Araripecarcinusferreirai is and dorsal spines probably were of about known from a single, almost complete spec­ equal length, and both are devoid of serra­ imen, associated with the fish Vinctifer in a tions. Romualdo Member concretion (Martins- The carapace is rounded, not laterally com­ Neto, 1987). Araripecarcinus was referred to pressed, and its dorsal surface is marked by the superfamily Portunoidea because the fifth grooves, apparently the larval equivalent of pereiopod has a flattened and bladelike dac- 10 AMERICAN MUSEUM NOVITATES NO. 3132 tylus distally. This crab is very small, with a ber environment could not have been a "nor­ carapace less than 10 mm long. It is quite mal" (i.e., open) sea, however, because most possible that the larva described here pertains pelagic and benthic marine invertebrates such to that species, but this cannot be demon­ as ammonites, belemnites, brachiopods, strated. We have refrained from erecting a sponges, and corals are absent. Instead, the formal taxon based on such indeterminate environment (which trophic data from fishes larval material. suggest was semienclosed; Maisey, 1994) was This discovery of a fossil protozoea larva able to support only a taxonomically impov­ is significant for several reasons. Firstly, it erished and highly filtered marine arthropod- represents the first recorded instance of a fos­ pelecypod-gastropod community. sil brachyuran larva, and is therefore the old­ Finally, finding a zoea larva in the stomach est known. Secondly, these larvae confirm the of the gonorynchiform fish Tharrhias strong­ presence of brachyurans in the Romualdo ly suggests that it was a plankton feeder, and Member of the Santana Formation; until now, differed in its feeding habits from Rhacolepis crabs were represented only by the holotype which was a predator of benthic shrimps and of Araripecarcinus ferreirai. small fishes. Tharrhias and Rhacolepis pos­ Thirdly (and perhaps most importantly), sess very different mouthparts; the former has this discovery demonstrates that waters with­ a small subterminal mouth and is toothless, in the semienclosed Araripe Basin supported whereas the latter has elongate jaws with many a planktonic community, which was utilized pointed teeth (fig. 5). Elsewhere it has been as a food source by some fishes. Micropale- noted that Rhacolepis and Tharrhias do not ontological data (particularly from dinoflag- usually occur together in the same horizon ellates and ostracods) suggest a pulsative ma­ (Maisey, 1991, 1994), and it was speculated rine ingression sustaining a mixihaline en­ that this probably had some ecological sig­ vironment (Arai and Coimbra, 1990). nificance. This suggestion can be expanded Brachyuran planktonic larvae were evidently by the present observations, because we now capable of surviving in this environment, de­ have evidence that these fishes were not in spite the general absence of other expected direct competition for the same food source. planktonic taxa such as foraminifers. In both Instead, it may be that their principal food anomuran and brachyuran crabs the zoea sources were not usually found in the same represents the first free-swimming larval stage, location (suggesting some local compartmen- corresponding to the mysis stage in penaeid talization of the environment) or at the same shrimps; earlier developmental stages of these time (suggesting some seasonality or longer- crabs are passed within the egg (Bliss, 1982). term changes in the environment). These ob­ Some brachyurans were thus able to repro­ servations offer tantalizing clues about the duce in the sea which occupied the Araripe paleoenvironmental dynamics of the semien­ Basin. closed sea that occupied the Araripe Basin at Fourthly, because zoea larvae characterize that time. the life cycle of many marine decapods, but are atypical of nonmarine taxa, the fossil zoea larvae provide further evidence that the en­ NEW SPECIMENS OF vironment was marine to brackish. There are BEURLENIA ARARIPENSIS FROM several groups of freshwater brachyurans to­ THE CRATO MEMBER day (e.g., families Potamidae, Pseudothel- ORDER DECAPODA LATREILLE, 1903 phusidae, Trichodactylidae), but modern portunoids are mostly marine. Callinectes INFRAORDER CARIDEA BURKENROAD, 1963 (blue crabs) can survive in almost fresh water, FAMILY PALAEMONIDAE? but do not reproduce there (G. Bishop, per­ RAFINESQUE, 1815 sonal commun., Oct. 1994). The presence of the supposed portunoid Araripecarcinus to­ Beurlenia gether with brachyuran zoea larvae, from the Martins-Neto and Mezzalira, 1991 same member of the Santana Formation, sug­ EMENDED DIAGNOSIS: Palaemonid? with gests a marine habitat. The Romualdo Mem­ antennae longer than body; antennae and an- 1995 MAISEY AND DE CARVALHO: BRACHYURAN CRAB LARVAE 11

Fig. 5. Outline reconstructions of two fishes from the San tana Formation that preyed upon decapods: (A) Rhacolepis buccalis, a teleost with elongate, toothed jaws, which ate the sergestid Paleomattea; (B) Tharrhias araripis, a gonorynchiform with a short, edentulous mouth, and which ate planktonic brachy- uran larvae. tennules with two flagellae; antennal scales (1991) on the basis of a single, almost com­ short; rostrum extending beyond level of an­ plete specimen, and its morphology was de­ tennal peduncles, with six rostral and six scribed in some detail. Certain anatomical postrostral spines dorsally and three rostral features were not well preserved, however. spines ventrally; second pereiopod enlarged, Additional specimens in the Department of chelate; pereiopods posterior to second slen­ Invertebrates, American Museum of Natural der, without chelae; pleopods multisegment- History, New York, reveal additional fea­ ed; telson with single apical process, appar­ tures, especially in the rostral region, pre­ ently lacking paired spines or setae. sented here (figs. 6, 7). TYPE SPECIES: Beurlenia araripensis Mar- The largest new specimen (AMNH 44984) tins-Neto and Mezzalira, 1991: 157. is approximately 45 mm overall length, with a carapace length of almost 12 mm excluding the rostrum (18 mm with rostrum; figs. 6A, Beurlenia araripensis 7A). Its antennules are approximately 29 mm Martins-Neto and Mezzalira, 1991 long, and antenna length was at least 45 mm DIAGNOSIS: AS for genus. although the distal extremities are not pre­ HOLOTYPE: CD-I-161, Desiree Collection, served. These dimensions are slightly greater Rio de Janeiro; complete specimen in lami­ than those of the holotype, but the two spec­ nated lacustrine limestone; Aptian, Santana imens are of approximately equal size. Formation, Crato Member, Chapada do Ar- AMNH 44990 represents a slightly smaller aripe, Ceara. individual with an overall length of approx­ REFERRED MATERIAL: AMNH 44984, imately 35 mm (figs. 6B, 7B), and AMNH 44990, 44991, stratigraphic and locality data 44991 is even smaller (approximately 28 mm as for holotype. overall length; figs. 6C, 7C). DISCUSSION: Beurlenia araripensis was first The rostrum is incomplete in the holotype described by Martins-Neto and Mezzalira and in AMNH 44990 and 44991, but is com- 12 AMERICAN MUSEUM NOVITATES NO. 3132

'i

B y y jr

Fig. 6. Specimens of Beurlenia araripensis'. (A) AMNH 44984, x 2.16; (B) AMNH 44990, x 2.28; (C) AMNH 44991, x 2.05. pletely preserved in AMNH 44984 (fig. 7A). about 6 mm in front of it. The dorsal margin In this specimen the rostrum arises some 4 of the rostrum bears six pronounced spines mm behind the level of the eye and extends proximally, and there are six more postros- 13 1995 MAISEY AND DE CARVALHO: BRACHYURAN CRAB LARVAE

38P Fig. 6. Continued.

tral spines on the carapace (the posteriormost The second pereiopod is enlarged in com­ postrostral spine is very small). Each spine is parison with the others (Martins-Neto and evenly spaced from the next, giving the ros­ Mezzalira, 1991) and in AMNH 44990 its trum a serrated appearance dorsally. The dis­ chela is clearly visible. It does not bear a tal half of the rostrum has no dorsal spines. terminal brush of hairs. The third and sub­ Ventrally the rostrum has three spines, the sequent pereiopods lack chelae. It is difficult first lying almost halfway along the ventral to observe the coxal and basal segments of margin, below the anteriormost dorsal spines. the pereiopods, but the visible ones seem to The other two ventral spines are located an­ lack exopods. Multisegmented pleopods like terior to all the dorsal spines. Rostral mor­ those of the holotype are visible in AMNH phology in Beurlenia is profoundly different 44984 and 44991. from that found in Paleomattea from con­ According to Martins-Neto and Mezzalira cretions in the Romualdo Member of the (1991) the telson in Beurlenia has a single Santana Formation. apical "spine," and this is confirmed in both