Imagocnus zazae (MacPhee and Iturralde-Vinent, 1994)

R D E Macphee, M A Iturralde-Vinent & Eugene S Gaffney, 2003, Domo de Zaza, an Early Miocene Vertebrate Locality in South-Central Cuba, with Notes on the Tectonic Evolution of Puerto Rico and the Mona Passage, American Museum Novitates 3394, pp. 1-43 : 19-32

Imagocnus zazae

(Tardigrada, Megalonychidae , Megalocninae )

ORIGINAL DESCRIPTION: MacPhee and Iturralde­Vinent, 1994.

MATERIAL: Holotype is an edentulous palate ( MNHNCu – P 3014 ; fig. 6) with partial alveoli for posterior molariforms, but lacking rostral end.

Referred material ( figs. 7–10) now comprises two fragmentary molariforms ( MNHNCu – P 3100, 3013); a right maxillary fragment ( MNHNCu –P 3165), preserving parts of two alveoli; a partial pelvis ( MNHNCu –P 3021), comprising most of the centra of sacrals 2 and 3, part of the sacral portion of the vertebral canal, and (on the left side only) part of the ilium; another sacral fragment, consisting only of the bodies of sacral?2 and?3 ( MNHNCu – P 3017); an anterior caudal vertebra ( MNHNCu –P 3050), lacking a portion of the centrum, parts of the transverse processes, and the spinous process; and a left distal humerus ( MNHNCu –P 3027), diaphysis only.

The palate ( fig. 6) is distinctively slothlike in having a thick cross­section, a narrow interalveolar space, large numbers of palatal nutrient foramina, and subparallel rows of tooth sockets. It preserves three half­alveoli on each side, the alveolar rows being separated by a very narrow interalveolar distance. The toothrows are very slightly divergent caudally. Ventrally, there is a pronounced mesial ridge bordering the alveolar rows, and also a midsagittal torus along the intermaxillary suture. A small portion of the right maxillopalatine suture is preserved; it joins the intermaxillary suture at the transverse level of the alveolus of the third molariform. In the Quaternary Puerto Rican megalonychid Acratocnus , the mesial ridges are better developed and the midsagittal torus is less developed. In the extant megalonychid Choloepus , both kinds of ridges are negligible even in large specimens, and the toothrows are divergent rostrally rather than caudally.

Degrees of toothrow convergence/divergence vary within extinct families, and it may be that row conformation is a poor character. Weber (1928) stated that the toothrows in megatheriids are parallel, while in mylodontids they can be parallel ( Scelidotherium ) or divergent ( Mylodon ). Englemann (1985) regarded anterior divergence of toothrows as probably primitive for sloths, but all three families have representatives in which they are subparallel. It may be that he was counting anterior, caniniform teeth as part of the toothrow, in which case anterior divergence is probably primitive. However, restricting attention to the molariform teeth indicates that the subparallel condition is more likely to be plesiomorphous.

The channels for the palatine vasculature and nerves can be seen in section in figure 6A. In the primitive case, sensory branches from the palate travel within palatal soft tissues and leave through foramina located in the rear of the mouth, to the pterygopalatine ganglion. The greater palatine artery or its equivalent, ultimately a branch of the maxillary artery, also travels unenclosed and supplies the palatal mucous membrane, glands, and gums. In tardigradans, by contrast, the bone of the palate tends to partly enclose the pathways of the palatine nerves and arteries, so that fibers and distributaries have to travel through individual small foramina to gain or leave their trunks. In some taxa, such as Imagocnus , the nerve and arterial trunks are completely isolated within long tubes. In others, channels are less deeply buried or take the form of simple grooves on the palatal surface (as in Choloepus ).

In addition to small, rather randomly arranged emissary foramina, in tardigradans there is normally a pair of larger, vaguely symmetrical foramina (anterior palatine foramina) in the rostral part of the palate. In most megalonychid taxa, the anterior palatine foramina are situated at the transverse level of the first molariforms. In a few taxa, Acratocnus and Parocnus included, they are regularly situated much farther forward. There is probably individual variation in this feature. The situation in Imagocnus is not completely certain because the portion of the palate anterior to the first molariforms was not preserved, although it is clear that no large paired foramina are represented on the part that is left. We accordingly infer that the anterior palatine foramina were situated well in advance of the first molariforms, as in Acratocnus and Parocnus . When they are distinguishable in Choloepus they are also rostrally situated.

The dorsal surface of the palate, which would have functioned as the floor of the nasal cavity, also bears a midsagittal ridge, on which the cartilaginous nasal septum presumably rested. Laterally, there is a well­ marked fossa, typical of megalonychids, in advance of the first molariform.

The two recovered teeth ( MNHNCu –P 3013 and MNHNCu –P 3100; fig. 7) are clearly megalonychid. In morphology the teeth closely resemble mandibular molariforms of the Quaternary Cuban species Parocnus browni , but metrically they are intermediate between this species and the larger Megalocnus rodens (for additional systematic commentary, see White and MacPhee, 2001).

The referred maxillary fragment (MNHN­ Cu–P 3165, not illustrated) consists only of the external surface of the bone covering two molariform alveoli. It clearly accommodated teeth within the size range of known molariforms of Imagocnus , but its condition does not warrant detailed description.

The partial pelvis ( MNHNCu –P 3021; fig. 8C, D) was found in situ, exposed on a manmade terrace cut into the canal wall ( fig. 3, lower star). The specimen is in poor condition, perhaps because it was weathered before it was originally buried. The ventral surface of MNHNCu –P 3021 is dominated by the fragmentary centra of upper sacral vertebrae, whose lateral masses are solidly fused with the contiguous portions of the ilia. The Zaza pelvis is actually slightly larger and more robust than AMNH –VP 49976, a nearly complete (although partly reconstructed) pelvis of Quaternary Megalocnus rodens ( fig. 8A, B). Sacral vertebral bodies are flanked by ventral sacral foramina of extraordinary size and the sacral canal is enormous, again larger than in Quaternary Megalocnus (table 5).

Extreme ventral projection of sacral centra is characteristic of Antillean sloths and is not seen to so marked an extent in Hapalops and other South American megalonychids. Welldeveloped ridges on the ventral surfaces of the centra (presumably for iliacus m. attachment) are present, as in Quaternary Antillean sloths. Foramina for basivertebral veins, seen perforating lumbar and sacral vertebrae in members of all major tardigradan taxa ( de Burlet, 1922), can also be seen in MNHNCu –P 3021 (but rather indistinctly because of damage).

On the left side, the dorsal aspect of the pelvis (not illustrated) bears remnants of two major longitudinal crests (crista sacralis la­ teralis and crista sacralis medialis, for attachment of epaxial musculature), in the gutter between which are stacked the dorsal apertures of the sacral foramina. In Quaternary Antillean megalonychids, these cristae are positioned fairly close together, and the gutter defined between them is a deep groove. In the Lagunitas fossil, the cristae are much farther apart (40% more separation than in Megalocnus rodens ; table 5), and the gutter is relatively less deep. Whether these features mean that the complete Lagunitas pelvis was considerably larger than that of M. rodens cannot be settled, but they certainly show that proportions were different.

The anterior caudal vertebra ( MNHNCu – P 3050; fig. 9) is imperfect, but it resembles anterior caudals of large Quaternary Antillean sloths in detail and there can be no question about its allocation. The four hemal facets on the ventral aspect of the centrum and the associated bridges enclosing large vascular foramina are diagnostic, occurring in much the same way in Parocnus and other Quaternary Cuban sloths ( Fischer, 1971). The precise position of this vertebra within the skeleton of the tail cannot be fixed, but from the marked separation of its pre­ and postzygapophyses it is probably one of the anteriormost. The extremely robust tail skel­

TABLE 5 Measurements of Megalonychid Posteranials from Domo de Zaza, in mm

etons of Quaternary sloths consisted of approximately 20 elements ( Fischer, 1971). Representative measurements (table 5) indicate that the Zaza specimen is in the size range of Megalocnus rodens .

A fragmentary distal humerus found in 1994 ( MNHNCu –P 3027; fig. 10) is here referred to Megalonychidae , although the specimen consists of the diaphysis only. In general conformation, the specimen agrees in detail with humeri of Quaternary Parocnus and Acratocnus (cf. White and MacPhee, 2001). It differs from the humeri of P. browni and P. serus in possessing an entepicondylar foramen (the aperture of the foramen is gone, but the diaphyseal sulcus is recognizable). It additionally differs from Parocnus in that the deltopectoral eminence is situated low (= distally) on the bone and lacks any lateral ‘‘flare’’. Acratocnus is similar in these regards, but all Quaternary Acratocnus species are smaller than MNHNCu –P 3027. Megalocnus is much larger, and Neocnus is much smaller. Assuming it was fully maturated at the time of death (the distal break occurred above the position of the epiphysis), the humerus suggests a smaller animal than does the pelvis.

DISCUSSION: Although significant differ­ ences in the body masses of individual sloths are implied by some of the Zaza specimens (e.g., Parocnus ­sized molariforms vs. Megalocnus ­sized or larger pelvis), the systematic significance of this is hard to evaluate. It may be that more than one species of large sloths existed in Cuba during the Early Miocene. Alternatively, body sizes and proportions may have varied substantially within Imagocnus zazae , as they are known to have done in many Quaternary Antillean sloth species (MacPhee et al., 2000c; White and MacPhee, 2001). Until sample sizes improve, this issue cannot be usefully addressed. For the same reason, it is equally difficult to fix the phylogenetic position of Imagocnus . Although it displays no exclusive resemblance to any single Quaternary sloth, on the whole it seems closer to Megalocninae than Choleopodinae as organized by White and MacPhee (2001). We therefore tentatively assign Imagocnus to the former subfamily without, however, deciding whether it is clos­ er to Megalocnus or Parocnus (or is the sister taxon to both, which is probably more likely).

CETACEA (ODONTOCETI)

Remarkably, Domo de Zaza has yielded evidence of toothed whales, although the material is too unsatisfactory to yield much in­ sight into the taxa represented. However, as a completely new group to the West Indian fossil record, their presence is worth recording.

Cetacean Species A

MATERIAL: Isolated tooth ( MNHNCu –P 3090) recovered in 1994 by M. Iturralde­Vinent in lag deposit ( fig. 11).

ATTRIBUTION: The tooth’s size (76.7 mm, greatest length; 19.7 mm, greatest width) and other distinctive features immediately preclude attribution of this specimen to Sirenia or, indeed, any mammalian group other than Odontoceti. The likeliest allocation is Physeteridae , although no exact match could be found in the literature.

The tooth is moderately curved, entirely free of enamel, with no waisting or other feature marking the transition from root to crown. A thick (> 1 mm) coating of cemen­ tum, broken off in places, covers the root. The crown end bears numerous, very distinct ‘‘anular’’ growth lines, and terminates as a smooth­surfaced conus with a well­marked wear facet. The apical end of the root, ovoid in section, rapidly narrows near the terminus; the pulp cavity is open but mostly filled in, so that it extends as a vacuity only a few millimeters into the substance of the tooth. This may indicate that the tooth came from an aged animal (E. Fordyce, personal commun.).

Except for this last feature, all of these characters occur in Orycterocetus (L. Miocene Calvert Fm; cf. Kellogg, 1965) and its close relatives in Physeteridae . Orycterocetus fossils display long, open pulp cavities (?age related), and at least some teeth are longitudinally fluted (Kellogg, 1965). Fluting is very slightly indicated in the Zaza specimen. The wear facet on the posterior surface may indicate the presence of opposing teeth, although this condition can be found in mandibular teeth of modern Physeter (which lacks erupted maxillary teeth) as the result of contact with abrasive food ( Boschma, 1938).

Although the specimen probably represents a species distinct from those currently recognized for Orycterocetus , the similarities are strong enough to support the idea that a closely related sperm whale lived in the Caribbean region in the Early Miocene. Physeterids are known from the Middle Miocene in North America and Early Miocene in South America ( Fordyce and Barnes, 1994); the Zaza specimen is apparently the first sperm whale of significant antiquity to be reported from the West Indies (and one of the few reports of a fossil cetacean of any sort from this area; cf. Ray, 1964).

This specimen was compared to the tusk of the rytiodontine dugongid Dioplotherium manigaulti as illustrated and described by Domning (1989a). MNHNCu –P 3090 resembles ‘‘typical’’ Dioplotherium tusks in having moderate curvature, cementum covering, prominent growth lines, shallow pulp cavity, and no or very little enamel. However, Dioplotherium tusks are typically thicker, longer, and lozenge­shaped (‘‘bladelike’’) in crosssection, among other distinguishing features. However, according to Domning (1989b) rytiodontines differ among themselves as to whether there is an apical wear facet and ‘‘self­sharpening edge’’ on tusks, and there is some shape similarity between MNHNCu – P 3090 and Domning’s so­called ‘‘variant’’ specimen, YPM 21334. While some degree of morphological convergence between rytiodontine tusks and sperm whale teeth is apparent—unsurprisingly, given the simple geometrical shapes involved—we feel confident that MNHNCu –P 3090 is better referred to the latter than to the former.

Cetacean Species B

MATERIAL: Isolated tooth ( MNHNCu –P 3086) recovered by the 1994 field party ( fig. 12).

ATTRIBUTION: We offer this specimen as a cetacean with some hesitancy, as it is far from complete and there is certainly room for disagreement until better material comes to hand. However, it is critical to note that there is a definite crown­root junction, and the crown itself is composed of enamel. In our opinion, this combination of features indicates that the specimen is mammalian rather than crocodylomorph, the only other reasonable possibility.

The enamel is black and densely wrinkled. The preserved part of the specimen is essentially acorn­shaped and is very slightly compressed (maximum width at base of crown, 14.5 mm; maximum width at right angle to previous measurement, 13.6 mm; height of crown as preserved, 15.0 mm). There is a slight keel or carina, and the apex of the crown is slightly abraded. There is no cingulum or suggestion of accessory cusps. Presumably, the crown capped a long root, now broken off, of the sort seen in many odontocetes.

The relationships of this fossil are quite unclear, although its size and some details of its construction make odontocetes the likeliest candidate. Ewan Fordyce (Univ. Otago), who has kindly examined photographs of this specimen, noted that the keel could have supported tiny denticles, now worn away—a rather ancient feature (cf. better­developed ‘‘cusps’’ or denticles of archaic odontocetes, e.g., Squalodontidae , Kentriodontidae ), archaic mysticetes (e.g., Aetiocetidae ), and archaeocetes (e.g., Basilosauridae ). Wrinkled enamel has been reported for archaic physeterids (e.g., Scaldicetus ) and some other archaic cetaceans (basal mysticetes, later archaeocetes), although in other respects the Zaza tooth is sui generis. However, Fordyce doubts that any fine allocation of the tooth is possible at this point, although he rules out any described group of delphinoids, platanistoids, and ziphiids. It is just possible that it is a physeterid.

SIRENIA

MATERIAL: Skull cap ( MNHNCu –P 3124; fig. 13); squamosal with zygomatic process ( MNHNCu –P 3316; fig. 14); proximal epiphysis of humerus ( MNHNCu –P 3091); caudal vertebra ( MNHNCu –P 3024); numerous ribs and rib fragments (accessioned as MNHNCu – P 3010); and a large number of unaccessioned vertebral fragments and other osteosclerotic/ pachyostotic fragments not securely identified as to element. Only the first two elements not­ ed above are illustrated here. Recovered by various field parties between 1990 and 1998.

ATTRIBUTION: The sirenians of Zaza seem to have been exclusively dugongid, as would be expected given that trichechids did not invade the Caribbean area until later in the Miocene ( Domning, 1989a). Two dugongids are possibly represented in the fossil material, one of which may be Metaxytherium (or some other hydrodamaline). However, as the material is quite poor we have refrained from making taxonomic allocations at this time.

The Atlantic–Caribbean area was the site of significant dugongid diversity during the early and middle Neogene ( Domning, 1988, 1989a, 1989b, 1990, 2001; MacPhee and Wyss, 1990). Although sirenian fossils are known from a number of localities in Cuba (Iturralde­Vinent, personal obs.), virtually none of this material has been described. At present, the only named Tertiary sirenian from Cuba is the hydrodamaline Metaxytherium riveroi (Varona, 1972, 1974), based on the jaw of an immature animal recovered from E.–M. Miocene Güines Fm near Matanzas. Few characters distinguish this species from M. calvertense as described by Kellogg (1966); indeed, it may not be distinguishable at all if dental homologies were misinterpreted by Varona, as seems likely (D. Domning, personal commun., cited by MacPhee and Wyss, 1990).

DESCRIPTION: In view of their generally poor condition, the Zaza sirenian fossils do not merit extensive description.

The skullcap ( fig. 13) consists, as is usual in sirenians, of the co­ossified parietals and supraoccipital. The temporal ridges are low, widely separated, and nearly parallel (as opposed to rostrally convergent as in Halitherium , or divergent as in Metaxytherium and many other taxa). At their closest approach, the ridges are separated by a gap of ~ 40 mm (vs. 15 mm in M. calvertense ). Skullcap length (external occipital protuberance to anterior border of parietals, in midsagittal plane) is 70.3 mm (cf.> 100 mm in M. calvertense ), while minimum biparietal width (at right angle to foregoing measurement, immediately anterior to position of lambdoidal suture) is 60.0 mm (cf. 65 mm in M. calvertense ). This suggests that the Zaza skullcap is proportioned differently from that of Metaxytherium . Measurements are closer to those of the? Caribosiren skullcap from the Miocene of Puerto Rico figured by MacPhee and Wyss (1990), although here also temporal line curvature differs from that of the Zaza specimen. Recently described rytiodontines from the Caribbean–Atlantic area ( Domning, 1989a, 1989b, 1990) also differ notably from the Zaza specimen for the few characters that can be compared.

The partial temporal ( fig. 14) consists of most of the zygomatic process of the squamosal, glenoid area, and squama. The petrous portion is not preserved. In this case, correspondence to Metaxytherium is actually quite close, especially in the conformation of the glenoid and postglenoid regions (cf. illustrations presented by Kellogg, 1966; Domning, 1989a).

The humeral head (MNHNCu–P 3091), not illustrated, is very poorly preserved and is noteworthy only for its size: maximum dimensions of the head are 83.5 mm by 75.0+ mm, which correspond well with measure­ ments reported by Kellogg (1966) for M. calvertense but are somewhat larger than those for M. floridanum provided by Domning (1989a). Morphologically all that can be said is that the head is highly rounded, as is typical for sirenians. The caudal vertebra (MNHNCu–P 3024), also not illustrated, is quite small (centrum height, 24.4 mm), but otherwise corresponds closely to Kellogg’s (1966) description of caudal vertebra morphology in M. calvertense (which had at least 20 caudals).

In conclusion, although there are some indications that a dugong closely allied with if not identical to Metaxytherium sp. is represented in the Domo de Zaza collections, decisive evidence (e.g., strongly downturned snout, dental features listed by Domning [1989a]) will be required before any systematic allocation can be confidently made. Presence of seagrass feeders in the Zaza area 17 Ma would certainly be expected, given the wide expanse of shallow waters framing the Cuban archipelago at that time.

TESTUDINES (PLEURODIRA, PELOMEDUSOIDES)

MATERIAL: Carapace ( MNHNCu –P 3012) lacking nuchal, pygal, and anterior peripherals; plastron lacking right mesoplastron, most of right hypoplastron and xiphiplastron, and posterior margin of left xiphiplastron.

Left half of carapace ( AMNH –VP 27465; fig. 15A) with seven neurals and adjacent proximal parts of right costals; peripherals 1– 9 and left edge of nuchal present; plastron lacking anterior margin, right epiplastron, right half of entoplastron, anterior portion of right hyoplastron, most of right bridge, posterior edges of both xiphiplastra; right and left pelves present.

Plastron ( AMNH –VP 27466; fig. 15B) badly fractured but lacking only anterior margin; left bridge and parts of left xiphiplastron; carapace fragments.

Portion of plastron ( AMNH –VP 27467).

ATTRIBUTION: The four listed specimens are very similar to one another and we conclude that they almost certainly belong to the same species. The species in question is clearly a pleurodire, because the pelvis is fused to the carapace and plastron—a diagnostic synapomorphy of Pleurodira ( Gaffney and Meylan, 1988). Small, laterally placed mesoplastra are characteristic of the Pelomedusoides, and this may be provisionally considered a shell synapomorphy of the group. This attribution could be placed beyond doubt if it could be established that the cervical scale is absent (a pelomedusoid synapomorphy; Gaffney and Meylan, 1988), but unfortunately none of the specimens pre­ serves a complete nuchal bone. Overall, the strong similarity of the shell of the Cuban species to known podocnemidids, such as Podocnemis , supports the identification offered here. The higher taxonomy of pleurodires follows Gaffney and Wood (2002) in recognizing the Hyperfamily Pelomedusoides, which contains the families Podocnemididae , Bothremydidae , and Pelomedusidae sensu stricto (containing only Pelusios and Pelomedusa ).

Despite the fact that the shell of the Cuban species is almost completely known, there are serious problems in assigning it to a more specific taxonomic level. This is partly due to the disarray of pelomedusoid systematics occasioned by the traditional emphasis on shell characters as a basis for diagnosing chelonian taxa. For some turtle groups in which shell characters have been adequately surveyed and analyzed, and complete (or nearly complete) specimens are available for study, the traditional emphasis is perhaps not misplaced. Most pelomedusoids, however, are strikingly conservative in shell morphology (Wood and Diaz de Gamero, 1971; Gaffney and Zangerl, 1968), and therefore the exclusive use of shell characters practically invites the creation of paraphyletic groups. An outstanding example of this is the systematic history of the genus Podocnemis , which retains a generalized pelomedusoid shell type. A large number of fossil species, mostly defined on the basis of shell characters, have been assigned to this genus; these taxa range in age from Cretaceous through Tertiary, and in area from North and South America to Africa and Asia. When cranial material is available, however, in most cases the basis for assignment to Podocnemis sensu stricto collapses (e.g., Gaffney and Zangerl, 1968; Wood, 1970; Pritchard and Trebbau, 1984; Gaffney and Wood, 2002). Podocnemis (sensu Gaffney, 1979, excluding Erymnochelys and Peltocephalus ) has a distinctive skull and cervical morphology, and all validly ascribed living and extinct species are restrict­ ed to South America.

Having said this, it is necessary to observe that although the Cuban species has a shell that closely resembles that of Podocnemis ( P. expansa in particular; see Description), there is no assurance or even a high likelihood that the Cuban form possessed the cranial and vertebral synapomorphies currently used to define this genus. No pelomedusoids have been described from Cuba heretofore, although there are records from the Tertiary of Puerto Rico (Wood, 1972; Wood and Gaffney, 1989; Gaffney and Wood, 2002). The latter consist of a fragmentary shell ( AMNH –VP 1836), identifiable only to Pelomedusoides, from the late Oligocene San Sebastian Fm (Wood, 1972; see also Mac­ Phee and Wyss, 1990), and a series of wellpreserved skulls and shells ( Gaffney and Wood, 2002) attributed to the new podocnemidid genus, Bairdemys . The Puerto Rican record is Bairdemys hartsteini , a form most closely related to Stereogenys (Eocene–Oligocene of Egypt) and Shweboemys (Miocene–Pliocene of Pakistan and Burma), equivalent to the unnamed taxon B9 of Gaffney and Meylan (1988). It is probably from near­shore marine sediments. The other Puerto Rican record, AMNH –VP 1836, is thought to have been deposited ‘‘partly under brackish and partly under nearshore marine conditions’’ (Wood, 1972: 2). Much the same thing could be said of the Cuban pelomedusoid. Bairdemys is also found in presumed near­shore sediments in Venezuela (Sánchez­ Villagra et al., 2000; Gaffney and Wood, 2002). The living Podocnemis itself is restricted to freshwater, but fossil evidence indicates that marine­adapted Podocnemididae were common in the past. The Cuban form is therefore ambiguous as an environmental indicator.

DESCRIPTION: Carapace ( fig. 15A) similar in size and shape to Podocnemis expansa ; oval, peripherals flaring posteriorly, approximately 68–70 cm minimum length (restored length, based on AMNH VP 27465). Nuchal, suprapygal, and pygal not known, remaining carapace bones similar to Podocnemis . Seven neurals present: neural 1 four­sided, neurals 2–5 six­sided and coffin­shaped, neural 6 sixsided and equidimensional, and neural 7 fivesided and smaller than others. Eighth costals meeting at midline, seventh costals separated by neural 7. Bairdemys venezuelensis (Wood and Diaz de Gamero, 1971; Gaffney and Wood, 2002) is unusual among Pelomedusoides in lacking neural bones in the carapace. The fragmentary Puerto Rican shell material, possibly associated with Bairdemys , does have at least some neurals. The presence of neurals in the Cuban species does not preclude it from belonging to Bairdemys or a near relative, as this feature can vary between species of turtles in the same genus.

Carapacial sulci similar to Podocnemis except that vertebral scales 2–4 narrow posteriorly, producing thereby a distinct but somewhat irregularly curved edge (similar to P. erythrocephala as figured by Pritchard and Trebbau, 1984: fig. 8). Little of vertebral 5 preserved.

Plastron completely known ( fig. 15B) and similar in size and shape to Podocnemis expansa . Anterior lobe curved, not truncated as in Kenyaemys; anterior lobe shorter than posterior lobe, as in Podocnemis . Intergular scale single, distinctly larger than gular scales; completely separates gulars and partly separates humerals. Gulars are slightly small­ er than their equivalents in most P. expansa and are entirely situated on epiplastra. Intergular extends onto entoplastron only for a short distance, as in P. erythrocephala as figured by Pritchard and Trebbau (1984: fig. 8). Humeral–pectoral sulcus lies entirely on hy­ oplastron, just posterior and parallel to epiplastron–hyoplastron suture. Mesoplastra longer than wide, with pectoral–abdominal sulcus lying across anterior portion as in Bothremys , in contrast to many (but not all) Podocnemis specimens. Axillary buttress strong, reaching peripheral 3 and extending nearly halfway across costal 1. Inguinal buttress distinctly weaker than axillary, as in P. expansa , only slightly attached to costal 5. Margins of posterior lobe straight but tapering. Xiphiplastron and xiphiplastral (anal) notch best seen in AMNH–VP 27466. Posterior margin of xiphiplastron rounded as in Podocnemis and Bothremys , in contrast to the acute margin found in Taphrosphys . On dorsal surface of xiphiplastron, pubic scar oval, not as elongate as in Podocnemis . Ischiac scar triangular, near edge of xiphiplastron, as in Podocnemis . Xiphiplastral notch acute rather than rounded. Known portions of pelvis indistinguishable from Podocnemis expansa .

DISCUSSION: The close similarities of the Zaza turtle with Podocnemis and other podocnemidids suggest identifying it with this group. Although it is likely that the Zaza pelomedusoid is a podocnemidid, until better material is found, it should be treated as Pelomedusoides incertae sedis ( Testudines, Pleurodira, Pelomedusoides ).

ELASMOBRANCHIA

MATERIAL: Isolated elements such as teeth, vertebrae, and spines of various fish groups have been recovered from Canal de Zaza and other places where Lagunitas Fm outcrops. The elasmobranchs have been worked up systematically by Iturralde­Vinent and coworkers (1996, 1998), and the reader is referred to these papers for further details. The elasmobranchs found at Zaza are essentially cosmopolitan species and therefore shed little light on the specific nature of the marine biota living around central Cuba in the Early Miocene. However, the species assemblage suggests the existence of warm, shallow, nearshore marine environments at this time ( Iturralde­Vinent et al., 1996), which is consistent with the presence of extensive marine limestone and marl facies in the Lagunitas Fm.

Shark teeth were collected from marine rocks and lag deposits derived from marine beds in various parts of the Domo de Zaza exposure. Four shark and two ray species have been identified in the collection so far.

ATTRIBUTION: All material consists of isolated teeth. Except where indicated, galeomorph material was recovered from lag deposits at Domo de Zaza in August 1990 by a team composed of R. MacPhee, M. Smith, R. Rojas, and M. Iturralde­Vinent. Myliobatid material was collected from the same place and also (in the case of MNHNCu –P 3121 and 3122) from nearby Zaza Dug­Out.

Galeomorpha: Carcharhinidae

Carcharhinus obscurus ( MNHNCu –P 2219, 3033)

Hemipristis serra (MNHNCu–P 2221, 3034 and 3036; plus fragmentary tooth MNHN 2268 collected by M. Iturralde­Vinent and R. MacPhee in 1996).

Negaprion brevirostris (MNHNCu–P 3037).

Galeomorpha: Sphyrnidae