Fossil Cenozoic Crassatelline Bivalves from Peru: New Species and Generic Insights

Discoveries of new fossil Cenozoic crassatellines in Peru provide a new phylogenetic perspective on “large” Neogene genera, in which four lineages are considered to have arisen independently from different Paleogene Crassatella ancestors. Latest Oligocene and early Miocene species of the new genus Tilicrassatella gen. nov.―T. ponderosa, T. torrens sp. nov., and T. sanmartini sp. nov. from the East Pisco Basin―probably evolved from the late Eocene species, Crassatella rafaeli sp. nov., which itself differed in significant respects from slightly older species of the East Pisco Basin, C. neorhynchus and C. pedroi sp. nov. The paciphilic genus, Hybolophus, is raised to full generic status. Added to its ranks are the East Pisco Miocene species H. maleficae sp. nov., H. terrestris sp. nov., and the oldest species of the genus, the late Eocene or Oligocene H. disenum sp. nov. from the Talara Basin of northern Peru. Kalolophus gen. nov., encompassing circum-Caribbean fossil species, the extant species, K. speciosus, and the trans-isthmus species, K. antillarum, appears to have evolved from the early Oligocene Floridian species, Crassatella portelli sp. nov. The genus Marvacrassatella is a western Atlantic Miocene lineage most likely descended from Kalolophus. The genus Eucrassatella is restricted to Australian and New Zealand taxa. The Eocene New Zealand species, Spissatella media, is transferred to Eucrassatella and deemed a candidate for the most recent common ancestor of younger Eucrassatella and all Spissatella species. In the southern Pacific Ocean, the circum-Caribbean region, and tropical western America, crassatelline lineages developed one or more of the following characters: large resilifers, smooth ventral margins, and an extended left anterior cardinal tooth. Some of these late Paleogene convergent character changes might have countered increased shear forces exerted on the crassatelline valves while burrowing into finer-grained and more cohesive sediments in deeper or quieter water.


Introduction
The bivalve subfamily Crassatellinae Férussac, 1822, is represented on the west coast of South America by two living species, Eucrassatella gibbosa (Sowerby, 1832) and E. antillarum (Reeve, 1842), and up until now, one Eocene species (Olsson 1931); Miocene, Pliocene, and Pleistocene species in northern Peru and Ecuador (Spieker 1922;Olsson 1932;Marks 1951;DeVries 1986); and one early Miocene species in Chile (Philippi 1887). Discoveries of new fossil species in Peru provide an opportunity to re-examine systematic relationships among American crassatellines and between American and Australian/New Zealand crassatellines.

Geological setting
The East Pisco Basin, source of most of the new crassatelline material, stretches 350 km along the desert coast of south-central Peru (Fig. 1B, C). Nearshore and outer shelf sands were deposited in this forearc basin from the middle Eocene through the Pliocene (DeVries 1998). The fossil record of the East Pisco Basin is notable for its cetaceans and penguins, but mollusks are also present (Lisson 1925;Rivera 1957). Forty km south lies the Sacaco Basin, extending 50 km from El Jahuay to Yauca (Fig. 1B, D), where upper Miocene and Pliocene nearshore marine strata contain a diverse assemblage of marine vertebrates and mollusks (Muizon and DeVries 1985), including one crassatelline species.
The Talara Basin of northern Peru (Fig. 1A, B), source of additional crassatelline material, is best known for its fossil mollusks (Olsson 1931(Olsson , 1932. Many thousands of meters of turbiditic, neritic, deltaic, and fluvial sediments accumulated from the Paleocene through the Pliocene (Higley 2004). Lower and middle Pleistocene marine terrace deposits now cover much of the older outcrop (DeVries 1988).
The Cenozoic stratigraphy for the East Pisco Basin has been described by DeVries (1998) and for the Sacaco Basin by Muizon and DeVries (1985). Ages are based on the occurrence of diatoms and radiolarians and K-Ar and 40 Ar-39 Ar radiometric dates from volcanic ash beds. Stratigraphic correlations across the basin are based on the author's fieldwork. Ages and stratigraphy for the Talara Basin are based on Higley (2004) and Martinez et al. (2005) and, for the Pliocene and Pleistocene, DeVries (1986,1988). Period and stage boundaries are taken from the ICS International Chronostratigraphic Chart, version 2015/01. Stages are cited when absolute ages can be estimated. Depositional sequence names for the East Pisco Basin (Paracas, Lutetian to Bartonian; Otuma, Priabonian; Chilcatay, Chattian to Langhian; Pisco, Serravallian to Zanclean; see DeVries 1998) are used instead of the eponymous formation names.

Material and methods
Most of the new fossil crassatellines from Peru were found by the author. Comparative material was made available by the USNM, USGS, UF, PRI, UWBM, SBMNH, Museo de Historia Natural (Santiago, Chile), GNS Science (Lower Hutt, New Zealand), Department of Earth Sciences, University of California (Riverside, USA), and International Fossil Shell Museum (Utrecht, The Netherlands). Most figured specimens were coated with ammonium chloride before being photographed. Types and figured specimens of crassatellines are deposited in the UWBM and the Departamento de Paleontología de Vertebrados, Museo de Historia Natural, Lima, Peru.
Dimensional measurements (in mm) enclosed within parentheses apply to broken specimens. Locality-sample numbers of the author begin with "DV" (e.g., "DV 509-1"). Most locality-samples also carry a locality number for the UWBM (e.g., "B8315"). Full locality-sample descriptions are listed in Appendix 1. Hinge characters have been used to distinguish crassatellines at the generic level since Lamy (1917). A lack of agreement about such basic metrics as the number of cardinal teeth and presence or absence of lateral teeth has been an impediment to understanding crassatelline taxonomy. The scheme for naming hinge characters employed here is strictly descriptive and keyed to images (Fig. 2).
Terminology for lateral hinge structures has often been contradictory (Darragh 1965;Chavan 1969;Coan 1984;Wingard 1993). The left valve exhibits a thickening of the lunule margin distal to the beak and a thickening of the posterior ventral margin of the hinge plate (vmHP). The right valve exhibits a thickening of both the anterior vmHP and the distal end of the escutcheon margin. These "lateral ridges" (Wingard 1993) are variably developed and not used here for classification.
The left valve of crassatellines has two cardinal teeth. The posterior margin of the ligamental pit is sometimes thickened beyond the edge of the escutcheon in the left valve, creating a third ridge that has been termed the posterior cardinal tooth by Collins (2013). Most often, the ridge is entirely coalescent with the margin of the escutcheon or obsolete. The right valve has one cardinal tooth, but does have an anterior ridge on the hinge plate that can (i) join the dorsal ends of the cardinal tooth and lunule margin ( Fig.  2C 2 ), (ii) diverge anterioventrally from the lunule margin, or (iii) extend parallel to and even coalesce with the lunule margin ( Fig. 2B, D 2 ). The right valve also has a lamellar posterior ridge that can diverge posterioventrally from the midpoint of the cardinal tooth at (i) a large angle, passing the resilifer ventrally to create a trigonal socket next to the vmHP (Fig. 2B), (ii) a small angle, passing the resilifer anteriorly to create a small elongate socket ( Fig. 2C 2 ), or (iii) emerge from beneath the beak and diverge posterioventrally from the cardinal tooth ( Fig. 2D 2 ). Both ridges, which often lack corresponding sockets, are termed pseudocardinal teeth (Wingard 1993). The inclination of hinge teeth is measured from the beak with respect to the dorsoventral axis.
Following Wingard (1993), the exterior angulation extending from the beak to the posterior margin is termed the posterior ridge; one or two can exist (Fig. 2D 3 ). Small teeth along the inner ventral margin are termed crenulations ( Fig. 2A, B).

Results
The genus Crassatella Lamarck, 1799, has a tortuous nomenclatural history that commenced with a misidentified type species (Stewart 1930;Vokes 1973;Wingard 1993). Crassatella is accepted here, as it was by Coan (1984), in the expectation of an overdue and successful formal petition to the International Commission on Zoological Nomenclature to validate the name. The type species is Crassatella tu-mida Lamarck, 1805 for reasons enumerated by Wingard (1993). The genus Crassatellites Krüger, 1823 is an unavailable homonym created for fossil crassatellids (Iredale 1921;Stewart 1930;Vokes 1973).
Diagnoses of Crassatella typically reference two characters: (i) a resilifer that in large specimens extends only about half way from the beak to the vmHP ( Fig. 2A, B), and (ii) a crenulate inner ventral margin ( Fig. 2A, B; Stewart 1930;Chavan 1969;Wingard 1993). The short resilifer appears in Cretaceous and Paleogene species and incongruously in modern crassatellines from South Africa, Japan, and Brazil (e.g., the genera Indocrassatella Chavan, 1952, Nipponocrassatella Kuroda and Habe, 1971, and Riosatella Vokes, 1973. In some late Paleogene specimens and species of Crassatella, however, the resilifer extends nearly to the vmHP (Wingard 1993;this paper). The full-length resilifer is typically associated with a distinctive Cenozoic genus, Bathytormus Stewart, 1930, and with Neogene crassatelline genera.
The crenulate inner ventral margin is more reliable for diagnosing Crassatella, although crenulations appear only on the largest specimens of some species (Wingard 1993;this paper). Rarely, Crassatella-style ventral crenulations incongruously occur together with a full-length resilifer, e.g., an Eocene species from New Zealand (Collins 2013) and an early Miocene species from Chile (Philippi 1887). (Note: a third example, the Californian Pleistocene species, Eucrassatella lomitensis [Oldroyd, 1924], is incorrectly described. The species in fact lacks ventral crenulations, both on the holotype [Coan 1984;Nigel Hughes, personal communication 2015] and on specimens examined by the author; see Fig. 2E).
The genus Eucrassatella was proposed by Iredale (1924) to replace Lamarck's (1799) Mactra-based Crassatella and Krüger's (1823) fossil genus, Crassatellites. The Recent Australian species, Crassatella kingicola Lamarck, 1805 (Fig. 2C), was designated the type species. Having incorrectly inferred that Iredale (1924) restricted Eucrassatella only to large living Australian species, Stewart (1930) proposed that the genus should also encompass tropical American taxa and set forth two diagnostic characters for the genus: a resilifer extending to or nearly to the vmHP and a smooth inner ventral margin.
Complicating the assignation of Eucrassatella to American taxa is a diagnostic character overlooked by all except Vokes (1973): the dorsal truncation of the left anterior cardinal tooth by the lunule. For most species of Cretaceous and Paleogene Crassatella and fossil and modern Eucrassatella from Australia and New Zealand (and some species of the genus Spissatella Finlay, 1926), the dorsal end of the lunule separates the left anterior cardinal tooth from the beak. In contrast, for most American species assigned to Eucrassatella (as well as a few species of southwestern Pacific Spissatella), the left anterior cardinal tooth extends from the vmHP to the beak with little or no separation by the lunule (compare Fig. 2C 1 and 2D 1 ).
Stewart (1930) noted the flattened umbo of American crassatelline species, contrasting it with the rounded umbo on a preponderance of Australian and New Zealand species (compare Fig. 2C and 2D). Rarely, specimens of the Australian Eucrassatella kingicola have a flattened umbo, but contrary to Stewart's (1930) inference based on a single specimen, most do not. A flattened umbo is therefore more diagnostic of American crassatellines than credited by Stewart (1930).
Having rejected the diagnostic value of a flattened umbo, Stewart (1930) (Stewart 1930;Darragh 1965;Coan 1984). The two orientations need not be the same. Eucrassatella species from Australia and New Zealand, for example, have orthogyrate umbones but the beaks may be prosogyrate (E. kingicola) or orthogyrate (E. pulchra [Reeve, 1842]). Stewart (1930) called attention to the strongly opisthogyrate flattened umbo of the modern American species, Eucrassatella gibbosa (Sowerby, 1832) (Fig. 2D), and designated it the type species of a new subgenus, Hybolophus, to which he also assigned the species E. antillarum (Reeve, 1842) (Fig. 3A, B), with a less opisthogyrate umbo, and a rumored Peruvian Miocene species, likely Spieker's (1922) account of E. nelsoni (Grzybowski, 1899). Oddly, Stewart (1930) also implied that fossil American Eucrassatella lacked opisthogyrate umbones, a claim belied by the existence of the Miocene species E. berryi (Spieker, 1922) and E. elassa Woodring, 1982 (Fig. 3C, D). Olsson (1932) proposed that Eucrassatella (Hybolophus) include Miocene species of Eucrassatella similar to E. gibbosa. Darragh (1965) added to E. (Hybolophus) any American Neogene species with an orthogyrate umbo, e.g., E. antillarum and allied species. Woodring (1982) divided E. (Hybolophus) into an "elongate" group, represented by E. gibbosa with its strongly opisthogyrate umbo, and a "high" group, represented by E. antillarum with its weakly opisthogyrate or orthogyrate umbo. Species of the "gibbosa" group are strongly inflated ventral to the umbo with the inflation diminishing rapidly anteriorly and steadily posteriorly. Species of the "antillarum" group are only modestly inflated ventral to the umbo and hardly less so posteriorly as far as the posterior ridge.
Eucrassatella mediamericana (Brown and Pilsbry, 1913), an early Miocene (Kirby et al. 2008) species from the La Boca Formation of Panama, was retained by Woodring (1982) in Eucrassatella sensu stricto. Inspection of Woodring's (1982) specimens of E. mediamericana (Fig. 3E), as well as specimens of early Miocene (Bryant et al. 1992) E. chipolana (Dall, 1903) from the Chipola Formation of Florida, middle Miocene (Huddlestun 1984) E. densa (Dall, 1900) from the Shoal River Formation of Florida, a late Miocene unnamed crassatelline from the Gatun Formation of Panama (Woodring 1982) (Fig. 3F), Pliocene (Donovan 1998) E. jamaicensis (Dall, 1903) from the Bowden Formation of Jamaica (Fig. 3G), and examination of high-resolution photographs of the early Pleistocene (McGregor 2011) E. mansfieldi MacNeil, 1936 from the Waccamaw Formation of South and North Carolina shows that all share external and internal characters with the modern species, E. speciosa (Adams, 1854) (Fig. 3H). Specimens of E. speciosa differ from those of E. antillarum principally in this regard: the external sculpture of the latter consists of irregular commarginal growth lines, whereas the former exhibits regularly spaced commarginal ribs over much of the shell (compare Fig. 3A 1 and 3H 1 ).
A discovery in the collections of the Florida Museum of Natural History of undescribed crassatellines from the lower Oligocene (Bryan 1991) Suwannee Limestone of Florida, herein named Crassatella portelli sp. nov., casts new light on the classification of North American crassatellines. Suwannee specimens have the crenulate inner ventral margin and the widely divergent posterior pseudocardinal ridge typical of Crassatella and resilifers that extend either half-way or nearly to the vmHP but, typical of the American taxa Eucrassatella and E. (Hybolophus), a flattened umbo and left anterior cardinal tooth that extends nearly or entirely to the beak (  (UF 18403, UF 18427, UF 18458, UF 18586), probably referable to the species C. ocordia Harris, 1951, have a rounded umbo, crenulate ventral margin, truncated left anterior cardinal tooth, and in most cases, a resilifer extending less than entirely to the vmHP-i.e., a typical Crassatella in nearly every respect.
The Ocala, Suwannee, Chipola, Gatun, and Bowden crassatellines and Recent Eucrassatella speciosa constitute a graded morphological series ranging from typical Paleogene Crassatella to the New World Neogene crassatellines of Woodring's (1982) "high" group. They stand apart morphologically from the strongly inflated "elongate" group of Woodring (1982). In accord with MacNeil's (1936: 528) prescient suggestion and Ward and Blackwelder's (1987: 151) implicit recommendation, a new genus, herein named Kalolophus gen. nov., is proposed for the "high" group, with Crassatellites chipolanus Dall, 1903 designated the type species. The paciphilic genus Hybolophus, comprising American species of Woodring's (1982) "elongate" group, which ranges from at least the early Oligocene (H. disenum sp. nov.) to the Recent (H. gibbosus), is elevated to full generic rank. Per Darragh's (1965) emphatic declaration, the name Eucrassatella is applied only to Australian and New Zealand species.
Diagnostic characters for the above-mentioned crassatelline taxa (Table 1) are the basis for generic assignments used in this paper. It should be understood that one or more specimens of a species, particularly species that are transitional between genera, can exhibit a character state contrary to that which otherwise would establish its generic placement.
Exceptions to this generic arrangement are large thickshelled crassatellines from the lower to middle Miocene Navidad Formation of central Chile (Tavera 1979) and Chattian to Langhian material from southern Peru. These specimens have a rounded and orthogyrate umbo, a resilifer that extends to the vmHP, a crenulate inner ventral margin on larger specimens, and a left anterior cardinal tooth separated from the beak by the lunule. This set of character states fits neither Crassatella, Eucrassatella, Hybolophus, nor Kalolophus. By reason of this character suite (Table 1) and associated characters described in the systematics section, a new genus is proposed for Philippi's (1887) species and two new species from southern Peru: Tilicrassatella gen. nov., with the type species designated as Crassatella ponderosa Philippi, 1887.  (Spieker, 1922), USNM 562399, early Miocene, lower Zorritos Formation, Zorritos, Peru; exterior of left valve (C 1 ), dorsal view of paired valves (C 2 ), anterior at right. D. Hybolophus elassa Woodring, 1982, USNM 647424, late Miocene, Gatun Formation, Panama; exterior of left valve. E. Kalolophus mediamaricanus (Brown and Pilsbry, 1913), USNM 647421, early Miocene, La Boca Formation, Panama; exterior of right valve. F. Kalolophus sp., USNM 647423, late Miocene, middle Gatun Formation, Panama; exterior (F 1 ) and interior (F 2 ) of left valve. G. Kalolophus jamaicensis (Dall, 1903) A few American Neogene taxa, assigned tentatively to Hybolophus, do not fit the definitions of these five genera. The Quaternary Californian species, H. fluctuatus, has a widely divergent posterior ridge, a foreshortened posterior, and dorsally truncated left anterior cardinal tooth (Fig. 2E), one or more characters of which are shared with Pacific specimens of the late Eocene or early Oligocene northern Peruvian H. disenum, the early Miocene (Marks 1951) or early middle Miocene (Deniaud et al. 1999) Ecuadorian species, H. carrizalensis (Marks, 1951), the late Miocene (Landau et al. 2012) Ecuadorian species H. picaderus (Olsson, 1964) (Fig. 5A, B), and Caribbean specimens of the late Miocene (Bold 1966) Colombian species, H. tuberus (Olsson, 1964) (Fig. 5C). The phylogenetic position of these veneriform morphological outliers is addressed in the systematics and discussion sections. Description.-Shell to 40 mm long, trapezoidal, not inflated, L:H ratio 1.3, T:H ratio 0.2, maximum inflation ventral to beak. Anteriodorsal and posteriodorsal margins slightly concave; anterior margin broadly rounded, posterior margin weakly produced, broadly truncate; ventral margin evenly rounded. Strongly angular posterior ridge diverging 20−25°, secondary posterior ridge 5° from posteriodorsal margin. Lunule and escutcheon elongate, equally wide, both about 40% length of respective dorsal margins. Beak prosogyrate. Umbo flattened, slightly prosogyrate to slightly opisthogyrate, with widely spaced and pronounced commarginal ribs; remainder of exterior with closely spaced commarginal ribs; ribs weak or absent posterior to primary posterior ridge. Resilifer extending about half to full distance from beak to vmHP. Left anterior cardinal tooth narrow, wedge-shaped, inclined anteriorly 30-40°, extending to beak or nearly so. Left posterior cardinal tooth wedge-shaped, orthocline, extending to beak or nearly so. Receptor pit obsolete. Right anterior pseudocardinal tooth coalescent with lunule margin. Right cardinal tooth thick, wedge-shaped, inclined anteriorly 15−20°. Right posterior pseudocardinal tooth lamellar, diverging 30−60° from midpoint of cardinal tooth, passing resilifer ventrally. Inner ventral margin crenulate.
Remarks.-Specimens of Crassatella portelli exhibit characters of Crassatella (widely diverging right posterior Crassatella neorhynchus (Olsson, 1931)    Description.-Shell to 60 mm long, trapezoidal, inflated, L:H ratio 1.3, T:H ratio 0.3−0.4, maximum inflation posterior to beak. Anteriodorsal and posteriodorsal profiles nearly straight, anterior margin sharply rounded, posterior margin broadly truncate, slightly produced in largest specimens; ventral margin evenly rounded or slightly angular. Remarks.-External characters of East Pisco Basin specimens match those of the Talara Basin specimens; internal features of the latter are not visible. Large specimens are more elongate than inferred by Olsson (1931) (Fig. 6A). The stratigraphic placement of Crassatella neorhynchus in Chira valley is unambiguous, although the accepted age of the Chira shales has changed from early Oligocene (Olsson 1931) to late Eocene (Higley 2004). The stratigraphic placement of C. neorhynchus near Mancora is less certain. The lower Oligocene Mancora Formation, cited by Olsson (1931) (Olsson, 1964), Ecuador. A. USNM 643826, holotype, Picaderos Formation, Picaderos; exterior (A 1 ) and interior (A 2 ) of left valve. B. USNM 645393, paratype, Mompiche-Portete; interior of right valve. C. Hybolophus tuberus (Olsson, 1964), USNM 643827, holotype, Tubera Formation, Tubera-Puerto Caiman, Colombia; exterior of left (C 1 ) and right (C 2 ) valves, dorsal margin of paired valves (C 3 ), anterior at right. Scale bars 10 mm. as the type horizon for C. neorhynchus, is not mapped near Mancora, despite being described from that stratotypic locality (Palacios 1994). Olsson's (1931) description of the area from which specimens of C. neorhynchus were collected corresponds to outcrop mapped as the upper Eocene Mirador Formation (Palacios 1994 Material.-B8329, six broken specimens. Remarks.-Despite the absence of visible hinge characters, the paired specimen of Crassatella pedroi can be distinguished from specimens of C. neorhynchus by its anterior compression (Fig. 6G 1 ) and commarginal ribs extending well beyond the umbo (Figs. 6G 2 , G 3 ). Furthermore, the lunule is narrowly cordate (Fig. 6G 1 ) and both it and the escutcheon are proportionally longer than their counterparts on specimens of C. neorhynchus. The widely spaced umbonal ribs distinguish C. pedroi from the slightly younger C. rafaeli sp. nov. from the Otuma depositional sequence.
Diagnosis.-Umbo with closely spaced commarginal ribs. Resilifer extending half way or entirely to vmHP. Large specimens with crenulate inner ventral margin.
Remarks.-Specimens of Crassatella rafaeli are distinguished from those of the slightly older C. neorhynchus and C. pedroi by their closely spaced commarginal ribs on the umbo (Fig. 7F 1 ). Large specimens of C. rafaeli become increasingly elongate and produced (Fig. 7F 2 , G). The crenulate ventral margin on the largest specimens is also visible externally as anteriodorsally-oriented parallel subsurface lineations that fade dorsally (Fig. 7C). Irrespective of shell size, some resilifers extend nearly fully from the beak to the vmHP (Fig. 7E, G). In contrast, for all specimens of all species of the younger Tilicrassatella, resilifers are broader and always extend entirely to the vmHP or nearly so.    Remarks.-Like specimens of Crassatella rafaeli, those of Tilicrassatella ponderosa have closely spaced commarginal ribs on the umbo. Specimens of T. ponderosa differ from those of T. torrens sp. nov. by their anterior inflation and markedly inequilateral valves, resulting in a steeply descending anteriodorsal margin. The only known specimen of T. sanmartini is more elongate and less rostrate posteriorly and its resilifer is directed in a mostly posterior direction. An anterioventral fragment of a juvenile left valve of T. ponderosa (UWBM 101845) shows no evidence of ventral crenulations. A paired Miocene specimen of about 43 mm length (SGO. PI. 115), previously termed Crassatella medinae (Philippi, 1887) by Frassinetti (1974) and recently photographed by Leonardo Pérez (Universidad Austral de Chile, Valdivia), may be a juvenile example of T. ponderosa. Small specimens with exposed hinge plates are needed to make such a determination.
Remarks.-This specimen differs from its sympatric congeners, Tilicrassatella ponderosa and T. torrens, by its extreme posterior elongation and posteriorly directed resilifer, features shared by many taxa attributed to Bathytormus, including the Eocene Argentinian Bathytormus longior (Ihering, 1897). One specimen of B. longior in particular, figured by Santelli and del Río (2014: fig. 6.7), greatly resembles the Peruvian specimen, although it is proportionally longer (L: H 1.7) with a less robust escutcheon, lunule, and umbo.
In addition to a full-length resilifer and crenulated inner ventral margin, Chavan (1939) attributed to Bathytormus a rostrate posterior, and Wingard (1993), a pronounced inequilaterality at all growth stages. Neither of the these characters describes the Peruvian specimen.
Chattian-Burdigalian specimens from the Guadal Formation of southern Chile attributed questionably to Bathytormus longior by Frassinetti and Covacevich (1999) are less elongate and more truncate posteriorly than specimens of B. longior described by Ihering (1897) and Santelli and del Río (2014). The Guadal specimens, with so few umbonal commarginal ribs (9−10 per 10 mm radially), are not examples of the contemporaneous Tilicrassatella, all three species of which have 17−25 umbonal ribs in the same radial interval.
Genus Hybolophus Stewart, 1930  Remarks. -Stewart (1930) equivocated in his description of Hybolophus, focusing finally but with reservations on the opisthogyrate beaks (umbones?) of the new taxon. Thus, Hybolophus is better viewed through the descriptive prism of the type species, H. gibbosus. MacNeil (1936) and Ward and Blackwelder (1987) consequently leaned towards excluding less inflated and less posteriorly produced species that are herein placed in Kalolophus. Similarly, Marks (1951) expressed uneasiness about placing in Hybolophus his early Miocene species from Ecuador, which, being only slightly produced posteriorly and possessing a widely divergent and single posterior ridge, presents a veneriform profile. Marks (1951) would have been uneasier yet if Olsson (1964) had already published accounts of his late Miocene Ecuadorian Eucrassatella picadera and Colombian E. tubera. For now, considering the scarcity of exposed hinge plates and specimens of veneriform crassatellines, it is premature to assign the veneriform crassatellines to a separate genus or subgenus; they are included here within a more broadly conceived Hybolophus.
Stratigraphic and geographic range.-Upper Eocene to Recent, western North and South America, Caribbean.
Hybolophus disenum sp. nov. Remarks.-Field notes of Woodring (1958: 8-9) put locality WP27 and the type locality of Hybolophus disenum, WP 28, close to the town of Mancora in soft sandstone beds of the basal Heath Formation, which were observed to overlie massive sandstone and conglomerate of the Mancora Formation. The two formations are lower and upper Oligocene, respectively (Higley 2004). Neither formation, however, is now mapped near Mancora (Palacios 1994;also, César Chalcatana [INGEMMET], personal communication 2015). Woodring's (1958) localities do plot in an area mapped by Palacios (1994) as the upper Eocene Mirador Formation. Adding to the confusion, distinctive black carbonaceous conglomerate near Mancora is attributed to the Mancora Formation by both Olsson (1931) and Palacios (1994).
Turritella woodsi Lisson, 1925 (= T. conquistadorana Hannah andIsraelsky, 1925;see DeVries 2007) is found at locality WP 27. The gastropod also occurs about 30 km northeast at localities WP 31 and WP 32, near the mouth of Quebrada Plateritos (= Quebrada Culebra), an area mapped as the Mancora Formation by Palacios (1994). In southern Peru, T. woodsi is found in Priabonian to Chattian strata of the East Pisco Basin (DeVries 2007). Thus, specimens of Hybolophus disenum could have an age between late Eocene and latest Oligocene. The stratigraphic and paleontological evidence for locality WP 28 indicates a late Eocene or early Oligocene age.
In a related aside, Deniaud et al. (1999) assigned the Ecuadorian Subibaja Formation, in which are found specimens of Turritella woodsi and Marks's (1951) type specimens of Hybolophus carrizalensis, to the Langhian. In the Talara and East Pisco basins of Peru and Progreso Basin of Ecuador, T. woodsi-bearing beds underlie beds with specimens of the Langhian T. infracarinata Grzybowski, 1899(Marks 1951DeVries 2007). Therefore, H. carrizalensis might have an Oligocene to early Miocene age, not Langhian.
Hybolophus disenum may be contemporaneous or nearly so with Crassatella neorhynchus, being from the same out-crop area west of Mancora (Olsson 1931;Woodring 1958). Specimens of H. disenum have a flattened umbo, typical of Hybolophus but absent in Crassatella; are closer to being equilateral with dorsal margins descending ventrally less steeply than on specimens of C. neorhynchus; and have a less angular and more widely divergent posterior ridge. The damaged left anterior cardinal tooth (Fig. 9C) may be separated from the beak by the lunule, a feature typical of Crassatella but present in more than one species of Hybolophus. The inner ventral margin of H. disenum is not preserved and so cannot be compared with the crenulate margin in large specimens of East Pisco C. neorhynchus.
Stratigraphic and geographic range.-Upper Eocene or lower Oligocene, Mirador or Mancora Formation, northern Peru.
Hybolophus terrestris sp. nov.  Hybolophus maleficae sp. nov. Etymology: From Latin malefica, witch; referring to the "La Bruja" (witch in Spanish), vertebrate level (Muizon and DeVries 1985), in which many specimens of this species were found. Type material: Holotype, complete adult left valve with hinge exposed: UWBM 101858, B8338, L 44.5, H 38.8, T 11.4 Description.-Shell to 70 mm long, trapezoidally ovate, moderately inflated, posterior variably elongate and produced, beak located two-fifths of length from anterior end, L:H ratio 1.2, T:H ratio 0.3, maximum inflation ventral or anterior to beak. Anteriodorsal and posteriodorsal profiles straight or slightly concave. Anterior margin rounded to bluntly rounded, ventral margin rounded and slightly angled, posterior margin bluntly truncate or produced. Moderately angular primary posterior ridge diverging 20° from posteriodorsal margin. Weak secondary posterior ridge nearly coincident with escutcheon margin. Lunule cordate, half the length of anteriodorsal margin. Escutcheon two-thirds the length of posteriodorsal margin, narrow, one-third width of lunule. Beak prosogyrate. Umbo flattened, prosogyrate to opisthogyrate with widely spaced commarginal ribs, latter sometimes extend ventrally across anterior of valve; remainder of exterior with irregularly spaced commarginal growth lines. Resilifer narrow. Left anterior cardinal tooth short, wedge-shaped, inclined anteriorly 30-35°, separated from beak by lunule. Left posterior cardinal tooth narrower, nearly orthocline. Right anterior pseudocardinal tooth short, diverging from lunule margin. Right cardinal tooth short, straight, wedge shaped, inclined anteriorly 20°. Right posterior pseudocardinal tooth lamellar, diverging 20° posterioventrally from midpoint of cardinal tooth. Inner ventral margin smooth.
Emended diagnosis.-Shell inflated. Anteriodorsal margin straight to slightly convex. Posterior produced. Primary and secondary posterior ridges diverging by more than 10°, latter close to margin of escutcheon.
Remarks.-Hybolophus nelsoni encompasses Miocene specimens from several localities in northern Peru. The oldest are attributed to the lower Miocene lower Zorritos Formation by Spieker (1922), although the specimens may have been collected from the middle Miocene Cardalitos Shale (Grzybowski 1899;Woods 1922;Hanna and Israelsky 1925;Olsson 1932). The youngest were collected from the upper Miocene Tumbez Formation (Olsson 1932).
Previous authors offer conflicting conclusions about the relationship between Hybolophus nelsoni and H. gibbosus. Quaternary specimens from northern Peru have closely spaced primary and secondary posterior ridges, producing an exceptionally narrow posterior truncation (i.e., the posterior nearly comes to a point), a deeply concave posteriodorsal margin, and an exceedingly opisthogyrate umbo; these are assigned to H. gibbosus. Specimens of H. nelsoni have a wider divergence between the primary and secondary posterior ridges, with the latter close to the margin of the escutcheon. The result is a broader posterior truncation, a straighter posteriodorsal margin, and a less opisthogyrate umbo. Hybolophus gibbosus (Sowerby, 1832)   Remarks.-Specimens of Hybolophus gibbosus are found in northern Peru on the Mancora, Talara, and Lobitos tablazos of early, middle, and late Pleistocene age, respectively, and in the Golf Course Member of the upper Pliocene Taime Formation ( DeVries 1986DeVries , 1988. A large unfigured specimen is reported from the Pliocene Canoa Formation of southwestern Ecuador (Pilsbry and Olsson 1941). Hybolophus aviaguensis peruviana (Olsson, 1932) from the middle Miocene (Dunbar et al. 1990) Montera Formation of the northern Peruvian Sechura Basin has an equally concave posteriodorsal margin and equally strongly opisthogyrate umbo, but the beak is more anteriorly located and the anterior margin consequently is more tightly circular in profile.  Dall, 1903; Crassatellites densus Dall, 1900; Crassatellites (Scambula) jamaicensis Dall, 1903;Eucrassatella mansfieldi MacNeil, 1936;Crassatellites mediamericanus Brown and Pilsbry, 1913; Crassatella speciosa Adams, 1854.
Diagnosis.-Shell trapezoidal. Posterior variably produced, bluntly truncate. Weakly to moderately inflated ventral to beak, flattening towards posterior ridge. Umbo orthogyrate to slightly opisthogyrate. Escutcheon often wider than lunule. Resilifer extending nearly or entirely to ventral margin of hinge plate. Left anterior cardinal tooth not separated from beak by lunule. Inner ventral margin not crenulate.
Stratigraphic and geographic range.-Lower Miocene to Recent, Florida, Caribbean, and western North and South America.
Remarks.-Kalolophus antillarum had become established in the Pacific and Atlantic during the Pliocene . Specimens of the modern sympatric crassatelline in the Pacific Ocean, Hybolophus gibbosus, are more inflated anteriorly, more constricted posteriorly, and their two pos-terior ridges are more closely spaced, thereby creating a posterior margin that is almost pointed. Specimens of the modern sympatric congener in the Atlantic Ocean, K. speciosus, resemble those of K. antillarum in most respects, but the exterior is entirely covered with regularly spaced and pronounced commarginal ribs (Fig. 3H 1 ).
Stratigraphic and geographic range.-Pliocene-Pleistocene, Gulf of California, Venezuela. Recent, Gulf of California to Guayas, Ecuador, and Venezuela.

Discussion
Crassatelline phylogeny.-Crassatella neorhynchus from the East Pisco Basin inhabited the southern end of a Lutetian tropical faunal province centered in the Talara Basin (Olsson 1931;Rivera 1957;DeVries 2004). The East Pisco C. pedroi sp. nov., seven million years younger but nearly identical, is the likely descendant of C. neorhynchus. Crassatella rafaeli sp. nov., living in the East Pisco Basin less than one million years after C. pedroi, differs significantly from its East Pisco predecessors and most other American crassatellines because of its closely spaced commarginal umbonal ribs. A phylogenetic connection with Australian/New Zealand crassatellines cannot be ruled out, since many southwestern Pacific taxa have similar umbonal sculpture. The only contemporaneous southwestern Pacific species that, like C. rafaeli, also had a crenulate inner ventral margin was the New Zealand Eocene Triplicitella australis (Hutton, 1873) (Table 1). However, T. australis has a broad resilifer that always reaches the vmHP, a character present only on some specimens of C. rafaeli.
The only western American taxa with umbonal ribs identical with those of Crassatella rafaeli are species of the East Pisco and Chilean Tilicrassatella gen. nov. The uniquely shared umbonal sculpture of Priabonian C. rafaeli and Chattian−Burdigalian species of Tilicrassatella indicates that Tilicrassatella species may have evolved from C. rafaeli. Such a scenario is consistent with the shift from a full-length resilifer in just a few specimens of C. rafaeli to its invariable presence in species of Tilicrassatella.
The shape of Tilicrassatella species diversified dramatically in south-central Peru during the early Miocene. The gibbose species, T. ponderosa, flattened and nearly equilateral species, T. torrens sp. nov., and posteriorly elongate species, T. sanmartini sp. nov. mimic comparably shaped crassatelline genera Hybolophus, Kalolophus gen. nov., and Bathytormus, respectively. Before the end of the middle Miocene, all species of Tilicrassatella were extinct.
Species of crassatelline genera in Argentina fared little better than species of Tilicrassatella. Entirely different than the crassatelline fauna of western South America, the Argentine fauna (Santelli and del Río 2014) included veneriform and strongly crenulate Crassatella species that persisted until the late Miocene; early Miocene taxa as-signed by the two authors to Talabrica Iredale, 1924 and Spissatella, presumed to be emigrants and immigrants to and from New Zealand, respectively; and species assigned to Bathytormus, present from the Paleocene through the early Miocene. Every Argentinian crassatelline taxon was extinct by the end of the Miocene except Riosatella, an extant monospecific Crassatella-like genus of uncertain phylogenetic affinity (Vokes 1973).
Crassatelline lineages in the circum-Caribbean and tropical western America have proven longer lived than those in southern South America. Kalolophus, which probably evolved in Florida during the late Oligocene from the morphologically transitional Crassatella portelli sp. nov., has occupied the circum-Caribbean from the early Miocene (K. chipolanus) to the present (K. speciosus). Populations of K. antillarum, which appeared in the Caribbean during the Pliocene , had spread into the equatorial Pacific Ocean before the late Pliocene shallow-water closure of the Isthmus of Panama Leigh et al. 2014). Kalolophus densus, occurring in middle Miocene strata of Florida, exemplifies a morphology transitional between that of the small and thin K. chipolanus and the large, thick, and posteriorly elongate species from the Atlantic margin of North America assigned by Ward and Blackwelder (1987) to Marvacrassatella (Table  1). The oldest species of Marvacrassatella is the middle Miocene (Andrews 1976) M. melinus (Conrad, 1832); the youngest, the early Pleistocene (Blackwelder 1981) M. kauffmani Ward and Blackwelder, 1987 A phylogenetic tree diagram for crassatelline genera discussed in this paper is presented in Fig. 11. Notably different than other treatments is the separation of large Neogene taxa into four independent lineages. The arrangement of New Zealand genera is adopted from Collins et al. (2014) except that the late Eocene species, Spissatella media (Marwick, 1926), is transferred to Eucrassatella based on a dorsally truncated left anterior cardinal tooth present on a specimen from the Geology Department, University of Otago (New Zealand) (OU8648, McCullough's Bridge; Katie Collins, personal communication 2015). The New Zealand species, E. subobesa (Marshall and Murdoch, 1919) and Australian E. maudensis (Pritchard, 1903) and (suggested herein) E. oblonga (Tenison Woods, 1876), all late Oligocene species (Darragh 1965;Collins et al. 2014) and all with a left anterior cardinal tooth extending to the beak, are candidates for the transitional taxon connecting late Eocene Eucrassatella and late Oligocene to middle Miocene Spissatella, some species of which have a left anterior cardinal tooth extending to the beak (Table 1).
The two sympatric lineages in the circum-Caribbean and tropical western America exhibit a number of convergent characters, including a flattened umbo, a full-size resilifer, a smooth inner ventral margin, and a dorsally extended left anterior cardinal tooth. These characters had appeared in Hybolophus species by the late Eocene or early Oligocene but were not all present in Kalolophus species until after the early Oligocene. Only one of the three characters was fully realized in the relatively short-lived latest Oligocene to middle Miocene Tilicrassatella.   (Table 1). The resilium, which nestles between facing resilifers, exerts an outward force to maintain a shell gape, counteracting a force exerted by the two adductor muscles to pull the valves closed (Kauffman 1969). The resilium also helps align the two valves and resists shear forces parallel to the plane of commissure that might arise while burrowing or being predated, as do the cardinal teeth (Stanley 1970). Ventral crenulations may also assist in aligning valves; they can also impede passage of sediment and debris when the valves are slightly open (Vermeij 2013), resist compressive pressure applied by predators (Thomas 2013), and thwart attempts by predators to drill or enter the bivalve through the valve margin (Vermeij 1983).
Two specimens with an anomalous cardinal tooth deserve mention. A left valve of the species Tilicrassatella torrens (UWBM 101852; see Fig. 8G 1 ) has a thin anterior cardinal tooth and anterior groove extending entirely to the beak, separated from a gracile lunule, much different than the truncated anterior cardinal tooth and robust lunule of all other left valves of all Tilicrassatella species. An extended left anterior cardinal tooth and anterior groove is also present on a specimen (USNM 618238; see Fig. 9B 2 ) of the species Hybolophus disenum, in which the two left cardinal teeth remain separated by an exceptionally wide socket terminating bluntly at the beak. Perhaps the expression of such phenotypic oddities reflects an underlying genetic propensity for an extended left anterior cardinal tooth, eventually realized as a ubiquitous character in the genera Spissatella and Kalolophus and present in the majority of Hybolophus taxa.
The late Eocene−Oligocene trend in crassatellines towards a ventrally elongated resilifer (and hence larger resilium) and dorsally extended left anterior cardinal tooth would have enhanced an animal's ability to align its valves and resist rotational forces in the plane of commissure. Ventral crenulations, already small in Crassatella, might have become less useful. The loss of crenulations might also signify that the passage of foreign material and predators across the shell margin had become less of a problem. A shift among crassatellines to burrowing in finer-grained and more cohesive sediment would be consistent with the morphological changes seen. Such sediment, probably mediumand fine-grained sand, would be encountered more often on the continental shelf or in lagoons and embayments than in the surf zone or well above wave base.
In the East Pisco Basin, specimens of Crassatella neorhynchus and C. pedroi, often paired, are found in coarse-grained and crossbedded transgressive sandstone beds of the upper Eocene Paracas and Otuma sequences, in the former case associated with the non-siphonate bivalve genera Nucula, Ostrea, Glycymeris, Cucullaea, and Cyclocardia. Specimens of Tilicrassatella species, often paired, similarly occur in coarse-grained bioclastic sandstone and gravelly sandstone of the uppermost Oligocene−lower Miocene Chilcatay depositional sequence and similar sediments of the basal Pisco depositional sequence, either in the basal transgressive unit, gravelly bioclastic intervals, or atop foreset beds. The Tilicrassatella specimens are associated with byssate, cemented, and non-siphonate Ostrea, Glycymeris, Atrina, and pectinid bivalves. Specimens of C. rafaeli, present throughout the Otuma depositional sequence and usually paired, occur in coarse-grained sandstone lenses interspersed with lenses of mudstone and large stranded tree trunks. Associated bivalves include Ostrea, Cyclocardia, Corbula, lucinids, and venerids. Based solely on sedimentological evidence, all the above-mentioned sediments are interpreted to have been deposited in shallow and agitated water, subject to reworking by waves, wave-induced currents, or in the case of beds with C. rafaeli, estuarine currents.
The oldest known and possibly transitional species of Hybolophus, the late Eocene or early Oligocene H. disenum, has a matrix of poorly sorted, bioclastic, coarse-grained sand and gravel, consistent with a high-energy depositional setting. Langhian and Serravallian specimens of Hybolophus maleficae from the East Pisco Basin, in contrast, occur in massive, bioturbated, and moderately well-sorted sandstone above basal transgressive deposits of the Pisco depositional sequence, associated with in situ paired valves of Chione and in situ paired valves and concentrated lags of disarticulated Mulinia and Anadara sechurana. A more diverse assemblage of in situ and transported venerids (Chione, Chionopsis, Dosinia, Amiantis) is associated with Tortonian paired and disarticulated valves of H. maleficae in winnowed tuffaceous sandstone of the Sacaco Basin and Messinian paired and disarticulated valves of H. terrestris in rippled and scoured tuffaceous silty sandstone in the East Pisco Basin. These middle and upper Miocene sediments, which in the Messinian example contain pelagic diatoms and clupeoid fish scales, are interpreted to have been deposited at inner and mid-shelf depths above storm wave base.
Modern specimens of Hybolophus gibbosus, H. antillarum, and H. fluctuatus are found in sand and less often gravel in water depths of 5−110 m and 5−206 m, respectively (Coan 1984;Coan and Valentich-Scott 2012), depths which are consistent with the occurrence of H. gibbosus in the upper Pliocene Taime Formation in massive, bioturbated, and moderately well-sorted medium-grained sandstone, associated with innumerable molds of Mulinia and Pitar valves (DeVries 1986). Single valves of H. gibbosus also constitute a small fraction of the bioclastic debris on the surface of Mancora and Lobitos tablazos of northern Peru and northern Peruvian modern beaches.
In the circum-Caribbean region, the Crassatella-bearing upper Eocene Ocala limestone consists of packstone, wackestone, and grainstone with abundant foraminifera, echinoids, and mollusks (Carr and Alverson 1959;Scott et al. 2001), indicating a shallow high-energy depositional environment (Portell and Hulbert 2011). Shallow to moderate depth is indicated for the lower Oligocene Suwannee limestone and its fauna, including the transitional Crassatella portelli (Carr and Alverson 1959). The sandy and clayey limestone of the lower Miocene Chipola Formation, containing the oldest Kalolophus, K. chipolanus, are attributed to a nearshore shelf environment (Bryant et al. 1992) or, in the case of the Farley Creek deposits with specimens of K. chipolanus, a quiet-water embayment (Brown et al. 2013). Specimens of modern K. speciosus in collections of the Department of Invertebrate Zoology at the Smithsonian Museum of Natural History (e.g., USNM 833713, USNM 855600) were collected at depths of 10−60 m. The quieter waters successively occupied by Floridian crassatellines is reflected in the morphological transition from Crassatella to Kalolophus involving the cardinal teeth, resilifer, and ventral crenulation.
A similar pattern of crassatellines shifting to deeper waters during the Eocene and Oligocene is not so clearly demonstrated with the southwestern Pacific fossil record, since no Paleogene Crassatella have been found in Australia and New Zealand. It has been argued that species of Spissatella, a largely Neogene clade with full-sized resilifers, left anterior cardinal teeth extending to the beak, and smooth ventral margins, occupied deeper shelf environments than older and contemporaneous species of Eucrassatella, which retain truncated left anterior cardinal teeth (Beu and Maxwell 1990;Collins 2013).
Why crassatellines should have had less evolutionary success in shallow and/or more agitated environments following the Eocene is not known. Ocean cooling may have played an indirect role (Berggren and Prothero 1992), or a shift from predominantly benthic to planktonic primary production (Vermeij 2011), or the rapid diversification at the generic level of competing infaunal siphonate bivalves (Stanley 1968). Why Neogene crassatellines were able to occupy deeper or quieter water niches that they had not previously favored is also not known.

Conclusions
Evidence has been presented that most American, Australian, and New Zealand large crassatelline bivalves evolved from Crassatella forbears in one of four independent lineages, represented by the genera Eucrassatella, Hybolophus, and two new genera, Tilicrassatella and Kalolophus. During the late Eocene and early Oligocene, convergent changes in hinge structure and the ventral margin in two, three, or all four lineages coincided with and might explain the greater success of Neogene and modern large crassatellines in deeper or quieter waters, whether on the continental shelf or in lagoons and bays.
The classification scheme presented in this paper excludes a number of crassatellines and proposes phylogenetic connections that would bear testing by other means, including statistical cladistic methodologies and a comparison of mitochondrial DNA sequences. For example, comparing the genetic sequences of extant circum-Caribbean and tropical western American species (H. gibbosus, H. fluctuatus, K. speciosus, K. antillarum) would constitute an independent test of the phylogeny of Hybolophus and Kalolophus proposed here. Similar tests could be applied to extant Crassatella-like species found in the western Pacific, the northeastern Pacific (Hybolophus fluctuatus), the southwestern Atlantic, and South Africa. It is not known if these taxa are related to one another, are descendants of Neogene genera, or are independent lineages derived from Paleogene ancestors. Lastly, a stratophenotypic cladistic analysis of the diverse fauna of fossil and modern crassatellines in Australia, including a Crassatella of Miocene age (Darragh 1965), might show that the phylogeny of Australian crassatellines is more complex than previously perceived, if independent crassatelline lineages truly arose as readily as this study indicates.