Cenozoic Methane-Seep Faunas of the Caribbean Region

We report new examples of Cenozoic cold-seep communities from Colombia, Cuba, the Dominican Republic, Trinidad, and Venezuela, and attempt to improve the stratigraphic dating of Cenozoic Caribbean seep communities using strontium isotope stratigraphy. Two seep faunas are distinguished in Barbados: the late Eocene mudstone-hosted ‘Joes River fauna’ consists mainly of large lucinid bivalves and tall abyssochrysoid gastropods, and the early Miocene carbonate-hosted ‘Bath Cliffs fauna’ containing the vesicomyid Pleurophopsis, the mytilid Bathymodiolus and small gastropods. Two new Oligocene seep communities from the Sinú River basin in Colombia consist of lucinid bivalves including Elongatolucina, thyasirid and solemyid bivalves, and Pleurophopsis. A new early Miocene seep community from Cuba includes Pleurophopsis and the large lucinid Meganodontia. Strontium isotope stratigraphy suggests an Eocene age for the Cuban Elmira asphalt mine seep community, making it the oldest in the Caribbean region. A new basal Pliocene seep fauna from the Dominican Republic is characterized by the large lucinid Anodontia (Pegophysema). In Trinidad we distinguish two types of seep faunas: the mudstone-hosted Godineau River fauna consisting mainly of lucinid bivalves, and the limestone-hosted Freeman’s Bay fauna consisting chiefly of Pleurophopsis, Bathymodiolus, and small gastropods; they are all dated as late Miocene. Four new seep communities of Oligocene to Miocene age are reported from Venezuela. They consist mainly of large globular lucinid bivalves including Meganodontia, and moderately sized vesicomyid bivalves. After the late Miocene many large and typical ‘Cenozoic’ lucinid genera disappeared from the Caribbean seeps and are today known only from the central Indo-Pacific Ocean. We speculate that the increasingly oligotrophic conditions in the Caribbean Sea after the closure of the Isthmus of Panama in the Pliocene may have been unfavorable for such large lucinids because they are only facultative chemosymbiotic and need to derive a significant proportion of their nutrition from suspended organic matter.


Introduction
Methane seeps on the deep-sea floor harbor dense faunal communities whose dominant members rely on chemosynthetic bacteria for nutrition [1,2]. First discovered in the Gulf of Mexico in the 1980s, they are now recognized at virtually all continental margins worldwide [3,4]. Although the rise of the modern, mollusk-dominated vent and seep fauna began during the Cretaceous, the most common taxa at present-day vents and seeps originated in the early Cenozoic [5][6][7][8][9]. Biogeographically the Cenozoic fossil record of methane seeps is strongly skewed toward the active continental margins of the Pacific Ocean [10][11][12][13]; fossil occurrences in the Atlantic realm are restricted to the Caribbean region [14,15] and the Mediterranean basin [16,17]. The fossil record of the Caribbean region is of particular interest in this context because this area has long served as a gateway for faunal exchange between the Atlantic and Pacific oceans [18,19].
Unusual faunal assemblages have been reported for a long time from the Caribbean region, including examples from Trinidad that were dominated by the enigmatic bivalve Pleurophopsis Van Winkle, 1919 [20,21], from Cuba, which was considered as a mix of marine and freshwater taxa [22], from Colombia, where the 'Pleurophopsis fauna' was reported from the Oligocene of the Sinú River basin [23], and a fauna from the Joes River area in Barbados, which "appears to be a specialized one, perhaps requiring an unusual environment for its existence" [24]. These and other, similar faunas were subsequently identified as ancient methane seep faunas [14,15,25], but the still poor age determinations and taxonomic identifications of these faunas have so far prevented a rigorous analysis of the evolution of seep faunas from the Caribbean/ Gulf of Mexico region, as well as their role in the biogeographic evolution of the seep fauna in general.
Here we report new seep faunas from Colombia, Cuba, the Dominican Republic, Trinidad, and Venezuela, and our attempts to improve the dating of some of the known faunas from Barbados, Cuba, Trinidad, and Venezuela (Fig 1), based on strontium isotope stratigraphy. A comprehensive taxonomic account on the mollusk species of the Caribbean Cenozoic seep faunas is beyond the scope of the present paper and will be published separately.
Sr isotope stratigraphy is based on the observation that (i) marine carbonates record the 87 Sr/ 86 Sr-ratio of the seawater in which they formed, and (ii) those ratios have changed through Earth's history. Thousands of measurements of samples with known biostratigraphic ages have been compiled into a Phanerozoic Sr-isotope curve that can be used, within certain limits, to infer the age of a marine carbonate sample based on its 87 Sr/ 86 Sr-ratio [26]. We have recently tested whether this approach can be applied to fossil methane-seep carbonates [27]. Methane seeps pose the potential risk that the seeping fluid carries Sr with a 87 Sr/ 86 Sr-ratio quite different from that of ambient seawater [28], which could corrupt the utility of the Sr isotope signature of the seep carbonate for stratigraphic purposes. However, this does not seem to the case as most tested Recent seep carbonates carry the marine Sr isotope signature [29,30], and our tests with a wide range of Phanerozoic seep carbonates indicated that diagenetic alteration, not initial contamination, is the main issue in this approach [27].

Material and Methods
The new faunas reported here are from the collection of the Paleontological Research Institution (PRI) in Ithaca, USA, and the Naturhistorisches Museum Basel (NMB), Switzerland. All taxonomic identifications and assignments presented here were made by SK, except when noted otherwise. Additional material for Sr isotope stratigraphy and thin sectioning was provided by the Smithsonian Natural History Museum (USNM) in Washington, DC, USA, for the Palmar-Molinera-road site in Colombia, and the Elmira asphalt mine site in Cuba [15]. The fossil occurrences are identified as ancient methane seeps based on (i) their faunal content, typically large lucinid, vesicomyid, and mytilid bivalves, (ii) the carbon isotope values (δ 13 C) of the associated carbonate, which tend to be very negative, and (iii) distinct carbonate fabrics of the associated carbonate (when available). The oxygen isotope signature (δ 18 O) of the associated carbonate is used as an indicator for the degree of diagenetic alteration of the carbonate, which is used to assess the reliability of the Sr isotope signature [27,31]. No permits were required for the described study, which complied with all relevant regulations.
Strontium isotopic compositions of representative samples were analyzed on unspiked samples. The samples were weighted into Teflon vials and dissolved in a mixture of 5 ml HF and HNO 3 (3:2) with a PicoTrace digestion system. This form of digestion may imply that tiny amounts of detrital and/or meteoric material are dissolved as well, and therefore the obtained ratios represent maximum ages. The solutions were processed by standard cation-exchange techniques for purification of the Sr fractions. Sr was loaded with 0.5N H 3 PO 4 on pre-conditioned double Re filaments. Measurements of isotopic ratios were performed on a ThermoFinnigan Triton mass spectrometer in static mode (GZG Göttingen, Dept. Isotope Geology). The mean 87 Sr/ 86 Sr ratio obtained for the Sr standard SRM NBS987 during the period of analytical work was 0.710272 ± 0,000039 (n = 8, 2σ). All Sr isotopic ratios were normalized to an 86 Sr/ 88 Sr ratio of 0.1194 over the course of this study. Total procedure blanks were consistently below 150 pg. 87 Sr/ 86 Sr ratios given as " 87 Sr/ 86 Sr [corr.]" in Table 1 are corrected for blank and mass fractionation, and after these corrections the 87 Sr/ 86 Sr ratios were adjusted to 0.710248 for the NBS987 which is the normalization ratio for the LOWESS curve and look-up version 5.0 [26]. All 87 Sr/ 86 Sr ratios are reported with their 2 σ internal precision. For further details see [32]. Samples for Sr isotope analyses were either taken from the shells of mollusks, or from matrix micrites or rim cements of the seep carbonates that contain the mollusks. The latter carbonate phases often preserve an unaltered Sr isotope signal [27] and the samples were checked for diagenetic alteration using their oxygen isotope signature (Table 1). Samples for carbon and oxygen isotope analyses were extracted using a hand-held microdrill and carbonate powders were reacted with 100% phosphoric acid at 75°C using a Kiel III online carbonate preparation line connected to a ThermoFinnigan 252 mass spectrometer. All values are reported in per mil relative to V-PDB by assigning a δ 13 C value of +1.95‰ and a δ 18 O value of -2.20‰ to NBS19. Reproducibility was checked by replicate analysis of laboratory standards and is better than ± 0.05‰.

Barbados
Fossil seep faunas in Barbados have been the subject of various studies, but the identity of the taxa and their stratigraphic ages are still poorly understood. An unusual mollusk fauna was reported from Eocene, oil-indurated mudstones of the Joes River Formation in Barbados [24]. Later, Harding [25] reported an early Miocene carbonate-hosted seep fauna with various bivalves and worm tubes from the nearby Bath Cliffs area, and identified the 'Joes River' fauna as seep related and as being of Miocene instead of Eocene age. A new geologic framework was adopted for these faunas by Gill et al. [14]: the fauna of the Joes River Formation was regarded as belonging to the diapiric mélange, and the carbonate-hosted fauna reported by [25] as belonging to the Suboceanic Fault Zone. Furthermore, they suggested that both faunas could have any geologic age between Eocene and Miocene [14]. To improve the dating of these faunas, five Sr isotope measurements were made (Table 1): three on the micritic matrix of carbonates associated with Pleurophopsis-and Bathymodiolus-like bivalves from the Oceanic Formation (samples Sr-1 to Sr-3), now the Suboceanic Fault Zone (Fig 2; Basel Museum locality 10147; from the East facing side of Bissex Hill near Cambridge, considered Middle-Early Eocene), and two on a mudstone-hosted shell of an abyssochrysoid gastropod from the Joes River Formation (samples Sr-4 and Sr-5), now diapiric mélange (Basel Museum locality 10039; western branch of Spa River, between Spa and Richmond Ridge, considered as middle Eocene).
The two samples from the gastropod (samples Sr-4 and Sr-5) indicate an early late Eocene age (basal Priabonian, 37.2 to 37.6 Ma). This is consistent with the previously suggested Eocene age [24] for the fauna associated with the Joes River Formation, which consists chiefly of large lucinid bivalves, a large bivalve of unknown affinity, the nuculanid bivalve Nuculana senni Kugler, Jung and Saunders, 1984, and a large abyssochrysoid gastropod now assigned to the genus Ascheria [24,33]. This late Eocene mudstone-hosted association will be referred to as 'Joes River fauna' in the following text. The three samples from the carbonates of the 'Oceanic Formation' (samples Sr-1 to Sr-3) indicate an early Miocene age, the two more reliable ones indicate a lower early Miocene (Aquitanian) age, consistent with the age given by [25]. These early Miocene faunas will be referred to hereafter as the 'Bath Cliffs fauna'.

Colombia
One Cenozoic seep deposit has been reported from Colombia so far, called Palmar-Molineraroad locality, based on a small collection deposited in the Smithsonian Natural History Museum in Washington, DC. It consists mainly of a solemyid bivalve, two lucinid bivalves including Elongatolucina peckmanni Kiel, 2013 and the mussel Bathymodiolus palmarensis Kiel, Campbell and Gaillard, 2010 [15,34,35]. Here we report new petrographic data for this deposit, and two new seep faunas based on collections made by Axel A. Olsson in the 1940s, now deposited in the PRI.
The Palmar-Molinera-road site Paleoecology and stratigraphic age. Thin sections made from voucher material from this locality show a micritic matrix with glauconitic grains, oval fecal pellets, detritus such as foraminiferan tests and gastropod shells, and signs of bioturbation ( Fig 3A). Former voids with banded and botryoidal rim cements and filled by sparry calcite are common. In an attempt to improve the stratigraphic age of the Palmar-Molinera-road locality, three Sr isotope measurements were made from rim cements in two different samples (samples Sr-10 to Sr-12). They indicate an early Miocene age (late Burdigalian, 16.5 to 17.3 Ma) rather than an Oligocene age as indicated on the label associated with this collection.

Mata Cana
Location and stratigraphic age. The labels for this collection read "Mata Cana" or "Matacona", and "San José ls., Field along cut rd., Lower Oligocene". Considering Olsson's (1940) statement about the frequent occurrence of carbonate lenses with the Pleurophopsis fauna in the Sinú River basin mentioned above, this could possibly refer to Mata de Caña, about 20 km S of Lorica in the Department Córdoba, along the Sinú River, at 9°04'30" N, 75°49'30"W.
The carbonate consists mostly of a peloidal micrite with microdetritus and abundant signs of bioturbation. It also contains a few small vugs that lack rim cements but are filled by sparry  calcite ( Fig 3B). The carbon isotope values range from -37.2 to -29.6‰ (Fig 4), indicating methane seepage [36]. The corresponding oxygen isotope values have a narrow range from -3.4 to -1.8‰ and thus indicate some diagenetic alteration; the two Sr isotope measurements (samples Sr-6 and Sr-7) fall below the Sr isotope curve for the Cenozoic (Table 1).
Fauna. Few fossil specimens are preserved with recrystallized shell; most are internal molds. Crustacean claws, possibly of Callianassa Leach, 1814, are abundant in this collection, and there is a crustacean carapace. Bivalves are also common and comprise the lucinids Elongatolucina Gill and Little, 2013 and a rounded-oval species with fine concentric sculpture, a very elongate solemyid resembling Solemya belensis Olsson, 1931 from Oligocene seep deposits in Peru [37], and the nuculid Truncacila Schenck, 1931; there is a high-spired gastropod, a nautiloid, and a crustacean carapace belonging to a? goneplacoid crab. The fauna is summarized in Table 2 and illustrated in Fig 5. Sta. Clara, loc. 303 Location and stratigraphic age. The label associated with this collection reads "Sta. Clara, sinú, 7225 i, flfs. ls., 303, 22-IV-40". A screening of Axel Olsson's field notes (Xerox copies in the USNM Cenozoic mollusk type collection) revealed no information about these potential locality numbers. But in a report on the Oligocene of western Columbia Olsson (1940, p. 250) writes that "The (middle) Oligocene beds are succeeded by shales, very similar to the Uscari and Tapaliza of Panamá and Costa Rica and at several places, particularly in the Sinú, contain limestone lenses carrying the Pleurophopsis fauna." [23]. The collection includes elongate vesicomyid bivalves that were typically referred to as Pleurophopsis by Olsson [37] and it is therefore assumed to be from the Sinú River basin. Indeed, Oligocene accretionary prism sediments crop out over wide areas of the Sinú River basin and are referred to as Floresanto Formation [38].
The carbonate consists of fine peloidal and detrital micrite, and shows evidence of bioturbation. Pyrite-rimmed dissolution features are common, and there are occasional small voids that lack rim cements ( Fig 3C). The carbon isotope values fall within a narrow range of -24.5 to -22.9‰ (Fig 4), indicative of hydrocarbon seepage [36]. The corresponding oxygen isotope values range from -6.3 to -5.7‰ and thus indicate considerable diagenetic alteration and as above, the two Sr isotope measurements (samples Sr-8 and Sr-9) fall below the Sr isotope curve for the Cenozoic (Table 1).
Fauna. Fossils are preserved either with recrystallized shell or as internal molds with chalky surface. The fauna consists of the vesicomyid bivalves Pliocardia sp. (common) and Pleurophopsis sp., the lucinid Elongatolucina sp., an unidentified, small oval bivalve that externally resembles Nucinella Wood, 1851, and the thyasirid Conchocele adoccasa (Van Winkle, 1919). This latter species was originally regarded as belonging to Thyasira Lamarck, 1818 [21] but its large size, rectangular outline and two posterior ridges and sulcus clearly place this species in Conchocele Gabb, 1866 [39,40]. The two gastropods from this site are a neogastropod and the seguenzoid Cataegis godineauensis (Van Winkle, 1919); the latter species was originally regarded as belonging to Solariella Wood, 1842, but with its strong spiral sculpture and fine commarginal ribblets it clearly shows affinities to extant species of Cataegis [41]. A summary of the fauna is provided in Table 3     original report on the locality, Cooke wrote [22] that "The fossils are doubtfully referred to the Oligocene", without providing any basis for this assignment. The fossil assemblage at this site is rather unusual compared to those described here and elsewhere for the Caribbean Cenozoic and it may therefore indeed be older than Oligocene. Even the early Eocene age is not contradicted by the fauna: the Unio? bitumen is as-yet difficult to place taxonomically, the only other occurrence of the gastropod Elmira Cooke, 1919 is at a late Cretaceous seep deposit in Japan (Takami Nobuhara and SK, unpublished observation), and gastropods resembling the abyssochrysoid from the Elmira asphalt mine are mostly found in Cretaceous and Eocene seep deposits [43][44][45].

Cantera Portugalete
Location and stratigraphic age. This material was collected by Dorothy K. Palmer during her pioneering work on the stratigraphy and geology of Cuba. All specimens have the number 802 written onto them, which most likely is Palmer's locality number. The description of this locality is "Habana Province, Cantera Portugalete, 1 km. N of Jamaica. Ls chips. 1/9/32" [46]. The village of Jamaica is just NW of San José de las Lajas, about 30 km SE of Havana, at 22°5 8'44" N, 82°10'11" W. The sediments in this area belong to the early Miocene Husillo Formation and represent a "carbonate and carbonate-terrigenous sequence" that was deposited in a deep marine channel between present-day western and central Cuba [47,48]. The Husillo Formation contains indurated, dark gray, massive rocks [47,49] like those associated with the fossils concerned here. Two samples for Sr isotope stratigraphy were taken from the carbonate that forms the internal molds of two bivalves (samples Sr-1 and Sr-17), and both indicate a basal Burdigalian age (19.4 to 19.5 Ma), consistent with the early Miocene age of the Husillo Formation (Table 1).
Fauna. Few fossils preserve recrystallized shell material, most are internal molds. The most conspicuous species is the large (up to 120 mm long) lucinid bivalve Meganodontia sp. The other species are an up to 90 mm long vesicomyid bivalve resembling Pleurophopsis lithophagoides Olsson 1931, a small (44.5 mm) lucinid resembling the extant Myrteopsis? lens Verrill and Smith, 1880 from the western Atlantic Ocean, and a mytilid that reaches 56 mm in length. The mytilid species is unusual among seep-inhabiting mussels because it does not resemble any extant or fossil bathymodiolin, but instead has a terminal umbo and the curved outline of species belonging to Brachidontes or Mytilus. A species of Brachidontes is here reported from a seep deposit at La Piedra in Venezuela (see below), which is also associated with a large Meganodontia. The Cuban species may thus also belong to Brachidontes. The fauna is summarized in Table 4 and illustrated on Fig 7. Paleoecology. The carbon isotope signature of the carbonate attached to the fossils or composing the internal molds ranges from -10 to -4.4‰ (Fig 4). These are less negative values compared to the other seep deposits reported herein and are not necessarily indicative of a seep limestone. However, during the precipitation of seep carbonate the typically very negative original isotope signature of the seeping methane is 'diluted' by mixing with other carbon sources (i.e., marine bicarbonate) [36]. Considering that (i) the Husillo Formation is largely a carbonate sequence the addition of marine carbonate was likely to be very high, and (ii) the fauna consists largely of chemosymbiotic bivalves, the Cantera Portugalete fauna is here interpreted to have lived at a methane seep, possibly with low rates of seepage and diffuse fluid flow.

Fauna and paleoecology
The fauna consists of numerous large (up to 115 mm long) specimens of Anodontia (Pegophysema) sp., possibly a precursor to the very similar Recent Caribbean A. (P.) schrammi (Crosse, 1876) (Fig 8). One of the specimens has some hard micritic matrix adhering to it, which shows a carbon isotope signature of -28 to -20‰ (Fig 4). Especially the lower value indicates the oxidation of hydrocarbons including methane during the formation of this carbonate.

Trinidad
The Cenozoic seep faunas of Trinidad reported so far were found along the shore of Freeman's Bay to the West of San Fernando, in the vicinity of the mouth of Godineau River. Here they were collected as float material on the beach, as mentioned in various reports [14,20,21,51] and indeed, many of the samples investigated here have numerous small extant barnacles attached to them. The fossils were reported to be associated with sediments of the Lengua Formation, which were either considered as middle Miocene [52] or late Miocene [53]. They were reported from two different lithologies (Fig 9) associated with two different preservational Fossil Caribbean Seep Faunas styles: (i) from a large, isolated, impure limestone block on the costal mudflat where fossils occur as internal molds, and (ii) from often brecciated, oil-impregnated calcareous mudstone lenses surrounded by greenish, unctuous and slickensided clays in the adjoining coastal section where the fossils occur as black, oil-impregnated specimens often with original shell material [14,51]. The impure limestone block was given its own lithostratigraphic unit, the "Freeman's Bay limestone Member", the oil-impregnated calcareous mudstone lenses were interpreted as having formed at mud diapirs [14]. Despite these differences, paleoecologic interpretations were considered as difficult to reach and the available fossils were reported as a single seep fauna [14]. Based on the material available at the NMB and the PRI, we separate two ecologically distinct types of seep faunas in the Miocene of Trinidad: the dark, bitumen-impregnated fauna is coined 'Godineau River fauna' because the boxes containing this fauna at the PRI were labeled Godineau River, whereas the fauna associated with the white limestone is referred to as 'Freeman's Bay fauna'. Two further collections with a typical Freeman's Bay fauna were recovered, Bronte Estate and Jordan Hill, from an area to the East of San Fernando, towards Princess Town. Our Sr-isotope work indicates a late Miocene age for the Godineau River and Jordan Hill faunas (Table 1), which supports the stratigraphic scheme of Erlich et al. [53].

Godineau River fauna
Stratigraphic age. Two Sr isotope measurements were made: one sample was from the calcareous mudstone and indicates a middle Messinian age (sample Sr-19, 6.35 Ma), the other is from the aragonitic shell of the possible vesicomyid bivalve, which gave a basal Tortonian age (sample Sr-18, 11.1 Ma). The corresponding oxygen isotope value of the latter sample (Table 1) indicates slight diagenetic alteration of this material by meteoric water, thus the Messinian age derived from the apparently unaltered calcareous mudstone is here considered the more reliable. However, both dates suggest a late Miocene age for the Godineau River fauna.
Fauna and paleoecology. The Godineau River fauna consists exclusively of very large, infaunal or semi-infaunal chemosymbiotic bivalves (Table 5, Fig 10): the thyasirid Conchocele adoccasa reaching 87 mm length, the lucinids Nipponothracia sp. reaching 167 mm in length and Elliptiolucina sp. reaching 85 mm length, and an unidentified species that may (or may not) belong to the Vesicomyidae, reaching 140 mm length. In thin section, the oil-impregnated calcareous mudstone is composed of monotonous micrite with rare planktonic foraminiferans (Fig 9A). Its carbon isotope signature ranges from -25.8 to -20.2‰, with corresponding oxygen isotope values ranging from 3.7 to 4‰ (Fig 11), indicating that the limestone formed due to methane oxidation with a considerable amount of marine bicarbonate mixed in [36].

Freeman's Bay
Stratigraphic age. Two Sr isotope measurements were made on two differently colored micrites (samples Sr-20 and Sr-21, Table 1). The accompanying oxygen isotope values were slightly negative, suggesting diagenetic alteration by meteoric waters, and the ages derived from the Sr isotope ratios were younger than reasonable considering the biostratigraphic framework of this locality. Fauna and paleoecology. The fauna consists largely of the vesicomyid bivalve Pleurophopsis unioides Van Winkle, 1919 and P. u. var. fernandensis Van Winkle, 1919 that reach just over 80 mm length. The validity of the genus Pleurophopsis had been questioned because of its poorly preserved type material [54], but was considered valid by other authors [55,56]. The material investigated during the present study indicates that it is indeed a valid and distinctive genus; further details will be published in a forthcoming paper. Minor elements of the Freeman's Bay fauna include the lucinid bivalves Cubatea sp. and an unidentified lucinid, which are here reported for the first time from this fauna, Conchocele adoccasa, which is the only species shared between the Freeman's Bay and Godineau River faunas, a nuculanid bivalve, and four gastropods: Cataegis godineauensis (Van Winkle 1919), Cantrainea sp., Provanna sp., and a limpet (Table 6, Fig 12). In addition, a bathymodiolin mussel has been reported [14,21].
The Freeman's Bay fauna was characterized by Van Winkle as "very peculiar and unlike any known" [21] and later interpreted as a fossil seep fauna [14,57]. Evidence for this interpretation has never been provided, and it was presumably based on the abundance of chemosymbiotic bivalves. The carbon isotope signature of the limestone ranges from -20 to -14.5‰, with corresponding oxygen isotope values ranging from -2.8 to -1.7‰ (Fig 11), consistent with the interpretation that methane oxidation was involved in the formation of the carbonate.

Fossil Caribbean Seep Faunas
Petrographically the Freeman's Bay limestone shows abundant traces of bioturbation and bioclasts (Fig 9B), unlike the Godineau River material.

Bronte Estate
Location and stratigraphic age. The locality description for NMB locality 10228 reads "Lengua Formation, Freeman's Bay limestone Member. Naparima area, Bronte Estate, Manager House. Quarry near trigonometric station W of Naparima." Today there is a 'Bronte village, Naparima' just south of the Jordan Hill Presbyterian school, exactly where the Lengua Formation crops out according to Kugler's map [51]. It is therefore assumed that the fossils reported here were collected in this area. Due to the (assumed) close proximity to the Jordan Hill site, this locality is considered to be of late Miocene age.   (Table 7, Fig 13). The carbonate adhering to the fossils is whitish-ochre colored micrite; material for thin sectioning was not available. Its carbon isotope signature ranges from -30.2 to -23‰, clearly indicating than methane oxidation was involved in the formation of the carbonate. The corresponding δ 18 O values range from -5.8 to -4.8‰ (Fig 11).

Jordan Hill
Location and stratigraphic age. The labels of this small collection read "Jordan Hill House" and "Trinidad GDH", suggesting that it was collected by Gilbert Dennison Harris. Today there is a Jordan Hill Presbyterian primary school on the Manahambre road, about halfway between San Fernando and Princes Town, at 10°15'22"N, 61°24'29"W, and it is assumed that the material was collected around here. Kugler's map [51] indicates the upper Cipero Formation in this area, which ranges into the late Miocene [53]. However, the Lengua Formation crops out just a kilometer or so to the SW and SE [51], and could thus also be a source of the material reported here. Two samples for Sr isotope stratigraphy indicate a late Tortonian age (samples Sr-22 and Sr-23, 7.7 and 8.8 Ma).
Fauna and paleoecology. The fauna ( Table 8, Fig 14) is very similar to the Freeman's Bay fauna except for a small, possible vesicomyid bivalve having the general outline of the genus Isorropodon Sturany, 1896 (Fig 14C). Today this genus has a wide geographic distribution but has so far not been reported from the western Atlantic Ocean [58]; the genus was also present in an early Miocene seep deposit in the northeastern Pacific Ocean [59]. The lithology of the Jordan Hill seep deposit differs from both the Freeman's Bay and Bronte Estate seep deposits by containing more diverse carbonate phases and especially by the presence of various vugs lined with banded rim cements; such cements have not been seen in the Freeman's Bay and Bronte Estate deposits (Fig 9C). Furthermore, the Jordan Hill seep carbonate shows a wider range in its carbon isotope signature, ranging from -43.7 to -12.3‰ (Fig 11), consistent with its more diverse carbonate phases. The very negative carbon isotope signature clearly indicates that biogenic methane (rather than thermogenic methane) was involved in the formation of the carbonate [36].

Venezuela
Cenozoic seep faunas in Venezuela are found in the Falcon Basin at or near the present-day coast of the Caribbean Sea, and here the moderate to deep-water deposits within the Agua Salada Group. A single Cenozoic seep deposit has so far been reported, from Puerto Escondido in Estado Falcón, consisting of solemyid, bathymodiolin, thyasirid, lucinid and vesicomyid bivalves, and a few gastropods, considered as being of Miocene age [14,60]. The new material reported here is from the collection of the NMB and includes four new sites and new data on the Puerto Escondido site reported earlier [14]. All new sites are presumed to be from the Agua Salada Group in the Estado Falcón, including three from the Pozón Formation, one from the Guacharaca Formation, and one from the Riecito Limestone. They are considered to range in age from Early Oligocene to middle Miocene. Most locality data are from the NMB locality register, addition information and opinions were provided by Oliver Macsotay (pers. comm. 2014), a specialist of the Cenozoic stratigraphy of Venezuela [61].

Paleoecology
The Venezuelan fossils in the NMB collection had very little or no carbonate matrix adhering to them, thus thin section preparation was not possible and only carbon and oxygen isotope data are available for environmental reconstructions. These isotope data were very similar among the sites (with the exception of La Piedra, discussed below). Therefore, their implications are summarized here, instead of reiterating them at each site description. The δ 13 C values range from -26.5 to -9.5‰, with the majority of values between -22 and -15‰ (Fig 11). These values indicate carbonate precipitation induced by the oxidation of hydrocarbons, but are not necessarily indicative of methane as sole hydrocarbon [15,36]. However, the characteristic, very negative δ 13 C signature of methane can be diluted by marine carbonate in carbonate-rich settings [62]. The Venezuelan sites reported here were formed in tropical marl-and limestone, middle to outer shelf settings where a strong input in marine carbonate can be expected. Hence we assume that the sites reported here formed at ancient hydrocarbon seeps where methane formed an important part of the hydrocarbons.

Buenavista de Maicillal
This small collection (Basel loc. 15781) obtained by Paul Leuzinger is from Buenavista de Maicillal in the Mirimire area in the Estado Falcón. It belongs to the Pozón Formation of the Agua Salada Group and here to "its Husite member, composed of marly clays interbedded with  suggests an uppermost Aquitanian (early Miocene) age; however, the oxygen isotope signature of the samples indicates some diagenetic alteration ( Table 1). The fauna consists of ten moderately sized (up to 77 mm long) specimens of a lucinid bivalve possibly belonging to Meganodontia (Fig 15).

Corro Colorado
The labels associated with this locality read "Corro Colorado, near Maicillal de la Costa, due South on road"and "[unreadable] between Corro Colorado onto road to San Francisco". According to Oliver Macsotay (pers. comm. 2014) "this locality, by cartography, belongs to the Husite Member of the Pozón Formation. The age is Early Middle Miocene." Two samples (Sr-24 and Sr-25) were used for Sr isotope stratigraphy and suggests an uppermost Aquitanian to   Fig 16).

La Piedra
Location and stratigraphic age. This small collection (NMB loc. 15780) obtained by Louis Vonderschmidt is from La Piedra in the Isidro area in Estado Falcón. It belongs to the Pozón Formation of the Agua Salada Group and was regarded as either middle to late Oligocene (NMB locality register) or mid-late Miocene (label). Oliver Macsotay (pers. comm. 2014) believes that "the fossils came from Policarpo member, composed of marly limestones, clays, with glauconititc beds at its base. Iron concretions are frequent. [. . .] The paleodepth data given by [65] based on benthic foraminifera should extend from 200 to 600 meters." The Policarpio "Greensand" Member is of late Aquitanian (early Miocene) age [65]. Two samples (Sr-30 and Sr-31) were used for Sr isotope stratigraphy and broadly suggest a late Miocene age (lower Tortonian to Messinian; Table 1), but with δ 18 O values as low as -6.5‰ the samples appear diagenetically altered and hence the Sr isotope age should be treated cautiously. Fauna and paleoecology. The fauna consists of a large lucinid resembling Meganodontia, a small lucinid, and a mytilid belonging to Brachidontes (Table 11, Fig 17). The carbon isotope values of the associated carbonate are very heterogeneous, ranging from -11.5 to +12.4‰, with more homogenous corresponding oxygen isotope values between -7.5 to -5.9‰. While the negative carbon isotope values indicate the oxidation of hydrocarbons, the very positive values can only be explained my local methane formation, because the CO 2 pool utilized during archaeal methanogenesis becomes enriched in 13 C [36,66,67]. Such 13 C-enriched carbonate phases are interpreted as having formed after the precipitation of the 13 C-depleted carbonate phases, probably during early burial [36].

Puerto Escondido
The carbonate adhering to a vesicomyid and a Bathymodiolus shell from this locality (NMB loc. 13968) has a carbon isotope signature as low as -26.5‰ (Fig 11), supporting the suggestion of [14] that the Puerto Escondido fauna formed at an ancient methane seep deposit. The age of   Table 1).

Stratigraphy of the Caribbean seep faunas
The exact geographic location and therefore also the stratigraphic ages of many of the Caribbean seep communities are still not well constrained, and our current understanding is summarized in Fig 18. Although our Sr isotope dating of known and new sites provided new insights into their stratigraphic ages, it also revealed a number of persistent problems and uncertainties. The Cuban Elmira asphalt mine site, hesitantly considered Oligocene in previous papers [15,22], may be as old as early Eocene, based on our Sr isotope stratigraphy. This would make it the oldest seep community of the Caribbean region so far. An Eocene age would be consistent with the presence of the relatively large-sized abyssochrysoid gastropods at this site, which occur also the late Eocene 'Joes River fauna' of Barbados [24], and are generally a common feature of late Mesozoic to Eocene seep faunas [9,33,34,43,68].
Fossil seep faunas from Barbados are known from two different rock units: the mudrockhosted Joes River fauna from the diapiric mélange, and the carbonate-hosted Bath Cliffs fauna from the Suboceanic Fault Zone [14]. These faunas are taxonomically distinct and were recently considered to be of any age between the Eocene and the Miocene, based on the wellestablished ages of the enclosing units [14]. Our Sr isotope data, although limited, indicate that at least some of these faunas are also stratigraphically distinct: our samples of the Joes River fauna with its large lucinid bivalves and the tall abyssochrysoid gastropods indicate a late Eocene (basal Priabonian) age, and an early Miocene age for the Bath Cliffs fauna with Pleurophopsis, Cataegis, and associated taxa. However, considering that seep faunas that are ecologically similar to the Joes River and Bath Cliffs faunas, respectively, occur together in late Miocene sediments in Trinidad indicates that these different faunal types may just be the result of different ecological settings [14]. Thus, we cannot exclude the possibility that mudstoneand carbonate-hosted faunas have coexisted in Barbados from Eocene to Miocene time and we have just accidentally sampled an Eocene example of the mudstone-hosted fauna and an early Miocene example carbonate-hosted fauna for our Sr isotope work. Further stratigraphic and taxonomic work on these deposits and their faunas may solve this question and could certainly provide further insights into the evolution of the Caribbean seep fauna.
In Colombia, our attempt at Sr isotope stratigraphy failed in cases of the Sta. Clara and Mata Cana sites. In case of the Palmar-Molinera-road site, it suggests an early Miocene instead of an Oligocene age as indicated by the sample label. This early Miocene age should be treated cautiously, because the samples show evidence of diagenetic alteration. However, the similarity of Bathymodiolus palmarensis to the bathymodiolin found in the Miocene Freeman's Bay and Jordan Hill seep limestones in Trinidad supports a Miocene age of the Palmar-Molinera-road site. Clearly, more detailed stratigraphic work on the Colombian seep faunas would provide more detailed insights into the evolution of the Cenozoic seep faunas of the Caribbean region. Sr isotope stratigraphy has proven useful in many, but not all, cases in the present study, another potential way forward may be dinoflagellate biostratigraphy, as recently used for Cretaceous seep carbonates found as float boulders on beaches in northern New Zealand [68].
Among the five Venezuelan sites, a bewildering array of contradicting stratigraphic ages is indicated by the labels and locality information associated with these fossils, by the personal assessment of an experienced regional geologist, and our Sr isotope stratigraphy approach. Most ages, however, seem to center around the Miocene (Fig 18).

Biogeography & Evolution
Through the Cenozoic, seep faunas in the Caribbean and Gulf of Mexico region show three successive faunal associations:  Geologic ranges of mollusk genera from Cenozoic seeps in the Caribbean region. Note break in time scale below 40 Ma; dotted lines indicate occurrences outside the seep environment; data for Bathynerita and 'Calliotropis' from [14], for Ascheria from [33]. doi:10.1371/journal.pone.0140788.g019 Fossil Caribbean Seep Faunas bivalves, are known from various Cretaceous to Eocene seep deposits worldwide [34,35,43,44,68], and S. Kiel and T. Nobuhara, unpublished observations]. The disappearance of the large gastropods from the Caribbean seeps after the Eocene coincides with their disappearance worldwide, while the lucinid bivalve genera survived into the Oligocene and Miocene, both in the Caribbean region and in general. Thus the Eocene Caribbean seep fauna is clearly derived from, and shares a common fate with, the global seep fauna of its time.
ii. The Oligocene to Miocene Caribbean seep faunas consist partly of members of widely distributed lucinid genera [35], and they also saw the rise of the Pleurophopsis-Cataegis-fauna. These taxa have not been reported from other well-studied seep faunas of this age, namely in the western USA [69,70], Japan [59,71], and New Zealand [72]. However, this fauna does show links to the Pacific Ocean, especially during the Oligocene, when Pleurophopsis and other taxa reported here from Colombia are known from seep deposits in northern Peru [15,37].
iii. Arguably the most remarkable faunal change among Caribbean/Gulf of Mexico seep faunas can be seen between the late Miocene and the modern fauna, when Pleurophopsis disappeared along with virtually all lucinid genera (Fig 19).
The latter is noteworthy because present-day Caribbean/Gulf of Mexico seeps are inhabited mostly by a distinct group of medium-sized (up to 70 mm length) lucinids which so far have no fossil record [35,73,74]. Although the fossil record currently lacks the resolution to narrow down the exact timing of this extinction, it roughly coincides with the closure of the Isthmus of Panama [75]. This event resulted in increasingly oligotrophic conditions in the Caribbean Sea [76,77] and it is tempting to speculate that this had a negative effect on the large lucinids: Lucinids are only facultative chemosymbiotic and derive a significant proportion of their nutrition from suspended organic matter [78]. Thus a decrease in food availability may have led to the disappearance of the large lucinids that were used to more eutrophic conditions, and those that replaced them could not realize such large size. Indeed, at least two of the large lucinid genera that inhabited Miocene seeps in the Caribbean region (Elliptiolucina and Meganodontia) are today restricted to the more eutrophic waters of the central Indo-Westpacific [79,80]. To the best of our knowledge, there are no studies on the size of suspension feeders on either side of the Isthmus of Panama before and after its closure, but a general decrease in the maximum size of tropical lucinid bivalves through the Cenozoic was attributed to decreasing productivity [81]. Likewise, the generally larger body size among soft-bottom feeding guilds in the western compared to the eastern North Atlantic Ocean was considered related to the higher productivity of the western North Atlantic Ocean [82].
(London) for advise on lucinid bivalves, Emilio Garcia (Lafayette) for providing extant material from the Gulf of Mexico, Adiël Klompmaker (Gainesville) for help identifying crustacean fragments, Jim Goedert (Wauna) for critically reading an early version of this manuscript, and Greg Dietl and three anonymous referees for their constructive criticism that very much improved this manuscript.