A minute ostracod (Crustacea: Cytheromatidae) from the Miocene Solimões Formation (western Amazonia, Brazil): evidence for marine incursions?

A huge wetland (the ‘Pebas system’) covered western Amazonia during the Miocene, hosting a highly diverse and endemic aquatic fauna. One of the most contentious issues concerns the existence, potential pathways and effects of marine incursions on this ecosystem. Palaeontological evidences (body fossils) are rare. The finding of a new, presumably marine ostracod species (Pellucistoma curupira sp. nov.) in the upper middle Miocene Solimões Formation initiated a taxonomic, ecological and biogeographical review of the genus Pellucistoma. We demonstrate that this marine (sublittoral, euhaline), subtropical–tropical taxon is biogeographically confined to the Americas. The biogeographical distribution of Pellucistoma largely depends on geographical, thermal and osmotic barriers (e.g. land bridges, deep and/or cold waters, sea currents, salinity). We assume an Oligocene/early Miocene, Caribbean origin for Pellucistoma and outline the dispersal of hitherto known species up to the Holocene. Pellucistoma curupira sp. nov. is dwarfed in comparison to all other species of this genus and extremely thin-shelled. This is probably related to poorly oxygenated waters and, in particular, to strongly reduced salinity. The associated ostracod fauna (dominated by the eurypotent Cyprideis and a few, also stunted ostracods of possibly marine ancestry) supports this claim. Geochemical analyses (δ18O, δ13C) on co-occurring ostracod valves (Cyprideis spp.) yielded very light values, indicative of a freshwater setting. These observations point to a successful adaptation of P. curupira sp. nov. to freshwater conditions and therefore do not signify the presence of marine water. Pellucistoma curupira sp. nov. shows closest affinities to Caribbean species. We hypothesize that Pellucistoma reached northern South America (Llanos Basin) during marine incursions in the early Miocene. While larger animals of marine origin (e.g. fishes, dolphins, manatees) migrated actively into the Pebas wetland via fluvial connections, small biota (e.g. P. curupira sp. nov.) were phoretically freighted and developed freshwater tolerance over long timescales. http://zoobank.org/urn:lsid:zoobank.org:pub:886C6476-393D-4323-8C0E-06BB8BD02FD9

Aside from mangrove-related pollen, dinoflagellate cysts, foraminiferal linings and remains of several marine vertebrate clades, body fossils that are specific for marine environments are scarce and restricted to thin intervals (e.g. Linhares et al. 2011;Boonstra et al. 2015). Remarkably, the highly endemic aquatic invertebrate fauna of the 'Pebas system' is strongly dominated by the abundance of pachydontine bivalves and the cytheroid ostracod Cyprideis. Otherwise common freshwater and marine taxa are rare and typical marginal marine molluscs (e.g. arcids, oysters, mangrove cerithioidean snails) are absent (e.g. Whatley et al. 1998a;Wesselingh 2006Wesselingh , 2007Gross et al. 2013Gross et al. , 2014. Based on the mollusc and ostracod faunas, brackish waters (e.g. Purper 1979;Nuttall 1990;Whatley et al. 1998a) or extensive marine transgressions (Sheppard & Bate 1980) were proposed. However, geochemical investigations performed on the shells of these biota consistently indicate freshwater conditions. Elevated salinity (»5 PSU) is only evident for a few localities (e.g. Vonhof et al. 2003;Kaandorp et al. 2006;Gross et al. 2013). Wesselingh (2007) considered the 'Pebas system' to be a predominantly freshwater environment, in which adaptations to predation pressure and a muddy, poorly oxygenated substrate triggered speciation, as well as habitat dominance by pachydontine bivalves. Gross et al. (2013) suggested that it was a locally unstable but on a regional scale long-lived wetland, where euryoecious biology, passive dispersal predispositions and reproduction modes (brood care) favoured the success of the genus Cyprideis. Thus, a conclusive, 'simple' explanation about the nature of this 'mega-wetland' is still pending.
The current study was initiated by the finding of a very small-sized, marine ostracod (Pellucistoma) in late middle Miocene sediments of western Brazil (Fig. 1). This posed three central questions: (1) does this record prove the existence of marine incursions, thousands of kilometres away from the next (palaeo-)coastline; (2) which provenance could Pellucistoma be from; and (3) what are the potential migration pathways? It is certainly tempting to approve the first question by applying uniformitarian principles, to suppose a Caribbean origin (as suggested for e.g. molluscs ;Wesselingh & Ramos 2010) and to relate our record directly to far-reaching marine incursions (e.g. Boonstra et al. 2015). However, in view of the conflicting discussions about this topic, we performed an extensive taxonomic, ecological and biogeographical appraisal of Pellucistoma. Based on this, we explore possible answers to the above-mentioned questions.
Well 1AS-10-AM was continuously cored down to 400.25 m. Its lithology consists of alternations of semiindurated clay and silt. Up to metre-thick, sandy as well as decimetre-thick, lignite intercalations occur subordinately. Recently, Gross et al. (2014) studied the Figure 1. Location of the studied well 1AS-10-AM in western Amazonia. A, overview map; B, position of exploration wells (after Maia et al. 1977); star D herein investigated core; compare Gross et al. (2014). microfauna (in particular the ostracod genus Cyprideis). For more detailed core descriptions and an illustration of the section, we refer to this work. The herein discussed findings stem from sample AM10/30 (depth D 141.2 m), which is a clayey silt, rich in mollusc remains. According to Gross et al. (2014) this sample is biostratigraphically dated to the Cyprideis obliquosulcata ostracod zone sensu Muñoz-Torres et al. (2006), corresponding to a late middle Miocene age (Wesselingh & Ramos 2010).
For stable isotope analyses (d 18 O, d 13 C), ostracod valves were additionally washed with distilled water and rinsed in ethanol. From core 1AS-10-AM (sample AM10/ 30) adults of three species were measured: Cyprideis machadoi (Purper, 1979), Cyprideis multiradiata (Purper, 1979) and Cyprideis sulcosigmoidalis (Purper, 1979) (for taxonomy see Gross et al. 2014). The number of specimens required for analyses (»50 mg) varied between 1 and 3 per measurement. A Thermo-Finnigan Kiel II automated reaction system and a Thermo-Finnigan Delta Plus isotope-ratio mass spectrometer were used to conduct the analyses (University of Graz; standard deviation D 0.1% relative to NBS-19; results in per mille relative to the Vienna Pee Dee Belemnite (VPDB) standard).
All specimens are housed in the micropalaeontological collection of the Museu Paraense Em ılio Goeldi, Bel em .

Systematic palaeontology
Suprageneric classification follows Horne et al. (2002). Description. Shape. Subrhomboidal in lateral view; anterior margin moderately infracurvate, dorsal margin almost straight and subhorizontal, ventral margin with slight concavity below the mandibular scars, posterior margin with a blunted subdorsal caudal process; valves anterodorsally, posteroventrally and posteriorly laterally flattened; lensshaped in dorsal view with beaked posterior end.
Ornamentation. The very thin-shelled valves are basically smooth except for shallow, wrinkle-like ridges along the free valve margin as well as in the centro-and ventrodorsal area, forming a weak reticulum anteroventrally; at the caudal process an additional, oblique ridge is always present.
Inner lamella. Anterior and posterior wide; anterior vestibulum large, posterior one narrow-elongated, extending up to the caudal process; several inner lists developed; selvage subperipheral, inconspicuous anteriorly, forming at the ventral concavity a bulge, which fits into a groove of the other valve (groove more prominent in right valves).
Marginal pore canals. Widened at their base, leading to an irregular line of concrescence; occasionally bifurcated, some branches developed as false pore canals.
Hinge. Right valve À anterior element with four roundish sockets (the most anterior being the largest, the most posterior one is barely developed) and a spatulate anti-slip tooth below; median element consists of a smooth groove, which is deepened at its posterior end; posterior hinge element with three elongated teeth, succeeded by a fourth, indistinct tooth, which merges backwards into a thin expansion of the posterior margin; left valve À anterior element with a trilobate tooth and a further weak tooth postjacent; median element consisting of a smooth bar, which forms an elongated tooth-like structure at its posterior end; posterior element with three, elongated, shallow sockets followed by a fourth, indefinite, elongated socket fading out towards the posterior end.
Central muscle scars. A row of four, slightly posteriorly inclined adductor scars; two oval mandibular scars, one irregularly ovate frontal scar; numerous dorsal muscle scars; well above the row of adductor scars a row of four dorsal scars is developed, probably corresponding to the 'lucid spot' (Morkhoven 1963;Sandberg 1969), which, however, is not a discrete, single spot here.
Eye-spot. Slight eye-spot developed. Sexual dimorphism. Unclear; a few specimens (e.g. Fig. 3B) are slightly larger and display a somewhat higher posterior valve proportion, which could be related to sexual dimorphism.

Remarks.
Generic classification. The current species imitates several genera of different families by its subrhomboidal outline and almost smooth shells.
Pseudoconcha has a bipartite hinge (right valve: anterior element formed by a strong bar, with a groove below; posterior half with groove and bar below), a less developed caudal process, a well-punctate surface and a narrower inner lamella (Witte 1993;Sarr et al. 2008).
Although some variability in details of the hinge, muscle scar patterns and pore canals seem to be present in Elofsonia (Aiello & Szczechura 2002), this genus differs by its less prominent, less pointed caudal process (except Elofsonia sp. in Keyser & Sch€ oning 2000) and its more simple hinge (right valve: dorsally crenulated anterior socket; smooth median groove; weak posterior tooth ;Whittaker 1973;Athersuch & Horne 1984;Athersuch et al. 1989).
Originally, Phlyctocythere was characterized by its inflated, almost spherical carapaces with a peripherally compressed zone and an obtuse, subdorsal caudal process. Its surface is smooth, lacks eye-spots and the valves are very thin-shelled. The hinge is adont (right valve: curved, smooth bar), marginal pore canals are simple, and one frontal muscle scar is developed (Keij 1958; compare also Morkhoven 1963). Subsequently, several species were included in Phlyctocythere, which blur the prime generic diagnosis. For example: (1) outline: Phlyctocythere hamanensis Ikeya & Hanai, 1982 (more elongated, less arched dorsal margin), Phlyctocythere japonica Ishizaki, 1981 (subovate), Phlyctocythere recta Bold, 1988 (straight dorsal margin), Phlyctocythere sicula Sciuto & Pugliese, 2013 (more elongated) and Phlyctocythere stricta Bold, 1988 (straight dorsal margin); (2) ornament: Phlyctocythere curva Bold, 1988, Phlyctocythere retifera Bonaduce, Masoli & Pugliese, 1978, P. sicula and P. stricta display a faint reticulation and/or longitudinal ridges/wrinkles; (3) Phlyctocythere curva, P. recta, P. stricta and probably P. fennerae Mostafawi, 1992 have a slight eye-spot; (4) for P. fennerae, P. japonica and P. retifera few (anteroventrally) branched marginal pore canals are described and these structures are often observed to be bifurcated in P. curva; (5) for P. hamanensis normal pores are 'presumably' of sieve-type; (6) in P. hamanensis two frontal scars and one elongated mandibular scar are mentioned; in the illustration of P. retifera a double frontal scar is indicated; in P. sicula the adductor scars are very elongated; and (7) hinge: in Phlyctocythere caudata Hartmann, 1979, P. curva, Phlyctocythere pellucida (M€ uller, 1894), P. retifera and P. sicula right valves display a smooth median groove (crenulated in P. curva) and (two) terminal sockets (in P. curva: teeth); P. hamanensis has a reduced gongylodont hinge. (Note: Phlyctocythere hartmanni Omatsola, 1970 is attributed to Elofsonia (Athersuch & Horne 1984;Schornikov 2011) or to Pseudoconcha (Witte 1993). Phlyctocythere pellucida is discussed as belonging to Loxocauda Schornikov, 1969(Athersuch & Horne 1984Schornikov 2011)). Consequently, Phlyctocythere is either quite variable or has turned into a collective genus due to inclusion of profuse species. An in-depth revision is obviously needed but is beyond the scope of the present work. In particular, P. retifera from the Red Sea is similar but it is more ovate (but note sexual dimorphism displayed in Mostafawi (1992) for P. fennerae), it diverges in details of the hinge (as far as reproducible) and has two frontal scars. However, by following the original diagnosis of Phlyctocythere (Keij 1958;Schornikov 2011) especially its outline (much more arched dorsal margin), its adont, right-bar hinge and simple marginal pore canals are considered herein to exclude an assignment of the current specimens to that genus.
Some authors (Bold 1950(Bold , 1958Benson et al. 1961;Morkhoven 1963) have discussed a possible synonymy of Javanella Kingma, 1948with Pellucistoma Coryell & Fields, 1937 (see also Gou & Chen 1988;Howe & McKenzie 1989;Ayress 1996). Lately, Bergue & Coimbra (2007) revised Javanella, revalidated it and reassigned it into the family Cytheridae Baird, 1850. According to this work only two species are left in Javanella, which clearly differ in outline (more elongated; caudal process below the middle of valves height) and in details of the terminal hinge elements from the present material.
Amongst the Cytheromatidae Elofson, 1939, the genus Paracytheroma Juday, 1907 is closely related to Pellucistoma. Nevertheless, Paracytheroma can be differentiated from the latter by lacking strong terminal anti-slip hinge elements, the absence of a caudal process and À to a minor degree À by missing a complex marginal zone with branched marginal canals (Hartmann 1978;Ayress 1990;compare also Sandberg 1969;Keyser 1976;Garbett & Maddocks 1979). Based on those features, our specimens do not belong to Paracytheroma but fit best with Pellucistoma as originally defined by Coryell & Fields (1937; for genus definition compare also Edwards 1944;Morkhoven 1963;Sanguinetti 1979). A few, minor differences concern the valves' hinge and surface ornament.
For the hinge of the left valve an "anterior long bladelike triangular tooth" is indicated (Coryell & Fields 1937, p. 17), being trilobate in the present specimens. The median element is formed by a "serrated bar", which "terminates at the posterior cardinal angle" (Coryell & Fields 1937, p. 17). Here, that bar is smooth À at least as preserved. As far as described or perceptible on the provided figures, a smooth median element occurs in Pellucistoma scrippsi Benson, 1959and Pellucistoma bensoni McKenzie & Swain, 1967(Benson & Kaesler 1963McKenzie & Swain 1967;Swain & Gilby 1967, 1974. For the type species, P. howei, the drawings of Bold (1967) and Teeter (1975) do not show such a crenulation. The crenulation of the median hinge element is in some species of Pellucistoma probably very delicate or indeed not developed.
A posterior hinge element, consisting of four elongated sockets/teeth as in our examples, has not been mentioned for Pellucistoma so far. However, based on the dorsal view of a left valve in Coryell & Fields (1937), behind the thickened, tooth-like terminal end of the bar, a shallow groove may be present which might analogously receive tiny teeth of the right valves. Moreover, the illustration of P. scrippsi in Swain & Gilby (1967) implies the presence of small posterior teeth in the right valve (note that this feature is not indicated in e.g. Benson 1959;Benson & Kaesler 1963;McKenzie & Swain 1967;Swain 1967). The description and illustration of the hinge structure of Pellucistoma spurium Bold, 1963 (p. 406: "In the left valve the selvage curves around the sockets and forms the upper border of the groove") also hints at the presence of a posterior element. Thus, the subtle posterior sockets/teeth, clearly visible in our species (under the SEM), seem to be present in other Pellucistoma species equally. Garbett & Maddocks (1979, p. 871) carefully described a similar posterior hinge structure for Paracytheroma stephensoni (Puri, 1954) of which Pellucistoma atkinsi Hall, 1965 is a synonym (Keyser 1976): "[a] posterior tooth formed by the expanded end of the posterior margin." That resemblance mirrors the close relation between Pellucistoma and Paracytheroma again (see above).
The surface of Pellucistoma is described as "finely perforated" (Coryell & Fields 1937, p. 17) and "smooth or finely punctate" (Morkhoven 1963, p. 436). Here, the valves are basically smooth, but display a weak reticulate pattern anteroventrally, some wrinkle-like ridges dorsocentrally and ventrocentrally, as well as a characteristic, oblique, light ridge on the caudal process. Shallow, posterocentral and posteroventral ridges, which converge towards the caudal process, can be seen on P. magniventra in Garbett & Maddocks (1979). True eye-spots have not been included in the genus definition so far. However, a slight eye-spot À like in P. curupira sp. nov. À is recognized in P. scrippsi (Swain 1967;Swain & Gilby 1974).
The type species, Pellucistoma howei Coryell & Fields, 1937 (first record: Panama, latest middleÀearly late Miocene), is quite similar but differs by: its more ovate outline; its more projecting posteroventral margin; a more acuminate caudal process; a narrower anterior vestibulum (which seems to be almost restricted to the lower half of valve height); the ventral snap-mechanism is less developed; and its larger size (holotype: length/height D 0.48/ 0.27 mm; Coryell & Fields 1937;Bold 1967; see also e.g. Teeter 1975;Bold 1988). For differences in hinge and ornamentation see above.
Pellucistoma magniventra Edwards, 1944 (first record: North Carolina, Pliocene) is distinguished by (largely based on the redescription of Garbett & Maddocks 1979): its more arched, upwards rising dorsal margin and its strongly projecting posteroventral margin, respectively (if not aligned to the base line D reversal points in front and backwards of the ventral concavity); its much more infracurvate anterior margin; its more pointed and acuminate caudal process; its ornament (see above); its anterior vestibulum, which is largely restricted to the anteroventral area (for variability see Garbett & Maddocks 1979); details of the hinge (crenulated median element; simple anterior and posterior sockets/teeth; lack of posterior hinge elements); and its larger size (holotype: length/ height D 0.62/0.33 mm ;Edwards 1944;Garbett & Maddocks 1979; see also e.g. Puri 1960;Bold 1963;Benson & Coleman 1963;Hall 1965;Morales 1966;Grossman 1967;Sandberg 1969;Cronin 1979;Krutak 1982;King Lyon 1990).
Pellucistoma scrippsi Benson, 1959 (first record: Baja California, Recent) has more convex dorsal and ventral margins and a less oblique anterior margin as well as a smooth surface. The posterior vestibulum is almost absent (but see Swain & Gilby 1967); the hinge is slightly different (simple anterior socket (right valve) and tooth (left valve); posterior teeth are lacking (except illustration of Swain & Gilby 1967)); marginal pore canals are missing on the apex of the caudal process; and it is larger (holotype: length/height D 0.  Benson & Kaesler 1963;McKenzie & Swain 1967;Swain 1967;Swain & Gilby 1967, 1974. Pellucistoma bensoni McKenzie & Swain, 1967 (first record: Baja California, Recent) differs in its strongly arched dorsal margin resulting in a subtriangular outline and a posterocentrally located caudal process, its smooth surface, its almost absent posterior vestibulum; and it is larger (holotype: length/height D 0.44/0.25 mm).
Pellucistoma spurium Bold, 1963 (first record: Trinidad, late Miocene) has a more convex dorsal margin; a more accentuated caudal process; and a hinge with a simple knob-like anterior tooth (left valve) and minutely crenulated median elements. The anterior vestibulum is restricted to the anterocentral area, the posterior vestibulum is only developed at the caudal process; and it is larger (holotype: length/height D 0.49/0.25 mm; compare also Pellucistoma? spurium of Bold (1988)).
Pellucistoma santafesinensis Zabert, 1978 (first record: Argentina, middleÀlate Miocene; correct spelling according to Kempf (2008): P. santafesinense; note: in the following we refer to correct spellings of species names but retain the original spellings in this work) is similar to Pellucistoma gibosa Sanguinetti, 1979 (see below) and possibly both are synonyms. However, it differs significantly in outline (subtriangular-elongate in lateral view; ventromedian long, pointed caudal process) and hinge structures from the current material. Size of holotype: length/height D 0.53/0.26 mm.
Pellucistoma gibosa Sanguinetti, 1979 (first record: southern Brazil, late Miocene; correct spelling according to Kempf (1986): P. gibosum) has an extremely humped (right valve) dorsal margin and an acuminate, long caudal process well below the half valves' height. It is smooth; it has a hinge with a simple, strong anterior tooth (left valve) and a slightly serrated bar; the anterior vestibulum is restricted to lower half of valves height; and it is larger (holotype: length/height D 0.51/0.23 mm).
Due to their simple marginal pore canals, the species Pellucistoma? sp. (Bold 1958(Bold , 1972a, Pellucistoma? compactum Bold, 1972a andPellucistoma? kingmai Bold, 1972a, that are questionably attributed to Pellucistoma, differ in outline, hinge structure and development of the inner lamella. Only Pellucistoma sp. in Bold (1970Bold ( , 1972bBold ( , 1988 is rather similar in its shape to the current species. Nevertheless, it has a smooth surface and is larger (length/height D 0.48/0.25 mm). Unfortunately, neither the hinge nor inner characters are accessible, which does not enable further comparisons.
Pellucistoma tumida Puri, 1954 (correct spelling according to Kempf (1986): P. tumidum) from the Pliocene of Florida is poorly described and a re-examination of the type material already failed (Bold 1988). Although its outline is comparable with the present individuals (except the equicurvate anterior margin), without additional traits (see also Bold 1963;Hulings 1967) a generic assignment or a species-specific identification is unfeasible. Bold (1988) discussed possible congruence with his Phlyctocythere sp. 2, which again demonstrates the superficial similarity of that loxoconchid genus with Pellucistoma (see above).
Pellucistoma ovaliphylla Hu, 1981 from the Plio-/Pleistocene of southern Taiwan has been recognized by Hu (1984) to belong to Paradoxostoma Fischer, 1855. Two further Taiwanese species, Pellucistoma magnolioidea Hu & Tao, 2008 and Pellucistoma chushunshui Hu & Tao, 2008, differ notably in outline, especially due to the almost lacking caudal process, from both the current as well as from other Pellucistoma species (important internal characters are not accessible because of missing illustrations and descriptions). Most likely, these species belong to another genus (Paracytheroma?) but this claim needs additional investigations. Ayress (1990Ayress ( , 1996 described Pellucistoma coombsi Ayress, 1990, Pellucistoma fordycei Ayress, 1990 and Pellucistoma punctata Ayress, 1996 (correct spelling according to Kempf (2008): P. punctatum) from New Zealand and the Tasman and Coral Seas. These species diverge significantly from the present species and the genus Pellucistoma in general (outline: much more elongated-rectangular; hinge structures: e.g. P. coombsi and P. punctata have a right-bar hinge; inner lamella: much wider; ornament: P. punctata), which shed doubt on their generic allocation. However, those species are not comparable with the material described herein.
The illustration and description of Pellucistoma sp.  (Brandão et al. 2015), actually belongs to the former genus (Yasuhara et al. 2013).
To conclude, all the above compared species can be clearly differentiated from Pellucistoma curupira sp. nov. and are noticeably larger. Most similar are P. howei from the Miocene of Panama and Pellucistoma sp. of Bold (1970Bold ( , 1972bBold ( , 1988; Miocene: Antilles and Panama), however, the latter is little known and a closer examination is not possible. Most likely, the Australasian and Taiwanese species (Ayress 1990(Ayress , 1996Hu & Tao 2008) do not belong to Pellucistoma. The records from the Pitcairn Islands and southern Chile (Whatley et al. 1997b(Whatley et al. , 2004 need additional affirmation. Currently, we assume that the genus Pellucistoma is confined to the Americas.

Stable isotope (d 18 O and d 13 C) analyses
Due to the minute size of Pellucistoma curupira sp. nov., geochemical analyses were performed on three Cyprideis species, co-occurring in the same sample (AM10/30). All measurements provided very light values with a range for d 18 O from À7.25 to À10.41% and for d 13 C from À8.74 to À13.54% (Fig. 4).

Spatiotemporal distribution and autecology of hitherto known Pellucistoma species
We evaluated all published Pellucistoma records (fossil and Recent) known to us ( Fig. 5; for references and details see Supplemental Material 1 and 2). This review might be incomplete due to overlooked literature and, probably more importantly, because of sampling biases. Partially poor stratigraphical assessment of sampling sites and the inclusion of reports without illustrations may additionally blur our results. Nevertheless, the dataset reveals some significant biogeographical and ecological patterns, which further enable a discussion of the erratic occurrence of P. curupira sp. nov. in the Miocene of western Amazonia.
Amongst extant Pellucistoma species, P. magniventra has a continuous Holocene record along the east coast of the USA (Maryland (»38 N) to southern Florida) and in the northern and southern Gulf of Mexico (Fig. 5). This is the 'core area' of P. magniventra today. Further scattered indications come from the Pacific coast of Mexico (Sinaloa, Nayarit, Gulf of Tehuantepec) and Colombia (Bah ıa de Tumaco), from Puerto Rico, Venezuela and Trinidad and the Brazilian equatorial shelf (»2 S; the latter as P. ex gr. magniventra). To summarize the ecological data (Supplemental Material 1), P. magniventra prefers shallow marine (inner sublittoral, »10À20 m water depth), subtropicalÀtropical (minimal water temperature of coldest month (T min ) »10 C; average annual water temperature (T av ) >20 C), oxygenated, euhaline waters (»30À40 PSU; e.g. Valentine 1971; King Lyon 1990; Fig. 5). Fossil occurrences (late MioceneÀPleistocene) of P. magniventra match with the 'core area' as outlined above. Additionally, it is reported from Cuba (middle MioceneÀPliocene), Trinidad (early MioceneÀPliocene) and northern Brazil (early Miocene). Pellucistoma aff. magniventra is noted from Costa Rica (late Miocene) and Hispaniola (middle MioceneÀearly Pliocene). Palaeoecological data for these fossil occurrences coincide with the ecology of Recent P. magniventra.
Pellucistoma scrippsi, up to now only with Recent and sub-Recent records, is restricted to the Pacific coast of North and Meso America (»Santa Barbara/USA (»34 N) to San Juan del Sur/Nicaragua (»11 N)). The rare Recent and Pleistocene findings of P. bensoni plot within the distribution area of P. scrippsi. For P. scrippsi and P. bensoni the following autecological data can be summarized: shallow marine (inner sublittoral, »10 m water depth), warm temperateÀtropical (T min »12 C, T av >15 C), euhaline.
Pellucistoma howei, although principally a fossil species (Panama and Dominican Republic; middleÀlate Miocene), is recorded from the Holocene of Belize as well as with uncertainty (P. cf. howei or P. aff. howei) from the Gulf of Mexico (Recent) and the Miocene of Costa Rica. Ecological data characterize it as a shallow marine (sublittoral, »5À50 m water depth), tropi-calÀsubtropical, euhaline species.
Except for a few further Recent Pellucistoma sp. records (Panama, Trinidad, Rocas Atoll/Brazil), P. elongata remains the last extant species to be discussed. It is confined to the Argentinean coast from about the Isla de Los P ajaros (»42 S) towards the north (»36 S) to the R ıo de la Plata estuary (with one Pleistocene report: Mar Chiquita). Further in the north, P. cf. elongata is noted off Cabo Frio (Rio de Janeiro, »23 S; subrecent). Based on the available information, P. elongata is a shallow marine (littoralÀinner sublittoral; littoral rock pools to »14 m water depth), euhaline species. It is reported from shallower settings than the species treated above, however, at least T min can be assumed to be comparable to the occurrences of other species (e.g. Mar del Plata: T min »10 C, T av »15 C; www.seatemperature.org, accessed 12 January 2015).
In summary, Pellucistoma is unquestionably a marine (sublittoral, euhaline) taxon of subtropicalÀtropical, oxygenated waters with a seasonally lower temperature limit of about 10 C. All fossil records point in the same direction. So far, there is no evidence for a different autecology in the geological past.

Constraints of dispersal and potential dispersal modes of Pellucistoma
Dispersal capacity of organisms depends on various biotic and abiotic factors (e.g. autecology, reproduction mode, predation, competition, active/passive dispersal capacity, medium of transport). In the case of Pellucistoma, water temperature, depth and salinity appear to be the most important physico-chemical parameters (e.g. Valentine 1971Valentine , 1976Bold 1974;Cronin 1979;Cronin & Dowsett (Gross et al. 2013). (Note: the indicated range for modern rivers and floodplain lakes is based on aragonitic mollusc shells , which give somewhat heavier values for the same environmental parameters compared to ostracod calcite (Grossman & Ku 1986)). 1990). Little is known about its reproduction; however, due to the proof of female and male individuals in some species, sexual reproduction can be assumed. As female carapaces lack an apparent brood pouch (for storage of eggs and/or early instars), brood care seems unlikely.
Planktonic larval stages À like in all marine podocopid ostracods À are missing (Titterton & Whatley 1988). Presumably, Pellucistoma belongs to the marine meiobenthos. Synecological information (e.g. predation, competition) is not available for this genus. By considering the small size of Pellucistoma and the lack of planktonic stages, active dispersal can be expected to be very slow (Sandberg 1964). Land bridges (e.g. Panamian isthmus), deep-water areas (e.g. Cayman Trench), cold waters/ocean currents (e.g. Peru Current) as well as massive river discharge (e.g. Amazon River; lowering of salinity, instability of the seabed) will be effective dispersal barriers for Pellucistoma (compare Cronin 1987;Coimbra et al. 1999;Iturralde-Vinent & MacPhee 1999).
Birds can transport ostracods (adults, juveniles, eggs) on e.g. their feet or feathers, preferably encased in sediment. In addition, intestinal transport (and survival) has been successfully demonstrated (e.g. L€ offler 1964;Frisch et al. 2007;Brochet et al. 2010). Although bird-mediated transport is conceivable for Pellucistoma, three counterarguments should be mentioned (Teeter 1973): (1) species of this genus (except P. elongata) prefer sublittoral settings, which makes their adhesion to shorebirds/waterfowls difficult; (2) Pellucistoma is rather small and thinshelled (P. curupira sp. nov. is very small and very thinshelled) and thus prone to digestion (also by fishes); and (3) torpid stages, desiccation-resistant eggs (as in freshwater Cypridoidea; e.g. Horne 1993; Rossi et al. 2011) and brood care (as e.g. in the cytheroid Cyprideis; Sandberg 1964; Bold 1976) facilitate aerial dispersal via birds, but the presence of such features is implausible for Pellucistoma. Comparable arguments seem to rule out transport by wind.
Human induced dispersal by ships (e.g. in ballast water or on fouling of the hull) has been inferred to influence modern ostracod distribution (Teeter 1973;Witte & Harten 1991). Although the evidence is quite poor, we would refer to the possibility of phoretic dispersal via larger aquatic animals. For instance, manatees (Sirenia: Trichechidae) carry diverse epibionts on their skin (e.g. barnacles, copepods, algae and ostracods; Hartman 1979; Su arez- Morales et al. 2010;Marsh et al. 2011). Possibly, such animals act À like vessels À as means of dispersal for shallow marine ostracods. In this case, the sublittoral lifestyle of Pellucistoma, its susceptibility to intestinal digestion and decease due to desiccation, will be no constraint, as in bird transport. In addition, stream, tidal and ocean currents can be surmounted.
In conclusion, due to the lack of specific, dispersalfavouring traits and its autecology, the colonization of new habitats is subject to more restrictions compared to, for example,theeuryoecious,broodcarepracticingCyprideis.

Origin and dispersal of Pellucistoma
Based on his extensive works on Caribbean ostracods, Bold (1974) previously outlined the dispersal of Pellucistoma. Bold suggested an early Miocene origin in northern South America and its subsequent spread towards the Lesser Antilles and Panama before the onset of the middle Miocene. Following our evaluations, the potentially earliest records stem from the Lares Formation of Puerto Rico (upper OligoceneÀlower Miocene; Fig. 6).
Either way, early Miocene occurrences are noted from Trinidad (Brasso Formation), Panama (Culebra Formation) and Brazil (Pirabas Formation), which support a substantial dispersal event in the early Miocene (Bold 1974). Afterwards (middle Miocene), Pellucistoma colonized the Greater Antilles (Hispaniola, Cuba) and arrived during the late Miocene at the south-eastern coasts of North America and the Pacific coast of Costa Rica. Amazingly, it also has been spread far to the south, to southern Brazil (Pelotas basin) and north-eastern Argentina (Paran a basin) at that time. During the Pliocene, Pellucistoma expanded along the Atlantic coast of the USA (up to North Carolina) and settled the Gulf of Mexico (P. magniventra).
Its Miocene and Pliocene dispersal in the Caribbean, Gulfian and Carolinian provinces (sensu Cronin 1987) can be explained by repeated, short-distance transport via floating water plants between the islands and along the shoals of the American continent, following the generally west and north-west directed sea surface currents (Bold 1974;Cronin 1987;Iturralde-Vinent & MacPhee 1999). More difficult to assert is the 'rapid' early Miocene spread of Pellucistoma towards the south-east (Pirabas Formation; »2000 km south-east of Trinidad), directed against the modern North Brazil Current. However, during the Miocene the Panama isthmus was open, causing ocean current patterns different from today. Due to the inflow of Pacific waters through the Central American Seaway, a reversed North Brazil Current, flowing along the northeast South American coast towards the south-east, has been proposed (Prange & Schulz 2004;Heinrich & Zonneveld 2013 and references therein). As neither the Orinoco River (D ıaz de Gamero 1996) nor the Amazon River (Hoorn et al. 2010b) were fully developed, a significant, riverine dispersal barrier (like the modern Amazon; Coimbra et al. 1999) was not installed during the early/ middle Miocene. Tentatively, a reversed Miocene North Brazil Current facilitated the south-eastward dispersal of Pellucistoma along the Brazilian shelf. Comparably, the colonization of areas far in the south (Pelotas and Paran a basin) in the late Miocene could have been triggered by the south-west directed Brazil Current (Wood et al. 1999). At this time, the cold Falkland Current is assumed to have been less effective (Coimbra et al. 2009;Le Roux 2012), which enabled the proliferation of Pellucistoma in the Paranaense Sea. Although highly speculative, larger animals (e.g. sirenians) might have accelerated the expansion along the east coast of South America by acting as ectophoretic vectors. This hypothesis receives some support by the proposed northÀsouth dispersal of sirenians along the eastern South American coast and their invasion of the Paranaense Sea from the south (V elez-Juarbe et al. 2012).
In warmer periods of the Pleistocene, Pellucistoma extended further northwards on the west Atlantic coast (P. magniventra; e.g. Valentine 1971;Cronin 1979;Cronin & Dowsett 1990) and started to settle the Pacific coast of the USA (P. bensoni; Valentine 1976). In southern South America, a new species appeared (P. elongata ;Ferrero 2009). Based on comparative morphology, P. elongata is much more closely related to P. magniventra than to the late Miocene species P. gibosa and P. santafesinensis. Thus, P. elongata is probably a descendant of P. magniventra and derives from a pre-Holocene dispersal event of the latter.
During the Holocene, Pellucistoma achieved its present distribution. Obviously, its latitudinal extension is limited by water temperature. While on the US Atlantic coast (»38 N) the cold Labrador Current forms a thermal barrier, on the east Pacific coast the California Current confines its northward dispersal (»34 N). On the south-west Atlantic coast (Argentina), the cold Falkland Current restricts its migration (P. elongata) further to the south. P. elongata has the highest latitudinal occurrence (»42 S) of all Pellucistoma species but it is also the species with the shallowest records (at 42 S: littoral rock pool). As it is found in restricted bays, locally significantly warmer waters can be assumed, which permit its survival. A substantial northward migration could be hindered by the freshwater discharge of the R ıo de la Plata (Whatley et al. 1998b).
Based on the fossil and Recent records, a huge distributional gap is obvious (Fig. 5). Pellucistoma is missing along practically the entire Pacific coast of South America (there is only one P. magniventra record (one valve); Bah ıa de Tumaco, Colombia). Potentially this is an enormous sampling bias; however, we expect the Peru Current impeded a successful settlement of Pellucistoma. The Peru Current has been in existence at least since middle Miocene times (Le Roux 2012 and references therein). Its cold waters and northward-directed drift probably hampered the dispersal of Pellucistoma on the western coast of South America from the Miocene up to present times.
Recent P. scrippsi is restricted to the west coast of North and Meso America. Off Sinaloa and Nayarit (Mexico), it co-occurs with P. magniventra. This sympatric occurrence could be a taxonomic artefact or À speculatively À the result of a quite recent, passive, manmade dispersal event, tracing major ocean lanes from the Panama Canal towards California (Teeter 1973). Comparably, the single P. magniventra record on the western Colombian coast might reflect such a scattered displacement.
The enigmatic occurrence of Pellucistoma curupira sp. nov. in western Amazonia: proof of marine incursions?
The spatiotemporal distribution pattern of thus far known Pellucistoma species can be explained by vicariant barriers (land bridges, water temperature, depth, salinity), as well as by passive transport (sea currents (drifting matter) and, possibly, phoresy). However, the crucial questions of the palaeoenvironmental implication and the provenance of our new species remain to be discussed.
Our appraisal of records of Pellucistoma demonstrates that it is a shallow marine taxon. No evidence of substantial deviations in habitat preferences or adaptations to non-marine environments has been reported so far. Consequently, the finding of P. curupira sp. nov. appears to be amongst the most solid (body fossil) biotic evidence for proposed marine incursions, affecting the centre of Amazonia in Miocene times.
Autecology of Pellucistoma curupira sp. nov. The new species, Pellucistoma curupira sp. nov., is evidently dwarfed in comparison to all other Pellucistoma species (in length »20À50% smaller) and very thin-shelled. There are multiple causes of size reduction or dwarfism in ostracods (temperature, oxygenation, salinity, food resources, etc.;e.g. Neale 1988;Majoran et al. 2000;Yin et al. 2001;Hunt & Roy 2006;Finston 2007;Hunt et al. 2010;Scheihing et al. 2011;Yamaguchi et al. 2012). Here, we cannot identify a single or several factors to be the reason for the small size of P. curupira sp. nov. Nevertheless, it obviously had to cope with some kind of environmental stress and lived close to its tolerance limits. Its extremely thin-shelled valves also support this theory (e.g. Frenzel & Boomer 2005 for references; compare also Vermeij & Wesselingh 2002 for marine-derived gastropods). During the preparation of this paper, Boonstra et al. (2015) reported Pellucistoma (conspecific with P. curupira sp. nov.; MIFR pers. obs.) from western Amazonia (middle Miocene; Nuevo Horizonte), accompanied by euryhaline foraminifers. Based on the low diversity foraminiferal assemblage and the high proportion of abnormal tests, these authors proposed a highly stressful habitat with poorly oxygenated bottom waters, close to freshwater conditions (Vonhof et al. 2003: <1 PSU). Hence, it seems plausible that low oxygenation ) and, in particular, low salinity are the main abiotic stress factors leading to the dwarfism and the poorly calcified valves of P. curupira sp. nov.
Pellucistoma curupira sp. nov. is associated in sample AM10/30 with a typical Pebasian ostracod fauna (Whatley et al. 1998a;Gross et al. 2013Gross et al. , 2014, totally dominated by Cyprideis (12 sympatric species; »99% of the total ostracod fauna). This ostracod genus is holoeuryhaline, able to survive hypoxic periods and especially successful in stressful settings (e.g. Gross et al. 2008 and references therein). The only other ostracods found in this sample are Perissocytheridea ornellasae (Purper, 1979), Rhadinocytherura amazonensis Sheppard & Bate, 1980 and Skopaeocythere tetrakanthos Whatley et al., 2000 (MG in prep.). These species are À like P. curupira sp. nov. À endemic for western Amazonia and supposedly of marine origin. Interestingly, they are also of minute size (»0.30À0.35 mm in length). An adaptation to freshwater settings during the late Miocene has been demonstrated for R. amazonensis and Perissocytheridea, as well as for Cyprideis spp. (Gross et al. 2013).
Stable isotope analyses (d 18 O, d 13 C) performed on three Cyprideis species associated with P. curupira sp. nov. yielded very light values (Fig. 4). Such depleted d 18 O and d 13 C ratios are indicative of a freshwater system (Leng & Marshall 2004). Our results are consistent with previous isotopic data obtained from outcrop material (ostracods: Gross et al. 2013). Earlier O/C-isotopic investigations (mainly molluscs) yielded closely comparable results (Vonhof et al. 2003;Kaandorp et al. 2006;Ramos 2006;. The morphology of P. curupira sp. nov. (very small and extremely thin-shelled), the associated ostracod fauna (cooccurrence of further stunted ostracods of potentially marine ancestry) and our geochemical evidence, suggest that this Pellucistoma species À exceptionally À managed to adapt to freshwater conditions in the late middle Miocene.
Provenance of Pellucistoma curupira sp. nov. We presume: (1) the genus Pellucistoma originated in the Caribbean realm around the Oligo-/Miocene boundary; (2) Pellucistoma curupira sp. nov. is most closely related to Caribbean species; and (3) Pellucistoma is a shallow marine clade but adapted (P. curupira sp. nov.) over geological timescales (about 10 million years) to freshwater conditions in western Amazonia. Based on this, we explore possible migration scenarios: (A) Migration of Pellucistoma curupira sp. nov. via aerial (bird) transport. Bird-mediated transport plays a certain role in ostracod dispersal. In such a case, Pellucistoma (and other small aquatic invertebrates, e.g. foraminifers) could have entered western Amazonia without aquatic connections (neither fluvial nor marine). By considering the palaeobiogeographical distribution of Pellucistoma, there are two potential sources (Fig. 6): the Caribbean and the Amazonian Sea. Due to the close relationship of P. curupira sp. nov. to Caribbean species and half the distance to travel, the first source seems more likely. Although less well developed than today, north-Àsouth bird migration was already in existence during the Miocene (Tambussi & Degrange 2013).
Nevertheless, Pellucistoma is not predisposed for aerial dispersal, and successful transfer as well as 'ad hoc' colonization of new (freshwater) habitats appears to be demanding. (Note: such a mode of migration is much more conceivable for the eurypotent Cyprideis. This ostracod achieved evidently a 'habitat monopoly' in the Miocene of western Amazonia, possibly causing additional biotic stress for other, rather stenopotent arrivals like Pellucistoma). However, a multitude (over millions of years) of short-distance transport 'accidents' through a patchy structured wetland is possible, enabling a stepwise, longdistance spread and gradual adaptation to freshwater. Even if this scenario applies for small invertebrates, it is hardly appropriate for larger, marine-derived vertebrates occurring in Amazonia (e.g. Lovejoy et al. 2006;Boonstra et al. 2015 and references therein).
(B) Migration of Pellucistoma curupira sp. nov. through marine incursions. Several marine pathways have been proposed to have linked western Amazonia with the sea during the Miocene (for comprehensive discussions see e.g . Nuttall 1990;Rebata et al. 2006;Wesselingh & Salo 2006;Hovikoski et al. 2010). Such connections have been envisaged towards the north (Caribbean Sea), the east (Amazonian Sea), the south (Paranaense Sea) and the west (Pacific, southern Ecuador).
Concerning Pellucistoma, we have no evidence for an eastern Pacific source, which corroborates low faunistic affinities in the mollusc record (Wesselingh & Salo 2006). Pellucistoma occurs in the Pirabas Formation (early Miocene) and the Amazonian Sea could be a potential source. Nevertheless, marine incursions, originating from the eastern Brazilian Atlantic, are unlikely since the 'Purus arch' formed a significant watershed at least until the late Miocene (e.g. Figueiredo et al. 2009;Hoorn et al. 2010b;Latrubesse et al. 2010;Dino et al. 2012).
Relations between aquatic biota of the Paranaense Sea and the Pebas system are poor (e.g. Marengo 2000;Hulka et al. 2006;Wesselingh & Salo 2006;Nicolaidis & Coimbra 2008). Similarly, P. curupira sp. nov. is not closer related to the species of the Paran a and Pelotas basins, for which we already suggested migration along the eastern coast of South America. Potential marine connections with the Paranaense Sea are dated to the late Miocene (Hovikoski et al. 2007;Uba et al. 2009). Thus, P. curupira sp. nov. (late middle Miocene) pre-dates the southern South American species and immigration from the south is improbable.
Most authors agree with a linkage between the earlyÀmiddle Miocene Pebas wetland and the Caribbean Sea through corridors in the Llanos basin (e.g . Nuttall 1990;Hoorn et al. 1995;Anderson et al. 2006;Lovejoy et al. 2006;Wesselingh & Macsotay 2006). Morphological affinities of P. curupira sp. nov. to the Caribbean also support this claim. Hence, a CaribbeanÀLlanosÀPebasdispersal pathway is most plausible; however, fossil evidence is pending for Pellucistoma in the Llanos region.
Marine incursions, deriving from the Caribbean realm, have been assumed to affect western Amazonia's environments and biota (e.g. Boonstra et al. 2015). At first glance, our Pellucistoma record is a profound confirmation of such far-reaching marine influences, as it could have followed the ingressions. But, as argued above, this record does not confirm the influx of marine waters. To clarify, we do not reject the possibility of marine incursion throughout the entire Miocene history of western Amazonia. For instance, the probably slightly older middle Miocene Pellucistoma records from oligo-/mesohaline layers of Nuevo Horizonte (Boonstra et al. 2015) could be interpreted to mirror an incursion-related immigration as well as the stepwise freshwater adaptation of P. curupira sp. nov. Nevertheless, in our case direct marine connections are not mandatory.
(C) Migration of Pellucistoma curupira sp. nov. through aquatic pathways. The Caribbean-derived Pellucistoma probably reached the Llanos Basin (like other marine biota) during sporadic marine incursions in the early/middle Miocene (e.g. Wesselingh & Macsotay 2006;Jim enez & Hammen 2007;Bayona et al. 2007;Gomez et al. 2009;Boonstra et al. 2015 and references therein). Larger animals of marine ancestry (e.g. fishes, dolphins, manatees; Lovejoy et al. 2006;Lundberg et al. 2010;Bianucci et al. 2013) were able to migrate actively into the Pebas system via fluvial pathways. Smaller biota (e.g. ostracods, foraminifers, mollusc larvae) could have entered this 'mega-wetland' actively too but, more likely, could have been freighted ectophoretically and gradually developed freshwater tolerance over long timescales (e.g. P. curupira sp. nov.). Thus, "it would not seem to be necessary for the connection between the sea and the heart of the basin to be direct at any one time. A series of lakes continually splitting and merging with each other, or perhaps becoming reconnected by streams, would enable taxa to progress gradually from one area to another." (Nuttall 1990, p. 351;compare Lundberg et al. 2010).
Based on our investigations and the data available, we favour hypothesis (C). However, we cannot explicitly reject hypotheses (A) and (B), which might also contribute to the enigmatic occurrence of Pellucistoma curupira sp. nov.

Conclusions
Based on our in-depth evaluation of a new species of the ostracod genus Pellucistoma from the late middle Miocene of western Amazonia, we conclude that this genus is: (1) biogeographically restricted to the Americas; (2) in general a typical shallow marine clade; and (3) of Oligocene/early Miocene Caribbean origin. We assume that Pellucistoma entered the Llanos Basin during the early Miocene, migrated into the fluvio-lacustrine Pebas mega-wetland by phoresy through aquatic (fluvial) connections and adapted to freshwater conditions. Our finding emphasizes again that palaeoenvironmental interpretations based on a straightforward application of uniformitarian principles are problematical for the endemic biota of western Amazonia (Wesselingh 2006;Gross et al. 2013). Thus we conclude that this record of Pellucistoma is not evidence for marine incursions.