Saline Waters in Miocene Western Amazonia – An Alternative View

Before the onset of the modern Amazon River system, northwestern South America was shaped by an extensive wetland during the Miocene. This “Pebas mega-wetland” kept a well-renowned endemic mollusk and ostracod fauna, which initiated a persisting debate about marine ingressions reaching the center of Amazonia at that time. Due to high endemism, uniformitarian principles are hardly applicable to this biota, but also other paleontological, sedimentological, and geochemical information led to ambiguous paleoenvironmental interpretations. Here, we investigate the ostracod and foraminifer assemblages and the oxygen and carbon stable isotope compositions of their biogenic calcite from an outcrop at the cutbank of the Amazon River (NE-Peru, ∼55 km S of Iquitos). While ostracods (e.g., the genus Cyprideis) are able to calcify their carapaces along the entire salinity range, at least low saline conditions are a prerequisite for the biomineralization of calcareous foraminiferan tests. Hence, the finding of calcareous foraminifers (Ammonia, Elphidium), associated mainly with brackish water ostracods indicates the presence of saline waters. In contrast, δ18O and δ13C analyses performed on co-occurring ostracod valves and foraminiferan tests yielded constantly very light ratios. Such values refer to a pure freshwater environment and are incompatible with the interference of isotopically heavier, marine waters or a stable isotope enrichment by evaporation. Based on these contrasting data, we hypothesize that the Pebas mega-wetland was episodically influenced by saline but isotopically light groundwater discharge. Possibly, the resulting specific hydrochemistry not only contributed to the evolution of the endemic Pebasian fauna but also facilitated the sporadic settlement of euryhaline foraminifers, which mimics short-lived marine incursions.

The conflict about the salinity regime of the Pebas wetland is well exemplified by Wesselingh et al. (2006b). Within one outcrop (Santa Rosa de Pichana; ∼39 km SSE Pebas; Figure 1A), mollusk assemblages and O/C isotopes from mollusk shells indicate a freshwater environment, whereas ichnofossils point to mesohaline conditions (compare Gingras et al., 2002 andHovikoski et al., 2007 for Pebasian ichnofossils). These authors proposed two explanations: (i) ichnofossils and mollusks lived in separate time intervals and environments, not resolved by the samples or by the preserved body fossils; or (ii) the producers of ichnofossils have adapted to freshwater like an array of other marine-derived biota (e.g., Bloom and Lovejoy, 2017).
Here, we investigate for the first time the δ 18 O-and δ 13 C-isotopy of calcareous ostracod and foraminifer shells cooccurring within the same layer of one outcrop (Porvenir, ∼55 km S Iquitos; Figure 1B). By this bed-by-bed analysis of microfossils, we avoid potential sampling gaps (Wesselingh et al., 2006b;Hovikoski et al., 2007Hovikoski et al., , 2010. While ostracods (bivalved crustaceans) settle almost every aquatic habitat, calcareous foraminifers (unicellular eukaryotes) are restricted to marine habitats or saline inland lakes (e.g., Horne et al., 2002;Iglikowska and Pawłowska, 2015). Hence, the assumption of an adaptation of calcified test bearing foraminifers ("supraliminal evolution"; Myers, 1960;Wesselingh, 2007) to a freshwater environment is much more challenging than for ostracods (Gross et al., 2013(Gross et al., , 2016. Based on our again puzzling paleontological and O/C-isotope data, we explore the possibility of marine incursions affecting the Pebas wetland as well as alternative explanations for the occurrence of foraminifers more than 1,000 km away from the next assumed paleocoast (Boonstra et al., 2015).

Description of the Porvenir Section
Bed 1 consists of more than 1.15-m-thick, green-gray, massive clayey silt. Centimeter-thick Thalassinoides burrows occur in the upper 0.30 m and are filled with dark gray pelite and mollusk shells. Bed 2 comprises 0.80-m-thick, mollusk-rich, dark gray, massive pelite, which is burrowed by Thalassinoides in the uppermost part. Above (bed 3) follow 0.95 m of green gray, partly brown mottled, massive fine sandy pelite with abundant mollusk shells at the top and base. Centimeter-scale calcic concretions and rare turtle remains have been observed. Bed 4 is formed by 2.50 m of dark brown to gray, massive clay with coaly (i.a., leaf remains) and silty fine sand layers in the lower 0.80 m. The uppermost 0.10 m contain coaly layers (i.a., leaf fragments) and merge gradually into a brown, laminated lignite (bed 5). Up-section (bed 6), 0.70 m of mollusk-rich, gray to gray brown, massive silty fine sand and 0.25 m of brown gray, massive fine sandy silt (bed 7), rich in mollusks, follow; 0.30 m of mollusk-rich, green gray, massive fine sandy silt form bed 8. Bed 9 comprises 2.80 m of blue gray, massive clay with some discontinuous fine sand layers,  Table S1). Several species are left in open nomenclature due to limited material or represent presumably new species. A more detailed treatment of these specimens is beyond the scope of the current study.
Three species of Rhadinocytherura Sheppard and Bate, 1980 have been found at Porvenir among them one could be determined on species level (Rhadinocytherura amazonensis Sheppard and Bate, 1980). Rhadinocytherura sp. 1 is similar to R. amazonensis but displays a distinct difference in ornament and is probably a new species. Muñoz-Torres et al. (1998, pl. 6, Fig. 10) figured valves under R. amazonensis, which are identical with Rhadinocytherura sp. 2 herein. However, because of marked divergences in outline and ornament, we consider Rhadinocytherura sp. 2 as a separate species, which will be described elsewhere. Rhadinocytherura, a genus endemic for western Amazonia, is supposed to be a marine to brackish water taxon (Sheppard and Bate, 1980).
Skopaeocythere tetrakanthos Whatley et al., 2000 corresponds to the specimens from Porvenir and those in Linhares et al. (2017). It is the only species of this genus, endemic for western Amazonian and considered to be a brackish water form (Whatley et al., 2000).

Results of Stable Isotope (δ 18 O-and δ 13 C) Analyses
Well-preserved, translucent (foraminifers) or translucent to slightly milky (adult ostracods) specimens were selected for geochemical analyses from layer PO-N-7 and PO-N-8 (Table 1, Figure 5). Both layers contain the richest, most diverse, and best-preserved calcareous microfossil fauna with an admixture of typically freshwater, brackish water, and marine taxa. All measurements furnished very light δ 18 O and δ 13 C ratios (for details, see Table 1)

Calcareous Microfossils
Charophyte gyrogonites, found at Porvenir in all productive samples, normally point to shallow freshwater bodies (Gennari et al., 2011). However, a few stoneworts live in (or tolerate) brackish, marine, and also hypersaline waters, which limits their paleoecological indication without a more detailed determination (e.g., Soulié-Märsche, 2008; Soulié-Märsche and García, 2015). Species of the genus Cyprideis strongly dominate the ostracod faunas of Porvenir (Figure 2). Other taxa occur in minor amounts. All recovered ostracod species and the extinct genera Rhadinocytherura and Skopaeocythere are endemic for western Amazonia. Extant Cyprideis and Perissocytheridea are typical brackish water elements; darwinulids (here Alicenula) dwell usually freshwaters, and Pellucistoma is a full marine taxon (for details, see Systematic Notes on Ostracods and Foraminifers). Rhadinocytherura and Skopaeocythere are allegedly marine to brackish water ostracods (Sheppard and Bate, 1980;Whatley et al., 2000).
Deduced from sedimentological and geochemical data, Gross et al. (2011Gross et al. ( , 2013 proposed the occurrence of Cyprideis, Perissocytheridea, and Rhadinocytherura in fluvial (freshwater), post-Pebasian deposits. Also an exceptional adaptation of Pellucistoma (and consequently of the co-occurring Skopaeocythere) to freshwater conditions has been discussed (Gross et al., 2016). Hence, none of the recorded ostracod species provides a doubtless clue for the presence of brackish waters or marine incursions, which is, however, clearly contested by the finding of calcareous foraminifers.
Hence, among a wealth of endemic forms, few freshwater and marine taxa co-occur at Porvenir (Figure 2). This is well comparable to the microfossil record and a common pattern for Pebasian strata (e.g., Wesselingh et al., 2006b;Hovikoski et al., 2007Hovikoski et al., , 2010. As already emphasized by Wesselingh et al. (2002;Wesselingh, 2007), the endemic mollusk (and ostracod) faunas indicate a long-lived, somehow isolated aquatic setting and conflicts with regularly open, marine connections (e.g., estuaries). A generally brackish water interpretation (e.g., lagoons, saline inland lake) is at odds with the highly diverse endemic faunas, the frequent occurrence of in situ stenohaline freshwater taxa, and the lack of otherwise common marginal marine mollusks. "Simple" floodplain environments (like modern Amazon's "várzeas") mismatch by their low diverse, obligate freshwater and terrestrial faunas too (e.g., Wesselingh et al., 2002;Wesselingh, 2007). These authors concluded a dynamic, overall freshwater megawetland, influenced by brief episodes of marine ingressions (see also Hoorn et al., 2010a;Jaramillo et al., 2017). Hovikoski et al. (2007) studied the trace fossil assemblages of Porvenir. The beds containing herein reported foraminifers were interpreted as deltaic and lagoonal deposits within a bay-margin succession. Among others, the ichnofossil Thalassinoides, related to thalassinid shrimp burrows, has been inferred to indicate mesohaline to marine salinity conditions (Gingras et al., 2002;Hovikoski et al., 2007). These authors refused other decapods as producers of these Thalassinoides traces, as well as an adaptation of their constructors to freshwater (Wesselingh et al., 2006b).

Ichnofossils
However, the only more thoroughly known decapod remains from the Pebas/Solimões Fm. belong to today exclusively freshwater-dwelling trichodactylid crabs (Klaus et al., 2017; FIGURE 5 | δ 18 O-and δ 13 C-isotopic ratios of foraminiferan tests, ostracod valves, and mollusk shells from Porvenir (for data, see Table 1). Ellipses indicate the range of modern Amazonian river and floodplain lake dwelling mollusks (modified from Wesselingh et al., 2006b; note: mollusk aragonite furnish slightly heavier ratios than ostracod calcite within the same environment; Grossman and Ku, 1986).

Paleoenvironmental Inferences From the Geochemical Record
In contrast to the paleontological record, geochemical data display a uniform pattern. Consistently, O/C-stable isotope analyses performed on mollusk, ostracod, and foraminiferan shells resulted in very light values (e.g., Vonhof et al., 1998Vonhof et al., , 2003Kaandorp et al., 2005Kaandorp et al., , 2006Wesselingh et al., 2006b;Gross et al., 2013Gross et al., , 2016 and Figure 5 herein). Significantly diverging O/C-isotopic compositions of rainwater, as, for example, modeled for the Eocene of the Tibetan Plateau (Botsyun et al., 2019) are not projected for the Miocene of western Amazonia. The impact of slightly differing paleolatitude or higher Mid-Miocene temperatures and atmospheric CO 2content are assumed negligible or irresolvable to date. The source of vapor (Atlantic Ocean), its distance from the ocean (causing "continental effects"), and amount (despite probably lower Andes) of precipitation are considered comparable to modern ones (Kaandorp et al., 2005(Kaandorp et al., , 2006; see also the modeling of Insel et al., 2010 andSepulchre et al., 2010). Nevertheless, the effect of the enormous Pebas wetland itself to the Amazonian hydrological cycle remains incalculable so far due to vague boundary conditions (i.e., water surface area and water depth at a specific stratigraphic interval; Kaandorp et al., 2006 and S. Richoz, pers. comm.).
A diagenetic alteration of the isotopic signal is improbable for the mineralogically pristine material and due to the preserved seasonal δ 18 O-cycles (Vonhof et al., 1998(Vonhof et al., , 2003Kaandorp et al., 2006). Biologically induced offsets ("vital effects") of δ 18 O-values from isotope equilibrium may range in the order of about + 2 but will not substantially affect the general isotope signal (Marco-Barba et al., 2012;Holmes and De Deckker, 2017;Bright et al., 2018a). Variations of δ 18 O recorded in the growth increments of Pebasian mollusks were related to distinct annual changes in precipitation (Kaandorp et al., 2005(Kaandorp et al., , 2006, which might have also affected the chemistry of ostracod valves. Because the isotopy of foraminifers corresponds well with that of the ostracods, the measured O/C ratios could reflect the dry season with the most evaporated and most 16 O-depleted waters. All here and previously measured biogenic carbonates furnished well-comparable O/C-values. This indicates that all biota calcified within the same ambient water and precludes reworking or faunal mixing.
An extensive 18 O depletion of marine waters reaching western Amazonia by incursions due to strong freshwater input (e.g., Andean runoff, precipitation) is possible. Although such diluted marine waters will be low saline, the amount of essential solutes (e.g., Ca +2 ) could be sufficient for euryhaline organisms to survive (Bright et al., 2018a,b). However, 87 Sr/ 86 Sr-isotope ratios of Pebasian mollusks do not suggest the influence of seawater (except arguable results from Buenos Aires/Colombia; Vonhof et al., 1998Vonhof et al., , 2003. Hence, the obtained light O/C ratios are incompatible with the influx of marine, isotopically heavier waters and exclude substantial evaporative effects ( 18 O-enrichment) as well (Wesselingh et al., 2002;Leng and Marshall, 2004;Hovikoski et al., 2010). From O/C-isotopic point of view, the analyzed taxa dwelled in a pure freshwater environment.
To sum up, within the same layers an admixture of usually freshwater, brackish-water, and marine-water biota occurs at Porvenir. This is probably best explained by a brackish (oligohaline) salinity regime, like in an estuarine system (Boonstra et al., 2015). However, a fully freshwater O/C-signal, earlier Sr-isotope analyses (Vonhof et al., 2003) and the faunal composition conflict with this interpretation and render marine incursions questionable. Short-term salinity oscillations not resolved by sampling, an alteration (including vital effects) of the geochemical signal, and mixing of faunas can be excluded (Wesselingh et al., 2006b;Hovikoski et al., 2007Hovikoski et al., , 2010. Therefore, other interpretations about the nature of the Pebas system with its unique fauna and freshwater geochemical signature are discussed below.
Notably, western Amazonian Cyprideis consistently lack hollow protuberances ("nodes") on their shells and bear round sieve pores (e.g., Muñoz-Torres et al., 1998;Gross et al., 2014; for extremely rare exceptions, see Linhares et al., 2011). Whereas a dominance of unnoded Cyprideis valves usually points to mesohaline (>14 PSU) salinity regimes, round sieve pores occur characteristically in oligohaline (<1 PSU) waters (Frenzel et al., 2012, 2017 andreferences therein). This discrepancy could be taxon-specific or related to a peculiar ion composition of the ambient water (e.g., van Harten, 2000;Pint et al., 2012;Grossi et al., 2015). For instance, adequate Ca 2+ supply has been demonstrated to reduce the proportion of noded Cyprideis valves compare Keyser, 2005;Keatings et al., 2007). Perhaps, also the SO 4 2− content of the host water plays an additional role in node formation as the sulfate (and chloride) concentration is negatively correlated to node frequency .
Hence, not salinity-the total amount of dissolved solidsper se but the ready presence of specific ions could be the controlling parameter for the occurrence of marine ostracods and foraminifers in the Pebas system (Forester, 1983(Forester, , 1986Flako-Zaritsky et al., 2011;Pint et al., 2015;Bright et al., 2018b).
If we assume a continental freshwater environment, shaped by precipitation and Andean runoff, and exclude marine ingressions and significant evaporative processes, possibly the inflow of saline, Andean, and/or basement-derived waters accounts for the solute composition of the Pebas wetland.
Until the late Miocene, the Huallaga basin, for example, was open to the Pebas system (Hermoza et al., 2005;Roddaz et al., 2005b). It is possible that saline surface waters, originating from Andean evaporite dissolution, affected the shallow Pebas wetland. Nevertheless, high freshwater input (rainwater, surface runoff) probably diluted such waters quickly over long distances (e.g., distance Huallaga evaporite domes-Iquitos > 450 km) and obliterated any effect on Pebasian environments.
Another possibility is that mineralized but O-isotopically light subsurface waters (Jørgensen and Holm, 1994;Wennrich, 2005) have influenced the Pebas wetland. Synsedimentary faults due to forebulge dynamics (e.g., Iquitos arch; Roddaz et al., 2005a,b; could have enabled the ascent of ion-rich groundwater during deposition of the Pebas Fm. An upward migration of brackish to hypersaline waters through faults can be still observed in western Amazonia nowadays (Stallard and Edmond, 1983;Rosário et al., 2016). Presumably, such groundwater inferences would be rather local phenomena. However, they could have occurred repeatedly and widespread in the Pebas system, mimicking short-lived marine incursions.
Possibly, the mixture of groundwater and freshwater resulted in a distinct hydrochemistry of Pebasian waters suitable only for few euryoecious biota. For example, athalassic ostracod distributions are sensitive to specific ion compositions (e.g., Ca 2+ , Cl − , SO 4 2− ; Stout, 1981;Keatings et al., 2007;Pint et al., 2015). Especially, Cyprideis seems to thrive in sulfate-and chloride-rich continental waters (Mezquita et al., 1999;Keatings et al., 2007;Bright et al., 2018a). The episodic influx of dissolved CaSO 4 (and NaCl) could have favored the occurrence of Cyprideis but hampered the settlement of cypridoid freshwater ostracods and ordinary freshwater mussels Keatings et al., 2007;Wesselingh, 2007). Perhaps, an unusual Ca 2+ , Cl −, and SO 4 2− supply not only prevented the formation of noded Cyprideis valves but also facilitated the erratic occurrence of marine ostracods and foraminifers.

CONCLUSION
In accordance to earlier studies on the mollusk contents, the microfossil fauna of Porvenir is by far dominated by endemic species of the "brackish water" ostracod Cyprideis. Ordinary freshwater and marine taxa (including foraminifers) occur in minor proportions. Solely based on actualistic evidence, a brackish water paleoenvironment is indicated, which agrees with repeatedly suggested, brief marine connections of the Pebas wetland to the Llanos basin and further to the Caribbean Sea.
Nonetheless, a simple application of uniformitarian principals to the highly endemic Pebasian fauna (and trace fossils) remains ambiguous. Geochemical data (O/C-and Sr isotopes) clearly point to a non-marine-influenced, freshwater environment.
Admittedly, the herein studied Porvenir section covers only a very small portion of the entire spatiotemporal extent of the Pebas Fm. and Pebas system, respectively (∼10 m vs. several hundreds of meters' thickness; several thousands to few hundred thousands of years vs. several millions of years; e.g., Hoorn et al., 2010a;Jaramillo et al., 2017). However, similar observations were reported from a variety of sites distributed over a wide area of the Pebas system. Thus, our interpretation of such opposing data could be a clue for the whole system or, at least, a novel perspective.
We hypothesize that the Pebas wetland received some influx of "saline" but isotopically "fresh" groundwaters resulting in a complex water chemistry. Such scenario would explain (i) the high endemicity of the Pebasian mollusk and ostracod fauna, (ii) the dominance and diversification of only few groups, (iii) the lack of other common marginal marine taxa, (iv) the low abundance of otherwise widespread freshwater elements, (v) the freshwater isotope signature of calcareous shells, and (vi) the sporadic existence of marine ostracods and foraminifers close to their tolerance limits far inland.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
MG and WP investigated the stratigraphical section. MG studied the fossil content, prepared all figures and wrote the manuscript. WP performed stable isotope analyses and contributed by discussions to the final version of the manuscript.