Abstract

Foraminiferal assemblages in the stratigraphically lower part of the Bouse Formation in the Blythe basin (lower Colorado River corridor, western USA) indicate marine conditions, whereas assemblages in the upper part of the Bouse Formation indicate lacustrine conditions and suggest the presence of a saline lake. Benthic foraminiferal assemblages in the lower part of the Bouse Formation are similar to lagoonal and inner neritic biofacies of the modern Gulf of California. Evidence suggesting a change from marine to lacustrine conditions includes the highest occurrence of planktic foraminifers at an elevation of 123 m above sea level (asl), the change from low diversity to monospecific foraminiferal assemblages composed only of Ammonia beccarii (between 110 and 126 m asl), an increase in abundance of A. beccarii specimens (above ∼110 m asl), increased number of deformed tests (above ∼123 m asl), first appearance of Chara (at ∼85 m asl), lowest occurrence of reworked Cretaceous coccoliths (at ∼110 m), a decrease in strontium isotopic values (between 70 and 120 m), and δ18O and δ13C values similar to seawater (between 70 and 100 m asl). Planktic foraminifers indicate a late Miocene age between 8.1 and 5.3 Ma for the oldest part of the Bouse Formation in the southern part of the Blythe basin. Benthic and planktic foraminifers correlate with other late Miocene sections in the proto–Gulf of California and suggest that the basal Bouse Formation in the Blythe basin was deposited at the northern end of this proto-gulf. After the marine connection was restricted or eliminated, the Colorado River flowed into the Blythe basin, forming a saline lake. This lake supported a monospecific foraminiferal assemblage of A. beccarii until the lake spilled into the Salton Trough and the Colorado River became a through-flowing river.

INTRODUCTION

The depositional setting of the Bouse Formation, preserved as a series of basins along the lower Colorado River corridor, is controversial and is variously attributed to a lacustrine origin in a series of lakes, or a marine incursion related to the proto–Gulf of California, which affected the Bouse Formation in the Blythe basin. The lacustrine origin suggests deposition of the Bouse Formation in a series of lakes that filled and spilled from one basin to the next until system was integrated and the Colorado River was a through-flowing river (House et al., 2005a, 2005b, 2008; Spencer et al., 2005, 2008). Arguments in favor of a lacustrine origin of the Bouse Formation are based on isotopic analyses (Spencer and Patchett, 1997; Buising, 1990; Poulson and John, 2003; Roskowski et al., 2010; Spencer et al., 2013), step-like maximum elevations of the Bouse paleolakes suggesting that no uplift or southward tilting has occurred since deposition of the unit (Spencer et al., 2008, 2013), and sedimentological evidence of floodwater influx derived from northern sources immediately preceding Bouse Formation deposition in the Mohave and Cottonwood Valleys (House et al., 2008; Spencer et al., 2013). An alternative interpretation suggests that the Bouse Formation accumulated in lakes along the northern part of the lower Colorado River corridor, but in the Blythe basin, the formation was initially deposited in a marine environment at the northern end of the proto–Gulf of California (Buising, 1988, 1990). Arguments in favor of the marine origin for the lower part of the Bouse Formation in the Blythe basin are the presence and distribution of marine fossils.

The purpose of this paper is to document the presence of foraminifers in the Bouse Formation of the Blythe basin and determine if they existed in a marine, saline lake, or mixed environment. This goal was achieved by reexamining previously collected micropaleontology slides and residues that could be found (∼74) and examining new samples collected in the past 15 yr (∼100). The Bouse foraminiferal faunas were compared to foraminiferal faunas in saline lakes and estuaries. In addition, the presence of Charophytes and reworked microfossils was noted. Arguments supporting the presence of a saline lake were examined in light of the foraminiferal analysis. Flood deposits, which are cited as evidence of lake deposition, do not occur or have not been found in the Blythe basin; mapping, sedimentary studies, and structural analysis of the area are inadequate to consider uplift or the lack of uplift as a viable means of proving deposition in a marine or saline lake environment. Many of these studies are currently ongoing. Isotopic data are frequently used to advocate a lacustrine origin for the Bouse Formation. Although the Sr data suggest deposition in a lacustrine environment, the data are limited for the older or basal Bouse Formation, and in the interval in which marine microfossils and high Sr values overlap there is evidence of mixing of marine and river water. The high strontium isotope values characteristic of the Bouse Formation are considerably higher than modern Colorado River values, suggesting that additional study is needed to understand the source and concentration of the strontium before the marine or nonmarine signal can be determined. Carbon and oxygen isotopic ratios are also limited and the marine or nonmarine interpretation of these values is reexamined.

GEOLOGIC SETTING

The Bouse Formation (Metzger, 1968) unconformably overlies a Miocene fanglomerate and is overlain by the Bullhead alluvium in the Blythe basin, and is composed of a basal limestone overlain by silts, clays, sands, and a tufa (Fig. 1). Metzger (1968) designated 233.8 m of sediments penetrated in the U.S. Geological Survey test well Mf11684 as the type section for the Bouse Formation (Fig. 2). Two additional reference sections were also designated: (1) outcrops south of Bouse Wash and east of the Colorado River flood plain, where 65.5 m is exposed; and (2) outcrops in sections 9 and 16, southeast of Cibola, where 30.5 m of the basal limestone is exposed (Metzger, 1968). Approximately 46 m of the reference section south of the Bouse Wash was resampled for paleomagnetics and microfossils (D. Malmon, 2008, personal commun.; Malmon et al., 2011). Only spot samples were taken from the reference section southeast of Cibola, although a section has been recently sampled for microfossils in Hart Mine Wash near Cibola (this paper; Fig. 2).

The Bouse Formation was defined in the Blythe basin and its geographic extent has since been expanded to include the Mohave-Cottonwood, Chemehuevi, and Bristol basins along and near the lower Colorado River corridor in Arizona, California, and Nevada (Metzger, 1968; Olmsted, 1972; Mattick et al., 1973; Metzger et al., 1973; Metzger and Loeltz, 1973; Olmsted et al., 1973; Carr and Dickey, 1980; Buising, 1988; Turak, 2000; House et al., 2008). Except for a disputed outcrop near the Laguna Diversion and Imperial Dams (sample locality J219 of Winterer, 1975; see also Olmsted et al., 1973, fig. 15 therein), the Bouse Formation is identified only in the subsurface in the Yuma area of Arizona (Olmsted et al., 1973; McDougall, 2008a; Spencer et al., 2013). Numerous core holes and wells that encountered the Bouse Formation in the subsurface of the Blythe basin were summarized previously (Metzger et al., 1973; Olmsted et al., 1973; P.J. Fritts, 1975, unpublished report, provided by K. Howard, 2012, personal commun.; Winterer, 1975; Turak, 2000). The Bouse Formation is identified in the Blythe basin between –173 m below sea level (bsl) and 330 m above sea level (asl), (Metzger et al., 1973; Buising, 1990; Spencer et al., 2008, 2013).

The Bouse Formation was tentatively assigned to the Pliocene based on post-Miocene fossils in sections near Yuma (Metzger, 1968; Smith, 1970; Metzger et al., 1973). An ash bed stratigraphically located in the upper part of the Bouse Formation in the southwest Blythe basin, (near Buzzards Peak) and in the Bristol basin (near the town of Amboy) correlates with the 4.83 Ma Lawlor Tuff (Reynolds et al., 2008; Sarna-Wojcicki et al., 2011). A lower age limit for the Bouse Formation in the Blythe basin is from the underlying Osborne Wash strata of Buising (1988, p. 12) in the western Buckskin Mountains. A basalt near the top of the Osborne Wash strata (Buising, 1988) is dated as 9.60 ± 0.60 Ma (Fugro, Inc., 1975, cited inReynolds et al., 1986; Buising, 1988) and a tuff bed from the upper clastic unit (i.e., equivalent to Osborne Wash strata of Buising, 1988) is 9.2 ± 0.3 Ma (Buising and Beratan, 1993). These dates suggest that deposition of the Bouse Formation in the Blythe basin began after 9.2 Ma but ceased after ca. 4.83 Ma, when the water breached the paleodam and spilled into the next basin.

PALEONTOLOGY

Early paleontologic studies suggested that the Bouse Formation was deposited in a marine embayment at the northern end of the Gulf of California during the latest Miocene to Pliocene. This interpretation is supported by the presence of benthic and planktic foraminifers (Durham and Allison, 1960; Hamilton, 1960; Metzger, 1968; Smith, 1960, 1969, 1970; Winterer, 1975; Buising, 1988, 1990; McDougall, 2008a), as well as other invertebrates (Metzger, 1968; Taylor, 1983; A. Cohen, 2012, personal commun.) and vertebrates (Crabtree, 1989). Other fossils in the Bouse Formation include diatoms, ostracodes, coralline algae, barnacles, mollusks, gastropods, crabs, and fish (Table 1). The distribution of these groups is not addressed in this paper and needs additional study, as these groups include marine, brackish, and freshwater species. Our study focuses primarily on the foraminifers, which are primarily a marine group. Nonmarine occurrences of foraminifers are limited and distinctive, and therefore easily recognized.

Micropaleontologic samples collected and analyzed from various research efforts are documented in Tables 2–11. The taxonomy was evaluated and notes are given in the Supplemental File1. Some of the foraminiferal slides from outcrops and wells drilled in conjunction with the Lower Colorado River Project (LCRP) and examined by P.B. Smith (Metzger et al., 1973; Smith 1970) are available in the U.S. Geological Survey Micropaleontologic collection (Flagstaff, Arizona). Previously unpublished data were added to the checklists. Some of the material used to generate the published checklists has not been located at the U.S. Geological Survey or at the Arizona Geological Survey (Rauzi, 1999) and remains missing.

A couple of stratigraphic sections were examined (Bouse Wash, Big Maria Quarry, and Hart Mine Wash; Fig. 2), but most of the samples available are spot samples and have little stratigraphic information. The units in the Bouse Formation are time transgressive, so the stratigraphic correlation between sections and spot samples is difficult to infer. In this study and in previous studies (Spencer et al., 2013), elevation is used as a proxy to identify stratigraphic position because of the difficulty in defining stratigraphic position based on lithology. Although elevation is not a good proxy, until the amount of faulting and subsidence noted by Metzger et al. (1973) in the Blythe basin is understood and corrected for, elevation, at least, allows the samples to be placed in relation to each other. Metzger et al. (1973) noted regional downwarping of the Palo Verde Valley (southern Blythe basin) centered at the city of Blythe, California, and small-scale displacements near the mountains southeast of Cibola that result in a differential movement of 300–600 m.

Foraminiferal assemblages in the Bouse Formation consist of up to 16 benthic and 7 planktic foraminiferal species. The benthic foraminiferal assemblages include species of Ammonia, Bolivina, Cibicides, Elphidium, Eponidella, Protoelphidium, Quinqueloculina, and Rosalina (Smith, 1969, 1970; Winterer, 1975; McDougall, 2008a, 2011). Agglutinated species are less common, but include Haplophragmoides and Trochammina. A rare occurrence of Uvigerina in Mf11676 is also noted. The most diverse assemblages are found in the lower part of the LCRP wells and Hart Mine Wash section, whereas assemblages from spot samples at higher elevations contain a single species, Ammonia beccarii.

Benthic foraminiferal diversity in the Bouse Formation of the Blythe basin increases from north to south and decreases with increased elevation (i.e., the younger portion of the Bouse Formation) (Fig. 3). Diversity is defined here as the number of species per sample, but since no standardization of sample size or picking techniques were used, this should be considered the species richness. Geographically, there are three noticeable breaks in the diversity: (1) in the Chemehuevi, Mohave, and Cottonwood basins, north of Parker Dam, where foraminifers have not been found; (2) in the northern part of the Blythe basin, from Parker Dam to the Big Maria Mountains, where diversity is five or fewer species; and (3) the southern part of the Blythe basin, where as many nine species per sample are found (Fig. 3). With few exceptions, diversities are limited to one species per sample at any elevation above sea level. Exceptions to the north-south and elevation patterns are: (1) five benthic foraminiferal species noted in the northern part of the Blythe basin at 63–65 m asl in Mf11677 and Mf11684, (2) two species at 110 m asl in the Big Maria Mountains area (Big Maria Quarry and sample Mf11680), and (3) nine to five species in the southern part of the Blythe basin at 110–123 m asl (Hart Mine Wash, Mf11682, and Mf11683).

Higher benthic diversities are associated with the presence of planktic foraminifers in the Hart Mine Wash section (110–113 m, possibly in the sample at 123 m asl) and Mf11680 (∼116 m asl). Planktic foraminiferal diversity is low (≤2 species) in most samples. A maximum diversity of four species occurs in Mf11676 (sample at –80 m bsl) and Mf11682 (110 m asl). In the lower part of the Bouse Formation, foraminiferal diversity patterns and species are more consistent with marine environments similar to estuaries. An estuary is defined as a coastal body of water with free communication to the ocean and within which ocean water is diluted by freshwater derived from land (Valle-Levinson, 2010). Above 110–116 m asl in the Bouse Formation of the Blythe basin, foraminiferal diversity patterns and species are more consistent with lacustrine environments similar to a saline lake. A saline lake is defined as a large body of water without connection to the sea in geologically recent time and with high salt concentrations (Hammer, 1986).

Benthic foraminiferal composition indicates that deposition initially occurred at shallow inner neritic depths (≤50 m), based on the presence of Ammonia beccarii, Bolivina subexcavata, Elphidium poeyanum, Eponidella palmerae, Neoconorbina terquemi, Protoelphidium sp., Rosalina columbiensis, and various agglutinated forms. This fauna, together with rare planktic foraminifers, occurs at the lowest elevations of the southern Blythe basin and last appears at 123 m asl (Fig. 4). Above 126 m the foraminiferal fauna is restricted entirely to a single species, A. beccarii.

These shallow inner neritic faunas occur in the Hart Mine Wash section at elevations of 110–126 m asl. Differential movement between the mountains east of Cibola and the center of the Palo Verde Valley (under the city of Blythe) is between 300 and 600 m, thus the Hart Mine Wash sediments probably correlate with sediments at a lower elevation and the original elevation at the time of deposition is unknown.

The foraminiferal faunas found at the lowest elevations and assumed to be the oldest are similar to faunas observed in the modern Gulf of California in the lagoons, beach, and inner neritic faunas (Bandy, 1961; Walton, 1955). In the San Francisco Bay, similar living assemblages are found in the deep western bay estuarine cluster of McGann (inChin et al., 2010) at water depths ranging from 14 to 49 m. Arnal et al. (1980) described a fauna including Ammonia beccarii, Elphidium, rare Bolivina and Buliminella, in which agglutinated species are rare or absent, in a deep channel of the southern San Francisco Bay, where water depths range from 12 to 22 m and ocean water is present year round. Both areas and assemblages contain rare planktic foraminifers. In general this group of species is commonly found in shallow (water depth 0–12 m) to intermediate (water depth 12–30 m) assemblages along the west coast of North America (Lankford and Phleger, 1973). In the Puget Sound area, the foraminiferal assemblage includes more agglutinated species than currently found in the Bouse Formation (Cushman and Todd, 1947; Cockbain, 1963; R.A. Harman, 2001, personal commun.).

Ammonia beccarii is abundant and dominates the upper part of the Bouse Formation in the Blythe basin (Fig. 5). In samples at elevations below 100 m asl, A. beccarii rarely exceeds 100 specimens or few to rare specimens in samples where only relative abundance data exist. Samples where relative abundances indicate an increase in the number of A. beccarii specimens are clustered in two intervals: –129 to –164 m bsl in Mf11678 and at –121 to –152 m bsl in Mf11676; and at 49 and 64 m asl in Mf11684 and 55–56 to 63 m asl in Mf11677. Maximum abundances of 137–179 specimens per sample were found in Mf11678 samples at –143 to –146 m bsl. At elevations above 100 m in the Blythe basin (especially in the northern part), the abundance of A. beccarii ranges from zero to much greater than 100 specimens per sample (maximum abundance >2000 specimens per sample). The higher abundances of A. beccarii are probably due to environmental changes that eliminated other benthic foraminiferal species, because A. beccarii is an extremely poor competitor and thrives best where it lacks competition (Arnal, 1961). The presence of juvenile specimens of A. beccarii throughout the Bouse Formation indicates that temperature and salinity conditions remained within the limits tolerated by this species. Laboratory experiments and field observations indicate that normal growth and reproduction in A. beccarii occurs between temperatures of 17 and 32 °C, and salinities of 15‰–40‰ (Bradshaw, 1957, 1961; Schnitker, 1974; Walton and Sloan, 1990).

In the Blythe basin, some A. beccarii specimens were deformed with abnormalities that include double tests, protuberances on the spiral side, or an unusual chamber arrangement. The number of deformed A. beccarii specimens in the Bouse Formation increases from an average of <1% of the population at elevations below sea level to 2.9% of the population at elevations between sea level and 100 m asl, and decreases to 1.72% of the population above 100 m asl (Fig. 6). The higher number of deformities in the interval between sea level and 100 m asl may reflect the unstable salinities in this interval. Test abnormalities in A. beccarii are common when there are variations in environmental parameters such as temperature or salinity (Arnal, 1955; Resig, 1974; Cann and de Deckker, 1981; Almogi-Labin et al., 1992; Stouff et al., 1999; Debenay et al., 2001; Wennrich et al., 2007). Under normal marine salinity, A. beccarii populations have a low number of abnormalities (∼1%), and abnormalities of 5% or more are reported from tide pools and coastal lagoons with intermittent marine connections (Arnal, 1955), whereas under hyposaline or hypersaline conditions abnormalities can affect 50% or more of the population (Stouff et al., 1999). Increased percentages of abnormal tests are noted in saline lakes worldwide (e.g., Resig 1974; Cann and de Deckker, 1981; Almogi-Labin et al., 1992; Le Cadre et al., 2003), and the identification of this pattern in the upper part of the Bouse Formation could suggest the presence of a similar environmental setting. However, additional study is needed before test deformity can be used to determine salinity changes in the Bouse Formation, because many samples are missing.

Planktic foraminifers in the Bouse Formation are primarily restricted to southern part of the Blythe basin (Tables 2–11; Fig. 3). The specimens are usually rare; some are well preserved, while others are fragmented but still preserve a distinctive delicate wall texture. Species belonging to the genera Globorotalia, Neogloboquadrina, Streptochilus, and Tenuitellita are recognized in the foraminiferal assemblages. Specimens identified as Globigerina sp. in Big Maria Quarry, in several of the LCRP wells, and in the Bristol basin (Danby and Cadiz wells) by Smith (1970) are in fact juvenile A. beccarii. Only one planktic foraminifer was identified in the northern part of the basin (Mf11677), but the isolated occurrence and poor preservation makes this identification questionable (Plate 1). Scanning electron microscope photographs revealed that some of the specimens first identified as the benthic foraminifer Bolivina subexcavata by Smith (1970) are the planktic foraminifer Streptochilus, based on the aperture and wall texture (Plate 1).

Streptochilus has been ignored in planktic assemblages and biostratigraphic analysis because of its misidentification as a benthic foraminifer, its small size (63–125 µm), and previously unproven stratigraphic value (Kennett and Srinivasan, 1983; Resig, 1989). A complete and well-preserved fossil record in the western Pacific and Indian Ocean has allowed recognition of the biostratigraphic relevance of this genus (Resig, 1989, 1993).

In the Blythe basin, Streptochilus cf. S. subglobigerum specimens have a broken last chamber. Although wall texture in the final chamber is an important characteristic with which to distinguish this species from the younger species S. globigerus, other characteristics that allow the taxonomic identification are the good preservation of finely cancellate wall texture (images 24 and 25 in Plate 1), straighter sutures, and less globular chambers than S. globigerus specimens (Plate 1). These characters were described by Resig (1989) for juvenile stages of S. subglobigerum and later forms transitional to S. latus species are also present in the Blythe basin.

Laboratory experiments indicate that some species of planktic foraminifers can tolerate greater variations in salinity than occurs in modern oceans (Hemleben et al., 1989; Bijma et al., 1990; Lee et al., 2010); e.g., Neogloboquadrina dutertrei can tolerate salinities ranging from 24‰ to 46‰ in the laboratory (Bijma et al., 1990). Regardless of this broader tolerance range, planktic foraminifers are absent from saline lakes that have no connection with marine waters, are extremely rare in saline lakes with marine connections, and are rare in estuarine environments. For example, living globigerinids are present in the Straits of Juan de Fuca, near the U.S.-Canadian border along the Pacific coast, but are rare in the Rosario Straits and are not present in the Straits of Georgia or in the southern part of Puget Sound (Cushman and Todd, 1947; Cockbain, 1963; R.A. Harman, 2001, personal commun.). In San Francisco Bay, northern California, Globigerina bulloides occurs rarely in the entrance to the bay and in the central bay, just west of the Golden Gate Bridge (Arnal et al., 1980; Chin et al., 2010). In the Gulf of California, planktic species are common in outer shelf water depths and deeper, but are missing from large areas of the Gulf of California, including the Colorado River delta, due to rainfall and run off, which cause salinity variations (Walton, 1955; Bandy, 1961). Planktic species most tolerant of hyposaline conditions (salinities range of 30.5‰–31‰) are Globigerina bulloides, G. quinqueloba, Globigerinita uvula, Globigerinoides ruber, Neogloboquadrina pachyderma (f. superficiaria), and Orbulina universa (Be and Tolderlund, 1971). These hyposaline tolerant species are absent in the Bouse Formation in the Blythe basin, whereas species common in normal marine salinities like Streptochilus are present (Darling et al., 2009; De Klasz et al., 1989). Living Streptochilus are present principally in surface marine waters (Darling et al., 2009; De Klasz et al., 1989; Smart and Thomas, 2006), although oxygen isotope values suggest that Streptochilus can live deeper in the upper water column in temperate to tropical, highly productive waters or in shallow surface waters close to upwelling zones in coastal regions (Resig and Kroopnick, 1983; Hemleben et al., 1989; Smart and Thomas, 2006).

The change to monospecific foraminiferal assemblages (∼110 m asl), increased abundance of Ammonia beccarii (∼110 m asl), and absence of planktic foraminifers above 123 m asl in the Bouse Formation coincide with the lowest occurrences of Chara, a green algae. Chara were observed in foraminiferal residues in a sample at the top of the Hart Mine Wash section (∼126 m asl), Milpitas Wash (85 m asl), Big Maria Quarry, (110 m asl), and Bouse Wash (116–152 m asl) (Fig. 7). This algae is common in shallow freshwater, and some species can tolerate brackish-water conditions with salinities approaching 20‰ (Olsen, 1944; Wood, 1952; Proctor, 1961; Burne et al., 1980). Its presence supports a saline lake interpretation for the upper part of the Bouse Formation.

The presence of reworked Cretaceous microfossils derived from the Mancos Shale in the Bouse Formation indicates the presence of Colorado River water. According to Winterer (1975), D.G. Metzger (1975, personal commun. inWinterer, 1975) reported Cretaceous coccoliths in Bouse Formation outcrops from Needles to Chemehuevi Valley. Paul Fritts (1975, personal commun., inWinterer, 1975; and Fritz, 1975, unpublished report provided by K. Howard, 2012, personal commun.) reported reworked Cretaceous coccoliths in the Bouse Formation from the subsurface near Blythe, Arizona. No depths or locations were given. Smith (1970; and inLucchitta, 1972) reported that there are no reworked microfossils in the subsurface samples she examined of the Bouse Formation. Winterer (1975) also reported coccoliths from three samples: one surface sample in the Needles area, and two subsurface samples from the Vidal wells (B-30, J40 and B-30, J64) correspond to elevations of 188 and 157 m asl. Winterer (1975) reported that there are no coccoliths in the basal marl of the Bouse Formation near Cibola (locality J217, samples C2–12, elevation 140.2 m asl). Lucchitta (1972) implied that although no Mancos-type foraminifers have been found in the Bouse Formation, coccoliths were found in surface exposures of the unit; however, no locations were given. In a memo from P. Smith to D. Bukry dated 20 May 1968 (USGS Micropaleontology collection, Flagstaff, Arizona), Smith indicated that Cretaceous coccoliths were found in the Big Maria Quarry samples (Mf441 to Mf456) at an elevation of 110 m asl.

The presence of reworked Cretaceous coccoliths at elevations of 110 m asl and higher, and primarily in the northern part of the Blythe basin, indicates the presence of sediment derived from the Mancos Shale, most likely by the Colorado River. This distribution resembles the distribution of Chara (Fig. 7), and the occurrences are in the interval interpreted to be deposited in a saline lake by the foraminiferal assemblages. The absence of reworked microfossils at lower elevations suggests that the Colorado River was not eroding the Mancos Shale or it was not a major source of sediment in the Blythe basin.

SALINE LAKE VERSUS MARINE ORIGIN OF THE BOUSE FORMATION

Although the Bouse Formation foraminiferal associations suggest an initial marine environment, it has been argued that the foraminifers were introduced by avian transport to a saline lake (Spencer and Patchett, 1997; Spencer et al., 2005, 2008, 2013). Saline lakes occur in a wide variety of settings and many contain foraminiferal faunas (Fig. 8). A survey of saline lakes worldwide shows that if foraminifers are present, they were introduced either as an in situ component (i.e., the lake was originally part of the sea or is periodically flooded by the sea) or through avian transport. Avian transport of foraminifers can be a successful dispersal mechanism to colonize new habitats (Patterson et al., 1997); a small population of foraminifers is transported by birds and even a smaller number survive the new environmental conditions to colonize the new habitat. Although this small population may give way to a much larger population, the probability is low that the initial foraminifers will survive and become part of the fossil record. Criteria to differentiate saline lakes with transported foraminifers from a marine environment include diversity patterns, faunal composition and distribution patterns of benthic foraminifers, abnormal tests, endemism, and the presence or absence of planktic foraminifers. These criteria are based on comparisons of saline lakes and several marine estuaries (Straits of Juan de Fuca, Straits of Georgia, Puget Sound, San Francisco Bay, and the Gulf of California) as listed in Table 12.

Foraminiferal faunal diversities in saline lakes are generally low and contain many of the same species as the local marine environment, whereas estuaries have higher diversities that increase away from the shoreline. The number of foraminiferal species in shallow saline lakes is consistent across the lake, whereas in larger lakes with greater depths, diversity decreases toward the center due to the increase in organic carbon content and the decrease in pH of the sediment (Arnal, 1961; Chin et al., 2010). With few exceptions the foraminiferal fauna in saline lakes that have had no connection with marine waters usually contain five or fewer species (Bachhuber and McClellan, 1977; Cann and de Deckker, 1981; Anadon, 1989; Kázmér, 1990; Patterson et al., 1997; Giralt et al., 1999; Boudreau et al., 2001; Wennrich et al., 2007). Exceptions include the Salton Sea with 21 species (Arnal, 1954, 1958, 1961) and the Salt Lake in Hawaii with 41 species (Resig, 1974). Lakes that have a marine connection at some point in their existence have higher diversities, including some species not found in the isolated saline lakes. Estuaries such as the Gulf of California, San Francisco Bay, and Puget Sound have much higher diversities and diversity increases with depth. Diversities of 20 or more species occur in the central Gulf of California (Bandy, 1961). In San Francisco Bay, diversities increase from ∼9 species/sample in the nearshore areas to 14 or more species/sample in the deeper central areas (Arnal et al., 1980; Chin et al., 2010). A similar pattern occurs in Puget Sound, where the highest diversities are found in the central area of the straits and away from the shoreline (Cockbain, 1963).

Distribution and composition patterns of foraminifers from estuaries exhibit biofacies patterns that parallel the shoreline and change with depth or distance from the shoreline. In contrast, patterns in saline lakes tend to be more uniform, do not change with depth or distance from the shoreline, and are affected primarily by the influx of river water (Arnal, 1961; Chin et al., 2010). Agglutinated faunas are common in marine estuaries and embayments (Bandy, 1961; Cockbain, 1963; Ellison and Nichols, 1970; Arnal et al., 1980; R.A. Harman, 2001, personal commun.).

The presence of planktic foraminiferal species in the foraminiferal fauna is a good indicator for discriminating marine shelf, estuarine, and saline lake deposits. Planktics are absent in saline lakes, rare in estuarine environments, and common in the outer shelf biofacies of marine environments. They are sensitive to salinity changes and are therefore missing from large areas of the Gulf of California because of river input and runoff. A similar situation occurs in the Straits of Georgia, British Columbia, Canada, where planktics and many benthic species are not present in the marine assemblages north of the Fraser River delta. In the San Francisco Bay, planktic foraminifers occur only rarely east of the Golden Gate Bridge. In both cases the absence is due to the influx of freshwater.

Abnormal, deformed, or dwarfed tests can also be used to differentiate saline lakes from marine estuaries. Abnormal or deformed Ammonia tests occur ∼1% of the time in normal saline conditions but there is a high rate of abnormalities under both hypersaline and hyposaline conditions (Stouff et al., 1999; Debenay et al., 2001; Wennrich et al., 2007). Dwarfism, which is present in marine environments, is usually the result of low oxygen conditions and is frequently accompanied by other indicators of low oxygen such as low oxygen species and enlarged pores on the tests (Douglas, 1979). Dwarfed benthic foraminiferal faunas can be found living below sill depth in marine basins in the Gulf of California (Bandy, 1961).

Endemism and isolated occurrences can be a diagnostic criterion of a saline lake. If the benthic foraminiferal species adapt to the environmental conditions of the saline lake after introduction and if the populations evolve isolated, with time an endemic fauna could develop. However, if environmental conditions do not promote survival after introduction, the species will be represented by a single or isolated occurrence.

Applying the criteria to differentiate marine from saline lake environments to the Bouse Formation suggests that in Blythe basin both environments were present. Fossil diversity, composition, abundances, and deformities indicate that the Bouse Formation was initially deposited in a marine environment that changed to saline lake environment before deposition ended.

Benthic foraminiferal diversity in the Bouse Formation samples is low. The highest diversities are common in the center of the southern Blythe basin (Mf1168 and Mf11676) and in Hart Mine Wash, whereas the lowest diversities are to the north and along the sides of the basin near Parker and in the Bristol basin. The highest diversities are found in the oldest sediments and diversities decrease toward the top of the outcrop sections and wells (Figs. 3 and 4). This configuration suggests that initially the Bouse Formation was deposited in a marine estuary rather than a saline lake, because the diversities increase toward the center of the basin, away from the edges and away from the influx of river water. The diversity decline with elevation indicates that the topographically higher (and probably younger) part of the Bouse Formation was deposited in a different sedimentary environment, possibly a saline lake.

Microfossil sampling of the Bouse Formation has been random, so there are insufficient samples to determine detailed distribution patterns of benthic foraminifers. Nevertheless, faunas in the lower elevations (and probably older) part of the Bouse Formation (Fig. 4) are most similar to the inner shelf biofacies of the Gulf of California. In contrast, exposures at higher elevations are dominated by Ammonia beccarii (Fig. 4), and are most similar to the lagoonal biofacies of the Gulf of California (Bandy, 1961) or the saline lake biofacies described from the Salton Sea (Arnal, 1961). The distribution of planktic foraminiferal species is restricted to the lower part of the formation in the southern Blythe basin (Fig. 3). The highest occurrence of planktic foraminifers, at 123 m asl, approximates the highest occurrence of benthic foraminiferal diversities of 2 or more species (116 and 110 m asl) and changes in the abundance and percent deformed A. beccarii as well as the lowest occurrences of Chara (Fig. 7). These changes suggest a conversion from marine to saline lake conditions.

The predicted salinity model for Lake Blythe (Spencer et al., 2008, 2013) indicates that as the lake level reached an elevation of –125 m bsl, salinity would range from 0‰ to 100‰ and while filling from an elevation of –125 bsl to 64 m asl, salinity would be at 27.6‰. Seawater salinities (35.0‰) would not be achieved until shortly before lake level reached 329 m asl. A final salinity of 35.4‰ just before spillover was suggested by Spencer et al. (2008). Based on this modeling, marine conditions would not be achieved until the lake was nearly filled, whereas patterns observed in the benthic foraminifers indicate that marine salinities were present during the initial deposition of the Bouse Formation in the Blythe basin. Saline lake conditions with salinities lower than normal seawater are evident only in the faunal assemblages found at the higher elevations of Bouse Formation.

ISOTOPES AND MARINE OR LACUSTRINE DEPOSITION

An additional argument used to support the lacustrine origin of the Bouse Formation is based on strontium, carbon, and oxygen isotopic ratios. Isotope studies (Spencer and Patchett, 1997; Spencer and Pearthree, 2001; Poulson and John, 2003; Roskowski et al., 2010; Spencer et al., 2013) concluded that the Bouse Formation was deposited in a lake composed of Colorado River water. However, because of the sampling distribution, the isotope data do not contradict a marine origin for at least part of the Bouse Formation. Foraminiferal data suggest that marine conditions persisted to an elevation of 123 m asl. Therefore, the overlap of the marine and freshwater conditions suggested by the strontium isotopes is limited to an interval between 70 and 123 m asl. Although isotope samples were taken from the same sections, they are not the same samples, and there has not been any coordination between the studies.

Strontium values from the Bouse Formation range from 0.7105 to 0.7112 (Buising, 1990; Spencer and Patchett, 1997; Spencer and Pearthree, 2001; Roskowski et al., 2010; Spencer et al., 2013) and are located at elevations ≥70 m asl with one exception (Fig. 9); Spencer and Patchett (1997) quoted a value of 0.71075 for the Colorado River based on an analysis by Goldstein and Jacobsen (1988). This value was modified by the analysis of Gross et al. (2001) that showed that the modern Colorado River value at Lake Mead (Hemenway Harbor, Nevada) was 0.710437 and 0.710274 at the Cibola Wildlife Refuge, Arizona. Compared to these modern strontium isotopic values for the Colorado River, the late Miocene to Pliocene values are much higher and a better understanding of the source and concentration of strontium is needed.

A single Sr sample analyzed by Spencer et al. (2013) from a snail in Mf11684 is from an elevation of –62 m bsl. This specimen and others found at –87 m bsl may be Pyrgulopsis avernalis (see Spencer and Patchett, 1997), which is known to inhabit freshwater springs (Hershler, 1994). Its presence in Mf11684 suggests a freshwater spring or transport to this location. In addition, the presence of several freshwater snails (Hydrobiidae?, Fontelicella, Physa, and Pyrgulopsis; Smith, 1970; Winterer, 1975; Spencer and Patchett, 1997; Turak, 2000; Spencer et al., 2008, 2013) was noted near the base of the Bouse Formation in Mf11684 (–64, –88, –105, and –119 m bsl) and Mf11677 (–80 and –83 m bsl) at the northern end of the Blythe basin. These freshwater snails occur below the first occurrence of foraminifers, suggesting that initial deposition in this part of the basin may have occurred in a spring or lake prior to the marine incursion or the development of saline conditions. Additional study is needed to resolve whether the snails in Mf11684 and therefore the strontium analysis represent conditions at the site or conditions from another location where the snail actually lived. If the strontium datum from the Pyrgulopsis snail is excluded, the remaining strontium samples were taken at elevations above 70 m asl.

Strontium isotope ratios from the marls and tufas in the interval between 70 and 120 m asl elevation in the southern part of the Blythe basin systematically decrease with elevation and then increase above 100 m elevation (Fig. 10). Roskowski et al. (2010) attributed this change to changing 87Sr/86Sr of water inputs into the river system and the fact that the lake was only 25% of capacity so these changes would be clear. Because diverse marine benthic and planktic foraminiferal assemblages occur in the southern Blythe basin at elevations between 110 and 123 m asl, the decrease in the strontium values may alternatively record a mixing of seawater and river water. The faunal changes above this elevation suggest a saline lake, as do the strontium values. Overall strontium isotopic values in the Bouse Formation are higher than modern river values but lower than Hualapai Limestone Member values (0.713654–0.719536), and suggest that there may be additional sources of strontium that were not considered (Crossey et al., 2013).

Like strontium, the oxygen and carbon isotopic analyses are primarily from the upper part of the Bouse Formation (Fig. 8). Samples analyzed for δ18O and δ13C from the Cibola area and elevations ranging from 70 to 100 m asl plot near seawater, whereas samples from Parker, Milpitas Wash, Palo Verde Hills, and Limekiln Wash are from elevations higher than 100 m and are further removed from the seawater values (Poulson and John, 2003; Roskowski et al., 2010) (Fig. 11). Although Poulson and John (2003) found anomalies in their data that indicated an estuarine origin, they concluded that overall the data were more consistent with a lacustrine origin. Roskowski et al. (2010) argued that the high negative δ18O values indicate a continental origin for the waters and evaporation of lake waters. The values that plot between –5‰ and 0‰ are believed to represent an interval when lake evaporation was high. Samples in the Cibola area at elevations of 100 m or lower contain negative δ18O (–2.76‰ to –0.22‰) and δ13C (–0.84‰ to –0.71‰) values approaching zero. Although interpreted as representing an interval of high evaporation of freshwater by Roskowski et al. (2010), these values and a trend toward zero (seawater) value could also be interpreted as including a seawater component.

Other isotope samples from elevations lower than 100 m include a sample from Milpitas Wash (95 m) and one from the Palo Verde Hills (87 m), from which oxygen and carbon isotopes are strongly negative and approach Colorado River values. Additional isotopes analysis from Milpitas Wash, Palo Verde Hills, and Parker are at elevations higher than 100 m and have negative δ18O values and low δ13C values. These values may indicate a lacustrine environment. Isotopic analyses from the bedding plane associated with the fish Colpichthys regis have high negative values for δ18O and low δ13C values (Roskowski et al., 2010). Although these samples have elevations (89–99 m asl) similar to the other Cibola area isotope samples, they are northwest of the other Cibola samples, and the isotope values seem to reflect evaporation and/or a continental signal, as suggested by Roskowski et al. (2010).

UPLIFT, FLOOD DEPOSITS, AND COLORADO RIVER SANDS

Other arguments in favor of a lacustrine origin for the Bouse Formation include the presence of flood deposits attributed to lake overflow or breakout and the lack of evidence for uplift. Flood deposits are present in northern Mohave Valley stratigraphically beneath the Bouse Formation (House et al., 2005a, 2005b, 2008). These deposits were interpreted to record lake spilling from the north. However, such deposits have not been found in the Blythe basin, suggesting that the arrival of the river was gradual and not catastrophic or that the deposits have not yet been found. The presence or absence of these deposits is not an indicator of marine or lacustrine deposition.

The portion of the Bouse Formation that contains evidence of marine deposition is today at elevations of 123 m asl and lower. Planktic foraminifers in the Hart Mine Wash section were found at an elevation of 112–123 m asl and correlative planktic species in Mf11676 are at elevations of –106 m bsl. The elevation difference between these assemblages (218–229 m) could be attributed to faulting and downwarping that resulted in differential movement of 300–600 m (Metzger et al., 1973). Planktic foraminiferal assemblages found at Mf11680 at an elevation of 115 m asl may also be correlative and the elevation may also be affected by faulting and downwarping. At the time of deposition, samples containing planktic foraminifers were below sea level, which was at least 25 m above present sea level in the late Miocene and Pliocene (Kominz, 1984; Haq et al., 1987a, 1987b; Johnson et al., 2005; Muller et al., 2008; Miller et al., 2012).

A small amount of regional or local uplift would have been necessary to accommodate a marine incursion and elevate the Bouse Formation to its current level. A postdepositional uplift of 500–1000 m was proposed (Metzger, 1968; Smith, 1970; Lucchitta, 1972, 1979; Buising, 1990). While this amount of uplift is not consistent with studies that conclude that uplift of the Colorado Plateau occurred in the Eocene, it is consistent with other models that suggest that uplift is ongoing (Karlstrom et al., 2008, 2012; Levander et al., 2011).

AGE AND CORRELATION

The presence of Streptochilus latus in sample Mf11682 and the Hart Mine Wash section indicates a late Miocene age between 8.1 and 5.3 Ma (Resig, 1989, 1993). This age is confirmed by the occurrence of Streptochilus cf. S. subglobigerum in the lower part of Mf11676 and sample Mf11682 that suggests a late Miocene age between 10.4 and 7.4 Ma corresponding to planktic foraminiferal zones N16 and N17 of Blow (1969; as modified and correlated by Gradstein et al., 2004; Gradstein and Ogg, 2005). If the S. subglobigerum identification is confirmed with more specimens, the co-occurrence with S. latus constrains the age of the Bouse Formation in the southern Blythe basin to late Miocene, between 8.1 and 7.4 Ma in the upper half of the late Miocene planktic foraminiferal zone N17. If the questioned specimens are identified as S. globigerus, then the co-occurrence with S. latus would constrain the age of the Bouse Formation in the southern Blythe basin to late Miocene, between 5.6 and 5.3 Ma. At present, the only reliable age for the Bouse Formation in the Blythe basin is based on the presence on well-preserved S. latus and is late Miocene, between 8.1 and 5.3 Ma.

The late Miocene age based on the planktic foraminifers agrees with the previously reported ages constraining Bouse Formation in the Blythe basin. A tuff (9.2 ± 0.3 Ma) and a basalt (9.6 ± 0.60 Ma) in the underlying Osborne Wash strata of Buising (1988) (Fugro, Inc., 1975, cited inReynolds et al., 1986; Buising, 1988; Buising and Beratan, 1993) suggest that the Bouse Formation was deposited after 9.2 Ma. The age of the Lawlor Tuff from the upper part of the Bouse Formation found at the Buzzards Peak and Amboy localities (Metzger et al., 1973; Buising, 1988; Sarna-Wojcicki et al., 2011; Spencer et al., 2013) indicates that deposition ended by 4.83 Ma. Both normal and reversed magnetic polarity have been noted in the Bouse Formation (Kukla, 1976; Malmon et al., 2011). Preliminary analysis of the Bouse Wash section by Malmon et al. (2011) indicated a paleomagnetic reversal; unfortunately, there are insufficient data to date this reversal.

Similar ages were reported from the Imperial Formation in the Cabazon, Whitewater, Garnet Hill, and Fish Creek sections in the Imperial Valley (Fig. 12). Strata in the Cabazon and Whitewater sections were assigned to planktic foraminiferal zones N17–N18 based on the occurrence of Globigerinoides obliquus, G. extremus, and Globigerina nepenthes and to calcareous nannoplankton zones CN9a–CN10 based on the occurrence of Discoaster brouweri, D. aff. D. surculus, Reticulofenestra pseudoumbilicata, Sphenolithus abies, and S. neoabies (McDougall et al., 1999; McDougall, 2008a). These assignments are supported by K/Ar dates from the underlying Coachella Fanglomerate (10.1 ± 0.2 Ma; Peterson, 1975) and overlying Painted Hill Formation (6.04 ± 0.18 and 5.94 ± 0.18 Ma; Matti et al., 1985; Matti and Morton, 1993). The Imperial Formation exposed in the Garnet Hill section is between 8.0 or 7.6 to 5.32 Ma based on a tephra correlation (Rymer et al., 1994, 1995) and the presence of Amphistegina gibbosa, which indicates deposition prior to the influx of sediment derived from the Colorado River (5.32–4.83 Ma; Spencer et al., 2013; Dorsey et al., 2007, 2011). The older part of the Imperial Formation exposed in the Fish Creek section also contains Amphistegina gibbosa and probably represents the same interval, but does not contain diagnostic microfossils.

Paleomagnetic correlations from the Fish Creek section suggest that this section is between 6.27 and 5.89 Ma, but also assumes that the upper megabreccia was emplaced with no break in sedimentation. The benthic foraminiferal assemblages indicate that water depths increased from shallow inner neritic (<50 m) to upper bathyal (150–500 m) before normal sedimentation returned after the megabreccia was emplaced. The disparity implies a nonconformity and that some time is missing. Therefore it is possible that the Imperial Formation below the megabreccia is at least one chron older than reported by Dorsey et al. (2007, 2011); e.g., either C3An.2n (6.57–6.27 Ma) or C3Bn (7.09–6.94 Ma) subchrons.

Other sections containing marine sediments similar in age in the Gulf of California include the Bouse Formation in the Yuma basin, the Boleo Formation in the Santa Rosalía basin, and the Diatomita Santiago Formation in San José de los Cabos, Mexico (Carreño, 1981, 1992). The Bouse Formation in the Yuma basin occurs primarily in the subsurface and contains many of the same species as the Bouse Formation along the lower Colorado River (Smith, 1970; Olmsted et al., 1973; Winterer, 1975): Ammonia beccarii, Elphidium gunteri, Eoponidella palmerae, and Rosalina columbiensis. Slides and material are not available to verify the presence of planktic foraminifers, and the checklists provided by Winterer (1975) are too vague to make a confident assessment of ages.

The most probable age for the Boleo Formation in the Santa Rosalia basin, Baja California Norte, is late Miocene and ranges from 7.09 to 6.93 Ma (Geomagnetic Polarity Time Scale subchron C3Bn) for the base and 6.27–6.14 Ma for the top of the formation (Holt et al., 2000). Foraminiferal assemblages from the Boleo Formation (A. Miranda, personal data) contain a diverse assemblage including Amphistegina gibbosa and could be correlative with the Imperial Formation found in the Whitewater, Cabazon, and Garnet Hill sections. The assemblages are more diverse than those found in the Bouse Formation, but contain many of the same species.

CONCLUSIONS

The foraminiferal associations indicate that the basal part of the Bouse Formation in the Blythe basin was marine and that marine sediments are found up to elevations of ∼123 m asl (Fig. 13). Above this elevation, the limited fauna (a single benthic foraminiferal species) coupled with the absence of planktic species indicates that the marine water is replaced by saline lake water. Chara, a green algae found in fresh to slightly brackish water, first appears at 85 m asl but becomes more common at elevations of 110 m and higher, and reworked Cretaceous coccoliths from the Mancos Shale appear at ∼110 m asl and higher. The dominance of Ammonia beccarii is low throughout the lower part of the Bouse Formation, but it becomes the dominant (or only) foraminiferal species between 110–126 m asl and higher elevations. This dominance suggests a change from marine to brackish water such as in a saline lake. Deformed tests compose 1% or less of A. beccarii tests in the lower elevations of the Bouse Formation and suggest marine water. At an elevation of 123 m and higher, the number of deformed tests in the A. beccarii population increases and can include as much as 5% of the population. These higher values suggest salinities differ from normal marine, but more data are needed to develop this line of evidence.

Arguments in favor of a lacustrine origin of the Bouse Formation are based on the presence of a saline lake, avian transport of foraminiferal species, isotopes, no evidence of uplift, and a variety of sedimentological arguments such as flood deposits and the presence of Colorado River–derived sediments. Faunal composition (benthic foraminiferal assemblages and the presence or absence of planktic foraminifers), and characteristics (abundance and deformity of A. beccarii) of the Bouse assemblages suggest that deposition was initially in a marine estuary and that a saline lake developed when the sediments now found at elevations between 110–123 m and higher were deposited. Salinity models (Spencer and Patchett, 1997; Spencer et al., 2013) that assume the Blythe basin was a lake filled by Colorado River water indicate that initially salinities were highly variable, ranging from fresh to hyposaline, and did not approximate marine salinities until just before spillover of the lake. The isotopic data are interpreted as indicating lacustrine deposition, and since most of the isotope samples are from the upper portion of the Bouse Formation, this is expected. Strontium isotope samples from elevations of 70–123 m decrease from the high Colorado River values, indicating that some mixing of different water masses occurred. Values increase above this and remain high. Oxygen and carbon isotopes were also sampled primarily from the saline lake portion of the formation and indicate a continental source. Samples from the lowest elevations (∼80–90 m) and from the southern part of the Blythe basin near Cibola have a marine signal. Perhaps with additional study of the Sr sources and concentration, and more isotope analysis, a marine signal will be detected in the isotopes from the lower part of the Bouse Formation. Although no evidence of uplift was observed, when differential movement due to faults and downwarping is accounted for, the marine portion of the Bouse Formation may be found to be below late Miocene sea level. Reworked Cretaceous microfossils, which are often used as evidence of Colorado River sediments, are present in the Bouse Formation at elevations of 110 m and higher.

Planktic foraminifers indicate a late Miocene age between 8.1 and 5.3 Ma for the oldest part of the Bouse Formation in the southern part of the Blythe basin. This age is between ages derived from other sources that limit the age of the Bouse Formation to younger than 9.2 Ma and no older than 4.83 Ma in the Blythe basin. Benthic and planktic foraminifers that correlate with other late Miocene foraminiferal assemblages suggest that the Bouse Formation was deposited at the northern end of the proto–Gulf of California.

We thank Elizabeth Mennow at the U.S. Geological Survey Flagstaff Micropaleontology Laboratory for her work and assistance in preparing and locating samples and Debra Block for assisting with geographic information system work. We also thank Kyle House and Sue Beard for listening and discussing various ideas, and for encouragement, and Yolanda Hornelas Orozco, in Laboratorio de Microscopía Electrónica de Barrido, Instituto de Ciencias del Mar y Limnología, Universidad Nacional Autónoma de México (UNAM), for taking scanning electron microscope photomicrographs. We appreciate the reviews and comments provided by Keith Howard and Charles Powell II of the U.S. Geological Survey, and Andy Cohen of the University of Arizona. Miranda appreciates the financial support provided by PAPIIT (Programa de Apoyo a Proyectos de Investigación e Innovación Tecnológica) UNAM (IN102211 to A.L. Carreño) and CONACYT (Consejo Nacional de Ciencia y Tecnología, scholarship 172854).

1Supplemental File. Taxonomic notes and sample locations. If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES00975.S1 or the full-text article on www.gsapubs.org to view the Supplemental File.