The cause of Cenozoic uplift of the Colorado Plateau is one of the largest remaining problems of Cordilleran tectonics. Difficulty in discriminating between two major classes of uplift mechanisms, one related to lithosphere modification by low-angle subduction and the other related to active mantle processes following termination of subduction, is hampered by lack of evidence for the timing of uplift. The carbonate member of the Pliocene Bouse Formation in the lower Colorado River Valley southwest of the Colorado Plateau has been interpreted as estuarine, in which case its modern elevation of up to 330 m above sea level would be important evidence for late Cenozoic uplift. The carbonate member includes laminated marl and claystone interpreted previously in at least one locality as tidal, which is therefore of marine origin. We analyzed lamination mineralogy, oxygen and carbon isotopes, and thickness variations to discriminate between a tidal versus seasonal origin. Oxygen and carbon isotopic analysis of two laminated carbonate samples shows an alternating pattern of lower δ18O and δ13C associated with micrite and slightly higher δ18O and δ13C associated with siltstone, which is consistent with seasonal variation. Covariation of alternating δ18O and δ13C also indicates that post-depositional chemical alteration did not affect these samples. Furthermore, we did not identify any periodic thickness variations suggestive of tidal influence. We conclude that lamination characteristics indicate seasonal genesis in a lake rather than tidal genesis in an estuary and that the laminated Bouse Formation strata provide no constraints on the timing of Colorado Plateau uplift.

The Colorado Plateau and eastern Mojave Desert of the American Southwest were part of cratonic North America during Paleozoic time as indicated by the thin (∼1 km), largely marine Paleozoic strata that blanketed the area and that are so well exposed in the Grand Canyon (Stone et al., 1983; Sloss, 1988). Uplift of the Colorado Plateau from near sea level to modern elevations of ∼2 km resulted from post-Paleozoic tectonic processes that are incompletely understood and the topic of many investigations. Uplift has been attributed to removal of all or part of the dense, underlying mantle lithosphere during Late Cretaceous and Paleogene low-angle subduction (Bird, 1988; Spencer, 1996) and hydration of mantle lithosphere or lower crust by slab-derived fluids (Humphreys et al., 2003; Jones et al., 2015; Porter et al., 2017; Levandowski et al., 2018). Uplift has also been attributed to post-subduction processes including uplift driven by the dynamic pressure of rising asthenosphere (Moucha et al., 2008, 2009), detachment and sinking of dense mantle lithosphere (van Wijk et al., 2010; Crow et al., 2011; Levander et al., 2011; Karlstrom et al., 2012; Walk et al., 2019), and heating following inflow of underlying asthenosphere (Roy et al., 2009). Uplift related to low-angle subduction would have occurred during or immediately following slab fallback or meltback as indicated by the westward sweep of mid-Cenozoic arc magmatism (e.g., Coney and Reynolds, 1977; Spencer, 1996) and by calculations of changing slab thermal and mechanical integrity during low-angle subduction (Severinghaus and Atwater, 1990; Spencer, 1994). In contrast, uplift related to post-subduction processes would be late Cenozoic and likely ongoing (e.g., Moucha et al., 2009; Karlstrom et al., 2012). As a result of these contrasting inferred uplift ages, determination of the time of Colorado Plateau uplift is critical to understanding which broad class of uplift mechanisms is primarily responsible for Plateau uplift.

The Cenozoic paleoelevation history of the Colorado Plateau has been difficult to determine despite the application of a variety of techniques over many decades (e.g., Sahagian et al., 2002; Huntington et al., 2010). The lower Pliocene Bouse Formation in the lower Colorado River Valley (Fig. 1) consists of limestone and fine-grained clastic strata that have been interpreted as estuarine and therefore were deposited below sea level (Lucchitta, 1979; Buising, 1990; Lucchitta et al., 2001; McDougall and Miranda-Martínez, 2014; O'Connell et al., 2017, 2020; Dorsey et al., 2018; Gardner and Dorsey, 2020). These strata are now exposed at elevations of up to 330 m in southern exposures and 555 m in northern exposures, which has led to the interpretation that uplift of the Colorado Plateau occurred in the past 5–6 Ma and was greater closer to the plateau (Lucchitta, 1979; Lucchitta et al., 2001). In this article, we present new data from laminated Bouse marl and claystone in the basal carbonate member of the southern Bouse Formation that support lacustrine rather than estuarine deposition. This, in turn, detracts from the interpretation that the Bouse Formation is relevant to the timing of Colorado Plateau uplift.

Bouse Formation

The lower Pliocene Bouse Formation in the lower Colorado River Valley of southeastern California, western Arizona, and southernmost Nevada (USA), consists typically of 1–20 m of bedrock-coating travertine, bedded marl, and bioclastic limestone overlain conformably by claystone, siltstone, and fine silty sandstone. These strata are in turn overlain unconformably by Colorado River sand and gravel and alluvial fan sediments (Metzger et al., 1973; Metzger and Loeltz, 1973; Buising, 1990; House et al., 2008; Pearthree and House, 2014; Homan, 2014; Gootee et al., 2016; Dorsey et al., 2018; O'Connell et al., 2017, 2020; Gardner and Dorsey, 2020). The Bouse Formation onlaps onto alluvial fan sediments and bedrock hillslopes and represents abrupt inundation of a previously subaerial environment. Blythe basin is the southernmost and largest of the basins in which the Bouse Formation was deposited (Fig. 1). The Colorado River flows through the axis of the basin where Bouse Formation has been recognized in wells of up to 280 m below sea level (Metzger et al., 1973; Cassidy et al., 2018). The axial valley may be a product of transtension in a broad, right-lateral shear zone associated with development of the San Andreas transform plate boundary during the past 10–12 Ma (Richard, 1993; Thacker et al., 2020).

Inundation leading to deposition of the Bouse Formation has been attributed to regional subsidence resulting in marine incursion during early opening of the nearby Gulf of California (Lucchitta, 1979; Buising, 1990; Lucchitta et al., 2001; McDougall and Miranda-Martínez, 2014; Dorsey et al., 2018) or to filling of closed basins by first-arriving Colorado River water (Spencer and Patchett, 1997; House et al., 2008; Spencer et al., 2008, 2013; Pearthree and House, 2014; Crow et al., 2021). A lacustrine origin is supported by Sr, O, and C isotopic data from basal carbonates (Spencer and Patchett, 1997; Poulson and John, 2003; Roskowski et al., 2010; Bright et al., 2018a), consistent maximum elevations of Bouse deposits within proposed paleolake basins (Spencer et al., 2013; Pearthree and House, 2014), palaeoecological analysis of fossil communities in Blythe basin (Bright et al., 2018b), and sedimentological evidence of floodwater influx from northern sources immediately preceding Bouse deposition in northern Mohave Valley (House et al., 2008; Pearthree and House, 2014). A marine origin is supported by the presence of several typically marine organisms including foraminifera, barnacles, and fish that are represented by shells and fossils from low elevations in the axis of Blythe basin (Smith, 1970; Todd, 1976; Crabtree, 1989; McDougall, 2008; McDougall and Miranda-Martínez, 2014; Dorsey et al., 2018), although a variety of factors allow for the possibility that all fauna lived in a brackish lacustrine environment (Bright et al., 2018b). In addition, some sedimentological features have been interpreted to indicate a tidally influenced estuarine origin (Buising, 1990; O'Connell et al., 2017, 2020; Dorsey et al., 2018; Gardner and Dorsey, 2020).

In the lacustrine interpretation, deposition of the Bouse Formation resulted from the first arrival of Colorado River water and sediment to a string of closed basins in the Basin and Range province. It marks the top-down initiation of a new river and incision of the modern Grand Canyon (e.g., Spencer and Patchett, 1997; Spencer and Pearthree, 2001; House et al., 2008). In the estuarine interpretation, the Bouse Formation records a phase of subsidence associated with early rifting in the Gulf of California that is not obviously or necessarily related to Colorado River arrival (e.g., Buising, 1990; Dorsey et al., 2018; O'Connell et al., 2020; Gardner and Dorsey, 2020). Furthermore, the presence of Bouse Formation of marine or estuarine origin at significant modern elevation would indicate the approximate amount of tectonic uplift since deposition (Lucchitta, 1979; Lucchitta et al., 2001), ignoring the influence of eustatic sea-level fluctuations since the early Pliocene (Winnick and Caves, 2015; Gasson et al., 2016; Raymo et al., 2018). Proposed uplift has been tied to the timing of uplift of the Colorado Plateau and is a major reason for broad interest in the Bouse Formation.

The study by O'Connell et al. (2017) of the southernmost area of Bouse Formation exposures (Figs. 12) identified several types of sedimentological features that were interpreted as representing the activity of tides and led to the conclusion that some carbonates and related sediments in the basal carbonate unit of the Bouse Formation (Fig. 3) were deposited in an estuary or tidal flat. Layered sequences of laminations or thin beds that are plausibly the result of daily (diurnal) or twice daily (semidiurnal) tidal cycles were analyzed using Fourier spectral analysis to identify periodicities associated with tides. One laminated carbonate sequence from the basal carbonate unit in the Cibola area (Figs. 23), with 82 lamination couplets (paired light and dark layers), yielded two peaks with one that is twice the frequency of the other (fig. 3B in O'Connell et al., 2017). These were interpreted as representing superimposed diurnal and semidiurnal tidal cycles. Least squares fitting of sine waves over a range of wavelengths identified longer wavelength periodicities at 4, 5, 6, 8, 11, and 13 couplets per cycle that were interpreted as incomplete records of 14-day, spring-neap tidal cyclicity in tidal flat environments at three different sample sites in the Cibola area (O'Connell et al., 2017).

Our study was designed to test the tidal hypothesis for lamination genesis proposed by O'Connell et al. (2017) by further examining southern Bouse laminated sediments, including identification of tidal signatures from lamination thickness data. Sedimentation rates of hundreds of millimeters per year, equivalent to tens of meters per century, are implied by the tidal interpretation of laminated marl in the Cibola area. This interpretation is difficult to reconcile with a previous study (Homan, 2014) that concluded that laminations in nearby Milpitas Wash were deposited annually as varves, which implies a sedimentation rate hundreds of times slower than the rate for tidal deposition. Considering that the laminations in the two areas occur in the same basal carbonate unit of the Bouse Formation and are less than 10 km apart, it seemed likely to us that they were both produced by similar processes and that additional study was warranted.

We visited multiple exposures of laminated Bouse strata in the lower Milpitas Wash area in southeastern California and the Cibola area in western Arizona (Figs. 1C and 2) and collected samples from two areas where laminations are numerous, in continuous sequence of tens to hundreds of laminations, and have sharp boundaries so that thicknesses could be accurately measured. Laminated marl was sampled in lower Milpitas Wash (site “C1” in Fig. 1C), where lamination boundaries are generally sharp (Fig. 4C). Laminations at this site, located at the top of the basal carbonate unit and just below the siliciclastic unit (“sample 6B” in Figs. 3 and 4C), were interpreted as varves (Homan, 2014) deposited annually in a “quiet offshore subtidal environment” (Dorsey et al., 2018). Samples were also collected from upper Cibola Wash (an informal name, also known as “Marl Wash”; Figs. 3 and 4E–4F) at the same outcrop where evidence of tidal activity was proposed by O'Connell et al. (2017). Polished thin sections of laminated samples from both areas were scanned with a Cameca SX100 Ultra electron microprobe to produce images revealing element abundance, mineral composition, and grain angularity. Lamination thicknesses were measured from one Milpitas Wash sample and evaluated for thickness periodicity with Fourier-transform spectral-power analysis.

Two samples of laminated marl from the Cibola area were analyzed for oxygen and carbon isotope ratios. For each sample, 12 alternating white and dark laminations in continuous sequences were sampled with a microscope-mounted 0.3-mm-diameter dental-drill bur. Carbonate in each microsample was analyzed for oxygen and carbon isotopes using an automated carbonate preparation device (KIEL-III) coupled to a gas-ratio mass spectrometer (Finnigan MAT 252). Powdered samples were reacted with dehydrated phosphoric acid under vacuum at 70 °C. Phosphoric acid does not react with silicates or organic matter in samples. One sigma uncertainty of isotope measurements is ± 0.10‰ for δ18O and ± 0.08‰ for δ13C based on repeated measurements of NBS-19 and NBS-18. NBS-19 (n = 5) is processed with every daily batch of samples. NBS-18 is run weekly and whenever operating conditions or focusing parameters are changed. Assigned values relative to Vienna Pee Dee Belemnite (VPDB) (δ13C, δ18O) used for these standards are +1.95‰ and −2.20‰, respectively, for NBS-19, and −5.01‰ and −23.2 ‰, respectively, for NBS-18. Sample analysis data are reported in the Supplemental Material1. All analyses are reported relative to the VPDB standard (Coplen, 1994).

Electron-microprobe maps of total electron backscatter (BSE) and of calcium and silicon abundance in a polished thin section of laminated marl from lower Milpitas Wash show clear alternations in the ratio of calcite to clay with very few identifiable silicic grains (Figs. 5A5C; see  Appendix for methods). A sequence of BSE images at progressively greater magnification (Figs. 5D5F) shows that calcareous laminations consist largely of 2–10 μm calcite grains. A sequence of 21 photographs of a polished thin section of this unit (Spencer et al., 2018) revealed ambiguity in assigning lamination boundaries to five of 32 laminations, including two calcareous laminations that are weakly defined and only slightly more calcareous than bounding claystone laminations.

Electron-microprobe maps of calcium, silicon, and potassium in a polished thin section of laminated marl from upper Cibola Wash show clear alternations of micrite and siltstone in which calcium is concentrated in micrite and silicon and potassium are concentrated in siliciclastic layers (Fig. 6). Magnesium, sodium, and aluminum are also associated with silty layers in electron microprobe images as is expected for siliciclastic grains. It is also apparent in the microprobe images that the sharpness of lamination transitions is variable and that some laminations have slightly irregular boundaries. Gradational boundaries for some laminations are especially apparent at higher magnification (Fig. 6, lower). At even higher magnification, ∼2–20 μm calcite grains with minor clay and other minerals make up a calcareous lamination (Fig. 7A). This contrasts with the relatively coarse (40–150 μm) angular minerals in a silty lamination (Fig. 7B). The high angularity of siliceous grains, with highly elongate mica flakes, indicates minimal transport by fluvial processes, wave action, or repetitive tidal currents. Siliciclastic mineralogy suggests a granitic source, which is consistent with Jurassic granitoids exposed in the adjacent range front less than 1 km away (Fig. 1).

Individual laminations from two laminated silty marl samples collected from the basal carbonate unit in upper Cibola Wash (Figs. 23) were analyzed for oxygen and carbon isotope ratios. Resulting analyses revealed alternating δ18O and δ13C with positive covariance in which the higher δ18O and δ13C are associated with silty laminations (Fig. 8). A continuous, high-resolution sequence of samples, from the center of a dark silty layer through a light calcareous layer to the overlying dark silty layer, yielded the same relationship in which isotopic variation corresponds to minor variations in color such that darker colors correspond to higher δ18O and δ13C (Fig. 9). Positive covariance of isotopic composition is apparent in a diagram of δ18O versus δ13C in which lines link adjacent laminations (Fig. 10).

Positive covariance of δ18O and δ13C is characteristic of lake sediments from other lacustrine settings, especially within closed basins, where the range of isotopic compositions is greater over larger stratigraphic intervals (Fig. 11; Talbot, 1990; Drummond et al., 1995). Included in Figure 11 are data on the basal 5 cm of Bouse Formation marl over a range of elevations in Cibola Wash (Roskowski et al., 2010). Also plotted are δ18O and δ13C from a set of laminations from Milpitas Wash (sample 6B in Figs. 3 and 4B), where only two of seven dark, silty layers yielded enough carbonate for analysis (square points in Fig. 11). All six analyses, with δ18O at approximately −9‰ to −10‰, fall in a cluster slightly to the left of the stratigraphically lower, more calcareous laminations in Cibola Wash (Fig. 11). All analyses from both sites are characterized by δ18O of less than −7‰ VPDB; such values are not found in unaltered marine carbonates and are extremely rare in estuarine systems (e.g., Ingram et al., 1996; Sampei et al., 2005).

Milpitas Wash site C1 includes laminated strata with generally sharp, planar boundaries between darker claystone laminations and lighter calcareous laminations (Fig. 4C). Thickness measurements of a sequence of 69 laminations, in which each lamination consists of a couplet of lower calcareous and upper clay-rich components, identified a bimodal distribution of thicknesses with greater thicknesses above about layer 50 (Fig. 12A). There is little evidence of other trends or an overarching pattern in this data set except perhaps for slightly lower thicknesses for, approximately, layers 20–45. The simplest general interpretation is that something changed in the geologic environment at about the time of lamination 50 that resulted in thicker laminations (e.g., Cohen et al., 2006).

The 69-layer thicknesses were analyzed by Fourier spectral analysis using PAST© software (Hammer et al., 2001), which yielded a simple periodogram showing identified dominant frequencies in the thickness sequence. Only one peak, with a period of 60.4 laminations, is statistically significant in this analysis. That peak is apparent in Figure 12A, where the slight trough in the middle of the graph and the upturn at the left and, especially, the right end are reflected in a sine wave with a frequency of 60.4 layers per cycle. To evaluate the significance of the Fourier spectral analysis, we subtracted this sinusoidal component from the thickness measurements and recalculated the correlation coefficient associated with a least-squares fit to a sloping line. As shown in Figure 12B, the correlation coefficient is greater. However, only 1.14 cycles are represented by the power spectrum peak at this frequency. Consequently, it is not possible to distinguish between a truly periodic component to the thickness data and an upturn at lamination 50 that reflects a non-periodic geologic event or process.

To further evaluate the significance of periodic tidal signatures in Bouse strata, we applied Fourier spectral analysis to data represented in figures 3B–3C of O'Connell et al. (2017). As with Milpitas Wash sample 6B, we combined light and dark layers (dark on top) before analysis. Fourier power-spectrum analysis of the two lamination sequences did not identify any spectral peaks with statistical confidence above the 95% level (Spencer et al., 2018). This lack of statistical significance is possibly exacerbated by a problem with the lamination thickness data reported by O'Connell et al. (2017) in which the sample 3B laminations include an exact duplication of thickness measurements for laminations 54–89 and 90–125. With Fourier spectral analysis, the null hypothesis is that all spectral power is random noise without any periodic signal that can be extracted from the complex power spectra. Failure to disprove the null hypothesis at the 95% confidence level is interpreted here to indicate that there are no statistically significant periodicities in the lamination data reported by O'Connell et al. (2017). Finally, we note that the Fourier transforms presented in figures 3B–3C of O'Connell et al. (2017) each include a peak at ∼2 cycles per couplet. Since a couplet represents two thickness measurements, a peak at ∼2 cycles per couplet indicates spectral power at twice the sampling frequency. Representation of frequencies higher than half the sampling frequency is inappropriate as sampled frequencies are beyond the Nyquist limit (Press et al., 1986).

The composition of the marl sampled in Milpitas Wash, with alternating laminations of claystone and micrite, is consistent with an origin as annual varves reflecting seasonal changes as inferred by Homan (2014). This interpretation is based in part on the compositional contrast between the dark and light portions of each lamination as there is no plausible mechanism to trigger carbonate production on the short time frame of diurnal or semidiurnal tidal cycles. In addition, there is a complete lack of evidence for sorting of clastic material by tidal currents as there are almost no silt or sand grains or bioclastic debris. Bioclastic debris is abundant in nearshore Bouse carbonates but completely absent in the Milpitas Wash sample. Lamination composition is consistent with clay and fine calcite settling out in deep water without any influence from tidal or other currents and with only seasonal variation in calcite production (e.g., Hodell et al., 1998; Teranes et al., 1999; Trapote et al., 2018). In this interpretation the 1–3 mm lamination couplets indicate a sedimentation rate of 1–3 mm/yr. In contrast, tidal rhythmite lamination couplets are deposited once or twice per day and reflect periodic sediment transport and deposition with changing grain size during each tidal cycle (e.g., Archer and Johnson, 1997; Williams, 2000; Coughenour et al., 2009). Associated deposition rates would be in the range of 0.5–2.0 m per year with currents strong enough to transport siliciclastic grains from areas submerged by high tides to sites of tidalite deposition. The almost complete lack of siliciclastic grains argues against influence from recurring tidal or shallow-water currents. Furthermore, the Milpitas Wash sample was collected from just above the distinctive clay layer that was interpreted as marking the time of maximum water depth and the initiation of lake overflow and outflow channel incision through the Chocolate Mountains to the south (Bright et al., 2016). At this time, the Milpitas Wash sample site would have been far from shore and beneath as much as 245 m of water (Fig. 13).

Laminated marl from the Cibola Wash sample site is similar to the laminated marl at the Milpitas Wash sample site but with generally less clay, more carbonate, and the addition of silt and fine sand in the darker half of the lamination couplets. We interpret the siliciclastic component as having been derived from Jurassic granitoids exposed in the mountain front directly to the east (Fig. 1C; Tosdal and Wooden, 2015) with delivery of sand and silt to the lake margin during winter storms and associated runoff. Depth could have been much shallower than at the Milpitas Wash sample site but deep enough to be unaffected by tides or other shallow water currents.

The alternation of δ18O and δ13C values between micritic and siliciclastic laminations in the two Cibola Wash samples is readily explained as a consequence of seasonal processes. Freshwater influx to lakes generally delivers water low in 18O and 13C, whereas evaporation of water and dissolved CO2 lead to enrichment in these heavy isotopes in remaining lake water. Preferential incorporation of 12C into organic carbon during photosynthesis and removal through settling of organic matter, generally during times of increased sunlight and evaporation, also lead to 13C enrichment in remaining lake water. These processes often operate together to produce positive covariation in oxygen and carbon isotopic signatures in calcareous sediments, especially in closed, semi-closed, or stratified lakes (e.g., Talbot, 1990; Drummond et al., 1995; Li and Ku, 1997). Diverse environmental situations can also affect lake water to produce negative covariance (Hodell et al., 1998; Teranes et al., 1999), but in either case covariance is a consequence of seasonal changes.

The Colorado River drains a large upland area that receives voluminous winter precipitation and likely did so in the early Pliocene (Ibarra et al., 2018). Runoff would have been depleted in 18O and 13C compared to downstream lake water that had been enriched in 18O and 13C by in situ evaporation and precipitation of 12C-enriched organic carbon over perhaps hundreds to thousands of years. It is also possible that runoff was enriched in 12C due to equilibration with near-surface soil carbon composed largely of 12C-enriched organic matter. Colorado River base flow derived from deeper groundwater also could have added water with higher δ18O and δ13C than spring and early summer runoff (Crossey et al., 2015). Our interpretation is entirely consistent with the “brackish lake–freshwater plume model for the southern Bouse Formation” of Bright et al. (2018a, p. 1902), who proposed that lower δ18O values “in the micrite represent more seasonally limited epilimnic calcium carbonate production in paleolake Blythe during spring/summer floods on the early Colorado River.”

Significant post-depositional alteration of just one of the two isotopes would have eliminated the isotopic covariation. We therefore interpret the isotopic data as reflecting the original composition of the carbonate rather than post-depositional alteration. Similar but negative covariation across growth bands in a Bouse barnacle shell was interpreted as evidence that carbon and oxygen isotopes are unaltered (Roskowski et al., 2010).

The Cibola Wash samples were collected from a 4-m-thick interval of laminated silty marl that is both underlain and overlain by bioclastic limestone (Fig. 3; Homan, 2014). The bioclastic limestone was deposited in nearshore environments where waves and currents produced well-sorted, sub-rounded to well-rounded calcarenite and fossil hash and a variety of types of cross beds (Gootee et al., 2016; Dorsey et al., 2018). The juxtaposition of laminated marl above bioclastic limestone in this sequence indicates increasing water depth but not so deep as to prevent seasonal delivery of silt and fine sand from the nearshore environment. Laminated marl deposition was followed by decreasing water depth and deposition of the upper bioclastic limestone layer in a nearshore environment. Overlying marl and stratigraphically higher claystone and siltstone indicate return of the deep-water environment (Bright et al., 2016).

The cause of inferred shallowing associated with the upper bioclastic limestone subunit in upper Cibola Wash is uncertain. Numerical modeling of filling and spilling in a sequence of Bouse lakes indicates lake water rise followed by spillover and rapid decline with inflow and outflow channel incision rate as the primary variable controlling lake level and salinity evolution (Fig. 13; Spencer et al., 2008, 2013). Measured sections indicate substantial stratigraphic diversity for the basal carbonate unit in the Cibola area such that most stratigraphic sections do not have distinctive lower and upper bioclastic limestone subunits (Homan, 2014), which raises the possibility that shallowing leading to deposition of the upper bioclastic limestone subunit in upper Cibola Wash was a minor event. One possibility is that a brief lake lowering occurred when rising Paleolake Blythe spilled over the saddle between the lower Colorado River Valley and the basin containing Ford Lake and Palen Lake playas west of Blythe (Fig. 1B). With modern topography, this would result in only 1–2 m of water level decline, but the basin containing Ford Lake and Palen Lake playas was likely substantially deeper at 4.8 Ma and perhaps accommodated a spillover that caused an abrupt decline in lake level of 3–6 m or more. The saddle, now at 140 m elevation, is higher than the Cibola sample site at 122 m, but the saddle would have been lower before alluvial fan aggradation, and the Cibola site would have been higher prior to downfaulting on the nearby normal fault directly to the east. Maximum offset of the Bouse Formation along this pre-Quaternary, west-side-down fault is ∼30 m with offset decreasing to the north over 5 km of intermittent exposure (Figs. 1C and 2; Gootee et al., 2016).

Fourier spectral analysis of 69 lamination thickness measurements from Milpitas Wash identified a statistically significant periodicity of 60.4 laminations per cycle. Lamination thicknesses plot in a horizontal band with the primary deviation represented by an abrupt increase in lamination thicknesses at about lamination 50. This deviation appears to be the primary cause of the identified periodicity. Subtracting the periodic signal from the thickness data yields a more linear trend with a higher correlation coefficient, which supports the interpretation that a periodic signal is superimposed on the linear trend. Lack of data over multiple cycles, however, renders the interpretation of a periodic tidal signature suspect. Non-periodic geologic processes could have influenced or caused the abrupt increase in lamination thicknesses without any periodic influence. The step in the data is clearly not random noise, as is apparent both from visual inspection and Fourier spectral analysis, but Fourier analysis has not demonstrated that it is periodic. This should serve as a caution about identifying periodicity in any Fourier power spectrum without determining the number of cycles represented by power spectrum peaks.

  1. Laminated marl in Milpitas Wash consists of alternating layers of claystone and micrite in which almost all calcite grains are less than ∼10 μm in diameter. Internal compositional variation in each lamination is due primarily or exclusively to varying abundance of fine calcite grains. There is no evidence of sorting by grain size as is expected for transport and deposition by tidal or other nearshore or shallow water currents. We infer that the Milpitas Wash laminated strata were gently and gradually deposited in deep water on a seasonal rather than tidal schedule, which is consistent with the interpretation of Homan (2014).

  2. Laminated marl in upper Cibola Wash at the site of purported tidalites (O'Connell et al., 2017) is similar in composition and thickness to laminations at Milpitas Wash except that carbonate content is greater in the micrite layers, and texturally immature silt and fine sand are a component of the siliceous layers. Siliciclastic grains consist primarily of angular quartz and feldspar and elongate mica flakes that were not significantly rounded during transport. Granitoids exposed in the range front <1 km to the east are the obvious source of the angular grains. We infer that the siliciclastic component was derived from this range front and deposited in water that was deep enough to be unaffected by shoreline waves and currents but not so deep as to be out of reach of silt and fine sand delivered to the lake margin by winter-storm runoff.

  3. Oxygen isotope and carbon isotope analyses of carbonate from 12 laminations in each of two Cibola Wash samples indicate alternating δ18O and δ13C between silty and calcareous laminations. We interpret this alternating pattern to reflect seasonal variations in a lake rather than estuarine tidal ebb and flow on a daily or twice daily schedule. Specifically, runoff from the upper Colorado River basin produced a shallow, freshwater plume each spring and early summer, and carbonate produced by shallow water biological activity during these seasons reflects the low δ18O and δ13C of the runoff. In contrast, older, more evaporatively modified lake water with higher δ18O and δ13C influenced the composition of less abundant carbonates deposited during fall and winter along with fine sand that was transported by local runoff during winter storms. This interpretation is consistent with the concept of an isotopically stratified water column identified by Bright et al. (2018a) based on oxygen and carbon isotopic analysis of carbonates and ostracods with lake water stratification maintained by seasonal influx of low-δ18O and low-δ13C runoff from the upper Colorado River catchment.

  4. Variation of δ18O and δ13C in carbonate both between and within Bouse laminations reflects the absence of post-depositional alteration as these fine-scale variations would have been homogenized if significant alteration had occurred. We conclude that post-depositional alteration was minimal or non-existent as have earlier studies of δ18O and δ13C in barnacle-shell growth bands (Roskowski et al., 2010) and 87Sr/86Sr in Bouse carbonates and post-depositional carbonate veins, crusts, and caliches (Spencer and Patchett, 1997).

  5. Fourier spectral analysis of a sequence of 69 lamination thicknesses from a sample of Milpitas Wash calcareous claystone identified a periodicity at 60.4 laminations per cycle that could be distinguished from random noise with more than 95% confidence. However, a step in thickness data at about lamination 50 likely has a geologic origin that is not random, but it is not necessarily periodic. We conclude that inference of a periodic signal in a lamination-thickness sequence is unwarranted unless multiple cycles have actually been sampled.

  6. Spectral analysis of two lamination thickness data sets from Cibola Wash, measured by O'Connell et al. (2017) and interpreted as indicators of tidal activity, did not identify any statistically significant periodicities (Spencer et al., 2018), and one of the two data sets contains a data duplication that appears to be erroneous. Furthermore, Fourier spectral analysis by O'Connell et al. (2017) inappropriately identified periodicities higher than half of the sampling rate (Press et al., 1986).

  7. The high sedimentation rates required by the tidal interpretation of laminations in upper Cibola Wash, and the extreme contrast with annual deposition of laminations in Milpitas Wash, are unnecessary in our analysis, which indicates similar, 1–3 mm/yr sedimentation rates in both areas.

  8. We conclude that lamination thicknesses, petrology, and O and C isotopic character in the Milpitas Wash and Cibola Wash sites support seasonally varying deposition in a lacustrine environment without any influence from tides. This is consistent with the concept that the Bouse Formation is irrelevant to the timing of Colorado Plateau uplift.

We thank Andy Cohen and Jordan Bright for comments on an earlier draft that led to significant improvements, and Daniel Ibarra, David Miller, an anonymous reviewer, and associate editor Karl Karlstrom for constructive reviews that led to greater clarity and better focus. Thanks! J. Spencer thanks Jon Patchett, Phil Pearthree, Brian Gootee, and Kyle House for informative discussions over many years regarding the Bouse Formation. Geologic mapping by J. Spencer in the Cibola 7.5′ Quadrangle in 2015 was jointly funded by the Arizona Geological Survey and the U.S. Geological Survey under STATEMAP assistance award #G13AC00374.

1Supplemental Material. Oxygen and carbon isotope analyses of laminated Bouse marl. Please visit to access the supplemental material, and contact with any questions.
Science Editor: Andrea Hampel
Guest Associate Editor: Karl E. Karlstrom


Backscattered electron (BSE) imaging and chemical element X-ray mapping were performed using the CAMECA SX100 Ultra electron microprobe at the University of Arizona, equipped with an LaB6 filament electron source and five wavelength dispersive spectrometers for detecting X-rays emitted from the sample. BSE images were obtained by either scanning the electron beam over an area of the sample or, for larger areas, by moving the sample back and forth under a stationary electron beam and recording the spatial location and number of high-energy “backscattered” electrons produced by the sample. The number of backscattered electrons scattered by the sample (and thus the brightness of the BSE image) increases with the average atomic number of the material in the sample. An image is thereby produced showing areas that differ in average atomic number, which is often useful in distinguishing different mineral species.

Chemical X-ray maps were produced by setting each spectrometer to detect the appropriate X-ray wavelength for a specified element. The sample was then moved back and forth under a stationary electron beam, and the number of X-rays produced at each point in the sample area was recorded. The number of characteristic elemental X-rays produced at any location (and thus the brightness of the X-ray image) is directly proportional to the concentration of that element at that location. As a result, the X-ray map displays the spatial variation in the chemical concentration of that element within the sample area. This technique is useful in revealing minerals with similar chemical compositions as well as chemical gradients both within individual minerals and across larger areas.

Gold Open Access: This paper is published under the terms of the CC-BY-NC license.