Subsidence history of the Ediacaran Johnnie Formation and related strata of southwest Laurentia: Implications for the age and duration of the Shuram isotopic excursion and animal evolution

Ediacaran strata record a critical period in Earth history (635–541 Ma), during which metazoan life first appeared (Knoll et al., 2004, 2006; Narbonne et al., 2012). They also record a significant rise in atmospheric and oceanic oxygen (Fike et al., 2006; Canfield et al., 2007; McFadden et al., 2008; Sahoo et al., 2012), which was a prerequisite to metabolic function in animals (Knoll and Carroll, 1999; Och and Shields-Zhou, 2012). Neoproterozoic oxygenation resulted in atmospheric oxygen levels generally interpreted as similar to those of the present day (Holland, 2006; Kump, 2008). Today, atmospheric oxygen levels are maintained by photosynthesis from land plants and marine organisms in roughly equal proportions (e.g., Field et al., 1998). It has therefore long been enigmatic that land plants are not preserved in rocks older than ca. 400 Ma, or ∼150 m.y. later than the first appearance of animals. For that reason, it is widely presumed that the rise of animal life required sufficient oxygen production, from either marine photosynthesis, or perhaps some sort of “bootstrap” mechanism from animals themselves, to survive (e.g., Butterfield, 2009; Lenton et al., 2014). In any event, progress toward understanding the fundamental question, “what is the origin of animals?,” hinges in part on understanding how and when oxygen became sufficiently available to make animal metabolism possible (e.g., Nursall, 1959).

Among the most fruitful avenues of research along these lines to date has been exploration of proxies for the chemistry of seawater in which animal life first appeared, primarily the stable isotope geochemistry of shallow-marine carbonate strata. The best-preserved Ediacaran strata around the globe that contain carbonate all feature a singularly large (by about a factor of two) negative anomaly in the isotopic composition of carbon, which has been attributed primarily to the isotopic composition of ancient seawater itself (Fike et al., 2006; McFadden et al., 2008; but for an alternative view, see Swart and Kennedy, 2012). The anomaly is best preserved and documented in the Ediacaran Shuram Formation in Oman (Burns and Matter, 1993; Le Guerroué et al., 2006a, 2006b; Osburn et al., 2015), and it is generally referred to as the “Shuram excursion,” taking its name from the discovery formation. A similar excursion has been documented in Neoproterozoic sections on five of Earth’s seven modern continents, and it occurs only once in each section: Africa (Kaufman et al., 1991; Halverson et al., 2005), Asia (Burns and Matter, 1993; Condon et al., 2005; Melezhik et al., 2005; Fike et al., 2006; McFadden et al., 2008; Macdonald et al., 2009; Osburn et al., 2015), Australia (Calver, 2000; Husson et al., 2015), Europe (Melezhik et al., 2005; Prave et al., 2009), and North America (Myrow and Kaufman, 1999; Corsetti et al., 2000; Corsetti and Kaufman, 2003; Kaufman et al., 2007; Bergmann et al., 2011; Petterson et al., 2011; Verdel et al., 2011; Macdonald et al., 2013). The Shuram excursion is the largest known Neoproterozoic or younger carbon isotope anomaly (Grotzinger et al., 2011), and its magnitude is among the largest recorded in Earth history (see, for example, the Paleoproterozoic Lomagundi-Jatuli excursion; Bekker and Holland, 2012).

Global chemostratigraphic expression of the Shuram excursion is a remarkable discovery from at least three perspectives. First, it represents a presumably isochronous fingerprint of a specific interval of time from sections with notoriously sparse age constraints. Second, it implies that a geologically extreme event of uncertain origin occurred at the same time as the rise of animals. Last, the singular magnitude of the excursion contributes to the goal of creating a global composite time series of secular variations in marine carbon isotope ratios. In regard to the third point, the duration of the anomaly raises the potential for using the shapes of the curves, rather than simply the magnitudes of the excursions, as a correlation tool from section to section; this of course presumes a relatively constant sedimentation rate at the hundred-meter scale (Halverson et al., 2005; Saltzman and Thomas, 2012). For the Shuram excursion, δ¹³C values rapidly descend with stratigraphic position to <–11‰, followed by a recovery that is at first gradual and then moderate in slope, with the change occurring near –4‰ (fig. 2 inCondon et al., 2005; fig. 3 inPrave et al., 2009; fig. 16 inVerdel et al., 2011; fig. 3 inGrotzinger et al., 2011; fig. 13 inMacdonald et al., 2013; fig. 1A inHusson et al., 2015).

At present, the most significant impediment to understanding Ediacaran biostratigraphy is the lack of internal age control in most sections around the globe. The ages of the boundaries of the Ediacaran Period are well defined radiometrically in multiple sections. The base is defined by the lithologically distinctive post-Marinoan cap carbonate sequence, which is associated with a –6‰ δ¹³C excursion in carbonate and is precisely dated at 635 Ma in Namibia and China (Hoffmann et al., 2004; Condon et al., 2005). The top is defined by the first appearance of the trace fossil Treptichnus pedum (541 Ma), which is also associated with a –6‰ δ¹³C excursion in carbonate. Other than the first appearance of large Ediacaran body fossils, which usually occurs rather high in most sections relative to the Ediacaran-Cambrian boundary, the Shuram excursion has emerged as the single most distinctive stratigraphic datum that is globally recognized. However, its precise age is poorly constrained, precluding any attempt to meaningfully subdivide some 94 m.y. of Ediacaran time, and creating first-order uncertainties in the relative timing of major environmental and biostratigraphic events (Xiao et al., 2016). A second major stratigraphic feature, largely restricted to sections in the North Atlantic region, is the Gaskiers glaciation (Myrow and Kaufman, 1999), which, in contrast to the Shuram event, is precisely dated at 579 Ma (Bowring et al., 2003a, 2003b; Pu et al., 2016). The mismatch between sections with glaciogenic rocks and precise radiometric ages on one hand, and the Shuram excursion in carbonate strata on the other, has left it uncertain whether or not these two events are correlative (Xiao et al., 2016). A 580 Ma age for the Shuram excursion provides an obvious correlation between the two most conspicuous events in the Ediacaran record (e.g., Xiao et al., 2004; Fike et al., 2006; Zhou et al., 2007; Halverson et al., 2005, 2010; Loyd et al., 2012; Schiffbauer et al., 2016). Alternatively, the stratigraphic proximity of the Shuram excursion to the Precambrian-Cambrian boundary, and a 551 Ma ash bed near the apparent upper zero crossing of the excursion in the Doushantuo Formation of China, suggest that it may be as much as 20–30 m.y. younger than the Gaskiers glaciation (Condon et al., 2005; Bowring et al., 2007; Cohen et al., 2009; Sawaki et al., 2010; Narbonne et al., 2012; Macdonald et al., 2013; Tahata et al., 2013; Xiao et al., 2016).

One chronological tool that has heretofore only been sparingly applied to Ediacaran strata is thermal subsidence analysis (e.g., Le Guerroué et al., 2006b). It is well known that thermal subsidence associated with seafloor spreading is a useful chronometer that can predict the age of the ocean floor based on the exponential decay of its elevation with respect to the abyssal plains for lithosphere older than 20 m.y. (e.g., equation 22 in Parsons and Sclater, 1977). The same principle also applies to models of the subsidence history of passive-margin basins, which include an initial thickness of newly stretched continental crust and substantial sediment loading (McKenzie, 1978). The decay is predicted by laws of diffusive heat transport of physical rigor that are on par with laws of closed-system radioactive decay used to date the timing of crystallization of minerals. The principal limitations in using thermal conduction as a chronometer are (1) the requirement that subsidence records thermal relaxation without significant mechanical modification of the lithosphere, such as extension, flexural loading, instability of a thermal boundary layer, or unmodeled sources of dynamic topography; and (2) corrections of the observed stratigraphic subsidence for the compaction and lithification of sediment after deposition, and for water depth and changes in sea level (e.g., Steckler and Watts, 1978; Allen and Allen, 2005).

In comparison with Phanerozoic sedimentary basins, published subsidence analyses of Ediacaran strata have been limited, with most of the effort thus far concentrated on the western Laurentian continental margin (Stewart and Suczek, 1977; Bond et al., 1983; Armin and Mayer, 1983; Levy and Christie-Blick, 1991; Yonkee et al., 2014). The focus on this region as a testing ground for thermal subsidence modeling was due to the fact that it is perhaps the best-preserved example of an ancient passive margin, analogous to present-day Atlantic-type margins, but with virtually complete surface exposure of apparent synrift and postrift sedimentary archives spanning several hundred million years (Stewart, 1972; Gabrielse, 1972; Burchfiel and Davis, 1972, 1975; Stewart and Poole, 1974; Dickinson, 1977; Monger and Price, 1979). Because these sequences span the Ediacaran-Cambrian boundary, such that roughly half their thickness is Proterozoic in age, temporal control on subsidence has been restricted mainly to the Phanerozoic portion of subsidence curves. The lack of age control on the lower part of the section precludes precise definition of the transition from mechanical extension to pure thermal subsidence. Fortunately, the accurate definition of an exponentially decaying system, in particular, extrapolating stratigraphic age backward in time from a curve with known ages, is independent of the timing of onset and total amount of purely thermal subsidence.

Here, we address the problem of the correlation and age of the Shuram isotopic excursion through lithostratigraphic and chemostratigraphic study of the type locality of the Ediacaran Johnnie Formation in the Spring Mountains of southern Nevada. The Johnnie Formation is at least 1800 m thick at the type locality, and it makes up more than half of the maximum known thickness of ∼3000 m of total Ediacaran strata exposed in this region. The underlying Noonday Formation provided the first isotopic match between the Marinoan cap carbonate sequence in Namibia (Hoffman et al., 1998) and a section from another continent (Petterson et al., 2011). The overlying Stirling and Wood Canyon Formations contain Ediacaran and Lower Cambrian fossil assemblages that define the Cambrian-Precambrian boundary within the lower part of the Wood Canyon Formation (Corsetti and Hagadorn, 2000; Hagadorn and Waggoner, 2000), 1200 m above the top of the Johnnie Formation in the Spring Mountains. The uppermost 300 m of section of the Johnnie Formation contains the best expression of the Shuram excursion in Laurentia (Corsetti and Kaufman, 2003; Kaufman et al., 2007; Bergmann et al., 2011; Verdel et al., 2011). Therefore, to the extent that the section was deposited at or very near sea level on a thermally subsiding continental shelf, subsidence analysis may be used to estimate the age of the Shuram excursion and perhaps even broadly constrain the overall age of the Johnnie Formation.

Neoproterozoic–Cambrian strata in western Laurentia are divisible into two principal components, including a lower diamictite and volcanic sequence, and an upper terrigenous detrital sequence (Stewart and Suczek, 1977; Poole et al., 1992). The Johnnie Formation is the lowest siliciclastic formation in the upper terrigenous detrital sequence, forming the basal deposits of a westward-thickening continental margin terrace wedge, widely regarded to have developed in the wake of late Neoproterozoic rifting of the Rodinian supercontinent (Li et al., 2008, 2013). The formation is a few hundred meters thick near its eastern pinchout beneath Lower Cambrian cratonic strata, systematically increasing to at least 1500 m thick in its westernmost exposures, where the base is not definitively exposed (Stewart, 1970; this report). Lithologically, it is primarily variegated siltstone and very fine-grained sandstone that contains varying amounts (10%–40%) of carbonate and orthoquartzite, distinguishing it from the carbonate-dominated Noonday Formation below and coarse siliciclastic rocks of the Stirling Formation above (Fig. 1).

The Johnnie Formation was first defined and described in the northwest Spring Mountains in the Johnnie Wash area (Fig. 2; Nolan, 1924, 1929), where its contact with the underlying Noonday Formation is apparently not exposed, and hence its thickness is a minimum for this location. Nolan’s (1924) thickness and description were included in the regional stratigraphic synthesis of Stewart (1970). The type locality was subsequently mapped and briefly described by Burchfiel (1964, 1965), and relatively complete lithostratigraphic sections were measured by Hamill (1966) and Benmore (1978). The type locality has since received little attention in comparison to the much thinner sections in the Nopah Range and environs 70 km to the south, or equivalents 100 km to the west in the Panamint Range, where its basal contact with the Noonday Formation is extensively exposed (e.g., Hazzard, 1937; Wright and Troxel, 1966; Labotka et al., 1980; Albee et al., 1981; Benmore, 1978; Summa, 1993; Fedo and Cooper, 2001; Corsetti and Kaufman, 2003; Kaufman et al., 2007; Verdel et al., 2011). With the exceptions of detailed studies of parts of the formation (Summa, 1993; Abolins, 1999; Bergmann et al., 2011), no systematic attempt has yet been made to describe and interpret the entire formation at its type locality in terms of key bed forms, depositional environments, sequence architecture, or chemostratigraphy, at the level of more southerly or westerly sections.

The uppermost part of the Johnnie Formation, the Rainstorm Member, is a lithostratigraphically distinctive unit that can be correlated with confidence over a broad region of southwestern North America, including eastern California and southern Nevada (Stewart, 1970), and it probably occurs as far south as northern Sonora, Mexico, where it forms a part of the Clemente Formation (Stewart et al., 1984). The basal strata of the Rainstorm Member are its most distinctive part. They include a thin (∼2 m), siltstone-enveloped, regionally extensive oolitic marker bed known as the “Johnnie oolite” (e.g., Bergmann et al., 2011). The oolite is underlain by greenish gray siltstone, and it is overlain by distinctive pale-red, fine-grained sandstone with or without associated sandy or silty micrites (“liver-colored limestones”). The overlying units characteristically contain groove marks, flute casts, intraformational conglomerate, and other indicators of shallow-water, high-energy currents. These carbonates record the onset and most negative part of the Shuram excursion in eastern California and southern Nevada (Corsetti and Kaufman, 2003; Kaufman et al., 2007; Verdel et al., 2011), as well as in the Sonora sections (Loyd et al., 2012). Similar to formation-scale thickness variations in the terrigenous detrital sequence as a whole, the Rainstorm Member generally thickens westward from as little as 20 m in the thinnest measured section to more than 300 m in the thickest sections (Stewart, 1970; Verdel et al., 2011).

Lower and middle Johnnie Formation strata are sufficiently variable in their lithostratigraphy that recognition of regionally mappable members is not as straightforward as in the case of the Rainstorm Member. As noted by Summa (1993), sub–Rainstorm Member depositional settings of the Johnnie Formation are interpreted as inner-shelf to tidally influenced nearshore environments that were highly susceptible to sea-level fluctuation (Benmore, 1978; Fedo and Cooper, 2001; Schoenborn et al., 2012). As we describe herein, depositional environments tend to be more landward to the south and east in these units, as suggested by the abundance versus absence of desiccation features, fluvial versus marine deposition, and medium- to coarse-grained sandstones versus fine- to medium-grained sandstones. Although this variability complicates simple lithostratigraphic correlation, if interpreted correctly, it can be used as an effective indicator of sea-level rise and fall.

Reported age constraints from the Johnnie and correlative Clemente formations include (1) a 640 Ma U-Pb age from a single detrital zircon grain in sub–Rainstorm Member siltstones in the Panamint Range of eastern California (Verdel et al., 2011), and (2) potential Ediacaran body and trace fossils (e.g., Cyclomedusa plana and Palaeophycus tubularis, respectively) ∼75 m below the oolite in the Clemente Formation (McMenamin, 1996). The U-Pb age, because it is based on a single grain, is subject to the uncertainty of contamination during mineral processing and needs to be confirmed with duplicate analyses. The putative fossils have been questioned after examination by other paleontologists (e.g., J.W. Hagadorn, 2017, personal commun.), and they have generally not been accepted in subsequent stratigraphic studies of the region (e.g., Loyd et al., 2012, 2015). Latest Ediacaran fossils have been recovered from the uppermost Stirling Formation and the Lower Member of the Wood Canyon Formation in the Spring Mountains and neighboring Montgomery Mountains to the south (e.g., Cloudina and Swartpuntia; Hagadorn and Waggoner, 2000; Smith et al., 2017), from sections in stratigraphic continuity with the type Johnnie Formation. These are succeeded immediately upward by Lower Cambrian trace fossils (Treptichnus pedum), which places the Ediacaran-Cambrian boundary in the Lower Member of the Wood Canyon Formation (Fig. 1; Corsetti and Hagadorn, 2000).

The underlying Noonday Formation has been interpreted as the cap carbonate sequence of the Marinoan “snowball Earth” glaciation (Petterson et al., 2011), which by definition would place its base at the beginning of the Ediacaran period (635 Ma; Knoll et al., 2004, 2006; Narbonne et al., 2012). The Johnnie Formation’s basal contact with the Noonday Formation is lithostratigraphically gradational, transitioning from sandy dolostones of the upper Noonday Formation (Mahogany Flats Member of Petterson et al., 2011) to interstratified dolomitic sandstone and orthoquartzite in the lower Johnnie Formation (Transitional Member of Stewart, 1970). Although traditionally regarded as a conformable contact on the basis of this gradation (Hazzard, 1937; Stewart, 1970; Wright and Troxel, 1984), the identification of local karstic surfaces along the contact raises the possibility that it is a disconformity with a substantial depositional hiatus (Summa, 1993).

In terms of chemostratigraphic constraints on age, the conspicuous excursions to –6‰ at the base and top of the Ediacaran section are well expressed in the south Laurentian sections (e.g., Petterson et al., 2011; Smith et al., 2016). The presence of the Shuram excursion in the Rainstorm Member, despite its value as a correlation tool, does little to constrain the depositional age, because unlike the tightly constrained boundary excursions, hard chronological constraints are lacking, as noted already.

For almost a century, the terrigenous detrital sequence has been studied extensively on many different levels. Much of the early work focused on stratigraphic group-level packages that record the transition from Precambrian to Cambrian time (Nolan, 1929; Burchfiel, 1964; Stewart, 1970). More recent work on the Johnnie Formation has focused largely on outcrops in eastern California (Summa, 1993; Fedo and Cooper, 2001; Verdel et al., 2011; Schoenborn and Fedo, 2011; Schoenborn et al., 2012), or on specific features related to the Rainstorm Member, such as an incision-related conglomeratic member (Summa, 1993; Abolins, 1999; Abolins et al., 2000; Clapham and Corsetti, 2005; Verdel et al., 2011), giant ooids (Trower and Grotzinger, 2010), or detailed chemostratigraphy of the Johnnie oolite (Bergmann et al., 2011). The lower and middle portions of the Johnnie Formation have not been given as much detailed attention, except in areas close to the craton miogeoclinal hinge, where the Johnnie Formation is only a few hundred meters thick. The 1600 m stratigraphic thickness of sub–Shuram excursion Johnnie Formation at the type locality exceeds the thickness of any globally correlative Ediacaran strata of which we are aware. Furthermore, total Ediacaran stratigraphic thickness in southwest Laurentia measures over 3000 m, greater than the approximate thicknesses of sections in Australia (2500 m), Oman (1500 m), and China (300 m). Strata of the lower and middle Johnnie Formation at its type locality therefore represent one of the best opportunities among sections globally to provide a relatively complete record of Ediacaran time prior to the Shuram excursion. An important gap in our understanding of Ediacaran chemostratigraphy is the paucity of carbonate strata below the Shuram anomaly in most sections. Of the major global sections that contain it, only the Oman example contains abundant carbonate in immediately underlying strata, the Khufai Formation. Discovery of correlative carbonate-bearing strata in one or more sections around the globe would thus represent a significant step in expanding the global inventory of chemostratigraphic time series for a critical interval in Neoproterozoic time.

To identify a structurally intact section of the Johnnie Formation, we performed geologic mapping at 1:10,000 scale in the northwest Spring Mountains, Nevada, both of the type locality at Johnnie Wash, and in an area ∼4 km to the southwest near Nevada Highway 160, 3 km west-southwest of Mount Schader (Fig. 2). We used the Mount Schader, Nevada 1:24,000 quadrangle map (U.S. Geological Survey, 1968) as a topographic base. Our field mapping spanned 9 d total between 21 April 2015 and 2 May 2015. The geologic maps were used to identify optimum transects for measuring stratigraphic section. The Mount Schader section was measured and sampled in detail using a Jacobs staff mounted with a Brunton® compass set to the dip of bedding. For each stratigraphic subunit, we recorded: (1) fresh and weathered color of lithology using a Munsell color chart; (2) grain size; and (3) bedding thickness (supplemental text¹). Section was measured to the resolution of ∼0.5 m (or finer in some instances, if warranted). The Johnnie Wash section was measured using geologic cross sections, and the general lithologic characteristics were recorded in the field during geologic mapping (see Appendix for unit names and descriptions).

For carbon and oxygen isotope chemostratigraphy, we collected samples at 0.3–1 m resolution in carbonate units. Samples from the upper ∼400 m of sub–Rainstorm Member lithostratigraphic units were collected from the Mount Schader section during stratigraphic logging. Samples from two prominent carbonate horizons that occur below the deepest exposed strata of the Mount Schader section were collected in the Johnnie Wash locality of the Spring Mountains, and at a location ∼3 km north of Johnnie Wash (locality A in Fig. 2A). In total, 107 centimeter-scale sample chips were collected for carbon and oxygen isotopic analysis, including 36 from the Johnnie Wash section and locality A and 71 from the Mount Schader section. In the laboratory, sample chips were sliced open using a diamond-bladed wet saw to expose fresh, unweathered surfaces. From the fresh surfaces, a high-speed rotary tool with a diamond-tipped drill bit was used to powder the sample. We carefully extracted ∼0.1 mg of analyte from each sample chip, taking care to avoid any visible alteration or veining. Sample powder was loaded into vials, the air was purged and replaced with helium gas, and then digested in phosphoric acid at 72 °C for at least 1 h to evolve sufficient CO₂ gas for analysis. Carbon and oxygen isotope ratios were measured at Caltech using a Delta V Plus isotope ratio mass spectrometer (“gas bench”). Our values for δ¹³C and δ¹⁸O are reported relative to the Vienna Peedee belemnite (VPDB) standard in per mil notation. We used Caltech’s laboratory working standards, which were calibrated to NBS 18 and NBS 19 and have uncertainties of ±0.1‰. Standards were measured once for every nine samples to assess systematic error.

Our tectonic subsidence analysis is based on stratigraphic thicknesses compiled from multiple sources for the northwest Spring Mountains, Nevada. The inner-shelf to fluvial-deltaic facies of virtually all units within the terrigenous detrital sequence in this region suggest shallow-water deposition, removing the need for paleobathymetric correction (Levy and Christie-Blick, 1991). We used thicknesses from this study combined with thicknesses for overlying formations principally based on Stewart (1970) and Burchfiel et al. (1974). Our analysis encompasses known time points ranging from the Ediacaran-Cambrian boundary in the Lower Member of the Wood Canyon Formation (Corsetti and Hagadorn, 2000; Hagadorn and Waggoner, 2000; Smith et al., 2016, 2017) at 541 Ma, up through the Devonian-Mississippian boundary at the top of the Devils Gate Formation (Burchfiel et al., 1974) at 359 Ma (Ogg et al., 2016). The only previous attempt at a geohistory analysis of the Spring Mountains (Levy and Christie-Blick, 1991) was temporally constrained mainly by the Lower-Middle and Middle-Upper Cambrian boundaries, which were then deemed to be ∼30 m.y. older than their currently accepted ages. We followed methods described in Allen and Allen (2005) to delithify and progressively unload (backstrip) the stratigraphic column in order to obtain the tectonic component of subsidence. Delithification parameters for siliciclastic rocks were taken from table 9.1 inAllen and Allen (2005), and parameters for carbonate rocks were taken from Equation 3inHalley and Schmoker (1983). Tectonic subsidence curves were calculated using Backstrip, an open-source software for decompaction and tectonic subsidence calculations (Cardozo, 2009). Results for our earliest model runs were verified by hand using a spreadsheet program (e.g., Larrieu, 1995).

The most salient feature of the Johnnie Formation in the Johnnie Wash type locality (Fig. 3) is that, although very fine-grained sandstone and siltstone are present in all mappable units (distinguishing it from the overlying Stirling and underlying Noonday formations), three intervals are characterized by an abundance of fine- to medium-grained sandstone (identified with Roman numerals I, II, and III on Fig. 4). The sand-rich intervals range from 160 to 430 m thick, form distinct, resistant ridges within the otherwise recessive Johnnie Formation, and establish a basis for subdividing it into mappable units. Each sand-rich interval exhibits characteristics that readily distinguish it from the other two, in terms of either bed forms (intervals I and II) or parasequence architecture (interval III). Further subdivision of the formation is afforded by a conspicuous, 30–40-m-thick cherty carbonate unit near the middle of the section, and by lithological variation within sand-rich interval III. Our subdivision into map units includes the Rainstorm Member at the top, underlain by 12 informal units designated A through L (Fig. 4; Fig. S1 [^{footnote 1}]). Sand-rich intervals I and II define units B and D, and their enveloping sand-poor strata define units A, C, and E. The cherty carbonate marker and overlying sand-poor strata define units F and G. The upper sand-rich interval exhibits rhythmic variations of sandstone, siltstone, and carbonate that are divided into five units, H through L, each of which is defined at the base of a 30–100-m-thick sand-rich subinterval (Figs. 3 and 4; Fig. S1 [^{footnote 1}]).

At the Johnnie Wash type section, bedding strikes approximately north-south and dips moderately to steeply eastward (Fig. 3). The total thickness of sub–Rainstorm Member strata is 1595 m. The lowest stratigraphic unit (unit A) encountered is a recessive, slope-forming phyllitic siltstone that contains a continuous cleavage at high angle to bedding. The base of unit B is defined by the lowest occurrence of meter-scale orthoquartzite beds, which are abundant in the unit. Unit B is readily distinguished from higher sand-rich intervals by pervasive penecontemporaneous deformation. Nearly every sandstone horizon is affected, principally by ball-and-pillow structure, so much so that individual orthoquartzite beds are difficult to trace along strike for more than a few tens of meters. Individual ball-and-pillow structures are up to meter-scale in size (roughly equal to orthoquartzite bed thickness) and occur where fine-grained sandstone and siltstone underlie coarser sandstone. The ball-and-pillow structure is manifested in some places as simple load casts with folded lamination (Fig. 5A) and in others as completely detached sand bodies that have slumped downward into the underlying siltstone, surrounded by flame structure developed within the siltstone (Fig. 5B). At the type locality, the occurrence of ball-and-pillow structure in the Johnnie Formation is restricted to unit B, but it was also observed along one horizon at the top of the Rainstorm Member in the Mount Schader section (Figs. 6 and 7).

Unit C marks a return to generally inconspicuous, slope-forming siltstone with a prominent orthoquartzite marker horizon near the middle of the unit. The uppermost beds include a brown, resistant, 2-m-thick dolostone bed, which marks the lowest occurrence of carbonate in the type section.

Unit D includes orthoquartzite and less abundant siltstone. The sedimentary characteristic that distinguishes unit D from the other two sand-rich intervals is abundant high-angle cross-stratification, preserved in medium- to thick-bedded orthoquartzite (Fig. 8A). Millimeter- to centimeter-scale laminae or thin beds are preserved in foresets within decimeter- to meter-scale beds that can be followed for tens of meters along strike. Foreset lamination is consistently truncated at high angles, ranging from 20° to 30°, by overlying beds (Fig. 8B). In stratigraphic coordinates (corrected by tilting to horizontal about the line of strike), poles to foreset lamination are strongly unimodal, dispersed in trend by more than 90° around a mean vector of ∼N30°E 65° (Fig. 8B). We measured grain-size variation with stratigraphic height across a sequence of about 10 foreset layers (Figs. 8C and 8D) to test for the presence of reverse grading, which is characteristic of dry grain flows on the lee side of dunes (e.g., Hunter, 1977; Boggs, 2012). The results indicated that the mean grain size of ∼200 µm varies little through the sample, if anything fining slightly upward. In general, the lamination is not defined by concentrations of detrital heavy minerals. Opaque phases in these quartzites are largely diagenetic and relatively uniformly distributed throughout the rock.

The boundary between units D and E is among the most readily mappable in the area and is also associated with a color change on remote imagery from dark brown to light brown, which is the most conspicuous color contrast in the section (Fig. 2A). Unit E is ∼280 m thick, and it is dominated by very fine-grained sandstone and siltstone, generally lacking the mature, fine- to medium-grained quartzitic sandstone characteristic of unit D. Near the top, unit E contains an interval of ∼30 medium-bedded cycles that alternate between massive, immature fine-grained sandstone and laminated siltstone. Unit F is a conspicuous, 30–40-m-thick, gray cherty dolostone (Fig. 5C) that can be followed for at least 5 km along strike, albeit with some minor faulting. Unit G is 135 m thick, and it returns to siltstone and very fine-grained sandstone similar to unit E, with a few inconspicuous orthoquartzite beds.

The overall lithostratigraphic character changes beginning at the base of unit H, from relatively thick, homogeneous sandstone-, siltstone-, or carbonate-dominated units below to the far more compositionally heterogeneous units above. From the base of unit H up to the base of the Rainstorm Member, the section contains abundant orthoquartzite, defining the uppermost of the three sand-rich intervals in the Johnnie Formation (Fig. 4). For mapping purposes, the most straightforward subdivision of sand-rich interval III in the Johnnie Wash area is defined by five quartzite-dominated subunits ranging from 10 to 100 m thick, which define the lower parts of units H, I, J, K, and L (Fig. 4; Fig. S1 [^{footnote 1}]). Each of these subunits is overlain by variable thicknesses of recessive, variegated siltstone (Fig. 5E). The occurrence of carbonate is sporadic. In the section in Johnnie Wash, units I, J, and K are each capped by a resistant, 1–3-m-thick subunit of brown-weathering, laminated dolostone (Fig. 5D), and units H and L do not contain carbonate (Fig. 4). The Mount Schader section contains a lesser proportion of quartzitic sandstone and a greater proportion of carbonate and siltstone (Figs. 6 and 7), which is the basis for selecting it, instead of the type locality, for detailed measurement and chemostratigraphic sampling. Even within the Johnnie Wash area, the distribution of quartzite, siltstone, and carbonate changes along strike, on a scale of a few kilometers (Abolins, 1999). Although orthoquartzite beds in units H through L locally exhibit some high-angle cross-stratification in the Johnnie Wash section, they contrast with unit D (sand-rich interval II) in mainly being parallel-bedded or, in the case of the Mount Schader section, hummocky cross-stratified. Orthoquartzite in units H–L is generally fine-to medium-grained and appears to contrast with lower sand-rich intervals in containing a greater proportion of medium-grained and locally coarse-grained sand.

Our informal unit nomenclature ends at the base of the formally defined Rainstorm Member, which, as noted earlier, is readily identified throughout the region on the basis of lithologic characteristics. In the northern Spring Mountains, the Rainstorm Member contrasts with the underlying units H through L in lacking fine- to medium-grained orthoquartzite beds. At the base of the Rainstorm Member, a fissile, phyllitic siltstone is overlain by the ocher-colored, 2-m-thick Johnnie oolite. The ooids are up to ∼2 mm in diameter (Fig. 5F) and exhibit local cross-stratification. The oolite horizon has erosional basal and upper contacts, locally including intraformational breccia and conglomerate, which includes cobbles and small boulders of the oolite. Above the Johnnie oolite, pale-red limestones locally contain dispersed, coarse quartz grains interstratified with carbonate-cemented, fine-grained sandstone. These carbonate-rich rocks are overlain by argillaceous mudstone with interbedded limestones, with scattered horizons of intraformational conglomerate. The uppermost part of the Rainstorm Member in the Mount Schader section contains a 4-m-thick triad of quartzite, siltstone, and dolostone, in which the quartzite is disrupted by locally intense ball-and-pillow structure (Figs. 5G and 5H). The base of the overlying A Member of the Stirling Formation is marked by highly resistant, massively textured to cross-stratified, medium- to thick-bedded, medium- to coarse-grained orthoquartzite. The principal contrast between the Stirling Formation’s A Member and any of the orthoquartzites in the Johnnie Formation is the coarse grain size, including the common occurrence of granules and small pebbles of vein quartz and jasper. Neither the Johnnie Wash section nor the Mount Schader section appears to preserve the incised valley fill characteristic of the conglomeratic member of the Johnnie Formation (Abolins, 1999; Verdel et al., 2011).

Carbon isotope ratios range from a low of –4.4‰ (VPDB) to a high of 4.9‰ (Table S1 [^{footnote 1}]). The data are mainly concentrated in carbonate beds in units H through L, which constitute the uppermost 400 m of pre–Rainstorm Member strata (Fig. 7). Within the underlying ∼1100 m of exposed Johnnie Formation strata, carbonate intervals are present only in units C and F, 1440 m and 860 m below the base of the Rainstorm Member, respectively (Figs. 4 and 9). The lowest and highest values of δ¹³C occur in the stratigraphically highest samples, and they define a strong negative trend, beginning near the top of unit K and ending at the top of the Johnnie oolite bed. Below this marked trend in the data, there is otherwise no general trend. More than 90% of the values recorded from the bottom of unit K to the unit C carbonate are positive, averaging 1.5‰, with a standard deviation of 1.2‰. The scatter in values within individual carbonate intervals is approximately the same as variations in the average values between carbonate intervals (Fig. 7). However, the variation in δ¹³C values with stratigraphic position within each relatively thin carbonate interval does not appear to be entirely random (Fig. 10). For example, carbonates from units C and the lower part of unit K show decreasing δ¹³C values stratigraphically upward (R² = 0.74, 0.88 respectively), whereas values from the lower two carbonates in unit H and the upper carbonate interval in unit J suggest increasing δ¹³C values stratigraphically upward (R² = 0.26–0.83).

Oxygen isotope ratios range from a low of –16.0‰ (VPDB) to a high of –5.0‰, with an average value of –9.4‰ (Table S1 [^{footnote 1}]). There is no general trend in the mean values for each individual carbonate interval with stratigraphic position (Fig. 7C). The range of values within the carbonate intervals is as great as 6‰, i.e., greater than the variation of mean values for each interval. Correlation of δ¹⁸O and δ¹³C is poor for the data set as a whole (Fig. 11). Plots of δ¹⁸O versus stratigraphic position and side-by-side comparison with δ¹³C values are presented in the Supplemental Items (Fig. S2 [^{footnote 1}]). Correlation of δ¹⁸O values with stratigraphic position within each interval is also generally poor. Of 12 beds with more than three samples, R² > 0.5 only for beds Zjj1, Zji2, and Zjc (see Table 1 for nomenclature). With regard to correlation of δ¹⁸O with δ¹³C, only the carbonate in unit C shows good positive correlation (R² = 0.9), but this interval only has four data points. Intervals with 10 or more data points all show poor (R² < 0.1) intrabed correlation of δ¹⁸O with δ¹³C (Fig. S2 [^{footnote 1}]).

Our analyses focused on modeling the tectonic component of subsidence for strata in the Spring Mountains section: Johnnie unit A through the Devonian Devils Gate Formation (Table 1). The model results define the relationship between the stratigraphic thickness S, and the tectonic component of subsidence Y (Table 2), which yields a resulting curve for the function Y(S) (Fig. 12). This curve depends on parameters that describe lithification and isostatic adjustment due to sediment loading (Tables 2 and 3), and it is independent of time (Eqs. 1 and 2inSteckler and Watts, 1978). We model the time dependence of subsidence in the Discussion section below.

Our determinations of Y(S) include the effects of some 3000 m of Mississippian through Triassic overburden that lay above the Johnnie–Devils Gate interval during Jurassic and Early Cretaceous time (Giallorenzo et al., 2017). They also include two major sources of uncertainty. The first is the possible effect of a significant sedimentary substrate, predating Johnnie unit A, on the calculated tectonic subsidence. The substrate may either have been: (1) limited to the Noonday Formation or its equivalents, which are at most a few hundred meters thick and may be represented by the lowest units of the Johnnie Wash section (values Y_ns indicate “no substrate”); or (2) a thick succession of Proterozoic Pahrump Group strata (Crystal Spring through Kingston Peak Formations), which could be present at depth beneath the northwest Spring Mountains (values Y_ws indicate “with substrate”; note that in Fig. 12, Y_ws values were plotted using the base of the Spring Mountains section as a datum for zero, for a direct comparison to Y_ns). The oldest Pahrump Group strata, the Crystal Spring Formation, were ∼500 m.y. old in Ediacaran time, and therefore these models may somewhat overestimate the effect of sedimentary substrate on late Cryogenian–Ediacaran subsidence. The second major source of uncertainty lies in the resulting density of the delithified sediment column (Bond and Kominz, 1984; Bond et al., 1988). We simulated this error by varying sediment grain density by ±5%, and we note that the effect of sediment unloading is such that the lowest assumed density results in the highest tectonic component of subsidence, and vice versa. This density range yields variations in values of Y for a given S (Y_low or Y_high; Table 4) that are similar to those obtained by Bond and Kominz (1984) and Levy and Christie-Blick (1991).

The resulting plots for Y(S) (Fig. 12) show a decreasing ratio of tectonic subsidence per meter of sediment thickness, with slopes (ΔY/ΔS) ranging from values near 1.0 at the base of the section for the “no substrate” curves, to as little as 0.1 near the middle of the section for the “with substrate” curve. More typically, slopes range from 0.3 to 0.6. There is an abrupt change in slope at S ≈ 3500 m, where the section transitions from predominantly siliciclastic to predominantly carbonate sedimentation. On the no-substrate curve, the slopes defined by the five values closest to S = 3500 m are 0.5 (S < 3500 m) and 0.2 (S > 3500 m), with each of the two arrays appearing quite linear. Corresponding values on the “with substrate” curve are 0.4 and 0.1. Thus, although there is a degree of gradual curvature above and below S = 3500 m, most of the flattening of Y(S) is associated with the lithologic transition.

The effect of including a thick substrate of Pahrump Group strata is to greatly reduce our estimate of Y for any given S. In other words, by not accounting for the substrate, we overestimate the tectonic component of subsidence by 50% or more, particularly in the early phases of subsidence. Physically, the reason for this is that the no-substrate model inadvertently places incompressible basement rocks where a compacting substrate exists in the event that there is substrate, and therefore the model incorrectly assigns the compaction of the substrate to tectonic subsidence, resulting in an overestimate.

The uncertainties in Y due to sediment grain density are generally in the ±10%–15% range. In addition, to the extent that a thick sedimentary substrate is present below the lowest exposures of the Johnnie Formation in Johnnie Wash, tectonic subsidence may be overestimated by several tens of percent. Despite the sensitivity of both the density and substrate effects on the absolute value of Y, as we will discuss in the next section, the effect on estimating the age of tectonic subsidence is not large, because these estimates depend mainly on relative, not absolute values of Y. Specifically, (1) errors arising from density and substrate are correlated, such that Y(S) retains its shape even though Y may vary significantly; and (2) the exponential equation describing the time dependence of Y is defined by ratios between values of Y, rather than their absolute magnitudes.

Perhaps the most basic question in regard to the origin of the Johnnie Formation is whether the sub-oolite interval contains recognizable subunits that can be correlated across its region of exposures, and the extent to which the section contains major unconformities. These issues are best addressed through lithostratigraphic characteristics and comparisons between the Spring Mountains section and the two other major sections in the region, the Desert Range to the north and the Nopah Range to the south. A second important question is whether or not the sub-oolite (sub–Shuram excursion) interval is a chemostratigraphic correlative with the sub–Shuram excursion Khufai Formation in Oman. A third significant issue is whether continuous Johnnie Formation (and subsequent) deposition occurred through most or all of Ediacaran time, because this aspect is critical to dating the Johnnie Formation using thermal subsidence modeling. A time-dependent exponential thermal subsidence model applied to our decompacted and backstripped subsidence model, Y(S) (Fig. 12), implies continuous sedimentation along the southwest Laurentian passive margin through the whole of Ediacaran time (i.e., from basal Noonday to early Wood Canyon time, or 635–541 Ma). If such a model is correct, it provides an independent estimate of the age and duration of the Shuram excursion, and whether or not it occurred near the time of the Gaskiers glaciation.

Although lithostratigraphic correlation of sub–Rainstorm Member Johnnie Formation units is not as straightforward as for the overlying intervals, neither is it particularly complex. The two thickest sections, which both lie in Nevada, the northern part of the Johnnie outcrop belt, include the northern Spring Mountains and Desert Range sections. Both sections are readily divisible into alternating sand-rich and siltstone/carbonate-rich intervals, each on the order of 100 m to a few hundred meters thick (I, II, III in Fig. 13 on the northern Spring Mountains section). The three sand-rich intervals of the Johnnie Formation are succeeded by four additional sand-rich intervals that have long been recognized as regionally correlative units (IV–VII in Fig. 13 on the northern Spring Mountains section), of which the top three have paleontological age constraints. The two sections each contain three sand-rich intervals below the Rainstorm Member that have proportionate relative thicknesses. Further, the siltstone/carbonate-rich interval between sand-rich intervals II and III contains an ∼40-m-thick cherty dolostone unit in both sections, strengthening correlation, as noted by Stewart (1970) and Benmore (1978). We correlate sand-rich interval I of the Spring Mountains (unit B) with the Carbonate member in the Desert Range section, on the basis of stratigraphic position. We note, however, that the pervasive soft sediment deformation in unit B has not been reported from orthoquartzites in the Carbonate member, and that unit B does not contain carbonate. Unit A, which is predominantly siltstone, would therefore correlate with siltstones and oolitic limestones underneath the Carbonate member. The oolitic limestone unit at the base of the Desert Range section has been considered to be correlative with the Noonday Formation (Longwell et al., 1965; Gillett and Van Alstine, 1982), implying that unit A in the Spring Mountains may also be a Noonday correlative (Fig. 13).

The southern Nopah Range section is approximately half the thickness of the northern Spring Mountains and Desert Range sections, and it contains a number of subaerial erosion surfaces that thus far have not been observed in the thicker Nevada sections (Summa, 1993). Like the Nevada sections, however, it does contain three sub–Rainstorm Member sand-rich intervals, suggesting lithostratigraphic correlation (Fig. 13). Specifically, the lower part of the Transitional, Quartzite, and Upper carbonate-bearing members of Stewart (1970) would correspond to sand-rich intervals I, II, and III, respectively, in the Spring Mountains. The correlation is strengthened by: (1) the alternating orthoquartzite/carbonate cycles evident in sand-rich interval III in both the northern Spring Mountains and southern Nopah Range sections; (2) the lack of carbonate and abundance of high-angle cross-stratification in interval II in all three sections (Quartzite member = upper part of Lower quartzite and siltstone member = unit D, Fig. 13); (3) the consistency of unimodal, south-southwest–directed paleoflow directions in pre–Rainstorm Member orthoquartzites in the Spring Mountains and Desert Range sections (Fig. 14); and (4) the lithological similarity between the lowest sand-rich intervals in the southern Nopah Range and Desert Range sections, both of which contain a mixed carbonate-siliciclastic assemblage. Militating against these lithostratigraphic correlations are the observations that: (1) the interval III correlative in the Desert Range lacks carbonate; (2) interval I in the Spring Mountains (unit B) also lacks carbonate; and (3) the proposed Noonday substrate of interval I is lithostratigraphically dissimilar in all three sections, ranging from pale-gray quartz-rich dolomite boundstone in the Nopah Range, to phyllitic siltstone in the Spring Mountains, to medium-gray oolitic limestone in the Desert Range. Regardless of the details of these correlations, the most important facets of the two sections in Nevada are the following: (1) The sub–Rainstorm Member sections are at least twice as thick as the Nopah Range section; and (2) evidence for subaerial erosion, such as grikes, paleosols, channel scour, and desiccation cracks, which is conspicuous in the Nopah Range section, appears to be lacking. Although significant depositional hiatuses within the Nevada sections cannot be ruled out, the overall lithostratigraphic uniformity or “monotony” of these sections (siltstone and fine- to medium-grained sandstone and orthoquartzite, with sporadic thin carbonate beds) is consistent with conformable sedimentation on a stably subsiding continental shelf (Stewart, 1970; Fedo and Cooper, 2001; Schoenborn et al., 2012).

The pervasive ball-and-pillow and other paleoliquefaction structures in sand-rich interval I (unit B) are most simply interpreted as reflecting a period of high sediment flux during early Johnnie Formation deposition. These structures may have significance for the timing of the transition from mechanical stretching of the lithosphere to purely thermal subsidence, because (1) rapid subsidence is characteristic of both the rift phase and early thermal subsidence phase of passive-margin formation (e.g., Sawyer et al., 1982), and (2) such structures could be evidence for seismic shaking (e.g., Sims, 2012). The observation that essentially the entire 160 m thickness of unit B is affected implies that, whatever its cause, it was persistent over a sustained period of time. The other significant observation is that with only one exception, paleoliquefaction structures do not appear anywhere else higher in the section, despite the ubiquity of meter-scale interbeds of fine- to medium-grained sandstone overlying fine-grained sandstone or siltstone throughout the section. Thus, the cause (or causes) of soft-sediment deformation appears to be temporally restricted to, at most, sand-rich interval I and enveloping siltstone units A and C, and it presumably ended by the time of deposition of sand-rich interval II (unit D). If it is assumed that the cause is earthquakes, then sand-rich intervals I and II record a transition from frequent seismic shaking to apparent seismic quiescence. Such an interpretation is consistent with previous suggestions that the end of mechanical stretching may have occurred near the base of the Johnnie Formation (Summa, 1993; Fedo and Cooper, 2001; Schoenborn et al., 2012). A ready alternative to a seismic trigger, however, is the effect of pressure contrasts from storm waves, which have also been shown to induce liquefaction and soft sediment deformation, including ball-and-pillow structure (Alfaro et al., 2002).

A composite plot of δ¹³C values of carbonate from the Johnnie Formation in southwest Laurentia (Verdel et al., 2011; this study) yields an overall pattern that is similar to profiles in Oman that contain the Shuram excursion, including a period of positive values as high as 4‰–6‰, rapid descent to values as low as –11‰ to –12‰, and a more gradual rise back to positive values (Fig. 15). The uniformly positive δ¹³C values below the excursion in southwest Laurentia, generally of 1‰–3‰, invite detailed comparison with chemostratigraphic profiles in the carbonate-rich Khufai Formation in Oman, which lies immediately below the type Shuram excursion. The stratigraphic thickness of units between the zero crossings of the Shuram excursion in Oman and southwest Laurentia are similar, ∼500–700 m (Verdel et al., 2011). We therefore compared our profile to those from Oman without any modification to the vertical scaling (stratigraphic height), fixing the zero crossings at the base of the Shuram excursion at the same height. The Khufai sections in general are positive in δ¹³C and show considerable variation, depending on the degree of diagenetic alteration. In least-altered sections (Mukhaibah Dome area), maximum values range up to 6‰, averaging 4‰–5‰ (Fig. 16A), i.e., considerably more positive than the Johnnie Formation profile. In more-altered sections (Buah Dome area; Fig. 16B), the profiles are quite similar to that of the Johnnie Formation. Given the close correspondence between the Johnnie profile and most of the Oman profiles (Fig. S3 [^{footnote 1}]), we conclude that the data are consistent with, but do not absolutely demonstrate, temporal correlation between the upper part of the sub–Rainstorm Member Johnnie Formation (units H through L in the Spring Mountains) and the Khufai Formation.

The least-altered Khufai sections are generally considered to be representative of seawater carbon isotopic composition, defining a prolonged interval of δ¹³C values in seawater near 6‰. Therefore, it seems clear that subsequent diagenesis was primarily responsible for reducing δ¹³C values, in both Oman and the sub–Rainstorm Member Mount Schader section, by as much as 4‰–5‰. In the Pleistocene environment, such reduction has been shown to result from carbon isotopic exchange between carbonate beds and meteoric water, often resulting in δ¹³C values decreasing stratigraphically upward at the scale of a few meters in beds exposed to erosion (Allan and Matthews, 1977, 1982; Quinn, 1991; Melim et al., 1995, 2001). The strong intrabed variations in δ¹³C values in 1–2-m-thick carbonate intervals in the Johnnie Formation (Fig. 10) could potentially be explained by this mechanism, although δ¹³C values of meteoric water at that time are poorly constrained and may not have been as strongly negative as modern values. Further, the intrabed trends in δ¹³C values both increase and decrease downward, and there is no evidence of subaerial exposure on the tops of any of the beds. As with most Neoproterozoic carbonates, determining the mechanisms of depletion of δ¹³C values and their relationship to diagenetic textures and the biosphere is a difficult and controversial issue (Knauth and Kennedy, 2009; Derry, 2010a, 2010b; Grotzinger et al., 2011), and it is beyond the scope of this paper to resolve. One thing we can say, however, about the Mount Schader data set is that it displays no clear correlations between δ¹⁸O and δ¹³C (Fig. 11), as predicted by various isotopic exchange models (fig. 4 inOsburn et al., 2015). Despite this controversy, the good match between the type Johnnie Formation sub–Rainstorm Member section and the Khufai Formation supports the hypothesis that regardless of the origin of the anomalies, they nonetheless appear to be a useful correlation tool (Grotzinger et al., 2011). Tectonic reconstructions of the Neoproterozoic continent Rodinia put both the Shuram and the Johnnie formations roughly at the equator in Ediacaran time, but the two formations were located anywhere from 10,000 to 15,000 km away from each other (Li et al., 2008, 2013), making the isotopic correlation of the Shuram and sub-Shuram intervals all the more impressive.

One of the hallmarks of Neoproterozoic glacial cap carbonates is their frequent occurrence as thin, isolated intervals amid large thicknesses of enveloping strata that are entirely siliciclastic. Below unit H, there are two such isolated carbonate intervals, one in unit C and the other composing the entirety of unit F. Given their stratigraphic position between the Marinoan cap carbonate sequence and the base of the Cambrian, it is possible that either one of these units represents postglacial carbonate “rainout,” for example, as might be expected in the more southerly latitudes in the wake of the Gaskiers glaciation at 579 Ma (e.g., Pu et al., 2016). The generally positive δ¹³C values in the unit C and unit F carbonates, averaging between 1‰ and 2‰, argue strongly against either of these intervals representing a Gaskiers cap carbonate, which in Newfoundland yielded δ¹³C values of –8‰ to –2‰ (Myrow and Kaufman, 1999). Further, textural features widely described from cap carbonates (e.g., sheet cracks, tubes, teepee structures, etc.) are not observed in either of these intervals.

The substantial thickness of the Johnnie Formation, lack of evidence for unconformities in the Nevada sections, and the strengthened isotopic tie to the type Shuram excursion, motivate the hypothesis that the Noonday through lower Wood Canyon interval records continuous deposition through most or all of Ediacaran time. In the last section, backstripping and decompaction defined tectonic subsidence Y as a function of stratigraphic position S, independent of time. In this section, we model the element of time as exponential subsidence, assuming that Johnnie Formation and subsequent deposition of the passive-margin wedge occurred as a result of conductive cooling of rifted lithosphere. Subsidence analysis with well-defined ages at the Cambrian-Precambrian boundary (541 Ma) and at the base of Cambrian Age 5 (509 Ma) creates a considerably improved basis over previous studies for estimating stratigraphic age in Ediacaran strata by extrapolating the subsidence history back in time.

In the case of mid-ocean ridges, where β = ∞ and r = 1, E₀r is empirically shown to be within a few percent of 3.2 km (Parsons and Sclater, 1977). We note that this value does not correspond to the actual ridge elevation above the abyssal plain, which is much higher for oceanic crust less than 20 m.y. old. The characteristic time τ, which depends on the thermal diffusivity and thickness of equilibrium lithosphere, shows somewhat greater variation depending on the ridge (±10% for the best-constrained ridges; table 1 inParsons and Sclater, 1977), but a generally accepted range of values in subsidence analyses of passive margins is 50–65 m.y. (McKenzie, 1978; Allen and Allen, 2005). This corresponds to a “half-life” of thermal subsidence of 35–45 m.y. Even though this key parameter may vary significantly, we can estimate τ directly from our subsidence model, as an independent test of the hypothesis that the margin is in a state of exponential thermal subsidence, comparable to well-studied Mesozoic and Cenozoic examples. If our estimate of τ lies significantly outside the range of 50–65 m.y., it would falsify the thermal subsidence hypothesis.

The differencing of the two equations eliminates E₀r, and hence the most important parameters in estimating both τ and the thermal subsidence curve itself are the elevation of two points relatively well separated in time from each other, and an estimate of zero elevation, i.e., where e(∞) = 0 or the slope of Y(t) is negligible. The thermal subsidence curve is then presumably applicable back in time to whatever point in the section at which we are still confident that the margin is in a state of pure thermal subsidence. As noted above, this level is probably no higher in the section than the lower part of the Johnnie Formation, and it may be much deeper, perhaps within the upper part of the underlying Pahrump Group.

Temporal constraints on the younger part of the subsidence curve are fairly similar to those used by Levy and Christie-Blick (1991), with the exception of their two oldest points, the base of the Cambrian Age 5 (approximately Middle Cambrian, 509 Ma; Walker et al., 2013) and the base of the Paibian (approximately Upper Cambrian, 497 Ma; Walker et al., 2013), which at the time were estimated to be 540 Ma and 523 Ma, respectively. Critically for this study, both the position and age of the Ediacaran-Cambrian boundary are well defined, lying within the Lower Member of the Wood Canyon Formation with an age of 541 Ma (Corsetti and Hagadorn, 2000). The 541 and 509 Ma constraints thus function as points (e₁, t₁) and (e₂, t₂), respectively, in our initial analysis, defining an exponential subsidence curve. As the oldest reliable temporally constrained points on the curve, they are the strongest constraints on extrapolating the curve back in time.

Points younger than 509 Ma are also well dated. These points clearly postdate the Sauk marine transgression, which marks a transition from predominantly siliciclastic to carbonate sedimentation, due to flooding of the craton through middle and late Cambrian time. Associated with the transgression, the average deposition rate S(t) increases markedly from ∼20 m/m.y. from 541 to 509 Ma to ∼80 m/m.y. from 509 to 497 Ma (Table 4). Clearly, a fourfold increase in accumulation rate appears incompatible with any form of exponential subsidence. As explained below, the remarkable increase in subsidence rate owes its origin to the combination of sea-level rise and carbonate sedimentation, not renewed tectonism. The important point here is to note that the 541 and 509 Ma data points occur within the Lower Wood Canyon and Lower Carrara formations, respectively, both of which are shallow-water, mixed carbonate-siliciclastic facies associations that were probably deposited at similar points in global sea level. Both were deposited during highstand intervals relative to their transgressive substrates (the Stirling E Member and Zabriskie Formation, respectively). In the case of the Lower Member of the Wood Canyon Formation, the system evolved into a glacial drawdown of sea level (Smith et al., 2016). In the case of the lower Carrara Formation, sea level kept rising to a level that generally exceeded that of Ediacaran–early Cambrian time (Palmer, 1981).

The late subsidence history is characterized by very slow accumulation in Silurian and Early Devonian time (<3 m/m.y.; Table 4), and hence the difference between Silurian and Devonian values of Y compared to those at 541 and 509 Ma provides firm estimates of e₁ and e₂. We note that with these constraints, the precise values of time and elevation for Paibian through Upper Ordovician strata provide little additional constraint on the form of the exponential subsidence curve.

We present a temporal model of both observed subsidence S(t) (i.e., stratigraphic thickness) and tectonic subsidence Y(t) using a novel mode of presentation that orthogonally projects Y(t) and t(S) onto a graph of the numerically determined function Y(S) (Fig. 18). In this approach, Y(S) is plotted in the upper-left corner, Y(t) is plotted in the upper-right corner, t(S) is plotted in the lower-left corner, and S(t) is plotted in the lower-right corner. The plot shows a simultaneous projection of Y and S onto their respective temporal models, graphically showing the influence of the slope of Y(S) on the observed subsidence rates. The graph shows that, between 509 and 485 Ma, the increase in compressibility of the carbonate sediment [lower slope on Y(S)] combined with the accelerated schedule of subsidence caused by the flooding of the craton [higher slope on Y(t)] resulted in a dramatic increase in sediment accumulation rate [lower slope on t(S) and higher slope on S(t)], even though exponential thermal subsidence was slowly decreasing. This result is critical, because it obviates the primary reason that most previous workers have cited in favor of Cambrian rifting along western Laurentia (e.g., Bond and Kominz, 1984; Levy and Christie-Blick, 1991; Yonkee et al., 2014).

Estimates of the exponential time constant τ vary according to two main uncertainties, (1) the sediment grain density assumed in our delithification model, and (2) whether or not a thick substrate of Pahrump Group strata is present at depth beneath the exposed Spring Mountains section. We calculated values of τ for values of tectonic subsidence Y in a series of models that encompass these parameter variations (Table 5). In addition, we defined Y according to two different assumptions for the point at which mechanical stretching ends and purely thermal subsidence begins, where Y = 0 (i.e., e = E₀r). One is at the lowest exposed stratum (base of unit A), and the other is within unit C, above the youngest ball-and-pillow structure at the top of unit B, assuming seismic shaking ended near this point. In Table 5, models with no Pahrump Group substrate are designated Y_ns, and those that include the substrate are designated Y_ws; intermediate-density models contain no additional subscript, and low- and high-density models are also subscripted “low” and “high,” respectively. Models with “**” define Y = 0 within unit C, and models with no superscript assume Y = 0 at the base of unit A. We defined the value of Y for which e = 0 to be the average of Y(393 Ma) and Y(383 Ma), designated Y(c. 388) in Table 5. The results are insensitive to this choice because there is so little variation in Y between 444 and 383 Ma. We cannot choose the next younger point in the subsidence profile (359 Ma), because it clearly reflects the onset of subsidence associated with Antler foredeep sedimentation.

The contrast in τ between models Y_ns and Y_ws is only 3 m.y., with τ = 55 and 52 m.y., respectively. As expected from Equation 4, the definition point of Y = 0 has no effect, because we define e₁ and e₂ on the basis of differences in Y values late in the subsidence history. For models with no substrate, varying the density between Y_{ns, low} and Y_{ns, high} (corresponding to the assumption of high and low sediment grain density, respectively) has a substantial effect on τ, which ranges from 42 to 65 m.y., respectively. For models Y_{ws, low} to Y_{ws, high}, the sensitivity is even greater, with τ varying from 36 to 63 m.y., respectively. Clearly, the upper part of these ranges accords with subsidence patterns in Mesozoic and Cenozoic basins. Despite the nearly 30 m.y. variation in τ among these models, we note that there is relatively little variation in the modeled age and duration of the Shuram excursion (Table 5). Among this suite of models, the onset varies by 12 m.y. (from 569 to 581 Ma), the termination varies by 9 m.y. (from 566 to 575 Ma), and the duration varies by 3 m.y. (from 3 to 6 m.y.).

A further consideration in estimating τ is the fact that because the Sauk transgression was well under way by 509 Ma, relative sea level may have been slightly higher than at 541 Ma. To the extent that it was, a significant systematic error is introduced in our estimate of τ. For example, for the model with no substrate and intermediate density, a correction in Y(509) of just –50 m to account for a change in sea level (+50 m in e₂) changes the estimate of τ from 55 to 69 m.y. (Eq. 1inSteckler and Watts, 1978; Eq. 4 herein). Thus, although the suite of models used for our sensitivity analysis may suggest an estimate of 50 ± ∼15 m.y. for τ, the systematic error introduced by the Sauk transgression, and the range of values indicated by models of Mesozoic and Cenozoic basins both suggest a value toward the upper end of this range. We further note that the earliest empirical fits to long-term seafloor subsidence data suggested a value of 62.8 m.y. (Eq. 22 inParsons and Sclater, 1977; table 1 inMcKenzie, 1978).

Based on these considerations, we developed a second suite of models for estimating pre–541 Ma ages of various horizons within the Johnnie Formation. For this suite, we chose a “midrange” model using what are perhaps the simplest set of assumptions: (1) intermediate sediment grain density values; (2) negligible Pahrump Group substrate; and (3) a time constant of τ = 65 m.y. Using these assumptions, we tied the subsidence curve to the oldest dated point at 541 Ma, minimizing both the amount of extrapolation back in time, and the degree to which the data reflect sea-level rise due to the Sauk and subsequent cratonic flooding events. The resulting subsidence model for the Spring Mountains section is shown in Figure 18 and in Table 6 in the second column from the right-hand side. The remaining columns in Table 6 and Table S2 (^{footnote 1}) demonstrate the sensitivity of our resulting age estimates for the Shuram excursion and other horizons in the Johnnie and Stirling formations to variations in density, τ, and the presence or absence of a substrate. Varying only sediment grain density, limits on the Shuram excursion (tops of units Zjr1 to Zjr2) are 581–575 Ma (low-density case), 585–579 Ma (intermediate density), and 592–585 Ma (high density; Fig. S4 [^{footnote 1}]). Thus, we see that the end of the Shuram excursion varies from 575 to 585 Ma, a range that is centered on the timing of the Gaskiers glaciation. We also note that the duration of the excursion is 6–7 m.y., and it is therefore insensitive to variations in sediment grain density. With regard to sensitivity to the time constant, the end of the Shuram excursion for intermediate density values is 579, 576, and 573 Ma for τ = 65, 60, and 55 m.y., respectively (Fig. S5 [^{footnote 1}]). In terms of the error introduced by the presence of a Pahrump Group substrate (Fig. S6; Table S2 [^{footnote 1}]), for τ = 65 m.y. and intermediate values of density, the Shuram excursion occurs from 584 to 578 Ma, which is only 1 m.y. later than, and of the same duration as, the case of no substrate. The duration of the Shuram excursion across all models in this suite ranges from 4 to 7 m.y. If we exclude with-substrate models (Table S2 [^{footnote 1}]), the variation decreases to 5–7 m.y., and if we further restrict the time constant to 65 m.y., it decreases to 6–7 m.y. These estimates are consistent with recent estimates of 8–9 m.y. for the Johnnie Formation, South Australia (Wonoka), and central China (Doushantuo) sections, based on rock magnetic chronostratigraphy (Minguez et al., 2015; Minguez and Kodama, 2017; Gong et al., 2017). These estimates are all considerably shorter than the subsidence-based estimate of 50 m.y. for the Shuram excursion in Oman (Le Guerroué et al., 2006b), which has been called into question on the basis that the Khufai/Shuram interval was probably not deposited on a thermally subsiding continental shelf (Bowring et al., 2007).

In sum, because the timing of the Shuram excursion is within ∼0.5τ of 541 Ma, varying parameters in the exponential subsidence model yields variations in our estimate of age and duration of the Shuram excursion of just a few million years. The fact that a fairly broad range of parameters leads to estimates of the end of the Shuram excursion centered on 579 Ma suggests that the valleys incised into the Rainstorm Member are indeed a manifestation of the Gaskiers glaciation at equatorial latitudes. To conclude otherwise strains credulity, because Johnnie/Stirling sequence architecture is relatively uneventful for 400–500 m both above and below the Rainstorm Member (Stirling Member A/B and Johnnie units H through L, respectively). If incision was unrelated to the Gaskiers glaciation, this requires: (1) that the most dramatic stratigraphic event in the Johnnie/Stirling interval was close in time, but unrelated to, glaciation; and (2) that the Gaskiers glaciation itself had virtually no impact on the section. In essence, the subsidence analysis provides a relatively coarse estimate of age that “registers” the section with possible correlatives elsewhere. The detailed stratigraphy then fine tunes the age estimate based on a specific correlation with well-dated events elsewhere, in this case, shelf-incision and the Gaskiers glaciation.

The overall consistency of exponential subsidence models with the hypothesis that incision of the Rainstorm Member shelf is an expression of the Gaskiers glaciation suggests that modeled ages of other horizons in the Johnnie/Stirling interval may also be accurate to within a few million years. The overall accuracy of this model can be further tested by assessing how well it estimates the age of the lowermost Johnnie and Noonday interval. As noted above in our discussion of the possible correlation of unit A with the Noonday Formation, we would expect the age of this unit to be close to the age of the base of the Noonday Formation, or 635 Ma (Petterson et al., 2011). The ranges of modeled ages for the base of unit A are 639–608 Ma (Table 6), with the “midrange” model shown in Figure 18B predicting an age of 628 Ma. As shown in Figure 18D, linear extrapolation below the deepest exposed strata of unit A, assuming a linear deposition rate, would require only an additional 288 m of “subunit A” strata to bring the section to the base of the Noonday Formation and the Ediacaran Period. This thickness plus the 125 m thickness of unit A yields a total thickness of 413 m, which is consistent with maximum known thicknesses of the Noonday Formation (Petterson et al., 2011). The apparent success of exponential subsidence models in predicting the age of both the Gaskiers event and the base of the Ediacaran Period at their most likely stratigraphic levels supports the hypothesis that Ediacaran deposition on the southwest Laurentian margin was largely continuous, and that the Noonday through Wood Canyon interval in its thickest, most basinal exposures does not contain unconformities with significant depositional hiatuses.

Lithostratigraphic and chemostratigraphic details of the Johnnie Formation at its type locality in the northwest Spring Mountains of southern Nevada provide a basis for regional lithostratigraphic correlation, global chemostratigraphic correlation, and subsidence analysis of the southwest Laurentian continental margin. The regional lithostratigraphy of Ediacaran through Cambrian Age 4 strata defines seven sand-rich intervals separated by siltstone- and carbonate-rich intervals, the upper two of which are the Sauk I and Sauk II sequences of Cambrian age (Palmer, 1981). The great overall thickness of the Johnnie Formation at its type locality (∼1800 m), and the apparent absence of subaerial exposure surfaces or other evidence of erosion that are well expressed in more cratonic sections, such as the Nopah Range section, support the hypothesis of continuous deposition. Nonetheless, the lithostratigraphy is strongly cyclic at the kilometer scale, suggesting that significant hiatuses, or at least greatly reduced rates of sediment flux, may be associated with the base of each of the seven sand-rich intervals, even in the more basinal sections. We therefore caution that the true slope of the observed sediment accumulation curve S(t) is almost certainly more variable than that shown in Figure 18D, especially for the segment between 635 Ma and 541 Ma. This concern is tempered by the fact that sediment flux was sufficient in Ediacaran and Paleozoic time to fill the accommodation space to within a few meters to a few tens of meters of sea level, implying that the tectonic component of subsidence was fully recorded.

Carbon isotopic data from sub–Rainstorm Member (sub–Shuram excursion) units in the Mount Schader section are generally positive and support correlation of Johnnie Formation units H through L with the Khufai Formation in Oman, but they do not require it. If correlative, the Mount Schader section would provide the first confirmation of an extended period (represented by 300–400 m of section) of positive δ¹³C values prior to the Shuram excursion in both Oman and Nevada.

The Gaskiers glaciation marks the beginning of widespread preservation of macroscopic Ediacaran animals (Xiao et al., 2016), and the Shuram excursion is the largest known carbon isotopic excursion in the geological record. A central issue in animal evolution is thus whether or not the Shuram excursion was approximately synchronous with the Gaskiers event, because it suggests that the Shuram excursion, whatever its cause, was genetically related to creating a surface environment that could support the metabolic requirements of macroscopic animals. A second consequence of Shuram-Gaskiers correlation is that it places the transition from diverse, ornamented acritarchs to a lower-diversity, unornamented assemblage in synchronism with the appearance of macroscopic animals, rather than at some later time. The issue is addressable in southwest Laurentia, to the extent that deposition of Johnnie Formation and related strata occurred more-or-less continuously on a thermally subsiding passive margin.

Based on this assumption, subsidence analysis strongly suggests that the end of Johnnie Formation deposition, at the time of valley incision and subsequent fill with the conglomeratic member, was correlative with the Gaskiers glaciation at 579 Ma. The analysis also suggests that the onset of the Shuram excursion near the base of the Rainstorm Member occurred at ca. 585 Ma. The implied 6 m.y. duration of the Shuram excursion is consistent with paleomagnetic and other proxies from sections in Laurentia and Australia. The subsidence analysis further indicates that if the assignment of the Gaskiers event to uppermost Johnnie time is correct, then the base of the Johnnie Formation is ca. 630 Ma. If so, then the Johnnie through lower Wood Canyon interval in the Spring Mountains represents a relatively complete, 3000-m-thick section that records most or all of Ediacaran time.

We are grateful to Gillian Anderson, Leah Sabbeth, Fenfang Wu, and the late Lindsey Hedges for assistance in the field and laboratory, and to Associate Editor Christopher J. Spencer and reviewer Tony Prave for insightful and constructive reviews. This material is based upon work supported by the National Science Foundation Graduate Research Fellowship Program under grant 1144469 awarded to R. Witkosky, and grant EAR 14-51055 awarded to B. Wernicke.

Descriptions apply to geologic maps and stratigraphic columns shown in Figures 3, 4, 6, and 7.

Qa: Alluvium and colluvium in active/ephemeral channels and piedmont-forming slopes.

QTl: In Johnnie Wash, a topographically prominent ridge of coarse, poorly sorted debris, here interpreted as a landslide deposit.

Zsa: Stirling Formation, A Member (labeled “Zs” on maps). Very pale-orange, grayish-black–weathering, medium-grained orthoquartzite, laminated to massive, medium to thick bedded, with trough cross-stratification. Contains some interbedded carbonate-cemented sandstone. In places, bedding is destroyed by secondary brecciation and recementation, forming irregular dark-weathering masses. Unit forms resistant ridges relative to underlying Johnnie Formation.

Zjr: Johnnie Formation, Rainstorm Member. Includes four distinct subunits, from bottom to top: (1) green phyllitic siltstone, (2) highly resistant, ocher-colored oolitic dolostone (“Johnnie oolite,” indicated by red open-dotted line, ∼2 m thick), (3) pale-red, carbonate-cemented, fine-grained sandstone and sandy limestone, and (4) a heterogeneous upper unit that includes siltstone, carbonate-rich sandstone, flaser-bedded sandy carbonate, and intraformational limestone breccia. In the Mount Schader section, an ∼4-m-thick triad of orthoquartzite, siltstone, and dolostone immediately underlies the Stirling Formation. Orthoquartzite is affected by meter-scale ball-and-pillow structure.

Zjl: Orthoquartzite and variegated siltstone. Generally a recessive/slope-forming unit. Red dotted line on Mount Schader map indicates a resistant, laminated, brown dolomitic marker bed, ∼2 m thick.

Zjk: Orthoquartzite, variegated siltstone, and dolostone. Orthoquartzite is parallel bedded and forms erosionally resistant base; siltstone locally contains ripple laminations; resistant, brown, dolomitic marker bed, indicated by red dotted line in Mount Schader map area, is hummocky cross-stratified and contains chert in its lower portion.

Zjj: Orthoquartzite, variegated siltstone, and minor dolomitic sandstone. Orthoquartzite and siltstone occur in ∼5 m cycles. Orthoquartzite is white, resistant, and locally granular, and it contains high-angle (∼20°) cross-stratification; dolomitic sandstones are thick, brown, resistant, fine-to medium-grained beds. Red dotted line on both maps indicates a hummocky cross-stratified, brown dolomite marker bed.

Zji: Orthoquartzite, variegated siltstone, and dolomitic sandstone, parallel bedded in the Johnnie Wash section, and hummocky cross-stratified in the Mount Schader section. Orthoquartzite is fine-grained and occurs as conspicuous thick-bedded intervals in 5–10 m cycles with siltstones. Red dotted line on Mount Schader map indicates a hummocky cross-stratified, brown dolomitic sandstone marker bed.

Zjh: Orthoquartzite and variegated siltstone. Red dotted lines on Mount Schader map indicate brown, hummocky cross-stratified dolomite marker beds that contain stromatolitic mounds.

Zjg: Variegated fine-grained sandstone and siltstone, weakly cemented; also occasional interstratified orthoquartzite, fine to medium grained, medium to thick bedded.

Zjf: Dolostone with centimeter- to decimeter-thick, centimeter- to meter-long chert nodules and lenses. The dolostone forms a conspicuous pale-weathering ridge. It is finely laminated to massive. Microcrystalline varieties weather gray; coarser-grained, secondary dolomite weathers dark gray to brown.

Zje: Massive, fine-grained sandstone and laminated siltstone. Upper part contains several dozen rhythmic cycles, ∼2 m thick, of sandstone and siltstone. Each cycle has a sharp, load-casted bottom and grades upward from sandstone to siltstone. Generally an olive-hued, recessive/slope-forming unit, with occasional beds of cross-stratified sandstone, similar to unit D.

Zjd: Well-cemented, fine-grained orthoquartzite in meter-scale beds featuring cross-stratification with steep truncation angles (up to ∼30°), interstratified with medium-grained, weakly hematite-cemented ferruginous sandstone and variegated siltstone. Unit forms resistant ridge that is conspicuously darker weathering than unit E.

Zjc: Siltstone (as Zja) with a calcareous, medium-grained orthoquartzite marker bed, similar to unit D, indicated on Figure 3 by green dotted line. The lowest carbonate in the Johnnie Wash section appears near the top of this unit as a brown, medium-grained, fabric-retentive dolostone.

Zjb: Interstratified fine-grained orthoquartzite and phyllitic siltstone. Orthoquartzite occurs in meter-scale beds with parallel lamination, pervasively disrupted by soft sediment deformation, primarily ball-and-pillow structure, such that individual beds are difficult to trace along strike.

Zja: Phyllitic siltstone, with a distinct crenulation cleavage at high angle to bedding, generally a recessive unit.