Over 150 years ago, Alexander Murray used “gas-escape” structures to correlate Neoproterozoic rock units for geological surveys across Newfoundland (Murray, 1869). Murray’s “gas-escape” structures are now recognized as Aspidella terranovica, some of the oldest known occurrences of macroscopic fossils possibly related to animals in the sedimentary record (Billings, 1872; King, 1980; Benus, 1988; Gehling et al., 2000; Menon et al., 2013). Organic interpretations of additional sedimentary features in Precambrian rocks were not widely accepted until a 15-yr-old girl, Tina Negus, discovered frond-like fossils in the Charnwood Forest of England, later described and named Charnia by Trevor Ford (Ford, 1958). The realization that Upper Neoproterozoic rocks contained some of the oldest known animal-like macroscopic body fossils (Sprigg, 1947; Glaessner, 1959) eventually led to further biostratigraphic and geochronologic studies and the formal establishment of the Ediacaran Period (635–541 Ma), the first such addition to the International Geologic Time Scale in over a century (Knoll et al., 2006).

In addition to containing the oldest known fossils of macroscopic metazoans (Knoll and Carroll, 1999; Xiao and Laflamme, 2009; Droser and Gehling, 2015; Erwin, 2015; Bobrovskiy et al., 2018; Dunn et al., 2018), Ediacaran rocks document the termination of Cryogenian “snowball Earth” glaciation in the form of a basal cap carbonate (Hoffman et al., 1998; James et al., 2001; Kennedy et al., 2001; Kennedy and Christie-Blick, 2011; Ahm et al., 2019), protracted rifting of the supercontinent Rodinia and associated rift-drift transitions (Cawood et al., 2001; Cawood and Pisarevsky, 2017), the largest negative carbon isotope excursion measured in the rock record (Shuram-Wonoka excursion; Burns and Matter, 1993; Kaufman and Knoll, 1995; Grotzinger et al., 2011; Xiao and Narbonne, 2020), and possible evidence of multiple post-Cryogenian glacial episodes (Gaskiers and other unnamed events; Narbonne and Gehling, 2003; de Alvarenga et al., 2007; Chumakov, 2009; Hebert et al., 2010; Halverson and Shields-Zhou, 2011; Hoffman et al., 2012; Vernhet et al., 2012; McGee et al., 2013; Etemad-Saeed et al., 2016; Pu et al., 2016; Letsch et al., 2018; Linnemann et al., 2018; Yang et al., 2019; Chen et al., 2020). This is all followed by the appearance of almost all modern animal phyla during the Cambrian radiation (Knoll and Carroll, 1999; Valentine, 2002; Marshall, 2006; Erwin et al., 2011; Briggs, 2015).

Great strides have been made in understanding the Ediacaran and its critical transition into the Phanerozoic. Sequence stratigraphy has served as an important interpretive framework in this regard by environmentally contextualizing and facilitating the correlation of various proxy records (Sarg, 1988; Van Wagoner et al., 1988; Christie-Blick et al., 1995; Le Guerroué et al., 2006; Jiang et al., 2011; Macdonald et al., 2013; Smith et al., 2016a; Xiao et al., 2016; Shahkarami et al., 2020; Xiao and Narbonne, 2020). Although interpretations of proxies are always subject to revision and refinement, all of the substantive data used for understanding the Earth system are anchored by their physical positions within the rock record itself. The quantitative spatiotemporal distribution of rocks, their lithological compositions, and their relationships to other younger, older, and coeval rocks can also contain process signals that integrate the interactions among tectonics, climate, and biology. Thus, a quantitative understanding of the distribution and composition of rocks on continental and global scales over the duration of the entire Ediacaran has the potential to complement our understanding of Earth systems evolution by better contextualizing our sampling of Earth history and by providing additional data with which to detect and test for process signals, which are often difficult to discern at the field scale.

For example, the “Wilson cycle” of ocean basin opening and closure (Wilson, 1966), and associated supercontinent breakup and assembly, has left an unambiguous and overwhelmingly strong signal in the continental-scale, first-order (~10⁸ yr) distribution and extent of Phanerozoic sedimentary rock globally and in North America (Dewey and Spall, 1975; Vail et al., 1977; Ronov et al., 1980; Nance et al., 1988; Condie, 1998, 2000, 2004; Groves et al., 2005; Miller et al., 2005; Peters, 2006; Hawkesworth and Kemp, 2006; Cawood et al., 2013; Nance et al., 2014; Peters and Husson, 2017; Wu et al., 2017; Merdith et al., 2019). Superimposed on this first-order Phanerozoic supercontinent cycle, there are second-order (~10⁷ yr), continental-scale sequences, first identified by Larry Sloss in North America as “tectonostratigraphic” units (Sloss, 1963), which formed in response to the accretion of arcs and other marginal tectonic events (Walcott, 1972; Dott, 1983; Haq et al., 1987; Ross and Ross, 1987; Gurnis, 1992; Burgess and Gurnis, 1995; Burgess et al., 1997; Miller et al., 2005; Burgess, 2008; Haq and Schutter, 2008). Although the first-order Wilson cycle is not readily apparent at the field-scale, second-order unconformities defining Sloss sequences are readily identified in the field. One of Sloss’s key insights was that these unconformities bound “genetically related” bodies of rock, establishing a conceptual foundation for sequence stratigraphy (Sloss, 1988; Christie-Blick and Driscoll, 1995; Catuneanu et al., 2011; Dott, 2014). Superimposed on Sloss sequences, there are third-order, ~1 m.y. sequences, in some cases formed during relatively short-lived, global drops in base level, like that which occurred during Late Ordovician glaciation, and others reflecting more local, basin-scale controls on subsidence and sediment supply. So it continues, on to fourth- and fifth-order sequences, which are superimposed on all lower-order sequences and form in response to a different set of shorter-duration shifts in the balance between accommodation and sediment supply, reflecting a mixture of global and regional processes (Miall, 2016).

Macrostratigraphy (Peters, 2005, 2006, 2008; Peters and Heim, 2011b; Husson and Peters, 2017; Peters and Husson, 2017; Peters et al., 2018, 2021) was developed as an analytical framework with which to quantitatively measure the entire range of variability in the rock record, including that which is qualitatively expressed by sequence stratigraphic architecture. Decades of work have enabled the construction of a descriptive and correlative framework for Ediacaran successions in North America (Narbonne et al., 2012; Xiao and Narbonne, 2020), and here we build on this work by integrating published descriptions of Ediacaran successions into a comprehensive, macrostratigraphic data set capable of quantifying many different regional- and continental-scale properties of the known Ediacaran rock record. Our primary goal herein is to frame Ediacaran Earth systems within the variability of the rock record itself. This objective is particularly important in the case of the Ediacaran because this system in North America is restricted to the modern-day continental margins, whereas the overlying Cambrian System covers much of the continent. Thus, the relationships among Ediacaran glaciations, carbon isotope excursions, and biological evolution leading into the Cambrian explosion should be interpreted within the broader context of a geologic record that changes dramatically from the Neoproterozoic to Paleozoic.

Our general methodology followed that of macrostratigraphy (Peters, 2006; Peters et al., 2018), which uses as its fundamental currency the duration and properties of rock units that are lithologically and/or geochronologically distinct at a given location and at a given scale of resolution. Multiple physically and chronologically juxtaposed rock units comprise columns. In the case of sedimentary rocks, a column might consist of multiple conformable units defined by different lithologies that are vertically juxtaposed, thereby establishing relative chronological control. This group of contiguous and continuous sediments might then be distinguished from another such group of sediments by a “gap,” establishing two “gap-bounded” successions. In the case where all sedimentary rocks are included, macrostratigraphy provides a quantitative summarization of sequence stratigraphic architecture. However, macrostratigraphy is more flexible in defining what constitutes a gap than sequence stratigraphy, meaning it can be applied to subsets of lithologies within sequences (e.g., carbonates only, with all other rock types treated as “gaps”; Peters, 2008) and to igneous rocks (Peters et al., 2021). Macrostratigraphic quantities, like total rock area and volume, are derived by aggregating the data for the target lithology (e.g., shales, all sediments, or all plutonic rocks) and analyzing them over all columns in the focal area.

In this study, columns were specifically constructed to capture the entire Ediacaran System of North America, along with the rocks immediately bounding the Ediacaran geochronologically. These columns were compiled from rock unit descriptions and regional syntheses in journal and survey publications as well as from geologic maps. The distinction between rock units was generally made at the greatest level of detail available, but operationally the effective resolution was “lithostratigraphic” units (e.g., at some locations, units could be subdivided into relatively thin member-level designations, whereas in other areas, units were defined as extensive undifferentiated group-level units). Interformational gaps (unconformities between two distinctly separate rock units) were included when identified and reported in the literature. Intraformational gaps (unconformities identified within a single rock unit) were only included if there was a distinct lithologic or environmental change between the bounding units. Less than 10% of units, primarily in regions with structural complexity or limited exposure and borehole data, had no clear published thickness estimate. In cases where thickness was unknown, the thicknesses of rock units of similar age in the surrounding areas were used to generate a median thickness estimate. The resulting compilation of units in each column has an effective temporal resolution sufficient to resolve most second- and third order-scale changes in temporal continuity.

The boundaries of the geographic area for each column were estimated by using the combined known surface/subsurface extent of its constituent geologic units. If there was a substantive lateral change in the properties of at least one geologic unit, then a new column was generated. Substantive changes initiating the definition of a new column included significant lateral facies transitions, the disappearance or “pinching out” of major units, and the appearance of distinct geologic units. The principle of lateral continuity was applied to provide complete coverage between control areas. This assumption was made conservatively, particularly in areas with greater structural complexity. For example, columns from the east coast of North America have areas that broadly conform to regional patterns of deformation, not a palinspastic reconstruction. Polygon areas were drawn manually using the free, open-source utility geojson.io in preparation for entry into the geographic information system (GIS) component of the Macrostrat database.

Columns were grouped into paleotectonic terranes, with further subdivisions of the large Laurentian block (Fig. 1). The terranes (“West Avalonia,” Ganderia, and Carolinia) were interpreted to be volcanic arcs and microcontinents rifted from the margin of Gondwana, all of which were accreted to Laurentia’s margin during the early to mid-Paleozoic (Schofield et al., 2016; Murphy et al., 2019). Columns located in regions associated with the Neoproterozoic paleocontinents of Baltica (north-central to northwest Alaska; Fig. 1) and Amazonia (southern Central America) were also included in this data set, though they were only included in the aggregate Ediacaran analyses. We refer to western, southeastern, and northeastern Laurentia as “realms” within the same continental block. The northeast Alaska columns located in the North Slope subterrane and northwest Canada (peri-Laurentia) were included in the aggregate Ediacaran results but were not included in the individual paleotectonic terrane/realm results due to contention over their association with western Laurentian or northeastern Laurentian strata (Fig. 1; Gibson et al., 2021). The terrane and realm groupings used here reflect usage in the literature, though the tectonic affinity of northeastern Laurentia (Nunavut, Greenland, and Svalbard) also remains ambiguous (Faehnrich et al., 2019).

All 201 columns compiled for this study are shown in Figure 1 in comparison to the geographic footprint of the 212 columns that contain Ediacaran rocks in the lower-resolution Macrostrat database project for the entire rock record of North America (e.g., Peters and Gaines, 2012; Husson and Peters, 2017; Peters and Husson, 2017). The complete ‘mesostrat’ data set for the Ediacaran System of North America, compiled as described above, consists of 1357 rock units from 201 columns (Fig. 1). A complete list of the references used in this study is provided in the Supplemental Material and linked to data records in Macrostrat (see Supplement S1: Extended References).¹

After compiling basic physical properties for all units and columns, additional attributes, including geochronological and stable isotopic measurements, fossil occurrences (microscopic and macroscopic), sedimentary structures (including concretions and other distinctive features), and paleoenvironmental interpretations, were compiled and linked to the entered rock units. Published correlations for each geologic unit were also recorded, when available, and used in the refinement of age models within and between columns (see below).

All data were entered into the Macrostrat database (macrostrat.org) with a distinct project designation, making it possible to leverage the existing infrastructure for data management and dissemination (Peters et al., 2018). This system also generates an initial continuous-time age model using the constraints provided by basic superposition and correlations to chronostratigraphic time bins that are either part of, or that can be correlated to, the current International Commission on Stratigraphy time scale (ICS; Cohen et al., 2013, updated, chart version 2020/03). Changes to any of the ICS boundary ages, such as a proposed revised age of ca. 539 Ma for the Ediacaran-Cambrian boundary (Linnemann et al., 2019; Hodgin et al., 2021), automatically propagate to the relative age model for columns and to all data linked to those rock units. A more complete description of Macrostrat’s architecture, features, accessibility, and initial age model construction steps can be found in Peters et al. (2018).

Once all pertinent data were compiled and entered into Macrostrat, the initial naïve age model was modified using published regional correlations and geochronologic measurements. In total, 146 radioisotopic dates (Fig. 2; Table 1; Fig. S2 [see ^{footnote 1}]) were linked to Ediacaran units (a complete set of radioisotopic dates and the references utilized for this purpose are provided in Supplement S2: Radioisotopic Dates [see ^{footnote 1}]). Published regional correlations were used conservatively when modifying the initial age model. For example, correlations based solely on lithologic similarity, such as the oolite marker beds in the Johnnie Formation of Death Valley, California, and a similar oolite horizon in the Clemente Formation of Mexico, were not forced into time equivalence, but they may still emerge as such. We also avoided forcing time equivalence of rock units containing Ediacaran macrofossils due to their enigmatic relationships. Global correlations of Ediacaran assemblages are typically based on their interpreted affinity with Avalon, White Sea, or Nama groupings, which have questionable utility for biostratigraphy (Gehling and Droser, 2013). In this manner, we constructed an age model for the Ediacaran of North America that incorporated the basic law of superposition, with constraints on time provided by correlations to calibrated chronostratigraphic time bins, stratigraphically positioned geochronologic measurements, and correlations between key strata with time significance. The purpose of the age model used here was to provide an up-to-date working hypothesis for the correlation of all Ediacaran successions in a continuous-time representation that made as few assumptions as possible. Like all models, our age model contains inaccuracies and is subject to ongoing revision as new data and new constraints are acquired.

It is important to note that we do not assert that the data set compiled for this study is free of inaccuracies of multiple different kinds at the level of individual rock units, sequences, and columns. However, our goal herein was not to use this compilation to make definitive statements about the age of any specific unit, event, or measurement that has been linked to Ediacaran rocks and time. Instead, we analyzed the entire Ediacaran System of North America quantitatively in order to generate summary statistics that characterize the system, as a whole, in a self-consistent fashion and as we understand it today. This statistical, aggregate approach means that it is possible for the majority of our individual unit data points to be incorrect in detail but for the underlying aggregate patterns to still be accurate reflections of what would have been measured if each individual unit were perfectly resolved, so long as errors are randomly distributed among units (for an excellent example of this phenomenon in a macroevolutionary analysis of trilobite diversity, see Adrain et al., 2000). This does not, however, mean that all of the quantities we derive are straightforward to interpret. For example, many of our columns are located in regions that have been structurally deformed. This makes some derivative aggregate quantities, like area and volume, more complicated in their interpretation. However, even in regions that have been deformed, sedimentary successions have some thickness, and the areas of our columns do incorporate many structural boundaries of note. Our aggregate statistics do, we believe, well characterize the aggregate properties of the Ediacaran System in North America as it exists today and provide a basis for reconstruction (see also the description below of scaling for comparison to the comprehensive lower-resolution North American data set).

All published carbonate stable carbon (δ¹³C) and oxygen (δ¹⁸O) isotope measurements from North American Ediacaran-aged sediments were also compiled, with geographic location of the samples linking them to the compiled units, and positions of samples relative to compiled stratigraphic boundaries linking them to relative positions within units. Using these positions, each measurement could be assigned an individually precise age (and, in aggregate, accurate age in the case of randomly distributed errors) from the age model using linear interpolation between lithological boundaries serving as anchors. It is important to note that carbon isotope values, often interpreted as part of the Shuram-Wonoka excursion, were not explicitly used in the construction of the age model. However, published age interpretations are typically influenced by chemostratigraphic correlation when it is available, so it is likely that there is an indirect effect of this excursion on published correlations and, therefore, the age model used here. Compiled carbonate stable carbon (δ¹³C) and oxygen (δ¹⁸O) isotope measurements from North America Ediacaran sediments are included in the Supplemental Material (S3: Carbonate Geochemistry) and are available through the measurements route of Macrostrat’s application programming interface (https://macrostrat.org/api).

Having captured lithology, thickness, and other basic descriptive attributes of the Ediacaran rock record using the Macrostrat GIS and temporal age model system, we then computed aggregate metrics describing rock quantity as a volume flux (derived by summing over the product of thickness and column area and dividing the result by the duration of each unit intersecting each time increment). Single units assigned multiple lithologies were assigned volumes based on the approximate proportionality of lithologies within each unit. In most of the results presented here, lithologies were grouped into four main classes: (1) intrusive igneous rocks, (2) volcanic/volcaniclastic rocks, (3) siliciclastic rocks, and (4) carbonates. Each one of these classes was further subdivided into more specific lithologies (e.g., dolomite, limestone, shale, basalt). To allow more direct comparison of our new, higher-resolution Ediacaran compilation to the existing Macrostrat compilation in the Phanerozoic, column areas of the data set compiled for this study were scaled by the mean of the percent difference for each million-year time step between the two Ediacaran data sets (factor of 1.85; see Fig. S1 [see ^{footnote 1}]).

Of the 1357 rock units compiled herein, 539 have an Ediacaran age, and the remaining 818 represent the succession of overlying or underlying non-Ediacaran units. Of the 539 Ediacaran units, 428 contain sediment/metasedimentary rocks, 99 contain volcanic/volcaniclastic rocks, 160 contain intrusive igneous rocks, and 50 contain crystalline metamorphic rocks (note that one unit can have multiple lithologies). The total area covered by the 201 Ediacaran-bearing columns compiled here is ~1,409,500 km². Of these, 141 columns covering ~1,393,000 km² contain sedimentary or metasedimentary rock units, which have a total combined Ediacaran-aged volume of ~2,789,000 km³. This represents a complete accounting of all Ediacaran (94 m.y. in duration) rocks over ~17% of Earth’s continental crust (i.e., North America; Fig. 1; Table 1). By comparison, sedimentary deposits in the shorter-duration (55.6 m.y.) Cambrian System in the original Macrostrat compilation in North America cover almost eight times more area at ~10,809,000 km² and have almost four times more volume at over 9,550,000 km³, representing a total net rate of Cambrian sediment accumulation that is over three times that of the Ediacaran, even after adjusting the Ediacaran estimates herein upward to minimize differences with the original Macrostrat North America data.

Below, we first summarize results from within each of the paleotectonic blocks sampled here. Together, these regional records combine to make the aggregate quantitative summary of the Ediacaran System of North America, with which we conclude.

Avalonia comprises parts of Great Britain (not included in this study), southeastern Canada, and northeastern New England in the United States (Fig. S1D of this study for West Avalonia; fig. 2 of Schofield et al., 2016 for Great Britain’s East Avalonia). Avalonia is thought to have been located on the active subducting margin of Gondwana that underwent a transition from subduction to rifting (for various reconstructions, see figures in Murphy et al., 2019). Indeed, this terrane served as one of the foundations in the development of the Wilson cycle model (Wilson, 1966; Murphy et al., 2019; Wilson et al., 2019).

Over the past half-century, four distinct tectonic phases have been identified in Avalonia: (1) pre–main phase magmatism (ca. 760–650 Ma), (2) main-phase arc magmatism (ca. 640–570 Ma), (3) arc to platform transition (ca. 600–540 Ma), and (4) platformal (continental) succession from the latest Ediacaran through to the Early Ordovician (Murphy et al., 2019). A comparatively high area of plutonic units throughout the early Ediacaran, pulses of increased rock volume with high proportions of volcanics from 635 to 570 Ma, and an increasing proportion of siliciclastics reflect an arc to platform transition after ca. 580 Ma, consistent with current interpretations of Avalonia’s latest Neoproterozoic tectonostratigraphic history (Fig. 3A; Murphy et al., 2019; van Staal et al., 2020). These quantities reflect the signatures of the transition of the subcontinent from subduction arc magmatism to rift magmatism as it and other terranes (i.e., Carolinia, Ganderia, Meguma) were rifted from Gondwana, with a period of transtensional displacement. Full separation of Avalonia from Gondwana did not occur until as late as ca. 490 Ma (Bevier et al., 1993; Mancuso et al., 1996; O’Brien et al., 1996; Pe-Piper et al., 1996; Murphy et al., 1997; Currie and McNicoll, 1999; Hatcher, 2010; Franke et al., 2017). The Avalonian terrane was accreted to Laurentia during the Early Ordovician to Early Devonian Appalachian-Caledonide orogeny (Domeier, 2016; Waldron et al., 2019; van Staal et al., 2020). The terrane and entire marine platform of the Laurentian margin, including Avalonia, was subsequently deformed during the Pangea-forming late Carboniferous Variscan orogeny (Winchester et al., 2002; Smit et al., 2018), and again by the breakup of Pangea and the opening of the Atlantic basin, which resulted in the separation of the terrane into east and west subterranes.

Carbonates are not a significant component of sediments in Avalonia, likely due to the large supply of clastic sediment available to fill available accommodation. This has made it difficult to place the Shuram-Wonoka excursion within Avalonia’s well-constrained geochronology (Myrow and Kaufman, 1999), though measurements of carbonate material in siliciclastic rocks have been used to identify and constrain its occurrence (Canfield et al., 2020). The Gaskiers Formation (Newfoundland, Canada), which is the primary evidence of the Gaskiers glaciation between 580 and 579 Ma (Pu et al., 2016), is found within the Avalonia terrane. However, the Gaskiers glaciation does not appear to be associated with any pronounced expression in rock quantity or type (Fig. 3A), perhaps because Avalonia was a very active portion of Gondwana during the Ediacaran in comparison to its more proximal counterparts in Ganderia and Carolinia. The development of Avalonian platformal successions occurred from the late Ediacaran to the Early Ordovician in extensional and transtensional basins, in which the distinctive Avalonian fossil assemblages, such as the Mistaken Point fauna (Newfoundland) or the Charnwood Forest fauna (Great Britain), have been documented (Landing, 1996; O’Brien et al., 1996; Liu et al., 2015; Murphy et al., 2019). The occurrences of Avalonian fossil assemblages coincide with the arc to platform transition, which is marked by a decrease in the volume of volcanics and a subsequent rise in siliciclastics (Fig. 3A). Overall, macrostratigraphic patterns in Avalonia, and the Gondwanan terranes preserved in North America, reflect primarily proximal tectonic events, with little or no superimposed global signal (Figs. 3A and 3D).

Ganderia and Carolinia have similar Ediacaran tectonic histories, though they are distinct tectonostratigraphic terranes. Ganderia currently encompasses strata exposed in Great Britain (not included in this study), southeastern Canada, and New England in the United States. Carolinia is currently located in the southeastern United States (Fig. S1; van Staal et al., 2012). Both Ganderia and Carolinia were in close proximity as a part of the Gondwanan margin during the Ediacaran, were rifted from Gondwana in the early Paleozoic, and were accreted to the Laurentian margin during the Taconic orogeny and the closing of the Iapetus Ocean (Murphy et al., 2004; van Staal et al., 2012). Avalonia, Carolinia, and Ganderia were all deformed and metamorphosed by orogenic events associated with the construction of Pangea, though Carolinia and Ganderia were located on the Taconic orogen collision front with Laurentia. Similar to Avalonia, Ganderia was separated into an east and west component by Pangea rifting and the formation of the Atlantic Ocean, whereas Carolinia has only been identified in the southeastern United States (van Staal et al., 2012).

Ganderia and Carolinia have similar macrostratigraphic patterns, though they differ in overall lithologic expression. Ganderia exhibits little preserved rock volume with a volcanic-dominated pulse from ca. 560 to 550 Ma (Fig. 3B). Similarly, Carolinia has little preserved rock volume besides a siliciclastic pulse from ca. 557 to 541 Ma (Fig. 3C). Ganderia’s rock volume flux is dominated by volcanics and coincides with an increase in plutonic unit area during the latter half of the Ediacaran. Ganderia’s dominantly igneous Ediacaran record lacks significant sediment volume, due either to unroofing or, perhaps more likely, to an original lack of accommodation space. By contrast, Carolinia’s dominantly siliciclastic pulse of preserved rock volume (ca. 557 Ma) suggests that accommodation within the terrane was created by renewed tectonic activity. Ganderia lacks Ediacaran fossil assemblages, due to either a lack of suitable environments for the recruitment of organisms or conditions favorable to preservation. Carolinia strata contain Ediacaran macrofossils in the terminal Ediacaran (Cid and Floyd Church Formations, North Carolina), when preserved rock volume is at a maximum (Fig. 4B; Weaver et al., 2006). As with Avalonia, low total rock volume and a volume flux dominated by proximal tectonic activity in Ganderia and Carolinia overprint any potential quantitative signature of Gaskiers glaciation. Despite relatively limited preserved rock volume from the Ediacaran of Avalonia, Ganderia, and Carolinia, the stratigraphic records of these terranes contribute to our understanding of the active volcanic margin of Gondwana throughout the Ediacaran, a condition which led to a rock record in these regions that is dominated by local tectonic activity (Figs. 3B–3D).

The Ediacaran of western Laurentia (W Laurentia) is dominated by the Windermere Supergroup and its correlatives, which stretch from the Sonora region of northwest Mexico to northern Alaska (Fig. 1C; Ross, 1991; Yonkee et al., 2014). Ediacaran sediments of the Mackenzie, Wernecke, and Ogilvie mountains in northwestern Canada, Death Valley (California), and the Sonora region of Mexico preserve multiple occurrences of macrofossil assemblages and signatures of the Shuram-Wonoka carbon isotope anomaly (Corsetti and Kaufman, 1994; Narbonne et al., 1994; McMenamin, 1996; Corsetti and Hagadorn, 2000; James et al., 2001; Corsetti and Kaufman, 2003; Hurtgen et al., 2004; Sour-Tovar et al., 2007; Bergmann et al., 2011; Macdonald and Cohen, 2011; Petterson et al., 2011; Verdel et al., 2011; Loyd et al., 2012; Macdonald et al., 2013; Smith et al., 2016b; Smith et al., 2017; Eyster et al., 2018; Witkosky and Wernicke, 2018; Moynihan et al., 2019; Hodgin et al., 2021). Neoproterozoic through Lower Cambrian strata of W Laurentia record signals of protracted Rodinia rifting and the development of an early Paleozoic passive margin (Ross, 1991; Cecile et al., 1997; Lund, 2008; Yonkee et al., 2014; Moynihan et al., 2019). However, the overall tectonic setting of W Laurentia during the Cryogenian through early Cambrian is a matter of ongoing debate, due in part to a relative lack of direct geochronologic controls in this volcanic-poor tectonic realm (Figs. 2 and 4A). Uncertainty has largely involved the timing of regional Rodinia rifting, whether there was one or two rift-to-drift transitions from the latest Neoproterozoic to the middle Cambrian, and the timing of the development of a passive margin (Stewart, 1972; Christie-Blick and Levy, 1989; Ross, 1991; Dalrymple and Narbonne, 1996; Colpron et al., 2002; Ross and Arnott, 2007; Post and Long, 2008; Macdonald et al., 2012; Yonkee et al., 2014; Strauss et al., 2015; Moynihan et al., 2019; Brennan et al., 2020, 2021).

Lower Ediacaran (635–585 Ma) W Laurentia rock volume flux is characterized by a comparatively low volume of rock dominated by mud-size sediments (Fig. 4A). There is an increase in the volume of sediments and a shift to sand- and carbonate-dominated composition at ca. 585 Ma. The uppermost Ediacaran (ca. 555–541 Ma) in W Laurentia exhibits relatively low preserved rock volume dominated by sand-size sediments and a terminal Ediacaran low that is consistent with an extensive hiatus obscuring the Ediacaran-Cambrian boundary in this tectonic realm (Yonkee et al., 2014). Compared to the other terranes and realms that make up the Ediacaran of North America, carbonates make up a high proportion of W Laurentia rock volume flux. Volcaniclastic sediments and igneous rocks are rare in W Laurentia sequences, though there is an increase in the area of plutonic units during the latter half of the Ediacaran. A significant carbonate volume increase in the Middle Ediacaran of W Laurentia at ca. 585 Ma signifies increasing accommodation along the continental margin (Fig. 4A). Increased accommodation would likely have been caused by subsidence-driven transgression, perhaps indicating continental separation along at least a portion of the W Laurentia margin during the mid- to late Ediacaran.

Of the tectonic realms/terranes discussed in this study, W Laurentia strata likely have the greatest potential to preserve signals of global geologic processes (such as that of glaciation) due to their overall coverage and a relative lack of proximal tectonic overprints. While no unambiguously glacially influenced sediments of Ediacaran age are known from W Laurentia, there is a decrease in rock volume that coincides with the ca. 580 Ma Gaskiers glaciation (Fig. 4A). This could indicate global glacioeustatic influence on the accumulation of the sedimentary record at this time. It is likely that any ice sheets present during deposition of the Gaskiers Formation were centered on continents other than Laurentia, much like Gondwanan-focused ice sheets during the late Paleozoic ice age (Montañez and Poulsen, 2013; Kent and Muttoni, 2020). Deposition of rock units bearing Ediacaran macrofossils in W Laurentia coincided with an overall sustained rise in rock volume and an increase in carbonate volume from ca. 575 to 550 Ma; macrofossil-bearing rock units terminated during decreased rock volume ca. 550 Ma (Fig. 4A). Increases in carbonate volume ca. 588, 580, 565, and 553 Ma likely indicate marine transgression of W Laurentia.

Ediacaran strata of the southeastern Laurentian (SE Laurentia) realm form a chain of exposures currently stretching from eastern Labrador, Canada, to the north-central portion of Georgia, southeastern United States (Fig. 1C). Stratigraphy of SE Laurentia is generally split by its encompassing physiographic provinces (i.e., Blue-Ridge and Piedmont), or it is simply referred to as “Laurentia” when discussed in relation to peri-Gondwanan terranes such as Avalonia (Hibbard et al., 2007; Pollock et al., 2009). Neoproterozoic–Paleozoic successions of SE Laurentia typically contain metasediments and have fewer geochronologic constraints due to deformation during tectonic events throughout Pangea’s assembly and rifting (Rankin, 1975; Thomas, 1991; Hatcher, 2010; Bailey et al., 2017; Waldron et al., 2019; van Staal et al., 2020). SE Laurentia’s quantified Ediacaran rock record predominantly reflects mid-Ediacaran proximal tectonic activation of a passive Laurentian margin with low preserved rock volume before 590 Ma (Fig. 4B). After 590 Ma, rock volume continuously increased, a trend that proceeded across the Ediacaran-Cambrian boundary. Volcanics, sand-size sediments, and greater than pebble-size sediments dominated during the initial rise in volume flux but gave way to an increasing proportion of mud-size sediments, signaling a transition from proximal sediment sources to more distal sediment sources in the late Ediacaran. Similar to W Laurentia, the Neoproterozoic SE Laurentian margin was associated with Rodinia rifting and a rift-to-drift transition that is thought to have occurred sometime in the early Cambrian (Hatcher et al., 2007; Hatcher, 2010). In this context, our results suggest an initiation of SE Laurentia rifting in the mid-Ediacaran. Unlike the W Laurentian realm, SE Laurentia does not exhibit a decrease in rock volume at the terminal Ediacaran, and increased carbonate flux suggests that there was increasing accommodation, perhaps due to subsidence of the continental margin after initial rifting. Like Ganderia, SE Laurentia lacks Ediacaran fossil assemblages, due to either a lack of suitable environments for the recruitment of organisms or conditions favorable to preservation.

The Fauquier Formation (Virginia) contains carbonates bearing a carbon isotope excursion that has been correlated with the Shuram-Wonoka anomaly and interpreted as a cap carbonate overlying a diamictite horizon, potentially of Gaskiers age, or perhaps related to a separate, younger glaciation (Hebert et al., 2010). However, there is little evidence of Ediacaran glaciated sediments in other SE Laurentia successions, and any quantitative signature of glaciation that might have been present has been overwritten by proximal tectonic activity (Fig. 4B).

Northeastern Laurentia (NE Laurentia) is a composite realm of Ediacaran strata from northeastern Nunavut (Canadian Arctic Islands) to northern and eastern Greenland and Svalbard (Figs. 1A and 1B). Detailed studies of the tectonic affinities of NE Laurentia Neoproterozoic rocks are limited due to difficult field accessibility, a lack of outcrops (Harrison, 1995; Dewing et al., 2004; Beranek et al., 2013), few geochronologic constraints (Fig. 2; Beranek et al., 2013; Willman et al., 2020), and deformation during both the Caledonian orogeny (Ordovician–Early Devonian) and Eocene (Beranek et al., 2013). Exposed ancient strata from Ellesmere Island of Nunavut to northeastern Greenland are known colloquially as the Franklinian Basin, and eastern Greenland to Svalbard Neoproterozoic strata are interpreted as Caledonian fragments. These strata are thought to represent a passive margin during the Ediacaran and Cambrian that formed from Rodinia rift-related Cryogenian (ca. 720 Ma) continental flood basalts (Dewing et al., 2004; Denyszyn et al., 2009; Bédard et al., 2012; Beranek et al., 2013). A comparatively low rock volume and high proportion of carbonates support interpretations of NE Laurentia as a passive margin during the Ediacaran, though increased rock volume with a high proportion of larger than pebble-size sediments at ca. 585–563 Ma suggests proximal topographic relief (Fig. 4C). The Shuram-Wonoka excursion has recently been identified in carbonates of northern Greenland in addition to Doushantuo-like phosphatized microfossils (Willman et al., 2020). While no definitive conclusions can be drawn with so few geochronologic constraints, there appears to be no quantitative signature of glaciation associated with the Gaskiers interval in NE Laurentia.

Median duration (m.y.), thickness (m), and simple sedimentary rates (m/m.y.) of unconformity-bounded intervals of rock in this study and from the comprehensive North American Macrostrat database are compared in Figure 5. One of the differences in this Ediacaran-centric data set is an overall decrease in the median duration and thickness estimates for Ediacaran strata, revisions that have made this compilation of the Ediacaran more comparable to the Macrostrat Phanerozoic data. This reflects an overall increase in the resolution of our new compilation and the many recent advances that have been made in Ediacaran correlation and geochronology (Figs. 5A and 5B). The sharp decrease in the median duration of unconformity-bounded intervals ca. 585 Ma is generally consistent with increasing geochronologic and biostratigraphic control in Upper Ediacaran strata (Fig. 5A). However, median thickness also increases after a nadir in the mid-Ediacaran, leading to increasing median sedimentation rate in rock units above the Gaskiers horizon (Fig. 5).

In both this study’s data set and the original Macrostrat data set, the unconformity-bounded intervals and area time series are in broad agreement, although there is an increase in the number of unconformity-bounded intervals in the Upper Ediacaran in the new compilation that is not reflected by area (Fig. S1). This is due to the influence of the tectonically complex Gondwanan terranes, where common ashes, tuffs, and volcanics in Ediacaran rocks have been used to subdivide sections into shorter-duration units, though the areal extent of the columns in that region are comparatively small (Figs. 1–2; Fig. S2).

In aggregate, rock volume flux shows a Lower Ediacaran (ca. 623 Ma) increase after the termination of snowball Earth glaciations, a fall and sustained low through to the Middle Ediacaran, a sustained increase from ca. 585 to 544 Ma, and a more rapid decrease at 544 Ma that results in an Upper Ediacaran low comparable to the Lower Ediacaran (Fig. 6D). At the largest of scales, this macrostratigraphic pattern is likely the quantitative expression of the rifting of Rodinia and a transition to a continental drift state, setting the stage for Paleozoic transgression and sediment deposition in continental interiors. The initial increase of sedimentary rock volume in the North America combined data set (Fig. 6D) at ca. 585 Ma coincides with increased volcanic/volcaniclastic rock volume until the sharp decrease in overall rock volume at ca. 544 Ma (Fig. 6D), reflecting to a large extent the contribution of terranes later accreted to Laurentia and the southeast margin of the continent (Figs. 3 and 4). Mud-size sediment dominance in the Lower Ediacaran may indicate more distal sediment sources, perhaps due to a dearth of proximal topography after Cryogenian glaciation, a lack of accommodation in landward settings and subsequent erosional loss of proximal sediments, overall net progradation, or some combination of these factors (Fig. 6D). An increasing proportion of nonmarine rock volume in the Middle-Upper Ediacaran (Fig. 7) suggests a role for accommodation limitation on Laurentia overall. Higher proportions of sand and larger than pebble-size sediments in addition to increasing nonmarine deposition throughout the mid-Ediacaran (ca. 590–575 Ma) suggest more proximal sediment-generating sources, possibly associated with tectonic uplift in the southeast part of the continent (Figs. 6D and 7). Carbonate volume pulses, most prominently developed in western Laurentia, indicate decreases in siliciclastic input, likely caused by subsidence-induced marine transgression. Notably, the cap carbonate phase of the lowermost Ediacaran does not exhibit a marked increase of carbonate volume or area, though there is an increased proportion of carbonate volume (Figs. 6D and 8A–8C). This may reflect a lack of accommodation space in continental settings, the destruction of Lower Ediacaran depositional sequences, or a combination of these factors. The terminal Ediacaran (ca. 544–541 Ma) records a sharp drop in preserved rock volume (Fig. 6D) and a return to Lower and Middle Ediacaran-like ratios of carbonate to clastic sediments, suggesting filling of available accommodation space and possibly a minor drop in base level.

The number of units that contain macrofossils increases with increasing rock volume at ca. 580 Ma before undergoing a decrease that coincides with the terminal Ediacaran drop in preserved rock volume at ca. 544 Ma (Figs. 6C and 6D). While direct geochronologic constraints on Ediacaran fossiliferous units are rare (Xiao and Narbonne, 2020), units bearing the earliest Ediacaran-type macrofossils within regions (e.g., Avalonia and W Laurentia) are consistently correlated with units bearing the Shuram-Wonoka carbon isotope excursion above the Gaskiers horizon, as reflected in the age model presented in this study (Fig. 6; Macdonald et al., 2013).

In total, 5410 individual stable carbon (and oxygen) isotopic measurements from nine regions were compiled from North America, representing what we believe to be a complete accounting of published results (Figs. 1 and 6A). For reference, schematics of key stratigraphic columns are plotted on the same time series as the carbon isotope values (Fig. 6B). The δ¹³C measurements from the “cap carbonate” at the base of the Ediacaran vary significantly, ranging from ~+9‰ to −9‰. The Lower to Middle Ediacaran section has few measurements due in part to lower volumes of carbonate, but it is evident that average δ¹³C measurements tend to climb, with some locations reaching as high as ~11‰ by the middle Ediacaran (Fig. 6A). A decrease to a minimum of almost −14‰ from a preceding period-high in δ¹³C marks the Shuram-Wonoka anomaly in North America. Interestingly, the anomaly coincides with an increase in total rock volume, an increase in the proportion of volcanics, the greatest increase in carbonate volume observed during the Ediacaran, an increase in stratigraphic units containing macrofossil occurrences, and an increase in the proportion of carbonates identified as limestone. A subsequent smaller but otherwise similar pulse in the volume and composition of carbonates at ca. 565 Ma is not associated with a similar coordinated drop in carbon isotope values, though some locations do exhibit negative excursions associated with this carbonate pulse.

The timing of the Shuram-Wonoka excursion that emerged in this analysis generally agrees with current constraints (fig. 4 of Rooney et al., 2020). The δ¹³C value recovers and “stabilizes” to generally more positive values by ca. 568 Ma, until another less-well-developed negative excursion (possible prelude to the basal Cambrian isotope excursion) begins during the terminal Ediacaran (Fig. 6A). The δ¹⁸O values measured from the same samples were also compiled, and there is little discernible temporal trend in the aggregate of these measurements (Fig. S3).

It is conceivable that not using carbon isotope stratigraphy to directly inform our correlations prior to constructing the age model reduced the precision of our model and, therefore, spuriously increased the range of temporal and spatial variability that is observed in δ¹³C (Fig. 6A). However, given the large degree of δ¹³C variability observed throughout the entire Ediacaran and from location to location across key intervals, we suspect that our synthesis is a reasonably accurate reflection of variability in δ¹³C at this scale.

Area and volume flux for the Ediacaran compilation (measured in km² and km³/m.y., respectively) are plotted adjacent to Ediacaran-normalized Phanerozoic Macrostrat sediment flux values in Figure 8. In keeping with previous results, which identified a large, stepwise increase in surviving sediment quantity from the Neoproterozoic to the Cambrian (Ronov et al., 1980; Peters and Gaines, 2012; Peters and Husson, 2017; Keller et al., 2019), this study found that Ediacaran rock volume is much diminished relative to comparable intervals in most of the Phanerozoic. We did identify a rise in preserved rock volume in the Upper Ediacaran, but the large increase to Paleozoic-like quantities takes place during the Cambrian and after a terminal Ediacaran decline (Figs. 8A and 8B). The composition of Ediacaran sediments further distinguishes them from the succeeding Paleozoic sediments, with the Ediacaran sediments exhibiting increased proportions of volcanics and decreased proportions of carbonates (Fig. 8C). This overall motif of low rock quantity and relative proportions of lithologies is not repeated until the late Permian through Triassic (Fig. 8).

The total rock volume flux of Ediacaran sediments from North America does not show a marked volumetric or lithologic shift associated with the Gaskiers glaciation (Figs. 6D and 8A), with the exception of a minor decrease of rock volume before ca. 580 Ma in W Laurentia (Fig. 4A). This could be a potential signature of glacioeustatic sea-level fall. However, the decrease in rock volume has a relatively short duration and could be more reflective of a general lack of geochronologic constraints for the Ediacaran at this time. Decreases in median thicknesses and accumulation rates of unconformity-bounded intervals do coincide with the timing of the Gaskiers glaciation (Figs. 5B and 5C), and there is a modest coarsening of the siliciclastic component (Fig. 6D). The timing of the Gaskiers glaciation coincides with early stages of the Shuram-Wonoka excursion, but the nature of its relationship to this anomaly is unknown.

The last section of the Ediacaran chapter in the Geologic Time Scale 2012 asks the question, “Ediacaran—Last period of the Proterozoic or first period of the Phanerozoic?” (Narbonne et al., 2012, p. 429 b). The large increases in rock volume and area that occur across the Ediacaran-Cambrian boundary and the changing lithological composition may provide a rock record–grounded basis for answering that question (Fig. 8), at least in North America. Ediacaran rock volume is decisively low, and its composition is quite different in comparison to the Cambrian and remainder of the Paleozoic (Fig. 8B). Ediacaran patterns of preserved rock volume and lithology do, however, bear some resemblance to the uppermost Permian through Lower Mesozoic (Figs. 8B and 8C). Thus, from the point of view of macrostratigraphy, the interpretation of the Ediacaran as the terminal period in a transition between eras is appealing.

The Wilson cycle of opening/closing ocean basins and supercontinent breakup and formation exerted a first-order control on Phanerozoic continental sedimentation patterns, and that is evident here, as it is in all other compilations of aggregate rock properties at Phanerozoic and longer time scales (Wilson, 1966; Ronov et al., 1980; Cloetingh et al., 1984; Peters, 2005, 2006; Meyers and Peters, 2011; Peters and Husson, 2017; Zaffos et al., 2017; Husson and Peters, 2018). Broadly speaking, the opening of ocean basins results in sediment accumulation on continental margins for several reasons, one of which is because it enables the thermal subsidence of rifted margins. Signals of at least regional subsidence are discernible throughout the Phanerozoic macrostratigraphy of North America. For example, Cretaceous strata of North America exhibit increased area associated with transgression, but it is also associated with a more pronounced rock volume increase (Figs. 8A and 8B), a reflection of the enhanced subsidence in western North America occurring in response to crustal flexure and the effects of the subducting Farallon slab and spreading center subduction on dynamic topography (Gurnis, 1993; DeCelles, 2004; Liu et al., 2010). The Carboniferous exhibits a similar signature of increased rock volume flux with a lesser magnitude area increase (Figs. 8A and 8B), representing a signature of crustal flexure along the ancient Appalachian mountains during Pangea’s assembly (Wilson, 1966; Nance et al., 2008; van Staal et al., 2009; Hibbard et al., 2010; van Staal et al., 2012). In contrast, Cambrian and Ordovician strata have a significantly increased area, reaching an all-time high in the Phanerozoic, but that maximum is not associated with a high in rock volume flux (Figs. 8A and 8B). This is because whatever induced the transgression that flooded most, if not all, of Laurentia in the Cambrian, and even more so during the Ordovician, left behind a relatively thin, widespread, carbonate-rich veneer of sediment, presumably because of a lack of subsidence (and/or sediment supply).

The Upper Ediacaran (ca. 590–541 Ma) stratigraphic record bears some resemblance in pattern and duration to Phanerozoic Sloss sequences (Meyers and Peters, 2011). The muted increase of total column area in the Upper Ediacaran compared to the coincident increase in volume flux (Figs. 8A and 8B) may be indicative of margin subsidence–induced accommodation formation, with comparatively limited flooding of the continent. Based on the overall similarity in pattern and time scale, and a possible link to marginal continental dynamics, we suggest that the Ediacaran rock record contains a Sloss sequence–like signal comparable to the macrostratigraphy-redefined Sloss sequences of Meyers and Peters (2011). For convenience, we refer to the Ediacaran sequence as the “Mackenzie sequence” (Fig. 8A; named after the Ediacaran carbonates of the Mackenzie Mountains in NW Canada, where the rise and fall in rock quantity and carbonate fractions thereof are well expressed).

From the point of view of macrostratigraphic quantities describing the physical, spatiotemporal variability of the rock record, the Ediacaran System is well placed at the boundary with the Sauk sequence and as a transition between the Neoproterozoic and Phanerozoic Eons (Landing, 1994). The closest quantitative analogue to the Ediacaran System in North America, from the point of view of macrostratigraphic quantities describing rock quantity and composition, and likely overall tectonic state, is the Triassic (Fig. 8).

The Shuram-Wonoka anomaly is the largest-magnitude marine carbonate δ¹³C excursion occurring in a series of perturbations observed during the Neoproterozoic (ca. 720 Ma Garvellach and ca. 650 Trezona anomalies) through the Ediacaran–Cambrian (basal Cambrian isotope excursion) and extending into the Lower Paleozoic (Cramer and Jarvis, 2020). It has been proposed that a δ¹³C gradient existed between shallow-marine shelf environments and deeper-marine settings along Ediacaran shelf-to-slope profiles of continental margins (Li et al., 2010; Macdonald et al., 2013; Schrag et al., 2013; Cui et al., 2015, 2017; Shields et al., 2019; Husson et al., 2020; Laakso and Schrag, 2020; Li et al., 2020). A globally synchronous migration of a δ¹³C gradient may have been a key mechanism behind the initial positive excursion and subsequent negative drop that together are a signature of the Shuram-Wonoka anomaly in marine carbonates. Global marine transgression during the Shuram-Wonoka excursion, evidenced here by a very large (more than quadrupling) and relatively abrupt (~10 m.y.) increase in carbonate volume and area (Figs. 4A, 6A, 6D, and 8A), is consistent with sequence stratigraphic interpretations (Christie-Blick et al., 1995; Le Guerroué et al., 2006; Macdonald et al., 2013; Cochrane et al., 2019). Thus, our data support the broad expansion of carbonate shelf area, likely caused by marine transgression, as an environmental correlative of the Shuram-Wonoka anomaly.

Macroevolutionary trends have long been recognized as correlating with preserved sedimentary rock volume throughout geologic time (Raup, 1972, 1976). A matter of continuing debate is whether the correlation between biodiversity and preserved rock quantity is evidence for preservation bias that has distorted patterns in the fossil record or is instead a signal of geological processes exerting direct or indirect controls on the timing of biological evolution and physical environmental changes that affect the pattern of sedimentation (Newell, 1959). The latter is known as the “common-cause” hypothesis (Crampton et al., 2003; Peters, 2005; Peters and Heim, 2011a; Peters et al., 2013; Holland, 2017; Husson and Peters, 2018; Nawrot et al., 2018). The appearances and disappearances of Ediacaran-type macrofossils are subject to this same debate (Seilacher, 1984; Amthor et al., 2003; Johnston et al., 2012; Laflamme et al., 2013; Cohen and Macdonald, 2015; Darroch et al., 2015; Sperling et al., 2016; Bowyer et al., 2017; Darroch et al., 2018; Tarhan et al., 2018; Gehling et al., 2019), particularly given their appearance around the Gaskiers horizon (Pu et al., 2016; Xiao and Narbonne, 2020) and the ascending limb of Ediacaran marine rock quantity, as well as given their disappearance during the terminal Ediacaran decline in rock quantity (Figs. 6C and 6D).

Enigmatic taxonomy and the overall rarity of fossil occurrences make it particularly difficult to measure biological diversity in the Neoproterozoic. However, Ediacaran eukaryotes have been shown to have increased within-assemblage diversity compared to the preceding Cryogenian and Tonian (Cohen and Macdonald, 2015). Counts of geologic units that contain Ediacaran macrofossils in North America serve as a coarse proxy for either the number of depositional environments capable of supporting the earliest known assemblages and/or our sampling thereof (Figs. 6B and 6C). Interestingly, units that contain the oldest known Ediacaran macrofossils are correlative with periods of increased preserved rock volume in their respective terranes/realms (Figs. 3A, 4A, and 4B). In aggregate, units that contain Ediacaran macrofossils have initiations correlative with the onset of marine transgression at ca. 585 Ma (Fig. 6). Subsidence-driven marine transgression could have functioned as a common-cause mechanism, driving accommodation-based expansion of preserved rock volume and increased habitable ecospace for the oldest known Ediacaran macroscopic organisms. Increased burial and long-term sequestration of carbon associated with continental flooding in W Laurentia (Fig. 4A; also evidenced by the global presence of carbonates from which the Shuram-Wonoka excursion has been identified around this time) may have contributed to increased pO₂ during the late Ediacaran (Canfield et al., 2007; Sahoo et al., 2012; Ling et al., 2013; Partin et al., 2013; Sperling et al., 2013, 2015b; Lee et al., 2016; Husson and Peters, 2017; Wang et al., 2018; Williams et al., 2019), possibly contributing to the development of more complex macroscopic organisms (Sperling et al., 2013; Mills et al., 2014; Sperling et al., 2015a; Zhang and Cui, 2016; Yang et al., 2018). Alternatively, the coinciding increase in rock units with Ediacaran macrofossils and increased rock volume could simply reflect a shift in the extent of facies suitable for fossil preservation and sampling. The disappearance of Ediacaran macrofossils in North America coincides with decreases in preserved rock volume (Figs. 3A, 4A, 6C, and 6D), perhaps indicating a regression-based influence on both preserved rock volume and available habitable ecospace or associated environmental shifts that resulted in a decrease in the extent and abundance of shelf life. Alternatively, it could reflect a facies control on sampling bias or destruction of terminal Ediacaran sediment cover during extensive Cambrian transgression (Peters and Gaines, 2012; Shahkarami et al., 2020). More detailed data on assemblage-level biodiversity (similar to data presented by Cohen and Macdonald [2015], but with higher resolution for the Ediacaran) and more detailed stratigraphic and environmental context for those fossils and the surrounding unfossiliferous successions will yield further insight into the common-cause hypothesis, which finds some support in the convergence of quantitative stratigraphic, geochemical, and biological patterns documented here (Fig. 6).

Ediacaran-age sediments are restricted to the margins of modern North America and were even more marginally located in the context of Laurentia (Fig. 1). Cambrian–Ordovician sediments, by contrast, extend across most of Laurentia’s continental area today (Peters and Gaines, 2012; Husson and Peters, 2017; Peters and Husson, 2017) and probably originally covered essentially all of the continent. This difference in sedimentary cover distribution constitutes the most important and distinguishing feature of the Great Unconformity. Multiple events spanning a long period of time contributed to the crustal-scale signatures of continental denudation that are evident among most of the highly heterogeneous rocks beneath the Great Unconformity (Flowers et al., 2020). Cryogenian snowball Earth glaciations have been hypothesized to be among the events contributing to that sub–Great Unconformity erosion. A more interesting aspect of this hypothesis is that it provides a mechanism for the abrupt creation of the accommodation required for the large increase in the volume (and area) of sediments capping the Great Unconformity (Keller et al., 2019). The isostatic aspect of this crustal-thinning hypothesis has the advantage of making several testable predictions, including the timing of increasing continental sedimentary coverage after Cryogenian deglaciation. When this is modeled and compared to the original Macrostrat data for North America (fig. 4 of Keller et al., 2019), there is a temporal lag between the predicted and observed expansion of sedimentary cover. Indeed, the glacial denudation model predicts that peak sediment coverage should have occurred during the latest Ediacaran and earliest Cambrian. However, expansion in continental sedimentary coverage was much more protracted; it did not occur before the middle Cambrian, and sediment coverage of North America (and the global extent of shallow seas; Ronov et al., 1980) seems to have continued to expand even into the Ordovician (Fig. 8A).

There are multiple possible explanations for the accommodation–sedimentation timing offset in the Keller et al. (2019) model. One possibility is that sampling and/or chronological bias in the original Macrostrat database for North America was grossly misrepresenting the actual volume and extent of Ediacaran–early Cambrian rocks. However, our new higher-resolution and up-to-date Ediacaran compilation demonstrates that this is not likely (Fig. S1), and even identifies a Phanerozoic-like Sloss sequence cycle within the Ediacaran (Fig. 8). Another possible explanation for the sedimentation lag was that Ediacaran depositional systems may have been sediment starved due to a lack of proximal topography. This might be expected if Cryogenian glaciations essentially “planed off” continental surfaces and pushed large quantities of sediment onto oceanic crust, where it was recycled. This would mean large areas of the continents were flooded after deglaciation, but little or no sediment was deposited. The dominance of mud-size sediments in the Lower Ediacaran is intriguing, but the presence and increasing proportion of nonmarine sedimentary deposits throughout the Ediacaran, on the margins of Laurentia, as well as coarser sediment deposition after ca. 585 Ma, indicate that such a sediment-starvation scenario is unlikely for at least the youngest half of the Ediacaran (Figs. 6D and 7). Instead, the weight of evidence suggests that Ediacaran sedimentary systems were accommodation limited, likely due to a lack of continental sedimentary basin formation and broader continental flooding, as may have been the case for much of the preceding Neoproterozoic and Mesoproterozoic.

Our quantitative macrostratigraphic perspective on the Ediacaran System in North America provides a geologic framework within which tectonostratigraphic, geochemical, and paleobiologic patterns can be quantified, analyzed, and compared to better-known intervals of the Phanerozoic. There are several striking similarities between the Ediacaran and Lower Mesozoic of North America that likely reflect a common position in the Wilson cycle and a superimposed Sloss sequence–scale accommodation cycle. The geographic and temporal extent of Ediacaran glaciation and the ways in which it relates to first occurrences of macrofossils and δ¹³C trends of marine carbonates remain unclear. There is little signature of Ediacaran glaciation in North American macrostratigraphy, but it is also possible that large ice sheets were located on other continents and left more distinct signatures in those successions and in the higher-order temporal structure of sedimentation than is resolvable by our third-order (10⁶ yr) and lower analysis.

The extent to which the Ediacaran fossil record and apparent timing of initial diversification of animals are overprinted by a rock record–related sampling bias is unclear, but it seems much more likely that there is a common cause that links temporal patterns in sedimentation to the evolution of the biosphere, either directly through something like a species-area effect, or indirectly through an effect on biogeochemical cycling and the protracted rise of oxygen to Phanerozoic levels (Partin et al., 2013; Lee et al., 2016; Husson and Peters, 2017; Wang et al., 2018; Williams et al., 2019).

Compilations of sedimentary and igneous rocks, assembled over continental and global scales, have made it abundantly clear that there are shared global signals carried by most large continental blocks (Ronov et al., 1980; Husson and Peters, 2017). However, each continent, each domain within a continent, and each basin and section within that continent are also, at some level, unique. The extent to which the macrostratigraphy for the Ediacaran System of North America contains a global signal remains unknown, but the close correspondence between the biggest increase in shelf carbonate sedimentation in North America and the largest ostensibly global carbon isotope excursion is quite compelling. This suggests that a global signal is likely to be present in these data, and that it was likely related to global continental flooding, as it was in the Phanerozoic. The quantitative signature of rock destruction is muted in these data and in the overlying Phanerozoic surviving rock quantities. However, erosion and loss of sediments have clearly occurred, and the extent to which Ediacaran sediments have been lost from the interior regions of North America is not known. However, the presence of thin Cambrian and Ordovician capping sediments over nearly the entire continent constrains the timing of any such loss of Ediacaran deposits; based on the observed distribution of facies and environments, we believe that significant quantities of Ediacaran sediment did not originally extend far beyond the limits that are defined by their current positions along the margins of the continent. However, any Ediacaran sediments deposited on adjacent oceanic crust would have been completely recycled by the end of the Paleozoic. Extension of our effort to other continents, and additional analyses of this data set that are focused on specific lithological and environmental components of the system, will expand our perspective of the Ediacaran Period in North America and provide an important quantitative chronostratigraphic framework within which to integrate and interrogate proxy records at the dawn of animal life.

D.C. Segessenman was supported by funding from the University of Wisconsin–Madison Geoscience Department and through the Morgridge Distinguished Graduate Fellowship. We would like to thank the Macrostrat development team and B. Hupp for their feedback on early versions of this study. Macrostrat infrastructure development was supported by U.S. National Science Foundation grant EAR-1150082 and EarthCube grant ICER-1440312. We would also like to thank N. Christie-Blick and an anonymous reviewer for their thorough, constructive comments that greatly improved this paper.