Continental-scale drainage reorganization during Mesoproterozoic orogenesis: Evidence from the Belt Basin of western North America Open Access

The Mesoproterozoic evolution of Laurentia involved several orogenic events that transformed the configuration of the North American plate and its margins. Continental growth along the southwestern margin of ancestral North America occurred at 1710–1300 Ma by accretion of multiple magmatic arc terranes that constitute the Yavapai, Mazatzal, and Mojave crustal provinces (e.g., Whitmeyer and Karlstrom, 2007; Daniel et al., 2023a), and this was accompanied in part by the 1490–1340 Ma intrusion of A-type granites (Goodge and Vervoort, 2006, and references therein). Pervasive Grenville orogenesis at 1200–900 Ma during assembly of Rodinia generated deformation and continental construction along the eastern and southeastern margins of ancestral North America (Rivers, 1997, 2008; Mosher, 1998; Gower and Krogh, 2002; Swanson-Hysell et al., 2023). The fundamental tectonic and metamorphic records of these orogenic phases are well established, and the widespread preservation of zircon grains generated during orogenesis attest to large-scale magmatism and crustal thickening (Spencer et al., 2015). However, a lack of widespread Precambrian sedimentary basins with protracted depositional records has greatly limited the ability to reconstruct surface processes and sedimentation pathways during the Mesoproterozoic evolution of Laurentia.

The Belt-Purcell Supergroup, a remarkably thick (5–20 km) metasedimentary succession of Mesoproterozoic basin fill, defines a >100 m.y. record of principally clastic sediment accumulation in the Belt Basin of the northwestern United States and southwestern Canada. These rocks provide a key opportunity to assess sedimentary depositional systems and their relationships to orogenesis and potential interactions with continental blocks during the Precambrian evolution of Laurentia. The basin has been variably attributed to intracratonic rifting within the North American plate or to basin development along the ancestral western continental margin during collisional interactions with Australia, Antarctica, or Siberia following extensional breakup of the Paleoproterozoic Columbia supercontinent and prior to the late Mesoproterozoic–early Neoproterozoic Grenville orogenesis and assembly of Rodinia (Zhao et al., 2002, 2004; Ross and Villeneuve, 2003; Sears et al., 2004; Medig et al., 2014; Winston, 2016; Jones et al., 2015; Brennan et al., 2021; Parker and Hendrix, 2022). Despite its importance in the evolution of North America and in reconstructions of supercontinent assembly and sediment dispersal, there remain fundamental uncertainties in the depositional age and sediment provenance record of the Belt-Purcell Supergroup.

Previous studies of the Belt Basin reported provenance assessments that largely focused on exotic, non–North American sources in the context of several hypothetical Precambrian supercontinent configurations. Proposed western sources of exotic detritus focused on Australia and East Antarctica (Dalziel, 1991, 2013; Ross et al., 1992; Stewart et al., 2010; Halpin et al., 2014; Jones et al., 2015; Brennan et al., 2021), with alternative models proposing sources in Siberia or China (Li et al., 1995, 2002; Sears and Price, 2003; Sears et al., 2004; Link et al., 2007; Zhang et al., 2012; Meert and Santosh, 2017). Relatively less emphasis has been placed on ancestral North American sources and the potential linkages between Laurentian orogenesis and sediment dispersal patterns. Two principal categories of North American provenance signatures have been recognized: (1) Archean detritus from nearby sources in the Wyoming province and smaller western Laurentian blocks (Mueller et al., 2016; Ronemus et al., 2020) and (2) late Paleoproterozoic to Mesoproterozoic contributions from the Yavapai, Mazatzal, and/or Mojave provinces that may have been related to newly recognized continental-scale deformation (e.g., Jones et al., 2015; Link et al., 2016, 2017; Doe and Daniel, 2019; Daniel et al., 2023b).

Improved constraints on the absolute ages and rates of deposition in the Belt Basin are essential for consideration of basin-forming mechanisms and comparison with other basins and tectono-magmatic records across Laurentia. A broad depositional age range between 1500 Ma and 1400 Ma is commonly adopted for the Belt-Purcell Supergroup (Sears et al., 1998; Evans et al., 2000). Despite a lack of detailed chronostratigraphic constraints, high rates of sediment accumulation have been envisaged for most of the succession (e.g., Lydon, 2000; Sears, 2007b; Parker and Hendrix, 2022). In general, detrital zircon (DZ) geochronologic analyses of the Belt Basin have been limited by: (1) an insufficient number of analyses per sample to be statistically sound based on contemporary standards; (2) pooling of results from stratigraphically correlative but lithologically distinct units with potentially contrasting provenance; (3) insufficient sampling of different stratigraphic intervals, such that thousands of meters of section may be defined by a single sample; and (4) a focus on particular age groups (e.g., non–North American ages) rather than the full spectrum of age distributions (Ross et al., 1991, 1992; Ross and Villeneuve, 2003; Link et al., 2007; Stewart et al., 2010). These issues highlight the need for a well-constructed evaluation of the complete stratigraphic succession with systematic sampling of multiple units and generation of a sufficiently large data set.

This study aimed to refine the chronostratigraphic framework and sediment provenance record for the Belt-Purcell Supergroup by assessing all major stratigraphic units within a single continuous section. Well-exposed, unambiguous units in the northeastern Belt Basin of northwestern Montana, USA, and southwestern Alberta, Canada, comprise an uninterrupted succession ~5 km thick. We sampled all major formations across four localities within the structurally exhumed Lewis thrust salient. DZ U-Pb geochronologic data were acquired for 27 samples by laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS). We then qualitatively and quantitatively compared our U-Pb results to the igneous and metamorphic zircon ages of major Precambrian tectono-magmatic provinces across Laurentia compiled from the literature using multidimensional scaling (MDS) plots and a statistical model to evaluate the contribution of clastic material from Laurentia to the Belt Basin. Our objectives were to: (1) estimate the age, duration, and rate of sediment accumulation; (2) evaluate sediment provenance, routing patterns, and regional paleogeography; and (3) determine the occurrence and timing of potential shifts in source regions within the available Mesoproterozoic framework for ancestral North America. These results have implications for Laurentian paleogeographic reconstructions, the extent and configuration of Mesoproterozoic drainage systems, possible episodes of continental-scale drainage reorganization, the position of past geographic divides, and possible influence of pre-Grenville orogenesis.

The Belt Basin contains Mesoproterozoic metasedimentary rocks belonging to the Belt Supergroup in the United States and the equivalent Purcell Supergroup of Canada (Fig. 1). Preserved basin fill covers an area of ~200,000 km², with ~90% exposed in Montana, Idaho, and Washington, USA, and the remainder in Alberta and British Columbia, Canada. The basin’s western depocenter near the Mission Range, Montana, records some ~20 km of predominantly fine-grained siliciclastic and carbonate rocks (Harrison et al., 1974; Winston et al., 1989), with gradual thinning to ~4.5 km on the northeastern basin margin near North Kootenay Pass, Alberta (Price, 1964; Fermor and Price, 1983; Sears, 2007a; Fuentes et al., 2011). The Belt-Purcell Supergroup overlies Paleoproterozoic quartzites and Archean to Paleoproterozoic crystalline basement rocks of the Wyoming craton, Selway terrane, Medicine Hat block, Great Falls tectonic zone, and Priest River complex (Foster et al., 2006; Brennan et al., 2021), and it was translated >100 km east-northeastward along the Lewis thrust during Late Cretaceous–early Eocene crustal shortening in the Cordilleran (Sevier) fold-and-thrust belt (Fermor and Price, 1983; van der Velden and Cook, 1994; Constenius, 1996; Sears, 2007a; Fuentes et al., 2011; Yonkee and Weil, 2015). Burial and regional metamorphism subjected the Belt-Purcell Supergroup to low-grade sub-greenschist-facies conditions, but the sedimentary character of the rocks is little changed. They also are distinguished by well-preserved sedimentary structures (Horodyski, 1993; Lydon, 2007). Reflecting this low-grade metamorphism, Belt-Purcell Supergroup literature commonly uses the terms quartzite, siltite, and argillite to describe the well-indurated sandstones and mudstones found in the study area. We use the latter to emphasize the grain size of the clastic rocks as they relate to this DZ study.

The Lewis thrust salient (Fig. 2) within the Mesozoic–early Cenozoic Cordilleran (Sevier) orogenic belt contains the northeasternmost exhumed portion of the Belt Basin, as exposed in the United States (Glacier National Park [NP], Lewis and Livingstone Ranges) and Canada (Waterton Lakes NP and Castle Wildland Provincial Park [PP], Clark Range). The allochthonous Lewis thrust sheet has been the subject of numerous geological studies that highlight the tectonic evolution of the North American Cordillera (Bally et al., 1966; Mudge and Earhart, 1980; Fermor and Price, 1987; van der Velden and Cook, 1994). The Lewis thrust salient overlies a bedding-parallel detachment surface with a regional westward dip of 3°–6° (Ross, 1959; Bally et al., 1966; Fermor and Price, 1983; Yoos et al., 1991; van der Velden and Cook, 1994); the salient is truncated in the west by the Flathead fault, a southwest-dipping listric normal fault (Dahlstrom, 1970; McMechan and Price, 1980; McMechan, 1981; Constenius, 1988, 1996; Whipple, 1992; Stockmal and Fallas, 2015).

The Belt-Purcell Supergroup consists of the Lower Belt, Ravalli Group, Piegan Group (Middle Belt Carbonate), and Missoula Group (Fig. 3). Stratigraphic thicknesses and lithologic descriptions have been documented for the Lewis, Livingstone, and Clark Ranges and correlative strata in Montana and Idaho (Ross, 1959; Fermor and Price, 1983, 1987; Winston et al., 1989; Pratt, 2001; Pratt and Ponce, 2019; Pratt and Rule, 2021). Contrasting nomenclature exists among stratigraphic equivalents in the western depocenter versus eastern basin exposures, and in the northern versus southern sides of the USA-Canada border. This study uses the accepted formation names according to relative sample locations north or south of the international border. Accordingly, the Lower Belt within the Lewis thrust salient in the United States is subdivided into the Waterton, Altyn, and Appekunny Formations, and the Ravalli Group is subdivided into the Grinnell and Empire Formations. Up section, the Piegan Group consists of the Helena Formation, and the Missoula Group in Canada is subdivided into the upper Siyeh, Purcell Lava, Sheppard, Gateway, Phillips, and Roosville Formations. For regional stratigraphic correlations of the Belt-Purcell Supergroup, the reader is referred to Lonn et al. (2020).

The Waterton Formation is a predominantly mixed carbonate and siliciclastic succession interpreted as a carbonate ramp formed on the northeast margin of the Belt Basin (Pratt and Rule, 2021). The Waterton Formation within the vicinity of Waterton Lakes NP contains more than 180 m of gray and greenish-gray interbedded dolomite, limestone, and mudstone (Daly, 1912; Fermor and Price, 1983, 1987; Pratt and Rule, 2021).

The Altyn Formation overlies the Waterton Formation and consists primarily of light-gray to green sandy and dolomitic mudstone and sandy dolomite ~250 m thick (Willis, 1902; Fermor and Price, 1983, 1987; White, 1984; Pratt and Rule, 2021). The Waterton-Altyn Formation succession, along with other units of the Belt-Purcell Supergroup, contain crack arrays filled with injected sediment that were formerly regarded as mud cracks due to syneresis (e.g., Fenton and Fenton, 1937; Winston, 1986) and more recently as the product of oscillatory wave action and later water escape during burial (Winston and Smith, 2016). Alternatively, it has been proposed that the Waterton and Altyn Formations contain units that experienced synsedimentary brittle and ductile deformation, expressed as fractured stromatolites, microfaults, breccias, cataclasites, ball-and-pillow structures, and microfolds that are interpreted as seismites (Pratt, 1994, 2017; Pratt and Rule, 2021). In addition, the Altyn Formation, in particular, contains evidence for the impact of repeated tsunamis and strong tidal currents (Pratt and Rule, 2021).

The Appekunny Formation consists of green, gray, and red mudstone interbedded with white and green sandstone with a thickness of ~550–660 m (Willis, 1902; Fermor and Price, 1983, 1987; Winston et al., 1989; Slotznick et al., 2016). The Appekunny Formation paleocurrents display a variable array of azimuths (McMechan, 1981).

The more marginal carbonates of the Lower Belt correlate with deeper-water deposits of the Prichard and Aldridge Formations in the western depocenter of the Belt Basin. Several studies hypothesized that coarse- and fine-grained deposits in these units were proximally and distally sourced, respectively (Frost and Winston, 1987; Pratt and Rule, 2021). In this scenario, coarse-grained basin-margin sediments were derived from the nearest exposed craton, and fine-grained siliciclastic sediments were derived from greater distances and reached the deeper basin localities via hyperpycnal and turbidity currents. The fine-grained clastic sediments are attributed to exotic western sources on the basis of isopach thicknesses, paleocurrent data (Cressman, 1989), and DZ ages that demonstrate the presence of non–North American grains with ca. 1610–1490 Ma ages (Frost and Winston, 1987; Ross and Villeneuve, 2003; Link et al., 2007; Brennan et al., 2021). Furthermore, mafic sills—such as the Paradise sill (1457 ± 2 Ma; Sears et al., 1998), Plains sill (1469 ± 2.5 Ma; Sears et al., 1998), Crossport sill (1433 ± 10 Ma; Zartman et al., 1982), and Moyie sills (1445 ± 11 Ma and 1468 ± 2.5 Ma; Höy, 1993; Anderson and Davis, 1995)—are hypothesized to have intruded wet unconsolidated sediments of the Prichard and Aldridge Formations, which places a rough estimate for earliest deposition in the Belt Basin at ca. 1470 Ma (Höy, 1989; Sears et al., 1998; Poage et al., 2000). In the Lewis salient, a mafic sill that intrudes the Appekunny Formation yielded an isotope dilution–thermal ionization mass spectrometry (ID-TIMS) baddeleyite age of 1436.2 ± 1.1 Ma (Pană et al., 2018).

The Grinnell Formation contains ~600 m of interbedded maroon, red-brown, and green mudstone and white sandstone beds (Fermor and Price, 1983, 1987; Winston et al., 1989; Pratt and Ponce, 2019). The Grinnell Formation of the Lewis thrust sheet is considered to be correlative with the Burke, Revett, and St. Regis Formations near the depocenter (Montana-Idaho border region) to the southwest and the Creston Formation in southern British Columbia. Harrison et al. (1974) proposed that the distinctive white sandstones of the Grinnell Formation were easterly derived, and the siltstones were sourced from the west. Revett-Creston isopach thicknesses and paleocurrent directions indicate a south or southwestern provenance direction for the mud (Harrison, 1972; Höy, 1993). Winston et al. (1989) and Winston (2016) interpreted these sediments as involving fluvial sheetflood deposits, and the presence of ripple marks and mud cracks likely indicates periodic episodes of subaerial exposure. Alternatively, Pratt and Ponce (2019) presented evidence for low-energy subaqueous conditions punctuated by off-surge of episodic tsunamis that carried in sand from the east and evidence that the mud cracks were seismically generated. The Empire Formation (lower Siyeh Formation in Canada) is composed of green mudstone interbedded with gray dolomitic sandstone with a thickness of ~125 m (Fermor and Price, 1983, 1987).

The Helena Formation (middle Siyeh Formation in Canada) is ~800 m thick and consists of cliff-forming gray and buff silty, argillaceous limestone and dolomite, minor gray mudstone with stromatolitic intervals, and rare sandstone (Fermor and Price, 1983, 1987; Winston et al., 1989; Pratt, 2001). The formation is commonly riddled with molar-tooth structure, which is expressed as thin ribbons and blobs of microcrystalline calcite within argillaceous carbonate rocks. Their origin is debated, and two hypotheses for the Belt-Purcell Supergroup units have been proposed. Gas escape structures were proposed by Furniss et al. (1998), whereas Pratt (1998, 2001, 2023) attributed them to subaqueous shrinkage, liquefaction, and injection of lime mud subject to synsedimentary earthquake-induced shaking. Harrison (1972) reported that provenance of the Helena Formation was from the east, south, and southwest, in what he referred to as a heterogeneous and compound sedimentary prism. Paleocurrent inferences have been taken from the long axis of Baicalia and Conophyton stromatolites that are inclined to the north-northeast (Horodyski, 1977; Winston et al., 1989). Additionally, an unusual mudstone in the upper part of the Helena Formation at Logan Pass, Glacier NP, interpreted as a meta-bentonite (Moe et al., 1996), yielded a U-Pb zircon age of 1454 ± 9 Ma (Evans et al., 2000). This age has been widely quoted (e.g., Luepke and Lyons, 2001; Link et al., 2007; Lydon, 2007; Stewart et al., 2010; Bookstrom et al., 2016).

The upper Siyeh Formation (Snowslip Formation in the United States) consists of ~360 m of green dolomitic mudstone and fine-grained sandstone interbedded with green and red mudstone (Figs. 3 and 4; Fermor and Price, 1983, 1987; Whipple and Johnson, 1988). The alternating sequences of carbonate and sandstone were interpreted by Whipple and Johnson (1988) and Winston et al. (1989) to represent northward progradation of subaerial sand flats and alluvial aprons. Isopach thicknesses for the upper Siyeh–Phillips Formations suggest coeval sedimentation and tectonism (Harrison, 1972; Winston et al., 1989; Höy, 1993).

The Purcell Lava overlies the upper Siyeh Formation and is composed of ~9–15 m of green chloritized pillow basalt at its base and ~0–54 m of pahoehoe flows (Daly, 1912; Burling, 1916; Hume, 1932; McGimsey, 1985; Verstraeten-Gless, 1987). Evans et al. (2000) reported a sensitive high-resolution ion microprobe (SHRIMP) U-Pb zircon age of 1443 ± 7 Ma from what they considered to be a thin flow unit of rhyolite or quartz latite in the McGillivray Range, north of Libby, Montana. Reevaluation of the dated flow unit revealed that it was actually a peperite formed by the volatile interaction of basalt and water-laden sediment and that the zircons were derived from the sediment, not the basalt. Therefore, the cited age is the maximum depositional age (MDA) of siliciclastic sediments intercalated with the basalts (Constenius et al., 2017).

The overlying Sheppard Formation (Shepard Formation in the United States) consists of ~165 m of yellowish-gray and gray-green dolomitic sandstone and mudstone (Fermor and Price, 1983, 1987; Whipple et al., 1985), and it is overlain by the Gateway Formation (Mount Shields Formation in the United States), which is up to ~800 m thick. The Gateway Formation is composed of red to red-brown mudstone and sandstone with minor purplish-red, gray, and green mudstone and dolomitic mudstone with stromatolites (Fermor and Price, 1983, 1987).

The Phillips Formation (Bonner Formation in the United States) is pink to maroon, cross-bedded to laminated, coarse- to fine-grained sandstone and mudstone that attain a thickness of 250 m (Fermor and Price, 1983; Whipple et al., 1985). These sandstones have been variously interpreted as braided fluvial channels (Whipple et al., 1985; Winston et al., 1986, 1989) or sheetflood deposits (Winston, 2016). Provenance studies of the Missoula Group by Stewart et al. (2010) and Link et al. (2016) suggested a southern source in the Yavapai-Mazatzal provinces.

The Roosville Formation (McNamara and Libby Formations in the United States) consists of ~450 m of red-brown, green, and gray mudstone, dolomitic mudstone, and lesser sandstone (Fermor and Price, 1983). These predominantly fine-grained siliciclastic rocks are interpreted to have been deposited in a shallow, subaqueous to subtidal paleoenvironment (Whipple et al., 1985). A mudstone unit thought to be a tuff from the basal part of the formation near Libby, Montana, yielded a zircon age of 1401 ± 6 Ma (Evans et al., 2000).

DZ U-Pb geochronologic data were acquired through LA-ICP-MS analyses of 27 samples (17 sandstones, 8 mudstones, 1 oolitic grainstone, 1 meta-bentonite) spanning all stratigraphic units except the Empire Formation (see Table S1¹). The relative stratigraphic position for each sample is denoted by a numerical prefix (1–27; Fig. 3). Samples of ~2–5 kg were prepared according to standard mineral separation techniques, with zircon extraction conducted at ZirChron, LLC, Tucson, Arizona, USA, using a jaw crusher, disc miller, electric pulse disaggregator, Wilfley water table, Frantz magnetic separator, and heavy liquids. Zircon separates were mounted in epoxy pucks, polished to expose grain interiors, and analyzed at Washington State University, Pullman, Washington. We sought to analyze a minimum of 120 grains per sample. All U-Pb results are reported as ²⁰⁷Pb/²⁰⁶Pb ages, along with associated 2σ uncertainties, for individual grains with <20% normal discordance and <5% reverse discordance.

The DZ U-Pb data set was assessed using detritalPy software (Sharman et al., 2018) in order to estimate the stratigraphic age, depositional duration, and average rates of sediment accumulation. Calculations of the MDA for individual samples has proven to be an effective tool in chronostratigraphic studies of basin fill with syndepositional volcanic input (Dickinson and Gehrels, 2009; Spencer et al., 2016; Sharman and Malkowski, 2020; and references therein). Whereas MDA estimates based on the youngest single grain (YSG) are effective for Cenozoic successions, Precambrian strata typically require multiple overlapping DZ ages to evaluate MDA values. Accordingly, this study reports MDA calculations based on the youngest cluster of at least three grains with 2σ uncertainty (YC2σ[3+]) and the youngest cluster of at least two grains with 1σ uncertainty (YC1σ[2+]) that displayed an up-section trend that decreased in age; YSG ages are only reported where applicable. The MDAs were calculated using only those analyses that were 1:1 concordant, consisting of data for which the 2σ uncertainty ellipse overlapped the 1:1 concordia line, which ultimately filtered out grains that were approximately >3% normally and >3% reversely discordant (Spencer et al., 2016).

Geochronologic data for potential sediment source regions were compiled into a database of igneous and metamorphic zircon U-Pb ages (n = 15,474) that spanned Laurentia during the Precambrian (Fig. 5). Geochronologic data from 150 references characterize 22 major tectono-magmatic Laurentian provinces, including the Slave craton, Taltson magmatic zone, Rae and Hearne cratons, Athabasca intrusive igneous rocks, Churchill province, Trans-Hudson orogen, Sask and North Atlantic cratons, Makkovikian orogen, Medicine Hat, Wyoming, and Superior cratons, Abitibi greenstone belt, Penokean and Grenville orogens, A-type (anorogenic) granites, as well as the Mazatzal, Mazatzal-Yavapai, Yavapai, Yavapai-Mojave provinces, and the Mojave crustal block. Additional information regarding the location (latitude, longitude), lithology, method of data acquisition (SHRIMP, ID-TIMS, LA-ICP-MS), zircon ablation position (core, rim), igneous and metamorphic U-Pb ages (older than 1000 Ma), uncertainty (2σ), and references therein is summarized in Table S2.

Non-Laurentian sources were not included in the compilation due to the extremely small proportion of non–North American (ca. 1610–1490 Ma) zircon grains analyzed in this study, and the disagreement over the identity of potential exotic sources (including multiple provinces in Australia, Antarctica, Siberia; Ross et al., 1992; Ross and Villeneuve, 2003; Sears and Price, 2003; Lewis et al., 2010; Lonn et al., 2020; Brennan et al., 2021). Formations such as the Prichard, Creston, and Revett Formations from the western side of the Belt Basin are notable for having large numbers of non–North American zircons that form distinct age peaks (Ross and Villeneuve, 2003; Brennan et al., 2021; Perelló et al., 2021). In contrast, our data from the northeastern side of the basin have a small percentage of grains (~5%) of this age scattered through this time interval (filtered <20% normal and <5% reverse discordance). Notably, these grains were mainly found in the seven mudstone units that yielded smaller zircons (<100 µm) and required analysis with a smaller beam size than the sandstone samples (15–20 µm vs. 25–30 µm), which resulted in larger uncertainties. In a reassessment of these ages using a more restrictive discordance filter (<10% normal and <5% reverse discordance) and only accepting ages with uncertainties that fell completely within the non–North American time interval, <1% would be considered exotic to ancestral North America.

The provenance of the Belt-Purcell Supergroup was evaluated qualitatively with MDS plots generated with detritalPy developed by Sharman et al. (2018). These illustrations, where the axes are dimensionless, are graphical representations of the similarity or dissimilarity between the U-Pb age distributions of samples within study (Vermeesch, 2013). Samples with zircon age distributions that resemble one another will plot in tight clusters, whereas dissimilar samples will yield a scattered appearance. For this reason, we incorporated MDS plots to qualitatively evaluate the similarity or dissimilarity between detrital samples from the Belt Basin and potential Laurentian sediment sources.

Laurentian clastic contributions to the Belt Basin were quantitatively determined using a statistical model. The relative mixing proportions of sediments from ancestral North American sources were determined using the MATLAB-based inverse Monte Carlo model DZmix developed by Sundell and Saylor (2017). The DZmix software calculates the proportional (weighted) contribution of the 22 potential igneous and metamorphic terranes to our detrital samples, such that a noncontributor is 0%, and the sum of all contributors is 100%. The software randomly generates many combinations (or mixtures) of the potential sources, in variable weighted proportions, and then iteratively defines the range of best-fit solutions (Sundell and Saylor, 2017; Sundell et al., 2018). The model considered all Laurentian regions, thereby minimizing the potential bias from previous interpretations of proximal versus distal derivation and without making assumptions about the original (syndepositional) extent of actively eroding Precambrian igneous and metamorphic basement. Results from the cross-correlation test were plotted based on the sensitivity and discriminatory power with the number and proportion of age distributions (Sundell and Saylor, 2017), and the coefficient of determination (R²) was used as a numerical estimate of the degree to which the cross-correlation model’s weighted proportions matched the actual (or analyzed detrital) sample (R² = 1 is perfect; R² = 0 is imperfect). For this study, when R² > 0.7, we inferred that the sediment sources were acceptably characterized. Conversely, when R² < 0.7, we suggest the observed age distribution could not be replicated by the program using the sources considered. Potential explanations for a poor fit include undocumented source areas (due to incomplete source characterization or erosion of source materials) and sedimentary recycling (such that the grains were not delivered as first-cycle sediments directly from the igneous or metamorphic sources to the depositional basin). For the Belt Basin, this approach enabled an independent assessment of the potential importance of non-Laurentian (ca. 1610–1490 Ma) grains, such that a poor fit (R² < 0.7) would be expected for any sample with significant amounts of such grains. This study reports DZmix results in terms of the mean, standard deviation, and coefficient of determination values for the cross-correlation model, representing the top five best-fit model outputs from 50,000 random combination iterations with 2σ uncertainty over the 3500–1300 Ma age interval (see Table S6).

New detrital U-Pb results for the Belt Basin and compiled igneous and metamorphic U-Pb data for potential Laurentian source regions were binned according to Proterozoic and Archean ages. Previous provenance analyses have referred to various age intervals according to the dominant geologic province in which those zircon ages most commonly occur; however, this nomenclature can propagate interpretive bias. For example, the 2000–1800 Ma age range has commonly been linked to the Trans-Hudson orogen (e.g., Laskowski et al., 2013), but zircon grains of that age interval are also observed in the Makkovikian orogen; the Churchill province; and the Slave, Rae, Hearne, and Wyoming (includes its modified NW edge) cratons. This study used the Precambrian chronologic divisions from the Geologic Society of America time scale (Walker et al., 2013), including the late Calymmian to Ectasian (LC-ET; 1480–1340 Ma), early Calymmian (EC; 1600–1480 Ma), late Statherian (LS; 1700–1600 Ma), early Statherian (ES; 1800–1700 Ma), Orosirian (OS; 1920–1800 Ma), and Siderian to Rhyacian periods (SD-RY; 2500–1920 Ma), and Archean Eon (AR; 3500–2500 Ma) (see Table S4).

Sediment source regions to the north, northeast, and east of the Belt Basin comprise the bulk of the Canadian Shield (Fig. 6). These geologic domains contain igneous and metamorphic zircon populations dominated by Archean and Paleoproterozoic ages, with no Mesoproterozoic groups.

In the north, potential source regions include the Slave craton and Taltson magmatic zone. The Slave craton is predominantly Archean (57%) and Orosirian (28%) with subordinate amounts of Siderian–Rhyacian (13%) ages. The Taltson zone is dominated by Siderian–Rhyacian (65%) and Archean (23%) ages with subordinate Orosirian (12%) ages. Northern source regions thus likely shed siliciclastic material with Archean (3500–2500 Ma) and early (2500–1920 Ma) to middle Paleoproterozoic (1920–1800 Ma) zircon age signatures.

Potential sources to the northeast include the Rae and Hearne cratons, Athabasca intrusive igneous rocks, Churchill province, Trans-Hudson orogen, Sask and North Atlantic cratons, and Makkovikian orogen. The Rae craton contains Archean (59%), Siderian–Rhyacian (21%), and Orosirian (20%) dates; the Hearne craton has a nearly identical distribution with Archean (62%), Siderian–Rhyacian (21%), and Orosirian (16%) ages. The Athabasca igneous rocks predominantly contain Siderian–Rhyacian (72%) and Orosirian (24%) ages. The Churchill province exhibits Archean (51%), Orosirian (27%), and early Statherian (13%) age signatures. The Trans-Hudson orogen contains a high proportion of Orosirian (38%), Archean (29%), and Siderian–Rhyacian (23%) ages with subordinate early Statherian (10%) dates. The Sask craton has an age spectrum nearly identical to the Trans-Hudson orogen but with lesser or greater concentrations of early Statherian (37%), Archean (26%), Siderian–Rhyacian (24%), and Orosirian (13%) ages. The North Atlantic craton is composed almost entirely of Archean (95%) dates. Finally, the northeastern Makkovikian orogen has a comparatively younger age spectra with Orosirian (66%), early Statherian (13%), and late Statherian (12%) ages. From these observations, potential northeastern sediment sources can be characterized as having dominantly middle (1920–1800 Ma) and early Paleoproterozoic (2500–1920 Ma) and Archean (3500–2500 Ma) zircon age populations.

Potential sediment sources directly east of the Belt Basin consist of the proximal Medicine Hat and Wyoming cratons and the distal Superior craton and Abitibi greenstone belt. The Medicine Hat craton contains Archean (84%) and Siderian–Rhyacian (16%) age signatures. Primarily Archean (68%) and Orosirian (20%) dates characterize the Wyoming craton (includes the Little Belt arc). Farthest east, the Superior craton and Abitibi greenstone belt contain almost exclusively Archean (89% and 100%, respectively) ages. In summary, source regions east of the Belt Basin contain high concentrations of Archean (3500–2500 Ma) ages.

Potential sediment sources south and southeast of the Belt Basin are grouped into a single category referred to as the southern provinces (Fig. 6). We consider this category to include the Penokean and Grenville orogens, A-type granites, Mazatzal province, Mazatzal-Yavapai transition, Yavapai province, Yavapai-Mojave intermediate zone, and Mojave crustal province. Primarily early Statherian (73%) and late Statherian (14%) dates characterize the Penokean orogen. High proportions of late Calymmian–Ectasian (38%), early Calymmian (24%), and late Statherian (20%) ages characterize the Grenville orogen. The age spectra for A-type granites is predominantly late Calymmian–Ectasian (91%). The Mazatzal province resembles the Penokean orogen in that it is dominated by late Statherian (80%) and early Statherian (11%) ages. The age distributions of the Mazatzal-Yavapai transition overlap with the Grenville province but contain greater amounts of late Calymmian–Ectasian (55%), late Statherian (27%), and early Calymmian (11%) dates, whereas the Yavapai province exhibits late Statherian (74%) and early Statherian (22%) zircon. The Yavapai-Mojave transition contains early Statherian (54%), late Statherian (20%), and Orosirian (16%) ages, and the Mojave crustal block contains early Statherian (31%), late Statherian (29%), and Orosirian (19%) dates. To summarize, the southern provinces are characterized primarily by late Paleoproterozoic (1800–1600 Ma) and early Mesoproterozoic (1480–1340 Ma) igneous and metamorphic zircon U-Pb ages and negligible proportions of the early Paleoproterozoic (2500–1800 Ma) and Archean (3500–2500 Ma) dates that dominate the Canadian Shield.

In total, 3123 individual grain analyses from 27 samples covering all stratigraphic units of the northeastern Belt Basin (excluding the Empire Formation; Fig. 7) yielded 530 that were >20% normal and >5% reversely discordant, leaving a filtered total of 2593 ages (see Table S3).

In the lower Belt Supergroup succession, samples 1–13, DZ age distributions for the Waterton to lower Helena Formations showed a high proportion of Paleoproterozoic (2500–1800 Ma) to Archean (3500–2500 Ma) ages (Fig. 7). The lowest stratigraphic sample, sample 1, from the Waterton Formation at Waterton Lakes NP, Canada, contained early Statherian (10%), Orosirian (31%), Siderian–Rhyacian (15%), and Archean (36%) grains. Up section, sample 2 of the overlying Altyn Formation yielded early Statherian (20%), Orosirian (31%), and Archean (41%) ages. Compared to other lower Belt Supergroup succession samples, sample 3 from the Altyn Formation was the only sample with a higher proportion of early Calymmian (11%) ages, in addition to early Statherian (11%), Orosirian (16%), Siderian–Rhyacian (17%), and Archean (38%) ages. Further up section, a third Altyn Formation sample, sample 4, was one of two lower Belt Supergroup succession samples that showed a higher percentage of late Statherian (10%) ages, along with early Statherian (34%), Orosirian (24%), Siderian–Rhyacian (13%), and Archean (15%) ages. Sample 5 from the Appekunny Formation had ages that fall into the Orosirian (13%), Siderian–Rhyacian (13%), and predominantly Archean (66%) ranges. Directly up section, sample 6 was composed almost entirely of Orosirian (13%) and Archean (78%) grains. Sample 7 was characterized by early Statherian (19%), Orosirian (28%), and Archean (39%) age populations. The overlying upper Appekunny to middle Grinnell Formations (samples 8–11) showed closely similar probability density plots (PDP) in that they all contained varying proportions of early Statherian (17%, 16%, 14%, 16%), Orosirian (40%, 40%, 41%, 39%), Siderian–Rhyacian (12%, 11%, 18%, 14%), and Archean (31%, 34%, 24%, 29%) age distributions. The mudstone sample from the middle Grinnell Formation (sample 12) contained the second instance of late Statherian (16%) ages within the lower Belt Supergroup succession, along with early Statherian (16%), Orosirian (28%), Siderian–Rhyacian (18%), and Archean (14%) ages. In upper levels of the lower Belt Supergroup succession, the lower Helena Formation (sample 13) had an age distribution dominated by Orosirian (48%) ages, in addition to Siderian–Rhyacian (14%) and Archean (28%) ages. In summary, the lower Belt Supergroup succession contains a high proportion of middle Paleoproterozoic to Archean (averaging 94%, 3500–1700 Ma) grains.

DZ U-Pb results for the middle Belt Supergroup succession (upper Helena to Sheppard Formations; samples 14–18) revealed considerable diversity in age populations. In the upper Helena Formation, sample 14 represents the first significant appearance of late Calymmian–Ectasian (17%) and early Calymmian (20%) grains, along with late Statherian (31%) and early Statherian (23%) ages. This interval also coincides with a dramatic decrease in the Orosirian, Siderian–Rhyacian, and Archean populations that are common in lower Belt Supergroup succession samples. Additionally, the PDP for sample 15 showed similarity to sample 14 due to the presence of late Calymmian–Ectasian (11%), early Calymmian (15%), primarily late Statherian (42%), and early Statherian (23%) grains. However, fine-grained upper Helena Formation sample 16 was unique as a possible meta-bentonite (Moe et al., 1996; Evans et al., 2000), and it showed the first occurrence of a dominantly unimodal age population that spans the late Calymmian–Ectasian (69%) and early Calymmian (29%) periods. This unimodal pattern was followed by a reintroduction of early Statherian (26%) to Orosirian (37%) and Archean (35%) distributions in sample 17 of the upper Siyeh Formation. Up section, above the Purcell Lava, coarse-grained sample 18 of the Sheppard Formation also displayed a mostly unimodal age spectrum that demonstrates similarity to sample 16 with late Calymmian–Ectasian (67%) to early Calymmian (12%) ages. To summarize, the middle Belt Supergroup succession has heterogeneous DZ populations with primarily early Mesoproterozoic (averaging 48%, 1600–1300 Ma), late Paleoproterozoic (averaging 29%, 1800–1600 Ma), and middle Paleoproterozoic–Archean (averaging 23%, 3500–1800 Ma) age groups in this part of the Belt Basin.

The upper Belt Supergroup succession (Gateway to Roosville Formations; samples 19–27) showed relatively uniform results in exhibiting predominantly late Paleoproterozoic (1920–1600 Ma) populations. Sample 19 of the Sheppard-Gateway transition contained primarily late Statherian (28%) and early Statherian (42%) ages with subdued late Calymmian–Ectasian (10%) signatures. Sample 20 of the Gateway Formation contained late Statherian (22%), early Statherian (31%), and Orosirian (22%) ages, with sample 21 displaying similar late Statherian (10%), early Statherian (44%), and Orosirian (16%) ages in combination with an Archean (15%) component. Samples 22 and 23 of the upper Gateway and lower Phillips Formations showed nearly identical distributions dominated by late Statherian (24%, 19%), early Statherian (53%, 57%), and Orosirian (11%, 12%) ages. Mainly late Statherian (10%) and early Statherian (66%) ages were recorded in sample 24 of the upper Phillips Formation. Roosville Formation sample 25 represents a reintroduction of late Calymmian–Ectasian (11%) and early Calymmian (10%) ages, in addition to late Statherian (18%) and early Statherian (54%) ages. Roosville Formation sample 26 was nearly identical, with late Calymmian–Ectasian (12%), late Statherian (20%), and early Statherian (44%) components. However, sample 27 of the Roosville Formation displayed a wide age distribution spanning late Calymmian–Ectasian (15%), early Calymmian (11%), late Statherian (13%), early Statherian (20%), Orosirian (16%), Siderian–Rhyacian (11%), and Archean (13%) populations. In summary, all units in the upper Belt Supergroup succession contain a high quantity of late Paleoproterozoic (averaging 75%, 1920–1600 Ma) zircons and comparatively few Mesoproterozoic (1600–1340) or early Paleoproterozoic–Archean (3500–1920 Ma) grains.

The MDAs were calculated for multiple stratigraphic horizons using only those analyses that fell within 2σ uncertainty of concordia (Spencer et al., 2016; defined as analyses that overlapped the 1:1 concordance line). The youngest grain populations for individual samples were calculated in detritalPy using the youngest cluster of at least three grains with 2σ uncertainty (YC2σ[3+]), the youngest cluster of at least two grains with 1σ uncertainty (YC1σ[2+]), and (when applicable) the youngest single grain (YSG) (Fig. 8; Sharman et al., 2018). An initial assessment indicated that 16 samples exhibited MDAs that are older than 1500 Ma or diverge >20 m.y. from the projected sediment accumulation curve and, therefore, were not interpreted to contain syndepositional zircon. Of the remainder, preferred MDAs were selected from synsedimentary samples by first selecting a YC2σ(3+) or YC1σ(2+) estimate for which the range of uncertainties (not the mean age) displayed an upward trend that decreased in age (Fig. 9). After a reliable sediment accumulation history was determined, a YSG value was included when it was within error of stratigraphically adjacent calculations (see Table S5).

DetritalPy values indicated that 11 samples contained syndepositional zircon (Figs. 8 and 9). Sample 1 from the Waterton Formation yielded a YC2σ(3+) and YC1σ(2+) age of 1484.6 ± 7.7 Ma (mean square of weighted deviates [MSWD] = 0.04, n = 4). Sample 3 of the Altyn Formation yielded a YC1σ(2+) age of 1494.9 ± 9.5 Ma (MSWD = 0.39, n = 8). Two samples from the upper Helena Formation, samples 15 and 16, yielded MDAs of 1433.0 ± 16.1 Ma (YSG) and 1430.7 ± 3.2 Ma (YC2σ[3+], MSWD = 0.78, n = 27), respectively. Samples from the Missoula Group began with Sheppard Formation sample 18 with an MDA of 1419.5 ± 4.7 Ma (YC1σ[2+], MSWD = 0.14, n = 7). YC1σ(2+) ages from samples 19 and 21 of the Gateway Formation yielded MDAs of 1427.6 ± 11.8 Ma (MSWD = 0.60, n = 2) and 1421.0 ± 8.1 Ma (MSWD = 0.47, n = 5), respectively. Sample 20 had a YSG age of 1432.0 ± 10.0 Ma. The upper Phillips Formation sample 24 produced an MDA of 1386.7 ± 10.5 Ma (YC1σ[2+], MSWD = 1.45, n = 2), followed by Roosville Formation samples 26 and 27 with 1375.4 ± 10.5 Ma (YC2σ[3+], MSWD = 4.20, n = 3) and 1386.3 ± 11.2 Ma (YC1σ[2+], MSWD = 1.32, n = 2).

Collectively, the MDA results revealed a systematic decrease in age up section consistent with basin stratigraphic ages of ca. 1495–1380 Ma (Fig. 9). These values suggest a depositional duration of at least ~115 m.y. The results indicate that the ~4600-m-thick succession exposed along the Lewis thrust salient recorded an average sediment accumulation rate of ~40 m/m.y. Extrapolating this stratigraphic chronology to the 15–20-km-thick western depocenter, which includes the Prichard Formation and seismically imaged Aldridge Formation (Waterton and Altyn Formation equivalents; Cressman, 1989; van der Velden and Cook, 1994; Lydon, 2007), the average sediment accumulation rate is markedly greater at ~155 m/m.y. (Waterton to Roosville Formations). Consideration of the effects of compaction require that these accumulation rate estimates are minimum values.

Sample 16 (n = 131), a meta-bentonite from the Helena Formation at Logan Pass, Glacier NP (Moe et al., 1996), offers new chronologic control for the Belt-Purcell Supergroup. This unit has a distinct lithology and a unimodal ca. 1.4 Ga U-Pb zircon age population (Fig. 7), in contrast to most other samples (Figs. 10 and 11). Nonmagnetic, heavy mineral separates revealed two populations of zircons, with one having numerous needle-like zircon grains with crystallographic c:a axis ratios of 20–15:1, consistent with a volcanic origin (Hirtz, 2021). Evans et al. (2000) also considered it to be an altered volcanic ash, and they reported a zircon U-Pb SHRIMP age of 1454 ± 9 Ma (n = 18). However, their data displayed a spread of values from 1492 Ma to 1420 Ma, and reanalysis of their data in detritalPy yielded new MDAs of 1423 ± 8.5 Ma (YSG), 1423.7 ± 4.7 Ma (YC1σ[2+]), and 1437.5 ± 2.8 Ma (YC2σ[3+]). Therefore, based on our estimate of 1430.7 ± 3.2 Ma (YC2σ[3+]; Fig. 8) from this same unit, we consider their reported age to be ~30 m.y. older than their four youngest grains and ~23 m.y. older than our MDA assessment. Multiple studies of the Belt Basin have relied on the previous chronology (e.g., Luepke and Lyons, 2001; Link et al., 2007; Lydon, 2007; Sears, 2007b; Stewart et al., 2010), highlighting the significance of the new age interpretation. In light of these findings, further analysis is warranted for other previously dated samples, such as the Purcell Lava and Libby tuff (e.g., Evans et al., 2000), with implications for reevaluation of the Belt-Purcell Supergroup chronostratigraphic framework.

An MDS plot (final stress = 0.096) demonstrates that samples from the lower Belt Supergroup succession share internally similar age distributions and are distinct from samples of the upper succession (Fig. 10; Table S7). In lower stratigraphic levels, the Waterton to lower Helena Formations (samples 1–13) are closely clustered in MDS plots because they uniformly contain middle Paleoproterozoic (1920–1800 Ma), early Paleoproterozoic (2500–1920 Ma), and Archean (3500–2500 Ma) ages. This sample set includes nine sandstone and four mudstone samples with comparable age distributions (Fig. 7), demonstrating that contemporaneous coarse- and fine-grained detrital fractions in the Belt Basin were derived from the same sources. In upper stratigraphic levels, the Gateway to Roosville Formations (samples 19–27) plot in close proximity to each other because they consistently contain late Paleoproterozoic ages of the late Statherian (1700–1600 Ma) and early Statherian (1800–1700 Ma) periods. By contrast, in intermediate stratigraphic levels, the upper Helena to Sheppard Formations (samples 14–18) yield heterogeneous patterns in which individual samples are scattered in the MDS plots, indicative of considerable variation in sample age distributions. Furthermore, the age spectra for sample 17 resembles the lower Belt Supergroup succession, samples 14 and 15 share similar features with the upper Belt Supergroup succession, and samples 16 and 18 are markedly different (most isolated in MDS plots) due to their predominantly unimodal Mesoproterozoic ages of the late Calymmian–Ectasian (1480–1340 Ma) and early Calymmian (1600–1480 Ma) periods.

A separate MDS plot (final stress = 0.127) was generated comparing all samples from this study to potential Laurentian sediment sources, illustrating their similarity or dissimilarity (Fig. 11; Table S7). The major tectono-magmatic domains throughout the Canadian Shield have age distributions dominated by middle Paleoproterozoic (1920–1800 Ma), early Paleoproterozoic (2500–1920 Ma), and Archean (3500–2500 Ma) ages, resulting in a cluster of data points in the MDS plot. The southern provinces plot as a separate, distinct cluster in the MDS plot due to their predominantly late Paleoproterozoic (1800–1600 Ma) populations with subordinate early Mesoproterozoic (1480–1340 Ma) signatures. In this plot, lower Belt Supergroup deposits (samples 1–13) plot near the Canadian Shield domain, and the upper Belt Supergroup units (samples 19–27) plot in proximity to the southern provinces domain. The transitional middle Belt Supergroup interval (samples 14–18), on the other hand, exhibits considerable scatter, with diverse clustering with the southern provinces (samples 14 and 15), the Canadian Shield (sample 17), and A-type granite (samples 16 and 18) fields.

An inverse Monte Carlo mixture model quantifying potential contributions of Precambrian Laurentian source regions gave the five best-fit model PDPs representing the top cross-correlation test outputs from 50,000 iterations, the relative contributions (%) of 22 identified Laurentian sources, and a coefficient of determination (in which R² >0.7 represents accurately characterized sources, and R² <0.7 represents incompletely characterized sources; Fig. 12).

Significant contributions for the lower Belt Supergroup succession demonstrated derivation from distal parts of the Canadian Shield currently exposed to the northeast. At the base, zircon grains from a Waterton Formation mudstone (sample 1; R² = 0.73) appear to have been derived from the Sask craton (29%), Slave province (17%), and Churchill orogens (12%). The basal Altyn Formation sample 2 (R² = 0.82) reflects sediment sources in the Sask craton (39%), Slave province (23%), and Churchill province (20%). Altyn sample 3 (R² = 0.61) was predominantly derived from the Sask craton (63%); the highest Altyn sample 4 (R² = 0.94) contained zircons from the Sask (21%), Mojave (21%), and Yavapai-Mojave (19%) provinces. Up section, the lowest Appekunny Formation sample 5 (R² = 0.56; lowest of all 27 samples) exhibited Hearne (32%), Wyoming (21%), Sask (18%), and Churchill (11%) provenance. Appekunny sample 6 (R² = 0.76) was uniquely dominated by detritus from the Superior craton (65%), in addition to an input from Churchill (14%) sources, and sample 7 (R² = 0.85) contained sediments derived from the Sask (31%), Makkovik (15%), and Churchill (10%) regions. The mixture modeling results show that the relative proportions for the upper Appekunny to lower Grinnell Formations (samples 8–10) indicate consistent sediment sources with relatively uniform contributions. Samples 8 (R² = 0.93), 9 (R² = 0.86), and 10 (R² = 0.87) indicate Churchill province (45%, 45%, 37%) and Makkovikian (30%, 29%, 39%) source regions. Upper Grinnell sample 11 (R² = 0.86) resembles the middle Appekunny Formation (sample 7), in that it was derived from Churchill (36%), Makkovik (20%), and Sask (11%) rocks, whereas uppermost Grinnell Formation sample 12 (R² = 0.82), a mudstone, contained primarily Makkovik (24%), Sask (14%), Mazatzal (14%), and Grenville (12%) signatures. The highest sample from the lower Belt Supergroup section, from lower Helena Formation sample 13 (R² = 0.89), contained detritus from Makkovikian (47%) and Churchill (38%) regions.

The complex provenance record for the middle Belt Supergroup interval showed substantial variations between successive samples, potentially due to rapid episodes of drainage reorganization and/or competing pulses of siliciclastic input from the southern provinces, Canadian Shield, and a potentially unidentified source. The model results for the upper Helena Formation showed sustained contributions from the southern provinces. Sample 14, a mudstone collected below the Conophyton zone (Fig. 3; R² = 0.76), appears to have had sediment provenance from the Grenville orogen (41%), Mojave (18%) and Mazatzal (11%) provinces, and A-type granites (10%); sample 15 (R² = 0.83) was dominated by Mazatzal (41%), Mojave (20%), and Yavapai-Mojave (15%) signatures. Up section, cross-correlation model results for sample 16 (meta-bentonite; R² = 0.63) suggest a major contribution from the Grenville (72%) and A-type granite (21%) domains, but the low coefficient of determination indicates that the model does not fully replicate the observed age distribution, potentially indicating incomplete characterization of all sources that fed the Belt Basin and suggesting contributions from unidentified sediment sources. The upper Siyeh Formation sample 17 (R² = 0.90) showed derivation from the Canadian Shield, including the Churchill province (45%) and Sask craton (14%). A single sample from the Sheppard Formation, sample 18 (R² = 0.90), was dominated by provenance identical to the uppermost Helena Formation (Grenville, 48%; A-type granites, 37%) but with a higher confidence of model likeness.

In sharp contrast to the lower and middle levels, the upper Belt Supergroup succession showed detrital contributions from the southern provinces (Fig. 12). For the lower Gateway Formation, sample 19 (R² = 0.90) showed primary contributions from the Mojave (43%), Yavapai-Mojave (25%), and Yavapai (16%) regions, with sample 20 (R² = 0.85) showcasing the greatest degree of siliciclastic input from the Mojave crustal bock (78%). Three coarse-grained samples from the upper Gateway to lower Phillips Formations (samples 21, R² = 0.92; sample 22, R² = 0.94; and sample 23, R² = 0.94; highest coefficients of determination) showed similar influence from southern sources, including the Yavapai-Mojave transition (21%, 53%, 57%) and Mojave block (32%, 30%, 19%), with sample 21 uniquely showing a subdued influence from the Sask craton (17%). Two sandstone samples from the upper Phillips to lower Roosville Formations (sample 24, R² = 0.81; sample 25, R² = 0.86) indicated almost exclusive derivation from the Yavapai-Mojave transition (78%, 72%). In contrast, Roosville Formation sample 26 (R² = 0.90), a mudstone sample, exhibited primarily Mojave (38%), Yavapai-Mojave (25%), and Grenville (15%) distributions. Last, the stratigraphically highest example in this study, sample 27 (R² = 0.86), a mudstone from the Roosville Formation, showed a cosmopolitan age distribution with considerable Mojave (33%), Grenville (11%), and A-type granite (10%) influence.

Collectively, provenance modeling results for 24 of the 27 samples exhibited acceptable source characterizations (R² > 0.7), suggesting that Laurentian rock units identified in our source compilation served as the dominant contributors to the northeastern Belt Basin. This is consistent with the very low proportion of non–North American (ca. 1610–1490 Ma) zircon grains analyzed in this study (~5% of all analyses using open discordance filter of <20% normal and <5% reverse discordance). In two of the three samples exhibiting poor model fits (R² < 0.7), non-Laurentian grains constituted a minor component of the age distributions. Although these ca. 1610–1490 Ma grains could be considered to have a non–North American origin, recent studies suggest the possibility of such early Mesoproterozoic rocks in western Laurentia, now exposed in the southwestern and northwestern United States (Doe et al., 2013; Jones and Daniel, 2023; Skotnicki and Gruber, 2019; Holland et al., 2020; Brennan et al., 2022). Further investigation is required to pinpoint the origin of this intriguing but extremely limited detritus in the northeastern Belt Basin.

New DZ U-Pb geochronologic results for 27 samples from a thick continuous succession along the Montana-Alberta border provide new chronostratigraphic and provenance insights for the Mesoproterozoic Belt-Purcell Supergroup. MDAs along with the age of an interbedded tuff require revision of commonly accepted age assignments for the Belt Basin. Quantitative assessments of DZ provenance data suggest a continental-scale drainage reorganization, which can be linked to emerging evidence for large-scale deformation across southwestern Laurentia, manifested as the Picuris orogeny and associated tectono-magmatic processes.

MDAs from the new comprehensive U-Pb data set spanning the Belt-Purcell Supergroup constrain sedimentation to ca. 1495–1380 Ma, as supported by the absolute age of a single interbedded tuff (meta-bentonite) in the upper Helena Formation. Reanalysis of published SHRIMP data (Evans et al., 2000) yielded new MDAs of 1437.5 ± 2.8 (YC2σ[3+]), 1423.7 ± 4.7 Ma (YC1σ[2+]), and 1423 ± 8.5 Ma (YSG). These values are consistent with the MDA of 1430.7 ± 3.2 Ma (YC2σ[3+]) from our calculation for sample 16.

Our chronostratigraphic assignments are in broad agreement with assessments of the Belt succession by McMechan and Price (1982), Sears et al. (1998), Anderson and Davis (1995), Doughty and Chamberlain (1996), and McFarlane (2015), who roughly bracketed (1) a minimum age for the onset of basin accumulation from dated magmatic intrusions in the Prichard Formation (Paradise sill, 1457 ± 2 Ma; Plains sill, 1469 ± 2.5 Ma; Crossport sill, 1433 ± 10 Ma; Moyie sills, 1468 ± 2.5 Ma) that correlate to the Waterton and Altyn Formations and (2) the final stages or cessation of basin accumulation to the intrusion of the Salmon River diabase, Idaho (1379 ± 1 Ma), and the Hellroaring Creek stock, British Columbia (1365 ± 5 Ma), which are magmatic elements of the East Kootenay orogeny. Study of high-grade metamorphic and igneous rocks associated with the Hellroaring Creek stock and the Matthew Creek metamorphic zone by McFarlane (2015) led him to conclude that the East Kootenay orogeny initially involved crustal shortening and thickening ca. 1365 Ma followed by later extension ca. 1334 Ma. These events contrast markedly with rift-related Belt Basin formation and are consistent with the hypothesis that the initiation of the East Kootenay orogeny defines the end of Belt-Purcell Supergroup sedimentation. The ages of the Salmon River diabase and the Hellroaring Creek stock are contemporaneous with or postdate deposition of the Roosville Formation.

The accumulation curve for synsedimentary detrital samples demonstrates a systematic up-section reduction in age through their range of uncertainties (Fig. 9). This internal consistency demonstrates that the MDA values offer a high-fidelity assessment of depositional ages throughout the stratigraphic succession. The new chronostratigraphic framework permits a robust appreciation of long-term sedimentation (and subsidence), yielding an average sediment accumulation rate of ~40 m/m.y. for the ~4.6-km-thick section in the northeastern Belt Basin study region (extrapolated to ~155 m/m.y. for the ~20-km-thick basin depocenter) over a total basin duration of ~115 m.y. These sedimentation rates differ with those reported by Lydon (2007) and Sears (2007b), who portrayed high rates of sediment accumulation for the Lower Belt and low rates for the Missoula Group. Specific rates interpreted by Sears (2007b) are ~540 m/m.y. for the Lower Belt and ~30 m/m.y. for the Missoula Group (Fig. 13). Contrary to Sears’ (2007b) order-of-magnitude reduction in accumulation rates, our calculations suggest sustained accommodation generation at moderate to low rates throughout basin development.

The new DZ U-Pb results for 27 samples (Fig. 7) constrain depositional ages and sediment dispersal patterns during Mesoproterozoic evolution of the northeastern Belt Basin near the western edge of Laurentia. Our synthesis identified three phases of basin evolution represented by the lower (Waterton–lower Helena Formations), middle (upper Helena–Sheppard Formations), and upper segments (Gateway-Roosville Formations) of the Belt-Purcell Supergroup (Fig. 14).

The lower Belt Supergroup succession was deposited from ca. 1495 Ma to 1440 Ma with sediment dispersal from distal Laurentian sources located northeast of the basin (present coordinates; Fig. 14A). Provenance signatures indicative of the Churchill, Makkovik, and Sask domains support regional rather than local sources, which implies a continental-scale drainage network that delivered vast quantities of the predominantly fine- and very fine-grained siliciclastic material observed throughout the basin. These materials potentially traversed up to 3000 km via a Mesoproterozoic transcontinental fluvial system, possibly aided by winds. This interpretation differs from past inferences of a southwestern to western, non–North American source for the Lower Belt (Prichard Formation), as proposed for the central to southern part of the Belt Basin on the basis of isopach, paleocurrent, and DZ U-Pb data (Cressman, 1989; Ross and Villeneuve, 2003; Link et al., 2007). Most of these past results from the Lower Belt correlate with the Waterton to lower Helena Formations analyzed in this study. In addition, several authors (Ross and Villeneuve, 2003; Winston 2016; Pratt and Ponce, 2019; Pratt and Rule, 2021) have suggested that provenance of the Belt-Purcell Supergroup in the eastern side of the basin can be estimated roughly by grain size, where coarse-grained sediments of the Altyn and Grinnell Formations had a proximal North American source, and, conversely, fine-grained fractions were derived from a distal exotic domain. This hypothesis was tested by analyzing both sandstones and mudstones from these units. For the lower Belt Supergroup succession, we interpret major contributions of both coarse- and fine-grained material from shared proximal (Sask) and distal (Churchill and Makkovik) Laurentian sources to the northeast. A lack of non–North American–aged detritus in eastern exposures suggests a different sediment source for the Prichard Formation in the western depocenter. Given the intersample consistency, we infer a prolonged phase (at least ~55 m.y.) of drainage stability, where transcontinental rivers flowed from northeastern Laurentia.

The transitional middle Belt Supergroup succession recorded alternating input from the Canadian Shield and the newly contributing southern provinces. These observations lend support to the notion that the Helena Formation had multiple sources, and the lower Missoula Group represents a period of major tectonic adjustment (Harrison, 1972). We infer a continental-scale shift in sediment provenance from the Canadian Shield to the southern provinces, but this transition was punctuated by several instances of drainage instability and local basaltic volcanism lasting ~20 m.y. from ca. 1440 Ma to 1420 Ma.

The upper Belt Supergroup interval, deposited from ca. 1420 Ma to 1380 Ma, records a switch to sediment delivery almost exclusively from the Yavapai, Mazatzal, and Mojave provinces of southwestern Laurentia (Fig. 14B). This finding agrees with recent work suggesting that Missoula and Lemhi Group sediments were derived from exhumed magmatic arc assemblages and the newly recognized Picuris orogen spanning Arizona, New Mexico, and Colorado, USA, with probable continuation of the orogenic system into midcontinent regions (Daniel et al., 2013, 2023a, 2023b; Jones et al., 2015; Link et al., 2016; Mahatma et al., 2022). We propose that regional uplift along the southern Laurentian margin occurred during the Picuris orogenic event (Daniel et al., 2013; Jones and Daniel, 2023), in which basement deformation deeply exhumed previously accreted magmatic arcs of the Yavapai and Mazatzal provinces (e.g., the “Big White arc” of Link et al., 2016). These processes helped to establish a new continental-scale drainage system where transcontinental rivers flowed from southern Laurentia (lasting at least ~40 m.y.) that resulted in a major provenance shift in the Belt Basin.

The early Mesoproterozoic provenance shift at ca. 1440–1420 Ma during deposition of the upper Piegan to lower Missoula Groups (upper Helena to Sheppard Formations) indicates continental-scale drainage reorganization within ancestral North America. Specifically, the lower Belt Supergroup succession, including the Waterton to lower Helena Formations (within the Lower Belt, Ravalli, and lower Piegan Groups), was sourced predominantly from Archean and early Paleoproterozoic basement sources to the east and northeast. This pattern was disrupted during deposition of the upper Belt Supergroup succession, with a switch to southern Laurentian sources of late Paleoproterozoic and Mesoproterozoic crustal provinces that provided sediment to the upper Missoula Group (Gateway to Roosville Formations).

Our provenance reconstructions utilized the known distribution of tectono-magmatic terranes across Laurentia to support the presence of two major topographic divides from at least ca. 1495 Ma (and possibly older) to 1440 Ma (Fig. 14A). These northern and southern divides separated fluvial drainage systems that fed the Belt Basin from other drainages in northwestern and southern Laurentia. The northern divide is inferred on the basis of the PR1 unit of the lower Fifteenmile Group, Yukon (Medig et al., 2014; Medig, 2016), for which provenance considerations and a dominantly unimodal age distribution point to local derivation during rapid exhumation of batholith rocks. This suggests that sediments from other sources were hindered from reaching the lower Fifteenmile Group, consistent with a drainage divide north of the Belt Basin at ca. 1495–1440 Ma. A separate southern divide is recognized on the basis of results from the Piedra Lumbre Formation, Trampas Group, New Mexico (Daniel et al., 2013). This unit was derived from proximal sources in the southwestern United States, with no contributions from the Canadian Shield. Instead, our results show that age-equivalent strata in the Belt Basin show the opposite pattern, with material from the Canadian Shield but not the southwestern United States. Therefore, we propose the existence of a second geographic divide south of the Belt Basin.

Beginning at ca. 1420 Ma, sediment delivery to the Belt Basin changed such that the influence of the northern and southern geographic divides became negligible, and sediment was overwhelmingly supplied by southwestern Laurentia (Fig. 14B). Our data indicate that siliciclastic input from the southern provinces was unhindered during deposition of the Gateway Formation and continued for the remainder of basin accumulation. We propose that the northern and southern drainage divides were no longer barriers to sediment dispersal by ca. 1420 Ma, and the pronounced rearrangement of continental-scale drainage patterns may have persisted into the Neoproterozoic with final assembly of Rodinia during the Grenville orogeny. This interpretation is supported by evidence for sediment transport from the Grenville orogen toward the west/northwest by a transcontinental fluvial system at ca. 1000 Ma (Rainbird et al., 2017).

The early Mesoproterozoic provenance shift that lasted ~20 m.y. represents one of the few documented instances of continental-scale drainage reorganization within North America. This duration is comparable to major catchment reversals such as the Cretaceous–Paleocene ancestral Mississippi River (Blum and Pecha, 2014) and the Miocene Amazon River (Hoorn et al., 1995), and it highlights the likelihood that continental-scale drainage reorganizations are gradual rather than instantaneous (Parker and Hendrix, 2022).

The pronounced provenance switch may have been linked to regional tectonic processes. Following the pronounced shift from east and northeast to southern sources at ca. 1440 Ma, the uniformity of DZ age distributions within the upper part of the Belt Supergroup succession indicates the persistence of the newly established drainage patterns throughout the continued basin filling. These changes in provenance and drainage conditions suggest that regional deformation generated and sustained positive topography for tens of millions of years.

The provenance shift involved the disappearance of Archean and appearance of dominantly Paleoproterozoic to early Mesoproterozoic zircons source regions to the south and southeast of the Belt Basin, pointing to construction of positive topography across a potentially wide swath of southern Laurentia. Because sedimentation in the Belt Basin generally postdates widespread magmatism in southern Laurentia, rock uplift and exhumation were likely due to tectonic rather than magmatic processes during the Mesoproterozoic.

A series of arc terranes composing the Yavapai and Mazatzal crustal provinces provided siliciclastic material from igneous sources during late-stage evolution of the Belt Basin. For example, the Lemhi subbasin (Missoula Group) likely received sediment from the Big White arc, an exhumed accretionary magmatic arc and batholith system of the Yavapai-Mazatzal terranes with a unimodal ca. 1740–1710 Ma zircon population (Link et al., 2016). However, while our upper Belt Supergroup succession samples also exhibit high concentrations of ca. 1800–1650 Ma DZ ages, accretion and orogenesis related to the Big White arc occurred during the late Paleoproterozoic and were therefore not drivers of the continental catchment reversal we interpret at ca. 1440–1420 Ma.

Rather than magmatism, the principal driver of topographic growth in southern Laurentia is interpreted to have involved crustal deformation and metamorphism associated with the Picuris orogeny, which affected the southwestern United States to midcontinent region. This event generated high-grade (amphibolite-facies) metamorphic conditions affiliated with regional contraction and syndeformational sedimentation between ca. 1500 Ma and 1350 Ma (Doe et al, 2012; Daniel et al., 2013, 2023a; Holland et al., 2016; Aronoff et al., 2016; Doe and Daniel, 2019; Kuiper et al., 2022). We propose that development of the Picuris orogen affected the evolution of the Belt Basin by triggering wholesale continental drainage reorganization, recorded by the shift from northeastern Laurentian to southern Laurentian source regions. Sustained topographic growth within the Picuris orogenic belt likely provided a persistent source of southern Laurentian detritus. We speculate that this Mesoproterozoic episode may have similarly affected other continental-scale patterns of exhumation, erosion, and basin evolution across Laurentia.

The Belt Basin of northwestern United States and southwestern Canada contains the longest and most-complete record of Mesoproterozoic sedimentary processes in North America. Based on 27 new samples that yielded 2593 concordant DZ U-Pb ages, this study presents a large geochronologic data set for northeastern exposures of the Belt Basin that refines the chronostratigraphic and provenance framework for the Belt-Purcell Supergroup. We conclude that deposition in the Belt Basin spanned ~115 m.y., from ca. 1495 Ma to 1380 Ma, with an average sediment accumulation rate of ~40 m/m.y. (extrapolated to ~155 m/m.y. in the basin depocenter). A volcanic tuff from the upper Helena Formation provides new absolute age control of 1430.7 ± 3.2 Ma. Comparison of the new DZ U-Pb data to previously reported ages for 22 major Precambrian Laurentian basement provinces (compiled from 150 publications encompassing 15,474 igneous and metamorphic zircon U-Pb ages) supports a new paleogeographic reconstruction of regional drainage patterns. Key features are as follows: (1) Sediment derivation for the Waterton to lower Helena Formations from the Canadian Shield (Churchill, Sask, and Makkovik regions) was succeeded by sediment delivery to the Gateway to Roosville Formations from the southern provinces (Mojave and Yavapai-Mojave terranes); (2) two contrasting transcontinental river systems and two Calymmian geographic divides formed key components of the early Mesoproterozoic paleotopography of Laurentia; (3) Laurentia experienced a continental-scale drainage reorganization at ca. 1440–1420 Ma during deposition of the upper Helena to Sheppard Formations; and (4) this major change in drainage patterns was likely driven by topographic modifications due to crustal deformation and uplift associated with the Picuris orogeny in southwestern Laurentia.

This study was supported by a student research grant from the American Association of Petroleum Geologists. We are grateful to Leona Constenius for logistical support, Jorge Gomez for meticulous mineral separation work, and to the staff of Waterton Lakes NP, Glacier NP, and Castle Wildland PP for issuing research and collection permits (to B.R. Pratt). Discussions with Sam Johnstone, Tomas Capaldi, Mike Doe, Glenn Sharman, and Kurt Sundell improved our provenance assessments. We appreciate the thorough and thoughtful comments and reviews by Chris Spencer, Stuart Parker, Chance Ronemus, Joel Saylor, and Andrew Zuza.