Bulk samples were collected from six sites during 2016 and 2017 where stratigraphical sections were also logged, forming a west–east transect from John’s Canyon to Comb Wash (Fig. 1). The samples were processed in a buffered 10% solution of acetic acid by one of us (C.M.H.) at the University of Calgary, and results were only briefly mentioned by Huttenlocker et al. (2018, 74). Samples (numbered 1349-1 through 1349-8 in Table 1) were recovered from marine carbonate marker beds distributed throughout the lower Cutler beds, spanning approximately 160 m of strata from the McKim limestone to the first major aeolian beds of the Cedar Mesa Sandstone Formation. Several prominent bench-forming limestones have been recognized as informal units in these lower Cutler beds, the first of which represents the uppermost unit sampled by Ritter et al. (2002): (1) the former ‘Shafer limestone’ exposed in the middle Rico Formation at the head of Honaker Trail, and which is not equivalent to the Shafer limestone of the Canyonlands and Moab districts (see O'Sullivan 1965; ‘Unnamed Bed’ of Soreghan et al. 2002a, fig. 4); (2) the ‘McKim limestone’ of Wengerd (1950), a thick (~1–6 m), fossiliferous bed whose top forms the Rico–Halgaito contact in John’s Canyon, Cedar Point and the central portion of Valley of the Gods; and (3) the ‘A limestone’ of O'Sullivan (1965), a thin (~1 m or less) but resistant capping unit that forms the ‘dome’ of Lime Ridge, and whose top forms the uppermost Rico–Halgaito contact east of Valley of the Gods and in the Lime Ridge area. These minor beds represent periodic, short-lived transgressions that occurred within a longer-term glacioeustatic-controlled regression that spanned into earliest Permian times. As such, the Rico–Halgaito transition is time transgressive, so that the upper contact of the Rico Formation west of W109° 50’ (top of the McKim limestone) is older than it is towards the northeast (top of the ‘A’ limestone) (Orkild 1955) (see Fig. 1).

A precise chronostratigraphy for the lower Cutler beds is hindered by few stratigraphical age controls. For example, magnetostratigraphy is hindered within the C–P interval by a long-lived period of reversed polarity, the Kiaman Superchron (Gose & Helsley 1972). Thus, micro-fossil indices from the marine marker beds may provide vital constraints on the age of the Rico–Halgaito transition. Whereas Baars (1962; Baars et al., 1967) suggested an unconformity between the Missourian and Wolfcampian strata, Loope et al. (1990) called this unconformity into question and Sanderson & Verville (1990) reported a succession of Missourian, Virgilian and Wolfcampian fusulinids from the lower Cutler beds of Emery County. However, the fusulinid assemblage of Sanderson & Verville (1990) was recovered from a core taken well outside our study area. The conodont sequence stratigraphical framework of Ritter et al. (2002) correlated the former ‘Shafer limestone’ of the Glen Canyon Recreation Area to the South Bend limestone (Lansing Group) of the North American Midcontinent at the Missourian–Virgilian stage boundary. This assignment was based primarily on the deep-water conodonts Streptognathodus pawhuskensis and S. firmus. The fossil vertebrate localities in the vicinity of Valley of the Gods occur mainly within the Halgaito Formation, but their associations with bounding marine marker beds make broader correlations possible. Scott (2013) considered the McKim limestone late Virgilian, and the ‘A’ limestone as either latest Virgilian or earliest Wolfcampian, reporting the xenacanthid Orthacanthus texensis from just beneath the level of the ‘A’ limestone at Castle Butte. The Birthday bonebed – a multitaxic bonebed positioned in the Halgaito Formation above the McKim but below the level of the ‘A’ limestone (Fig. 1 – includes taxa known from the Cobrean and Coyotean land vertebrate faunachrons (LVFs) of New Mexico and was interpreted as late Pennsylvanian (Virgilian) (Sumida et al. 1999a, 1999b, 1999c; Lucas 2006, 2018; Huttenlocker et al. 2018). Finally, Soreghan et al. (2002a, 2002b) reported a single detrital zircon age from loessites in the Halgaito Formation, but the grains were low-yield and were not analysed using methods that minimize effects of lead-loss, producing an unexpectedly young age of 283 ± 4 Ma (boundary interval between Artinskian–Kungurian global stages). Thus, portions of the lower Cutler beds may be as old as the Missourian and perhaps as young as the Leonardian North American stage. Overall, the conodont succession recovered here suggests a maximum age range from upper Gzhelian to mid-Asselian (ca. 300 to 296 Ma).

Of the major facies belts distributed throughout the Cutler Group, our study interval is centred on the southwestern carbonate platform – represented by the Rico Formation – and a minor clastic sediment wedge at its southern border that drained towards the north – represented by the Halgaito Formation. The carbonate systems can be further sub-divided into local lithofacies (listed in Table 1) on the basis of their matrix, skeletal grains and fabrics. These generally included: (1) skeletal wackestone to packstone reflecting shallow reef to offshore marine facies; and, (2) grainstone or microbial bindstone, generally associated with back-reef lagoon to intertidal settings (nearshore). A third carbonate lithofacies that is more common in the Cedar Mesa Sandstone is represented by the thin, brecciated, non-fossiliferous limestones reflecting freshwater or playa lake settings (Mountney & Jagger 2004), but these are not represented in our samples. Likewise, carbonate-cemented sandstones are also common both in marginal marine facies of the Rico Formation and in floodplain facies of the Halgaito Formation. In stream channels and estuaries, environmentally informative micro-fossils have also been encountered, including tiny microconchid tubeworm shells (~ 3 mm diameter) and teeth of small xenacanthid chondrichthyans, which have been interpreted as brackish- to freshwater-dwelling based on their facies associations and strontium isotope compositions (Huttenlocker et al. 2018).

Compared to the lower horizons analysed by Ritter et al. (2002), relatively few fossils are found in the shallow-water carbonates of the upper Rico–Halgaito Formation transition. However, echinoid spines are characteristically abundant in the McKim limestone at multiple localities (John’s Canyon, Valley of the Gods), as well as brachiopods and gastropods. In Valley of the Gods, along the base of West Lime Creek, the McKim limestone crops out as a skeletal wackestone to packstone with abundant gastropod accumulations recording evidence of monsoonal, shallow-water storm deposits (e.g. echinoderm- and gastropod-dominated ‘proximal tempestites’; Samira et al. 2018). In addition to the echinoid spines, field observations of fossils have included the Carboniferous–Permian brachiopod Hystriculina cf. H. wabashensis and the bellerophontid gastropod Euphemites cf. E. graffhami, both reported previously by Scott (2013).

Micro-fossil samples were collected from the McKim limestone in John’s Canyon and in the west side of Valley of the Gods near Moki Dugway where the top of this limestone forms the Rico–Halgaito contact (Fig. 1). Conodont elements are represented by mostly shallow-water taxa, including Ellisonia, Hindeodus and Adetognathus, of which Ellisonia appears to have been the most abundant (Figs 1, 2; Table 1). At least 15 ramiform elements of Ellisonia conflexa (von Bitter & Merrill 1983) were produced from a sample taken from northeast of Cedar Point along the Valley of the Gods loop road adjacent to the Moki Dugway. Ellisonia conflexa is a nearshore conodont best known from the Late Carboniferous of southwestern North America (e.g. Desmoinesian to Virgilian stages of New Mexico; von Bitter & Merrill 1983; Armstrong et al. 1994; Krainer et al. 2003; Krainer & Lucas 2013). However, the species also ranges into the Wolfcampian of North American stages and to the upper Artinskian in the Canadian Arctic (Henderson 1989; Wamsteeker 2009) and Svalbard (Nakrem 1991). Its rarity in Wolfcampian deposits in the American southwest may be due to poor sampling effort in Permian-aged rocks in this region, given that prior studies have focused almost entirely on the carbonate-rich Carboniferous units (e.g. Armstrong et al. 1994; Krainer et al. 2003; Orchard et al. 2004; Barrick et al. 2013; Krainer & Lucas 2013). Hindeodus elements, not diagnostic at the species level at this time, were also encountered, but are similarly limited in their stratigraphical utility due to the long temporal range of the genus (Carboniferous to lowermost Triassic). Finally, Adetognathus elements (most of which are not identifiable to species) occurred in similar shallow-water environments (Clark 1974; Davis & Webster 1985) and have a long stratigraphical range (lower Pennsylvanian to Sakmarian of the Lower Permian). Several formal and informal species were used to develop a shallow-water zonation for the Moscovian to Sakmarian of the Canadian Arctic where fusulinids and small benthic foraminifers also constrained ages (Henderson et al. 1995; Pinard & Mamet 1998). The zonation was confirmed elsewhere in the Canadian Arctic (Wamsteeker 2009; Beauchamp et al. 2020), the Russian Arctic (Sobolev & Nakrem 1996) and in east-central British Columbia (Zubin-Stathopoulos et al. 2013). Species within this zonation have included Adetognathus lautus (Bashkirian to lowermost Gzhelian), Adetognathus sp. A (Gzhelian; Henderson et al. 1995; anterior inner parapet exhibits flexure), Adetognathus sp. B (uppermost Gzhelian to lower Asselian; Henderson et al. 1995; anterior inner parapet is thickened and smooth in adult forms), Adetognathus sp. C (Asselian; Henderson 1989; free-blade and outer parapet are disjunct) and Adetognathus paralautus (uppermost Asselian to Sakmarian; Orchard & Forster 1988; Henderson 1989; wide flattened parapets with numerous transverse ridges). Though not sampled in Ritter et al.’s (2002) study of the underlying ‘Rico’ beds, we show that the McKim limestone marks the local first occurrence of Adetognathus sp. B (Fig. 1; Table 1).

Adetognathus sp. B (Henderson 1989) is also found in the highest shallow marine limestones locally, including the ‘A’ limestone. The ‘A’ limestone is fossiliferous only in a few areas, particularly where it pinches out in the Valley of the Gods and in isolated areas on Lime Ridge farther to the northeast. The few macrofossils suggest a bivalve- (Aviculopecten) and gastropod-dominated fauna (O’Sullivan 1965) with sparse fenestrate bryozoans and occasional fishes such as the ctenacanthiform Glikmanius (CM 90266) and palaeoniscoid actinopterygians. None of these fossils provide useful age constraints. Sedimentary features of the ‘A’ limestone include laminations and microbialite fabrics typical of autochthonous bindstone and, together with the fauna, likely reflect sub-tidal to intertidal environments. Based on a personal communication from E. L. Yochelson, O’Sullivan (1965:30) reported: ‘The abundance of shell fragments and the water-worn condition of many of them implies shallow-water and suggests that this may have been a [near shore] accumulation … Abundance of mollusks and the absence of all other fossil groups except a small fragment of a bryozoan might indicate deviation from normal marine salinity.’

Samples collected in the east side of Valley of the Gods (Setting Hen Butte) and Lime Ridge yielded a few micro-fossils, represented by Ellisonia, Hindeodus and Adetognathus (Figs 1, 2; Table 1). Ramiform elements of Ellisonia conflexa were most common (more than 13 elements), and rare platform elements referable to Adetognathus sp. and, notably, to Adetognathus sp. B were also encountered. Collectively, this conodont fauna is indicative of warm, shallow-water biofacies (Driese et al. 1984) and possibly fluctuating salinity conditions (Hermann et al. 2015). Adetognathus sp. B, as in the McKim, indicates an upper Gzhelian to lower Asselian age. High-resolution zonation of this time interval is defined by species of the genus Streptognathodus (Henderson 2018), which is normally found in more offshore biofacies, lacking in this region. Biofacies play a key role in the recognition of conodont biozones. Conodont biofacies boundaries are fuzzy as shallow-water taxa were abundant in near shore settings, but increasingly rare or absent in offshore, deeper water settings.

To the north of our sample, between Comb Wash and Arch Canyon where the Halgaito Formation grades into predominantly marine carbonates of the former ‘Elephant Canyon Formation’ (Baars 1962, 1967), the laterally equivalent facies of the ‘A’ limestone contain more abundant fossils that reflect offshore marine and shallow reef biofacies. This is consistent with the interpretation that the Halgaito drainage was towards the north (Baars 1962; Vaughn 1962, 1967). Fossils from the shallow-water facies in the Arch Canyon area include brachiopods, fenestrate bryozoans, crinoids and disarticulated elements of small vertebrates (including chondrichthyans, actinopterygians, rhipidistians and the aïstopod Phlegethontia) (Vaughn 1967; Sumida et al. 1999a, 1999b). Micro-fossils include only sparse fusulinaceans and the conodont Adetognathus sp. B. The record of Adetognathus sp. B in Arch Canyon further supports that these facies are laterally equivalent to the uppermost Rico–Halgaito transition in the south and that this conodont may be widespread in both nearshore sub-tidal/intertidal and some offshore reef facies (Wamsteeker 2009).

Another direction for increased chronostratigraphical resolution involves astronomical tuning of the rock record in combination with radiometric dates (Schmitz & Davydov 2012). Beauchamp et al. (2020) recognize numerous long eccentricity 405 Kyr cyclothems from Gzhelian to Asselian in the Canadian Arctic on the basis of carbonate sedimentology and sequence stratigraphy and link these with conodont temporal ranges. For example, the range of Adetognathus sp. B can be tuned to 300 to 297.7 Ma in the Canadian Arctic. Ritter (1995) shows 47 shale-limestone rhythms from the Eudora Shale (approximately base-Gzhelian) to Bennett Shale (base-Permian). The Gzhelian is 4.8 Myrs in duration (Fig. 3) suggesting that these rhythms represent 100 Kyrs short eccentricity cycles. Preliminary tuned ages can be provided for some key taxa in the mid-west succession including Streptognathodus virgilicus (302.3 to 299.6 Ma) and S. pawhuskensis (303.7 to 299.6 Ma); both of these taxa are immediately overlain by S. brownvillensis (299.6 to 299.3 Ma). This suggests that any samples in which these Streptognathodus species co-occur with Adetognathus sp. B, must be very close to the FAD of the latter species.

The conodont records further constrain ages assigned to the major marker beds in the lower portion of the Cutler Group (Fig. 3). Prior work has identified a maximum age of the middle Rico Formation by placing the Missourian–Virgilian stage boundary at the former ‘Shafer limestone’ (Ritter et al. 2002; ‘Unnamed Bed’ of Soreghan et al. 2002a, fig. 4), correlative to the South Bend limestone (Lansing Group) of the Midcontinent. Preliminary sampling further up section by Huttenlocker et al. (2018) targeted the McKim and ‘A’ limestones and suggested minimum ages of latest Pennsylvanian (Virgilian) for these units on the basis of Ellisonia and Adetognathus. Elsewhere in the Four Corners region, these taxa have been reported from the Virgilian Atrasado Formation and lower Red Tanks Member of the Bursum Formation, central New Mexico (Orchard et al. 2004; Barrick et al. 2013; Krainer & Lucas 2013; Lucas et al. 2013), as well as the Oso Ridge Member underlying the Abo Formation, Zuni Mountains, west-central New Mexico (Armstrong et al. 1994; Krainer et al. 2003). Orchard et al. (2004) illustrated Streptognathodus cf. virgilicus and S. pawhuskaensis deflectus as well as a transitional form of Adetognathus sp. A (their fig. 2.11-12) in the upper Atrasado Formation and correlate to mid-Virgilian; the cycle tuned age based on the Streptognathodus species (Ritter 1995) is 302.4–302.1 Ma. Chernykh (2005) described a comparable fauna in Bed 13 (mid-Gzhelian) of the Usolka section in the Urals of Russia; the interpolated radiometric age (Schmitz & Davydov 2012) is 301.2 Ma. A specimen of Adetognathus sp. B comparable to our material was reported from the Red Tanks Member (Orchard et al. 2004; fig. 2.6-7) below a second Streptognathodus horizon that included Streptognathodus cf. S. variabilis (their fig. 2.1-2) and S. cf. S. bellus (their fig. 2.3-4). Chernykh (2005) describes a comparable fauna from Usolka section Bed 15 (uppermost Gzhelian), which has an interpolated radiometric age (Schmitz & Davydov 2012) of 299.5. Therefore, the Red Tanks occurrence of Adetognathus sp. B is upper Gzhelian; however, the taxon is also recognized higher in the Asselian of Arctic and western Canada where it is more common (Henderson 1989; Zubin-Stathopoulos et al. 2013; Beauchamp et al. 2020). Consistent with the newly reported conodonts, the vertebrate assemblage that is bracketed by the McKim and ‘A’ limestones also supports a Virgilian age, including Cobrean and Coyotean LVF taxa: Sagenodus copeanus, basal diadectomorphs and the synapsids Ophiacodon navajovicus and Sphenacodon, among others (Huttenlocker et al. 2018, figs 15,16). The Coyotean LVF, whose base is defined by the first appearance of the abundant Sphenacodon, spans the C–P transition in the Four Corners region (Lucas 2006, 2018).

The ‘A’ limestone, which forms the uppermost bed of the Rico Formation east of W109° 50ʹ (Orkild 1955) includes marine vertebrates (palaeoniscoids, the ctenacanthiform Glikmanius) and invertebrate macro- and micro-fossils indicative of a C–P age. Moreover, Adetognathus sp. B is also found in this unit. The first appearance of Adetognathus sp. B is upper Gzhelian and is common in the Lower Permian in Siberia and the Sverdrup Basin of the Canadian Arctic (Fig. 3). The lack of Adetognathus sp. A in these samples could support an Early Permian age, but additional samples with more specimens would be necessary to confirm this absence as evidence. Therefore, it is reasonable to interpret that the Carboniferous–Permian boundary is found within the range of Adetognathus sp. B, which occurs from the McKim to the ‘A’ limestone. It is also consistent with the hypothesis that the C–P boundary is placed in the middle-to-upper Halgaito Formation as suggested by Scott (2013) (most likely above the level of the Birthday bonebed near or encompassing the ‘A’ limestone). The co-occurrences of Ellisonia conflexa, Hindeodus and Adetognathus sp. B within the ‘A’ limestone correspond to Sverdrup Basin P3 zone of Henderson (1989), which is now interpreted to span the C-P boundary (S. longus-A. sp. B to lower S. constrictus zones; Fig. 3).