Abstract
Metoposaurid-dominated bonebeds are relatively commonplace in Upper Triassic continental deposits with at least ten monodominant, densely populated bonebeds globally. The biostratinomy of several classic localities in India, North America, and Poland have been explored in detail, however, variability in methods and resultant conclusions point to the need for a more nuanced approach to understanding both the taphonomic and the ecological origins of metoposaurid-dominated bonebeds. Here we present the first monodominant metoposaurid mass mortality assemblage from the Late Triassic Popo Agie Formation and the stratigraphically lowest known record of several fauna from the Popo Agie Fm including the first occurrence of the Buettnererpeton bakeri in Wyoming. We employ previously used binning methods based on perceived hydrodynamic equivalence (“Voorhies groups”) to assess pre-burial skeletal sorting. We suggest a simple counting and normalization method that avoids the inherent bias introduced by the interpretation of hydrodynamic equivalence of skeletal elements in taxa that lack actualistic experimental data. In contrast to other North American metoposaurid bonebeds, the sedimentology and skeletal sorting analyses of the Nobby Knob quarry support an autochthonous origin of this assemblage in a fluvio-lacustrine system with relatively little pre-burial sorting. Despite differences in underlying assumptions regarding the dispersal potential of specific skeletal elements, binning methods tend to follow similar trends regardless of framework used to assess different assemblages.
Introduction
Metoposaurid stereospondyls are common in mid-low latitude Laurasian and mid latitude east Gondwanan non-marine Upper Triassic deposits. They are known from innumerable single occurrence localities and no less than ten bonebeds (Fig 1), and the taphonomy of several of these metoposaurid-dominated bonebeds has been explored in detail. Romer (1) provided the first interpretation of a monodominant metoposaurid bonebed–known as the Lamy amphibian quarry–as a “drying pond” scenario. Subsequent workers presented evidence of transport of the skeletons from the site of death and suggested that the Lamy amphibian quarry is an allochthonous assemblage that resulted from overbank flooding (2–4). Metoposaurid bonebeds range from hydrodynamically sorted and nearly complete disarticulation of skeletal elements (2,3,5–7), to uncommon partial articulation dominated by disarticulated remains (8,9), to fully articulated skeletons (10). Differences between individual metoposaurid size within these bonebeds varies from narrow (5,7,10–12) to broad size ranges (10,13,14). The taxonomic composition of these bonebeds varies from monotaxic (5,7), to monodominant (2,3,10,11), to multitaxic (8,13,15). Some metoposaurid-bearing bonebeds cannot be distinguished between a time-averaged accumulation (e.g. Krasiejów clay pit, Rotten Hill bonebed, and Lamy amphibian quarry: 2,3,8,13) and those considered to represent mass mortality events (e.g., Site XIII; 10). Given the variability between modes of deposition and taxonomic composition, a nuanced understanding of the biostratinomic processes involved in the formation of metoposaurid-dominated bonebeds is required to interrogate the paleobiology of fossil taxa with no modern analog.
Methods previously used to investigate the biostratinomy of metoposaurid-dominated bonebeds can be summarized into three categories: (1) paleocurrent indicators or proxies, (2) subaerial exposure indicators, and (3) skeletal sorting. Paleocurrent is typically measured through the alignment of primary sedimentary structures (e.g., ripple marks, imbrication, flute casts; see 16 for a summary). However, in the absence of primary sedimentary structures, some workers have used the hydrodynamic preference of long bones to orient parallel to current direction to infer the presence and direction of flow (17; but see 18 for complications). Proxies for subaerial exposure can range from mudcracks in sediment to splintering or cracking of cortical bone induced by desiccation (5,19). Subaerial or subaqueous exposure can also be indirectly inferred through the presence of bioglyphs such as toothmarks of terrestrial or aquatic consumers (20). Disarticulated vertebrate skeletons have been used to infer the degree of skeletal sorting prior to final burial (e.g. 2,5,13,21,22) similar to assessments of winnowing in sediments (23,24), microfossils (25), and shell-beds (26) inferred by distributions of grain size. Actualistic flume experiments with mammal skeletons allowed skeletal elements to be categorized into early, intermediate, and late dispersal groups (22). Subsequently, workers have modified the group assignments of Voorhies to assess the skeletal sorting of temnospondyl assemblages (2,13,27). Flume experiments with non-mammal skeletons have demonstrated a continuum of dispersal potential that aligns with complexity of form and density and not strictly with homology as expected given the diversity and disparity of vertebrate skeletons (see 28). In the absence of actualistic flume experiments for temnospondyls, the a priori assignment of temnospondyl skeletal elements to “Voorhies groups” is subjective and must be reassessed in an empirical framework of skeletal completeness based on known assemblages (e.g. 29).
Highly fossiliferous bonebeds are widespread in non-marine Upper Triassic localities of North America (1,6,13,30–35), India (5), Morocco (10), Europe (11,36), and South America (37,38). All previously reported bonebeds from non-marine Upper Triassic strata of the western United States have been restricted to the Dockum Group and the Chinle Formation (2,13,15,30,33,34,39). In contrast, vertebrate remains from the upper Chugwater Group (i.e., Crow Mountain, Jelm, and Popo Agie formations sensu 40) are only known from isolated individuals, a few co-occurring conspecifics, or limited, disarticulated remains of a few taxa recovered from a single locality (see Table 1). Most published vertebrate fossils from the Popo Agie Formation are restricted to the ocher unit of the upper Popo Agie Fm and lower carbonate unit of the lower Popo Agie Fm (40) with only fragmentary remains published from the intervening purple unit (Table 1). Recent radioisotopic detrital zircon ages (Fig 1A) from the upper Popo Agie Fm place the purple unit within the Carnian age of the Late Triassic which, along with the vertebrate fossil occurrences therein, provides a window into non-marine Carnian vertebrate communities of the western United States that are otherwise unknown or ambiguous in the south and southwest USA (40).
*Minimum number of individuals determined based on personal observation. **See supplemental information for reasoning behind renaming this site. ***Dolichobrachium has not been formally considered a nomen dubium. Instead, it has not been considered in the evaluation of “rauisuchians” since at least 1985 due to either being missing or disposed of due to damage beyond repair.
Here we report on the highly fossiliferous site Nobby Knob from the lowermost purple unit of the Popo Agie Formation that contains a bonebed we infer to be a monodominant metoposaurid mass mortality assemblage (Fig 1). We present a biostratinomic analysis of the assemblage and compare it with two other metoposaurid-dominated bonebeds: the Elkins Place bonebed, a monotaxic metoposaurid bonebed with predominantly disarticulated remains from the lower Dockum Group, and Site XIII, a monodominant metoposaurid bonebed with articulated skeletons from the Timezgadiouine Formation in Morocco. In order to avoid binning biases based on perceived hydrodynamic equivalency or homology rather than actualistic experimental data, we use a simple counting and normalization method that can be used to assess skeletal sorting in historic collections lacking detailed site mapping data.
(A) A generalized stratigraphic column of the upper Chugwater Group sensu (40) (triangles = CA-ID-TIMS detrital zircon ages [Lovelace et al., in press]). (B) Geographic setting; first inset = areal extent of Chugwater Group, second inset = geological map (colors match the GTS) with the Nobby Knob locality denoted (red star). (C) A stratigraphic column of the local exposure of the Popo Agie Formation at the Nobby Knob locality. (D) Paleogeographic map with regional basins that contain Late Triassic metoposaurid-bearing bonebeds (* = monodominant metoposaurid bonebeds). Green portions of the map have more annual precipitation relative to tan (modified after 62). Abbreviations: Ar, Archosauromorpha indet., Bb, Buettnererpeton bakeri, Br, Brachychirotherium, Nd, Ninumbeehan dookoodukah, Ph, Phytosauria indet., Rd, Redfieldiidae indet.
Nomenclatural Note: The historic sites “Sq*** Creek” and “Between Baldwin Creek and Sq*** Creek” (47) were named after a nearby creek that has since been renamed due to the derogatory nature of the original name. Places using the same derogatory term on federal land in Wyoming have undergone name changes since 2021 (Secretary of the Interior, Order No. 3404, 2021). The aforementioned creek was renamed to Popo Agie Creek, so we propose modifying the name of these sites in order to not maintain the use of this derogatory term. The two modified site names are: “Popo Agie Creek” to refer to the type locality of “Paleorhinus” bransoni and “Between Baldwin Creek and Popo Agie Creek” to refer to the type locality of Angistorhinus grandis.
Institutional Abbreviations: UWGM, University of Wisconsin Geology Museum, Madison, WI, USA.
Materials and methods
Stratigraphy
Three partial, but overlapping, trenched stratigraphic sections were measured in 2018 using a modified Jacob’s staff, a compass, a laser pointer, and an early version of the iOS app “Jake for stratigraphy” (63). Sites were chosen based on quality of exposure, distance from the Nobby Knob quarry, and accessibility (including a section measured directly through the NK quarry). Each section’s composition was observed and recorded, including grain size, sedimentary structures, color (64), pedogenic features, and body/trace fossil occurrences. Based on characteristic compositions, stratigraphic bodies were grouped into several facies associations described in detail elsewhere (65).
Specimen accessibility, preparation, and photography
During the summer of 2014, a field party from the University of Wisconsin Geology Museum (UWGM) collected the first in situ remains from the purple unit of the lower Popo Agie Formation (Chugwater Group) in Fremont Cty, WY (Fig 1) from a site named Nobby Knob (NK) referencing both the fictional Discworld character “Nobby” Nobbs as well as the knob-like topography of the outcrop. During subsequent field seasons, an increased effort to excavate the NK locality proceeded with excavations in 2016, 2018, and 2019. The NK fossils were prepared with contemporary methods of fossil preparation such as pneumatic oscillating air scribes and air abrasion with sodium bicarbonate. The fossils were consolidated and repaired when necessary primarily with paraloid B-72 and in some cases with Butvar B-98 if especially friable. Specimens were photographed with either a DSLR camera or a Dino-Lite Edge 3.0 USB Digital Microscope. Photographs were adjusted for color correction and focus stacking when necessary with Adobe Lightroom and Adobe Photoshop, respectively, and figures were compiled with Adobe Illustrator. A complete list of specimens prepared as of December 2024 is available in S1 Table.
Specimens from the Elkins Place (EP) bonebed were observed and counted firsthand for comparison due to clear skeletal sorting (see below) and sediment winnowing (6). The EP locality is situated in the Camp Springs Conglomerate of Texas, a high energy depositional system of the basal Dockum Group (7,66). These specimens are reposited in the University of Michigan Museum of Natural History Paleontology Collection (S2 Table). Specimens from Site XIII were counted via published photographic plates (10) and are reposited in the Muséum National d’Histoire Naturelle collection (S3 Table).
Taphonomy
A preliminary analysis of the taphonomy of the NK locality demonstrated little to no transport of skeletal elements in the NK locality (67), but here a more detailed analysis is undertaken. Three criteria are considered to evaluate hydrodynamic influence on the NK locality: (1) sediment grain size and distribution, (2) sedimentary structures or long bone orientation indicative of flow direction or regime, (3) hydrodynamic sorting of skeletal elements.
Metoposaurid remains from the NK, EP, and Site XIII localities were identified and categorized into Voorhies groups following previous work (Table 2) based on the inferred surface area:volume ratio of temnospondyl remains contrasted with the mammalian remains originally tested (22). To assess the validity of these Voorhies group assignments, skeletal elements were counted, categorized into bins based on anatomical identification, and counts of elements were normalized to the expected number of skeletal elements based on the minimum number of individuals (MNI) for all localities. Due to a lack of complete articulated skeletons of North American metoposaurids, skeletal content was based on articulated specimens of Dutuitosaurus (10) and partially articulated specimens of Metoposaurus krasiejowensis (9,14). The presented method assumes similar bone density across all skeletal elements which is rarely the case for air-breathing aquatic tetrapods (68), but in the absence of actualistic flume experiments for temnospondyls, we consider it sufficient for comparison with previous work.
Original “Voorhies groups” (dispersal potential groups: 1=early; 2=intermediate; 3=late) of mammal skeletons (22), modifications for metoposaurid temnospondyls (2,3,13), modifications for capitosaurid temnospondyls (27), and modifications for metoposaurid temnospondyls (21). Numbers separated by a forward slash were considered intermediate between dispersal potential groups.
*Mandibles dispersed later if articulated in Voorhies’ experiments (22). **Implicitly included as part of a larger element (22).
Five portions of the NK locality in field jackets (NK16 J3 2-3, NK16 J8 1-2 [PL-011519], NK16 J10 3-5 [PL-015], NK18-D1-000 [NKAC], and NK19-C2-709.3 [NKAD]) were used to gather spatial data (S1–S6 Figs in S1 File). A quarry map was made for metoposaurid fossils collected in field jackets. Because elongate skeletal elements (N=207, here defined as bones with length ≥ 4 times width following 2) tend to orient with the long axis parallel to the direction of flow if a current is present during deposition (17), the azimuth of the long axis of each elongate, disarticulated element was measured. For all elements, the azimuth is bidirectional because it cannot be determined if either end of each bone would trend downstream or upstream. A Hermans-Rasson test and a Pycke test were performed to test for a multimodal distribution of the orientations of long bones using the functions “HR_test” and “pycke_test”, respectively, in the R package CircMLE version 0.3.0 (69,70).
Results
Sedimentology and stratigraphy
Triassic stratigraphy of the upper Chugwater Group, Wyoming, includes the Jelm Formation which is unconformably(?) overlain by the mid-late Carnian-aged Popo Agie Formation. There is a roughly 25 Ma hiatus between the Popo Agie Formation and the overlying Bell Springs Formation (Upper Triassic), or more depending on regional stratigraphy (e.g., Jurassic-aged Gypsum Springs, Nugget Sandstone, or Sundance formations; 40,65). The stratigraphy of the Popo Agie Formation is best known from the Lander area (65,71) where the lower carbonate, purple, ocher, and (locally) upper carbonate are well represented, but outcrops are visible along the entirety of the western slope of the Wind River Range.
The larger scale stratigraphic architecture and geological context of the lower Popo Agie Formation between Lander and Dubois, WY, USA was described by Deckman and colleagues (65) where the authors identified 15 discrete lithofacies. Four of these facies are present in the lower Popo Agie strata associated with the Nobby Knob locality (Table 3) and are consistent with a distal splay facies association of a distributive fluvial depositional system (65).
In the Nobby Knob area there is a notable absence of conglomeratic facies typically associated with the lower carbonate unit (40,65). As such, following the interpretation of Deckman and colleague (65) we place the Jelm–Popo Agie contact at the base of the purple unit and the top of the underlying Jelm Formation (Fig 2) which regionally consists of paleosols with indicators of strong seasonality (e.g., redox mottling, prominent Bk horizons; 65).
(A) Facies associations 1–4 (see Table 3) of the uppermost Jelm and lowermost Popo Agie formations, Dubois, WY, USA (after 65). (B) Photographs of the Nobby Knob outcrop (2014 field party for scale), (C) excavated cross section above the Nobby Knob quarry, and (D) interpretation of major bedding. Red dotted line marks the upper surface of the NK quarry, red solid lines = gilgai microrelief, yellow lines indicate fining-upward foresets of a fluvial channel. Note: channel cuts into underlying paleosol and the upper fluvial surface exhibits an erosional contact with overlying pedogenically modified mudstones. (E) Clay-rich vertic paleosols directly overlying (F) the bonebed layer which exhibits rhizoliths, redox mottling, and concentration of vertebrate remains.
The paleosol surface just below the purple unit at the NK locality has a prominent 30–50 cm thick carbonate nodule-rich horizon where nodules are partially coalesced forming a relatively resistant platform. The lower carbonate unit (considered absent at NK) is commonly expressed as a microconglomerate composed of reworked pedogenic carbonate clasts and minor amounts of vertebrate material (58). We hypothesize that the paleosol surface at the top of the Jelm is the most likely source of the carbonate nodule clasts found in the conglomerates of the lowermost Popo Agie Formation. Although the lower carbonate is not present at NK, it has been observed at multiple locations within 1–5 km of the quarry. Although it is possible that differences in sedimentation distal to typical lower carbonate deposition in the Lander area are obscuring this otherwise prominent unit, we hypothesize that the carbonate nodule dominated horizon at the top of the Jelm represents a remnant of the surface that is more commonly recycled into the higher-energy deposits of the lower carbonate unit of the Popo Agie Formation.
Locally, the majority of the lower purple unit is a pedogenically modified mudstone with disrupted primary sedimentary structures (e.g., facies 4, Table 3). At the NK locality the first 1–2 m of the purple unit consists of a silty mudstone that exhibits low angle crossbedding and >1 cm fining upward layers of low-angle laminations (silty mudstone to mudstone). Between 2–4 m above the base, 0.25–1 m scale wedge-shaped peds with prominent slicken-sides, redox features, vertebrate remains, root traces, and minor amounts of plant material are common (Fig 2). Lateral accretion sets of a channel-form consisting of very-fine to fine grained muddy-sandstone fills and incised incises at least 2 m into mudstone deposits between 5–7 m from the base of the purple unit, which is consistent with other observations of intermixed fluvial sandstones and floodplain mudstone deposits regionally as well as the remainder of the NK section (65). The upper Popo Agie Fm is present downdip of the NK locality (<100 m), but the top of the local “knob” ends at the purple-ocher transition.
Taphonomy
Bivalve preservation
The mode of preservation of the bivalves at Nobby Knob is unusual in that: (1) only the external morphology appears to be preserved (i.e., there are no internal molds/impressions, even in “butterflied” specimens, (2) there is no measurable thickness to specimens (e.g., shell material [nacre] is not present, nor a mold filled with secondary mineralization or preserved as a void), and (3) a dark “stain” is visible across the surface of some specimens which is mirrored on both halves when observed (usually after splitting sediment along a fracture plane during preparation or excavation of vertebrate material). Preliminary EDS analysis shows a miniscule calcium component (∼0.6 normalized Wt%) along the preserved impression as well as the surrounding clay-rich matrix (S7 Fig in S1 File).
Spatial distribution of skeletal elements
The majority of elongate disarticulated elements considered for the azimuth analyses were between 4–6 cm in length with the exception of mandibles disarticulated from skulls and a few isolated large ribs. These elongate bones show no preferred orientation (Fig 3) based on both the Hermans-Rasson test (T= 1.477, P = 0.910) and the Pycke test (T= −1.772, P = 0.943). There are several instances of articulation or close association, but the majority of the elements are disarticulated. A partial skeleton crosses the field jacket NK19-C2-709.3 with over one dozen nearly articulated intercentra, several articulated neural arches and ribs, and some girdle and limb elements with similar preservation and loose association with the axial skeleton (Fig 3; S6 Fig in S1 File). Additional elements found throughout the bonebed, including two partial, articulated hindlimbs (UWGM 7044) and multiple skulls, exhibit a steep and oblique orientation relative to the bedding plane. For instance, the hindlimbs are nearly vertically oriented, effectively “mired” in the sediment, and this orientation is more likely to occur in a low energy deposit such as a channel fill (17). There are also several ribs that plunge into the bonebed which has been previously interpreted as evidence for trampling in ceratopsian dinosaur bonebeds (72), although none of the bones are broken as is often the case in the dinosaur bonebeds. Three of the plate-like skulls are oriented obliquely to the horizontal plane of the bonebed, two of which are entirely perpendicular to the majority of the bones. At least two of these skulls have articulated mandibles, and at least one of those also has a partially articulated pectoral girdle.
Individual metoposaurid elements were mapped from their respective field jackets. Note that all field jackets were prepared from the “field down” side, so the jacket maps are reflected to show their position from a “field up” view. Colors indicate anatomical identification as follows: red=skull roof, basicranium, or palate, orange=mandibular, brown=vertebral, blue=costal, yellow=pectoral girdle and forelimb, and purple=pelvic girdle and hindlimb. Field jackets with dashed outlines are in estimated positions, and those with crosses are unprepared. The rose diagram in the upper right indicates the orientation of disarticulated long bones (length ≥ 4 times width; N = 207). Detailed field jacket maps and the full list of metoposaurid elements from the NK locality as of December 2024 are available in S1 Table. Grid squares are 1 m2.
Sorting of skeletal elements
The assignment of skeletal elements to Voorhies groups (Table 2) follows previous studies with some minor modifications to the counts to adjust for errors in limb element counts because the humerus and fibula were considered to sort differently from other limb elements by previous workers (2,13,21,22,27). However, some flume experiments have demonstrated that the skeletal elements of freshwater turtles and alligators are not freed from the bed in discrete groups, but rather, the competent velocities (velocity of water at which the element is dislodged) of the elements lie along a continuum with “intermediate” dispersal elements overlapping in competent velocity with early and late dispersal elements (28,73). This was similarly the case in Voorhies’ original flume experiments on coyote and sheep skeletons where he assigned several elements to intermediate positions within his early, intermediate, and late dispersal group framework (22). Despite this, Blob (73) suggested that the use of discrete Voorhies groups was more practical than tabulating competent velocities for all skeletal elements. Some collection bias was introduced by a tendency of the field party to focus on jacketing the fragile, plate-like elements of the pectoral girdle and skull, however, all elements were considered both from field jackets and the surrounding trenches in the counts.
The MNI for EP is 13 based on 12 interclavicles of similar size and one large frontal among a few other exceptionally large non-duplicated elements. The MNI of the subset of Site XIII used in this study is 20 based on the maximum number of any individual skull roof element. The MNI of Nobby Knob is 19 based on 19 interclavicles and right clavicles. Skeletal elements normalized to the expected number of each element based on MNI revealed a pattern in each of the three bonebeds similar to what would be expected based on flume experiments of non-mammal skeletons (28,73): a clear overrepresentation of low-profile, plate-like elements (e.g., skulls, interclavicles, and clavicles) and underrepresentation of diminutive skeletal elements (e.g., autopod bones, neural arches, and hemal arches). However, the skeletal elements of the NK locality do not exhibit evidence of significant hydrodynamic sorting (Fig 4) that would be expected if deposited in a channel-lag in a fluvial system (17,22). In contrast, the Elkins Place bonebed exhibits a strong overrepresentation of the flat plate-like skulls, clavicles, and interclavicles that would be expected with the loss of earlier dispersing elements coincident with the loss of fine-grained sediment in the conglomeratic facies of the Camp Springs Formation. The figured blocks from Site XIII also exhibit overrepresentation of the skulls, clavicles, and interclavicles but with slightly better representation of the rest of the skeleton than the Elkins Place bonebed. However, this sample of Site XIII is not comprehensive and may not be representative of the entire locality.
(A) Elkins Place elements normalized to expected value based on an MNI of 13. (B) Site XIII elements normalized to expected value based on an MNI of 20. (C) Nobby Knob elements normalized to expected value based on an MNI of 19. (D) Ternary diagram showing the expected proportion if all skeletal elements were present (circle) and the observed proportions for NK (hexagon), Site XIII (triangle), and EP (square) under different dispersal potential hypotheses (colors).
Uniquely among metoposaurid bonebeds, the NK locality preserves articulated and disarticulated denticulate palatal plates (Figs 5E–G). Denticulate plates were embedded in the soft tissue of the palate in life (74) and have been reported from other metoposaurid assemblages (7,14) but have not been found in articulation within the palate. Disarticulated palatal plates have only been found underneath skulls, but articulated plates have been found in skulls with the palate facing up.
(A) Photograph of partial skull UWGM 7211 in dorsal view. (B) Interpretive drawing of the same. (C) Photograph of partial skull UWGM 7211 in ventral view. (D) Interpretive drawing of the same. (E) Photograph of articulated denticulate palatal plates in dorsal view UWGM 7574 and (F) close-up inset image of the same. (G) Photograph of disarticulated denticulate palatal plates associated with UWGM 7211. (H) Representative metoposaurid silhouette. Stippling represents non-cranial bones adhered to UWGM 7211; diagonal lines represent broken surfaces. Abbreviations: c, choana; ec, ectopterygoid; f, frontal; j, jugal; l, lacrimal; m, maxilla;, n, nasal; on, otic notch; p, parietal; pal, palatine; pm, premaxilla; po, postorbital; pof, postfrontal; pp, postparietal; prf, prefrontal; sq, squamosal; st, supratemporal; sta?, partial stapes?; t, tabular; v, vomer. Scale bar for A–D equals 5 cm, scale bars for E and G equal 1 mm, scale bar for F equals 1 cm. Metoposaurid silhouette by A. J. Fitch used with permission.
Bone modifications
The majority of postmortem fracturing and deformation is unidirectional and consistent with sediment compaction and the shrink-swell cycle of surrounding clays. For example, the dorsal processes and laminae of several clavicles are broken and compressed against the ventral blade, and the posterior end of several skulls is splayed and flattened or sometimes mediolaterally compressed if lying on its side. Several bones are broken, or in some cases sheared along minor faults with slickensides (argillipedoturbation). One mandible has an unusual fracture pattern with a crushed glenoid region and the postsplenial split off from the overlying elements at about the mid length (UWGM 7568; S8 Fig in S1 File) differing from the typical unidirectional modification we attribute to sediment deformation. This could be due to trampling or bioturbation, but it is an outlier within the context of the rest of the bonebed. No elements exhibit weathering or abrasion consistent with long term exposure at the surface or fluvial transport (e.g. 75).
The Elkins Place (EP) bonebed is situated in the Camp Springs Formation (a.k.a. Santa Rosa Sandstone) has commonly been interpreted as a fine to coarse sand-sized fluvial deposit interbedded with coarser grained conglomerates (6,66,76). No site map of the EP bonebed exists so any comparison of spatial relationships between skeletal elements or orientation is impossible (7). The representation of skeletal elements is biased toward plate-like bones of the skull and pectoral girdle with very poor representation of autopodial elements and vertebral arches (Fig 4A). Many of the skull bones are disarticulated and only three skulls are essentially complete, and there is no clear evidence of abrasion consistent with prolonged saltational transport on any bones (7,17,77).
Systematic paleontology
BIVALVIA Linnaeus, 1758
UNIONOIDA Gray, 1854 (sensu Newell, 1965)
ANTEDIPLODON Marshall, 1929 (sensu Good, 1998)
ANTEDIPLODON sp.
(Fig 6)
Referred specimens.—UWGM 7567, 7571–7573, 7584.
Description and rationale for taxonomic assignment.—Unionoid bivalves have been reported from the Popo Agie Formation and were considered similar to Unio dumblei (82) (=Antediplodon dumblei, sensu 80) from the Dockum Group of Texas (60). None of the specimens from the NK locality reveals the entire morphology of the external valve surface, but some taxonomically informative anatomy can be determined in aggregate. The annuli are apparent, and the nearly complete external molds have an oblong ovate shape with an abrupt anterior end (Fig 6). Several riblets radiate from the umbo, and there are at least 17 ridges in a “pseudo-radial” pattern sensu (83)although none of the specimens reveals the umbo in its entirety. Similar to Antediplodon dockumensis (82) sensu (81), all specimens lack lirae between the annuli. Species of Antediplodon can also be differentiated by the thickness of the shell which cannot be assessed here due to the nature of preservation.
Remarks.—All bivalves from the NK locality are preserved as molds, all of which appear to be external molds due to the presence of only external morphology on both the positive and the negative relief (see Fig 6A–B). Bivalve molds are found below, within, and above the bonebed, but they appear more common in the layers underlying the bonebed. There may be additional bivalve taxa at NK, but only more complete specimens will be conclusive. UWGM 7567 is notable in lacking any apparent riblets radiating from the umbo similar to Triaslacus (84). UWGM 7584 is anteroposteriorly elongate with at least 7 riblets radiating from the umbo (Figs 6C–D). In addition to the specimens included here, there are dozens of uncatalogued bivalve impressions from the NK locality all exhibiting similar morphology to those described.
(A) UWGM 7571 part. (B) UWGM 7571 counterpart. (C) UWGM 7584 part. (D) UWGM 7584 counterpart. (E–G) partial external mold UWGM 7573 preserving the anterior left valve and the posterior right valve in (E) left lateral, (F) dorsal, and (G) right lateral views. (H) UWGM 7572. (I) Unionoid bivalve silhouette. Upper left scale bar for A–B, lower left for C–D, upper right for E–G, and lower right for H. Scale bars equal 1 cm. Silhouette by AMK based on UWGM 7572.
STEICHTHYES Huxley, 1880 (sensu Nelson et al., 2016)
ACTINOPTERYGII Woodward, 1891 (sensu Goodrich, 1930)
REDFIELDIIFORMES Berg, 1940 (sensu Schaeffer, 1984)
REDFIELDIIDAE Berg, 1940 (sensu Hutchinson, 1973)
REDFIELDIIDAE gen. et sp. indet.
(Fig 7)
Referred specimens.—UWGM 7575, 7576, 7579, and 7586.
Description and rationale for taxonomic assignment.—Actinopterygian remains from this site may be referable to Redfieldiidae although only the mold of an element that most closely resembles a redfieldiid supracleithrum (e.g. figs 5, 8b–f: 92) is preserved with most of the element lost during preparation (Fig 7F–G). Several isolated ganoid scales, conical teeth, and fragments of similar bone were recovered. The ganoid scales are diamond-shaped and lack any apparent ornamentation (Figs 7C–D) similar to Lasalichthys otischalkensis (92). A fragmentary dentigerous element appears to show teeth angled in multiple directions (Fig 7E) much like the characteristic morphology of the rostral bone of Redfieldiiformes (92).
Remarks.—Isolated actinopterygian remains have been recovered from the bonebed and appear more common in the underlying sediment, but they are rare and fragmented compared to the metoposaurid remains. It is unclear if a probable actinopterygian fin from NK (Figs 7A–B) is from a redfieldiid or an as yet unidentified actinopterygian.
(A) UWGM 7576 possible actinopterygian fin with inset (B) showing elongate structures (=lepidotrichia?). (C) UWGM 7579 ganoid scale part. (D) UWGM 7579 scale counterpart. (E) UWGM 7586 tooth-bearing element (rostral?). (F) UWGM 7575 right supracleithrum counterpart. (G) UWGM 7575 partial right supracleithrum remaining part. (H) Representative redfieldiid silhouette. Scale bars equal values indicated in figure. Left scale bar for A, middle scale bar for C–E, and right scale bar for F–G. Redfieldiid silhouette by AMK based on Lasalichthys stewarti (93) sensu (92).
TETRAPODA Jaekel, 1909
TEMNOSPONDYLI von Zittel, 1887–1890 (sensu Schoch, 2013)
METOPOSAURIDAE Watson, 1919 (sensu Buffa et al., 2019)
BUETTNERERPETON Gee & Kufner, 2022
BUETTNERERPETON BAKERI Case, 1931 (sensu Gee & Kufner, 2022)
(Figs 5, S8)
Referred specimens.—See S1 Table for complete list.
Description and rationale for taxonomic assignment.—The vast majority of vertebrate remains from the NK locality are referable to Metoposauridae, and the more complete cranial remains are indistinguishable from Buettnererpeton bakeri (7). Some of the salient features of the skull that demonstrate this taxonomic assignment are: (1) lacrimal separated from the orbital margin by contact of the prefrontal and the jugal, (2) deep otic notch, (3) well-developed tabular horn, and (5) anterior margin of orbit posterior to anterior margin of interpterygoid vacuity. Additional features of the postcranial skeleton (not figured) that demonstrate this taxonomic assignment are: (1) large areas of reticulate ornamentation on both the interclavicle and the clavicle and (2) a sensory groove along the ventral side of the posteromedial margin of the clavicle.
Remarks.—Specimens from the type locality of Buettnererpeton bakeri (Elkins Place) are primarily from similar-sized individuals (midline skull length ∼30 cm) with the exception of at least one fragmentary, large individual (7). The NK locality preserves a relatively wide size range of nearly complete skulls (<20 – ∼40 cm midline length) as well as a greater representative sample of the postcranial skeleton than the type locality of B. bakeri.
DIAPSIDA Osborn, 1903
ARCHOSAUROMORPHA Huene, 1946 (sensu Benton, 1985)
ARCHOSAUROMORPHA gen. et sp. indet.
Referred specimens.—UWGM 7569, 7570, and 7585.
Description and rationale for taxonomic assignment.—Three isolated, shed archosauromorph teeth were recovered from the NK bonebed. The two most complete teeth are shown in Figs 8A– B. One tooth (Fig 8A; UWGM 7585) is recurved and conical with fine serrations (=denticles) on its mesial and distal margins. Another tooth (Fig 8B; UWGM 7569; possibly a maxillary or posterior dentary tooth) is subtriangular in labial and lingual views and mediolaterally compressed. Both the mesial and distal margins are slightly convex and lined with serrations, and the tooth is not expanded at its base. The presence of both mesial and distal serrations suggests archosauriform affinities (102), however, some malerisaurine azendohsaurids convergently acquired tooth serrations (33,103) restricting our assignment of isolated serrated teeth to Archosauromorpha. Previously described archosauromorphs from the Popo Agie Formation include phytosaurs, a poposaurid, a loricatan, and two dinosauriforms (Table 1).
ARCHOSAURIFORMES Gauthier et al., 1988
PHYTOSAURIA Meyer, 1861 (sensu Doyle & Sues, 1995)
PARASUCHIDAE Lydekker, 1885 (sensu Kammerer et al., 2016)
PARASUCHIDAE gen. et sp. indet. (Figs 8D–J)
Referred specimens.—UWGM 1995, 7578.
Description and rationale for taxonomic assignment.—Fragments of a phytosaur mandible (UWGM 1995, 7578) were recovered from above the NK bonebed or as weathered fragments around the hill. UWGM 1995 is made up of several fragments collected as “float” including a segment of a right dentary. The fragment of dentary bears at least seven alveoli and a prominent median platform that rises above the level of the lateral margin of the element. This segment is D-shaped in cross section similar to the cross section of the anterior portion of UWGM 7578. (Figs 8I–J). The medial surface of the dentary has posteriorly dipping interdigitations of the symphyseal plate (Fig 8E). The largest fragment is made up of the angular and the splenial from the portion of the mandible ventral to the mandibular fenestrae (Figs 8I–J). The labial surface of this element is ornamented with raised ridges at its posterior extent. The lingual surface is smooth, flat, and lacks any symphyseal suture morphology such as that seen in UWGM 1995. There is a depression on the lingual surface of the angular of UWGM 7578 at the posterior extent of the preserved element (Fig 8J) bounded ventrally by a medially inflected ridge.
Remarks.—These fragments were all recovered ex situ and are most likely from a horizon above the primary NK bone layer, exhumed during removal of overburden. The bone-bearing horizon above the primary bone layer of NK could not be located, however, the horizon can be stratigraphically constrained to within 5 m above the NK bonebed, below the purple-ocher transition zone.
(A) Photographs of UWGM 7585 from left to right in lingual, labial, mesial, and distal views. (B) Photographs of UWGM 7569 from left to right in labial, lingual, and mesial or distal views. (C) Representative silhouettes of two archosauromorphs with serrated dentition known from the Popo Agie Formation, phytosaur (top) and poposaurid (bottom). (D–F) Photographs of partial right phytosaur dentary UWGM 1995 in (D) lateral, (E) medial, and (F) dorsal views. (G–H) Photographs of partial left(?) prearticular UWGM 7578 in (G) lingual(?) and (H) labial(?) views. (I–J) Partial right mandible UWGM 7578 in (I) labial and (J) lingual views. (K) Shaded parasuchid skull to show approximate positions of specimens. Alveoli in D–F marked with an asterisk (*). Abbreviations: a. ar., articulation with the articular; an, angular; d, depression; emf, margin of the external mandibular fenestra; imf, margin of the internal mandibular fenestra; pa, prearticular; sa, surangular; sym. p., symphyseal plate. Scale bars equal 1 cm. Phytosaur and poposaurid silhouettes by Scott Hartman used under Creative Commons Attribution 3.0 Unported https://creativecommons.org/licenses/by/3.0/ from phylopic.org. “Paleorhinus” skull illustration modified from work by Scott Hartman used with permission.
Incertae sedis
Plant megafossils.—Some plant remains have been recovered at the NK locality including abundant fragments of weathered petrified wood littering the surrounding area. Potential in situ plant remains are concentrated below the bonebed including possible leaves (Figs 9A, D–E), stems (Figs 9B–C), and roots (Fig 9F). Root traces are present throughout the interval as graygreen reduction zones, and some of these reduction zones contain carbonized root material (e.g., Fig 9F). Plant material is often not well preserved at the NK locality probably due to pedogenic overprinting of this interval. Previous reports of plant megafossils from the Popo Agie Formation are restricted to the area around Lander, WY and the upper Popo Agie–the ocher unit or higher (60,108).
(A) Block with plant fragments, UWGM 7588. (B) Possible stem, UWGM 7580, in part. (C) Possible stem, UWGM 7581. (D) Plant material, UWGM 7587, in part. (E) Plant material, UWGM 7589, in part. (F) Root fossil, UWGM 7577, under cross-polarized light. (G) Coprolite fragments, UWGM 7582. (H) Coprolite fragments, UWGM 7583. Scale bar for A equals 5 cm, scale bars for B–E and G–H equal 1 cm, scale bar for F equals 1 mm.
Ichnofossils—Despite the presence of isolated non-metoposaurid teeth among the metoposaurid remains, no bioglyphs indicative of scavenging or predation (e.g., tooth or claw scores) have been identified. However, some isolated coprolites were recovered within the bonebed layer (Figs 9G–H). A lack of diagnostic features such as longitudinal striations or spirals precludes taxonomic identification. The coprolites are oblong and compressed in one direction resulting in an elongate oval cross section; this compression may be a taphonomic artifact based on similar compression of skeletal elements in this interval. Based solely on size and other taxa present, we think they were likely produced by tetrapods, but this is equivocal.
Discussion
Sedimentology, stratigraphy, and environmental interpretations
Identification of the Popo Agie-Jelm contact is complicated by the absence of a carbonate- or quartz-grain conglomerate or a “chert pebble horizon” found elsewhere in the Wind River Basin (109,110). In the Dubois section we consider the contact to be above the Serendipity bed (Fig 2; 61). Below this contact, pedogenesis is less common and fluvial-dominated deposits such as laterally accreting bars and splay deposits are abundant, although weak pedogenic overprinting is present throughout. These crevasse splay deposits are similar to those found below the Channel Facies Association, indicating they may derive from a similar depositional system. Above the contact, paleosols are dominant, with few unaltered deposits. The much thicker and more developed paleosol deposits above the Serendipity bed horizon indicates that there is much slower deposition in the region compared to downsection deposits.
The Nobby Knob bonebed is situated in pedogenically modified floodplain deposits interpreted as the distal portion of a larger distributive fluvial system (65). Fluvial processes inferred from lateral accretion sets incised into floodplain deposits above and below the bonebed interval is the primary depositional agent. Fine laminations in the immediate interval surrounding the bonebed support a localized lower energy depositional system such as an oxbow lake or pond. Pedogenic overprinting partially obscures primary sedimentary structures, however remnant structures such as low angle fining upwards cross-bedding are consistent with an initial fluvial origin (e.g., point bar or splay deposits). Difficulties in determining lateral relationships along this interval of outcrop prevent a more detailed analysis. The Dubois area is considered to be dominated by Facies 4 (Table 1), with minor mud-rich splay deposits.
The overall interpretation of a fluvio-lacustrine depositional system for the lower Popo Agie Formation sensu (40) has not drastically changed since early work in the unit (111–113), although recent work rejects the hypothesis of a large persistent single lacustrine system as previously suggested for the upper Popo Agie Formation (e.g. 113) in favor of smaller temporally variable lacustrine environments across the broader floodplain more common in medial and distal splay deposits of a distributive fluvial system (65).
The Elkins Place conundrum
Visual inspection of previously published images EP specimens containing residual matrix (figs 7, 10, & 18b: 7) and outcrop images of conglomeratic beds from the Santa Rosa Sandstone (figs 2.10a, b, and d: 66), clearly demonstrate very poor sorting, as mentioned by these and other authors (6,76). The larger quartz-rich (and bone) clasts are common but isolated and surrounded by a sand-sized matrix.
With the caveat that we lack detailed field observations of our own, we would like to point out that these matrix-supported clasts appear to be more consistent with debris (or possibly hyperconcentrated) flows than the more typical clast-supported conglomerates of braided streams they were inferred to represent. The matrix-supported “conglomeratic” deposits within the Santa Rosa Sandstone may be better explained by debris flow deposition which can follow local topography, occur both subaerially and subaqueously, and travel over short distances and low gradients (115–117). Additionally, debris flows, a type of hyperconcentrated flow, would explain the unusual deposits near the base of the Santa Rosa Sandstone with extremely large clasts supported in a sandy matrix (fig 2.10d: 66). Entrainment in a hyperconcentrated flow during a flash-flooding event could explain the lack of abrasion and the presence of small (early dispersal; Voorhies Group I) and large (late dispersal; Voorhies Group III) elements in what was previously described as a high-energy conglomerate. If so, this would represent a relatively novel depositional environment, though one consistent with interpretations of a strongly seasonal climate, and possibly one of the oldest debris flow-hosted vertebrate bonebeds (118).
Bivalve preservation in a paleosol
The presence of unionoid bivalve molds and isolated redfieldiid bones (Figs 6 and 7) through the interval surrounding the bonebed provides additional support for a freshwater origin of the assemblage with low hydrodynamic influence, but the preservation of the bivalves is atypical. Although we do not have a clear explanation for the unusual mode of the observed bivalve preservation, we hypothesize that this phenomenon is a result of peri- or post-depositional dissolution of biominerals coupled with the higher initial preservation potential of the outermost organic layer. The very low concentration of measurable Ca+ in the clay of both the mold and the surrounding matrix (S7 Fig in S1 File) is consistent with a lack of Bk horizons in the Nobby Knob vertic paleosol and the absence of carbonate minerals. Although vertisols do not typically have a pH lower than 6.0 and more commonly between 7.0 and 8.5 (119,120), surface (or pore) water pH of 6–7.5 would be sufficient to dissolve the aragonitic biominerals within the outer prismatic layer and organic-mineral layered nacre (121) in pore waters undersaturated in calcium.
Dissolution of the outer prismatic aragonitic shell and inner nacre in a pH environment between 6–7.5 (121) would potentially leave behind organic structures and residues (i.e., periostracum and/or narcre-embedded organic layers [121]). We hypothesize the robust protein structure of the periostracum (composed of conchiolin) would effectively preserve the external morphology of the shell, once buried within the sediment, even after biomineral dissolution. Given the relatively uniform thickness of the periostracum, a mirror of that morphology at the periostracum/prismatic layer boundary would be recorded as a negative. The loss of the biomineral portion of the shell would result in deflation of unlithified sediment through expansion and contraction (i.e., shrink-swell of clay component in vertic soils). The absence of measurable organic material (e.g., elemental carbon during energy dispersive spectroscopy analysis; see S7 Fig in S1 File) suggests subsequent diagenetic decomposition and removal of any organic residues (122). Although, elemental carbon is at the edge of detectable spectra in the instrument used.
Skeletal sorting and transport
The Nobby Knob bonebed exhibits a much lower degree of pre-burial sorting and, as a result, a higher representation of the skeleton than other monodominant North American metoposaurid bonebeds (Figs 4A, C; 2,3,13). There is also no evidence for alignment of long bones with a current (Fig 5) unlike the Lamy amphibian quarry (2,3) and possibly the Rotten Hill bonebed (13). Unfortunately, direct comparisons of skeletal sorting between the three presented sites and two previously studied North American metoposaurid bonebeds are not possible because either raw data on skeletal element counts are not provided or incomplete (2,3) or skeletal element counts were not done (13). Outside of the late dispersal group, previously assigned Voorhies groups for temnospondyls fit poorly with observed skeletal completeness in any of the bonebeds assessed here (Figs 4A–C).
Proposed Voorhies group assignments of temnospondyls have relied on qualitative assessments of surface area:volume and complexity of morphology. One recent study used nearly the same dispersal potential groups proposed by Voorhies (1969) despite the dissimilarity between many temnospondyl skeletal elements and their homologs in mammal skeletons (21). Others have modified the Voorhies group assignments of temnospondyl skeletal elements (2,3,13,27), however, some of the Voorhies group assignments differ between these studies. For example, the metoposaurid clavicle has been treated as a late dispersal element (Voorhies group III; 2,3,13), but Rinehart and Lucas (27) considered the clavicle of the capitosaur Eocyclotosaurus to be an intermediate dispersal element (Voorhies group II) despite nearly identical morphology and thus surface area:volume. In addition, the relatively high density of metoposaurid dermal bones would be expected to increase competent velocity (123). Regardless, there have been no actualistic experiments to determine the dispersal potential of temnospondyl skeletal elements.
Skeletal elements with similar shape such as skulls, interclavicles, and clavicles with large, flat regions tend to be represented in similar proportions (Fig 4). The humerus, fibula, and ischium were considered either intermediate or late dispersal elements in previous Voorhies groups applied to temnospondyls (2,3,13,27), but the representation of each of these elements in actual assemblages does not fit a late dispersal model but instead may fit intermediate dispersal (Fig 4). Diminutive skeletal elements such as vertebral arches, metapodials, and phalanges are poorly represented in all of the assessed bonebeds–in line with early dispersal and loss from the site of deposition. The intermediate dispersal elements are highly variable between the three sites, and it is not clear that their relative abundance conveys any useful information with regard to sorting or transport. Interestingly, the estimated skeletal completeness from Site XIII with articulated skeletons appears more similar to the Elkins Place bonebed (Fig 4), however, this assessment is based solely on published photographic plates of blocks that may have been selectively favored for having the more diagnostic skulls and pectoral girdles (10). The relative completeness of the EP bonebed and Site XIII appear to follow a power curve, and preliminary results from flume experiments seem to show a similar trend in the competent velocities of skeletal elements in simulated dinosaur bones (124).
All parts of the skeleton of B. bakeri (except the epipterygoid) are represented in the NK bonebed including diminutive elements such as the distal phalanges and denticulate palatal plates (S1 Table). The underrepresentation of autopodial elements, neural arches, and hemal arches from the NK bonebed may be due to incomplete preparation of several of the quarry blocks since many of these elements are ∼1 cm in maximum dimension. Preparation of NK material is ongoing and is expected to continue to yield more diminutive skeletal elements. The NK bonebed more than doubles the minimum number of individuals of B. bakeri and preserves a wide size range of individuals from a single site that can provide insight into the ontogeny of metoposaurids.
In addition to the presence/absence of skeletal elements, the alignment of elongate skeletal elements can be used to indicate the presence and direction of a current (2,17). The NK bonebed exhibits no evidence of alignment of long bones (Fig 3). Articulation of metoposaurid elements is particularly rare in North America with only a few specimens previously reported (3,6,7,15). Some skeletal elements from the NK bonebed are articulated including several skulls with mandibles, some of the postcranial axial skeleton, pectoral girdles and forelimbs, hindlimbs, and denticulate palatal plates. The presence of articulated denticulate palatal plates that would have been embedded in the soft tissue of the palate (74) implies the syndepositional presence of soft tissue in at least some individuals. The differences in articulation between individuals could indicate differing degrees of decomposition prior to burial, or an effect of shielding by the plate-like skulls and pectoral girdles from post-depositional movement or shearing (i.e., shrink-swell, trampling or other bioturbation).
There are a number of large, plate-like elements that are preserved in opposition to the bedding plane (oblique, and near-vertical alignment) positioned in such a manner that is rarely observed in a low-energy depositional setting similar to what we propose. We suggest these unique instances are most likely a result of localized or isolated specimens that were mobilized due to the shrink-swell actions of clay-rich soil prior to lithification. It is unlikely that these result from some form of hyperconcentrated flow considering the absence of matrix supported clasts (e.g. 123,124), or due to large tetrapod trampling given the lack of features typically associated with that scale of bioturbation (e.g. 72).
The near monotaxicity of the NK site taken together with sedimentological and taphonomic evidence of contemporaneous deposition suggests that the NK bonebed is the result of a biotic aggregation of metoposaurids (77,127). Several possible scenarios can result in biotic aggregations of a single taxon such as breeding colonies (128), predator bone accumulation (129), or mass mortality (77,127). No evidence of predation or scavenging is present on any of the metoposaurid remains and no osteological proxies for temnospondyl breeding are known. A monsoonal climate has been proposed for the underlying upper Jelm Formation resulting in seasonal aridity that could have dried isolated bodies of water (61). The more developed paleosols of the lower purple unit indicate an increased hydrologic cycle with higher water table relative to the underlying Jelm Fm, but the presence of vertic features such as gilgae topology and slickensides, as well as variable redoximorphic features (see Fig 2) indicate periodic (seasonal?) saturation and desaturation (130). The additional vertebrate remains within the bonebed interval appear to be incidental and possibly time-averaged or transported (e.g., disarticulated ray-finned fish remains and shed archosauromorph teeth). The concentration of metoposaurids at this locality is almost certainly biotically induced and plausible scenarios observed in modern amphibians include: seasonal breeding and subsequent die-off (e.g. 126), ontogenetic niche partitioning (e.g. 129), or anatomical and/or physiological limitations that prevent exodus from a drying body of water (e.g. 130).
Comparison with other metoposaurid sites
The Nobby Knob bonebed (NK) is the first unequivocal metoposaurid mass mortality assemblage from the Carnian-aged lower Popo Agie Formation. Two additional sites from the upper Popo Agie Fm are monodominant or monotaxic metoposaurid bonebeds that could be interpreted as mass mortality assemblages sensu (77): Bull Lake Creek and Willow Creek. The stratigraphic position of Bull Lake Creek is uncertain, but it is from the upper Popo Agie Formation and may be situated as low as the purple-ocher transitional zone (43,44). The skeletal elements from Bull Lake Creek are from no less than five individuals and are almost entirely large, flat elements such as skulls, mandibles, interclavicles, and clavicles with the exception of one ilium and one intercentrum (44). The Willow Creek locality is located within the ocher unit based on residual matrix on the holotype skull of Anaschisma browni, a referred skull (the holotype of “Anaschisma brachygnatha” 41), and a referred mandible (42). There is a strong bias toward inferred late dispersal elements of the skull and pectoral girdle and a paucity of postcranial skeletal elements at both of these sites (42,44). Neither of these bonebeds can be distinguished from time-averaged assemblages or highly sorted assemblages with loss of early and intermediate dispersing elements and should not be considered mass mortality assemblages.
Site XIII and the Elkins Place (EP) bonebed could be considered two end members of metoposaurid skeletal deposits with nearly complete articulated skeletons at Site XIII and completely disarticulated and sorted skeletons at EP. Brief descriptions of the taphonomy of Site XIII have been provided in previous studies (3,13), but an overall assessment of the taphonomy of this site has not been done. In addition to abundant articulated skeletons, some details of the communications provided by (3) suggest that Site XIII represents a “drying pond” scenario as proposed by (1) including: (1) large individuals were concentrated at the center of the deposit with small individuals forming a ring around them, (2) there is no “imbrication” of bones, (3) there are no mud cracks in the bonebed layer, but they are present in the same layer surrounding the quarry, and (4) no calcrete pebbles are present in the bonebed. Two of these criteria (1 and 3) require complete excavation of a site for proper comparison, and thus are not comparable with NK in its current state of excavation (i.e. the limits of the NK bonebed have not been found). However, the lack of Bk horizons (e.g., absence of calcrete pebbles) at both sites suggests some similarity in depositional regime, specifically a relatively consistent degree of sediment saturation.
At the EP site, the concentration of associated vertebrate material, the overrepresentation of late dispersal elements (Voorhies group III), and the lack of transport induced modifications (e.g. 75) is suggestive of an autochthonous assemblage that has been winnowed. Similarly, the presence of disarticulated denticulate palatal plates in one skull from EP (fig 16: 7) suggests either these plates were trapped in place under a skull unmoved from the site of death or the entire skull was transported with soft tissue intact. Considering the inferred seasonality during Camp Springs fluvial deposition (66) the specimen was likely a desiccated fragmentary carcass. Desiccation or draping of the bone-bearing palatal integument would have allowed for the retention of the denticulate plates leading up to final deposition where these diminutive elements were ultimately preserved in place within the interpterygoid vacuity. Taken together with our observations of matrix supported clasts associated with EP, these data do not exclude the possibility of a debris flow-hosted bonebed (118,125). However, in the absence of a quarry map and other firsthand data, these observations remain inconclusive at this time.
Implications for the taphonomy of metoposaurids and other large temnospondyls
The skeletal composition of Voorhies dispersal groups (VDG) previously used for temnospondyls have varied (2,3,13,21,27). Ultimately, each version of VDG applied to temnospondyls ((21) [unmodified VDG]; (2,3,13) [metoposaurid modified VDG]; (27) [capitosaurid modified VDG]) result in similar enrichment in late dispersal elements in hydrodynamically sorted deposits despite different underlying assumptions in dispersal potential of specific elements. Regardless of which version of VDG are implemented, the overall trends are comparable (Fig 4D). Further interrogation of subtle differences in depositional regimes is possible using the framework we have presented through the inclusion of additional bonebeds coupled with actualistic studies.
In the current absence of flume experiments on temnospondyl skeletons, freshwater turtles or crocodilians may provide a much needed modern analog to better understand the taphonomy of aquatic temnospondyls (mostly stereospondyls). Bones of freshwater turtles subjected to flow in flume experiments displayed an array of preferred orientations which broadly fall into two categories: (1) plate-like elements (e.g., stereospondyl crania and pectoral girdle elements), vertebrae, and skulls showed no preferred orientation, (2) while limb bones and ribs tended to orient with long axis parallel to flow direction (73). Bones of alligators in flume experiments demonstrated a trend in which the interaction of height and weight of each element correlated well with dispersal potential, and in general, tall and light elements were most likely to disperse and flat and heavy objects were least likely to disperse with an array of possible morphologies in between (28). In addition, some aquatic turtles, metoposaurids, and other stereospondyls exhibit osteosclerosis or a skeletally-variable increase in bone density for buoyancy regulation that could increase competent velocity (68,123). These complex interactions between morphology, bone density, and dispersal potential serve to further highlight the need to conduct actualistic experiments to better understand the underlying processes in the dispersal of temnospondyl bones.
Biochronological implications
Nearly all previous reports of vertebrate fossils from the Popo Agie Fm were restricted to the lower carbonate unit and the ocher unit with little known from the intervening purple unit. The Nobby Knob locality from the purple unit yields the stratigraphically lowest diagnostic metoposaurid material and the first occurrence of Buettnererpeton bakeri from the Popo Agie Formation (Fig 5). Buettnererpeton was previously only known from the Camp Springs/Santa Rosa and lower Cooper Canyon formations in Texas (7,133) and the Wolfville Formation in Nova Scotia (134). The only metoposaurid previously reported from the Popo Agie Fm is Anaschisma browni from the upper Popo Agie Fm or the lower-upper transitional zone (42,44). Buettnererpeton bakeri is consistently found below the lowest known occurrence of A. browni in North American Upper Triassic strata now including the Popo Agie Formation with no reported stratigraphic overlap (7,133,134). This succession of Buettnererpeton–Anaschisma is also present in the lower Dockum Group (7,133) possibly including the Otis Chalk quarries (pers. obsv. AMK; 15).
The NK locality also yields the lowest known occurrence of a phytosaur in the Popo Agie Formation, but this specimen cannot be referred beyond Parasuchidae indet. (Fig 8). A previous report of phytosaur material from the “unnamed red beds” or lower carbonate unit in eastern Wyoming (135) has not been confirmed despite exhaustive study of material from the Clark Locality (57). The presence of parasuchids in the Popo Agie Fm has been used to correlate it to the Otischalkian holochronozone of other Upper Triassic strata of the western USA, but the duration of Popo Agie deposition is unclear due to a lack of leptosuchomorph phytosaurs that define the succeeding Adamanian holochronozone (40,136). The absence of leptosuchomorph phytosaurs in the Popo Agie Fm renders the whole of the Popo Agie a topless teilzone which precludes definitive correlation of estimated holochronozone boundaries in the lower Popo Agie Fm and the lower Dockum Group in the current biostratigraphic framework (136). The base of the Otischalkian teilzone can confidently be extended to ∼5 m above the base of the local purple unit in the Popo Agie Fm due to the presence of a partial parasuchid phytosaur mandible at this level (Fig 8). Others have mentioned the presence of ‘phytosaur’ remains in the lower Popo Agie Fm of Wyoming (lower carbonate unit: (112) and northeastern Utah (purple unit: (137); ‘coarser parts’: (138), however the material referenced is either lost, never curated, or possibly not collected. Detrital zircon radioisotopic ages from the overlying ocher unit provide a definitively late Carnian age of the lower purple unit (ca. 230 Ma) and suggest a Carnian age for correlative strata in the lower Dockum Group (Lovelace et al., in press). The succession of the metoposaurids Buettnererpeton and Anaschisma may provide additional constraint, but the current taxonomy of metoposaurids in Texas, particularly the Otis Chalk quarries, is in need of revision.
Conclusion
The Nobby Knob locality from the lower purple unit of the Popo Agie Formation preserves a dense association of a nearly monotaxic assemblage dominated by the oldest known North American metoposaurid Buettnererpeton bakeri with minor additional faunal elements. We interpret the NK bonebed as a mass mortality event with attritional accumulation of elements surrounding the bonebed interval. Unlike other metoposaurid-dominated bonebeds in North America, the NK locality exhibits little evidence of hydrodynamic sorting. The combination of fine-grained sediment, fluvio-lacustrine characteristics of surrounding stratigraphy, partially articulated (or closely associated) skeletons, and preservation of miniscule elements in articulation are consistent with a low energy deposit such as an oxbow lake or a pond. The NK bonebed significantly increases the sample size of B. bakeri and may provide crucial information for elucidating intraspecific variation within the oldest known North American metoposaurids. The taxonomic composition of the NK locality is similar to sites from the lower Dockum Group of Texas with the occurrences of the metoposaurid B. bakeri, an indeterminate redfieldiid ray-finned fish, abundant unionoid bivalves cf. Antediplodon, and an indeterminate parasuchid phytosaur improving biostratigraphic correlations of the Popo Agie Fm and the lower Dockum Group.
sample Supporting information
S1 Fig. Schematic drawing of field jacket NK16 J10 3-5. High resolution jacket map as included in quarry map (top) and with numbers that correspond to “Prep Lab Number” in S1 Table (bottom). Some elements are slightly transparent to show underlying bones used in the azimuthal and/or total element count analyses. Colors indicate anatomical position as follows: red=cranial, orange=mandibular, brown=vertebral, blue=costal, yellow=pectoral girdle and forelimb, and purple=pelvic girdle and hindlimb. Scale bar equals 10 cm.
S2 Fig. Schematic drawing of field jacket NK16 J3 2-3. High resolution jacket map as included in quarry map (top) and with numbers that correspond to “Prep Lab Number” in S1 Table (bottom). Note that two additional elements of this jacket are present but could not be mapped. Colors indicate anatomical position as follows: red=cranial, orange=mandibular, brown=vertebral, blue=costal, and purple=pelvic girdle and hindlimb. Scale bar equals 5 cm.
S3 Fig. Schematic drawing of field jacket NK16 J8 1-2. High resolution jacket map as included in quarry map (top) and with numbers that correspond to “Prep Lab Number” in S1 Table (bottom). Note that additional elements of this jacket are present but could not be mapped. Some elements are slightly transparent to show underlying bones used in the azimuthal and/or total element count analyses. Colors indicate anatomical position as follows: red=cranial, orange=mandibular, brown=vertebral, blue=costal, yellow=pectoral girdle and forelimb, and purple=pelvic girdle and hindlimb. Scale bar equals 5 cm.
S4 Fig. Schematic drawing of field jacket NK18-D1-000. High resolution jacket map as included in quarry map (top) and with numbers that correspond to “Prep Lab Number” in S1 Table (bottom). Colors indicate anatomical position as follows: red=cranial, orange=mandibular, brown=vertebral, blue=costal, yellow=pectoral girdle and forelimb, and purple=pelvic girdle and hindlimb. Scale bar equals 5 cm.
S5 Fig. Schematic drawing of field jacket NK19-A2-706.21. High resolution jacket map as included in quarry map (top) and with numbers that correspond to “Prep Lab Number” in S1 Table (bottom). Colors indicate anatomical position as follows: brown=vertebral, blue=costal, and yellow=pectoral girdle and forelimb. Scale bar equals 5 cm.
S6 Fig. Schematic drawing of field jacket NK19-C2-709.3. High resolution jacket map as included in quarry map (top) and with numbers that correspond to “Prep Lab Number” in S1 Table (bottom). Some elements are slightly transparent to show underlying bones used in the azimuthal and/or total element count analyses. At the time of submission, this jacket was still undergoing preparation, so several skulls are only partially revealed. Colors indicate anatomical position as follows: red=cranial, orange=mandibular, brown=vertebral, blue=costal, yellow=pectoral girdle and forelimb, and purple=pelvic girdle and hindlimb. Scale bar equals 5 cm.
S7 Fig. Scanning electron microscope (SEM) image and energy dispersive spectroscopy (EDS) X-ray spectra. (A) SEM image with circles marking the spots from which spectra were obtained. (B) EDS of the bivalve mold. (C) EDS of the matrix.
S8 Fig. Photographs of UWGM 7568, a damaged mandible from the NK bonebed. (A) labial view, (B) lingual view, and (C) occlusal/dorsal view. Note the crushed glenoid region and the postsplenial split off from the ramus–damage inconsistent with unidirectional and uniform sediment compaction. Scale bar equals 5 cm.
S1 Table. Metoposaurid fossils from the Nobby Knob locality cf. Buettnererpeton bakeri. Composite elements such as crania and mandibles were separated into their constituent elements to estimate the completeness of crania and mandibles when isolated elements were present. Jacket numbers associate specimens collected as hand samples with the number of a field jacket. Field numbers represent either a field jacket number or a hand sample number; field jackets may have a “lab shorthand” indicated by a number or abbreviation in parentheses. Prep lab numbers are purely for bookkeeping purposes of this study, and the catalog numbers are the numbers of record (which record all numbers used for a given specimen throughout collection history).
S2 Table. Metoposaurid fossils from the Elkins Place bonebed–the type locality of Buettnererpeton bakeri. Composite elements such as crania and mandibles were separated into their constituent elements to estimate the completeness of crania and mandibles when isolated elements were present.
S3 Table. Metoposaurid fossils from Site XIII–the type locality of Dutuitosaurus ouazzoui. Composite elements such as crania and mandibles were separated into their constituent elements to estimate the completeness of crania and mandibles when isolated elements were present.
S1 File. Supplemental figures and extended methods for skeletal sorting and SEM. S2 File. R script for azimuthal analyses and plotting of skeletal sorting and Voorhies groups.
Acknowledgments
Thanks to C. Eaton and A. Goncalves for assistance with the UW Geology Museum collection and use of the collection space and equipment. Thanks to B. Wathen for use of the photo microscope and to W. Schneider for use of the scanning electron microscope. Thanks to R. Rofkar for helpful discussion on place names in Wyoming. Our most sincere gratitude to the UWGM field crews (2014–2019) and UWGM fossil preparation lab volunteers too numerous to name who have made the study of the Nobby Knob material possible. We acknowledge the Eastern Shoshone people and their stewardship of these lands to which they have belonged since time immemorial. We also acknowledge the violation of the sovereignty of the former Shoshone Reservation (now Wind River Reservation) by western researchers whose resultant museum-based specimens we have referenced. The specimens housed at UWGM were collected under BLM permits PA16–WY–252 and PA16–WY–254 granted to DML.
References
- ↵
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.
- 17.↵
- 18.
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.
- 32.
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.
- 46.
- 47.↵
- 48.
- 49.
- 50.
- 51.
- 52.
- 53.
- 54.
- 55.
- 56.
- 57.↵
- 58.↵
- 59.
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.
- 79.
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.
- 86.
- 87.
- 88.
- 89.
- 90.
- 91.
- 92.↵
- 93.↵
- 94.
- 95.
- 96.
- 97.
- 98.
- 99.
- 100.
- 101.
- 102.↵
- 103.↵
- 104.
- 105.
- 106.
- 107.
- 108.↵
- 109.↵
- 110.↵
- 111.↵
- 112.↵
- 113.↵
- 114.
- 115.↵
- 116.
- 117.↵
- 118.↵
- 119.↵
- 120.↵
- 121.↵
- 122.↵
- 123.↵
- 124.↵
- 125.↵
- 126.↵
- 127.↵
- 128.↵
- 129.↵
- 130.↵
- 131.
- 132.
- 133.↵
- 134.↵
- 135.↵
- 136.↵
- 137.↵
- 138.↵