Archosaur footprints (cf. Brachychirotherium) with unusual morphology from the Upper Triassic Fleming Fjord Formation (Norian–Rhaetian) of East Greenland

Abstract The Ørsted Dal Member of the Upper Triassic Fleming Fjord Formation in East Greenland is well known for its rich vertebrate fauna, represented by numerous specimens of both body and ichnofossils. In particular, the footprints of theropod dinosaurs have been described. Recently, an international expedition discovered several slabs with 100 small chirotheriid pes and manus imprints (pes length 4–4.5 cm) in siliciclastic deposits of this unit. They show strong similarities with Brachychirotherium, a characteristic Upper Triassic ichnogenus with a global distribution. A peculiar feature in the Fleming Fjord specimens is the lack of a fifth digit, even in more deeply impressed imprints. Therefore, the specimens are assigned here tentatively to cf. Brachychirotherium. Possibly, this characteristic is related to the extremely small size and early ontogenetic stage of the trackmaker. The record from Greenland is the first evidence of this morphotype from the Fleming Fjord Formation. Candidate trackmakers are crocodylian stem group archosaurs; however, a distinct correlation with known osteological taxa from this unit is not currently possible. While the occurrence of sauropodomorph plateosaurs in the bone record links the Greenland assemblage more closer to that from the Germanic Basin of central Europe, here the described footprints suggest a Pangaea-wide exchange. Supplementary material: Three-dimensional model of cf. Brachychirotherium pes–manus set (from MGUH 31233b) from the Upper Triassic Fleming Fjord Formation (Norian–Rhaetian) of East Greenland as pdf, ply and jpg files (3D model created by Oliver Wings; photographs taken by Jesper Milàn) is available at https://doi.org/10.6084/m9.figshare.c.2133546

During the Geocenter Møns Klint Dinosaur Expedition in July 2012, a new site with archosaur footprints was discovered by one of the authors (OW) in the lowermost part of the Ørsted Dal Member of the Upper Triassic Fleming Fjord Formation in mountain slopes facing the Carlsberg Fjord in East Greenland (Fig. 1). These archosaur footprints add to the knowledge of vertebrate life in the ancient lake and steppe system, and supplement previous descriptions of tetrapod footprints and other vertebrate trace fossils from the Upper Triassic deposits (Jenkins et al. 1994;Clemmensen et al. 1998;Milàn et al. 2004Milàn et al. , 2012Milàn & Bromley 2006;Niedźwiedzki et al. 2014;Sulej et al. 2014). They increase the diversity of tetrapod footprint assemblages from the Fleming Fjord Formation, reflecting a typical Late Triassic community with dinosaur and stem-crocodylian archosaurs. This is in congruence with the osteological record, which has an equivalent in assemblages of the Germanic Basin of central Europe (Jenkins et al. 1994). The new material, consisting of abundant and well-preserved imprints of identical shape, shows an anatomically based morphological peculiarity, and therefore requires a detailed discussion of the ichnotaxonomy and the possible trackmaker.

Geology and the depositional environment
The Jameson Land Basin lies in central East Greenland at 728 N (Fig. 1). The basin contains a seemingly complete (up to 300 m thick) Upper Triassic succession of lake deposits (Clemmensen et al. 1998). Stratigraphically, these sediments belong to the Fleming Fjord Formation, which is composed of a lower Edderfugledal Member, a middle Malmros Klint Member and an upper Ørsted Dal Member (Fig. 2) (Clemmensen 1980a, b). The latter is divided into a lower unit of red mudstones and sandstones (the Carlsberg Fjord beds), and an upper unit of light greyish dolomitic limestones and variegated mudstones (the Tait Bjerg Beds) (Clemmensen et al. 1998).
When the sediments of the Fleming Fjord Formation were deposited, the Jameson Land Basin lay at about 408 N, at the margin of the dry interior of the supercontinent Pangaea (Kent & Clemmensen 1996;Clemmensen et al. 1998;Kent & Tauxe 2005). Lake deposition was likely to have been influenced by seasonal, as well as orbitally controlled, precipitation changes (Clemmensen et al. 1998). The lake sediments record a gradual, long-term (a few million years) change in climate from semi-arid to humid, probably reflecting a slow northwards drift of the continent in latest Triassic times (Clemmensen et al. 1998;Kent & Tauxe 2005).
Deposits of the Fleming Fjord Formation are particularly well exposed in mountain slopes facing the Carlsberg Fjord. The tracks in question were found in the lowermost part of the Carlsberg Fjord beds in the Ørsted Dal Member at the Lepidopteriselv and Tait Bjerg sections (Fig. 2). The Carlsberg Fjord beds are here composed of red mudstones repeatedly interbedded by thin silt-and sandstones with wave and current ripples. The sedimentary characteristics of these deposits indicate deposition in a relatively shallow lake that frequently dried up (Clemmensen et al. 1998). The base of the unit contains two fluvial sheet sandstones and one lenticularshaped channel sandstone. The tracks from Lepidopteriselv locality were found in loose blocks at the base of the fluvial channel sandstone, and the sedimentary characteristics of the blocks indicate that they belong to the uppermost part of the composite channel sandstone (Figs 2 & 3).
Magnetostratigraphical investigation of the Fleming Fjord Formation suggests that the tracks from the basal part of the Ørsted Dal Member are about 210 myr old and, hence, of Norian age (Clemmensen et al. 1998). A diverse assemblage of fossil vertebrates also indicates that the main part of the Ørsted Dal Member is of Norian age (Jenkins et al. 1994).

Material and methods
Three slabs with footprints have been recovered (Fig. 1). Two of them have been found at a locality (Lepidopteriselv) at the eastern margin of the Jameson Land Basin in East Greenland (Fig. 1). These are loose blocks (consisting of several pieces), but sedimentological data indicate that they originate from the topmost part of a fluvial channel sandstone in the lowermost part of the Ørsted Dal Member of the Fleming Fjord Formation (Fig. 3). The third sample was collected at Tait Bjerg (Fig. 1). These footprints were seen in situ at the top of a fluvial sheet sandstone in the lowermost part of the Ørsted Dal Member. The material is stored in the collection of the Natural History Museum of Denmark (Copenhagen) (MGUH) under catalogue numbers MGUH 31233a -c, MGUH 31234 and MGUH 31235. Concave imprints were outlined with chalk, drawn onto transparency film and digitized using Adobe Illustrator CS3 software. Sillicone moulds  Clemmensen et al. (1998). New evidence from high-precision U-Pb geochronology, however, places this boundary at about 205.5 Ma (Wotzlaw et al. 2014), suggesting that the Norian-Rhaetian boundary should be moved upwards in the East Greenland section. and plaster casts were taken from the best-preserved footprints. Photographs were taken under artificial and natural light from both original slabs and casts, and all measurements were made from the original material only (Table 1).
Untextured 3D models from the slabs and highresolution models from the best-preserved tracks were produced with photogrammetry. Digital photographs were subsequently processed using the methods described by Mallison & Wings (2014) using the photogrammetric software Agisoft Photo-Scan Professional 1.0.4 build 1847 (64 bit) with generally high settings.

Formerly known material
The Ørsted Dal Member of the Fleming Fjord Formation has provided a rich tetrapod ichnofauna, including trackways of theropods with tridactyl footprints up to 28 cm long that can be assigned to the Eubrontes-Anchisauripus-Grallator plexus (Olsen et al. 1998;Gatesy et al. 1999;Milàn et al. 2004Lucas et al. 2006;Milàn & Bromley 2006). Large imprints of a quadruped, up to 53 cm in length, were also described (Jenkins et al. 1994;Clemmensen et al. 1998;Lockley & Meyer 2000). Recently, Niedźwiedzki et al. (2014) and Sulej et al. (2014) reported new discoveries of trackways of large quadrupeds (up to 50 cm pes length) from the Malmros Klint Member of the Fleming Fjord Formation and attributed them to sauropodomorph dinosaurs. These large tracks from Greenland have been compared with Eosauropus, a purported sauropodomorph ichnotaxon known from Upper Triassic deposits of North America and Europe (Lockley & Meyer 2000;Lockley et al. 2006Lockley et al. , 2011Niedźwiedzki et al. 2014). Until now, they were figured without details or only symbolically on a map of the track surface (Jenkins et al. 1994;Sulej et al. 2014). Details of their morphology and documentation using close-up photographs and sketches were not given by these authors, and their identification as large chirotheres, therefore, cannot presently be excluded. Jenkins et al. (1994, p. 19) also mentioned trackways of a smaller quadruped from the Ørsted Dal Member and small (2.5-5 cm) tetradactyl imprints showing 'four relatively short digits with rounded ends' from the Malmros Klint Member. The latter they consider as probable amphibian tracks. Again, no figures were provided and these authors did not indicate the whereabouts of the material. Probably, these footprints were left in the field.

New discoveries
The material described here consists of about 100 distinct pes and manus imprints made by similarly sized individuals distributed on three main slabs (

Discussion
Morphologically, the imprints described here show a strong similarity with chirotheriid footprints, in particular with the ichnogenus Brachychirotherium Beurlen, 1950 from the Late Triassic of Europe (Beurlen 1950;Karl & Haubold 1998Petti et al. 2009  This is unusual for Brachychirotherium and for all chirotheriids, especially in the pes imprint. Brachychirotherium occasionally lacks a trace of digit V due to substrate conditions and pes posture (Karl & Haubold 1998), but the large number of tracks from Greenland suggests a relationship of this feature to the pes anatomy of the trackmaker, which obviously had a functionally tetradactyl pes. However, the compact digit group I -IV in the Greenland tracks is typically chirotheriid. The only non-chirotheriid tracks of a quadruped that are similar to Brachychirotherium but show a functionally tetradactyl pes imprint is Batrachopus, mainly known from the Lower Jurassic but also from the Late Triassic (Fig. 11g). Batrachopus is considered to be a crocodylomorph track (Olsen & Padian 1986;Klein & Lucas 2010b). However, the manus imprint of Batrachopus is strongly rotated outwards relative to the pes imprint: whereas, in the material from Greenland, the manus imprint is inwards or slightly rotated outwards with respect to the pes imprint (cf. Fig. 11f & g). In typical Brachychirotherium tracks from the Late Triassic, the manus imprint is turned slightly more outwards than the pes (Fig. 11b-d).
Considering the overall congruence of the Greenland tracks with the ichnogenus Brachychirotherium, as well as the morphological peculiarity and lack of a (diagnostic) fifth digit trace, we tentatively refer the material described here to archosaur footprints cf. Brachychirotherium. Possibly, the lack of digit V generally characterizes juvenile Brachychirotherium (see below), requiring an emendation of the diagnosis of the ichnogenus. Late Triassic Brachychirotherium footprints of small size (,5 cm) are hitherto known only from a few imprints with a preserved fifth digit from a single locality in Argentina, South America (Fig. 11e). Small 'Brachychirotherium' tracks from the Middle Triassic of Germany (Haubold 1967(Haubold , 1971a also show a fifth digit, but the correct identification of the ichnogenus in these strata (originally Late Triassic) is doubtful (Karl & Haubold 1998;Klein & Haubold 2004). Alternatively, the tracks from Greenland could represent a new ichnotaxon, but this cannot be demonstrated here with confidence. Establishing a new ichnotaxon would require the presence of a juvenile Brachychirotherium and early ontogenetic stage to be excluded, and, instead, the presence of a small adult form to be proven. However, a new ichnotaxon would be based weakly on the lack of a fifth digit only.

Trackmakers
Brachychirotherium footprints have most probably been left by stem-crocodylian archosaurs (sensu Brusatte et al. 2010;Nesbitt 2011). The pes of the latter is distinguished from that of dinosaur bird-line archosaurs (Avemetatarsalia sensu Benton 1999) by the broader configuration of the metatarsals. This feature is reflected in the typically broad pes imprints of Brachychirotherium. An exception to this rule is the pes of the stem-crocodylian Poposaurus from the Late Triassic of North America, which has a slender shape developed convergent with that of dinosauromorphs (Farlow et al. 2014). Brachychirotherium has been attributed to rauisuchians, crocodylomorphs and aetosaurs by different authors Avanzini et al. 2010;Heckert et al. 2010;Lucas & Heckert 2011;Desojo et al. 2013).
Small Brachychirotherium from Upper Triassic deposits are uncommon, and the tracks from Greenland thus far could represent the smallest known. A detailed study of different-sized specimens and their morphologies has never been undertaken. Chirotheriid footprints reflecting juvenile individuals have been described from the Middle Triassic (Anisian) of Germany (Haubold 1967(Haubold , 1971a. Avanzini & Lockley (2002) gave a detailed analysis of different Isochirotherium size classes from the Middle Triassic of the Southern Alps (Italy), with some specimens having a pes length of 2.7 cm. However, these very small imprints show all features diagnostic of the ichnogenus Isochirotherium and of chirotheriids in general, including an impression of the fifth pedal digit. The Greenland tracks represent the second occurrence of small-sized cf. Brachychirotherium footprints from Upper Triassic deposits. They could possibly represent juveniles and ontogenetic stages with a still less-developed (not impressed) digit V or functionally tetradactyl pes. The wide variation in the position and orientation of the manus imprint (Fig. 10) might reflect the still unconsolidated gait of juvenile individuals of basal archosaurs. Possibly, juveniles also walked with a more digitigrade posture of the pes, where the short fifth digit did not touch the ground and therefore left no impression. During growth and increase in body weight, this might have changed towards a semi-plantigrade or plantigrade posture, leaving footprints with distinct sole and digit V impressions. Allen et al. (2009Allen et al. ( , 2010 demonstrated the shift of the centre of mass (COM) during ontogeny in extant crocodylians and birds. Crocodylians show a craniodorsal shift of the COM and an increase in loading on the hip joints that essentially influence locomotor mechanics (Allen et al. 2009). For extant crocodylians, Farlow & Britton (2000) showed that during growth the autopodia become shorter relative to the glenoacetabular length. However, comparing the lengths of the toe region and the plantar/palmar portion of the feet relative to total leg length, the digits show slightly negative allometric growth, whereas values for the sole indicate isometric growth (Farlow & Britton, 2000). This would mean that the sole becomes slightly longer relative to the digit portion. For the theropod dinosaur Allosaurus, Foster & Chure (2006) showed negative allometric growth of metatarsal IV relative to the tibia. In any case, the increase in weight on the hind limbs, combined with the relative shortening of metatarsals, as in some digitigrade bipeds, or the elongation of the palmar/plantar portion in more plantigrade quadruped forms, affect the biomechanics and function of the pes. Whether the same patterns observed in extant archosaurs were present  in the Greenland trackmakers cannot be proved with certainty, but the degree of contact of pedal digit V with the substrate, and the preservation of its impression, was probably controlled by differential growth of limb parts.
The osteological tetrapod record of the Fleming Fjord Formation thus far yields temnospondyls, possible sphenodonts and lepidosaurs, turtles, phytosaurs, a possible rauisuchian, aetosaurs, pterosaurs, prosauropods, theropods, and mammals (Jenkins et al. 1994;Mateus et al. 2014;Sulej et al. 2014). From these, only stem-crocodylian archosaurs (rauisuchians, aetosaurs) might be cautiously considered as potential trackmaker candidates (Karl & Haubold 1998;Heckert et al. 2010;Lucas & Heckert 2011). An affinity of Brachychirotherium to aetosaurs was suggested by Lucas & Heckert (2011), who found several synapomorphies of footprints and pes skeletons. Skeletal remains of aetosaurs are known from the Fleming Fjord Formation by the taxa Aetosaurus ferratus and Paratypothorax andressi (Jenkins et al. 1994). The size of the pes of Aetosaurus is in the range of the tracks described (Schoch 2007): however, the skeleton shows a relatively long digit IV in the pes and manus, whereas the tracks have a short pedal and manual digit IV (IV , II). Unfortunately, no trackway is preserved in the chirotheriid material from Greenland, and a comparison with the pattern of Brachychirotherium trackways from other localities is not possible. It is important to mention that the occurrence of body fossils and footprints in the same stratigraphic unit does not necessarily indicate a relationship. For example, the Olenekian-Anisian (Lower-Middle Triassic) Moenkopi Formation/Group is dominated by amphibian body fossils, whereas the tetrapod footprint record shows only a few amphibian, but abundant reptile, tracks (Peabody 1948;Klein & Lucas 2010a). This is probably due to alternating ecological conditions with different precipitation and water supply under a semi-arid climate and/or a taphonomic effect due to the different conditions under which skeletons or footprints can be preserved.
Following Lucas (1998Lucas ( , 2010, the vertebrate body fossil assemblage in Greenland belongs to the Revueltian land-vertebrate faunachron (LVF), which can be cross-correlated with the Carnian -Norian as it is similar to typical associations in Central Europe (e.g. from the German Keuper). In particular, the occurrence of the prosauropod Plateosaurus resembles assemblages from the Germanic Basin, whereas the Triassic bone record from North America lacks prosauropods. Against the background of the close proximity of what is today Greenland and North America in Late Triassic times, this can possibly be explained by the existence of larger barriers such as mountain ranges and/or climatic obstacles. In a reconstruction of climate belts on the Late Triassic supercontinent Pangaea, Kent et al. (2014) placed the Germanic Basin and the Jameson Land Basin in a relatively humid climate belt, while the North American localities are situated in a relatively arid climate belt. This climatic zonation may have controlled faunal variation on the continent. However, tetrapod footprints of the ichnogenus Eosauropus, originally described from the Upper Triassic of North America and possibly present in the Flemingfjord assemblage (see above), have been attributed to sauropodomorphs (Lockley et al. 2001(Lockley et al. , 2011. Interestingly, the Late Triassic record of the Argana Basin of Morocco, which was positioned close to eastern North America prior to the Atlantic rifting, also lacks sauropodomorph body fossils. Purported prosauropods from Morocco (Azendohsaurus: Jalil 1999) have more recently been referred to basal archosauromorphs (Flynn et al. 2010). In any case, the footprint record, and the presence of the characteristic and widely distributed Upper Triassic morphotype Brachychirotherium in the Fleming Fjord Formation of Greenland, indicates a distinct faunal exchange of trackmakers within Triassic Pangaea.
In Late Triassic times, the Jameson Land Basin lay at 408 N in a transition zone between the relatively dry interior of the supercontinent Pangaea and the more humid peripheral part of this continent (Kent & Tauxe 2005). Much of the interior of the continent was influenced by monsoonal rain, and shallow lakes were formed. The archosaur footprints were found in a fluvial channel sandstone in the lowermost part of the Carlsberg Fjord beds during a period when the overall climate of the basin was relatively dry (Kent et al. 2014). The formation of a relatively thick channel sandstone may record intense monsoonal rain transporting sand into the marginal part of the lake system (Clemmensen et al. 2015).
The lake deposits have been shown to contain a high-resolution record of orbitally forced climate change controlling the intensity of monsoonal rain with time (e.g. Kemp & Coe 2007;Volmer et al. 2008). Previous studies of sedimentary cycles in the Late Triassic lake deposits of the Jameson Land Basin suggest that this lake system too was influenced by orbitally forced climate change (Clemmensen et al. 1998). Detailed studies of facies, colour variation, measurements of magnetic susceptibility, gamma-ray intensity and carbonate content were used in a recent study by Frobøse & Clemmensen (2014) to detect statistical cyclicity in the uppermost part of the Carlsberg Fjord beds. These new data indicate, in agreement with Clemmensen et al. (1998), the existence of a composite cyclicity in the Carlsberg Fjord beds, with significant cycles. This cyclicity matches well the theoretically expected cyclicity in a lake system that experienced orbitally controlled variation in precipitation and lake environment (Clemmensen et al. 2015).
The track-bearing Carlsberg Fjord beds form part of a lake succession that indicates a long-term shift in climate from semi-arid to humid at the transition from the Norian to the Rhaetian (Clemmensen et al. 1998(Clemmensen et al. , 2015. This long-term shift in climate has been explained by the slow northwards drift of the continent in the Late Triassic (Clemmensen et al. 1998;Kent & Tauxe 2005).

Conclusions and perspectives
Clarification of the question of whether the hereindescribed tracks represent a very unusual morphotype of Brachychirotherium, possibly an early ontogenetic stage or a new ichnotaxon, must await additional data. Studies of ontogenetic growth in the autopodia of basal archosaurs based on footprints are little known. Future discoveries of differentsized specimens in the Fleming Fjord Formation and a potential distinction of different size classes would be of great importance. They might shed light on the differential growth of chirotheriid trackmakers, and, furthermore, permit a re-evaluation of the extremely large forms reported by Jenkins et al. (1994), Niedźwiedzki et al. (2014) and Sulej et al. (2014) from the same unit (?Brachychirotherium or ?Eosauropus) that could possibly represent adult individuals.