Sedimentology and ichnology of the Mafube dinosaur track site (Lower Jurassic, eastern Free State, South Africa): a report on footprint preservation and palaeoenvironment

Geological background and stratigraphy of the tracksite

The Mafube dinosaur track site is located between Fouriesburg and Clarens in the eastern Free State, South Africa (Fig. 1) on the properties of Mafube Mountain Retreat (formerly known as Groot Verwulf 274, and prior to that as eastern Blackwood 329). The region is dominated by the Upper Triassic to Lower Jurassic clastic sedimentary rocks of the Stromberg Group (Molteno, Elliot and Clarens formations) as well as the 183 ± 1.0 Ma old intrusive and extrusive mafic rocks of the Drakensberg Group (Duncan et al., 1997; Fig. 1).

Stratigraphically, the track site is ∼35 m below the base of the Clarens Formation, and within the last of last sandstone of the Elliot Formation (Fig. 1C). This Upper Triassic-Lower Jurassic fluvio-lacustrine unit has an unconformable, sharp, regionally traceable lower contact with the fluvio-lacustrine and coal-bearing Molteno Formation and a conformable, chiefly gradational upper contact with the mainly aeolian Clarens Formation (Bordy, Hancox & Rubidge, 2004). The Elliot Formation crops out as a ring-shaped belt surrounding the Drakensberg Plateau and has a maximum thickness of nearly 500 m in the south that thins to <30 m in the north (Botha, 1967; Smith & Kitching, 1997). Near the study site, the unit ranges in thickness from ∼160–185 m in the region south of Mafube, and in northern Lesotho to about ∼120 m just east of Fouriesburg. Within the Elliot Formation there are major sedimentary facies differences which allow the subdivision of the formation into two informal units, namely the lower and upper Elliot formations (lEF and uEF; Bordy, Hancox & Rubidge, 2004).

Regionally, the upper Elliot formation consists of silty mudstones with intermittent sandstones that have a distinctive deep red to maroon colour with erratic, light grey mottles and an abundance of other pedogenic features (Bordy, Hancox & Rubidge, 2004). The sandstones of uEF are sheet-like, tabular and multi-storied bodies that are tens of metres wide and up to 6 m thick. Individual beds within the sandstone bodies have thicknesses ranging between 0.2–1 m and are separated by flat, internal erosional surfaces with geometries similar to the basal bounding surface of the multi-storied sandstone bodies. The bounding surfaces in the uEF sandstones are laterally continuous, parallel, and devoid of topographical irregularities greater than tens of centimetres (Bordy, Hancox & Rubidge, 2004). Internally, the tabular fine to very finegrained sandstones of the uEF are dominated by massive beds, horizontal lamination, low-angle cross-bedding, parting lineations, ripple cross-lamination, flaser and wavy bedding, mud draped surfaces, small-scale soft sediment deformations and bioturbation (Bordy, Hancox & Rubidge, 2004). Channel lags, comprising pedogenic nodule conglomerates, commonly occur at the base of sandstone bodies. These nodules range from rounded to sub-angular, are moderately sorted and white to red in colour. In addition to the pedogenic glaebule conglomerates (a diagnostic lithology of the uEF) mud pebble conglomerates may also form stingers at the base the upward fining successions.

The uEF sandstones are separated from one another by 0.5–10 m thick mudstone units that range from pure claystone to fine-sandy siltstone which are mostly massive but sometimes horizontally laminated (Bordy, Hancox & Rubidge, 2004). The mudstones contain in situ pedogenic calcareous concretions, irregular mottles, desiccation cracks, falling water level marks and mud drapes (Bordy, Hancox & Rubidge, 2004). The mudstones also contain a large diversity of vertebrate body fossils, including sauropodomorph dinosaurs, turtles, fishes, amphibians, crocodylians, advanced therapsids and early mammals as well as crustaceans (conchostracans) petrified wood, and carbonised and calcretised root traces (Bordy & Eriksson, 2015). The informal lithostratigraphic units of the Elliot Formation coincide with the informal biostratigraphic units of Kitching & Raath (1984), namely the ‘Euskelosaurus’ and Massospondylus Range Zones with the lEF and uEF, respectively. The Mafube dinosaur track site therefore falls within the Massospondylus Range Zone, and this is supported by the fossil remains reported from Mafube, which include the postcranial remains (scapula, coracoid and tarsal) of Massospondylus sp together with crocodylians remains (Protosuchus haughtoni) and a jaw from a large, basal archosaurian reptile (‘thecodont’, ?rauisuchid; Kitching & Raath, 1984). Less than 9 km from the Mafube track site, the following fossil remains have been reported in Kitching & Raath (1984): (a) dinosaurs (mostly Massospondylus sp.) from farms Glen Skye (121), Helpmekaar (251), La France (379), Saaihoek (194), Bloemhoek (330), Naudes Lust (334), Mizpah (164), De Villiersdrift (338), Langkloof (34), and Foutanie (331); (b) cynodonts (only Tritylodon sp.) from farms Saaihoek (194), Bloemhoek (330), Naudes Lust (334), and Langkloof (34); (c) basal archosaurian reptile (‘thecodont’, ?rauisuchid) from farms Foutanie (331); and (d) amphibians (skull and jaw fragments) from farms De Villiersdrift (338) and Langkloof (34). It is noteworthy that a large partial femur of the “Highland Giant,” a herbivorous sauropod was discovered only <17 km ENE of Mafube at the same stratigraphic level as the Mafube track site in close proximity to the contact of the Elliot and Clarens formations (McPhee et al., 2016).

Description of the host rock sedimentology

The Mafube tracks are preserved in a 300 m long, 2 m thick, very fine grained, yellow to cream-coloured, tabular sandstone unit within the uEF (Fig. 1). Internally, this host sandstone unit consists of up to ten distinct tabular sandstone beds each 20–25 cm thick. The last three sandstone beds are separated by exposed palaeosurfaces that contain vertebrate tracks, invertebrate traces and casts of desiccation cracks (Figs. 3–8). The lowermost palaeosurface, with an exposed surface area of ∼1 m by ∼20 m, approximately 40 cm below the top of the sandstone unit contains the most numerous and best-preserved tracks at Mafube and is therefore the focus of this study.

The primary sedimentary structures in the sandstone unit are horizontal laminations, low-angle crossbedding (Figs. 3A and 3D), ripple cross-lamination, ripple marks (Fig. 7A) and massive beds. Horizontal lamination is the dominant primary sedimentary structure and low-angle cross-bedding is more commonly observed than ripple cross-lamination (Figs. 3A, 3D and 7A). Secondary sedimentary structures penetrating into the sandstone unit are casts of desiccation cracks and bioturbation (Figs. 3E–3G). The primary sedimentary structures are locally disrupted by soft sediment deformation features, the majority of which are expressed as wavy structures of variable length (Figs. 3B and 3C). Some of these are considered undertracks, because of the distinct concave-up bending of the strata directly underneath tracks on the palaeosurface (Fig. 3C).

Table 1:

Anatomical measurements of key tracks discussed in text relating to the size differential between morphotypes A and B at the Mafube dinosaur track site.

See Fig. 2B for how the interdigit (i.e., divarication) angle, length (t), width (w), length of rear of phalangeal part of foot (r) have been measured in this study (cf. Olsen, Smith & McDonald, 1998). All distance measurements are in centimetres; angles are in degrees. N/A: Measurements could not be determined due to e.g., absence of digit impressions; absence of heel impression; presence of the natural cast obscuring the track See Table S1 for additional track measurements.

Footprint #	3	7	16	27	31	57	58	60	67	68	69	70
Length	33.2	34	16.6	16.9	33.4	33.2	N/A	25.4	24.73	27	N/A	28
Width	21.8	24.9	10.8	11.8	26.3	N/A	N/A	21.4	N/A	N/A	N/A	21
Ratio of length/width	1.52	1.37	1.53	1.43	1.27	N/A	N/A	1.19	N/A	N/A	N/A	1.33
r	21.2	20.2	9.2	9.4	22.5	20.1	N/A	17.8	N/A	N/A	N/A	N/A
Interdigit angle II∧IV	55.1	66.7	56.1	67.8	65.91	N/A	N/A	57.7	N/A	N/A	N/A	N/A
Hypex shape II∧III	V	U	V	V	V	U	Wide U	N/A	U	N/A	N/A	N/A
Hypex angle II∧III	43	N/A	72.59	N/A	46	N/A	N/A	N/A	N/A	N/A	N/A	N/A
Interdigit angle II∧III	24.8	35.91	29.94	35.06	30.13	35.28	N/A	N/A	N/A	N/A	N/A	N/A
Hypex shape III∧IV	U	V	V	V	V	N/A	Wide U	N/A	V	N/A	V	N/A
Hypex angle III∧IV	32	37.48	71.57	N/A	65	N/A	N/A	N/A	N/A	N/A	N/A	N/A
Interdigit angle III∧IV	30.3	30.8	26.16	32.75	35.78	N/A	N/A	N/A	N/A	N/A	35	N/A
Morphotype and Remarks	A	A	B	B	A	A	A	A; metatarsal impression	A	?A; Sediment collapse	?A; Desiccation cracks	?A; Ripple marks
Figure # in text	4A, 5A; 8C	4A	4B; 6C	4B	4B, 6A, 8A	4C, 5C; 8D	4C, 5C	4C, 7B	5B	7D	7C	7A

DOI: 10.7717/peerj.2285/table-1

Hexagonal casts of desiccation cracks that penetrate into the sandstone are preserved on the exposed track palaeosurface (Figs. 3E and 3F) and on the surface of some tracks (Figs. 5A and 7C). The casts of desiccation cracks are larger on the track surface (maximum diameter of 11 cm) than the smaller casts of desiccation cracks within the tracks (maximum diameter of 2.5 cm). Bioturbation on the track palaeosurface is due to invertebrate burrowing that comprises of cylindrical tubes that are non-branching, unlined, unornamented, and have uniform diameter along their length. The burrows are both horizontal and vertical (Figs. 3F and 3G) and lack internal structures (i.e., are filled with massive, very fine grained sandstone). The horizontal invertebrate traces are slightly curving to straight, ∼0.5 cm in diameter, and range in length from ∼3 to ∼22 cm. Surface texturing within some of the bedding plane depressions and tracks (Fig. 5C) comprise pits with variable diameters ranging from a few mm to 1 cm with a mean pit diameter on a mm-scale (∼4 mm). The subdued relief of the pits prevents more accurate measurements.

Description of the track site

Over 80 tracks have been identified at the Mafube dinosaur track site (Fig. 4) over a lateral distance of ∼80 m. The degree of track preservation varies and some of the better preserved tracks show claw marks as well as desiccation cracks, ripple marks, etc. (Figs. 5–8). Based on key biological and sedimentological features the lowermost palaeosurface is sub-divided into three surface tracts, named: Surface I, II and III (Fig. 4). There are additional tracks preserved on the same lowermost palaeosurface, but they are sparse and far apart (Fig. 7). Directional information obtained from 50 tracks of the palaeosurface is presented in a rose diagram (inset in Fig. 4). The orientation of the tracks is multi-directional and no individual trackways are recognised on the palaeosurface

Description of selected tracks

The summary below accounts for the key characteristics of the best preserved large (e.g., #3, 31, 57, 58, 67 of morphotype A) and small (#16 of morphotype B) tridactyl tracks as well as tracks of particular interest (#60, 68, 69, 70; see Figs. 4–8 for illustrations and Table 1 and Table S1 for measurements).

Track #3, a right footprint is one of the best preserved tracks and shows desiccation cracks indicating that it is a true track (Fig. 5A; cast in Fig. 8C; Table 1). Moreover, the track has a natural cast at the tips of digits II and IV, a 2.5 cm long claw mark on digit II and slightly curved, V-shaped digits, especially in case of digit II. The track walls are steep and the heel is clearly defined.

Track #67 retains most of its natural cast, therefore the depth of the impression is unknown (Fig. 5B; Table 1). A claw mark on the rightly-curved digit IV is present.

Tracks #57 and 58 are superimposed on one another (Fig. 5C) and point in different directions (angle of difference between digits III is ∼78°). Track #57 lacks digit IV (Fig. 8D; Table 1) and so its total width is unknown. Digits II and III curve slightly towards the tips. Pitted textures (∼4 mm in diameter) are observed in track #57 (Figs. 8D and 8F). The less clearly defined, shallow impression (not suited for length and width measurements) associated with track #58 preserves three rounded, straight digits.

Track #31 is one of the largest tracks at the Mafube site, a right pes (Fig. 6A; Table 1) with its digit II alone being 21 cm long. The digit impressions are deeper than the poorly defined heel and are relatively straight with distinct curvature towards their V-shaped tips. Curvature of digit II and IV is divergent. Steep track walls surround digit III which retains some of its natural cast.

Track #6 is a left pes which forms a depression of variable depths because its natural cast is still largely present (Fig. 6B; Table 1). All three digits are V-shaped. The cast (Fig. 8B) was made with leftover casting mixture which did not fully capture all the properties of the track, thus limiting accurate measurements.

Track #16 is half the size of the larger tracks being only 16.6 cm long and 10.8 cm wide (Fig. 6C; Table 1). This small and very shallow impression has straight digits with U-shaped tips.

Track #70 is isolated, morphologically indistinct and preserves ripple marks consisting of very fine grained sandstone (Fig. 7A; Table 1). The small ripple marks are asymmetrical and <1 cm high with a wavelength of 1.6 cm.

Track #60 preserves claw marks and a poorly preserved impression of the metatarsal (Fig. 7B; Table 1). Claw marks are preserved on all three digits and terminate in very acute angles. The natural cast obscures numerous morphological characteristics (e.g., shape of digits and hypex; inter-digit angles).

Track #69, a modified true track, preserves casts of desiccation cracks with desiccation polygons that range in diameter from 1.9–2.7 cm (Fig. 7C; Table 1). Although the indistinct morphology of the track prevents accurate length and width measurements, the following morphological characteristics can be observed: (1) impressions of digits II (?) and III; (2) the increasing depth of the impression towards the heel; (3) slight upward and downward curvature of digit III and IV, respectively; (4) V-shaped hypex; and (5) interdigit angles of 35°.

Track #68 lacks distinct morphology, however digits III and ‘ridges’ within digit II or IV (digit cannot be adequately identified) may be inferred (Fig. 7D; Table 1). Sediment-collapse features include the semi-parallel ridges surrounding U-shaped digit III as well as ridges in the impression of digit IV.

Genesis and preservation of the track surface

Interpretations of the high energy, flash-flooding and episodic drying out of the stream/river channels in the uEF (Bordy, Hancox & Rubidge, 2004) are well-illustrated at the Mafube dinosaur track site. It preserves low energy sedimentary structures (bioturbated, desiccated and cracked palaeosurfaces) that overlie and are separated from upper flow regime, very fine grained horizontally laminated sandstones. Ephemeral streams/rivers typical of semi-arid environments are also characterised by fairly sudden reductions in the volume of running water (due to evaporation and seepage), transforming them into very shallow ponds or pools with still-standing or slow moving water. It is within these stagnant water bodies that sediments may be rippled and colonized by microbial organisms and other invertebrates before desiccating. Sedimentary structures on the track surface show that the tracks were made in a dense, but soft, water-saturated sediment and, in some places, below the surface of a shallow water layer. This is because modified tracks show small, asymmetrical ripple marks, soft sediment deformation features (e.g., expulsion rims sediment-collapse structures), invertebrate burrows, and casts of desiccation cracks (Figs. 3–8).

Asymmetrical ripple marks within and around track #70 suggest that some impressions were likely made in an underwater substrate over which gentle, unidirectional currents moved (Fig. 7A). Moreover, the cast of desiccation cracks on the track palaeosurface (e.g., Figs. 5A and 7C) suggest that the sediment surface was covered in a thin, wet mudfilm that dried out and was ultimately covered by sandy sediments that also infilled the cracks and formed a natural casts (Fig. 9). The smaller desiccation cracks within tracks indicate that track depressions (e.g., tracks #3, 69 in Figs. 5A and 7C) were able to collect muddy floodwater. Upon evaporation, the muddy floodwater left behind a thin, mud-cracked sediment veneer similar to that observed in a modern track after a summer thunderstorm (Fig. 9).

Overall, the sedimentary structures preserved in and around the tracks show that the exposed surface is largely the original one on which the theropods moved, essentially contemporaneously, in the Early Jurassic. Therefore, the best-preserved tracks are faithful replicas of the anatomical features of the feet of these Early Jurassic dinosaurs. Taxonomically less valuable undertracks (e.g., Fig. 3C) have also been identified in the form of deformation features in the sediment layers below the original track surface. Such features are a combined result of the size and locomotion style of the trackmaker (e.g., large running adult) as well as the rheological properties of the sediment (i.e., sediment consistency) on which the animals moved (e.g., very fine grained, plastic, water-saturated sand layer) (Gatesy, 2003; Milàn, 2006; Milàn & Bromley, 2006).

From north to south, the tracks on the surfaces become less morphologically distinct resulting in a track morphology that is highly variable over the length of the study site. For instance, several tracks of similar size show variable depths of imprinting (e.g., tracks #3, 26, 60, 69 in Figs. 4, 5A and 7C). Furthermore, the amorphous quality of tracks on Surface III (Figs. 4E and 4F) is likely due to the different consistency of the substrate in that area of the palaeosurface (e.g., more water saturated conditions). The faint, amorphous nature of the isolated tracks within the southern portion of the site further south of Surface III may also be due to sediment-collapse features that obscured the original track as the substrate was too wet to allow the production and preservation of well-defined tracks. Ripple marks, expulsion rims and other soft sediment deformation features resembling squish marks (Fig. 7) are only observed together with the isolated tracks found in the southern section of the site. In contrast, tracks express more detail (e.g., claw marks, desiccation cracks; track #3 in Figs. 4A, 4B and 5A) in the north. This trend in track preservation with tracks to the south having a generally lower degree of anatomical preservation and higher abundance of associated soft sediment deformation features (such as expulsion rims, e.g., tracks #31 and 68 in Figs. 6A and 7D, respectively) is evidence for the changing consistency of the substrate. This is specifically indicative of an increasingly moist substrate along the palaeosurface to the south, possibly due to the proximity of a water pool. The preservation of metatarsal impressions (track #60 in Fig. 7B) are likely related to the animals’ movement in the more water saturated substrate (cf. Gatesy, 2003). The presence of metatarsal impressions and slipping tracks (e.g., tracks #60, 68 on Figs. 7B and 7D) on Surface II reinforces the interpretation of a water saturated substrate.

Despite the lack of trackways within the multi-directional tracks at Mafube, when quantified (inset in Fig. 4), the track orientations reveal a dominant movement direction to the south-east and a less dominant one to the north-west. Locomotion towards the south-east is compatible with the position of a potential water source (e.g., ephemeral water body) to the south-east of the palaeosurface as indicated by the sedimentological and track taphonomic results. Therefore, it may be assumed that the tracks indicate animal movement towards this water source for drinking and/or feeding purposes, and that not all animals returned along the same route (to the north-west). Furthermore, the close spacing and multi-directionality of the tracks (Fig. 4) suggest a moderate degree of trampling especially in the area of the track palaeosurface shown in Figs. 4C and 4D.

Events following the formation of a track are crucial for its preservation, because exposed tracks are highly degradable via erosional processes acting on the sediment surface (e.g., wind or water currents). Therefore, at Mafube, the solidification of the imprinted layer had to be relatively rapid, because the deposition of the overlying sediment layer occurred without producing erosional features. Instead, during the deposition of the overlying sediment layer, sandy sediment infilled (and formed natural casts in) the delicate desiccation cracks and depressions between ripple marks as well as theropod foot impressions. Given that the sedimentary structures in the overlying very fine grained, horizontally laminated sandstones indicate upper flow regime conditions, the pervasive solidification of the palaeosurface was likely enhanced not only by subaerial exposure (i.e., moisture evaporation), but potentially also by microbial mats that biostablised the granular, sand-rich sediment layer, and ultimately produced a more cohesive and thus less erodible substrate (cf. Carvalho, Borghi & Leonardi, 2013; Fig. 10). The pitted texture in Surface III (Figs. 4E, 4F, 5C, 8D and 8F) somewhat resembles both experimentally produced biofilm covered surfaces (e.g., Mariotti et al., 2014) and “wrinkled textures” described by Parizot et al. (2005; see their Fig. 15), Schieber et al. (2007) and Mariotti et al. (2014; see their Fig. 3A) which have been interpreted as the remnants of microbial mats reinforcing sandy, granular substrates. The partially subaqueous nature of the Mafube palaeosurface (explained in detail below) may have supported microbial growth, especially in the tracks that may have acted as pools for stagnant water (Fig. 10). These puddles would have allowed for localized algal blooms despite rapid drying or absorption on the bedding surface as a whole. Nonetheless, the quality of the pitted texture is poor and variable within and between tracks and therefore it is hard to show conclusively that microbial mats were widespread or even present at Mafube. All in all, there is no credible evidence to wholly support this texture as being part of a microbially induced sedimentary structure, and at best is should be characterised as a sedimentary surface texture sensu Davies et al. (2016).

Identity of the possible trackmaker

Tracks found to date at the Mafube are tridactyl, barring several which are poorly preserved (e.g., Track #69) giving the impression of being didactyl. The anatomical detail of the tracks decreases towards the southern end of the site. The lack of manus impressions is indicative of obligate bipeds and eliminates quadrupedal trackmakers. We link the Mafube tracks to theropods based on the following criteria that distinguishes tracks of tridactyl theropods from ornithischians: (1) long, slender digits, that are asymmetrical, taper; and (2) often end in a sharp claw impression or point (e.g., track #3 in Figs. 4A, 4B and 5A); and (3) the tracks being longer than broad (Olsen, Smith & McDonald, 1998; Raath & Yates, 2005).

If the interpretation of the Mafube tracks being made by large and small theropod dinosaurs is correct they can then be compared to body fossils of theropods within the uEF and Clarens Formation in South Africa and Lesotho. To date, there are only two known theropod dinosaurs capable of producing such tridactyl tracks in the Lower Jurassic geological record of the southern Africa: Dracovenator regenti (Yates, 2005) and the coelophysid theropod, Coelophysis rhodesiensis (formerly Syntarsus rhodesiensis Raath, 1977; Bristowe & Raath, 2004). The latter is known from the Lower Jurassic Karoo-aged deposits in Zimbabwe (Raath, 1977), whereas fragmentary remains of Coelophysis sp. have been identified in the uEF of South Africa (Kitching & Raath, 1984; Munyikwa & Raath, 1999). Globally, the type species of Coelophysis (Coelophysis bauri) and an anomalously high abundance of specimens are known from the Upper Triassic Chinle Formation (Schwartz & Gillette, 1994) and remains have also been reported in the Kayenta Formation (Glen Canyon Group; Rowe, 1989) and Moenave Formation (Lucas & Heckert, 2001) in the Lower Jurassic of the USA.

Potential ichnotaxa

The Mafube track surfaces preserve tridactyl tracks which can be distinguished on the basis of their size. The commonly occurring morphotype A has a length averaging 32 cm in contrast to rare morphotype B which has a length between 10–13 cm (e.g., tracks #16, 47 in Figs. 4C, 4D and 6C). Both Mafube morphotypes can be attributed to theropod trackmakers. Morphotype A represents medium to large animals and morphotype B represents smaller animals or juvenile individuals of the larger trackmaker. Given the track morphology, only two theropod dinosaur ichnotaxa are possible candidates for the tridactyl tracks at Mafube. Morphotype A may be related to the ichnotaxon Eubrontes based on morphological characteristics while morphotype B may be considered more representative of the ichnotaxon Grallator.

Comparison to Lower Jurassic theropod tracks of southern Africa

The density, preservation and diversity of Lower Jurassic dinosaur tracks in Lesotho far outnumber other sites in South Africa. Tridactyl tracks ≤35 cm in length have been described by Ellenberger (1970), Ellenberger (1972) and Ellenberger (1974) from the lower and upper Elliot and Clarens formations of Lesotho and been assigned to various ichnogenera such as: Kainotrisauropus isp. Qemetrisauropus isp., Prototrisauropus isp., Deuterotrisauropus isp. However, several authors (e.g., Lucas et al., 2006; D’orazi Porchetti & Nicosia, 2007) consider all these to be synonymous with Eubrontes. Furthermore, Olsen & Galton (1984) synonymized all occurrences of Kainotrisauropus with Grallator.

The larger Eubrontes-like Mafube morphotype A shows similarities in size and general morphology with a number of Lower Jurassic tracks in South Africa (Uniondale trackway; Raath & Yates, 2005), Lesotho (Moyeni tracksite; Wilson, Marsicano & Smith, 2009) and Zimbabwe (Ntumbe River, lower Zambezi Valley; Munyikwa, 1996). The five successive tracks forming the Uniondale trackway were considered to be similar to Grallator as well as the ichnogenera Kainotrisauropus (of Ellenberger, 1970) by Raath & Yates (2005). These Uniondale tracks are morphologically similar to tracks from Ntumbe River (Zimbabwe) which Munyikwa (1996) assigned to Eubrontes. Moreover, the recently re-analysed Moyeni tracks in the uEF that were originally referred to as Neotrisauropus by Ellenberger (1974) have been incorporated into the ichnogenus Grallator by Wilson, Marsicano & Smith (2009). Although the Moyeni tracks are larger (∼28 cm in length) than typical tracks assigned to Grallator Wilson, Marsicano & Smith (2009) argued that the tracks are below the maximum lengths of similar tracks from the Newark Supergroup, and considered them to be produced by a theropod similar to Dracovenator regenti.

The foregoing is reflective of the on-going ichnotaxonomic debate on Upper Triassic-Lower Jurassic tridactyl, non-avian theropod tracks (e.g., Olsen, 1980; Olsen, Smith & McDonald, 1998; Lockley, 1998; Rainforth, 1997; Rainforth, 2005; Lucas et al., 2006; Lockley, 2010). The difficulty with qualitatively distinguishing these tracks is mainly caused by the simplicity of the track morphology and the relatively conservative foot anatomy of non-avian theropods over most of the Mesozoic. Among others, this ichnotaxonomic discourse was examined by Lockley (1998), who briefly accounted for the various quantitative methods for characterising the morphological differences in tridactyl tracks. Some of these approaches have been gaining momentum in more recent studies (e.g., geometric morphometrics in Castanera et al., 2015).

Currently there are four widely used Lower Jurassic ichnogenera of tridactyl, non-avian theropod tracks (Grallator, Anchisauripus, Eubrontes and Kayentapus) despite the current ichnotaxonomic debate. Several other ichnotaxa are considered either behavioural/preservational variations (e.g., Gigandipus) or junior synonyms of the four commonly considered valid ichnogenera above. Size has long been the key separating criteria in the GrallatorAnchisauripusEubrontes or GAE allometric plexus (Olsen, Smith & McDonald, 1998; Rainforth, 2005; Petti et al., 2011), and synonymising the three ichnogenera into a single valid ichnogenus (Grallator or Eubrontes) has been long considered and debated (e.g., Olsen, 1980; Olsen, Smith & McDonald, 1998; Rainforth, 2005; Lucas et al., 2006). The ichnogenera Kayentapus and Anchisauripus have been incorporated into the GrallatorEubrontes spectrum largely because of their intermediate size between the two ichnogenera (Olsen, Smith & McDonald, 1998; Lockley, 1998; Milner et al., 2009). The former, with its wide divarication angles and gracile form has been either synonymised within Eubrontes (Lucas et al., 2006) or considered as a valid ichnogenus by several authors based on its morphology (Lockley, 1998; Milner et al., 2009; Lockley, Gierlinski & Lucas, 2011).

The foregoing also highlights the plight of assigning relatively large tridactyl tracks to the GrallatorEubrontes spectrum, two ichnogenera that are in essence morphologically similar, except for their size. Because track size is naturally variable its ichnotaxonomic utility in this case is questionable (e.g., Rainforth, 2005; Lucas et al., 2006). For example, considering its average length of ∼32 cm, the Mafube morphotype A could be placed between the Moyeni (length 28 cm) and the Uniondale (length 39.7 cm) tridactyl tracks which are the two current end-members of the GrallatorEubrontes spectrum in southern Africa. In spite of this problematic, size-based distinction between the otherwise morphologically similar Grallator and Eubrontes ichnogenera, we consider the larger Mafube dinosaur tracks (morphotype A) to be more evocative of Eubrontes and the smaller morphotype B to be more similar to Grallator.