Two thin-toed tridactyl tracks in a fluvial sandstone bed of the Eumeralla Formation (Albian) at Dinosaur Cove (Victoria, Australia) were likely made by avian trackmakers, making these the oldest known fossil bird tracks in Australia and the only Early Cretaceous ones from Gondwana. These tracks, which co-occur on the same surface with a slightly larger nonavian theropod track, are distinguishable by their anisodactyl form, hallux impressions and wide digit II–IV divarication angles. A lengthy hallux impression and other deformational structures associated with one track indicate foot movement consistent with an abrupt stop, suggesting its tracemaker landed after either flight or a hop. The single nonavian theropod track is similar to other tracks described from the Eumeralla Formation at another locality. The avian footprints are larger than most Early Cretaceous avian tracks recorded worldwide, indicating sizeable enantiornithine or ornithurine species in formerly polar environments of Australia. The avian tracks further supplement scant body fossil evidence of Early Cretaceous birds in southern Australia, which includes a furcula from the Wonthaggi Formation. Because of this discovery, Dinosaur Cove, previously known for its vertebrate body fossils, is added to a growing list of Early Cretaceous vertebrate tracksites in southern Australia.
Despite intensive study of vertebrate body fossils from Aptian–Albian rocks of coastal Victoria, Australia, during the past 30 years (Molnar et al. 1981; Rich et al. 1988; Rich and Vickers-Rich 1994, 1999; Currie et al. 1996; Rich and Vickers-Rich 2000, 2003; Rich et al. 2002, 2010), fossil vertebrate tracks are rare in these strata. Researchers have documented only one notable dinosaur track locality to date: Milanesia Beach in the Eumeralla Formation (Albian) along with individual footprints from a few other sites (Rich and Vickers-Rich 2003; Martin et al. 2007, 2012). The least common vertebrate body fossils from the Aptian–Albian of Victoria are those of birds; their fossil record may consist only of several feather impressions and a furcula from the Wonthaggi Formation (Aptian) (Close et al. 2009). However, considering the large number of nonavian feathered theropods now known (Norell and Xu 2005; Xu et al. 2009, 2010; Zelenitsky et al. 2012), the identity of these feathers as solely avian is also suspect. In other words, one furcula may be the only definite body fossil evidence of Early Cretaceous birds in southern Australia. Avian remains are likewise rare in Early Cretaceous strata elsewhere in Australia, found only in Albian rocks of New South Wales (Molnar, 1999) and Queensland (Molnar 1986; Kurochkin and Molnar 1997).
This paper reports two relatively large avian footprints and a nearby nonavian theropod track from the Lower Cretaceous Eumeralla Formation. These tracks are preserved in a loose block originating from a fluvial sandstone only a few metres above the bone bed at Dinosaur Cove (Rich and Vickers-Rich 2000). The avian tracks are the oldest known in Australia, the first reported from the Early Cretaceous of Gondwana, and among the oldest in the Southern Hemisphere, depending on the identity and age of avian-like tracks from South Africa (Ellenberger 1972, 1974). These footprints therefore signify another Early Cretaceous vertebrate tracksite in Victoria, and only 9 km from a recently discovered dinosaur tracksite at Milanesia Beach (Martin et al. 2012). The tracks are also at the same location as many dinosaur body fossils (Rich and Vickers-Rich 2000). Furthermore, one of the tracks displays evidence consistent with flight made by landing, further supporting its avian identity.
Study area, discovery, initial diagnosis
The tracks were on an isolated sandstone slab among other displaced blocks on a marine platform, located about five metres south-east of the entrance to the Slippery Rocks Tunnel of Dinosaur Cove, Victoria. The tunnel, from which Museum Victoria and Monash University excavated a bone bed at Dinosaur Cove in 1984–1994 (Rich and Vickers-Rich 2000, 2003), is now sealed and marked by a commemorative plaque (Fig. 1A). The block likely came from a thick sandstone bed immediately above the bone bed. The palaeontological content and geological context of Dinosaur Cove is well documented and was the primary site for dinosaur exploration in Victoria before 1994 (Rich et al. 1988; Currie et al. 1996; Rich and Vickers-Rich 1999, 2000; Rich et al. 2002, 2010). Still, the resulting collection of body fossils continues to yield surprises, including the oldest crayfish body fossils from Gondwana (Martin et al. 2008) and the only known tyrannosauroid from Australia (Benson et al. 2010). However, before this study, neither invertebrate nor vertebrate trace fossils had been noted at this locality.
All exposed strata along the marine platform at Dinosaur Cove are of the Eumeralla Formation (Albian), which is composed of fluvial sandstones, conglomerates, breccias, mudstones and coals formed in the Otway Basin (Rich et al. 1988; Bryan et al. 1997; Tosolini et al. 1999; Miller et al. 2002). During the Early Cretaceous, the southern margin of Australia was still connected to Antarctica and this area lay close to the South Pole, at about 75–80 degrees south palaeolatitude (Veevers et al. 1991, Veevers 2006). Hence, its fossil assemblage is part of a circumpolar biota and is the most complete biota of its kind described from Mesozoic rocks in the Southern Hemisphere (Rich et al. 2002). The bone bed is interpreted as an overbank splay deposit (Rich et al. 1988; Rich and Vickers-Rich 2000). Strata both below and overlying it, including the probable source bed for the tracks, are likely fluvial floodplain deposits, which is corroborated by our sedimentological analysis of the track-bearing sample.
Discovery, initial diagnosis and methods
Sean Wright discovered the sandstone slab containing the tracks while he and Alan Tait were prospecting at Dinosaur Cove on 29 November 2010 (Fig. 1B). They photographed its surface and noted its location near the Slippery Rocks Tunnel. The slab had dislodged from a bed directly above the former excavation site and was evidently transported by waves a few metres laterally along the marine platform. Tait recovered the tracks on 31 March 2011 by splitting the uppermost part from the main slab, which he then broke into four pieces to facilitate transport on foot out of Dinosaur Cove. Tait contacted one of the authors (THR) after collecting the specimens and brought the pieces to Museum Victoria, where all were collectively assigned museum specimen number NMV-P231650 (Fig. 2). On 4 April 2011, THR contacted another of the authors (AJM) to seek identification of the trace fossils depicted in Tait's field photographs. AJM verified that these trace fossils were likely vertebrate tracks, similar to ones he had identified from Milanesia Beach (Martin et al. 2012). AJM and THR directly examined the tracks at Museum Victoria 11–12 July 2011. The lithology was described then by AJM. Tracks were measured using Mitutoyo digital callipers, and each footprint was sketched before photographing. MH reexamined the lithology in more detail on 23 September 2013 and based his analysis on a section cut from the slab (Fig. 2).
Track measurements were rounded to the nearest 1 mm. For all three tracks, track length was measured from the distal to proximal ends of digit III, and width was based on the distance between the outermost impressions of digits II and IV. All measurements of the track interior were taken from a minimum outline, where changes in slope between the track bottom (‘floor’) and sides (‘wall’) were used as reference points for measuring. Where digits may have bent, as observed in one of the specimens, digit length was measured as the straight-line distance from the proximal to the distal end. Divarication angles were measured in the same manner as described and illustrated by Scott et al. (2012).
The tracks are preserved in fine-grained, lithic (volcaniclastic) sandstone with a rippled upper surface. Ripple amplitude is 1–2 cm, and thickness of the slab varies from 5.5 to 9.9 cm. Based on a cut sample from the specimen (Fig. 2), the lower portion of the bed is a laminated fine-grained sandstone that, in a 10-mm interval towards its top, grades upward into a 2-mm-thick layer of darker coloured, very fine-grained sandstone. A 12-mm-thick layer of fine-grained sandstone overlies this, with foresets that are likely small-scale cross-beds. This interval is capped by 12–13 mm of laminated, fine-grained sandstone that also could be part of another thin foreset bed. Silt and clay are apparently minor components of the sediment. The top few centimetres of the bed also contain sparse subhorizontal burrows, which are 5–11 mm wide, sand-filled and with circular cross-sections expressed on the top surface of the bed. At least one observed burrow had internal menisci, akin to Taenidium isp., but with no other distinguishing features.
Tracks are evident on the top surface as three negative-relief impressions with positive-relief structures (ridges) on their peripheries and interiors. As mentioned before, the slab was broken into four pieces when collected, with one track on the each of two pieces (Tracks 2 and 3) and the third (Track 1) split between the remaining two parts. This breakage, travelling along the long axis of the track, resulted in the loss of a small part of the upper sandstone surface. Fortunately, this damage did not adversely affect primary features of the track.
This track, interpreted as a right pes impression, has four digit impressions (I–IV) and several prominent structures directly associated with these impressions (Fig. 3A, B). The track is slightly wider (108 mm) than long (91 mm), yielding a length/width ratio of 0.84 (Table 1). However, digit III length is likely truncated where its distal part coincides with the outermost boundary of the rock; hence, 91 mm represents a minimal length and the original length/width ratio was probably larger than 0.84. Using the entirety of the hallux impression (digit I) would result in a total track length of about 17 cm, but this measurement does not accurately reflect the actual length of the tracemaker's foot, as explained later. This track also has forwardly pointing digits II–IV, with a symmetrical plane defined by digit III, constituting a mesaxonic condition (Leonardi 1987; Lockley et al. 2009). A medially offset hallux further defines this track as anisodactyl (Elbroch and Marks 2001). Divarication between digits II–IV is 120 degrees, combining 59 degrees (II–III) and 61 degrees (III–IV), and the interdigital angle between digits I–II is 86 degrees. Other digit lengths were 21 mm (digit I), 62 mm (digit II) and 68 mm (digit IV). Digit I length was difficult to discern precisely, but estimable from where the digit and its ungual left an impression (Fig. 3C–D). Digits are relatively thin compared to overall track size, with widths ranging from 6.1 to 6.7 mm. Thin (1- to 2-mm-wide) grooves, denoting ungual impressions, are associated with digits I and III, but are not visible on the distal ends of digits II and IV. Pad impressions were not observed on any of the digit impressions, and interdigital webbing is not clearly demonstrated, either.
Table 1. Measurements of fossil tracks from Eumeralla Formation, Dinosaur Cove, Victoria
W, total width; L, total length; I-W, digit I width; I-L, digit I length; II-W, digit II width; II-L, digit II length; III-W, digit III width; III-L, digit III length; IV-W, digit IV width; IV-L, digit IV length; I-II, interdigital angle between digits I and II; II–III, interdigital angle between digits II and III; III–IV, interdigital angle between digits III–IV; II–IV, total divarication between digits II and IV; NA, not applicable.
Lengths and widths in cm.
Pressure-release structures are present in the track, evident as variations in relief around and within the track. Pressure-release structures (sensu Martin 2009; Martin et al. 2012), or ‘indirect features’ (Gatesy et al. 1999; Gatesy 2003), are variations in relief caused by the foot deforming sediment within and outside of the main footprint outline. The most prominent of these structures is a heightening of sand as a ridge around the anterior and distal parts of digit IV, which is aligned with the length of the hallux impression (Fig. 3A, B). Smaller ridges are also associated with the anterior edge of the lengthy digit I impression, the distal end and proximal edge of digit II, the right side of digit III and the middle of the digit III impression. The last of these structures curves to the right and is bisected by a thin (1- to 2-mm-wide) medial groove, which is attributed to an ungual impression of digit III (Fig. 3D). Track depths below the main sandstone surface range from 1 to 7 mm, whereas pressure-release structures are as high as 6 mm.
This track is a left pes impression and, like Track 1, has four digit impressions (I-IV), is anisodactyl and mesaxonic. It is wider (114 mm) than long (107 mm), with a length/width ratio of 0.94 (Fig. 4; Table 1). One of the digit impressions is of a retroflexed hallux (digit I) posterior to digit II. Divarication between digits II–IV is 95 degrees, with II–III at 46 degrees and III–IV at 49 degrees, and the interdigital angle between digits I–II is 75 degrees. Digit widths ranged between 6.5 and 7.5 mm, and digits II and III end with narrow (<1-mm-wide) ungual impressions. Digit III length was the same as track length (107 mm). Digit I, II and IV lengths were 29, 77 and 83 mm, respectively. The right posterior margin of the track has prominent deformation structures, which complicated measurements of digit I width, as well as digit II and its interdigital angle with respect to digit III. Consequently, measurements reported here represent best estimates.
Pressure-release structures directly associated with the negative-relief digit impressions include a ridge surrounding the posterior edge and distal part of digit IV; a broad plate on the anterior edge of digit IV; a thin ridge on the left edge of digit III that connects with a plate around the left distal part of that same digit; a subcircular mound on the posterior part of the digit IV impression; a curved ridge along the anterior edge of digit II; and an irregular ridge posterior of digit II. The last of these is associated with soft-sediment deformation features that may be additional pressure-release structures. Track depths below the main sandstone surface range between 3 and 5 mm, and pressure-release structures are as high as 8 mm. The most obvious morphological difference between this track and the other two is a pronounced curvature of digit II. Posterior to this digit is another apparent impression of it, but nearly linear.
An invertebrate burrow, one of the few evident on the top surface of the sandstone, is towards the proximal end of digit IV (Fig. 4B). The burrow is tentatively assigned to the ichnogenus Taenidium isp. based on meniscae observed in similarly sized burrows in the top few centimetres of the bed, but it had no other distinguishing features. We could not discern whether this burrow cross-cut the track, or the trackmaker had stepped on the burrow.
This track is a right pes impression, tridactyl, mesaxonic and slightly larger than tracks 1 and 2. It is wider (137 mm) than long (118 mm), with a width/length ratio of 0.86 (Fig. 5; Table 1). Although the posterior margin of the track is associated with deformational structures, there is no clear evidence of a digit I impression. The track also may represent a double impression, based on a linear groove posterior to the clearest digit IV impression and considerable deformation to the left of digit III. Hence, divarication and digit measurements were taken from the second, more forward impression. Divarication between digits II–IV is 87 degrees, with II–III at 44 degrees and III-IV at 43 degrees. Digits were thin compared to overall track size, ranging between 7.5 and 7.8 mm wide, and all three have narrow (<1-mm-wide) grooves at their distal ends, interpreted as sharp ungual impressions. Digit III length was the same as track length (118 mm), and digit II and IV lengths were 98 and 102 mm, respectively.
Pressure-release structures directly associated with the negative-relief digit impressions include a ridge around the distal part of digit II that connects to another ridge on the anterior edge of that same digit; a subcircular mound on the posterior part of digit II; two subparallel linear ridges to the left of digit III; a broad plate on the right posterior edge of digit III; linear ridges both anterior and posterior of digit IV; and broad, low-relief, overlapping plates posterior to digit IV. Track depths below the main sandstone surface vary between 2 and 7 mm and pressure-release structures are as high as 9 mm. The distal end of digit II also may be an elongation of that impression (Fig. 5B).
Sediment grain sizes, internal bedding and rippled top surface of NMV-P231650 imply that sediments were deposited in a fluvial floodplain. This diagnosis is supported by: (1) a uniform thickness to the probable source bed for the tracks, which overlies the Dinosaur Cove bone bed, an overbank splay deposit (Rich et al. 1988; Rich and Vickers-Rich 2000); (2) fine-grained sandy sediment with small-scale cross-bedding (Stow 2006); (3) small invertebrate trace fossils in the top few centimetres of the rock, indicating postdepositional colonization by infauna; and (4) a rippled upper surface imprinted with terrestrial vertebrate tracks. These features are similar to sedimentary structures associated with dinosaur track-bearing surfaces at Milanesia Beach (Martin et al. 2012). Terrestrial vertebrate tracks on the top surface further suggest subaerial conditions, albeit with a relatively moist substrate (sensu Jackson et al. 2010; Martin et al. 2012). However, further evidence of subaerial conditions, such as mudcracks and raindrop impressions, is lacking.
All three tracks are interpreted as either shallow undertracks or poorly defined surface tracks made in moist, fine-grained sand that also deformed sediment within and around the footprints, further obscuring specific features. Undertrack preservation of the tracks is supported by: (1) no skin or digital pad impressions; (2) incomplete track outlines; (3) deformational structures in and outside of tracks; and (4) only partial preservation of ungual impressions (Milàn and Bromley 2006; Martin et al. 2012). Sediment moistness is interpreted on the basis of visual categories defined by Jackson et al. (2010), which match these tracks to those formed in sand with about 30% moisture content. A lower moisture content would have resulted in better defined tracks, whereas tracks made in completely saturated sands would have collapsed on themselves or filled with water, thereby further distorting track outlines (Gatesy et al. 1999; Jackson et al. 2010). We have no evidence of a ‘soupy’ substrate, in which tracemakers immersed and then extracted their entire feet, as described for Late Triassic theropods by Gatesy et al. (1999). Furthermore, one of the tracks (Track 2, Fig. 4) has a prominent plate of sand similar to those formed by tridactyl feet pushing against fine-grained moist sand, which causes a false interdigital ‘webbing’ (Falkingham et al. 2009). However, because all digits on the Eumeralla tracks are definable, these were probably made on an emergent surface. Modern exposure of the fossil tracks on a wave-dominated marine platform may have also eroded these and obscured further anatomical traits. A similar preservational mode was inferred for theropod tracks in the Eumeralla Formation at nearby Milanesia Beach (Martin et al. 2012).
Invertebrate burrows in the uppermost few centimetres of the sandstone, absent below laminated sediments, indicate a probable waning flow that formed both the bedding and allowed for colonization by infauna (Buatois and Mángano 2007; Martin 2009; Martin et al. 2012). This interpretation is further supported by low-angle cross-bedding just below the track-bearing surface. However, without clear evidence of cross-cutting relations, for example a track stepping on a burrow or a burrow crossing a track feature, we cannot definitely say whether the infauna colonized the sediment while it was still subaqueous or after exposure.
Tracks 1 and 2
We interpret Tracks 1 and 2 as avian tracks, based on criteria applied to Mesozoic and Cenozoic fossil bird tracks, as well as modern ones (sensu Elbroch and Marks 2001), using several key features. Lockley et al. (1992), Wright (2006) and Falk et al. (2011) summarized the traits required for distinguishing Mesozoic avian tracks from those of nonavian theropods, which include: (1) relatively small size; (2) thin digits with undefined digital pads; (3) wide (>110 degrees) divarication between digits II and IV (with some exceptions); (4) retroflexed hallux; (5) thin clawmarks; (6) clawmarks curving distally from the main axis of the footprint. Tracks 1 and 2 have nearly all of these traits, save for distally curving clawmarks. Both tracks are slightly smaller compared to some nonavian theropod tracks from the Eumeralla Formation at nearby Milanesia Beach (Martin et al. 2012); have thin digits lacking digital pads; a wide digit II–IV divarication (120 and 95 degrees in tracks 1 and 2, respectively); a hallux impression associated with an anisodactyl foot; and thin (sharp) ungual marks. Of these traits, track size can be variable, whereas the other characteristics are diagnostic of bird tracemakers (Lockley et al. 1992; Wright 2006; Falk et al. 2011). Although Track 1 does not show distally curving clawmarks, the distal end of digit III is likely truncated and ungual impressions are not readily observable on digits II and IV. A thin clawmark, however, is recorded from digit III, bisecting a sediment mound in the middle of the track (Fig. 3C–D), which was probably formed with retraction of that digit as the bird stepped forward. Nonetheless, perhaps the most important qualitative criterion indicating an avian origin for a Mesozoic track is the presence of a well-defined retroflexed hallux. Both tracks 1 and 2 show a hallux, as well as high angles between digits I and II (86 degrees in Track 1, 75 degrees in Track 2), which is typical of other Mesozoic avian tracks (Lockley et al. 1992, 2009; Wright 2006; Falk et al. 2011).
Wright (2006) further distinguished Mesozoic tracks made by wading birds as thin-toed; having high divarication between digits II–IV; with a probable hallux; and webbing. The Eumeralla tracks are thin-toed, have high divarications and well-defined halluces, but evidence for proximal webbing between digits II–III and III–IV is unclear. Shorebirds, in contrast, lack or have a reduced hallux or show full webbing (Elbroch and Marks 2001; Lockley 2009; Falk et al. 2010); these criteria were applied to the interpretation of Early Cretaceous shorebird tracks in Korea (Lim et al. 2000; Lockley et al. 2009, 2012; Falk et al. 2011). Moreover, although the tracks may belong to enantiornithine birds, their overall form and size are similar to those of ornithurines (Falk 2011), specifically birds belonging to the modern clade Ardeidae, such as egrets and herons (Elbroch and Marks 2001; Lockley et al. 2009). If so, these tracks would greatly extend the palaeogeographical range of Early Cretaceous ornithurine footprints (Falk 2011). These tracks would also supplement an Aptian–Albian fossil record in Australia only represented previously by enantiornithines (Close et al. 2009). Ornithurines dominate the Early Cretaceous waterbird body fossil record in China, but are less common elsewhere (Zhou and Zhang 2007).
Both avian tracks are slightly smaller in overall length and width compared to the other track on the same surface, but are significantly larger than most Early Cretaceous avian ichnogenera, such as Jindongornipes and Ignotornis (Matsukawa et al. 2006; Lockley et al. 2009, 2012; Kim et al. 2012). Consequently, we hesitate to assign ichnotaxa to these specimens, despite their gross morphological resemblance to Ignotornis (Currie 1981; Lockley et al. 2009). For instance, the tracks are more than twice the length and width of the holotype and average-sized Ignotornis, described first from Aptian rocks of Colorado, but also reported from the Aptian–Albian of Korea (Currie 1981; Lockley et al. 1992, 2009, 2012; Kim et al. 2006, 2012). A comparable ichnogenus in overall size and form is Hwangsanipes from the Late Cretaceous of Korea, but this ichnogenus shows clear evidence of interdigital webbing (Yang et al. 1995), whereas the Eumeralla tracks apparently lack webbing and are about 20–25% larger. Regardless of ichnotaxonomic considerations, the relatively large dimensions of all three tracks compared to most Early Cretaceous avian tracks support a presence of sizeable enantiornithine or ornithurine bird species in polar environments of Australia.
The lengthy hallux impression in Track 1, coupled with a prominent deformational structure associated with the distal end of digit IV and edges of other digits, further suggests that the total length of this trace was elongated by tracemaker behaviour. We interpret these structures as a result of dragging the hallux through the sediment with forward foot movement, and then placing the remainder of the foot as it stopped abruptly. Such an action would have caused sedimentary structures on the anterior edge of the track analogous to thrust faults and other tectonic features (Graversen et al. 2007), pushing moist, cohesive sand around and in front of the foot. However, when this track is compared to modern flying bird traces (volichnia, sensu Müller 1962 and Reineck 1981), these traits also match those for tracks made by a bird landing after flight (Martin 2013; Fig. 6). This interpretation was likewise applied to avian-like tracks from Argentina (Genise et al. 2009), which were formerly interpreted as Late Triassic – Jurassic (Melchor et al. 2002), but recently reinterpreted as Eocene (Melchor et al. 2013). Falk (2009) likewise interpreted volichnia from Early Cretaceous bird tracks of the Haman Formation of South Korea. In each study, some tracks had elongated hallux impressions, which in modern birds are caused by tracemakers landing with their legs pointing forward (Genise et al. 2009; Martin 2013).
The ideal method for testing a volichnion hypothesis for fossil bird tracks would be to study an entire sequence of tracks, beginning with the first two formed by landing and ending with the last two made at take-off. Unfortunately, as a single footprint, this specimen cannot be tested in this way. If this right-side track were part of a right–left pair, the left-side track may have been offset far enough ahead that it was not preserved on the sandstone-block surface. Still, we note that the features of this single track are consistent with a volichnion, as observed in modern avian tracks (Martin 2013) and likewise interpreted in fossil tracks (Falk 2009, Genise et al. 2009).
Pressure-release structures associated with Track 2 also provide some evidence of foot movements, with prominent ridges and plates on the anterior and left sides of digits II–IV. These show that pressure was exerted against the sediment forward and left, with an anticlockwise twisting of the foot before its removal from the surface, perhaps as it stepped to its right foot. Track 2 also has a prominent plate of sediment mimicking distal interdigital ‘webbing’, a trait associated with deformation of a fine-grained, water-saturated sediment and imparted by foot movement (Falkingham et al. 2009). Another distinctive feature of this track is a pronounced curvature of digit II, which may have resulted from a bending of the digit as it interacted with the substrate. Posterior to this part of the track is another possible impression of the same digit, but which is nearly linear, further suggesting flexure associated with forward movement of the foot.
If made by the same species of tracemaker, possible sources of variation in dimensions between tracks 1 and 2 could have been through spreading of the digits with forward movement of a foot, whether with landing from flight or interaction of the foot with a soft, wet substrate. These factors would have resulted in a higher divarication angle and overall width of the track, as opposed to a footprint made during normal walking. Lockley et al. (2009) noted similar variations in divarications and track widths for the Early Cretaceous avian ichnogenus Ignotornis, which they attributed to local differences in substrate and behaviour. Alternatively, variations in size and divarications between Tracks 1 and 2 may be biologically linked, such as from different ages or genders of the same species or different species.
Track 3 differs from tracks 1 and 2 in its greater size, smaller digit II–IV divarications and lack of an obvious hallux impression. As a result, we could not confirm with confidence that this was an avian track. However, it conforms well with expected traits of nonavian theropod tracks, particularly those of coelurosaurs, which are normally longer than wide, have narrow digits relative to track length, sharp ungual impressions and relatively low angles of divarication (35–75 degrees) between digits II and IV (Lockley et al. 1992; Lockley 2009, Mateus and Milán 2010). Avian tracks without hallux impressions have been interpreted from the Early Cretaceous, but these have digit II–IV divarications of 100–125 degrees and most are considerably smaller than Track 3 (Currie 1981; Lockley et al. 1992; Lockley et al. 2001; McCrea and Sarjeant 2001; Falk et al. 2010; Kim et al. 2012; Lockley et al. 2012). Hence without further persuasive evidence, we would rather err on the side of caution and interpret this track as that of a nonavian theropod, while also acknowledging an avian affinity. Ornithopods were also ruled out as potential tracemakers for this track and the other two tracks, despite their common representation as body fossils in the Eumeralla Formation, particularly from Dinosaur Cove (Rich and Vickers-Rich 1999, 2000; Rich et al. 2010). This elimination was based on relatively narrow digits ending with sharp clawmarks on all tracks, which are more typical of avian or nonavian theropods (Lockley et al. 1992, Lockley 2009).
Track 3 also has ridges on the anterior sides of digits II and IV, a mound posterior to digit II and an imbricated plate on the anterior-left side of digit III. These structures are consistent with the right foot rotating and lifting just before the tracemaker stepped to its left foot. However, this interpretation remains speculative in the light of how each track is disconnected from a trackway, and hence cannot be tested further.
As was the situation with theropod tracks interpreted from Milanesia Beach (Martin et al. 2012), this track defies an easy ichnotaxonomic designation, owing to its vague outline and lack of anatomical detail. Based on tridactyl tracks illustrated by Lockley (2009), the closest ichnogenera comparable in size and form are Columbosauripus or Irenichnites (Sternberg 1932; Currie 1981). Theropod tracks at Milanesia Beach, interpreted as ornith-omimid or oviraptorsaurid, fell into three statistically distinct size classes, with one significantly smaller than the other two, and the other two differing only slightly but designated as ‘intermediate’ or ‘large’ (Martin et al. 2012). Track 3 would have fallen into the ‘intermediate’-size category applied to the Milanesia Beach theropod tracks, and is nearly identical in form, implying similar tracemakers at each locality. Using the method refined by Henderson (2003) of 4.0 × track length = hip height, and assuming a theropod tracemaker, the calculated hip height for the tracemaker would have been 47 cm. This is smaller than inferred sizes for nonavian theropods represented by body fossils in the Eumeralla Formation, such as Timimus (Rich and Vickers-Rich 1994), an unnamed tyrannosauroid (Benson et al. 2010) and an unnamed spinosaurid (Barrett et al. 2011). Still, it is within the range of sizes calculated for theropod trackmakers at Milanesia Beach (Martin et al. 2012). Ornithomimids, oviraptorsaurids and other small maniraptors have been interpreted in the Eumeralla Formation on the basis of a few body fossils, including some from Dinosaur Cove (Rich and Vickers-Rich 1994; Currie et al. 1996; Benson et al. 2012). Hence, a coelurosaur is reasonable as an inferred tracemaker.
The palaeontological significance of these avian tracks from the Eumeralla Formation is twofold. For one, they would be the oldest known in Australia, which has a poor fossil record in general for Mesozoic birds (Molnar 1986, 1999; Kurochkin and Molnar 1997; Close et al. 2009). Secondly, the tracks indicate the presence of a large enantiornithine or ornithurine birds during the Early Cretaceous of Australia. Although we can also state that the Eumeralla tracks are the first reported from the Early Cretaceous of Gondwana, we cannot say for sure they represent the oldest avian tracks in the Southern Hemisphere. At one point, tracks from the Santo Domingo Formation of Argentina, originally interpreted as Late Triassic – Jurassic, were identified as possibly avian or at least represented convergence of small theropod traits with those of avians (Melchor et al. 2002; de Valais and Melchor 2008; Genise et al. 2009). This problematic extension of avian-like theropods to the Late Triassic – Jurassic was rendered moot once refined age-dating methods revealed their age as Late Eocene (Melchor et al. 2013), confirming the alternative explanation that the Santo Domingo Formation was younger than previously surmised (Genise et al. 2009).
Additionally, Late Triassic – Early Jurassic tracks of South Africa described by Ellenberger (1972, 1974) also resemble bird tracks, although these have not been confirmed since (Genise et al. 2009). An avian interpretation for the South African tracks, however, is controversial because of the significant gap represented between these and the earliest known body fossil record of birds (Chiappe and Dyke 2006; Xu et al. 2011). On the other hand, the Eumeralla footprints, coming from well-dated Albian strata at Dinosaur Cove (Rich and Vickers-Rich 2000), are well within the currently agreed-upon geological range of birds. Hence, their assignation as avian is less contentious.
Another key point related to Track 1 is its features suggesting that its maker landed from flight. Unfortunately, the slab containing the track lacks a second, opposite-side track typical of a landing pattern (Martin 2013). Consequently, we can only speculate that landing from flight is one explanation for the unusual features of this track.
The presence of three tracks on a small area (about 650 cm2) of the same bedding plane is also noteworthy, showing vertebrates were living directly above the Dinosaur Cove bone bed (Rich and Vickers-Rich 2000, 2003). These footprints qualify Dinosaur Cove as another vertebrate tracksite in the Eumeralla Formation, one of only a few found thus far (Rich and Vickers-Rich 2003; Martin et al. 2007, 2012). Indeed, the tracks echo the opening statement on the plaque commemorating the site: ‘significant fossils were discovered at this locality, Dinosaur Cove, in 1980’, only now the date can be amended to include 2010. In short, the trace fossil record for Early Cretaceous birds in Australia and the remainder of Gondwana begins with these steps.
Two of three closely associated vertebrate tracks in a sandstone bed from the Eumeralla Formation (Albian) of Dinosaur Cove, Victoria, were likely made by relatively large avians, either enantiornithine or ornithurine. These are the first reported avian tracks from the Early Cretaceous of Australia and the remainder of Gondwana and are the oldest in Australia.
One of the avian tracks has sedimentary features consistent with an abrupt stop, such as those caused by dragging and halting of the foot upon landing after flight or a hop.
The theropod track, which matches the size and form of recently reported tracks from nearby Milanesia Beach (Victoria), was likely made by coelurosaur.
These tracks add to the palaeontological importance of Dinosaur Cove and confirm another vertebrate tracksite in the Eumeralla Formation.
We heartily thank Sean Wright and Alan Tait for discovering the tracks at Dinosaur Cove, supplying yet another example of the indispensable role volunteers play in the continuing discovery of scientifically important fossil specimens in the Cretaceous rocks of Victoria. The tracks were collected under a Victoria Parks and Wildlife permit granted to PVR. David Pickering (Museum Victoria) provided access to specimen MV-P231650 during AJM's study of the tracks in July 2011. He also cut a sample of the specimen for MH to examine for its sedimentological content. Rod Start (Museum Victoria) photographed the tracks for some of the figures used in this paper. Travel for AJM to Victoria was covered by the Center for International Programs Abroad of Emory University. We are grateful to two anonymous reviewers for their thorough and detailed feedback on the first submitted version of this manuscript, to Matteo Belvedere and Jenni Scott for their in-depth reviews of the second version and to Palaeontology editors Svend Stouge, Kenneth Angielczyk, Sally Thomas, and Andrew Smith for their guidance and input. We also give thanks to the National Geographic Society for their long-term support to the palaeontological studies of THR and PVR for more than 30 years.