Smallest known raptor tracks suggest microraptorine activity in lakeshore setting

Ongoing studies of a multiple track-bearing horizons from massive excavations in the Jinju Formation (Lower Cretaceous) of South Korea have yielded a remarkable diversity of avian, non-avian dinosaur, pterosaur, crocodilian and mammal tracks, many very small and well preserved. Here we report diminutive, didactyl tracks (~1.0 cm long) assigned to a new dromaeosaurid ichnogenus Dromaeosauriformipes, which resembles the larger, but still quite small, ichnogenus Dromaeosauripus, also from the same formation only 30 km away. These diminutive tracks are consistent with the foot size of smaller dromaeosaurid taxa like Early Cretaceous Microraptor from China, and may represent diminutive species or juveniles of larger species. The association of tracks with lakeshore sediments is consistent with the evidence that Microraptor was a fish eater. Two trackways and isolated tracks indicate variable trackmaker gaits and speeds. If oviparous, as assumed for most non-avian dinosaur neonates, the trackmakers must have hatched from tiny eggs. Previous studies of the Korean Cretaceous indicate the presence of other diminutive (~1.0 cm long) theropod tracks (Minisauripus). Such occurrences strongly suggest that small tracks attributed to juveniles, or very small tetrapod species, are more common than previously supposed especially where suitable preservation conditions prevailed.

The remarkable abundance of Cretaceous tracks being discovered in Korea [6][7][8] and China [9][10][11][12][13][14] , and the number of new ichnotaxa attributable to so called functional-didactyl 'raptors' (inferred dromaeosaurs) 6-14 raises two intriguing, but challenging questions which we attempt to address below: 1) what is the size range of raptor tracks based on footprints or inferred from skeletal remains? and 2) how might diminutive tracks of juveniles be distinguished from those of adults? . Among the rescued specimens, cut out with large rock saw equipment 4,5 , is Chinju National University of Education (CUE) specimen CUE JI-2E Dr001 (Fig. 2), a slab containing two trackways of a diminutive didactyl biped and additional isolated tracks. All these footprints are approximately 1.0 cm long ( Table 1). The tracks interpreted as continuous trackway segments are designated Trackways 1 and 2 with seven and three tracks respectively . There are at least eight other isolated tracks labelled as T1-T8. The surface also reveals two tridactyl tracks with quite different morphologies. One consisting of three parallel scratch marks up to 5 cm long, the other a large, more robust tridactyl track 16.2 cm long and 12 cm wide (SI Fig. SI3). All tracks were mapped and photographed in situ 4,5 and in many cases replicated before the original specimens were excavated. In the case of CUE JI-2E Dr001 a replica was made of Trackway 1 which has been donated to the University of Colorado Museum of Natural History (UCM 214.309) for analysis as part of this study. The length (L), width (W) and steps (St) and strides (Str) between individual tracks were measured (Table 1) using the original specimens, replicas, tracings and calipers calibrated to the nearest 0.1 mm (Table 1). These measurements are shown graphically in Supplementary Information and Table 1.

Results
Description of tracks and trackways. The holotype trackway of Dromaeosauriformipes rarus ichnogen.  The eight isolated footprints that do not belong to Trackways 1 and 2 are numbered T1-T8. They are similar in size to the individual footprints in Trackways 1 and 2 but not obviously mutually associated with one another in trackways, although footprints T5-T6 and T7-T8 might represent steps of ~23.0 and 18 cm in two other trackways. Thus, the 18 tracks collectively represent an estimated 6 to 10 individual trackways possibly made by a similar number of different individuals. This inference is based on the convention that it is more parsimonious to count one individual per trackway (not per track) 15 , especially where footprint size varies, than to assume one individual making repeat crossings of a small area, at different speeds and in as many as ten different directions. The similarity in track depths at the Jinju site also suggests activity over a short period of time, rather than repeat crossings of the area by few individuals over a longer duration.

Systematic Ichnology
Systematic notes. The naming of tetrapod ichnotaxa at the ichnogenus and ichnospecies level is based primarily on morphology of the footprints, and trackway configurations, which in turn are influenced to various degrees by biological factors such as trackmaker foot morphology, size and behavior, but also by geological (physical environment) factors such as substrate conditions. There are well established conventions and guidelines that govern the introduction of new ichnotaxonomic names 16 and aim to constrain variable or subjective interpretations. For example, as discussed below (Supplementary Information), differences used in naming new ichnotaxa at the ichnogenus or ichnospecies level may result, respectively, in a more conservative 'lumping' or a more liberal 'splitting" approach.
Tracks of diminutive Dromaeosauriformipes rarus ichnogen. et ichnosp. nov., described here, are clearly distinct from all representatives of the much larger, previously-named ichnogenera (Velociraptorichnus, Dromaeopodus, Dromaeosauripus, Menglongipus, and Sarmientichus) attributed to functionally didactyl dromaeosaurid theropods 6,7,[9][10][11][12][13][14] . However, it is morphologically more similar to Dromaeosauripus, represented by three ichnospecies, than to any representatives of the four other ichnogenera. D. rarus differs from the first-named Dromaeosauripus ichnospecies (D. hamanensis) 7 in size, shape and digital pad configuration, and differs from D. jinjuensis 6 from the Jinju Formation which is almost nine (8.96) times larger than D. rarus and much less elongate. However, the footprint to step length ratio between diminutive D. rarus and large D. jinjuensis is very similar. The choice of the ichnogenus label Dromaeosauriformipes recognizes both the similarities, in trackway proportions, and the differences, in footprint shape and size, between these two ichnotaxa.
Etymology: Ichnogenus named for similarity in "form" (shape) to the didactyl raptor track Dromaeosauripus, previously defined as meaning "dromaeosaur foot. " Ichnospecies from Latin rarus meaning both small and infrequently found.  Diagnosis: Didactyl tracks of a diminutive biped more than twice as long as wide, with two equidimensional, elongate parallel traces. Trackway about as narrow as individual tracks, with step length highly variable. Differs from Dromaeosauripus in size and degree of footprint elongation, and in the absence of any trace of digit II characteristic of other dromaeosaurid ichnotaxa.
Holotype, locality and horizon: as for ichnogenus. Description: The holotype slab contains 18 tracks of a diminutive biped, preserved as natural impressions (concave epireliefs). These include the holotype series designated as Trackway 1 (Fig. 2), containing a seven-track sequence, a second three-track, paratype series, Trackway 2, and eight isolated tracks. All are remarkably consistent in size. The holotype averages 10 Table 1.
The dimensions of holotype Trackway 1 reveal a mean step and stride length of 4.62 cm and 10.00 mm, respectively, indicating a step about 4.5 times footprint length and slow progression (0.6 m/sec, or 2.16 km/hr). By contrast paratype Trackway 2 has a mean step or pace length (PL) of 27.82 cm (stride of 55.63/2) indicating steps about 26.49 times footprint length and much higher estimated speed of 10.5 m/sec, or 37.8 km/hr. However, both trackways are very narrow, with correspondingly high pace angulation values (168°-188°).

Discussion
The deinonychosaurid/dromaeosaurid track record. All previous reports of didactyl, deinonychosaurid (dromaeosaurid) tracks including several from Korea [6][7][8][9][10][11][12][13][14][17][18][19] have dealt with footprints an order of magnitude larger than those described here. These reports include the report of Dromaeosauripus jinjuensis 6 , which reveals footprints which are morphologically similar to the tracks described here, except for a huge size disparity. D. jinjuensis footprints average 9.26 cm long and 6.75 cm wide, thus being and order of magnitude larger. They are  Fig. SI4A-C). The D. jinjuensis pace angulation averages 174.7° compared with 188° for Trackway 1 described here for D. rarus (Figs 3 and 4). Trackways 1 and 2 indicate the trackmakers were highly mobile, and capable of progression at variable speeds. As noted above, speed estimates for Trackway 1 indicate slow progression at ~0.6 m/s (2.16 km/hr). The speed estimates for Trackway 2 (~10.5 m/s = 37.8 km/h: Table 1) 20 , suggest a striking variation in locomotor range (gait) for this trackmaker. However, this is within the range of estimates for other dinosaurs, and is consistent with the maximum running speed estimates for a dinosaur such as adult Velociraptor 21 . Likewise Minisauripus from Korea and China also indicate a wide range of variation in step or pace length, relative to footprint length (PL/FL) [22][23][24] and thus considerable variation in trackway-derived speed estimates. For example, PL/FL for Minisauripus ranges from 5.18 to 9.07, 9.52 to 11.12 and 7.73 to 15.00 respectively for three samples from China 9,23,24 and 6.14 to 18.0 for samples from Korea 8,24 : see SI Table 1.
It is well-established that deinonychosaurid tracks vary in both size and shape 18 giving rise to the naming of several ichnogenera. Among these, in order of naming, Velociraptorichnus 9 , Dromaeopodus 10 , and Dromaeosauripus 7 are based on well preserved types associated with trackway segments, and tracks showing digital pad traces. All are from Lower Cretaceous type localities in Asia. Ichnogenus Dromaeosauripus is represented by three aforementioned ichnospecies 6,7,12 and indeterminate ichnospecies from the Lower Cretaceous of North America 19 . Menglongipus, also from the basal Cretaceous of China 11 is based on material with suboptimal preservation. The Velociraptorichnus ichnospecies V. zhangi is also from the Lower Cretaceous of China 13 . Recent work suggests that ichnogenus Sarmientichnus is an extramorphological nomen dubium representing a poorly-preserved didactyl dinosaur track of deinonychosaur affinity 14 . Previous studies indicate footprint lengths range between ~6.7 cm for Menglongipus 11 and 28 cm for Dromaeopodus 10 . Thus the tracks described here (Figs 2-5 and SI4-5) are far smaller than any previously reported. Among named and well defined dromaeosaurid track morphotypes there are at least two preservational and/or behavioral morphotypes: 1) where the traces of digits III and IV converge and unite towards the heel, notably in Velociraptorichnus isp. and Dromaeopodus isp. and 2) those where separate toe traces are not joined at the heel, as in all examples of Dromaeosauripus jinjuensis 6 , diminutive D. rarus, and some D. hamanensis 7 and D. yongjingensis trackways 12 . The variation in the large sample of trackways of D. yongjingensis suggests that the separate paired traces first reported for D. jinjuensis are a preservational or behavioral variant of deinonychosaurian tracks with digit traces which may be united in the heel [10][11][12][13][14] . If the lack of heel traces is due to a more digitigrade gait the explanation is behavioral, but if it is the result of substrate influences then the explanation is preservational. Both factors may be factors in any given cases (SI).
Based on trace morphology we conclude that at least 22 occurrences of deinonychosaurian tracksites, exhibiting traces of functionally didactyl bipeds were reported prior to the present discovery. Of these 14 were from China 9-14 , three from Korea 6-8 , three from North America 17,19 and two from Europe 18 . The aforementioned differences in morphology, indicate that the diminutive tracks described here resemble Dromaeosauripus isp. in some features (separate traces of digits III and IV and proximal trace of digit II is not impressed) 6 . However, striking differences in size and elongation (L/W), and comparisons with other inferred dromaeosaurid tracks, 14,18,19 , marks Dromaeosauriformipes rarus as a new ichnotaxon distinct from any other existing ichnotaxon.
Other potential trackmakers considered. It is important to consider the possibility of other potential, non-deinonychosaurid tracemakers. Based on the size of the individual traces, and their preservation as surface impressions (concave epireliefs) they are clearly surface trails. However, their small size suggests they could have been made by invertebrates. We have examined a number of surface trails attributed to invertebrates, mostly arthropods and illustrate them for comparison (SI Fig. SI4), showing both trackway configurations, and the morphology of individual tracks or traces that repeat within the trackways. These include the ichnogenera Bifurculapes, Diplichnites, Hamipes, Kouphichnium, and Permichnium 25 While it is possible to see similarities between some of these individual traces (SI Fig. SI4) and those here labeled as small Dromaeosauriformipes rarus ichnogen. et ichnosp. nov., none of these ichnogenera have similar trackway configurations. In the case of the small didactyl traces described here, there is no evidence, other than size that they are of invertebrate origin. Moreover we are constrained by visible trace morphology to name traces according to ICZN conventions and guidelines 16 and available literature 6,7,[9][10][11][12][13][14][17][18][19] (Suppl. Info). Thus, we infer, based on morphology, that the tracks were most likely made by a juvenile deinonychosaurid or a very small species representative of that clade. Again ICZN convention 16 stresses that inferences about trackmaker identity are secondary to the description of track morphology.

Interpreting the preservation and affinity of small trackmakers. Previous studies of the Korean
Cretaceous have unequivocally established that preservation conditions in fine-grained deposits of the Sindong and Hayang groups, notably the Jinju, Haman and Jindong formations have provided excellent substrate conditions for the preservation of both small and large tetrapod tracks. These include the dinosaur track Minisauripus 8,24 , and the lizard track Neosauroides 26 from the Haman Formation and the mammal track Koreasaltipes from the Jinju Formation 3 . Currently Koreasaltipes, is the smallest fossil mammal track known, Korean Minisauripus, which also occurs in China, is the smallest representative of this ichnogenus, and the Dromaeosauriformipes rarus, described here, is clearly the smallest known didactyl track inferred to represent the dromaeosaurid clade. As there is no paleoecological evidence to suggest that Korea was home to exceptionally small tetrapods, the reason for these occurrences probably pertain to exceptional, preservation conditions. Such positive biases favoring preservation of small tracks have important paleontological implications. The significance of small tracks and the biases against their preservation have been discussed by several authors, often inconclusively 8,23,[26][27][28][29] . Possible and partial interpretations for lack of small tracks are as follows: 1) in comparison with large tracks, small tracks are rarely preserved, 2) trackmakers grew rapidly, perhaps not leaving their nests until having reached relatively large sizes, 3) small species were rare, 4) before burial and/or after present day exhumation and exposure to erosion, small tracks are easily eroded, 5) it is not easy to find smaller tracks due to observer oversight.
The first argument is compelling given that many substrates are unsuitable for preserving tracks of small, lightweight trackmakers. This argument is also supported by the concentration of small tracks in certain optimal preservation facies (Konservat-Lagerstätte type), such as the lake basin settings dealt with here.
The second argument pertaining to trackmaker growth most likely applies to relatively large species, and depends on finding evidence of a range of track sizes pertaining to particular ichnotaxa. For example, the Korean Cretaceous has produced a relatively full range of sauropod track sizes 30 . There are also differences, for example, between birds (avian-theropods) which are precocial and leave the nest early with track making capability when juvenile and small, and altricial species which may be relatively large by the time they leave the nest.
The third argument, that small species were rare, although refuted by the skeletal record, raises the question of whether small tracks represent small species or juveniles of larger species. Differentiating between such interpretations depends, partly, on the range of size of tracks of a particular morphotype (ichnotaxon). For example, it has been inferred 8  However, in the case of the Dromaeosauriformipes rarus tracks described here, we have good evidence of large tracks, Dromaeosauripus jinjuensis (Fig. 5), from the same formation, albeit at only one site 30 km away. Thus, we could either infer that the tracks represent juveniles, or a very small species. There is no evidence to reject the latter possibility of multiple, co-existing species. On the contrary, there is growing evidence that a large number of Asian dromaeosaurid (microraptorine) species were small [31][32][33] . These include two species of Microraptor from China 31-33 , the smaller of which (M. zhaoianus) had feet only ~2.5 cm long, based on total length of digit III phalanges. These measurements are much closer to the D. rarus tracks described here than to any of the larger tracks (Figs 2-5, SI Figs SI4 and 5). There are also very small North American species including Hesperorhynchus 34 and Bambiraptor 35 , the former of which has been estimated to have a body weight of only 1900 grams (~2.0 kg). Evidence that Microraptor was a fish eater 33 provides support for the proposed microraptorine origin of the D. rarus tracks described here, since they were found in a lakeshore setting.
The inference that small tracks are rare [27][28][29] has been increasingly challenged by more finds and ongoing studies of the Korean Cretaceous where small tracks of avian theropods (birds), non-avian theropods, pterosaurs, lizards and mammals are abundant. The growing database offers the opportunity consider the ontogeny of various trackmaking groups, and such paleobiological questions as hatchling or birth size and related questions of oviparity and viviparity. However, the data base is still small and it is recognized that without compelling evidence that juveniles and adults were active together, such as in social gregarious groups 29 the occurrence of Dromaeosauripus jinjuensis 6 and Dromaeosaurifromipes rarus, described here, in the same formation, but 30 km apart, is not an association from which unequivocal biological or ontogenetic inferences can be drawn. We can only infer that presumed dromaeosaur track makers, much smaller than the maker of D. jinjuensis tracks, and with more elongate feet, were active at about the same time, and that Chinese discoveries confirm that small microraptorines [31][32][33] are known in the greater region (SI Figs 5 and 6). Given reports that small theropod tracks tend to be narrow, or elongate, whereas large tracks tend to be wide (Suppl. Info), there is precedent for inferring that the feet of dromaeoasurid trackmakers also grew wider during ontogeny.
Many extant squamates (lizards and snakes) are viviparous, but all "other extant Reptilia (turtles, crocodilians, spenodontids, and birds) are entirely oviparous 36 . " It is therefore assumed that non avian dinosaurs were also oviparous 36 . It has been suggested that some sauropsid dinosaurs were also oviparous 37,38 , but given the abundant evidence for eggs and nests, this speculation has not been well received 36 . Thus, the weight of evidence does little to support speculations of dinosaur viviparity. However, most known dinosaur eggs are relatively large (>10 cm diameter), and in many cases produced hatchlings that would have produced trackmakers larger than those inferred here. However, in the case of Minsauripus 8.23,24 there is unequivocal evidence that some such small trackmakers registered tracks only ~1.0 cm long. Minisauripus tracks are abundant in the Haman Formation but were also recently discovered, as exceptionally well-preserved specimens, in the Jinju Formation (Suppl. Info), again supporting the wealth of evidence for diminutive trackmakers. Such small tracks (L: 1.0-2.5 cm) suggest diminutive trackmakers that likely hatched from smaller eggs. Alternatively, one might infer they belonged to small species that gave live birth. If altricial, with a period of growth in the nest, as in the case of many small extant birds (avian theropods) like passerines, the eggs may have been as small as those laid by small sparrow-sized SCiEntifiC REPORtS | (2018) 8:16908 | DOI:10.1038/s41598-018-35289-4 birds 8 . If the species was precocial, as in the case of many extant shorebirds (charadriformes) the eggs may have been larger, with size presumably depending on the size of the adult egg layer 38,39 . Despite these considerations which suggest oviparity throughout the Dinosauria recent evidence from the study of Dinocephalosaurus (an archosauromorph) suggests "no fundamental reason archosauromorphs could not achieve live birth" 36 . The evidence that some early archosauromorphs were viviparous does little to support or refute the idea that diminutive theropod tracks might represent viviparous species. Likewise, given that most extinct viviparous reptiles, including Dinocephalosaurus, were marine adapted 40 there is no compelling support for viviparity among small terrestrial theropods. Thus, on balance, D. rarus tracks described here are attributed to small hatchlings of a species of unknown adult size, but consistent with several of the smaller microraptorine dromaeosaurid dinosaurs known from the Asian fossil record.