The midline structure in SMF R 4970 consisting of paired scales has the shape and restricted position to the posterior part of the abdomen commensurate with the morphology of the umbilicus (umbilical scar) in extant amniotes. Unlike most mammals, reptiles (including birds) do not have a true umbilical cord—which is an elongate allantois-derived structure linking the placenta and the embryo [13]. Instead, the reptilian embryo is in direct communication with the yolk sac and allantois (or chorioallantoic membrane [14]) via a longitudinal midline aperture in the abdominal wall. This aperture closes over to form the umbilicus, which, in most extant reptiles (including birds), persists for only a few days to weeks and is therefore a common indicator of an individual’s relative maturity (e.g., [15,16,17]). In late-stage embryonic birds, the umbilicus forms a circular aperture in the abdominal wall that is reduced to a small circular scab that typically drops off within a few days of hatching (Fig. 2E; Additional file 1: Table S1; see also Fig. 5 in Kenny and Cambre [18]). Although scarring does not normally occur, it may persist in some adult birds (e.g., rock pigeon, Columba livia [19]) as a linear scar on the lower abdominal wall [19]. Scarring in lepidosaurs (snakes and lizards) is not obvious because of the presence of scales; however, the position of the umbilicus in the adults of some species of snake is retained as a faint line or crease in the ventral scales [15] (Fig. 2B; Additional file 1: Table S2). In lizards, scarring is typically absent but the umbilicus may be demarcated and clearly recognized by a change in abdominal scale morphology. This is typically reflected as a longitudinal row of paired scales that are generally larger and of a different shape to the surrounding abdominal scales and which remains visible throughout life, very similar to the condition in Psittacosaurus (Fig. 2; Additional file 1: Table S2). Between the paired scales, a clear line of interstitial tissue delimits the midline and the former umbilical aperture (Fig. 2A), which is observable also in Psittacosaurus even where the paired scales more closely match the size and shape of the surrounding abdominal scales (Fig. 1B, C). Such a distinct line of interstitial tissue is not visible anywhere else on the abdomen of SMF R 4970. Where they do occur in modern lizards, paired scale rows are often moderately long and linear features (approx. 7% of SNV; Additional file 1: Table S2), although we have also observed some as long as 11% of the total snout-vent length (SVL) (e.g., spiny-tailed lizard, Uromastix sp.; Fig. 2A; Additional file 1: Table S2)—which is similar in size to that of SMF R 4970 (14% of SNV) and to late-stage embryonic birds (approx. 13% of SNV; Additional file 1: Table S2). These values contrast with the relatively shorter umbilici in snakes (approx. 2% of SNV; Additional file 1: Table S2), which is influenced by the elongation of the trunk region. Among crocodylians, the umbilicus is generally absent in adults but remains distinct in Alligator mississippiensis, forming an elongate, spiderweb-like arrangement of small scales between the transverse bands of larger quadrangular scales [20]. This arrangement is unique to Alligator mississippiensis [20] although it is variably present in other crocodylians as a more subdued line in some individuals with or without marginal paired scales (Fig. 2D). In both crocodylians and lepidosaurs that lose their umbilicus, the pattern of abdominal scales remains undisturbed along the midline (Figs. 2F and 3). Based on the close structural and topographic similarity, particularly to some crocodylians and squamates, we interpret the ventral midline structure in Psittacosaurus SMF R 4970 as delimiting the umbilicus, which remained visible for many years, at least into sexual maturity (6+ years).
Where present, the persistence of umbilical scarring into adulthood among reptiles and birds may be species-specific and its position even sexually dimorphic in some cases [15]. Scarring may also be linked to yolk sac infections (omphalitis) in farmed birds and crocodylians, which is connected to improper conditions pre- or post-hatching (e.g., hygiene, diet, temperature/humidity imbalance), although these adverse conditions are rarely present in the wild [21, 22]. Therefore, the presence of an umbilical scar in Psittacosaurus does not imply its presence in mature (or near-mature) individuals of all non-avian dinosaur species, although we cannot reject this possibility at this time.
The regular sizes, smooth margins and disposition of the umbilical scales along the midline of Psittacosaurus—which is similar to those in wild-caught reptiles (Additional file 1: Table S2)—as well as the absence of regenerative tissue, all rule out a traumatic or pathological origin [21]. Because of the haphazard way in which they are inflicted, traumatic injuries, in contrast, can truncate/bisect individual scales and heal by forming granulation tissue—a type of smooth, scale-free connective tissue—over the open wound [23,24,25]. In reptiles, scales may not regrow, leaving a central region of the smooth dermis that may be surrounded by a disrupted region of small, irregular scales that differ from the surrounding ‘normal’ scales [23, 24]. Trauma-induced granulation tissue surrounded by a ring of disrupted scales has been reported in the skin of at least one dinosaur [26], but none of these characteristic features are observable in SMF R 4970.
Remnants of the soft tissues associated with embryogenesis and early post-hatching development (including the umbilicus) are rarely captured in the fossil record. To our knowledge, an umbilical scar has not been previously reported in any fossil amniote [27], although the ‘umbilical cord’ was described in a Devonian-aged viviparous placoderm fish [28]. Embryonic skin has been reported in only two non-avian dinosaurs [29,30,31], neither of which show indications of an embryonic umbilicus, and few other specimens preserve naked integument in this region of the abdomen that can be compared with SMF R 4970. Two unique specimens that do—the hadrosaurid Edmontosaurus annectens ‘mummy’ AMNH FARB 5060 and the juvenile ankylosaurian Liaoningosaurus paradoxus IVPP V12560—show no indication of any structure that could be interpreted as an umbilicus, even though the latter was less than a year old at the time of death [32]. The skin of Liaoningosaurus [33], however, does not perfectly intersect with the midline of the animal and the umbilicus, if present, may simply have not been preserved. The linear row of paired quadrangular scales in an embryonic titanosaur from the famous Auca Mahuevo locality is from an unknown position on the body that was presumed to be along the dorsal midline [30]. Although a ventral midline position is also tenable, an umbilical origin for this structure in the Auca Mahuevo titanosaurs can be rejected as the animals were still in ovo, during which time the abdominal aperture would still have been open.
