Spinal fracture reveals an accident episode in Eremotherium laurillardi shedding light on the formation of a fossil assemblage

The Toca das Onças cave is one of the most important Quaternary mammal deposits of Brazil. Two different hypotheses have been proposed to explain the preservation mode of its skeletal remains: either the animals climbed down into the cave, or it could have functioned as a natural trap. Evaluation of pathological modifications on three articulated vertebrae of a single adult giant ground sloth Eremotherium laurillardi reveals a particular type of bone fracture caused by compressive force on the vertebral column, which split the vertebral bodies in the sagittal plane. This diagnosis suggests that the animal accidentally fell into the cave, in accordance with the second hypothesis proposed to the incorporation mode of skeletal remains into the cave.


Results
The three vertebrae investigated (DGEO UFPE 5769, DGEO UFPE 7167 and DGEO UFPE 9048) belong to the same individual and show lesions with similar appearance, location, and size. All lesions are in the caudal endplate of each vertebra near the left edges of the vertebral bodies running in the dorsoventral direction with a gently oblique position (Fig. 1). Each lesion appears as a deep and narrow bone discontinuity which has its www.nature.com/scientificreports/ gap widening at the midpoint before narrowing again, something that is most noticeable in the vertebra DGEO UFPE 9048 (Fig. 1a). Each lesion has approximately the following measurements of length: (1)  www.nature.com/scientificreports/ rounded edges (see images in detail in Fig. 1). There is no macroscopic evidence of new bone formation on the surface of the vertebrae.

Discussion
Since the bone discontinuities noted in the three vertebrae analyzed show no clear sign of bone overgrowth, it is pivotal to rule out the possibility that we are dealing with preservation damages before proposing an accurate diagnosis for the lesions. The close-up view examination of the abnormalities shows that their edges have clear signs of smoothing and rounding ( Fig. 1), which represent important evidence of osteoblastic activity 18,19 . Additionally, the similar color of the cortical damage and normal bone can be used as secondary evidence to rule out post-mortem processes as a possible origin of the alterations, since recent destructive processes are lighter than the rest of the bone 19 . Therefore, as taphonomic processes can be ruled out, the pointed evidence strongly suggests that the discontinuities observed are of pathological origin. More specifically, these breaks found in all three vertebrae are indicative of bone fracture. Based on fracture analysis criteria applied here 20 , which consider the location and morphological pattern of the fractures, we classified the fractures noted in all vertebrae as traumas belonging to Type A (vertebral body compression), Group A2 (split fractures), and subgroup A2.1 (sagittal split fracture). This diagnosis implies that the traumatic episode was likely caused by a compressive force on the vertebral column, which split the vertebral bodies in the sagittal plane. This type of injury is considered stable-i.e., the fracture does not have a tendency to displace after reduction-and neurological deficit is uncommon 20,22,23 . Although stable traumas cause only moderate pain, without generating significant movement limitations 20 , the Eremotherium individual here analyzed died with unhealed bones, as there is no evidence of callus formation.
The absence of other skeletal signs that point to the presence of another type of disease concomitantly to the fractures allows us to reject the possibility that they have been generated as a result of a pre-existing disease (e.g., infection, neoplasm). We also consider that the vertebral injuries were not caused by repetitive force (stress fractures) because this type of injury is commonly characterized as a nondisplaced line or crack in the bone, called hairline fracture 3 . Those refer to situations where the broken bone fragments are not visibly out of alignment and exhibit very little relative displacement 21 . Although the Eremotherium vertebrae fractures' can be described as nondisplaced, they also have a noticeable gap between their edges that is mostly narrow with wider parts in the middle, something found in split fractures 20 but that is not characteristic of hairline fractures. Lastly, the subgroup C1.2.1 (rotational sagittal split fracture) might be a source of confusion due to similar morphological pattern with subgroup A2.1 (sagittal split fracture). However, in subgroup C1.2.1 there are compressive and rotational forces acting simultaneously, producing total separation into two parts 20 , which clearly did not occur in the vertebrae analyzed here.
In humans, compression fractures are most commonly caused by osteoporosis, although infection, neoplasm and trauma can also be etiological factors [23][24][25] . However, as aforementioned, the absence of other pathological skeletal marks is an important characteristic to take note as it serves to disregard the possibility of the fractures' genesis to be secondary to another pathology. As such, in this case, osteoporosis, infection and neoplasm are unlikely etiologies. On the other hand, a compression fracture in a healthy individual is commonly generated after a severe traumatic event such as a fall from great height 23,26 . This scenario seems to better explain the origin of the vertebral fractures in the case of the Eremotherium ground sloth herein studied.
The three fractured vertebrae were recovered in the Toca das Onças site (Fig. 2), a small cave considered as one of the richest paleontological sites of the Brazilian Quaternary 15 . Two complete skeletons of Eremotherium laurillardi and fragments belonging to at least thirteen other individuals, together with several other bones assigned to different smaller species are known to this cave 14 . It comprises of a single dry chamber that can only be entered through vertical entrances approximately 4.5 m high (Figs. 2b-d and 3). Two different hypotheses concerning the depositional process of Toca da Onças were previously proposed: (1) the animals climbed down into the cave in search of water 14 ; or (2) due to the vertical character of the cave entrance, it could have functioned as a natural trap where animals accidentally fell into the cave 15 .
The first hypothesis would indicate that the animal fell into the cave during an attempt to climb down. However, there is no report in the literature indicating that Eremotherium laurillardi could have been a climbing animal. In addition, the vertical morphology of the cave entrance would be a limiting factor for climbing behavior (see Fig. 3).
Therefore, based on the type of fracture (compression sagittal split fracture) observed in the three vertebrae of Eremotherium as well as the inferred origin mechanism (fall from a great height), the presence of the individual here analyzed in the fossil accumulation of Toca das Onças is more likely explained by the second hypothesis. This idea is not particularly new as 'entrapment due to fall' has been described as a fossil accumulation mode to several other caves worldwide (e.g., 27,28 ). However, the use of bones fractures as an indicator of fossil accumulation mode is an interesting novelty. Of course, a detailed taphonomic investigation in the Toca das Onças still needs to be conducted in order to accurately interpret the formation of this important Quaternary fossil accumulation from Brazil.
In sum, we suggest that the animal accidentally fell into the cave, fractured at least three sequential vertebrae (12th, 13th thoracic vertebrae and 1st lumbar vertebra) after the impact on the ground, survived for a while, but succumbed trapped inside the cave without food and water (Fig. 4). Other animals found in the cave, but without signs of bone fracture, may have fallen and not fractured their bones or not survived after the fall, especially the smaller ones. Finally, the proposal of falls to explain the unusual record of giant ground sloth fossils preserving much of its skeleton in caves, as reported for Toca das Onças site, contrasts with the better-documented pattern of skeletal accumulation via hydraulic action.

Materials and methods
The Each vertebra was macroscopically examined to characterize the pathological changes. We adopt here the fracture classification system proposed by Magerl et al. 20 to describe the lesions studied. This classification system www.nature.com/scientificreports/ was created specifically for thoracic and lumbar vertebral fractures and is based on morphological patterns that can be divided in a set of three main categories of lesions (Type A, Type B and Type C). Each category reflects the effect of a force and the mechanism that was involved in the origin of the type of lesion and its subtypes. The system presents a hierarchy of types, groups and subgroups that are ranked according to its progressive severity and are accompanied by detailed morphological features (see 20 for more details). The length of each fracture was measured using the open-source platform Fiji 30 .