Absolute Ontogenetic Age Calculation

To calculate absolute ontogenetic ages for the development of particular morphological features in ancient species, it is necessary to find a structure (e.g., a tooth) that has a growth period that overlaps the development of the features of interest, and to have data on the timing of growth initiation and the growth rate of the key feature. It is possible to use C¹ growth rate to calculate the absolute ontogenetic ages of initiation and conclusion of eruption for nearly all of the permanent teeth of S. fatalis because the eruption of these teeth occurs while the C¹ forms in this species [13]. What is more, the C¹ growth rate can be used to calculate the absolute ontogenetic timing of other coinciding developmental events (e.g., cranial suture closures and loss of most of the deciduous dentition).

Determining the absolute ontogenetic age at which C¹ growth began is necessary for calculating the absolute ontogenetic ages of development for particular features of S. fatalis. Because the C¹ growth appears to have begun prior to birth [24], a conservative estimate for the duration of gestation in S. fatalis can be used to estimate the absolute ontogenetic timing of C¹ growth initiation. In mammals, gestation is correlated with body mass and the duration of gestation tends to be rather consistent within carnivoran families [21, 25]. The gestation periods of the extant large felids Panthera tigris (104 to 106 days), P. leo (110 days), P. onca (101 days), P. pardus (88 to 112 days), P. uncia (90 to 103 days), Puma concolor (82 to 96 days), and Acinonyx jubatus (90 to 95 days), exemplify this consistency [26–32]. The hyaenids, Hyaena hyaena, H. brunnea, and Crocuta crocuta, have periods of gestation that average 91 days, 97 days, and 110 days, respectively [33–36]. The gestation periods for large viverrids are 60 to 81 days for Civettictis civetta and approximately 91 days for Arctictis binturong [21, 37–38]. Considering that gestation is associated with body mass and that the extant large felids and other feliforms exhibit relatively similar gestation periods, it seems unlikely that the gestation period of S. fatalis deviated drastically from those observed in the extant large felids (Table 1) [21, 25]. Because the estimated body mass of S. fatalis (approximately 160 to 280 kg) [39] is most similar to the body mass of P. tigris [26, 29], P. tigris appears to be the best modern analog for S. fatalis gestation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Body Mass and Gestation of Feliforms.

https://doi.org/10.1371/journal.pone.0129847.t001

Studies have generally shown that deciduous tooth mineralization and permanent tooth mineralization do not begin during the early stages of gestation [45–49]. For example, in the small felid, Felis silvestris catus, tooth buds begin developing at about 28 to 32 days of its 63 day gestation period [29, 46]. Similarly, calcified tooth germs are not perceptible in fetuses of the rodent Myocaster coypus before 100 to 105 days of its 125 day gestation period [50]. Given that approximately 1cm of the forming C¹ crown is present in S. fatalis at about the time of birth [24] and given that the average C¹ enamel growth rate is 5.8mm/month [19], it appears likely that the C¹ of S. fatalis began to mineralize roughly one month prior to birth. Utilizing one month prior to birth as the point of initial C¹ growth, absolute ontogenetic age ranges were calculated for developmental events using a C¹ enamel growth rate range of 5 to 7.5 mm/month determined previously through examination of stable oxygen isotope values incorporated into tooth enamel [19, 51] and C¹ crown lengths previously assigned to particular S. fatalis developmental events [13] (see S1 Text).

Although it is possible to assign specific values to the timing of developmental events, ontogenetic age ranges are more appropriate for describing the timing of developmental events in the life history of S. fatalis. Tooth development and the fusion of cranial sutures are known to occur at various ontogenetic ages for different individuals of other felid species due to factors such as sexual dimorphism [20, 52]. For instance, the timing of initial permanent tooth eruption in P. leo differs by as much as 6 weeks between males and females [53]. In S. fatalis, studies suggest that there is either no sexual dimorphism in skull size or that the degree of size sexual dimorphism in S. fatalis is less than that observed in P. leo [54–56]. However, S. fatalis skulls do appear to exhibit a degree of shape variation that is equivalent to the degree of morphometric sexual dimorphism in Panthera [56]. The exact nature of S. fatalis sexual dimorphism notwithstanding, the timing of developmental events in the life histories of mammalian species can be influenced by other differences that exist between individuals, such as disparities in nutritional health that alter the rate of cranial suture fusion [57]. Thus, even though there may be a popular desire to bestow fixed ontogenetic ages to the developmental events in the life history of S. fatalis, we believe that calculated ontogenetic age ranges serve as more appropriate guidelines.

C¹ Eruption Rate Calculation

In S. fatalis, C¹ enamel formation was completed before the loss of the dC¹ and the completion of C¹ eruption. Thus, another approach, beyond using the canine growth rate calculated using stable oxygen isotopes [19], was necessary to determine the absolute ontogenetic ages for these two developmental events, and that was the calculation of the C¹ eruption rate. It should be noted that C¹ growth rate and C¹ eruption rate are closely connected processes, but dissimilar from one another in that C¹ growth rate describes the rate of tooth formation, whereas C¹ eruption rate describes the rate at which the distal-most portion of the C¹ moves in relation to its alveolar border. It was not possible to calculate the C¹ eruption rate by merely measuring the C¹ length in specimens of different ontogenetic stages. Preservation of the un-mineralized portion of this tooth can be inconsistent between specimens. In addition, because the formation of tooth enamel is a two-part process consisting of matrix formation and subsequent mineralization that affects stable oxygen isotope values [58], C¹ eruption rate calculation using the previously published C¹ growth rate [19] should be more accurate if based on C¹ mineralized enamel lengths (S1 Text). Therefore, calculation of the C¹ eruption rate required determination of the proximodistal lengths of C¹ mineralized enamel, which were determined ultimately by using micro-computed tomography (μCT) to examine the proximal portion of the C¹ of each specimen (see Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Smilodon fatalis specimen UCMP 152565 with partially formed permanent upper canine.

(A) Photograph of UCMP 152565. Scale bar represents 1 cm. (B) Image of the 3D model created from the uCT data of UCMP 152565. The model represents the proximal-most portion of the canine. Dark brown color indicates completely mineralized enamel, whereas light brown is unfinished or unmineralized portion of canine. Scale bar represents 1cm. (C) A slice from the uCT data of UCMP 152565.

https://doi.org/10.1371/journal.pone.0129847.g001

Although the number of identified specimens and number of individuals of S. fatalis at Rancho La Brea is quite large, certain criteria (e.g., C¹ of different ontogenetic stages) were needed to calculate the C¹ eruption rate, which severely limited the number of specimens that were relevant for analysis. ZJT and RSF identified a total of six specimens (LACM A-3748, LACM A3775, LACMHC 2001–9, LACMHC 2003-R-1, UCMP 152565, and UCMP 152566) as suitable for C¹ eruption rate calculation on the basis that each of these specimens has at least one intact, erupting C¹. Due to the spatial constraints of the μCT bore, the large size of the specimens, and similarity in ontogenetic stage between some specimens, only three specimens (LACM A-3748, UCMP 152565, and UCMP 152566) were able to be scanned. For this study, a SCANCO Medical Viva CT40 μCT scanner was utilized at the Center for Biotechnology and Interdisciplinary Studies at Rensselaer Polytechnic Institute (Troy, NY). Specimens were scanned at an energy of 70kVp, an intensity of 114 uA, and a 199 ms integration time. Scanning was performed near the proximal portion of the C¹ with voxel size of 19 μm. The resulting image stacks had maximum dimensions of 2048 x 2048 pixels. 2100 slices were produced for UCMP 152565, which was scanned in two parts. 3036 slices were produced for UCMP 152566, and 5039 slices were produced for LACM A-3748. Three-dimensional reconstructions of enamel and dentin were created using Mimics version 13.1. Because of the spatial constraints of the μCT, two specimens (UCMP 152565 and UCMP 152566) yielded data (S1 and S2 Datasets) that were sufficient for creating three-dimensional reconstructions usable in this study. LACM A-3748 was able to be scanned; however the data were incomplete because the specimen was too large to fit properly within the μCT bore, thereby making it impossible to generate a complete three-dimensional reconstruction for LACM A-3748 (see S3 Dataset).

Density differences in the form of intensity grayscale values (with fully mineralized enamel appearing white, and dentin and other bones appearing as darker shades of gray) were utilized to segment the dentin versus enamel structures in the μCT data. As a consequence of limitations in computing resources, the reconstructions were built from reduced datasets that were 50% of original resolution in the x and y dimensions, and 10–30% in the z direction. The reductions resulted in an effective x-y pixel size of 38 μm and interslice distance of 57–190 μm, providing a maximum error of +/-190 μm (i.e., the exact location of the proximal borders of dentin and mineralized enamel are not resolved below the interslice distance of the reduced dataset) in the placement of measurement points.

To determine the C¹ mineralized enamel length of each specimen, the distance from the most proximal border of preserved dentin to the most proximal border of mineralized enamel along the anterior edge of the C¹ was subtracted from the overall C¹ length. Measurements of the distance between the most proximal border of preserved dentin and the most proximal border of mineralized enamel along the anterior edge of the C¹ were made within the 3D window in Mimics to the nearest 0.01 mm. Digital calipers were used to measure the overall C¹ length of each specimen, as well as the length of erupted C¹ from the alveolar border. The alveolar border, defined as an anteroposterior straight line across the alveolus on the maxilla and premaxilla at its distal-most point, represented 0.0 mm. Using the previously published average C¹ enamel growth rate of 5.8 mm/month (range = 5 to 7.5 mm/month) [19, 51], we estimated S. fatalis C¹ eruption rate, utilizing the following equation: (1)

The C¹ mineralized enamel lengths of two specimens (UCMP 152565 and UCMP 152566) were used to calculate the C¹ eruption rate. The C¹ eruption rate and associated lengths of erupted C¹ (which are known via the previously published tooth replacement sequence of Tejada-Flores and Shaw (1984) [13]) were used to determine the period of time (in months) from initial C¹ eruption to the loss of the dC¹, as well as the period of time from initial C¹ eruption to the completion of C¹ eruption. The period of time from initial C¹ eruption to the loss of the dC¹ was added to the absolute ontogenetic age at which the C¹ begin to erupt (calculated using the C¹ enamel growth rate range of 5–7.5mm/month [19]) in order to calculate the absolute ontogenetic age range of dC¹ loss. Similarly, the period of time from initial C¹ eruption to the completion of C¹ eruption was added to the absolute ontogenetic age at which the C¹ begin to erupt in order to calculate the absolute ontogenetic age range for the completion of C¹ eruption.

Species Comparison

Determining the timing of tooth development and eruption in S. fatalis makes it possible to compare aspects of the life history of this saber-toothed species to those of several extant conical-toothed species. The data suggest that dC¹ eruption, and thus eruption of the deciduous dentition, in S. fatalis did not conclude until 11.5 to 18 months of age. In contrast, the entire deciduous dentition is completely erupted at about 1.5 to 2 months in P. tigris and by 2 to 3 months in P. leo [20, 26]. S. fatalis and P. leo are also dissimilar in that the parietals appear to have fused at 11.5 to 18 months in S. fatalis as opposed to 26 to 28 months in P. leo [20]. Tejada-Flores & Shaw (1984) [13] point out that the closure of this suture and the growing sagittal crest in S. fatalis indicate that this relatively young age class possessed a large attachment area for the temporalis muscles. Permanent tooth eruption began at a slightly later age in S. fatalis (9–14 mo.) compared with many of the extant conical-toothed species such as P. concolor (8 mo.), P. leo (8-9 mo.), and P. tigris (8.5-9.5 mo.) [20, 26, 27]. Conversely, permanent carnassial eruption finished at a younger age in S. fatalis (12–19 mo.) than it does in P. leo (18–24 mo.) [20]. Notably, permanent carnassial eruption coincides with permanent incisor eruption in S. fatalis, whereas permanent carnassial eruption takes place after permanent incisor eruption in P. leo [13]. Nevertheless, on the whole, S. fatalis appears to have had delayed dental development compared to dental development in similar-sized extant felids [20, 26]. Another one of the more distinctive features of S. fatalis tooth development is that the dC¹s appear to have been retained until approximately 23 to 30 months of age, which suggests that S. fatalis possess both the erupted dC¹s and the erupting C¹s for a period of about 11 months. This period of time is greater than those observed in several extant conical-toothed felids (e.g., P. leo possess both the erupted dC¹s and the erupting C¹s for a period of about 3 months) [20, 27, 52, 53, 59–62]. Considering that the upper canines of saber-toothed carnivores seem to have been at a greater risk of fracture from medio-laterally directed forces than are the conical upper canines of extant felids [63], this relatively prolonged period of time that the deciduous upper canines remain in place alongside the erupting permanent upper canines may be noteworthy. Further exploration is currently underway to evaluate the possibility that this pattern of canine replacement (i.e., deciduous canines remaining in position alongside permanent canines for a large percentage of the permanent canine eruption process) functions as a protective mechanism that reduces the likelihood of saber-tooth fracture.

Behavioral and Ecological Implications

Although the S. fatalis fossil record at Rancho La Brea is extensive, there is a scarcity of S. fatalis specimens from younger age classes, which Tejada-Flores and Shaw (1984) suggest is not merely the result of the poor preservation of more fragile juveniles. Explanations for this scarcity include parental behavior that kept cubs in close proximity to den sites and that preferred den sites were not common near Rancho La Brea [13]. P. leo cubs become somewhat mobile at approximately 4 to 6 weeks old, which is also the approximate age at which they are introduced to the rest of pride [53]. Similarly, P. tigris cubs tend to remain at den sites until they become fairly mobile at 1 to 2 months old and P. onca cubs do not leave den sites until about 2 months of age [28, 59, 64]. Our results suggest that S. fatalis individuals of less than 4 to 7 months of age were not frequently preserved at Rancho La Brea, and are consistent with the ideas that den sites were rarely located near Rancho La Brea and that parental behavior restricted the cubs to the den sites [13]. The results also seem to be consistent with the concept that there may have been a prolonged period of parental care upon consideration that it appears S. fatalis juveniles did not possess fully erupted deciduous upper canines until 11.5 to 18 months of age. Watts et al. (2009) [65] showed that for an extant dietary specialist, the bone-cracking C. crocuta, the protracted development of the specialized feeding apparatus affects the rapid feeding performance of juveniles not only on carcasses containing bone, but also on softer foods, which represents a handicap to feeding in general and makes maternal aid crucial to juvenile survival while the specialized feeding apparatus is still developing. In a similar way, the relatively delayed completion of deciduous upper canine eruption in S. fatalis could have impacted hunting in juveniles by rendering them less effective as predators prior to possessing their deciduous upper canines, which could have made S. fatalis juveniles more reliant on parental support while the deciduous upper canines were still developing. In addition to the explanations that parental care restricted the cubs to the den sites and that den sites were rarely located near Rancho La Brea [13], the dearth of S. fatalis specimens from younger age classes may be due to predator avoidance behavior that typically prevented S. fatalis cubs from leaving cover to directly feed on carcasses mired in the asphalt seeps. P. tigris cubs are known to be present at kill sites by as early as 3 months of age, but the cubs remain concealed within cover while tigresses make the kills alone; subsequently tigresses lead the cubs to the kill sites [59, 64]. On other occasions, tigresses transfer the kills, meeting their cubs at alternate locations [59]. Similarly, P. leo adults exhibit the behavior of moving kills to their cubs and P. leo adults sometimes move kills into cover while they are in the presence of other predators [53]. Given that there is a relative abundance of carnivores preserved at Rancho La Brea that has been attributed to predators becoming trapped in the asphalt seeps as they attempted to feed on the carcasses of mired animals, the asphalt seeps may have attracted an array of predators, potentially making it an especially hazardous area for young S. fatalis cubs to move out of cover [23, 66–73]. Thus, it appears plausible that infrequent den site selection near Rancho La Brea, parental behavior that restricted cubs to the den sites and predator avoidance behavior that prevented cubs from directly visiting carcasses mired within the asphalt seeps all contributed to the infrequent preservation of S. fatalis individuals from younger age classes.

Saber-Tooth Development

The data, as expected, suggest that the upper canines of S. fatalis finished erupting at an older absolute ontogenetic age (34–41 mo.) than the absolute ontogenetic age at which they finish erupting in P. leo (24–36 mo. [20]). Yet, the difference in these absolute ontogenetic ages at which the C¹s finish erupting is not as different as one might expect if predictions were made on the basis of C¹ crown heights alone (43–52 mm in P. leo vs. ~130 mm in S. fatalis) [13, 30]. The S. fatalis C¹ crown growth rate (5.8 mm/mo.) was faster than that of P. leo (2.9 mm/mo.)[19] and the data from this study suggest that the C¹ eruption rate of S. fatalis was faster than the C¹ eruption rate of P. leo [20, 30]. It is apparent, however, that C¹ growth rates are not uniform for all saber-toothed carnivores. The C¹ growth rates for Smilodon (5.8 mm/mo. for S. fatalis and 6.3 mm/mo. for Smilodon gracilis) are faster than the C¹ growth rate of the scimitar-toothed sabercat, H. serum (2.8 mm/mo.) [19, 51, 74]. What is more, there are marked distinctions in the types of tooth morphology [75, 76] and tooth growth exhibited by saber-toothed carnivores (e.g., unlike the C¹s of S. fatalis, the C¹s of Thylacosmilus atrox appear to have been hypselodont [77]). Despite these differences, the evolution of saber-teeth may be based upon similar evolutionary changes. In particular, it is thought that regulation of the dental epithelial stem cell niche might have the evolutionary flexibility to produce variation in tooth shape, growth rates, and timing of root development [78, 79]. The loss of fibroblast growth factor 10 (Fgf10) signaling appears to be associated with a shift from crown formation to root development; a delay in this change from crown formation to root formation could lead to the development of hypsodont teeth and the absence of this transition to root formation could generate hypselodont teeth [78, 80]. Therefore, future research may reveal that despite the substantial differences in tooth crown height, growth rate, and growth duration among the various saber-toothed and conical-toothed carnivores, the evolution of saber-teeth may involve similar changes to dental epithelial stem cell niche regulation.

This study emphasized the calculation of absolute ontogenetic age for the development of the dentition of S. fatalis, but this technique for absolute ontogenetic age determination should be applicable in some other species. In particular, if a starting ontogenetic age (i.e., initiation of growth) and growth rate can be calculated for a particular morphological feature, it is possible to calculate its absolute ontogenetic age of completion and the absolute ontogenetic ages of any features that develop simultaneously. Although, hypothetically, it is possible to construct absolute ontogenetic age estimates via linking together multiple developmental processes (e.g., the sequential growth periods of several different teeth), we suspect that approach may prove to be less accurate. Absolute ontogenetic age estimates contingent on a linked series of developmental processes are likely subject to a greater degree of inaccuracy because each additional developmental process includes another range of values due to the variation that exists between individuals of a given species. The technique utilized in this study (focusing on C¹ growth and eruption) may not be useful for absolute ontogenetic age determination in all ancient species, as C¹ growth and eruption may not coincide with the majority of tooth developmental events, but it would seem that studies on taxa with tusks (e.g., proboscideans, walruses, and narwhals) would yield accurate ontogenetic age determinations akin to this study. Rountrey et al. (2012) proposed a similar methodology for aging young woolly mammoths (Mammuthus primigenius) [81]. Another useful technique for ontogenetic age determination that has been used to study ancient species, including S. fatalis, is the analysis of pulp cavity closure using radiographs of mandibles and lower canines [55, 82, 83]. It appears that this technique was difficult to apply to S. fatalis because this saber-toothed felid has a relatively reduced lower canine, whereas the technique was more effective for studying the conical-toothed felid, Panthera atrox [55]. All in all, absolute ontogenetic age determinations should be possible within taxa that preserve anatomical features that record incremental growth, and whose growth trajectories overlap with those of other structures.