Humerus
Left humeri display a non-stationary (i.e., spatially variable in intensity) distribution of tooth marks, with inhomogeneous intensity and with a clustering trend, especially on the proximal end. Tooth marks cluster on the proximal epiphyses, both on the tubercles as well as the articular surface area and proximal metadiaphyses. They also occur in the vicinity of the deltoid crest. Shafts present more abundant modifications on the caudal and medial sides. The cranial and lateral distal shafts show very few tooth marks in comparison. This distribution shows a connection between tooth mark occurrence and areas of muscle and ligament insertions. Tooth marks were probably created during defleshing and limb detachment from the trunk. They are most abundant on the neck junction between the articular head and the proximal metadiaphysis (Fig. 2).
Examples of three-dimensional tooth mark distribution from the lion-consumed carcass sample on each of the four long bones. Distribution of marks is shown on bilateral representation.
Both, the K function and the pair-correlation function indicate an overall trend of clustering. This is nuanced by the other functions. The near-neighbour G function shows a slight clustering trend in short distances and a general asymptotic trend of dispersal in longer distances. The empty-space F function suggests a trend towards clustering within an overall CSR pattern (Fig. 3).
Three-dimensional plot of the distribution of tooth marks on the left humerus. (A) K-function plot. (B) G near-neighbour function plot. (C) F empty space function. (C) Pair-correlation function. The F function suggests a pattern non-differentiable from CSR. The other three functions suggest a mild clustering trend in short distances. Key to (A,B,D) Dotted red line shows the Poisson Complete Spatial Random (CSR) process and the gray band shows its confidence envelope. Black line shows the point process of the target sample (here, tooth marks on humerus). When above the CSR Poisson process, it indicates a clustering trend. When below, it indicates a regular scattering trend. The interpretation is reverse for (C) (F empty space function). Same interpretation applies to equivalent figures in the Supplementary Information.
Right humeri show a similar tendency of mark clustering around the neck under the articular surface with both tubercles impacted. Marks on the medial shaft are slightly more abundant than on the lateral shaft, while the latter shows higher concentrations around the deltoid crest. Interestingly, marks on the shaft cluster on the proximal and distal portions and the mid-shaft is mostly devoid of marks, regardless of orientation. The cranial side, especially the shaft, is again the least impacted by lions. All the functions show a moderate tendency to clustering; so much so in the K and pair-correlation functions because the latter is a modified version of the former (by using rings within the distance radius). The G function shows a very slight clustering trend in short distances, which in the F function is barely outside the CSR envelope (Fig. S4).
In sum, the slight clustering trends in both sides indicate a redundant pattern of tooth mark location. This shows that mark distribution in humeri is not random, since it is repeated across all the carcasses studied.
Femur
Left femora do not show a more widespread distribution of tooth marks than documented in both humeri. Most tooth marks also occur on the proximal half of the element. Most distal tooth marks appear concentrated on the epiphyses. They occur mostly on the medial condyle (on its medial facet) and on the medial portion of the trochlea. Marks on the proximal end occur on the trochanters and also on the spiral line of the neck. The lateral sides of the shaft are the least modified, followed by the caudal distal shaft. Tooth marks on the caudal shaft occur on both sides of the line aspera. As was the case with the humerus, a large portion of marks appear at or near muscle insertion areas. All the functions show a slight clustering trend in short distances and a CSR pattern in longer distances (Fig S5).
Right femora appear substantially more toothmarked than the left ones. Again, the proximal and distal ends exhibit the highest amount of marks. Both trochanters and the proximal metadiaphysis contain large numbers of modifications. Marks on the distal epiphysis occur both on the medial facet of the throclea and on both condyles. Marks on the caudal shaft, along the linea aspera, are more abundant than on the cranial shaft. All functions coincide in finding a moderate clustering trend, which indicates that BSM are not following a CSR pattern (Fig. S6).
As was the case for humeri, the non-random and moderately clustered pattern shows that there are locations, mostly coinciding with tendon and muscle insertions, that are more prone to be impacted by lions during carcass consumption than others.
Radius-ulna
Radii from carcasses consumed by lions are generally left unmodified12,26,27. Most of the damage concentrates on the olecranon of the ulna (Fig. S7). Only a few tooth marks have been documented scattered on the proximal metadiaphysis, some under the articular facet of the lateral epiphysis. The rest occur mostly in the form of isolated marks, without any specific preference for clustering or side. The left radius shows this distribution. Marks outside the ulna are very few and occur on the cranial and lateral sides of the proximal metadiaphysis, in proximity to the articular facet. Scattered marks can be observed on the distal end. In contrast with the stylopodials, the left radius-ulna shows more intense clustering of tooth marks, as denoted by the K,G, F and pair-correlation functions (Fig. S4). This may be the effect of the intense damage on the olecranon.
The right radius appears also very slightly toothmarked, despite the large number of carcasses involved. Most tooth marks concentrate on the ulnar olecranon, with very few scattered along the ulnar shaft and even less so on the radial shaft. The few tooth marks documented on the shaft appear on the uppermost cranial shaft and a couple on the lower caudal shaft. As was the case with the left radius, the second-order functions indicate a clear clustering of tooth marks in slightly longer distances than documented in the stylopods (Fig. S8).
In sum, marks in radii are few and mostly clustered on the ulna. Those on the radial shaft are scattered but also seem to be in connection with damage on the proximal end imparted during defleshing by lions.
Tibia
The left tibia shows a concentration of tooth marks on the proximal end, more specifically, on the epiphysis and, especially, on the crest. Marks on the shaft are not common and they cluster mostly on the lateral and medial sides and on the lateral portion of the caudal side. Marks in the lower half of the shaft are uncommon regardless of orientation (Fig. S9). This element exhibits the lowest frequency of marks of the whole long bone set. The second-order functions indicate a very minor clustering trend in short distances, probably caused by redundancy in damage in the proximal portion of the element, but most of the shaft, where the few scattered marks occur, seems very similar to a Poisson process. This suggests that damage to the tibia (with the exception of the crest and proximal end) is more stochastic than on the other elements.
Right tibiae are only slightly more toothmarked than the left tibiae. Given its overall greater length than other long bones, its low toothmarking frequencies renders them the least impacted elements in number of tooth marks. Most marks cluster on the proximal end, more specifically, on the tibial crest. The lateral side is more damaged than the other sides. In the whole collection, only one tooth mark was found in the distal half of the shaft (Fig. 2). Again, most of the damage on the caudal side was concentrated on the proximal lateral side, coinciding with the more intensive damage on the lateral portion of the cranial side. The second-order functions suggest also a very minor clustering trend, slightly more marked than on left tibiae, probably because all tooth marks documented concentrate on the proximal half of the element (Fig. S10).
In sum, tibiae show some of the least intense point processes resulting from toothmarking by lions on long bones. Marks occurring on the shaft are usually isolated and more random than on other elements, where they are more spatially recurrent.
Bilateral element comparison
Left and right humeri display a similar pattern in the location of most damage as the three-dimensional coordinates of the PCA show (Fig. 4). This is reinforced by the bivariate wavelet analysis, which shows that both sides of humeri show a strong correlation (> 0.8) in the location of most tooth marks in specific locations (Fig. 5). Both humeri display high frequencies of tooth marks (remember, the lower the frequency, the higher the scale) and a clear clustering on the proximal epiphysis and proximal metadiaphysis, as well as on the distal shaft. Most of the mid-shaft shows almost no tooth marks and when they do, they occur in very low frequencies. High frequency marks have only been documented on the proximal epiphyseal portion (Fig. 5). The frequency distribution also shows that the medial and caudal sides bear more marks than the lateral and cranial sides. Most cranial marks are concentrated in the articular surface, tubercles and metadiaphyseal portion of the proximal end.
Principal component analysis (PCA) of each of the four long bones (humerus, femur, radius-ulna and tibia) according to side (left–right) showing point distribution according to components generated by compressing the three-dimensional coordinates. A 95% confidence ellipse per side shows variation and similarity of toothmark patterns in each of the bones. Percentages shown are for the first and second component respectively.
Bivariate wavelet coherence plot showing the correlation of most tooth mark damage on the proximal and distal sections of left and right humeri in low frequencies. Arrows indicate that in these two high-correlation areas, both humeral sides are in phase (i.e., the covary together in the same direction). In the distal area, the right humerus is leading (arrows pointing to the right-down or left-up) and in the proximal area, the left humerus leads (arrows pointing to the right-up or left-down). Binning of histograms is described in Table 3. (A) frequency of marks from distal end (left) to proximal (right) end; (B) frequency of marks from lateral (left) to medial (right), and, (C) frequency of marks on caudal (left) to cranial (right).
Left and right femora also display a similar toothmarking pattern (Fig. 4). Both 95% confidence PCA ellipses overlap in most of their areas. The wavelet coherence analysis shows that both sides display a high correlation (> 0.8) in toothmarking on proximal and distal ends as well as on the shaft when the frequency of marks is low or moderate. Most marks occur on the proximal portion of the element, with a higher impact on the cranial side and more medial for the left femur and more lateral for the right one (Fig. 6). Femoral mid-shafts, thus, appear more highly toothmarked than humeral shafts. Interestingly, the wavelet analysis also shows that when modifications are abundant, there is correspondence between left and right sides only at the distal end. This seems to respond to bone and muscle insertions and ways in which lions deflesh carcasses at this part of the limb. A moderate correlation (> 0.6) between both sides of the element can be found at the level of the proximal articular neck (metadiaphysis) and surrounding the trochanter section (see yellow islands at the level of the 20th-23rd bins in Fig. 6).
Bivariate wavelet coherence plot showing the correlation of tooth mark damage on the proximal and distal sections of left and right femora in moderate frequencies. Arrows indicate that in these two high-correlation areas, both femoral sides are in phase (i.e., the covary together in the same direction). The right femur is leading (arrows pointing to the right-down or left-up). Binning of histograms is described in Table 3. (A) frequency of marks from distal end (left) to proximal (right) end; (B) frequency of marks from lateral (left) to medial (right), and, (C) frequency of marks on caudal (left) to cranial (right).
As was the case of the upper limb bones, radii-ulnae also exhibit a localized tooth mark pattern. The 95% confidence PCA ellipses overlap for both sides is more intense even than with the stylopodials. The only points falling outside the confidence ellipse are those that appear in the form of single marks and are caused stochastically. The wavelet coherence analysis indicates a strong pattern between both sides, with marks clustering in the proximal epiphysis and strong correlation in the exhibition of low-impact modifications (i.e., few isolates marks) in most of the shaft (high scale = low frequency). There is a high frequency of modifications on the proximal end (see black line sloping upwards in Fig. 7), which decreases as we go down the shaft. The low frequency is maintained throughout the length of the shaft. Only because a few more marks have been documented on the distal and proximal ends, do we see a lower scale (i.e., higher frequency) at the beginning and end of the plot. The high correlation spread along the element shaft indicates that both the right and left radii-ulnae display virtually the same modification pattern.
Bivariate wavelet coherence plot showing the correlation of tooth mark damage on the proximal and distal sections of left and right radius-ulna in moderate to high frequencies. Arrows indicate that in these two high-correlation areas, both femoral sides are in phase (i.e., the covary together in the same direction). The right radius-ulna is leading (arrows pointing to the right-down or left-up). Binning of histograms is described in Table 3. (A) frequency of marks from distal end (left) to proximal (right) end; (B) frequency of marks from lateral (left) to medial (right), and, (C) frequency of marks on caudal(left) to cranial (right).
Tibiae also show similar tooth-marking patterns when comparing right and left sides of the skeleton. A PCA shows that a 95% confidence ellipse of samples from both sides overlap in most of their areas (Fig. 4). However, it should be remarked that there is more coordinate variation (i.e., variation in distribution) of tooth marks in tibiae compared to the other long bones. The reason may be double. On the one hand, the tibia exhibits the longest length dimensions of the appendicular skeleton. On the other side, the occurrence of tooth marks outside the area surrounding the tibial crest is commonly in the form of isolated marks that are more prone to occur randomly during defleshing because no muscle insertions occur on the cranial aspect of the element. Only in the proximal caudal side are tooth marks more prone to cluster because of the muscle insertions on that side. A wavelet coherence analysis shows that tibiae show a low density of modifications, similar to radii-ulnae but over a more widespread area. This creates a situation of high correlation between the left and right sides in the location of the few scattered marks (Fig. 8). The correlation is also similar in the proximal and distal ends when modifications are more clustered. Overall, the lack of intensive (i.e., abundant clustering) modifications on the shaft, makes both tibial sides to lack a pattern, with the exception of the lateral and caudal proximal shafts. This moderate clustering there creates the small peninsula between bins 5 and 11 of Fig. 8.
Bivariate wavelet coherence plot showing the correlation of tooth mark damage on the proximal and distal sections of left and right tibiae in moderate to high frequencies. Notice different location of proximal and distal ends compared to the other elements. Arrows indicate that in these two high-correlation areas, both tibial sides are in phase (i.e., the covary together in the same direction). Binning of histograms is described in Table 3. (A) frequency of marks from distal end (right) to proximal (left) end; (B) frequency of marks from lateral (right) to medial (left), and, (C) frequency of marks on caudal(left) and cranial (right).
In summary, the humeri, femora and radii-ulnae exhibit strong patterning on how lions modify them after consumption, as reflected in tooth mark distribution on both sides of the same elements. The tibiae display a more variable pattern, which overall is reflected on fewer modifications, especially along the shaft. Given the commonly isolated nature of most marks created along the shaft, these respond more to stochastic processes and reflect higher variability than in the other elements. Exceptions to this observation are found in BSM observed on the tibial crest and proximal caudal-lateral portions of the shaft.
Multi-element comparison
The information contained in the three-dimensional coordinates of the toothmark pattern documented on each of the elements, when approached through the holistic consideration of the mean values of their global interrelation (as documented through the second-order functions), provides identity information (i.e., element-specific identification) for each of the bones analyzed. On a different scale, this could be applied to individual assemblages instead of individual elements as done here. In the comparison among the different elements and their sides, the way marks were distributed in each of their respective point processes (considering their intensity and distances per element) contained sufficient information to differentiate four different clusters corresponding to the four different elements (Table 2, Fig. 9). Within each element set, both sides were contained within the same node. This is of utmost interest, because in the variables used for this analysis, it is the patterns and not the raw coordinates of marks on each element that were used. This enabled the relativization of the actual location of marks on the different long bone elements and only the emergent properties of the mark assemblage in each of them (understood as individual point process) was considered. Thus, multi-element comparison was possible and different bones were successfully differentiated (Fig. 9).
Hierarchical clustering of the selected variables from the second-order functions, intensity, and nearest-neighbour distance. A phylogenetic dendrogram was used. Four groups were identified (different colors) corresponding to each of the four elements analyzed. Key: lHum (left humerus); rHum (right humerus); lFem (left femur); rFem (right femur); lRad (left radius-ulna); rRad (right radius-ulna); lTib (left tibia); rTib (right tibia).
Source: Ecology - nature.com
