METRIC DATA OF THE ATLAS BROWN BEAR

PHALANGES AND METAPODIALS

A total of 45 first phalanges, 23 second phalanges and 30 third phalanges of the Atlas bear were included in the study (see details in Appendix 1). Metapodials are very rare in the Atlas sites, and not all of them are measurable. We have been able to collect measurements of 24 metacarpals and 30 metatarsals. For most of them, we only have the total length and the width of the diaphysis (calculated in many cases based on the robustness index, seeAppendix 3). The statistical summary of the measurements is shown in Table 2. Detailed measurements as shown in Appendix 2 are in Appendix 4.

DISCUSSION

METRIC COMPARISON

The description of the Atlas bear provided by Crowther and reported byBlyth (1841) states of the specimen – a female – that: “Both its toes and claws were remarkably short (for a Bear), the latter being also particularly stout.” With the data obtained so far, it is possible to compare the dimensions of the metapodials and phalanges of the Atlas brown bear with those of Holocene brown bears from the Cantabrian Mountains in the Iberian Peninsula, and, to a lesser extent, with the Syrian brown bear (Ursus arctos syriacus), for which data are more limited.

Considering only the total length of the metapodials, both the metacarpals and metatarsals of the Atlas bears are shorter than those of the Cantabrian brown bears (Fig. 2).

In brown bears, the metatarsals increase in size progressively from the first to the fifth, as described by Torres Pérez-Hidalgo (1988a) in a large sample from the Iberian Peninsula.

This pattern is mirrored in both the Atlas and Cantabrian bears, suggesting that, despite the small number of elements available, the expected trend is maintained. In contrast, the metacarpals of the forelimb do not follow the same pattern of progressive increase in length. While the first metacarpal is clearly shorter, the lengths of the remaining metacarpals are more uniform (Torres Pérez-Hidalgo 1988b). This trend is also evident in the comparison between the Atlas and Cantabrian bears.

The figure also reveals a certain bimodal tendency in the length of some metapodials, which may reflect sexual size dimorphism. Sexual dimorphism is a well-documented trait in brown bears. At present, in Europe, males weigh approximately twice as much as females in the spring and around 1.7 times as much in the fall (Swenson et al. 2023). Although it does not affect the entire skeleton equally, sexual dimorphism can also be expressed in the metapodials, as observed in the Pleistocene cave bear Ursus spelaeus (see for instance Grandal-d’Anglade 1993; Baryshnikov & Puzachenko 2017).

In some cases, outliers are observed in the dataset, as in as in the fourth metacarpal and fourth metatarsal of the Atlas bears, or in the third, fourth, and fifth metatarsals of the Cantabrian bears. Whether these outliers are due to exceptionally large individuals within each population, or to a higher representation of females compared to males, cannot be determined based solely on length. The bivariate plots of total length and transverse shaft width (Fig. 3) provide greater detail regarding size and proportions.

We could hypothesize that the Atlas bear sample is composed primarily of females, and that the individuals whose lengths appear as outliers are males. Even so, their overall size remains smaller than that of the Cantabrian bears, although they appear equally robust, or possibly even more so. Although the number of cases is small, we see this trend towards greater robustness of the Atlas bears in the first and fourth metacarpals, in the fifth metatarsal and especially in the fourth metatarsal, of which there are also more elements in the analysis.

Therefore, although the number of elements analyzed is limited, and although we only have two dimensions for comparison, we can see that the metapodials of the Atlas bear are shorter and, in some cases, clearly more robust, proportionally, than those of the Cantabrian brown bear. This could explain why the Atlas bear’s paws seemed so short, as Crowther mentioned.

As for the phalanges, several caveats complicate their metric characterization. First, the phalanges of each digit differ in size and robustness. According to Koby & Fritz (1950), among the first phalanges of brown bears, that of digit I is always the longest. In a small sample like the one under study, it is likely that the phalanges of each digit are not equally represented. Moreover, front and hind phalanges are not the same size. This is especially evident in the third phalanges, which support the claws; in brown bears, the forelimb phalanges are significantly longer than those of the hindlimbs. Since it is not possible to confidently distinguish the phalanges by digit or by limb, each dataset necessarily represents a heterogeneous mix. Finally, it must be remembered that sexual dimorphism may also affect phalangeal size.

Nevertheless, Figure 4 shows that the first and second phalanges of the Atlas bear tend to be shorter than those of the Cantabrian brown bear, and also slightly shorter than the Syrian bear. A t-test for the length of the first phalanges shows significant differences in the mean (p same mean = 2.90E- 03), and the same is true for a Mann-Whitney test for equal medians (p same median = 4.9102E- 06), when comparing the Atlas and the Cantabrian groups. In the second phalanges, on the other hand, the differences are close to the significance threshold, without quite reaching it. In contrast, the distribution of the third phalanges is markedly different. The bimodality is especially pronounced in the Atlas bear compared to the Cantabrians. The Syrian bear may also follow this bimodal trend, but the available data are insufficient to confirm it.

Additionally, Atlas bears show the most extreme lengths. In this case, we might suggest that the size difference between front and hind claws is even greater in the Atlas bear than in the Cantabrian one. This observation seems to contradict Blyth’s account, “Both its toes and claws were remarkably short”, but it is important to recall that the described specimen was a female, and its dimensions were likely less pronounced than those of the males.

When examining the bivariate plots of the phalanges, comparing proximal widths to total lengths (Fig. 5), a significant difference emerges in the third phalanges. These show not only extreme lengths – both shorter and longer than those of the Cantabrian bears – but also a much greater proximal height, especially in the longer (presumably forelimb) phalanges. The increased height in this proximal area would result in particularly robust claws, which aligns perfectly with Crowther’s observations as described by Blyth. The most robust are those of the Kaf Taht el-Gar cave, but in general, all the Atlas sites show this trend.

In summary, the metric comparison highlights differences in size and proportions between the Atlas bears and Cantabrian brown bears, with the former generally showing shorter but equally or more robust skeletal elements.

CHRONOLOGY AND BODY SIZE

The four new direct radiocarbon dates presented in this study, together with one already available (Fontugne et al. 2012) do not cover all the sites included in the analysis, but only three of them: Ifri Oussaïd, Kehf el-Hammar, and Hattab II. None of them go beyond the Holocene. Thus, we do not have any bear remain directly dating to the Pleistocene.

However, there are two cases where the stratigraphic chronology points to slightly older ages. At Kehf el-Hammar the carbon dating of the levels from which the bear remains were recovered are from the very end of the Late Pleistocene. The oldest level with bear remains (Barton et al. 2005) is around 19 500-19 000 y cal BP. At Kaf-That-el-Gar, 14C dating of charcoals from the archaeological sequence (Daugas et al. 1998, 2008; Martínez Sánchez et al. 2021) shows that the lower level also has an End-Pleistocene date, around 15000-16 500 y cal BP.

This is probably not a long enough time span to assess whether there was a decrease in bear size over time, as suggested by Bourguignat (1870) and endorsed by Arambourg (1933). In fact, based primarily on the dimensions of the molars – of which there is a larger number of remains – Ouchaou (2000, 2008) concludes that bears from the Neolithic and Protohistoric periods are smaller than those from the Paleolithic and Epipalaeolithic. Comparing the scarce End-Pleistocene materials with the Holocene ones (Fig. 6) there does appear to be a decrease in the length of Holocene second phalanges and second, third and fifth metatarsals with respect to the Pleistocene ones. However, very few brown bear elements come from the oldest levels, and in the metatarsals, only one or two elements: clearly, an insufficient number of data to draw a conclusion.

The key issue, however, remains the accurate determination of the chronological context of the materials under analysis. At Hattab II, our dating results yield similar ages regardless of the stratigraphic level from which the samples were taken (see Table 1). This strongly suggests that the sequence is disturbed, as noted by Barton et al. (2008), who attribute some of the mixing to animal burrowing. A similar issue has been observed at other sites, such as Ghar-Khal, where recent excavations have revealed signs of disturbance (Ouchaou 2021). Likewise, bioturbation likely contributed to the mixing of Neolithic and older Epipalaeolithic materials at KafTaht el-Ghar (Daugas et al. 1998). However, in this case, a combination of radiocarbon dating, and ceramic decorative styles supports the integrity of the stratigraphic sequence (Martínez Sánchez et al. 2021). In contrast, the Keft el-Hammar sequence shows a good correlation between the depth of the levels and the radiocarbon ages of the carbon samples, despite the presence of some burrows along the two-metre stratigraphic profile (Barton et al. 2008; Bouzouggar et al. 2008). Altogether, these findings underscore the importance of obtaining a substantial number of direct radiocarbon dates when attempting to compare size or other attributes over time.

DETAILS ON HISTORIOGRAPHY

The main doubts regarding the accuracy of Crowther’s account, as transcribed by Blyth, revolve around two key issues. First, the apparent absence of knowledge among the local population about the presence of bears. As Aucapitaine noted just twenty years later, the inhabitants of the region did not even seem to know the animal’s name (Aucapitaine 1860). This would suggest that the extinction of the Atlas bear in North Africa may have occurred within that short time frame.

Second, Cabrera (1932) almost jokingly suggested that what Crowther actually saw might have been no more than the skin of another animal, perhaps a large wild boar. Blyth indeed comments that it was not possible to preserve the skin of the specimen. However, Crowther’s account gives the impression that he saw the freshly killed body of a hunted animal rather than just its hide. In addition to describing the fur, he comments on characteristics such as overall size and robustness, the short and broad face, and the pointed shape of the muzzle, all of which are difficult to infer from a loose skin alone.

Regarding the size, Crowther claimed that the specimen he observed was “smaller than an American black bear”. This comparison is intriguing, as it implies that Crowther was familiar with the American species, an animal from a different continent. It is possible he encountered one during military service abroad, or perhaps he had seen one at the Zoological Gardens in London, where a juvenile black bear had been on display since 1829 (Zoological Society of London 1829). Access to the Gardens, however, was limited to Society members until 1847. Another possibility for seeing a black bear was the Tower of London Royal Menagerie, a collection of exotic wild animals that existed from 1 200 to 1835, which housed at least one yellow American black bear or “cinnamon bear” in 1821 (Griffith 1821: 236), to which another American black bear, probably with common fur, was added in 1824 (Bennet et al. 1829: 120).

In any case, it’s difficult to determine how Crowther was able to make a reliable comparative assessment of the Atlas bear’s size, just as it remains difficult today to estimate the overall dimensions of the Holocene North African population due to the fragmentary nature of the fossil record. No available measurements allow us to reconstruct the width of the face or the exact shape of the muzzle. However, we have seen that the toes of Atlas bears are generally shorter than those of coeval Cantabrian bears, and their claws, if not shorter, are certainly more robust, details that are consistent with Crowther’s description.

As for behavioral traits, the metric analysis offers limited information. According to Crowther, the Atlas bear “feeds on roots, acorns, and fruits. Does not climb with facility.” This information likely came from local informants, as such details cannot be gleaned from observing a dead animal. Alternatively, Crowther may have inferred the diet based on what was already known about European brown bears at the time. Schinz, for example, offered a detailed description:

“The bear’s diet is chosen from the plant and animal kingdoms, and among all the carnivores it is least restricted to meat, as evidenced by its fine molars, which are adept at chewing and grinding. […] Despite feeding on plants, it is a true predator and its strength allows it to attack large animals” (Schinz 1824: 97).

In contrast, the Atlas bear’s diet, as inferred from both Crowther’s report, seems to have been more plant-based. Notably, there is no mention of animal prey. This suggests a possibly more herbivorous lifestyle compared to its European relatives.

To approach the diet of fossil bears with greater objectivity, stable isotope analysis of bone collagen can be used. This technique has already been applied to fossil Cantabrian brown bears, revealing a strong plant-based component in their diet (García-Vázquez et al. 2018). For Atlas bears, only one such analysis is currently available: a specimen from Ifri Oussaïd (García-Vázquez et al. 2024), which, although limited in scope, similarly suggests a low-carnivorous dietary pattern.

In short, although the nineteenth-century accounts of the Atlas bear must be read with caution, the partial alignment with current data suggests that it was not a mere confusion, but a real population, poorly understood and already absent by the time scientific interest began to grow.