I have studied six skulls of Cave bears coming from the Cova Eirós site (Triacastela, Lugo), stored at the University Institute of Geology of the University of A Coruña (Spain). It is a population that dates back from around 24 000 BP years, with morphological and metric traits similar to those of any typical population of this species and a marked sexual dimorphism (Grandal-d’Anglade and López González, 2005 and Grandal-d’Anglade and Vidal Romaní, 1997). The stable isotope studies made in bones of this population show a mainly herbivorous feeding and a long hibernation period, relating to a time of cold weather at the end of the Marine Isotopic Stage 3 (MIS 3) or beginning of MIS 2 (Fernández-Mosquera et al., 2001). The six skulls are depicted to a same scale in Fig. 1, showing the typical polymorphism of the species, and the main dimensions are in Table 1.

In this work, I have made a theoretical calculation of the BF following the “dry skull method” proposed by Thomason (1991). In this method, the skull is modelled as a third-order lever, in which the effort (jaw closing muscles) is nearer the fulcrum (cotylus) than the load (food item, BF), So: Load   ( BF ) = effort × in - lever   arm out - lever   arm \[ \text{Load}\text{\hspace{0.17em}}(\text{BF})=\text{effort}\times \frac{\text{in}-\text{lever}\text{\hspace{0.17em}}\text{arm}}{\text{out}-\text{lever}\text{\hspace{0.17em}}\text{arm}} \]

The force is calculated by using cross-sectional areas for the major jaw adductors: the temporalis and masseter-pterygoideus muscle complexes, considering the force vectors act transversally to the centroid of the muscle area. These data are obtained by means of image analysis from photographs of the skull in lateral, inferior and posterodorsal norm (Fig. 2).

High-resolution photographs of the skulls were measured using the software ImageJ, a public domain Java image-processing program developed at the National Institutes of Health (Rasband, 1997-2008). Each area or length was measured 10 times, and then extreme values were discarded and mean values of measurements were calculated and used in this study.

That is how I obtain the area of the section of both muscle complexes and the distance from the centroid of each muscle area to the cotyle, which, multiplied by the muscle force, allows us to calculate the force of each group of muscles. As in other similar studies on carnivores’ skulls (Christiansen, 2007a, Christiansen and Adolfssen, 2005 and Christiansen and Wroe, 2007), I have taken the value of 37 N/cm² as the force a mammal's muscle is able to develop.

The arm of resistance is calculated according to the distance between the cotyle and the region of the upper dentition for which one wishes to calculate the BF. In this paper, I have calculated the BF of canine at crown basis, upper carnassial, first upper molar, and second upper molar (at the protocone in all three cases).

Thus: BF = Masseter area × 37 × distance   cotyle - masseter   centroid + temporalis   area × 37 × distance   cotyle - temporalis   centroid Distance   cotyle - teeth \[ \text{BF}=\frac{\left(\text{Masseter area}\times 37\times \text{distance}\text{\hspace{0.17em}}\text{cotyle}-\text{masseter}\text{\hspace{0.17em}}\text{centroid}+\text{temporalis}\text{\hspace{0.17em}}\text{area}\times 37\times \text{distance}\text{\hspace{0.17em}}\text{cotyle}-\text{temporalis}\text{\hspace{0.17em}}\text{centroid}\right)}{\text{Distance}\text{\hspace{0.17em}}\text{cotyle}-\text{teeth}} \]

Estimated BF for ursids and other carnivores already published (Christiansen, 2007a, Christiansen and Adolfssen, 2005 and Christiansen and Wroe, 2007) show striking differences for the same species, mainly in the case of ursids. These differences may be caused by the heterogeneity of the sample, at least in the case of the polymorphic brown bear, whose skull dimensions can dramatically vary along its geographical distribution. Another problem with bears is sexual dimorphism that, in our opinion, makes it necessary to separate sexes for studies related to size, BM or BF. Due to the heterogeneity of the published data, I have decided to use for comparison just those in Christiansen (2007a) and data on large cats taken from Christiansen (2007b).

Table 2 shows the muscle areas and distances to jaw hinge obtained from the photographs of each skull studied. Estimates of muscle forces and BF for each dental piece are offered in Table 3. BF differ noticeably between sexes, pointing to the convenience of studying males and females separately in this high dimorphic species. Male skulls are larger than female ones and, besides, they present reinforced areas of muscle attachment, mainly the occipital area and the zygomatic arches, (Grandal-d’Anglade and López González, 2005), so such a difference was expected.

Morphometric studies (Grandal-d’Anglade and López González, 2005) have shown that in males the musculature involved in biting is absolutely and relatively more developed than in females. Our results show that, in absolute values, males show BF that are very similar to each other, regardless of their size, and clearly higher than those of females (Fig. 3) that form a more heterogeneous group. The differences between the average values of BF in both sexes for each dental piece are statistically significant (t-test, P < 0.001), although the scarce number of individuals (three of each sex) prevent us to make this difference extensive to all Cave bear populations.

Due to the sexual size dimorphism, it is also interesting to compare BF relative to cranial dimensions. For this purpose, I calculated relative BF by dividing the absolute BF by the skull dimensions.

Values of BF relative to total skull length (a), bizygomatic breadth (b), anchura occipital breadth (c) and frontal breadth (d) yield similar intersexual differences (Fig. 4). The difference of mean values for all BF relative to skull dimensions between sexes are aswell statistically significant (t-test, P < 0.001).

It is worth to mention that the BF difference is more marked in relation to frontal breadth (Fig. 4d). In the Cave bear, the anterior temporal muscles are not inserted in an anterior sagittal crest, as in other carnivores, but into high, broad and robust frontal bones. The expanded forehead of the male Cave bear could be a result of the need of a large area for muscle attachment, leading to an anisometric growth of this skull region.

However, when the BF is contemplated with regard to the BM estimated according to Van Valkenburgh (1990), the females’ values come closer to the males’ ones or even surpass the male values (Fig. 5). In a t-test comparing mean values for males and females, the values of P range from 0.465 to 0.757, depending on each dental piece, but in this case, the mean values of relative BF of females are slightly higher than those of males for all the dental pieces excepting the canine.

A morphometric study of Cave bear skulls conducted to determine trends in sexual size dimorphism (Grandal-d’Anglade and López González, 2005) showed that adult female skulls are smaller, albeit not necessarily more slender than male ones, probably due to the fact that the neurocraneus must be conservative because of the similar size of the braincase in both sexes. Male skull are larger than female ones and, besides, they present reinforced areas of muscle attachment, mainly the occipital area and the zygomatic arches, but also the frontal region. Female jaw muscles are less developed than males’ ones, but the shorter skull and jaw of the females allow them larger relative BF, because in females the molars are placed closer to the jaw hinge.

When comparing the Cave bear's BF values with other large carnivores I will focus on the BF in the upper carnassial, because this tooth is used for processing food, unlike the canine, of which function is for killing prey in carnivores and with no feeding function in bears – although it was suggested that bear species that use the canines for tearing trunks or bamboo show a special canine morphology (Christiansen, 2008). Also because published data on extant bears do not include BF for postcarnassial molars.

The Cave bear's elongated skull allows for a lower BF in the carnassial than that of large felines but similar to that of the Giant panda and higher than that of the polar bear, if we consider the mean value for males and females. In Fig. 6 the BF in the carnassial are represented against the skull length (a) and the estimated BM (b).

Both graphics show that the Cave bear (both males and females) fall on a regression line for all other Ursinae ursids, including the almost herbivore Spectacled bear Tremarctos ornatus, the mainly insectivore Sloth bear Melursus ursinus and all the omnivore bears, excluding the outlier Sun bear Ursus malayanus with large BF for its size, perhaps related to its frequent behaviour of tearing into trees and termite mounds (Christiansen, 2007a and Christiansen, 2008).

The herbivore Giant panda, on the contrary, shows higher BF for its skull length and BM than all other ursids, clearly derived from its adaptation to feeding on tough, fibrous bamboo leaves. This adaptation takes place early in its phylogeny, what is confirmed by the morphology of a small fossil panda, Ailuropoda microta from the Late Pliocene of China. Teeth, jaw and skull morphology of this primitive panda indicates that the species was already adapted to a diet of bamboo, at least two to three million years ago (Jin et al., 2007). Its specializations include increased cuspation of the cheek teeth, and a robust expansion of the posterior cranium for enhanced temporal musculature. This is the main difference with the Cave bear, in which the cranium is expanded frontally for attachment of the anterior temporalis muscles, besides having a larger development of the masseter muscle than of temporalis, as can be seen in the values for cross-sectional areas in Table 3. This is a good example of how phylogeny constrains the adaptation to herbivore feeding in different lineages.

On the other hand, the Polar bear, despite of its well-known carnivore feeding, does not show specially high BF for its size and BM, probably due to its short evolutionary history (Christiansen, 2007a).

The Cave bear BF in the postcarnassial molars, which are the pieces commonly used in chewing (according to the degree of wearing in adult individuals), exceed largely the BF values in the carnassials of the Tiger and the Lion. This means a high BF, however concentrated in the backmost part of the jaw, where gape is restricted and only small items can be processed.

In any case, the BF does not express the ability to process certain types of food by itself. The application of that force is done by means of the dental cusps, which in the Cave bear are blunt, not slicing (Grandal-d’Anglade and López González, 2004, Kurtén, 1976 and Rabeder, 1999). The most marked difference with regard to the Brown bear's dentition is the loss of the three anterior premolars and the total loss of the carnassials’ slicing character. Eventually, along the evolutionary line of the Cave bear, it was observed an increase in the molars’ size, with wider and wider occlusal surfaces, with numerous additional blunt cusplets (Grandal-d’Anglade and López González, 2004 and Rabeder, 1999). Therefore, the powerful Cave bear's BF is widely dispelled on a large occlusal surface of a grinding character, and with restricted gape, all of which point at a feeding based on tough vegetable matter.

In the Cave bears from Cova Eirós, males show higher BF than females in absolute terms but more similar with regard to BM, which partly compensates for the females’ smaller size.

The Cave bears studied show lower BF in the upper carnassial than large cats if we consider the average values for both sexes, similar to the one calculated for the Giant Panda and higher than that of the Polar bear.

The estimated BF, together with the skull and teeth morphology, point to a mainly herbivore feeding for the Cave Bear.