Osteometry , Variability , Biomechanics and Locomotion Pattern of the Cave Bear Limb Bones from Croatian Localities

Previously, biomechanical analyses have been widely applied in anthropology and primatology and sporadically in the study of caviomorph rodents. Recently, because of the rich fossil record and contemporaneous presence of similarly designed, yet differently adapted forms, K U N S T (1996) applied biomechanical analyses on femora of Pleistocene European bears. Along with other skeletal traits, femoral cross-sectional shape was interpreted in Ursus spelaeus as being indicative of limb bone structure designed primarily for static stability. Therefore, the previous study of the cave bear limb bones from Croatian sites based on the variability of bone lengths, width of epiphysis and diaphysis (K R KL E C , 1997; J A M B R E π I Ê et al., 2000) is now extended to the analysis of the inner structure, because the internal bone structure yields some information on bone loading which can be analyzed and interpreted in bioOsteometry, Variability, Biomechanics and Locomotion Pattern of the Cave Bear Limb Bones from Croatian Localities


INTRODUCTION
Previously, biomechanical analyses have been widely applied in anthropology and primatology and sporadically in the study of caviomorph rodents.Recently, because of the rich fossil record and contemporaneous presence of similarly designed, yet differently adapted forms, K U N S T (1996) applied biomechanical analyses on femora of Pleistocene European bears.Along with other skeletal traits, femoral cross-sectional shape was interpreted in Ursus spelaeus as being indicative of limb bone structure designed primarily for static stability.Therefore, the previous study of the cave bear limb bones from Croatian sites based on the variability of bone lengths, width of epiphysis and diaphysis (K R K-L E C , 1997; J A M B R E π I Ê et al., 2000) is now extended to the analysis of the inner structure, because the internal bone structure yields some information on bone loading which can be analyzed and interpreted in bio-

Osteometry, Variability, Biomechanics and Locomotion Pattern of the Cave Bear Limb Bones from Croatian Localities
Gordana JAMBRE©I∆ and Maja PAUNOVI∆ mechanical terms.Thus, the aim of this paper is to test a hypothesis of convergent adaptation that was registered on the metapodial bones from Mixnitz (S I V E R S , 1931).
The results are analysed together for each site because of the low statistical frequency of bones in stratigraphically different levels (Table 1).(1997) with a digital calliper.The data obtained were compared and also used for computation of the robusticity index (Ir), representing the ratio of the mean value of the smallest diaphysis diameter and maximal bone length, and bone index (BI) representing the
Diaphyseal length of ulna, fibula and tibia c o r r espond to the maximal bone length.
For selected bones from each site radiographs were taken with the aim of measuring outer diameter (D), inner diameter (M), lateral cortical thickness (CL), and medial cortical thickness (CM) at 50% of the diaphyseal length from the distal end and perpendicular to the central axis (Fig. 1).These parameters have been used for calculation of the general cortical index (CI), medial cortical index (CIM), lateral cortical index (CIL) and cortical area (CA).

STRATIGRAPHY AND PALAEOGEOGRAPHY
The present analysis was performed in order to try to confirm the hypothesis that the difference in morphology of the cave bear limb bones, opposite to the variability registered on the teeth, reflects distinct environmental conditions and not different geological age.Namely, considering the age, the studied material was found in levels ranging from the OIS 7 or 6 to the OIS 2. Thus,  for better understanding of the problem, the stratigraphy of the investigated material must be reconsidered.
The sample from the Vindija cave derives from deposits divided into 13 stratigraphic levels: the levels M and L appear to date to the OIS 7 or 6, unit K was previously correlated with the Eemian (OIS 5e), levels J-D encompass the Last Glacial (OIS 5d-2), while the upper three levels are Holocene (P A U N O V I ∆ et al., 2001).The cave bears from the Velika peÊina cave were found in level k (probably stadial phase of the Middle Würm), but also in levels j-i (interstadial phase of the Middle Würm), e-h (Late Würm), and d (Last Glacial Maximum).In the CerovaËke peÊine cave (P A U N O V I ∆ et al., 2001) only one fossiliferous level was registered and dated to the Last Glacial, i.e. to the OIS 3. Supported by palaeontological evidence the uppermost Pleistocene levels d and e from the Veternica cave are ascribed to the OIS 2, and the levels f, g and h to the OIS 3 (PAUNOVI∆ et al., 2001).Also, the radiometrical data (Table 2) obtained from Croatian sites (WILD et al., 2001), as well as from Austrian and Italian caves (LEITNER-WILD & STEFFAN, 1993) show clearly the different ages of analysed populations.
At the same time, considering the geographical characteristics of the investigated sites, the analysed material was found in the caves belonging to: (1) the hilly-lowlands part of NW Pannonian Croatia (Vindija, Velika peÊina, Veternica) and (2) the Dinarides -mountainous part of E Central Croatia (CerovaËke peÊine).During the Last Glacial the northwestern part of Croatia was quite near to the ice cover of the Alps, while the northeastern area was part of huge steppe areas of the Pannonian Basin which featured a different vegetation, and was characterised by the dominance of sand steppe meadows, grassy steppes, riparian forests, and mixed deciduous and coniferous forests in the hilly regions.At  the same time, the Dinarides (as the southern extension of the Alps) played the role of a physiographic barrier between north and south.Although not glaciated today, they have supported isolated glaciers and were characterised by high mountain (?alpine) vegetation developed on a karstic relief.

Sexual dimorphism
The first step in recognizing the problem of expressed sexual dimorphism registered among bears (P E R K I N S , 1973;STIRLING, 1993;REISINGER & HOHENEG-GER, 1998), required separation of the limb bones samples into those representing males and females.
The histograms and dendrograms (Fig. 3) showed a clear bimodal distribution for all bones -especially in the case of greatest and physiological length -in each site.
Thus, the differences in sex ratio obtained by cluster analysis strengthen the clear differentiation into sexes obtained by univariate analyses.While in the investigated 6 limb bone elements from Velika peÊina the ratio is 73 : 61, in all other sites it is approximately 71 : 61.

Biometric analyses
Because of the danger of allometric effects, i.e. size dependence of proportions which is especially relevant in materials representing variability in size and distributions (sex ratio), the frequency distribution for each bone from each site was estimated (Fig. 4).It appears that all samples exhibit the same allometry, that the deviations from the regression are a result of modification, and that the single change deduced from the present data is size difference between geographically different populations.
The Hierarchical Joing (tree clustering) method showed the size variability of all limb elements, and a similar disposition for both sexes within the cluster of higher values for Vindija, Velika peÊina and Veternica, and the cluster of lower values for CerovaËke peÊine, Conturines, Herdengel and Gamssulzen (Fig. 5).
A robusticity index, ratio of the mean value of smallest diaphysis diameter and maximal bone length, was calculated for both sexes together, by reason of comparison with relevant but unsexed data from some European localities.
The strongest bones are the humerus and tibia.At the same time, the Ir of femora most probably indicates different types of locomotory and loading pattern: values higher than 10.0 are registered among material from Veternica, Vindija, Velika peÊina and are similar to the indices from Azé, Berzé and the Iberian caves.On the other hand values lower than 10.0 are characteristic for bones from CerovaËke peÊine, Conturines, Herdengel and Gamssulzen (Fig. 6).Thus, biometric analyses, including robusticity index as an indicator of body size, show variability of the limb bones and the difference in body mass between geographically distinct populations of the cave bear.

Biomechanical and locomotory analysis
The analysis consists of calculation and comparison of the bone index (BI), cortical index (CI) and cortical area (CA).The bone index (BI) reflects the ratio between the mean value of the smallest diaphysis diameter and diaphyseal length.Unlike the maximal length (used in Ir), representing linear distance between two punctuate points on the epiphyses, diaphyseal length strictly connote maximal length of a central axis and therefore rep-resent the length of the most laden part of the bone.Accordingly BI could indirectly reflect a type of bone loading.
The estimated BI for CerovaËke peÊine and Conturines, Herdengel and Gamssulzen caves are very similar, while the BI of the bears from the Veternica and  Velika peÊina caves are higher, approximately 5-10% for males and 10-12% for females (Table 4, Fig. 7).
Calculated general cortical index (CI), medial cortical index (CIM), lateral cortical index (CIL) and corti-cal area (CA), which also reflect the limb loading type, posture and locomotory pattern (Table 5), were examined by the Hierarchical Joing (tree clustering) method.The dendrograms clearly show two main clusters con-   sisting of males and females, and within each of them a disctinct sub-cluster of the CerovaËke peÊine cave (Fig. 8).
The analysis of the CIM and CIL calculated for the cave bear bones and of CIM and CIL of the recent brown bear sample, shows that there is a difference in the loading pattern of limbs: on the forelimbs loading is stronger laterally, and medially on the hind limbs.But, there is no distinction between various populations of the cave bear as well as between these two species.

DISCUSSION
Because the vertebrate body is defined by dimensions, weight, constitution, and proportions, its characteristics are closely related to the configuration, geometry and distribution of the mass, thus with inner mechanical forces deriving from the activity of the organism and/or caused by external forces connected with the environmental conditions (NIKOLI∆ & HUDEC, 1988).For example, performed anthropological studies (© E ∆ E-R O V , 1973) showed differences between the mountain and lowland populations expressed in the different width of the cortical bone and robusticity of the skeletal elements.These differences were explained with the subsistence and activity pattern of investigated human groups in different surroundings.The investigations of the human long bones were performed using, among other characteristics, the fact that the diaphyseal crosssection is non-homogeneous, and that the stress and tension is stronger on the outer contour of the bone than near the medullar cavity.Thus, using the actualistic theory, we have applied the same methods to study the biomechanical and locomotory pattern of the fossil sample, especially because the cave bear was previously treated as a plantigrad animal, and the obtained results showed a clearly different pattern of ecologically different populations.
Therefore, if external mechanical forces depend on the activity of the organism, but in the first place reflect the ecological influences, the present study leads to the following conclusions: -The statement that large sized males are numerous at low altitudes, while females seem to be more abundant in higher regions (REISINGER & HOHENEG-G E R, 1998) must be reconsidered.Our oppinion is that the sex ratio of the cave bears depends in the first place on the site morphology (exogene or endogene cave), i.e. its function (dense or periodically visited shelter), geological processes and human activity.
-The estimated bone index is lower for the material from the mountainous CerovaËke peÊine caves than the bone index for hilly-lowland sites Vindija, Velika peÊina and Veternica indicating that limb loading was caused by the smaller body mass, i.e. the type of movement was similar to the locomotion of the recent brown bear.
-The RTG-osteometry of the studied bones showed the biggest difference in a cortical area (amount of cortical bone that is proportional to the axial rigidity, i.e. closely related to the body weight) indicating that the hilly-lowland forms from Vindija, Velika peÊina and Veternica were heavier and bigger and therefore featured clumsier locomotion.
-Comparison of medial and lateral cortical indices of all samples leads to the conclusion that plantigrady (sensu D E N D A L E T C H E , 1986) described by P R A T & THIBAUT (1976) and C H A G N E A U (1986) is not a characteristic of the cave bears.From the obtained results it is obvious that the cave bear, similarly to the recent brown bears, was a semiplantigrad with somewhat more expressed plantigrady on the hind limbs.
Thus, because differences in biometry and biomechanical characteristics between geologically older and younger populations were not registered, the obtained results confirmed only the discrepancy between the hilly-lowlands-and mountain-populations of the cave bear, and coinciding with the study of V I R A N T A (1994) indicated different limb loading and locomotory patterns.The smaller and more intensively movable forms (?cursorial) of the cave bear were adapted to harsher environmental conditions with insufficient food sources.
Still, a more integrated approach, i.e. careful morphological (not only metrical) analysis (e.g.articular shape, cross-sectional morphology, muscular scars, etc.) of skeletal elements, especially of the distal elements, is required for future studies of the cave bear because the effects of altitude, local morphology, pleistocene climate and fodder quality upon bear size/morphology remain only partially understood and not straightforward.

Fig. 3
Fig. 3 Examples of sex separation by K-means clustering method.

Fig. 8
Fig. 8 Example of radiograph analysis performed on material from the caves Veternica, Vindija, Velika peÊina and CerovaËke peÊine by Joing tree clustering method.

Table 5
General cortical index (CI), medial cortical index (CIM), lateral cortical index (CIL) and cortical area (CA) estimated for the cave bear long bones from the Veternica, Vindija, Velika peÊina, CerovaËke peÊine caves and the long bones of the recent brown bear.