Geometric morphometric assessment of the fossil bears of Namur, Belgium: Allometry and ecomorphology

The Namur area in Belgium is useful to study brown (Ursus arctos) and cave bears (Ursus spelaeus) as the assemblage contains little temporal and no geographical variation. Here, we aim to assess ontogenetic allometry within cave bears, as well as ecomorphological differences between adult brown bears (n = 9), adult cave bears (n = 5) and juvenile cave bears (n = 3). Landmarks for 3D digitization of the mandible were chosen based on the taphonomical damage of the specimens. Extant brown bears and extinct Pleistocene brown and cave bears were digitized with a Microscribe G2. Generalized Procrustes superimposition was performed on the coordinates. Allometry was studied using regression analysis. Principal component analysis (PCA) was conducted to assess ecomorphological differences between the groups. 61% of the shape variance within juvenile and adult cave bears was predicted by size (n = 8, p < 0.01). The juvenile cave bears have relatively deep horizontal rami. In adult cave bears, the horizontal ramus is much narrower dorsoventrally. Juvenile cave bears have a small masseteric fossa and a short coronoid process, whereas both are larger, relative to mandible size, in adult cave bears. This made juvenile cave bears likely less effective masticators than fully grown cave bears. In the PCA, principal component (PC) 1 accounts for 45.0% of the total variance and PC2 accounts for 27.6%. Fossil U. arctos from Namur fall within the 95% confidence interval of modern North American U. arctos on both PCs, but are more similar to cave bears than the average extant brown bear. From the similarity of fossil and modern brown bears, it can be deduced that the diet of fossil brown bears was probably also within the range of their modern North American conspecifics, although they might have been more efficient at masticating plant matter.

Fossil bears display evolutionary allometry through time (Baryshnikov & Puzachenko 2019) and at the subspecies level (Baryshnikov & Puzachenko 2020).Static allometry has also been shown in cave bears and extant bears (Kurt en 1955;Gould 1971;van Heteren et al. 2016; but see Fuchs et al. 2015).Additionally, ontogenetic allometry has been detected in the mandib-ular and cranial shape of brown and cave bears, but with a mixed geographical sample (Fuchs et al. 2015).
The brown bear is the closest living relative of the cave bear (H€ anni et al. 1994;Bon et al. 2008).Bocherens et al. (2011) described finds of brown bears in caves that were inhabited contemporaneously by U. spelaeus.Fortes et al. (2016) stated that movement and migration between predominantly cave bear caves and brown bear caves were possible, and the two species were sympatric.In general, admixture between mammalian species is not rare (Shurtliff 2013).For Ursidae in general, geneflow and hybridization were frequent events in their evolutionary history (Kumar et al. 2017).In addition, admixture and hybridization are reported for U. arctos and U. maritimus (Cahill et al. 2015(Cahill et al. , 2018)).Hybridization between brown and cave bears has also been shown (Barlow et al. 2018) and might lead to intermediate morphologies (Estraviz-L opez et al. 2021).Despite possible admixture, adult extant and fossil U. arctos and fossil U. spelaeus have been interpreted to have different diets based on their differential morphologies (Mattson 1998;van Heteren et al. 2009van Heteren et al. , 2014van Heteren et al. , 2016;;P erez-Ramos et al. 2019).Therefore, it may be possible to observe the morphological consequences of niche partitioning between contemporaneous U. spelaeus and fossil U. arctos in the Namur area.Here, we aim to elucidate static and ontogenetic allometry, as well as functional adaptations to the Pleistocene environment, which was dynamic with cycles of climatic change, and temperatures were much lower than today.The Namur bears, encompassing members of both species, including juvenile cave bears (Fig. 2), form a suitable assemblage to analyse allometry and ecomorphology.

Material
To minimize possible temporal and spatial confounding of results, we restrict our analyses to samples from the Late Pleistocene collected at Namur, Belgium.Therefore, all fossil specimens in this analysis come from the same region and time period: Late Pleistocene, Namur, Belgium.
The present paper focuses on the differences and similarities between U. spelaeus (three juveniles and five adults) and fossil U. arctos (n = 2) from the same area.Mandibles are relatively well represented in the Namur sample and, therefore, are the focal element of this study.Seven modern brown bears were also included in the analysis, as well as a fossil brown bear from Goyet and one from Trou des Nutons (Table S1).Europe currently has a very mild climate, partly because of the warm Gulf Stream.It is expected that modern brown bears from Alaska, Canada and other parts of North America experience a climate much more akin to that of Pleistocene Europe than the modern European brown bears do.Additionally, modern European brown bears are the only extant bears in Europe, but modern North American brown bears share much of their range with the much less carnivorous American black bear, potentially creating competition similar to one between fossil cave and fossil brown bears in Europe.As such, to minimize the effect of climate and competition on the shape of the mandible, the modern brown bear sample (n = 7) in the present study comes from North America.At first sight, the fossil brown bears do appear to be most similar to modern brown bears from Arctic and mountainous areas.Mandibles of adult and juvenile bears from Namur and extant adult brown bears were digitized with a Microscribe G2 desktop digitizing system (Immersion Corporation, San Jose, CA).When both hemimandibles were present, the more complete was chosen for digitization.Landmarks for 3D digitization were chosen to reflect functional aspects of the mandibular corpus (Fig. 3, Table 1).
The data were loaded into MorphoJ version 1.07a (Klingenberg 2010a) using the function 'Create new project'.The datawere loaded in three dimensions from a Morphologika file.Under 'Preliminaries' using 'New Procrustes fit', raw 3D coordinates were scaled, rotated and translated by Procrustes superimposition, which is a form of statistical shape analysis used to analyse the distribution of a set of shapes (Gower 1975;Rohlf & Slice 1990;Goodall 1991).MorphoJ uses a full Procrustes fit with reflection to remove the effect of laterality and projects the data onto the tangent space by

BOREAS
Fossil bears of Namur, Belgium: allometry and ecomorphology orthogonal projection (Klingenberg 2010b).The specimens were aligned by principal axes for visualizations, but the choice of alignment does not influence the statistical results.

Allometry
Juvenile and adult U. spelaeus were compared to adult U. arctos.Within the latter, only static allometry should be present, whereas in the former, both static and ontogenetic allometry are expected to influence morphology.The allometric trajectories cannot be assumed to be the same for static and ontogenetic allometry (Cheverud 1982).The data set was split into juvenile cave bears, adult cave bears, adult fossil brown bears and adult modern brown bears without performing a new Procrustes fit.This resulted in all the subsets being in the same morphospace.Regression analyses of Procrustes coordinates onto log centroid size were performed on each of the subsets.To assess allometry, the regression vector directions in morphospace were compared between the two adult groups (static allometry) and between the juvenile and adult cave bears (ontogenetic allometry).

Ecomorphology
The differences between juvenile and adult U. spelaeus and adult U. arctos were assessed by performing principal components analysis (PCA) on the Procrustes coordinates.Only PCs that explain a minimum of 15% of the variance are interpreted.MorphoJ was used to construct 95% confidence intervals around the PC scores of juvenile and adult U. spelaeus and adult modern U. arctos.

Allometry
The regression of Procrustes coordinates onto log centroid size for juvenile cave bears shows that 55% of the variance is predicted by size (n = 3, p = 0.17).The same analysis for adult cave bears shows that 23% is predicted by size (n = 5, p = 0.48).The angle between the juvenile regression vector and that of the adult cave bear regression is 87°(p = 0.44).Since the vectors of both analyses do not have significantly different directions in morphospace, the two groups were combined for an ontogenetic static allometry regression (Fig. 4).In this combined regression, 61% of the sums of squares is predicted by size (n = 8, p < 0.01).In terms of shape differences, the juvenile cave bears have relatively deep horizontal rami, because they are short in terms of size.In adult cave bears, the horizontal ramus is much narrower dorsoventrally.Juvenile cave bears have a small masseteric fossa and a short coronoid process, whereas both are much larger in adult cave bears.
The regression of Procrustes coordinates onto log centroid size for modern adult brown bears shows that 19% of the sums of squares is predicted by size (n = 7, p = 0.37).When modern and fossil brown bears are taken together, 20% is predicted by size (n = 9, p = 0.12).The latter has slightly more statistical power due to a larger sample size.Although neither result is significant, they are similar enough to reasonably use the combined brown bear data set from here on.The angle between the regression vectors of the combined brown bear sample and the cave bears in morphospace is 67°(p = 0.07).

Ecomorphology
The ecomorphology of the Goyet bears was assessed through a PCA (Fig. 5).PC1 accounts for 45.0% of the total variance and PC2 accounts for 27.6%.Together, this is 72.5%.Subsequent PCs were not considered for interpretation.Adult U. spelaeus have high PC1 and PC2 scores.Juvenile U. spelaeus have low PC1 and PC2 scores.The three groups are relatively well separated; only the 95% confidence intervals of adult U. spelaeus and U. arctos overlap slightly.U. arctos, both fossil and modern, have high PC1 scores and low PC2 scores.Fossil U. arctos clearly fall outside the 95% confidence interval of U. spelaeus and within the confidence interval of modern U. arctos.Adult U. spelaeus resemble both fossil and modern U. arctos with respect to the shape of the horizontal ramus and the position of the masseteric fossa.The coronoid process of adult U. spelaeus, however, is directed much more rostrally than U. arctos and resembles juvenile U. spelaeus.Juvenile U. spelaeus are most different from the other three groups with avery dorsoventrally deep horizontal ramus and a masseteric fossa, which is relatively small, consistent with the allometric results above.

Allometry
Ontogenetic allometry is important for understanding development.In this study, more than half of the shape variance within juvenile cave bears is predicted by size.This result is not significant, but with a sample size of three individuals, it is important to be careful about Type II errors (Cardini & Elton 2007;Brown & Vavrek 2015), especially since in the combined ontogenetic static analysis the results are significant.The shape changes between juvenile and adult cave bears can be found in Fig. 4. The juvenile cave bears have relatively absolutely short horizontal rami, making them relatively deep compared to length.In adult cave bears, the horizontal ramus is much narrower dorsoventrally.A possible explanation is that the deep mandible is to provide space for the development of the permanent dentition underneath the milk dentition.Another possible explanation is that the muzzle is still relatively short, as is common in juvenile mammals.Juvenile cave bears have a small masseteric fossa and a short coronoid process, whereas both are much larger in adult cave bears.The mammalian jaw adductor system is a Class III lever system (Jain 2017).A longer coronoid process and an rostrally expanded masseteric fossa correspond to increased muscle attachment areas and longer in-levers, which is In this analysis, adult and juvenile U. spelaeus are clearly separated from each other Fig. 5.However, the transition from being cubs to being adult cave bears must have been continuous.The seasonality of birth and hibernation and a distinct drop in mortality after surviving the first winter, which has been observed in eastern European cave bears (Debeljak 2007), may be responsible for the absence of certain size classes found at the Goyet cave.The morphology of subadult cave bears is expected to have been intermediate between that of adult and juvenile cave bears.Thus, for example the rostral extent of their masseteric fossa and the length of their coronoid process would have been intermediate.
Static allometry has been shown to be present in extant bears and cave bears (van Heteren et al. 2016(van Heteren et al. , 2019)).Here, we aimed to assess static allometry in a single population.It should be noted, however, that, in the case of the Goyet cave bears, it is not absolutely certain that the specimens under study belong to a population equivalent to modern populations.For cave bears, it has been shown that at least the females display site fidelity (Abelova 2006;Fortes et al. 2016), so the sample can probably be regarded as a chrono-species (Baryshnikov & Puzachenko 2019).The static allometric vectors of adult brown bears and adult cave bears are not significantly different from each other in the present study.This is in line with previous research (van Heteren et al. 2016).

Ecomorphology
Fossil U. arctos mandibles are sometimes mistaken for small or subadult U. spelaeus when hidden in a large assemblage of U. spelaeus.U. arctos and U. spelaeus, however, have different diets (Mattson 1998;van Heteren et al. 2009van Heteren et al. , 2014van Heteren et al. , 2016;;Naito et al. 2016;van Heteren & Figueirido 2019;P erez-Ramos et al. 2019) and are, therefore, expected to be distinguishable based on the morphology of their mandibles.Although there is some overlap between the 95% confidence intervals of adult cave bears and modern brown bears, the fossil brown bears clearly fall within the 95% confidence interval of modern brown bears only.Subadult cave bears are not part of the present sample, but are predicted to have an intermediate morphology between juvenile and adult cave bears.In terms of PC2 values, they would be similar to fossil brown bears, but fossil brown bears would have much higher PC1 scores than subadult cavebears (Fig. 5).In terms of morphology, although subadult cavebears are expected to have a similar coronoid process morphology to brown bears, their mandibular corpus is expected to be much deeper.Of course, brown bears and cave bears should be easily distinguishable by their dentition, but this part of the mandible is not always preserved.
The two fossil U. arctos fall well within the 95% confidence interval of modern North American U. arctos, but have relatively high PC2 scores and fairly low PC1 scores.From the similarity of fossil and modern brown bears, it can be deduced that the diet of fossil brown bears was probably also within the range of their modern North American conspecifics.Particularly, the angle of the coronoid process with the tooth row of fossil U. arctos approaches that of U. spelaeus.In the cave bear configuration, the mechanical advantage of the temporal muscle during a small gape is much larger than the mechanical advantage during a large gape (van Heteren et al. 2016).In the modern brown bear configuration, the mechanical advantages during a large gape and a small gape are much more similar.Accordingly, modern brown bears have a jaw configuration that is suitable for chewing at different gape angles, as would be expected for an omnivore, and cave bears have a jaw configuration that is relatively unsuitable for large gape angles, making them more efficient at smaller gape angles.Fossil brown bears, approaching cave bears in morphology, were less efficient at chewing or biting with a large gape than the average modern brown bear, suggesting an increased specialization to grinding plant matter.There are two possible interpretations.Since cave bears are generally considered to have been herbivorous (Bocherens et al. 1994(Bocherens et al. , 1997;;Stiner et al. 1998;Christiansen 2007;van Heteren et al. 2009), it may be inferred from the PCA that fossil brown bears were on average slightly more herbivorous than their modern North American conspecifics, but still much more omnivorous than cave bears.A second interpretation is that fossil brown bears did not necessarily consume a larger amount of plant matter overall than their extant North American counterparts, but that their fallback food, defined as a food source of relatively poor nutritional quality that becomes an especially important dietary component during periods when preferred foodstuffs are scarce (Marshall et al. 2009), was the same as that of cave bears.One reason why the fossil brown bear diet likely included either more plant matter than extant North American brown bears or contained similar elements to that of the cave bear might be that fossil brown bears and cave bears both lived in the same area and in the same geological time period and, therefore, similar food sources were available to both species.Juvenile cave bears have a different morphology to that of adult cave bears (see also the section on allometry above).Part of the reason for this could be purely developmental, but there is likely also an ecomorphological component.The observations that all juvenile bears in the present study are small with centroid sizes approximately one half of that of the adults (note that Fig. 4 displays the logarithm of centroid size) and that there are no subadults in this Belgian collection suggest that the juveniles at Goyet might still be (partly) dependent on their lactating mothers for nutrition.The shorter masseteric fossa and coronoid process of the juveniles compared to the adult cave bears suggest that they could not chew as efficiently as their adult counterparts.
Dental microwear analysis is commonly used to infer aspects of diet (Molleson et al. 1993;Anyonge 1996;Pinto Llona 2006;Schmidt 2008;Estebaranz et al. 2009;Peign e et al. 2009;Romero et al. 2012Romero et al. , 2013;;€ Ozdemir et al. 2013).When much energy is needed to break a material, the material is tough, whereas solids with surfaces that deform easily are soft, but these characteristics are not mutually exclusive (Strait 1997).Dental microwear of bears is characteristic of their diet; for example, bears with a tough diet consisting of foliage have a high number of fine scratches, mostly with the same orientation, whereas bears with a soft mast diet have a lower number of fine scratches and fine pits, and bears with a hard mast diet have the smallest average number of pits and an intermediate number of scratches (Pappa et al. 2019).No difference in tooth microwear was found between adults and juveniles of Ursus rossicus, a small extinct species of cave bear that lived in the steppe regions of northern Eurasia during the Pleistocene, and Ursus ingressus, an extinct species of cave bear that lived in central Europe during the Late Pleistocene (Baryshnikov & Foronova 2001;Ramirez Pedraza et al. 2022).Although these were different species to the cave bears from Belgium, all three species are closely related, since all three belong to U. spelaeus sensu lato (Gimranov & Kosintsev 2022) (Konyovska et al. 2020), and had a similar diet, generally characterized by tough plant matter such as leaves (van Heteren et al. 2012;P erez-Ramos et al. 2019;Silaev et al. 2020;Fig. 4. A combined ontogenetic static allometry analysis using a regression of the Procrustes coordinates onto log centroid size for adult (turquoise) and juvenile (blue) Ursus spelaeus.Gimranov et al. 2022).Assuming individuals of all taxa within U. spelaeus s.l.shared a similar dietary development through maturation, it seems likely that the juvenile cave bears from Belgium supplemented suckling with tough plant foods (e.g.foliage, Williams et al. 2005), rather than soft mast.Their jaw morphology, therefore, might be related to the spatial needs for the development of their adult dentition underneath their milk dentition.
Subadult cave bears are missing in this Belgian sample.Assuming a direct trajectory through morphospace, their morphology would have been intermediate between those of juvenile and adult cave bears, with a regression score of approximately À0.03 (Fig. 4) and PC scores around PC1 = À0.06 and P2 = 0.00 (Fig. 5), resulting in a masseteric fossa that would have been less rostrally expanded and therefore would have had a shorter moment arm.The subadults were likely not capable of masticating as efficiently as their adult counterparts due to the biomechanics of their jaw musculature, despite already having been weaned.Studies on rats have shown that soft foods during development reduce the sizes of the masseter and temporal muscle (Ikeda 1998), whereas hard foods result in larger muscle size and better dentition in pigs (Ciochon et al. 1997).Similar effects have been suggested for humans (Le R ev erend et al. 2014).As such, subadult cave bears would have needed to masticate tough foods to properly develop their craniofacial bones and muscles, despite a lack of efficiency.Subadult animals with lower masticatory efficiency may have compensated for that by increasing the number of chewing cycles (Fujishita et al. 2015).

Conclusions
Juvenile cave bears were probably less efficient at processing their food orally than their adult counterparts.This may be explained by their (partial) dependence on milk, and the fact that their jaws needed to accommodate the development of their permanent dentition.
Fossil U. arctos probably had a very similar diet to modern North American brown bears, but likely incorporated relatively more or relatively tougher plant matter into their diet than extant North American brown bears.It is possible to distinguish fossil U. arctos and small or subadult U. spelaeus based on the morphology of the lower jaw. of this study are available from the corresponding author upon reasonable request.

Fig. 2 .
Fig. 2. Cave bear ontogeny.A. Close-up of the erupting dentition of RBINS 2835-3; eruption stage 6-month-old; note that the M 3 is oriented vertically in the ascending ramus B. RBINS 2835-6; eruption stage 8-to 9-month-old.C. RBINS 2835-1; eruption stage 11-monthold.D. RBINS 2705-4 (not included in this study); eruption stage 12month-old.E. RBINS 2705-3 (not included in this study); eruption stage 14-month-old.F. RBINS 2705-5; eruption stage corresponds to an adult brown bear.Eruption stages correspond to the equivalent ages of brown bears (Dittrich 1959), actual ages may be different due to differential development rates.B and C have the same scale, and D through F have the same scale.

Fig. 3 .
Fig. 3.The two fragments D2555-0001 and D2555-0002 with the landmarks used in the geometric morphometric analyses.The landmarks were chosen to correspond with existing data sets (van Heteren et al. 2014, 2016) with slight alterations.