Middle Pleistocene Steppe Lion Remains from Grotte de la Carrière (Têt Valley, Eastern Pyrenees)

Late Pleistocene cave lions are one of the most iconic species of Northern Hemisphere Quaternary taphocoenoses. Despite their often-scarce record in cave environments, their ubiquitous distribution across Eurasia and North America assemblages attests to their position as top ice-age predators. Nevertheless, the origins of these former large felids, their distribution during the Middle Pleistocene, and their paleoecology during co-existence with the scimitar-toothed cat Homotherium remain debated. Here we describe for the first time an abundant collection of large-sized and stout felid remains from the recently discovered site of Grotte de la Carrière in Eastern Pyrenees, with an estimated age corresponding to MIS 9. Our results highlight the larger size of Middle Pleistocene lions compared to Late Pleistocene ones as well as a trend of decreasing in size, which has been previously stated by other authors. Grotte de la Carrière steppe lions have similar morphological and biometrical parameters to those of other samples from MIS 11–9, being larger and stouter than younger latest Middle Pleistocene-Late Pleistocene forms and slightly smaller than older MIS 15–12 forms.


The Earliest Lions and their European Record
Nowadays, the African lion (Panthera leo) lives mainly in scattered populations across Sub-Saharan Africa (Stuart and Lister 2011;Bellani 2020). However, Middle to Late Pleistocene lions were one of the most widely distributed terrestrial carnivores (Kahlke 1999;Barnett et al. 2009;Stuart and Lister 2011;Bellani 2020), being recorded in Europe, Asia, and North America in the Late Pleistocene under different Panthera specific attributions, such as Panthera spelaea, P. fossilis, P. intermedia, and P. atrox. Since the split of their ancestors in eastern Africa, which has been dated to the earliest Early Pleistocene (ca. 2.5 Ma) (Werdelin and Lewis 2005;Argant et al. 2007;Sabol 2011), lions spread over the Holarctic domain, where they became one of the top predators of the European carnivore guild (Turner 2009;Lewis et al. 2010;Bocherens et al. 2011). They survived until the early Holocene, when the last cave lion lineage went extinct ca. 12 ka (Stuart and Lister 2011).
Historically, the Middle Pleistocene lion fossil record is not as well-known as the Late Pleistocene one due to the scarcity of Middle Pleistocene sites. Sites have mainly been known in Southern and Central Europe (Fig. 1) such as Breccia de Château (Argant 1988) and Mauer and Mosbach (Schütt and Hemmer 1978). This scantiness does not help clarify the evolutionary history of Pleistocene lions. The first documented Eurasian lions come from Pakefield (ca. 750 ka; Lewis et al. 2010), Kozi Grzbiet (ca. 750 ka; Marciszak et al. 2021), and Isernia la Pineta (ca. 650 ka ;Sala 1990). Recent paleontological works have pointed to eastern Africa as the origin of lion dispersion around 800 ka (Argant 1991;Burger et al. 2004;Argant et al. 2007;Sabol 2011). However, the presence of lions in the latest Early Pleistocene of Iberia (i.e., Vallparadís Section and Cueva Victoria, Spain) suggests that the complex history of European lions may have started earlier (Madurell-Malapeira et al. 2014. Furthermore, an entrance from Western Siberia should not be ruled out, as Sotnikova and Foronova (2014) recently described P. fossilis in the Kuznetsk Basin (Russia) at ca. 1.0 Ma.
Unfortunately, the taxonomy of the cave lion lineage still generates debate in the scientific community (Sabol 2011;Marciszak et al. 2014). According to some authors, skull similarities with modern lions indicate that the Pleistocene forms are a large extinct subspecies of P. leo (Ballesio 1975(Ballesio , 1980Turner 1984). In contrast, others support the existence of two valid chronospecies: Panthera fossilis (von Reichenau, 1906) and Panthera spelaea (Goldfuss, 1810) (Sotnikova and Nikolskiy 2006;Sabol 2014; Barnett et al. 2016). Other scholars regard both of these species as subspecies of P. spelaea (Argant et al. 2007;Barycka 2008;Marciszak and Stefaniak 2010). Recent DNA studies reveal that the cave lion putatively dispersed over Europe ca. 600 ka after becoming isolated from Asian and African populations (Burger et al. 2004; Barnett et al. 2009) and diverged from modern lions around 1.89 Ma (Barnett et al. 2016), becoming a clear monophyletic outgroup (Barnett et al. 2016;deManuel et al. 2020). Other more recent studies have supported the existence of three different clades of cave lions that diverged from extant species ca. 1.85 Ma. One of these clades was putatively restricted to Beringia, while others were prevalent in Eurasia (Stanton et al. 2021).
All these phylogenetic problems are partly a consequence of the high adaptability of European lions to diverse environmental conditions and biomes (Sabol et al. 2018). Following this interpretation, it seems plausible that P. fossilis was distributed throughout European lands (Fig. 1), inhabiting a great variety of biotopes, from open lowlands and woodlands to mountainous boreal forest (Turner 2009;Baryshnikov 2011;Sabol et al. 2018). Their large size qualifies them to be predators of a wide prey spectrum, including large-sized herbivores and occasionally cave bears (Hemmer 2004;Baryshnikov 2011;Bocherens et al. 2011;Diedrich 2011;Stuart and Lister 2011;van Valkenburgh et al. 2016). In general, regional differences attributed to habitat and prey availability, in addition to sexual dimorphism and the scantiness of fossil record, constrain attempts to correctly resolve phylogenetic relationships (Schütt 1969;Turner 1984;Argant et al. 2007;Diedrich 2011;Christiansen and Harris 2012;Marciszak et al. 2014Marciszak et al. , 2017Sabol et al. 2018).
The field surveys performed the last five years in the newly discovered site of Grotte de la Carrière inside the Reséau Lachambre (Têt Valley, Eastern Pyrenees) have provided new Middle Pleistocene lion material chronologically constrained to MIS 9 (Fig. 2) and previously tentatively attributed to P. fossilis Llenas 2018, 2019;Prat-Vericat et al. 2019, 2020. The aim of the present paper is to provide a detailed description of all unearthed specimens, determine their morphological and taxonomical position close to the latest known occurrences of P. fossilis, and report new knowledge about the paleobiology of this European Lion.  The Reséau Lachambre, located in the surroundings of Villefranche-de-Conflent village, is a karstic system with more than 25 km of galleries and several accessory cavities (Madurell-Malapeira and Llenas 2018). In one of its accessory caves, namely Grotte de la Carrière, five systematic excavations recovered more than 8,000 well-preserved large mammal remains in two distinct stratigraphic Sondages , 2020.
The cavity located at 540 m of altitude and 75 m above the Lachambre main cave is a Middle to Late Pleistocene cave that was inhabited by the cave bear Ursus deningeri in the lower layers (Sondage 1) and Ursus spelaeus in Sondage 2 (Madurell-Malapeira and Llenas 2019; Prat-Vericat et al. 2020). According to preliminary data, the lion remains have been found in the lower layers of Sondage 1 (Fig. 2)

Material
All the specimens described herein were restored by the preparation lab team of the Institut Català de Paleontologia Miquel Crusafont (ICP, Spain) and are under supervision of the Direction Regionalle des Affaires Culturelles (DRAC) of Occitanie (Montpellier, France) and the Conflent Spéléo Club de Prades and housed in Prades-de-Conflent (Occitaine, France).

Methods
The statistical approach for the systematic classification of Grotte de la Carrière (GDC) individuals was based on classic morphometric analyses performed from dental morphology and represented by bivariate graphs and boxplot performed in PAST v. 3 (Hemmer et al. 2001). We follow Schmid (1940) for terminology (dental and osteological) and the measurements taken with a digital caliper to the nearest 0.1 mm ( Fig. 4; Table 1). To test the normal distribution of the analyzed variables such as the homogeneity of variance across the groups, a Shapiro-Wilk test was carried out on data sets including at least two specimens. Finally, we test the variables that showed significant intergroup differences in a one-way ANOVA (for parametric groups of dental specimens) or Mann-Whitney pairwise (for non-parametric groups of post-cranial specimens).
The Mann-Whitney test was run to determinate the statistical relevance for the astragalus (n = 3). Additionally, we tested how several dimensions are affected by sexual dimorphism in various skeletal elements throughout a oneway ANOVA for parametric groups. Finally, we tested the putative climatic and regional differences throughout time  in all the available fossil sample; however, the number of complete fossil bones coming from each locality is normally low. If we separate in glacial/interglacial stages or by latitude, the sample size is short to assess the results or to be statistically significative. Furthermore, to realize an accurate latitudinal distinction, we must take into consideration also the differences among biomes.

Description
Dental Remains Only one left c1 (GDC2016-D6-37-Nv6) was recovered. It belongs to an adult specimen. The tooth displays a moderate degree of wear. The apical part is partially fragmented, which prevent us from taking measurements (Fig. 5a). Nevertheless, two clear longitudinal ridges can be seen on the buccal side seen. One left (GDC2016-D5-29-Nv1) and one right (GDC2016-D6-225-Nv6) p3 were studied ( Fig. 5b-c). The shape of the buccal margin is slightly convex on its whole mesiodistal length. Both p3 protoconids are low and asymmetric in buccal view. Mesiolingually, a small and rounded paraconid is visible in occlusal view. Distally, a well-developed accessory cusp is characterized by a rounded morphology. Finally, a very incipient distal cingulum is visible, especially in GDC16-B5-29-Nv6 (Fig. 5b).
Regarding lower forth premolars, only right GDC16-B4-r-Nv6 (Fig. 5d) is complete. In the middle of the crown in buccal view, a high protoconid is well-separated from  (left) and distal articulation facet (right) views. f. GDC2020-C5-393-Nv4, left navicular in proximal articulation facet (left) and distal articulation facet (right) views. g. GDC2020-C5-313-Nv4, right navicular in proximal articulation facet (left) and distal articulation facet (right) views. h. GDC2020-C5-548-Nv4, cuneiform III in proximal articulation facet (left) and distal articulation facet (right) views mesiobuccally placed paraconid and distal metaconid. Both are much smaller than the protoconid and relatively equally developed. A second tiny distal accessory cusp is visible attached to a distinct distal thin cingulum (Fig. 5d). The mesial margin is slender and uniform without any visible cingulum.
The two well-preserved isolated m1s were studied (Tables S6-S7; Fig. 5f-g). The protoconid is higher than the stouter paraconid, being comparable in mesiodistal length. Distally to the protoconid there is a small slightly developed talonid, both cusps are separated by a relatively deep notch. In occlusal view, this valley forms a V shape. In the middle of the lingual side, a tiny tubercle can be observed at the base of the crown. Distally, both m1s display a developed cingulum without any accessory cusp ( Fig. 5f-g). In occlusal view, the crown is convexly curved. Its enamel presents a zigzag structure. Biometrically, p4 is practically equal in size to both studied m1s, being slightly larger (32.5 mm vs. 30.6 and 31.7 mm in GDC16-B4-r-Nv1, GDC15-Pou-176, and GDC16-D6-64-Nv6, respectively).
Axial Skeleton Two complete and well-preserved caudal vertebrae were recovered. The systematic description of caudal vertebrae (or the vertebral column in general) does not report any taxonomic information at the species level. However, the large size of both specimens suggests a large lion (Fig. S1b-c).

Appendicular Skeleton
The lion forelimb skeleton specimens come from the lowest layers of Grotte de la Carrière and are represented by a right ulna and eleven metacarpals from at least two individuals: one Mtc I, three Mtc II, four Mtc III, two Mtc IV and one Mtc V (Table S5; Figs. 6 and 7).
Ulna The well-preserved fragment of a proximal right ulna includes the sigmoid cavity and the olecranon. The sigmoid cavity is concave and deep and forms a C shape in lateral view. An especially robust and elongated olecranon is proximal to the sigmoid cavity (75.4 mm and 30.4 mm anteroposterior and transverse diameter, respectively; Table 1), which displays clear scoring marks (Binford 1981) and pits (Online Resource 1, Fig. S1a). On the cranial surface of the proximal epiphysis is observed a significant depression for the triceps muscle insertion. The anterior part is broken. In general terms, GDC15-D6-37-Nv4 is anteroposteriorly stout.
Metacarpals A stout and short right Mtc I was recovered (GDC-D6-638-Nv4c; Fig. 7F). Two almost complete stoutly built right Mtc II ( Fig. 6a-b) from distinct individual from that of the Mtc I (GDC-D6-638-4c) were preserved. The diaphysis of Mtc II is comparatively less straight and proximodistally large. The medial side of the proximal epiphysis is rectilinear. The shape of proximal epiphysis is triangular in proximal view, displaying an elliptical longitudinally developed facet on the lateral proximal side for Mtc III articulation. On the anterior side, a diagonal depression crosses the proximal epiphysis (strongly marked in GDC16-D6-216-Nv6; Fig. 6a).
Four Mtc III are preserved including two that are almost complete (right GDC-C6-17-Nv4 and left GDC-D7-21-Nv4; Fig. 6c-d) and one that is partially fragmented on the lateral side of proximal epiphysis (GDC-D6-276-Nv8; Fig. 6c). The two well-preserved proximal epiphyses (Fig. 6d) display a deep central depression in proximal view for the capitate carpal joint. The lateral articular facets for the Mtc IV articulation consist of one circular surface on the lateral anterior side and another one, elliptical and longitudinally expanded, on the lateral posterior side, similar to Mtc II. All the three metacarpals display a strongly marked depression on the anterior side of the proximal epiphysis. Contrary to Mtc II and Mtc III, the diaphyses of the two recovered Mtc IV are slender ( Fig. 6e-f). The proximal epiphysis has a stout morphology. The single recovered Mtc V is well-preserved, the shaft is distally curved, and both epiphyses are stout (Fig. 7e). The medial side of the proximal epiphysis presents a large surface with a wide circular facet for the Mtc IV articulation (Fig. 6e).

Fibula
The single right fibula (GDC2015-D6-182-Nv6) is partially fragmented on the proximal epiphysis. Only the distal epiphysis is relatively well-preserved. It shows an irregular rectangular shape with great relief in the distal tibiofibular joint, suggesting a powerful muscle insertion.
Calcaneus The posterior part of the calcaneus is fragmented (Fig. 8c). The calcaneal tubercle measures 87.63 mm in length, and on the superior surface, a strongly marked calcaneal tuberosity is present. On the inferior surface of this calcaneal tubercle, the sustentaculum tali emerges prominently from the anteromedial portion of calcaneus body. In general terms, the calcaneus is large and slender.
Astragalus All three preserved astragali display rounded borders in the neck and the body (Fig. 8a, b, d); the malleolar articulation, visible on the medial side of lateral trochlea, is partially fragmented (except in GDC-B7-579-Nv4), which prevents the measurement of maximal transverse diameter (Fig. 8). Between the lateral and the medial trochlear crests, there is an accentuated depth depression for the superior tibia facet. The lateral trochlear crest is relatively distally oriented, towards the medial one, and is significantly larger. On the posterior part of astragalus body, the wide articular facets for the calcaneus insertion minimizes the calcaneal sulcus space. On the distal part, an especially stout neck and a large surface for the insertion of the navicular are visible (Fig. 8).
Metatarsal Three almost complete large and stoutly-built Mtt III were recovered (Fig. S7a, b). The asymmetrical "T" shape of the proximal epiphysis for the third cuneiform articulation is well-preserved in two of them (Fig. 7a, b). On the proximolateral side (Fig. 7a, b), the proximal epiphysis displays a deep fossa for the Mtt IV articulation. The fragmented right Mtt IV GDC-s1-C5-314-Nv4 (Fig. 7d) preserves only the proximal epiphysis. In proximal view, the proximal epiphyseal surface for the cuboid bone insertion is asymmetrically quadrangular (Fig. 7c). This Mtt IV is associated with Mtt III GDC-C5-11-Nv4 and cuboid GDC-C5-312-Nv4. Finally, the only recovered Mtt V (Fig. 7d) is distally fragmented and presents strong traces of weathering and corrosion on the bone surface along the diaphysis, precluding accurate morphological description (This issue is present in all the bones recovered but is especially prominent in Mtt V).
For further information on the studied specimens unearthed from the Middle Pleistocene layers of Grotte de la Carrière, the reader is referred to Online Resource 1.
Despite the high reliability of teeth for taxonomic purposes, the postcranial skeleton may help when the dentognathic remains are scarce (Barycka 2008;Sotnikova and Foronova 2014;Marciszak et al. 2019). The interspecific differences of limb bones have been observed both dimensionally and proportionally. According to Argant and Brugal (2017) and Marciszak et al. (2019Marciszak et al. ( , 2021, the chronologically older P. fossilis showed an increase in size until MIS11-9, when the largest individuals are recorded. From this moment on, size in fossil lions decreased until they reached a size similar to that of extant African lions. The main problem in considering this simplified size-reduction trend is that it does not seem to respond only to chronological proxies. Indeed, it is necessary to consider the different biotope inhabited, which depends on the climate, the faunal communities, and intraspecific differences, especially sexual dimorphism (see Table S8 for our analysis on the sexual dimorphism of extant lions).
Focusing our attention in the morphological characters, the specimens described in this study share the diagnostic features reported for P. fossilis, namely (Schütt 1969;Schütt and Hemmer 1978;Argant 1988Argant , 1991Argant et al. 2007;Hankó 2007;Barycka 2008;Marciszak and Stefaniak 2010;Marciszak et al. 2019Marciszak et al. , 2021: markedly wide m1 at the base of the crown, larger premolars (Tables S9 and S10), high and sharp protoconid edge and short rounded paraconid edge in m1, mesially oriented edge of the protoconid of m1, small lingual protuberance in the medial part of m1 crown, the zigzag structure in the m1 crown's enamel -taxonomically used by some authors who considered that a common character related to Middle Pleistocene lions (Marciszak and Stefaniak 2010) but questioned by others (Sabol 2014), proportionally short paraconid and oval and slightly convex buccal margin in both p3, presence of abrupt curve in the buccal enamel in m1 that rises below the protoconid, B/T ratio close to P. fossilis average (Table S1), stoutly-built astragalus (Fig. 8), large calcaneus, anteroposteriorly stout ulna, robust proximal and distal epiphyses of third and fourth metacarpals and third metatarsal (Figs. 6 and 7). On the other hand, the analyzed specimens differ from previously reported P. fossilis in: the slightly developed posterior talonid in m1, absence of the groove at the top of the medial lingual protuberance of the m1 (Schütt 1969;Schütt and Hemmer 1978;Argant and Brugal 2017), slender calcaneum, and proportionally proximodistally short metatarsus and metacarpus.
In summary, based on morphological characteristics described and compared above, the specimens from Grotte de la Carrière can be referred here to Panthera fossilis. The combination of morphological similarities with P. spelaea in the dental remains and of the tendency of postcranial remains such as tarsal and metapodial bones to be stouter, as in P. fossilis, further suggest the presence of intermediate features towards late Pleistocene European lions.

Biometrical Comparisons with European Pleistocene Lions
The morphometric differences between European lion groups over geological time normally follow the diagnostic features related to dentognathic remains (Schütt 1969;Argant 1991Argant , 2010Hankó 2007;Sabol, 2014). Alternatively, several researchers have also included the proportions of postcranial in statistical analyses (Dietrich 1968;Ballesio 1980;Bona and Sardella 2012;Baryshnikov 2016;Sotnikova and Foronova 2014;Marciszak et al. 2019). In the present study, biometrical comparisons are focused on lower dental remains and several morphometric characters of postcranial bones, such as tarsals and metapodials. To maximize intergroup differences and to avoid intraspecific variation, we chose those variables with significant intergroup differences supported by statistical tests have been observed: one-way ANOVA (p < 0.05) for parametric variables and Mann-Whitney pairwise (p < 0.05) for non-parametric. The complete measurements are shown in Table 1 and the statistically significant comparisons in Table S11.

Dental Remains
In general terms, European lions display a trend of decreasing size over time from the Middle Pleistocene forms to their most recent relatives (Marciszak and Stefaniak 2010;Marciszak et al. 2019). Nevertheless, the decrease of middle Middle Pleistocene lions to the two younger groups of Late Pleistocene and the extant African lions is not always observed in the geologically older group from MIS 15-12. These forms usually show smaller values compared with those from MIS 11-9 (Marciszak et al. 2019). Notwithstanding, the extant relatives and the MIS 5e-2 lions were significantly smaller than the massive, large lions from the early and middle Middle Pleistocene (Fig. 9a).
The above-mentioned chronological decreasing trend of size was firstly observed in the lower dental remains (Fig. 9a). In the lower m1 B/TL ratio (Fig. 9a), this trend is easily followed from MIS 15 to latest Pleistocene ones, with the GDC specimens better grouped with MIS 11-9 forms, as expected. The box plots depicted from the premolars (Fig. 9a) show that the smaller specimens are the ones of the MIS 15-12 group. Moreover, MIS 5e-2 is larger than MIS 8-6 in the p4 B/TL (Fig. 9a). GDC values fit in the middle Middle Pleistocene box, much closer to the median point of MIS 11-9 box (Fig. 9a).
Concerning interspecific differences, the studied specimens were compared to both European lions, P. fossilis and P. spelaea, as well as extant P. leo. We included the P. intermedia m1 values from Igue de Rameaux in the P. spelaea group, as they are biometrically indistinguishable. Complete measurements of Grotte de la Carrière specimens can be found in Table 1. In Fig. 9b are represented the scatter plot diagrams for maximum breadth against total length for m1, p4, and p3. The one-way ANOVA corroborates the statistical significance of results (see Table S11). Fig. 9 a. Boxplots of total breadth to total length B/TL ratio of lower dental remains from Grotte de la Carrière, Middle to Late Pleistocene localities and extant relatives showing the chronological trend. b. bivariate graphs of total width against total length of lower dental remains In the scatter diagrams for p3 and p4 are observed some clear intergroup differences (Fig. 9b). The set of P. fossilis p3 variables are in the uppermost part of the plot, separated from P. spelaea and P. leo values (Fig. 9b). Concerning the p4 bivariate graph, P. fossilis values are located in the uppermost part (Fig. 9b). In this case, overlapping groups of species can be observed in the lowermost part of the convex hull. In both premolars, the GDC specimens overlap the uppermost part of P. fossilis variance (Fig. 9b). As regards the m1 (Fig. 9b), GDC m1 fits within the P. fossilis cloud, near the uppermost P. spelaea area (population from Lherm and Azé I-3), in the fringe between both species. In detail, GDC values are located closest to P. fossilis from Westbury, Petralona, Aragó, and Vértesszȍlȍs II and P. spelaea from Lherm and Azé I-3.

Tarsal Bones
Among the postcranial elements, the tarsal and metapodial bones have significant results for interspecific correlations (Sabol 2014). In tarsal bones, the chronological trend of decreasing size over time seen in dental remains is also normally displayed (Fig. 10a). However, lions from MIS 5e-2 experienced an increase in size compared to the chronologically older MIS 8-6 group, a tendency also observed between MIS 11-9 and MIS 15-12 for the calcaneus values (Fig. 10a). The tarsal bones were compared to both classical Pleistocene European lions: P. spelaea and P. fossilis. Complete measurements are in Table 1. In Fig. 10b are represented the bivariate graphs for astragalus/calcaneus, plotting the maximum mediolateral diameter against total proximodistal length, with the graphical representation of MD/L and TL ratio in boxplots.

Astragalus
The statistically significant results for the maximum mediolateral diameter to the total length ratio MD/ TL for the astragalus point out a clear division between the two species and reflect how Late Pleistocene P. spelaea differs from the Middle Pleistocene ones based on smaller size (Fig. 10a). Contrary to the observed analysis for the dental remains, the set of variables from MIS 11-9 shows a decreasing size tendency with respect to the older MIS 15-12. The MIS 8-6 group instead, has significantly reduced size compared to the most recent group, MIS5e-2. GDC values are located at the top level of the MIS 11-9 box and at the uppermost MIS 5e-2. In the resulting bivariate graph, the values are perfectly separated into two clouds of points (Fig. 10b). In the uppermost part of the graph, where GDC specimens fit, are located the P. fossilis individuals. GDC-s1-Pou-63, in which the partially rounded bone affected the measurement of maximum transversal diameter, is located outside the P. fossilis range, near one specimen from Zá Hájovnou Cave.

Calcaneus
In the MD/TL ratio for calcaneus, the trend is not as clear as the one for the astragalus (Fig. 10a). The differences are also not statistically significant. Nevertheless, the P. fossilis groups from middle Middle Pleistocene differ statistically in the total length of calcaneus from P. spelaea and P. intermedia from MIS 8-6 and MIS 5e-2 (Fig. 10a). In the bivariate graph, the uppermost part corresponds to P. fossilis values and the lowermost part to P. spelaea (Fig. 10b). The single GDC specimen fits within the P. fossilis variability, near the edge of the P. spelaea cloud where the highest values are located, for Lezetxiki VI and within the Château range.

Metapodial Bones
The taxonomic reliability of the metacarpus and metatarsus is rather poor. These features present an intraspecific variation related generally to the locomotor apparatus in relationship with the habitat (Turner 2009). In general, autopodial bones are practically indistinguishable from one to another. In the comparison below, we included European and North American fossil lions from MIS 15-2 to recent lions and focused only on variables in which the intergroup differences were supported by statistical tests (Mann-Whitney test, p > 0.05; Tables S1, S2, S7 and S11). In Figs. 11 and 12 are represented in boxplots the significant interspecific indices.
In general terms, the P. spelaea from MIS 5e-2 and the extant lions are less stoutly-built than the ancient P. fossilis and North American P. atrox. The recent lions differ from the rest of the groups in the slenderness of both epiphysis and the shorter diaphysis. In P. fossilis from Poland MIS 9-10 and P. spelaea from MIS 6 of Lherm (France), relatively larger values have been noticed in comparison to others of their own groups in the maximum anteroposterior diameter of the proximal epiphysis and the proximodistaly larger diaphysis.
The intergroup differences that are statistically significant for the Mtc III include total length and maximum anteroposterior diameter of proximal and distal epiphysis (Fig. 11). For Mtc IV, they are total length, maximum anteroposterior diameter of proximal epiphysis, maximum transverse middle shaft diameter and maximum transversal diameter of distal epiphysis (Fig. 11). For Mtt III, they are: total length, maximum anteroposterior and transversal diameter of proximal epiphysis, and maximum transversal diameter of distal epiphysis (Fig. 12). Checking the resulting boxplots for the ratio between Dpap/TL and Ddap/TL of Mtc III and Mtc IV (Fig. 11), the midpoint of the GDC data is closer to the top or midpoint of the P. fossilis box, except Ddap/TL of Mtc IV, which are near the bottom (Fig. 11).
According to the performed analyses, the GDC specimens are significantly distinct from P. spelaea in maximum anteroposterior diameter of the proximal epiphysis of metacarpals III and IV (Fig. 11) and maximum anteroposterior and mediolateral diameter of the proximal epiphysis of metatarsal III (Fig. 12). None of the metacarpals nor metatarsals from GDC shows a statistically significant distinction compared with P. fossilis.
In summary, the trend observed shows that middle Middle Pleistocene P. fossilis are significantly distinguishable in proportions from Late Pleistocene European and North American lions and their extant African relatives. Generally, the studied specimens differ from Middle Pleistocene lions P. fossilis in total length but are similar to them in anteroposterior and mediolateral diameter of the proximal epiphyses, a feature especially prominent in Mtt III (Fig. 12).

Grotte de la Carrière Lions in the Context of Middle Pleistocene European Lions
As discussed in the biometrical comparisons section, the fossil remains of P. fossilis from Grotte de la Carrière reveal stout Fig. 10 a. Boxplots of select ratios of tarsal bones from Grotte de la Carrière, Middle to Late Pleistocene localities showing the chronological trend. b. bivariate graphs of total width against total length of astragalus (left) and calcaneus (right) and relatively large specimens, fitting within the size range and morphology of this species from coeval sites (middle Middle Pleistocene, MIS 11-9). The specimens from Sima de los Huesos from the Atapuerca complex (Iberian Peninsula), Vértesszȍlȍs 2 (Hungary), and Zá Hájovnou from Moravia (Czech Republic) are the similar to Grotte de la Carrière individuals and also proportionally larger than Petralona (Greece), Torre in Pietra (Italian Peninsula), Kent's Cavern (Great Britain), and Lunel-Viel and Orgnac (France) samples. In contrast, those individuals from the Polish localities of Biśnik and Wierzchowska, predominantly from steppe-tundra and boreal forest landscape, are especially large. The harsh and cold ecosystems in which they lived could explain their larger size following Bergmann's Rule (Bergmann 1847). Bergman's Rule proposes a heat balance explanation for the differences in body size depending on the latitudinal distribution of species (Bergmann 1847). In general, the gradient of body size increase with the latitude (in accordance with increasingly cold conditions). Not all species fit Bergman's Rule because there are other factors to take into consideration. In this case, the differences in the biomes such as the Mediterranean vegetation (e.g., Torre in Pietra, Sima de los Huesos, and Petralona) and steppe/tundra (Biśnik and Wierzchowska) or alpine boreal forest (Zá Hájovnou) may affect the biocenosis.
Traditionally in the paleontological literature, the steppe lion has been considered a predator adapted to wide variety of ecosystems (Hemmer 2004;Turner 2009). Bearing in mind that each biotic community is naturally adjusted according to its local climate and that the locomotor apparatus is Fig. 12 Boxplots of select ratios and measurements of Grotte de la Carrière metatarsal bones (marked in red) compared to Panthera fossilis, Panthera spelaea, Panthera atrox and values for extant relatives linked to predatory strategies (Moore and Biewener 2015), species adaptations to distinct biomes may reflect regional differences in the skeleton. For example, European terrestrial biomes usually encompassed grassland, forest, and tundra. Consequently, the prey spectrum available and hunting strategies are distinct for each one. In particular, for ambushing prey hunters wait for the prey and launch one quick strike (Moore and Biewener 2015), so this type of predatory behavior strategy has specific skeletal requirements. To illustrate, recent studies report that extant male lions use the savanna vegetation for ambushing prey instead of pursuit in open grasslands like female lions (Loarie et al. 2013).
Regarding the case of Biśnik Cave, the fossil remains were deposited on the cavity during the Odra Glaciation, when the typical steppe tundra dominated these habitats, a landscape dominated by wooly rhinoceros (Coelodonta antiquitatis) and steppe bison (Bison priscus). These conditions favored long proportions, especially in the metapodials, whereas the proportions of Grotte de la Carrière lions resemble those from the inhabitants of forest biomes. The presence of species well adapted to rocky environments such as the genus Hemitragus or the roe deer (Capreolus capreolus) (Demay et el. 2020 and references therein) suggest a mountainous forest, probably with steep cliffs. The mammal assemblage from MIS 9 layers of Caune de l' Arago (50 km northeast of Grotte de la Carrière), formed by Cervus elaphus, Bison priscus, Equus ferus, and Stephanorhinus (Moigne et al. 2006), spans a wider spectrum of species adapted to more open grasslands, the same conditions as in Moravian localities. To sum up, the lions from the abovementioned Polish localities and surrounding areas probably tended to be stouter and larger, conditioned by the surrounding biome.

Comparisons with the Earliest Lions from Middle Middle Pleistocene MIS 15-12
Following the size trend observed by Marciszak et al. (2019), the studied material from Grotte de la Carrière should be in general larger than major part of chronologically older lions from MIS 15-12. This implicit trend is fulfilled by most fossiliferous sites, except the northeastern France locality of Breccia de Château and the Central European sites of Moggaster and Poludniowa, in which they are larger and stouter. The studied specimens from Westbury and Caune de l' Arago share similar characteristics with GDC lions.
The intraspecific distinctions between the larger specimens from Central Europe could be a response to climatic dynamics and geographical settings, as has been seen with the coeval sites. Traditionally, climatic events strongly influenced the biogeographical distributions of mammalian species and influenced the faunal associations of regional biomes (Rodríguez et al. 2004). The earliest glaciation period (MIS 16-13) from the Middle Pleistocene occurred at the same time for all European regions (Head and Gibbard 2005). It is characterized by non-glacial cooling during cold stages (Candy et al. 2010(Candy et al. , 2014Lindner et al. 2013), succeeded by less interglacial warming compared to MIS 11-9 (Szymanek and Julien 2018). The Mid-Brunhes Transition (ca. 430 ka) marks the increase in the amplitude of the climate cycles to 100 kyr (Jansen et al. 1986;Elderfield et al. 2012). So, the strong glacial dynamics and the development of regular icesheet covering during cold stages started in MIS 12-10 ( Böse et al. 2012) with the arrival of the socalled Elster glaciation across northern Europe (Kahlke et al. 2011;Ehlers et al. 2011), also known as Mindel in the Alpine region and Anglian in Great Britain. Consequently, the increase in body size may be related to these climatic fluctuations, as an adaptation to the resulting new cold conditions and the latitudinal changes in the biomes (Rodríguez et al. 2004). Indeed, that could explain the coincidence among the large and stout lions' occurrence with the strong climatic events of MIS 12-10.
In Central and Northern Europe, the cold-adapted large mammals of the Mammuthus-Coelodonta steppe complex were established then for the first time (Kahlke et al. 2011;Szymanek and Julien 2018). Some studies have revealed that the increase in body size may be related to the expansion of this so-called 'Mammuthus-Coelodonta steppe' as an adaptation to the resulting new biome. The mammal assemblage of Mosbach and other localities from the Great European Plains suggest a sequence of changes in ecological conditions usually dominated by dry steppe-tundra landscapes occasionally interrupted by the extension of deciduous and mixed forest during warmer and wet climatic episodes (Kahlke et al. 2011). Instead, in southwestern Mediterranean regions, including southern France and the Great Britain site of Westbury, the climate during MIS 13-7 intervals was generally temperate (Szymanek and Julien, 2018;López-García et al. 2021), less cold than in Central Europe. That favored the expansion of deciduous areas. Precisely, the largest individuals from Moggaster and Poludniowa come from the steppe-tundra, areas in which extreme climatic conditions, prey availability (and thus the hunting strategies), and interspecific competition forced individuals to be stouter.
Unfortunately, all the factors that affect the body size structure of large mammals are difficult to analyze due to the lack of available specimens in the known fossiliferous sites.

Comparisons with Transitional Forms from Late Middle Pleistocene MIS 8-6
According to Marciszak and Stefaniak (2010), the forms between MIS 9-6 normally possesses intermediate characteristics, especially those around MIS 7-5e. The specimens from GDC presents some intermediate features, especially in the biometric dimensions the of metapodial bones and the proportions of m1, which resemble the ones of P. spelaea. Argant and Brugal (2017) describe the lion from Igue-de-Rameaux as a distinct species with intermediate features, namely Panthera intermedia, and include in its hypodigm individuals from the French localities of Abimes de la Fage (MIS 7, France), Romain la Roche (MIS 6), and Orgnac III (MIS 9). Compared to them, the Grotte de la Carrière sample has some morphometrical resemblances including the reduced talonid and the absence of a tiny groove at the top of the medial lingual protuberance of m1 and the proximodistally shorter metacarpals and metatarsals in the case of Romain la-Roche and Orgnac III specimens. However, the size of dental remains is proportionally larger than in the Abimes de la Fage and Romain la Roche samples and similar to large-sized specimens from Igue de Rameaux. The lion remains from the rest of the fossiliferous sites of Central-Eastern Europe MIS 8-6 are attributed to P. fossilis, whereas in the western France localities of Lherm, Santenay, and Azé I-3, the lions are referred to the P. spelaea group. This complexity in the characters of populations, considering the chronology of the studied sites, is expected. Notwithstanding, the lions from Lherm differ substantially from the rest by their larger dimensions, similar to the P. fossilis of the Middle Pleistocene, almost equal to Grotte de la Carrière lions. As seen above, geography is not the only factor that influences species variability. Considering the chronology of Lherm, those lions lived in glacial conditions. In this case, Bergman's rule should not be ruled out, for which the mammals in cold regions have the trend to be stouter and larger than the individuals of the same species living in warmer sites. Finally, it should not be forgotten that due to European regional differences, the replacement P. fossilis-P. spelaea probably occurred at different times (Marciszak et al. 2014).

Conclusions
On the basis of the morphological and morphometric analysis performed in the present study, the Grotte de la Carrière large felid remains are included in the hypodigm of the species Panthera fossilis (von Reichenau, 1906). The studied lion from the Réseau Lachambre karstic complex, with an estimated chronology corresponding to the MIS 9 interglacial, shares several dental diagnostic features and body proportions related to previously recorded Middle Pleistocene forms from the interval MIS 11-9. Nevertheless, some minor features more closely resemble later P. spelaea European forms from MIS 8-6. Such later forms have also been called 'intermediate forms' by previous authors, between the two species of the European lion lineage. The reported data can shed new light on the still scarcely recorded Northern Hemisphere lions and their size trends over the Pleistocene putatively related to climatic shifts and environmental parameters.
Acknowledgements The authors want to express their gratitude to Philippe Galant, Jean Louis-Perez, Michel Gomez and Jêrome Durbet for their help and availability during the field work at the Têt Valley and to Darin Croft, the associate editors and two anonymous reviewers for improve an early version of this manuscript.

Authors Contributions
All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Maria Prat-Vericat, Adrian Marciszak and Joan Madurell-Malapeira. The first draft of the manuscript was written by Maria Prat-Vericat and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Funding Open Access Funding provided by Universitat Autonoma de Barcelona. This work has been funded by the Direction Regionalle des Affaires Culturelles of the French Government (DRAC Occitaine) and the Agencia Estatal de Investigación-European Regional Development Fund of the European Union (CGL2017-82654-P, AEI/FEDER-UE) and the Generalitat de Catalunya (CERCA Program). J.M.-M. is member of the consolidated research group 2017 SGR 116 (AGAUR, Generalitat de Catalunya). LS is supported by the FI AGAUR fellowship (ref. 2020 FI_B1 00131) funded by the Secretaria d'Universitats i Recerca de la Generalitat de Catalunya and the European Social Fund.

Availability of Data and Material
The datasets generated during and/or analyzed during the current study are available in the Online Resource 1.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.