Functional anatomy of the forelimb in Promegantereon ogygia (Felidae, Machairodontinae, Smilodontini) from the Late Miocene of Spain and the origins of the sabre-toothed felid model

Abstract

We examine the functional anatomy of the forelimb in the primitive sabre-toothed cat Promegantereon ogygia in comparison with that of the extant pantherins, other felids and canids. The study reveals that this early machairodontine had already developed strong forelimbs and a short and robust thumb, a combination that probably allowed P. ogygia to exert relatively greater forces than extant pantherins. These features can be clearly related to the evolution of the sabre-toothed cat hunting method, in which the rapid killing of prey was achieved with a precise canine shear-bite to the throat. In this early sabre-toothed cat from the Late Miocene, the strong forelimbs and thumb were adapted to achieve the rapid immobilization of prey, thus decreasing the risk of injury and minimizing energy expenditure. We suggest that these were the major evolutionary pressures that led to the appearance of the sabre-toothed cat model from the primitive forms of the Middle Miocene, rather than the hunting of very large prey, although these adaptations reached their highest development in the more advanced sabre-toothed cats of the Plio-Pleistocene, such as Smilodon and Homotherium. Although having very different body proportions, these later animals developed such extremely powerful forelimbs that they were probably able to capture relatively larger prey than extant pantherins.

Introduction

The development of sabre-shaped upper canines has occurred in several groups of carnivorous mammals including marsupials of the family Thylacosmilidae (such as Thylacosmilus atrox, from the Pliocene of South America), paleogene creodonts (such as Machaeroides eothen, from the Middle Eocene of North America), and Carnivora of the families Nimravidae (such as Hoplophoneus mentalis, from the Late Eocene of North America or Eusmilus bidentatus, from the Oligocene of Europe), Barbourofelidae (such as Barbourofelis fricki, from the Late Miocene of North America), and Felidae, the so-called sabre-toothed cats. These species developed huge, laterally compressed and, in many cases, crenulated upper canines (Turner & Antón, 1997). This convergence has been classically explained as an adaptation to the hunting of large prey, because the presence of upper canines was always accompanied by the development of strong forelimbs and other morphological traits that resulted in very strongly built animals (Gonyea, 1976a; Akersten, 1985; Turner & Antón, 1997). However, the evolutionary mechanisms leading to the appearance of this recurrent predator model have never been studied in depth. An additional problem when attempting to understand these processes is that it is the most derived or crown taxa in each group of sabre-toothed mammals that are best known from the fossil record, whereas the basal taxa are usually unknown or known only on the basis of very fragmentary remains. The case of the sabre-toothed felids has been no exception, with early taxa such as Promegantereon ogygia traditionally known from very fragmentary fossils. The situation changed with the discovery in 1991 of the Cerro de los Batallones fossil sites in central Spain (Morales et al. 2000, 2004; Salesa, 2002; Antón et al. 2003, 2004a; Salesa et al. 2003, 2005), which have yielded almost complete skeletons of several individuals. With a very similar body size to extant, medium-sized pantherins, P. ogygia exhibits a number of derived characters in its skull, dentition and appendicular skeleton foreshadowing those seen in later machairodonts (Salesa, 2002; Salesa et al. 2005).

Because sabre-toothed cats were generally strongly built animals, it has been suggested that they specialized in taking larger prey than extant pantherins (Gonyea, 1976a; Emerson & Radinsky, 1980; Akersten, 1985; Rawn-Schatzinger, 1992; Turner & Antón, 1997). But this can only have been possible in the case of the later machairodonts, such as the large Smilodon or Homotherium. The size of P. ogygia, similar to that of a puma, Puma concolor, makes it more likely that the origin of the machairodont adaptations was related to the advantage of reducing the time of prey immobilization, rather than to changes in prey size (Salesa et al. 2005, 2006).

The machairodont killing technique is thought to have involved a well-aimed bite to the throat of an immobilized prey so as to damage blood vessels and trachea and thus produce a rapid death (Antón & Galobart, 1999; Antón et al. 2004a, 2005). Once the prey was subdued, the sabre-toothed cat used its long and compressed upper canines to cut the throat of prey, by way of a head-depression movement in which the mandible served as an anchor (Turner & Antón, 1997; Antón et al. 2004b). This hunting method is radically different from that of the large felines, in which the killing of large prey is achieved by a bite lasting several minutes, which suffocates the prey. This bite is applied to the throat or, if the prey is very large, to the muzzle; if the prey is small enough, the kill is achieved by way of a nape bite (Turner & Antón, 1997). In spite of the differences between the feline and sabre-toothed bites, both can be derived reasonably from an ancestral felid trend to bite at the throat of prey while controlling it with the forelimbs, a trait that would distinguish cats from other carnivores that tend to attack prey by simply biting at it (Leyhausen, 1979).

The sabre-toothed cat hunting method would have reduced the energetic cost of hunting, while providing more precision and a greater margin of safety during the kill, reducing the risk of canine breakage due to the struggles of a downed animal (Salesa et al. 2005). To accomplish this kind of attack, it was essential to possess very strong forelimbs to immobilize the prey. In this paper we analyze the anatomy of the forelimb of P. ogygia in comparison with the range of variation seen in a variety of extant and fossil species of Felidae, and show that several traits indicative of such increased strength are clearly evident in the material now available for study.

Materials and methods

The fossils of P. ogygia analyzed in this study belong to the extensive sample from the fossil carnivore trap of Batallones-1 (Late Miocene, Vallesian, MN 10, Madrid, Spain), housed in the collections of the Museo Nacional de Ciencias Naturales-CSIC (Madrid, Spain). This large sample has allowed us to study all the bones of the skeleton of this early sabre-toothed cat, and to discern the degree of variability in the features in a way that is essential to tackling biomechanical inferences. Thus the morphological features that we discuss are similarly developed in all the available specimens. We did not include the magnum and unciform bones in our study due to the great similarity that we found in these bones across the range of species analyzed. Comparisons with extant carnivores were made using the collections of the Museo Anatómico de la Universidad de Valladolid (Spain), Museum National d’Histoire naturelle (Paris) and Museo Nacional de Ciencias Naturales-CSIC (Madrid, Spain), which provided complete skeletons of the felines Leopardus wiedii (margay), Acinonyx jubatus (cheetah), Panthera leo (lion), Panthera tigris (tiger), Panthera pardus (leopard), Panthera onca (jaguar) and Puma concolor (puma), the viverrids Genetta genetta (genet) and Cryptoprocta ferox (Malagasy fossa), the ailurid Ailurus fulgens (red panda), and the canids Canis lupus (wolf) and Lycaon pictus (African wild dog). Comparisons with other fossil Felidae, such as Panthera atrox (American lion), Smilodon fatalis, Smilodon gracilis, Pseudaelurus quadridentatus, Megantereon cultridens, Proailurus lemanensis and Homotherium latidens were made using published data from several authors.

Measurements (Supplementary Tables S1–S8) were taken with digital calipers, as shown in Fig. 1. The anatomical position and articulation between the studied bones are showed in Fig. 2. The quantitative data were analyzed using one-way anova of the statistics software package spss 11.0. We also calculate the post-hoc tests of each anova, using the Tukey test, to find which means were significantly different from one another. Morphological differences observed in the study, but less adequate for quantification, are the subject of qualitative observations. Finally, the functional and biomechanical implications of all observed differences are discussed.

Fig. 1.

Measurements for each skeletal element (Third Phalanx: ML, maximum length; H, height; Pisiform: TW, tubercle width; ML, total length; Scapholunar: ASW, proximal surface width; ASL, proximal surface length; DG, major diagonal of the articular proximal surface; PTL, posterior tubercle length; Pyramidal: MLW, medio-lateral width (not represented); PDH, proximo-distal height; Metacarpal I and Second Phalanx of Dew Claw: PEL, proximal epiphysis length; PEW, proximal epiphysis width; ML, maximum length; Trapezoid: MLW, medio-lateral width; PDH, proximo-distal height; Trapezium: APL, antero-posterior length; PDH, proximo-distal height).

Fig. 2.

Illustration of the left forelimb of Panthera leo in antero-lateral (A) and antero-medial (B) views, showing the articulations between the different bones: hu, humerus; mg, magnum; McI, first metacarpal; McII, second metacarpal; McIII, third metacarpal; McIV, fourth metacarpal; McV, fifth metacarpal; ps, pisiform; py, pyramidal; ra, radius; sc, scapholunar; th, second phalanx of dewclaw; trp, trapezoid; tr, trapeze; ul, ulnae; un, unciform (modified from Ellenberger et al. 1956).

Statistical results

The results of the different post-hoc tests of each one-way anova calculated for the quantitative data are shown in Tables 1–8. The bivariate plots of the analyzed measurements are shown in Figs 3–6, and the regression equations are shown in the Supplementary Table 9. The most remarkable results were as follows:

Table 1.

Results of the Tukey post-hoc tests of the anova for the variable TW/ML of the pisiform; values in bold indicate the existence of significant differences between the species.

Species	Pogy	Sfat	Ponc	Ppar	Ptig	Ajub	Pleo	Pcon	Patr
Pogy		0.166355	0.999966	0.924130	1.000000	0.000264	0.413335	0.000799	0.163506
Sfat	0.166355		0.591501	0.019153	0.537241	0.000137	0.002983	0.000140	0.001035
Ponc	0.999966	0.591501		0.883402	1.000000	0.000448	0.459042	0.002611	0.223715
Ppar	0.924130	0.019153	0.883402		0.980670	0.002250	0.990516	0.020151	0.855360
Ptig	1.000000	0.537241	1.000000	0.980670		0.001411	0.744218	0.012142	0.476081
Ajub	0.000264	0.000137	0.000448	0.002250	0.001411		0.013888	0.860090	0.046459
Pleo	0.413335	0.002983	0.459042	0.990516	0.744218	0.013888		0.148704	0.999602
Pcon	0.000799	0.000140	0.002611	0.020151	0.012142	0.860090	0.148704		0.418982
Patr	0.163506	0.001035	0.223715	0.855360	0.476081	0.046459	0.999602	0.418982

Tukey HSD Test. Variable: TW/ML (PISIFORM). Probabilities for post-hoc tests. Error: Between MS = 0.00045; df = 39.000.

Pogy, Promegantereon ogygia; Sfat, Smilodon fatalis; Ppar, Panthera pardus; Ponc, Panthera onca; Pcon, Puma concolor; Ajub, Acinonyx jubatus; Ptig, Panthera tigris; Patr, Panthera atrox; Pleo, Panthera leo.

Table 8.

Results of the Tukey post-hoc tests of the anova for the variable APL/PDH of the trapezium; values in bold indicate the existence of significant differences between the species.

Species	Pogy	Sfat	Ppar	Ponc	Pcon	Ajub	Ptig	Patr	Pleo
Pogy		0.997048	0.000139	0.000139	0.000139	0.000182	0.000139	0.025827	0.000139
Sfat	0.997048		0.000156	0.000139	0.000273	0.008799	0.000140	0.531555	0.000204
Ppar	0.000139	0.000156		0.088487	0.999984	1.000000	0.542632	0.058895	1.000000
Ponc	0.000139	0.000139	0.088487		0.077075	0.414178	0.996289	0.000369	0.206555
Pcon	0.000139	0.000273	0.999984	0.077075		1.000000	0.453916	0.174379	0.999973
Ajub	0.000182	0.008799	1.000000	0.414178	1.000000		0.825757	0.525122	0.999999
Ptig	0.000139	0.000140	0.542632	0.996289	0.453916	0.825757		0.003096	0.727840
Patr	0.025827	0.531555	0.058895	0.000369	0.174379	0.525122	0.003096		0.098028
Pleo	0.000139	0.000204	1.000000	0.206555	0.999973	0,999999	0.727840	0.098028

Tukey HSD Test. Variable: APL/PDH (TRAPEZIUM). Probabilities for post-hoc tests. Error: Between MS = 0.00216; df = 55.000.

Pogy, Promegantereon ogygia; Sfat, Smilodon fatalis; Ppar, Panthera pardus; Ponc, Panthera onca; Pcon, Puma concolor; Ajub, Acinonyx jubatus; Ptig, Panthera tigris; Patr, Panthera atrox; Pleo, Panthera leo.

Fig. 3.

Bivariate plots of (A) TW (tubercle width) and ML (maximum length) of the pisiform and (B) MLW (medio-lateral width) and PDH (proximo-distal height) of the pyramidal, in different species of fossil and extant felids, showing the regression line for each species.

Fig. 6.

Bivariate plot of APL (antero-posterior length) and PDH (proximo-distal height) of the trapezium in different species of fossil and extant felids, showing the regression line for each species.

Table 9.

Relative weight of the forearm muscles in some species of Felidae and Canidae. The numbers indicate the percentage of each muscle weight over the total weight of forearm muscles (modified from Gambaryan, 1974).

Muscle	Acinonyx jubatus	Panthera pardus	Uncia uncia	Panthera onca	Puma concolor	Panthera leo	Lynx lynx	Felis chaus	Canis lupus	Lycaon pictus
Triceps brachii (long branch)	10.5	7.0	7.4	6.6	9.0	9.1	9.2	8.8	10.3	9.9
Triceps brachii (lateral branch)	3.5	3.1	4.0	2.3	2.9	3.4	3.6	3.7	3.4	2.7
Triceps brachii (medial branch)	1.8	3.2	3.3	3.1	2.6	2.2	3.4	4.1	3.5	4.8
Flexor digitorum profundus	2.6	3.2	3.9	3.0	3.5	3.5	3.4	4.9	2.6	2.3
Abductor pollicis	0.2	0.5	1.0	0.7	0.7	0.7	0.3	0.7	0.2	0.2
Ulnaris medial	0.7	1.2	1.5	1.6	1.0	1.4	1.2	1.2	0.2	0.2
Ulnaris lateral	0.4	0.7	0.9	0.7	0.7	0.9	0.6	1.0	0.5	0.7
Biceps brachii	2.8	2.6	2.8	3.4	2.6	2.9	2.5	3.8	2.1	2.0
Brachiocephalicus	4.3	9.4	5.6	8.3	7.8	8.5	6.8	5.7	5.2	4.7

Fig. 5.

Bivariate plots of (A) ML (maximum length) and PEW (proximal epiphysis width) of the first phalanx of the dewclaw and (B) ML (maximum length) and PEW (proximal epiphysis width) of the metacarpal I in different species of fossil and extant felids, showing the regression line for each species.

1. The shape of the posterior tubercle of the pisiform, in relation to the antero-posterior length of this bone, shows significant differences between P. ogygia and the most cursorial extant felids, P. concolor and A. jubatus, whereas S. fatalis shows significant differences from P. pardus, P. atrox, P. pardus, A. jubatus and P. concolor. This is demonstrated by a one-way anova for an index between the width of this tubercle (TW) and the maximum length of pisiform (ML) (Table 1).

2. The width of the anterior face of the pyramidal shows significant differences between P. ogygia and the felids P. pardus, P. leo, P. atrox, P. concolor and A. jubatus. Smilodon fatalis only shows significant differences from A. jubatus. This is demonstrated by a one-way anova for an index between the medio-lateral width of the pyramidal (MLW) and its proximo-distal height (PDH) (Table 2).

3. The postero-medial tubercle of the scapholunar of P. ogygia is relatively longer than that of the pantherins P. concolor and A. jubatus, as shown by a one-way anova calculated for the index between the length of this tubercle (PTL) and the proximal area (PA) of the scapholunar, calculated by multiplying the length (ASL) and width (ASW) of the articular surface of this bone (Table 3).

4. The metacarpal I of P. ogygia is relatively shorter and stronger than that of the pantherins P. concolor and A. jubatus, as shown by a one-way anova for an index between the proximal width of the metacarpal I (PEW) and its maximum length (ML) (Table 4).

5. In P. ogygia, the pollex, or dewclaw, shows a significantly more robust first phalanx and larger second or ungual (claw) phalanx than those of the pantherins. (Note that, as in humans, the first digit of the manus has only two phalanges, including the ungual phalanx, whereas digits 2–5 have three.) This is shown by an anova for an index between the proximal width of the first phalanx of the pollex (PEW) and its maximum length (ML) (Table 5); another one-way anova was calculated for an index between the lateral area of the second phalanx of the first and the third phalanx of the fifth digits; these lateral areas (1 LA for digit 1, and 5 LA for digit 5) are the product of the maximum length (1 ML for digit 1, and 5 ML for digit 5) and height (1 H for digit 1, and 5 H for digit 5) of the ungual phalanges, and are used as an estimation of the relative size of the second phalanx of the pollex (Table 6).

6. Promegantereon ogygia differs in the proportions of the trapezoid from P. pardus, P. onca, P. atrox, and P. leo, and also from S. fatalis. These differences are significant, as shown by a one-way anova for an index between the medio-lateral width of the trapezoid (MLW) and its proximo-distal height (PDH) (Table 7).

7. Concerning the trapezium, there are significant differences in its proximo-distal height between a group comprising P. concolor, A. jubatus and the pantherins, and another group formed by the sabre-toothed cats S. fatalis and P. ogygia, the latter having lower trapezia. This is shown when an index between the antero-posterior length (APL) and the proximo-distal height (PDH) is analyzed by a one-way anova (Table 8).

Table 2.

Results of the Tukey post-hoc tests of the anova for the variable AML/PDH of the pyramidal; values in bold indicate the existence of significant differences between the species.

Species	Pogy	Sfat	Ppar	Ponc	Pcon	Ajub	Ptig	Pleo	Patr
Pogy		0.097806	0.002084	0.546651	0.000749	0.000139	0.276888	0.001437	0.000141
Sfat	0.097806		1.000000	0.973721	0.912733	0.010176	0.998715	0.999109	0.178277
Ppar	0.002084	1.000000		0.776419	0.820253	0.001941	0.967095	0.998649	0.040774
Ponc	0.546651	0.973721	0.776419		0.216130	0.000460	0.999965	0.529267	0.006156
Pcon	0.000749	0.912733	0.820253	0.216130		0.110863	0.433846	0.991010	0.855511
Ajub	0.000139	0.010176	0.001941	0.000460	0.110863		0.001081	0.010780	0.723928
Ptig	0.276888	0.998715	0.967095	0.999965	0.433846	0.001081		0.820341	0.019239
Pleo	0.001437	0.999109	0.998649	0.529267	0.991010	0.010780	0.820341		0.227293
Patr	0.000141	0.178277	0.040774	0.006156	0.855511	0.723928	0.019239	0.227293

Tukey HSD Test. Variable: AML/PDH (PYRAMIDAL). Probabilities for post-hoc tests. Error: Between MS = 0.00135; df = 37.000.

Pogy, Promegantereon ogygia; Sfat, Smilodon fatalis; Ppar, Panthera pardus; Ponc, Panthera onca; Pcon, Puma concolor; Ajub, Acinonyx jubatus; Ptig, Panthera tigris; Patr, Panthera atrox; Pleo, Panthera leo.

Table 3.

Results of the Tukey post-hoc tests of the anova for the variable PTL/DG of the scapholunar; values in bold indicate the existence of significant differences between the species.

Species	Pogy	Ppar	Pleo	Ptig	Ponc	Ajub	Pcon
Pogy		0.000131	0.000131	0.000133	0.000131	0.000144	0.000131
Ppar	0.000131		0.571075	0.201562	0.526479	0.992017	0.660368
Pleo	0.000131	0.571075		0.980798	0.999962	0.999454	0.999995
Ptig	0.000133	0.201562	0.980798		0.998717	0.944887	0.944865
Ponc	0.000131	0.526479	0.999962	0.998717		0.995965	0.999260
Ajub	0.000144	0.992017	0.999454	0.944887	0.995965		0.999929
Pcon	0.000131	0.660368	0.999995	0.944865	0.999260	0.999929

Tukey HSD Test. Variable: PTL/DG (SCAPHOLUNAR). Probabilities for post-hoc tests. Error: Between MS = 0.00007; df = 39.000.

Pogy, Promegantereon ogygia; Sfat, Smilodon fatalis; Ppar, Panthera pardus; Ponc, Panthera onca; Pcon, Puma concolor; Ajub, Acinonyx jubatus; Ptig, Panthera tigris; Patr, Panthera atrox; Pleo, Panthera leo.

Table 4.

Results of the Tukey post-hoc tests of the anova for the variable PEW/ML of the metacarpal I; values in bold indicate the existence of significant differences between the species.

Species	Pogy	Ppar	Pleo	Ptig	Ponc	Pcon	Ajub
Pogy		0.000133	0.000133	0.000133	0.000133	0.000133	0.000133
Ppar	0.000133		0.999987	0.427640	0.956070	0.999799	0.167609
Pleo	0.000133	0.999987		0.705215	0.992232	1.000000	0.175144
Ptig	0.000133	0.427640	0.705215		0.989482	0.772291	0.011778
Ponc	0.000133	0.956070	0.992232	0.989482		0.996869	0.075682
Pcon	0.000133	0.999799	1.000000	0.772291	0.996869		0.146959
Ajub	0.000133	0.167609	0.175144	0.011778	0.075682	0.146959

Tukey HSD Test. Variable: PEW/ML (Mc I). Probabilities for post-hoc tests.

Error: Between MS = 0.00071; df = 59.000.

Pogy, Promegantereon ogygia; Sfat, Smilodon fatalis; Ppar, Panthera pardus; Ponc, Panthera onca; Pcon, Puma concolor; Ajub, Acinonyx jubatus; Ptig, Panthera tigris; Patr, Panthera atrox; Pleo, Panthera leo.

Table 5.

Results of the Tukey post-hoc tests of the anova for the variable PEW/ML of the first phalanx of thumb; values in bold indicate the existence of significant differences between the species.

Species	Pogy	Ponc	Ppar	Ptig	Pleo	Ajub	Pcon
Pogy		0.000131	0.949913	0.001448	0.002042	0.000131	0.000242
Ponc	0.000131		0.000195	0.498342	0.168008	0.764808	0.959640
Ppar	0.949913	0.000195		0.035949	0.075830	0.000140	0.003894
Ptig	0.001448	0.498342	0.035949		0.998062	0.052508	0.976294
Pleo	0.002042	0.168008	0.075830	0.998062		0.012204	0.780533
Ajub	0.000131	0.764808	0.000140	0.052508	0.012204		0.283521
Pcon	0.000242	0.959640	0.003894	0.976294	0.780533	0.283521

Tukey HSD Test. Variable: PEW/ML (FIRST PHALANX THUMB). Probabilities for post-hoc tests. Error: Between MS = 0.00066; df = 64.000.

Pogy, Promegantereon ogygia; Sfat, Smilodon fatalis; Ppar, Panthera pardus; Ponc, Panthera onca; Pcon, Puma concolor; Ajub, Acinonyx jubatus; Ptig, Panthera tigris; Patr, Panthera atrox; Pleo, Panthera leo.

Table 6.

Results of the Tukey post-hoc tests of the anova for the variable 1LA/5LA of the third phalanx; values in bold indicate the existence of significant differences between the species.

Species	Pogy	Ponc	Ppar	Pcon	Ptig	Pleo
Pogy		0.000130	0.000130	0.000159	0.000140	0.000462
Ponc	0.000130		0.988556	0.806939	0.051156	0.000590
Ppar	0.000130	0.988556		0.433483	0.001692	0.000131
Pcon	0.000159	0.806939	0.433483		0.855126	0.159141
Ptig	0.000140	0.051156	0.001692	0.855126		0.494782
Pleo	0.000462	0.000590	0.000131	0.159141	0.494782

Tukey HSD Test. Variable: 1LA/5LA (THIRD PHALANX). Probabilities for post-hoc tests. Error: Between MS = 0.02086; df = 27.000.

Pogy, Promegantereon ogygia; Sfat, Smilodon fatalis; Ppar, Panthera pardus; Ponc, Panthera onca; Pcon, Puma concolor; Ajub, Acinonyx jubatus; Ptig, Panthera tigris; Patr, Panthera atrox; Pleo, Panthera leo.

Table 7.

Results of the Tukey post-hoc tests of the anova for the variable MLW/PDH of the trapezoid; values in bold indicate the existence of significant differences between the species.

Species	Pogy	Sfat	Ppar	Ponc	Pcon	Ajub	Ptig	Patr	Pleo
Pogy		0.041297	0.000142	0.323583	0.000244	0.230972	0.979250	0.000546	0.000268
Sfat	0.041297		0.998685	0.998694	0.995869	1.000000	0.751171	0.974079	0.999644
Ppar	0.000142	0.998685		0.829811	1.000000	0.999984	0.154229	0.999478	1.000000
Ponc	0.323583	0.998694	0.829811		0.809114	0.998731	0.985534	0.691244	0.903260
Pcon	0.000244	0.995869	1.000000	0.809114		0.999864	0.180408	0.999989	0.999998
Ajub	0.230972	1.000000	0.999984	0.998731	0.999864		0.849562	0.997519	0.999996
Ptig	0.979250	0.751171	0.154229	0.985534	0.180408	0.849562		0.143929	0.256578
Patr	0.000546	0.974079	0.999478	0.691244	0.999989	0.997519	0.143929	0.143929	0.999321
Pleo	0.000268	0.999644	1.000000	0.903260	0.999998	0.999996	0.256578	0.999321

Tukey HSD Test. Variable: MLW/PDH (TRAPEZOID). Probabilities for post-hoc tests. Error: Between MS=0.01910; df = 62.000.

Pogy, Promegantereon ogygia; Sfat, Smilodon fatalis; Ppar, Panthera pardus; Ponc, Panthera onca; Pcon, Puma concolor; Ajub, Acinonyx jubatus; Ptig, Panthera tigris; Patr, Panthera atrox; Pleo, Panthera leo.

Morphological observations on the forelimb anatomy of P. ogygia

Promegantereon ogygia was found to differ from other felids in several morphological features of the forelimb bones. The scapula of P. ogygia has a longer coracoid process than that of pantherin cats, in addition to several differences in detail of muscle and ligament attachment areas. The humerus differs in aspects related to the attachment areas of the forearm flexor muscles in the distal epiphysis of the bone, with a greater projection of the medial border. The ulna differs in the relative development of the olecranon tubercles related to the attachment areas of muscles anconeus lateralis and medial branch of the triceps brachii; the magnum resembles that of S. fatalis and L. wiedii in lacking an articulation facet for metacarpal II, unlike the case in H. latidens and the pantherins.

Scapula

The scapula of P. ogygia has a very similar overall morphology to that of the pantherin cats. However, it has a coracoid process longer than that of the latter, and a supraglenoid tubercle with a smaller facet for the insertion of the muscle biceps brachii (Fig. 7). The coracoid process is the origin area of the muscle coracobrachialis, whose tendon covers both the process and the coraco-humeral and coraco-acromial ligaments (Taylor & Weber, 1951). In pantherin cats the surface below this process is smooth, whereas in P. ogygia there is a very well developed fossa, which increases the attachment area for the coraco-humeral and coraco-acromial ligaments.

Fig. 7.

Anterior view of the left scapula of Panthera pardus (A) and Promegantereon ogygia from Batallones-1 (B). The arrows indicate the attachment area for the muscle biceps brachii.

Humerus

The proportions and morphology of the humerus of P. ogygia are very similar to those of a pantherin cat. There are some differences on the medial border of the distal epiphysis, related to the attachment areas of several flexor muscles of the forearm. Basically, this area projects much more medially in P. ogygia than in any other of the pantherins. This morphology is different from that observed in the humerus of other, more derived sabre-toothed cats, such as S. fatalis or Homotherium serum. Among these forms the humerus also shows strong differences in robustness, proportions, and morphology of the distal epiphysis. Thus, whereas S. fatalis had a very robust humerus, with a strong medial projection of the distal epiphysis (Merriam & Stock, 1932; Gonyea, 1976a), the humerus in H. serum is slender and with a distal epiphysis showing a moderate medial projection (Rawn-Schatzinger, 1992; Antón et al. 2005).

Radius

The radius of P. ogygia resembles that of a pantherin cat both in general proportions and in detailed morphology, suggesting similarity in function.

Ulna

The olecranon process of the ulna displays a pair of tubercles that correspond to the attachment areas for several muscles involved in extension of the forearm, mainly the anconeus lateralis, which attaches on the lateral tubercle, and the medial branch of the triceps brachii, which attaches on the medial tubercle (Gonyea, 1978; Barone, 2000). As has been pointed out (Hopwood, 1945; Ginsburg, 1961), these tubercles show great differences in size and shape within the Carnivora, and even among the Felinae. In the cheetah (A. jubatus), as well as in canids, the lateral tubercle has less proximo-distal development than the medial one, whereas in other felines either both have similar development or the lateral tubercle surpasses the level of the medial one (Fig. 8). Acinonyx jubatus has a long branch of the triceps brachii proportionally larger than the rest of felines, whereas the medial branch of triceps brachii is relatively smaller (Gambaryan, 1974). In the case of the medial branch, the attachment facet is much smaller in the felines than in P. ogygia, whereas the attachment area for the long branch has a very similar morphology.

Fig. 8.

Anterior view of the right ulna of (A) Panthera pardus, (B) Acinonyx jubatus, and (C) Promegantereon ogygia from Batallones-1.

The ulna of P. ogygia displays well-developed olecranon tubercles, the medial one having a greater proximo-distal height than the lateral one, as in A. jubatus (Fig. 8). Although rare among extant felids, this pattern has been observed by us in the fossil felids P. lemanensis (Later Oligocene-Early Miocene of Europe), P. quadridentatus (Middle Miocene of Europe), M. cultridens (Pliocene-Early Pleistocene of Eurasia, Africa and North America), and S. gracilis (Pliocene-Early Pleistocene of North America). However, there are still differences within this rare pattern. The cheetah has a very narrow, strongly reduced medial tubercle, whereas in the four fossil felids it is wide and robust. This would be related to the fact that the medial branch of the muscle triceps brachii, which inserts on the medial tubercle, is much reduced in A. jubatus in relation to other large felines (Table 9). Besides this, the attachment facet for the medial branch of the triceps brachii in P. ogygia is a large and marked facet, very different from the small one seen in the felines.

Pisiform

The posterior tubercle of the pisiform constitutes the attachment area of the muscles ulnaris medialis (or flexor carpi ulnaris) and ulnaris lateralis (or extensor carpi ulnaris) (Davis, 1964; Barone, 2000). The shape of this tubercle shows significant differences between two groups of felids that could be respectively considered cursorial and less cursorial. The primitive morphology, observed in G. genetta, corresponds to an ellipsoidal, clearly proximo-distally compressed posterior tubercle, and it appears in S. fatalis, P. ogygia, P. onca, P. pardus, P. tigris and P. concolor. On the other hand, those cursorial felids, such as P. leo, P. atrox, H. latidens and A. jubatus, have rounded tubercles, very similar to those of the Canidae.

Pyramidal

The width of the anterior face of the pyramidal of the sabre-toothed cats P. ogygia and S. fatalis is relatively narrower than that of pantherins, and very similar to the model present in G. genetta.

Scapholunar and metacarpal I

The postero-medial tubercle of the scapholunar is relatively longer in P. ogygia than in any of the pantherins. In addition, this tubercle in P. ogygia and S. fatalis has a proximal orientation, instead of the distal orientation typical of pantherins and G. genetta. It is remarkable that those two species have such a different tubercle morphology from that of either cursorial or less cursorial felines. Even when the scapholunar of P. ogygia is compared with that of a similar-sized pantherin, such as P. pardus, the differences in the length and orientation of the tubercle are clearly evident (Fig. 9).

Fig. 9.

Posterior view of the left scapholunar of Panthera pardus (A) and Promegantereon ogygia from Batallones-1 (B) showing the different shape and orientation of the posterior tubercle.

Metacarpals II–V and phalanges

The overall morphology of the metacarpals II, III, IV and V, and the phalanges of P. ogygia is similar to that of the extant pantherins. Nevertheless, there are some evident differences in the relative size of the first digit, such as the presence of a significantly more robust first phalanx and a larger second phalanx in P. ogygia.

Magnum

In P. ogygia the medial face of the magnum lacks an articulation facet for metacarpal II, as in S. fatalis and L. wiedii, whereas the pantherins and H. latidens have this facet (Fig. 10). The absence of this facet indicates that the magnum only articulates medially with the trapezoid, whereas its presence is due to an articulation with the trapezoid and metacarpal II (Salesa, 2002). In the first group, when metacarpal II and III articulate, their proximal border is situated at the same level, so that metacarpal II does not overlap metacarpal III, and the magnum does not make contact with metacarpal II. Primitive viverrids, such as G. genetta, show this condition, so we considerer this absence to be a plesiomorphy.

Fig. 10.

Medial view of the right magnum of Panthera pardus (A) and Promegantereon ogygia from Batallones-1 (B) showing the articulation facets for metacarpal III (1), trapezoid (2) and metacarpal II (3).

Trapezoid

The trapezoid shows different proportions between pantherins and the sabre-toothed cats P. ogygia and S. fatalis. In the two latter species, the trapezoid has an ellipsoid shape in anterior view, lengthened in the medio-lateral direction, whereas in pantherins the shape is more or less rounded, because the bone is higher proximo-distally than in the two sabre-toothed cats.

Trapezium

Despite being a small bone, the trapezium is an important articulation joint for metacarpal I. In relation to the proximo-distal height of this bone, there are significant differences between felines and the smilodontins S. fatalis and P. ogygia, with the latter having lower trapezes. A proximo-distally low trapezium is also present in G. genetta, which indicates that this morphology is plesiomorphic for machairodontines and viverrids, with the felines showing the derived state.

Functional implications of the anatomy of P. ogygia

Scapula

The observed differences in the development of the coracoid process are probably related to the stresses suffered by the humero-scapular articulation, which would be higher in P. ogygia than in the pantherins; this increased stress would require stronger ligaments in the sabre-toothed cat, with a larger insertion area. The lengthening of the coracoid process of P. ogygia also produces an increase in the available area for the insertion of the muscle coracobrachialis, which would be larger than that of a pantherin of similar size and is also related to the presence of stronger shoulder ligaments.

The development of the attachment area for the muscle biceps brachii on the scapula also raises some interesting questions. In some groups of mammals, such as primates (Kimura & Takai, 1970) and some ursids and procyonids (Windle & Parsons, 1897) this muscle has a single proximal branch, composed of a strong tendon attached on the cranial margin of the supraglenoid tubercle, on a well-developed and large facet. In contrast, the distal attachment of the muscle is divided into two branches from its distal half (Taylor & Weber, 1951; Barone, 1967, 2000). In P. ogygia and other machairodontines, such as S. fatalis, the attachment facet for this muscle is smaller than that of G. genetta and the pantherins, and restricted to the medial apex of the anterior surface of the supraglenoid tubercle. In the pantherins the attachment area occupies the whole anterior face of the supraglenoid tubercle. This state, which is present in primitive Viverridae, is thus considered plesiomorphic for the Felidae. What do these differences in the development of scapular structures imply? When the muscle biceps brachii has two proximal branches, the long one is attached just over the glenoid cavity and the short one, which is absent in extant Felinae (Windle & Parsons, 1897; Barone, 2000), inserts on the tip of the coracoid process, just beside the tendon of the muscle coracobrachialis (Davis, 1964; Cuenca, 1995; Barone, 2000). Given the great development of the coracoid process in P. ogygia and other sabre-toothed cats, it might be tempting to propose that the short branch of the muscle biceps brachii was present in this group. However, given the absence of this short branch in extant felids, it is more reasonable to assume that the large coracoid process of sabre-toothed cats is related to the presence of a much stronger muscle coracobrachialis. Contraction of the muscle coracobrachialis produces adduction and pronation of the forearm, whereas contraction of muscle biceps brachii produces flexion and supination (Barone, 2000). Felids use forelimbs to capture and to hold prey, and for that reason these muscular actions are well developed in this group (Gonyea, 1976b; Akersten, 1985; Anyonge, 1996; Bryant et al. 1996). If machairodontines had to immobilize their prey completely before applying the canine shear-bite as we suggest, then an increase in the efficiency in these actions would reduce the time necessary to complete this action, making hunting safer for the predator.

Humerus

The greater development of the medial border of the distal epiphysis of the humerus in P. ogygia in relation to pantherins could be related to the presence of stronger flexor muscles in the former. Later sabre-toothed cats, such as S. fatalis and H. latidens, developed a very derived humerus and radius, although in very different ways – short and very robust in the former, and longer and slender in the latter – reflecting the disparity of biomechanical adaptations of the terminal lineage of each group of sabre-toothed cats (Gonyea, 1976a; Van Valkenburgh, 1987; Rawn-Schatzinger, 1992; Anyonge, 1996; Turner & Antón, 1997).

Ulna

The observed differences in the morphology of the tubercles of the ulna have interesting biomechanical implications. In felids, the normal quadrupedal standing posture implies that the humerus–radius articulation is moderately flexed, and, as a consequence of this, the fibres of the triceps brachii are oriented at a nearly perpendicular angle relative to the main axis of the forearm. In these conditions, and for a similar length of the olecranon, an increase in the proximo-distal development of the medial tubercle produces an increase in the length of the muscle. In the same way, an increase in the distance between the rotation point of the forearm (the elbow) and the attachment point of the medial branch of the muscle triceps brachii, increases the effort arm of this muscle. Following the basic principles of leverage, this configuration also reduces the force that this muscle has to produce to extend the forearm. Such lengthening of the muscle also produces an increase in the length of the fibres, which augments the velocity of contraction (Eckert et al. 1990). Because of this, felids with a medial tubercle that is relatively higher proximo-distally should have the longest medial branch of the triceps brachii, although this does not imply a relatively heavier muscle. In medium-sized pantherins, such as P. onca or P. pardus, the medial branch of the triceps brachii is about 3.2% of the total weight of the forearm muscles, almost double that in A. jubatus (Table 9). It is remarkable that in the canids such as C. lupus or L. pictus, which are cursorial animals like the cheetah, the relative weight of the medial branch of the triceps brachii is even higher than in the pantherins. This difference between A. jubatus and the pantherins is probably not due to a reduction in the relative size of the medial branch of the triceps brachii, but to the presence of an exceptionally large long branch of this muscle, which is 13–39% heavier than in any of the pantherins (Table 9). In this case, the huge increase in the relative size of the long branch of the triceps brachii in A. jubatus would have produced a decrease in the relative importance of the medial branch within the muscles of the forearm. This hypothesis seems more reasonable when analyzing other sub-cursorial felines, such as P. leo or P. concolor, which show similar although less marked differences between the three branches of the triceps brachii (Table 9). This increase in the relative size of the long branch of the triceps brachii, seen also in canids, is probably related to the development of cursorial abilities, as Gambaryan (1974) pointed out.

If the long branch of the triceps brachii is longer in A. jubatus than in the pantherins, but its relative weight is smaller, this is probably because its function in the forearm extension is to increase the velocity of contraction, something advantageous for a high-speed runner. Obviously, the relative weight of the muscles of P. ogygia is unknown, but we can compare the medial tubercles of the latter with those of A. jubatus. While the medial tubercle has a higher proximo-distal development than the lateral one in both species, in A. jubatus the medial tubercle has a narrow and smooth attachment area for the medial branch of the triceps brachii, whereas in P. ogygia this medial tubercle is wide and very rough, and the attachment facet for the medial branch is much larger (Fig. 11).

Fig. 11.

Detailed views of the olecranon of Panthera pardus (A,C) and Promegantereon ogygia from Batallones-1 (B,D) in medial (A,B) and lateral (C,D) views, showing the differences in the development of the attachment areas of the muscles anconeus lateralis (al) and medial branch of triceps brachii (m-tb). In views (C) and (D) the dotted line represents the surface of the attachment area occupying the lateral tubercle, whereas line indicates the complete inferred attachment area.

All these differences would imply that P. ogygia had a medial branch of the triceps brachii relatively larger than that of A. jubatus and larger and longer than that of the pantherins. In that case this muscle would be capable of quicker and more powerful contractions in P. ogygia than in similarly sized pantherins such as P. onca or P. pardus.

But there are more differences. When the olecranon processes of P. ogygia and the pantherins are compared in proximal view (Fig. 12), it can be seen that the lateral tubercle of P. ogygia is markedly narrower medio-laterally, thus showing a very distinct morphology from that of a pantherin. This lateral tubercle is part of the attachment area of the muscle anconeus lateralis, whose main action is to assist the triceps brachii in the forearm extension (Evans & de Lahunta, 1991; Barone, 2000). From this morphology, it can be inferred that in P. ogygia the number of fibres attaching on the anterior margin of this lateral tubercle would be proportionally lower than in a pantherin, which would basically make this muscle act from a more lateral position than in a pantherin. As the forearm extends by the action of the muscle triceps brachii, the olecranon approaches the posterior face of the distal epiphysis of the humerus, which effectively limits the contraction of the anconeus lateralis; thus, if the attachment of the anconeus lateralis on the olecranon is mainly lateral, the axis of its contraction is placed farther from the humerus, and the range of contraction would be higher. This small difference could represent an increase in the extension range of the forearm, as well as in the efficiency of the anconeus lateralis assistance to the triceps brachii in this action.

Fig. 12.

Proximal view of the right ulna of Panthera pardus (A) and Promegantereon ogygia from Batallones-1 (B), showing the different width of the lateral tubercle (l. t.).

All these changes in the ulna suggest that P. ogygia had already evolved some of the characters of the forelimb that reached their highest development in the Pleistocene sabre-toothed cats. The American S. fatalis from the Pleistocene of Rancho La Brea had extremely powerful forelimbs (Merriam & Stock, 1932; Gonyea, 1976a) and the olecranon tubercles of the ulna were very large, showing a similar proximo-distal development. Other, more primitive smilodontins, such as S. gracilis and M. cultridens, have the same olecranon pattern as P. ogygia. These differences between the primitive and derived smilodontins can be related to the great robustness of S. fatalis, a species that probably had very strong triceps brachii and anconeus lateralis muscles.

In summary, the morphology of the olecranon process of P. ogygia and primitive sabre-toothed cats probably points towards the presence of a medial branch of the triceps brachii that was more powerful than that of pantherins, and a muscle anconeus lateralis that was more efficient in its assistance of the triceps brachii. These differences would be related to the necessity of a quick immobilization of prey, and to the development of a powerful forepaw grasp.

The presence in P. lemanensis of similar olecranon morphology to that of P. ogygia poses some questions. Because of the great similarity between the postcranial skeleton of P. lemanensis and some extant climbing carnivores, it has been proposed that the former would also have been a mainly arboreal animal (Turner & Antón, 1997; Agustí & Antón, 2002). Most of these tree-climbing carnivores, such as G. genetta, L. wiedii, C. ferox and A. fulgens, have olecranon tubercles with similar proximo-distal development. So it is possible that the olecranon morphology of P. lemanensis does not necessarily or exclusively reflect an arboreal adaptation. It is feasible that this species was a mainly terrestrial or scansorial predator, and if it developed strong forelimbs for hunting activities this would reflect a parallel appearance of a machairodontine trait, or even perhaps a closer relationship with the sabre-toothed cats than previously supposed.

Pisiform

The differences in the development of the posterior tubercle could be related to the higher cursorial abilities of the open terrain felids in relation to the forest felids. In this respect, P. leo shows a pisiform morphology similar to that of the most cursorial felid, A. jubatus, indicating that P. leo has developed some adaptations for more efficient terrestrial locomotion, and also represents a morphological difference from the other big cat of the genus, P. tigris, which is basically a forest dweller. Interestingly, the radius of P. leo is longer relative to the humerus than in most extant or extinct felids, yet another indication of a moderate cursorial adaptation (Turner & Antón, 1997,Fig. 4: 11).

Fig. 4.

Bivariate plots of (A) PTL (posterior tubercle length) and PA (proximal area) of the scapholunar and (B) PDH (proximo-distal height) and MLW (medio-lateral width) of the trapezoid, in different species of fossil and extant felids, showing the regression line for each species.

The distal attachment of the muscle ulnaris lateralis is on a small area of the dorso-lateral vertex of the posterior tubercle, and on the lateral face of the proximal epiphysis of the fifth metacarpal, whereas the muscle ulnaris medialis is attached distally only on the medial half of the tubercle (Fig. 13). In those species with ellipsoidal tubercles (see list above), both attachment areas are slightly displaced laterally, because of the presence of a marked facet on the medial margin on which the two rough branches of the pisi-metacarpian ligament are attached. Those species with a rounded tubercle lack this facet, and the complete muscular attachment area occupies most of the posterior surface of the pisiform tubercle.

Fig. 13.

Posterior view of the left pisiform of Promegantereon ogygia from Batallones-1 (A) and Panthera leo (B), showing the attachment areas of the pisimetacarpian ligament (l. p. m.) and the muscles ulnaris lateralis (u. l.) and ulnaris medialis (u. m.).

As Table 9 shows, in canids such as C. lupus, the relative weight of the ulnaris lateralis is double than that of the ulnaris medialis, whereas in felines this relationship is the opposite. Although there are no differences in the relative weight of these muscles among the analyzed felids, the different shape of the tubercle of the pisiform and the relative medial displacement of the attachment areas (pointed out above) place the muscular fibres of the ulnaris lateralis closer to the extension–flexion plane of the wrist. One of the main functions of this muscle is to bend the hand in a lateral direction (Barone, 2000), so the medial displacement of its attachment area, closer to the plane of movement, limits the range of this action and reduces the risk of wrist dislocation during running. These modifications are also present in ungulates, in which the strong and fibrous insertions of this muscle passively limit the extension of carpals and metacarpals when the forelimb contacts with the ground during running (Barone, 2000). In this respect, P. ogygia and other felids display the primitive state, whereas A. jubatus, P. leo, P. atrox and H. latidens show a convergent morphology in this trait, probably due to their more cursorial habits.

Pyramidal

The relatively narrower anterior face of the pyramidal in P. ogygia and S. fatalis implies that, when articulating with the pyramidal, the angle between the antero-posterior axis of the pisiform and the proximo-distal axis of the forearm is greater in the sabre-toothed cats than in the felines; that is, the pisiform is oriented more proximally in the latter. The main consequence of this difference is that the distance between the rotation point (the wrist) and the attachment point of the muscles ulnaris medialis and ulnaris lateralis is greater in the sabre-toothed cats and G. genetta, probably increasing the strength of the muscular contraction (Taylor, 1974) and the range of wrist rotation. This different angle can be related to the different needs of cursorial vs. non-cursorial carnivores; the former need to control lateral movements of the wrist, whereas for the latter, strength is more important (Taylor, 1974).

Scapholunar and metacarpal I

Two muscles, abductor pollicis brevis and flexor pollicis brevis, are developed from the tubercle of the scapholunar to the proximal epiphysis of the first phalanx of the dewclaw (Davis, 1964). If the length of the metacarpal I were the same in P. ogygia and in the pantherins, the different morphology of the scapholunar tubercle would mean that these muscles would be relatively longer in P. ogygia than in a pantherin. The functions of the muscle abductor pollicis brevis are abduction of the dewclaw and rotation of the hand in a medial direction, whereas the muscle flexor pollicis brevis is mainly a flexor of the dewclaw (Davis, 1964; Barone, 2000). Nevertheless, the metacarpal I of P. ogygia is relatively shorter and stronger than that of the pantherins and A. jubatus (Fig. 14). Therefore these modifications cannot be related to a lengthening of these muscles. The more plausible interpretation is that these differences point to a displacement of all the abduction–flexion mechanisms of the dewclaw, with a consequent increase in the angle formed between it and the rest of the digits. Sabre-toothed cats have relatively larger dewclaws than pantherins (Salesa, 2002), and it is not surprising that this group have dewclaws with greater abduction and flexion capacities than pantherins, in keeping with the more powerful overall anatomy of their forelimbs.

Fig. 14.

Anterior view of the right metacarpal I of Promegantereon ogygia from Batallones-1 (A) and of the left one of Panthera pardus (B) for comparison.

It is remarkable that one of the most arboreal felids, L. wiedii (Alderton, 1998), has the same scapholunar pattern as G. genetta and the pantherins, but other arboreal carnivores, such as A. fulgens, although having a scapholunar tubercle distally oriented, also have a proximal prolongation. Ailurus fulgens uses its claws to grip the branches (Taylor, 1989) so it needs a high degree of abduction and flexion of the dewclaw; this probably explains this convergence with sabre-toothed cats.

The larger machairodontines, such as S. fatalis and H. latidens, have scapholunar tubercles with a similar length to those of similar-sized pantherins such as P. leo (Salesa, 2002). In fact, P. ogygia shows a different pattern from that of the more derived sabre-toothed cats, which could be due to allometry. Because P. ogygia is a medium-sized felid, and less powerful than larger machairodonts, it is feasible that it needed relatively stronger dewclaws than S. fatalis or H. latidens to achieve the machairodont attack, which required the complete immobilization of prey before biting its throat.

Metacarpals II–V and phalanges

The presence of a significantly more robust first phalanx and a larger second phalanx in P. ogygia indicates that this primitive sabre-toothed cat had a relatively larger and more powerful claw in the first digit compared with the pantherins, including the larger species such as P. leo. The dewclaw of felids has an important role in the hunt, because it is partially opposable and helps the animal control and subdue its prey (Gonyea, 1978; Turner & Antón, 1997). The presence in P. ogygia of a more powerful dew claw than that of pantherins indicates its importance in the machairodont hunting method, this character being present in subsequent species of sabre-toothed cats.

Magnum

The observed morphology of the magnum is probably related to the degree of mobility of the metacarpals, because if these bones overlap, the lateral movement is highly restricted, whereas if they do not overlap there is more range of movement between them. This lateral movement is more restricted in pantherins and H. latidens than in P. ogygia, S. fatalis, L. wiedii and G. genetta. Pantherins and H. latidens would have developed this facet for the metacarpal II in the magnum independently, because of the necessity for restricting this lateral movement. Promegantereon ogygia and S. fatalis show the primitive condition, with a higher degree of lateral mobility than pantherins, and this could have been advantageous for increasing the area of the paw during deployment of the claws. Homotherium latidens shows the derived morphology probably related to the cursorial adaptations of this sabre-toothed cat (Rawn-Schatzinger, 1992; Turner & Antón, 1997), which requires strong metacarpal contact to decrease the risk of luxation.

The facet for articulation with metacarpal II in the magnum is also present in canids, which supports its connection with cursorial adaptations. The overlap of the proximal borders of metacarpal II and III has been associated with digitigrady (Ginsburg, 1961; van Valkenburgh 1987). The forest pantherins such as P. pardus or P. onca, show restriction of lateral movements of the metacarpals in spite of the absence of cursorial adaptations. The origin of pantherins in very poorly understood (Turner & Antón, 1997), but it is interesting that this group bears the facet for metacarpal II in the magnum (derived state). It is feasible that pantherins have this morphology because their origin lies among a group of early felines that inhabited open environments (Agustí & Antón, 2002). Leopardus wiedii shows the primitive state, as well as viverrids, reflecting the arboreal habits of both groups.

Trapezoid

The presence of a high trapezoid in pantherins could be related to the relative enlargement of the carpal elements, which is typical of the cursorial mammals. As a result of this, when the carpals articulate, metacarpal II is clearly situated higher than metacarpal III (Ginsburg, 1961; Gonyea, 1976b). This difference also produces the lengthening of the lever arm of the forelimb, and an increase in the power of muscles related to running. That is the reason why cursorial felids, in contrast with the sabre-toothed cats, are expected to have trapezoids with a greater proximo-distal height.

Trapezium

The presence in G. genetta of a proximo-distally low trapezium is probably related to the great mobility of the dewclaw of this arboreal viverrid (Taylor, 1974), whereas smilodontins retained this low trapezium because of their hunting method, in which it is necessary to have a robust dewclaw with a wide abduction range. The modification of this character in felines would be related to their general cursorial adaptations, which implied the modification of the carpal and metacarpal morphology, as we pointed out above.

Conclusions

The overall anatomy of the anterior limb of P. ogygia shows important differences when compared with that of the pantherins. This primitive sabre-toothed cat shows early adaptations for powerful extension and supination movements in the forearm, a reinforcement of the elbow and shoulder stabilization, and a very robust dewclaw, all of them related to the evolution of the specialized hunting method of the machairodontines. This method, as has been discussed in depth (Akersten, 1985; Turner & Antón, 1997; Antón & Galobart, 1999; Antón et al. 2004a, 2005) was based on the rapid immobilization of prey, to apply the ‘canine shear-bite’ to the throat with the laterally compressed upper canines, cutting blood vessels and trachea and producing almost immediate death of the prey. This action had an evident risk for the hunter if the prey struggled because it could cause significant injuries such as the breakage of the upper canines, or even the mandible. The reduction in the duration of this initial phase was achieved in the primitive sabre-toothed cats, such as P. ogygia, by the development of strong forearms and a powerful dewclaw, which are the weapons used by felids to hold and subdue prey. The shortening of this phase would also save a significant amount of energy expenditure by the predator, and thus this advantage in reducing energetic costs and risks was probably the major evolutionary pressure leading to the appearance of the specialized killing method of the sabre-toothed cats, rather than the killing of larger prey (Salesa et al. 2005). These adaptations, producing such strongly built predators, may have made it possible for the later sabre-toothed cats, such as the tiger-sized S. fatalis or H. latidens, to subdue larger prey than those of similarly sized extant pantherins. However, the presence of adaptation for a strong forelimb grip in the much smaller P. ogygia emphasizes that the sabre-toothed model was not tied to large and thick-skinned prey as viewed in traditional theories, a point that was already indicated by the small size of some derived sabre-toothed carnivores such as the lynx-size nimravid E. bidentatus or the creodont M. eothen, which was not much larger than a domestic cat. In fact, in all known groups of sabre-toothed predators the basal taxa, where known, were considerably smaller than the crown species and yet they displayed the essential traits of the sabre-toothed complex.

In our view, the reduction of energetic costs and risks of the hunt would be the main adaptive advantage of the forelimb adaptations for strength, in combination with sabre-like upper canines (Salesa et al. 2005) in the primitive sabre-toothed cat P. ogygia, whereas other aspects of its anatomy, such as the mastoid morphology or the shape of the coronoid process in the mandible, remained in relatively plesiomorphic stages (Salesa et al. 2005). However, similarities in detail between the forelimb anatomy of P. ogygia and S. fatalis are difficult to interpret as a mere combination of convergence and shared retentions, and actually seem to strengthen the case for a close phylogenetic relationship, confirming the status of P. ogygia as close to the ancestry of Plio-Pleistocene smilodontins.

Supporting Information

Additional Supporting Information may be found in the online version of this article:

Table S1. Measurements in mm of the pisiform of the analyzed felid sample.

Table S2. Measurements in mm of the pyramidal of the analyzed felid sample.

Table S3. Measurements in mm of the scapholunar of the analyzed felid sample.

Table S4. Measurements in mm of the metacarpal I of the analyzed felid sample.

Table S5. Measurements in mm of the first phalanx of thumb of the analyzed felid sample.

Table S6. Measurements in mm of the third phalanx of the analyzed felid sample.

Table S7. Measurements in mm of the trapezoid of the analyzed felid sample.

Table S8. Measurements in mm of the trapezium of the analyzed felid sample.

Table S9. Regression equations for each analyzed bone (Abbreviations: F Ph Thumb, First phalanx of thumb).

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.