European remains of M. aphanistus outside the Iberian Peninsula are generally scarce and fragmentary. This species is mainly recorded from MN9 to MN10 localities, although there are some records from MN11; they include: Eppelsheim (MN9, type locality; Beaumont, 1975), Höwenegg (MN9; Bernor et al., 1988) and Dorn-Dürkheim 1 (MN11; Peigné et al., 2005) in Germany; Montredon and Soblay (MN10; Beaumont, 1975) in France; Charmoille (MN9; Beaumont, 1975) in Switzerland; Zillingdorf (MN9 or MN10; Beaumont, 1975 and Peigné et al., 2005) in Austria; Csákvár (MN11; Morlo, 1997) in Hungary; Nessebar (MN9 or MN10; Spassov et al., 2006) in Bulgaria; and Kemiklitepe D (MN11; Bonis, 1994, Peigné et al., 2005 and Spassov et al., 2006) and Mahmutgazi (MN11, Peigné et al., 2005) in Turkey.

In the Iberian Peninsula, M. aphanistus is recorded from several Spanish localities, including Los Valles de Fuentidueña (MN9, Duero Basin; Ginsburg et al., 1981), Batallones-1, 3 and 10 (MN10, Madrid Basin; Abella et al., 2011 and Morales et al., 2008), and several Vallès-Penedès localities (Agustí et al., 1984, Agustí et al., 1997, Alba et al., 2011a, Beaumont and Crusafont-Pairó, 1982, Crusafont and Truyols, 1954, Crusafont Pairó, 1959, Crusafont Pairó, 1964, Crusafont-Pairó and Golpe-Posse, 1972, Crusafont Pairó and Kurtén, 1976, Fraile et al., 1997, Golpe-Posse, 1974, Golpe-Posse et al., 1979 and Pons-Moyà, 1990).

From older to younger, the previously known Vallès-Penedès localities are the following: Creu Conill 22, Can Poncic 1, Santiga and Can Llobateres 1, from MN9; and Torrent de Febulines and La Tarumba 1, from MN10. The remains from Los Valles de Fuentidueña were used to erect a distinct species, M. alberdiae Ginsburg et al., 1981, which is currently considered a subjective junior synonym of M. aphanistus (Peigné et al., 2005). The most abundant and complete remains of the species have been recovered from Batallones (Antón et al., 2004, Salesa et al., 2005 and Turner et al., 2011).

The Iberian record of M. aphanistus in the Vallès-Penedès Basin (Fig. 1) is thus more restricted than in Europe as a whole, where this species persists into MN11, being apparently replaced by its purported descendant, A. giganteus, from MN11 onward. Here, we describe several unpublished remains of M. aphanistus from different localities of the Vallès-Penedès Basin, namely Can Mata (late MN7+8–MN9), Can Llobateres 1 (MN9), Viladecavalls (MN10) and Ronda Oest Sabadell ROS-D3 (MN10), further providing new data on the remains from the previously known localities of Can Poncic 1, Santiga, Can Llobateres 1, La Tarumba 1 and Torrent de Febulines. Finally, we discuss the presence of M. aphanistus in Creu Conill 22, putatively considered the first record of the Hippotherium–Machairodus association in the Iberian Peninsula (Agustí et al., 1997 and Casanovas-Vilar et al., 2006).

Institutions and fossil collections: ICP, Institut Català de Paleontologia Miquel Crusafont, Universitat Autònoma de Barcelona; IPS, collections from the ICP.

Fossil sites: CM: Can Mata; CNN: Creu Conill; CP: Can Poncic 1; SA: Santiga; CLL: Can Llobateres; LTR: La Tarumba; VC: Viladecavalls; TF: Torrent de Febulines; ROS: Ronda Oest Sabadell.

The fossil remains described in this paper (Fig. 2, Fig. 3, Fig. 4, Fig. 5 and Fig. 6) are housed at the ICP. The comparative sample includes fossil remains of M. aphanistus from Los Valles de Fuentidueña housed at the ICP, as well as data from other Late Miocene European machairodontines based on data taken from the literature (Anton et al., 2004; Beaumont, 1975, Peigné et al., 2005, Salesa et al., 2012, Sardella and Werdelin, 2007 and Sotnikova, 1992). Dental and mandibular measurements of the studied specimens were taken with a digital caliper to the nearest 0.1 mm.

Can Mata: IPS30910, complete right talus.

Can Llobateres 1: IPS13169, complete left C1; IPS13178, crown fragment of right C1; IPS13166, complete left P3; IPS13164, complete left P4; IPS13170, right P4 lacking the protocone; IPS13168, right P4 lacking paracone and metacone; IPS15042, right P4 without parastyle and paracone; IPS15123, partial right P4; IPS13167, complete left p4; IPS13165, complete right m1; IPS15032, distal fragment of right m1; IPS15094, partial left scapula; IPS13176, complete right second metacarpal; IPS15045, proximal fragment of left second metacarpal; IPS15061, proximal fragment of right third metacarpal; IPS11381, left third metacarpal lacking the head; IPS13172, right fourth metacarpal; IPS15046, proximal fragment of left fifth metacarpal; IPS11020, right scapholunar; IPS11341, diaphysis of right femur; IPS13173, complete right third metatarsal; IPS15043, proximal fragment of right fourth metacarpal; IPS13174 and IPS13175, two complete right fifth metatarsals; IPS13171, right calcaneus; IPS15077, right navicular; IPS11183, left cuboid; IPS15021, IPS15022, IPS15015, and IPS15007, proximal phalanges; IPS15068, IPS15008, and IPS15017, middle phalanges.

Can Poncic 1: IPS15033, right premaxillary fragment with I2–I3; IPS13177, complete right p4.

Santiga: IPS13185, rostral fragment with left I1–M1 and right I1–M1.

Torrent de Febulines: IPS28929, two crown fragments of C1; IPS28932, left scapholunar; IPS28904, complete left fifth metatarsal.

Viladecavalls: IPS36058, proximal phalanx lacking the trochlea.

La Tarumba 1: IPS16286, right third metacarpal lacking the head.

Ronda Oest Sabadell ROS-D3: IPS62083, nearly complete left hemimandible with c1–m1 and associated complete right hemimandible with c1–m1.

Measurements: See Table 1, Table 2 and Table 3.

Upper teeth: The left P3 from CLL1 (IPS13166) displays a high protocone, slightly inclined lingually. The two accessory cusps are well-developed, the distal cusp being mesiodistally longer than the mesial cusp. A marked cingulum runs from the distal portion of the P3 crown to the protruding distolingual expansion, as in Amphimachairodus and Panthera leo (Antón et al., 2004; Fig. 3D–F), thus resembling the condition displayed by IPS13185 from SA (Beaumont and Crusafont-Pairó, 1982). IPS13166 differs from Amphimachairodus and P. leo in the concave buccal profile and the presence of a weak cingulum on the distolingual corner of the crown (Antón et al., 2004).

Mandible: The mandible from ROS-D3 (IPS62083) is long, with relatively deep and robust corpora, a relatively slender and moderately verticalized symphysis that is higher than the corpus, and a postcanine diastema longer (42.2–43.3 mm) than in other specimens of M. aphanistus, thus closely resembling those of A. giganteus (Antón et al., 2004, Sardella and Werdelin, 2007 and Sotnikova, 1992). In the right hemimandible, there are three mental foramina, two below the diastema and another one below the p3 and close to the inferior margin of the mandible; on the left side, the two former foramina are more closely packed and partially fused with one another (Fig. 2B, E). The mandibular flange is only weakly developed, as it is characteristic of M. aphanistus, whereas in contrast A. giganteus has a better developed mandibular flange (Fig. 2B; Antón et al., 2004 and Sardella and Werdelin, 2007). The corpus is thinner and shallower at the level of the distal portion of the diastema, and thickest and deepest at the level of the p4. The masseteric fossa is deep, reaching anteriorly the level of the m1 protoconid. The angular process is clearly individualized from the corpus and posterolingually projected. The coronoid process is rostrocaudally wide, posteriorly inclined and superiorly projecting well above the uppermost symphyseal level (Fig. 2A, D).

Lower teeth: Only the canines and cheek teeth are preserved in the mandible from ROS-D3 (IPS62083). All these teeth are completely preserved, except for the missing tip of the right c1. The cheek teeth display a moderate degree of wear, with dentine exposure in both m1 and in the protoconid of the left p4 (Fig. 2B, E). c1 is high-crowned, proportionally larger than in A. giganteus, and implanted obliquely relative to the postcanine tooth row. The crown displays an oval (longer than broad) occlusal profile and is tilted distally (with its apex almost reaching the distalmost level of the crown base). The crown may be divided into a markedly convex buccal portion and a slightly concave distolingual portion. The latter is delimited by two (distal and mesiolingual) serrated crests that run from cervix to the apex. The mesiolingual crest is thicker than the distal one, and, unlike the latter, it ends shortly before reaching the crown apex; next to the former there is mesial shallow sulcus that runs from the cervix to about crown mid-height (Fig. 2A, B).

The premolars are tricuspid and somewhat broader distally than mesially. p3 displays a low and finely-serrated protoconid that is slightly tilted distalwards, well-developed mesial and distal accessory cuspids (the mesial one slightly smaller and more lingually situated than the distal one), and a distinct distal cingulid (Fig. 2C). p4 displays a similar morphology but is much larger than the preceding premolar. The protoconid has a similar occlusal morphology, although it is much higher, the mesial and distal accessory cuspids are comparatively better developed and similar in size to one another (being as high as the p3 protoconid), and the distal cingulid is even more conspicuous. As compared with the p4 from CLL1 (IPS13167) and CP (IPS13177), the p4 from ROS-D3 shows better developed and higher accessory cuspids and a stronger distal cingulid, thus more closely resembling the condition observed in A. giganteus than the previously described specimens (Beaumont and Crusafont-Pairó, 1982, Geraads et al., 2004 and Salesa et al., 2012).

The lower carnassial (m1) displays two main cuspids, the paraconid being lower and mesiodistally shorter than the protoconid, which is markedly asymmetrical (very tilted distally). The m1 further displays a short but distinct talonid with a distinct metaconid (Fig. 2C), which is relatively well-developed as compared with other specimens of M. aphanistus (Antón et al., 2004). Another m1 is available from CLL1 (IPS13165). It displays a similar occlusal morphology as the carnassials from the ROS-D3 mandible (IPS62083), although that from CLL1 displays a less developed talonid that further lacks a distinct metaconid (Fig. 3A–C).

Scapula: A partial right scapula (distal fragment) from CLL1 (IPS15094; Fig. 4A–C) preserves the glenoid cavity, the neck and the supraglenoid tuberosity. The glenoid cavity displays a subcircular profile as in Smilodon fatalis and Homotherium latidens from Europe, thus differing from H. serum, H. ischyrus and extant large felids, which display a transversely wider cavity that is rather ovoid (Ballesio, 1963, Martin et al., 2011, Merriam and Stock, 1932 and Rawn-Schatzinger, 1992). The supraglenoid tuberosity is mediolaterally wider than in Homotherium, thus resembling the condition observed in extant P. leo.

Carpus: One scapholunar from CLL1 (IPS11020; Fig. 4H, I) shares several characters with Homotherium. The radial facet is clearly convex and narrower than in extant large felids and Smilodon, thus being more similar to the condition seen in Homotherium. The proximal process is palmomedially situated, whereas in extant large felids and Homotherium, this process is more palmarly projected, and in Smilodon, it is clearly projected medially. The facet for the radial sesamoid is large and rounded, as in Homotherium (Ballesio, 1963, Hearst et al., 2011, Martin et al., 2011, Merriam and Stock, 1932 and Rawn-Schatzinger, 1992). The scapholunar has a distal rectangular outline, with the facet for the radial sesamoid oriented along the dorsoplantar plane. The facet for the trapezoid, situated on the dorsomedial corner of the bone, is triangular and slightly convex, being separated from the facet for the trapezium by a narrow dorsopalmar ridge. The facet for the trapezium extends towards the palmar margin, as in Homotherium and P. leo, but unlike in Smilodon (Ballesio, 1963; Hearst et al., 2011; Merriam and Stock, 1932 and Rawn-Schatzinger, 1992). The facet for the magnum is very concave, running from the palmar to the dorsal side. The facet for the unciform is rectangular and narrower than in extant large felids and Homotherium, thus being similar to the condition displayed by Smilodon (Ballesio, 1963, Hearst et al., 2011, Merriam and Stock, 1932 and Rawn-Schatzinger, 1992). The scapholunar from CLL1 (IPS11020) displays the same morphological features as that previously published from TF (IPS28932; Fig. 4F, G; Pons-Moyà, 1990), although the latter is somewhat larger in size (Table 3).

Metacarpus: There are two second metacarpals available from CLL1: IPS13176 (Fig. 5P–T), which is complete; and IPS15045, which only preserves the proximal portion. IPS13176 is stouter (proximodistally shorter and mediolaterally wider) than in Homotherium, Smilodon and P. leo, thus more closely resembling the condition of Xenosmilus (Fig. 8). The diaphysis is slightly curved palmarly, as in P. atrox but unlike in the other taxa mentioned above (Ballesio, 1963, Martin et al., 2011, Merriam and Stock, 1932 and Rawn-Schatzinger, 1992). The base displays a triangular proximal profile, as in Homotherium, Smilodon and extant large felids (Christiansen and Adifssen, 2007; Merriam and Stock, 1932 and Rawn-Schatzinger, 1992). The articular facet for the trapezoid is markedly concave, as in extant large felids, Megantereon and Smilodon, thus differing from the flat facet displayed by Homotherium (Ballesio, 1963, Christiansen and Adolfssen, 2007, Merriam and Stock, 1932 and Rawn-Schatzinger, 1992). The facet for the trapezium is triangular and relatively small, as in Homotherium, whereas in Xenosmilus it is more circular and in Smilodon larger and more elongated (Ballesio, 1963, Martin et al., 2011, Merriam and Stock, 1932 and Rawn-Schatzinger, 1992). The articular facets for the magnum, present on the lateral side of the metacarpal base, are poorly preserved in the two available specimens. The articular facet for the third metacarpal is very conspicuous and distally-projected, similar to the morphology displayed by extant large felids and Homotherium.

The third metacarpal is represented by three specimens: IPS11381 from CLL1 (Fig. 5U–Y) is complete, whereas IPS15061 from CLL1 and IPS16286 from LTR (Fig. 5K–O) only preserve the proximal portion. The base is poorly preserved in all the available specimens, although it can be discerned that the facet for the magnum is concave, as in Xenosmilus, Megantereon and Smilodon, whereas in contrast the facet for the magnum is convex, as in Homotherium and Panthera (Christiansen and Adolfssen, 2007, Martin et al., 2011 and Rawn-Schatzinger, 1992). The proximolateral facet for the unciform is posteriorly elongated and less triangular than in P. leo, thus resembling the morphology displayed by Homotherium, Meganteron and Xenosmilus (Christiansen and Adolfssen, 2007, Martin et al., 2011 and Rawn-Schatzinger, 1992). The dorsalmedial facet for the second metacarpal is dorsodistally elongated compared to Homotherium, Megantereon and Xenosmilus, thus resembling the condition displayed by P. leo (Ballesio, 1963, Christiansen and Adolfssen, 2007 and Martin et al., 2011, Rawn-Schatzinger, 1992). The diaphysis is slender (mediolaterally narrow relative to length) and subtriangular in cross-section.

A proximal fragment of a fifth metacarpal from CLL1 (IPS15046; Fig. 5E′–I′) shows that the facet for the unciform is rectangular and very convex, thus being similar in shape but less posteromedially inclined than in Homotherium, and more clearly differing from the triangular morphology displayed by Smilodon, P. atrox and extant P. leo (Ballesio, 1963, Merriam and Stock, 1932 and Rawn-Schatzinger, 1992). The medial articular tubercle for the fourth metacarpal is less developed than in extant large felids and P. atrox, thus resembling the condition seen in Smilodon and Homotherium (Merriam and Stock, 1932 and Rawn-Schatzinger, 1992). The proximolateral tuberosity is flattened or depressed, as in Smilodon and Homotherium, whereas in extant large cats and P. atrox it is generally convex (Merriam and Stock, 1932 and Rawn-Schatzinger, 1992).

Tarsus: The right talus from CM (IPS30901; Fig. 4D, E) is complete and well preserved. The trochlea is grooved and shallower than in extant large felids, thus resembling the condition observed in Homotherium, Amphimachairodus and Smilodon. In contrast, the neck is relatively longer than in Homotherium, Amphimachairodus, Smilodon and P. leo (Ballesio, 1963, Berta, 1987, Merriam and Stock, 1932 and Salesa et al., 2012). On the plantar side, the astragalar foramen can be discerned, although it is poorly preserved. The sustentacular and navicular facets are also poorly preserved, being separated by a deep and wide groove from the rectangular (mediolaterally wide) and strongly convex astragalocalcanear facet, resembling the morphology of the various species of Homotherium (Ballesio, 1963, Berta, 1987 and Merriam and Stock, 1932).

A right navicular from CLL1 (IPS15077; Fig. 4J–K) shows that the astragalar facet is rounded and slightly concave, as in other Machairodontini. The ectocuneiform facet is slightly larger than the mesocuneiform facet, further being merely separated from one another dorsally by a ridge as in Smilodon and Homotherium. The cuboid and calcaneal facets are contiguous but separated by a weak ridge, as in Smilodon and Homotherium (Ballesio, 1963, Berta, 1987, Martin et al., 2011 and Merriam and Stock, 1932). The morphology of the described specimen fits perfectly with that previously reported for the same element from Los Valles de Fuentidueña (IPS12664; Crusafont-Pairó and Ginsburg, 1973).

The left cuboid from CLL1 (IPS11183; Fig. 4L–Q) similarly shows the same morphology displayed by the specimen from Los Valles de Fuentidueña (IPS12661) reported by Crusafont-Pairó and Ginsburg (1973). The calcaneal proximal facet is quadrangular and slightly convex, thus differing from the more rectangular and flat morphology displayed by Homotherium. IPS11183 is proximodistally higher than in Homotherium and Smilodon, more closely approaching the condition observed in extant large felids (Martin et al., 2011, Merriam and Stock, 1932 and Rawn-Schatzinger, 1992). The ectocuneiform facet is dorsoventrally high and located more dorsally than in Homotherium. The small and triangular navicular facet is located on the dorsodistal corner of the bone, like in Homotherium. In lateral view, the oblique, deep and wide groove for the insertion of the m. peroneus longus is more marked and deeper than in the specimen from Los Valles de Fuentidueña (Crusafont-Pairó and Ginsburg, 1973), thus more closely approaching the condition of Homotherium (Martin et al., 2011). The facets for the fourth and fifth metatarsals are visible on the distal portion of the bone, although poorly preserved.

Metatarsus: A proximal fragment of fourth metatarsal from CLL1 (IPS15043, Fig. 5Z–D′′) shows that the proximal facet for the cuboid is rectangular and slightly convex, further being mediolaterally broader on its plantar than on its dorsal portion, as in other Machairodontini, but unlike in pantherines (Martin et al., 2011 and Rawn-Schatzinger, 1992). There is only a single, deep and distally-projected facet for the fifth metatarsal, whereas in Homotherium, Smilodon, Amphimachairodus and pantherines there are two deep but less distally-projected facets (Ballesio, 1963, Berta, 1987, Merriam and Stock, 1932, Rawn-Schatzinger, 1992 and Salesa et al., 2012). The fourth metatarsal displays two medial facets for the third metatarsal: a subcircular one, situated dorsoproximally; and a more rectangular and inclined one, situated more palmarly. The circular dorsal facet is generally situated more proximally in Homotherium, Smilodon and P. leo than in the studied specimen, whereas the posterior facet in the latter resembles that observed in P. leo (Ballesio, 1963, Berta, 1987, Merriam and Stock, 1932 and Rawn-Schatzinger, 1992). Three fifth metatarsals are available: one from CLL1 (IPS13175) published by Beaumont and Crusafont-Pairó (1982), another from TF (IPS28904, Fig. 5F–J) published by Golpe-Posse et al. (1979) and Pons-Moyà (1990), and a third one CLL1 (IPS13174, Fig. 5A–E) that remained unpublished. All these specimens display the morphology already described by Beaumont and Crusafont-Pairó (1982) and Pons-Moyà (1990), although that from TF is comparatively stouter and further displays a very deep ligamental groove on its proximoplantar portion that is not observed in the specimens from CLL1 (Fig. 8).

Phalanges: Four complete proximal phalanges from CLL1 (IPS15015, IPS15007, IPS15022 and IPS15021; Fig. 4R–Y) and a proximal fragment from VC (IPS36058) are available. Their comparison with extant P. leo and the extinct H. latidens specimens do not enable their attribution to either the manus or the pes, or to any specific ray. These specimens display the same morphology as that previously described from Los Valles de Fuentidueña (IPS12667; Crusafont-Pairó and Ginsburg, 1973), being characterized by a mediolaterally robust diaphysis, a vertically oriented proximal articular surface and a deep ligamental groove on its proximal portion, and thus resembling the morphology displayed by Amphimachairodus, Homotherium and Smilodon (Ballesio, 1963, Berta, 1987, Gaudry, 1832, Merriam and Stock, 1932 and Salesa et al., 2012).

The comparison of three middle phalanges from CLL1 (IPS15068, IPS15008 and IPS15017; Fig. 4A–H) with those of pantherines and H. latidens do not allow us to provide with an anatomical attribution to either manual or pedal phalanges. IPS15068 (Fig. 4E–H) and IPS15008 (Fig. 4A–D) have slender and longer shafts than in IPS15017. In general terms, the reported middle phalanges seem slightly stouter than those of P. leo. The same elements of Homotherium and Smilodon seem generally slenderer and proximodistally shorter (Ballesio, 1963, Berta, 1987 and Merriam and Stock, 1932).

Creu Conill: Agustí et al. (1997) reported the presence of M. aphanistus from CCN22 (MN9; 11.1 Ma) on the basis of an undescribed left P4 (IPS28746; Fig. 6A–C). In fact, the morphology of this P4, which does not preserve the protocone, is quite different from those of M. aphanistus. The parastyle in IPS28746 lacks the well-developed buccal cingulum that is characteristic of M. aphanistus, and which can be observed in the P4 of this taxon available from the Vallès-Penedès Basin from CLL1 (IPS13170; Fig. 3G–I) and SA (IPS13185). Moreover, IPS28746 from CCN22 displays a slightly marked lingual cingulum running from the distal aspect of the metacone to the mesial portion of the paracone (Fig. 6B), which cannot be observed in the Vallès-Penedès material of M. aphanistus, but which is displayed by medium-sized Late Miocene hyaenids from this basin, such as Thalassictis. We therefore remove IPS28746 from the hypodigm of M. aphanistus and attributed it to Hyaenidae indet. (see a discussion of the biostratigraphic implications below).

Can Mata: The talus from the classical collections of CM (IPS30901) shares several characters with those of P. leo (such as a moderately large neck, and a deep and mediolaterally compressed trochlea). In these features, this taxon differs from Homotherium and Amphimachairodus, which usually display a relatively shorter talar neck, together with a narrow and mediolaterally wide trochlea (Ballesio, 1963, Berta, 1987, Merriam and Stock, 1932 and Salesa et al., 2012). The characters displayed by the studied talus therefore agree with the purported less derived morphology of M. aphanistus compared with both Amphimachairodus and Homotherium.

Can Poncic, Can Llobateres and Santiga: The remains of Machairodus from CP1, CLL1 and SA were published by Beaumont and Crusafont-Pairó (1982), who attributed them to M. cf. aphanistus. Here, we describe previously unpublished material from CLL1, mostly consisting of postcranial remains. The latter closely resemble in morphology the postcranial material of M. aphanistus from Los Valles de Fuentidueña (Crusafont-Pairó and Ginsburg, 1973), although the metacarpal shafts and the phalanges are stouter than in other machairodontins (more similar to the stoutly-built metacarpals and phalanges of P. leo).

La Tarumba and Viladecavalls: The felid material from LTR, in the area of Viladecavalls, was originally described and published by Villalta Comella and Crusafont Pairó, 1943 and Villalta Comella and Crusafont Pairó, 1948, who attributed it to Felis antediluviana and Felis sp. Villalta Comella and Crusafont Pairó (1943), in particular, reported from this locality a corpus fragment with p4–m1, whereas Villalta Comella and Crusafont Pairó (1948) reported a few postcranials (a second, third and fourth metacarpals, and a proximal phalanx) that they considered might belong to the same taxon recorded by the dentognathic fragment. The morphology of the right third metacarpal (IPS16286) is clearly that of a felid, and the proportions and morphology of the base allows us to include it in the hypodigm of M. aphanistus (Table 3), together with other postcranial specimens from the same locality. In contrast, the other metacarpals and the proximal phalanx described by Villalta Comella and Crusafont Pairó (1948) are mediolaterally wider and proximodistally shorter, which together with the morphology of the articular facets indicates an alternative attribution to an ursid, probably Indarctos vireti. In contrast, the mandibular fragment (IPS35093) described by Villalta Comella and Crusafont Pairó (1943) is attributable to Paramachaerodus sp., on the basis of its shallow mandibular corpus, the presence of a well-developed accessory cuspids and a distinct buccal cingulid in the p4, and an m1 protoconid clearly longer mediodistally than the paraconid (J.M.-M. and J.M.R., unpublished data). Here, we also report a previously undescribed proximal phalanx (IPS36058) coming from an undeterminate stratigraphic horizon from the area of Viladecavalls. This specimen can be attributed to M. aphanistus based on similarities (robust shaft and deep ligamental groove) with the previously published material from Los Valles de Fuentidueña (Crusafont-Pairó and Ginsburg, 1973).

Ronda Oest Sabadell: The mandible from ROS-D3 is ascribed to M. aphanistus based on several characters displayed by material of this taxon from other Eurasian localities, such as Batallones-1 (Antón et al., 2004, Salesa et al., 2005 and Turner et al., 2011), CLL1 (Beaumont and Crusafont-Pairó, 1982; this study), and Eppelsheim and Charmoille (Beaumont, 1975). According to the emended diagnosis of this species provided by Antón et al. (2004), these features include: large lower canines of oval cross-section; large lower premolars with a complete set of accessory cuspids; well-developed metaconid-talonid complex in m1; thick and high mandibular ramus; high and posteriorly inclined coronoid process; and undeveloped mandibular flange. The ROS specimen can be thus distinguished from A. giganteus by the possession of larger c1, the relatively broader p4, the non-reduced talonid-metaconid complex in the lower carnassial, and the poorly-developed flange. The specimen from ROS thus merely shows a few slight differences compared to the remains from other Eurasian localities, namely: relatively longer postcanine diastema; shorter p3 relative to p4 (Fig. 2E); better developed premolar accessory cuspids (especially in p4); and better developed m1 metaconid. Amphimachairodus displays several derived dental features relative to Machairodus (Geraads et al., 2004), which support the purported phyletic lineage constituted by the Vallesian M. aphanistus, the Turolian-Ventian A. giganteus, and the Plio-Pleistocene Homotherium (Kurtén and Anderson, 1980 and Sotnikova, 1992). Dental evolution throughout this lineage would reflect the progressive acquisition of several traits already displayed to some degree by the Turolian-Ventian species (Geraads et al., 2004), including the reduced c1, shorter p3 (usually absent in Homotherium), larger p4 with a well-developed mesial accessory cuspid, and loss of the m1 talonid.

In the above-mentioned regards, the ROS specimen displays some derived traits compared to typical M. aphanistus, including the somewhat reduced p3 (Fig. 2A) and the longer postcanine diastema. Given that the ROS specimen fits with the known chronostratigraphic range of M. aphanistus from the Vallès-Penedès Basin and elsewhere in Eurasia, these slight differences support the view that the characters employed to distinguish Vallesian from Turolian-Ventian machairodontins are more variable than customarily recognized (Beaumont, 1975 and Sotnikova, 1992), further including symphyseal height, development of the mandibular flange, length of the postcanine diastema, and relative size between the premolars. We therefore conclude that the above-mentioned slight dentognathic differences merely reflect the intraspecific variability of M. aphanistus. This is also supported by the fact that the ROS mandible still retains a well-developed p4 mesial accessory cuspulid as well as a distinct m1 talonid with a large metaconid (even larger than in typical M. aphanistus), which according to Geraads et al. (2004) should be interpreted as primitive features. Only a few complete mandibles of M. aphanistus are known from Eurasia, except from the large collection from Batallones, which in the future will hopefully enable a more complete assessment of intraspectific variability in this species.

Torrent de Febulines: The felid remains from TF were originally published by Golpe-Posse et al. (1979), who attributed the following remains to M. aphanistus: two C1 crown fragments (IPS28932), one scapholunar (IPS28932), a fifth metacarpal (IPS28904), a first metacarpal and an acetabulum fragment. The two latter specimens were not found among the ICP collections by the authors, and subsequently Pons-Moyà (1990), who restudied the Machairodus remains from TF, only described the remains that are reported here. According to Pons-Moyà (1990), these fossils should be attributed to Amphimachairodus giganteus, based on the buccolingual compression and crenulated morphology of the two canine fragments, as well as the morphological similarities of the fifth metatarsal with the material of this taxon from Concud (Pons-Moyà, 1990). Our assessment of the material indicates that the TF specimens display the same morphology as those from CLL1 and SA, merely differing from the latter by the slightly larger dimensions (and hence stouter appearance) of the former (Table 3). The scarce available material from TF thus lacks clear resemblances with Amphimachairodus, leading us to conclude that the above-mentioned differences in size compared to specimens of M. aphanistus from SA and CLL1 are merely attributable to the large degrees of interspecific variability and sexual dimorphism characteristic of medium to large-sized felids.

M. aphanistus is mainly known on the basis of craniodental material, including several mandibles from various European sites and even complete skulls from Batallones-1 (Antón et al., 2004). In contrast, the postcranial anatomy of this early scimitar-toothed cat is practically unknown. Only scarce postcranial material was published by Crusafont-Pairó and Ginsburg (1973), Beaumont and Crusafont-Pairó (1982) and Pons-Moyà (1990), who nevertheless did not provide any inferences on the locomotor behavior of this species. The newly described postcranial remains of M. aphanistus (basically, carpals, metacarpals, tarsals, metatarsals and phalanges) allow us to provide important new data for better understanding the locomotor and predatory behavior of this taxon. The metacarpals and metatarsals are clearly stouter (mediolaterally wider) than in Homotherium, thus more closely resembling those of P. leo (Fig. 7 and Fig. 8; Ballesio, 1963 and Rawn-Schatzinger, 1992). The proximodistally-elongated calcaneus, and the relatively long neck as well as deep and mediolaterally compressed trochlea of the talus are also more similar to those of P. leo than to those of Homotherium (Ballesio, 1963 and Rawn-Schatzinger, 1992). In contrast, the cuboid articular facets for the calcaneus and metatarsals are parallel, as in Acinonyx and Homotherium, instead of diverging as in P. leo (Ballesio, 1963, Hearst et al., 2011 and Rawn-Schatzinger, 1992). The middle phalanges are proximodistally long, as in P. leo, rather than short as in Homotherium, but like those of the latter they display a lesser degree of mediolateral asymmetry than in P. leo (Ballesio, 1963, Hearst et al., 2011 and Rawn-Schatzinger, 1992).

The long calcaneus and the morphology of talus (with a moderately long neck and a mediolaterally compressed and deep trochlea), similar to those of P. leo, indicate a digitigrade locomotion for M. aphanistus (Ballesio, 1963 and Rawn-Schatzinger, 1992). In contrast, Homotherium displays a proximodistally short calcaneum and a bear-like talus with a proportionally short neck, these features being rather associated with a semiplantigrade stance with the ability to place the foot nearly flat on the ground (Ballesio, 1963 and Rawn-Schatzinger, 1992).

The morphology of the studied cuboid and the lesser degree of asymmetry in the middle phalanges (which implies a smaller degree of claw retraction; Gonyea, 1976) resemble the condition seen in H. serum and H. latidens (Ballesio, 1963, Hearst et al., 2011 and Rawn-Schatzinger, 1992). A smaller degree of claw retraction is usually related to the condition seen in Acinonyx jubatus and canids, which have non-retractile claws so as to combine high speed with catching ability (Rawn-Schatzinger, 1992). In the case of Homotherium and probably M. aphanistus, the lesser degree of asymmetry in the middle phalanges may be associated with a high degree of traction during hunting at high speeds (Eaton, 1974, Ewer, 1973 and Rawn-Schatzinger, 1992).

Several authors have previously suggested long-distance travel capabilities in open habitats for Homotherium (Ballesio, 1963, Hearst et al., 2011 and Rawn-Schatzinger, 1992). The detailed study of the postcranial remains of Homotherium shows that clearly cursorial features (the proportions of the hindlimbs and the non-retractile claws) are associated with more generalized terrestrial features (such as the morphology of the talus and calcaneum). This evidence suggests moderate cursorial abilities and a body plan adapted for running in Homotherium (Rawn-Schatzinger, 1992). In contrast, M. aphanistus appears as a more stoutly-built species that generally more closely resembles P. leo, including the morphology of the tarsal bones. This is evidenced, among other features, by the stoutness of the metapodials of M. aphanistus, which are mediolaterally much wider than in the more lightly-built Homotherium. The morphology of the middle phalanges, proximodistally long (as in P. leo) and with a low degree of asymmetry (resembling Homotherium), implies a moderate capacity of traction during hunting at high speeds, and foreshows to some degree the more derived traits displayed in this regard by Homotherium (Rawn-Schatzinger, 1992) Overall, thus, M. aphanistus displays several postcranial features more primitive than in Homotherium, with the former being apparently less adapted for long-distance travel and high-speed hunting than its putative descendant.

Before this study, the earliest record of the genus Machairodus in the Vallès-Penedès Basin was thought to be in the early MN9 locality of CCN22, being slightly younger than the locality of CCN20, which records the first appearance datum of the equid Hippotherium in this basin, with an estimated age of 11.1 Ma (Agustí et al., 1997 and Garcés et al., 1997). In fact, Agustí et al. (1997) considered that the putative joint dispersal of Hippotherium and Machairodus might be a major dispersal event of eastern immigrants, marking the beginning of “MN9a” (Megacricetodon ibericus+Hippotherium biozone). Agustí and Galobart (1998) reported the presence of Machairodus sp. from CCN20, although an inspection of the ICP collections showed that only a single carnivoran remain superficially resembling this taxon was available from locality CCN22, as reported by Agustí et al. (1997) and Casanovas-Vilar et al. (2006). Here, we show that this specimen belongs in fact to a hyaenid, indicating that the first appearance datum of M. apahanistus in the Vallès-Penedès might be younger than previously assumed.

On the other hand, the previously undescribed talus from CM described here probably represents the oldest Vallès-Penedès record of this genus. This specimen comes from the historical collections amassed decades ago by Miquel Crusafont et al. around Can Mata de la Garriga in els Hostalets de Pierola. This area records the Aragonian/Vallesian transition at 11.1 Ma (Alba et al., 2006, Alba et al., 2011b, Casanovas-Vilar et al., 2011a and Moyà-Solà et al., 2009), so that uncertainties concerning the precise stratigraphic horizon of provenance precludes determining with certainty whether the described talus is MN7+8 or MN9 in age. However, this specimen indicates at least that M. aphanistus was already recorded in this basin by the Earliest Vallesian. All the remaining Vallès-Penedès material of this taxon comes from younger localities, the oldest ones being CP1 and SA, with an estimated age comprised between 9.9 and 10.4 Ma (MN9). The new material from ROS-D3 (9.7–8.7 Ma, MN10), in contrast, is dated to the Late Vallesian, being probably coeval or slightly younger than that from LTR (ca. 9.6 Ma, MN10). However, the certain last occurrence datum of M. aphanistus in the Vallès-Penedès Basin corresponds to the site of TF (ca. 9.1 Ma, MN10). Overall, the chronostratigraphic range of M. aphanistus in the Vallès-Penedès Basin agrees with its known chronological distribution in the rest of Europe, where it is mainly recorded in MN9–MN10, with doubtful and scarce records in the Early Turolian (MN11) that might be alternatively attributable to A. giganteus (Antón et al., 2004).