The Capo Vaticano–Monte Poro sedimentary basin is located in the southwestern sector of the Calabria–Peloritani Arc. The basin overlies a crystalline substratum, Palaeozoic in age, and is characterised by a stratigraphic succession, attributable to the late Miocene, that outcrops with different thickness and facies throughout the area (Nicotera, 1959).

The succession consists of the following four informal units, from bottom to top: •

“dark argillaceaous sands with Ostrea and Cerithium”, alternating with coarse sandstones, attributed to lagoonal deposition (Neri et al., 2005);

The succession is covered by Messinian and Plio-Pleistocene deposits. At the Cessaniti site (Fig. 1), quarry works have exposed the stratigraphic sequence and have released a rich paleontological record. At the Gentile's Quarry, the outcrop of the late Miocene succession is complete and at a maximum thickness (Fig. 2). The succession has been interpreted as a single transgressive event (Carone and Domning, 2007, Gramigna et al., 2008 and Neri et al., 2005), until recent research that has suggested a more complex evolution of the basin (Marra et al., 2017). In particular, the recognition of soils and fluvial deposits in the “Clypeaster sandstones” (FL1 to 3 in Fig. 2) has allowed reinterpretation of the unit as having been affected by temporary falls in sea level within the general transgressive trend, probably under control of tectonics (Marra et al., 2017). The most relevant fall of the relative sea level is documented by the thick fluvial deposit FL1, which sharply cuts the lagoonal deposit LG (Fig. 2). The shoreface deposit SH1 represents a subsequent marine ingression, marked by a ravinement surface (RS in Fig. 2) overlying FL1. Two minor fluvial phases (FL2 and FL3) are interlaid on the shoreface deposits SH1–SH3, marked at the bottom by ravinement surfaces (RS, Fig. 2). The shoreface sandstones SH4 (“Heterostegina yellow sandstones”) indicate a definitive setting of frankly marine conditions, with a subsequent increasing of depth documented by the overlying marls OT (Offshore Transition; “Orbulina marls”). The maximum age range of the Cessaniti succession, from the lagoonal deposits to the top of “Heterostegina yellow sandstones”, ranges between 8.1 Ma (by the attribution of LG to the Chron C4n; Marra et al., 2017) and 7.2 Ma (by the attribution of the “Orbulina marls” to the nannoplankton zone CNM17; Marra et al., 2017).

At Cava Gentile, mammalian remains arise mainly from the “Clypeaster sandstones” (Fig. 2) and are attributed to Metaxytherium serresii (Carone and Domning, 2007 and Carone et al., 2013), Bohlinia cf. attica (Marra et al., 2011), Cetaceans (Carone and Marra, 2014 and Marra et al., 2016), Rhinocerothidae indet., Anthacotheridae indet. and Bovidae indet. (Marra et al., 2017 and Marra et al., 2011). Stegotetrabelodon syrticus is recorded in the brackish mud deposits LG, as well as in the “Clypeaster sandstones” (Fig. 2; Ferretti et al., 2017 and Ferretti et al., 2003). Giraffidae fossils, attributable to Samotherium cf. boissieri, have been recovered in the “Clypeaster sandstones” that form outcrops at Zungri, near Cessaniti (Fig. 1; Marra et al., 2011).

The hemimandible attributed to Tragoportax cf. rugosifrons in this paper, and the tentatively referred radius, astragalus and anterior phalanx, have been recovered at Cava Gentile and come from sandstones SH2–FL3 (Fig. 1 and Fig. 2). Other post-cranial bones, tentatively referred to Tragoportax cf. rugosifrons, come from Papaglionti (Fig. 1), a village near Cessaniti, where the succession thickness is reduced compared to the thickness at Cessaniti. The succession outcropping near the village of Papaglionti is located on the “Serre” hill, whose northeastern side released the studied materials. The succession, from bottom to top, consists of: •

grain-coarse sandstones, correlatable to FL1 of Cava Gentile;

The thickness of the outcrop is lower than that at Cessaniti, and the sandstones do not show evidence of fluvial phases. The remains, tentatively attributable to Tragoportax, have been recovered in the uppermost layers of the Clypeaster sandstones.

Order Artiodactyla Owen, 1848

Family Bovidae Gray, 1821

Subfamily Bovinae Gray, 1821

Tribe Boselaphini Knottnerus-Meyer, 1907

Genus Tragoportax Pilgrim, 1937

Tragoportax cf. rugosifrons (Schlosser, 1904)

HOLOTYPE. — Skull (Schlosser, 1904: pl. 12, fig. 6).

TYPE LOCALITY. — Samos

Referred material: hemimandible (Inventory Number: 4 (CES)vc6);

Tentatively referred material: radius (Inventory Number: 13 (CES)vc6); humerus (Inventory Number: 26 (PAP)vc2); anterior phalanx (Inventory number: 123 (CES)vc7); metatarsal (Inventory Number: 119 (PAP)vc2); posterior phalanx (Inventory number: 124 (PAP)vc2); astragalus (Inventory Number: 120 (CES)vc7).

Repository: Civico Museo di Paleontologia di Ricadi (MURI), via Strada Provinciale, 89866 Santa Domenica di Ricadi (VV) Italy.

The main criticism in Boselaphini taxonomy lies in features of the Tragoportax–Miotragocerus complex, represented by two genera instituted after the invalidation of the genus Tragocerus and included in the Tribe Tragoportacini (Bibi et al., 2009). The distinction between the two genera is not often definite: new genera have been erected in recent times and species have been moved among them (Bibi, 2011, Bibi et al., 2009 and Gentry, 2010). At the present state of the art, the genus Tragoportax includes species signalled in the late Miocene–Pliocene of Eurasia and Africa, with an opened debate on the generic attribution and on possible synonymies. Diagnostic characters for the determination of the species are derived mainly from the horn-cores; however, some data on the lower teeth can be considered for taxonomic attributions of the Boselaphine from Cessaniti.

Spassov and Geraads (2004) indicated morphological and biometrical differences between the genera Tragoportax and Miotragocerus using the sample from Hadjidimovo (Bulgaria; 8.1–7.1 Ma), and they referred to the species T. rugosifrons and M. (Pikermicerus) gaudryi. Although the main differences are in the body size (larger in Tragoportax) and in the morphologies of the horn-cores, the authors indicated distinctive characters in teeth and post-cranial bones as well.

As observed in T. rugosifrons from Hadjidimovo (Spassov and Geraads, 2004), the specimen from Cessaniti has a premolar row that is shorter than the molar row; the metaconids of p3 and p4 are large but less enlarged lingually; the anterior valleys of the premolars are open and well developed; and the premolar row is short compared to the molar row. These characters are distinctive for the genus Tragoportax sensu Spassov and Geraads (2004). Furthermore, tooth dimensions and proportions clearly separate the two genera (Fig. 2 in Kostopoulos, 2009).

The tooth measurements and proportions of the specimen from Cessaniti lie within the variability described for the genus Tragoportax. The premolar/molar (p/m) index of the specimen from Cessaniti is 0.67, which is comparable with the values for Tragoportax rugosifrons from Hadjidimovo (0.65, 0.64% and 0.58; Spassov and Geraads, 2004), but differs from the values for Miotragocerus (Pikermicerus) gaudryi from the same site (0.68 to 0.78; Spassov and Geraads, 2004).

The distinction of the species belonging to the genus Tragoportax is highly debated and further complicated by the moving of some Tragoportacini species from one genus to another, as well as by erecting new species based on scanty materials. Given the characters of the Cessaniti assemblage, comparisons have mainly addressed the Tragoportax species of the Greco-Iranian bioprovince and of Africa.

The tooth dimensions and proportions of the Cessaniti specimen fall within the morphological and biometrical variability of Tragoportax rugosifrons from Hadjidimovo (Spassov and Geraads, 2004; Fig. 4), and are similar to those of T. rugosifrons from Samos (Greece; Solounias, 1981; Fig. 4) and from Perivolaki (Greece, 8.1–7.1 Ma; Kostopoulos, 2006; Fig. 4). However, the tooth measurements and proportions of Tragoportax rugosifrons from Perivolaki are very close to those of T. amalthea (Kostopoulos, 2006; Fig. 4). An interesting observation is that Kostopoulos (2006) noticed more advanced features for the sample from Perivolaki, where 8 of the 13 specimens showed a molarised p4 in addition to an elongated p2 and a decrease of the paraconid in p3. These observations could explain why the premolar row was slightly more elongated than in T. rugosifrons from Hadjidimovo and Tragoportax from Cessaniti, where these characters are not reported, as well as providing a reason for the partial overlap with the p/m indexes vs. molar length of T. amalthea from Nikiti-2 (Greece, 8.7–8.0 Ma; Kostopoulos, 2016), a species characterised by proportionally longer premolars (Fig. 4).

The occurrence of a second Boselaphine tentatively referred to the genus Miotragocerus at Perivolaki and at Nikiti introduces an intriguing hypothesis (Kostopoulos, 2006 and Kostopoulos, 2016). According to Kostopoulos (2006), some cranial materials from Perivolaki are determined to belong to Miotragocerus sp.; however, the possibility that some females of Tragoportax also had horns leaves the taxonomic attribution open. In addition, some horn-core features considered useful for the distinction between T. rugosifrons and T. amalthea may also be due more to ontogenetic variability than to interspecific differences (Kostopoulos, 2006, Kostopoulos, 2016 and Kostopoulos, 2009). Specimens from Nikiti, which were dubitatively attributed to ?Miotragocerus sp., show features that are not decisive for their attribution to this genus (Kostopoulos, 2016). These specimens show a frequently molarised p4, and their dental measurements and proportions fall within the variability of T. amalthea from the same site (Kostopoulos, 2016) and of T. rugosifrons from Perivolaki (Fig. 4). Therefore, open questions remain due to the uncertain taxonomic position of these specimens, together with the suspicion of a significant incidence of sexual dimorphism and ontogenetic stages for the characters of Tragoportacini that are presently considered taxonomically distinctive.

Compared to Tragoportax amalthea from Maragheh, Iran (T. cf. amalthea in Kostopoulos and Bernor, 2011, T. amalthea in Kostopoulos, 2016), the p4 of the specimen from Cessaniti seems less advanced, as it shows a protruding, but slightly antero-posteriorly extended, metaconid and an open anterior valley. The dental sample from Maragheh, Iran (Age: 8.16–7.68 Ma) is considered more advanced than those of T. amalthea from Pikermi and T. rugosifrons from Hadjidimovo, while it is comparable to T. rugosifrons from Perivolaki (Kostopoulos and Bernor, 2011; Fig. 4). The marked T or Y shape of the p/4 metaconid, as well as the close anterior valley, as observed in the sample from Maragheh, are not present in the specimen from Cessaniti.

The specimen from Cessaniti is comparable to Tragoportax rugosifrons from the Macedonian sites of Ravine des Zouaves 5 (about 8.2 Ma; Koufos, 2006 and Sen et al., 2000), Vathylakkos 2 and Prochoma (the latter are dated back to about 7.5 Ma, Koufos, 2006 and Sen et al., 2000) (Bouvrain, 1994). The horizontal ramus of the mandible is higher under the molars than under premolars in the Cessaniti and the Macedonian specimens (Bouvrain, 1994). The lower molars from Ravine des Zouaves 5, Vathylakkos 2 and Prochoma, and Cessaniti show the presence of strong ectostylids. The metastylid of m3, absent in the specimen from Cessaniti, is also often absent in specimens from the three Macedonian sites (Bouvrain, 1994). The wear stage of the specimen from Cessaniti does not allow verification of the presence of a strong mesostylid or of a feeble goat fold, as these characters are not observable in the inferior 3/4 of the crown (according to Bouvrain, 1994). Bouvrain (1994) reports a certain variability in the premolar characters. The fourth premolar from Cessaniti has a paraconid that is more developed than the parastylid, as in 64% of the specimens from Ravine des Zouaves (n. 41 specimens) and in 54% of the specimens from Prochoma (n. 11 specimens; Bouvrain, 1994). The third lower premolar from Cessaniti shows a paraconid that is more developed than the parastylid, whereas in the samples from Macedonian sites, this morphology is present in 42% (Ravine des Zouaves, n. 27 specimens) and 22% (Prochoma, n. 2 specimens; Bouvrain, 1994). The tooth dimensions and proportions of the specimen from Cessaniti are within the variability of the Macedonian samples (Bouvrain, 1994; Fig. 4).

T. cyrenaicus (formerly Miotragocerus cyrenaicus) has been found at Sahabi (Libya, 6.7 Ma; Bernor and Rook, 2008; Lehmann and Thomas; Thomas, 1979). Lehmann and Thomas (1987) pointed out that this species is larger than T. rugosifrons and shows marked differences in its lower dentition. However, the differences indicated by Lehmann and Thomas (1987) (paraconid and parastylid distinct from parastylid on premolars and not sub-equal in size, open valley between paraconid and metaconid, metaconid and entoconid not fused) are among the characters emphasised by Spassov and Geraads (2004) for distinguishing the genus Tragoportax, and they are present in T. rugosifrons. The metaconid is T-shaped, whereas the metaconid in the specimen from Cessaniti is moderately laterally expanded. In both, the anterior valley is open and the metaconid and entoconid are not fused. The teeth are larger in T. cyrenaicus than in T. from Cessaniti. T. cyrenaicus falls within the variability of T. rugosifrons–T. amalthea (Bouvrain, 1994, Kostopoulos, 2016, Kostopoulos, 2009, Lehmann and Thomas, 1987, Solounias, 1981, Spassov and Geraads, 2004 and Thomas, 1979). The mandibular horizontal ramus is higher in the specimen from Cessaniti than in T. cyrenaicus.

T. abyssinicus is a species recently proposed by Haile-Selassie et al. (2009) on the basis of the Mio-Pliocene materials from the Middle Awash (approx. 5.6 my; Ethiopia). The metrics of its teeth fall within the variability of T. rugosifrons–T. amalthea (Fig. 4). The tooth morphology has similarities with that from T. rugosifrons from Hadjidimovo and from the specimen from Cessaniti: the metaconid of the premolars points backwards; the metaconid is well separated from the entoconid in the premolars; small basal pillars are present in the molars; and deep valleys are separated by cusps. The molars differ by the presence of developed flanges but not so much as to represent a true “goat fold”. In T. rugosifrons, the flanges are very weak.

The spread of Tragoportacini to sub-Saharan Africa probably occurred later than the spread to Eurasia and North Africa (Bibi, 2011). T. acrae (previously Mesembriportax acrae and Miotragocerus acrae; Gentry, 1974 and Gentry, 1980) from Langebaanweg, South Africa (dated about 4.5 Ma), shows more advanced features in its lower dentition: a small goat fold on the lower molars, long premolar rows, and the hypoconid of p/4 sometimes projecting. The teeth are measurably larger than those of other Tragoportax species (Fig. 4).

Post-cranial bones allow few comparisons due to the scanty available data. The characters of the right humerus from Cessaniti differs from that from Alcelaphini (see Gentry, 1974): the dorsal margin of the medial condyle is smoothed and uniform, without interdigitations; the trochlea has a shallow groove.

The radius from Cessaniti differs from that of T. acrae (Gentry, 1974) in terms of the overall shape of the proximal articulation and the lack of a central fossa in its central part. The distal end is abraded, but the distal facet for the cuneiform seems to be reduced, as in T. rugosifrons (Buvrain, 1994), rather than wide like in T. acrae (Gentry, 1974). The distal end of the radius from Cessaniti shows flat facets. The measurements of the metatarsal are within the variability of T. rugosifrons of Hadjidimovo (Spassov et al., 2004).

The specific attribution of the Tragoportax specimens from Cessaniti is difficult because of the limited taxonomic significance of teeth. Moreover, some species have been moved and/or synonymised into the genus Tragoportax, adding to the complexity of the picture and often based on scanty materials.

The lower dentition of the Cessaniti mandible is quite comparable to that of the specimens of T. rugosifrons from Hadjidimovo, Bulgaria (Spassov and Geraads, 2004), and from some Macedonian sites (Ravine des Zouaves 5, Vathylakkos 2 and Prochoma; Bouvrain, 1994), whereas it differs from that of Tragoportax amalthea from Nikiti 2, Greece (Kostopoulos, 2016) by having a shorter premolar row. However, a trend towards the molarisation of p/4 and an elongation of the premolar row has been detected in T. rugosifrons from Perivolaki, Greece (Kostopoulos, 2006). For this reason, the variability in the lower dentition of T. rugosifrons shifts into the variability of T. amalthea, which is characterised by longer premolar rows (Fig. 4). Molarised p4 are present in some of the specimens from Nikiti that are dubitatively attributed to?Miotragocerus sp. (Kostopoulos, 2016). The tooth proportions of the specimens of ?Miotragocerus sp. from Nikiti overlap those of T. rugosifrons from Perivolaki (Kostopoulos, 2016; Fig. 4) and T. amalthea from Nikiti (Kostopoulos, 2016; Fig. 4), in contrast to the marked separation of tooth proportions and dimensions usually observed between the two Tragoportacini genera (Kostopoulos, 2009). The complex picture of morphometric variability and the taxonomic assessment of the species T. rugosifrons and T. amalthea remain unclear. T. cyrenaicus also falls into this wide variability, as does T. abyssinicus, whereas the more derived species, T. acrae from South Africa, shows a clear trend toward larger size.

As noted above, tooth morphology is not particularly conclusive. The comparisons of morphological characters and biometrical data, however, suggest that the specimen from Cessaniti can be assigned to the genus Tragoportax, while its specific attribution to T. rugosifrons remains provisional. The specimen from Cessaniti is, by and large, comparable to the less derived forms of Tragoportax rugosifrons, while it differs from T. amalthea. By contrast, comparisons with T. cyrenaicus can only be based on scarce materials and are unreliable; albeit, the tooth measurements are larger in T. cyrenaicus than in the specimen from Cessaniti. For these reasons, the specimen from Cessaniti can be considered conformis to T. rugosifrons.

The genus Tragoportax was common in Eurasia and Africa during the late Miocene (Bibi, 2011 and Kostopoulos, 2009), so the occurrence of T. cf. rugosifrons at Cessaniti confirms that the assemblage from Cessaniti represents a peculiar bioprovince in the late Miocene of the central Mediterranean, including North African species associated with Pikermian (Greco-Iranian bioprovince) ones. Cessaniti probably records the westernmost occurrence of some taxa belonging to the Pikermian biome. The extent of the emerged lands of Cessaniti and their possible connection to North Africa are still not fully elucidated by geological studies.