The age of the fossil mammals of the Tusco-Sardinian PB cover a time span of ∼1.6 Ma (∼8.3–6.7 Ma; Fig. 2) (Benvenuti et al., 2015 and Rook et al., 2011; see also Cirilli et al., 2016). The palaeobioprovince is, however, much older, and late Miocene palaeocommunities were the result of several dispersal events. The last two occurred: (1) at ∼8.3–8.1 Ma, age of the V0 assemblage, attested by the presence of a non-endemic European murid (Engesser, 1989), and (2) after ∼7.1–6.7 Ma, age of the V2 assemblage, attested by the presence of new taxa of European affinity (Benvenuti et al., 2001 and Benvenuti et al., 2015). Older dispersal events from Europe, and probably from Africa, are attested by other endemic taxa (Abbazzi et al., 2008). Nevertheless, until now, it was impossible to date them and, consequently, to understand when the Tusco-Sardinian PB first became isolated.

A revision of the stem lagomorph Paludotona provided a minimum age constraint for the onset of insular conditions in the Tusco-Sardinian PB. The continental ancestor of Paludotona was individuated in a European species not younger than MN1–2 (Angelone et al., 2017). As lagomorphs seem incapable to actively cross water straits (Angelone, 2007), MN1–2 may represent the upper constraint for the age of the separation of the Tusco-Sardinian PB from continental Europe, or the age of the (?)last landbridge connection to Europe before MN13.

During the Miocene, the Tusco-Sardinian PB received dispersals from mainland, as the composition of the assemblages attests to a polyphasic origin (see Masini et al., 2008 with references). Nevertheless, no other lagomorph apart Paludotona is present in its fossil record. Particularly remarkable is the absence of Prolagus, a very widespread genus in European Miocene (López Martínez, 2001) and possibly a pioneer taxon, capable to disperse in insular domains immediately after an emergent land connection is available (see Sections 3.3 Mainland Italy and 4.2 Sardinia for examples relative to Italy). If this may indicate that the Tusco-Sardinian PB was not connected to European mainland through a landbridge connection (= corridor dispersal sensu Van der Geer et al., 2010) since MN1–2 until MN13 (= at least 13 Ma), is a hypothesis that it will be worth to test. When Tuscany was joined to the Italian mainland (∼6.5 Ma), new lagomorphs dispersed in the area (Section 3.3).

The geographical distribution of the Paludotona-bearing fossil sites is peculiar, as they are concentrated in modern maritime Tuscany only. No report of Paludotona is available from Fiume Santo, a fossil site located in Sardinia, i.e. the western side of the Tusco-Sardinian PB. The fossil content of Fiume Santo stands out also for its rodent composition (Casanovas-Vilar et al., 2011). This pattern can be explained by an ecological difference between the two sides of the palaeobioprovince, but Angelone et al. (2017) argued that a more suitable hypothesis is to imagine the Tusco-Sardinian PB as divided into two subdomains (Fig. 1A): (1) a Sardinian subdomain corresponding to modern Sardinia + Corsica, and (2) a Tuscanian subdomain comprising maritime Tuscany. The geographical hindrance between the two subdomains that prevented the dispersal of lagomorphs and made the interchange of rodents difficult, in spite of the geographical closeness, may have been related to the opening of the Tyrrhenian Sea. According to Mauffret et al. (1999), the opening of the northern Tyrrhenian Sea started in the Oligocene–early Miocene, coeval with the rifting of the Ligurian-Provençal Basin; it stopped almost immediately after, and then restarted again in the late Tortonian (Moeller et al., 2014). This is congruent with the existence of a seaway that encumbered lagomorph dispersal through the entire palaeobioprovince since its establishment as an isolated domain.

Further insights about the palaeogeography of the Tuscanian subdomain emerged after the study of the phylogeny and evolution of Paludotona. Actually, the revision of the genus (Angelone et al., 2017) highlighted some apparent inconsistencies: (1) Paludotona etruria from local biochron V1 is more advanced and larger than the younger species Paludotona minor recovered from local biochron V2; (2) the oldest known specimens of the genus, classified as P. aff. minor, were more similar to the younger ones (P. minor, V2) than to those that should have been the “intermediate” stage (P. etruria, V1). In a small insular area as the Tuscanian subdomain, and in a limited time span as the one covered by the available findings (∼1.6 Ma), this can be explained only assuming the fragmentation of the domain in an archipelago, in which the different islands experienced episodes of connection and subsequent isolation that led to allopatric speciation. The hypothesis of a Tusco-Sardinian palaeoarchipelago was briefly hinted by Engesser (1989), though never developed or tested.

The age(s) of the faunal dispersal(s) into the Abruzzi–Apulia PB still is a matter of discussion (De Giuli et al., 1987, Freudenthal and Martín-Suárez, 2010, Freudenthal et al., 2013, Masini et al., 2002, Masini et al., 2010, Masini et al., 2013, Mazza, 2013, Mazza and Rustioni, 2008, Patacca et al., 2008a, Patacca et al., 2008b, Savorelli et al., 2017 and Van den Hoek Ostende et al., 2009 with references).

As for lagomorphs, the hypothesis of a Balkan origin of the insular endemics from the Terre Rosse of Gargano (Prolagus apricenicus and Prolagus imperialis; Mazza, 1986) was validated both phylogenetically and palaeogeographically: •

cladistic analyses confirm that P. apricenicus and P. imperialis pertain to an eastern European clade that dates back at least to the early part of the late Miocene (Angelone and Veitschegger, 2015 and Angelone et al., 2015);

The inferred age of dispersal is consistent with the lack of lagomorphs in the oldest localities of Gargano (Masini et al., 2013). Thus, among the models proposed for the origin of the Gargano Terre Rosse faunal assemblages, lagomorphs support the ‘polyphasic’ hypothesis, which assumes that the assemblage was built up after several diachronous dispersal events (Masini et al., 2002 and Savorelli et al., 2017 with references).

Lagomorphs confirm the fragmentation of the Abruzzi–Apulia PB in an archipelago suggested since the earliest papers on the subject (De Giuli et al., 1987, Freudenthal, 1971 and Freudenthal, 1976). The existence of a hypothetical transitional form between P. apricenicus and P. imperialis never found in the fossil record is quite unlikely. In fact, the “sudden” appearance of P. imperialis and its “superposition” with P. apricenicus in youngest infillings (= no substitution of the older species by the younger one) definitively excludes anagenesis. These evidences fit instead with the dispersal of P. imperialis to Gargano from another island of the archipelago. Cladistics analyses indicate P. apricenicus and P. imperialis as sister species (Angelone et al., 2015), a phylogenetic scenario consistent with an allopatric speciation of populations derived from a common European ancestor, due to the fragmentation of the palaeobioprovince in several islands.

In Fig. 1B, we represented the Abruzzi–Apulia PB during MN13 as divided in three main domains corresponding to Abruzzi, Gargano, and Murge. As indicated above, Gargano yields continental vertebrates of MN13 age, thus there is no doubt it was dry land. Tectonic and sedimentological data attest that the Murge area was an emerged land in MN13 (see De Giuli et al., 1987). No studied record of continental vertebrates is available from the Murge area, but Freudenthal (1971, p. 5) affirms that faunas “comparable if not identical” to those of the Gargano assemblages were found “in a limestone quarry between Barletta and Andria, on the Murge highland”. As for the Abruzzi, the only evidence that there still were land mammals in this sector, at least until the early pre-evaporitic Messinian, is the partial skeleton of Prolagus found in shallow marine sediments just above the Bulimina echinata FOD (∼6.8 Ma in the Mediterranean; Mazza et al., 1995). The specimen was attributed to P. cf. apricenicus on a biochronological and palaeogeographical basis, and represents the sole report of Prolagus out of Gargano in the Abruzzi–Apulia PB. Further studies of the specimen may provide data to refine the reconstructions of the Abruzzi–Apulia PB, and verify (1) if Abruzzi was separated from Gargano during the whole Messinian, and (2) if this subdomain may be the source of P. imperialis.

The leporids of the early Pleistocene of Gargano (Section 5.1) are not related to MN13 assemblages. Actually, the latest Miocene assemblages of Abruzzi–Apulia PB became extinct in the earliest Pliocene due to a marine transgression that submerged the area (De Giuli et al., 1987).

The building of the Italian Peninsula dates back to the pre-evaporitic Messinian (MN13). The appearance of eastern and western European taxa in northern and central-western Italy provides indirect evidence of a connection with mainland through a north–west pathway (Fig. 1B). Both leporids and ochotonids were recorded in the oldest post-connection fossil sites of the Italian Peninsula (Angelone, 2007 and Angelone and Rook, 2011), as well as several other land mammal dispersed from mainland Europe, attesting that the land connection was “immediately” configurated as a corridor.

MN13 ochotonids of the Italian mainland are represented by Prolagus sorbinii and related forms (P. cf. sorbinii) found in northwestern and central Italy (Angelone et al., in press). A systematic revision of Prolagus sp. from the post-evaporitic Messinian of Molise (Cosentino et al., 2018) will clarify if the Messinian distribution of P. sorbinii has to be extended also to southeastern Italy. At the present state of the art, the palaeodistribution of P. sorbinii appears disjoint: the only reliable record of P. sorbinii outside the Italian peninsula is from northern Greece, but the locality is younger than Italian reports (Maramena, Messinian/Pliocene boundary; De Bruijn, 1995). Reports of P. sorbinii from the Ruscinian of Moldova and Ukraine (Tesakov and Averianov, 2002) must be reconsidered. The mainstream hypothesis is that P. sorbinii arrived from eastern Europe. The alternative is an “out-of-Italy” hypothesis for the dispersal path of P. sorbinii, which implies a dispersal towards Italy of a non-identified European species in the earliest Messinian, its immediate speciation into P. sorbinii, and its dispersal from Italy to the southern Balkans. It is more parsimonious to hypothesize that P. sorbinii originated in eastern Europe not after the earliest Messinian (due to the presence of a mesial hyperloph on P2, a trans-specific feature not acquired before the Messinian; Angelone and Čermák, 2015). The fact that there is no trace of the westward dispersal path of P. sorbinii may be explained by the poor knowledge of eastern and central European lagomorphs. We exclude that P. sorbinii entered mainland Italy using a “shortcut” through the Abruzzi–Apulia PB. Apart for its connection with the Balkans in the early Messinian (Section 3.2), the Abruzzi–Apulia PB was an isolated domain, and island hopping is not congenial to lagomorphs. Moreover, cladistic analyses evidence that P. sorbinii and Prolagus spp. from Gargano are phylogenetically distant, being placed in different clades (Angelone et al., 2015).

Alilepus, the first leporid to reach Eurasia in MN11 (Flynn et al., 2014), is also the oldest Italian leporid (Fig. 2): Alilepus meini is present in the pre-evaporitic Messinian of southern Tuscany. The erection of this species excluded the presence of Hypolagus in Italy in pre-Pliocene times (Angelone and Rook, 2011). Alilepus meini, as well as other coeval mammals of southern Tuscany, is a continental endemic not inherited from the insular endemic Tusco-Sardinian PB. The MN13 palaeophysiography of central Apennine, characterized by several intermontane basins, explains the high incidence of continental endemisms. Alilepus meini is phylogenetically close to A. turolensis (Angelone and Rook, 2011). The geographical distribution of A. turolensis was recently reconsidered as limited to the MN12–13 of Spain and MN13 of France, as the recorded occurrences in the MN12–MN13 of northern Greece need to be re-evaluated (Čermák et al., 2015 and references therein). Thus, A. meini can be considered a taxon of western European origin.

The oldest Italian report of Trischizolagus, an early leporid of Eurasian distribution, occurs in the post-evaporitic Messinian of eastern-central Italy (De Giuli, 1989), i.e. slightly later than that of Alilepus. Another report of Trischizolagus is recorded at the Mio/Pliocene boundary in Tuscany (Angelone and Rook, 2012). At present, the specific attribution of Italian Trischizolagus is not available, thus its phylogenetic affinities remain problematic. Two possible source species are Trischizolagus crusafonti and Trischizolagus maritsae, from localities coeval or slightly younger than Italian ones (latest MN13 or MN13/MN14; age of La Alberca and Alcoy N: Garcés et al., 1998 and Montoya et al., 2006). Nevertheless, the distribution of the European species is scattered from southwestern Spain to Greece (Čermák et al., 2019); thus the dispersal pathway of Trischizolagus to Italy is unknown.

Undetermined leporids of post-evaporitic Messinian of northwestern and southeastern Italy (Moncucco, Cava Stingeti; Angelone et al., 2011 and Cosentino et al., 2018) may pertain to Trischizolagus, as the distribution of Italian Alilepus is limited to endemic contexts of the pre-evaporitic Messinian of southern Tuscany. Further taxonomic insights are needed to avoid unsupported palaeobiogeographic statements.

The last occurrence of Prolagus sorbinii/P. cf. sorbinii is in the earliest MN14 of Tuscany (Borro Strolla upper synthem). The fossil record of Prolagus has an apparent gap until the appearance of P. aff. sorbinii in the early part of MN16 (Fig. 2) of Tuscany (Arcille), due to the extreme scarcity of early Pliocene continental deposits. The close phylogenetic relationship between P. sorbinii and P. aff. sorbinii excludes another dispersal of Prolagus in the late Pliocene. If P. aff. sorbinii is a synonym of the coeval “P. savagei”, the geographical distribution of the species will encompass northern and central Italy. It is necessary to verify if P. depereti (MN15, southern France) is another synonym of P. savagei (Angelone and Rook, 2012). At the present state of the art, P. savagei/P. aff. sorbinii seems an Italian endemic taxon.

The leporid Hypolagus appears in northwestern Italy in the early Villányian (Arondelli; Berzi, 1967). The sole available tooth was referred to H. petenyii, quite widespread in central Europe during the late Ruscinian–early Villányian (Čermák, 2009). Palaeogeographically, this means that the northwestern passage activated in MN13 continued to be active during the Pliocene, though it is impossible to state if early MN16 represents the exact moment of the attempted dispersal of H. petenyii from central Europe to the Peninsula. The extreme scarcity of Italian material of H. petenyii may suggest that ecological/climatic factors hindered its dispersal.

The oldest Sardinian lagomorphs date back to the earliest part of the late Pliocene (Capo Mannu D1) (Fig. 2). Very likely, the undetermined leporid from Capo Mannu D1 is phylogenetically related to the early Pleistocene Sardolagus obscurus, thus the discussion about the arrival of leporids in Sardinia will be addressed in Section 5.2.

Taxonomic revisions, further validated by cladistic analyses, recognized the other Capo Mannu D1 lagomorph, the ochotonid Prolagus aff. figaro, as a descendant of P. sorbinii. Consequently, its source area should be the Italian peninsula (Angelone et al., 2015). Prolagus aff. figaro is very peculiar, because its morphology is intermediate between its continental ancestor and its direct descendant, the insular endemic P. figaro (Section 5.2). This stage is extremely rare to be recorded, because in an insular environment the characters of the source species are obliterated in a few generations (see Mein's, 1983 biphasic model). Capo Mannu D1 records the “exact” moment of Prolagus dispersal from mainland. Angelone et al. (2015) argue that an emergent landbridge was activated during the early/late Pliocene marine regression.

There is no Pleistocene report of Prolagus in northern Italy (Fig. 2), probably a local effect of the southwards contraction of the geographical distribution of Prolagus, which began in the Pliocene (López Martínez, 2001). In the early Pleistocene, while P. aff. sorbinii persists in Tuscany, a continental endemic Italian species, P. italicus, is recorded in Umbria and Latium (Angelone, 2008b and Bona et al., 2015). Cladistic analyses suggest that P. italicus derived from the basal stock of P. sorbinii, which dispersed into the Italian Peninsula at the beginning of the Messinian (Angelone et al., 2015). If P. sorbinii and P. italicus existed as separate species already since the Messinian, a palaeobiogeographical hypothesis to be tested is that P. sorbinii inhabited northwestern and northern central Italy, whereas P. italicus occupied central and southern central Italy. Further taxonomic insights of Prolagus sp. from the post-evaporitic Messinian of Molise (Cosentino et al., 2018) may shed light on this topic. The reason of such an early divergence among the populations of P. sorbinii that dispersed to Italy is unknown, but the complex Messinian physiography of the Apennine may have played a pivotal role. Prolagus aff. sorbinii occupied part of the distribution area of its ancestor P. sorbinii (Tuscany, but not Piedmont and Romagna), probably due to the above-mentioned southward contraction of the geographical range of Prolagus. Prolagus did not survive after ∼1.5 Ma in continental Italy.

In the earliest Biharian, Hypolagus appears again in Italy (Marchetti et al., 2000). The findings of H. brachygnathus from northeastern Italy (Fig. 1D) fit in with the westward expansion of Hypolagus brachygnathus. This species appears in the late Villányian of Hungary and its distribution area widens at the expense of H. petenyii, reaching western France (Čermák, 2009). Hypolagus brachygnathus is larger and more advanced than H. petenyii and has more cursorially-structured limbs, an advantageous character for the survival in an epoch (Biharian) characterized by a marked shift towards open habitats (Fladerer and Reiner, 1996).

The other Biharian report of Hypolagus in the Italian Peninsula is from southeastern Italy (H. aff. brachygnathus, Pirro Nord, late early Biharian). It belongs to a balanced faunal assemblage established after the sinking of the Garganic portion of the Miocene Abruzzi–Apulia PB and its eventual re-emersion and connection to mainland Italy. The tooth morphology of H. brachygnathus from northwestern Italy fits in with that of early European populations (Angelone, 2013). The tooth morphology of H. aff. brachygnathus, instead, is advanced also compared to that of the European H. brachygnathus, but its tooth size is smaller (Angelone, 2013) and it has a lesser cursorial adaptation (Fladerer and Fiore, 2003), suggesting an origin from a basal stock of H. brachygnathus. The dispersal that led the ancestor of H. aff. brachygnathus to southeastern Italy is very probably distinct from the dispersal related to the northeastern record of H. brachygnathus. In fact, in the early Pleistocene, the Padanian Gulf would have prevented a direct descent of Hypolagus from northeastern to southeastern Italy. The obligatory route to the Italian peninsula was the northwestern passage activated in the early Messinian. The timing of the dispersal of Hypolagus into the peninsula is difficult to constrain, as in late MN17 Hypolagus brachygnathus already reached the eastern side of the Pyrenean chain (Chaline et al., 2000).

Oryctolagus appears in the Italian fossil record in the late Villányian (∼2.3–2.1 Ma; Angelone and Rook, 2012). Oryctolagus valdarnensis, apparently endemic of the Italian Peninsula, is distributed in the late Villányian–early Biharian in northern-central to southeastern regions. Oryctolagus valdarnensis possibly stemmed from O. lacosti, a western European species with isolated findings in Hungary and Greece (Angelone and Rook, 2012 and López Martínez, 2008). Indeed, the appearance of O. valdarnensis occurs slightly later than the first appearance of O. lacosti (∼2.5 Ma). The first occurrence of the genus Oryctolagus in Italy is delayed with respect to Europe. The oldest species of Oryctolagus is O. laynensis (∼3.5–2 Ma), distributed in southwestern Europe plus some reports from the MN15 of Greece (these latter, however, deserve careful revisions) (Koufos, 2006 and López Martínez, 2008). As Italy and Europe were connected since the latest Miocene, the ecological requirements of O. laynensis did not fit in with the environments of the Italian peninsula, or its dispersal was prevented by a local hindrance in northwestern Italy.

The occurrence of O. valdarnensis and H. brachygnathus in the Italian Peninsula follows an opposite pattern with respect to their closest European mainland relatives. In Europe, H. brachygnathus is quite frequent, whereas the distribution of O. lacosti is scattered. By contrast, in Italy, O. valdarnensis is quite common compared to Hypolagus.

The dispersal of Lepus into the Italian peninsula at the end of the early Biharian (L. terraerubrae, now nomen dubium, pers. comm. S. Čermák, 2019; see Angelone et al., in press; Monte Peglia lower levels, ∼1.0 Ma or possibly slightly earlier; Van der Meulen, 1973; see Angelone et al., in press for age reference) is related to the oldest westward migration of Lepus into Europe, started at ∼2 Ma. This event, which can be followed during the Biharian of eastern and central Europe (Averianov, 2001), is probably related to climatic/environmental events. Monte Peglia lower levels record a warm phase (Van der Meulen, 1973) and the slightly older assemblage of Deutsch-Altenburg 4B, where Fladerer (1987) reports L. terraerubrae, is also related to a warm, dry phase (climatic data from Ivanov, 2007).

Several other late Biharian leporids of northeastern Italy plus one in southern Italy (Bon et al., 1991 and Masini et al., 2005) need to be revised to verify their relationship with the oldest Italian Lepus.

The endemic insular Prolagus figaro characterizes the?latest Pliocene–late early Pleistocene of Sardinia. No dispersal event is required to justify the “sudden” appearance of P. figaro in the fossil record as previously thought (López Martínez and Thaler, 1975). In fact, P. figaro is the direct descendant of the earliest late Pliocene P. aff. figaro (Section 4.2), and its absence in the Sardinian fossil record of Prolagus for ∼1 Ma is due to the extreme scarcity of fossiliferous sites of late Pliocene age (Angelone et al., in press).

At ∼0.8–0.7 Ma, Sardinia records a most remarkable species turnover and the appearance of Prolagus sardus (Angelone et al., 2008; age inferred from Palombo, 2009: fig. 2) (Fig. 2). Prolagus sardus is another endemic insular ochotonid of Sardinia (recorded also in Corsica), known since the middle part of the early Pleistocene to the Holocene. The evidence of a direct phylogenetic relationship P. figaro → P. sardus eliminated the long-debated problems of the continental ancestry of P. sardus, of the timing of its presumed arrival in Sardinia, and of the purported presence of a suitable land connection with mainland between Middle and Late Pleistocene.

The?latest Pliocene–late early Pleistocene of Sardinia yielded the leporid Sardolagus obscurus. A direct phylogenetic relation between this taxon and Leporidae indet. from Capo Mannu D1 is the most parsimonious hypothesis to cover the ∼1 Ma gap in the fossil record (Angelone et al., 2018). This means that there are no dispersal events involving leporids between ∼3.6 Ma (Capo Mannu D1) and ∼2.1/1.9 Ma (the onset of the Capo Figari/Orosei 1 faunal sub-complex, in which Sardolagus is first recorded). The ancestor of Sardolagus obscurus underwent such deep changes in the insular domain, that it is impossible to link it to any continental leporid genus. Lacking phylogenetic data, the arrival of the ancestor of Sardolagus in Sardinia can not be temporally framed. Angelone et al. (2018) examined several possibilities: the most probable palaeobiogeographical scenario is a dispersal of Hypolagus from the Italian mainland at ∼3.6 Ma through the landbridge that allowed Prolagus to disperse to Sardinia (Fig. 1C). Indeed, Hypolagus, though extremely rare, was present in Italy in the early part of the late Pliocene (one report, coeval or slightly younger than Capo Mannu D1; Section 6.1).

The sole assemblage referable to the Monte Pellegrino FC (Biharian) is composed by heterogeneous elements, both for their degree of endemism and for their geographical origin (Marra, 2013 and Masini et al., 2008). Part of the assemblage, represented by elements of European and African affinity that arrived in Sicily during the late Miocene, underwent very marked morphological and dimensional modifications due to the permanence in an insular environment. The modifications are so pronounced that it is difficult to individuate the continental source species. Another part of the assemblage is of European origin and shows less marked modifications. The endemic Sicilian leporid Hypolagus peregrinus pertains to this latter group. According to Fladerer and Fiore (2003), H. peregrinus is related to early representatives of the European taxon H. brachygnathus, which dispersed through Italy and eventually to Sicily. The peninsular endemic H. aff. brachygnathus from the late early Biharian of southeastern Italy (Section 6.1) may represent the closest relative of H. peregrinus. Very likely, once they entered Italy, the above-mentioned basal forms of H. brachygnathus underwent different degrees and patterns of morphological changes in response to the different palaeogeographical settings they settled in. For example, in the Italian peninsula, where the gene flow with the European source species was possible but obviously difficult, Hypolagus started to diverge morphologically from European coeval populations, giving rise to a possible new species (H. aff. brachygnathus). However, H. aff. brachygnathus did not develop the morphological and dimensional traits typical of insular endemic mammals as did instead H. peregrinus, the species that developed in Sicily after its connection with mainland was over. The Italian remains of Hypolagus are insufficient to provide a reliable age constraint to the dispersal of Hypolagus either into peninsular Italy (Section 6.1) or Sicily. Fladerer and Fiore (2003) argue that Hypolagus reached Sicily through a sweepstake route, but in our opinion the dispersal of Hypolagus into Sicily is better explained hypothesizing the presence of an emergent land connection between Calabria and Sicily in the latest Villányian–early Biharian (= time-span of the maximum diffusion of H. brachygnathus in Europe).

Ochotona pusilla, an extant Asian taxon typical of cold, dry steppe is recorded in northeastern Italy in the early Toringian and also in the late Toringian (Abbazzi et al., 2000 and Bon et al., 1991). Such reports do not attest to a continuous Toringian record, but represent two different moments of the expansion of the distribution area of O. pusilla related to two different cold phases (MIS13–12/MIS11–10, and probably MIS4). Ochotona pusilla reached the Iberian Peninsula at 63.4 ± 5.5 ky (Laplana et al., 2015), but it has not been recorded in the northwestern part of the Italian Alps. European Alpine mountain ranges could have presented a barrier for the dispersal of the steppe pika. However, during the cold intervals of the Middle–Late Pleistocene, the forest cover was reduced and large areas of the continental shelf were emerged due to the drop of sea level associated with glaciations (Laplana et al., 2015). The northern part of the Adriatic Sea, which emerged during the glacials, represented an easy dispersal path for Ochotona from eastern domains to the southern borders of the Alps.

In the earliest Toringian (∼0.6 Ma), Oryctolagus appears again in the Italian Peninsula after the gap that followed the extinction of O. valdarnensis. The new dispersal is related to a modern stock of Oryctolagus, also encompassing the extant rabbit (Angelone, 2013), and is characterized by the appearance in the western part of northern and central Italy of O. burgi (Nocchi and Sala, 1997). Oryctolagus burgi is an Italian endemic form of quite large size not related to the extant O. cuniculus but phylogenetically closer to O. lacosti (López Martínez, 2008). Its dispersal into the Italian Peninsula occurred from western domains of Europe. Oryctolagus burgi eventually disappeared and, in the Italian record, there is another gap of ∼150 ky (Angelone et al., in press). The extant rabbit, O. cuniculus, appeared in southern Spain at ∼0.6 Ma, associated with a relict faunal assemblage of warmth-loving elements (López Martínez, 2008), and from there invaded France. In the late early Toringian, it is recorded in central Italy (Torre in Pietra lower levels; Caloi and Palombo, 1978). Also for this taxon, the northwestern corridor is the obligatory way to enter the Italian Peninsula. Oryctolagus cuniculus is recorded in several Italian fossil sites until MIS3 (Angelone et al., in press). In France, it disappeared earlier, and was absent from the Late Pleistocene to the Early Holocene (Callou, 2003). At the end of the Last Glacial Maximum, its distribution area was limited to the Iberian Peninsula, and probably southernmost Italy and Sicily (Sections 6.3 Sicily and 6.4 Minor islands).

In the early Toringian, Lepus corsicanus is recorded in northeastern Italy (Visogliano A lower levels; Abbazzi et al., 2000). Lepus corsicanus is linked to the basal radiation(s) of hares in Europe and its extant distribution is a relict one (Alves et al., 2008 and Averianov, 2001). The report of Visogliano is outside the current Italian distribution of the species (central and southern Italy and Sicily), as a consequence of the arrival of L. europaeus, which restricted the area of L. corsicanus (Vismara et al., 2014). A reliable palaeodistribution of early L. corsicanus in Italy would require a revision of early Toringian specifically undetermined remains of Lepus, in order to understand if they are related to L. corsicanus or to L. europaeus.

The oldest reports of Lepus europaeus are difficult to frame: the taxonomic assignment of possible early forms is controversial and the age and correlations with the fossil localities from which they are reported are dubious. The Caucasus seems to be the area from which the species radiated to western Europe. Oldest reports may be of MIS 22–21 (∼800 ky, Akhalkalaki; Tappen et al., 2002; age in Baryshnikov, 2002). During the Last Glacial Maximum, the distribution of L. europaeus shrunk to southern Europe and was fragmented in at least three refugia: the Italian Peninsula, the Balkans, and Anatolia (Fickel et al., 2008). The oldest confirmed reports of L. europaeus in the Italian mainland date back to the earliest late Toringian (the several latest early Toringian remains from Veneto need a systematic revision; Angelone et al., in press). The Italian distribution of L. europaeus in the late Toringian covers the northern and northern-central sectors, whereas its presence in southern-central and southern Italy needs to be confirmed by a revision.

In the late Toringian, Lepus timidus is recorded in northern and central Italy plus one report from southeastern Italy (Angelone et al., in press with references), but, at present, we can not determine if this record represents a continuum or if we are observing different cold pulses. In fact, L. imidus is related to cold climates (tundra and taiga environments; Angerbjörn and Flux, 1995), quite different climatic/environmental requirements than L. europaeus. Angelone et al. (in press) considered dubious Italian reports of L. timidus together with L. europaeus. The oldest European record of the species goes back to the Middle Pleistocene (Sudmer-Berg 2; Koenigswald, 1972). During the Late Pleistocene, L. timidus was widespread from Portugal to western Ukraine (Averianov, 2001). Its dispersal into Italy may have occurred from both northwestern and northeastern passages, but the northeastern passage was possibly preferential, as, in the cold stages of the Late Pleistocene, ample portions of northern Adriatic Sea were emerged land.

During the Toringian, the endemic Prolagus sardus continued to thrive in Corsica and Sardinia, which represent the last refugia of the genus (Angelone et al., 2008).

The connection of Sicily to mainland Italy, occurred during the Last Glacial Maximum (Marra, 2013), allowed the dispersal of several taxa, and of Lepus among them. Those findings are known in the literature as L. europaeus, but they very probably pertain to L. corsicanus (Angelone et al., in press).

Probably, Sicily represented a refugium for Oryctolagus during the Last Glacial Maximum: the presence of O. cuniculus is reported during the transition Castello FC/Holocene (Masini et al., 2008), in levels of supposed tardiglacial age (the layers may be reworked, though; Mangano and Bonfiglio, 1998), and from levels of early post-glacial age (Villari, 1986).

A combination of eustasy and tectonics caused the connection of Capri with mainland, as attested by an early late Toringian (∼250–150 ky) non-endemic, balanced assemblage yielding Oryctolagus cuniculus. Between ∼150–60 ky, the balanced assemblage was replaced by an insular endemic fauna (Capasso Barbato and Gliozzi, 1995).

In other cases, the connections are mostly driven by eustatic factors (Angelone et al., in press with references): •

Palmaria (La Spezia Archipelago): balanced fauna with two/three leporids (possibly Oryctolagus and one/two species of Lepus), attesting to a connection with mainland during a non-identified cold interval of the Late Pleistocene;

In Italy, the number of endemic lagomorphs is particularly high. Among the lagomorphs taxa recorded since the early–middle Turolian to the Toringian, ∼65–70% are endemics (Table 1). In pre-Pliocene times, the majority of the endemic taxa were insular endemics (the early–middle Turolian stem lagomorphs of the Tuscanian subdomain of the Tusco-Sardinian PB: Paludotona etruria, P. minor, P. aff. minor; the Messinian ochotonids of the Abruzzi-Apulia PB: Prolagus apricenicus and P. imperialis), whereas one or two were continental endemics (the leporid Alilepus meini, developed in the early Messinian intermontane basins of southern Tuscany due to the peculiar local physiographic factors, and possibly the late Messinian leporid Trischizolagus sp.). Apparently, Prolagus sorbinii, known in the Italian territory since the earliest phases of its continentalization process, was in pre-Pliocene times the only Italian non-endemic lagomorph, as its geographic distribution encompassed also southern Balkans (see Angelone, 2007 and Angelone et al., 2019 with references). Thus, at the end of the Miocene, endemic lagomorphs were almost 90% (more than 60% insular endemics and about 30% continental endemics) of the total number of species.

After the Mio-Pliocene boundary, insular endemic lagomorphs can be found only in the current Italian main islands (P. aff. figaro, P. figaro, P. sardus, Sardolagus obscurus in Sardinia, Hypolagus peregrinus in Sicily). However, the Italian Peninsula continued to host endemic taxa. In fact, during the Villafranchian (= Villányian and most part of the early Biharian), continental endemic lagomorph species still outnumbered those with a wider geographical distribution. Continental endemics represented 40% of the total number of Villafranchian Italian lagomorph species vs. the 20% represented by non-endemics. This means that continental endemics represented almost the 70% of the species that inhabited the Italian Peninsula during the Villafranchian. Interestingly, non-endemic species (Hypolagus petenyii, H. brachygnathus) occupied the northernmost sectors of Italy, whereas continental endemic taxa were recorded in central and southern Italy (Prolagus italicus, Hypolagus aff. brachygnathus, Oryctolagus valdarnensis; Prolagus savagei/P. aff. sorbinii is considered here as an endemic peninsular Italian species, and possibly a vicariant of P. savagei/P. aff. sorbinii, pending a revision of MN14-MN16 Prolagus of western Europe).

Since the Galerian (late early Biharian + late Biharian + first part of the early Toringian), the lagomorphs of the Italian Peninsula have lost their endemic fingerprint. In fact, the recorded species have a European distribution (Lepus spp., Oryctolagus cuniculus) or even Eurasian ranges (Ochotona pusilla), except for the continental endemic Oryctolagus burgi and Lepus corsicanus, and the endemic insular Prolagus sardus. Endemics at this point represent about 37% of the total number of species (12% insular endemics and 25% continental endemics).

Thus, it is evident that in Italy endemic lagomorph taxa underwent a reduction since the late Miocene. Insular endemic taxa were affected more significantly than continental endemics. The reduction was progressive, but a couple of main watersheds can be individuated at the Mio-Pliocene boundary and at the beginning of the Galerian. In the first case, the profound palaeobiogeographical changes occurred in the Italian territory during the latest Miocene (i.e. the merging of late Miocene isolated domains into a peninsula connected to European mainland) drastically affected the number of endemic taxa, especially insular endemic ones, and explain the drop in the number of endemisms. In the second case, the framework is more complex. The beginning of the Galerian marked the transition to cooler and more arid climates. This led to major changes in vegetation composition and, consequently, to significant faunal turnovers (Masseti and Sala, 2007 and Masseti and Sala, 2011 with references). Thus, indeed the turnover highlighted among lagomorph taxa at the beginning of the Galerian is in line with the well-known Galerian turnover of Italian and European faunas. However, this does not explain satisfactorily the decrease of endemic taxa in the Italian Peninsula. Climate changes affected not only the environment, but also the distribution of emerged lands. In particular, the northeastern passage to Italy became more and more accessible for trans-alpine vertebrates through the Middle and Late Pleistocene (the first steps of the building up of the Po alluvial plain are recorded at the beginning of the Galerian, at ∼1 Ma; Bona and Sala, 2016). This is attested by the noticeable increase of the dispersals into Italy recorded during the Middle and Late Pleistocene (Bona and Sala, 2016 with references), as well as by the homogenization of Italian mammal faunas with those of the European mainland.

Prior to the Pliocene, tectonics was the main factor that controlled lagomorph dispersals into the Italian territory, fragmented in several isolated palaeobioprovinces. Italy started its continentalization process connecting to Europe during MN13. However, since the Messinian and until the early Pleistocene, the only dispersal pathway to Italy was the northwestern passage. For Prolagus, though arriving from southeastern Europe, it was an obligatory passage. For Alilepus and Oryctolagus, which came from western European sectors, the northwestern passage was the “natural” dispersal pathway. It is worth to notice that, in spite of the fact that the pathway was activated during the Messinian, some taxa have been having difficulties to reach the Italian peninsula for several hundreds of thousands years. For example, Oryctolagus appeared for the first time in the Italian Peninsula with a noticeable delay (more than 1 Ma) with respect to Europe. The same occurred with Lepus terraerubrae and, possibly, this was also the case of Hypolagus petenyii. Very possibly, the northwestern passage acted as a physiographic and environmental filter due to its position at the piedmont of the Alpine chain.

The northeastern passage was activated relatively recently, thanks to the accumulation of sediments in the Po plain and to the eustatic variations triggered by climate changes. The northeastern passage was very extended during the coldest phases of the Late Pleistocene and especially during the Last Glacial Maximum, when the northern portion of the Adriatic Sea became dry land. This dispersal pathway progressively gained importance and possibly had a major role for the dispersals, during the Toringian, of Ochotona and Lepus.

Since the Pliocene, Sardinia and Sicily could be reached by lagomorphs only in correspondence of marine regressions. This is also the case of minor Italian islands: the appearance of lagomorphs in their Middle and Late Pleistocene fossil record has to be linked to eustatic variations, with the exception of Capri, where tectonics played a significant complementary role.

As far as lagomorphs are concerned, the Tusco-Sardinian PB and the Abruzzi–Apulia PB were “closed systems”: they received a dispersal from mainland and, after that, speciations and dispersals were restricted within the archipelago. In the Abruzzi–Apulia PB, there was no actual lagomorph turnover, as the younger species, Prolagus imperialis, coexisted with the older one, P. apricenicus. In the Tusco-Sardinian PB, each local biochron is characterized by a different species of Paludotona; however, the temporal successions of the species can not be interpreted as a series of turnovers, as it represents a series of “snapshots” taken at different times and in different parts (= islands, probably) of the Tuscanian subdomain palaeoarchipelago.

During the early Messinian, when the Tusco–Sardinian PB was disrupted, the territories of the Tuscanian subdomain connected to mainland and recorded a turnover related to the arrival of European lagomorphs, whereas Sardinia continued to experience the “closed system” pattern, except for one episode at the early/late Pliocene boundary (arrival of the forerunner of P. figaro and P. sardus). Nevertheless, not being an archipelago, the speciation pattern which characterizes Sardinia since the late Pliocene to the Holocene is different than the one observed in the Tusco-Sardinian PB and the Abruzzi–Apulia PB. Actually, in Sardinia, after the earliest late Pliocene dispersal from mainland, the species replacements in leporids and ochotonids were anagenetic. Climatic/environmental changes could have been the trigger of the replacements, but this causal relationship was tested only in the case of P. figaro → P. sardus.

In the Italian Peninsula, Alilepus, Trischizolagus, and Prolagus did not experience turnovers with European congeneric species, nor with other lagomorphs. Within the genus Prolagus, after the early specific divergence (already in the Messinian?) of P. italicus, an anagenetic substitution occurred during the early Pliocene within the other lineage (P. sorbinii → P. aff. sorbinii).

The turnovers observed among Italian Plio-Pleistocene leporids (Hypolagus, Oryctolagus, Lepus) seem not to be a consequence of the dispersal of congeneric species from Europe. In the case of Oryctolagus and Lepus, the turnovers occurred at the beginning of the Toringian, but the early Pleistocene species (O. valdarnensis, L. terraerubrae) had already became extinct in the late Biharian. In the case of H. petenyii/H. brachygnathus, due to the poor fossil record, it cannot be stated with certainty if the older species was already extinct when the new dispersal into Italy occurred. The Plio-Pleistocene lagomorph turnover also involved the arrival of Ochotona and was very likely related to climatic and environmental changes.

Within the extant species of Lepus, appeared in Italy during the Toringian, it is not possible to individuate actual turnovers, but fluctuations of their palaeodistribution related to climate shifts: L. europaeus replacing L. corsicanus in northern Italy during the late Toringian, and the expansion (probably at the expense of other congeneric species) and the subsequent reduction of the distribution area of Lepus timidus in the late Toringian–Holocene.