The Cetopirus specimen here studied (MSNUP I16903) was collected by A.V. from a rocky cliff (known as “Il Fascio”) outcropping in the area of Porto Craulo, inside the port of Otranto (Salento, southern Italy) (Fig. 2a). The rocks exposed at Il Fascio belong to the Uggiano La Chiesa Formation (UCF), a Plio-Pleistocene unit deposed within the Adriatic foredeep (Bossio et al., 2002).

From a lithostratigraphic viewpoint, in southeastern Salento the UCF mainly consists of fossiliferous, well stratified biodetritical limestones and yellowish calcareous sands and marls deposed in an open shelf environment (Bossio et al., 1993). The benthic assemblages indicate typical depths of the inner neritic zone, although slightly greater depths have been recorded occasionally (Bossio et al., 2005). In the Otranto area, the UCF shows a maximum thickness of ca. 35 m and ranges from the Uppermost Piacenzian stage (D’Alessandro et al., 2004) to the basal part of the Calabrian stage (Bossio et al., 1987).

At the port of Otranto, the UCF is well exposed. In particular, the outcrop of Il Fascio (Fig. 2b) is characterized by levels of marlstones which alternate with intensively bioturbated, decimetric calcarenitic layers. Main fossils are decapod crustaceans (including Lobocarcinus sismondai) and ichthyoliths (Menesini, 1967, Varola, 1965 and Varola, 1981); they are common and particularly concentrated in the marly levels. Brachiopods, echinoids, bivalves, gastropods, corals, and turtle remains have been also recognized. Whale barnacle shells and isolated plates belonging to the fossil species Coronula bifida are known from three marly beds sited near the top of the succession (Fig. 2c); the specimen here studied was found in the middle marly bed.

For more details on the geological setting see the supplementary online material file Cetopirus_ESM.pdf.

The assemblage observed in the sample contains: common to abundant specimens of small (between >2 and <7 μm) unidentified Reticulofenestra; few specimens of Calcidiscus leptoporus, Calcidiscus macintyrei, Coccolithus pelagicus, Cricolithus jonesi, small (<3 μm) unidentified Gephyrocapsa with closed central area, Helicosphaera carteri, Pontosphaera japonica, Pontosphaera sp., Pseudoemiliania lacunosa, Pseudoemiliania ovata, Syracosphaera histrica; rare specimens of Braarudosphaera bigelowii and Helicosphaera sellii.

Based on the occurrence of Calcidiscus macintyrei and the absence of Discoaster brouweri, the analyzed assemblage was attributed to the Calcidiscus macintyrei subzone of Gartner (1977); furthermore, the absence of large Gephyrocapsa oceanica, allows the identification of Mediterranean Neogene Nannoplankton zone 19a of Rio et al. (1990). The sample is dated to the late Gelasian-early Calabrian (early Santernian substage of the ‘Italian Marine Stages’ regional scheme, following Gibbard and Cohen (2008)), with absolute ages ranging from 1.95 Ma to 1.73 Ma as reported by Raffi et al. (2006), respectively for the highest occurrence of D. brouweri and the lowest occurrence of large G. oceanica in the Mediterranean area.

Class Maxillopoda Dahl, 1956

Subclass Cirripedia Burmeister, 1834

Order Sessilia Lamarck, 1818

Suborder Balanomorpha Pilsbry, 1916

Family Coronulidae Leach, 1817

Genus Cetopirus Ranzani, 1817

Type species: Lepas complanata Mörch, 1853, Late Pleistocene to Recent.

Distribution: Early Pleistocene to Recent. Known as fossil from Lower Pleistocene deposits of Italy, Upper Pleistocene deposits of Spain, and Holocene deposits of Argentina and the Netherlands. Currently known as an exclusive symbiont of Eubalaena spp., living in temperate seas.

Emended diagnosis: Body within a dome-shaped shell, consisting of 6 sub-equal compartments; circumference circular in apical view; rounded-hexagonal orifice of the body chamber, not larger than the basal opening; opercular valves present, much smaller than the orifice; sheath short and smooth, whose basal edge does not project freely; ala square and thin; compound radius moderately thick to very thick, whose closely spaced, completely branched sutural septa are inclined with respect to the true radius and originate from a main septum running along the outer edge of the radius; external radius rather narrow and transversely striated; paries thin, ornamented with broad longitudinal ribs having simple, T-shaped terminations; terminal transverse loops of the ribs flattened towards the outside and affected by longitudinally elongated tubes or tubules; core of the ribs solidly calcified, no tubes or tubules present in the core of the ribs outside the lateral lobes of the terminal transverse loops of the ribs; ribs flattened, ornamented by transverse lamellae (externally) and weak longitudinal striae (both externally and internally); apex of the shell presenting four ribs forming three cavities; secondary branching very symmetrical and frequent, occurring near the apex of the shell and causing the basal edge of each plate to present a treelike pattern.

Cetopirus fragilis, sp. nov. (Fig. 3 and Fig. 4)

Holotype and only referred specimen: MSNUP I16903, a complete left compartment (latus or carinolatus).

Type horizon and locality: Lower Pleistocene marlstones at Il Fascio (Latitude: 40°09’09” N; Longitude: 19°09’29” E), Porto Craulo (Otranto, Apulia region, Italy), characterized by the presence of fossil remains of Coronula bifida. The type horizon, which is part of the Plio-Pleistocene Uggiano La Chiesa Formation, is attributed to the Mediterranean Neogene Nannoplankton zone 19a sensu Rio et al. (1990), that is, to the 1.95 Ma - 1.73 Ma time span (according to Raffi et al., 2006).

Derivation of the name: From the Latin “fragilis” (fragile), considering its rather thin radius with respect to the very thick radius of the extant species Cetopirus complanatus.

Diagnosis: Radius moderately thick, able to fill only the external half of the cavity between two neighbouring plates; terminal transverse loops of the ribs whose lateral lobes present subcylindrical tubules affecting the solid surface originated by the calcification of the space sited between the opposite inner faces of the inner lamina.

Description, remarks and comparisons: This medium-sized specimen (Fig. 3) is very well preserved, and various details of the shell ornamentation are exquisitely observable. A weak degree of erosional smoothing is limited to the upper, more convex part of the compartment; fine features of the ribs are instead perfectly preserved in the lower part of the compartment. In fact, the upper part of the specimen corresponds to the portion of the shell which in life emerged from its host's skin, thus explaining the moderate smoothing there observed (Bianucci et al., 2006a and Dominici et al., 2011).

The well developed treelike pattern of the basal edge of MSNUP I16903 means that it belonged to a still-growing mature individual rather than to a juvenile (Seilacher, 2005), as also the overall size of the plate indicates.

Although the type material consists just of a single compartment, various features which on the whole characterize the well recognizable genus established by Ranzani (1817), distinguishing it from the other whale barnacles, have been detected by us; some of these features are, to our knowledge, exclusive of Cetopirus. The observed characters include: •

the flat and broad ribs, which bifurcate abundantly and symmetrically (Darwin, 1854), thus generating the treelike pattern of the basal edge of the shell (Seilacher, 2005);

According to Scarff (1986), the coronulid species most similar to Cetopirus complanatus is Coronula reginae. MSNUP I16903 differs from C. reginae, the former suggesting a significantly convex shell, the latter being typically conical and straight-shaped outside (Darwin, 1854); moreover, the compound radius of MSNUP I16903 is remarkably thicker and narrower than in C. reginae. Anyway, the compound radius of MSNUP I16903 is significantly thinner than observed in any C. complanatus specimen (in which the radius virtually fills the cavity between neighbouring compartments, Pastorino and Griffin (1996); see Fig. S3 of the supplementary online material file Cetopirus_ESM.pdf) and, in its middle portion, it does not reach the basis of the sheath by about half the thickness of the compartment. Darwin (1854) observed that in the genus Cetopirus the terminal transverse loops of the ribs are remarkably flatter towards the outside compared to other whale barnacles. In C. complanatus these loops are affected by thin, well-spaced and rather symmetrical planar septa which connect the opposite inner faces of the inner lamina, thus defining tubes whose transverse section is sub-rectangular and rather wide (Fig. 4a; Darwin, 1854). These septa affect even the very central portion of the loop. On the contrary, MSNUP I16903 displays a few very fine tubules affecting only the lateral lobes of the loops, far from the medial plane of symmetry of the rib (which instead presents a tear-shaped, smooth and septa-free area). These distal tubules (Fig. 4b) are roughly cylindrical and thick-walled (namely, they can be described as asymmetrical and irregular cylindrical openings within a solid surface originated by the calcification of the space sited between the opposite inner faces of the inner lamina).

The fossil coronulid species Cetolepas hertleini (from the Lower Pleistocene of the San Diego Formation, California, USA) shows striking similarities with Cetopirus complanatus. MSNUP I16903 mainly differs from Cetolepas hertleini by showing: •

a significantly convex sutural edge of the radius (instead of a simple, linear one);

The coronulid genus Cryptolepas includes two species: the extant C. rhachianecti and the fossil C. murata; the latter presents strong affinities with both Cetopirus complanatus and MSNUP I16903 (Zullo, 1961). However, MSNUP I16903 presents a moderately thick compound radius, while Cryptolepas spp. presents a thick, C. complanatus-like radius (Davis, 1972 and Pilsbry, 1916). Moreover, three features associate MSNUP I16903 to C. complanatus while distinguishing them from both C. murata and C. rhachianecti. First, the texture of the sutural edge of the radius, which presents closely spaced, completely branched sutural septa that project inward and downward and originate from a main septum running along the outside edge of the radius; in turn, in Cryptolepas spp. the pattern is crenulate and originates from a central septum running along the inner edge of the radius (Davis, 1972 and Zullo, 1969). Second, the core of each of the ribs (”buttresses proper” sensu Davis (1972)) is solidly calcified (with the important exception of the terminal transverse loops of the ribs, which show the above reported tube-like features), while Davis (1972) and Zullo (1961) report on longitudinal septa, forming longitudinal tubes within the largely uncalcified buttresses proper of Cryptolepas spp. Third, the sheath is remarkably smooth, while it is affected by deeply incised transverse grooves in Cryptolepas spp. (Zullo, 1961 and Zullo, 1969). Finally, both C. complanatus and MSNUP I16903 strongly differ from C. rhachianecti by presenting T-shaped terminations of the ribs which build up a complete outer wall (see Pilsbry, 1916).

As reported above, MSNUP I16903 belongs to the genus Cetopirus owing to a number of features which distinguish the genus established by Ranzani (1817) from the other coronulids; anyway, the above mentioned differences between MSNUP I16903 and the recent species C. complanatus (i.e., the presence in MSNUP I16903 of a thin radius and a peculiar inner structure of the terminal transverse loops of the ribs) allow and recommend us to establish a new species – Cetopirus fragilis – within the genus Cetopirus. Since the coronulid Bauplan - paraphrasing Darwin (1854): six compartments of equal size, having the same outline, arranged in a crown-shaped, more or less depressed, shell - is well known and suffers no remarkable variations among the various species of Coronulidae, a complete single plate is enough to provide a reliable reconstruction of its complete skeleton and consequently to describe a new taxon.

The fossil record of the whale barnacles is quite fragmentary and localized in few areas of the world. In the Mediterranean Basin, fossil whale barnacles are known from various Pliocene to Pleistocene Italian and Cypriot shallow marine deposits (De Alessandri, 1895, De Alessandri, 1906, De Gregorio, 1895, Dominici et al., 2011, Menesini, 1968 and Seguenza, 1873). This record also includes some C. bifida specimens from the Uggiano La Chiesa Formation (Bossio et al., 1993). However, only whale barnacles belonging to the genus Coronula (namely: C. bifida, C. diadema, and the dubious C. (Flabelcorona) ficarazzensis) have been recorded from the Mediterranean Basin to this day, with the exception of some fragments of Tubicinella major and Cetopirus complanatus found in the Late Glacial settlement of Nerja Cave (South Spain) which have been regarded as an indirect evidence of human feeding on a displaced southern right whale (Álvarez-Fernández et al., 2014). The only other fossil records of Cetopirus complanatus published to this date are from Holocene deposits of Argentina (ca. 7500–2000 years ago; Pastorino & Griffin, 1996) and the Netherlands (10th century A.D.; Holthuis et al., 1998). Therefore, the Early Pleistocene plate here described (MSNUP I16903) allows us to extend back the global fossil record of Cetopirus for more than 1.7 My; moreover, it represents the first record of Cetopirus for the Mediterranean fauna before the Late Glacial period.

Considering that: •

today Coronula spp. live preferentially on the skin of the humpback whale Megaptera novaeangliae;

Now, MSNUP I16903 suggests that the diversity of the migratory mysticetes in the Mediterranean during the Pleistocene was greater than previous suggested by the fossil record of Coronula spp. In fact, we propose that MSNUP I16903 was hosted by a balaenid whale, since Cetopirus complanatus is currently known as an exclusive symbiont of both the North Atlantic and the southern right whales and considering that the general high host specificity of the whale barnacles is believed to date to the Pliocene (e.g., Bianucci et al., 2006a and Dominici et al., 2011). We suggest that a right whale migratory route was active between the Mediterranean (the putative breeding ground) and the Northeast Atlantic (the putative feeding ground) during the Plio-Pleistocene. In this regard, the presence in Italian Pliocene deposits of: •

at least three genera of balaenids (including Eubalaena, see Bisconti, 2002 and Bisconti, 2009);

The Plio-Pleistocene balaenid migratory path here proposed (Fig. 5) could represent the fossil analogous of the historical (and possibly no longer extant) Eubalaena glacialis migration route spanning between the Biscay Bay and the Northeast Atlantic (Brown, 1986). In this framework, the modern occurrence of the North Atlantic right whale in the Biscay Bay wintering area (Aguilar, 1986) could be a recent product of the northward repositioning of the Gulf Stream which is believed to have occurred during the Last Glacial Termination and, possibly, during all other Late Quaternary deglaciation phases (Keffer et al., 1988).

MSNUP I16903 also indicates the persistence of Balaenidae in the Mediterranean Basin during the Early Pleistocene, at least until the Gelasian-Calabrian transition which corresponds to the Mediterranean Neogene Nannoplankton zone 19a sensu Rio et al. (1990) spanning in time from 1.95 Ma to 1.73 Ma (according to Raffi et al., 2006). Until today, fossil balaenids from the Mediterranean were reported from Upper Miocene to Pliocene deposits (Bianucci and Landini, 2007 and Bianucci et al., 2009). The coexistence of both Coronula and Cetopirus (documented by fossil shells of Coronula bifida from the horizon in which the holotype of Cetopirus fragilis was found) suggests that Balaenidae and Balaenopteridae shared the same breeding ground: the shallow sea covering the easternmost part of Salento during the Early Pleistocene. This observation finds an analogous in the distribution of the extant Eubalaena japonica, which is believed to breed and calve south of Baja California (Gendron et al., 1999 and Gilmore, 1956), just as the preferential host of Coronula spp., the rorqual Megaptera novaeangliae, does (Gendron and Urban, 1993 and Scammon, 1874). A similar sympatric distribution during the cold breeding season (a most unusual case among extant mysticetes) has been hypothesized for the Hawaiian waters (Herman et al., 1980). In this respect, it's noteworthy that E. japonica and M. novaeangliae have occasionally been observed while interacting in their winter low-latitude grounds (Salden and Mickelsen, 1999 and Scarff, 1986).

The fossil coronulids from Otranto suggest the preservation of a rather high baleen whale biodiversity in the Mediterranean during the Early Pleistocene, despite the scarcity of fossil mysticetes (Bianucci et al., 2009). Recent findings from South Italy, comprising two mysticetes from the Lower Pleistocene beds of Cutrofiano (Margiotta and Varola, 2007) and a large balaenopterid (approaching the size of Balaenoptera musculus) from the Calabrian deposits of Matera (Bianucci et al., 2012), partially fill the gap in the Pleistocene fossil record of the Mediterranean baleen whales, thus supporting the continuation of a diverse mysticete fauna.

Finally, it should be noted that, in contrast to some tens of specimens of Coronula spp., only one isolated specimen belonging to the genus Cetopirus has been collected from the Lower Pleistocene deposits of the Mediterranean Basin. This observation does not necessarily indicate that the abundance of Balaenopteridae was much greater than that of Balaenidae: in fact, field research suggests that, while virtually every extant humpback whale hosts at least some Coronula diadema individuals (e.g., Nishiwaki, 1959), reaching eventually more than 450 kg of coronulid hosts (Fertl, 2002), on extant right whales the frequency and abundance of Cetopirus complanatus seem to be much lower, and roughly comparable to those of the more host-generalist symbiont Coronula diadema (Scarff, 1986).

This work provides the first unambiguous report of Cetopirus prior to the Late Glacial Period, thus greatly extending back the stratigraphic distribution of this genus and significantly contributing to expand our scarce knowledge about the fossil history of the whale barnacles. On the basis of some anatomical differences (regarding the thickness of the radius and the inner structure of the terminal transverse loops of the ribs) with the Recent species Cetopirus complanatus, MSNUP I16903 was here described as the holotype of the new fossil species Cetopirus fragilis.

Since Cetopirus is currently regarded as exclusively inhabiting the skin of the balaenids, the Cetopirus fragilis specimen MSNUP I16903 strongly suggests the persistence of at least one population of Balaenidae in the Mediterranean Basin during the early Pleistocene, as well as the Coronula bifida remains suggest with regard to Balaenopteridae. Therefore, since Cetopirus is currently known as a symbiont of the North Atlantic right whale, MSNUP I16903 supports the existence of a baleen whale migratory route active between the Mediterranean and the North Atlantic during the Plio-Pleistocene.

According to Dominici et al. (2011), the fossil coronulid Coronula bifida is the direct ancestor of the most common extant species Coronula diadema; a similar model could possibly be suggested for Cetopirus fragilis - Cetopirus complanatus. No phylogenetic analyses comprising both extant and fossil coronulid taxa have been published to this date, and the evolutionary relationships between the extant whale barnacles and their Plio-Pleistocene relatives is unknown and largely speculative. Future work should undertake comprehensive phylogenetic analyses of Coronulidae that would further enhance our understanding of the evolutionary links between baleen whales and their barnacles during the Late Neogene and the Quaternary.

A.V. collected the specimen MSNUP I16903. A.C., M.B., and G.B. undertook the systematic study and elaborated the palaeobiogegraphical hypotheses. S.M. provided the geological setting and R.C. the biostratigraphical framework. A.C. wrote the paper and prepared the illustrations. All authors discussed the results and commented on the manuscript at all stages.