Persufflatius renefraaijeni, Bosselaers & Munsterman, 2022

Bosselaers, Mark & Munsterman, Dirk K., 2022, The discovery of a Balaenomorpha (Persufflatius renefraaijeni n. gen., n. sp.) from the upper Miocene of the Netherlands sheds new light on the cranial anatomy of archaic rorqual relatives, Geodiversitas 44 (30), pp. 933-973 : 937-959

Persufflatius renefraaijeni n. gen., n. sp.

( Figs 1-25 ; Tables 1 -6; Appendix 2)

urn:lsid:zoobank.org:act:B1833429-144C-44FF-B7FF-AC45E8552437

HOLOTYPE. — MAB 010293 . A partial cranium consisting of the right half of a neurocranium including the squamosal (without the anterior part of the zygomatic process of the squamosal), the supraoccipital, the exoccipital (the medial part, bearing the condyle and the foramen magnum, is missing), the parietal, a small fragment of the pterygoid, the periotic and the compound posterior process of the petrotympanic. Part of the vertex is preserved, although eroded. The bulla tympani, part of the basicranium, the palate, the suborbital region and the rostrum are missing.

TYPE LOCALITY. — The skull MAB 010293 has been discovered on the spoil piles of the Liessel sand mining (brick producing) company Hoogdonk, in the 1980-90s. The factory was situated at ‘De Brink’, in between the villages Liessel and Deurne , province of Noord Brabant, the Netherlands (51°25’40”N, 5°49’46”E); ( Fig. 1 ).

ETYMOLOGY. — Named after the first and last name of Dr René H. B. Fraaije ( Fig. 6 ), founder and director of the Oertijdmuseum at Boxtel, the Netherlands, who discovered the specimen. In honour of his long-lasting efforts in favour of the study of fossils in general (especially Decapoda) and of the Liessel cetaceans in particular, and for his outstanding efforts to bring particularly (school) children in contact with natural history, c. q. fossils and dinosaurs, in the Oertijdmuseum.

TYPE HORIZON AND GEOLOGICAL AGE. — Borehole B52C1978, interval 0-44.5 m, directly adjacent to the sand pit (at RD-coordinates: (X) 185.627 and (Y) 382.024) was drilled in 2001. The results of the lithostratigraphic and palynological interpretations are reported by Munsterman (2007) (see also: Peters 2009: 98-105; Marx et al. 2016a: 4; Bisconti et al. 2020: 3-5).

LITHOLOGICAL UNIT. — The specimen was dragged from sands assigned to the Diessen Formation. This is a shallow marine deposit consisting of glauconiferous sands, sandy clays and clays. The Diessen Formation is present throughout the Netherlands and includes part of the Dutch late Miocene succession (Tortonian-Messinian) ( Munsterman et al. 2019).

PALYNOFACIES AND AGE- ASSESSMENT ( FIGS 2 ; 3 )

The assemblage is dominated by continental influence (63% of the total sum palynomorphs, i.e., spores, pollen and dinoflagellate cysts). The non-bisaccate pollen group is in minority (23% of the total sum sporomorphs). Most of the sporomorphs are bisaccate pollen (77% of the total sum sporomorphs). Bisaccate pollen are formed by conifers, gymnosperms (Gymnospermae). Bisaccate pollen have a higher aerial and aquatic buoyancy than other sporomorphs, indicating a relatively distal position from the coast. The relatively distal facies is confirmed by the concentration and composition of marine dinoflagellate cysts. The most common genus, Spiniferites (42% of the total dinocyst sum), has a preferential orientation for open marine conditions. Barssidinium graminosum on the contrary, is also well-represented (20% of the total sum dinoflagellate cysts). This taxon has a temperate to tropical distribution in neritic (and especially, inner neritic) waters. Heterotrophic genera like Barssidinium , Selenopemphix and Trinovantedinium (25% of the total dinocyst sum) refer to nutrient-rich water. As a whole, the marine dinocyst assemblage is relatively variegated, indicating nutrient-rich neritic conditions.

An age-diagnostic taxon is Labyrinthodinium truncatum . This taxon has a last occurrence datum (LOD) in the late Miocene, late Tortonian, Zone SNSM14 ( Munsterman & Brinkhuis, 2004). This event is also used to define the DN9 Zone by de Verteuil & Norris (1996) (originally described for the East Coast of the USA and generally adopted in Belgium), in Germany by Köthe (2012) and for the Hystrichosphaeropsis obscura Zone on- and offshore Denmark by Dybkjaer & Piasecki (2010). Marker taxa indicating possible older zones, like e.g. the dinocysts Systematophora placacantha (LOD in Zone SNSM13) and Palaeocystodinum golzowensis (LOD in Zone SNSM12) are absent. The presence of Operculodinium janduchenei with a maximum age range (FOD) in the Tortonian confirms the dating. Also present in the microflora is Achomosphaera andalousiensis , having a slightly older FOD. In conclusion the age assessment of the present assemblage is early late Miocene, late Tortonian SNSM14 Zone, c. 8.1-7.5 Ma ( Munsterman et al. 2019) ( Figs 2 ; 3 ).

DIAGNOSIS

Absolute diagnosis

Small-sized non-balaenid balaenomorph mysticete, differing from all other basal non-balaenid Balaenomorpha (BnBB) in having i) a notably inflated squamosal, in particular the posteriorly oriented postglenoid process, which is also large compared to the size of the neurocranium; ii) a rounded bulbous exoccipital that is confluent with the posterior portion of the zygomatic process of the squamosal ( Duboys de Lavigerie 2020: ch. 74), resulting in an anteroposteriorly long squamosalexoccipital complex; iii) a bell-shaped supraoccipital without an external occipital crest (Bisconti, chs 140, 141, 142; Duboys de Lavigerie: ch. 112), that anteriorly overhangs the temporal fossa; and iv) a very wide, rounded posterior temporal crest (anterior width c. 15 cm). The periotic is Pelocetus -like in general aspect, in having v) a small pars cochlearis, that is not cranially elongated, in having vi) an overall inflated aspect, vii) having a strong and wide, dorsally protruding tuberosity, continuously running over the whole length of the dorsal surface of the pars petrosa and the anterior process. But contrary to the latter, it has (amongst others), viii) the rodlike compound posterior process widely exposed laterally (L: 34 mm) and ix) the cochlear aqueduct, the dorsal vestibular area and the endocranial opening of the facial canal aligned. The estimated bizygomatic width, including the anterior zygomatic process of the squamosal (missing in the holotype) is about 720 mm ( Table 1 ).

Differential diagnosis ( Fig. 7 )

Persufflatius renefraaijeni n. gen., n. sp. is closely related to Pelocetus calvertensis, Uranocetus gramensis and Parietobalaena palmeri in having a rod-like posterior process of the petrotympanic. It differs from eomysticetids, balaenids, Atlanticetus patulus , Diorocetus hiatus and the cetotheriids Brandtocetus chongulek Gol’din & Startsev, 2014, Cetotherium rathkii Brandt, 1843 , Heterocetus affinis and Piscobalaena Pilleri & Siber, 1989 in having a squamosal cleft (Bisconti: ch. 114; Duboys de Lavigerie: ch. 107), and from all known mysticetes except Parietobalaena , in this cleft being ‘smiley’-shaped (semi-circular, dorsally concave). It further differs from eomysticetids in having a more anteriorly projected supraoccipital and parietal; from balaenids in having a concave anterior supraoccipital dorsally (Bisconti: ch. 135; Duboys de Lavigerie: ch. 113), a small but prominent attachment surface (‘paired tubercles’) for the neck muscles on the anterior supraoccipital (Bisconti: ch. 85; Duboys de Lavigerie: ch. 109), a posterodorsally inclined superior part of the posterior wall of the squamosal fossa, and a wide temporal fossa (Bisconti: ch. 110); from ‘ Balaenoptera ’

ryani Hanna & McLellan, 1924 in having a wide semi-circular posterior temporal crest (Bisconti: ch. 107), a posterodorsally inclined superior part of the posterior wall of the squamosal fossa (Bisconti: ch. 113), a parietal that anteriorly overhangs the temporal fossa (Bisconti: ch. 78), and a relatively large periotic having a prominent, deep groove for the tensor tympani muscle (Bisconti: ch. 196; Duboys de Lavigerie, chs 160, 161), the cochlear aqueduct being not confluent with the fenestra cochlearis (Bisconti, chs 215, 216; Duboys de Lavigerie: ch. 174), and a small, not-inflated pars cochlearis (Bisconti: ch. 213); from balaenopterids and eschrichtiids in having the parietal broadly exposed at the vertex (Bisconti, chs 82, 83) and a Pelocetus -like periotic with a pars cochlearis that is not cranially elongated (Bisconti: ch. 211; Duboys de Lavigerie: ch. 144); from Parietobalaena palmeri and P. laxata by the large, strongly inflated and ventrally expanded postglenoid process; from Atlanticetus patulus , Diorocetus hiatus, Pelocetus calvertensis, Tiucetus rosae , ‘ Diorocetus ’ shobarensis, ‘ Diorocetus ’ chichibuensis, Isanacetus laticephalus and Parietobalaena in having a bell-shaped supraoccipital (Bisconti, chs 140, 141, 142; Duboys de Lavigerie: ch. 112), large inflated postglenoid processes and wide rounded exoccipitals, confluent with the squamosal (Duboys de Lavigerie: ch. 74); from Uranocetus gramensis in having a large inflated postglenoid process, anteroposteriorly much longer squamosals (squamosal + exoccipital length: 170 mm) and wide rounded exoccipitals; from Isanacetus , Parietobalaena and ‘ Diorocetus ’ chichibuensis in them having a narrow body of the periotic with no sign of lateral inflation; from Diorocetus hiatus in lacking a ridge posteriorly bordering the facial sulcus on the compound posterior process (Duboys de Lavigerie: ch. 185); from all cetotheriids except Cephalotropis nectus and Joumocetus shimizui in having parietals that are well exposed on the skull vertex (Bisconti, chs 82, 83); from Diorocetus , Uranocetus , Tiucetus and Parietobalaena

in lacking an external occipital crest on the anterior supraoccipital (Bisconti: ch. 81; Duboys de Lavigerie: ch. 114); from all BnBB in having a bell-shaped supraoccipital (except maybe for Uranocetus ( Steeman 2009) ; due to breakage the exact shape of the latter’s supraoccipital is not very clear) (Bisconti, chs 140, 141, 142; Duboys de Lavigerie: ch. 112); differs from Atlanticetus patulus by the deeply excavated tensor tympani fossa (Bisconti: ch. 196; Duboys de Lavigerie: ch. 160), the

different rod-like posterior process (Duboys de Lavigerie: ch. 188), slit-like aqueducts (Bisconti: ch. 217; Duboys de Lavigerie: ch. 172) and the strongly inflated postglenoid process of the squamosal and from ‘ Plesiocetus ’ dubius (IRSNB M 652), ‘ Plesiocetus ’ burtinii (IRSNB M 676), Mesocetus latifrons (IRSNB M 567) and Idiocetus longifrons (IRSNB M 719) in having a small cochlea, a slit-like cochlear aqueduct foramen (except ‘ P ’. dubius) (Bisconti: ch. 217; Duboys de Lavigerie: ch. 172), a very different rod-like compound posterior process (Duboys de Lavigerie:ch. 188) and a strongly inflated postglenoid process of the squamosal. Differs from Atlanticetus patulus , Heterocetus affinis and Cephalotropis nectus in that all three taxa have a (sub-)diamond-shaped compound posterior process of the petrotympanic in ventral view; in addition, the shape of the anterior process of the periotic and the cranial openings of the promontorium also differ (Duboys de Lavigerie, chs 168, 169). The species is smaller than Pelocetus calvertensis, Uranocetus gramensis and Atlanticetus patulus , and larger than Tiucetus rosae ( Fig. 7 ).

DESCRIPTION OF THE HOLOTYPE

A (partial) reconstruction of the holotype MAB 010293 was performed by mirroring the preserved part along the medial axis. Compared to almost all other mysticetes, the postglenoid process, the shape of the exoccipital, the anteromedioventral squamosal and the supraoccipital are more rounded, bulbous and, ‘inflated’ ( Figs 4 ; 5 ).

PRESERVATION, TAPHONOMY, ONTOGENETIC STAGE, BODY SIZE ESTIMATE, FAUNAL ASSEMBLAGE

The specimen shows several recent fractures, most likely the result of dredging. These are: the fractures at the base of the zygomatic process and, anteroventrally, at the squamosal ( Figs 8-10 ) and the pterygoid ( Fig. 11B ), at the posteromedial supraoccipital, and across the cranial cavity, just in front of the squamosal-parietal suture, in the middle of the supraoccipital ( Fig. 8 ). All other damage seems to be due to natural processes. The fossil shows erosion at all edges (all of them being rounded and many of them are polished, exposing deep blue-black apatite). Erosional features are particularly strong at the vertex ( Fig. 8 ), at the posterodorsolateral side of the squamosal, at the supramastoid crest (lateral temporal crest) on the posterior part of the zygomatic process ( Figs 10-14 ) and along the ventral and lateral edge of the postglenoid process ( Figs 12 ; 13 ).

The specimen is to be considered a young adult, judging from the state of merging/fusion of the cranial sutures (see below)( Marx et al. 2016a: 21; Walsh & Berta 2011; Bouetel & de Muizon 2006: 11; MB pers. obs. of several ontogenetic stages of Balaenoptera acutorostrata Lacépède, 1804 ).Moreover,the periotic has impressive medial and anteromedial secondary (hypertrophied) bone growth medial to the dorsal vestibular area. The posterior process of the petrotympanic is mediolaterally wide ( Fig. 12 ). Some sutures on the squamosal surface are not fully merged yet. On the contrary, the “smiley”-shaped squamosal cleft is hard to observe externally, unless viewed in grazing light ( Figs 10 ; 14 ). It is definitely at an advanced stage of merging; in addition this cleft is no longer discernible in internal view.Likewise,the dorsal part of the parietal-squamosal suture and the parietal-supraoccipital suture are in an advanced state of merging, both being almost completely merged(though still discernible as a fine line)( Fig.11A, B, D ). In turn, the ventral part of the parietal-squamosal suture, which is exposed due to breakage of the ventral part of the parietal, is not in an advanced state of merging yet ( Fig. 11B ; 14 ). The parietal-frontal suture shows a fragment of merged frontal of about 26 × 26 mm posteroventrally, but the rest of the suture is certainly not merged ( Fig. 11C ). As the parietal-frontal suture is filled with sediment, it looks like the frontal detached (i.e. broke off) prior to fossilization ( Fig. 11C ). The posteromedial part of the squamosal-pterygoid suture is preserved. This suture is well merged and is locally hard to discern ( Fig. 14 ).

The zygomatic width of the specimen, as preserved and thus without the anterior zygomatic process, is about 330 mm. Based on superpositions of photos of Persufflatius renefraaijeni n. gen., n. sp. and of three closely related species (Pelocetus calvertensis, Uranocetus gramensis and Atlanticetus patulus ), we estimated the full bizygomatic width to be about 720 mm ( Table 1 ). Based on this estimated bizygomatic width, the body size has been calculated, using the general mysticete equation of Lambert et al. 2010 (equation ‘1’), the ‘stem mysticete equation’ (equation ‘2’) and the ‘stem balaenopterid equation’ (equation ‘3’) of Pyenson & Sponberg (2011) ( Table 2 ). This resulted in an estimated body size of roughly 650 cm, somewhat larger than in adults of the pygmy right whale Caperea marginata (Gray, 1846) , the smallest extant mysticete ( Baker 1985; Kemper 2008; Reidenberg & Laitman 2008; MB pers. obs. 2008; Appendix 4).

The type and only specimen exhibits clear, specific characters, that are quite different from all other BnBB-species. More in particular, the neurocranium exhibits an inflated general aspect. Especially the postglenoid process of the squamosal, the exoccipital and the posterolateral base of the zygomatic process are swollen. Also, the bell-shaped supraoccipital gives the neurocranium a more inflated aspect than all of the other BnBB that have a narrow triangular supraoccipital ( Figs 4 ; 5 ; 8 ; 24 ). The anterior third of the supraoccipital is dorsoventrally thick ( Table 3) and consists of dense bone (compactness about 0.83) ( Table 4).

Frontal ( Figs 4 ; 8 ; 9 ; 11 ; 14 )

The frontal is almost completely missing. The parietal-frontal suture preserves a remnant of merged frontal of about 26 × 26 mm posteroventrally in the top of the curvature. The lower part of the suture is missing in most places, apparently as it was already fused when the frontal broke off ( Fig. 4D, E ). The parietal-flange that covers the posterodorsal corner of the descending supraorbital process of the frontal has broken off. Most of the parietal-frontal suture is preserved and the missing flange covering the posterodorsal corner of the supraorbital process of the frontal must have been small; judging from the preserved adjacent bone, it is estimated to have been maximum a few centimetres long ( Figs 8 ; 9 ; 11 A-C; 13; 14).

Parietal ( Figs 4 ; 8-11 ; 13-15 )

The right parietal is almost completely preserved. Only the ventral part of the anteroventral flange that was covering the descending supraorbital process of the frontal and its anterodorsal part at the skull vertex are missing. The parietal bulges

slightly into the temporal fossa posterolaterally; anterolaterodorsally it is strongly concave ( Figs 10 ; 11A ). Anteriorly, the dorsal border of the parietal supports the supraoccipital; it is slightly inflated and projects laterally, forming the ventral component of the temporal crest that overhangs the temporal fossa for about 25 mm as preserved, but the size of the overhang is estimated to have been originally maximum 40 mm wide (based on the preserved adjacent bone) ( Figs 10 ; 11A ). The parietal’s internal structure is exposed on a roughly 40 × 70 mm (fracture) surface anterodorsomedially ( Fig. 11A ); it consists of spongy bone with large, elongated cavities (mean H: 1.586 mm; mean W: 0.636) (Appendix 2) and a relatively low compactness (about 0.55) ( Table 4).

At the level of the dorsal part of the squamosal-parietal suture, the temporal crest does no longer overhang the temporal fossa. The dorsal part of the parietal-squamosal suture is dorsoventrally broadly straight and runs vertically. The suture is crenate (wavy and rough). It is in an advanced state of fusion ( Fig. 11A, B ; 14 ). The ventral part is missing (due to breakage). It runs inclined in anteroventral direction (at an angle of 135° with respect to the vertical part), contacting the preserved squamosal side of the suture. This lower part of the suture is not fused ( Fig. 11B ). The anterior part of the parietal is anteroventrally inflated and bulges (inverse U-shaped) to accommodate for the missing dorsally convex descending supraorbital process of the frontal ( Figs 10 ; 11B ; 13 ; 14 ).

Parietal on vertex

In dorsal view, the parietal is broadly exposed on the vertex ( Figs 8 ; 11A, B ; 13 ; 14 ), but the exact shape and length of the exposure is unknown, as the dorsal-most part of the vertex is missing (due to fracture and/or erosion). The preserved part is posteriorly about 30 mm wide; in the middle 10 mm and anteriorly about 15 mm wide, as preserved; the maximum length is about 58 mm.

Parietal-frontal (coronal) suture

The dorsal part of the parietal-frontal suture on the vertex is preserved (c. q. the parietal overlapping the posterodorsal frontal). It is (half) inverse U-shaped in anteroventral view. At the (curved) top 6 cm it features about nine irregular rough and deep lamellae ( Figs 11C ; 14 ). As this part of the suture is locally covered by sediment, it looks like the frontal broke off prior to fossilisation. The lowermost (ventrolateral) 6 cm of the suture are damaged in most places (probably as it was already fused when the frontal broke off). Where it is still preserved, it shows a much finer interdigitation ( Figs 11C ; 14 ). The suture was generally not fused anterodorsally, but locally remnants of a fused fragment of the frontal are present (over a 26 × 26 mm surface).

Dorsal border/anterior nuchal crest

The dorsal parietal border contacting (supporting) the supraoccipital is bell-shaped in dorsoposterior view. Anterolaterally, the dorsal border overhangs the temporal fossa. Below this border, the parietal is rather deeply concave

( Figs 10 ; 11A ). Anteroventrally, it becomes strongly convex and inflated (the ventral-most part of the flange that was covering the descending supraorbital process of the frontal is missing), providing the anterior part of the parietal with a saddle-like shape in anterior view. Anteriorly, the parietalsupraoccipital suture is V-shaped, due to a wide and deep sutural lamella ( Figs 10 ; 11A ). Dorsally, the anterior 9 cm of this suture are irregular, wavy and widely open (up to 13 mm separation). The anterolaterodorsal extremities of both the supraoccipital and the parietal (estimated to be less than 15 mm wide) have broken off and are missing ( Figs 10 ; 11A, B ). Medially, the parietal-supraoccipital suture has completely merged, resulting in a fine line, running parallel to the dorsal supraoccipital surface. The supraoccipital is about 28 mm thick at this medial surface, and as such is very robust ( Table 3). Internally (posteriorly), in the cranial cavity (imprint of the frontoparietal lobe of the brain), the suture is visible as a fine, very irregular wavy white line ( Figs 11D ; 15 ). Here also the suture has completely merged.

Squamosal ( Figs 4 ; 8-14 )

The squamosal is massive and inflated, especially laterally and anteriorly. It is generally robust, and together with the exoccipital, this part of the skull is anteroposteriorly relatively longer (170 mm) than in most other BnBB (such as Pelocetus and Uranocetus ( Fig. 7C, F, J ). The dorsal edge of the squamosal (temporal crest) gently slopes downwards in lateral direction; centrally, it slopes down in anterior direction, to form the dorsal squamosal fossa (similar to Pelocetus, Uranocetus , Parietobalaena and Tiucetus , but it is, especially posteriorly, a lot wider in Persufflatius n. gen. ( Figs 7 ; 8 ; 10 ; 14 ). This posterior wall (‘floor’) of the temporal fossa is continuously convex and bulges into the temporal fossa. Basally and medially, at the level of the pseudoval foramen, the squamosal is anteroposteriorly short (57 mm). The squamosal fossa is anteroposteriorly elongate. Its floor is concave, especially anteriorly, though it becomes less concave posteriorly as it approaches the posterior apex of the temporal crest. The zygomatic process of the squamosal has broken, exposing its robust tear-shaped transverse section (ventral width: 87 mm; height: 141 mm). It is ventrally and laterally inflated and rounded, and consists entirely of spongy bone with, very small cavities (mean mediolateral W: 0.354; mean H: 0.478) (Appendix 2), covered externally by a thin layer of cortical bone ( Figs 13 ; 14 ). The compactness of the bone at this section is c. 0.39 ( Table 4). The temporal crest

is strong, obtuse and rounded dorsal to the parietal; on the squamosal it gradually becomes a 90° edge. Laterodorsally, at the base of the zygomatic process of the squamosal, the supramastoid crest (lateral temporal crest) is eroded over the 70 lateral-most millimetres, judging by the curvature of the remaining surfaces, about 2 cm of bone is missing dorsally at this lateral-most point ( Figs 10 ; 13 ; 14 [dotted purple/green line at the left]). Laterally, only the posterior 71 mm of the zygomatic process of the squamosal are preserved. This posterior remnant of the zygomatic process of the squamosal is unusually high ( Table 5A, B ). On this surface, contrary to other BnBB, there is no sign of a supramastoid crest or fossa; the surface is laterally smooth over the dorsal 60 mm and is possibly abraded by erosion ( Figs 8 ; 13 ). Anteroventrally, the pseudoval process protrudes relatively far into the temporal fossa ( Figs 4E ; 9 ; 12 ). The bone is spongy, with big mediolaterally elongated cavities (mean mediolateral W: 2.142; mean H: 0.824) (Appendix 2), having a compactness of c. 0.54 ( Table 4). The anteroventral part of the dorsal squamosal fossa (= the posterior wall of the temporal fossa) bulges into the temporal fossa rising upwards at about 65° with respect to the horizontal plane; the posterodorsal part rises in posterior direction at some 45° to form the squamosal fossa. Posteriorly, the dorsal squamosal fossa is notably wider than in all other BnBB. Ventral to the dorsal squamosal-parietal suture, on the lateral wall of the temporal fossa, the ventral squamosal bulges more strongly into the temporal fossa, being somewhat inflated at this point. The bone has a c. 20 mm thick layer, composed of alternating thin layers of spongy and compact bone. More dorsally, along the dorsal part of the squamosalparietal suture, the surface slopes in dorsomedial direction at about 120° with respect to the horizontal plane. In ventral view, the periotic fossa is deeply excavated and it is anteriorly abruptly restricted by a steep wall at the level of the posterior side of the pseudoval foramen ( Fig. 12 ).

Pseudoval process, squamosal window and falciform process We use the new term ‘pseudoval process’ ( Figs 9 ; 12 ; 14 ) to indicate the anteromedial process of mysticete squamosals, that projects medioventrally and that bears the pseudoval foramen ( Figs 9 ; 12 ) and the falciform process ( Figs 9 ; 12 ; 20 ); In ventral view, the pseudoval process is laterally limited by the medial border of the ventral squamosal fossa; posteriorly it is limited by the anterior border of the periotic fossa. Anterodorsally, the pseudoval process is confluent with the squamosal body; the imaginary line running from the medial squamosal cleft to the medial border of the ventral squamosal fossa delimits the pseudoval process there. Medially and medioventrally the pseudoval process has a big sutural contact surface for the pterygoid ( Fig. 14 ); mediodorsally it (most probably) contacted the (missing) alisphenoid. The pseudoval process is quite robust but anteroposteriorly short (71.8 mm high at the anterior side of the pseudoval foramen and 56.4 mm long anteroposteriorly). The pseudoval foramen is about 35 mm wide transversally (widening to 42 mm posteriorly). It is weakly curved ventrodorsally (about 9 mm high in the middle, as preserved; both the falciform process and the anterior process being damaged ventrally). The pseudoval process is mediolaterally relatively short (roughly 85 mm including the pseudoval foramen but excluding the falciform process). Only the dorsal part (roof) of the pseudoval foramen is preserved ( Figs 4E, F ; 9 ; 12 ). This preserved dorsal part is exclusively formed by the squamosal. As the ventral part is missing, it is unclear whether the foramen was entirely formed by the squamosal or whether the pterygoid made up part of the ventral side. Medially, a small fragment of the pterygoid is preserved. The preserved bone fragment is firmly fused with the squamosal ( Fig. 14 ); it does not contact the pseudoval foramen. The falciform process is about 15 mm thick. It is at least 39 mm long (anteroposteriorly), but the exact dimen-

sions are uncertain, as it is damaged at both the anterior and posterior sides. We designate the concave surface of the squamosal between the posterior falciform process and the anterior border of the external meatus as ‘squamosal window’ (new term; Figs 4E, F ; 9 ; 12 ; 20 ). The squamosal window exposes part of the lateral side of the anterior process of the periotic. In ventral view, the squamosal window is anteroposteriorly short (about 28 mm long, which is shorter than in Parietobalaena ) and continuously curved (contrary to the condition in Metopocetus hunteri Marx, Bosselaers & Louwye, 2016 , where the squamosal window is squared ( Marx et al. 2016a: 12-13, fig. 5/6a); it is rounded also in (most) other BnBB mysticetes like Parietobalaena palmeri ( Kellogg 1968: 182, fig. 83, 84, pl. 67), Diorocetus hiatus ( Kellogg 1968: plate 50), Parietobalaena campiniana ( Bisconti et al. 2013: 111, RBINS M 399, fig. 19 [hiat]), ‘ Parietobalaena ’ laxata (IRSNB M 712, 713 and 727) and Idiocetus longifrons (IRSNB M 718) (MB pers. obs.).

Squamosal-parietal suture

The dorsal part of the squamosal-parietal suture descends anteroventrally from the temporal crest and is almost perfectly straight ( Figs 8 ; 10 ; 11A, B ; 13 ; 14 ). In this dorsal part, the suture-line itself is crenate, still visible as a 3 to 4 mm wide furrow externally; internally, the suture is not discernible anymore. slightly below mid-height the suture inclines about 135° in anteroventromedial direction. The squamosal side of this lower part of the suture is preserved, but the opposite parietal side has broken off and is missing. This ventral part of the squamosal-parietal suture (anterodorsal border of the squamosal) is slightly curved (concave anterodorsally). This part of the suture is not fused and, contrary to the upper part, is finely denticulate. At the point where the squamosal-parietal suture changes direction, it also contacts the squamosal cleft (laterally).

Squamosal cleft

The squamosal cleft is ‘smiley’-shaped (dorsally concave); the suture is irregular, shallow and in places only visible in

grazing light due to an advanced state of merging ( Figs 11B ; 14 ). It looks similar to the squamosal cleft of Parietobalaena sp. ( Parietobalaena laxata (IRSNB M 712 and M 727); MB pers. obs.). It is mediolaterally 70 mm wide. Internally, on the anterior border and on the anteromedial border of the periotic fossa, the remnant of the internal squamosal cleft suture is visible as a sharp edge, medially terminating in a narrow furrow, running upwards from the anteromediodorsal edge of the periotic fossa. It seems to continue in anterior direction, to ‘surface’ as a narrow foramen (8.5 mm) at the posteroventrolateral corner of the parietal flange covering the dorsal supraorbital process of the frontal (white arrows in Fig. 14 ).

Postglenoid process

The postglenoid process is wide and semi-circular to half-teardrop-shaped in posterior view ( Figs 4A, B ; 9 ), protruding far ventrally (about 100 mm high, as preserved posteriorly; another c. 15 mm is missing ventrally, based on the shape, curvature and orientation of the preserved surfaces, totalling about 115 mm

dorsoventral height). Posteriorly, it is flat to slightly concave; anteriorly, it is globose and inflated (a strange condition for the surface that articulates with the convex mandibular condyle). Lateroventrally, the postglenoid process is also strongly inflated ( Figs 12 ; 13 ). This is somewhat similar to the condition observed in Atlanticetus patulus , but the anterior surface is more inflated and is mediolaterally wider, being massive and large compared to the size of the skull. The postglenoid process consists entirely of spongy bone (its internal compactness is about 0.35; Table 4) with medium-sized cavities (mean W: 1.01 mm; mean H: 0.765 mm; Appendix 2). Externally, it is covered by a thin layer of cortical bone. The posterodorsal base of the postglenoid process, the central edge in the external auditory meatus (EAM), is 188 mm wide mediolaterally (medial border of the cranial hiatus till the lateral U-shaped edge of the foramen for the facial nerve). The base-line is mediolaterally continuously curved.

Ventral squamosal crest

The anteroventral border of the squamosal fossa is delimited by a ventrally protruding, sharply edged bony crest (ventral squamosal crest) that separates the ventral surface of the squamosal (ventral squamosal fossa) from the temporal fossa, thus limiting the manoeuvrability of the mandibular condyle medial to the postglenoid contact surface ( Figs 4E ; 9 ; 12 ).

External auditory meatus (EAM)

The EAM is narrow (medially, 17 mm; centrally, 15 mm; laterally, at the level of the lateral exoccipital, c. 31 mm), slightly widening medially, recaling the condition in Cephalotropis nectus and Heterocetus affinis , giving it the appearance of a slender elongated spoon. Far laterally the borders of the EAM are not well defined and it widens fan-shaped ( Fig. 21 ). Anterolaterally, a furrow continues far on the posterior side of the postglenoidal process, terminating laterodorsally in a distinct narrow U-shaped sulcus (11.2 mm wide and 8.22 mm deep), that most probably accommodated the facial nerve ( Figs 4A, B ; 12 ; 13 ; 21 ).

Sternomastoid fossa

Dorsolaterally, above the postglenoid process, there is no deep well-defined sternomastoid fossa. This fossa is very pronounced in Parietobalaena , but in the same position on the squamosum, there is no sign of a fossa in Persufflatius renefraaieni n. gen., n. spec. Instead, there is an inflated protrusion, having an irregular, coarse, somewhat eroded surface, not covered externally by cortical bone. The sternomastoid fossa may have been completely abraded, or, the sternomastoid fossa was positioned more anteriorly, on a part of the squamosal that is missing due to fracturing ( Figs 8 ; 13 ).

Ventral squamosal

The periotic fossa (the anterolateral, lateral and laterodorsal border of the cranial hiatus) is deeply excavated and ovoid (anteroposteriorly oblong) and anteriorly steeply restricted by the dorsal wall of the pseudoval foramen (width 52.5 mm; length: 47 mm; deep: 20.1 mm with respect to the EAM border; 33 mm with respect to the lateral border). Whereas the periotic fossa is long, the pseudoval foramen is shorter than usual ( Fig. 12 ).

Pterygoid

The anteromedioventral part of the squamosal-pterygoid suture is preserved on the squamosal. Medioventrally, this suture seems to be in an advanced state of fusion and its shape is difficult to observe in full detail. The preserved part of the fused bone is 35 mm wide and 25 mm high. On the posterior side the bone has broken and the trapezoid suture, delimiting this roughly 50 × 30 mm sized bone is clearly visible ( Figs 11B ; 14 ). The bone is spongy with a compactness of c. 0.59 ( Table 4).

Occipital

The occipital is (half) bell-shaped in posterodorsal view. The medioventral part of the supraoccipital, the exoccipital and the basioccipital are missing ( Figs 4B, C ; 7F ; 8 ; 10 ; 11A ; 14 ; 15 ). The lateral exoccipital is well preserved.

Supraoccipital

Part of the right half of the supraoccipital is preserved. Judging from the right half, the complete supraoccipital shield must have been (sub-)bell-shaped, with a sigmoid temporal crest (somewhat similar to Uranocetus but more rounded and bellshaped), the supraoccipital being transversely compressed along its anterior half.The anterodorsal surface of the supraoccipital is slightly to strongly concave. The medial part of the exoccipital, bearing the condyles and the foramen magnum, is missing, but the approximate position of the right occipital condyle can be reconstructed ( Fig. 8 ).

The exact shape of the apex is unclear, due to erosion of this part of the skull. It was most probably pointed (triangular) or narrowly trapezoid. Medially, the suture between

the supraoccipital and the parietal is visible as a fine straight line, about 25 mm below the dorsal edge of the supraoccipital. Centrally, in the c. 25 mm-thick-supraoccipital, a layer of spongy bone, about 7 to 9 mm thick, is present (compactness c. 0.51; Table 4). The bone above and below this strip consists of dense, hypermineralized bone (compactness c. 0.81; Table 4), which is most probably redeposited ( Figs 11D ; 15 ). Such a thick anterior supraoccipital with two thick layers of dense bone is unusual. The anterior supraoccipital can be equally thick or even thicker in Herentalia nigra Bisconti, 2014 -like cetotheriids, such as Metopocetus hunteri (with the same bizygomatic width), but the bone is internally uniformly spongy and becomes quickly thinner in posterior direction, while in Persufflatius n. gen. it stays equally thick over 85 mm ( Table 3) and becomes even more dense posteriorly (compactness c. 0.84; Table 4). In extant Balaenopteridae the anterior supraoccipital is thin. In specimens of Balaenoptera acutorostrata with the same bizygomatic width, the supraoccipital is anteriorly about 12 mm thick and becomes slightly thinner posteriorly (MB pers. obs.; Table 3).

Anterolaterally,the supraoccipital is roughly indented with the parietal. This anterolateral part of the supraoccipital overhangs the temporal fossa laterally; posteriorly, it does not overhang this fossa (similar to Titanocetus Bisconti, 2006 , but in the latter it also overhangs the temporal fossa posteriorly). The lateral overhanging edge is c. 25 mm wide as preserved. The anterolaterodorsal extremities (estimated to have been about 15 mm wide) of both the supraoccipital and the parietal broke off and are missing ( Figs 8 ; 10 [white lines]). The width of the overhang is thus estimated to have been maximum c. 40 mm wide origi- nally, based on the orientation and curvature of the preserved surfaces of the lateral supraoccipital and the parietal adjacent to the fracture. The parietal-supraoccipital suture is rough at the level of the overhang; anteriorly, it is deeply indented with a triangular descending V-shaped point ( Figs 10 ; 11A ; 14 ).

Anterodorsolaterally on the supraoccipital there is a prominent, broad, oblique crest, running from the central axis in posterolateral direction, roughly parallel to the lateral border of the bone ( Figs 8 ; 10 ; 11A ). Dorsally, the crest is eroded, polished and obtuse. This crest is quite peculiar and does not, to our knowledge, occur in any other mysticete. Apart from this crest there is no other obvious attachment surface for neck musculature preserved. The preserved part of the supraoccipital has no external occipital crest; the anteromedial part of the supraoccipital is concave (the external occipital crest is missing also in Pelocetus calvertensis and Atlanticetus patulus , contrary to Parietobalaena , Tiucetus , Uranocetus and Diorocetus hiatus ).

Cranial endocast ( Figs 15-17 )

Medioventrally, the imprints of the frontoparietal and the temporal lobes of the cerebrum are clearly visible ( Figs 11C, D ; 15 ; 16 ; 17 ). Because the internal neurocranium is well preserved and looked different from most neurocrania we know, we decided to make a cranial endocast (see Material and methods). We estimate the volume of the brain between 1500 and 1800 cm 3. The imprint of the temporal lobe of the cerebrum is short and deeply excavated (extending about 35 mm more lateral than the lateral-most point of the frontoparietal lobe). The temporal lobe and the adjacent perisylvian cortex are strongly developed in whales, indicating that the auditory signal processing is important ( Fig. 16 ; Ölschläger & Öls-

chläger 2002: 137, 138). The temporal lobe is anteroposteriorly compressed (foreshortened; Morgane et al. 1980), making it transversely ovoid (L: 63 mm; H: 82 mm; W: c. 53 mm). The top 21 mm of this lobe are situated in the supraoccipital. The ventral part is situated in the parietal and the squamosal. The suture between the parietal and the squamosal runs over the middle of the temporal lobe, from the left top corner to the right bottom corner ( Figs 16 ; 17 ). The surface of the bone near this suture is damaged due to fracturing. It is visible in the stereo-image ( Fig. 16 ) as a diagonal, slightly elevated dark grey strip. The posteroventral part ( Fig. 17 ), just below the centre of the image, which is situated within the squamosal, has a wrinkled surface. The largest tubular imprints are terminal branches of venous sinuses, presumably linked to the inferior anastomotic vein of Labbé (black lines in Fig. 17 ), some of which are crossed by fine thread-like superimposed imprints, being terminal branches of smaller cortical veins (magenta lines onFig. 17) (obs. and pers. corr.Dr Yannick Van Ael; MB pers. obs.). At this posterior base of the lobe, surface features are well preserved and more marked than on the rest of the cerebral surface. The anterior side of the temporal lobe and the frontoparietal lobe have a smooth surface, due to the partial ossification of the meninges ( Ölschläger & Ölschläger 2002: 133; Marx et al. 2016b: 81); therefore, cortical gyri are not present on the imprint/cast. The curved border on the right side of the temporal lobe ( Fig. 17 [top right, green line]) is the sylvian fissure ( Breathnach 1955; Ölschläger & Ölschläger 2002: 135; Bisconti et al. 2021b). The deep imprint of the sylvian fissure ventrally, hints at a sizeable superficial sylvian vein. The rectangular part at the bottom left, below the lobe, is not part of the cranial imprint. It is the medial

surface of the squamosal that connects the fossa for the periotic (just above the lateral side of the anterior process of the periotic) with the fossa for the temporal lobe. The imprint of the frontoparietal lobe is about 100 mm high and the diagonal length is 80 mm as preserved. The top 30 mm are situated in the supraoccipital; the ventral part is completely situated in the parietal. Mediodorsally, internally, there is no indication of a internal central occipital crest ( Fig. 11C, D ).

Exoccipital ( Figs 4 A-C; 8; 12; 13; 15)

The exoccipital is mediolaterally wide compared to all other BnBB; it is strongly inflated, posterolaterally globose and very rounded (anteroposterior length: 43 mm at the paroccipital

cavity; 53 mm laterally, just dorsal to the posterior process of the petrotympanic). The lateral and dorsal border of the exoccipital is poorly defined, as it is almost completely merged with the posterior side of the zygomatic process of the squamosal, hinting at a young adult ontogenetic age. In dorsal view its lateral border is just lateral to the level of the medial base of the zygomatic process.This lateral part consists of spongy bone with a low compactness (c. 0.41; Table 4). The exoccipital is tapering at the lateral side, dorsal to the posterior process of the petrotympanic.The posterior process of the petrotympanic is laterally confluent with the exoccipital border and it is surrounded by the exoccipital dorsally, posteriorly and posteroventrally. This suture is also merging dorsally, where it is difficult to observe. Ventrally, the jugular notch and the paroccipital concavity are partially preserved. Only the lateral part of the jugular notch is preserved over c. 22 mm (posteriorly). The anteroposterior length of the exoccipital is c. 35 mm at the level of the jugular notch, the surface of which is only slightly concave. The jugular notch is laterally constricted by a 19 mm wide and 38.5 mm long robust crest, running over the entire length of the exoccipital, that separates it from the paroccipital concavity. The ventral edge of this crest is missing due to erosion. The jugular notch widens anteriorly, approaching the posterior cranial hiatus, and reaches c. 32 mm there, but due to the fracture of the anteromedial part, the real shape can only be estimated.The surface of the jugular notch is strongly inclined in anterodorsal-posteroventral direction (c. 30° with respect to the horizontal plane) ( Fig. 12 ). The bone of the jugular notch is spongy, with a compactness of c. 0.48 ( Table 4). As the jugular notch is adjacent to the occipital condyle, we can roughly estimate the bicondylar width at c. 150 - 170 mm. The condyles must have been relatively smaller than in Tiucetus , but larger than in Pelocetus and probably similar in size to those of Uranocetus ( Fig. 7C, F, G ). The paroccipital cavity is weakly excavated. Its shape is not clear due to erosion, but the concavity was at least 30 mm long at its longest point and about 42 mm wide posteriorly, at its widest point ( Fig. 12 ).

Periotic ( Figs 4F ; 12 ; 13 ; 18-20 ; 22 )

In mysticetes, the periotic often gives an important indication of the species’ overall taxonomic affinities ( Steeman 2010; Ekdale et al. 2011). The periotic of Persufflatius renefraaijeni n. gen., n. sp. closely resembles that of Pelocetus calvertensis and, to a lesser extent, that of Atlanticetus patulus . It resembles Pelocetus in having a generally inflated aspect, having a strong and wide, dorsally protruding tuberosity, continuously running over the whole dorsal surface of the pars petrosa (H: 22 mm). It continues (somewhat less inflated) over the anterior process (H: 10 mm). The lateral tuberosity is relatively small compared to the overall volume of the periotic. This process appears bladelike in the images, but it should be noted that a spherical dorsal part that belongs to it, is still attached to the ventral side of the squamosal (the circumference of which is marked with a dashed line inFig. 18A, B. The full extent of this lateral process can be judged from the reconstruction in ( Fig. 22 [purple-grey spot at the left]). The lateral tuberosity is positioned very far posteriorly, at the level of the fenestra vestibuli (oval window).

Pars petrosa ( Figs 18 ; 19 )

The pars petrosa is mediolaterally wide (W: 62 mm at the level of the lateral process), and especially wide compared to Parietobalaena palmeri , Heterocetus affinis and Cephalotropis nectus ( Fig. 19E ). A c. 15 mm long part of the anterior process is missing. The preserved length of the anterior process,

anterior to the mallear fossa, is 42.3 mm, so originally it must have been about 57 mm long ( Fig. 22 [right purple-grey spot]). Ventrodorsally it is very high (about 37 mm anteriorly). The length of the periotic as preserved is 71.5 mm (anterolateral process till posterior cochlear crest). The original length is estimated to have been c. 83-87 mm. The periotic is medially and anteromedially hypertrophied, dorsal and anterodorsal to the dorsal vestibular area. This type of bone growth is often due to aging (ontogenetic hypertrophy), indicating the specimen is certainly a young adult at least. At the dorsal side, the dorsal tuberosity is reminiscent of that in Parietobalaena , being anteroposteriorly elongated (L: c. 38 mm; W: c. 27 mm) and inflated, its dorsal surface being irregularly rippled and bumped, without a clear crest. Anteromedially, on the pars petrosa (anteromedial to the promontorium), there is an anteromedially oriented, ventrodorsally thick, relatively long and slender, pointed anteromedial process ( Fig. 18 A- D), about 36 mm wide (anteromedial to the dorsal bulge); 45 mm long anteromedial to the anteromedial corner of the promontorium and 27 mm thick (apparently as a result of ontogenetic secondary bone growth also).

Mallear fossa

The mallear fossa is poorly defined and has a somewhat rough surface (c. 6 mm wide). It is situated just posterior to the anterior pedicle. Medial to this surface, there is a narrow

(L: 11 mm; W: 3.3 mm), relatively deep and well-defined incudal fossa, discernible just lateral to the base of the lateral (distal) facial canal, medially and posteriorly constricted by a clear steep crest ( Figs 18A ; 19F ; 20 ). This fossa is similar to the one in Atlanticetus patulus , but narrower.

Compound posterior process of the petrotympanic

The compound posterior process of the petrotympanic (hereafter posterior process) is mediolaterally wide (long), slender and rod-like. Laterally the posterior process widens a bit to become slightly inflated and sub-rectangular ( Figs 13 ; 21 ). But its lateroventral border lacks a flange or other protrusions ( Figs 12 ; 20 ). Posterodorsal to the EAM, dorsal and far lateral to the lateral end of the posterior process, the postmeatal crest forms a deep, prominent, well-defined canal for the transit of the facial nerve laterally ( Fig. 13 ).

Medially, the contact surface between the pars petrosa and the posterior process is large, measuring 26 × 15 mm, and roughly semi-circular, the circular side pointing dorsally. In ventral view, the ventral surface of the posterior process shows a narrow cusp posteriorly, for the transit of the facial nerve, that is anteriorly bounded by a weak, elliptical anteroventral flange (sensu Marx et al. 2016a: 16). It is c. 23 mm wide, c. 11.7 mm long and c. 6 mm high. Anteriorly, this groove is defined by a robust and distinct upright crest delimiting the lateral facial canal. The groove for the facial nerve disappears more laterally ( Fig. 21 ). Medially, on its posterior edge, the posterior process has an elongated, mediolaterally oriented,

kidney-shaped posterior pedicle that contacts the tympanic bulla. The anterior and posterior pedicles to contact the tympanic bulla are closely separated (about 25 mm), hinting at a small (anteroposteriorly short) tympanic bulla ( Table 1 ; Figs 12 ; 20 ), possibly similar to that of Parietobalaena campiniana .

The shape of the dorsal pars petrosa and of the anterior process and the long, straight and slender posterior process, with just a simple anteroventral flange medially and no protrusions laterally, indicate a close relationship to some BnBBmysticetes, such as Pelocetus, Uranocetus and Parietobalaena .

Lateroventrally on the posterior process, centrally in the concave depression, there is an elongated tube-like structure. This structure is an ichnofossil, made of very hardened sediment. It is probably a burrow (L: 60 mm; W: c. 12 mm), though no clear openings or surface structures are present to confirm this ( Figs 9 ; 12 ; 13 ).

Anterior process of the periotic

The anterior process occupies the anterolateral, lateral and anterodorsal borders of the cranial hiatus. It was probably anteroposteriorly short. Anteriorly, the process is damaged: a 19 mm thick flange has broken off. The posterior side of the pseudoval foramen anteriorly limits the possible length of the missing part to maximum 17 mm, thus limiting the maximum original length of the complete anterior process to maximum 59 mm. Posterolaterally, almost adjacent to the posterior process, is a small (diameter about 13 mm), rounded, lateral tuberosity. It appears as blade-like in the

posterior view ( Fig. 19E ), but is in fact inflated, globular, as the dorsal part of the process broke off and is still embedded in the sediment on the ventral border of the squamosal (H: 11 mm) ( Fig. 22 [left purple-grey spot]). Anterior to the lateral tuberosity, there is a 2.5 mm deep fossa for the lateral facial nerve (Table 6). More anteriorly, a fragment of the lateral border of the anterior process has broken off and is also still attached to the squamosal body. Therefore, the lateral side is slightly convex ( Fig. 22 [central purplegrey spot]); the ventral surface is flattened. Anterior to the promontorium, the anterior process bears a distinct groove to accommodate the tensor tympani muscle, separating the pars cochlearis clearly from the anterior process. The anterior pedicle for the bulla contact is slender, anteroposteriorly short and circular (about 6.5 mm diameter; Table 1 ). It is positioned far posteriorly, just in front of the lateral process and the mallear fossa (and thus about 10 mm anterior to the fenestra vestibuli).

Pars cochlearis

The fenestra vestibuli is sub-circular and small (L: 4.5 and H: c. 3 mm). The promontorium (pars cochlearis) is equally small and not inflated, being about 11 mm high (dorsal to the fenestra vestibuli) and about 33 mm long, including the long, shelf-like posterior cochlear crest. Dorsal to the latter, there is a high but anteroposteriorly short stylomastoid fossa; mediolaterally wide, posteriorly restricted by the compound posterior process and dorsally not roofed. Ventrally, there is a prominent promontorial groove (L: 14 mm; at less than 10 mm from the lateral border; H: 3.3 mm). The promon-

torium is mediolaterally strongly restricted (11 mm wide, ventral to the promontorial groove).

Cranial foramina of the pars cochlearis ( Fig. 18C, D ; Table 6) The dorsal vestibular area and the endocranial opening of the facial canal are separated by a thin transverse crest (1.3 mm wide). The endocranial opening of the facial canal is oval and big (almost as big as the dorsal vestibular area) and tapers anteriorly to make it appear tear-shaped. The dorsal vestibular area is circular. The foramen for the vestibular aqueduct (endolymphatic foramen) runs vertical and is slit-like, as well as that for the cochlear aqueduct (perilymphatic foramen). The dorsal vestibular area, the endocranial opening of the facial canal and the cochlear aqueduct are aligned, the vestibular aqueduct being slightly more dorsal. This configuration is exceptional for this type of basal BnBB (diameter dorsal vestibular area: 6.9 mm; endocranial opening of the facial canal: W: 9.3 mm, H: 4.3 mm; H cochlear aqueduct: 9.4 mm, H vestibular aqueduct: 7.1 mm; both slit-like and vertical; Table 6). The fenestra cochlearis (fenestra rotunda) is sub-rectangular and strongly compressed mediolaterally. The cranial openings differ from Parietobalaena sp. and Heterocetus affinis by being aligned, whereas they form a equilateral triangle in the latter species; moreover, both the cochlear aqueduct and the endocranial opening of the facial canal are slit-like in the latter species. Dorsal to the dorsal vestibular area, the pars cochlearis has a strong, hypertrophied process posteromedially (= process of tegmen tympani of Luo & Gingerich 1999 and Ekdale et al. 2011). This extensive medial bone growth indicates we almost certainly deal with a nearly adult speci- men, even though it must be noted that the frontal was not completely merged to the parietal and also the anterior part of the parietal-squamosal suture was not merged. There is no hole anterodorsal to the endocranial opening of the facial canal and there are no other holes in the dorsal part of the periotic. The cranial openings differ from Atlanticetus patulus in the vestibular aqueduct being rounded in the latter and the endocranial opening of the facial canal being circular rather than tear shaped. In addition, the vestibular aqueduct, the dorsal vestibular area and the endocranial opening of the facial canal are not aligned in the latter ( Steeman 2010: 66, fig. 1).

SOME NOTES ON PACHYOSTOSIS, OSTEOSCLEROSIS, BONE EXPANSION AND BONE COMPACTNESS IN THE NEUROCRANIUM OF PERSUFFLATIUS N. GEN. ( FIGS 23-25 ; Tables 4, 5 ; APPENDIX 2)

Non-pathological pachyosteosclerosis often occurs in postcranial bones during early stages of adaption to life in water of secondarily aquatic tetrapods, such as the ribs in basilosaurids; this condition is typified by dense and inflated bone ( de Buffrénil et al. 1990; de Buffrénil & Lambert 2011; Houssaye et al. 2015; van Vliet et al. 2019). In a later evolutionary phase, bone thickening usually diminishes as swimming efficiency and speed increase ( Dewaele et al. 2021). As the pachyostosis observed in several bones of the neurocranium of Persufflatius renefraaijeni n. gen., n. sp. does not display irregular or discontinuous internal transitions, and as the external surfaces are smooth and unaltered, we consider the observed increase in spongy bone volume to be nonpathologic (contrary, for instance, to infectious spondylitis: Fettuccia et al. 2013; Kompanje 1999). In several baleenbearing mysticetes, non-pathological pachyostosis and/or osteosclerosis re-emerged during the Miocene; postcranial osteosclerosis, without pachyostosis, is known from Miocene mysticetes such as Diorocetus hiatus Kellogg, 1968 and, to a lesser extent, also in other contemporaneous chaeomysticetes such as Metopocetus Cope, 1896 and Parietobalaena Kellogg, 1924 ( van Vliet et al. 2019). The late Miocene Cetotherium riabinini Hofstein, 1943 , Brandtocetus chongulek and other marine mammals from the Paratethys have pachyostotic/ pachyosteosclerotic ribs and/or vertebrae, presumably as an adaptation to increased salinity (the Badenian salinity crisis; Gol’din 2014; Dewaele et al. 2021). Furthermore, extant mysticetes at least occasionally develop bone thickening. Especially the anterior and anterodorsal parts of the squamosal body (medial to the zygomatic process) form very thick sections of non-remodelled, apparently cyclically deposited multi-layered bone in extant rorquals (MB pers. obs. 2014), consisting of a succession of thin, alternating layers of compact and trabecular bone. Several extinct cetotheriids, such as: cf. Herentalia nigra (ZMA.MAM.5069); cf. ‘ Plesiocetus ’ burtinii (IRSNB M 676) and Mesocetus latifrons (IRSNB 567), show similar sections of multi-layered cortex (MB pers. obs. 2017). So far, little has been published about the pachyostosis, osteosclerosis and pachyosteosclerosis of cranial elements of mysticetes. To our knowledge, the degree of inflation of the neurocranium of Persufflatius renefraaijeni n. gen., n. sp. is

layered bone

spongy bone

more pronounced than it is in any other chaeomysticete discovered so far. The preserved neurocranium of the holotype is relatively intact. Therefore, the internal bone structure is only visible on some eroded surfaces and fracture zones. We could thus preliminarily document several bone structures in the inflated areas of the type skull ( Figs 23 ; 24 ; 25 ) and

calculate the compactness for the locations involved. We further documented the shape and size of the cavities and the thickness of the trabecular rods in the studied areas of spongy bone (Appendix 2). It must be noted however that these areas are randomly oriented and hence these measurements should be treated with caution.

We could discern three different types of bone: 1) very dense bone; 2) “inflated” spongy bone; and 3) layered bone, consisting of alternating thin layers of trabecular and compact bone ( Fig. 23 ). Most of the extra bone volume is due to extra spongy bone with a relatively low to intermediate compactness (0.32 to 0.59; Appendix 2). The anterior supraoccipital consists of ventrodorsally thick, compact bone (compactness 0.81 to 0.84; Tables 3, 4).

The postglenoid process and the posterior zygomatic process are inflated (pachyostotic) in the holotype and consist internally of spongy bone only, externally, they are covered by a thin layer of cortical bone. Anteroposteriorly, the postglenoid process is long ( Table 5A, B ). It is longer than in many other fossil whales, such as Herentalia nigra , Herpeocetus sendaicus , Metopocetus hunteri , and Parietobalaena campiniana (MB pers. obs.; Table 5A, B ). At the posterolateral side (as preserved), at the jugular notch, at the exoccipital and below the fossa for the compound posterior process is just spongy bone covered by a thin (sub-millimetric) layer of cortical bone ( Fig. 23 ).

When we look at the sections ( Fig. 25 ), we see a varied picture, with different structures. In some places, the cavities are elongated and run in a specific direction ( Fig. 25A, C, G 2 ); in other places we see roughly circular, evenly spaced cavities ( Fig. 25B, H, I ). It must be noted however that the observed differences could be, at least in part, a side effect of the fact that the sections are randomly oriented. There is also a clear difference in size: in some places we encounter only small structures ( Fig. 25H ); elsewhere, small and large structures are mixed ( Fig. 25A, E, I ), and in still other places, large structures prevail ( Fig. 25B, C, G 2 ). The size of the trabecular rods and plates varies widely. At the postglenoid process and the posterior base of the zygomatic process, trabecular rods (about 0.14 mm thick and elongated) and plates (smaller than 0.42 × 0.37 mm) are thin and small, resulting in a low compactness (less than 0.40). In other places, the thickness of the trabecular rods varies between 0.3 and 0.7 mm, resulting in a compactness between 0.55 and 0.63 ( Table 4; Appendix 2).

Figures

FIG. 1. — Location of the holotype of Persufflatius renefraajieni n. gen., n. sp.: A, in the Netherlands (green), the blue square indicates the type locality;B, the vicinity of Liessel. The holotype of Persufflatius renefraaijeni n. gen., n. sp. was discovered at ‘De Brink’ (bright blue spot, 51°25’40”N, 5°49’46”E).

FIG. 2. — Dinoflagellate cyst zonation cf. Munsterman et al. 2019. Abbreviations: Ma, million years ago; Tor, Tortonian; Aan, Achomosphaera andalousiensis; Cpa, Cannospaeropsis passio; Ede, Erymnodinium delectabile; Ltr, Labyrinthodinium truncatum; Pgo, Palaeocystodinum golzowensis; Spl, Systematophora placacantha; Sso, Sumatradinium soucouyantiae. Red lines indicate the limits of the dinozone Southern North Sea DN9, the dinozone the specimen is dated from.

FIG. 3. — Distribution chart of palynomorphs in the sandy sample, the newly defined specimen (MAB 010293) was dated from. The sample, consisting of fine glauconitic sand, was intercalated between the periotic and the squamosal of the holotype. Abbreviations: SP, sporomorphs;MP, miscellaneous palynomorphs.

FIG. 4. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293, holotype, partial right neurocranium: A, posterior view; B, posterodorsal view; C, dorsal view; D, anterior view;E, anteroventral-ventral view (with periotic removed);F, anteroventral view (with periotic in place).Colour images: original; grey images: reconstruction of the left side by mirroring the original. Scale bar: 30 cm.

FIG. 5. — Persufflatius renefraaijeni n. gen., n. sp.; MAB 010293, holotype. The neurocranium of Persufflatius n. gen. compared to the outline drawings of Pelocetus, Atlanticetus, Parietobalaena, Piscobalaena (a typical Cetotheriidae) and Balaenoptera acutorostrata (a typical derived Balaenopteridae). Notice the rounded, bulbous, ‘inflated’ shape of the postglenoid process, the exoccipital, the anterior squamosal and the supraoccipital (arrows).

FIG. 6. — Dr René H. B. Fraaije, ferociously attacked by one of his dinosaurs at the entrance of the ‘Oertijdmuseum’ at Boxtel, the Netherlands. Anterior view. u, up; ll, left lateral. Scale bar: 15 cm.

FIG. 7. — Persufflatius renefraaijeni n. gen., n. sp.; MAB 010293, holotype and some closely related species. Line drawings in dorsal view, on the same scale. a, anterior; rl, right lateral. A, Isanacetus laticephalus MFM 28501, holotype; B, ‘Balaenoptera’ ryani CASG 1733, MFM 28501; C, Pelocetus calvertensis USNM 11976, holotype; D, Atlanticetus patulus USNM 23690, holotype; E, Morenocetus parvus MPL 5-11, holotype; F, Persufflatius renefraaijeni n. gen., n. sp., MAB 010293,holotype (original: right, purple-grey side; mirrored: left, green side); G, Tiucetus rosae MNHN.F.PPI261, holotype; H, ‘Diorocetus’ chichibuensis SMNH- VeF-68, holotype; I, ‘Diorocetus’ shobarensis HMN-F00005, holotype; J, Uranocetus gramensis MSM p 813, holotype; K, Parietobalaena palmeri USNM 10668, holotype; L, USNM 16119, referred specimen; M, Diorocetus hiatus USNM 16783, holotype. Scale bar 20 cm.

FIG. 8. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293, holotype. Posterodorsal view. a, anterior; l, lateral. Scale bar: 10 cm.

FIG. 9. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293, holotype. Anteroventral view. a, anterior; m, medial. Scale bar: 10 cm.

FIG. 10. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293, holotype: A, sections from right to left at 2, 25, 65, 90, 150 and 190 mm from the anterior tip, superimposed on an anterolateral view of the neurocranium. B, the same sections, superimposed on a posterodorsal view, from bottom to top at 2, 25, 65, 90, 150 and 190 mm from the anterior tip. Clear blue lines, sections as preserved; dotted clear blue lines, fractures as preserved; white lines, estimated shape of missing parts based on adjacent preserved parts; dotted black lines, estimated shape of the lateral temporal crest dorsal to the base of the zygomatic process. a, anterior; l, lateral. Scale bar: A, not to scale; B, 10 cm.

FIG. 11. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293, holotype. Detailed views of the parietal and the anterior supraoccipital: A, anterodorsal view; B, anterior view; C, ventral view; D, posteromedial view. Dark blue lines, sutures; clear blue lines, sections; dotted magenta lines, zones; black hatched surface, broken surfaces of the parietal. a, anterior; d, dorsal; l, lateral; m, medial. Scale bars: 5 cm.

FIG. 12. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293, holotype. Ventral view, with the periotic in place. l, lateral; p, posterior. Dashed magenta line, (maximum) estimated shape of the missing tip of the anterior process, based on the limited room left by the posterodorsal part of the pseudoval foramen and the shape and curvature of the preserved part of the anterior process of the periotic; pink surface, pseudoval process. Scale bar: 10 cm.

FIG. 13. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293, holotype. Posterolateral view. Scale bar: 10 cm.

FIG. 14. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293,holotype. Anterolateral view with all preserved sutures indicated (yellow lines). White arrows, narrow posteroventrolateral parietal foramen. d, dorsal; m, medial. Scale bar: 30 cm.

FIG. 15. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293, holotype. Medial view without periotic. d, dorsal; l, lateral; p, posterior. Stereo image. Scale bars: 10 cm.

FIG. 16. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293, holotype. Lateral stereo view and annotated line draw of the endocast of the right cerebral imprint as preserved. d, dorsal; a, anterior. Grey areas, missing fragments due to fracture. At the right side is the imprint of the frontoparietal lobe. To the left is the protruding imprint of the laterally protruding temporal lobe. Scale bars: 5 cm.

FIG. 17. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293, holotype. Cast of the lateral side of the temporal lobe of the cerebrum: interpretation of the surface structures. Antrolateral view. a, anterior; d, dorsal. Bright green line indicates the position of the almost vertical sylvian fissure (the separation between the frontoparietal and the temporal lobe). The wavy violet line in the middle of the lobe indicates the position of the internal parietal-squamosal suture.The black lines indicate terminal branches of the venous sinuses (reasonably large blood vessels); they are crossed by smaller cortical veins (magenta lines). Hatched lines, fracture. Scale bar: 30 mm. (obs. and pers. comm. Dr Yannick Van Ael; MB pers. obs.).

FIG. 18. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293,holotype.Right periotic: A, ventral view; B, dorsal view; C, posteromedial view; D, medial view. a, anterior; d, dorsal; l, lateral; m, medial; p, posterior; pl, posterolateral. Scale bar: 5 cm.

FIG. 19. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293, holotype. Right periotic: E, posterior view; F, lateroventral view. a, anterior; d, dorsal; l, lateral. Scale bar: 5 cm.

FIG. 20. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293, holotype. Right periotic in situ in squamosal.Light blue line, burrow. Ventral view.

FIG. 21. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293, holotype. Posterior view. d, dorsal; l, lateral. Scale bar: 10 cm. TABLE 6. — Measurements of periotic foramina in mm.

FIG. 22. — Persufflatius renefraaijeni n. gen., n. sp., MAB 010293, holotype. Periotic, lateral view (anterior to the right): additions in purple-grey; right purple-grey spot, reconstruction of the broken anterior process (maximum size possible); middle purple-grey spot, the addition of the fragment of the lateral side of the anterior process that is still attached to the squamosal; left purple-grey spot, addition of the fragment of the lateral tuberosity that is still attached to the squamosal.

FIG. 23. — Anteroventral (left) and posterodorsal (right) views of Persufflatius renefraaijeni n. gen., n. sp., with a schematic indication of expanded (‘inflated’) regions (compared to the neurocranium of Pelocetus, Uranocetus and Piscobalaena). Note the sub-triangular ventral squamosal fossa (1), with a strong ventrally protruding wall (D) and the wide,rounded dorsal squamosal fossa (2). A, dense mediodorsal parietal; B, dense anterodorsal supraoccipital; C, layered anteromedioventral squamosal body; D, inflated ventral squamosal wall; E, inflated and laterally expanded anterior surface of the postglenoid process; F, inflated and laterally expanded posterolateral postglenoid process;G, inflated laterodorsal squamosal and posterolaterodorsal zygomatic process;H, inflated and mediolaterally expanded exoccipital. Scale bars: 10 cm.

FIG. 24. — Series of bone samples of the neurocranium of the holotype (MAB 010293), showing some different bone structures. Left, neurocranium of holotype: A, anterior view; B, posterior to posterodorsal view; C, posterodorsal view. a, anterior; d, dorsal; m, medial; l, lateral. Magenta ovals show positions where the bone samples were taken. Scale bar: 10 cm exept photos of the bone samples (on the right), to the same scale (10 mm).

FIG. 25. — Camera lucida drawings of the bone samples shown in Fig. 24: A, anterior parietal (at vertex); compactness: 0.545; B, pterygoid-squamosal suture;compactness:0.631;C, anteroventral squamosal border;compactness: 0.594; D, posterior fracture of anterior supraoccipital; compactness: 0.836; E, postglenoid process; compactness: 0.348; F, jugular notch; compactness: 0.461; G, medial anterior supraoccipital shield; compactness: G1, top: 0.808; G2, middle: 0.509; G3, bottom: 0.775; H, base of the zygomatic process; compactness: 0.391; I, lateral exoccipital; compactness: 0.413. The bone compactness was calculated using ‘Bone ProfileR’ (Girondot & Laurin 2003). Scale bar: 10 mm. See also Table 4.

Tables

TABLE 2. — Persufflatius renefraaijeni n. gen., n. sp.; MAB 010293, holotype. Body size estimates (in cm), based on the general mysticete equation of Lambert et al. 2010 (equation ‘1’), the ‘stem mysticete equation’ (equation ‘2’) and the ‘stem balaenopterid equation’ (equation ‘3’) of Pyenson & Sponberg 2011.

Estimated body length for bizygomatic width 72 cm
Equation 1	658 cm
Equation 2	658 cm
Equation 3	741 cm