A critical revision of the fossil record, stratigraphy and diversity of the Neogene seal genus Monotherium (Carnivora, Phocidae)

Historically, Monotherium had been one of the few genera of extinct Phocidae (true seals) that served as a wastebin taxon. Consequently, it did neither aid in understanding phylogenetic relationships of extinct Phocidae, nor in understanding seal diversity in deep time. This urged the reassessment of the genus. Before our review, Monotherium included five different species: Monotherium aberratum, Monotherium affine, and Monotherium delognii from Belgium; Monotherium gaudini from Italy; and Monotherium? wymani from the east coast USA. In this work we redescribe the fossil record of the genus, retaining the type species M. delognii. Monotherium aberratum and M. affine are reassigned to the new phocine genus Frisiphoca. Monotherium gaudini is renamed and considered a stem-monachine (Noriphoca gaudini). The holotype of the monachine M.? wymani requires further study pending the discovery of new fossil material that could be attributed to the same taxon. Reinvestigating the stratigraphic context reveals that N. gaudini most likely represents one of the two oldest named phocid seals, or even the oldest, dated to the late Oligocene–earliest Miocene. Our results allow questioning the widespread idea that Phocidae originated in the western Atlantic and better appreciate their palaeobiogeography during the late Oligocene–Miocene interval in the North Atlantic realm.


Introduction
The extinct genus Monotherium Van Beneden, 1876 (Monatherium in Van Beneden, 1877) is a particular taxon among Phocidae (true seals). Of all currently known extant and extinct phocid seal taxa, most are considered monospecific or include only two species [1], and only two genera include three species: the extant Pusa and the extinct middle to late Miocene Praepusa [2,3]. However, the Miocene monachine genus Monotherium surpasses these and currently includes five species: Monotherium aberratum Van Beneden, 1876; Monotherium affine Van Beneden, 1876; Monotherium delognii Van Beneden, 1876; Monotherium gaudini (Guiscardi, 1870); and Monotherium? wymani (Leidy, 1853), from Belgium (first three species), Italy and the east coast of the United States, respectively. Furthermore, historically other species have been assigned to this genus, e.g. Monotherium maeotica [Cryptophoca maeotica] (Nordmann, 1860), and Monotherium rugosidens (Owen, in Adams, 1879) [holotype is odontocete tooth] [4,5]. On the other side, M. aberratum, M. affine and M. delognii have been recently considered nomina dubia [6], but without any evidence provided. A major issue for the taxonomy of Monotherium is that the fossil record of the different species within the genus varies strongly in terms of preserved skeletal elements: most Monotherium species cannot be directly compared to other species within the genus. For example, M. gaudini is known from a partial skull [7], while no cranial bones have been attributed to the other species of the genus.
In the light of recently discovered specimens showing affinities with part of the species of this genus, a reinvestigation is essential, with a description of the new material and reassessment of the previously described species.

Terminology and comparative material
This study follows the anatomical nomenclature of the recent publications of Amson & Muizon [8], Berta et al. [6] and Dewaele et al. [9,10]. Whenever terms have not been used in any of the aforementioned publications, anatomical nomenclature follows Evans & Lahunta's description of the dog [11]. Length measurements were taken to the nearest 0.1 mm, using analogue calipers and are presented as tables 2, 4 and 5 and electronic supplementary material (Supplementary Information 1: tables S1-S3). For reasons of consistency, these measurements were taken following the same scheme as Koretsky [2], which has more recently been applied to other extinct phocids [6,9,10]. Comparative specimens of extant and extinct taxa are listed as electronic supplementary material (Supplementary Information 1: lists 1 (extant taxa) and 2 (extinct taxa)).

Dinoflagellate cyst biostratigraphy
The palynological preparation of the sediments followed standard techniques described by Louwye et al. [12]. Acid treatments with HCl and HF were applied for the removal of carbonates and silicates, respectively. Sieving of the organic residue was carried out on a nylon screen with a 10 µm mesh size. The residue was placed on glass slides with glycerol gelatin jelly. The microscopic analysis was carried out with a transmitted light microscope Zeiss AxioImager A1 under 400× magnification. The entire slide was scanned in non-overlapping traverses. The taxonomy of the dinocysts and acritarchs follows Fensome et al. [13]. A table showing all the observed dinocyst and acritarch taxa is presented as electronic supplementary material ( Supplementary Information 1: table S4).   Table 2. Measurements of the humerus of Frisiphoca aberratum and Frisiphoca affine (in mm). Measurements based on the scheme presented by Koretsky [2].    [71]. The robustness R is calculated as the ratio of l/L or (l 1 + l 2 + l 3 + l 4 )/L, with l 1 = maximum transverse width of the proximal epiphysis, l 2 = minimum transverse width of the diaphysis, l 3 = maximum transverse width of the distal epiphysis, l 4 = maximum anteroposterior width of the diaphysis at the level of the deltoid tuberosity; L = maximum length of the humerus. Measurements with an asterisk are estimations. Measurements in mm. Sources provided for measurements retrieved from the literature. Measurements by Koretsky [2] represent averages of multiple specimens.
taxon l 1 l 2 l 3 l 4 l L R= l/L Acrophoca longirostris [71] 65   and M. gaudini have been described prior to the erection of the genus Monotherium, but had been named Phoca wymani Leidy, 1853, on the basis of a few isolated cranial and postcranial specimens from the Miocene (presumably the Calvert formation) of Richmond, Virginia [15], and Phoca gaudini Guiscardi, 1870, on the basis of one partial skull and a mandible from 3 km east of Roccamorice, Abruzzo Region, Italy [7], respectively. Later, the genus Monotherium was erected by Van Beneden [16], including M. delognii, M. affine and M. aberratum based on isolated postcranial material from Antwerp (Belgium, southern margin of the North Sea Basin). Although somewhat vague and very concise, Van Beneden [16] provided descriptive elements for M. delognii ('similarities with Phoca barbata') and M. aberratum ('size greater than Monachus'), but not for M. affine. However, it is argued that M. affine comprises all material from the original collection that could not be assigned to either of the two other species. For neither of the three species, Van Beneden [16] provided illustrations or collection numbers.
Having met the requirements of naming a species under the International Code on Zoological Nomenclature (ICZN) Article 11, and being published before 1931, ICZN Article 12 applies to Monotherium and 'indications' suffice for naming taxa. Therefore, Monotherium is a valid name, irrespective of Van Beneden [17] using the name Monatherium one year later and implicitly providing the etymology of Monatherium, and rendering the original name of Monotherium a typographical error. Van Beneden corrected the spelling, providing the etymology, after observing affinities between the genus 'Monatherium' and the extant Pelagius monachus, which is a junior synonym of Monachus monachus. Therefore, it can safely be implied that the name Monatherium etymologically refers to Monachus monachus and that the former name of Monotherium by Van Beneden [16] is a typographical error; yet, Monotherium is the valid name for the taxon and Monatherium should be considered a junior synonym of it. In the 1877 publication, Van Beneden [17] described Monotherium delognii, Monotherium affine and Monotherium aberratum in much more detail and provided the collection numbers of specific specimens.
Van Beneden did not assign a type species to the genus Monotherium. In 1922, Kellogg retained the name Monotherium for all three taxa and assigned Monotherium delognii as the type species of the genus on the basis of page priority in Van Beneden [17] [18, p. 72]. He also regarded Monotherium affine as a junior synonym to M. delognii, stating 'Monotherium delognii is based upon too fragmentary material to distinguish it from Monotherium affine. Therefore, since Monotherium delognii has page priority, it is here interpreted to include Van Beneden's second species, Monotherium affine, as well.' Kellogg [18] also retained Monotherium affine as a separate taxon and renamed the Italian species Phoca gaudini to Monotherium gaudini.
More recently, Ray [15] studied the North American Phoca wymani in detail, removing it from the genus Phoca and tentatively placing it among Monotherium: Monotherium? wymani. However, M.? wymani is based on very fragmentary isolated specimens that bear little diagnostic value for comparison with other extinct Phocidae. Following Kellogg [18], Ray [15] did not use the generic name Monatherium, but instead used the first, and correct, name Monotherium for the three taxa described by Van Beneden [16,17].

Dinoflagellate cyst biostratigraphy of Monotherium from Belgium
Historically, the stratigraphic context of Monotherium aberratum, Monotherium affine and Monotherium delognii had been poorly defined. Van Beneden [17] assigned a 'Diestian' age to the entire record of Monotherium from Belgium. However, the 'Diestian' is a currently abandoned term and it had been shown that the term should not be used any more [19]. Moreover, shortly after the description of Monotherum by Van Beneden [16,17], the 'Anversien' (='Antwerpian') was erected and considerably restricted the extent of the 'Diestian' [19,20]. Nowadays, the same toponym is used for the upper Miocene Diest Formation, which is roughly the lithostratigraphical equivalent of the 'Diestian' stage. The Diest Formation is a diachronous formation deposited in a marginal marine setting [19]. Near the city of Antwerp, the deposits of the Diest Formation (Deurne Member) are of late Tortonian age [19][20][21][22][23].
Only two sediment samples could be recovered from bone cavities of specimens formerly attributed to the genus Monotherium: sample 1108LDW-1100Lab from the thoracic or lumbar vertebrae of Monotherium delognii (either from specimen IRSNB 1108 or from specimen IRSNB 1108-M255; Phocidae indet. in this study, see Supplementary Information 3) and sample 1132LDW-1102Lab from the cervical or thoracic vertebrae originally assigned to Monotherium aberratum (IRSNB 1132-M269; Phocidae indet. in this study, see Supplementary Information 3). The samples were palynologically analysed for organic-walled dinoflagellate cysts (dinocysts) and acritarchs (see Supplementary Information 1, table S4).
The preservation and diversity of the dinocysts in sample LDW1108-1100Lab is poor. A total of 10 dinocyst species and one reworked acritarch were recorded ( Supplementary Information 1: table S4).  [24] defined the Achomosphaera andalousiensis zone as the interval from the lowest common occurrence of the eponymous species to the lowest occurrence of Gramocysta verricula, and suggest an age of 13.2 Ma for the lower boundary of the zone. The dinoflagelatte cysts thus indicate a maximum age of 13.2 Ma (early Serravallian, late middle Miocene) for the sediment sample. The other recorded dinoflagellate cysts are stratigraphically long ranging species with no biostratigraphical value.

Dybkjaer & Piasecki
The preservation and diversity of the dinocysts in sample LDW1132-1102Lab is similarly poor. Only eleven dinocyst species and two acritarch were recorded. A maximum age for the sample is provided by the key species Habibacysta tectata. This species has a lowest occurrence in high latitudes dated at 14.2 Ma by Schreck et al. [25], and this datum was later confirmed by Quaijtaal et al. [26] in lower latitudes (Porcupine Basin, off southwest Ireland). A lowest occurrence of Operculodinium? eirikianum at ca 14 Ma (upper Langhian) is provided by Louwye et al. [27] in a low-resolution palynological study of the Miocene of the Porcupine Basin. Operculodinium? eirikianum has a persistent highest occurrence at the lower-upper Pliocene boundary at ca 2.617 Ma [28]. The lowest occurrence of Operculidinium tegillatum is located at the Tortonian-Messinian boundary (7. 25 Ma). The persistent highest occurrence is noted in the Zanclean at 3.7 Ma [28]. Quinquecuspis concreta has a poorly specified lowest occurrence in the upper Tortonian of Germany [29]. The dinoflagellate cysts in this second sample indicate an age situated between ca 7.25 Ma (or somewhat older in the late Tortonian), and 3.7 Ma (late Zanclean).

Diagnosis. As for the only included species
Etymology. From the Greek adjective 'noris' and the Greek noun 'phoke'. Meaning 'early' and 'seal', respectively, referring to the geologically old age of the species. An age interval of late Oligocene to early Miocene is presented here (see below). Hence, this taxon may possibly represent the first unquestionable phocid from the Palaeogene ( [10] versus [30,31]).

(figures 1 and 2)
LSID. urn:lsid:zoobank.org:act:0A15AA71-8B86-495B-BF9B-5FE3D11FB2CF Diagnosis. Large phocid, comparable in size to the leopard seal, Hydrurga leptonyx. Stem phocid, but still yielding typically monachine characters in having: an ascending process of the premaxilla that is (at least partially) within the nasal cavity and not visible laterally; roots of incisors not (or only weakly) laterally compressed. The anterior termination of the maxillary process of the jugal is located lateral to the infraorbital foramen, which is shared with extinct Monachinae (Acrophoca longirostris, Hadrokirus martini, Homiphoca capensis and Piscophoca pacifica) and the extant Monachinae Mirounga. In the phylogenetic analysis, the identification of Noriphoca gaudini as a separate taxon is supported by one unequivocal autapomorphy: the ventral edge of the zygomatic arch is level with the alveolar plane. Furthermore, the skull of N. gaudini differs from all other Monachinae by the presence of three upper incisors, and a paracone on the postcanine teeth that is low.
Holotype. MSNUN123, partial skull. Only the ventral and anterior portion of the skull are visible. The dorsal portion of the skull is missing and the preserved part is embedded in the matrix, inhibiting description of the specimen in dorsal view.
Stratigraphy and age. Guiscardi [7] noted that the specimen comes from calcareous deposits, rich in bitumen. Based on more recent literature, it is evident that the outcropping formations, 3 km east of Roccamorice are the Santo Spirito and Bolognano formations [32,33]. Because it had been hypothesized that the earliest Phocidae lived around 23   Formation are found in its lower part, mainly restricted to the Lepidocyclina Limestone dated to the Oligocene or earliest Miocene (Aquitanian) ( [33,35]; and references therein). Consequently, the exact age of the holotype specimen of M. gaudini is still unknown; but nevertheless, this holotype is most likely as old as, or even older than the Aquitanian. Hence, N. gaudini is most likely older than Afrophoca libyca, known from the Burdigalian of Libya [36].
Remarks. Other specimens from the same locality and from the same level, i.e. a partial mandible and isolated teeth, had been presented by Guiscardi [7], but these specimens are currently lost (Giovanni Bianucci 2017, personal communication  [6]) to the late Langhian-Serravalianearly Tortonian [40,41] (adopted by Deméré et al. [4]). Although these teeth are clearly monachine, we observe only few similarities between those and the teeth of the holotype of Noriphoca gaudini. Although both have strongly pronounced cusps and a crenulated enamel layer, the cusps of the teeth of N. gaudini are much less raised than in the specimens from the Stirone River. Therefore, we deem it impossible to identify these isolated teeth more precisely than Monachinae indet. and they will no further be considered in this study. The redescription of M. gaudini (as N. gaudini) is entirely based on observations made on the descriptions, images and drawings presented by Guiscardi [7]. We, the authors, have not studied the type specimen in person.

Description and comparison
The type and only known specimen of Noriphoca gaudini comprises an incomplete skull (figures 1 and 2). The known parts of the skull include the ventral part of the rostrum and parts of the basicranium and the right zygomatic arch. A number of maxillary teeth are known as well [7]. The dorsal portion of the skull is missing and only the anterior portion of the snout is visible in dorsal view. The remainder of the skull is embedded in the matrix.
Only the anterior portion of the premaxilla is preserved and it is restricted to the interior of the nasal cavity, which means that it is not visible in lateral view. This is a typically monachine characteristic [42,43]. The anterior alveolar plane faces anteroventrally and the canines have a strong anterior aspect in their orientation. Amson & Muizon [8] observed a similar condition in the monachine Homiphoca capensis, but we conclude that this is considerably more pronounced in N. gaudini than it is in H. capensis. In lateral outline, the nasal cavity is weakly curved and almost rectilinear, and has a strong dorsal aspect to its orientation. This corresponds, notably, with Acrophoca longirostris from the late Miocene of Peru, and Allodesmus spp. [42,[44][45][46]. Other Monachinae and pinnipedimorphs usually have snouts that face more anteriorly and that are either more strongly concave in lateral view (Phocidae), or convex (early pinnipedimorphs), or vary between concave or convex in lateral view within a single clade [8,42,[47][48][49][50][51][52][53][54][55][56][57].
The palate of N. gaudini is slightly constricted at the level of the first premolar, after which the combined tooth rows diverge posteriorly. This is a typically phocid characteristic [58] and it is far less expressed in stem pinnipedimorphs and other non-phocid pinnipeds, where this constriction is minimal or absent [8,[44][45][46][47][48][49][50][51][52][53][54][55]58]. The posterior divergence of the tooth row is minimal in early pinnipedimorphs, Odobenidae and Otariidae [47][48][49][50][51][52][53][54][55], but is present to varying degrees in Desmatophocidae and Phocidae [8,42,[44][45][46]. On the palate, a small, slit-like and narrow palatine fissure is located at the suture between the premaxilla and maxilla, at the level of P1. The shape of the palatine fissure varies among extinct Phocidae and is, for instance, small in Homiphoca capensis [56,57], but large in Hadrokirus martini and Piscophoca pacifica [8,42], and this palatine fissure is generally large in other extinct Pinnipedia and early pinnipedimorphs [47][48][49][50][51][52][53][54][55], although a strongly reduced palatine fissure has also been observed in some desmatophocids [44]. A significant portion of the maxilla is preserved. The alveolar process, bearing the teeth, is slightly raised over the palatine process of the maxilla. In lateral view, the alveolar process faces ventrally, as in other Pinnipedimorpha, except the Monachinae Hadrokirus martini, Ommatophoca rossii and Piscophoca pacifica [8]. The palate is slightly arching dorsally in N. gaudini. This condition varies within different clades of Pinnipedimorpha. Among stem pinnipedimorphs, for instance, it is arching in Enaliarctos spp., but nearly flat in Pinnarctidion bishopi [47]. Even among Monachinae, Muizon [42] noted variation in the degree of arching of the palate. The palatal groove on the palatine process of the maxilla is narrow and becomes gradually more pronounced towards the anterior palatal foramen, which is located at the level of the posterior foramen of P4. Among Phocidae, the location of the anterior palatal foramen varies from the level of P3 (Erignathus barbatus, Monachus spp.) to posterior to the level of M1 (Homiphoca capensis, Mirounga spp., Ommatophoca rossii and Pusa spp.) [56,57,59]. Among extant and other Pinnipedimorpha, the position of this foramen varies, but it is generally well anterior to the level of the last postcanine tooth, notwithstanding exceptions such as Desmatophoca brachycephala [46][47][48][49][50][51]. The maxillopalatine suture is transversely straight between the M1. Consequently, for N. gaudini, the maxillopalatine suture is located posterior to the anterior palatine foramina. Muizon [42] noted that in some Phocinae, the anterior palatine foramina are located on the maxillopalatine suture, while they are anterior to that suture in Monachinae. Differences in the terminology used combined with inadequate illustrations inhibit studying this trait in detail for other pinnipedimorphs, based on literature alone. the palatines is smoothly rounded [44][45][46][47][48][49][50][51][52][53][54]. Compared to all other pinnipedimorphs, including Phocidae (except the Monachinae Acrophoca longirostris and Hadrokirus martini), the (rounded) posterior margin of the palatines is located relatively anterior, with the anteriormost tip located little posterior to the last postcanine tooth and the anterior extremity of the orbit in ventral view. In all other pinnipedimorphs, the palatine extends much more posteriorly, reaching the anteroposterior level of the jugal-squamosal contact and much more posterior to the last postcanine tooth.
The anterior margin of the infraorbital canal on the antorbital process of the maxilla is located at the level of M1. In Monachinae, this compares to the extant Lobodon carcinophaga and Monachus spp., and the extinct Hadrokirus martini and Pi. pacifica, whereas this canal is located either anterior (e.g. Leptonychotes weddellii) or posterior (e.g. to M1) in other Monachinae. In Phocinae, the anterior margin of the infraorbital canal is located posterior to M1. In stem pinnipedimorphs, the anterior margin of the infraorbital canal is usually located at the level of M1 [46][47][48][49]. Only rarely is the anterior margin of the infraorbital canal located anterior to M1 (e.g. Enaliarctos spp.) [52,53]. Related to that, the last postcanine is as the level of the root of the jugal process of the maxilla in Noriphoca gaudini, as in many Monachinae. In Phocinae, the last postcanine tooth is located anterior to the jugal process. In other early pinnipedimorphs and early pinnipeds, the last postcanine tooth reaches the level of the posterior portion of the root of the jugal process of the maxilla or posterior. The maxillary process of the jugal contacts the maxilla, terminating dorsal to the infraorbital canal, as in extinct Monachinae and Mirounga. In extant Monachinae (except Mirounga) and Phocinae, this process terminates lateral to the infraorbital canal. The jugal is incomplete, but comparison with more complete monachine skulls show that the anterior portion of the arch of the jugal is flat to slightly oriented downwards in N. gaudini. This condition varies among Monachinae: flat in Acrophoca longirostris, Hadrokirus martini, and Mirounga, upward in Homiphoca capensis, and Piscophoca pacifica, and downward in other Monachinae. In Phocinae, the anterior portion of the arch of the jugal is directed flat to upward.
In the upper tooth row, Noriphoca gaudini is characterized by having three incisors. Among Phocidae, the presence of three upper incisors is generally considered a characteristic of Phocinae (except Cystophora cristata having two upper incisors), while Monachinae are characterized by having two upper incisors. Three upper incisors are also present in early pinnipedimorphs and desmatophocids [18,[44][45][46][47][48][49]53,54]. The lateral incisor I3 is larger than I1 and I2, which are similar in size, but still much smaller than the canine, and all incisors form a transversely straight row. In most Phocinae (except Halichoerus grypus) and early pinnipedimorphs, the lateral incisor is comparable in size to or only slightly larger than the medial incisor(s). A number of extinct Monachinae retain relatively small lateral incisors (Hadrokirus martini, Homiphoca capensis and Piscophoca pacifica), while the lateral incisor is clearly intermediate in size between medial incisors and canines in other extinct (Acrophoca longirostris) and extant Monachinae. Though, overall the incisors of N. gaudini are relatively smaller than the incisors in Monachinae (except Acrophoca longirostris and Ommatophoca rossii), while the incisors are almost always comparatively small in Phocinae (except H. grypus). The roots of the incisors of N. gaudini are not transversely compressed. In Monachinae and stem pinnipedimorphs, roots are not or only faintly compressed transversely, while in Phocinae, incisor roots are strongly compressed. Morphologically, there is little variation between the mesial incisors and the lateral incisors, apart from the size. The mesial incisors are more slender transversely than the lateral incisors. The incisors are single cusped and bear no cingulum. They are conical, but slightly recurved lingually and bear two occlusal facets on their lingual surfaces: one posterolateral and one posteromedial. Both canines are only partially preserved, but they are conical and appear labially curved. The canine alveolus is oval, i.e. slightly mediolaterally compressed.
The maxillary postcanine teeth include four premolars and one-noticeably smaller-molar. This is typically phocid, as other early stem pinnipedimorphs, desmatophocids, and many extinct odobenids and otariids have at least two upper molars [44][45][46][47][48][49][50][51][52][53][54][55]58]. Apart from the single-rooted P1, the postcanine teeth of Noriphoca gaudini are all double-rooted. This is common among Phocidae, in which only few taxa (Halichoerus grypus and Mirounga spp.) show a tendency towards single-rooted postcanine teeth. Whereas Devinophoca claytoni is the only known phocid to have a triple-rooted upper first molar [60], early stem pinnipedimorphs have triple-rooted upper first molars [46][47][48][49][52][53][54]. Desmatophocids, odobenids and otariids all show a trend towards reduction of the roots to single-rooted postcanine teeth during their evolution [44][45][46]50,58]. All postcanine teeth of N. gaudini are labiolingually broad and the lingual cingulum is well developed, yielding a semicircular or subtriangular shape in occlusal view. This is clearly a plesiomorphic character observed in early stem pinnipedimorphs, Hadrokirus martini, Monachus spp., and Piscophoca pacifica, while other (extant) Phocidae all display a specialized dentition with the cingulum being strongly reduced or absent [ The first premolar (P1) is subtriangular in occlusal view and is implanted parallel to the tooth row. The labial margin is slightly concave, while the lingual margin strongly projects lingually, yielding a subtriangular outline in occlusal view. The paracone (central cusp) and the ?metacone distal to the paracone are lowly raised and only little defined. The ?metacone is closely appressed against the paracone, further reducing the prominence of both. The term 'metacone' is based on usage in publications on other pinnipedimorphs [47], but contrasts with dedicated literature on dentition in carnivorans. For example, Solé et al. [61] stated that there is no metacone present on the maxillary premolars of Dormaalocyon latouri, the oldest known carnivoran. Anterior to the paracone, there is a strongly reduced and rounded cusp. Similar to the ?metacone, this cusp may be considered the ?preparacrista or the ?parastyle, following Solé et al. [61], and pending further studies on the evolution of the dentition in (early) Pinnipedimorpha.
P2-M1 are severely damaged, precluding detailed description. They are more elongate than P1, the lingual convexity is less pronounced, and the labial concavity is slightly more pronounced than in P1. P2-P4 are implanted slightly obliquely to the tooth row axis, with the distal extremity of P2 located labial to the mesial extremity of P3 and the distal extremity of P3 located labial to the mexial extremity of P4. P2-P4 are morphologically similar to P1, having a lowly-raised paracone and a ?metacone distal to it. The height of the cusp mesial to the paracone is strongly reduced but mesiodistally long and may be considered the ?preparacrista or the ?parastyle. Posterior to the ?metacone, there is a small protuberance that can be considered the ?postmetacrista or the ?metastyle.
The first upper molar (M1) is separated from and slightly smaller than the premolars. M1 is morphologically strongly similar to P1, and is clearly premolariform, having lost the trigonid morphology and the protocone that are still present in early stem pinnipedimorphs [46][47][48][49][52][53][54]. The prominence of the paracone and the ?metacone is greater than in the premolars.
The enamel of the postcanine teeth is 'wrinkled', as had been observed for H. martini by Amson & Muizon [8]. The protocone is not as prominent as in other Monachinae, but broad: contributing to the robust appearance of the teeth. The metacone is large, about half the size of the protocone. The paracone anterior to the protocone is strongly reduced. The robust upper dentition of N. gaudini is most similar to that of the extant Monachus and a number of extinct Monachinae, such as H. martini and Pl. etrusca, while extant Monachinae (except Monachus) have highly specialized teeth. The mandibles have been described for both Afrophoca libyca and N. gaudini, and both are geographically and possibly geochronologically close [7,36]. However, the state of preservation of the holotype, and only specimen, of A. libyca (ELNRP 2Z131) is poor, and the illustrated mandible of N. gaudini ([7]: fig. 6) is missing (Giovanni Bianucci 2017, personal communication). Therefore, formal comparison between both taxa is for now precluded.
Type horizon. Ray [15] elaborated on the probable type horizon of MCZ 8741, concluding on the Calvert Formation. However, the Calvert Formation spans across the entire early Miocene and into the late middle Miocene (ca 23.03-13. 8 Ma). At the Calvert Cliffs in Maryland, the oldest published record phocid fossils come from zone 10 of the Calvert Formation [64] which is dated to the early middle Miocene ( [65]; and references therein). This renders a pre-middle Miocene age for MCZ 8741 less likely.
Comments. Originally presented as Phoca wymani, Ray [15] tentatively placed the original material of the species in Monotherium? wymani, as well as newly described specimens. The fossil record of M.? wymani is difficult to assess: The holotype MCZ 8741, a cranial fragment including the ear region and malleus, is valuable for the differentiation between different taxa of Phocidae [8,59], but the fossil record of Phocidae from the North Atlantic and Paratethys contains only a few ear regions [2]. Hence, the diagnostic value of fossil phocid ear regions is undermined. Additionally, a study of tympanic bullae of elephant seals, Mirounga, has shown that intraspecific variation is prominent [52]. Because other specimens attributed to M.? wymani only include disassociated and/or postcranial bones, Ray [15] could only tentatively assign them to M.? wymani. Moreover, the fossil record of referred specimens of M.? wymani does not include either humeri or femora, despite these bones being the most valuable postcranial bones for the identification of Phocidae [2,15]. The lack of humeri in the fossil record of M.? wymani precludes any  . These characteristics clearly support the identification of MCZ 8741 as a monachine (e.g. [15,43] [15] stated that the specimens came from one single block and are probably of one individual, but he implicitly expressed his doubt. Indeed, their association can be considered questionable in the absence of other preserved parts of the skeleton. Pending the discovery of more complete fossil specimens of Monotherium? wymani, we propose to restrict the species M.? wymani to the holotype tympanic bulla (MCZ 8741), discarding the remainder of the fossil record proposed by Ray [15], because it cannot be compared to the holotype. The holotype of 'Phoca' wymani cannot be compared to the original Monotherium material from Belgium and Italy, i.e. the holotype skull of Noriphoca gaudini, the lectotype humeri of Frisiphoca aberratum and Frisiphoca affine, and the lectotype pelvis of Monotherium delognii. Apart from M.? wymani, tympanic bullae have been presented for three other Neogene seals from the North Atlantic: Leptophoca proxima [2], Terranectes magnus and Terranectes parvus [66]. However, Dewaele et al. whether the tympanic bullae referred to T. magnus and T. parvus can be securely attributed to these taxa. Thus, the holotype of M.? wymani cannot be compared to other contemporaneous Phocidae from the North Atlantic. Therefore, we consider M.? wymani to be a monachine of unknown affinities, pending the discovery of more complete specimens that can be attributed to the taxon. Despite being uncomparable to Monotherium delognii, we provisionally retain the genus name Monotherium with a question mark. It is unknown whether M.? wymani indeed belongs to the genus Monotherium. It is likely that the holotype will eventually be designated into another genus, but this cannot be ascertained based on the current fossil record. Contrastingly, Monotherium gaudini is given a new genus name in this study, based on the argumentation that the systematic comparison and phylogenetic analysis (see above and below) of the more complete type material places it as a stem monachine, while the phylogenetic affinities of the holotype of M.? gaudini with other Monachinae cannot be ascertained (see below). The current genus name Monotherium is preferred as placeholder over the genus name Phoca, which was used prior to Ray's redescription [15]. The name Phoca is currently restricted to Phoca largha and Phoca vitulina: two phocine seals that show no affinities with M.? wymani (see below).

Subfamily Phocinae Gray, 1821
Genus Frisiphoca nov. gen. Etymology. From the Latin pronoun 'Frisicum' and the Greek noun 'phoke'. 'Frisicum' refers to the historical region of Frisia and the much smaller modern Dutch province with the same name. Here, the term is used in reference to the Mare Frisicum, the Latin name for the North Sea, alluding to the geographical origin of the two species of the genus listed here. 'Phoké' means 'seal'.
Comments. Although Frisiphoca shares characteristics with both Monachinae and Phocinae (see diagnosis), the presence of an entepicondylar foramen is regarded as a characteristic uniting Frisiphoca with Phocinae. Unfortunately, this is not reflected in the phylogenetic analysis due to the poor scoring of the fragmentary fossil record of the genus.    Type horizon. Van Beneden (unpublished handwritten notes in the IRSNB archives) assigned the specimen IRSNB 1191-M266 to the 'Diestien' (Diestian). However, as mentioned above, the Diestian is currently considered an obsolete term and should not be used any more [20]. Different authors assign different ages and stratigraphic intervals to the Diestian (see [19]: table 1), but in general it is considered that the Diestian is roughly equivalent to the Deurne Sands Member of the Diest Formation. Louwye et al. [21] assigned a Messinian to Tortonian (late Miocene) age to the Diest Formation north of Antwerp and in the Campine area. In a more detailed description of the Neogene stratigraphy of the Antwerp area, Mourlon [68] mentions the occurrence of Frisiphoca aberratum (and possibly also Frisiphoca [Monotherium] affine and Monotherium delognii) in one of the strata he described. This unnamed stratum, composed of greenish glauconiferous sands overlays-also unnamed-darker green and black sands [68]. Extrapolating this to the current knowledge on the stratigraphy of the Antwerp area, these greenish glauconiferous sands most likely represent the Deurne Sands Member of the Diest Formation, while the underlying darker green sands and black sands represent the Antwerpen Sands Member of the Berchem Formation [69]. Therefore, it can be assumed that F. aberratum comes from the Deurne Sands Member of the Diest Formation (table 1). Comments. Although Van Beneden [17] assigned partially articulated specimens to Frisiphoca [Monotherium] aberratum, these specimens, including two partial pes associated with a baculum and three caudal vertebrae (IRNSB 1187-M273a-o); a phalanx and a fifth metatarsal (IRSNB 1188-M270a, b); a partial hind limb including a second and third metatarsal, a partial fibula, and an ectocuneiform (IRSNB 1189-M271a, b); and a thoriac and cervical vertebra (IRSNB 1132-M269a, b), bear little diagnostic value. Given the overall rarity of such bones in the fossil record, they cannot be compared with other extinct phocid taxa from the southern North Sea basin, and the lectotype humerus has been selected as the type specimen of F. aberratum, degrading the other specimens to Monachinae indet., Phocidae indet., or Phocinae indet. (see Supplementary Information 3). A sediment sample associated with the vertebrae IRSNB 1132-M269 was analysed biostratigraphically with dinoflagellate cysts for our study.
The identification of this specimen as F. aberratum is questioned (this study; considered Phocidae indet.), but it should be mentioned that this sediment sample returned an age range from 7.25 Ma (latest Tortonian) to 3.7 Ma (late Zanclean), thus providing a minimum age interval for IRSNB 1132-M269 that does not contradict the stratigraphic assignment of the lectotype of F. aberratum.

Description and comparison
Humerus (figure 4a-d, i-l). In the absence of more complete, e.g. cranial, material, the humerus is the most diagnostic bone in Phocidae [2]. The humerus IRSNB 1191-M266 is overall well preserved, only missing part of the deltopectoral crest and portions of the distal epiphysis. The bone is straight and moderately slender. Ray [70] already noted that the humerus is relatively straight in some early Phocidae, such as Leptophoca proxima, as well as early stem pinnipedimorphs and terrestrial carnivorans, while most extinct and recent Phocidae have a more strongly curved humeral diaphysis. A measure for the 'slenderness' or 'robustness' of the humerus of Monachinae has been provided by Muizon & Bond [71]. This has been expanded to include Phocinae (table 3) and shows that the humerus in Phocinae is generally more slender than in Monachinae, although there is noticeable overlap. The humerus of F. aberratum is moderately slender and falls within the range observed in extinct and extant Phocinae (table 3). The humeral head is small and strongly hemispherical. As in Acrophoca longirostris, Hydrurga leptonyx and Otariidae, the head faces relatively proximally (contra [42]), relatively more posteroproximal than in other Phocidae, including F. aberratum. Unlike other Phocinae, except the extinct Leptophoca proxima [10], and the extant Histriophoca fasciata and Pagophilus groenlandicus, the neck is poorly developed in the species of Frisiphoca.
The deltopectoral crest is slender in anterior view. Although incompletely preserved, it appears that the deltoid tuberosity must have been located approximately halfway the length of the bone. The bicipital groove bears a small but noticeable transverse bar, which is also observed in the Monachinae Lobodon carcinophoca, Monachus spp., Ommatophoca rossii, Piscophoca pacifica and Pliophoca etrusca, and in the phocine Frisiphoca affine [42]. The lesser and greater tubercles reach slightly below the level as the head (contra [42]; for the greater tubercle). Among Phocidae, most extant taxa (except Monachus) have a strongly developed lesser tubercle and a small greater tubercle. While this condition is variable among extant taxa, many early phocid taxa bear a relatively little-developed lesser tubercle (e.g. Leptophoca proxima, Nanophoca vitulinoides). Geologically younger extinct Phocidae (e.g. Homiphoca capensis, Pliophoca etrusca) tend to have a lesser tubercle that shows a degree of development intermediate between early extinct Phocidae and extant Phocidae [9]. On the lateral surface of the greater tubercle, the insertion area for the infraspinatus and supraspinatus muscles is well outlined. The deltopectoral crest is overall slender and tapers smoothly towards the coronoid fossa, distally, as is typical for many extinct monachines and phocines (see genus diagnosis) [10,42]. Just distal to the humeral head and lesser tubercle, on the posterior surface of the diaphysis, there is a prominent fossa for the origin of the triceps brachii muscles.
An entepicondylar foramen is present, which is a characteristic shared with other Phocinae; among Monachinae, only Homiphoca capensis has an entepicondylar foramen [42,58]. The supinator crest is reduced, as in Monachinae, but contrasting to other Phocinae, including F. affine. The medial epicondyle is broad and flaring, as in F. affine. The lateral margin of this supinator crest is rugose, providing an origin area for powerful manual extensor muscles. The medial epicondyle is broad at the distal extremity of the supinator crest and medial to the medial condyle. The olecranon fossa and coronoid fossa are strongly reduced, these regions being almost completely flat. When compared to other Phocidae (except Homiphoca), at the distal epiphysis the ulnar articular facet of the trochlea is prominent in relation to the radial capitulum. This contrasts with F. affine, in which the coronoid fossa is less reduced. Overall, the humerus of F. aberratum combines characters which are otherwise considered typically monachine, or typically phocine. The poorly developed supinator crest is considered a plesiomorphic character retained in Monachinae, while the presence of an entepicondylar foramen is regarded as a plesiomorphic character retained in Phocinae [8,58]. Comments. In his redescription of Monotherium? wymani, Ray [15] identified specimen USNM 214625, a partial humerus, as ?Monotherium aberratum ( [15]: figs 8-11, subset 1). Despite the poor state of preservation, its general shape and size, and more specifically the proximal projection of the humeral head, the morphology of the greater tubercle, the shape of the insertion pit for the spinatus muscles, and the weak development of the supinator crest indicate similarities with the lectotype of Frisiphoca aberratum (figure 5a). However, given the poor state of preservation of the specimen, we deem it best to consider the specimen Phocidae cf. Frisiphoca aberratum. Dall [72] and Ray [15] considered the Gay Head Greensand to be part of the St Marys Formation, which is nowadays considered to be Tortonian (upper Miocene) in age (dinocyst zones DN 8 and 9 in Kidwell et al. [65]; 11.2-7.6 Ma from Köthe [73]).  fig. 1) [9,10,17,67]. This type locality is derived from the original labels associated to the specimen, while Van Beneden [17,18] did not discuss the geographical provenance of all specimens, restricting the description to 'third section'.
Type horizon. Van Beneden (unpublished handwritten notes in the IRSNB archives) assigned IRSNB 1118-M260 to the 'Diestien' (Diestian). However, as mentioned above, the Diestian is currently considered an obsolete term and should not be used any more [19]; in general, it is considered roughly equivalent to the Deurne Sands Member of the Diest Formation, dated to Messinian to Tortonian (late Miocene; [21]). In addition to Monotherium aberratum, Mourlon [68] mentions the possible occurrence of Monotherium affine and Monotherium delognii in an unnamed stratum, composed of greenish glauconiferous sands overlaysalso unnamed-darker green and black sands that most likely represents the Deurne Sands Member of the Diest Formation [69]. Comments. When Kellogg [18] elected Monotherium delognii as the type species for the genus Monotherium, he considered the differences between M. delognii and Monotherium affine minimal and regarded M. affine as a junior synonym to M. delognii. Based on the current fossil record, it is clear that, apart from the partial pelvis IRSNB 1153-M257a, b, none of the bones originally assigned to M. delognii bears enough diagnostic characteristics to identify it beyond the family level. This partial pelvis is selected as the lectotype of M. delognii, showing similarities to the pelvis of Prophoca rousseaui. However, it cannot be compared with F. affine due to the lack of a preserved pelvis in the known fossil record of F. affine. Therefore, we do not follow Kellogg [18] and do not regard M. affine as a junior synonym to M. delognii.

Description and comparison
Humerus ( figure 4e-h, m-p). The lectotype humerus of Frisiphoca affine IRSNB 1118-M260 is the only known humerus for the species and it is moderately well preserved. The humerus of F. affine differs relatively little from the humerus of Frisiphoca aberratum. Overall, the humerus is also slender and straight, but it is noticeably longer than the humerus of F. aberratum ( humerus precludes quantification of the 'slenderness' or 'robustness' as has been done for F. aberratum (table 3). However, the similar shape allows assuming similar robustness values for both Frisiphoca species. The humeral head of F. affine is strongly hemispherical. The bicipital groove is moderately wide and relatively open, i.e. the margins of the bicipital groove almost form a straight angle and are not U-shaped in section. The bicipital groove bears a little-developed but noticeable transverse bar. This transverse bar was also observed in a large number of Monachinae and F. aberratum [42]. The lesser tubercle does not reach the level of the head and the greater tubercle reaches the same level as the head. Among other Phocidae, comparable conditions have been observed in extinct taxa, such as the middle Miocene phocines Leptophoca proxima and Pachyphoca spp. In extant phocids only Monachus has a somewhat reduced lesser tubercle (see above). As in F. aberratum, the insertion area for the infraspinatus and supraspinatus muscles is well outlined on the lateral surface of the greater tubercle. Distally, the deltopectoral crest tapers smoothly towards the coronoid fossa. As already pointed out by Muizon [42], the posterior surface of the diaphysis bears a prominent fossa just distal to the humeral head and lesser tubercle, for the origin of the triceps brachii muscles. This was also observed in the South American stem lobodontins Acrophoca and Piscophoca, as well as in F. aberratum [42]. An entepicondylar foramen is present. The supinator crest is well developed, as in other Phocinae except F. aberratum. Comments. All Belgian specimens considered as Phocidae cf. Frisiphoca affine in this study had originally been identified and illustrated as Monotherium affine [17]. However, these specimens have been found isolated and are generally considered of very little diagnostic value. Another radius (IRSNB 1138-M267), astragalus (IRSNB 1144-M272) and calcaneum (IRSNB 1187-M273d) have originally been assigned to Monotherium aberratum, which are much smaller than the referred specimens IRSNB 1121-M261b, IRSNB 1126-M262, IRSNB 1125-M263. Therefore, the fossil record of phocids from the 'Diestian' includes the larger bones referred to in this section, as well as comparatively smaller radius, astragalus and calcaneum. The larger set can be assigned to Phocidae cf. Frisiphoca affine, based on the larger size of the specimens better matching F. affine, presumably from the same lithological unit. The American specimen of Phocidae cf. Frisiphoca affine, specimen USNM 187410, had previously been considered as Monotherium? wymani [15].

Description and comparison of the Belgian material
Ulna (figures 5b,c and 6). The anteroproximal portion of a right ulna is preserved (IRSNB 1121-M261a) ( figure 6). The anconeal process is located relatively more distal on the proximomedial surface of the olecranon process than it is in any other phocid, except specimen USNM 187410 (figure 5b,c), formerly assigned to Monotherium? wymani [15]. This yields a strongly sloping appearance for the anconeal process. The prominence of this anconeal process is similar to that in other Phocinae, while it is generally much reduced in Monachinae (except Homiphoca) [57]. The greater sigmoid cavity for articulation with the humerus is mushroom-shaped, with the upper, greater facet facing anteriorly. Gradually curving around the sigmoid notch, this facet transits distally in a smaller facet facing medially. The lesser sigmoid cavity (=radial notch) is located anterodistally of the greater sigmoid cavity. This cavity is circular, flat, and faces anterolaterally. Again, this matches well M.? wymani, but it should be highlighted that little attention has historically been given to the description of the shape of the sigmoid notch of fossil Phocidae. Published drawings and descriptions have shown appreciable differences in the shape of the sigmoid notch among extant Phocinae [59], but no detailed description has been provided for Monachinae [42,74]. Overall, the position of the anconeal process suggests that the ulna is phocine. However, being only one characteristic combined with the poor state of preservation of the specimens raises doubt on the validity of this assumption. Therefore, it is safer to consider the specimens Phocidae cf. In addition, a number of differences can be observed, separating this astragalus from the indeterminate phocid astragalus IRSNB 1144-M272.
The tibial facet is small, and proportionally smaller than in IRSNB 1144-M272 and other phocid astragali. The lateral and medial tibial facets form a right angle in proximal view, as in Monachus sp., Phocinae and Piscophoca pacifica; but this angle is slightly larger than in IRSNB 1144-M272. In lateral view, the tibial facet is convex. The tibial facet is separated from the caudal process by a 'neck', which is more pronounced than in other phocids, except Acrophoca longirostris. The caudal process is long, which is a typically phocid characteristic [58], and plantardorsally elongate, but mediolaterally not as thick as in astragalus IRSNB 1144-M272. The dorsoplantar height of the caudal process of IRSNB 1126-M262 is 30.0 mm, and the mediolateral width of the process is 16.6 mm. The dorsoplantar height of the caudal process of IRSNB 1144-M272 cannot be measured due to the poor state of preservation, but the mediolateral width is 18.5 mm, which is somewhat wider than in IRSNB 1126-M262. Although the proximodistally elongate ectal facet is not completely preserved, it appears proportionally thicker and less elongate than in IRSNB 1144-M272. However, the elongation of the ectal and sustentacular facets is marked, uniting both specimens with Australophoca changorum, Monachus sp., Phocinae and Piscophoca pacifica. The elongate sustentacular facet is well developed and highly raised over the astragalar head. This facet is convex and facing ventrolaterally; it is separated from the ectal facet by an interarticular sulcus. This sulcus forms a distinct oval fossa just ventroproximally of the ectal facet. Although this sulcus is present in other Phocidae as well [57], it only forms a deep and narrow fossa in IRSNB 1126-M262. The sustentacular facet is separated from the facet for the navicular. The facet for the navicular is plantardorsally elongate and runs along the entire distal margin and the distal portion of the plantar margin of the astragalus. The long axes of the caudal process and of the head form an obtuse angle of approximately 120°, making the astragalus arched in lateral view, as in extant Lobodontini. Given the strong similarities between the astragalus in Phocinae and some Monachinae, it is difficult to elucidate whether IRSNB 1126-M262 is monachine or phocine.
Calcaneum ( figure 8). Similar to the astragalus, one isolated right calcaneum (IRSNB 1125-M263) can very tentatively be assigned to Phocidae cf. Frisiphoca affine, largely based on its relatively large size in comparison to another calcaneum from the 'Diestian' of Antwerp that was originally assigned to Monotherium aberratum (IRSNB 1187-M273d; Phocidae indet. this study, see Supplementary Information 3). Although it has been found isolated, the specimen is much larger than the adult male calcaneum IRSNB 1187-M273d (see Supplementary Information 3 [42] considered a long calcaneal tuber to be a plesiomorphic characteristic. IRSNB 1125-M263 bears a prominent medial process at its proximal end that is well developed, as in IRSNB 1187-M273d. Halfway on the dorsolateral margin of the calcaneum, there is an oval, concave facet for the articulation with the fibula. Such a facet has also been observed in IRSNB 1187-M273d, and is generally considered more prominent in Monachinae than in Phocinae [42]. The trochlear process extends across the dorsal surface of the calcaneum, anterior to the ectal facet (=proximal astragalar facet). The astragalar articular facets are relatively long, as in Phocinae. The sustentacular facet (=distal articular facet) is slightly less slender than in IRSNB 1187-M273d and the curvature of the facet in medial view is less pronounced. The ectal facet (=proximal articular facet) is oriented anterodorsally-posteroventrally, but nearly horizontal, and is shorter than in IRSNB 1187-M273d and slightly thicker. In IRSNB 1125-M263, the length of the ectal facet is 26.6% of the total length (20. (8.6 mm : 20.8 mm) and 37.9% in IRSNB 1187-M273d (5.5 mm : 14.5 mm). Anteriorly, the sustentacular facet transits into the cuboid facet. The concave and lozenge-shaped cuboid facet is higher than wide and, contrasting to IRSNB 1187-M273d, not restricted to the dorsal two-thirds of the distal margin of the calcaneum. Although sharing a number of characteristics with extant Phocinae and not with extant Monachinae, because extinct Monachinae and Phocinae tend to exhibit an overall intermediate morphology [42], it is impossible to confidentially assign this specimen to either of both subfamilies and its comparison with F. affine is largely based on its size.

Description and comparison
Sacrum ( figure 9). The sacrum IRSNB 1153-M257a is only very partially preserved: only the sacral wings, and the bodies of the first and second sacral vertebrae are incompletely preserved and severely abraded, inhibiting a detailed description. The promontory is dorsoventrally slightly compressed (43.9 mm : 60.9 mm). The sacral wings (alae) are strongly laterally enlarged, but not as pronounced as in Prophoca rousseaui or some other phocids [10]: the transverse width across the wings is 2.70 times the lateral width across the promontory (164.4 mm : 60.9 mm) (3.31× in P. rousseaui, [10]). Although past studies indicated that this ratio is higher in Monachinae than in Phocinae [42], Dewaele et al. [10] showed that there is considerable overlap between the ratio ranges of Monachinae and Phocinae, preventing clear distinction between both subfamilies. The lateral margins of the sacral wings are badly preserved, but they appear to have been originally rounded. The ventral margins of the sacral wings are remarkably straight horizontally, as in P. roussseaui, but unlike other Phocidae. Other similarities with P. rousseaui are the dorsal and posterior offset of the wings in relation to the promontory, and the wide base of the prezygapophysis.
Innominate ( figure 9). The partial left innominate IRSNB 1153-M257b associated with the partial sacrum IRSNB 1153-M273a is noticeably larger than the innominates IRSNB 1192-M276 and IRSNB M2234 (both assigned to Prophoca rousseaui [10] [58,75]. Kawas benegasorum and Leptophoca proxima are returned as stem Phocinae, as has been shown before [9,10]. However, Devinophoca claytoni and Nanophoca vitulinoides are not returned as stem Phocinae, contrasting with a recent analysis by Dewaele et al. [9]. Historically, there has been little consensus on the phylogenetic relationships of extant and extinct Monachinae [6,8,76]. This first phylogenetic analysis returns F. aberratum and F. affine as stem Phocinae  figure S1 and table S5). In the phylogenetic analysis, the identification of N. gaudini is supported by one unequivocal autapomorphy: the ventral edge of the zygomatic arch is level with the alveolar plane (character 13, state '0' to '1').
0.46, HI is 0.54, RI is 0.69 and RC is 0.32. The topology of this analysis differs from the fourth analysis for the Phocinae, with the clade composed of Kawas benegasorum + Leptophoca proxima returned as a sister clade to all other included Phocinae, and with Nanophoca vitulinoides returned as the last stem phocine before crown Phocinae. However, the phylogenetic position of Noriphoca gaudini as the earliest-branching stem monachine is unchanged.

Phylogenetic analysis and palaeobiogeographical considerations
The phylogenetic relationships among the Phocinae in the cladistic analyses correspond with the previously published trees by Dewaele et al. [9,10], with Leptophoca proxima and Kawas benegasorum as stem Phocinae. The phylogenetic relationships of extant and extinct Monachinae are less well established and different previously published analyses show different topologies [6,8,76]: the specific relationships among the four lobodontin species remain poorly resolved [34,43,47,86], and there is no agreement whether extinct Monachinae from the Southern Hemisphere form a distinct clade [8] or not [6,87], or whether they are stem Monachinae or not. The phylogenetic trees obtained here clearly show that the extremely fragmentary fossil record of Frisiphoca aberratum and Frisiphoca affine precludes any detailed phylogenetic analysis and that it is for now impossible to resolve the subfamilial affinities of these taxa based on the current phylogenetic analysis. However, the presence of an entepicondylar in the lectotype humeri of F. aberratum and F. affine supports assigning the genus Frisiphoca to the Phocinae subfamily. Monotherium? wymani, on the other hand, is returned as a stem monachine, but the holotype is very incompletely coded (7/80 characters), and the exact phylogenetic relationships vary between different analyses (figure 10b,c) and remain unclear. Therefore, M.? wymani is considered a monachine of unknown phylogenetic affinities.
Noriphoca gaudini is a stem monachine, a result that agrees well with its geological age. Assuming that the type specimen of N. gaudini had indeed been collected from the bituminous layers of the Lepidocyclina Limestone of the Bolognano Formation, this species can be dated to the Oligocene or earliest Miocene (Aquitanian) ( [33,35]; and references therein), which is older than the Burdigalian (ca 19 Ma) Afrophoca libyca from Libya [36]. Consequently, our phylogenetic analysis, together with the presence of A. libyca in the early Miocene Mediterranean, may suggest a Mediterranean origin for the subfamily Monachinae or even the entire family Phocidae. Ancestral non-phocid pinnipedimorphs were all restricted to the North Pacific realm [4,[44][45][46][47][48][49][50][51][52][53][54][55]. The oldest pinnipedimorphs are Enaliarctos spp. from the late Oligocene, and the previously oldest known crown pinniped is Desmatophoca brachycephala from the Aquitanian of the northeast Pacific. It is broadly accepted that the direct ancestors of Phocidae crossed the Central American Seaway, with an origin of Phocidae along the east coast of North America ( [4]; contra [88]). However, no fossils of these stem Phocidae have been found or recognized, either near the Central American Seaway, or on the east coast of North America, or in Europe. Hence, it can be hypothesized that direct ancestors of Phocidae or the earliest Phocidae crossed the North Atlantic Ocean from the Central American Seaway to the Mediterranean Sea during the late Oligocene or earliest Miocene, shortly after diversification of early Pinnipedimorpha and Pinnipedia in the northeast Pacific. Direct migration from the Central American Seaway to the Mediterranean may indeed explain (1) the record of a monachine mandible in the Burdigalian to Langhian of Libya (Afrophoca libyca) [36], (2) the early diversification of Phocidae taking place in Europe [9,10], and (3) the lower diversity of the Phocidae during the early and middle Miocene along the east coast of North America, as compared to Europe [10].

Conclusion
Formerly, the extinct phocid genus Monotherium included five different species: Monotherium aberratum Van Beneden, 1876; Monotherium affine Van Beneden, 1876; Monotherium delognii Van Beneden, 1876; Monotherium gaudini (Guiscardi, 1870); and Monotherium? wymani (Leidy, 1853). After a careful revision, only the type species M. delognii is retained, and considered Phocidae indet., sharing affinities with the late middle to late Miocene North Sea stem phocine species Prophoca rousseaui. Monotherium aberratum and M. affine are replaced into the new, most likely phocine, genus Frisiphoca. Most of the original material from the late Miocene of the North Sea is considered non-diagnostic, with only a lectotype humerus retained for F. aberratum, one lectotype humerus for F. affine, and one lectotype pelvis for M. delognii. Frisiphoca aberratum, F. affine and M. delognii are dated to the late Miocene, based on stratigraphic provenance in the literature.
First identified as Phoca wymani and being restricted here to its holotype ear region, the species Monotherium? wymani from the east coast of North America cannot be compared to M. delognii, but the type tympanic bullae could be considered diagnostic; the taxon is not treated in more detail, pending the discovery of more complete fossils. However, some specimens previously referred to Monotherium from the east coast of North America share affinities with F. aberratum and F. affine from the North Sea Basin, indicating a certain degree of faunal communication across the North Atlantic during the late Miocene.