Diversity of late Neogene Monachinae (Carnivora, Phocidae) from the North Atlantic, with the description of two new species

While the diversity of ‘southern seals’, or Monachinae, in the North Atlantic realm is currently limited to the Mediterranean monk seal, Monachus monachus, their diversity was much higher during the late Miocene and Pliocene. Although the fossil record of Monachinae from the North Atlantic is mainly composed of isolated specimens, many taxa have been erected on the basis of fragmentary and incomparable specimens. The humerus is commonly considered the most diagnostic postcranial bone. The research presented in this study limits the selection of type specimens for different fossil Monachinae to humeri and questions fossil taxa that have other types of bones as type specimens, such as for Terranectes parvus. In addition, it is essential that the humeri selected as type specimens are (almost) complete. This questions the validity of partial humeri selected as type specimens, such as for Terranectes magnus. This study revises Callophoca obscura, Homiphoca capensis and Pliophoca etrusca, all purportedly known from the Lee Creek Mine, Aurora, North Carolina, in addition to their respective type localities in Belgium, South Africa and Italy, respectively. C. obscura is retained as a monachine seal taxon that lived both on the east coast of North America and in the North Sea Basin. However, H. capensis from North America cannot be identified beyond the genus level, and specimens previously assigned to Pl. etrusca from North America clearly belong to different taxa. Indeed, we also present new material and describe two new genera of late Miocene and Pliocene Monachinae from the east coast of North America: Auroraphoca atlantica nov. gen. et nov. sp., and Virginiaphoca magurai nov. gen. et nov. sp. This suggests less faunal interchange of late Neogene Monachinae between the east and west coasts of the North Atlantic than previously expected.

. Line drawings showing the basic morphological differences in the humeri of two different genera of extant Monachine. Five specimens of the leopard seal (Hydrurga leptonyx) indicated by full lines and one specimen of the Weddell seal (Leptonychotes weddellii) indicated by a dashed line. All specimens rescaled to the same proximodistal size. Arrows indicate the most important differences in the humerus between both genera.

Intra-and interspecific long bone variation
Understanding intra-and interspecific variation in long bone shape among Monachinae is crucial to assess the degree of completeness required for an isolated fossil humerus to be considered diagnostic. A detailed study on intra-and interspecific monachine humerus shape is beyond the scope of this study. However, preliminary qualitative observations suggest that complete or nearly complete humeri can be treated as diagnostic bones to differentiate between different species of Monachinae. A very basic qualitative comparison of four humeri of (both sexes of) Hydrurga leptonyx and one humerus of Leptonychotes weddellii suggests that humeri are sufficiently diagnostic to distinguish taxa when complete (figure 1). This figure shows that the four complete humeri of H. leptonyx differ relatively little from one another, while the humerus of L. weddellii differs from the humeri of H. leptonyx on key morphological features, such as the curvature of the diaphysis and the morphology of the distal portion of the deltopectoral crest. However, the number of morphologically varying characters is limited, suggesting that complete or nearly complete humeri are required to differentiate between closely related species. All characters listed in the diagnoses of taxa below are considered to represent interspecific variation, intraspecific variation being discarded based on personal quantitative observations of extant Table 1. Measurements (in mm) of the holotype humeri of Auroraphoca atlantica and Virginiaphoca magurai. Measurements were taken to the nearest 0.1 mm using an analogous caliper. Measurement approach follows Koretsky [1].  . . height head 38.9 28.9 .    . . width proximal epiphysis 62.5 56 3. Geological background and dinoflagellate cyst biostratigraphy 3. 1

. Callophoca obscura
Callophoca obscura is known from the 'Scaldisian' of the southern margin of the North Sea Basin, in Belgium, and of the Yorktown Formation in the Lee Creek Mine, Aurora, North Carolina, USA [2][3][4]. However, as mentioned above, the Scaldisian is currently considered an obsolete term and its use has been discontinued [25]. Different authors assign different ages and stratigraphic intervals to the Scaldisian ([25]: table 1), but in general, it is considered that the Scaldisian is roughly equivalent to the early Pliocene Kattendijk Formation [26][27][28]. However, many phocid fossils from the 'Scaldisian' show signs of reworking [13]. Most of the vertebrate remains recovered from the Kattendijk Formation come from the basal gravel [29], suggesting reworking of many of the specimens from the underlying Deurne Two sediment samples (sample L15-1105 from the large (possible male) humerus of C. obscura, IRSNB M1156, originally described as Mesotaria ambigua Van Beneden, 1876 and sample L15-1108 from a small (possible female) humerus of C. obscura, IRSNB 1214) recovered from bone cavities were palynologically analysed for organic-walled dinoflagellate cysts (dinocysts) and acritarchs (electronic supplementary material). Unfortunately, the revision of C. obscura humeri proved that the latter specimen cannot be assigned to the species unambiguously. Both sediment samples return a Messinian age (latest Miocene), which can be regarded as a minimum age of the C. obscura specimens. However, because reworking of dinoflagellate cysts has been detected, it is impossible to elucidate an absolute age interval.  [30] show that the basal Sunken Meadow Member of the Yorktown Formation has been dated to 5.0 Ma using foraminifera biostratigraphy, and Akers [31] presented a radiometrically measured age of 4.4

Pliophoca etrusca
From Italy, only the type specimen of Pliophoca etrusca is known. Tavani [35] assigned multiple other isolated and associated bones to Pl. etrusca, but Berta et al. [12] attributed these to Pliophoca cf. Pl. etrusca. A Piacenzian age (late Pliocene) has been assigned to the holotype skeleton: the layer of the Pliocene Argille Azzurre Formation it came from has been dated to 3.19-2.82 Ma [12,36]. The geological and historical background of the North American specimens of 'Pliophoca etrusca' is identical to the background of the North American specimens of H. capensis: Pl. etrusca from North America is only known from the Yorktown Formation at the Lee Creek Mine, which had been dated to the Zanclean (early Pliocene) (see above). 3. 4. Auroraphoca atlantica nov. gen. et nov. sp.
Two humeri that were previously assigned to Pliophoca etrusca from North America (USNM 181419 and USNM 250290) [4] have both been recovered from the Lee Creek Mine in Aurora, Beaufort County, North Carolina. As has been noted above, the fossiliferous stratum at the Lee Creek Mine is the early Pliocene Yorktown Formation, with the basal Sunken Meadow Member corresponding with the 5.0 Ma foraminiferal zone D18 and the overlying Rushmere Member radiometrically dated to 4.4 ± 0.2 Ma [30,31] (see above). 3. 5. Virginiaphoca magurai nov. gen. et nov. sp.
One newly described humerus (USNM 639750) that has been dredged off the Nottoway River bed at Franklin, Virginia, is assigned to the new taxon Virginiaphoca magurai. The holotype and only known specimen, humerus USNM 639750, was found ex situ by Joseph Magura and was 'clean', i.e. no original sediment had been attached to the specimens that could be used in elucidating its lithostratigraphic or biostratigraphic origin. However, the specimen shows only few signs of rolling, and hence, fluviatile transport of the specimen must have been limited. The possible stratigraphic provenance covers the upper Miocene Cobham Bay Member (CBM) of the Eastover Formation and the lower Pliocene Yorktown Formation (figure 2), because both outcrop in the river banks and riverbed of the Nottoway River in the Franklin area [30]. A similar strategy had been employed to 'date' the fossils of the river dolphin Meherrinia isoni from the nearby Meherrin River in North Carolina [37]. The Yorktown Formation is dated to the Zanclean [30]. Ward   pattern of specimen USNM 639750. Currently, no specimens that have been recovered from the Yorktown Formation at the Lee Creek Mine can be attributed to V. magurai. Although the latter does not prove the absence of the species throughout the entire Yorktown Formation elsewhere, it supports the claim that the CBM is the more likely origin of the type specimen.   [4].

Systematic palaeontology
Type locality and age. Third section at Deurne, Antwerp, Belgium. The 'third section' follows Van Beneden's discretization of the nineteenth-century fortification constructions around the city of Antwerp, with the third section at Deurne being located southwest to the Deurne district of Antwerp ( [8]: fig. 1) [3,38]. However, it should be noted that this type locality is derived from the original labels associated with the specimen. In his original publications Van Beneden [2,3] did not discuss the geographical provenance of individual specimens of the original fossil record of C. obscura.
V. Beneden (1877, unpublished data) assigned the specimen IRSNB 1198-M203 to the 'Scaldisien' (Scaldisian). However, as mentioned above, the Scaldisian is currently considered an obsolete term and the name is no longer used [25]. In addition, many of the apparently Scaldisian phocid remains show signs of reworking and for Nanophoca vitulinoides it has already been shown that the actual fossil record is older than the Pliocene range that is commonly accepted for Scaldisian phocids ( [13] contra [4] Figure 5. over one thousand isolated bones of the North Atlantic Ocean, including a rare partial cranium from North America (USNM 475486) (e.g. [2][3][4]; C.M. Peredo and L. Dewaele 2017, personal observation). As a result, C. obscura has been considered in many phylogenetic analyses of Phocidae, and Monachinae in particular (e.g. [12,16,[39][40][41][42]). However, all specimens that have diagnostic value (i.e. cranium, humeri, femora) have been found isolated, and only humeri can be compared with the lectotype humerus. All other types of bones have only tentatively been assigned to C. obscura in the past on the basis of the size of the specimens and the relative abundances in collections (e.g. [3,4]). Although it is likely that most of these specimens indeed belong to C. obscura, we invoke a conservative approach and reassign all nonhumerus specimens from C. obscura as Monachinae indet. An individual reassessment of all specimens is beyond the scope of this study and we limit our research to the humeri of the species.
Koretsky & Ray [4] synonymized Callophoca obscura and Mesotaria ambigua because they observed no anatomical differences other than size, arguing that size difference alone represents sexual dimorphism: the smaller C. obscura are proposed to represent females whereas the larger M. ambigua would represent males of the same species. Apart [4] seem to belong to C. obscura. In addition, Berta et al. [12] noted marked differences between collections from the Eastern and Western Atlantic, considering C. obscura a nomen dubium because the taxon is based on isolated specimens, which they deemed dubious for reliable taxonomy.

Description and comparison
Humerus. Because the lectotype of Callophoca obscura is an isolated partial humerus (IRSNB 1198-M203), other humeri are the only bones that can unambiguously be compared and assigned to C. obscura (figures 3a,b, 4a,b, 5a,b and 6a,b). Four isolated humeri from Belgium and a few tens of specimens from the east coast of North America can be assigned to the taxon. The humeral head is slightly compressed proximodistally, as stated by Koretsky & Ray [4], and the head faces proximoposteriorly, while it projects more posteriorly in Pliophoca etrusca. The humeral neck is poorly developed in C. obscura, while it is much more prominent in the holotype of Pl. etrusca and strongly overhangs the diaphysis posteriorly. The lesser tubercle reaches proximal to the level of the humeral head. This strongly resembles extant Phocidae (except Monachus) which all have a strongly developed lesser tubercle, reaching higher than the level of the humeral head, while other late Neogene Monachinae (e.g. Acrophoca longirostris, Homiphoca capensis, Piscophoca pacifica and Pl. etrusca) all tend to have a lesser tubercle that approximately reaches the same level as the humeral head, or distal to it (e.g. [12,22,23,32,34]). The greater tubercle is also welldeveloped, reaching the proximal level of the humeral head. A similar condition has been observed in the Neogene Monachinae Ac. longirostris, Auroraphoca atlantica (newly described taxon in this study), H. capensis and Pi. pacifica, in which the greater tubercle reaches the same proximal level as the humeral head, or exceeds it. This contrasts with other Monachinae in which the greater tubercle does not reach the proximal level of the humeral head [22][23][24].
The deltopectoral crest is poorly preserved in the lectotype humerus IRSNB 1198-M203 of C. obscura, but it is preserved in North American specimens. It is separated from the lesser tubercle by a wide and deep intertubercular groove. In the intertubercular groove, a distinct transverse bicipital bar is present, as in most Monachinae, except the extant Hydrurga leptonyx, Leptonychotes weddellii and the extinct Ac. longirostris, Homiphoca sp., and the newly described Virginiaphoca magurai [22,32]. The deltopectoral crest of C. obscura differs from that in other Monachinae in that it is both broad (contra other extinct Monachinae) and rounded triangular (contra extant Monachinae) in lateral view ( [24]: fig. 4). As in other Monachinae, the deltopectoral crest of C. obscura is rounded and distally terminates smoothly just proximal to the coronoid fossa (compare [24]: fig. 4). In lateral view, the deltopectoral crest is widest proximally. The deltoid tuberosity is strongly pronounced on the lateral side of the deltopectoral crest and located just proximal to the middle of the bone. In anterior view, the deltopectoral crest is slightly curving laterally and slightly offset laterally, as in Homiphoca, while it is straight in Pl. etrusca. In lateral view, the deltopectoral crest of C. obscura is uniquely broad yet subtriangular; and diaphysis has only a minor curvature. Extant Monachinae all have a diaphysis with strong curvature, while many extinct Monachinae retain a straighter diaphysis, such as Ac. longirostris, Pi. pacifica and Pl. etrusca. Yet, a number of extinct Monachinae, including Homiphoca and Properiptychus argentinus have a rather strongly curving diaphysis [12,15,22,23,32,45].
At the distal extremity of the humerus of C. obscura, the coronoid fossa on the anterior side is weakly developed, as is the olecranon fossa on the posterior side. The supinator crest on the lateral epicondyle is only little developed, which is a typical monachine trait. The medial epicondyle is better developed and appears as a rounded obtuse triangle in anterior view. Callophoca obscura lacks an entepicondylar foramen, similar to all other Monachinae, except Homiphoca and V. magurai nov. gen. et nov. sp. At the trochlear notch, the anterior margin of the radial head strongly slopes distolaterally, as in Homiphoca. In Pliophoca etrusca, this margin is more proximally convex.
Homiphoca Muizon and Hendey, 1980 are small but distinct, while the M 1 usually lacks such cusps and is distinct in having a strongly recurved and sharp, pointed principal cusp. The M 1 is the largest of the cheek teeth, with the principal cusp slanted posteriorly, and often with a small accessory cusp low on the long anterior keel of the principal cusp. The interorbital region is broad and tapers posteriorly as in crabeater seal Lobodon carcinophaga, but unlike all other monachines. In the auditory region the tympanic bulla covers the petrosal, while the mastoid forms a lip overlapping the posterior border of the bulla.' Holotype Type locality and age. 'E' Quarry, Langebaanweg, Cape Province, South Africa. Originally from 'bed 2' [45], reassigned to the Langeberg quartzose sand and Muishond Fontein peletal phosphorite members of the Varswater Formation, with an age of 5.15 ± 0.1 Ma (Zanclean, early Pliocene) [33].
Comments. Historically, Homiphoca capensis was represented by more than 3000 specimens from the 'E' Quarry, Langebaanweg, South Africa [32,34], and a handful of isolated specimens from the Lee Creek Mine in North Carolina [4]. Recent research of the Homiphoca material of the 'E' Quarry by Govender et al. [34] and Govender [46] showed the existence of at least two different morphotypes of Homiphoca in the fossil record. Govender et al. [34] presented two morphotypes of the cranium, scapula, humerus, radius, ulna, innominate, femur, and tibia and fibula; with morphological variation apparently exceeding that of intraspecific variation based on extrapolation of extant Monachinae. Later, Govender [46] performed a detailed analysis of the known crania of Homiphoca, confirming the presence of two (and possibly three) different morphotypes of Homiphoca crania.
Moreover, there is considerable disagreement on the possible existence of a yet-undescribed species of Homiphoca. Some researchers suggest that the observed morphological differences between the different morphotypes exceeds intraspecific variation [34,46], while others rather invoke profound sexual dimorphism and consider it unlikely for multiple closely related species to occupy the same ecological niche in the same environment and at the same time (C. de Muizon 2017, personal communication; L. Dewaele 2017, personal observation). Because the fossil record of Homiphoca consists of isolated crania, mandibles and postcranial bones, it is impossible to relate the holotype cranium to any of the isolated postcranial specimens, or to compare the different isolated postcranial bones reciprocally. Therefore, it is impossible to ascertain the correlation between the different morphotypes to different types of bones. Consequently, we judge it appropriate to consider the entire fossil record of Homiphoca with extreme care and only consider the holotype cranium SAM-PQ-L15695 to belong to H. capensis unquestionably. Following the reasoning for Callophoca obscura, all postcranial specimens of H. capensis should be considered as indeterminate monachine. However, the stratigraphy of Homiphoca from South Africa differs from C. obscura from the North Atlantic. Virtually all cranial and postcranial specimens attributed to H. capensis come from the LQSM and MPPM levels of the Varswater Formation in South Africa [32][33][34]46]. These levels have a very limited temporal range, and are virtually bonebeds. Hence, it can be argued as to what extent multiple closely related taxa can occupy the same ecological niche in the same area and at the same time. Furthermore, the high number of specimens found at Langebaanweg, the only known South African locality of Homiphoca renders it very likely that at least one of both morphotypes of humeri belongs to H. capensis. Therefore, it is safe to assume that all isolated seal specimens, which can be grouped into two morphotypes, may actually represent one sexually dimorphic taxon. But, pending the discovery of more complete articulated skeletons of H. capensis, we judge it most appropriate to consider specimens Homiphoca sp., rather than H. capensis or Monachinae indet. The stratigraphic range of C. obscura, on the other hand, is completely different in that the stratigraphic range of C. obscura from Belgium is poorly delineated and that the Yorktown Formation in North America had been deposited over a longer time span [30,47] with no clear fossiliferous horizon and most fossils recovered from spoil piles with no detailed stratigraphic context other than the formation [ fig. 50E-H

Comments.
The holotype specimen of Homiphoca capensis, the sole species currently defined in the genus Homiphoca, is a partial cranium that has been found isolated [32,45]. Recent research by Govender et al. [34] and Govender [46] showed that virtually all cranial and postcranial specimens attributed to H. capensis have been found isolated and that at least two different morphotypes exist, hypothesizing the presence of at least a second species of Homiphoca. However, the presence of two closely related taxa in the same locality and occupying the same ecological niche is questionable, as noted above.
Description and comparison Humerus. Koretsky & Ray [4] assigned two humeri from the Yorktown Formation at the Lee Creek Mine to Homiphoca capensis: USNM 187228 and USNM 214550 (figures 3c,d, 4c,d, 5c,d, 6c,d). However, the whereabouts of the latter specimen are unknown and the current study only focuses on specimen USNM 187228. The humeral head of Homiphoca specimen USNM 187228 from the east coast of North America faces proximoposteriorly, similar to Callophoca obscura and Homiphoca from South Africa. The neck is well developed and it is morphologically similar to that of Homiphoca from South Africa, but unlike that of C. obscura in which it is less developed, or Pliophoca etrusca in which it is also well developed but more strongly overhanging the diaphysis. The lesser tubercle of Homiphoca from North America reaches the proximal level of the humeral head, as in C. obscura, Homiphoca from South Africa, and Virginiaphoca magurai nov. gen. et nov. sp. (see below) from the Nottoway River, but unlike Pl. etrusca and Auroraphoca atlantica nov. gen. et nov. sp. (see below) previously considered Pl. etrusca from North America, which has a lesser tubercle that exceeds the level of the humeral head. A reduced lesser tubercle is considered a plesiomorphic trait among pinnipeds [22,49]. The lesser tubercle diverts strongly off the diaphysis. The greater tubercle does not reach the proximal level of the humeral head, which is considered a derived character in Phocidae, but appears to be present in all extant Phocidae and only some extinct Phocidae (see, e.g. Dewaele et al. [11,13]).
The deltopectoral crest is moderately well preserved in USNM 187228. It is separated from the lesser tubercle by a wide intertubercular groove. In the intertubercular groove, there is no transverse bicipital bar present. As mentioned above, such a bicipital bar is present in most Monachinae, except the extant Hydrurga leptonyx, Leptonychotes weddellii and the extinct Acrophoca longirostris, Homiphoca sp. from South Africa and V. magurai [22]. The deltopectoral crest of USNM 187228 is rounded and terminates distally proximal to the coronoid fossa. On the lateral side of the deltopectoral crest, the deltoid tuberosity overhangs the diaphysis, along the entire length of the tuberosity. This condition varies among Neogene Monachinae from the North Atlantic. In anterior view, the deltopectoral crest is slightly curving laterally and slightly offset laterally, as in C. obscura and Homiphoca from South Africa, while it is straight in Pliophoca etrusca. In lateral view, the diaphysis has a strong curvature, as in Homiphoca from South Africa and extant Monachinae. Many other extinct Monachinae retain a straighter diaphysis, such as Ac. longirostris, C. obscura, Piscophoca pacifica and Pl. etrusca [12,22].
At the distal extremity of the humerus of USNM 187228, the coronoid fossa on the anterior side is weakly developed, as is the olecranon fossa on the posterior side. The supinator crest on the lateral epicondyle is better developed than in other Monachinae, except Homiphoca from South Africa, and hence, also better than in C. obscura and Pl. etrusca. The medial epicondyle bears an entepicondylar foramen, which is a symplesiomorphic trait shared with Phocinae and other Pinnipedia, except Monachinae. Among Monachinae, the presence of an entepicondylar foramen is unique to Homiphoca and V. magurai nov. gen. et nov. sp. (USNM 639750, see below). Monotherium also has an entepicondylar foramen and has been considered a monachine [12,22], but the humeri of Monotherium aberratum and Monotherium affine Van Beneden, 1876 are considered to be phocine in a recent study by Dewaele et al. [50]. Overall morphology and the description of the Homiphoca specimens from the Lee Creek Mine, show noticeable similarities with the Homiphoca humeri from South Africa. The study of Govender et al. [34] shows relatively little difference between the two humerus morphotypes of Homiphoca from South Africa. Two of the most prominent differences observed are the degree of development and curvature of the deltopectoral crest and the shape of the deltoid tuberosity. The deltoid tuberosity is very incompletely preserved in specimen USNM 187228. The shape of the deltopectoral crest of USNM 187228 corresponds to that of morphotype I from South Africa, in which the deltopectoral crest 'does not follow the curve of the bone' [34, p. 142]. Overall, the humerus USNM 187228 is strongly similar to that of morphotype I of Homiphoca from South Africa, and differences are negligible. Therefore, it is indeed safe to consider Homiphoca sp. to be present in the Yorktown Formation of the east coast of North America.
Pliophoca Tavani, 1941 Type and only included species. Pliophoca etrusca Tavani, 1941 Diagnosis. As for the species.  Tavani, 1941 (figures 3e, 4e, 5e, 6e) Emended diagnosis. We retain the holotype specimen, described by Tavani [15], and redescribed by Berta et al. [12] as the sole unquestionable specimen of Pliophoca etrusca. We only emend the part of the humerus in the diagnosis presented by Berta et al. [12] and exclude Leptophoca True, 1906 from the list of stem monachines, following recent recent phylogenetic analysis of the taxon by Dewaele et al. [11,13]: Pliophoca is distinguished from Monachus by medial tuberosity on premaxillary reduced or absent, lateral extension on premaxillary present, upper incisors transversely compressed, femur epiphyses in which the distal epiphysis is wider than the proximal epiphysis, greater trochanter of femur higher than the head, and calcaneum slightly longer than the astragalus. Pliophoca is distinguished from stem monachines (i.e. Acrophoca, 'Callophoca', Homiphoca and Piscophoca) by the following derived characters: humeral head strongly overhanging the humeral neck, supinator ridge on humerus absent or poorly developed and metacarpal I longer than metacarpal II. Pliophoca is distinguished from Mirounga and lobodontines in retention of the following primitive characters: medial tuberosity on the premaxillary reduced or absent, mastoid lip does not cover the external cochlear foramen, procumbent upper incisors absent and intercondylar region of femur narrow and deep. Comments. The holotype specimen MSNUP I-13993 is the only known specimen that can be attributed to the species Pliophoca etrusca [12]. However, this holotype is relatively complete, including a partial cranium, limb bones and vertebral bodies [12,15,35], making it one of the most completely known phocid fossils. Originally, Tavani [35] assigned additional material to the species: A fragment of a mandible from Orciano Pisano and specimens attributed to multiple individuals, including the left and right mandibles of a single individual and multiple other cranial, axial and appendicular bones from near Saline di Volterra, are housed at the Museo di Geologia e Paleontologia in Florence and were referred to Pl. etrusca by Tavani [35]. However, Berta et al. ([12]: e889144-16) approached the material more conservatively and only retained the holotype specimen as Pl. etrusca, arguing about the additional specimens that 'most of it is from a different locality than the holotype and it consists, in part, of some noncomparable elements to the holotype (e.g. mandible), we feel that it is more appropriately referred to Pliophoca cf. Pl. etrusca'. In the current study, we encourage a more conservative approach in phocid palaeontology, advocating against the widespread practice of assigning isolated incomparable specimens to one or another taxon. Therefore, we support Berta et al. [12] in considering that it is not appropriate to consider the very partial specimens from the Museo di Geologia e Paleontologia in Florence as Pl. etrusca. In their review of Pliocene Phocidae from the east coast of North America, Koretsky & Ray [4] presented 171 specimens of Pl. etrusca from different localities in North Carolina and Florida: mostly isolated bones, but also including one associated scapula, humerus, radius and ulna (USNM 250290), and one associated humerus and femur (USNM 374222). Furthermore, this collection includes 16 maxillae, 8 isolated maxillary teeth, 44 mandibles, 18 isolated mandibular teeth, 35 humeri, 24 radii and 17 femora, based on the study by Koretsky & Ray [4]. However, these specimens show little similarities with the holotype from Italy, and are all reassigned to other taxa (see below) or as indeterminate Monachinae. This has also been postulated by Berta et al. [ [30].

Description and comparison
Scapula-One very incomplete scapula can be assigned to Auroraphoca atlantica. Of this bone, from specimen USNM 250290, only the proximal portion is preserved, but its identification as Au. atlantica is based on its articulation with a humerus that can be identified as Au. atlantica. The specimen is not described here because it is too incompletely preserved, rendering the significance of a description futile.
Humerus-Humeri USNM 181419 and USNM 250290 differ morphologically from both Callophoca obscura and Pliophoca etrusca, although Koretsky & Ray [4] formerly assigned both specimens to Pl. etrusca (figures 3f, 4f, 5f, 6f ). Humerus USNM 181419 is nearly complete, missing only the anteroproximal portion of the deltopectoral crest. Humerus USNM 250290 is associated with a partial scapula and an ulna. This specimen is only partially preserved and is clearly not an adult, as can be judged from the clearly visible suture between the proximal epiphysis and the diaphysis. The humerus of Auroraphoca atlantica is overall relatively slender compared with other contemporaneous monachine humeri [ North Atlantic and Homiphoca from South Africa. The greater tubercle extends proximally to the humeral head, and the lesser tubercle reaches proximal to both the humeral head and the greater trochanter. Among Monachinae, and Phocinae in general, the development of the lesser and great tubercles of the humerus show evolutionary trends [11,13,22]. Among Phocidae, the development of the greater and lesser trochanter shows opposite evolutionary trends, with the greater trochanter usually being well developed in extant Phocidae (except Monachus monachus), and less well developed in extinct Phocidae (except Callophoca obscura and Pl. etrusca) ( [11,13]; this study). Dewaele et al. [11,13] and Muizon [22] noted the reverse for the development of the lesser tubercle of the humerus in Phocidae. The strong development of the lesser tubercle in the humerus of Au. atlantica corresponds more with extant Phocidae and C. obscura and Pl. etrusca than with most extinct Phocidae. A broad bicipital groove separates the lesser tubercle from the greater tubercle and the deltopectoral crest. In the intertubercular groove, there is a transverse bicipital bar. The deltopectoral crest extends for almost two-thirds along the anterior side of the diaphysis, terminating rather abruptly, distally. This abrupt distal termination separates the humerus of Au. atlantica from the humeri of extant and other extinct Monachinae, which all have a smooth and more gradual distal termination [49,51]. The deltoid tuberosity extends along the proximal two-thirds of the lateral side of the deltopectoral crest. This tuberosity overhangs the diaphysis laterally in two distinct places, anteriorly and posteriorly. This condition varies among different taxa of Monachinae from the Neogene of the North Atlantic. A similar condition to Au. atlantica is observed only in the newly described humerus USNM 639750 from Virginiaphoca magurai (see below). In anterior view, the deltopectoral crest of humerus USNM 181419 is slightly convex laterally, and the crest terminates above the coronoid process. In anterior view, the distal epiphysis of USNM 181419 is equally wide as the proximal epiphysis. This character is shared with Homiphoca and with the newly described humerus from the Nottoway River (see below). In C. obscura and Pl. etrusca, the proximal epiphysis is broader than the distal epiphysis. The medial epicondyle strongly reaches distally. The epicondylar crest is poorly developed, but forms a short and prominent protuberance distally, compared with the condition in other Monachinae, resulting in a strongly reduced attachment surface for multiple manual extensor muscles. This condition is unique in Phocidae. In extant Monachinae, Pl. etrusca, and V. magurai nov. gen. et nov. sp., the epicondylar crest is absent or poorly developed, while it is moderately well developed along the epicondyle in other extinct Monachinae [12,22]. There is no entepicondylar foramen, a synapomorphy among Monachinae. Among Phocidae, only the extinct monachines Homiphoca and Monotherium, and Phocinae have an entepicondylar foramen [2][3][4]22,32,45,49,51,52]. Correlated, the weak development of the epicondylar crest yields an apparently straighter humerus in lateral view than it does in Homiphoca or V. magurai nov. gen. et nov. sp. from the Nottoway River, but not as straight as in C. obscura or Pl. etrusca. The olecranon and coronoid fossae are strongly reduced, and the trochlea is small. The medial ulnar lip of the trochlea is well-developed, forming a sharp ridge, as in Pl. etrusca and V. magurai, but unlike C. obscura and Homiphoca.
Ulna-Contrasting to other types of bone, Koretsky & Ray [4] assigned only one ulna from North America to Pliophoca etrusca: USNM 250290, which is associated with a scapula and a humerus. In the current study, this specimen is reassigned to Auroraphoca atlantica (figure 7a-c). Although the proximal epiphysis of the specimen is fused to the diaphysis, the absence of the distal epiphysis argues that this specimen belonged to a skeletally subadult seal [23], as had been argued by Koretsky & Ray [4]. However, it appears that the olecranon process in USNM 250290 is more strongly sloping than in Homiphoca and Pl. etrusca. The proximal portion of the ulna of M. mauretanica is also strongly sloping, with a strongly concave proximal margin of the olecranon process [19]. The anconeal process on the medial side of the proximal portion of the olecranon process is weakly developed, as in Pl. etrusca Homiphoca, M. mauretanica and Pl. etrusca. Partly related to this strong curvature of the sigmoid notch in lateral view, the coronoid process of USNM 250290 is more strongly pronounced than in Homiphoca and Pl. etrusca. The greater sigmoid cavity is saddle-shaped and simple (reversed) tear-drop shaped in USNM 250290, while it is much more sigmoidal in the Pl. etrusca holotype MSNUP I-13993, in which the distal extremity of this cavity strongly deflects medially; and the lesser sigmoid cavity is circular and flat, while it is slightly concave in Pl. etrusca. Overall, the sigmoid notch of ulna USNM 250290 resembles that of M. mauretanica. However, comparison involved illustrations of M. mauretanica and no direct observations. Moreover, the limited diagnostic value of ulna for Phocidae precludes comparing USNM 250290 with M. mauretanica.
The diaphysis of the ulna USNM 250290 is robust and strongly curved. In lateral view, the diaphysis is thicker than in Homiphoca, M. mauretanica and Pl. etrusca. In anterior view, the diaphysis is approximately equal to the proximal portion of the coronoid process, in USNM 250290 and Homiphoca, while the proximal portion of the coronoid process is wider than the diaphysis in Pl. etrusca, and while the diaphysis is wider than the proximal portion of the coronoid process in M. mauretanica [30,31].
Comments. In addition to humerus USNM 639750, other phocid specimens, including two complete femora (USNM 639748 and 639749), two radii (USNM 639751 and 639752), a tibia (USNM 639753), and two metatarsals (USNM 639754 and 639755), as well as terrestrial mammal vertebra (USNM 639756) have been dredged off the Nottoway River bed. These show a similar state of preservation with respect to completeness of the bone and color. However, a detailed description of the other phocid specimens is beyond the scope of the present study.
Description and comparison Humerus-The left humerus USNM 639750 of Virginiaphoca magurai is completely preserved. The humerus is short and robust (table 1; figures 3g, 4g, 5g, 6g). The humeral head is hemispherical and slightly compressed proximodistally. The degree of proximodistal compression is higher than in other monachine humeri from the Neogene of the North Atlantic. The neck is strongly reduced, compared with other Monachinae from the North Atlantic, giving the entire humerus a moderately straight appearance in lateral view: straighter than in Homiphoca sp. from North America and South Africa, similarly straight as humerus USNM 181419 of Auroraphoca atlantica, but not as straight as in Callophoca obscura or Pliophoca etrusca. Extant Phocidae and other extinct Phocidae all have a relatively more curved humeral diaphysis in lateral view. The lesser tubercle almost reaches the same proximal level as the head and does not divert strongly medially from the axis of the humerus. In living monachines, except the monk seals of the genus Monachus, the lesser tubercle is strongly developed and exceeds the level of the humeral head. This condition varies among extinct Monachinae, with, e.g. C. obscura, Homiphoca sp. and Pl. etrusca having a well-developed lesser tubercle, while Acrophoca longirostris, Piscophoca pacifica and Properiptychus argentinus from South America all have a relatively small lesser tubercle ( [22,51]; this study). The bicipital groove is moderately deep and wide. In the intertubercular groove, there is no transverse bicipital bar. The greater tubercle reaches slightly lower than the humeral head, as in extant Phocidae [22]. However, a number extinct monachine taxa have a greater tubercle that exceeds the level of the humeral head, including C. obscura and Homiphoca sp. (this study). The deltopectoral crest is long, approximately two-thirds the length of the entire bone and reaching the level of the proximal portion of the medial epicondyle. In lateral view, the humerus USNM 639750 of V. magurai is semicircular. This condition is roughly intermediate between the less expanded deltopectoral crest of Ac. longirostris, morphotype II in Homiphoca, Pl. etrusca and Pr. argentinus; and the deltopectoral crest of C. obscura, morphotype I of Homiphoca and Piscophoca pacifica (see [12,15,22,23,34]). The deltoid rugosity on the lateral surface of the deltopectoral crest is proximodistally elongated, widest proximally and tapering distally, and slightly overhanging the lateral surface of the deltopectoral crest in two separate places. The latter condition is observed in the humerus of Au. atlantica, but otherwise this varies among extinct Monachinae. In anterior view, the deltopectoral crest of V. magurai does not deflect as much laterally as it does in other Neogene Monachinae from the North Atlantic, except Pl. etrusca, and largely remains within the anteroposterior plane through the axis of the humerus. Among Monachinae, the presence of an entepicondylar foramen is a feature unique to the genus Homiphoca [22,32,45,49]. An entepicondylar foramen is also known to exist in Monotherium [22]. However, a study contesting Monotherium as a monachine genus is in review. However, it is also present in USNM 639750. The distal portion of the humerus of V. magurai does not differ significantly from that of Homiphoca, with a broad and little-developed medial epicondyle. The epicondylar crest is little developed, as is common in all extant Monachinae, Au. atlantica, and Pl. etrusca, while it is moderately developed in other extinct Monachinae [12,22]. As Berta et al. [12] observed in Pl. etrusca, V. magurai has a narrow but well-marked attachment surface for the m. extensor carpi radialis on the posterolateral margin of the epicondylar crest. This attachment surface appears less strongly developed in C. obscura and Homiphoca. The coronoid fossa is rather shallow and the roughly oval olecranon fossa is moderately deep. The shared presence of an entepicondylar foramen in V. magurai and Homiphoca sp. and the absence of a transverse bicipital bar in the intertubercular groove may suggest a phylogenetic relationship between both taxa. However, the incompleteness of the fossil record of V. magurai inhibits a detailed phylogenetic analysis. Comments. Fossil phocid specimens that Koretsky & Ray [4] assigned to Pliophoca etrusca from North America are reconsidered in this study (see below). Our reassessment shows that some specimens are incomparable with the holotype (and currently only definitely known) specimen of Pl. etrusca, such as is the case for mandibles and the mandibular dentition ([4] versus [12]  North America that are comparable with the Italian holotype of Pl. etrusca show noticeable differences with the holotype, despite the designation by Koretsky & Ray [4] (see below). Consequently, given the overall incompleteness of the fossil record of Monachinae from the North Atlantic, most of these redescribed specimens are to be considered Monachinae indet.
In the current study, we argue that only nearly complete or complete humeri should be used in the designation of new monachine taxa, in the absence of more complete and articulated specimens. Indeed, the Auroraphoca atlantica humerus USNM 250290 is articulated with a partial scapula and an ulna. However, the specimen is clearly juvenile, which we consider inappropriate for holotype designation.

Description and comparison
Maxilla and maxillary postcanines (figure 8a-c)-Koretsky & Ray [4] assigned 16 isolated maxillae, of which some include maxillary teeth, and eight isolated maxillary teeth from different localities in North Carolina and Florida to Pliophoca etrusca. For the original material of Pl. etrusca from Italy, i.e. the holotype, no mandibles or mandibular teeth are known [12]. In the absence of articulated specimens having both mandibles and mandibular teeth as well as bones that are comparable to the holotype specimen, it remains impossible to assign isolated mandibles and mandibular teeth to the taxon. The maxilla, on the other hand, is known from the holotype of Pl. etrusca and can be compared with specimens from North America; although they are very incomplete.
The state of preservation of the referred and illustrated specimens from North America inhibits a detailed description. The postcanine tooth row is positioned on a lowly raised alveolar process.
postcanine tooth row is slightly diverging posteriorly, as in Homiphoca. However, the incomplete state of preservation of the holotype cranium of Pl. etrusca from Italy inhibits comparison between the referred specimens and the Pl. etrusca holotype. The alveoli for the maxillary teeth in specimen USNM 205397 show that P1 is single-rooted, while the other postcanine teeth are double-rooted. Spacing between the alveolus of P1 and C and P2 is large and shows the presence of a diastema.
On the ventral surface of maxilla USNM 205397, on the palatal process of the maxilla, there is a broad and shallow groove parallel and close to the tooth row. Posteriorly, this groove terminates into the anterior palatal foramen, located at the level between P4 and M1. This condition strongly differs from Homiphoca (narrow, deep groove and termination posterior to the anterior alveolus of M1) and from the holotype of Pl. etrusca (shallow groove terminating anterior at the level of the posterior alveolus of P3) Consequently, although the maxillae are very incomplete, both for the referred specimens from North America and the holotype of Pl. etrusca, there are a few differences that set both operational taxonomic units (OTUs) apart from each other. We consider it more conservative to consider the referred specimens from North America as Monachinae indet.
Radius-Koretsky & Ray [4] assigned 24 radii from the east coast of North America to the species Pliophoca etrusca (figure 9a-d). One subadult specimen, USNM 250290 has been found associated with a scapula, a humerus and an ulna [4] (see above). However, the radii Koretsky & Ray [4] assigned to Pl. etrusca show marked differences with the Pl. etrusca holotype radius MSNUP I-13993.
Overall, the radii from North America Koretsky & Ray [4] assigned to Pl. etrusca are slender and elongate, with smooth distal expansion of the diaphysis. This condition has also been observed in the holotype radius of Pl. etrusca, as well as in the morphotype II radius of Homiphoca (see Govender et al. [34]). For morphotype I, Govender et al. [34, p. 142] argue that 'the shaft of morphotype I is fairly short and wide with the distal expansion beginning higher on the shaft'; i.e., the distal expansion of the diaphysis of morphotype I is more pronounced in morphotype I of Homiphoca. Also, the holotype radius of Messiphoca mauretanica, MNHN.F.ORN1, has a short diaphysis that widens very strongly, distally. In lateral view, the curvature of the radii is moderate, intermediate to weakly curved radius of Homiphoca and the more strongly curved radius of Pl. etrusca. On the proximal epiphysis, the radii which Koretsky & Ray [4] assigned to Pl. etrusca bear a well-marked articular facet for the ulna, with a strongly developed distal lip on the posterior margin. This condition has also been observed in Homiphoca and Pl. etrusca, and only to a lesser extent in M. mauretanica. The bicipital tuberosity is subcircular and lowly raised, contrasting to Homiphoca and Pl. etrusca in which it is rather ovoid and well raised (see Berta et al. [12]). The bicipital tuberosity is separated from the proximal epiphysis and the position on the diaphysis varies, being located either medially (USNM 307537) or posteriomedially (USNM 181504) on the diaphysis. Among other Phocidae, all extinct Monachinae, including Homiphoca, and Pl. etrusca have a medially located bicipital tuberosity, and extant Monachinae have a posteromedially located bicipital tuberosity [12,22]. On the diaphysis, the considered radii bear a weak rugosity approximately halfway along the posterior margin. This is the insertion area for the radius-ulna ligament and is similarly developed in Homiphoca, but very prominent in Pl. etrusca. On the anterior margin of the diaphysis, the insertion area for the pronator teres muscle is weakly developed, while the insertion areas for the supinator and brachioradialis muscles are smooth, i.e. well developed. This corresponds with observations in extant Phocinae, but contrasts with both extant and extinct Monachinae, except Monachus monachus and Piscophoca pacifica. In lateral view, the distal portion of the diaphysis is relatively smooth, as in extant Monachinae, while it possesses relatively well-developed grooves for manual tendons in other extinct Monachinae, including Homiphoca and Pl. etrusca. The scapholunar facet on the distal extremity of the radius is concavoconvex, as in other Phocidae.
The referred radii of Koretsky & Ray [4] share a number of characteristics with Pl. etrusca, and Monachinae in general. However, it should be clear that there are numerous differences between the referred humeri and the holotype radius MSNUP I-13993 of Pl. etrusca. Consequently, it is highly doubtful that the referred radii from North America can be assigned to Pl. etrusca, but that they should rather be considered as Monachinae indet.
Femur-Koretsky & Ray [4] assigned 17 femora from the east coast of North America to Pliophoca etrusca (figure 10a-c). One specimen, USNM 374222, includes an apparently associated humerus and femur. All other femora from North America that have formerly been assigned to Pl. etrusca constitute isolated specimens. Although the holotype femur of Pl. etrusca, MSNUP I-13993, is only partially preserved, it is clearly distinct from the 'Pl. etrusca' femora described by Koretsky & Ray [4]. Both adult (figure 10a-c) and juvenile (figure 10d) specimens from North America have been considered, judging by the degree of epiphyseal fusion [53]. It can also be noted that there are noticeable differences among rsos.royalsocietypublishing.org R. Soc. open   the different specimens that Koretsky & Ray [4] consider for Pl. etrusca: e.g. the greater trochanter is much broader but lower in USNM 243686 (figure 10a) than in the other illustrated adult specimens ( figure 10b,c). Considering the absence of a detailed quantitative study on the femora of Phocidae, it is yet unknown if the latter variation can be considered intra-or interspecific.
All femora are typically pinniped in that they are longitudinally short and transversely broader than they are anteroposteriorly wide, resulting in a strongly elliptical transverse cross section. In the considered specimens, the minimum diaphyseal width of the shaft is in the proximal portion. Berta et al. [12] presented the same condition for the holotype femur of Pl. etrusca. However, we disagree and consider the minimum diaphyseal width of the shaft in both Homiphoca and Pl. etrusca rather in the middle portion of the bone. In the specimens described by Koretsky & Ray [4], the distal epiphysis is wider than the proximal epiphysis in anterior and posterior view; similar to the condition in Homiphoca and Pl. etrusca. Although the femoral head is not completely preserved, it is hemispherical, as in Homiphoca and Pl. etrusca. On the contrasts, the neck does not appear as strongly marked as in Homiphoca and the holotype of Pl. etrusca. The greater trochanter of the specimen USNM 243686 is only lowly raised and transversely broad, contrasting to the other illustrated specimens from North America, as well as the holotype of Pl. etrusca, in which the greater trochanter is transversely less broad, but more highly raised. In all specimens from North America, the greater trochanter bears a distinctive trochanteric fossa. While this trait is shared among all Phocinae, such a trochanteric fossa is rarely present among Monachinae (including Pl. etrusca), only the extant Lobodon carcinophaga, and the extinct Homiphoca and Piscophoca pacifica. The femora that have formerly been assigned to Callophoca obscura also have a well-developed trochanteric fossa.
In anterior view, the diaphysis of the femur gradually widens distally into the distal epiphysis. As a result, the distal portion of the femur attains a strongly triangular appearance. This corresponds with the general condition among Monachinae (L. Dewaele 2017, personal observation). Dewaele et al. [11] also noted a strongly triangular distal part in the femur of the phocine Prophoca rousseaui. Yet, generally, the distal portion of the diaphysis widens abruptly in Phocinae. The distal portion of the holotype femur of Pl. etrusca attains a rather phocine appearance. The epicondylar crest is lowly raised and robust, gradually merging into the diaphysis, proximally (distinct thin ridge in Pl. etrusca). The lateral epicondyle is weakly developed and smoothly tapers into the diaphysis, proximally; while the lateral epicondyle is much more prominent in Pl. etrusca, recurving proximally. The lateral condyle is large and rounded and the medial condyle is correspondingly large. Both condyles are widely spaced and separated by a wide intercondylar groove. The patellar facet is wider than it is high, as is widely observed among Monachinae [22].

Virginiaphoca magurai
Overall, the femora which Koretsky & Ray [4] identified as Pl. etrusca from North America show little similarity with the holotype femur of Pl. etrusca. Consequently, their designation as Pl. etrusca is contested. Nevertheless, because we advocate for a more conservative approach to specimen designation in this study, we reassign the femora from North America that have formerly been identified as Pl. etrusca, to Monachinae indet. It should be noted that these femora are strongly similar to each other and may belong to the same taxon, and that they superficially resemble the femora that have formerly been assigned to C. obscura. However, as stated above, we limit the current fossil record of C. obscura to the lectotype humerus and other humeri.

Discussion
The fossil record of all extinct Monachinae taxa that have been described from the Neogene of the North Atlantic Ocean (including the Mediterranean Sea) has been reassessed (table 2). As shown from the descriptions of the different taxa and specimens, monachine humeri are generally easily distinguishable and may be considered as an acceptable basis for identifying different Monachinae. However, it should be clear that this requires (nearly) complete specimens. Hence, both species of Terranectes are considered nomina dubia: the holotype of Terranectes magnus is a very incomplete humerus, and the holotype of Terranectes parvus is a femur and cannot be compared with most other Monachinae from the North Atlantic realm. In the absence of a more complete fossil record, i.e. with more complete and articulated specimens, we strongly advocate the consistent use of at least nearly complete humeri (and associated or articulated bones) as type specimens of fossil phocid taxa. The noticeable intraspecific variation observed, as well as the interspecific similarities on some parts of the humerus support our assertion that incomplete specimens are insufficiently diagnostic for species identification.
The question about the validity of Messiphoca mauretanica is more complex: the holotype humerus MNHN.F.ORN1 is only partially preserved; however, it is associated with an ulna, a radius and vertebrae. This entails a more complete type specimen than for many other phocids, but ulnae, radii and vertebrae are rarely articulated or associated with type specimens in other Neogene Monachinae from the North Atlantic, precluding comparison (see above). Consequently, we consider it appropriate to leave M. mauretanica in limbo, pending the discovery of new specimens prior to further investigation regarding the validity of this species.
Comparison of the fossil monachine fauna from the late Miocene and Pliocene of Europe (east Atlantic) with that from the late Miocene and Pliocene of the east coast of North America (west Atlantic) reveal strongly different biogeographic patterns than have previously been proposed. Koretsky & Ray [4] suggested a strong connection between both sides of the North Atlantic, considering the co-occurrence of Callophoca obscura and Pliophoca etrusca on both sides. However, in light of our taxonomic revision, this is called into question. The absence of Pliophoca in North America and the presence of two newly described species that are currently only known from the east coast of North America suggest that there was little 'interchange' between monachine faunas on the east coast of North America and the North Sea Basin (and South Atlantic, when including Homiphoca). However, the temporal differences between the late Miocene and Pliocene fossil phocid-bearing strata from the North Sea Basin, the east coast of North America, and the Mediterranean, joined by the incomplete fossil record of both sides of the North Atlantic, with many taxa known from exceedingly fewer specimens, precludes being conclusive on the question of the palaeodiversity of phocids in the North Atlantic during the late Neogene.
The current fossil record indicates that there was little faunal interchange between the seal faunas from the east coast of North America and the North Sea Basin, with C. obscura being the only monachine seal recorded at both sites. The fossil record of Monachinae from the Neogene of the Mediterranean is small [6,7,12,15,35,[54][55][56] and primarily consists of very fragmentary material [6,35,[54][55][56]: the skull of Monotherium gaudini from the late Oligocene-early Miocene of Italy [7,50], Afrophoca libyca from the Burdigalian of Libya [6], indeterminate Monachinae from the middle Miocene of Italy and Malta [54][55][56], and Pliophoca etrusca and Pliophoca cf. Pl. etrusca from the late Pliocene of Italy. This precludes drawing detailed conclusions on palaeobiogeographic changes. However, the fossil record suggests the apparent absence of any late Miocene-early Pliocene Monachinae in the Mediterranean realm, while Monachinae were diverse in the North Atlantic during that time interval. On the contrary, Pl. etrusca was present in the Mediterranean realm during the late Pliocene, while Monachinae seem to have disappeared from the North Atlantic at that time.
Pliophoca etrusca is widely considered to be a sister-taxon to the modern monk seals of the genus Monachus [12], phylogenetically supporting a Mediterranean origin of the monk seals. The research presented in the current study shows a higher diversity of Monachinae from the east coast of North America than previously expected, corroborating Berta et al.'s [12] alternative hypothesis of a western North Atlantic origin for the genus Monachus. This hypothesis assumes that the genus Monachus originated on the east coast of North America and 'monachinine ancestral stock then presumably dispersed in two directions: (1) to the Mediterranean evolving into Pliophoca etrusca; and (2) to the Pacific, which led to the evolution of Neomonachus schauinslandi in Hawaii before close of the Panama Seaway (about 3 Ma)' [12]. Both hypotheses depend on the discovery of more complete specimens of late Neogene Monachinae from the North Atlantic in order to perform phylogenetic analyses and to create a solid basis for morphological arguments. However, recent molecular studies by Scheel et al. [57] led to the separation of the three monk seals into two distinct genera, with the Mediterranean monk seal remaining Monachus monachus, and the Hawaiian and Caribbean monk seals renamed N. schauinslandi and Neomonachus tropicalis, respectively. Both genera appear to have been split 6

Conclusion
After a careful revision of the fossil record of the Monachinae from the North Atlantic, we contradict Koretsky and Ray's [4] observation of Pliophoca etrusca on the east coast of North America during the late Neogene, and we describe two new taxa, from North America. This leads us to conclude that similarities between the Monachinae faunas on both sides of the Atlantic Ocean were actually less than previously expected, with Callophoca obscura being found on the east coast of North America and in the North Sea Basin, and with Homiphoca sp. from the east coast of North America and Cape Province in South Africa (table 2) and the monachine fauna from North America suggests less faunal interchange than previously assumed.
Formerly, the vast majority of the fossil Monachinae (and by extent Phocidae in general) from the Neogene of the North Atlantic was composed of isolated bones. Most notably, the species C. obscura included thousands of isolated bones in private and public collections that are not at all comparable with the isolated lectotype humerus. Our study strongly advocates ending the formerly common practice to use such a syntypic approach to group isolated and incomparable specimens as it is based on weak and untestable arguments. We strongly encourage researchers only to consider complete or nearly complete humeri as potential type specimens for fossil Monachinae, in the absence of cranial specimens or more complete material.
Data accessibility. All data for this paper is available within this paper or as electronic supplementary material of this paper. Dinoflagellate cyst biostratigraphy results are provided as electronic supplementary material. This published work and the nomenclatural acts it contains have been registered in ZooBank. The LSID for this publication is urn:lsid:zoobank.org:pub:D9F119E9-64FB-48E8-8509-4D1EFCE0B6C9.