Appendicular skeleton

Humerus (Fig. 2): The lectotype IRSNB 1149-M274 is the only humerus currently known forProphoca rousseaui. This isolated humerus is poorly preserved, missing the distal extremity, and with a capitulum which is nearly completely abraded.

The humerus is relatively slender, with a size comparable to that ofL. weddellii. Although the proximal part of the deltopectoral crest is abraded, it appears that the partially preserved bicipital groove is wide and deep. In other Phocinae, the bicipital groove tends to be much narrower than inProphoca rousseaui. In Monachinae, the bicipital groove is generally very wide and shallow. To some extent, there is a low-raised bicipital ridge connecting the lesser tubercle and the deltopectoral crest. A bicipital ridge has been previously described in the monachine crabeater sealL. carcinophaga and, to a lesser extent, in the monachinesMonachus Fleming, 1822,MonotheriumVan Beneden, 1877, andP. pacifica (de Muizon, 1981).

The lesser tubercle is moderately well developed, proportionally slightly smaller than the lesser tubercle in the extinct monachineP. etrusca (seeBerta et al., 2015). It is definitely not as large as in most extant Phocidae, exceptMonachus spp., but comparable in dimension to a number of Neogene phocids, including the monachinesA. longirostris andC. obscura (Leonard Dewaele, 2016, personal observation; see, e.g.,Van Beneden, 1877;de Muizon, 1981;Koretsky & Ray, 2008). The lesser tubercle is roughly cylindrical, with a circular scarred surface for the attachment ofm. subscapularis facing posteriorly.

Compared to extant Phocidae, the deltopectoral crest is typically monachine inProphoca rousseaui in that it distally gradually curves toward the medial epicondyle (seeHendey & Repenning, 1972). In all extant Phocinae, the deltopectoral crest has an abrupt distal termination. However, many extinct Phocinae, such asC. maeotica,K. benegasorum,Phoca vitulinoides, andS. sintsovi (Cozzuol, 2001;Koretsky, 2001; Leonard Dewaele, 2016, personal observation) show varying conditions that are intermediate between extant non-phocid pinnipeds (Odobenus Brisson, 1762 and Otariidae Gray, 1825), monachines and extant phocines (Fig. 3). From what is present of the distal portion of the humerus, the deltopectoral crest ofProphoca rousseaui (andLeptophoca proxima) appears to terminate near the distal end of the diaphysis, as inO. rosmarus, otariids, and monachines (exceptL. weddellii andL. carcinophaga) and a number of extinct phocines (e.g.,C. maeotica,K. benegasorum,Phoca vitulinoides, andS. sintsovi). On the other hand, the distal end of the deltoid surface is clearly demarcated by a weak but abrupt angle in the deltopectoral crest, as in Phocinae, instead of a smooth and gradual distal termination. An abrupt distal termination of the deltoid surface is typical for extant phocines (Hendey & Repenning, 1972), where it is even more pronounced than inProphoca rousseaui. Also, inProphoca rousseaui the deltopectoral crest is only moderately raised over the diaphysis, contrasting with the very large and prominent deltopectoral crest in most extant Phocidae. The deltopectoral crest is less developed in non-phocid pinnipeds, the extant monachineMonachus spp., and the extinct monachinesA. longirostris,Monotherium spp.,P. pacifica andP. argentinus (de Muizon, 1981;de Muizon & Bond, 1982).de Muizon (1981) also mentionsP. etrusca, but this disagrees with a subsequent description (Berta et al., 2015). The humeri of many other extinct Phocidae display an intermediate state between these extinct and extant Phocidae. The state of preservation does not allow measuring the deltopectoral angle (sensude Muizon, 1981). On the lateral side of the deltopectoral crest, the robust deltoid rugosity corresponds to the raised insertion area for them. brachioradialis. The deltopectoral crest does not overhang the diaphysis laterally, differing from most extant Phocinae (exceptC. cristata and to a lesser extent alsoE. barbatus). Another prominence on the medial side of the deltopectoral crest and distal to the brachioradialis prominence serves for the attachment of them. pectoralis.

On the posterior surface of the diaphysis, just distal to the damaged capitulum and the lesser tubercle, there is a shallow fossa for the origin of them. triceps brachii caput mediale; a similar fossa is described inMonotherium spp. andP. etrusca (de Muizon, 1981).

Radius (Fig. 4): The proximal part of an isolated radius with the circumference of the humeral articular fovea noticeably damaged (IRSNB 1147-M275) has been tentatively assigned toProphoca rousseaui, based on its markedly different shape compared to the radius of the contemporaryLeptophoca proxima IRSNB 1145-M280a: IRSNB 1147-M275 smoothly widens anteroposteriorly in its distal region while the widening is much less pronounced in IRSNB 1145-M280a fromLeptophoca proxima and the diaphysis of the radius of the latter forms an angle at the level of the bicipital tuberosity. Hence, with currently no phocid species other thanLeptophoca proxima andProphoca rousseaui known from the Berchem Formation, radius IRSNB 1147-M275 can tentatively be assigned toProphoca rousseaui.

The preserved portion is robust, as in Monachinae, and the diaphysis gradually widens anteroposteriorly. Although incompletely preserved, the ulnar facet on the medial margin of the humeral articular fovea is prominent. The bicipital tuberosity is smooth and oval in outline, relatively little developed, and located medial to posteromedial on the proximal portion of the diaphysis. In other Phocinae, the location of the bicipital tuberosity is variable, while this tuberosity is located medially in Otariidae Gray, 1825, and the monachinesP. argentinus, andP. pacifica and located posteromedially in other Monachinae. This tuberosity is not located as high proximally as inP. etrusca (Berta et al., 2015), but a little separated from the proximal epiphysis, as in the extant Phocinae, the extant monachine Mediterranean monk sealM. monachus, and the extinct monachinesA. longirostris andP. pacifica (de Muizon, 1981;Amson & de Muizon, 2014). The diaphysis is straight along the bicipital tuberosity, similar to Monachinae. In Phocinae, the diaphysis exhibits a pronounced obtuse angle at the level of the bicipital tuberosity, delineating the location of the oblong rugosity.

The origin ofm. supinator is clearly outlined on the anterior margin of the diaphysis, indicating a rather strong development of this muscle. This corresponds with observations in extant and extinct Phocidae, except Lobodontini Gray, 1869, and Otariidae (de Muizon, 1981).

Innominate (Fig. 5): Three innominates are assigned toProphoca rousseaui.Van Beneden (1877) assigned two associated innominates toProphoca rousseaui (IRSNB 1192-M276b). These innominates have been found together with the sacrum IRSNB 1192-M276c and articulate perfectly with it. Hence, the sacrum and both innominates should be considered as belonging to the same individual. The third, isolated innominate IRSNB M2234 has been described byLouwye et al. (2010), but was identified as Phocidae aff.Prophoca based on minor morphological differences with the two aforementioned innominates; i.e., having a more slender anterior crest of the ilium and a mediolaterally more flattened anterior portion of the pubis and ischium.

In this paper, we tentatively assign this third, isolated innominate toProphoca rousseaui. AsLouwye et al. (2010) anticipated, we consider it very likely that the size and shape differences between this smaller innominate and the larger innominates IRSNB 1192-M276b may be attributed to sexual dimorphism. Sexual dimorphism and size differentiation is prominent among Phocidae (King, 1964).King (1964) also noted specific differences between male and female pinniped innominates, but these apply to the shape and size of the ischiac spine and the posterior portion of the ischium and pubis, which are neither preserved in IRSNB 1192-M276b nor in IRSNB M2234. Nevertheless, we consider the overall similarities between these innominates to outweigh the differences and tentatively consider IRSNB M2234 from Posthofbrug to representProphoca rousseaui. If so, the smaller, from Posthofbrug, would most likely represent a female and the larger a male, based on comparison with sexually dimorphic extant pinniped taxa (King, 1964). For all three innominates, only the ilium and the most anterior parts of the ischium and pubis are preserved.

In articulation with the sacral wing (at the auricular surface), the lateral surface of the innominate tends to face slightly dorsally. The same trait has been observed in the extant monachineM. monachus. The ilium is strongly elongate compared to most other Phocidae (see below). In general, Phocidae have a strongly reduced ilium compared to other carnivorans, which is presumably related to their aquatic locomotion, heavily relying on pelvic oscillations (e.g.,Berta & Wyss, 1994;Bininda-Emonds & Russell, 1996).de Muizon (1981) considered the relatively less reduced ilium of the monachinesM. monachus andP. pacifica to be plesiomorphic. The innominates ofProphoca rousseaui are incompletely preserved, precluding precise quantification of its shape. However, it appears that the ilium ofProphoca rousseaui is relatively elongated in relationship to the rest of the innominate, as compared to other Phocidae exceptM. monachus andP. pacifica. The ilium ofProphoca rousseaui has a markedly triangular shape (Van Beneden, 1877), having a pronounced iliac crest anteriorly and being dorsoventrally pinched posteriorly, as in phocines, contrasting with the roughly rectangular ilium of monachines (except the leopard seal,H. leptonyx, and the Ross seal,O. rossii) (de Muizon, 1981). The small lateral eversion of the ilium and the very shallow gluteal fossa inProphoca rousseaui are shared with Monachinae and the basal phocinesE. barbatus andK. benegasorum, and to a lesser extent withC. cristata. The anterior iliac crest is slightly laterally convex inProphoca rousseaui, which has also been observed in the extant Phocinae, the extant monachinesL. weddellii,L. carcinophaga, andM. monachus, and the extinct monachinesH. capensis andP. pacifica, but not in the extant monachinesH. leptonyx, O. rossii and the extinct monachineA. longirostris.

The posteroventral process of the ilium is well developed, as in other Phocinae (exceptC. cristata andK. benegasorum), and much better developed than it is in Monachinae (exceptMonachus spp.). However, it is still much better developed in the larger IRSNB 1192-M276b than it is in the smaller IRSNB M2234 from Posthofbrug.

The iliopectineal eminence is well developed. Hence, the ventral margin of the ilium between the iliopectineal eminence and the posteroventral process of the ilium, corresponding to the area of origin for them. psoas major andm. quadratus lumborum, is strongly concave.

On the lateral side of the ilium there is a well-developed fossa for the origin of them. rectus femoris. This fossa is best developed in the smaller Posthofbrug specimen. As has been pointed out byde Muizon (1981), in Monachinae, the anteroventral and anterodorsal processes of the ilium are at the same anteroposterior level, giving a vertical aspect to the anterior margin, whereas the anteroventral process is anterior to the anterodorsal process in Phocinae,Mirounga Gray, 1827,Monotherium delogniiVan Beneden, 1877, andProphoca rousseaui.

The acetabulum is very deep, as in other Phocinae andMonachus spp., and located high dorsally on the lateral side of the innominate.

The pubis and ischium are only very fragmentarily preserved in the three innominates described here. On the dorsal margin of the ischium, just posterior to the level of the anterior margin of the obturator window, there is a strongly developed, highly raised and anteroposteriorly elongated ischiac tuberosity.

Femur (Fig. 6): Within the material originally described byVan Beneden (1877), only one right femur is present (IRSNB 1150-M277a). This femur has been found in association with a partial tibia (IRSNB 1150-M277b). The capitulum is missing, as well as the anterodistal part of the medial epicondyle. A possible second femur only consists of a partial femoral capitulum.Louwye et al. (2010) tentatively referred this specimen to Phocidae aff.Prophoca, based on its stratigraphical occurrence and its size matching the innominates ofProphoca rousseaui IRSNB 1192-M276b. Based on the aforementioned arguments, we support this tentative designation fromLouwye et al. (2010).

The femur attains a nearly triangular shape in anterior view, with the minimum width of the diaphysis located proximally and strongly broadening distally. A roughly similar triangular shape can be observed in extant Lobodontini. The stout greater trochanter is strongly rounded in anterior view. The proximal part of the greater trochanter being abraded, if complete a lip would most likely have overhung the trochanteric fossa, similar to other extinct and extant phocines. The anterolateral surface of the greater trochanter is heavily marked with muscle scars for the gluteus muscles. The trochanteric fossa is deep and slit-like. A deep trochanteric fossa is characteristic for Phocinae (King, 1956;de Muizon, 1981). Among monachines, only the extantL. carcinophaga and the extinctC. obscura show a similarly deep trochanteric fossa (de Muizon, 1981;Koretsky & Ray, 2008; Leonard Dewaele, 2016, personal observation).

Medially, the epicondylar crest is low, only weakly projecting, as in other Phocinae; but it is transversely thick and straight along its entire length. The lateral epicondyle is equally indistinct as the epicondylar crest, a feature shared with Monachinae (de Muizon, 1981). Distally, the transition between both the epicondylar crest and the lateral epicondyle and the diaphysis is marked by two low, proximodistally oriented ridges (one medial and one lateral). Such ridges are not found in extant phocids, but are presumed to correspond to the origin for powerfulm. gastrocnemius andm. flexor digitorum superficialis.

The femur ofProphoca rousseaui has a much deeper and more clearly outlined suprapatellar fossa than extant Phocidae, similar to many other, extinct Phocinae, includingPhoca vitulinoides (Leonard Dewaele, 2016, personal observation). This suprapatellar fossa is slightly smaller in size than the patellar facet.

The lateral margins of the patellar facet are not preserved, but the facet appears slightly higher than wide. This corresponds with the condition in Phocinae and the extinctA. longirostris, and points toward an increased mobility of the knee joint (de Muizon, 1981). Monachinae other thanA. longirostris have a patellar facet that is wider than high, relating to a decreased mobility of the knee joint.

The intercondylar fossa is wide and deep. The medial condyle is smaller and less strongly curving than the lateral condyle. Although the medial condyle is not entirely preserved in IRSNB 1150-M277a, it is much smaller compared to the lateral condyle. Generally, the size difference between the lateral and medial epicondyles tends to be larger in Phocinae than in Monachinae (de Muizon, 1981).

Tibia (Fig. 7): The proximal portion of the tibia IRSNB 1150-M277b (associated to the femur IRSNB 1150-M277a) is assigned toProphoca rousseaui.Van Beneden (1877) originally assigned a second left tibia (IRSNB-VERT-3250-15) toProphoca rousseaui as well. However, for IRSNB-VERT-3250-15 only the proximal part of the tibia is preserved and only the little diagnostic diaphysis is visible. The tibial plateau is embedded in consolidated matrix. Hence, no morphological observations can be invoked to attribute the specimen toProphoca rousseaui unambiguously, and IRSNB-VERT-3250-15 is only tentatively assigned toProphoca rousseaui, on the basis of the overall similarity in size and morphology with the diaphysis of IRSNB 1150-M277b.

In IRSNB 1150-M277b, the suture between the diaphysis and the proximal epiphysis is visible, suggesting a skeletally almost adult specimen, based on observations in extant Phocinae (Storå, 2000). The tempo of fusion of the diaphyses and epiphyses of different bones varies within each taxon. Although the tempo of fusion of the diaphysis and epiphyses of different bones varies within each taxon, in phocine seals, the fusion of the diaphysis and proximal epiphysis of the tibia is one of the last steps to skeletal maturity (Storå, 2000).

The tibia only slightly curves laterally. This corresponds with Monachinae, in which the tibia generally curves less than it does in Phocinae (de Muizon, 1981). On the tibial plateau, anterior and posterior intercondyloid areas separate the articular condyles for the femur from each other. These intercondyloid areas are separated by the intercondyloid eminence. Both articular condyles are concave, yet the concavity of the medial condyle is faint. This difference may be due to the size difference between the two condyles, the lateral being considerably larger than the medial. The intercondyloid eminence is prominent and raised, similar to Phocinae and many extinct Monachinae such asH. capensis,Monotherium aberratumVan Beneden, 1877, andP. pacifica, (de Muizon, 1981; Leonard Dewaele, 2016, personal observation). The anterior intercondyloid area is well outlined inProphoca rousseaui and the popliteal groove in the posterior intercondyloid area is deep. A well-outlined anterior intercondyloid area is a phocine character that has also been described for the monachineH. capensis,Monotherium aberratum, andP. Pacifica (and also forProphoca Rousseaui) byde Muizon (1981).

Although poorly preserved, the patellar facet on the proximo-anterior margin of the tibia is shallow. In other Monachinae, this facet varies from low to high anterior to the main body of the tibia. The patellar facet is roughly semicircular in anterior view.

The anterior and posterior tibial fossae, form. tibialis cranialis andm. tibialis caudalis respectively, are both moderately excavated.

Axial skeleton

Lumbar vertebrae (Fig. 8); Currently, three lumbar vertebrae IRSNB 1192-M276a have been assigned toProphoca rousseaui. These vertebrae were found articulated with a sacrum and two innominata (IRSNB 1192-M276b-d). None of the three lumbar vertebrae is preserved (sub-) completely, but combined they give a good representation of the lumbar vertebrae ofProphoca rousseaui.

Based on the slightly reduced length of one vertebral body (54.6 mm) compared to the other vertebra with its body completely preserved (62.1 mm), the shorter one can tentatively be identified as the fifth, and last, lumbar vertebra (L5). Moreover, the shorter lumbar vertebra ofProphoca rousseaui lacks the two hemal processes, caudal on the ventral surface of the vertebral body, which is a general feature in Phocidae. The hemal processes are clearly present in the other completely preserved vertebral body. The two other vertebrae are more difficult to locate precisely. Not only does the mediocre state of preservation of the lumbar vertebrae inhibit to be conclusive on the subject, overall little variation between L1 and L4 in specimens of extant seals renders it difficult to extrapolate from them and pinpoint the exact position of individual lumbar vertebrae in extinct Phocidae.

The lumbar vertebrae ofProphoca rousseaui are anteroposteriorly elongated. The anterior and posterior articular surfaces are oval in L5 and reniform in the other two vertebrae.

The transverse processes are not preserved completely in any of the three vertebrae, but they appear to be robust, thick, and projecting anterolaterally. Although incomplete, the transverse processes in L5 ofProphoca rousseaui are proportionally relatively long, intermediate to the generally relatively short lumbar transverse processes of extant Monachinae and the relatively long lumbar transverse processes of extant Phocinae (Leonard Dewaele, 2016, personal observation).

The vertebral foramen is rectangular in shape and wider than it is high in all three specimens. We observed a similar state in the phocine harp sealPagophilus groenlandicus, and ringed sealPusa hispida, and in the monachine Mediterranean monk seal,M. monachus. In other extant Phocidae the lumbar vertebral foramina are semicircular or faintly dorsally pointed.

The lumbar prezygapophysis inProphoca rousseaui is large and robust, compared to extant Phocidae. Also contrasting to extant Phocidae, the teardrop-shaped articular surface of the prezygapophysis is relatively large in size inProphoca rousseaui. As a consequence, the blunt mammillary process is relatively small compared to the anterior articular surface inProphoca rousseaui. The mammillary process is slightly offset posteriorly to the center of the anterior articular surface.

Sacral vertebrae (Fig. 9): Of the original material assigned toProphoca rousseaui, one sacrum is present, with only the first sacral vertebra and the anterior part of the second sacral vertebrae preserved (IRSNB 1192-M276c). The preserved parts are poorly preserved, i.e., fragile processes and other elements are broken off, precluding a detailed description.

The sacral wings (alae) are strongly laterally enlarged inProphoca rousseaui, with a transverse width across the wings 3.31 times the lateral width across the promontory (131.5 mm/39.7 mm) (Fig. 10). In general, the lateral expansion of the sacral wings is considered as a synapomorphy distinguishing Phocidae from other Carnivora and among Phocidae these wings were considered more expanded in Monachinae than in Phocinae (de Muizon, 1981). However, the range of ratios observed in both living and fossil Phocinae and Monachinae (Supplemental Information 3) prevents clear distinction between the two subfamilies. The lowest ratio observed in our dataset of 56 specimens of 14 extant and extinct phocid species) (63.5:29.0 mm; ratio 2.19) is attributed to a specimen of the phocine harbor sealPhoca vitulina, while the highest ratio (147.2:41.7 mm; ratio 3.53) is attributed to a specimen of the fossil monachineC. obscura. The dataset shows intraspecific variability of the ratios, generally ranging between 10 and 40%, with the largest differences between highest and lowest ratios observed inPhoca vitulina (38.7%) andC. obscura (35.2%). Given the limited number of specimens included and because the sex of many specimens measured during this study remains unknown, no statistically supported differentiation between male and female could be established. Actual intraspecific variability is most likely higher than appears from the dataset (Supplemental Information 3). Notwithstanding, the sacral width ratio of 3.31 observed in the single sacrum attributed toProphoca rousseaui is at the upper limit of the entire range observed (seeSupplemental Information 3). Although the value of 3.31 forProphoca rousseaui exceeds all observed extant phocids, but larger ratios have been observed in certain specimens of extinct phocids, i.e.,C. obscura,Phoca vitulinoides, andPhocanella pumilaVan Beneden, 1877.

Published studies on the musculature of pinnipeds are scarce (e.g.,Bryden, 1971;Piérard, 1971;English, 1977), butde Muizon (1981) proposed that them. erector spinae originated on the anterior surfaces of the sacral wings ofA. longirostris. Contrastingly,Howell (1929) found that them. erector spinae originated on the anterior border of the iliac crest in the ringed seal,Pusa hispida (Phoca hispida) (Schreber, 1775). However,Howell (1929) noted the strong development of theerector spinae muscles inPusa hispida compared to the Californian sea lion,Zalophus californianus the latter having much smaller sacral wings; andKing (1964) observed the strong development ofm. iliocostalis lumborum, grouped within theerector spinae muscles, in phocids. Hence, the development of the sacral wings may tentatively be linked to the development of them. erector spinae and the ability to extend and rotate the dorsum.

The lateral margins of the sacral wings are not preserved, hampering proper description. Yet, based on preserved parts, the reconstruction of the lateral outline of the wings appears convex laterally. This differs from extant Phocinae and Monachinae in which the lateral margins being straight and are divergent ventrally (Phocinae) or vertical (Monachinae) (or even ventrally convergent inA. longirostris). The wings are slightly offset dorsally, relative to the promontory, and project laterally, i.e., they do not project ventrolaterally as in other Phocidae.

The prezygapophysis has a thick base on the sacral wing. The anterior articular surface of the prezygapophysis is large, with a rounded triangular outline. Right and left articular facets are perpendicular to each other. The mammillary processes are virtually absent, as in many Phocidae (except the phocinesP. groenlandicus andPusa spp. and the monachineL. carcinophaga).

The promontory is slightly compressed dorsoventrally in anterior view. The sacral canal is subtriangular in section, though with strongly rounded corners. Similar to extant Phocinae, the pelvic sacral foramina are relatively small compared to extant Monachinae; the spinous process has a mediolaterally thick base.

LeptophocaTrue, 1906

• Type and only included species—Leptophoca proxima (Van Beneden, 1877), by subsequent designation (this work).

• Diagnosis—Same as for the only included species.

• Leptophoca proxima (Van Beneden, 1877) n. comb.

• Prophoca proximaVan Beneden, 1877: 802, Plate XVIII.

• Leptophoca lenisTrue, 1906: 836–840, 2 plates.

• Leptophoca proximaRay, 1976a: 391–406.

Lectotype: Right humerus, IRSNB 1146-M279; illustrated and described byVan Beneden (1877, Plate XVIII, Figs. 12–14). Selected byKoretsky (2001) as lectotype ofProphoca proxima.

Type Locality: ‘Borgerhout, third section,’ Antwerp, Belgium. This section of the former fortification ring around the city of Antwerp runs from the district of Deurne to the former ‘porte de Borsbeek.’ The district of Borgerhout encompasses the northern portion of this section (approximately 51°12′47″N, 4°26′49″E Web Mercator).

Type Horizon and Age: ‘Anversien’ (Van Beneden, 1876,1877). The precise location and stratigraphical position is unknown.Deméré, Berta & Adams (2003) reassigned the species to the Berchem Formation without providing any evidence. Dinoflagellate cyst biostratigraphic analysis of a sediment sample from the lectotype humerus IRSNB 1146-M279 ofLeptophoca proxima provides a late Langhian–early Serravallian age (14.8–13.2 Ma, middle Miocene). As forProphoca rousseaui, the highly diachronous nature of the Neogene strata from Belgium precludes assigning the lectotype specimen ofLeptophoca proxima to a specific lithostratigraphic unit through biostratigraphic age determination. Also, the sediment sample from which the dinoflagellate cysts were recovered was too small to allow any comprehensive sedimentological analysis. Although a generally poor criterion, the dark color of the specimens corresponds rather to the gray–black Antwerpen Sands Member of the Berchem Formation than to the overlying, more brownish Deurne Sands from the Diest Formation. This does not contradict the results of the biostratigraphic analysis and would correspond to the stratigraphic position ofProphoca proxima as proposed byDeméré, Berta & Adams (2003).

Comments: Differing from the proposal ofRay (1976a) to reassignProphoca proxima to the genusLeptophoca, but to retainLeptophoca lenis as a separate species, in the current study we synonymizeProphoca proxima andLeptophoca lenis, and propose the new combinationLeptophoca proxima. Although the genus nameProphoca has age priority overLeptophoca, the distinct speciesProphoca rousseaui has been named type species of the genus byKellogg (1922) and has priority overProphoca proxima. Following article 23 of the ICZN, synonymsProphoca proxima andLeptophoca lenis are combined intoLeptophoca proxima n. comb.

Emended Diagnosis: Large phocine, similar in size to largeE. barbatus. The humerus ofLeptophoca proxima differs from all other phocines in the following unique combination of characters: lesser tubercle of humerus small not reaching the proximal level of the humeral capitulum (also present inPraepusa vindobonensis,Prophoca rousseaui andS. sintsovi); intertubercular groove wide and shallow (also present inC. cristata,E. barbatus andProphoca rousseaui); relatively straight posterior margin of the humeral capitulum; deltopectoral crest extending along the proximal two-third of humerus (also present inC. maeotica,Phoca vitulinoides,Prae. vindobonensis,Prophoca rousseaui andS. sintsovi); deltopectoral crest terminating abruptly, distally, but less abrupt than in extant Phocinae (also present inProphoca rousseaui); deltopectoral crest mediolaterally thin; lateral epicondyle thin and strongly projecting posteriorly; deep and well-outlined coronoid fossa (also present inPhoca spp. andPusa spp.)

Koretsky (2001) presented a diagnosis of the cranium and mandible ofLeptophoca proxima (asLeptophoca lenis). However, this diagnosis is based on isolated skulls and skull fragments and mandibles. Without any supported association toLeptophoca proxima, i.e., association with the humerus, the designation of any cranial or mandibular specimen toLeptophoca proxima remains doubtful. Therefore, we acknowledge the diagnosis byKoretsky (2001), but neither accept nor reject it. Similarly,Koretsky (2001) tentatively assigned a significant number of isolated postcranial bones toLeptophoca proxima. However, given the abundance of humeri from the Calvert Formation and other formations of the Chesapeake Group (Neogene of the mid-Atlantic coastal plain, Delaware, Maryland, North Carolina, and Virginia) assigned toLeptophoca proxima, it can relatively safely be assumed that other phocine bones that have been found in relatively large numbers in the Chesapeake Group, such as femora and tibiae, can be related toLeptophoca proxima as well. Nevertheless, no femora or tibiae from the Neogene of Belgium can be assigned toLeptophoca proxima. Hence, because the current study focuses on material from Belgium, neither the femur nor the tibia ofLeptophoca proxima will be treated in detail here.

Referred specimens: IRSNB 1146-M279, lectotype, left humerus, section no. 3 at Borgerhout, Antwerp, Belgium, Miocene, ‘Anversien’; USNM 5359, right humerus, holotypeLeptophoca lenis, Miocene, latest Burdigalian–Langhian, Calvert Formation, Shattuck zone 10, float from zone 10 at Plum Point, Calvert County, Maryland, USA; USNM 23450, left humerus, Miocene, late Langhian–early Serravalian, Calvert Formation, Plum Point Member, Shattuck zone 12, second hill south of Parkers Creek, approximately 30 cm above beach level, Calvert County; USNM 186990, right humerus, Miocene, middle Serravalian–late Serravalian, Choptank Formation, Shattuck zones 18–20, approximately 6 km NW of Cove Point, Calvert County; USNM 284721, right humerus, Miocene, late Langhian–early Serravalian, Calvert Formation, Plum Point Member, Shattuck zone 12, Pope’s Creek at Stratford Bluffs, Westmoreland County, Virginia; USNM 412115, left humerus, Miocene, Calvert Formation (Plum Point Member) or base of Choptank Formation, Shattuck zones 13–16, Scientists Cliffs, Calvert County, Maryland, US; IRSNB 1145-M280a-b, left ulna and left radius, fort no. 3 at Borsbeek, Antwerp, Miocene, ‘Anversien’; USNM 263648, right radius, Miocene, Calvert Formation (Plum Point Member) or base of Choptank Formation, Shattuck zones 11–16, north of Scientists Cliffs, Calvert County. The geological ages for specimens of North America listed above are deduced from associating the known geographic locations to stratigraphic charts and logs of locations fromKidwell et al. (2015).

Description

Humerus (Fig. 11): The lectotype left humerus ofLeptophoca proxima, IRSNB 1146-M279, is incomplete, missing the proximal portion of the deltoid crest and the distal epiphysis. It is larger than the original holotype ofLeptophoca lenis (USNM 5359), as noted byTrue (1906) (Supplemental Information 1).

The capitulum is located only slightly proximal to the greater tubercle. But both the capitulum and the greater tubercle reach much more proximal than the lesser tubercle. The round, knob-like lesser tubercle is small, especially compared to extant Phocidae. Many other extinct Phocinae also have a lesser tubercle smaller than in extant Phocinae, but they are still more prominent and reaching higher proximal than inLeptophoca proxima (seeKoretsky, 2001; Leonard Dewaele, 2016, personal observation). The posterior surface of the lesser tubercle bears clear scars for the attachment of them. subscapularis.Koretsky (2001), similarly considers that the lesser tubercle ofLeptophoca lenis is located distal to the capitulum and the greater tubercle, but also that both the greater and lesser tubercle are located distal to the capitulum inProphoca proxima. However, we tend to disagree with the latter, based on the incompleteness of the single humerus formerly attributed toProphoca proxima, which has a little damaged lesser tubercle but a strongly damaged greater tubercle. In IRSNB 1146-M279, the proximal part of the greater tubercle is indeed missing. If complete, its greater tubercle would approximately reach the same level as the capitulum and reach much more proximal than the lesser tubercle, as is the case in specimens attributed toLeptophoca lenis.

The intertubercular groove is very poorly preserved in the lectotype, but it is narrow and appears to be deep in better-preserved specimens (USNM 5359 and USNM 412115). However, the latter could not be ascertained due to the general poor preservation of the greater tubercle.

The deltopectoral crest is widest proximally and becomes very thin, blade-like, and distally. This has previously been observed for specimens attributed toLeptophoca lenis (True, 1906;Koretsky, 2001) and toProphoca proxima (True, 1906). The deltopectoral crest extends presumably for approximately two-thirds of the length of the bone. According toTrue (1906), the deltopectoral crest extends beyond half-length of the bone in bothLeptophoca lenis andProphoca proxima, whileKoretsky (2001) makes the distinction betweenLeptophoca lenis andProphoca proxima, stating that the deltopectoral crest extends for more than two-thirds of the total length of the humerus inProphoca proxima and less than two-thirds inLeptophoca lenis. Given the incompleteness of the humeri attributed to both species, we do not follow Koretsky’s separation of both species on the basis of this character. In extant Phocinae, the deltopectoral crest is generally restricted to the proximal half of the humerus and never attains a prominently blade-like shape as it does inLeptophoca proxima. However, Monachinae and certain fossil Phocinae, such asK. benegasorum,M. pontica andS. sintsovi (Cozzuol, 2001;Koretsky, 2001), have deltopectoral crests longer than half the length of the humerus. The medial side of the deltopectoral crest is straight; on the lateral side of the crest the deltoid tuberosity extends along the proximal half of the crest, as observed byKoretsky (2001) in specimens referred toLeptophoca lenis andProphoca proxima. Proximal on the lateral surface of the deltopectoral crest, there is a small but clearly visible round and shallow fossa for the spinatus muscles.

In posterior view, the diaphysis ofLeptophoca proxima is much thinner than in other Phocidae.True (1906) already noted this character in the holotype specimen ofLeptophoca lenis (and named the species accordingly:leptòs means “thin” or “slender” in Greek) and in the humerus ofProphoca proxima. Overall, the diaphysis ofLeptophoca proxima is straight.True (1906) noted the straighter diaphysis (shaft) inProphoca proxima compared toLeptophoca lenis. This is indeed true when comparing the lectotype humerus ofProphoca proxima (IRSNB 1146-M279) with the holotype humerus ofLeptophoca lenis (USNM 5359) and other, more recently found specimens (e.g., USNM 186990 and USNM 284721). However, other recently found humeri (e.g., USNM 23450 and USNM 412115) also have a straighter diaphysis than the holotype ofLeptophoca lenis.Ray (1976a) pointed out that a relatively straight humerus is a character state intermediate between terrestrial carnivorans and early Pinnipedimorpha (Berta, Ray & Wyss, 1989) on the one hand, and later-branching Phocidae on the other hand. Pinnipedimorpha are a monophyletic group includingEnaliarctosMitchell & Tedford, 1973 and the Pinnipediformes Berta, 1994, the latter includingPteronarctos Barnes, 1989, the three modern pinniped families: Odobenidae Allen, 1880, Otariidae and Phocidae, and the two extinct pinniped families Desmatophocidae Hay, 1930 and SemantoridaeOrlov, 1933. In other Phocinae, we only noted a relatively straight humeral diaphysis inC. cristata. On the diaphysis, there are two distinct fossae for muscle origins. On the posterior surface, just distal to the capitulum and the lesser tubercle, there is a distinct fossa for them. triceps brachii caput mediale. The surface for them. triceps brachii caput laterale just distal and lateral to the capitulum is clearly marked as well. The fossa proximal on the lateral surface of the diaphysis and the deltopectoral crest serves as the origin ofm. brachialis. Especially the cranial portion of this fossa is very deep, compared to other fossil and living Phocidae.

The lateral epicondyle is well developed but thin and strongly projecting posteriorly. The lateral epicondyle ofLeptophoca proxima is thinner than in other Phocinae. Proximally, the lateral epicondyle reaches the same level as the distal part of the deltopectoral crest, as has also previously been observed for specimens previously attributed toProphoca proxima andLeptophoca lenis (Koretsky, 2001). The medial epicondyle is not preserved on the lectotype ofLeptophoca proxima, but it is preserved in some North American specimens of that species, where it is not particularly strongly developed (e.g., USNM 5359, USNM 186990, USNM 284721, and USNM 412115). We follow Koretsky’s description (forLeptophoca lenis) that the medial epicondyle is anteroposteriorly compressed and not extending proximal to the coronoid fossa. The entepicondylar foramen is rather small (contraKoretsky, 2001; forProphoca proxima andLeptophoca lenis) and oval, with a wide bridge over it.

Neither the olecranon fossa nor the coronoid fossa is preserved in the lectotype, but it is in USNM 5359, USNM 186990, USNM 284721, and USNM 412115 from North America. The olecranon fossa ofLeptophoca proxima is wide and shallow, and the coronoid fossa is clearly outlined, subtriangular and deep, as already noted byKoretsky (2001) (forLeptophoca lenis).

Overall, the humerus ofLeptophoca proxima shows similarities to the Histriophocini tribe and particularly to the living ribbon sealH. fasciata. In bothLeptophoca proxima and the Histriophocini, the origin for them. brachialis forms a well-developed fossa.Leptophoca proxima andH. fasciata both have a relatively straight humeral diaphysis.

Radius (Fig. 12): Only one left radius from the collection at the IRSNB has formerly been assigned toProphoca proxima (IRSNB 1145-M280a), and, hence, currentlyLeptophoca proxima. This radius had been found in association with the partial ulna IRSNB 1145-M280b. Lacking its distal epiphysis, the radius is considered as a late subadult (seeStorå, 2000). Without the distal epiphysis fused, the total length of the radius is 114.7 mm. Similarly, specimen USNM 263648 previously referred toLeptophoca lenis and here identified asLeptophoca proxima also represents a juvenile to subadult specimen. Both specimens are of the same size.

The radius is slender and strongly curving in the parasagittal plane, similar to other Phocinae (exceptE. barbatus) and unlike Monachinae. The radius USNM 263648 is slightly more slender than IRSNB 1145-M280a.

The ulnar facet of the humeral articular circumference is lens-shaped, arching around the posterior and medial margins of the articular fovea. A small triangular facet for articulation with the humeral trochlea is located anterior to the aforementioned ulnar facet on the medial side of the radius. Circumferentially, the humeral articular circumference is not as large compared to the proximal part of the diaphysis, than it tends to be in extant Phocidae.

The bicipital tuberosity is large and rounded, asKoretsky (2001) already noted, but slightly more flattened in the Belgian specimen. The bicipital tuberosity is located posteromedially on the diaphysis. At the level of the bicipital tuberosity, the diaphysis forms an angle. This corresponds with other Phocinae and contrasts to Monachinae, which generally have a more smoothly curving diaphysis at the level of the bicipital tuberosity.

de Muizon (1981) noted well developed insertion surfaces form. supinator andm. brachioradialis and a weakly developed insertion surface form. pronator teres on the diaphysis of the radius in all extant Phocidae (except Lobodontini showing the reverse condition),P. pacifica, and Otariidae. InLeptophoca proxima we observed an intermediate state: the insertion surface form. supinator is moderately well developed, while on the other hand, the insertion surfaces form. pronator teres andm. brachioradialis are relatively little developed. However, the surface for the insertion of them. brachioradialis is located more distal on the anterior margin of the radius, as compared to modern Phocinae. It also protrudes more inLeptophoca proxima than it does in extant Phocinae. Our observations contrast withKoretsky (2001), stating that the insertion surface form. pronator teres is almost absent inProphoca proxima. We agree that the insertion surface form. pronator teres is indeed slightly better developed in the North American specimen USNM 263648, but we consider the insertion surface for both still better developed than in Lobodontini. The ridge guiding the tendon groove ofm. extensor digitorum communis, distal on the lateral side of the radius is very pronounced in comparison with extant and other extinct Phocinae, as noted byKoretsky (2001) forLeptophoca lenis.

Ulna (Fig. 13): Only one ulna (IRSNB 1145-M280b) had been attributed toProphoca proxima (Van Beneden, 1877); and is associated with the radius IRSNB 1145-M280a. Formerly, no ulna ofLeptophoca lenis had been described (seeKoretsky, 2001). Hence, this unique ulna has been assigned toLeptophoca proxima based on its association with a radius assigned to that species.

Only the diaphysis of the ulna is preserved. This portion is relatively large, pointing toward an estimated total length in the order of 15 cm when complete. The diaphysis is robust and the distal epiphysis is styloid, i.e., pointed. The oblong rugosity is faintly developed. This fragmentary bone does not differ significantly from the ulna of other Phocinae.

• Phocinae aff.Leptophoca proxima

Referred specimens—IRSNB M2233, right humerus, Miocene, Antwerp, Belgium; IRSNB M2232, right fibula, Miocene, Antwerp.

Description

Humerus (Fig. 14): One right humerus (IRSNB M2233), originally attributed toProphoca proxima byVan Beneden (1877) is significantly smaller than the aforementionedLeptophoca proxima humerus IRSNB 1146-M279; it better matches the size of femora attributed toLeptophoca amphiatlantica (no humerus known for the latter) and the North American humeri ofLeptophoca proxima. The humerus IRSNB M2233 misses its distal portion and is strongly abraded.

Although slightly abraded, the capitulum is roughly hemispherical and located high proximally. The capitulum is located slightly proximal to both the lesser and the greater tubercle and strongly projecting anteriorly. The round, knob-like lesser tubercle is small, but slightly more pronounced than in IRSNB 1146-M279 and USNM 412115, humeri identified asLeptophoca proxima. The posterior surface of the lesser tubercle bears clear scars for the attachment of them. subscapularis. The intertubercular groove is narrow and it appears to be deep.

The deltopectoral crest is wider proximally and extends presumably for the proximal half of the length of the bone. The medial side of the deltopectoral crest is straight and the deltoid tuberosity extends on the lateral side of the crest along its proximal portion, before tapering abruptly distally. Along the distal two-thirds, the deltopectoral crest is blade-like and mediolaterally thin.

As inLeptophoca proxima, the diaphysis is thin compared to the proximal epiphysis: less than one third at its thinnest, in posterior view. However, the diaphysis of IRSNB M2233 is significantly thinner than inLeptophoca proxima in lateral view. The posterior surface of the diaphysis is relatively straight, as inLeptophoca proxima. On the diaphysis, just distal to the capitulum and the lesser tubercle the fossa form. triceps brachii caput mediale is moderately well developed. Proximal on the lateral surface of the diaphysis and the deltopectoral crest, the origin of them. brachialis is clearly outlined and well developed.

This humerus IRSNB M2233 shows some marked affinities withLeptophoca proxima, but differs on some noticeable aspects. Similarities include the small lesser tubercle; the thin deltopectoral crest, profoundly extending distally; the narrow diaphysis; and the strong anterior projection of the capitulum. Marked differences are the extreme thinness of the diaphysis and the concavity of the posterior surface of the diaphysis. Due to the absence of the distal portion and the strong abrasion of IRSNB M2233, it is difficult to ascertain whether the differences between IRSNB M2233 andLeptophoca proxima can be attributed to intraspecific variation (including sexual variation) or not.

Fibula:Van Beneden (1877) identified one very fragmentary right fibula (IRSNB M2232) asProphoca proxima. Of this fibula, only the distal half is moderately well preserved. In its current condition, the fibula bears very little—if any—diagnostic information.

The fibula is short and transversely thin; based on the preserved parts it is comparable in size to the fibula of members of the genusPusa Scopoli, 1777. It is slightly curving sigmoidally, but more than is usual for Phocidae. The diaphysis is triangular in cross-section, having approximately straight lateral and anterior surfaces. The posterior surface is deeply concave, representing a well-developed fossa for the origin of them. flexor hallucis longus.

The distal epiphysis, i.e., the lateral malleolus, is square with clearly visible tendon grooves on its posterior surface. The distal articular surface for the astragalus is horizontal and only slightly concave. In extant Phocinae, the distal articular surface of the fibula is strongly sloping laterally and more strongly concave.

The size and general shape of IRSNB M2232 corresponds to that of specimen USNM 372547, found at the east coast of North America. However, the fibula is generally considered of very limited use to differentiate phocid species (Hodgetts, 1999). Combined with the poor state of preservation of IRSNB M2232, it is very difficult to unambiguously assign this specimen toLeptophoca.

• Comments onLeptophoca amphiatlanticaKoretsky, Ray & Peters, 2012

Leptophoca amphiatlantica was described only recently byKoretsky, Ray & Peters (2012). The species has been erected, based on four isolated partial femora: three specimens from the Calvert and St. Marys formations (lowermost Aquitanian to middle Tortonian) from the east coast of North America; and one specimen from the Breda Formation (uppermost Burdigalian to Langhian) from the Netherlands. However, we consider the evidence in support ofLeptophoca amphiatlantica as a distinct species relatively weak.

At the time of the formal description ofLeptophoca amphiatlantica, the only other known species in the genus wasLeptophoca lenis. As mentioned above, the holotype ofLeptophoca lenis is an isolated humerus, and more recently described specimens tentatively attributed to the species—including femora—have been found isolated (Koretsky, 2001). Therefore, the identification and diagnosis of the type femora ofLeptophoca amphiatlantica in relation toLeptophoca lenis is precluded due to non-overlapping type specimens.

The described overall size discrepancy betweenLeptophoca amphiatlantica andLeptophoca proxima does not appear to be supported by measurements (Supplemental Information 1;Table 2;Fig. 15). When comparing measurements of bothLeptophoca species, most measurements overlap between both species. Ranges do not overlap for only a few measurements: absolute length, length of the medial and lateral condyle, width of the medial condyle, height of the humeral head and height of the articular surface of the patellar surface. Because of the limited number of (often fragmentary) femora attributed toLeptophoca amphiatlantica andLeptophoca lenis, the statistical value of the biometric differences is very low. Indeed, given the often fragmentary nature of the specimens, not all characters could be measured for all specimens, and the numbers of measurements range from two to nine inLeptophoca lenis (Koretsky, 2001) and from one to four inLeptophoca amphiatlantica (Koretsky, Ray & Peters, 2012). For instance, although the absolute length range ofLeptophoca lenis exceeds that ofLeptophoca amphiatlantica, both ranges are based on only two specimens, and it can be suggested that the true intraspecific variation ofLeptophoca lenis must have been larger than the 119.0–120.0 mm range presented (Koretsky, 2001; see alsoSupplemental Information 1;Table 2) and may, hence have overlapped withLeptophoca amphiatlantica. Also, two left femora ofLeptophoca amphiatlantica, USNM 321926 and MAB 2129, bear clear signs of mechanical abrasion, further allowing questioning the value and validity of the presented measurements. Apart from the small number of measurements, the significance of non-overlapping measurement ranges can be questioned due to the nearness of the measurement ranges in both species. On an absolute length for the bone in the order of > 100.0 mm, the statistical significance of a condylar size difference in the order of a few millimeters, based on two specimens ofLeptophoca lenis and one specimen ofLeptophoca amphiatlantica, is questionable.

We supportKoretsky, Ray & Peters (2012) in the observation of certain differences between the holotype ofLeptophoca amphiatlantica, USNM 23227, and isolated femora attributed toLeptophoca proxima: i.e., the shape and position of the trochanteric fossa. Nonetheless,Koretsky (1987) observed sexual dimorphism at the level of the greater trochanter both in livingPusa spp. and in fossilM. pontica. Therefore, we question the validity of the differences at the level of the greater trochanter to separateLeptophoca amphiatlantica fromLeptophoca proxima.

Pending the discovery of more complete/better preserved specimens for bothLeptophoca proxima andLeptophoca amphiatlantica, we provisionally reassign the specimens originally attributed toLeptophoca amphiatlantica toLeptophoca cf.Leptophoca proxima, hereby confirming the presence of certain characters that differ fromLeptophoca proxima but considering them too weak to define a distinct species; accordingly we consider the species nameLeptophoca amphiatlantica anomen dubium. Because the specimens ofLeptophoca amphiatlantica cannot be unambiguously reassigned toLeptophoca proxima, we cannot considerLeptophoca amphiatlantica to be a proper junior synonym toLeptophoca proxima.

Middle Miocene taxa and their provenance (including Monachinae)

Historically, both the oldest unequivocal monachine and the oldest known Miocene phocine are known from the middle Miocene (Langhian) Calvert Formation of Maryland and Virginia (True, 1906;Ray, 1976a,1976b;Deméré, Berta & Adams, 2003). Presumably older specimens and species have been described (Koretsky & Sanders, 2002;Diedrich, 2011), but their age assignments are contested further in the current study.Leptophoca proxima (formerly asProphoca proxima andLeptophoca lenis) has commonly been regarded as the oldest known record for the family Phocidae (e.g.,Deméré, Berta & Adams, 2003;Koretsky, Ray & Peters, 2012). The holotype ofLeptophoca lenis (USNM 5359) had been retraced to “zone 10” of the Plum Point Marl Member of the Calvert Formation, which had been dated to the earliest Langhian (True, 1906;de Verteuil & Norris, 1996).Koretsky (2001) reconsidered the origin of USNM 5359 between Chesapeake Beach (zone 5) and Plum Point (zone 10) in the Calvert Formation, adjusting the oldest known occurrence ofLeptophoca to the middle Burdigalian (approximately 18 Ma). Other specimens ofLeptophoca proxima from the West Atlantic, presented byKoretsky (2001) can all be retraced to an age corresponding to the Langhian and Serravallian of the Plum Point Member of the Calvert Formation and the Serravalian and early Tortonian Choptank Formation. However, as stated above, the assignment of other isolated specimens toLeptophoca proxima should be considered with care. The Belgian lectotype ofLeptophoca proxima (IRSNB 1146-M279) has a minimum middle Langhian–early Serravallian age (14.8–13.2 Ma) according to dinoflagellate cyst biostratigraphy. ConsideringLeptophoca amphiatlantica asLeptophoca cf.Leptophoca proxima, the occurrence ofLeptophoca proxima in the East Atlantic can be dated to the early late Burdigalian–early Langhian Dutch Breda Formation (16.4–15.8 Ma).

The enigmatic monachineMonotherium wymani (Leidy, 1854) is known from only few isolated bones and for a long time its stratigraphic position and age remained doubtful.Ray (1976b) re-investigated this subject and identified the uppermost zones of the Calvert Formation at Richmond, Virginia, as the origin of theMonotherium wymani material. These zones have been dated to the late Langhian–early Serravallian (Kidwell et al., 2015).

Ray (1976a) also identified one specimen from the Calvert Formation as belonging toProphoca rousseaui, but he did not describe or illustrate it. We could not retrace it at the USNM, nor find any data on a more precise age designation. Hence, we will not consider it in the discussion. Belgian material ofProphoca rousseaui had been (re)assigned to the late Burdigalian–Langhian Antwerpen Sands Member of the Berchem Formation (Deméré, Berta & Adams, 2003). Two Belgian specimens ofProphoca rousseaui have an age between 14.2 and 13.2 Ma (late Langhian–early Serravallian) and 14.2–7.2 Ma (late Langhian–Tortonian/Messinian boundary). The isolated innominate IRSNB M2234 from the base of the Antwerpen Sands Member of the Berchem Formation at the Posthofbrug, rich in vertebrate remains (Louwye et al., 2010), is here assigned toProphoca rousseaui.Louwye et al. (2010) dated the base of the Antwerpen Sands Member of the Berchem Formation at the Posthofbrug to the latest Burdigalian.

The current body of evidence indeed indicates thatLeptophoca proxima is the oldest true seal currently known, but we have to admit that the currently limited fossil record ofMonotherium wymani andProphoca rousseaui may in fact not give a good representation of their stratigraphical distribution. WithLeptophoca proxima andProphoca rousseaui forming a distinctive group of early branching phocines, exactly distinguishing the related ages of each taxon is of subordinate importance.

Other geologically old, i.e., Burdigalian, Langhian and/or Serravalian, Phocidae areA. libyca,Miophoca vetustaZapfe, 1937,Monotherium gaudini, and other Phocidae from Libya and the Paratethys (Koretsky, 2001;Deméré, Berta & Adams, 2003;Koretsky, Ray & Peters, 2012;Koretsky & Domning, 2014;Berta et al., 2015) andK. benegasorum, andP. argentinus from Argentina (Ameghino, 1893;de Muizon & Bond, 1982;Cozzuol, 2001).A. libyca may presumably replaceMonotherium wymani as the oldest known monachine seal. Recovered from the Garat Jahanam Member of the Marada Formation in Lybia, the holotype and only specimen, a partial mandible, may be dated between 19 and 14 Ma (Koretsky & Domning, 2014).Monotherium gaudini is a monachine from Italy (Guiscardi, 1871).Deméré, Berta & Adams (2003) degradeMonotherium gaudini toMonotherium sp. indet. from the Bismantova Formation. However, there is no conclusion for the age of this formation.Odin et al. (1997) dated the Bismantova Formation to the Burdigalian–early Langhian, which had been adopted byBerta et al. (2015), whileDeméré, Berta & Adams (2003) consider the late Langhian–Serravalian-early Tortonian age presented by, e.g.,Pini (1999) andCarena et al. (2000).

Deméré, Berta & Adams (2003) regardPristiphoca vetusa (sic) (Thenius, 1950) as an early Sarmatian (Serravalian) monachine: the oldestz record of a monachine in the Paratethys. However, it should be pointed out that the species under consideration is in fact a junior synonym toM. vetusta, an enigmatic Badenian (middle Miocene) phocid from the Vienna Basin (Zapfe, 1937), and considered to be a “cystophorine” byKoretsky (2001).Piller, Harzhauser & Mandic (2007) considered the Badenian to be Langhian–early Serravallian in age. Nonetheless,M. vetusta can be considered anomen dubium, or asGrigorescu (1976: 416) states it: “M. vetusta (…) is known only by a fragment of the lower jaw, that can’t represent a satisfactory basis for detailed discussions.” However, a detailed revision ofM. vetusta and its age is beyond the scope of this study.

Properiptychus argentinus is an early monachine from the Paraná Formation of Argentina.de Muizon & Bond (1982) assigned a Friasian age to the species, which is a South American Land Mammal Age roughly spanning from 16.3 to 15.5 Ma, (late Burdigalian to the early Langhian;Flynn & Swisher, 1995). However, more recent research assigns a late Miocene age to the Paraná Formation on the basis of bivalve biostratigraphy (Pérez, Genta Iturrería & Griffin, 2010).

According toKoretsky (2001),Praepusa spp.,C. maeotica,S. sintsovi andM. pontica from the Central and Eastern Paratethys are all known from Sarmatian (Langhian and Serravallian) deposits. More recent stratigraphic research byPiller, Harzhauser & Mandic (2007) supported the agesKoretsky (2001) assigned to the Sarmatian stages.

Remarks on allegedly Eocene and Oligocene Phocidae

The early evolution of Phocidae is poorly known and the two oldest records are a matter of debate. The presumably oldest described phocid is the speciesPraephoca bendullensisDiedrich, 2011 proposed to date from the Eocene of northern Germany (Diedrich, 2011). The type—and single—specimen (an isolated femur) had allegedly been found in the early Lutetian (early middle Eocene; 49–45 Ma) Fürstenau Formation. This geologic age predates the formerly generally accepted earliest pinnipedimorph records by at least 15–20 Ma and the author even suggests a late Paleocene/early Eocene age for the specimen, based on its reworked nature. The oldest known pinnipedimorphs areEnaliarctosMitchell & Tedford, 1973 (late Oligocene Yaquina Formation of Oregon, USA;Berta, 1991),Potamotherium Geoffroy, 1833 (late Oligocene lake sediments of Rheinland-Pfalz, Germany;Mörs & von Koenigswald, 2000; phylogenetic affinities with pinnipeds still debated), orPuijilaRybczynski, Dawson & Tedford, 2009 (early Miocene Haughton Formation of Nunavut, Canada;Rybczynski, Dawson & Tedford, 2009). However,P. bendullensis has been proposed to represent a phocid, thus, antedating the oldest unambiguously known Phocidae by approximately 30 Ma (see below) and creating multiple Eocene and Oligocene ghost lineages for all pinnipedimorphs. It also antedates published molecular divergence dates for Arctoidea Flower, 1869 (e.g.,Higdon et al., 2007 as well as the oldest arctoid fossils, e.g.,Parictis Scott, 1893 from the late Eocene/early Oligocene of North America (e.g.,Hunt, 1998).

Diedrich (2012) formally described the Fürstenau Formation in another paper, published almost simultaneously to the description ofP. bendullensis. Currently only two partial sections of the Fürstenau Formation have been described, both in Germany, based on the correlation of a fossiliferous gravel bed (Diedrich, 2012): the lower part of the section at Dalum, and the upper part at Osteroden. The middle Eocene age proposed for the Fürstenau Formation is entirely based on the relative dating of the fossil content (terrestrial mammal teeth and shark teeth) of only one fossiliferous layer. This layer is a condensed basal gravel and the published fossil content of this fossiliferous layer may not be a good representative for the entire fossil content—and geologic age—of this layer. Indeed, the fossil content is strongly biased to sharks (approximately 95% of the material), with illustrated specimens showing signs of reworking and abrasion. A more thorough biostratigraphic investigation may potentially reveal a time averaged nature of the fossil content of this layer, possibly containing a mix of fossils from different Paleogene levels. The partial femur attributed toP. bendullensis had been collected during the 1980s, separately from the study presented byDiedrich (2011,2012). Furthermore, it is strongly abraded and was most likely reworked. Its stratigraphic origin in the Fürstenau Formation is thus at best questionable, considering the worldwide lack of any other Eocene pinnipedimorph records despite extensive sampling of other marine mammal fossils from Eocene marine deposits around the world, i.e., Indo-Pakistan, North Africa, southwestern USA, and New Zealand (e.g.,Thewissen, Williams & Hussain, 2001;Peters et al., 2009;Peredo & Uhen, 2016). No details are provided about the overlying Neogene layers in the Dalum section (Diedrich, 2012), and the potential origin of the specimen in another, younger gravel was not discussed by the author. UnlikeKoretsky & Domning (2014: 227) who stated that “the only known specimen, the proximal part of the femur, is so damaged that it is inadequate for its identification as a pinniped,” we support the identification of this specimen as representing a phocid by, considering the anteroposterior flattening of the diaphysis, the absence of a fovea capitis and the absence of a lesser trochanter (seeDiedrich, 2011). Based on the small size of the specimen, it may tentatively be related to the Miocene phocinesBatavipusa neerlandica orPhoca vitulinoides, known from the Netherlands and Belgium (Koretsky & Peters, 2008;Van Beneden, 1871,1876,1877), but the poor state of preservation of the specimen does not permit a detailed description and comparison; the definition of a separate taxon seems thus hazardous. Therefore, we considerPraepusa bendullensis anomen dubium.

Koretsky & Sanders (2002) described two indeterminate phocid femora from the upper Oligocene Ashley and Chandler Bridge formations of South Carolina, USA. Although the oldest formally described Phocidae from the east coast of North America date from the late early to early middle Miocene, it has been assumed that the Phocidae originated during the latest Oligocene or the earliest Miocene, based on molecular evidence (Higdon et al., 2007). Such evidence renders it possible that the specimens presented byKoretsky & Sanders (2002) are indeed from the Chandler Bridge Formation, but not from the Ashley Formation, based on age definitions of both formations (Weems et al., 2012). However, at the localities where these pinniped remains were found, just south to southwest of Summerville, SC, USA, the Ashley and Chandler Bridge formations only outcrop in ditches and scarps, and are covered by the “Ashley Phosphate Beds,” the Pleistocene Wando Formation and Penholoway Formation (Weems, Lewis & Lemon, 2014). Hence, in the absence of a firmly established stratigraphic framework for the specimens, it is most likely that these specimens, which have been found in ditches, have been washed to lower levels in the ditch from the overlying Ashley Phosphate Beds or Wando Formation and that the specimens are found ex situ. The Ashley Phosphate Beds are usually identified as the basal phosphatic layer of the Wando Formation and are rich in reworked Miocene and Pliocene fossils (Monsch, Fierstine & Weems, 2005). In the absence of a more precise geographic and stratigraphic position for both specimens, their Oligocene provenance should probably only be considered as a hypothesis to be confirmed. Keeping the fragmentary nature of the specimens in mind, a shallow trochanteric fossa has also been observed in the late Neogene phocinePhocanella from Belgium and the east coast of North America (Koretsky & Ray, 2008).

Early phocid paleobiogeographic evolution

While early-branching pinnipedimorphs, early Odobenidae, early Otariidae, and Desmatophocidae diversified in the North Pacific Ocean during the early and middle Miocene (see, e.g.,Deméré, Berta & Adams, 2003;Boessenecker & Churchill, 2015), specimens ofMonotherium wymani andLeptophoca proxima from the east coast of North America represent the earliest unambiguous records of Phocidae. Most scientists agree on the migration of a phocid ancestor from the northeastern Pacific to the northwestern Atlantic (seeDeméré, Berta & Adams, 2003). Of the two possible pathways, through the Central American Seaway or through the Bering Strait and Arctic Ocean, the first is the prevailing option (Costa, 1993;Bininda-Emonds & Russell, 1996;Deméré, Berta & Adams, 2003). During the early and middle Miocene, the Beringian land bridge prevented the marine connection of the North Pacific Ocean to the Arctic Ocean (Costa, 1993;Bininda-Emonds & Russell, 1996;Marincovich, 2000;Deméré, Berta & Adams, 2003). However, asDeméré, Berta & Adams (2003: 63) pointed out, “the absence of basal phocid fossils in the North Pacific and Caribbean is a weakness of this hypothesis.” Other possible regions of origin for Phocidae that have been considered include the Paratethys (Orlov, 1933;McLaren, 1960;Koretsky & Barnes, 2006) and the Arctic (Rybczynski, Dawson & Tedford, 2009). Those are difficult to unite with the paleobiogeography and chronology of early and middle Miocene diversification of other pinnipedimorphs (seeDeméré, Berta & Adams, 2003), given the generally accepted monophyly of pinnipeds and pinnipedimorphs both on morphological and molecular evidence (e.g.,Berta & Wyss, 1994;Bininda-Emonds & Russell, 1996;Higdon et al., 2007;Fulton & Strobeck, 2010;Árnason et al., 2006). However, a number of geologically old Phocidae have been found outside the East coast of North America:A. libyca,P. argentinus, several phocids from the Paratethys, andLeptophoca proxima specimens from the Netherlands may rivalLeptophoca proxima andMonotherium wymani as the oldest known phocids.

Although these findings put the hypothesis of the East coast of North America as the area of origin of Phocidae to the test, the widely recognized monophyletic origin of pinnipeds renders it difficult to consider the Paratethys as the origin of Phocidae in relation to its Pacific ancestors and relatives. However, early diversification of crown Phocidae presumably occurred during the late Burdigalian or early Langhian in the North Atlantic. This is supported by (1) the co-occurrence of both a monachine (Monotherium wymani) and a phocine (Leptophoca proxima) in the Langhian–lower Serravallian Calvert Formation, as well as (2) the presence of the phocinesLeptophoca proxima andProphoca rousseaui and the monachineA. libyca in the North Sea and in Lybia, respectively, during the same time interval.P. argentinus from Argentina may be a stem-monachine that ventured into the South Atlantic. Pending the discovery of other, more complete, fossil monachines from the Southern Hemisphere,P. argentinus and other may represent a basal monachine branching off the lineage that would ultimately diversify in the Southern Hemisphere, leading to later fossil monachines and modern members of the subfamily.

Specimens ofLeptophoca found in Belgium and the Netherlands (this work;Koretsky, Ray & Peters, 2012), on the southern part of the North Sea Basin indicate that shortly after their population of the east coast of North America,Leptophoca must have migrated across the North Atlantic Ocean. Given the poor fossil record ofProphoca rousseaui, it is unknown as to whetherProphoca originated along the North American east coast and migrated to the North Sea Basin, just asLeptophoca proxima, or thatProphoca diverged fromLeptophoca in the North Sea and migrated back across the Atlantic to North America. Based on the diversity of middle and late Miocene phocids, the North Sea Basin and the Paratethys must have been the regions where phocids diversified in the Northern Hemisphere during the middle and late Miocene (see,Koretsky, 2001;Deméré, Berta & Adams, 2003).

It should be noted that fossil marine mammal finds (including cetaceans) from deposits underlying the Berchem Formation in the Antwerp area are generally very rare. Hence, the absence of phocid specimens antedating these layers may be related to a preservational bias and not necessarily to the absence of Phocidae in the North Sea at that time. Published data on the relative abundance of pinnipeds in other fossil-rich marine strata, such as the Purisima Formation (Boessenecker, Perry & Schmitt, 2014) and the Sharktooth Hill bonebed (Pyenson et al., 2009) in California show relatively low abundances for pinnipeds in comparison to other vertebrate taxa. For the Neogene strata of Belgium, no detailed analysis has been undertaken about the relative abundances of fossil marine mammal taxa. However, a concise qualitative analysis byLouwye et al. (2010) from one horizon of the Berchem Formation at Posthofbrug also indicates that odontocete remains are much more common and diverse than phocid remains.

During the late middle Miocene, following the initial diversification in the North Atlantic and the Paratethys Phocinae ventured into the South Atlantic Ocean; roughly 5 million years younger than the monachineP. argentinus,K. benegasorum is currently the oldest and only known phocine seal from the Southern Hemisphere.