More than 23 extinct species and 10 extant species of the Balaenopteridae are known. Our knowledge of the family Balaenopteridae is increasing quickly, however, few fossil records support a circum-North Pacific distribution of balaenopterid genera and species. Because of limited preservations, most rorqual fossils reported from the western North Pacific can only be identified to the family level. A skull from the Shinazawa Formation (late Pliocene to Early Pleistocene) in Yamagata, Japan, is identified as aff. Balaenoptera bertae by possessing two diagnostic features of the species: large occipital condyles, and a posteriorly elongate postglenoid process. Combination of four more features also support that the specimen is a closely related to B. bertae. The specimen is probably a slightly older individual than the holotype of B. bertae, based on the estimated bizygomatic width and slightly longer posterior process of the tympanoperiotic. The first and only report of B. bertae was from the Pliocene Purisima Formation in California, USA. The specimen from Japan is incompletely preserved, but shows the occurrence of B. bertae in the western North Pacific for the first time, as many living balaenopterids are distributed across the North Pacific, such as Balaenoptera musculus, B. physalus, B. borealis, B. acutorostrata, and Megaptera novaeangliae.
Introduction
Our knowledge of the fossil record of the rorqual whales, family Balaenopteridae, has been improving recently, establishing new balaenopterids from Belgium, the Netherlands, Italy, California, Peru, Chile and Japan, and taxonomic revision of European problematic fossils (Deméré et al., 2005; Bisconti, 2007, 2010; Bosselaers and Post, 2010; Boessenecker, 2013a; Bisconti and Bosselaers, 2016, 2020; Marx and Kohno, 2016; Bisconti et al., 2019, 2020a). More than 23 extinct species and 10 extant species are known (Rosel et al., 2021). Among fossil balaenopterids, few records suggest a circum North Pacific distribution because there are only a few records of extinct balaenopterid species from the western side of the North Pacific.
The Miocene through Pleistocene fossil records of balaenopterid whales in Japan has improved recently (Oishi et al., 1985; Hasegawa et al., 2002; Kohno et al., 2007; Kimura et al., 2015; Tanaka and Taruno, 2019). Some records with species level identification are as follows; Megaptera novaeangliae from the Middle Pleistocene (about 0.12 to 0.15 Ma) at Chiba (Nagasawa and Mitani, 2004), Burtinopsis tatsunokuchiensis from the early Pliocene at Iwate (Hatai et al., 1963), and Miobalaenoptera numataensis from the late Miocene at Hokkaido (Tanaka and Watanabe, 2019). These limited fossil records from the western North Pacific are insufficient to document the past distribution of balaenopterids across the Pacific. However, fossils of other whale families such the pygmy right whale from Okinawa and the right whale from Taiwan provide their past global distribution (Tsai et al., 2017; Tsai and Chang, 2019).
Here, we describe a balaenopterid cranium from the late Pliocene to Early Pleistocene found at Yamagata, Japan, which was first reported by Nagasawa et al. (2003). The specimen is incompletely preserved but tentatively identified to species level. This record implies that there was a widely distributed extinct balaenopterid species across the Pacific.
Material and method
Morphological terminology follows Mead and Fordyce (2009). The specimen was measured using calipers in mm. Distances are either horizontal or vertical, unless identified as point to point (Table 1). Comparison was made with all named balaenopterid fossils with the skull (Table 2) by combination of examination of the literature and through our own examination of specimens.
Institutional Abbreviation.—YPM, Yamagata Prefectural Museum, Japan.
Systematic paleontology
Order Cetacea Brisson, 1762
Unranked taxon Neoceti Fordyce and de Muizon, 2001
Suborder Mysticeti Gray, 1864
Family Balaenopteridae Gray, 1864
Genus
Balaenoptera
La Cépède, 1804
Balaenoptera
aff.
bertae
Boessenecker, 2013a
aff. Balaenoptera bertae
Figures 2, 3, 4 and Table 1
Diagnosis.—YPM 11852 preserves the supraoccipital, right exoccipital, basioccipital, right squamosal, right alisphenoid, and parietals. YPM 11852 can be identified as a member of the Balaenopteridae by having combination of the temporal fossa invisible in dorsal view, which is overhung by the temporal and nuchal crests, widely expanded posterior portion and anteriorly converging supraoccipital in dorsal view, weak protrusion of the posterolateral corner of the exoccipital in ventral view, and posteriorly well projected post glenoid process in ventral view (Bisconti and Bosselaers, 2016).
YPM 11852 is identified as aff. Balaenoptera bertae by having two diagnostic features of the species stated in 2013 such as having large occipital condyles, and a posteriorly elongate postglenoid process (Boessenecker, 2013a). The ventral part of the postglenoid process of YPM 11852 is broken, but preserved part shows slender process with a strong excavation at the base of the postglenoid process, which is the same condition as the holotype of B. bertae. In addition, there are four shared features (see discussion) such a large fossa for the sternocephalicus, dorsally wider and ventrally narrower foramen magnum, occipital condyle bordering the lateral and ventral sides, but not dorsolateral sides of the foramen magnum, and straight nuchal crest at the level of the subtemporal crest in dorsal view. Comparison with the holotype of B. bertae, YPM 11852 has more slender zygomatic process and straight lateral border of the supraoccipital. As Tsai and Fordyce (2014) mentioned, different heterochronic processes characterize different mysticete clades. These differences might be affected by ontogenetic variation as these can be seen among Balaenoptera acutorostrata in Figure 2 of Nakamura and Kato (2014) (see more in ontogeny section).
Locality.—YPM 11852 was found ventral up in the Shinazawa Formation at the Shinazawa sandstone quarry, in Shonai Town (former Tachikawa Town), Yamagata Prefecture, Japan (Figures 1 and 2): Latitude 38°41′15.87″N, longitude 140°0′14.97″E on 10th October 2000 by a student of Yamagata University on a field trip (Nagasawa et al., 2003). The specimen was excavated by R. Abe and S. Oba on 4th November 2000.
Horizon.—Several fossiliferous stratigraphic units are exposed in the vicinity of the YPM 11852 locality near Shinazawa, including the middle Miocene Aosawa and Kusanagi Formations, the upper Pliocene Shinazawa Formation, and terrace deposits (Honda and Applied Geological Society of Yamagata, 2016).
The Shinazawa Formation is exposed in a small geographic area where YPM 11852 was discovered (Zinbo, 1965). This unit at the YPM 11852 locality is reported to be about 90 m in thickness and is overlain by unnamed unconsolidated Pleistocene sediments (Nagasawa et al., 2003). The Shinazawa Formation of the fossil locality can be divided into three parts. The lower part is about 40 m thick, composed mainly of massive fine sandstone with burrows and bioturbation, and intercalated thin silt and fine gravel layers. The middle part, about 20 m thick, is formed by mainly massive fine sandstone with burrows and bioturbation, shell fragments, carbonaceous materials, and a shell bed. Grain size in the middle part coarsens upwards. The upper part, about 30 m thick, consists of cross-laminated coarse sandstone at the bottom, alternating coarse sandstone and fine gravel at the middle, and fine sandstone with a lignite layer at the top.
YPM 11852 was collected from the lower part. The fossil horizon is approximately 10 m below the 20 cm thick shell marker bed. Other vertebrate remains have been reported, such as a skull (still unprepared), mandible, rib and an atlas of cetacean specimens belonging to different individuals from a different horizon of the lower and middle parts, and two sirenian ribs from the upper part of the unit at the Shinazawa quarry (Nagasawa et al., 2003). Foraminifera (Elphidium) from the lower part of the formation also indicate that the environment was shallow marine (Ozawa et al., 1986). The area was a shallowing upward sequence.
The Shinazawa Formation has not been directly dated, but is thought to be late Pliocene (Honda and Applied Geological Society of Yamagata, 2016), based on fossils such as a Pliocene plant Comptonia kidoi (Suzuki, 1961) and various mollusks belonging to the Omma-Manganzian fauna (Otuka, 1939), such as Yabepecten tokunagai, Mizuhopecten yessoensis, Turritella saishuensis and others (Sato, 2000). The Omma-Mangazian fauna is composed of mainly cold water species and existed from the Pliocene to the Early Pleistocene age in the northeastern Sea of Japan area (Amano, 2007). A previous geological map (Ozawa et al., 1986) did not use the term Shinazawa Formation, and recognized the fossil locality area as the Kan-nonji Formation (Lower Pleistocene) in the northern Shonai area based on similar lithology. At present, it is not possible to correlate the two formations lacking dating for the Shinazawa Formation. Based on the mollusc fauna, we estimate the age of YPM 11852 as late Pliocene, and also including possibly Early Pleistocene.
Table 1.
Measurements of skull, aff. Balaenoptera bertae (YPM 11852) in mm. + are maximum preserved measurements of broken parts. * means only right side.

Table 2.
Compared named fossil balaenopterids. * means directly observed specimens.

Figure 1.
A–B, Maps showing the locality of YPM 11852. The base map of A is from a topographic map published by the Geospatial Information Authority of Japan. C, Stratigraphic sections of the locality modified from Nagasawa et al. (2003).

General description
Ontogeny.—YPM 11852 can be identified as an immature individual. It has visible parietal/squamosal and alisphenoid/pterygoid sutures; and does not preserve other cranial sutures. The holotype of Balaenoptera bertae identified as an immature individual also has a visible parietal/squamosal suture (Boessenecker, 2013a). The posterior tympanoperiotic process grows during ontogeny (Bisconti, 2001). YPM 11852 shows a slightly longer posterior process of the tympanoperiotic (about 105.0 mm long, based on the fossa of the skull for the posterior process) than that of B. bertae (about 80.0 mm long). In addition, estimated bizygomatic width of YPM 11852 from the right side is 719 mm + . It is wider than that of the holotype of B. bertae: 614 mm (Boessenecker, 2013a). Thus, YPM 11852 possibly was slightly ontogenetically older than the holotype of B. bertae.
Supraoccipital, exoccipital and basioccipital.—The supraoccipital is triangular in dorsal view (Figure 3D) with an almost straight nuchal crest. The posterolateral part of the nuchal crest is expanded laterally. The occipital tuberosities are formed as horizontal ridges and depressions ventral to the ridges. The foramen magnum is teardrop shaped with a bluntly triangular ventral margin. The occipital condyles are wide and project posteroventrally beyond the surface of the occipital shield. The intercondyloid notch is shallow. Ventrally, the exoccipital has a weakly depressed paroccipital concavity (Figure 4). Medial to the paroccipital concavity, there is an anteroposteriorly running shallow jugular notch medially. The lateral end of the exoccipital stops medial to the postglenoid process and forms a strong posteroexternal angle.
Squamosal.—The zygomatic process projects anterolaterally and is bounded medially by a shallow and vertical squamosal crease. A shallow squamosal cleft is developed, running horizontally from the posterior end of the alisphenoid to the dorsal broken base of the zygomatic process. A small elliptical opening filled by matrix (24.5 mm long, 13.5 mm high) is located anterior to the squamosal cleft, ventral to the parietal/squamosal suture on the temporal fossa; this might be a junction of the parietal, squamosal and alisphenoid. The postglenoid process is slender (preserved length is 88.0 mm) and projects posterolaterally. The lateral surface of the zygomatic process is weakly excavated, and the anterior tip of the zygomatic process orients anterolateral. Posterior to the postglenoid process, there is a shallow transverse groove of the external auditory meatus. In ventral view, the external auditory meatus is narrow anteroposteriorly (about 14 mm long) but wider medially (about 27 mm long). On the lateral surface of the squamosal, dorsal to the external auditory meatus, there is a shallow depression of the fossa for the sternocephalicus. On the ventral surface and just lateral to a broken anterior process of the periotic, only the base of the falciform process is preserved.
Figure 3.
Skull of YPM 11852, aff. Balaenoptera bertae; A, ventral view; B, right lateral view; C, left lateral view; D, dorsal view; E, anterior view; F, posterior view.

Pterygoid.—The right pterygoid (Figure 3E) is exposed laterally within the temporal fossa, located anterior to the squamosal, and contacts the alisphenoid.
Parietal.—The parietal shows a weakly excavated surface and forms the temporal wall and a dorsoventrally thin nuchal crest. The parietal/squamosal suture is visible.
Alisphenoid.—Based on relative position with other structures, the presumed alisphenoid has a small exposure (17.9 mm long, 6.0 mm high) on the anterolateral surface of the temporal fossa, anteroventral to the squamosal and dorsal to the pterygoid. The alisphenoid contacts with the squamosal cleft (Figure 3E). Its anterior part was possibly damaged as mentioned above.
Periotic.—The incomplete right periotic preserves only the dorsal part of the anterior and posterior processes and showing fresh broken surface (Figure 4). The lateral end of the posterior process of the tympanoperiotic is damaged, and the posterior process was possibly exposed on the lateral surface of the skull. The anterior process was at least 25.5 mm long. The preserved pars cochlearis is 48.5 mm long anteroposteriorly, 43.5 mm wide mediolaterally. The ventral surface of the pars cochlearis is broken away, and the section of a presumed internal acoustic meatus is exposed, and its preserved anteroposterior length is 13.5 mm. The remnants of a long posterior process projects posterolaterally from the damaged pars cochlearis. Though incomplete, the posterior process appears laterally narrowing (maximum length of the part is 15.5 mm). The posterior process was probably about 105.0 mm long (preserved maximum length is 86.0 mm) when complete.
Discussion
Comparison with fossil balaenopterids and linkage in the North Pacific of the Balaenopteridae
As mentioned above, YPM 11852 is identified as aff. Balaenoptera bertae, because it has two diagnostic features of the species, such as large occipital condyles occuping one third of the bizygomatic width, and a posteriorly elongate postglenoid process (Boessenecker, 2013a). YPM 11852 and B. bertae also have four shared features such as having 1) a large fossa for the sternocephalicus on the lateral surface of the squamosal, 2) a dorsally wider and ventrally narrower foramen magnum, and 3) an occipital condyle borders the lateral and ventral sides, but not dorsolateral sides of the foramen magnum, and 4) a straight nuchal crest at the level of the subtemporal crest. YPM 11852 is closely related to B. bertae by the two diagnostic features and four shared features.
Elongate and posteroventrally projecting postglenoid processes can be seen also in Marzanoptera tersillae, which formed a clade with Balaenoptera bertae by phylogenetic analysis of Bisconti et al. (2020b). In comparison with presumed closely related Marzanoptera tersillae, YPM 11852 and Balaenoptera have a more slender postglenoid process, larger occipital condyles, and also a dorsally wide and ventrally narrow foramen magnum. On the other hand, M. tersillae shows an elliptical foramen magnum.
Of note, Bisconti et al. (2020b) established the genus Marzanoptera with combination of some diagnostic features on the skull and periotic, such a remarkably short premaxilla, strongly narrowed ascending process of the maxilla and so on. Balaenoptera bertae was transferred to the genus as Marzanoptera bertae. However, the premaxilla, maxilla and body of the periotic were not preserved on the holotype of Balaenoptera bertae, which is currently only one specimen of the species (Boessenecker, 2013a). Thus, we prefer to use the genus name Balaenoptera for Balaenoptera bertae here.
In comparison with all named balaenopterids preserving skulls (Table 2), “Balaenoptera” siberi also possesses a long postglenoid process (longer than in Norrisanima miocaena, Parabalaenoptera baulinensis, Nehalaennia devossi, Incakujira anillodefuego, Archaebalaenoptera castriarquati, Plesiobalaenoptera hubachi), but the postglenoid process of “B.” siberi is shorter than in B. bertae and YPM 11852. Archaebalaenoptera liesselensis also shows larger occipital condyles, occupying almost one half of the exoccipital width compare with other balaenopterids (“Balaenoptera” siberi, Norrisanima miocaena, Protororqualus wilfriedneesi, Fragilicetus velponi, Parabalaenoptera baulinensis, Nehalaennia devossi, Incakujira anillodefuego, “Balaenoptera” ryani, Marzanoptera tersillae, Protororqualus cuvieri and Plesiobalaenoptera hubachi).
A large fossa for the sternocephalicus can be seen on not only YPM 11852 and B. bertae, but also on Norrisanima miocaena, Protororqualus wilfriedneesi and Diunatans luctoretemergo. Parabalaenoptera baulinensis and Marzanoptera tersillae have a much smaller single fossa for the sternocephalicus. Archaebalaenoptera liesselensis, Fragilicetus velponi, Incakujira anillodefuego and Plesiobalaenoptera hubachi possess two fossae for the sternocephalicus. The foramen magnum is rounded to dorsoventrally long elliptical in most balaenopterids (Norrisanima miocaena, Protororqualus wilfriedneesi, Parabalaenoptera baulinensis, Nehalaennia devossi, Marzanoptera tersillae and Plesiobalaenoptera hubachi), but only YPM 11852, B. bertae and Incakujira anillodefuego have a dorsally wider and ventrally narrower foramen magnum. YPM 11852, B. bertae and most of other balaenopterids (“Balaenoptera” siberi, Norrisanima miocaena, Protororqualus wilfriedneesi, Fragilicetus velponi, Nehalaennia devossi, Incakujira anillodefuego, and “Balaenoptera” ryani) possess occipital condyles that are positioned ventrolaterally, and do not extend to the dorsolateral margin of the foramen magnum. Only Parabalaenoptera baulinensis, Marzanoptera tersillae and Plesiobalaenoptera hubachi have occipital condyles that completely encircle the foramen magnum. Among balaenopterids, the lateral margin of the exoccipital at the level of the subtemporal crest in dorsal view is posteriorly concave (Norrisanima miocaena, Archaebalaenoptera liesselensis, Diunatans luctoretemergo, Incakujira anillodefuego, Marzanoptera tersillae, Archaebalaenoptera castriarquati and Plesiobalaenoptera hubachi), posteriorly convex (Fragilicetus velponi, Nehalaennia devossi and Protororqualus cuvieri), or straight (YPM 11852, B. bertae, “Balaenoptera” siberi, Protororqualus wilfriedneesi and Parabalaenoptera baulinensis). The combination of the diagnostic features and other features support YPM 11852 is most closely related to B. bertae.
Presumed closely related specimens, YPM 11852 and the holotype of Balaenoptera bertae show that the extinct balaenopterid was distributed both the eastern and western North Pacific during the late Pliocene to possibly into the Early Pleistocene. Currently, many balaenopterids have circum North Pacific distributions, including Balaenoptera musculus, B. physalus, B. borealis, B. acutorostrata, and Megaptera novaeangliae (Jefferson et al., 2008). There are some other marine mammal species and/ or genera known to have a circum-North Pacific distribution during the Pliocene to Early Pleistocene; Eschrichtius spp. (Tsai and Boessenecker, 2015; Kimura et al., 2018), Balaenoptera physalus (Tsai and Boessenecker, 2017), Herpetocetus (Oishi et al., 1985; Boessenecker, 2013a, b) Callorhinus gilmorei and Thalassoleon (Repenning and Tedford, 1977; Kohno, 1992; Kohno and Yanagisawa, 1997), and Hydrodamalis (Domning and Furusawa, 1995). This new record/YPM 11852 indicates that the extinct rorqual species Balaenoptera bertae can be added to this list of marine mammals with a circum-North Pacific distribution during the pre-glacial Pliocene-Pleistocene interval.
Conclusion
We report a skull YPM 11852 from the Shinazawa Formation (late Pliocene to Early Pleistocene) in Yamagata, Japan, as aff. Balaenoptera bertae. The specimen shows two diagnostic features of the species such as having large occipital condyles, and a posteriorly elongate postglenoid process. The combination of four more shared features also support this. YPM 11852 is probably slightly ontogenetically older individual than the holotype of B. bertae, based on the estimated bizygomatic width and slightly longer posterior process of the tympanoperiotic. YPM 11852 suggests that fossil balaenopterids had a circum North Pacific distribution, like several modern balaenopterid species. YPM 11852 is incompletely preserved, however nonetheless demonstrates the occurrence of Balaenoptera bertae and/or closely related taxa in the western North Pacific for the first time.
Acknowledgments
We thank the student at Yamagata University who discovered YPM 11852, and R. Abe (Yamagata Paleontological Research Group) for excavating the study material, and T. Kawabe (Yamagata University), M. Inaba (GeoPuck Associates) and K. Amamo (Joetsu University of Education) for discussion on geology and molluscan fauna. We thank Y. Otomo and T. Kawabe (both Yamagata University) for their contribution to discovery of the specimen, and the Educational Board of the former Tachikawa Town for cooperation at the locality. We also thank Robert W. Boessenecker (College of Charleston) and Cheng-Hsiu Tsai (National Taiwan University) for their constructive comments on this manuscript.