Prenatal developmental sequence of the skull of minke whales and its implications for the evolution of mysticetes and the teeth‐to‐baleen transition

Abstract

Baleen whales (Mysticeti) have an extraordinary fossil record documenting the transition from toothed raptorial taxa to modern species that bear baleen plates, keratinous bristles employed in filter‐feeding. Remnants of their toothed ancestry can be found in their ontogeny, as they still develop tooth germs in utero. Understanding the developmental transition from teeth to baleen and the associated skull modifications in prenatal specimens of extant species can enhance our understanding of the evolutionary history of this lineage by using ontogeny as a relative proxy of the evolutionary changes observed in the fossil record. Although at present very little information is available on prenatal development of baleen whales, especially regarding tooth resorption and baleen formation, due to a lack of specimens. Here I present the first detailed description of prenatal specimens of minke whales (Balaenoptera acutorostrata and Balaenoptera bonaerensis), focusing on the skull anatomy and tooth germ development, resorption, and baleen growth. The ontogenetic sequence described consists of 10 specimens of both minke whale species, from the earliest fetal stages to full term. The internal skull anatomy of the specimens was visualized using traditional and iodine‐enhanced computed tomography scanning. These high‐quality data allow detailed description of skull development both qualitatively and quantitatively using three‐dimensional landmark analysis. I report distinctive external anatomical changes and the presence of a denser tissue medial to the tooth germs in specimens from the final portion of gestation, which can be interpreted as the first signs of baleen formation (baleen rudiments). Tooth germs are only completely resorbed just before the eruption of the baleen from the gums, and they are still present for a brief period with baleen rudiments. Skull shape development is characterized by progressive elongation of the rostrum relative to the braincase and by the relative anterior movement of the supraoccipital shield, contributing to a defining feature of cetaceans, telescoping. These data aid the interpretation of fossil morphologies, especially of those extinct taxa where there is no direct evidence of presence of baleen, even if caution is needed when comparing prenatal extant specimens with adult fossils. The ontogeny of other mysticete species needs to be analyzed before drawing definitive conclusions about the influence of development on the evolution of this group. Nonetheless, this work is the first step towards a deeper understanding of the most distinctive patterns in prenatal skull development of baleen whales, and of the anatomical changes that accompany the transition from tooth germs to baleen. It also presents comprehensive hypotheses to explain the influence of developmental processes on the evolution of skull morphology and feeding adaptations of mysticetes.

Introduction

The evolution of heterodont dentition, characterized by multicuspid postcanine teeth, represents a key innovation that allowed early mammalian lineages to occupy the vacant niches after the K‐Pg extinction (Luo, 2007; O'Leary et al. 2013). After this initial diversification, several groups of mammals independently reduced their tooth number and complexity, until they completely lost their adult dentition. In extant taxa, complete loss of teeth is observed in some monotremes (echidna), xenarthrans (anteaters), pholidotans (pangolins), and cetaceans (baleen whales) (Davit‐Béal et al. 2009). Baleen whales (Mysticeti) are the only living eutherian clade to have evolved a unique functional substitute for teeth in the form of keratinous plates and bristles, growing from at the edges of the palate, called baleen. The plates function as a sieve and allow these mammals to filter small prey from the water (Pivorunas, 1979; Jensen et al. 2017). The evolution of baleen allowed mysticetes to reach their impressive body size (Goldbogen & Madsen, 2018) and to occupy different feeding niches, with the four living families each displaying its own set of skull adaptations related to their bulk filter‐feeding style (Goldbogen et al. 2017; Werth et al. 2018). Rorquals (Balaenopteridae) are the most speciose group of modern baleen whales (Berta et al. 2016). Their filter‐feeding strategy known as lunge‐feeding involves engulfment of a large amount of water and prey, and the employment of baleen to filter out the water before swallowing prey (Goldbogen et al. 2017; Kahane‐Rapport & Goldbogen, 2018). This feeding style required these mammals to evolve a characteristic skull shape, with a broad and flat rostrum and an elongated mandible (Bouetel, 2005; Pyenson et al. 2013; Kahane‐Rapport & Goldbogen, 2018).

The progressive loss of teeth and acquisition of related skull adaptations is well documented in the fossil record of mysticetes, but the timing and mode of the teeth‐to‐baleen transition are still debated (e.g. Deméré & Berta, 2008; Boessenecker & Fordyce, 2015a; Peredo et al. 2017; Fordyce & Marx, 2018). Baleen whales split from toothed whales (Odontoceti) around 40 Ma ago (Berta et al. 2016). The earliest diverging mysticete lineages (Coronodon havensteini ~ 34–28 Ma, Llanocetidae ~ 37–33 Ma and Mammalodontidae ~ 28–24 Ma) had limited telescoping of the skull, defined as the overlapping of the bones of the braincase to allow for the migration of the external nares towards the posterior end of the skull (Miller, 1923), and a short rostrum, in contrast to the elongated head and mouth of modern baleen whales. Since they retained an ancestral heterodont dentition, they are generally believed to have been suction and raptorial feeders, although it has been proposed that some of these animals could have used teeth as a functional analog for baleen to filter‐feed, as has been observed in some modern seals (Fordyce & Marx, 2016, 2018; Geisler et al. 2017; Hocking et al. 2017). Later lineages (e.g. Aetiocetidae, ~ 31–24 Ma) evolved some skull features related to filter‐feeding, but still retained teeth in both jaws. It is debated whether some representatives of these groups could have had both teeth and baleen‐like structures, given the diastemata between teeth and the possible presence of bony correlates in the rostrum for baleen (Deméré et al. 2008; Marx et al. 2016; Peredo et al. 2017). The Eomysticetidae (~ 29–27 Ma) are the earliest clade to be functionally toothless and likely possessed baleen plates. They had a broad and flat rostrum, indicating that they probably employed a filter‐feeding strategy, and undisputable bony correlates for baleen in the rostrum and palate, although some species might have still retained vestigial teeth in the anterior portion of the rostrum and mandible (Boessenecker & Fordyce, 2015a,2015b, 2016; Berta et al. 2016).

All modern mysticete species still retain vestiges of their toothed ancestry in prenatal development, and still develop mineralized tooth germs in both jaws, with dentin but no visible enamel. The teeth never erupt from the gums and are resorbed before birth, with the neonates already bearing baleen plates (Van Dissel‐Scherft & Vervoort, 1954; Karlsen, 1962; Slijper, 1976; Ishikawa et al. 1999; Fudge et al. 2009; Thewissen et al. 2017). Despite many qualitative descriptions of prenatal specimens (e.g. Hilaire, 1807; Eschricht, 1849; Tullberg, 1883; Ridewood, 1923), and more recent histological studies specifically on tooth germs and baleen development (e.g. Karlsen, 1962; Ishikawa & Amasaki, 1995; Ishikawa et al. 1999; Thewissen et al. 2017), no conclusive data are available on how and when the transition from teeth to baleen occurs in ontogeny. Additionally, no full description of a developmental sequence is available for any mysticete species. Many of the available studies lack an evolutionary approach and do not explicitly relate skull ossification and development to the teeth‐to‐baleen transition in evolution.

The primary goal of this paper is to provide the first detailed account of skull ontogeny, with emphasis on the transition from tooth germs to baleen, for two sister species of rorquals, the Antarctic and the common minke whale (Balaenoptera bonaerensis and Balaenoptera acutorostrata) and relate these findings to the evolution and diversification of mysticetes. These species warrant a more exhaustive investigation not only from an evolutionary prospective but also from an ecological one. They are the main focus of scientific whaling and the lack of information on their biology and development causes many pregnant females to be caught every year (Bando et al. 2018). The implications of this activity are difficult to predict given the paucity of existing data, with the Antarctic minke whale only being recognized as a species since the 1990s (Rice, 1998). Very few descriptions of skull ossification in prenatal specimens have been published for either species (Eschricht, 1849; Julin, 1880; Sawamura, 2008). More data are available on their teeth‐to‐baleen transition during ontogeny, as Ishikawa & Amasaki (1995) and Ishikawa et al. (1999) conducted a histological analysis on a series of embryos and fetuses of Antarctic minke whale, providing a framework for the timing of tooth resorption and baleen development in all living mysticetes. Although they did report resorption of tooth germs and baleen ‘rudiments’ at the same time in a few specimens, they did not present a clear description of these observations. Therefore, it is difficult to generalize their findings or replicate their study.

Here, a detailed anatomical description of 10 minke whale fetal specimens (eight Antarctic and two common minke whales) is presented. I focused on the skull ossification sequence and shape changes, and the teeth‐to‐baleen transition. Data was collected using diffusible iodine‐based contrast enhanced computed tomography (diceCT; Gignac et al. 2016) and traditional CT scanning, based on the high‐quality results obtained using these non‐invasive methods in another mysticete species, the humpback whale (Lanzetti et al. 2018). Three‐dimensional (3D) geometric morphometric (GM) landmark analysis was used to quantify skull shape changes in a developmental series of minke whales, from the earliest fetal stages to adult. Growth curves constructed with data from the literature for the two minke whale species are also presented, to better visualize the major stages of development and the relative appearance of teeth and baleen during gestation. Finally, these results are evaluated in the context of previous research and hypotheses formulated on the relationship between skull shape changes and the teeth‐to‐baleen transition in development and evolution. These preliminary hypotheses will require further testing by investigating the ontogeny of these species with other methods such as histology or molecular analyses, and by examining the prenatal development of other baleen whales.

Materials and methods

Specimen acquisition and age estimation

The internal anatomy of eight fetal Antarctic minke whale (B. bonaerensis) specimens, housed at the National Museum of Nature and Science (NSMT) Research Facility in Tsukuba, Japan and two fetal common minke whale (B. acutorostrata) specimens from the Natural History Museum of Denmark (ZMUC) (Copenhagen, Denmark) is described (Table 1). Detailed external measurements following Lanzetti et al. (2018) were taken on all described fetuses up to 1.21 m in total body length due to the excessive weight and size of larger specimens (Supporting Information Table S1). Since the Antarctic minke whale was not recognized as a separate species until recently (Rice, 1998) , specimens at both institutions were labeled as either ‘minke whale’ or ‘B. acutorostrata’, as they were collected from the late 1800s to the 1960s, although most of the specimens had a known collection location, with fetuses from Japan being collected in the Southern Pacific Ocean and those from Denmark being collected in Greenland. The two minke whale species are known to be geographically separated, with the common minke whale mostly found in the Northern Hemisphere and the Antarctic minke only found in the Southern Hemisphere (Brownell et al. 2000). A dwarf subspecies of common minke whale is also found in certain regions in the Southern Hemisphere, although these whales are rarer than the Antarctic species and significantly smaller (Kato & Fujise, 2000; Perrin et al. 2018). Given this geographic separation and the size of the specimens, it was possible to assign the Japanese specimens to the Antarctic species confidently, whereas those from Denmark were assigned to the northern ‘common’ species. The size, coloration, and other external anatomical features of the specimens were also congruent with these assignments. Even if these specimens belong to distinct species, results are combined in this study because they are sister species (Gatesy et al. 2013) with similar gestation times (Ivashin & Mikhalev, 1978; Masaki, 1979). All specimens at NSMT are preserved in formalin 5% solution, and the specimens from ZMCU are preserved in ethanol 70%. Images of all specimens and relative museum labels are deposited in the Balaenoptera database (Supporting Information Table S2).

Table 1.

Prenatal Antarctic and common minke whales (Balaenoptera bonaerensis and Balaenoptera acutorostrata) specimens described

Specimen code	Catalog number	Age (months)	Age (% gestation time)	TL (cm)	Part of specimen CT scanned
Mf1	NSMT27154	4	42	28	Whole body
Mf2	NSMT27149	4.5	47	41	Whole body
Mf3*	ZMCU‐CN6x	5	48	48	Whole body
Mf4	NSMTblue10	5.4	57	74	Head only
Mf5	NSMTwhite11	6.3	66	110	Head only
Mf6	NSMTwhite15	6.3	66	110	Head only
Mf7	NSMT27171	6.5	68	115	Head only
Mf8*	ZMCU‐CN4x	7.7	73	125	Head only
Mf9	NSMT27175	7.6	80	182	Head only
Mf10	NSMT27174	8.1	85	212.5	Head only

All data acquired by A. Lanzetti (A.L.).

NSMT: National Museum of Nature and Science, Tsukuba, Japan. ZMU(C) Natural History Museum of Denmark, Copenhagen, Denmark.

^*Common minke whale specimens. All other listed specimens belong to Antarctic minke whale.

A Fetal Stage (FS) of development was assigned to each of the specimens based on previous work on the development of the humpback whale (Lanzetti et al. 2018) and the pantropical spotted dolphin (Stenella attenuata; Sterba et al. 2000; Thewissen & Heyning, 2007). New stages were added to best describe the teeth‐to‐baleen transition in the oldest specimens in the series (Mf6–Mf10). Applying this standard staging system based on external features allowed direct comparison of these specimens with fetuses of other cetacean species.

Growth curves were also compiled based on published data of total length and estimated gestational age for both species. They were constructed using the same methodology described in previous work (Frazer & Huggett, 1973; Roston et al. 2013; Lanzetti et al. 2018). Details on this procedure can be found in Supporting Information Fig. S1. The growth equations obtained from the growth curves were used to estimate the absolutes age of the samples, and the relative age, expressed as the percentage of total gestation time. This also made it possible to visualize the absolute timing of tooth formation and resorption described by Ishikawa & Amasaki (1995), by calculating the age of the specimens they presented according to the equation, making it easier to predict what could be observed in the fetuses described in this work of similar approximate age. This age estimation method is not unequivocally reliable since most of these data were compiled using information collected by whaling stations in the first half of the 20th century and later compiled by the International Whaling Commission, and much of this information has been found to have been misreported (Frazer & Huggett, 1973; Clapham & Ivashchenko, 2016, 2018). Therefore, although the growth curves represent a visual aid to understand growth patterns and the timing of the teeth‐to‐baleen transition, they should be used with caution when comparing specimens of different species based on their relative gestational age.

CT scanning and image reconstruction

A diceCT protocol was applied to the two smallest specimens in the series, Mf1 and Mf2. The iodine stain enhances the contrast difference between tissues that would otherwise be impossible to recognize (Gignac et al. 2016), especially in smaller specimens when using a medical‐grade CT scanner. The CT scan resolution was in fact too poor to create a full 3D reconstruction of the internal skeletal anatomy of these two specimens, but this technique at least allowed the presence of tooth germs to be recognized. The staining solution consisted of 1% iodine (1% KI – potassium iodide + 0.2% I₂E – metal iodine) in deionized water. The two specimens were immersed in the staining solution for 10 h (Mf1) and 24 h (Mf2) according to their size. After the CT scanning, these samples were immersed in a destaining solution of 3% sodium sulfite (Na₂SO₃) and deionized water for 1 h to eliminate the color of the stain and best preserve these rare specimens for future use. This diceCT protocol had already proven effective on other baleen whale fetal samples preserved in 70% ethanol (Lanzetti et al. 2018), and was only slightly modified to accommodate the different preservation medium (5% formalin). The other eight specimens described in this paper were not stained because of their larger size, which would have required very long periods of staining that might have damaged the tissues (Metscher, 2009; Gignac et al. 2016). The CT scans were performed at local institutions: the samples from the ZMUC were scanned at the Laboratory of Biological Anthropology of the University of Copenhagen, Denmark, using a Siemens Somatom Definition medical‐grade CT scanner, and those from the NSMT were scanned at the Nippon Veterinary and Life Science University in Tokyo, Japan, using a Toshiba Aquilion Prime medical‐grade CT scanner (Table 2).

Table 2.

CT scan details for minke whale prenatal specimens

Specimen code	TL (cm)	Voltage (kV)	Current (mA)	Slice thickness (mm)	Number of slices	MorphoSource DOI
Mf1*	28	120	25	0.3	951	https://doi.org/10.17602/m2/m65343
Mf2*	41	120	25	0.3	951	https://doi.org/10.17602/M2/M65345
Mf3	48	120	300	0.4	893	https://doi.org/10.17602/M2/M65390
Mf4	74	120	34	0.3	781	https://doi.org/10.17602/M2/M65350
Mf5	110	120	54	0.3	1041	https://doi.org/10.17602/M2/M65377
Mf6	110	120	83	0.3	1181	https://doi.org/10.17602/M2/M65379
Mf7	115	120	131	0.3	1281	https://doi.org/10.17602/M2/M65381
Mf8	125	120	300	0.4	849	https://doi.org/10.17602/M2/M65388
Mf9	182	120	175	0.3	2571	https://doi.org/10.17602/m2/m65384
Mf10	212.5	120	175	0.21	2939	https://doi.org/10.17602/m2/m65386

^*Specimens CT scanned after iodine staining (diceCT).

The CT images were created in DICOM format and were then converted to JPEG, cropped and contrast‐adjusted using imagej (Schneider et al. 2012). They were then imported in avizo lite 9.5 (Fei, 2018), where the ossified regions of the skull and postcranial skeleton, tooth germs and baleen tissue were highlighted and segmented as different materials. Ossification was determined by evaluating the relative gray‐scale values of the tissues when compared with the surrounding areas. Higher gray‐scale values (i.e. closer to white) indicate higher tissue density, therefore using this and knowledge of baleen whale anatomy it is possible to separate ossified areas from surrounding tissues and cartilage. Tooth germs contain dentin, a very dense tissue (Karlsen, 1962; Koussoulakou et al. 2009; Thewissen et al. 2017), and they appear as distinct, bright structures. Baleen, instead, is composed of keratin, which has only a slightly higher density than most tissues (Pivorunas, 1979), but it is possible to clearly separate it from the surrounding flesh (e.g. Ekdale et al. 2015). All CT images are deposited in the Balaenoptera database (Table 2).

Landmark data collection and morphometric analyses

To assess quantitatively shape change throughout the ontogenetic series and to be able to relate the growth of these species to others, landmark‐based 3D GM analyses were performed on the anatomy of the skull, using the same landmarks and workflow employed in a study of humpback whale development (Lanzetti et al. 2018). Landmarks were collected in two configurations: one including the entire skull (16 landmarks) and one only including the rostrum, from its anterior tip to the posterior border of the nasals (12 landmarks). The rostral portion of the skull is key to the evolution of mysticetes and contains major adaptations to filter‐feeding, and it has also been shown to change allometrically through postnatal development relative to body size (Nakamura & Kato, 2014). The second configuration allows more precise comparison of shape change in this region of the skull among specimens at different stages of development. The whole skull configuration is important to evaluate relative shape changes between rostrum and braincase, and the progression of telescoping. The dentaries were not included in the landmark analyses since they showed high levels of deformation and were broken in many specimens. The rest of the skull appeared to be un‐deformed, with its general shape matching those of fetal specimens of rorquals previously described using CT scan images (e.g. Hampe et al. 2015; Lanzetti et al. 2018). Unfortunately, there are no comprehensive qualitative descriptions of the skull of minke whale fetuses, only partial accounts (Eschricht, 1849; Julin, 1880), thus it is not possible to check for deformation by comparing these descriptions with direct accounts obtained using traditional methods (i.e. dissection). Consequently, deformation might have had an influence on the results of the analyses; more specimens of similar ontogenetic age are needed to confirm the observations presented in this work.

To examine a broader growth series, additional specimens of different ages (two neonates and two adults) were added to the dataset, all belonging to B. acutorostrata. The 3D models of these specimens were acquired using CT scanning and laser surface scanning by other researchers and their data was generously provided to use in this project (Table 3). Landmarks were collected on all specimens in avizo lite 9.5 (Fei, 2018) using the 3D models of the skulls. Landmark collection was done twice by the same operator (A. Lanzetti) to minimize errors. They were then imported in morphoj 1.06d (Klingenberg, 2011) for analysis. This software was chosen because it incorporates the generalized least‐squares Procrustes superimposition, which allows landmark configurations to be compared independent of specimen size or orientation (Rohlf & Slice, 1990). All analyses were conducted assuming object symmetry, since it has been shown that mysticetes have a symmetric skull (Fahlke & Hampe, 2015). Consistency between landmark takes was tested by performing an anova analysis on the shape coordinates, which showed no significant difference between the two replicate sets (Supporting Information Data S1). After this, the two sets were averaged to obtain a final landmark configuration for each specimen for use in the following analyses. This procedure was implemented separately for the two landmark configurations (whole skull and rostrum only).

Table 3.

Postnatal common minke whale specimens used in morphometric analyses

Specimen code	Catalog number	TL (m)	Growth stage	Digitalization method	Data provided by
Mn1	USNM593554	3.51	Neonate	CT scanning	Smithsonian Institution Bio‐Imaging Research Center
Mn2	LACM72507	3.08	Neonate	Surface laser scanning	G. Franci (Franci & Berta, 2018)
Ma1	HSU2670	7	Adult	Surface laser scanning	G. Franci (Franci & Berta, 2018)
Ma2	CAS23807	8	Adult	Surface laser scanning	G. Franci (Franci & Berta, 2018)

CAS: California Academy of Sciences, San Francisco, CA, USA. HSU: Natural History Museum Humboldt State University, Arcata, CA, USA. LACM: Natural History Museum of Los Angeles County, Los Angeles, CA, USA. USNM: National Museum of Natural History, Washington, DC, USA.

A discriminant function analysis (DA) was performed for both configurations using developmental stage as the classifier: ‘early fetal’, ‘late fetal’, ‘neonate’, ‘adult’. By analyzing the specimens qualitatively, the fetal period was divided into two stages: ‘early fetal’ and ‘late fetal’. The late fetal stage encompasses most of the second half of the gestation and it is clearly distinguishable from the previous early fetal period. In the late fetal stage, all major skull bones are almost completely ossified and only the sutures remain open. The tooth germs have clearly started resorption and are sparser and smaller. There is also a change in the rostrum length to braincase length ratio. Observing this pattern and dividing the specimens accordingly allowed better characterization of the skull shape changes during the prenatal development of the species, although this subdivision is approximate and should be evaluated for each species separately in an interspecific analysis. Following this growth stage characterization, the dataset is subdivided as follows: four early fetuses (Mf1–Mf4), six late fetuses (Mf5–Mf10), two neonates (Mn1–Mn2) and two adults (Ma1–Ma2). The DA analysis compares data divided in a priori groups to each other. The results assess whether the groups are significantly different from each other but also produce landmark configurations all scaled to the same selected ‘target shape’. Using the growth stage as classifier in the DA analyses allowed comparison of the average shape obtained by all specimens at that stage to the adult average, providing a direct way to visualize changes in shape through development, independent of size or individual variability (Timm, 2002; Klingenberg, 2011; Lanzetti et al. 2018).

Results

Stages of development in the minke whale

Both the external characteristics and internal skull anatomy of the 10 fetal minke whale (m.w.) specimens (Table 1) were examined. The growth curve summarized the position of the specimens in the ontogenetic series of each species and provided an estimate of the timing of the hypothesized occurrence of tooth germs or developing baleen based on Ishikawa & Amasaki (1995) (Fig. 1). Since the absolute age of the specimens is only an approximation, a list of the specimens divided by FS, and a brief description of the distinguishing and diagnostic features of each stage is provided, focusing on changes in head anatomy, skull ossification, suture closure, and major characteristics of teeth and baleen. The descriptions are based on the 3D reconstructions of the internal anatomy derived from the CT images. Therefore, their quality is influenced by imaging resolution, as well as by the status of preservation of the specimen. However, all bony elements can be recognized and described for each stage. The two specimens of common minke whale are described as part of the growth series and are labeled accordingly; all other specimens belong to the Antarctic species. The staging followed the descriptions provided by Thewissen & Heyning (2007) and Lanzetti et al. (2018), and stages new to this paper are marked as such and described in more detail. The presence of tooth germs and baleen is compared with the descriptions provided by Ishikawa & Amasaki (1995) for specimens of similar total length and approximate gestational age (Table 4). A detailed account of the changes relative to teeth‐to‐baleen transition is provided in the next section of the Results. Complete descriptions of each specimen including the ossification state and shape of each major bony element of the skull, and postcrania when possible, are provided in Supporting Information Data S2 and the detailed measurements in Table S1. Other aspects of the skull ontogeny are addressed in the Discussion.

Figure 1.

Calculated growth trajectories, using total length (cm) and estimated gestational age (months), for the two minke whale species (Balaenoptera bonaerensis – Antarctic minke whale and Balaenoptera acutorostrata – common minke whale). Squares = specimens included in this study and retro‐fitted to the growth curve; crosses = lengths at which important steps in the teeth to baleen transition were recorded in the species; diamonds = estimates of length at birth (m). Embryo/fetus line is based on the age estimate for balaenopterids listed by Roston et al. (2013) (2.1 months). Early fetus/late fetus line is based on observations of internal and external anatomy of the specimens and their Fetal Stage (FS 20–24: ‘early fetus’; FS 25–27: ‘late fetus’). Individual data points have been removed to make the graph more legible and highlight the specimens from this study. Complete graph with data points, equations and R‐values for the curves can be found in Fig. S1.

Table 4.

Teeth‐to‐baleen transition in prenatal minke whales

Specimen code	Fetal Stage	Tooth germs	Baleen	I&A (1995)* specimen number	I&A (1995)* Tooth germs	I&A (1995)* Baleen
Mf1	20/21	?	No	16	Yes	No
Mf2	21/22	Yes	No	18	Yes	No
Mf3	21/22	Yes	No	18	Yes	No
Mf4	23/24	Yes	No	20	Yes	No
Mf5	25	Yes	Rudiments	21‐22	Yes	Rudiments
Mf6	25	Yes	Rudiments	21‐22	Yes	Rudiments
Mf7	26	Yes	Rudiments	23	Yes	Rudiments
Mf8	26	Yes	Rudiments	24	Yes	Rudiments
Mf9	27	No	Yes	24	No	Yes
Mf10	27	No	Yes	n.a.	n.a.	n.a.

Fetal stages are referred to the specimens described in this study. Fetal Stages 20–24: ‘early fetus’ growth stage; Fetal Stages 25–27: ‘late fetus’ growth stage.

^*Ishikawa & Amasaki (1995) specimens with similar TL, numbers follow original publication.

Fetal Stages 20/21/22

Eyelids fused, umbilical hernia retracted/tactile hair present, blowhole in the last one‐third of head/eyelids separated, throat grooves present (Figs 2 and 3, Supporting Information Video S1)

Figure 2.

Mf2 (NSMT27149; TL 41 cm) external morphology and longitudinal CT image of the rostrum. (A) Left lateral view of external morphology, (B) CT slice of longitudinal section of rostrum of Mf2. Red dashed line in (A) represents CT slice location and orientation. tg, tooth germs.

Figure 3.

Mf3 (ZMCU‐CN6x; TL 48 cm) external and internal morphology. (A) Left lateral view of external morphology, (B) left lateral view of 3D rendering of the internal morphology of head, (C) head in posteroventral view, (D) head in dorsal view, (E) head in ventral view. Ossified skull bones in yellow, nasal bones in green, tooth germs in red. An animation of the 3D model of this specimen is shown in Video S1. Ossified elements are labeled in the figures. Abbreviations for Figs 3, 4, 5, 6, 7, 8, 9, 10, 11 and [Link], [Link], [Link]: ac, alveolar canal (maxilla and dentary); b, baleen; bhy, basihyoid; bo, basioccipital; br, baleen rudiments; bs, basisphenoid; cp, coronoid process (dentary); d, dentary (= mandibular rami); dif, dorsal infraorbital foramina; ect, ectotympanic; eo, exoccipital; f, frontal; hy, indeterminate hyoid elements; ip, infraorbital process (maxilla); j, jugal; inp, interparietal; l, lacrimal; m, maxilla; ms, mandibular symphysis; n, nasal; p, parietal; pg, postglenoid process (squamosal); pl, palatine; plf, palatine foramina; plp, palatine process (maxilla); pm, premaxilla; pt, pterygoid; shy, stylohyoid; so, supraoccipital; sop, supraorbital process (frontal); sq, squamosal; tg, tooth germs; thy, thyrohyoid; v, vomer; zp, zygomatic process (squamosal).

Specimens

Mf1, Mf2, Mf3‐common m.w.

Skull ossification

All major bones are recognizable as they started ossifying, including nasals in Fetal Stage 22. Infraorbital process of maxilla is ossified but not in contact with the frontal. Interparietal is visible at the top of the neurocranium. Ectotympanic has assumed a 3D shape but still lacks posterior and ventral portion. Fontanelles and sutures are open.

Tooth germs

Present at least from FS 21, form complete rows in both jaws.

Fetal Stages 23/24

Pigmentation present/throat grooves well developed, extend below the flippers (Fig. 4, Supporting Information Video S2).

Figure 4.

Mf4 (NSMTblue10, TL 72 cm) external and internal morphology. (A) Left lateral view of external morphology, (B) dorsal view of 3D rendering of the internal morphology of head, (C) head in ventral view, (D) head in left lateral view, (E) head in posteroventral view. Ossified skull bones in yellow, nasal bones in green, tooth germs in red. An animation of the 3D model of this specimen is shown in Video S2. Ossified elements are labeled in the figures. Abbreviations listed in Figure 3.

Specimen

Mf4.

Skull ossification

Rostrum is elongated compared with previous stages and has the same length as the mandible, whereas in FS 20–22 the mandible was longer than the rostrum. Nasals are elongated posteriorly. No major changes in the infraorbital process of maxilla or interparietal. Ectotympanic ossification advanced, but still lacks skull articulation. The supraoccipital shield is extended anteriorly, contacts the interparietal. Periotic is recognizable in the CT scans and ossified.

Tooth germs

Present, only a few visible but the bony correlates (open upper and lower alveolar canals) imply that they form a complete row in both jaws.

Fetal Stage 25 – new stage

Thick baleen gum develops medial to gingival ridge (Fig. 5, Supporting Information Fig. S2 and Video S3).

Figure 5.

Mf6 (NSMTwhite15, TL 110 cm) external and internal morphology. (A) Left lateral view of external morphology, (B) 3D rendering of internal morphology of head in dorsal view, (C) head in ventral view, (D) head in left lateral view, (E) head in posteroventral view. Ossified skull bones in yellow, nasal bones in green, tooth germs in red, baleen rudiments in blue. An animation of the 3D model of this specimen is shown in Video S3. Ossified elements are labeled in the figures. Abbreviations listed in Figure 3.

Notes on staging: this new fetal stage is characterized by the development of a thick gum in the medial part of the palate, especially visible in the posterior portion of the upper jaw (‘baleen gum’), which progressively becomes smaller anteriorly. In the anterior one‐third of the rostrum, only a groove is visible between the palate and the lip.

Specimen

Mf5, Mf6.

Skull ossification

Rostrum is flattened dorso‐ventrally. Nasals continue to elongate posteriorly. The infraorbital process of maxilla is not in contact with the frontal. Interparietal is sutured with parietal but its shape is still recognizable. Ectotympanic has a small portion lateroventrally that lacks ossification. Mandibular symphysis is probably present. Supraoccipital is in contact with interparietal anteriorly. Periotic is sutured to the skull.

Tooth germs/baleen rudiments

Present. A full tooth row is visible in the upper jaw, whereas only a few smaller tooth germs are visible in the lower jaw, where the alveolar canal is also closing. Denser tissue is present in the posterior one‐third of the upper alveolar canal, medial to the teeth and not connected to the bone. This tissue occupies the empty space between the dorsal (rostral) and ventral (palatal) processes of the maxilla.

Fetal Stage 26 – new stage

Ridge (keel) forms at the center of the baleen gums (Fig. 6, Supporting Information Fig. S3 and Video S4).

Figure 6.

Mf7 (NSMT27171, TL 115 cm) external and internal morphology. (A) Left lateral view of external morphology, (B) left lateral close‐up view of head, (C) cross‐section of 3D rendering of internal morphology of head, (D) head in dorsal view, (E) head in ventral view, (F) head in left lateral view, (G) head in posteroventral view. Ossified skull bones in yellow, nasal bones in green, tooth germs in red, baleen rudiments in blue. Ossified elements are labeled in the figures. Abbreviations listed in Figure 3.

Notes on staging: this new stage is characterized by the formation of a distinct cartilaginous ridge at the center of the baleen gum. It is most prominent at the posterior end of the rostrum. The gingival groove still borders the baleen gum laterally and it is the only structure present at tip of the rostrum.

Specimen

Mf7, Mf8‐common m.w.

Skull ossification

Rostrum is broad and flat. Nasals contact the frontal posteriorly, whereas the infraorbital process of the maxilla remains unconnected to the frontal. Interparietal is not visible anymore, as the supraoccipital shield has extended anteriorly to cover it. Ectotympanic only has a very small unossified segment ventrolaterally, but it is now sutured to the squamosal. Mandibular rami are united by a short and presumably ligamentous symphysis. Fontanelles are closed. Exoccipital is sutured to the squamosal, and basioccipital and basisphenoid contact the surrounding dermal bones. Periotic has assumed the typical ‘inflated’ shape of mysticetes.

Tooth germs/baleen rudiments

Tooth germs are present for the entire length of both jaws, although they are less evenly distributed. Upper and lower alveolar canals are narrower than in previous stages. Denser material (baleen rudiments) is present for the entire length of the rostrum, occupying most of the space between the dorsal and ventral processes of the maxilla. In the dorsal part of the rostrum, where the keel occurs externally, this material extends medially and laterally to the tooth germs, covering some of them.

Fetal Stage 27 – new stage

Transversally oriented rows of connective tissue and baleen visible (Fig. 7, Supporting Information Fig. S4 and Video S5).

Figure 7.

Mf10 (NSMT27174, TL 212.5 cm) internal morphology. (A) Dorsal view of 3D rendering of the internal morphology of head, (B) head in ventral view, (C) head in left lateral view, (D) head in posteroventral view. Ossified skull bones in yellow, nasal bones in green, baleen in blue. An animation of the 3D model of this specimen is available in Video S5. Ossified elements are labeled in the figures. Abbreviations listed in Figure 3.

Notes on staging: this new stage is characterized by the presence of the transversal ridges that will form the base of the baleen plates visible in the gums of the upper jaw. They are present for the entire length of the palate except at the anterior end. It is known that this structure represents developing baleen based on the work on blue whale fetuses by Tullberg (1883) (translated and commented on by Fudge et al. 2009).

Specimen

Mf9, Mf10.

Skull ossification

The skull has assumed the familiar shape observed in minke whale neonates. All the dermal bones are at least partially in contact, except the infraorbital process of the maxilla, which is still separated from the frontal on one or both sides of the skull. Ectotympanic is completely ossified and sutured to the squamosal for the entire length of its posterior process. Supraoccipital has assumed the characteristic triangular shape of postnatal minke, with its anterior end reaching the midpoint of the orbit and contacting the frontals anteriorly. Endochondral bones are still not sutured posteriorly, as this process takes places postnatally (Walsh & Berta, 2011).

Baleen

No tooth germs are present. The upper alveolar canal is narrow but open, and the lower alveolar canal is very shallow, although it is still marked for the entire length of the rami. Transversal process of the baleen appears to not have any internal connection with the rest of the skull, although this might be due to preservation issues and CT scan limitations. From both external and internal examination, it appears to be more developed posteriorly and gradually progressing anteriorly.

Anatomical aspects of the teeth‐to‐baleen transition in ontogeny

Tooth number, size and distribution

Specimens spanning from FS 20 to FS 24 (Mf1, Mf2, Mf3, Mf4) have a distinctive gingival ridge, or gum, that extends throughout each side of the upper jaw (Fig. 8). No distinct gum is recognizable on the lower jaw. The ridge is identifiable until FS 26 (Mf7, Mf8), although it undergoes significant structural changes after FS 24. Paired with the internal anatomical information from the CT imaging, the appearance of this ridge can help determine the phase of gestation and teeth‐to‐baleen transition that the specimen occupies.

Figure 8.

‘Tooth germ’ stage (FS21–FS24), represented using specimen Mf4. (A) Left lateral close‐up view of head, (B) close‐up view of gums in the upper jaw, (C) cross‐section of 3D rendering of internal morphology of head. Red dashed line in (A) represents cross‐section location and orientation, box represents magnified section of gums. Ossified skull bones in yellow, nasal bones in green, tooth germs in red. Ossified elements are labeled in the figure. Abbreviations listed in Figure 3.

As anticipated from the external appearance of their gums, starting from Mf2, all specimens have clearly recognizable tooth germs in both jaws, except the two older specimens at FS 27 (Mf9 and Mf10). As only medical‐grade CT scans were available, presence of teeth in Mf1 could not be assessed due to limited imaging quality, but it is likely that this specimen also had tooth germs based on its external anatomy and the level of development in the dentition of Mf2. The resolution of the CT images also does not allow for a precise count of the tooth germs in all fetuses or to describe their shape. As many as 35 rounded or conical tooth germs per side in the upper jaw of Mf6 were identified. This elevated number is comparable to the results in Ishikawa & Amasaki (1995) (Table 4). None of the lower jaws preserved a visible complete tooth row: this might be due to the poor preservation of these elements and to the relative size of the mandibular tooth germs, which have been found to be consistently smaller than those in the upper jaw among mysticete fetuses of numerous species (Ridewood, 1923; Karlsen, 1962; Thewissen et al. 2017; Lanzetti et al. 2018). From the presence of a marked alveolar canal along the entire length of the rami in all specimens from FS 22 to FS 26 and from previous accounts (Eschricht, 1849; Julin, 1880; Ishikawa & Amasaki, 1995), it can be assumed that all specimens up to Mf8 (FS 26) have a complete tooth row in the lower jaws as well. The characteristic ‘double teeth’, defined as secondarily fused or very closely spaced tooth germs (Thewissen et al. 2017), are only visible in the lower jaw of Mf8. Julin (1880) described ‘double teeth’ in a mandible of a young minke fetal specimen, tentatively assigned to FS 21/22. It is probable that this tooth morphology occurs in the minke whale from very early stages of development, as it has been reported for other mysticetes (Ridewood, 1923; Van Dissel‐Scherft & Vervoort, 1954; Karlsen, 1962; Thewissen et al. 2017).

The size of the tooth germs can be used as a proxy to determine when the tooth germs start resorption, especially given that the quality of the CT scans does not allow for direct comparison of other factors such as spacing between germs. The tooth germs stop increasing in size at FS 25 (Mf6) (Table 5), indicating the probable start of resorption after mid‐gestation. The specimens at FS 26 (Mf7, Mf8) show similar tooth germ proportions and size to Mf6, and their germs are smaller in proportion than those in FS 21/22 (Mf2). It is important to note that tooth germs do not steeply decrease in absolute size, but rather stop growing, becoming smaller in proportion due to the growth of the skull and the rest of the fetus. Overall, these observations seem to confirm the results of Ishikawa & Amasaki (1995), who delimited the end of tooth formation and the start of the degradation stage in a specimen close in absolute age to Mf4 and noted high levels of degradation in specimens comparable to Mf5 and Mf6 (Table 4).

Table 5.

Measurements of tooth germs (tg) size relative to skull and TL in minke whale prenatal specimens

Specimen code	Upper jaw tg diameter (mm)	Distance from tip of snout to eye (mm)	Relative size of tg/head (%)	TL (mm)	Relative size of tg/TL (%)
Mf2	1.8	75	2.40	410	0.44
Mf4	2.6	150	1.73	740	0.35
Mf5	2.6	200	1.30	1100	0.24
Mf6	4.2	230	1.83	1100	0.38
Mf7	4.8	255	1.88	1150	0.42
Mf8	4.8	240	2.00	1250	0.38

Measurements and proportions given in italics might be unreliable due to poor preservation of the specimen.

Baleen rudiment formation and tooth resorption

Beginning at FS 25, a series of external anatomical modifications marks the initiation of baleen formation. First, a thick gum develops in the upper jaw on the medial side of the gum that was visible in earlier stages (Fig. 9); in FS 26 this new gum then thickens laterally, fusing to the earlier gums (Fig. 10). A deep ridge or keel forms in the center of this expanded ‘baleen gum’, but the lower jaw does not present any significant changes. The keel appears denser than the normal gum tissue, having a cartilaginous texture. This ridge is more marked posteriorly and fades away in the anterior part of the rostrum, where only a groove is present between the palate and the lip. These structures have been previously described in a single Antarctic minke whale fetus of a comparable age to Mf7 and Mf8 (145 cm TL, ca. 7 months from conception, 74% of gestation) by Sawamura (2008). By dissecting the palate, Sawamura found that the remaining tooth germs occupy the center of the keel. Medial to this, he identified a thick palatal mucosa, which presents high vascularization and innervation, and he hypothesized that this tissue was a precursor to baleen formation. He stated that this mucosa is part of the periodontal tissue that produces teeth rather than the transversal palatine fold, which produces the soft tissue that covers the hard palate. The present findings confirm that this external morphology exists at this stage of development in both minke whale species and, given the position of the tooth germs in the CT images, it is likely that they occupy the center of the keel.

Figure 9.

‘Baleen gum’ stage (FS25), represented using specimen Mf6. (A) Left lateral close‐up view of head, (B) close‐up view of gums in the upper jaw, (C) cross‐section of 3D rendering of internal morphology of head. Red dashed line in (A) represents cross‐section location and orientation, box represents magnified section of gums. Ossified skull bones in yellow, nasal bones in green, tooth germs in red, baleen rudiments in blue. Ossified elements are labeled in the figure. Abbreviations listed in Figure 3.

Figure 10.

‘Keel’ stage (FS26), represented using specimen Mf8 (ZMCU‐CN4x; TL 125 cm). (A) Left lateral close‐up view of head, (B) close‐up view of gums in the upper jaw, (C) cross‐section of 3D rendering of internal morphology of head. Red dashed line in (A) represents cross‐section location and orientation, box represents magnified section of gums. Ossified skull bones in yellow, nasal bones in green, tooth germs in red, baleen rudiments in blue. Ossified elements are labeled in the figure. Abbreviations listed in Figure 3.

Along with these changes in the external anatomy, a denser material can be isolated from the CT images medial to the tooth germs starting from specimen Mf6 (Fig. 9). It is likely not recognizable in Mf5 due to the poor preservation conditions of the specimen. This tissue does not contact the rostral bones inside the medial part of the alveolar canal. It is present mostly in the posterior part of the rostrum, where most of the foramina are located and also where the enlarged gum is visible externally. In the following stage FS 26 (Fig. 10), this material is visible along most of the upper jaw and appears, although it is denser, in the posterior portions of the alveolar canal. It is now expanded laterally and in some regions it engulfs the tooth germs. It is clearly connected to the visible dorsal and ventral foramina in the posterior parts of the maxilla, as it occupies the space between them. A highly vascularized area was also identified by Sawamura (2008) in the FS 26 specimen he dissected. Based on this evidence, and on the findings of Ishikawa & Amasaki (1995), which identified ‘baleen rudiments’ in comparable specimens (Table 4), this tissue appears to be a precursor to baleen development, although it is difficult to infer its composition without histological analysis.

Eruption of the transversal process of baleen

The last two specimens of the series, Mf9 and Mf10 (FS 27), display a thick baleen ridge erupted from the upper jaw, with the lower jaw retaining most of the same external morphology as previous stages. No tooth germs are recognizable at this stage and the alveolar canal in both jaws has significantly reduced its size. The baleen ridge is composed of a series of transversal processes, small plates oriented transversally to the main axis of the rostrum, spaced by deep grooves (Fig. 11). The baleen is recognizable in the CT scan as a structure of intermediate density between the soft tissues and the bone. This tissue does not present any apparent physical connection with the alveolar canal or other parts of the palate. This is to be expected as baleen is not directly surrounded by maxillary bones but is connected to the palate by a network of blood vessels and nerves (Ekdale et al. 2015). The transversal rows of connective tissue appear to have a medial depression, as they ‘erupted’ from the center of the gum towards the sides.

Figure 11.

‘Baleen ridge’ stage (FS27), represented by specimen Mf10. (A) Left lateral close‐up view of head, (B) close‐up view of baleen ridge, (C) cross‐section of 3D rendering of internal morphology of head. Red dashed line in (A) represents cross‐section location and orientation, box represents magnified section of baleen ridge. Ossified skull bones in yellow, nasal bones in green, baleen in blue. Ossified elements are labeled in the figure. Abbreviations listed in Figure 3.

The morphology of the baleen ridge matches the description of Sawamura (2008) of a similar fetus of Antarctic minke whale (175 cm TL, ca. 7.5 months f.c., 80% of gestation). It is also possible to confirm his observation that the baleen appears to develop from the posterior towards the anterior end of the rostrum. This conclusion is supported by the relative thickness of the baleen ridge, which is thicker posteriorly and becomes progressively less pronounced anteriorly. Additionally, the ridge also appears to be better developed anteriorly in the oldest specimen, Mf10, than in Mf9. This pattern of development is congruent with the keel and dense tissue morphology observed in the previous stages, as both those structures are more pronounced posteriorly. The overall morphology of the transversal process of baleen also matches the observations reported by Fudge et al. (2009) on a blue whale (Balaenoptera musculus) fetus of 3 m in total length (6.5 months f.c., 55% gestation; Lanzetti et al. 2018).

Discriminant analysis (DA) of 3D skull shape at different growth stages

For both the whole skull and rostrum‐only configurations, no significant differences were found between growth stages, but this is most likely due to the low sample size for each stage. The detailed results of the DA analyses can be found in Supporting Information Data S3. The dataset includes only a few representatives of each stage, making it impossible to conduct reliable statistical tests, although several important shape changes during growth can be noticed by comparing the average landmark shapes for each growth stage scaled to the adult in dorsal view.

In the whole skull configuration (Fig. 12), several major patterns can be identified. As growth progresses, the rostrum appears to become relatively longer than the braincase. It also assumes a more tapered shape by progressively exhibiting a smaller mid‐length width (landmarks 2–11). The anterior end of the nasals appears to be moving posteriorly and the anterior end of the supraoccipital shield is shifting in the opposite direction, reducing the relative distance between these two landmarks (13–14). This pattern illustrates the process of telescoping. The braincase shows an apparent width reduction but its length does not change. Therefore, it seems that the rostrum follows a different developmental trajectory compared with neurocranium, but this should be tested using other methods with a larger sample size (Schlosser & Wagner, 2004; Goswami, 2007). Finally, the postglenoid process of the squamosal moves medially as growth progresses (landmarks 15–16). This movement is part of the general tapering pattern observed in the rest of the braincase.

Figure 12.

Shape change in the whole skull of minke whales at different growth stages, visualized using the 16‐landmark configuration in dorsal view with the adult configuration as ‘target shape’. Configurations are arranged from youngest to oldest from left to right. Progressive elongation of the rostrum is marked by the black dashed lines (landmarks 1–3). Rostral tapering is highlighted by the dash‐and‐dot line (landmarks 2–11). The two black dotted and dash‐and‐dot lines indicate telescoping, as the relative posterior movement of the anterior end of the nasals (landmark 13) and the anterior movement of the supraoccipital shield (landmark 14). The star highlights the medial movement of the postglenoid process of the squamosal on the left side (landmark 15). Lines that connect neonate and adult shapes are full, as there is no visible change between these two growth stages in the shapes.

In the rostrum‐only configuration (Fig. 13), the same pattern of rostrum tapering and elongation can be detected as noted in the whole skull configuration. The nasal bones also become more elongated, especially between the early fetal and late fetal stages (landmarks 11–12), with the anterior end moving posteriorly most noticeably in the later stages of growth, between the late fetus and neonate stages.

Figure 13.

Shape change in the rostrum of minke whales at different growth stages, visualized using the 12‐landmark configuration in dorsal view with the adult configuration as ‘target shape’. Configurations are arranged from youngest to oldest from left to right. Progressive elongation of the rostrum is marked by the black dashed lines (landmarks 1–3). Rostral tapering is highlighted by the dash‐and‐dot line (halfway between landmarks 2–9 and landmarks 3–8). The two black dotted and dash‐and‐dot lines indicate telescoping, as the progressive elongation of the nasal bones (landmarks 11–12). Lines that connect neonate and adult shapes are full, as there is no visible change between these two growth stages in the shapes.

Both looking at the whole skull and at the rostrum, it appears that only minimal modifications occur between the neonate and adult stages. The changes in postnatal ontogeny might be subtle and not detectable using this dataset.

Discussion

Ossification sequence of minke whale skull compared with other cetaceans and terrestrial artiodactyls

All major skull elements have already started to ossify by Fetal Stage (FS) 22. This is in agreement with previous research on the humpback whale (Megaptera novaeangliae), in which all the dermal and endochondral bones were identified beginning at FS 21 (Ridewood, 1923; Hampe et al. 2015; Lanzetti et al. 2018). In particular, given the extent of the ossification, it appears that the dermal bones of the anterior portion of the skull – premaxilla, maxilla, frontal and dentary – start ossifying earlier than the endochondral and intramembranous elements of the neurocranium. This was also found to be the case in humpback whales, as well as in the pantropical spotted dolphin (S. attenuata, Odontoceti – Moran et al. 2011), in the sperm whale (Physeter macrocephalus, Odontoceti – Kuzmin, 1976) and in terrestrials artiodactyls such as pigs (Sus scrofa – Hodges, 1953) and sheep (Ovis ovis – Harris, 1937). This pattern of early onset of ossification of rostral bones appears to be a general feature of all mammals (Smith, 1997). The mandible is the first element to complete ossification in both terrestrial artiodactyls and cetaceans, including the minke whale, given that it is the only skull element to be completely ossified in the earliest growth stage examined in detail (FS 22).

Among the braincase elements, the nasals appear to be the elements that start ossifying last, because of their small size and shape in FS 22. In the humpback whale, nasals are present at FS 20 (Lanzetti et al. 2018), therefore they might ossify earlier in that species or they might not be recognizable in the minke whale specimens due to the quality of the CT scan. In odontocetes (pantropical spotted dolphin and sperm whale), the nasals appear in the last embryonic stages, thus slightly earlier than in mysticetes (Kuzmin, 1976; Moran et al. 2011). Although both groups of cetaceans have a much earlier onset of ossification of these bones compared with terrestrial artiodactyls, as in the pig and sheep, they first appear in advanced fetal stages (Harris, 1937; Hodges, 1953). The first endochondral element to ossify is the supraoccipital, which is probably already present at FS 20/21 in the minke whale (specimen Mf2) and occupies most of the posterior part of the skull in Mf3. In the other cetaceans examined, including the humpback whale, it starts ossifying in the embryonic stages (Kuzmin, 1976; Moran et al. 2011; Lanzetti et al. 2018). This might be the case also in the minke whale, given the extensive ossification visible in Mf3, but younger specimens should be examined to confirm this hypothesis. In sheep, the supraoccipital begins ossifying relatively early as well, just after the anterior dermal bones (Harris, 1937). In pigs, however, it only starts ossifying with the nasal bones, after the parietal and when the rostral bones are completely ossified (Hodges, 1953). Overall, the rostral dermal bones appear to have a conserved ossification timing compared with the braincase elements in all artiodactyls, including cetaceans. For the braincase, the ossification of the minke whale is very similar to that observed in the humpback and in the odontocetes for which data is available, with the supraoccipital being the earliest element to ossify. In the pantropical spotted dolphin, the early ossification of this bone has been shown to be correlated with its higher EQ (encephalization quotient – a measure of brain size relative to body size), with only humans and some primates having comparable ossification timing (Koyabu et al. 2014). Mysticetes do not share the high EQ of toothed whales (Boddy et al. 2012), though two species have now been shown to exhibit early ossification of the supraoccipital as well. Therefore, one can hypothesize that this trait was acquired in the common ancestor of Neoceti (modern toothed and baleen whales) and retained in both lineages rather than being convergently evolved in connection with encephalization.

Telescoping instead has evolved convergently in both cetacean lineages, although in odontocetes it is more extreme and it has been associated with echolocation (Miller, 1923; Armfield et al. 2011; Churchill et al. 2018). Baleen whales possess an apomorphic structure, the infraorbital process of the maxilla, which probably does not allow for the complete overlap of the maxilla on the frontal as seen in toothed whales (Deméré et al. 2005). This process is already ossified in FS 22 in the minke whale but does not suture with the frontal posteriorly even in the oldest specimen (Mf10). This pattern was already observed in the humpback whale (Lanzetti et al. 2018) and therefore it can be assumed it is shared, at least, by rorqual whales. The lack of fusion between this process and the frontal, as well as the late closure of the fontanelles in the neurocranium, which only ossify at FS 26, are probably what allows the partial telescoping observed in mysticetes, together with the anterior movement of the occipital shield. The late closure of sutures is also possibly connected to rostral elongation, but further investigation of the late stages of prenatal development of baleen and toothed whales should be done to test these hypotheses. Most available information on suture closure comes from the analysis of postnatal ontogeny, where it is known that there are major differences between the two cetacean groups. In many dolphin species, sutures between braincase elements close in the early postnatal stages, whereas rostral joints only close in the late juvenile stages (Perrin, 1975; Perrin & Heyning, 1993). In mysticetes instead, the rostral bone sutures remain open for their entire life. In the neurocranium, all sutures are closed at birth, except those between the occipital elements (supraoccipital/exoccipital, exoccipital/basioccipital, basioccipital/basisphenoid) (Walsh & Berta, 2011). Therefore, analyzing in detail the timing of suture closure prenatally would probably provide more information to understand the differences in development that influence the skull anatomy in adult mysticetes and odontocetes, given that the ossification sequence of the bone elements appears mostly conserved (Roston & Roth, 2019).

The presence of the interparietal bone in the fetuses from FS 22 to FS 25 (Mf3–Mf6) was confirmed in this study. In older specimens, this bone appears to have been covered by the advancing anterior end of the supraoccipital (Mf7–Mf10). This finding is in line with the observations of Eschricht (1849), who noted the presence of this element in a fetus of minke whale. This bone has also been shown to be present in the fetal stages of sei and humpback whales (Ridewood, 1923; Lanzetti et al. 2018), although postnatally it has only been observed in adult Omura's whale (Balaenoptera omurai, Wada et al. 2003).

Teeth‐to‐baleen transition in minke whales and other mysticetes

The developmental series of minke whale fetuses described here provides new and more conclusive evidence on tooth germ formation, resorption and baleen development in these species. Even if limited by the poor resolution of the CT scans, these results confirm previous observations of histological studies (Ishikawa & Amasaki, 1995; Ishikawa et al. 1999) on the presence of tooth germs in minke whale fetuses beginning by, at least, early fetal stages (FS 21/22). The dentition in the lower jaw of fetal baleen whales is usually defined in the literature as heterodont, given the presence of ‘double teeth’ that in appearance resemble multicuspid teeth of extinct stem‐Cetacea (Archaeocetes) and stem‐Odontoceti (Kükenthal, 1893; Abel, 1914; Van Dissel‐Scherft & Vervoort, 1954). The tooth germs of the upper jaw instead appear to be similar in shape along the tooth row, although it is not possible to comment confidently on the precise shape of each visible tooth germ. ‘Double teeth’ in the mandible have been observed in all other fetal balaenopterids for which data are available (e.g. Van Dissel‐Scherft & Vervoort, 1954; Karlsen, 1962) and its absence in the upper jaw has recently been linked to baleen development in the bowhead whale (Balaena mysticetus; Thewissen et al. 2017). As dentition and baleen share part of their developmental pathway at least in this species, tooth germs in the upper jaw could still be under selection, although the same constraints do not apply to the mandibular tooth germs allowing for their more variable shape and size (Thewissen et al. 2017). Gene expression and focused histological research are needed to confirm that these conclusions apply to the minke whale as well, but these species could be perfect candidates for such studies, given the present knowledge on their tooth development.

The formation of baleen rudiments as a denser tissue medial to the tooth germs is first observed in FS 25. It is linked to external anatomical changes in the gingival ridge of the upper jaw, which enlarges medially before forming a distinct keel at FS 26. This is in line with previous histological evidence presented by Ishikawa & Amasaki (1995) and by the observations reported by Sawamura (2008) in Antarctic minke whales. Slijper (1976) also noticed the presence of an elevated ridge in early to late fetal specimens of fin whale (Balaenoptera physalus) while tooth germs are still present (Karlsen, 1962). Thewissen et al. (2017) noted that baleen tissue starts developing from the medial side of the dental lamina in the bowhead whale, supporting the hypothesis that baleen indeed developed from periodontal tissue rather than from the palatal mucosa. These authors also noted an increased vascularization in the posterior portion of the palate in specimens presenting early developing baleen. In the bowhead whale, early sign of baleen development can be observed in a specimen at FS 21 (less than 2 months in absolute age). This might be an intraspecific difference or might be due to the preservation of the specimens and different study methodologies.

More data are available on the early development of baleen plates once they are visible in the gums (FS27). In both the blue (Fudge et al. 2009) and minke whales, the connective tissue plates are placed transverse to the main axis of the palate, with grooves between them. There is also a depression at the center of the ridge, which might be connected with the eruption pattern of the baleen from the gum. A major difference between the observations of Fudge et al. (2009) and those made in the minke whale in this work and by Sawamura (2008) is that in the blue whale it appears that baleen starts developing at the center of the palate, with the posterior and anterior ends of the ridge appearing less thickened, whereas in the minke whale the transversal process appears consistently more developed at the posterior end. This might be an interspecific difference, although all hypotheses should be considered with caution given the very limited sample size available. Thewissen et al. (2017) noted that the zone of baleen growth (Zone 2) expands caudally in bowhead fetuses, possibly implying that this species also has a posterior‐to‐anterior baleen development pattern, even if the adult bowhead has a very different baleen rack structure compared with rorqual whales, due to their different filter‐feeding style (Werth et al. 2018). Given the available information and the phylogenetic position of these species, with the Balaenidae representing the earliest diverging mysticete lineage and the minke whale the earliest diverging lineage of Balaenopteridae (Gatesy et al. 2013), I tentatively hypothesize that the posterior‐to‐anterior development of baleen is the ancestral condition in mysticetes, and that in the blue whale and possibly in other later diverging rorquals it has been modified, with the center of the palate now developing first. More evidence is needed from other species of balaenopterids and different families (Eschrichtiidae and Neobalaenidae) to confirm this hypothesis.

Interspecific disparities between timing of the initiation of baleen development is more complicated to interpret, given the major size differences between the species and the unreliability of the growth data that were available to calculate comparable relative age (Lanzetti et al. 2018), although it appears that in the bowhead the baleen begins developing significantly earlier, with a fetus of 159 cm TL (6.5 months f.c., 45% gestation – Reese et al. 2001), already showing well developed bristles (Thewissen et al. 2017), not just the ridge observed in the minke whale. Also, in the blue whale, the transversal process appears earlier (55% gestation) then in the minke whale (80% gestation) but presumably later than in the bowhead (15–40% gestation). This difference might be due to the shorter gestation time of the minke whale (9.5 months Antarctic m.w. – Ivashin & Mikhalev, 1978; Masaki, 1979, 10.5 months common minke whale – Tomilin, 1967; Ivashin & Mikhalev, 1978), compared with the bowhead (14.5 months – Reese et al. 2001) and the blue whale (12 months – Tomilin, 1967). More reliable data on the gestation time and a larger sample are needed to test hypotheses on the correlation between gestation time and onset of baleen development.

Skull morphology development and implications for baleen whale diversification

By analyzing the landmark configurations of the whole skull and the rostrum in dorsal view at different ontogenetic stages (Figs 12 and 13), several major patterns stand out: rostral elongation, rostral tapering, telescoping (i.e. relative posterior movement of the nasals and anterior movement of the supraoccipital), and medial movement of the postglenoid process of the squamosal. These patterns are similar to what was observed in a similar analysis of humpback whale specimens (Lanzetti et al. 2018). Kükenthal (1914) also observed that the buccal cavity progressively elongates during growth, analyzing changes in the external morphology of humpback whale embryos and early fetuses. Another feature shared with the humpback whale is the progressive lateral curvature of the mandible: the minke whale specimens display a clear curvature starting at FS 21/22 (Mf3) and the ligamentous symphysis at the anterior end is present starting at FS 26 (Mf7, Mf8). Although the humpback whale specimens were probably too young to possess a distinguishable symphysis, the lateral curvature of the mandible was evident and close to the adult proportions at FS 25 (Lanzetti et al. 2018). Unfortunately, the mandibles are dislocated or broken in the majority of specimens, so geometric morphometrics analyses cannot be conducted on the lower jaws. In addition, the sample sizes of the present study and the humpback study are not sufficient to conduct statistical analyses on the overall skull shape development in these two species. More specimens and additional taxa should be added to this dataset to test quantitatively how these developmental patterns influenced the evolution of baleen whales (e.g. Goswami & Polly, 2010; Klingenberg & Marugán‐Lobón, 2013).

Given the current evidence, it can be hypothesized that these general skull shape growth patterns are shared by at least all Balaenopteridae, as they all share a similar filter‐feeding style (Berta et al. 2016). The other living families of Mysticeti employ alternative filter‐feeding strategies: bowhead and right whales (B. mysticetus and Eubalaena spp. – Balaenidae) and the pygmy right whale (Caperea marginata – Neobalaenidae) are skim‐feeders, as they capture plankton from the water by swimming slowly with their mouth open. Instead, the gray whale (Eschrichtius robustus – Eschrichtiidae) feeds prevalently on benthic invertebrates by filtering its prey from the sediments while lying on one side, utilizing lateral suction‐feeding (Berta et al. 2016; Werth et al. 2018). These different feeding modes have been shown to require major differences in filtration area and associated buccal size, as well as overall skull shape (Bouetel, 2005; Werth et al. 2018). As the adult mouth shape is under strong ecological and evolutionary selective pressures, several developmental processes likely played a role in the evolution of modern mysticete anatomy. A developmental process that has been shown to have driven the evolution of many mammalian linages is heterochrony, a broad concept that describes all variations in relative timing of development in different features between the ancestor and the descendant (McNamara, 1986; Smith, 1997; Klingenberg, 1998; Bininda‐Emonds et al. 2003; Goswami, 2007). It has been demonstrated that heterochrony, in the form of changes in skull shape development rate after birth, is an important driver of species divergence in many odontocete genera (Galatius, 2010; Sydney et al. 2012; del Castillo et al. 2017). Tsai & Fordyce (2014) hypothesized that this same process is connected to the diversification of modern baleen whale taxa. The authors speculated that the two main types of heterochrony, paedomorphosis and peramorphosis (McNamara, 1986), provided the basis for the characteristic skull shapes observed in two families of modern Mysticeti that occupy different feeding niches, Neobalaenidae and Balaenopteridae. They hypothesized that the pygmy right whale exhibits a paedomorphic skull shape, as the adult seems to retain many features of the juvenile morphology, whereas Balaenopteridae such as the humpback whale would display the opposite trend, peramorphosis, in which adults undergo a longer or accelerated development that allows them progressively to acquire new traits. Their study focused on a few specimens, mainly postnatal, and did not include the common ancestor of the species considered, and therefore the role of heterochrony in the diversification of baleen whales could not be assessed definitively (McNamara, 1986; Klingenberg, 1998). However, it demonstrated that major lineages of baleen whales tend to follow different developmental paths, with their distinctive skull shapes already noticeable in the fetal stages of ontogeny. Geometric morphometric studies that quantify cranial morphological changes during the development of baleen whales, such as the one presented here on the minke whale, and similar information on transformations in rostral morphology in fossil mysticetes are needed to directly test how heterochronic processes influenced the diversification of modern families into their feeding niches.

Developmental insights into tooth loss, baleen and the evolution of filter‐feeding

First appearance and specialization of baleen in stem‐mysticetes

From this detailed knowledge on the anatomical aspects of ossification sequence and shape development of the skull, and the transition from teeth to baleen in the ontogeny of minke whales, it is possible to revisit previous hypotheses on tooth loss and evolution of baleen and other traits related to filter‐feeding in mysticetes.

In minke whales, this study confirms that the resorption of tooth germs and tissue connected with baleen development are clearly present at the same time in the fetus, even if only for a limited period of time. This transitional phase also coincides with the appearance of an enlarged gum, or keel, in the posterior end of the rostrum. Though it is not possible to speculate directly on feeding adaptations of fossils using developmental data, the adult morphology of some stem‐mysticetes taxa might have mirrored this intermediate stage in the turnover between teeth and baleen. Previous studies have hypothesized that the extinct taxa, mostly belonging to the family Aetiocetidae, possessed enlarged gums, based on the presence of palatal foramina and other bony correlates (Marx et al. 2016, 2016; Peredo et al. 2017, 2018; Fordyce & Marx, 2018), whereas other researchers concluded that these structures implied the presence of baleen plates or at least some form of bristle‐like primitive baleen structures (Deméré et al. 2008; Ekdale et al. 2015). On one hand, developmental evidence points to the possibility of coexistence of proto‐baleen structures and tooth germs on both an anatomical level, as seen in this study, and on a gene expression level, as it was shown that genetic pathways commonly used for tooth growth in mammals are coopted for baleen (Thewissen et al. 2017). However, many features of the dentition of fossil aetiocetids do not resemble prenatal modern mysticetes: these taxa had heterodont teeth, in contrast to the homodont dentition found in the fetuses, and were not polyodont, retaining tooth counts similar to Archaeocetes, the ancestral group to both baleen and toothed whales, with fewer than 20 teeth per side of upper and lower jaws (Deméré & Berta, 2008; Marx et al. 2016, 2016). Most previous studies have focused on these apparent discrepancies to argue that present developmental evidence does not support the cooccurrence of teeth and baleen in Aetiocetidae (Marx et al. 2016; Peredo et al. 2017; Fordyce & Marx, 2018). Although it is beyond the scope of this paper to comment directly on the anatomy of fossil specimens, it is important to emphasize that fetal morphology does not directly relate adult anatomy, as seen in modern species that have teeth in the initial stages of development but were born with baleen plates. Therefore, it is not possible to determine whether polyodonty was present in this lineage before birth. Several known processes occurring in the ontogeny of extant taxa, such as higher levels of expression of certain growth factors (Thewissen et al. 2017) or the fusion of many single‐rooted teeth as observed in the lower jaw of the fetal specimens (e.g. Abel, 1914; Karlsen, 1962), could explain the heterodont dentition in the fossil adults, without precluding the possibility of presence of prenatal polyodonty and seemingly associated baleen tissue in adults. Moreover, prenatal polyodonty and homodonty might not be strictly necessary for baleen development and could have only been acquired after the loss of adult dentition as a developmental precursor to baleen rack formation (Thewissen et al. 2017). Aetiocetidae and similar taxa are hypothesized to only have borne proto‐baleen, primitive keratinous structures, which probably would not have presented the organized pattern seen in modern species, as also shown by the lack of other skull characters correlated with specific types of filter‐feeding (Deméré & Berta, 2008; Goldbogen & Madsen, 2018). It is also important to stress that Aetiocetidae might not represent the common ancestor of modern taxa, especially given their contentious phylogenetic position (Geisler et al. 2017; Fordyce & Marx, 2018; Gatesy et al. 2018; Tsai & Fordyce, 2018).

Considering all of these factors, it is possible to conclude that the present developmental evidence does not contradict any interpretation of Aetiocetidae feeding anatomy. The hypothesized occurrence of enlarged gums in some taxa might be correlated with the presence of some precursor of baleen, rather than be an alternative morphology (Peredo et al. 2017), with some proto‐baleen structure occurring concurrently with teeth. The reconstructed morphology of other toothed mysticete genera such as the tooth‐bearing Llanocetus (Llanocetidae – Fordyce & Marx, 2018) or the recently discovered toothless Maiabalaena (Peredo et al. 2018) is also in line with this hypothesis. Baleen precursors would have been found prevalently at the posterior end of the rostrum, where most of the palatal foramina are found in the fossil skulls, as previously hypothesized (Sawamura, 2008). The proto‐baleen structure would most likely not exhibit the specialized morphology observed in modern mysticetes, given the probable absence of growth factor expression during development, which would also cause polyodonty (Thewissen et al. 2017). Transformations in ontogeny, such as increasing rate or length of development or similar peramorphic processes, would have caused the progressive replacement of teeth with baleen in later diverging lineages, as previously proposed by Deméré & Berta (2008). This hypothesis would explain the peculiar growth pattern observed in modern mysticetes species (Lanzetti et al. 2018) and their disparate heterochronic development (Tsai & Fordyce, 2014).

The first functionally toothless stem lineage, Eomysticetidae, had been hypothesized to have borne baleen only in the posterior portion of the skull, with some specimens retaining shallow alveoli mostly at the anterior end of the rostrum (Boessenecker & Fordyce, 2015a,2015b, 2016). Teeth in Eomysticetidae are thought to be vestigial, possibly present only in early ontogeny and never fully erupted. They probably had peg‐like conical shape, based on an isolated tooth that was found near one specimen (Boessenecker & Fordyce, 2015b).

The pattern of baleen development from the posterior to the anterior end of the palate in the minke whale seems congruent with these previous observations in Eomysticetidae, especially if this is the ancestral direction of growth. The dental morphology of the fossils resembles the prenatal tooth germs of modern taxa, although the two are not directly comparable. The simplified conical shape of these vestigial teeth that might have been present along with baleen reinforces the idea of a connection between homodonty and baleen rack formation. The morphology of Eomysticetidae does not contradict interpretations of the anatomy of toothed stem mysticete groups, as these lineages were mostly coeval and might have occupied different feeding niches and evolved independently (Gatesy et al. 2013; Peredo et al. 2017; Tsai & Fordyce, 2018). These observations instead complement the hypothesis that progressive ontogenetic changes and acceleration of trait development both on a molecular and anatomic level have caused tooth loss and baleen rack formation in Mysticeti. More research needs to be conducted to test these hypotheses, but currently these early toothless taxa appear to represent a key step in the evolution of mysticetes, displaying transitional traits that bridge the gap from toothed ancestral forms and living toothless lineages that are also consistent with modern fetal morphologies.

Changes in rostral shape for filter‐feeding

Significant changes in the developmental pattern of skull shape accompanied and allowed the replacement of teeth with baleen and the transition to a filter‐feeding strategy in baleen whales. One of the major requirements for filter‐feeding, and in particular lunge‐feeding, is an enlarged buccal cavity, which is achieved by possessing a long and broad rostrum (Goldbogen et al. 2010; Werth et al. 2018). Trends such as progressive rostral elongation are well documented in the fossil record of mysticetes, and these processes are closely mirrored in prenatal ontogeny, as shown in the present study. An increase in shape development specifically in this part of the skull might have favored the evolution of their characteristic mouth shape. These peramorphic transformations would have paralleled the transition from teeth to baleen, which could also be a product of progressive acceleration of developmental changes. The magnitude of these changes in shape and how they relate to the presence of teeth and baleen should be evaluated quantitatively using large datasets and employing geometric morphometric analyses, similarly to what has been done in other groups such as birds (Bhullar et al. 2012) and crocodiles (Foth et al. 2017; Morris et al. 2019). Additionally, the development of this extremely large mouth probably heavily influences other aspects of skull development in mysticetes, and it is responsible for their unique external anatomy and increased growth rates relative to odontocetes (Armfield et al. 2011; Lanzetti et al. 2018).

Until information on the ontogeny of fossil mysticetes is available, or at least more complete ontogenetic sequences for all modern species are described, it is not possible to identify specifically how heterochronic process interacted with each other to produce the skull shape of extant baleen whales. However, by qualitatively comparing the current evidence on minke and humpback whale developmental sequences and with dolphins and terrestrial artiodactyls, it can be hypothesized that changes in the timing of suture closure between skull elements are probably the main driver of heterochronic shifts in mysticete evolution, since the ossification sequence is mostly shared with odontocetes. In particular, all mysticetes show some degree of paedomorphism, since the suture between the premaxilla and maxilla never ossify, and a flexible rostrum is needed for effective filter‐feeding (Walsh & Berta, 2011; Berta et al. 2016). Heterochronic processes have been shown to interact in complex ways at different scales in the evolution of many lineages, including toothed whales (Sydney et al. 2012; del Castillo et al. 2017), therefore the presence of individual paedomorphic traits does not exclude that a generalized peramorphic pattern of development has influenced the evolution of Mysticeti.

Conclusion

This research documents the teeth‐to‐baleen transition and associated skull changes in baleen whales. The first description of the developmental sequence of the Antarctic and common minke whales is presented, which helps illuminate the pattern of tooth resorption and baleen growth in these species, by confirming and expanding on previous data (Ishikawa & Amasaki, 1995). Tooth germs appear in the very early fetal stages; then, after mid‐gestation, a denser material, interpreted as a precursor of baleen development and referred to as ‘baleen rudiments’, starts developing medial to the resorbing tooth buds. By the start of the last one‐quarter of gestation, the transversal process of the baleen is erupted, starting at the posterior end, and no tooth germs are present (Fig. 14). This transition is matched by changes in the external appearance of the borders of the palate: first gums develop, then they enlarge medially, forming a distinct ridge at the center, at the same time as the denser tissue is visible internally, and finally only the ridge containing the baleen is visible (Fig. 14). Overall, the skull ossification pattern and shape changes are similar to those observed in the development of the humpback whale (Lanzetti et al. 2018), although no quantitative comparisons of shape were possible due to the limited sample sizes.

Figure 14.

Schematic representation of the external and internal changes during the teeth‐to‐baleen transition at different stages of development in minke whales. (A) Simplified lateral view of external morphology of the head of a minke whale fetus at different growth stages, (B) simplified cross‐section of internal morphology of the rostrum of a minke whale fetus at different growth stages. In (A), the gum region is highlighted in gray to best represent changes. In (B), ossified maxilla in gray, tooth germs in red, baleen rudiments area dashed in blue and white, baleen in blue. Dashed line in (A) indicates the location of the cross‐section in (B) for each stage. The drawings are not to scale.

The observations presented here have implications for the evolution of mysticetes as a whole, especially regarding the teeth‐to‐baleen transition in fossil lineages. Teeth and proto‐baleen could have been present in Aetiocetidae during at least some part of their ontogeny (Deméré et al. 2008) and thickened gums were probably also present at the same time (Peredo et al. 2017; Fordyce & Marx, 2018), as they appear to be correlated with baleen formation in modern fetuses. The anatomy of another lineage of extinct mysticetes, Eomysticetidae, which was previously suggested to only have borne baleen in the posterior portion of the rostrum (Boessenecker & Fordyce, 2015a,2015b, 2017), seems congruent with the posterior‐to‐anterior development of the baleen ridge observed in the minke whales. A progressive acceleration of mode and tempo of baleen development could be responsible for the progressive loss of teeth and acquisition of baleen‐aided filter‐feeding. However, all of these hypotheses are provisional as it is not possible directly to relate the morphology of modern prenatal baleen whales to the anatomy of fossil adult stem mysticetes.

Quantitative investigation of skull shape changes also shows that development potentially played a key role in the evolution of other characteristic traits of this lineage related to filter‐feeding, such as their large buccal cavity and elongated rostrum. A complex interaction of heterochronic patterns at different levels could explain the changes observed between extinct and extant taxa, and even differences among modern species (Tsai & Fordyce, 2014). This hypothesis is not directly testable until information on fossil ontogenies is available, or at least a large dataset is compiled that includes all extant families of mysticetes and closely related groups such as toothed whales and terrestrial artiodactyls, in order directly to compare ossification sequences and rates of development.

Future research should include histologic and gene expression analyses on the gum tissues of these specimens or comparable ones, to determine the composition of the denser tissue and confirm its connection with baleen development. A larger 3D skull dataset representing other species and families of modern baleen whales is also needed for quantitative analyses of shape change and to test for possible developmental drivers of evolution (e.g. heterochrony, allometry, modularity). Once such a dataset is available, it would be possible to include fossil specimens and draw more definitive conclusions about the relationship between prenatal and fossil skull morphology, although the present CT data still offers the possibility to investigate other aspects of prenatal skull ontogeny of the minke whale, such as patterns of suture ossification and soft tissue development.

Conflict of interest

The author has no conflict of interest to declare.

Author contributions

A.L. designed the study, acquired and analyzed the data, and wrote the manuscript.

Supporting information

Fig. S1. Calculated growth trajectories, showing all data points.

Fig. S2. Mf5 external and internal morphology.

Fig. S3. Mf8 external and internal morphology.

Fig. S4. Mf9 external and internal morphology.

Table S1. Detailed external measurements of prenatal specimens (Mf1‐Mf7).

Table S2. MorphoSource links for images of minke whale specimens.

Video S1. Animation of 3D model of Mf3.

Video S2. Animation of 3D model of Mf4.

Video S3. Animation of 3D model of Mf6.

Video S4. Animation of 3D model of Mf8.

Video S5. Animation of 3D model of Mf10.

Data S1. Results of ANOVA analyses on repeated landmark takes.

Data S2. Detailed descriptions of external and internal anatomy of all prenatal specimens included in the study.

Data S3. Results of DA analyses comparing skull shape at different growth stages.