Neuroanatomy and inner ear labyrinths of the narwhal, Monodon monoceros, and beluga, Delphinapterus leucas (Cetacea: Monodontidae)

Abstract

Narwhals (Monodon monoceros) and belugas (Delphinapterus leucas) are the only extant members of the Monodontidae, and are charismatic Arctic‐endemic cetaceans that are at risk from global change. Investigating the anatomy and sensory apparatuses of these animals is essential to understanding their ecology and evolution, and informs efforts for their conservation. Here, we use X‐ray CT scans to compare aspects of the endocranial and inner ear labyrinth anatomy of extant monodontids and use the overall morphology to draw larger inferences about the relationship between morphology and ecology. We show that differences in the shape of the brain, vasculature, and neural canals of both species may relate to differences in diving and other behaviors. The cochleae are similar in morphology in the two species, signifying similar hearing ranges and a close evolutionary relationship. Lastly, we compare two different methods for calculating 90var – a calculation independent of body size that is increasingly being used as a proxy for habitat preference. We show that a ‘direct’ angular measurement method shows significant differences between Arctic and other habitat preferences, but angle measurements based on planes through the semicircular canals do not, emphasizing the need for more detailed study and standardization of this measurement. This work represents the first comparative internal anatomical study of the endocranium and inner ear labyrinths of this small clade of toothed whales.

Introduction

The narwhal (Monodon monoceros Linnaeus, 1758) and beluga, or white whale (Delphinapterus leucas Pallas, 1776), are the two extant members of the Monodontidae. Both endemics to the Arctic Ocean, the elongate tusk found mostly in male M. monoceros (which may have inspired the myth of the unicorn) and the all‐white coloration of D. leucas contribute to the iconic nature of these species. Because of their unique morphologies, Arctic range and, in the case of D. leucas, routine presence in captivity, extant monodontids continue to fascinate humans, and their behavior and population ecology and genetics are relatively well studied. Anatomical descriptions representing the original geometry of the soft tissue within the skull are informative for comparative work with extinct species and contribute to a basic understanding of the biology of these animals. In fact, basic physiological traits such as myoglobin content and slow twitch muscle fibers in M. monoceros indicate that the species is unlikely to have the flexibility to respond to rapid climatic change (Williams et al. 2011). The growing importance of preserving arctic‐endemic species amid increasing concerns about climate change emphasizes the need for further understanding their biology through their anatomy. Here, we describe anatomical features of cranial and inner ear endocasts of extant monodontids as they relate to physiological and behavioral characteristics.

Currently, M. monoceros is restricted to the Atlantic portion of the Arctic Ocean, whereas D. leucas has a circumpolar distribution (Heide‐Jørgensen, 2018; O'Corry‐Crowe, 2018). The fossil record of monodontids is limited, but individuals have been found in lower latitudes, suggesting that extant species’ cold‐water adaptations and restrictions may have evolved relatively recently. The two described extinct species, Denebola brachycephala Barnes, 1984 and Bohaskaia monodontoides Vélez‐Juarbe & Pyenson, 2012; are from the Miocene Almejas Formation of Baja California, Mexico, and the Pliocene Yorktown Formation of Virginia, USA, respectively, indicating that monodontids were present at least by the late Miocene. Molecular divergence times suggest monodontids and phocoenids diverged 10.82–20.12 Ma, and D. leucas and M. monoceros diverged from each other 2.73–10.23 Ma, in agreement with the fossil evidence. Recent phylogenetic analyses using genotypic data reconstruct monodontids as sister to phocoenids (true porpoises, Phocoenidae), and subsequently sister to delphinids (oceanic dolphins, Delphinidae) (McGowen et al. 2009; Fig. 1). Phenotype‐only studies sometimes reconstruct a sister relationship of delphinids and phocoenids, with monodontids as a subsequent sister clade (e.g. see Geisler et al. 2011 and Racicot, 2018 for summaries; and Murakami et al. 2012 for a more recent analysis); thus, additional phenotypic data assessed from internal anatomy may enhance future combined and morphology‐only phylogenetic analyses. Future phylogenetic analyses may shed light on the morphological transformations required for transitions from temperate to colder water habitats.

Figure 1.

Phylogenetic relationships of the Monodontidae, modified from McGowen et al. (2009).

Both extant monodontid species exhibit unique characteristics and behaviors that suggest sensory system morphological correlates may be present in their skulls. The presence and secondary sexual characteristic of the elongate, spiraled tooth of M. monoceros is intriguing anatomically, but other aspects of its skull have received comparatively little attention. Adaptations to colder waters and an isolated lifestyle restricted to pack ice may require anatomical and sensory modifications that have yet to be described in detail. D. leucas employs behaviors such as spy hopping, tail slapping, and waving, and a wide array of facial muscular movements allowing it to alter the shape of the mouth and melon, sometimes producing bubble‐blowing displays (O'Corry‐Crowe, 2018). These behaviors and secondary sexual characteristics in monodontids potentially require more visual acuity compared with other delphinoids. Vogl & Fisher (1981a,b, 1982), using gross dissections, polyester resin injections, and histology, described the arterial retia associated with the central nervous system, the small size of the carotid canals (similar to other odontocetes), and arterial circulation of the spinal cord and brain in monodontids. This work includes some of the best photographic illustration of monodontid brains currently available in the literature (Vogl & Fisher, 1982), and suggests that diving adaptations may relate to some of the peculiarities of the vascular system of M. monoceros. The internal anatomy of the brain of D. leucas has been described (Morgane et al. 1980; Marino et al. 2001), but limited detailed comparative data have been presented of the gross external anatomy of the brain and associated neural canals, and arterial and venous supplies. Morphological correlates to behavioral traits may be inferred from detailed 3D digital cranial endocast data from our work, such as the relative importance of certain sensory pathways, which may in turn be useful in inferring aspects of the behaviors of extinct (fossil) taxa.

Digital cranial endocasts provide anatomical details not always available from extracted brains, particularly in terms of the relationship of certain canals and features with the surrounding skull, and also provide useful comparative neuroanatomical information for paleontologists. Cranial endocasts may more accurately represent the hydrostatic shape of the brain in life position (Colbert et al. 2005), and the volumes can be easily and readily obtained from digital data, usually extracted from X‐ray CT scans (Balanoff et al. 2015; Racicot, 2017). Another advantage of cranial endocasts is that they may be obtained from nearly any complete or partial skull, which is particularly important in the case of rare specimens, and in species which are protected and/or have controlled takes. Although the volumes of cranial endocasts are overestimates of actual brain volume (Colbert et al. 2005; Ridgway et al. 2017), in the case of fossils it is difficult to separate brain mass entirely from other aspects of the endocranial space. Endocranial volumes and other measurements from extant specimens, with comparisons with actual brain volumes, can inform estimates for fossil specimens, making broader comparisons more tenable.

The inner ear labyrinths and hearing abilities of D. leucas have been relatively well documented in the literature (e.g. Mooney et al. 2008; Sensor et al. 2015), making it a useful comparative sister taxon to M. monoceros. We analyze individual variation among a small sample of M. monoceros inner ear labyrinths, as these are comparatively understudied. Various measurements of the cochlea, the primary organ of hearing, correlate with hearing capabilities in mammals (e.g. West, 1985; Ketten & Wartzok, 1990), and nine specific measurements have been used recently to investigate hearing ranges across most artiodactyl clades, including cetaceans (Mourlam & Orliac, 2017). Including our data with previous work allows us to establish probable hearing ranges in M. monoceros. Environmental and behavioral preferences may also be inferred from these data (Gutstein et al. 2014; Racicot et al. 2016; Aguirre‐Fernández et al. 2017).

The utriculus, sacculus, and semicircular ducts may impart information on the habitat preferences and ecologies of mammals, as they represent the portion of the inner ear involved with balance and sense of space. The cetacean synapomorphy of reduced size relative to terrestrial mammals (Ekdale, 2013) and possible lack of comparability to terrestrial mammals (e.g. Spoor et al. 2002, 2007) have resulted in its exclusion from analyses until fairly recently (Ekdale & Racicot, 2015; Racicot et al. 2016). A method that has recently been employed for calculating rotational sensitivity as associated with locomotor behavior of mammals, 90var (Malinzak et al. 2012; Berlin et al. 2013), averages the variation or deviance from 90^° of each ipsilateral semicircular canal angle. This calculation is particularly useful because it is independent of body mass, a measure that is sometimes difficult to obtain for fossil and some extant marine mammal species. Semicircular canal deviation from 90^° is negatively correlated with vestibular sensitivity (Malinzak et al. 2012; Berlin et al. 2013). Recent studies have demonstrated the utility of 90var for inferring potential movement speeds and ecologies of some cetaceans. For example, Ekdale & Racicot (2015) found that the extinct archaeocete Zygorhiza and extant Balaena had 90var values lower than any of the other cetaceans examined, indicating higher rotational sensitivity. The authors hypothesized that reduced rotational sensitivity in baleen whales could result from the shift from raptorial, single‐prey feeding in earlier whales to slower‐moving bulk filter feeding (Ekdale & Racicot, 2015). In another study, rotational sensitivity as calculated with 90var was higher (lower 90var values) in extant phocoenid species that prefer pelagic habitats, and lower (higher 90var values) in species that prefer coastal habitats; these results were used to infer the habitat preferences of extinct phocoenid species (Racicot et al. 2016). The inferences for fossil species in that study were supported further by depositional environment based on locality and other lines of inferred behavioral evidence (e.g. Semirostrum ceruttii Racicot et al. 2014). For the present study, we build on the Racicot et al. (2016) dataset, separating extant groups into ‘coastal’ and ‘pelagic’ preferences, and include an additional preference of ‘arctic’ for the monodontids, which have a variety of habitat preferences including shallow coastal, pelagic, deep water/benthic (feeding), and movement along hydrographic fronts (between water masses in the case of D. leucas and near densely packed sea ice in the case of M. monoceros). We also test between two different measurement styles using software typically available for measurements of CT scans and endocasts, because the 90var method originally described by Malinzak et al. (2012) may not be feasible for researchers who do not have access to matlab. The matlab code from Malinzak et al. (2012) uses the original CT scan slices as proxies to obtain angular measurements, without a rendered view of the semicircular canals, which could introduce some level of error if points are placed incorrectly without reference to a 3D rendering. Recent publications on cetaceans that report 90var measurements (e.g. Ekdale & Racicot, 2015; Racicot et al. 2016; Aguirre‐Fernández et al. 2017; Costeur et al. 2018) used avizo, avizo lite, or related software to obtain angular measurements. Consequently, the methods used to take this measurement differ between research groups. The material described here therefore presents an opportunity to test whether two methods for 90var measurement produce significantly different inferences for cetacean ecology.

In this study we examine digital endocasts of the crania and the inner ear labyrinths of extant monodontids. The digital cranial endocasts inform on external neuroanatomical details that have not been addressed in the literature previously, provide an important addition to existing cetacean endocast data for a delphinoid clade, and provide endocranial volume estimations comparable with known brain masses that will be useful in broader comparisons of cetacean brain and endocranial evolution. The inner ear labyrinth investigations allow for reasonable estimation of hearing ranges in M. monoceros, for which no direct audiogram data are likely to become available, and for a detailed examination of the utility of 90var measurements from the semicircular canals as proxies for habitat preference. Perhaps most importantly, this study is the first detailed investigation of the inner ear labyrinths and cranial endocast of M. monoceros, one of the most iconic marine mammals. The Arctic‐restricted lifestyles of both extant monodontid species may be impacted by climate change (Williams et al. 2011), thus further understanding of their biology and behavior through basic anatomical data is an urgent endeavor.

Materials and methods

Specimens

The skulls and petrosals of representative individuals of D. leucas and M. monoceros were μCT and CAT scanned at various institutions (Supporting Information Table S1). The skulls were selected from the two available specimens (one from each species) located in the teaching collections at the Yale University Department of Geology and Geophysics. Odontocete ear bones (petrosals or tympanoperiotics) typically dissociate from the skull during osteological preparation or decay and fossilization, facilitating high‐resolution CT scanning and digital reconstruction of the inner ear labyrinths. In the case of the petrosals, wherever possible specimens were chosen based on availability of locality information and petrosals from the left and right sides of the same individual. Multiple individuals of each species were used for comparison and assessment of individual variation. Museum collections from which specimens were borrowed: Natural History Museum of Los Angeles County (LACM); National Museum of Nature and Science, Tokyo, Japan (NMST); National Museum of Natural History, Washington, DC (NMNH); and Yale Peabody Museum, New Haven, CT (YPM).

Scanning and segmentation

Scanning protocols varied depending on specimen type and scanner used, and full skull and petrosal scans were performed as well as ‘close‐ups’ of the endocranium and inner ear labyrinths to increase resolution for the endocasts. Digital endocasts of the endocranium and the inner ear labyrinths were extracted from the CT data of all specimens (Fig. 2) using vgstudiomax v. 2.2, following previous work (e.g. Colbert et al. 2005; Racicot & Colbert, 2013; Racicot et al. 2016). The 3D magic wand tool at varying thresholds and other region of interest (ROI) tools were used. All canals of the cranial and inner ear labyrinth endocasts were digitally ‘cut’ off as close to or at their apertures as possible using the ‘3D Polyline’ tool, unless the ‘close‐up’ CT scans cut them off before the full extent of the opening. Curved linear measurements were obtained using the ‘polyline length’ tool in vgstudiomax. Other measurements were obtained using angle and line tools in the ‘Measurement’ module of avizo v. 6 after separately loading the original CT and endocast data into the software. Volume measurements were obtained using the ‘Properties’ tool for regions of interest (ROIs), which comprise segmented subsets of the data, in vgstudiomax. Measurements using the endocasts and CT data follow previous work (Colbert et al. 2005; Racicot & Colbert, 2013; Ekdale & Racicot, 2015; Racicot et al. 2016; Mourlam & Orliac, 2017). Original CT data and 3D *.stl files of the endocasts will be made available through morphosource.org, those scanned at UTCT are available by request from UTCT and digimorph.org staff.

Figure 2.

Isosurface renderings from CT scans of example specimens used in this study showing the geometric relationships of anatomical features within the skull. Top images, from left to right: falsely colored Monodon monoceros skull (YPM MAM 013218) in left anterolateral view; skull rendered partially transparent with cranial endocast rendered at 100% opacity; digital cranial endocast with skull rendered transparent. Bottom images: Oblique anterodorsal views of a M. monoceros petrosal (NMNH 267958) from the right side of the body. From left to right: petrosal at full opacity; petrosal rendered at 50% transparency with digital endocast of the inner ear labyrinths at full opacity; digital endocast of inner ear labyrinths with petrosal rendered transparent.

Correlates for ecological and habitat preference

To test between different semicircular canal angle measurement styles used to calculate 90var (Malinzak et al. 2012), angles were measured directly upon the endocasts of the canals/vestibules (‘direct method’, Supporting Information Fig. S1a,c and Table S2), following Racicot et al. (2016) and on planes manually inserted through the canals with the ‘Oblique Slice’ module in avizo (‘plane method’, Supporting Information Fig. S1b,c and Table S3). Both methods were applied to the present digital endocast dataset of monodontids and a published dataset of the sister clade to monodontids, phocoenids (Racicot et al. 2016). Each measurement was repeated three times for each angle and averaged.

Correlates for hearing frequency

Two recent studies (Churchill et al. 2016; Mourlam & Orliac, 2017) have shown that nine measurements of the inner ear labyrinth in a variety of terrestrial, semi‐aquatic, and fully aquatic mammal groups correspond to hearing sensitivity, such that ultrasonic and infrasonic hearers show clear separation in a principal components analysis (PCA). To visualize the hearing ranges of our selected monodontid species alongside other artiodactyls, we built on these analyses by taking the same measurements as these previous authors, and added them to the PCA performed by Mourlam & Orliac (2017). The measurements were: cochlear length (Cl), secondary basilar membrane length (SBL), cochlear width (Cw), cochlear width perpendicular to Cw (W2), inter‐turn distance (ITD), spiral ganglion canal width at the first quarter‐turn (GAN), surface area of the fenestra cochleae (FC), and number of turns (T). The PCA was performed using the FactoMineR package (Le et al. 2008) in the r programming language (R Core Team, 2015). Following Mourlam & Orliac (2017), any measurements that could not be taken from the CT scans were imputed using the missMDA package (Josse & Husson, 2016), although these were very few (Supporting Information Table S4).

Cranial endocast anatomical abbreviations

CB, cerebellum; ccs, carotid canal of the sphenoid; cfo, cranial foramen ovale; cg, crista galli (impression); ch, cranial hiatus; CR, cerebrum; dss, dorsal sagittal sinus; dsss, dorsal sagittal sinus sulcus; efo, external foramen ovale; etc., ethmoidal canal; fg/oc, frontal groove/optic canal; fm, foramen magnum; fo, foramen ovale; fr, foramen rotundum; g, gyri; hyc, hypoglossal canal; hypf, cast of the hypophyseal fossa; iorm, internal ophthalmic rete mirabile; mca, middle cerebral arteries; mma, middle meningeal artery; ms, median sulcus; OFL, olfactory lobe; ORL, orbital lobe; ppv, vessels that extend into the perpendicular plate of the ethmoid; rm, rete mirabile; Sf, Sylvian fissure; sma, spinal meningeal artery/arteries; TL, temporal lobe.

Inner ear labyrinth endocast anatomical abbreviations

aa, anterior ampulla; asc, anterior semicircular canal; cc, canaliculus cochleae for membranous perilymphatic duct; co, cochlea; cr, common crus; er, elliptical recess of vestibule; fc, fenestra cochleae; fv, fenestra vestibuli; la, lateral ampulla; lsc, lateral semicircular canal; pbl, primary bony lamina; psc, posterior semicircular canal; sbl, secondary bony lamina; sr, spherical recess.

Results

Cranial endocasts

Comparisons of the M. monoceros and D. leucas endocast specimens included in our analyses are made with previous descriptions and measurements of cetacean cranial endocasts (Colbert et al. 2005; Racicot & Colbert, 2013; Racicot & Rowe, 2014), descriptions of cetacean brains (Morgane et al. 1980; Ridgway et al. 2017), and descriptions of monodontid arterial retia and related vasculature (Vogl & Fisher, 1981a,b, 1982). Previous cranial endocast descriptions are of particular interest because the most detailed relevant data comprise the sister clade to monodontids, the true porpoises or phocoenids (Racicot & Colbert, 2013; Racicot & Rowe, 2014), and a successive sister taxon, a delphinid Tursiops truncatus (Colbert et al. 2005). Standard neuroanatomical and osteological terms (sensu Mead & Fordyce, 2009) are used for the endocasts and associated canals, as well as species names for the individuals described in this study to simplify communication (Fig. 3).

Figure 3.

Labeled anterolateral views of the digital cranial endocasts of Monodon monoceros (YPM MAM 013218; left side of figure, A & C) and Delphinapterus leucas (YPM MAM 006765; right side of figure, B & D). Abbreviations listed in the text.

Overall, the surface features of the D. leucas endocast are less distinct than those of M. monoceros, possibly reflecting more extensive ossified meninges in D. leucas (Figs 4 and 5). Both skulls appear to have not entirely ossified in several places, indicating that the specimens may be juveniles or younger individuals, although no life history data were available for the specimens. In the D. leucas specimen, lack of ossification occurs around the sutures between the supraoccipital, exoccipital, and parietal on both sides of the skull. Incomplete ossification occurs in the M. monoceros specimen between the supraoccipital and exoccipital sutures, as well as on the exoccipital lateral to the dorsal condyloid fossa on both sides of the skull. The incomplete ossification in both specimens results in roughened or flat edges at the lateroposterior parts of each side of the cerebra, visible in posterior views (Fig. 5A,B), appearing similar to the posterior view of the cranial contents of a M. monoceros with no maturity information provided in Vogl & Fisher (1982). Both endocasts have a sharply inclined anterodorsal surface visible in lateral views (Fig. 5E–H), similar to those of other cetacean endocasts, but unlike extracted brains which tend to appear rounded.

Figure 4.

Labeled anterior (A,B) and ventral (C,D) views of digital cranial endocasts of Monodon monoceros (A,C) and Delphinapterus leucas (B,D). Abbreviations listed in the text.

Figure 5.

Posterior (A,B), dorsal (C,D), right lateral (E,F), and left lateral (G,H) views of digital cranial endocasts of Monodon monoceros (left side of figure) and Delphinapterus leucas (right side of figure). Abbreviations listed in the text.

The D. leucas cranial endocast volume (2458.69 cc) is approximately 18% larger than the average reported elsewhere (2086.91 g, Ridgway et al. 2017). The M. monoceros endocast volume (2550 cc) is smaller than average (2858 g, Ridgway et al. 2017) by approximately 11% (note that cc and g are used interchangeably when discussing vertebrate brain mass and volume; although these tissues do not have an exact density of 1g/cc – i.e. that of water – they typically do not deviate significantly from this value, often less than 5%). Although our sample size is too small to obtain possible correlations between skull and endocast volume measurements, linear measurements of the skulls should be useful for future reference for more taxonomically comprehensive studies (Supporting Information Table S5).

Cerebrum

As in other cetacean cranial endocasts, the separation between the left and right hemispheres (median sulcus) is relatively well demarcated by ossified falx cerebri in both specimens (Figs 3, 4, 5). The median sulcus resembles those of extant porpoises and T. truncatus and is not as deep as in S. ceruttii. Asymmetry is visible at the dorsal vertex in both specimens, with D. leucas having a taller and more conical left vertex than the smoothly rounded right hemisphere, whereas the dorsal vertices of M. monoceros appear as fairly symmetrical sharp bulges (Figs 4A and 5C).

Both endocast cerebral surfaces are relatively smooth, barring some faint impressions of gyri on the dorsal surfaces of the cerebra (primarily on the right side) of D. leucas (Figs 3D and 4B), contrasting with the highly gyrencephalic brains of both species, reflecting extensive preserved meningeal ossification. The M. monoceros frontal and orbital lobes appear more foreshortened than in D. leucas (Fig. 5E–H). In dorsal views, this manifests as a tapering, anteriorly rounded frontal lobe in M. monoceros and a more squared anterior appearance in D. leucas (Fig. 5C,D), similar to Phocoena phocoena and Phocoena spinipinnis.

The Sylvian fissure is more distinct on both sides of the cerebrum in M. monoceros than in D. leucas, in which the temporal lobes are nonetheless visible, albeit more so on the left sides (Figs 3 and 5C–H). Monodon monoceros appears to have broader, more anteriorly oriented temporal lobes (obscured slightly in dorsal views by the spinal meningeal arteries, see below) than D. leucas, which has more posteriorly oriented temporal lobes, although the left temporal lobe is directed more anteriorly than the right. Ventrally, the temporal lobes appear slightly asymmetric in each specimen, with the right side appearing larger and extending more laterally than the left. On the whole, the D. leucas cerebra appear more asymmetric than M. monoceros. (Fig. 4C,D).

Ventroanteriorly in M. monoceros a thin groove corresponding to a ridge along the surfaces of the mesethmoid and presphenoid bones separates the left and right hemispheres, similar to phocoenids. It is particularly prominent as a small, deep posteriorly directed dip, corresponding to a sharp, narrow ridge on the ethmoid, which is fused with the presphenoid, representing the crista galli (Fig. 4C). The same groove is less clear and broader in D. leucas (Fig. 4D). In M. monoceros, the thin groove separating the two cerebral hemispheres continues posteriorly toward a thin transverse trough, deeper on the left and right sides as opposed to the midline, representing the raised presphenoid and basisphenoid at the intersphenoidal synchondrosis, similar to all phocoenids previously described. The intersphenoidal synchondrosis is also distinct in D. leucas. The intersphenoidal synchondrosis trough provides a posterior border for the orbital lobes, and anterior border for the olfactory lobes, both of which are more defined in the M. monoceros endocast, and which may be obscured in part by rostral retia mirabilia, especially in D. leucas (described later; Fig. 4C,D). The right orbital lobe appears to project more anteriorly in a ventral view image of a female D. leucas shown elsewhere (Ridgway et al. 2017); it is difficult to determine whether the same is the case in our sample because the optic canal/frontal groove obscures this region slightly.

Hindbrain

Moderately distinct tentorium cerebelli differentiate the cerebellum from the cerebrum in both specimens (Figs 4C,D and 5A,B,E–H). The cerebella are relatively small compared with the cerebra, following the general condition of extant odontocetes. Neither ventral surface of the monodontid hindbrains appears to have a clearly delineated pyramidal tract or pontine impression (Fig. 4C,D), unlike some phocoenids and T. truncatus. The cerebella appear somewhat bilaterally asymmetric, with the right sides directed more ventroposteriorly than the left, visually more notable in D. leucas. The foramen magnum appears smaller and more circular in M. monoceros, whereas it is more undifferentiated in D. leucas.

Neurovascular canals and other features

In this section we describe features of the endocasts that represent casts of canals and grooves that would have conducted circulatory and neural tissue, and anything besides the cerebrum and cerebellum, including the hypophysis.

On the M. monoceros endocast, there is an interfingering linear bulge starting on the medial edge of the left cerebral hemisphere, representing the dorsal sagittal sinus sulcus. The dorsal sagittal sinus sulcus becomes confluent posterodorsally and medially with the dorsal sagittal sinus, coursing medially and towards the right cerebral hemisphere (Figs 4A and 5A,C). Small, occasionally paired canals extend dorsally from the dorsal sagittal sinus sulcus (Figs 3A,C and 4A). The dorsal sagittal sinus traverses posteriorly along the dorsomedial surface of the right cerebrum until it appears to descend deep into the endocast (Figs 4A and 5A–C). Only a faint impression of the dorsal sagittal sinus is visible in D. leucas tracing a similar route along the medial edge of the right cerebral hemisphere (Figs 4B and 5B,D). Evidence for the straight sinus is not present in either specimen owing to either less ossified or not as well preserved falx cerebri as found in phocoenids and T. truncatus. Likewise, transverse sinus canals are not prominent in either endocast, unlike many phocoenids and T. truncatus.

Spinal meningeal arteries emerge from the entrance of the foramen magnum and are expressed laterally along the surfaces of the temporal lobes. They are more prominent in M. monoceros, consisting of three to four branches that wrap ventrally towards the basicranium (Fig. 5E,G). Delphinapterus leucas has similar features, although they are more prominent along the right temporal lobe than the left (Fig. 5F,H). Middle meningeal arteries are present in both specimens; they are more defined in M. monoceros and reach or nearly reach the longitudinal fissure, with the left side appearing to enter the endocast by curving medioanteriorly towards the dorsal sagittal sinus sulcus (Figs 3,4C,D, and 5C–H). Faint impressions of middle cerebral arteries are notable in both specimens (Figs 3 and 5E–F).

Anterior to the dorsal sagittal sinus sulcus, several small canals extending dorsally are present in each specimen; these have been identified in phocoenids and T. truncatus as terminal nerve canals (CN 0) based on their location within the ethmofrontal suture (Figs 3, 4, 5). These canals are more visible in the CT scans than in the endocasts, particularly due to a small amount of smoothing of the D. leucas endocast, leading to visualization of only a portion of these small canals on the surface of the endocast. Relatively bilaterally symmetric anterolaterally directed canals representing the ethmoidal canals are present in M. monoceros (Figs 3 and 4A). The same canals are also present in D. leucas, and emerge from a more medial position than in M. monoceros (Fig 4B). Vessels that extend into the perpendicular plate of the ethmoid are present anteroventral to the terminal nerves and medial to the ethmoidal canals, more visible in the endocast of M. monoceros (Figs 3 and 4).

Delphinapterus leucas has thicker and more anteriorly oriented frontal grooves/optic canals than M. monoceros does, the latter having laterally positioned and smaller frontal grooves/optic canals (Figs 3 and 4C,D). The internal ophthalmic rete may be visible as somewhat thin channels along the frontal groove/optic canals in D. leucas (Fig. 4D). In anterior views the frontal grooves/optic canals of the right side in both specimens appear directed dorsolaterally, whereas those on the left course dorsoanteriorly (Fig. 4A,B). Small, medioanteriorly positioned, well defined, and medially directed sickle‐shaped canals are visible on the anterior surface of the orbital lobes of the D. leucas endocast (Figs 3B,D and 4B). Small canals issue anteriorly from these superficial canals, becoming confluent with the optic canals (visible in CT scans). In M. monoceros, a smaller canal that issues from a more medial position somewhat resembles these accessory canals, although it becomes confluent anterolaterally with the larger ethmoidal canal.

Posterolateral to the frontal grooves/optic canals in M. monoceros are rather large openings representing an undifferentiated confluence of the foramen rotundum, superior orbital fissure, and inferior orbital fissure (labeled ‘fr’ for foramen rotundum; Figs 3A,C and 4A,C). The alisphenoid, frontal, and orbitosphenoid constitute the borders of this opening. The majority of the participating bones have fused in D. leucas, resulting in limited openings confluent with the optic canal/foramen rotundum, representing the orbital fissure. Posterior to the optic canals or orbital fissures/foramina rotunda lie the canals leading to the foramina ovalia, which are thicker (the right side anteroposteriorly thicker but dorsoventrally thin), less defined, and more anteriorly directed in D. leucas than in M. monoceros. The foramen ovale on the left side of the D. leucas specimen is broader and appears confluent with other canals, possibly due to lacking development or a missing part of the tentorium cerebelli on this side. The canals leading to the foramina ovalia in M. monoceros are well defined, widening proximally/deeply, with medial notches representing small eminences on the lateral edges of the basisphenoid. The foramina ovalia in M. monoceros branch superficially/distally, with one smaller canal positioned dorsally; this condition is more obvious on the left side. The initial ramus and larger branch likely constitute the cranial foramen ovale, marking the exit of the mandibular nerve from the brain case, and the smaller canal is likely to be the external foramen ovale, which represents the exit of the mandibular nerve toward the mandible, sensu Mead & Fordyce (2009). In both specimens the carotid canals of the sphenoid, located just medial to the foramina ovalia, are thin and long, similar to T. truncatus and phocoenids. Between the carotid canals lies a fairly undifferentiated surface with an anterior and ventrally rounded left half that would have contained the hypophysis and likely portions of the carotid retia mirabilia in M. monoceros. In D. leucas this area appears more elliptical and peanut‐shaped with bilateral domes, resembling the same region in Phocoenoides dalli. Recent work has variably identified the retia in this area as the internal ophthalmic rete, from which it may originate, or simply rete mirabile. The general area, at least the portion laterally surrounding the carotid canals of the sphenoid, is likely to be the carotid rete mirabile sensu Vogl & Fisher (1982). It is not possible to identify exactly which portion of the retia these are using only osteological material, thus we continue to refer to them as simply the retia mirabilia with prominences representing the hypophyses. Anterior to this region and at the midline of the M. monoceros endocast lies a small canal extending ventrally into the basisphenoid that is not paired and does not exit the skull.

On the anteroventral surface of the hindbrain, just posterior to the hypophysis/carotid rete mirabile in M. monoceros, two spherical bulges, the left one having three to four possible circulatory branches extending posteriorly from it, are visible (Figs 3A and 4C). Similarly, in D. leucas, a small bulge appears just posterior to the left side of the hypophysis, although no branching is visible (Fig. 4D). Both are positioned on the basioccipital (which is fused with the basisphenoid). Presumably these are expressions on the hindbrain of the retia mirabilia. Lateral to these are larger bulges in D. leucas, possibly representing aspects of the cranial or carotid rete mirabile, similar to those shown in Vogl & Fisher (1982), whereas in M. monoceros the same area is flattened and appears to form partial canals that become confluent with other canals leading to the foramina ovalia.

Similar to other odontocetes, cranial hiati are present on the endocast which house the peribullary sinus, petrosal, and tympanic bulla or ‘tympanoperiotic’ bones, and the facial and vestibulocochlear nerves. The region is contained within a confluence of the jugular foramen, posterior lacerate foramen, internal acoustic meatus, and sometimes (although not in the case of these endocasts) the foramen ovale. The cranial hiati are positioned anteroventral to the cerebella and posterior to most other cranial nerves and openings. A lateroposterior notch along the length of the cranial hiatus, most differentiated on the left side of each specimen, outlines the ventral surface of the jugular foramen (Figs 3 and 5E–H). Endocranial grooves for the vestibulocochlear nerves are visible on dorsolateral surfaces of the cranial hiati (Figs 3 and 4). D. leucas appears to have a more anteriorly directed left cranial hiatus compared with M. monoceros, which has more ventrally directed cranial hiati on each side (Fig. 5E,F). Posterior to the cranial hiati are the small hypoglossal canals, which appear more anteriorly directed in D. leucas and more ventrally directed in M. monoceros. Small canals housed within the exoccipital posterior to the hypoglossal canals are visible on the endocast of D. leucas (Figs 3, 4, 5); similar canals are present in M. monoceros scan slices but were not digitally isolated for the endocast.

Inner ear labyrinths

General observations

The overall morphology of D. leucas specimens resembles previous data (e.g. Sensor et al. 2015; Figs 6 and 7). The canaliculus cochleae for the cochlear aqueduct is transversely thinner along its length in all specimens of D. leucas compared with any M. monoceros specimen (e.g. 2.01 mm at aperture cutoff for specimen NMNH 267958), with NMNH 572014 having an exceptionally thin canal (0.49 mm at aperture cutoff). The canals are also shorter in D. leucas (e.g. 6.55 mm for specimen NMNH 572014) than M. monoceros (e.g. 8.39 mm for specimen NMNH 267958). Roughened edges, likely representing extensive vascular supply, are visible along more of the canaliculus cochleae length in M. monoceros than in D. leucas, which appears mostly smooth except towards the aperture (Figs 7, 8, 9).

Figure 6.

Labeled digital endocast in anterior (top) and dorsal (bottom) views of the inner ear labyrinths of Monodon monoceras (NMST 267958) to facilitate comparisons across individuals and species. Abbreviations listed in the text.

Figure 7.

Digital endocasts of inner ear labyrinths of Delphinapterus leucas specimens from the right side of the body in four standard views.

Figure 8.

Digital endocasts of the inner ear labyrinths of Delphinapterus leucas and Monodon monoceros specimens from the left side of the body in four standard views.

Figure 9.

Digital endocasts of the inner ear labyrinths of Monodon monoceros specimens from the right side of the body in four standard views.

The number of cochlear turns for each specimen does not vary greatly, between 1.89 and 2.12, with D. leucas having the largest number of turns. The cochlear canals overall appear transversely and dorsoventrally thinner in D. leucas specimens than in M. monoceros. Monodon monoceros has more dorsally directed spiral laminae towards the second quarter turns, particularly noticeable in both the left and right sides of specimen LACM 72174 (Figs 8 and 9). Although not reflected in the inter‐turn distance at the first quarter turn measurements from our study (Table S4), visually D. leucas cochleae appear less ‘tightly’ coiled than M. monoceros (Figs 7, 8, 9). Previous measurements from Mourlam & Orliac (2017) better reflect this observed difference (Table S4).

The elliptical recess of the vestibule is thinner in NMNH 57204 than in any other individuals; otherwise it varies slightly in size and shape among all specimens. The aqueductus vestibuli is larger, has a broader opening, and more extensive roughened edges in M. monoceros than in D. leucas, although the overall conical shape and extremely small canal leading to a cone‐shaped aperture are similar to other cetaceans (e.g. Ekdale & Racicot, 2015).

PCA of cochlear measurements

The PCA results comparing cochlear measurements (Fig. 10A) illustrate that the spatial relationships described by Mourlam & Orliac (2017) are not significantly altered by the addition of monodontids from this study. The inner ear labyrinths of monodontids occupy a tightly constrained area nested within the odontocetes, showing clear separation from odontocetes that use narrow band high‐frequency (‘NBHF’) sounds, mysticetes, and terrestrial mammals. The variables factor map (inset in Fig. 10A) illustrates that most of the variation among monodontid inner ear labyrinths is described by aspects of the cochlear width (Cw and W2). The inner ear labyrinths from the left and right sides of the body in the same individuals generally plot close to one another. All samples from this study plot between the M. monoceros and D. leucas from the original dataset (Mourlam & Orliac, 2017) but are closer to, and share morphospace with, Acrophyseter sp. (Galatius et al. unpubl. obs.).

Figure 10.

Principle components analysis (PCA) of nine cochlear measurements including all monodontid individuals from this study (A), with variables factor map inset, and a close‐up of the region of the PCA including monodontids (B). Dataset modified from previous work (Churchill et al. 2016; Mourlam & Orliac, 2017) to include two Delphinapterus leucas individuals (n = 3) and five Monodon monoceros individuals (n = 9), both including petrosals from the left and right sides of the body, when available, to investigate variation.

Comparison of 90var measurements

We compared two methods for obtaining 90var measurements – the ‘direct’ method (Racicot et al. 2016; Fig. S1a,c) and the ‘plane’ method (Fig. S1b,c) – by splitting a dataset including phocoenids (Racicot et al. 2016) and monodontids into coastal, pelagic, and arctic biogeographic preferences, and statistically comparing the distributions (Fig. 11). Comparisons between binned measurements were performed using Kolmogorov–Smirnov tests; these provide a non‐parametric test for the hypothesis that two sets of measurements were drawn from the same distribution. The results illustrate large differences between the two methods. The ‘direct’ method recovers a significant difference between coastal and pelagic species at the 95% confidence interval. Coastal and arctic species also have highly dissimilar distributions, but this is significant only at the 90% rather than 95% confidence interval. In stark contrast, the ‘plane’ method does not recover any significant differences between any of the three groups.

Figure 11.

Distributions of 90var measurements taken two different ways – the ‘direct’ method (of Racicot et al. 2016), and the ‘plane’ method (cf. Malinzak et al. 2012) – split between coastal, pelagic, and arctic‐type habitat preferences. Distributions are compared using two‐sample Kolmogorov–Smirnov tests. The ‘direct’ method recovers significant differences between coastal and pelagic species, whereas the ‘plane’ method does not.

Discussion

Digital endocasts of the cranium and inner ear labyrinths yield valuable information pertaining to the physiology and evolution of vertebrates. These data are becoming more common in the literature, as the utility of endocasts in understanding sensory system evolution has been long recognized (e.g. Edinger, 1955; Luo & Eastman, 1995) and access to CT scanning facilities is becoming more widespread (Racicot, 2017). Although occasionally only single specimens are available for study, especially in the case of fossil specimens, X‐ray CT allows for extensive and detailed examination of these specimens, exponentially increasing the natural history data associated with them. Using CT effectively and efficiently to evaluate individual variation in endocranial spaces is among the next frontiers in morphological and anatomical studies (e.g. Macrini et al. 2007a,b; Macrini, 2009; Ekdale, 2010, 2011; Ahrens, 2014). Establishing levels of variation and areas of morphology more prone to variation in populations informs on the functional importance (or lack thereof) of certain features (e.g. Billet et al. 2012), and polymorphic characters may provide phylogenetic data (Weins, 1995). In this study, we used X‐ray CT to describe cranial endocasts and inner ear labyrinths of monodontids, a small cetacean clade with two iconic extant members. This approach provides quantitative data and descriptions that can be incorporated into broader‐scale studies and standalone insights about these charismatic species.

Cranial endocasts

The smaller volume of the M. monoceros specimen of this study compared with the average from Ridgway et al. (2017) may be related to the specimen representing a juvenile or younger individual, given its two incisors and small size. Reasons for a smaller overall endocast volume than average are difficult to assess fully because no further details about the M. monoceros specimens are provided in the Ridgway et al. (2017) study. Endocasts are expected to have larger overall volume estimates than measurements of extracted brains because of the existence of extraneural tissues in the endocranium (e.g. Colbert et al. 2005; Ridgway et al. 2017), thus the larger volume we find for D. leucas is expected. However, the specimen is likely to be a younger individual based on the lack of ossification in some parts of the skull. The relation between endocast volumes and brain volumes should be investigated further and in greater detail, as brain volume is commonly used as part of brain‐body‐size estimates including encephalization quotients (Jerison, 1973), and endocasts are sometimes the best or only source of data available, especially for extinct groups. To our knowledge, no studies have reported estimates of the amount of each type of extraneural components contained within the skull for different cetacean clades, making estimations of actual brain volume from endocasts difficult. When comparing lineages across deep time, in the case of fossils and rare species for which extracted brains are not available, the endocranial cavity may be our only window into the overall shape and mass of the brain, thereby making comparisons with endocasts of modern species crucial and quite relevant to answering questions about sensory system evolution in cetaceans. Further, digital endocast volumes are readily measurable at any time once they have been extracted. Although remaining aware of the possible discrepancy between actual brain volume and total endocranial volume, we nonetheless note that endocast shape and volumes on their own provide valuable information in terms of how brain shape is reflected by endocranial shape. Total endocast volumes of extant taxa, when incorporated into larger datasets including deep time components of extinct or rare cetacean species, provide more comparable information than extracted brain volumes from extant animals alone. A useful step forward would be a comprehensive dataset of average cetacean endocast volumes to match that of existing brain mass averages, which would allow for reasonable predictions of brain mass using endocasts. A predictive equation could then be made for brain volumes in fossils using known extant extraneural endocranial volumes.

The less defined endocast shape in D. leucas might relate to more highly ossified meninges, which may aid in keeping the brain more solidly in place during daily activities. Unfused cervical vertebrae in D. leucas allow for more lateral head and neck flexibility, which may provide the flexibility to move slowly in pursuit of prey in shallow waters 1–3 m deep (O'Corry‐Crowe, 2018). D. leucas have great variety in the range of habitats they explore, some with seasonal migration patterns, and behaviors they exhibit in search of food and molting grounds, including benthic and near‐shore feeding as well as following tidal glacial fronts (O'Corry‐Crowe, 2018). M. monoceros also have unfused cervical vertebrae, and while one might expect a need for highly ossified meninges in males with tusks, M. monoceros may nonetheless be more restricted in head and neck flexibility. The specimen used in this study is also potentially a juvenile and younger than the D. leucas specimen, which implies that less ossification of meninges might be likely. Behavioral preferences in M. monoceros regardless of tusk presence is primarily associated with long periods of isolation in high‐density pack ice (Heide‐Jørgensen, 2018), which may require less variety and flexibility in head movement.

Bilateral asymmetry in the bulk of the endocasts of both species may reflect skull torsion, as reported elsewhere (Fahlke et al. 2011; Racicot & Colbert, 2013), which in turn may impact biosonar abilities by facilitating bidirectional hearing. Asymmetry in the narwhal in particular may correlate with having a single large tooth, or one tooth smaller than the other, as in the case of the specimen used in this study.

The presence of an intersphenoidal synchondrosis groove in monodontids and phocoenids may either further reflect that specimens in this study were juveniles or, alternatively, that this groove is a useful morphological character for the clade including monodontids and phocoenids. Phocoenids possess many paedomorphic characters (e.g. Galatius et al. 2011), and the intersphenoidal synchondrosis suture and corresponding groove on the endocast may be yet another indicator of paedomorphic characteristics expressed in the skull. Many other skull bones are not fully ossified or fused in phocoenids as compared with some delphinids, so this is not unexpected.

With Cranial Nerve 0 present in monodontids as well as phocoenids (Racicot & Colbert, 2013; Racicot & Rowe, 2014) and T. truncatus (Colbert et al. 2005; Ridgway, 1987; Ridgway, 1990), our study has increased the number of recognized cetacean taxa with remnants of the vomeronasal system. Although certainly reduced, the vomeronasal and olfactory systems in cetaceans may still be used. A recent study reported that the cervical gland of the pygmy sperm whale, Kogia breviceps de Blainville, 1838 may be used to excrete water‐soluble peptides (Keenan‐Bateman et al. 2017). The system by which these peptides may be ‘smelled’ or otherwise sensed by conspecifics or other cetaceans is unclear. Although this type of organ is only described in kogiids, additional investigation into olfactory and chemoreception is broadly important, as intuitive assumptions about the loss of ‘unnecessary’ sensory apparatus and/or organs in whales because of their aquatic lifestyles may not be fully supported.

The thicker and more anteriorly oriented frontal grooves/optic canals in D. leucas imply good binocular vision, allowing for better depth perception than narwhals. The sickle‐shaped canals on the surface of the orbital lobes that extend anteriorly and become confluent with the optic canal may provide further innervation for optic sensory modalities. Further, D. leucas are able to move their facial muscles such that they can change the shape of the prominent melon and create a variety of facial expressions with their mouths (O'Corry‐Crowe, 2018). These behaviors along with spy hopping and other displays may require more dependence on visual signaling than some other cetaceans, while also requiring more intensely modified, larger musculature and optic nerve canals than in M. monoceros. More laterally oriented and smaller optic canals in M. monoceros may relate to reduced importance of vision and better monocular vision, allowing for a larger field of view, while also accommodating the presence of large tusks. Monocular vision may be more useful for preferred wintering areas with heavily consolidated pack ice, sometimes with only 3% open water (Heide‐Jørgensen, 2018). Monocular vision in conjunction with echolocation would more readily provide visibility of open water channels and breathing holes or cracks. The tusk of M. monoceros, although clearly a secondary sexual characteristic, has been found to be highly sensitive to changes in salinity and water temperature, and may therefore be used as an additional sensory apparatus (Nweeia et al. 2014). Dissections showed a direct connection with the maxillary division of the trigeminal nerve (CNV) (Nweeia et al. 2014). Correspondingly, larger trigeminal nuclei or maxillary portion of the trigeminal nerves may be expected on the endocasts of male M. monoceros or other individuals possessing tusks, as has been found in other mammals, such as the duck‐billed platypus and its extinct relatives (Macrini et al. 2006), which have increased sensory perception in the rostra. To our knowledge, whether M. monoceros males specifically have larger maxillary branches of the trigeminal nerves than other odontocetes has not been detailed and cannot be specifically identified with our data. The large foramina rotunda could indicate larger maxillary branches of the trigeminal nerves than in D. leucas, but could also be related to lack of ossification because of the young age of the individual. Individual variation in the presence of tusks exists and therefore speculating on alternative functions should be done with restraint. Because the tusk is found primarily in males, and females and neonates survive without it, it is not likely to be critical as an additional sensory apparatus. Further investigation of the endocasts and cranial nerves of males, females, and neonates will shed light on whether monocular vision is a key feature of M. monoceros associated with their habitat preferences or, the simpler explanation, that it helps to accommodate the presence of a tusk, or a combination of factors.

Inner ear labyrinths

The canaliculus cochleae for the cochlear aqueducts in M. monoceros are longer and appear more highly vascularized based on visibly roughened edges in the endocasts. The smaller and thinner size of the same canal in D. leucas was also visible in a previous study (Sensor et al. 2015). These differences may be useful phylogenetic differences to note for future morphological phylogenetic studies. They may also reflect differing degrees of change to the vascularization of the endocranium in each species. M. monoceros in particular seems to prefer deeper dives for longer intervals than are recorded for D. leucas, and is one of the deepest diving marine mammals with dives sometimes exceeding 1500 m (Heide‐Jørgensen, 2018). Deeper diving behavior and preference for pack ice during wintering months may require more extensive vascularization of the endocranium and inner ears; this may be an avenue of study requiring further detailed comparative soft tissue analysis.

The similar overall morphology and cochlear metrics as reflected in the PCA analysis allow us to infer similar hearing frequency ranges in M. monoceros and D. leucas. D. leucas audiograms show peak hearing frequencies at ~ 15–90 kHz (Mooney et al. 2012), and they have been observed to have a wide repertoire of whistles and pulsed calls in the 0.1–12 kHz range (O'Corry‐Crowe, 2018). Narwhal observations include whistles or pure tones from 18 to 300 kHz (Rasmussen et al. 2015), further supporting at least an overlap in hearing ranges.

Comparisons of the semicircular canal orientations as calculated from the 90var measurements show expected significant differences in habitat preference using the direct measurement method, but no significant differences were found using the plane measurement method. Although we split species into arctic, coastal, and pelagic preferences, we recognize that M. monoceros typically prefers open ocean and densely packed sea ice areas, whereas D. leucas varies in preference depending on time of year, maneuvering in extremely shallow coastal areas to open ocean areas. These variable preferences are the reason for choosing arctic over coastal or pelagic when comparing differences in semicircular canal measurements. Although the sample sizes for some of the comparisons used here are small, our results illustrate that care should be taken in how the ipsilateral semicircular canal angles are measured, as in this case study the two methods give very different results. The ‘direct’ method (Racicot et al. 2016) produces significant differences between pelagic and coastal taxa. The ‘plane’ method does not recover these same differences. We currently do not favor using one method over the other, but our results may highlight the need for a larger and more focused study on the semicircular canals of cetaceans, perhaps also including use of the matlab code from Malinzak et al. (2012) to obtain the semicircular canal measurements. The ‘plane’ measurements are closer to the originally described method of Malinzak et al. (2012), and therefore may be truly identifying a lack of informative environmental preference signal in odontocetes. Ecologies may be inferred from other aspects of the external and inner ear; Gutstein et al. (2014) found a correspondence in attributes of the cochlea (internally) and external pars cochlearis with environmental preference. Similarly, a study using 3D geometric morphometrics of odontocete inner ear labyrinths found that the semicircular canals are not statistically significantly informative for ecological preference, whereas the cochlea alone is a decent proxy for habitat (Costeur et al. 2018). Riverine, coastal, and arctic environments all have different requirements for sound production and reception because of differences in spatial dimensions including pack ice; as Costeur et al. (2018) found, the cochlea may be informative for certain aspects of sensory requirements in certain habitats.

In summary, we used CT scans of skulls and petrosals of the two extant monodontid species, M. monoceros and D. leucas, to identify internal anatomic characteristics that may indicate known environmental and behavioral differences, and characters that may be useful for future morphological or combined phenotype and genotype phylogenetic analyses. The neuroanatomic data show volumetric differences from brain‐only volumes that point to the need for additional study and comparison of brain volumes with endocast volumes. Differences in the shape of the optic canals/frontal grooves of each specimen may reflect their different behavioral and environmental preferences for isolated, high‐density pack ice in the case of M. monoceros and possibly more versatile and optically driven behaviors of D. leucas. The inner ear labyrinth results show that the two species are likely to have similar hearing frequency ranges, and that there may be environmentally driven differences reflected by semicircular canal angle measurements. Further examination of these measurements is needed, however, but cochlear measurements independent of the semicircular canals may be useful proxies for environmental preference. The data presented will be useful for comparisons with fossil specimens and for inclusion in broad investigations, including fossils, of cetacean sensory system evolution.

Author contributions

R.A.R. conceived the study, performed CT data analysis, and wrote the first draft of the paper. All authors facilitated CT scanning of specimens, performed statistical and other analyses, and contributed to critical review and writing the manuscript.

Supporting information

Fig. S1. Examples of angular measurements taken in avizo on Monodon monoceros (LACM 72548, left side).

Table S1. Museum number, specimen information, and relevant scanning parameters and locations for specimens used in this study.

Table S2. Semicircular canal angle measurements used to calculate 90var using the ‘direct’ method.

Table S3. Semicircular canal angle measurements used to calculate 90var using the ‘plane’ method.

Table S4. Data (measurements) used for principle components analysis, modified from Mourlam & Orliac (2017) by the addition of monodontid species from this study and additional species from a narrow‐band high‐frequency study.

Table S5. Several standard skull measurements taken from full CT scans of the Delphinapterus leucas and Monodon monoceros skulls.