Rostral Densification in Beaked Whales: Diverse Processes for A

C. R. Palevol 10 (2011) 453–468

Contents lists available at ScienceDirect Comptes Rendus Palevol

www.sciencedirect.com

General palaeontology, systematics and evolution (Vertebrate palaeontology) Rostral densiﬁcation in beaked whales: Diverse processes for a similar pattern

La densiﬁcation du rostre des baleines à bec : des processus variés pour un résultat similaire Olivier Lambert a,∗,b, Vivian de Buffrénil a, Christian de Muizon a a Département histoire de la terre, Muséum national d’histoire naturelle, UMR 7207 (paléobiodiversité et paléoenvironnements), (MNHN, CNRS, UPMC), CP 38, 57, rue Cuvier, 75005 Paris, France b Département de paléontologie, institut royal des sciences naturelles de Belgique, Bruxelles, Belgium article info abstract

Article history: As compared to other odontocetes (toothed whales), the rostrum of beaked whales (family Received 24 September 2010 Ziphiidae) often displays extensive changes in the shape, thickness, and density of its con- Accepted after revision 21 March 2011 stituent bones. Previous morphological observations suggested that these modiﬁcations Available online 31 May 2011 appeared in parallel in different ziphiid lineages. However, very few data were available on the compactness and histology of these rostral bones, which precluded the study of the pro- Written on invitation of the Editorial Board. cesses at work for the development of such structures, as well as the interpretation of their functional implications. In this work we review the bibliographic data on the anatomy of the Keywords: ziphiid rostrum and we add new observations on adults of several extinct and extant taxa. Odontoceti Ziphiidae These observations are based on CT scans and transverse histological sections. Our results Beaked whales conﬁrm that different bones (vomer, mesethmoid, premaxilla, maxilla) are involved in the Pachyostosis various morphologies displayed by ziphiid rostra. Strong density contrasts are detected Osteosclerosis between bones and/or inside the bones themselves; for example, parts of the rostrum Computed tomography reach densities in the range of Neoceti ear bones, which are among the densest bones Bone histology known hitherto. Furthermore, the histology of the pachyostotic and osteosclerotic bones proves to change from one taxon to the other; the degree of Haversian remodeling varies strongly between species: it can be absent (e.g. Aporotus recurvirostris), partial (e.g. aff. Ziphi- rostrum), or complete (e.g., Mesoplodon densirostris). The atypical secondary osteons known to be responsible for bone hypermineralization in the rostrum of M. densirostris occurred also in Choneziphius sp. Confronted with a phylogenetic framework, these anatomical and histological observations indicate that the acquisition of compact (osteosclerotic) and/or swollen (pachyostotic) bone is the result of a broad convergence between taxa, in response to common selective pressures. The functional dimension of this question is discussed with respect to what is known about extant ziphiid ecology. © 2011 Académie des sciences. Published by Elsevier Masson SAS. All rights reserved. résumé

Mots clés : Comparé aux autres odontocètes (cétacés à dents), le rostre des baleines à bec (famille des Odontoceti Ziphiidae) est souvent sujet à d’importantes modiﬁcations de la forme, de l’épaisseur, et Ziphiidae de la densité des os qui le constituent. Des observations préliminaires ont suggéré que Baleines à bec ces modiﬁcations se sont développées en parallèle dans plusieurs lignées de ziphiidés. Pachyostose Cependant, très peu de données étaient disponibles sur la compacité et l’histologie des os du

∗ Corresponding author. E-mail address: [email protected] (O. Lambert).

1631-0683/$ – see front matter © 2011 Académie des sciences. Published by Elsevier Masson SAS. All rights reserved. doi:10.1016/j.crpv.2011.03.012 454 O. Lambert et al. / C. R. Palevol 10 (2011) 453–468

Ostéosclérose rostre, ce qui a empêché l’étude des processus à l’œuvre dans le développement de Tomographie informatisée telles structures, ainsi que l’interprétation des implications fonctionnelles de celles-ci. Histologie osseuse Dans ce travail, nous passons en revue les données bibliographiques sur l’anatomie du rostre des ziphiidés et ajoutons des observations nouvelles sur les adultes de plusieurs taxons éteints et actuels. Ces observations ont été collectées à l’aide de deux moyens : la tomographie informatisée et la confection de lames minces transversales. Nos résul- tats confirment que des os différents (vomer, mésethmoïde, prémaxillaire, maxillaire) sont impliqués dans les changements morphologiques variés, observés dans différents taxons de ziphiidés. De forts contrastes de densité sont également détectés entre les os et/ou au sein des os eux-mêmes ; certaines parties du rostre atteignent, par exem- ple, des densités similaires à celle des os de l’oreille des Neoceti, qui figurent parmi les plus denses connus. De plus, l’histologie des os pachyostotiques et ostéosclérotiques s’est avérée changer d’un taxon à l’autre ; le degré du remaniement haversien varie fortement entre les espèces : il peut être nul (e.g., Aporotus recurvirostris), partiel (e.g., aff. Ziphi- rostrum), ou complet (e.g., Mesoplodon densirostris). Des ostéones secondaires atypiques, connus comme étant responsables d’une hyperminéralisation osseuse dans le rostre de M. densirostris, sont également détectés dans Choneziphius sp. Confrontées à un cadre phy- logénétique, ces observations anatomiques et histologiques indiquent que l’acquisition d’os compacts (ostéosclérotiques) et/ou renflés (pachyostotiques) est le résultat d’une convergence entre taxons, en réponse à des pressions de sélection communes. L’aspect fonctionnel de cette question est discuté, en rapport avec ce qui est connu de l’écologie des ziphiidés actuels. © 2011 Académie des sciences. Publié par Elsevier Masson SAS. Tous droits réservés.

1. Introduction 1997; Zylberberg et al., 1998). In several extinct taxa, the mesorostral groove is dorsally closed by the thickened and Beaked whales (Ziphiidae) form a diversified fam- compact premaxillae, which, in some cases, form elevated ily of odontocetes (toothed whales), including 21 extant longitudinal crests (Lambert, 2005), whereas large rostral species and six genera. Predominantly teuthophagous, maxillary crests are observed at the rostrum base of the most ziphiids are considered as deep divers, with dive extant Hyperoodon spp. At least in Hyperoodon ampullatus records at more than 1800 m for some species (Hooker the development of these crests is a sexually dimorphic and Baird, 1999; Schreer and Kovacs, 1997; Tyack et al., character (Mead, 1989b). 2006), depths at which prey are detected by echolocation The preliminary observation of the diverse morpholog- (Johnson et al., 2004). With exception for Tasmace- ical peculiarities of ziphiid rostra leads to the assumption tus shepherdi, all extant ziphiids display an important that pachyosteosclerosis, the combination of pachyostosis, reduction of their dentition related to suction feeding or swollen, protuberant periosteal cortices, and osteoscle- (Heyning and Mead, 1996; Werth, 2006): only one or rosis, an increase in bone inner compactness (de Buffrénil two pairs of mandibular teeth are retained, modified in and Rage, 1993; de Ricqlès and de Buffrénil, 2001), was a tusks that generally erupt in adult males only. Dental common feature in ziphiids, occurring independently in reduction is similarly observed in several of the extinct multiple lineages (Bianucci et al., 2007; Lambert, 2005). species known hitherto, whereas others (e.g., Messapicetus However, this assumption was based on a superficial obser- spp., Ninoziphius platyrostris) retain numerous homod- vation of the skulls, lacking a detailed analysis of their inner ont rostral and mandibular teeth (Bianucci et al., 2007, organization and histological features. Furthermore, vari- 2010; de Muizon, 1984; Lambert, 2005; Lambert et al., ous potential function(s) of a pachyosteosclerotic rostrum 2009). in ziphiids have been proposed in the past, in relation- In addition, the ziphiid skull, and specially its rostral ship with three different ecological or behavioral traits of region, is often modified, as compared to the typical con- extant taxa: echolocation, deep diving, and intraspecific dition of other odontocetes (Bianucci et al., 2007; Heyning, fights between males (de Buffrénil et al., 2000; Heyning, 1989a; Lambert, 2005). Various changes in the general 1984; Lambert et al., 2010a; MacLeod, 2002). Until now, shape, thickness, and inner compactness of its constituent none of these functional hypotheses can explain the full bones have been described in both extinct and extant range of patterns observed. taxa. The most common modification is a filling of the The present study investigates, from morphological and mesorostral groove by the thickened vomer (Bianucci et al., structural points of view, the diversity of forms displayed 2007; Fraser, 1942; Heyning, 1989b; Mead, 1989a). In the by the ziphiid rostrum bones. The aim is to elucidate the extant species Mesoplodon densirostris, the only ziphiid growth processes at work in the different lineages, and to to have been studied in detail from an osteohistological bring new data that could enrich the discussion of func- point of view, the filling of the groove is accompanied tional hypotheses. In addition to a review of published by a strong increase of the compactness and mineral- data, we bring new information on bone compactness ization rate of all rostrum bones (maxillae, premaxillae, and histology in several extinct and extant taxa thanks to and vomer), with strong implications for their mechanical two techniques: computed tomography (or CT scans) and properties (de Buffrénil and Casinos, 1995; Zioupos et al., ground sections. O. Lambert et al. / C. R. Palevol 10 (2011) 453–468 455

2. Material and methods rostrum base fragment, right side, ground sections. Based on the complete closure of the sutures between rostral 2.1. Remarks on information sources bones, we estimate that all the specimens examined were adult. Two information sources were exploited, bibliographic The stratigraphic context is unfortunately not known data and first hand observations. precisely for many of the fossil specimens discussed here. Either they have been reworked and found in basal gravel 2.1.1. Bibliographical data layers, or they were dredged from the bottom of the sea A rich set of bibliographic information dealing with (Bianucci et al., 2007; Lambert, 2005). Therefore, for most rostrum anatomy in various extant and extinct ziphi- of them we can give only an approximate geological age. ids, and specially stressing the relationships between the Even if the extinct species “M.” longirostris and “M.” modified rostral bones (premaxillae, maxillae, vomer, and tumidirostris have been described from isolated rostrum mesethmoid), was considered (Besharse, 1971; Bianucci elements (Cuvier, 1823; Miyazaki and Hasegawa, 1992) et al., 2007, 2008, 2010; Fraser, 1942; Glaessner, 1947; that were shown to be of low diagnostic value (Bianucci Hardy, 2005; Heyning, 1984, 1989a, 1989b; Lambert, 2005; et al., 2007), we choose to maintain the original attribu- Lambert et al., 2010a, 2010b; Leidy, 1877; MacCann, 1965; tion to the genus Mesoplodon (between quoting marks) to MacLeod and Herman, 2004; Mead, 1975a, 1989a, 1989b; stress their obvious affinities with this genus. We agree Miyazaki and Hasegawa, 1992; Moore, 1968; Post et al., that the final genus and species attribution of the speci- 2008; Raven, 1942; Reyes et al., 1991; True, 1910). A sec- mens displaying similarities with the original material of ond set of published works focusing on the histology and these taxa will depend of the description of more complete mechanical properties of the rostrum of the extant M. den- skulls, from the same stratigraphic levels. sirotris was also considered (de Buffrénil and Casinos, 1995; de Buffrénil et al., 2000; Rogers and Zioupos, 1999; Zioupos 2.2.2. Extant ziphiids et al., 1997; Zylberberg et al., 1998). Finally, a few recent Hyperoodon ampullatus MNHN 1881-1149, juvenile, CT works investigating the anatomy of some extant ziphiid scan; M. bidens MNHN 1975-112, adult female, CT scan; skulls by means of X-ray analyses were also taken into M. layardii MNHN 1984-038, adult male, CT scan; Ziphius account (Cozzi et al., 2010; Cranford et al., 2008). cavirostris MNHN 1934-253, adult male, CT scan.

2.1.2. First hand observation New information, mainly dealing with the inner struc- 2.2.3. Non-ziphiid extant odontocetes ture of the rostral region of ziphiid skulls, was collected Delphinus delphis MNHN 1934-367, adult, CT scan; Kogia along two main axes: breviceps MNHN 1976-37, adult, CT scan; Monodon monoceros MNHN 1983-103, adult female, CT scan. • three-dimensional mapping of bone density using computed tomography; 2.3. Techniques implemented • histological examinations based on ground sections. 2.3.1. Computed tomography 2.2. Biological sample and analyses performed Each skull was scanned with a standard medical X-ray tomograph (Siemens AS + 128 slices), producing transverse 2.2.1. Extinct ziphiids slices of 0.6 mm thickness. For all specimens, the scanner New observations of fossil material were made on the was operated at 120 kV, 350 mAs. Mimics software (ver- five taxa listed below. Aporotus recurvirostris (Neogene of sion 13.1) allowed the extraction of transverse sections and Antwerp area, Belgium) IRSNB 3810-M. 2012 a–b, right three-dimensional reconstructions of the skulls based on part of the rostrum, including the subcomplete premax- the sections. illa and adjacent remnants of the maxilla, ground sections; In addition to the identification of sutures between Choneziphius planirostris (Late Miocene of Antwerp area, bones (when not fused), CT scans provide access to Belgium) IRSNB 3775-M.1883, partial skull including ros- bone mineral density (BMD), a morphometric parameter trum and facial area, figured in Lambert, 2005, CT scan; commonly used in medicine, combining local bone com- Choneziphius sp. (Neogene of Antwerp area, Belgium) MB pactness (or trabecular volume) and mineral content of the CR-15, apex of rostrum, ground sections; aff. “Mesplodon” osseous tissue (Meunier and Boivin, 1997). This parame- longirostris (Neogene of Antwerp area, Belgium) IRSNB ter was not quantified in the present study (not calibrated 3804, partial skull including rostrum and facial area, ground with rods of known density), but considered in qualitative sections; aff. “M.” longirostris (Neogene of Antwerp area, terms only, through a visual assessment of the local varia- Belgium) MB CR-9, left part of the rostrum, ground sections; tion of X-ray opacity (white or grey values). For example, aff. “M.” longirostris (Neogene of Antwerp area, Belgium) a roughly similar X-ray opacity could be noted between MB CR-10, anterior part of the rostrum, ground sections; the ear bones and some portions of the rostrum in ziphi- Ziphirostrum marginatum (Neogene of Antwerp area, Bel- ids. Neoceti (Odontoceti + Mysticeti) ear bones are made of gium) MB CR-13, rostrum base fragment, right side, ground one of the densest bone tissues described hitherto (Currey, sections; Z. recurvus (Neogene of Antwerp area, Belgium) 2002; de Buffrénil et al., 2004; Nummela et al., 1999). Using IRSNB 3805-M.544, rostrum, figured in Lambert, 2005,CT the threshold algorithm in Mimics we isolated parts of the scan; aff. Ziphirostrum (Neogene of Antwerp area, Belgium), skull displaying a given density range and produced three- 456 O. Lambert et al. / C. R. Palevol 10 (2011) 453–468 dimensional reconstructions of these high-density parts, in Table 1 a way similar to Cranford et al. (2008). Data-matrix for the four characters discussed in the text. Tableau 1 Matrice de données pour les quatre caractères discutés dans le texte. 2.3.2. Ground sections 1 234 Ground sections were made from rostrum fragments Aporotus † 110- from extinct taxa (see list above), according to the clas- Beneziphius † 11?? sic technique for this kind of preparations (de Buffrénil and Messapicetus † 11?? Mazin, 1989). All sections were made transversely, i.e., per- Ziphirostrum † 0&1 1 1 0 Tasmacetus 00?? pendicular to the longitudinal axis of the rostrum. They Nazcacetus † 10–– × × were observed at low (25 ) and medium (200 ) power Archaeoziphius † ? ??? magnification of a microscope, in natural and polarized Berardius 20?? light. The quantification of bone compactness, or param- Microberardius † 20?? Tusciziphius † 1 11? eter C, was performed on sketches of the sections made † × Choneziphius 1 111 with a camera lucida (magnification 40–50 ; minimum Izikoziphius † 00?? size of cavities: 20 ␮m), using the software Bone Profiler Ziphius 00?? (Girondot and Laurin, 2003). Because all the sections were Xhosacetus † 00?? relatively large (at least 8–12 cm2) and displayed a very Pterocetus † 00?? Indopacetus 10–– homogeneous compactness, measurements were made for Africanacetus † 00?? × each section in six to nine fields 2.5 2.5 mm, evenly dis- Ihlengesi † 00?? tributed within the sectional area, and finally averaged to Hyperoodon 10–– give a mean indication. Mesoplodon 0021 † The terminology used in this study for describing bone aff. “M.” longirostris 0010 structure and histology refers to Francillon-Vieillot et al. 0: primitive state; 1, 2: derived states; ?: no data; –: not applicable; †: (1990). extinct taxa.

2.4. Institutional abbreviations 2.3.3. Phylogeny The phylogenetic background of the discussion is a com- IRSNB: Institut Royal des Sciences Naturelles de Bel- posite phylogeny based on several phylogenetic analyses gique, Brussels, Belgium; MB CR: private collection of Mark available in literature (Bianucci et al., 2007, 2010; Lambert, Bosselaers, Berchem, Belgium; MNHN: Muséum National 2005). It must be noted that the phylogenies taken into d’Histoire Naturelle, Paris, France; SAM: Iziko South African account refer to morphological characters only, including Museum, Cape Town, South Africa; USNM: United States those related to the closure or filling of the mesorostral National Museum of Natural History, Washington DC, USA. groove by rostral bones. The main features discussed in the text can be summarized as four characters: 3. Results • char. 1, filling of the mesorostral groove: vomer (0) – By far the most common (and much likely ple- absent (1) – mesethmoid (2); siomorphic) condition of the rostrum in odontocetes is a • char. 2, dorsal closure of the mesorostral groove by the mesorostral groove void of bone or other mineralized tis- thickened premaxillae: absent (0) – present (1); sue (Fig. 1a, b) and only filled with the mesorostral cartilage, • char. 3, degree of remodeling of the compact rostrum an anterior extension of the nasal septum (Cozzi et al., bones: absent (0); partial (1); complete (2); 2010; Mead and Fordyce, 2009). The groove is laterally and • char. 4, atypical osteons in remodeled compact rostrum ventrally bordered by the vomer, which is usually thin on bones: absent (0); present (1). the lateral walls of the groove (Fig. 1a, b). Dorsally, the groove is often partly overhung by the dorsal portion of Character states are given in Table 1. Using MacClade 4 the premaxillae that forms a thin plate with a transversely (Maddison and Maddison, 2005), we optimized the charac- convex, smooth and hard dorsal surface (i.e., porcelanous ters 1–3 on the composite tree under linear parsimony (see part in Mead and Fordyce, 2009). This porcelanous part Swofford and Maddison, 1987). The derived state for char- is often made of compact, intensely remodeled bone (de acter 4 is for now observed only in two taxa. Finally, with Buffrénil and Lambert, 2011), contrasting with the more Mesquite (Maddison and Maddison, 2010) we used random spongy aspect of the ventral part of the premaxilla and of taxon reshuffling on the reference phylogeny to assess the the maxilla, with exception for the palatal cortex of this presence of a phylogenetic signal for each of the charac- bone (e.g. Delphinus delphis). ters above (Laurin, 2004), a method that can also be used for discrete characters (Laurin, 2005). Ten thousands trees 3.1. Different bones, different shapes were randomly generated, and the number of steps for the given character has been compared between the reference A synthesis of previous works (Bianucci et al., 2007; phylogeny and the random trees. If the number of steps is Lambert, 2005), new external observations of the skull lower in the reference tree than in at least 95% of the gen- anatomy, and CT scans altogether suggest that the diverse erated trees, it can be concluded that a phylogenetic signal modifications of the rostrum shape in ziphiids can be clas- is detected for the studied character (Laurin, 2004). sified in several distinct groups. However, we acknowledge O. Lambert et al. / C. R. Palevol 10 (2011) 453–468 457

Fig. 2. Partial skull of Izikoziphius angustus SAM PQ 3004, Neogene of South Africa, in dorsal view, displaying the complete ﬁlling of the mesorostral groove by the vomer. Scale bar = 100 mm. Fig. 2. Vue dorsale d’une partie du crâne de Izikoziphius angustus SAM PQ 3004, Néogène d’Afrique du Sud, montrant le remplissage complet de la gouttière mésorostrale par le vomer. Barre d’échelle = 100 mm.

mesorostral groove are distinctly thickened. This condition provides to the mesorostral groove a V-shaped section Fig. 1. 3D reconstructions of the skulls of (a) the delphinid Delphinus (Fig. 3), instead of a U-shaped section as is commonly delphis MNHN 1934-367 and (b) the ziphiid Mesoplodon bidens MNHN observed in odontocetes with an unmodified groove (e.g., 1975-112 in dorsal view. The densest areas, mostly the porcelanous part Delphinus, Ninoziphius). Unfortunately we could not per- of the premaxillae on the rostrum and the left ear bones of M. bidens, are form a CT scan of the skull of Tasmacetus during this study. highlighted in yellow and red for D. delphis and M. bidens, respectively. Each reconstruction is accompanied by a virtual transverse section at In some cases the vomer reaches a dorsal level considerably mid-rostrum length displaying the variation of compactness (white areas higher than the premaxillary margins of the mesorostral represent the most compact bone) and the empty mesorostral groove. Scale bars = 100 mm. Fig. 1. Reconstruction 3D des crânes (a) du delphinidé Delphinus delphis MNHN 1934-367 et (b) du ziphiidé Mesoplodon bidens MNHN 1975-112 en vue dorsale. Les régions les plus denses, principalement la partie « à l’aspect de porcelaine » du prémaxillaire sur le rostre et les os de l’oreille gauche de M. bidens, sont colorées respectivement en jaune pour D. delphis et en rouge pour M. bidens. Chaque reconstruction est accompagnée d’une coupe transversale virtuelle à mi-longueur du rostre, montrant les varia- tions de compacité (les zones blanches représentent l’os le plus compact) et la gouttière mésorostrale vide. Barres d’échelle = 100 mm. that the limits of these groups are not always sharply defined. For example, in some cases where the premaxillae dorsally close the mesorostral groove (e.g., Choneziphius), the vomer is significantly thickened in the groove, and the maxillae display an increased width lateral and dorsolateral to the groove. In other cases the inner sutures are fused and the degree of involvement of each rostral bone is difficult to establish (e.g., M. densirostris; de Buffrénil and Casinos, 1995). The classification proposed here is the following.

3.1.1. Filling of the mesorostral groove by the vomer Africanacetus †, Ihlengesi †, Izikoziphius †, Khoikhoice- tus †, Mesoplodon,“M.” longirostris †,“M.” tumidirostris †, Microberardius †, Pterocetus †, Xhosacetus †, Ziphius (Fig. 2; char. 1, state 0). Sexual dimorphism has been demonstrated for this character in Ziphius cavirostris and at least some species of Mesoplodon, in which the ﬁlling is more advanced in adult males than in young males and females (Heyning,

1984, 1989b; MacLeod and Herman, 2004; Mead, 1989a). Fig. 3. Skull of a sexually mature female Tasmacetus shepherdi USNM Nevertheless, it should be noted that some degree of ﬁlling 484878 in anterodorsal view, displaying the thickened vomer on the walls has been observed in adult females of some species (Reyes of the mesorostral groove, giving the groove a V-shaped section at the et al., 1991). An incipient development of this condition rostrum base. Scale bar = 100 mm. Fig. 3. Crâne de femelle sexuellement mature de Tasmacetus shepherdi is observed in a sexually mature female Tasmacetus shep- USNM 484878 en vue antérodorsale, montrant l’épaississement du vomer herdi (USNM 484878, Mead and Payne, 1975), in which the sur les ﬂancs de la gouttière mésorostrale, donnant à la gouttière une lateral walls of the vomer in the posterior region of the section enVàlabase du rostre. Barre d’échelle = 100 mm. 458 O. Lambert et al. / C. R. Palevol 10 (2011) 453–468

Fig. 5. Partial skull of Choneziphius planirostris IRSNB 3774-M.1881, Miocene of the North Sea, in right laterodorsal view, displaying the dorsal closure of the mesorostral groove by the thickened premaxillae. Scale bar = 100 mm. Fig. 5. Vue latérodorsale droite d’une partie du crâne de Choneziphius Fig. 4. Partial skull of Berardius arnuxii SAM 39296 in anterior to planirostris IRSNB 3774-M.1881, Miocène de la Mer du Nord, montrant anterodorsal view, displaying a partial filling of the mesorostral groove la fermeture dorsale de la gouttière mésorostrale par les prémaxillaires by the spongy mesethmoid. The vomer and premaxillae are not signifi- épaissis. Barre d’échelle = 100 mm. cantly thickened. The anterior part of the rostrum is missing, as well as portions of the premaxillae above the groove. Scale bar = 100 mm. Fig. 4. Crâne partiel de Berardius arnuxii SAM 39296 en vue antérieure à either form joined high longitudinal crests, in contact with antérodorsale, montrant le remplissage partiel de la gouttière mésoros- (but not fused to) each other, as exemplified by Aporotus, trale par le mésethmoïde spongieux. Le vomer et les prémaxillaires ne sont pas significativement épaissis. La partie antérieure du rostre est man- or a single medial crest on the rostrum (e.g., Tusciziphius). quante, de même que des portions des prémaxillaires au-dessus de la In both cases, the volume of the premaxillae is significantly gouttière. Barre d’échelle = 100 mm. increased. groove (e.g. the extant Mesoplodon carlhubbsi, Heyning, 3.1.5. Combination of filling of the mesorostral groove by 1984). The high elevation of the vomer above the mesoros- the vomer and dorsal closure by the thickened tral groove at the rostrum base is associated with a distinct premaxillae (char. 1, state 0; char. 2, state 1) bulge in the extinct species “M.” tumidirostris (Miyazaki and Z. recurvus †. In this case the mesorostral groove is com- Hasegawa, 1992). pletely closed and the rostrum is massive and much higher than wide (see description of growth marks farther in the 3.1.2. No thickening or filling (char. 1, state 1; char. 2, text and Lambert, 2005, fig. 15). state 0) Indopacetus, Nazcacetus †, Ninoziphius †. Among the six 3.1.6. Development of high rostral maxillary crests, at the described specimens of the extant Indopacetus pacificus, rostrum base including only one possible adult male, none displays an Hyperoodon (Fig. 6). Sexual dimorphism has been extensive filling of the mesorostral groove (Dalebout et al., demonstrated for this character in at least H. ampulla- 2003). Only one specimen of Nazcacetus urbinai and two tus; adult males bear much higher crests (Hardy, 2005; specimens of Ninoziphius platyrostris were observed. The small samples available for these two extinct species leave open the question of sexual dimorphism; it is possible that only females were discovered until now.

3.1.3. Partial ﬁlling of the mesorostral groove by the mesethmoid Berardius, Microberardius? †, Nenga † (Fig. 4; char. 1, state 2). In the specimens of Berardius arnuxii in which this feature has been observed, the outer surface of the mesethmoid is made of spongy bone.

3.1.4. Dorsal closure of the mesorostral groove by the thickened premaxillae Aporotus †, Beneziphius †, Choneziphius †, Messapicetus †, Tusciziphius †, Ziphirostrum † (Fig. 5; char. 2, state 1). Fig. 6. (a) Skull of adult male Hyperoodon ampullatus MNHN 1872-491 in A mesorostral canal remains open, although its diameter right anterolateral view, displaying the high rostral maxillary crests. (b) varies from one species, or one specimen, to another. Above detail of the spongy surface of the right crest. Scale bars = 300 mm for a, the canal, the premaxillae either remain separated (e.g. 10 mm for b. Fig. 6. (a) Crâne de mâle adulte de Hyperoodon ampullatus MNHN 1872- Aporotus), or more often display a sutural contact, with a 491 en vue antérolatérale droite, montrant les hautes crêtes maxillaires varying degree of fusion between the premaxillae (Fuller rostrales. (b) détail de la surface spongieuse de la crête droite. Barres and Godfrey, 2007, ﬁg. 4). In some cases the premaxillae d’échelle = 300 mm pour a, 10 mm pour b. O. Lambert et al. / C. R. Palevol 10 (2011) 453–468 459

Fig. 7. CT scan of the skull of an adult male Mesoplodon layardii MNHN 1984-038. (a) 3D reconstruction in right lateral view. The densest areas, mostly on the rostrum and part of the vertex, are highlighted in red. (b–d) transverse sections displaying the variation of compactness, at the rostrum base (b), three quarters of the rostrum length (c), and mid-rostrum length (d). Scale bar = 100 mm. Fig. 7. CT scan du crâne d’un mâle adulte de Mesoplodon layardii MNHN 1984-038. (a) reconstruction 3D en vue latérale droite. Les zones les plus denses, principalement sur le rostre et une partie du vertex, sont colorées en rouge. (b–d) coupes transversales montrant la variation de compacité, à la base du rostre (b), aux trois quarts de la longueur du rostre (c), et à mi-longueur (d). Barre d’échelle = 100 mm.

Mead, 1989b). Here again, the increase in bone volume is 3.2. Bone compactness very important. Considerably lower crests medial or lateral to the antorbital notch have been described in various Based on the one hand, on CT scans of specimens of extinct and extant ziphiids (Mead and Fordyce, 2009, Dia- extant species and, on the other hand, on ground sections gram 2); the most conspicuous dome-like crests, in the of fossil specimens, we obtained, respectively, qualitative extinct Africanacetus ceratopsis, are much less developed and quantitative compactness assessments for the differ- than in Hyperoodon spp. (Bianucci et al., 2007). ent bones constituting the rostrum.

Fig. 8. CT scan of the skull of adult male Ziphius cavirostris MNHN 1934-253. (a) 3D reconstruction in left anterodorsolateral view. The densest areas, mostly the premaxillae, the vomer, and the left periotic, are highlighted in red. (b–e) transverse sections at mid-rostrum length (b), three quarters of the rostrum length (c), the rostrum base (d), and across the left periotic (e). Scale bar = 100 mm. Fig. 8. CT scan du crâne d’un mâle adulte de Ziphius cavirostris MNHN 1934-253. (a) reconstruction 3D en vue antérodorsolatérale. Les zones les plus denses, principalement les prémaxillaires, le vomer, et le périotique gauche, sont colorées en rouge. (b–e) coupes transversales à mi-longueur du rostre (b), aux trois quarts (c), à la base du rostre (d), et à travers le périotique gauche (e). Barre d’échelle = 100 mm. 460 O. Lambert et al. / C. R. Palevol 10 (2011) 453–468

(96.8–98.8%), whereas the maxilla is made of cancellous bone (C: 40–65%; de Buffrénil and Lambert, 2011). Skull parts that are most commonly preserved in fossil ziphiids are the rostrum and the dorsal part of the cranium, sometimes including the vertex (Bianucci et al., 2007, 2008; Lambert, 2005; Leidy, 1877). This typical preservation is much more frequent than in any other extinct odontocete group, and much likely results from the high compactness of these areas of the skull. More spongy bones simply dis- appear in the course of deposition, burial, and fossilization.

Fig. 9. Virtual transverse sections in the skull of juvenile Hyperoodon 3.3. Bone histology ampullatus MNHN 1881-1149. (a) section across the spongy rostral maxillary crests. (b) section across the posterior part of the cranium highlighting In the compact rostrum of the extant M. densirostris, the high compactness of the right ear bones. Scale bar = 100 mm. Fig. 9. Coupes transversales virtuelles dans le crâne d’un Hyperoodon all hypermineralized bones consist of a dense Haversian ampullatus juvénile MNHN 1881-1149. (a) coupe à travers les crêtes max- tissue, the secondary osteons of which display atypi- illaires rostrales spongieuses. (b) coupe à travers la partie postérieure du cal features intermediate between parallel-fibered and crâne mettant en évidence la compacité élevée des os de l’oreille droite. woven-fibered bone (this tissue is monorefringent in trans- Barre d’échelle = 100 mm. verse sections). This structure is very different from that of the lamellar bone usually observed in the walls of secondary osteons in adult mammals (de Buffrénil et al., 2000; As mentioned above, for rostrum bones in non-ziphiid Fig. 10e,f). In M. densirostris, the osseous tissue forming the odontocetes (e.g., D. delphis, M. monoceros) and in ziphiid walls of the osteons is characterized by a drastic reduc- specimens devoid of conspicuous local specialization of tion of its collagenous network, a feature that could explain skull bones (e.g., female M. bidens MNHN 1975-112), the extremely high mineralization rate and physical den- compact bone is often observed at the level of the thin pre- sity of the rostrum (2.612 to 2.686 g/cm3; de Buffrénil and maxillae above the mesorostral groove and on the palatal Casinos, 1995; Zylberberg et al., 1998). In addition to its portion of the maxillae (Fig. 1b). exceptional mineralization rate, this kind of bone proved In adult males of M. layardii and Z. cavirostris, the to be the stiffest and hardest ever observed, with a corre- mesorostral groove is filled by the vomer, and the dorsal sponding high brittleness (Zioupos et al., 1997). part of the vomer and the dorsomedial part of the pre- Similar histological features are observed in a trans- maxillae are much more compact than the surrounding verse section from the anterior part of the rostrum of bones (Figs. 7 and 8). In both cases, the limits of the bones Choneziphius sp. Here again the bone is completely remod- do not match the marked changes in compactness. Com- eled, with the same type of atypical, longitudinally oriented pact bone extends on the posterior part of the premaxilla, secondary osteons made of monorefringent, non-lamellar until the vertex, as observed in Z. cavirostris by Cranford osseous tissue (Fig. 10). et al. (2008). However, compact bone is predominantly In the extinct aff. “M.” longirostris, a part of the concentrated in the rostrum, becoming thinner towards the sutures between rostral bones (vomer-maxilla and vomer- vertex. An in situ periotic (ear bone) in the scanned skull premaxilla) is not fused (Fig. 11). The dorsal portion of of Z. cavirostris fits the range of compactness of the most the vomer in the mesorostral groove is distinctly less compact rostrum bones of this specimen. remodeled than the ventral portion, the maxilla, and the As already suggested by Hardy (2005), the maxillary premaxilla. This upper part of the vomer is mostly made crests of the young H. ampullatus are much less compact of a fibrolamellar complex with a predominantly vertical than the thin plate of premaxilla overhanging the mesoros- orientation of the primary osteons (equivalent to a radial tral groove and than the ear bones (Fig. 9). Unfortunately, orientation in a tubular long bone). The secondary osteons due to size constraints we could not scan the skull of a observed in the rest of the rostral bones are more often large adult male of this species (MNHN 1872-491), but the longitudinally or obliquely oriented. They differ strikingly external surface of its crests is distinctly spongy (Fig. 6b), from the secondary osteons observed in M. densirostris and corroborating the scans of the young specimen. Choneziphius sp. because they display an alternate bire- A previous study based on ground sections of M. fringence in polarized light, indicative of a true lamellar densirostris (de Buffrénil and Casinos, 1995) yielded an bone tissue characterized by an orthogonal alternation of extremely high value of compactness (99%) for all the lamellae (Fig. 11 g). bones of the rostrum, with completely fused sutures. The association of non-remodeled and strongly remod- Compactness is similarly high in rostrum fragments from eled compact bone is similarly observed in a dorsal Choneziphius sp. and aff. Ziphirostrum, and in the whole ros- fragment of the rostrum base of aff. Ziphirostrum. The trum of two specimens of the extinct aff. “M.” longirostris, location of these two types of tissue roughly corresponds where only some of the sutures can be distinguished on to the premaxilla and the maxilla, respectively (Fig. 12). the sections. In the rostrum of A. recurvirostris, a differ- The clear alternate birefringence of the numerous longi- ent condition is observed: the hyperplasic premaxilla (i.e. tudinally oriented secondary osteons forming the dense the cortex of this bone is much more developed than in Haversian tissue indicates that osteon walls are made of other taxa, which induces a swollen aspect) is very compact typical lamellar bone. O. Lambert et al. / C. R. Palevol 10 (2011) 453–468 461

Fig. 10. (a) Outline of the partial skull of Choneziphius planirostris IRSNB 3774-M.1881 in right lateral view, with the location of the transverse section performed in the rostrum fragment Choneziphius sp. MB CR-15. (b–d) details of the ground section in Choneziphius sp. MB CR-15. (b) numerous secondary osteons, some of them with a subcomplete closure of the vascular canal. Small rectangle = area enlarged in c–d. (c–d) atypical secondary osteons made of parallel-fibered to woven-fibered bone, in natural (c) and polarized (d) light. (e–f) details of a transverse ground section in the rostrum of Mesoplodon densirostris MNHN 1922-143, in polarized light, displaying similar atypical secondary osteons, lacking an alternate birefringence. Scale bars = 100 mm for a, 2 mm for b, 500 ␮m for c–f. Fig. 10. (a) Contour d’une partie du crâne de Choneziphius planirostris IRSNB 3774-M.1881 en vue latérale droite, montrant la position de la coupe transversale réalisée sur le fragment de rostre Choneziphius sp. MB CR-15. (b–d) détails de la lame mince de Choneziphius sp. MB CR-15. (b) nombreux ostéones secondaires, certains avec une fermeture subcomplète du canal vasculaire. Petit rectangle = zone aggrandie en c-d. (c–d) ostéones secondaires atypiques, constitués d’os à fibres parallèles et à fibres enchevétrées, en lumière naturelle (c) et polarisée (d). (e–f) détails d’une lame mince transversale dans le rostre de Mesoplodon densirostris MNHN 1922-143, en lumière polarisée, montrant le même type d’ostéones secondaires atypiques, sans biréfringence alternée. Barres d’échelle = 100 mm pour a, 2 mm pour b, 500 ␮m pour c-f.

From the medial area to the peripheral dorsolateral accretion rates of the bone tissue, with an initial high region of the swollen, hyperplasic premaxilla of A. recurvi- rate (reticular bone), an intermediate lower rate (laminar rostris, bone tissue type changes, within the general frame bone), and a final increase of the accretion rate (radiating of fibrolamellar complexes, from reticular to laminar, and bone). In the laminar bone, growth marks are conspicu- finally radiating bone, according to vascular orientation ous. The thickness of the growth layer groups (GLG) tends (Fig. 13). This succession likely reflects changes in the to increase from deep to superficial layers in the dor- 462 O. Lambert et al. / C. R. Palevol 10 (2011) 453–468

Fig. 11. (a) Outline of the partial skull of aff. “Mesoplodon” longirostris IRSNB 3800 in left lateral view, with the location of the transverse sections performed in the rostrum fragments aff. “M.” longirostris MB CR-10 (S1) and MB CR-9 (S2). (b) ground section in MB CR-10 with part of the sutures between bones visible. (c–g) details of the ground section in MB CR-9. (c) premaxilla-vomer suture area displaying the transition between the barely remodeled vomer (upper left) and the distincly remodeled premaxilla (lower right). (d–e) detail of the fibro-lamellar complex in the vomer with predominantly vertical primary osteons, in natural (d) and polarized (e) light. (f–g) detail of the longitudinally oriented secondary osteons in the premaxilla, in natural (f) and polarized (g) light. Scale bars = 100 mm for a, 10 mm for b, 2 mm for c, 500 ␮m for d–g. Fig. 11. (a) Contour d’une partie du crâne de aff. « Mesoplodon » longirostris IRSNB 3800 en vue latérale gauche, montrant la position des coupes transversales réalisées dans les fragments de rostre aff. « M. » longirostris MB CR-10 (S1) et MB CR-9 (S2). (b) lame mince dans MB CR-10, avec une partie des sutures entre les os visible. (c–g) détails des lames minces dans MB CR-9. (c) zone de la suture prémaxillaire-vomer montrant la transition entre le vomer à peine remanié (en haut à gauche) et le prémaxillaire distinctement remanié (en bas à droite). (d–e) détail du complexe fibro-lamellaire dans le vomer avec des ostéones primaires en majorité verticaux, en lumière naturelle (d) et polarisée (e). (f–g) détail des ostéones secondaires longitudinaux dans le prémaxillaire, en lumière naturelle (f) et polarisée (g). Barres d’échelle = 100 mm pour a, 10 mm pour b, 2 mm pour c, 500 ␮m pour d–g. sal and dorsolateral parts of the bone. This indicates that 1979), and occurred at 9 GLGs in one female M. densirostris sub-periosteal accretion rate was particularly fast in these (Ross, 1984). These comparative elements suggest that the directions, with a tendency to get still faster in time (de studied specimen of A. recurvirostris was a young adult or, Margerie et al., 2002, 2004). The premaxilla of Aporotus at least, a late subadult. Furthermore, an absence or a low is entirely made of primary bone devoid of any kind of degree of remodeling is commonly observed in the lami- inner (especially Haversian) remodeling, with exception nar bone tissue of adults in other mammals, for example for a narrow zone along the medial wall of the bone. Con- artiodactyls (Currey, 2002). versely, the spongy maxilla of the same specimen was The macroscopic examination of fracture surfaces in intensely remodeled (Fig. 13e), as is the general condi- the hyperplasic premaxillae of aff. Aporotus dicyrtus and tion for this bone in mammals. Remodeling of the maxilla Choneziphius planirostris (Fig. 14a–c; Lambert, 2005, fig. in A. recurvirostris is a first argument against the hypothe- 18), and the examination of CT scan sections of the sis of an age-related non-remodeling of the premaxilla. A premaxilla of Z. recurvus (Fig. 14d) reveal clear growth second argument is the observation of 8 to 9 GLGs in this marks in complete sequences. Therefore, it must be con- specimen, each of them supposed to represent one year cluded that these bones were not remodeled, like those of in the life of the animal (de Buffrénil and Lambert, 2011). A. recurvirostris. Therefore, this specimen must have died in its 9th or 10th To summarize, the degree of remodeling in ziphiid ros- year. Among extant ziphiids, the few data available indi- tra strongly differs between taxa: secondary osteons can be cate that sexual maturity is reached at 7–11 dental GLGs developed in the whole rostrum (M. densirostris), or only in in Hyperoodon ampullatus (Benjaminsen and Christensen, some bones (aff. “M.” longirostris, aff. Ziphirostrum). They O. Lambert et al. / C. R. Palevol 10 (2011) 453–468 463

Fig. 12. (a) Outline of the partial skull of Ziphirostrum marginatum IRSNB 3847-M.537 in right lateral view (reversed), with the location of the transverse section performed in the rostrum fragment aff. Ziphirostrum.(b) ground section in the maxilla and premaxilla. (c–d) detail of the maxilla-premaxilla suture area displaying the non-remodeled premaxilla (upper left) and the strongly remodeled maxilla (lower right), in natural (c) and polarized (d) light. (e–f) secondary osteons in the maxilla, in natural (e) and polarized (f) light, displaying the subcomplete closure of the vascular canals and the alternate birefringence typical of lamellar tissue. Scale bars = 100 mm for a, 1 mm for c–d, 500 ␮m for e–f. Fig. 12. (a) Contour d’une partie du crâne de Ziphirostrum marginatum IRSNB 3847-M.537 en vue latérale droite (inversée), montrant la position de la coupe transversale réalisée dans le fragment de rostre aff. Ziphirostrum.(b) lame mince dans le maxillaire et le prémaxillaire. (c–d) détail de la région de la suture maxillaire-prémaxillaire montrant le prémaxillaire non remanié (en haut à gauche) et le maxillaire fortement remanié (en bas à droite), en lumière naturelle (c) et polarisée (d). (e–f) ostéones secondaires dans le maxillaire, en lumière naturelle (e) et polarisée (f), montrant la fermeture subcomplète des canaux vasculaires et la biréfringence alternée typique du tissu lamellaire. Barres d’échelle = 100 mm pour a, 1 mm pour c–d, 500 ␮m pour e–f. can also be absent from the bones (otherwise pachyos- following results. The ancestral state for char. 1 (filling totic and osteosclerotic) of other species (A. recurvirostris). of the mesorostral groove) is equivocal, depending on Remodeling can also be restricted to a part only of a bone the choice of the Accelerated Transformation (AccTran) (e.g., vomer of aff. “M.” longirostris that is only remod- or Deleted Transformation (DelTran) hypothesis; either eled in its ventral portion). With the exception of the the filling by the vomer is ancestral (DelTran), subse- atypical osteons of Choneziphius sp. and M. densirostris, sec- quently lost in several clades and re-appearing in the ondary osteons in ziphiid rostra display a normal lamellar clade Izikoziphius + Ziphius, or it appeared independently organization in transverse section. In all rostra for which (AccTran) in numerous clades (Z. recurvus but not the ground sections of compact bone were available, a com- other species of the genus Ziphirostrum, Tasmacetus, plete or subcomplete occlusion of vascular canals occurs. Izikoziphius + Ziphius, Xhosacetus, Pterocetus, Africanace- Since the precursor cells of the osteoclasts, the monocytes, tus + Ihlengesi + Mesoplodon + aff. “M.” longirostris). The are brought in situ via blood vessels, this observation would derived filling of the groove by the mesethmoid appears mean that the remodeling process of ziphiid rostral bones, in the clade Berardius + Microberardius (unknown in though potentially intense, cannot proceed after canal Archaeoziphius). However, it should be noted that when closure and is thus limited in time (there is a trend to self- assessing the phylogenetic signal for char. 1, the number blockage of remodeling). This inference is confirmed by the of steps in the reference tree is lower than in 94.61% of the lack or great sparseness of open erosion bays in all the taxa randomly generated trees (probability 0.0539, Table 2), investigated. which is at the very limit of reliability for this character optimization. For char. 2 (dorsal closure of the groove by 3.4. Confrontation with phylogenetic framework the thickened premaxillae) the ancestral state is similarly equivocal; either the closure occurs in the common ances- The data compiled here are summarized in a com- tor of all ziphiids (AccTran) and is lost in several clades, posite phylogenetic tree in Fig. 15. The optimization or it appears independently (DelTran) in three clades of the ancestral state for the characters 1–4 gave the (Aporotus + Beneziphius + Messapicetus + Ziphirostrum, 464 O. Lambert et al. / C. R. Palevol 10 (2011) 453–468

Fig. 13. (a) Outline of the partial skull of Aporotus recurvirostris IRSNB 3812-M.1887 in right lateral view, with the location of the transverse section performed in the rostrum fragment A. recurvirostris IRSNB 3810-M.2012. (b) ground section in the hyperplasic premaxilla of A. recurvirostris IRSNB 3810- M.2012 displaying conspicuous growth marks (arrows) in the laminar bone. (c) detail of the radiating bone in the upper left area of the section. (d) detail of the laminar (left) and more reticular (right) primary bone. (e) ground section in the remodeled and spongy maxilla of A. recurvirostris IRSNB 3810-M.2012. Scale bars = 100 mm for a, 10 mm for b, 5 mm for e, 1 mm for c–d. Fig. 13. (a) Contour d’une partie du crâne de Aporotus recurvirostris IRSNB 3812-M.1887 en vue latérale droite, montrant la position de la coupe transversale réalisée dans le fragment de rostre A. recurvirostris IRSNB 3810-M.2012. (b) lame mince dans le prémaxillaire hyperplasique de A. recurvirostris IRSNB 3810- M.2012 montrant les marques de croissance (ﬂèches) dans l’os laminaire. (c) détail de l’os radiaire dans la zone supérieure gauche de la coupe. (d) détail de l’os primaire laminaire (à gauche) et réticulaire (à droite). (e) lame mince dans le maxillaire spongieux et remanié de A. recurvirostris IRSNB 3810-M.2012. Barres d’échelle = 100 mm pour a, 10 mm pour b, 5 mm pour e, 1 mm pour c–d.

Table 2 Choneziphius, Tusciziphius). For this character the assess- Probability that the distribution of the states of each character studied is ment of the phylogenetic signal provides a probability of random with respect to the phylogeny. This is assessed by determining the number of randomly generated trees (out of 10 000) that imply the same 0.0104, which indicates a reliable character optimization. number of steps (or less) than the reference tree. This test was performed The ancestral state for char. 3 (degree of remodeling of by Mesquite. A probability lower than 0.05 indicates a reliable optimiza- compact rostral bone) is much likely a partial remodel- tion of the given character (number of steps lower in the reference tree ing. Optimization proposes the absence of remodeling than in more than 95% of the random trees). in Aporotus and possibly Beneziphius + Messapicetus, Tableau 2 Probabilité que la distribution des états de caractères soit aléatoire par and complete remodeling in Mesoplodon and possibly rapport à la phylogénie. Cette probabilité est déterminée par le nombre Africanacetus + Ihlengesi. However, with a probability of 1 d’arbres aléatoires qui impliquent le même nombre de pas (ou moins) que the optimization cannot be considered reliable. A similar l’arbre de référence. Les 10 000 arbres aléatoires ont été générés à l’aide unreliable result is achieved for char. 4 (presence of atypi- de Mesquite. Une probabilité inférieure à 0,05 indique une optimisation ﬁable pour le caractère considéré (nombre de transitions sur l’arbre de cal osteons in remodeled bone), only observed for now in référence, plus bas que dans au moins 95 % des arbres aléatoires). two distantly related taxa (Choneziphius and Mesoplodon).

Probability Number of steps on reference tree 4. Discussion Char. 1 0.0539 7 Char. 2 0.0104 3 Although occurring in most ziphiids, be they extinct or Char. 3 1 2 extant, the morphological specialization of rostral bones Char. 4 1 2 is expressed in distinct ways among the various lineages. O. Lambert et al. / C. R. Palevol 10 (2011) 453–468 465

Fig. 15. Composite phylogenetic tree of the ziphiids based on several cladistic analyses (Bianucci et al., 2007, 2010; Lambert, 2005). †, extinct taxa. Specimens referred to the extinct “Mesoplodon” longirostris probably do not belong to the genus Mesoplodon, even if they are closely related. Fig. 14. Growth marks in the premaxilla, on dorsal surface of the ros- Optimization of the ancestral state is shown for characters 1–2. Char. 1 trum of Choneziphius planirostris IRSNB 3777-M.1882 (a), on transverse (filling of the mesorostral groove) on the left mirror-tree. Light grey, fill- fracture surface at the rostrum base of aff. Aporotus dicyrtus IRSNB 3807- ing by the vomer (0); dark grey, absence of filling (1); black, filling by the M.1889 (b–c), shown in dorsal view (b), in posterior view of the fracture mesethmoid (2). Char. 2 (dorsal closure of the groove by the thickened pre- surface (c), and on virtual transverse section at mid-length of the rostrum maxillae) on the right mirror-tree. Light grey, absence (0); black, presence of Ziphirostrum recurvus IRSNB 3805-M.544 (d). Arrows indicate the most (1). Only the DelTran hypothesis is illustrated for both characters. conspicuous growth marks. Scale bars = 50 mm for a-b, 10 mm for d. Fig. 15. Arbre phylogénétique composite des ziphiidés, basé sur plusieurs Fig. 14. Marques de croissance dans le prémaxillaire, en surface dorsale analyses cladistiques (Bianucci et al., 2007, 2010 ; Lambert, 2005). †, taxons du rostre de Choneziphius planirostris IRSNB 3777-M.1882 (a), sur une éteints. Les spécimens attribués au taxon éteint « Mesoplodon » longirostris surface de fracture transversale à la base du rostre de aff. Aporotus dicyr- n’appartiennent vraisemblablement pas au genre Mesoplodon, même s’ils tus IRSNB 3807-M.1889 (b–c), en vue dorsale (b) et en vue postérieure en sont proches. L’optimisation de l’état ancestral est montrée pour les de la surface de fracture (c), et en coupe transversale virtuelle à mi- caractères 1–2. Car. 1 (remplissage de la gouttière mésorostrale) sur longueur du rostre de Ziphirostrum recurvus IRSNB 3805-M.544 (d). Les l’arbre-miroir de gauche. Gris clair, remplissage par le vomer (0) ; gris flèches indiquent les marques de croissance les plus manifestes. Barres foncé, absence de remplissage (1) ; noir, remplissage par le mésethmoïde d’échelle = 50 mm pour a–b, 10 mm pour d. (2). Car. 2 (fermeture dorsale de la gouttière par les prémaxillaires épais- sis) sur l’arbre-miroir de droite. Gris clair, absence (0) ; noir, présence (1). Seule l’hypothèse DelTran est illustrée pour les deux caractères. From an anatomical point of view, the bones involved in the changes of shape of the rostrum (vomer, mesethmoid, maxilla, premaxilla) differ considerably from one taxon to remodeling may deeply change from one taxon to another another. Furthermore, the optimization of ancestral states (Fig. 15; Aporotus and Ziphirostrum, M. densirostris and aff. indicates a complex history with several convergences and “M.” longirostris). Furthermore, similar, atypical (monore- reversions (Fig. 15). From a histological point of view, fringent) secondary osteons can be observed in two taxa the limited sample of adult specimens studied hitherto belonging to distantly related lineages (Choneziphius sp. suggests that the remodeling pattern of compact cortices and M. densirostris). Due to the preservation state of fos- also varies among taxa (no remodeling, partial remodeling, sil specimens and the difficulty to gain access to samples complete remodeling). In a same lineage, the extent of bone of extant species, hypermineralization could be demon- 466 O. Lambert et al. / C. R. Palevol 10 (2011) 453–468 strated for now only in M. densirostris (Zylberberg et al., may therefore be considered more as an obstacle than as a 1998). With the exception of Choneziphius sp., the other reflector for the sounds to be transmitted forwards via soft taxa studied display osteons made of typical lamellar bone, tissues. Another even more problematic issue is the pref- a tissue known to have a high collagen content and a rela- erential development of compact bone in adult males of tively low mineralization rate (Boivin and Meunier, 2002). extant ziphiids. Indeed, the few data available hitherto do However, further investigation of the mineral and colla- not allow the distinction of different echolocation or div- gen content of extant ziphiid rostra would be necessary to ing abilities between adult males and females/juveniles in confirm this potentially important difference. a same ziphiid species (Hooker and Baird, 1999; Johnson et The high variation, at different levels of bone organiza- al., 2004, 2008; Madsen et al., 2005; MacLeod, 2002; Tyack tion (at least anatomical and histological), of the detailed et al., 2006). phenotypical expressions of rostrum specializations in the As mentioned above, the most obvious physical effect of Ziphiidae strongly suggests that distinct processes were developing pachyosteosclerotic rostrum bones (irrespec- at work for the acquisition of such features in the various tive of their external shape and histology) is a local increase ziphiid lineages. Indeed, even if, as proposed by character in mass. As deep divers, ziphiids might find a benefit in such optimization, the ancestral condition might be a mesoros- a ballast, positioned at the anterior extremity of the body, tral groove filled by the vomer or dorsally closed by the even if it is of minor importance, as compared to the total thickened premaxillae (both results equivocal), a great weight of the animals. This ballast could indeed help them diversity of anatomical and histological specializations maintaining a vertical position in the water column dur- appears within this clade. These features do not simply ing descent (dynamic ballast; de Buffrénil et al., 2000). The represent a plesiomorphic feature inherited from stem- most obvious criticism to this hypothesis is that the ener- ziphiids; they would rather reflect a broad convergence getic benefit during the descent would be lost during the between most ziphiid taxa (but not all taxa; cf. the cases way back to the surface (MacLeod, 2002). However, prelim- of Indopacetus, Nazcacetus, and Ninoziphius), in response inary observations on diving ziphiids showed that, at least to (a) common selective pressure(s) arising from similar, in some of the few dives recorded for M. densirostris and Z. but possibly not fully identical, ecological specializations cavirostris, the locomotion type changes from the descent (Bianucci et al., 2008). to the ascent; the studied animals more frequently per- This consideration raises, of course, the question of the formed passive and metabolically cheap gliding during the functional implication(s) of the rostral specializations. A descent, whereas swimming was more active, with more basic observation is that, beyond the diversity of processes, frequent tail strokes during the ascent (Tyack et al., 2006), a there is an obvious, factual result: increasing bone mass, behaviour similarly observed in several other groups of div- stiffness, and physical density in the rostral region of the ing marine mammals (Williams et al., 2000). Nevertheless, skull. as mentioned above, no significant intraspecific difference Several functions have been proposed for the compact could be observed for now between the swimming style rostrum bones of extant and extinct ziphiids. The most con- and diving depths of adult males, adult females, and juve- vincing hypotheses can be classified into three categories, niles. related to three main functional and ecological features: With their sexually dimorphic tusks, adult males of acoustics, deep diving, and intraspecific fights between many extant ziphiid species are supposed to engage in males (de Buffrénil et al., 2000; Cozzi et al., 2010; Heyning, intraspecific fights. Based on indirect observations (accu- 1984; MacLeod, 2002). mulation of unpigmented linear scars on the body, likely In odontocetes, the production of echolocation sounds made by the tusks), this putative behaviour leads to occurs in the forehead and the transmission is made the hypothesis that thicker and heavier rostrum bones through the melon (Cranford et al., 1996; Mead, 1975b). (especially the filling of the mesorostral groove by the Air sacs surrounding the phonic lips, the sound-producing vomer) may limit the risks of fractures caused by impacts organs in the forehead, may act as acoustic mirrors for (Heyning, 1984; MacLeod, 2002). However, the extremely echolocation and communication sounds. In the case of high mineral content of the hyperdense rostrum of M. deep-diving ziphiids (for example, individuals of M. den- densirostris, the only species on which mechanical tests sirostris and Z. cavirostris have been recorded at depths have been applied, results in a very brittle material, cer- up to 1251 and 1888 m, with dive durations reaching 57 tainly unsuited to impact loading (de Buffrénil et al., 2000; and 85 min, respectively; Tyack et al., 2006), hydrostatic Zioupos et al., 1997). Interestingly, a preliminary analysis pressure strongly reduces the volume of air spaces. Den- of the bone organization in the maxillary crests of H. ampul- sity disparity between forehead tissues and highly compact latus, the only species for which head-butting has been bone could therefore constitute an alternative acoustic actually observed (Gowans and Rendell, 1999), indicates reflector (Cranford et al., 2008). This model is in agreement that, as proposed by Hardy (2005), the crests are made of with the geometry of compact bone elements in the skull of spongy bone, in strong contrast with the compact bone of several extant and extinct ziphiids, even if compact bone many other ziphiid rostra. Such a condition would actu- is preferentially located in the anterior part of the skull. ally be reminiscent of the spongy aspect of the antlers and However, unusual shapes of hyperplasic bones in various other skull elements of head-butting terrestrial ungulates extinct ziphiid taxa are in poor agreement with this acous- (Currey, 2002). It nevertheless remains that changes in the tic model alone. For example, the elevated medial crests of thickness and shape of the rostrum bones, leading to the “M.” tumidirostris and Tusciziphius sp. are located anterior dorsal closure or the filling of the mesorostral groove, could to the area of production of the echolocation sounds, and have a positive effect on the mechanical strength of this O. Lambert et al. / C. R. Palevol 10 (2011) 453–468 467 part of the skull, especially in long-snouted species with Currey, J.D., 2002. Bones. Structure and mechanics. Princeton University apical lower tusks (Bianucci et al., 2008; Lambert et al., Press, Princeton, 456 p. Cuvier, G., 1823. Recherches sur les ossements fossiles 5 (1re partie). G. 2010a). However, unusual shapes of the rostrum bones, Dufour et E. D’Ocagne, Paris, 405 p. as exemplified by the medial crest in “M.” tumidirostris Dalebout, M.L., Ross, G.J.B., Baker, C.S., Anderson, R.C., Best, P.B., Cock- and Tusciziphius sp., are more difficult to understand in the croft, V.G., Hinsz, H.L., Peddemors, V., Pitman, R.L., 2003. Appearance, distribution, and genetic distinctiveness of Longman’s beaked whale, framework of this hypothesis. Indopacetus pacificus. Mar. Mamm. Sci. 19, 421–461. To conclude, the striking diversity of morphological and de Buffrénil, V., Casinos, A., 1995. Observations histologiques sur le rostre histological features documented here in the rostrum of de Mesoplodon densirostris (Mammalia, Cetacea: Ziphiidae): le tissu osseux le plus dense connu. Ann. Sci. Nat. 13e série 16, 21–32. extinct and extant ziphiids suggests a complex evolution- de Buffrénil, V., Lambert, O., 2011. Histology and growth pattern of the ary history for these characters within the family, and pachy-osteosclerotic premaxillae of the fossil beaked whale Aporotus poorly matches a single, univoqual functional explanation recurvirostris (Mammalia: Cetacea: Odontoceti). Geobios 44, 45–56. de Buffrénil, V., Mazin, J.M., 1989. Bone histology of Claudiosaurus ger- for now. Several behavioural/ecological factors may have maini (Reptilia Squamata) : données comparatives et considérations played a role in the development of pachyosteosclerotic fonctionnelles. Hist. Biol. 2, 311–322. bone, and the latter could possibly be a relevant solution de Buffrénil, V., Rage, J.C., 1993. La « pachyostose » vertébrale de Simo- for more than one selective pressure. A broader sampling liophis (Reptilia Squamata) : données comparatives et considérations fonctionnelles. Ann. Paleontol. 79, 315–335. for histological analyses, including new extinct species and de Buffrénil, V., Dabin, W., Zylberberg, L., 2004. Histology and growth of the juveniles, adult females, and adult males of extant species, cetacean petro-tympanic bone complex. J. Zool. London 262, 371–381. will certainly provide additional arguments to the discus- de Buffrénil, V., Zylberberg, L., Traub, W., Casinos, A., 2000. Structural and mechanical characteristics of the hyperdense bone of the rostrum of sion, as well as more detailed data about the ecology of Mesoplodon densirostris (Cetacea Ziphiidae): summary of recent obser- extant ziphiids (recognition of different feeding and diving vations. Hist. Biol. 14, 57–65. habits for males and females, observation of intraspecific de Margerie, E., Cubo, J., Castanet, J., 2002. Bone typology and growth ... rate: testing and quantifying “Amprino’s rule” in the mallard (Anas fights ). platyrhynchos). C. R. Biologies 325, 221–230. de Margerie, E., Robin, J.P., Verrier, D., Cubo, J., Groscolas, R., Castanet, J., 2004. Assessing a relationship between bone microstructure and Acknowledgements growth rate: a fluorescent labeling study in the king penguin chick (Aptenodytes patagonicus). J. Exp. Biol. 207, 869–879. We wish to thank A. Abourachid, M. Bosselaers, A. Folie, de Muizon, C., 1984. Les Vertébrés de la Formation Pisco (Pérou). Deuxième partie : les odontocètes (Cetacea Mammalia) du Pliocène C. Lefèvre, E. Pellé, and E. Steurbaut for the loan or gift of inférieur du Sud-Sacaco. Trav. Inst. Fr. Et. Andines 27, 1–188. specimens, L. Cazes and J.L. Lemoine for the production of de Ricqlès, A., de Buffrénil, V., 2001. Bone histology, heterochronies and the the ground sections, J.M. Lorphelin, P. Somma, F. Boumalk, return of tetrapods to life in water: where are we? In: Mazin, J.M., de Buffrénil, V. (Eds.), Secondary adaptation of tetrapods to life in water. and L. Jokiel from the Centre d’Imagerie Médicale de Bois- Verlag Dr. Friedrich Pfeil, München, pp. 289–310. Bernard and W. Coudyser from the Universitair Ziekenhuis Francillon-Vieillot, H., de Buffrénil, V., Castanet, J., Géraudie, J., Meu- Gasthuisberg Leuven for the CT scans, D. Germain for his nier, F.J., Sire, J.Y., Zylberberg, L., de Ricqlès, A., 1990. Microstructure and mineralization of vertebrate skeletal tissues , In: Carter, J.G. help with character optimization, D. Geffard-Kuriyama and (Ed.), Skeletal biomineralization: patterns, processes and evolutionary F. Goussard for their assistance with Mimics, and three trends., vol. 1. Van Nostrand Reinhold, New York, pp. 471–530. anonymous reviewers and the editor M. Laurin for their Fraser, F.C., 1942. The mesorostral ossification of Ziphius cavirostris. Proc. constructive comments about our manuscript. Zool. Soc. London B 112, 21–30. Fuller, A.J., Godfrey, S.J., 2007. A Late Miocene ziphiid (Messapicetus sp.: Odontoceti: Cetacea) from the St. Marys Formation of Calvert Cliffs, Maryland. J. Vert. Paleontol. 27, 535–540. References Girondot, M., Laurin, M., 2003. Bone Profiler: a tool to quantify, model and statistically compare bone section compactness profiles. J. Vert. Benjaminsen, T., Christensen, I., 1979. The natural history of the bottlenose Paleontol. 23, 458–461. whale, Hyperoodon ampullatus Forster. In: Winn, H.E., Olla, B.L. (Eds.), Glaessner, M.F., 1947. A fossil beaked whale from Lakes Entrance, Victoria. Behavior of marine mammals. Plenum Press, New York, pp. 143–164. Proc. R. Soc. Victoria 58, 25–34. Besharse, J.C., 1971. Maturity and sexual dimorphism in the skull, Gowans, S., Rendell, L., 1999. Head-butting in northern bottlenose whales mandible, and teeth of the beaked whale, Mesoplodon densirostris.J. (Hyperoodon ampullatus): a possible function for big heads? Mar. Mammal. 52, 297–315. Mamm. Sci. 15, 1342–1350. Bianucci, G., Lambert, O., Post, K., 2007. A high diversity in fossil beaked Hardy, M.T., 2005. Extent development and function of sexual dimor- whales (Odontoceti Ziphiidae) recovered by trawling from the sea phisms in the skulls of the Bottlenose whales (Hyperoodon spp.) and floor off South Africa. Geodiversitas 29, 561–618. Cuvier’s beaked whale (Ziphius cavirostris). Master thesis, School of Bianucci, G., Post, K., Lambert, O., 2008. Beaked whale mysteries revealed Biological Sciences, University of Wales, Bangor, U.K, 99p. by sea floor fossils trawled off South Africa. S. Afr. J. Sci. 104, 140–142. Heyning, J.E., 1984. Functional morphology involved in intraspecific fight- Bianucci, G., Lambert, O., Post, K., 2010. High concentration of long- ing of the beaked whale, Mesoplodon carlhubbsi. Can. J. Zool. 62, snouted beaked whales (genus Messapicetus) from the Miocene of 1645–1654. Peru. Palaeontology 53, 1077–1098. Heyning, J.E., 1989a. Comparative facial anatomy of beaked whales Boivin, G., Meunier, P.J., 2002. The degree of mineralization of bone tissue (Ziphiidae) and a systematic revision among the families of extant measured by computerized quantitative contact microtomography. Odontoceti. Contr. Sci., Nat. Hist. Mus. Los Angeles County 405, Calcif. Tissue Int. 70, 503–511. 1–64. Cozzi, B., Panin, M., Butti, C., Podestá, M., Zotti, A., 2010. Bone density Heyning, J.E., 1989b. Cuvier’s beaked whale Ziphius cavirostris G. Cuvier, distribution patterns in the rostrum of delphinids and beaked whales: 1823 , In: Ridgway, S.H., Harrison, R. (Eds.), Handbook of marine mam- evidence of family-specific evolutive traits. Anat. Rec. 293, 235–242. mals, vol. 4: river dolphins and the larger toothed whales. Academic Cranford, T.W., Amundin, M., Norris, K.S., 1996. Functional morphology Press, London, pp. 289–308. and homology in the Odontocete nasal complex: implications for Heyning, J.E., Mead, J.G., 1996. Suction feeding in beaked whales: mor- sound generation. J. Morph. 228, 223–285. phological and observational evidence. Contr. Sci., Nat. Hist. Mus. Los Cranford, T.W., McKenna, M.F., Soldevilla, M.S., Wiggins, S.M., Goldbo- Angeles County 464, 1–12. gen, J.A., Shadwick, R.E., Krysl, P., St Leger, J.A., Hildebrand, J.A., 2008. Hooker, S.K., Baird, R.W., 1999. Deep-diving behaviour of the northern Anatomic geometry of sound transmission and reception in Cuvier’s bottlenoose whale. Hyperoodon ampullatus (Cetacea: Ziphiidae). Proc. beaked whale (Ziphius cavirostris). Anat. Rec. 291, 353–378. R. Soc. Lond. B 266, 671–676. 468 O. Lambert et al. / C. R. Palevol 10 (2011) 453–468

Johnson, M., Hickmott, L.S., Aguilar Soto, N., Madsen, P.T., 2008. Echoloca- Mead, J.G., 1989b. Bottlenose whales Hyperoodon ampullatus (Forster, tion behaviour adapted to prey in foraging Blainville’s beaked whale 1770) and Hyperoodon planifrons Flower, 1882 , In: Ridgway, S.H., Har- (Mesoplodon densirostris). Proc. R. Soc. Lond. B 275, 133–139. rison, R. (Eds.), Handbook of marine mammals., vol. 4: river dolphins Johnson, M., Madsen, P.T., Aguilar Soto, N., Zimmer, W.M.X., Tyack, P.L., and the larger toothed whales. Academic Press, London, pp. 321–348. 2004. Beaked whales echolocate for prey. Proc. R. Soc. Lond. B 271, Mead, J.G., Fordyce, R.E., 2009. The therian skull: a lexicon with emphasis 383–386. on the odontocetes. Smithsonian Contr. Zool. 627, 1–248. Lambert, O., 2005. Systematics and phylogeny of the fossil beaked whales Mead, J.G., Payne, R.S., 1975. A specimen of the Tasman beaked whale Ziphirostrum du Bus, 1868 and Choneziphius Duvernoy, 1851 (Cetacea, Tasmacetus shepherdi, from Argentina. J. Mammal. 56, 213–218. Odontoceti), from the Neogene of Antwerp (North of Belgium). Geo- Meunier, P.J., Boivin, G., 1997. Bone mineral density reflects bone mass but diversitas 27, 443–497. also the degree of mineralization of bone: therapeutic implications. Lambert, O., Bianucci, G., Post, K., 2009. A new beaked whale (Odontoceti Bone 21, 373–377. Ziphiidae) from the Middle Miocene of Peru. J. Vert. Paleontol. 29, Miyazaki, N., Hasegawa, Y., 1992. A new species of fossil beaked whale, 911–922. Mesoplodon tumidirostris sp. nov. (Cetacea Ziphiidae) from the central Lambert, O., Bianucci, G., Post, K., 2010a. Tusk-bearing beaked whales from North Pacific. Bull. Natn. Sci. Mus. Tokyo 18, 167–174. the Miocene of Peru: sexual dimorphism in fossil ziphiids? J. Mammal. Moore, J.C., 1968. Relationships among the living genera of beaked whales. 91, 19–26. Fieldiana Zool. 53, 209–298. Lambert, O., Godfrey, S.J., Fuller, A.J., 2010b. A Miocene Ziphiid (Cetacea: Nummela, S., Wägar, T., Hemilä, S., Reuter, T., 1999. Scaling of the cetacean Odontoceti) from Calvert Cliffs, Maryland. U. S. A. J. Vert. Paleontol. 30, middle ear. Hearing Res. 133, 71–81. 1645–1651. Post, K., Lambert, O., Bianucci, G., 2008. First record of Tusciziphius crispus Laurin, M., 2004. The evolution of body size, Cope’s rule and the origin of (Cetacea, Ziphiidae) from the Neogene of the US east coast. Deinsea amniotes. Syst. Biol. 53, 594–622. 12, 1–10. Laurin, M., 2005. Embryo retention, character optimization, and the origin Raven, H.C., 1942. On the structure of Mesoplodon densirostris, a rare of the extra-embryonic membranes of the amniotic egg. J. Nat. Hist. beaked whale. Bull. Am. Mus. Nat. Hist. 80, 23–50. 39, 3151–3161. Reyes, J.C., Mead, J.G., Van Waerebeek, K., 1991. A new species of beaked Leidy, J., 1877. Description of vertebrate remains, chiefly from the whale Mesoplodon peruvianus sp. n (Cetacea: Ziphiidae) from Peru. Phosphate Beds of South Carolina. J. Acad. Nat. Sci. Philadelphia 8, Mar. Mamm. Sci. 7, 1–24. 209–261. Rogers, K.D., Zioupos, P., 1999. The bone tissue of the rostrum of a Meso- MacCann, C., 1965. The mesorostral groove in Ziphiidae, with special ref- plodon densirostris whale: a mammalian biomineral demonstrating erence to its closure by ossification in Mesoplodon – Cetacea. Rec. extreme texture. J. Mater. Sci. Lett. 18, 651–654. Dominion Mus. 5, 83–88. Ross, G.J.B., 1984. The smaller cetaceans of the South East coast of southern MacLeod, C.D., 2002. Possible functions of the ultradense bone in the ros- Africa. Ann. Cape Prov. Mus. Nat. Hist. 15, 173–410. trum of Blainville’s beaked whale (Mesoplodon densirostris). Can. J. Schreer, J.F., Kovacs, K.M., 1997. Allometry of diving capacity in air- Zool. 80, 178–184. breathing vertebrates. Can. J. Zool. 75, 339–358. MacLeod, C.D., Herman, J.S., 2004. Development of tusks and associated Swofford, D.L., Maddison, W.P., 1987. Reconstructing ancestral character structures in Mesoplodon bidens (Cetacea, Mammalia). Mammalia 68, states under Wagner parsimony. Math. Biosci. 87, 199–229. 175–184. True, F.W., 1910. An account of the beaked whales of the family Ziphiidae Maddison, D.R., Maddison, W.P., 2005. MacClade 4 Release Version 4. 08 in the collection of the United States National Museum, with remarks for OS X. Sinauer Associates, Inc, Sunderland, Massachusetts. on some specimens in other American museums. Bull. U. S. Nat. Mus. Maddison, W.P., Maddison, D.R., 2010. Mesquite: a modular system for 73, 1–89. evolutionary analysis. Version 2.73. http://mesquiteproject.org. Tyack, P.L., Johnson, M., Aguilar Soto, N., Sturlese, A., Madsen, P.T., 2006. Madsen, P.T., Johnson, M., Aguilar Soto, N., Zimmer, W.M.X., Tyack, P.L., Extreme diving of beaked whales. J. Exp. Biol. 209, 4238–4253. 2005. Biosonar performance of foraging beaked whales (Mesoplodon Werth, A.J., 2006. Mandibular and dental variation and the evolution of densirostris). J. Exp. Biol. 208, 181–194. suction feeding in Odontoceti. J. Mammal. 87, 579–588. Mead, J.G., 1975a. A fossil beaked whale (Cetacea: Ziphiidae) from the Williams, T.M., Davis, R.W., Fuiman, L.A., Francis, J., Le Boeuf, B.J., Horning, Miocene of Kenya. J. Paleontol. 49, 745–751. M., Calambokidis, J., Croll, D.A., 2000. Sink or swim: strategies for cost- Mead, J.G., 1975b. Anatomy of the external nasal passages and facial com- efficient diving by marine mammals. Science 288, 133–136. plex in the Delphinidae (Mammalia: Cetacea). Smithsonian Contr. Zioupos, P., Currey, J.D., Casinos, A., de Buffrénil, V., 1997. Mechanical Zool. 207, 1–67. properties of the rostrum of the whale Mesoplodon densirostris,a Mead, J.G., 1989a. Beaked whales of the genus Mesoplodon , In: Ridgway, remarkably dense bony tissue. J. Zool. London 241, 725–737. S.H., Harrison, R. (Eds.), Handbook of marine mammals., vol. 4: river Zylberberg, L., Traub, V., de Buffrénil, V., Alizard, F., Arad, T., Weiner, S., dolphins and the larger toothed whales. Academic Press, London, pp. 1998. Rostrum of a toothed whale: ultrastructural study of a very 349–430. dense bone. Bone 23, 241–247.