Return to the Sea: The Evolution of Marine Mammals

Abstract

Early Cetacea are classified as Archaeoceti (six families), and their fossils reveal the morphological and functional evolution of early, terrestrial Artiodactyla to fully aquatic Cetacea over 15 million years during the Eocene (~52 Mya). The evolution of modern Cetacea (Neoceti: Mysticeti and Odontoceti) began in the late Eocene (~37 Mya). As with Mysticeti, late Eocene Dorudontidae were a likely sister group of early Odontoceti. The evolution of Sirenia began in the early Eocene (56–48 Mya). Fossil pinnipedimorphs only extend to the late Oligocene (27–25 Mya). Morphological and molecular evidence support a monophyletic origin for the three extant families of pinnipeds (Otariidae, Odobenidae, and Phocidae) within the taxonomic order Carnivora. The earliest fossils of Enhydra sp. are from early Pleistocene (2.6–1.8 Mya). There is no fossil record to document the evolution of the physiological adaptations that occur in living marine mammals. The dive response and enhanced oxygen stores, which increased the aerobic dive duration, and blubber for thermal insulation may have been early adaptations in each of the marine mammal lineages. The exception are fur seals and the sea otter, which rely on waterproof fur. Adaptations to avoid the detrimental effects of pressure during deep diving may have come later.

To understand the morphological and physiological adaptations of marine mammals, it is helpful to place them into an evolutionary context and discuss their phylogenetic relationships. What is noteworthy is that convergent morphological adaptations for an aquatic life (e.g., fusiform body shape, flukes, and flippers) evolved separately in the three taxonomic orders of extant marine mammals (Artiodactyla [infraorder Cetacea], Carnivora [clade Pinnipedia and sea otters], and Sirenia). These adaptations evolved in other extant marine tetrapods (e.g., diving reptiles and birds) and extinct marine reptiles (e.g., ichthyosaurs, plesiosaurs, and pliosaurs). This fact alone indicates the omnipresent selection pressure on animals to function efficiently through drag reduction, enhanced thrust production, and decreased cost of locomotion in the aquatic environment (see Chap. 5).

Analysis of the fossilized bones (soft tissues rarely fossilize) of ancestral marine mammals has greatly improved our understanding of their morphological evolution, and osteological correlations provide some understanding of neural-sensory function (Domning 2005; Thewissen 2014; Berta et al. 2015; Marx et al. 2016a; Lindgren et al. 2018). However, fossils provide no direct information on their physiology, especially concerning diving and thermoregulation. Hence, we can only speculate about the evolution of the dive response, enhanced body oxygen stores, tolerance to pressure, and the ability to maintain a high and stable core body temperature in water.

2.1 Cetacea

2.1.1 Archaeoceti

The story of Cetacea began with the evolutionary radiation of placental mammals after the extinction of the dinosaurs and most large marine reptiles at the end of the Cretaceous about 66 million years ago (hereafter written Mya) (Fig. 2.1). The Paleocene (66–56 Mya) was the first epoch of the Cenozoic, a time when many placental mammals were small (<10 kg; Godinot 1994; Alroy 1999; Hu et al. 2005) and none larger than ~650 kg (e.g., the pantodontid Barylambda). During the Eocene (56–34 Mya), they evolved into diverse forms including a clade of even-toed (Artiodactyla) and odd-toed (Perissodactyla) ungulates that were digitigrade with hooves. Stem Cetacea (Archaeoceti) still had four limbs and a distinctive double-pulley astragalus (ankle bone or talus) in the hindlimbs, a key feature found in all fossil and extant terrestrial Artiodactyla. In contrast, mesonychids (order Mesonychia), which once were thought to be the progenitors of early Cetacea based on dental similarities and other morphological traits, lacked the distinctive astragalus.

Fig. 2.1

Phylogenetic trees of marine mammals. Heavy lines indicate the estimated stratigraphic range of each group, whereas thin lines indicate phylogenetic links. Stratographic range for Prorastomidae from Domning (pers. com.). (Adapted from Uhen (2007))

A stratocladistic analysis of morphological, molecular, and stratigraphic data found a close phylogenetic relationship between Artiodactyla and Archaeoceti but not with Mesonychia. Results from the analysis indicated that species of Diacodexis, a 50-cm-long ungulate belonging to the extinct family of Dichobunidae, were the basal progenitors of Artiodactyla including Cetacea and a sister group, the Hippopotamidae (Geisler and Uhen 2005, but see also Gingerich et al. 2001). Thewissen et al. (2001a, 2007) also excluded the Mesonychia from the Artiodactyla and argued that Raoellidae (semiaquatic digitigrade Artiodactyla such as the chevrotain-like Indohyus; Fig. 2.2a) and Cetacea were more closely related than either is to Hippopotamidae. This conclusion was based on the morphology of the tympanic bone (presence of an involucrum or thickened inner lip), osteosclerotic (denser) limb bones, and the ratio of stable isotopes of oxygen (O¹⁶/O¹⁸) indicating a semiaquatic habitat for Indohyus (Thewissen et al. 2009). Yet a third study of morphological and molecular data for Artiodactyla concluded that Indohyus was a close relative to Cetacea, but the authors admitted that their analysis was unstable and strongly affected by which taxa were included (Spaulding et al. 2009). The emerging consensus is that Cetacea evolved from primitive Artiodactyla in the early Eocene and therefore belong in the taxonomic order Artiodactyla (suborder Whippomorpha, infraorder Cetacea). However, in some taxonomic schemes, the term Cetartiodactyla is used instead of Artiodactyla.

Fig. 2.2

Skeletal reconstructions of: (a) Indohyus, (b) Pakicetus, (c) Ambulocetus, (d) Rodhocetus, and (e) Basilosauridae (Dorudon atrox). (Reprinted and adapted with permission: (a) Thewissen et al. 2007, (b) Thewissen et al. (2001a) (c) Thewissen (2014), (d) Gingerich et al. (2001), (e) Gingerich (2012). Not drawn to scale)

Early Cetacea are classified as Archaeoceti (parvorder Archaeoceti) for which there are five to six families: Pakicetidae, Ambulocetidae, Remingtonocetidae, Protocetidae, Dorudontidae, and Basilosauridae (the latter two are sometimes grouped together) (Thewissen and Williams 2002). Together, their fossils reveal the morphological and functional evolution of terrestrial Artiodactyla to aquatic Cetacea over ~15 million years during the Eocene (Fig. 2.1). Many Archaeoceti fossils come from geological strata that once formed floodplains and shallow waters of the ancient Tethys Ocean, with notable discoveries from the Indo-Pakistan region.

Pakicetidae (three genera) were the earliest Eocene Archaeoceti and primarily terrestrial with only a few morphological features that link them to modern Cetacea (Gingerich et al. 1983; Thewissen et al. 2001a). Hence, they were a transitional group from terrestrial Artiodactyla (viz., a Raoellidae) to semiaquatic, stem Cetacea in the floodplains of present-day Southern Asia ~50 Mya (Thewissen and Williams 2002; Thewissen et al. 2009; Clementz et al. 2006). They retained the heterodont dentition of generalized placental mammals with 44 teeth (upper/lower dental formula: incisors 3/3, canines 1/1, premolars 4/4, molars 3/3). However, the rostrum was elongated placing the incisors in line with cheek teeth, and the premolars were triangular blades while the molars retained a reduced number of cusps. Their diet may have included terrestrial and freshwater prey. The postcranial skeleton was adapted for terrestrial locomotion (Figs. 2.2b and 2.3) even though Gingerich (2012) portrays them as more amphibious. Regardless, they were probably waders in shallow freshwater and used quadrupedal paddling when swimming similar to that in other terrestrial mammals (Thewissen and Fish 1997; Thewissen et al. 2009). The middle ear had adaptations consistent with hearing in air (external auditory meatus and ear ossicles similar to those in terrestrial mammals) but also in water (a large, pachyosteosclerotic auditory bulla partially separated from the squamosal bone and an involucrum on the tympanic bone). As with Indohyus, Pakicetidae had osteosclerotic limb bones with a ratio of stable isotopes of oxygen indicating a semiaquatic habitat.

Fig. 2.3

The skeletons of Ambulocetus (top) and Pakicetus (bottom). (Reprinted with permission from the Thewissen Lab)^©

One of the oldest Ambulocetidae, Himalayacetus, was discovered in marine sediments of the ancient Tethys Ocean in what is now northwestern India (Bajpai and Gingerich 1998), although there is some uncertainty about its age. It was originally described as a terrestrial or semiaquatic species of Pakicetidae (based only on teeth and mandibular fragments) but reclassified as Ambulocetidae (three genera) after an 80% complete specimen of Ambulocetus natans (Latin for the walking and swimming cetacean) was discovered (Thewissen et al. 2001b; Thewissen et al. 2009). Weighing an estimated 200 kg, Ambulocetus was an amphibious transitional phase of cetacean evolution (48.6–40.4 Mya), which was more aquatic than Pakicetidae (Figs. 2.2c and 2.3; Thewissen et al. 1994). Their fossils have been found in nearshore marine deposits that may have been estuaries or bays. Limb proportions are similar to those in otters (greatly foreshortened, especially the femur) but with long hind feet, so they probably swam using dorsoventral undulation or pelvic paddling but may have moved onshore like a sea lion (Thewissen and Fish 1997; see Sect. 5.2.2). Their diet is unknown, but the morphology of their skull and teeth is similar to that in crocodiles, which are ambush predators. Whether or not they still had fur is unknown but a possibility.

The Remingtonocetidae was a short-lived family of Archaeoceti from the early-to-middle Eocene (contemporaneous with Ambulocetus) and known only from India, Pakistan, and Africa in geological strata of shallow water, marine origin (Kumar and Sahni 1986; Thewissen and Hussain 2000; Bebej et al. 2016). This family is conspicuous among Archaeoceti for elongated and narrow cranial rostra, heterodont dentition (i.e., incisors, canines, premolars, and molars), a small periotic bone and internal auditory meatus, and a reduction in the size of the semicircular canals of the vestibular apparatus similar to that in modern Cetacea (see Sect. 7.2.2). The mandibular foramen also is large similar to that in modern Cetacea and may have been associated with improved underwater hearing. The postcranial skeleton was primitive with long cervical vertebrae, a sacrum composed of four (instead of five) fused vertebrae, and robust forelimbs capable of bearing the animal’s weight on land (Gingerich et al. 1995). The morphology of Remingtonocetidae is therefore consistent with an amphibious lifestyle, but the extent of terrestrial locomotion will require better knowledge of the limbs and tail. As with Ambulocetus, they may have used some variation of pelvic paddling or dorsoventral undulation for aquatic locomotion (Thewissen and Williams 2002; Thewissen et al. 2009).

The next family of Archaeoceti was the middle Eocene (~46 Mya) Protocetidae (15 genera), the most basal family of Cetacea to be globally distributed. Their ability to colonize continents outside of Asia implies that some species were pelagic, although fossils also occur in strata that are near-coastal (Clementz et al. 2006; Thewissen et al. 2009). Although the Protocetidae were diverse, perhaps the best known is Rodhocetus, which weighed ~500 kg (Fig. 2.2d). The nares are posterior from the tip of the rostrum (over the upper canines or first premolars) on a path that would eventually produce the blowhole of modern Cetacea. The eyes were large and faced laterally. The teeth of Rodhocetus were simplified, but the dentition of Protocetidae varied depending on feeding ecology (Fig. 2.4a). The greatly enlarged mandibular foramen along with an enlarged auditory bulla that is less well connected to the skull provided better directional hearing. Although Rodhocetus had compressed cervical vertebrae and a short neck, as occurred in later Dorudontidae and some fast-swimming modern Odontoceti, it still had four limbs with long hands and feet that were modified for pelvic paddling or pelvic undulation. However, four sacral vertebrae were still fused with the pelvis suggesting some ability to move onshore (Gingerich et al. 2001). Later Protocetidae such as Protocetus atavus and Georgiacetus vogtlensis had one or no sacral vertebrae attached to the pelvis, which would have made movement on land difficult or impossible (Hulbert et al. 1998).

Fig. 2.4

(a) Skull cast of a Protocetidae (Protocetus atavus; note enlarged auditory bullae), (b) skull of a Dorudontidae (Cynthiacetus uvianus), and (c) skeleton of a Dorudontidae. (Reprinted and adapted with permission: (a) Funkmonk/Wikimedia Commons, (b) D. Gordon E. Robertson, (c) Jeff Major/Flickr)

The Dorudontidae and Basilosauridae families were contemporaneous and often included in a single family (Basilosauridae, seven genera) although their morphology differed (Fig. 2.2e). These were the last families of Archaeoceti to complete the transition to a fully aquatic existence 38–34 Mya in the middle-to-late Eocene: the last walking whales. As with the Protocetidae, they were globally distributed. The nares continued to retract and migrate posteriorly to become a blowhole, and the heterodont dentition (upper/lower dental formula: incisors 3/3, canines 1/1, premolars 4/4, molars 2/3) was aligned in the plane of the jaw with accessory denticles on the cheek teeth that resembled flat blades (Uhen 2008; Fig. 2.4b). The auditory bullae continued to become detached and acoustically isolated from the rest of the skull, but there was no anatomical indication of a melon or echolocation ability. Compared to Dorudontidae, the Basilosauridae were longer (up to 16 m) because of additional and longer vertebrae, but both groups lacked fused sacral vertebrae (Fig. 2.4c). The shape of the caudal vertebrae indicates the presence of flukes (although there is no fossil evidence) and caudal oscillation for propulsion. This lift-based mode of locomotion was more efficient than the drag-based modes of early and middle Eocene Archaeoceti (see Sect. 5.2.1). The appendicular skeleton is also modified with a reduction in size, shape, and movement of the forelimbs (now recognizable flippers) and hindlimbs (very small and associated with a vestigial pelvis), which would reduce drag but make land locomotion impossible. Hence, the selection for enhanced swimming efficiency drove the modification of the postcranial skeleton from the early terrestrial Artiodactyla to the fully aquatic Dorudontidae and Basilosauridae (Fig. 2.2e).

The evolution of modern Cetacea (Neoceti: Mysticeti and Odontoceti) began in the late Eocene (~37 Mya; Fig. 2.1), and the sister group was likely the Dorudontidae (Uhen and Gingerich 2000; Uhen 2008). Toward the end of the Eocene (34 Mya), cetacean morphology for efficient locomotion, which had evolved over the previous 15 million years, remained mostly the same during subsequent evolutionary changes (Thewissen and Williams 2002). Instead, the primary distinctions between the modern parvorders of Mysticeti and Odontoceti are associated with morphological adaptions for feeding.

2.1.2 Mysticeti

Late Eocene Dorudontidae, the likely predecessors of early Mysticeti, had teeth to capture large or intermediate size prey, while modern Mysticeti use baleen to filter large numbers of small prey (see Sect. 8.1.1.1). Hence, there must have been a transitional phase from toothed Dorudontidae that used raptorial biting and intraoral suction to modern Mysticeti that lack an adult dentition (Fig. 2.5a; Fitzgerald 2006; Demere et al. 2008; Uhen 2008; Hocking et al. 2017). This transitional phase may have involved intermittent or continuous suction feeding with teeth acting as a simple sieve to retain food (Marx et al. 2016b). Baleen may have evolved later to retain smaller prey during the water expulsion phase of suction feeding. Whether proto-baleen grew between the upper teeth or posterior to vestigial teeth in the rostrum is uncertain. Alternatively, suction feeding and the complete loss of teeth may have preceded the evolution of baleen (Peredo et al. 2017, 2018). At some point, teeth were replaced by baleen so that filtering no longer relied on suction leading to continuous ram or skim feeding in the Balaenidae and intermittent lunge feeding among the Balaenopteridae. Interestingly, grey whales still engage in suction and filter feeding in mixed sediments on the seafloor. However, the evolutionary steps leading to the replacement of teeth with baleen in archeo-mysticetes continues to be debated, and several transitional linages may have occurred simultaneously until one eventually lead to crown Mysticeti.

Fig. 2.5

(a) Reconstruction of Aetiocetus weltoni showing hypothesized simultaneous occurrence of teeth and baleen, (b) baleen of a Bryde’s whale. (Reprinted and adapted with permission: (a) Deméré et al. (2008), (b) Jason Thompson/Flickr)

Baleen is composed of keratin, a tough, fibrous protein that is synthesized by the cornified epithelial cells of the palate in the form of transversely oriented racks of vertical plates that hang from the upper jaw between the tongue and the lip of the lower jaw when the mouth is closed (Figs. 2.5b and 2.6a, c; Utrecht 1965). Keratinous filaments on the lingual surfaces of the continuously growing baleen plates form a dense mat that retains prey while water is expelled through the lateral sides of the mouth similar to a cross-flow filter (see Sect. 8.1.1.1). The transition to filter feeding in early Mysticeti also was associated with an increase in body size, especially the head, which increased the amount of water that could be filtered and the total mass of prey captured. However, most of the Mysticeti remained less than 15 m in length, and it may not have been until the Plio-Pleistocene (~3 Mya) and the onset of seasonally intensified ocean upwelling that gigantism was ecologically selected resulting in whales longer than 15 m such as the fin and blue whale (Slater et al. 2017). Although the dinosaurs were the largest terrestrial animals, we live in a time of giant marine mammals and the largest animals that have ever lived.

Fig. 2.6

(a) The skeleton of a bowhead whale showing the highly arched upper jaw with baleen and bowed lower jawbones, (b) Bryde’s whale with throat pleats expanded after feeding on a bait ball of sardines, (c) skull of a minke whale showing less of an arched upper jaw and shorter baleen. (Reprinted and adapted with permission: (a) Mira Mechtley/Flickr, (b) Doug Perrine/Nature Picture Library)

The evolution of Mysticeti and baleen as a feeding adaptation occurred during the Oligocene (33.9–23 Mya), but the details are unclear because of the scarcity of fossils from this period. In addition, some of the earliest known fossils of archaic Mysticeti without teeth are coincident with those that still have a heterodont dentition with multicusped teeth (Deméré et al. 2008; Uhen 2008). Although fossilized baleen is uncommon (Gioncada et al. 2016), anatomical features in the skull of early Mysticeti provide evidence that baleen was present. These features include foramina (openings in the bone) and associated sulci (grooves) for blood vessels on the lateral portions of the palate of some fossil Mysticeti that are homologous with similar foramina and blood vessels that perfuse the epidermal base of the baleen plates in extant Mysticeti. Hence, we have indirect fossil evidence that baleen formed a primitive filter between the widely spaced teeth in the upper jaw of early Mysticeti such as the Aetiocetidae and Mammalodon, although the shape and size of the baleen can only be estimated (Fig. 2.5a; Deméré et al. 2008; Berta et al. 2016). An alternative hypothesis is that some stem Mysticeti (e.g., Maiabalaena nesbittae) became edentulous suction feeders in the early Oligocene and that complete tooth loss preceded the evolution of baleen (Peredo et al. 2018).

An analysis of fossil and molecular evidence indicates that the evolutionary loss of teeth finally occurred in archaic toothless Mysticeti such as Eomysticetus whitmorei in the late Oligocene. Around this time, dentition became rudimentary and prey capture with teeth ended. Additional morphological adaptations for filtering large quantities of water would follow such as (1) a broad rostrum, (2) bowed lower jaw bones, (3) a ligamentous mandibular symphysis, (4) a highly arched upper jaw to accommodate the long baleen in right and bowhead whales (Balaenidae), and (5) a highly elastic throat pouch that is pleated externally in rorquals (Balaenopteridae) (Figs. 2.6a, b and 8.1).

Cetotheriidae were a large, diverse assemblage of archaic Mysticeti dating from the late Oligocene to the Pliocene that were small- to medium-sized baleen whales. Mysticeti such as Mauicetus were stem Balaenopteridae from the late Oligocene (28–23 Mya), while Morenocetus was a stem Balaenidae from the early Miocene (23 Mya). The early fossil record is less clear for the pygmy right whale (Neobalaenidae) and grey whale (Eschrichtiidae).

A phylogenetic analysis including known Oligocene cetacean fossils indicates a post-Oligocene radiation of crown Mysticeti (Geisler et al. 2011), which includes 14 extant species with 16 subspecies (Appendix 2). These species are currently grouped taxonomically into two superfamilies: Balaenoidea and Balaenopteroidea. The Balaenoidea (right whales) currently include two families: Balaenidae (4) and Cetotheriidae (1). The Balaenopteroidea (rorquals and grey whale) include two families: Balaenopteridae (8) and Eschrichtiidae (1). A broad synthesis of morphological, embryological, molecular, and paleontological factors indicates that Eschrichtiidae is an extant sister species of the Balaenopteridae and that Neobalaenidae along with the Balaenidae are sister taxa to the Eschrichtiidae and Balaenopteridae clade (Deméré et al. 2008). Another cladistic analysis concluded that Cetotheriidae, Balaenopteridae, and Eschrichtiidae should be grouped into a new clade named Thalassotherii, which excludes Neobalaenidae and Balaenidae (Bisconti et al. 2013). However, a recent study suggested that Neobalaenidae might be the last of the Cetotheriidae, thereby resurrecting this group from extinction (Fordyce and Marx 2012). If true, this would reclassify Neobalaenidae as Cetotheriidae and place it in the Balaenopteroidea (rorquals, grey whale, and pygmy right whale). It is likely that the phylogeny of the Mysticeti will continue to change as new fossils are discovered.

2.1.3 Odontoceti

As with Mysticeti, late Eocene Dorudontidae were likely predecessors of early Odontoceti. However, they retained, to varying degrees, their adult dentition, although not necessarily for feeding. Some species continued to use teeth to capture small and intermediate-sized prey, but others evolved suction feeding and lost the functional use of teeth (reduced in number or modified in size and shape) except among males for fighting. In both Mysticeti and Odontoceti, the premaxillary and maxillary bones are elongated so that the nasal bones and blowhole are far back on the head. However, the premaxillary and maxillary bones of Mysticeti are convex in shape (to varying degrees; Figs. 2.6a, c) to accommodate palatal baleen plates. In contrast, the facial plane of Odontoceti is concave and asymmetric (with the skew to the left side with the right side being larger) to accommodate the other major evolutionary adaptation in Cetacea: a melon, facial muscles and nasal sacs for directional sound generation associated with echolocation (Fig. 2.7a, b, c). Secondary characteristics included an enlarged mandibular foramen along with an enlarged auditory bulla not connected to the skull for directional hearing. Hence, the use of sound comes to dominate both the morphology of the skull and the method of prey detection and feeding ecology.

Fig. 2.7

(a) Lateral view of the skull of a bottlenose dolphin showing its concave shape to accommodate a melon, enlarged mandible, and polydont and homodont dentition, (b) dorsal view of the skull of a bottlenose dolphin showing the asymmetry (with the skew to the left side with the right side being larger), elongation of the premaxillary (1) and maxillary bones (2) and foreshortening of the nasal bones (3), which places the blow hole (4) farther back on the head, (c) the concave head of a sperm whale, which accommodates the spermaceti organ and junk for sound production, (d) lower jaw of a sperm whale showing the teeth and enlarged mandible

There was a significant diversification of Odontoceti species in the late Oligocene (25 Mya). The Xenorophidae (e.g., Xenorophus, Albertocetus, and Agorophius) are considered stem Odontoceti that were heterodont and polydont resulting from additional, undifferentiated cheek teeth (Uhen 2008). The Squalodontidae was another family of Odontoceti whose fossils date from the late Oligocene to the Miocene (25–6 Mya). They were small- to medium-sized dolphins with polydont, triangular cheek teeth that had serrated edges and a wrinkled surface (shark-like teeth). Hence, the evolution of polydonty becomes prevalent in these stem Odontoceti possibly resulting from the caudal expansion and expression of bone morphogenetic protein (Bmp4) that led to additional teeth with similar, simple crowns (Armfield et al. 2013). This is manifest in many of the crown Odontoceti, which are polydont and homodont (Fig. 2.7a), but not in the extant sperm whale and most of the extant beaked whales, which are suction feeders and have reduced adult dentition (Fig. 2.7c, d).

The Kentriodontidae was a family similar to small extant dolphins (but with a more primitive, symmetrical skull) whose fossils date from the late Oligocene to the Pliocene (25–5 Mya). They may have had high frequency hearing and the ability to echolocate, a major evolutionary trait characteristic of crown Odontoceti and one that occurs to the same degree of sophistication in no other mammalian order except the Chiroptera (bats) (Churchill et al. 2016).

As with the Mysticeti, a phylogenetic analysis indicates a post-Oligocene radiation for crown Odontoceti (Geisler et al. 2011), which include 74 extant species and 34 subspecies (Appendix 2). These crown Cetacea are currently grouped taxonomically into five superfamilies: Delphinoidea, Physeteroidea, Platanistoidea, Inioidea, and Ziphioidea. The Delphinoidea (oceanic dolphins, Arctic whales, and porpoises) include three families (number of species in parentheses): Delphinidae (37), Monodontidae (2), and Phocoenidae (7). The Physeteroidea (sperm whales) include two families: Physeteridae (1) and Kogiidae (2). The Platanistoidea include one family: Platanistidae (1). The Inioidea include two families: Iniidae (1) and Pontoporiidae (1). Finally, the Ziphioidea (beaked whales) include one family: Ziphidae (22).

2.2 Sirenia

As with the Cetacea, the evolution of Sirenia (order Sirenia) began in the early Eocene (56–48 Mya; Fig. 2.1). Sirenia belong to the Paenungulata clade that includes three extant mammalian orders: Hyracoidea (hyraxes), Proboscidea (including elephants), and Sirenia, (dugongs and manatees). It currently includes the extinct order Desmostylia (large aquatic quadrupeds), which, with the Proboscidea and Sirenia, are grouped in the Afrotherian clade of Tethytheria, a group named after the Tethys Ocean around which they originally evolved. However, a new cladistic analysis suggests that the Desmostylia, whose fossils have always been restricted to the Pacific Rim and never found in Afro-Arabia, may be more closely related to the Perissodactyla in the Laurasiatheria clade rather than Paenungulata (Cooper et al. 2014). Further cladistic analysis of morphological and stratigraphic data will be needed to establish their phylogenetic relationship.

The earliest stem Sirenia remains uncertain. One possibility is the Chambi sea cow (family indeterminate) from Tunisia, which dates from the early Eocene (~50 Mya) (Benoit et al. 2013). Unfortunately, this claim is based only on a fossil petrosal (the dense part of the temporal bone surrounding the inner ear), so no additional information is available on general morphology. The location of the discovery was in a bed of lacustrine limestone, which suggests a freshwater or euryhaline environment.

Fossils that are more complete have been found for amphibious, stem Sirenia in the family Prorastomidae also dating from the same period (~50 Mya) but from Jamaica (Fig. 2.1). Fossil bones (skull, mandible, and atlas of an adult) of the holotype Prorastomus sirenoides have an interesting history because they were described in 1855 by Sir Richard Owen and then lay in the British Museum of Natural History, receiving little attention until reanalysis starting in 1977 (Savage 1977; Savage et al. 1994). Nearly complete skeletal remains of another Prorastomidae (Pezosiren portelli) from Jamaica were described in 2001 (Fig. 2.8a: Domning 2001). The fossils were found in lagoon sediments that were estuarine or deltaic suggesting a semiaquatic lifestyle. The morphology of this pig-sized (~2.1 m in length) quadruped with a barrel-shaped trunk and short legs is considered transitional between terrestrial and aquatic locomotion (Fig. 2.8b). It had some skeletal characteristics (e.g., limb bones) resembling herbivorous Paleocene condylarths such as Ectoconus, but its skull, teeth, and ribs (pachyostotic and osteosclerotic) were typical of Sirenia. The heavy ribs provided ballast for a semiaquatic lifestyle similar to modern Hippopotamidae. They most likely swam using quadrupedal or pelvic paddling and walked along the bottom of shallow lagoons.

Fig. 2.8

(a) Reconstructed composite skeleton of Pezosiren portelli (shaded elements are represented by fossil bones), (b) artists’ rendition of the live animal. (Courtesy of Tim Scheirer © CalvertMarineMuseum.com), (c) dugong skeleton, (d) dugong. (Reprinted and adapted with permission: (a) Domning (2001)/Macmillan, (b) The Calvert Marine Museum, Solomons, Md by Tim Scheirer © CalvertMarineMuseum.com, (d) Doug Perrine/SeaPics.com)

The Protosirenidae were another family of stem Sirenia from the middle Eocene that were intercontinental across North America, Europe, North Africa, and Indo-Pakistan. The fossil skull of Protosiren frassi from Egypt, first described by Andrews (1906), dates from the early-middle Eocene (46 Mya) (Fig. 2.1; Gingerich et al. 1994). A more complete fossil skeleton of Protosiren smithae from Egypt dates from the middle Eocene (41–38 Mya) and may be a direct ancestor of the former (Domning and Gingerich 1994). Although Protosirenidae were still amphibious based on the short but well-developed limbs, whether they could lift their bodies off the ground is uncertain. As a result, their mode of locomotion on land is indeterminate, but they probably swam like Pezosiren. Morphological changes in the skull associated with feeding also are apparent. While Prorastomus and Pezosiren had nearly straight rostra, Protosirenidae developed a premaxillary deflection indicating that they fed off the bottom of lagoons and estuaries. Although more divergently specialized and less aquatically adapted than contemporary Dugongidae representing a more advanced stage of sirenian evolution, Protosirenidae illustrate the morphological transition from amphibious to more aquatic locomotion.

Early Dugongidae (family Dugongidae, dugongs) first appeared in the middle Eocene (41 Mya) and showed progressive reduction in the pelvis and hindlimbs (Fig. 2.1). By the end of the Eocene (34 Mya), Dugongidae had a fusiform body shape, foreshortened neck, forelimbs that were flippers, vestigial pelvis, hindlimbs that probably lacked the more distal bones and did not protrude externally, and flukes for locomotion using caudal oscillation similar to Cetacea (Fig. 2.8c, d). One group of early Dugongidae, the Halitheriinae, gave rise to two subfamilies: the Dugonginae (dugongs) and the Hydrodamalinae (with Metaxytherium as an intermediate genus). One feature was a reduction in cheek teeth (1–3 pairs of molars on either side of the upper and lower jaws). Dugonginae were once more diverse, with at least nine genera from the late Oligocene to Pliocene, but are now reduced to one species (Dugong dugon). The last member of the Hydrodamalinae, Steller’s sea cow, became extinct in the 1760s as a result of its accidental discovery by Vitus Bering’s expedition in the North Pacific in 1741 and the subsequent maritime fur trade for sea otters during which this large Sirenia was a food source for Russian sailors and hunters.

The Trichechidae (family Trichechidae, manatees) may have been a sister group of the early Dugongidae or Protosirenidae in the early Oligocene (~28 Mya), but the fossil record is incomplete (Fig. 2.1). Two subfamilies are now recognized: Miosireninae and the Trichechinae. Miosireninae occurred around northern Europe in the late Oligocene to early Miocene and may have fed on shellfish. The earliest species of Trichechinae (Potamosiren magdalenensis) dates from the middle Miocene (16–12 Mya) in South America, but the holotype (a single type specimen upon which the description and name of a new species is based) is only half a mandible and a third molar, so the animal’s morphology is uncertain. Likewise, Ribodon limbatus, dating from the late Miocene to early Pliocene (6–5 Mya), is based solely on mandibles and teeth (Domning 2005). However, what is noteworthy is that this species (in contrast to Potamosiren) had a continuous succession of horizontally replaced supernumerary teeth (although fewer than in Trichechus), making unlimited tooth replacement a relatively recent adaptation to an abrasive diet of siliceous grasses during the late Miocene. Ribodon occurred in estuarine and riverine regions of South America and was evidently an ancestor of modern manatees. At some point, there must have been sufficient genetic isolation for Amazon basin manatees to diverge from the coastal West Indian manatees (Domning 2005). Coastal manatees in South America extended their range northward through the Caribbean and North America and eastward to West Africa during the Pliocene and Pleistocene (Domning 1982). Genetic isolation eventually caused the African and West Indian manatees to diverge and resulted in two subspecies of the West Indian manatee in North America. Interestingly, as manatees expanded their range, Dugongidae became extinct in the western Atlantic, Europe, and the Mediterranean. The extant Trichechidae trace to the Pleistocene and include three species: Amazonian manatee (Trichechus inunguis), African manatee (Trichechus senegalensis), and West Indian manatee (Trichechus manatus). There are two subspecies of Trichechus manatus: Florida manatee (T. manatus latirostris) and Antillean or Caribbean manatee (T. manatus manatus) (Appendix 2).

2.3 Pinnipedia

Fossils of the arctoid ancestors of Pinnipedia (fur seals, sea lions, walruses, seals) can be traced to the Eocene (~45 Mya), although fossil pinnipedimorphs only extend to the late Oligocene (27–25 Mya) (Fig. 2.1). Morphological and molecular evidence support a monophyletic origin for the three extant families (Otariidae, Odobenidae, and Phocidae) within the taxonomic order of Carnivora (suborder Caniformia, infraorder Arctoidea). There was a controversy over whether Pinnipedia were more closely related to the families of Ursidae (bears) or Mustelidae (otters), but the emerging consensus is that the clade of Pinnipedia is a sister group to the Ursoidae, which includes the Ursidae (Rybczynski et al. 2009; Berta 2018). Fossils of possible pinnipedimorphs that predate the late Oligocene are rare. The fossils of Puijila darwini and Potamotherium miocenicum dating from the late Oligocene to early Miocene (24–21 Mya) may be sister groups to the semiaquatic stem Pinnipedia, which were a transitional link between early Arctoid carnivores and the more highly derived pinnipediformes such as Enaliarctos (Berta et al. 1989: Wang et al. 2005; Rybczynski et al. 2009). These otter-like animals probably swam quadrupedally using their webbed fore- and hindlimbs for propulsion, but they were more closely related to Ursidae whose ancestors at that time were smaller. Hence, stem Pinnipedia probably evolved monophyletically from Arctoid carnivores in the early Oligocene and went through an amphibious, otter-like transitional phase to become recognizable as Pinnipedia by the late Oligocene and early Miocene.

The Enaliarctidae (family Enaliarctidae, five species) from the late Oligocene to early Miocene (27–18 Mya) were a group of small (50–100 kg) stem Pinnipedia in the clade of Pinnipedimorpha, which originated in the eastern North Pacific (Berta et al. 1989; Deméré et al. 2003; Berta 2018; Fig. 2.1). Both sets of limbs were already modified as flippers, although the hindlimbs were longer than modern Pinnipedia because the tibias had not yet shortened (Fig. 2.9; Berta and Ray 1990). The spine was flexible with long transverse processes on the lumbar vertebrae for the attachment of the epaxial and hypaxial muscles associated with hindlimb propulsion (Bebej 2009). However, the size and shape of the scapulae indicate that forelimb propulsion also was used. Although Enaliarctidae could swim and walk, their mode of swimming may have been similar to that modern walruses, which use a combination of fore- and hindlimb propulsion. The skull had large orbits and nares but reduced olfactory bulbs. The dentition was transitional between the generalized heterodont pattern of arctoid carnivores with shearing carnassial teeth and the postcanine homodont dentition of modern Pinnipedia for catching and swallowing small prey. Fossils from later pinnipediformes such as Pteronarctos and Pacificotaria, which date from the early Miocene (19–15 Mya) from coastal Oregon, were more closely allied with modern Pinnipedia than with Enaliarctidae based on skull morphology (i.e., the maxilla of Pinnipedia makes a significant contribution to the orbital wall, which does not occur in terrestrial carnivores).

Fig. 2.9

(a) Reconstruction of the skeleton of Enaliarctos mealsi (shaded elements are missing), (b) artists’ rendition of the live animal. Estimated length 144–154 cm. (Reprinted and adapted with permission from Berta and Ray (1990))

The earliest Otariidae is Eotaria crypta (17.1–15 Mya) from southern California (Boessenecker and Churchill 2015). Other stem Otariidae include the middle-to-late Miocene Pithanotaria starri (11 Mya) and the large (100–300 kg) Thalassoleon (8–4 Mya). Thalassoleon macnallyae may have been a sister group to Callorhinus in the Pliocene. The Otariidae probably evolved in the North Pacific with the divergence of the Otariine (sea lions) and Arctocephaline (fur seals) clades in the late Miocene (~6 Mya) (Boessenecker and Churchill 2015).

The Phocoidea include the extinct Desmatophocidae and the extant Phocidae (Fig. 2.1). The family of Desmatophocidae diverged from Enaliarctidae in the early Miocene (~23 Mya), achieved maximum diversity in middle Miocene (18 Mya), and became extinct in early-late Miocene (13–11 Mya) (Berta 2018). They were distributed in coastal waters of the North Pacific. There were two genera: Desmatophoca (one species) and Allodesmus (three described species). The oldest known Phocoidea is Desmatophoca brachycephala from the early Miocene. There was controversy about whether the Desmatophocidae was more closely related to Otariidae or Phocidae as they have features of both. Desmatophoca brachycephala was originally ascribed to the Otariidae based on the morphology of the skull, even though there were general features that were convergent with Phocidae (Barnes 1987). However, Deméré et al. (2003) assigned it to the Phocoidea, which makes it a sister group of the Phocidae. Allodesmus is another Desmatophocidae that had features of both sea lions and seals and probably swam with the forelimbs, although not necessarily using pectoral oscillation like modern Otariidae (Bebej 2009). In addition, they may have been able to rotate their hindlimbs forward for walking on land, something not possible for Phocidae. Unfortunately, the fossil record for Phocidae is poor making it difficult to assign phylogenetic relationships. Current data indicate that they evolved in the North Pacific during the early Miocene from stem Phocoidea, with Phocidae and Desmatophocidae diverging before the early Miocene (~18 Mya) (Deméré et al. 2003). Subfamilies of Phocidae, the Monachinae and Phocinae, probably diverged several millions years later in the North Atlantic associated with long-distance dispersal, possibly through the Central American Seaway (Higdon et al. 2007). The transition to hindlimb propulsion remains uncertain and will require the discovery of more fossil Phocidae.

Odobenidae (walruses) monophyly is strongly supported, and molecular analysis places it in the Otarioidea despite earlier phylogenetic analysis that placed it in Phocoidea (Deméré et al. 2003; Arnason et al. 2006; Boessenecker and Churchill 2013; Berta 2018). Phylogenetic and stratigraphic data suggest that Odobenidae first evolved in the North Pacific sometime before the late-early Miocene (~18 Mya; Fig. 2.1). There are three subfamilies: Imagotarinae, Dusignathinae, and Odobeninae. (Berta 2018). Proneotherium repenningi is a stem Imagotarinae (~18 Mya) that had a mosaic dentition with both shearing carnassial teeth and molariform anterior premolars, a laterally flexible lumbar region, a broad and shortened femur with an elongated tibia, and a paddle-shaped foot (Mitchell 1968; Deméré and Berta 2001). Dusignathinae walruses had modest enlargement of the upper and lower canine teeth (Boessenecker and Churchill 2013). Odobeninae walruses included the primitive Aivukus from the late Miocene (~6.5 Mya). They had a slight increase in the length of the upper canines, which became increasingly longer in later species of male walruses, perhaps for social and sexual display-related behavior. Walruses may have migrated through the Central American seaway 8–5 Mya and dispersed through the North Atlantic Ocean. Further evolution took place in the Atlantic with enlargement of the canines into tusks. Odobenidae disappeared from the Pacific during the Pliocene and may have returned less than one million years later through the Arctic Ocean. However, another hypothesis is that Odobenidae first evolved in the North Pacific and then dispersed into the Arctic Ocean and North Atlantic during an interglacial event in the Pliocene or Pleistocene.

The taxonomy of crown Pinnipedia is still in flux as new genetic data become available. Rice (1998) recognizes 34 species and 18 subspecies, whereas Berta and Churchill (2012) recognize 34 species and 29 subspecies. The difference in the number of subspecies depends on, for example, the inclusion of an extinct species (Japanese sea lion, Zalophus japonicus) and recognition of three subspecies of walruses instead of two. If you do not count two extinct Pinnipedia (Caribbean monk seal and Japanese sea lion), then there are 33 species and at least 18 subspecies (Appendix 2). Regardless of the numbers, Pinnipedia evolved monophyletically from a common arctoid ancestor most closely related to the Ursidae (bears) in the late Oligocene (27–25 Mya). By this time, they were already recognizable as amphibious pinnipediformes.

2.4 Sea Otters

Similar to Pinnipedia, fossils of the carnivorous arctoid ancestors of sea otters can be traced to the Eocene (~45 Mya). However, whereas Pinnipedia are now thought to be a sister group of the Ursidae, sea otters are clearly in the family of Mustelidae.

Phylogenetic analysis resolved the Mustelidae (superorder Laurasiatheria) into seven primary divisions that include the otter clade (subfamily Lutrinae) and a sister clade comprising mink and true weasels (subfamily Mustelinae) (Koepfli et al. 2008). Stem Mustelidae first evolved during the Oligocene in Eurasia including the now extinct subfamily of Leptarctines, which spread to North America by the middle Miocene (Qiu and Schmidt-Kittler 1982). The Lutrinae, a sister group of the Leptarctines, appeared in the middle-late Miocene (~8 Mya) in Eurasia, and the ancestors of sea otters (Enhydra) diverged from other Eurasian otters in the early Pliocene (~5 Mya; Fig. 2.1). Two sister groups of modern sea otters occurred in the late Miocene to early Pliocene. One was Enhydriodon, which occurred in Eurasia and Africa. The other was Enhydritherium, which occurred in Eurasia and North America based on fossils from Spain and Florida, respectively, and reclassification of fossils from California originally assigned to Enhydriodon (Repenning 1976; Berta and Morgan 1985). One hypothesis for the distribution of Enhydritherium is that the European species (E. lleucai) spread around the northern rim of the Atlantic Ocean and into the Gulf of Mexico where the first fossils of E. terraenovae appear in Florida. From there, E. terraenovae dispersed into the Pacific through the Central American Seaway where presumably it gave rise to Enhydra, which is found only in the North Pacific Ocean. However, this dispersal would have occurred around the time that the Central American Seaway closed in the late Pliocene (2.76–2.54 Mya), so aspects of this scenario remain uncertain.

In size, E. terraenovae was similar to Enhydra (~22 kg; Lambert 1997). However, the paleoenvironment of the fossil sites in Florida and California indicate both nearshore marine and inland freshwater habitats instead of the strict marine habitats of Enhydra. The limbs of E. terraenovae show more similarity to river otters (Lutra and Lontra) than to Enhydra suggesting the forelimbs were used in aquatic (mode uncertain) and terrestrial locomotion (Lambert 1997). The thickened but heavily worn cusps on the carnassials of E. terraenovae indicate a diet of extremely hard food items such as molluscs similar to Enhydra, although soft foods such as fish may have also been consumed when in freshwater. Because E. terraenovae had attributes of both river otters and modern sea otters by the late Miocene to early Pliocene (5–4 Mya), it appears to be transitional and more of a habitat generalist than a marine specialist.

The alternative hypothesis is that Enhydra evolved in the North Atlantic (perhaps from Enhydritherium) and spread into the North Pacific through the Arctic Ocean and Bering Straits (Boessenecker 2016 and references therein). This hypothesis is supported by two fossil molars of Enhydra reevei that were found in strata dating from the early Pleistocene (2.2–1.7 Mya) in England (Willemsen 1992). The low, blunt, and inflated cusps of the teeth are more similar to those Enhydra than to Enhydritherium whose postcanine teeth have sharper ridges. Repenning (1983) provided further support for this hypothesis with a mandible of Enhydra sp. of similar age from the coast of the Chukchi Sea in northern Alaska. Unfortunately, the fossil record in the North Pacific is poor, so there is no evidence of Enhydra (e.g., Enhydra macrodonta) earlier than the middle Pleistocene (<0.7 Mya; Mitchell 1966; Kilmer 1972; Boessenecker 2016 revised the age of the Oregon femur in Leffler (1964) to 0.5–0.7 Mya, pers. com.). Hence, the dispersal of sea otters into the North Pacific Ocean during the intervening 1–1.4 million years remains uncertain. What is known is that Enhydritherium became extinct in the Atlantic and Enhydra became extinct in the Arctic, although the timing is uncertain. We will need additional fossil evidence to resolve the evolutionary origins of sea otters.

Today, there is one species of sea otter (Enhydra lutris) and three subspecies: (1) E. lutris lutris (northern Hokkaido to the Commander Islands in the western Pacific Ocean), (2) E. lutris kenyoni (Alaska and the Pacific west coast from the Aleutian Islands to northern Oregon), and (3) E. lutris nereis (coast of central California to northern Baja California and San Miguel Island). In many ways, subspecies designation is mostly a management tool because of the near extirpation of sea otters during the 18th and 19th century maritime fur trade, which fragmented the contiguous distribution from Japan to northern Baja California. Indeed, all or most of the otters that occur in Southeast Alaska, British Columbia, and Washington State are descended from otters that were translocated from the Aleutian Islands and Prince William Sound, Alaska in the 1960s and 1970s (Davis et al. 2019). Although the three subspecies vary in body size and in some skull and dental characteristics (Wilson et al. 1991), a phylogenetic analysis is warranted because sea otters will eventually reoccupy their pre-exploitation distribution in the North Pacific with overlap of the currently designated subspecies.

2.5 Evolution of Physiological Adaptations for Aquatic Life

As this chapter has shown, the evolutionary origins of extant marine mammals are becoming clearer through the stratocladistic analysis of fossil morphology dating as far back as the Eocene. The morphological changes that occurred as the terrestrial ancestors of marine mammals became increasingly adapted for an aquatic life are now well documented. However, there is no fossil record to document the evolution of physiological adaptations such as the dive response, enhanced body oxygen stores, tolerance to pressure, elevated resting metabolism, and thermal insulation. As with convergent morphological adaptations for an aquatic life (e.g., fusiform body shape, flukes, and flippers), convergent physiological adaptations in extant marine mammals indicate the strong selection pressure to function efficiently (i.e., energetically) and avoid the detrimental effects of pressure and enhanced heat loss because of the higher density of water compared with air (Table 1.1). We can only speculate on the course of events that led the ancestors of marine mammals to become aquatic, but food availability was a likely factor. However, we also have examples of living terrestrial mammals that enter the water to avoid predators such at the water chevrotain (mouse deer, Hyemoschus aquaticus). This small artiodactyl dives into a stream or pond and walks along the bottom to avoid large avian predators. Interestingly, the fossils of these small ruminants (family Tragulidae) date as far back as the Oligocene and are morphologically similar to Indohyus (Fig. 2.2a), a member of the extinct family of Raoellidae thought to be a sister group of early Archaeoceti. Whether searching for food or avoiding predators, all mammals stop breathing when submerged, and their ability to extend breath-hold duration, dive to deeper depths without experiencing the detrimental effects of pressure, and maintain a high and stable core body temperature through physiological adaptation extends available habitat and enhances biological fitness.

The dive response in marine mammals is the hallmark of physiological adaptation for breath-hold diving (see Sect. 6.1). It involves cardiovascular adjustments (bradycardia and peripheral vasoconstriction) that result in the efficient use of blood and muscle oxygen stores to maximize aerobic dive duration. The dive response is a modification of the primitive asphyxial response that prolongs life when oxygen absorption is interrupted and occurs, to varying degrees, in all vertebrates. Hence, it may have been an early adaptation for aquatic life in each of the marine mammal linages, other tetrapods (diving reptiles and birds), and extinct marine reptiles (e.g., ichthyosaurs, plesiosaurs, and pliosaurs).

Of all the adaptations that marine mammals exhibit for maintaining aerobic metabolism during dives, they never reacquired the ability to breathe water. As a result, they carry all of the oxygen that will be available during the dive in their blood, muscle, and lungs (see Sect. 6.3). Oxygen is bound to hemoglobin (Hb) and myoglobin (Mb) in the blood and muscle, respectively. Relative to terrestrial mammals, marine mammals have higher concentrations of Hb and Mb and a larger blood volume, so they have larger body oxygen stores, which enables longer aerobic dive durations, although this varies by species. Since Hb and Mb evolved over 550 million years ago (Pesce et al. 2002), increased concentrations of these globins in blood and muscle probably occurred early in ancestral marine mammals as an upregulation of globin expression.

When marine mammals dive to depth, the thorax and lungs are compressed resulting in potentially high partial pressures of pulmonary gases, especially nitrogen. The effects of pressure pose a potential risk of barotrauma, decompression sickness (DCS), and nitrogen narcosis, something that terrestrial mammals (except humans) never experience (see Sect. 3.6). Pressure can also affect cell membrane function resulting in high-pressure nervous syndrome (HPNS). Ancestral marine mammals were probably shallow divers initially. Nevertheless, as they evolved a dive response and enhanced body oxygen stores, breath-hold duration increased and the ability to dive deeper became possible. The pulmonary and neurological adaptations that protect marine mammals from barotrauma, DCS, nitrogen narcosis, and HPNS probably evolved after the dive response, and increased concentrations of Hb and Mb enabled marine mammals to hunt at greater depths, thereby making the mesopelagic zone (200–1000 m) accessible.

The high thermal conductivity of water posed a thermoregulatory challenge for the homeothermic ancestors of marine mammals. Unless they remained in a shallow, warm water habitat (which is still true for extant Sirenia), maintaining a high and stable core body temperature required elevated heat production (an upregulation of resting metabolic rate) and better insulation (primarily blubber but also waterproof fur in sea otters and fur seals) (see Sect. 4.2.1.4). Unfortunately, we have no indication as to when any of these adaptations occurred. However, the thermoregulatory challenges that living marine mammals face would have been similar for their early ancestors. As a result, these adaptations must have evolved early in the ancestral forms depending on geographic dispersal and prevailing water temperatures at the surface and at depth (viz., thermocline depth).

What is noteworthy about all of these adaptations is that they were modifications of pre-existing metabolic and physiological processes or morphology that may have allowed the relatively rapid evolution from a terrestrial to an aquatic life. Hence, these adaptations may have been present once the ancestors of each group of marine mammals (Cetacea, Sirenia, Pinnipedia, sea otters) had become entirely or largely aquatic, although they evolved separately in each ancestral group. For Cetacea, this could have been 48–40 Mya for Ambulocetus but very likely by 37–34 Mya for the Dorudontidae and Neoceti (Mysticeti and Odontoceti). Similar timing may have occurred for the Sirenia except for resting metabolic rate, which is much less (30%) than the allometric prediction in extant species, thereby necessitating warm water habitats or large body mass (e.g., Steller’s sea cow). For early Pinnipedia, the Enaliarctidae likely would have exhibited some or all of these physiological adaptations 27–25 Mya. Finally, early Eurasian otters may have exhibited some of these physiological adaptations, and Enhydritherium had attributes of both river otters and the modern sea otters ~5 Mya. In many cases, the ontogeny of marine mammals demonstrates critical aspects of their evolution from terrestrial to aquatic animals.

2.6 Summary

• 1. Convergent morphological adaptations for an aquatic life (e.g., fusiform body shape, flukes and flippers) evolved separately in the three taxonomic orders of extant marine mammals (Artiodactyla [infraorder Cetacea], Carnivora [clade Pinnipedia and sea otters], and Sirenia).

• 2. Cetacea evolved from primitive Artiodactyla in the early Eocene and therefore belong in the taxonomic order Artiodactyla.

• 3. Early Cetacea are classified as Archaeoceti (six families), and their fossils reveal the morphological and functional evolution of early, terrestrial Artiodactyla to fully aquatic Cetacea over 15 million years during the Eocene.

• 4. The Pakicetidae were the earliest Eocene (~50 Mya) Archaeoceti and primarily terrestrial with only a few morphological features that link them to modern Cetacea.

• 5. The Ambulocetidae (e.g., Ambulocetus natans) were an amphibious transitional phase of cetacean evolution (48.6–40.4 Mya) that probably swam using pelvic paddling but may have moved onshore like a sea lion.

• 6. The middle Eocene Protocetidae was the most basal family of Cetacea to be globally distributed. Early Protocetidae such as Rodhocetus still had four limbs with long hands and feet that were modified for pelvic paddling or undulation but still enabled movement on shore.

• 7. The Dorudontidae and Basilosauridae were the last Archaeoceti to complete the transition to a fully aquatic existence 38–34 Mya in the middle-to-late Eocene: the last walking whales. The shape of the caudal vertebrae indicates the presence of flukes (although there is no fossil evidence) and caudal oscillation for propulsion.

• 8. The evolution of modern Cetacea (Neoceti: Mysticeti and Odontoceti) began in the late Eocene (~37 Mya), and the sister group was likely the Dorudontidae.

• 9. Toward the end of the Eocene (34 MA), cetacean morphology for locomotion that had evolved over 15 million years remained more or less the same during subsequent evolutionary changes. Instead, the primary distinctions between the modern Mysticeti and Odontoceti are associated with morphological adaptions for feeding.

• 10. Late Eocene Dorudontidae, the likely sister group of early Mysticeti, had teeth to capture large or intermediate-sized prey. Because modern Mysticeti use baleen to filter large numbers of small prey, there was a transitional phase from a toothed ancestor to a mosaic intermediate with both teeth and baleen. The evolutionary loss of teeth finally occurred in archaic toothless Mysticeti in the late Oligocene.

• 11. In addition to the evolution of baleen in Mysticeti, other morphological adaptations included a highly arched upper jaw to accommodate the long baleen in Balaenidae and a highly elastic throat pouch that is pleated externally in Balaenopteridae.

• 12. Cladistic analysis including known Oligocene Cetacea fossils indicates a post-Oligocene radiation of crown Mysticeti that includes 14 extant species and 16 subspecies.

• 13. As with Mysticeti, late Eocene Dorudontidae were a likely sister group of early Odontoceti. However, they retained, to varying degrees, their adult dentition, although not necessarily for feeding. The facial plane of Odontoceti became concave to accommodate a melon, facial muscles, and nasal sacs for directional sound generation associated with echolocation.

• 14. The Xenorophidae and Squalodontidae were heterodont and polydont early Odontoceti whose fossils date from the late Oligocene (~25 Mya). The Kentriodontidae, whose fossils date from the late Oligocene to the Pliocene (25–5 Mya), were similar to small extant dolphins.

• 15. Cladistic analysis indicates a post-Oligocene radiation for crown Odontoceti, which include 74 extant species with an additional 34 subspecies.

• 16. The evolution of Sirenia began in the early Eocene (56–48 Mya). Prorastomidae were stem Sirenia dating from the early Eocene (~50 Mya) in Jamaica.

• 17. The Prorastomidae Pezosiren portelli from Jamaica was a pig-sized quadruped with a barrel-shaped trunk and short legs. It is considered transitional between terrestrial and aquatic locomotion.

• 18. The Protosirenidae from the middle Eocene (41–38 Mya) were still amphibious based on well-developed fore- and hindlimbs, but modes of locomotion in water and on land are uncertain.

• 19. Early Dugongidae first appeared in the middle Eocene (~41 Mya) and showed progressive reduction in the pelvis and hindlimbs. By the end of the Eocene (~34 Mya), Dugongidae had a fusiform body shape, vestigial pelvis, hindlimbs that did not protrude externally, and flukes for locomotion using caudal oscillation.

• 20. The Dugonginae were once more diverse with at least nine extinct genera from the late Oligocene to Pliocene but are now reduced to one species (Dugong dugon). The last member of the Hydrodamalinae, Steller’s sea cow, became extinct in the 1760s.

• 21. The Trichechidae may have been a sister group of the early Dugongidae or Protosirenidae in the early Oligocene (~28 Mya).

• 22. By the late Miocene to early Pliocene (6–5 Mya), Trichechidae had a continuous succession of horizontally replaced supernumerary teeth as an adaptation to an abrasive diet of siliceous grasses during the late Miocene.

• 23. The extant Trichechidae date from the Pleistocene and include three species (Amazonian manatee, West African manatee, and the West Indian manatee) and two subspecies of the West Indian manatee (Florida manatee and the Antillean or Caribbean manatee).

• 24. Fossils of the arctoid ancestors of Pinnipedia (fur seals, sea lions, walruses, seals) can be traced to the Eocene (~45 Mya), although fossil pinnipedimorphs only extend to the late Oligocene (27–25 Mya).

• 25. Morphological and molecular evidence support a monophyletic origin for the three extant families of Pinnipedia (Otariidae, Odobenidae, and Phocidae) within the taxonomic order Carnivora. The most likely sister group is the Ursoidae, which includes the Ursidae (bears).

• 26. The Enaliarctidae were a group of stem Pinnipedia, which originated in the eastern North Pacific during the late Oligocene (27–25 Mya). Both sets of limbs, which were modified as flippers, may have been used in aquatic locomotion.

• 27. The earliest Otariidae include the mid-to-late Miocene Pithanotaria starri (~11 Mya) and the large Thalassoleon (8–4 Mya).

• 28. The Otariidae probably evolved in the North Pacific by the middle Miocene (~11 Mya) with the divergence of the Otariine (sea lions) and Arctocephaline (fur seals) clades in the late Miocene (~6 Mya).

• 29. The Phocoidea include the extinct Desmatophocidae and the extant Phocidae, which were distributed in coastal waters of the North Pacific. The Desmatophocidae had features of both sea lions and seals and may have used forelimb propulsion.

• 30. Phocidae evolved in the North Pacific during the early Miocene from stem Phocoidea, with Phocidae and Desmatophocidae diverging before the early Miocene (~18 Mya). The transition to hindlimb propulsion remains uncertain.

• 31. Phylogenetic and stratigraphic data suggest that Odobenidae first evolved in the North Pacific sometime before the early Miocene (~18 Mya). Early Odobeninae walruses may have migrated through Central American seaway 5–8 Mya and dispersed through north Atlantic.

• 32. The taxonomy of extant crown Pinnipedia recognizes 33 species and at least 18 subspecies.

• 33. Fossils of the carnivorous arctoid ancestors of sea otters can be traced to the Eocene (~45 Mya) and belong to the family of Mustelidae and subfamily of Lutrinae (otter clade). The ancestors of sea otters (Enhydra) diverged from other Eurasian otters in the early Pliocene (~5 Mya).

• 34. The European otter Enhydritherium may have spread around the northern rim of the Atlantic Ocean and into the Gulf of Mexico where fossils first appear in Florida. From there, it dispersed into the Pacific Ocean through the Central American Seaway where presumably it gave rise to the Enhydra, which is found only in the North Pacific Ocean.

• 35. The earliest fossils of Enhydra sp. are from early Pleistocene (2.6–1.8 Mya) deposits along the coast of Oregon and California. Today, there is one species of sea otter (Enhydra lutris) and three subspecies.

• 36. There is no fossil record to document the evolution of the physiological adaptations that occur in living marine mammals. The dive response and enhanced oxygen stores, which increased the aerobic dive duration, and blubber for thermal insulation may have been early adaptations in each of the marine mammal linages. The exceptions are fur seals and the sea otter, which rely on waterproof fur. Adaptations to avoid the detrimental effects of pressure during deep diving may have come later.