otter, but very different in shape, with small feet and a long and power-
ful tail, which is flat from top to bottom. All those differences in feet and
tails relate to how these animals swim.
Minks, for instance, are a close terrestrial relative of otters, and are
probably similar to the ancestral otter from the time before they were
aquatic. Minks are land animals, but occasionally they do swim. Their
long, sleek body is well suited to dashing through the underbrush with-
out getting caught in branches. But minks are slow swimmers. They
paddle with all four feet;13 left fore and right rear beat at the same time,
as the animal struggles to keep its head above water in order to breathe.
River otters are different. In the zoo, one can see them dash unpredict-
ably left and right and up and down with a real or imagined playmate,
but that is not the kind of swimming that Frank can study. To measure
joint movements and swimming speed precisely, he needs to see the ani-
mals go in a straight line. River otters swim using different strokes,
depending on how fast they want to go.14 They move from quadrupedal
paddling and paddling with just the hind limbs (pelvic paddling), when
swimming at the surface, to dorsoventral undulation at faster speeds
underwater. The latter is the most efficient. Waves traveling through the
vertebral column propel mostly the tail, but also the hind feet. The sea
sea lion
seal
Pectoral
oscillation
odes, based
odern
odes of
Pelvic
oscillation
ing m
inda Spurlock.
otter shrew
mim
and into others, and
platypus
ings by L
ploying those m
wingro
nating pectoral
undulation
Lateral caudal
als em
e draw
m
ustelids (otters and their relatives)
Alter
Lateral pelvic
undulation
ostly otter relatives and m
am
odes evolved from
ode. M
ere used to infer the sw
usk rat
ing m
ilarly, m
m
odern m
polar bear
m
sim
ing m
im
m
im
im
wing
sw
ples of m
nating pectoral
paddling
nating pelvic
ro
how
Alter
Alter
als that sw
ith exam
s show
ore than one sw
als, w
rrow
ill be discussed in future chapters. Som
m
am
ploy m
ading
upedal
ventral
oxes contain anim
hales, and their body proportions w
ers em
Pelvic
paddling
caudal
m
Quadr
alking/w
undulation
odes in m
w
ventral
Dorso
Quadrupedal paddling
im
pelvic
odes. B
Dorso
undulation
Caudal
ing m
als by Frank Fish. A
e sw
oscillation
m
hales in this figure w
mink
m
ater w
im
am
Modern mammals
giant
otter
ing evolution of w
fresh
odern
m
hippo
ing m
river otter
im
al m
cetaceans
m
ploying those m
sea otter
im
e of the fossil w
als em
?
he evolution of sw
odern sw
. T
anim
0
lear bars indicate that som
odels for the sw
Fossil
icetids
whales
Basilosaurus
e 2
pak
orudon
Ambulocetus
D
ru ork on m
remingtonocetids
fig
on w
outlines show
cetaceans. C
are good m
extinct cetaceans . Som
56 | Chapter 4
figure 21. Skeleton of Ambulocetus, a forty-eight-million-year-old whale from
Pakistan, based on Thewissen et al. (1996). Soccer ball is 22 cm (8.5 inches) in
diameter.
otter swims underwater by moving its feet up and down, propelled by
sinuous movements of its body: pelvic undulation. The feet are enor-
mous and asymmetrical: they provide lift. Most interesting from a whale
perspective is that giant South American freshwater otter. It propels
itself with its long tail, which it swings through the water in an up-and-
down fashion: caudal undulation. Frank put it all together and pro-
posed that whales went through locomotor changes in their evolution
that are mirrored in the members of the modern otter group. And he did
this before any fossils documenting that transition were found.
That made the fossils the perfect way to check his results. If Frank
was right, then the locomotor skeleton of Ambulocetus should match
that of one of those otters. And indeed, Ambulocetus is proportionally
like a river otter.15 It is likely that the terrestrial ancestors of whales were
quadrupedal paddlers, since most land mammals swim that way. From
there, it is likely that swimming modes in whales changed a number of
times, going through stages represented by modern otters—alternating
pelvic paddling; simultaneous pelvic paddling and dorsoventral pelvic
undulation; caudal undulation—to finally end up as caudal oscillators.
Since that work, more fossil whales have been discovered. An Eocene
whale from India, Kutchicetus (discussed in detail in chapter 8), is
younger than Ambulocetus and has flat tail vertebrae
and short limbs,
suggesting that it was a caudal undulator.16 Other analyses elaborated
on this work. A complex mathematical analysis of whale skeletons geo-
logically younger than Ambulocetus confirmed that a hind-limb-domi-
nated phase of swimming preceded the tail-based phases.17 However,
the results of that study did not find a link to mustelids, probably
Learning to Swim | 57
Kala Chitta Hills
Chorlakki
Islamabad
Afghanistan
Kalakot
Simla
Sulaiman Range
Delhi
Pakistan
India
Raoellid artiodactyls
Pakicetid whales
Ambulocetid whales
Remingtonocetid whales
Indian
Protocetid whales
Ocean
Kutch
Basilosaurid whales
figure 22. Sites where fossils of Eocene whales and raoellid artiodactyls have been
found in Pakistan and western India. Pakicetid, ambulocetid, and remingtonocetid (see
chapter 8) whales, as well as raoellid artiodactyls (see chapter 14), are only known from
this part of the world.
because it did not include any data on the tail, which is probably of
importance because it is the propulsive organ of modern whales.
We already saw that the fluke of Dorudon indicates that it was a
caudal oscillator. Basilosaurus does not have an analogue among the
otters. Even though it retained the fluke of its ancestors, its vertebral
column was extremely flexible. It was probably an undulator,18 along
the lines of snakes and eels and different from any other cetacean.
That killer whale that looked at me as I entered its enclosure could
hardly be more different from the otters that Frank studied: screamingly
black and white, smooth and robotic, the size of small bus. However, when
making its rounds through the water, the differences blur. Both otters and
whales are perfectly at ease in water—gracious, fast, and acrobatic. The
up-and-down movement is apparent in both, even though one has a fluke
58 | Chapter 4
and the other big feet. Here, then, one can see the hidden connections that
evolution exposes. Modern otters’ swimming can teach us about swim-
ming evolution in whales. The present is the key to the past. And Ambu-
locetus formed the icing on the evolutionary cake, showing that the forms
predicted by inferences drawn from modern animals indeed existed.
ambulocetid whales
When Ambulocetus was discovered, it was the only whale known to
have limbs that could support the animal, unlike the rudimentary hind
limbs of basilosaurids and the internal hind limbs of modern cetaceans.
As a result, locomotion was the focus of the excitement about the new
species. However, the skeleton of Ambulocetus (figure 21) also repre-
sented an intermediate form in other respects, allowing us to study other
organ systems that changed as whales went from land mammals to obli-
gate marine swimmers. Ambulocetus is very different from all other fos-
sil or modern cetaceans, and is classified in its own family: Ambuloceti-
dae. Fewer than ten individuals of Ambulocetus natans have been
discovered, all from the Kala Chitta Hills in northern Pakistan. The fam-
ily includes two other genera, and both are from Pakistan and India
(figure 22). The first is Gandakasia, for which only a few teeth are known.
Those teeth were not recognized as being from a whale when they were
discovered at a site just a few miles from the Ambulocetus site.19 The
second is Himalayacetus, for which a single lower jaw was found in
the Indian Himalayas.20 Himalayacetus was thought to be the oldest
whale in the world, at 53.5 million years, but it appears that this dating
was based on associated fossils that washed in from older layers.21 It is
likely that all ambulocetids lived around forty-eight million years ago.
Nearly all of the specimens known for Ambulocetus are just frag-
ments, such as a single vertebra or a piece of jaw with a tooth. The only
specimen of Ambulocetus natans that tells us anything about the skele-
ton is the one initially found by Mr. Arif (figure 23). The size of the
skeleton indicates that Ambulocetus was about as large as a male sea
lion. Many bones are known for that specimen, but some important
parts are missing. For instance, the tip of the snout was never found,
and as a result we have to infer its length from the lower jaw (which was
found), and we do not know where the nose opening is.
For a present-day person to imagine Ambulocetus in its natural envi-
ronment, the best bet would be to travel to a coastal swamp in a hot
climate and study alligators (figure 19). Ambulocetus looked like a croc-
Learning to Swim | 59
figure 23. All known fossil bones of one individual of Ambulocetus natans (H-GSP
18507), with a hammer for scale.
odilian with its long snout, compact body, short forelimbs, and power-
ful, straight tail. The skin covering would be different— Ambulocetus
was a mammal, with hair (possibly a sparse coat), whereas reptiles have
scales. But just like alligators, Ambulocetus was probably an ambush
predator, too slow to pursue prey on land or in water but able to jump
on hapless prey that was close, either in the shallows, or at the water’s
edge getting ready to drink.
Feeding and Diet. With the tip of the snout missing, there is no way to
determine how many upper teeth there were, but we do know that there
were three upper molars, as can be expected for a basal placental mam-
mal. In the lower dentition, the incisors are lined up from front to back,
one canine, four premolars, and three molars. The lower molars of
Ambulocetus are much simpler than those of basilosaurids. Instead of a
row of cusps that decreases in height, Ambulocetus had a single high
cusp in the front of the tooth (the trigonid) and a single much lower
cusp in the back (the talonid). Analysis of the enamel of the teeth indi-
cates that it fed on animals, consistent with its crocodile-like looks.
Tooth wear in Ambulocetus is similar to that in other Eocene whales:
there are steep shear facets that indicate forceful tooth-to-tooth contact,
but not much wear caused by food that blunts the tips of the teeth. W
e
will get back to that in chapter 11.
60 | Chapter 4
Dog
Porpoise
Human
soft palate
epiglottis
epiglottis
soft palate
soft palate
epiglottis
larynx
Dog
Dolphin
Ambulocetus
hard palate
hard palate
hard palate
hyoid
hyoid
hyoid
figure 24. The paths of food (red) and air (blue) cross in the throat of mammals.
Top diagrams show sections through the midline plane. Note how the red arrow
passes to the side (laterally) of the blue one in all, but how different the relative
location of soft palate and epiglottis is. Bottom drawings show the same paths
superimposed on the skulls and hyoids of three mammals, showing the extension
of Ambulocetus’s palate to the back.
Swallowing
The most puzzling part of the skull of Ambulocetus is the area of
the throat. In most mammals, the bony part of the palate (the hard
palate) ends near the back of the teeth (figure 24). Behind that is
the soft palate, a wall of connective tissue and muscle that separates
the rear of the mouth (the oral cavity) from the rear of the nose (the
nasopharyngeal duct) just before both open into the throat.22 Food is
carried from mouth to the throat, and air is transported through the
nasopharyngeal duct to the throat. In humans, a little tissue flap hangs
from the back of the soft palate and is featured in many comic strips,
but most animals do not have it. The throat anatomy of Ambulocetus
is different from that of most mammals. The hard palate goes back
much beyond the teeth, all the way to the ears, and the nasopharyngeal
duct and hard palate flare down (ventrally), into the back of the oral
cavity. The soft palate does not fossilize, so we do not know about its
anatomy, but certainly mouth and nose were separated by bone much
farther back than in most mammals.
Areas deeper in the neck are also different. In most land mammals,
The Walking Whales Page 9
