Post by brobear on Apr 21, 2017 15:19:06 GMT -5
www.bearbiology.com/fileadmin/tpl/Downloads/URSUS/Vol_9/McLellan_Reiner_Vol_9.pdf
A REVIEW OF BEAR EVOLUTION
BRUCE MCLELLAN, Forest Sciences Research Branch, Revelstoke Forest District, RPO#3, Box 9158, Revelstoke, BC VOE 3KO
DAVID C. REINER, U.S. Fish and Wildlife Service, NS 312, University of Montana, Missoula, MT
Members of the bear family, Ursidae, currently
inhabit North America, Europe, Asia, and South
America. Using generic names suggested by Hall
(1981), Nowak and Paradiso (1983), Goldman et al.
(1989), and Wayne et al. (1989), species found in Asia
include the brown bear (Ursus arctos), Asiatic black
bear (U. thibetanus), sun bear (U. malayanus), sloth
bear (U. ursinus), and polar bear (U. maritimus). The
taxonomic status of the sixth Asian member, the giant
panda (Ailuropoda melanoleuca), remains in question,
although most evidence suggests that it belongs to
Ursidae. Brown bears and polar bears are found in
Europe and these 2 species plus the American black
bear (Ursus americanus) inhabit North America. The
spectacled bear (Tremarctos ornatus) is the only
member of Ursidae in South America.
Ursidae are placed in the order Carnivora but, except
for the largely carnivorous polar bear, bears are
omnivorous, feeding mostly on plant material, insects,
fish, and mammals. They are generally large, stocky,
and powerful animals. All bears are plantigrade,
walking on their entire foot. The radius and ulna and
the tibia and fibula are separate, which enables bears to
rotate their limbs, improving their ability to dig and
manipulate food, and facilitating climbing by some
species. Bears' teeth reflect their omnivorous habits by
lacking the carnassials common in other mammalian
carnivores and having flattened molars adapted to
crushing and grinding vegetation. Bears' premolars are
reduced in size and utility, creating a gap between
incisors and molars similar to that found in many
herbivorous mammals.
Bears are a young family, evolving from early canids
during the late Oligocene and early Miocene, about 20-
25 million years before present (MYBP). So recent is
this divergence that some taxonomists believe that
canids and ursids should be considered as one family
and dividing them is due to "custom of more than a
century" (Simpson 1945). The majority of students,
however, have separated these 2 families but frequently
disagree on where the line between canids and ursids,
and many other taxonomic boundaries, should be
drawn. Recently, the families Ursidae and Otariidae
have been placed in the superfamily Ursoidea. These
2 families have been joined with members of the
Canoidea superfamily, Canidae, Procyonidae,
Mustelidae, and Phocidae, into the suborder Caniformia
(Wozencraft 1989).
The purpose of this paper is not to create another
view of bear evolution and resulting taxonomy. Rather,
it is to summarize for the interest of biologists who
work with extant species the large body of knowledge
that has grown over the past century but is scattered in
journals and papers that many of us rarely encounter.
SUBFAMILIES OF URSIDAE
Although 2 major contributors to bear taxonomy,
Simpson (1945) and Erdbrink (1953), did not favor
subfamily divisions as suggested by Kragavlich (1926),
most systematicians divide the bear family into 3
(Kragavlich 1926, Kurten 1966) or 4 (Pilgrim 1932,
Thenius 1959) subfamilies without including
Ailuropodidae, the subfamily that includes the giant
panda. These authors generally disagree over the
inclusion of the subfamily Hemicyoninae, or dog-like-bears
(or bear-like-dogs), with either the canids or
ursids. Hendey (1980) splits the bears into 5
subfamilies and 7 tribes; his groups include different
genera than other students. We will include the giant
panda as a bear and discuss 5 subfamilies (Fig. 1):
(1) Hemicyoninae, (2) Agriotheriinae, (3) Tremarctinae,
(4) Ursinae, and (5) Ailuropodidae.
SUBFAMILY HEMICYONINAE
It is believed that bears evolved from the canid line
during the late Oligocene and early Miocene (Kurten
1966). The change from canids to ursids left a fossil
record relatively rich with intermediate genera and this
has led to various opinions on where to differentiate the
2 families. Frick (1926) separated some intermediate
forms from both the canids and ursids by creating the
family Hemicyoninae, in which he included the genera
Hemicyon, Dinocyon, Hyaenarctos, and Ursavus (since
Matthew [1929] Hyaenarctos is considered to be
Agriotherium). Frick (1926) specifically refrained from
including this new family in the Ursidae, although
Mitchell and Tedford (1973) thought that he had
presented enough evidence that suggested it did belong
with bears. Pilgrim (1932) placed Frick's family,
Hemicyoninae, plus earlier genera, Amphicyon and
Cephalogale, into Ursidae. This general division
between canids and ursids was favored by many authors
(Thenius 1959, Hendey 1972, Mitchell and Tedford
1973) although Kurten (1966) placed this group with
the canids and considers the genus Ursavus to be the
first bears.
Several students (Erdbrink 1953, Kurten 1966,
Mitchell and Tedford 1973, Thenius 1979) suggested
that the evolutionary line between the canid subfamily
Amphicynodontinae and the ursid subfamily
Hemicyoninae was through the genera Cephalogale and
Ursavus (Fig. 2).
Members of Hemicyoninae were relatively small
during their early history with Cephalogale being about
the size of a raccoon. As was often the trend with
Ursidae, however, they increased in size with time and
some became the size of the largest modem bears.
Hemicyoninae were largely carnivorous, but it appears
that Cephalogale became increasing omnivorous, which
is why it is considered to be the ancestor of all ursids.
Cephalogale occurred in Eurasia from the late
Oligocene and North America from the early Miocene
(Tedford et al. 1987). The much larger Hemicyon was
confined to Eurasia during its early history but became
very successful and spread to North America in the
mid-Miocene (Hendey 1972, Tedford et al. 1987).
Scott (1937) believed the bear-dogs were the dominant
canid type in North America during the late Miocene
and Pliocene. The extinction of Hemicyoninae has
been related to the radiation of a more advanced
subfamily of bears, Agriotheriinae (Hendey 1972,
Kurten 1971), or possibly due to competition with large
felids (Lydekker 1883, Frick 1926).
SUBFAMILY AGRIOTHERIINAE
The ursid subfamily Agriotheriinae as described by
Thenius (1959) and Kurten (1966) include the genera
Ursavus, Indarctos, and Agriotherium. The genus
Ursavus, which is believed to have evolved in Europe
from Cephalogale, appears to have given rise to its own
subfamily Agriotheriinae plus the subfamilies
Tremarctinae, Ursinae, and Ailuropodidae (Fig. 3).
Ursavus elmensis, which was about the size of a fox, is
thought to be the most primitive species of this genus
(Crusafont and Kurten 1976). It existed during the
early Miocene, when the climate in Europe was
relatively stable and sub-tropical.
From the small, primitive U. elmensis, 2 larger
species appear to have evolved, U. primaevus and U.
brevirhinus (Crusafont and Kurten 1976). Crusafont
and Kurten believe that these 2 species were sympatric
in Europe for a relatively long period and therefore
deserve specific distinction, which Stromer (1940) and
Thenius (1949, cited from Crusafont and Kurten 1976)
did not give them. From U. primaevus, Crusafont and
Kurten (1976) suggest that 2 larger species, U.
ehrenbergi and U. depereti evolved. Thenius (1949)
mentioned that U. ehrenbergi may have been close to
the line leading to the subfamily Ursinae, though
Crusafont and Kurten (1976) propose that the ursine
bears probably arose from a line closer to the more
primitive U. brevirhinus. Members of the genera
Ursavus were found in Eurasia for over 10 million
years. It appears that Ursavus, perhaps the debated
species U. pawniensis, was also in North America from
the early to mid-Miocene (Tedford et al. 1987).
Hendey (1980) suggests the genus Ursavus should be
their own subfamily, Ursavinae.
The transition from the genus Ursavus to Indarctos
during the early Pliocene appears to follow an increase
in body size (Hendey 1972). Crusafont and Kurten
(1976) proposed that the relatively lightly built
Indarctos vireti was the most primitive species of this
genus. The various species of Indarctos generally
continued to increase in size and spread from Europe
and Asia to North America, where the remains of I.
oregonensis, a very large bear, was discovered in
Oregon (Merriam et al. 1916) and Nebraska (Shultz and
Martin 1975).
For many years there has been disagreement on the
relationships of the third genus of the subfamily,
Agriotherium. Recently, Thenius (1959) and Hendey
(1972 and 1980) conclude that Agriotherium evolved
from Indarctos and may even have been congeneric
(Hendey 1980). Dalquest (1986), however, believes
that the 2 are not closely related. The long-standing
confusion over the evolutionary direction of the 2
species may be due to Agriotherium being more
carnivorous than Indarctos, which is against the general
trend of the Ursidae (Hendey 1980).
Remains of Agriotherium have been found in many
parts of the world including Europe, Iran, India, South
Africa, and North America (Hendey 1972). In North
America, its known range extended from California to
Florida and from Nebraska to southern Mexico
(Dalquest 1986). In North America Agriotherium
became larger than any extant species of bear (Scott
1937, Shultz and Martin 1975).
The extinction of Agriotherium due to competition
with early Ursinae, as was a probable cause of
extinction for other genera of Agriotheriinae, appears
less plausible because Agriotherium was thought to be
largely carnivorous. Competition with other carnivores
may have been involved (Hendey 1972). Many
carnivores had difficulties during the general extinction
of the late Hemphillian near the Miocene/Pliocene
boundary, when 60 mammalian genera disappeared
from North America (Martin 1984). The last known
representative of this subfamily, A. insigne, disappeared
after the Villafranchian in Europe (Kurten 1968) as the
climate began gradual cooling and warming oscillations,
foreshadowing the ice age.
SUBFAMILY TREMARCTINAE
The subfamily Tremarctinae includes the genera
Plionarctos, Arctodus, and Tremarctos (Thenius 1959,
Kurten 1966). There is 1 extant species, Tremarctos
ornatus of the South American Andes. Although the
fossil evidence leading to Tremarctinae is poor,
paleontologists suggest that Ursavus is likely the
ancestral genus (Thenius 1959, 1976; Kurten 1966).
Cytological and molecular methods indicate that
T. ornatus diverged from the genus Ursus 10.5-15
MYBP (O'Brien et al. 1985, Goldman et al. 1989),
when Ursavus sp. were common in Eurasia and
apparently present in North America. Although there
are several morphological and biochemical differences
between Tremarctinae and Ursidae, including a
different number of chromosomes (2n is 74 in Ursus
and 52 in Tremarctos), the fact that T. ornatus and U.
thibetanus have crossed in captivity questions placing
Tremarctos in its own subfamily (Mondolfi 1983).
The earliest Tremarctinae is Plionarctos and was
found in the upper Pliocene of California (Frick 1926)
although earlier roots in Asia are suspected (Kurten and
Anderson 1980). This genus is likely the ancestor of
the 2 other genera of the subfamily, Tremarctos and
Arctodus (Fig. 4). These 2 genera made their first
appearance in the Pleistocene of North and South
America (Kurten 1966).
The early history of both Arctodus and Tremarctos is
poorly recorded in the fossil record. Early students
such as Merriam et al. (1916) and Frick (1926) and
more recently Erdbrink (1953), believed Arctodus
evolved from a line close to Indarctos. Thenius (1959)
and Kurten (1966), however, believe that Arctodus was
a Tremarctinae.
Of the 5 species of Arctodus, A. pristinus,and A.
simus were in North America and A. bonariensis, A.
pamparus, and A. brasiliensis were found in South
America. Kurten (1967) suggested A. pristinus may
represent a relatively primitive form. It was smaller,
more lightly built, and probably less carnivorous than
the other species and so far has only been found in the
southeastern portion of North America.
More is known of A. simus, the great short-faced
bear, than other Arctoid bears, because the fossil record
is extensive in area, covering most of North America,
and relatively complete. A. simus was a very large
bear, with some individuals weighing at least 650 kg
(Emslie and Czaplewski 1985). This bear had long legs
and stood about 2 m at the shoulder, which suggests an
adaptation for fast movement. Kurten (1967) thought
that A. simus was not truly a cursorial predator, but
may have been capable of bursts of speed exceeding
those of U. arctos. Kurten (1967) suggested that its
short but wide jaws demonstrated convergence with the
great cats and indicated that it was largely carnivorous.
According to Kurten's interpretation, A. simus was by
far the most powerful predator during the Pleistocene
and possibly preyed on contemporary species of bison,
deer, horse, and ground sloth. Stirling and Derocher
(1990) and McLellan (1993) suggested that co-existence
with A. simus for or over 1 million years inflicted
significant constraints on the evolution of Ursus
americanus.
Emslie and Czaplewski (1985) disagree with the
conclusions of Kurten (1967). Based on characteristics
of the skull, body size, and relative lengths of distal and
proximal limb segments, Emslie and Czaplewski (1985)
suggest that A. simus was largely herbivorous.
A. simus disappeared at the end of the Wisconsin
glaciation, perhaps due to competition with Ursus arctos
(Kurten and Anderson 1974). These 2 bears, however,
apparently co-existed in Beringia for about 100,000
years, so if competition was the leading factor, other
conditions must have changed. The close proximity in
time between the disappearance of A. simus and the late
Pleistocene extinction, when 57 of the 79 species of
large mammal ( > 45 kg) in North America disappeared
(Marten 1984), suggests a correlation, particularly if A.
simus was largely carnivorous and much of its prey
disappeared. The peak of this extinction was about
11,000 YBP (Martin 1984) whereas the last A. simus
remains to be dated were more than 1,000 years older.
This date, however, was obtained on an Equus bone
found at the deepest level of a cave (Kurten and
Anderson 1980) and the bear was likely younger,
perhaps near the 10,000 years ago suggested by
Harrington (1973). Two competing hypotheses
explaining the late Pleistocene extinction, which
perhaps led to the extinction of A. simus, are overkill
by Paleoindians of the Clovis culture, and climatic
change that included a very dry period and a resulting
reduction of habitat diversity.
Relatively little is known of Arctodus in South
America because few fossils have been found. A.
brasiliensis was the smallest of the 3 species and, as it
more closely resembled the North American species,
may have been an intermediate. A. pamparus was also
relatively small, whereas A. bonariensis was very large,
rivaling A. simus in size (Kurten 1967). A. bonariensis
had large canines and carnassials but short posterior
molars, suggesting a carnivorous diet. Based on the
structure of their molars, Kurten (1967) suggested the
possibility of a mollusk-eating specialization for the
other South American species.
The genus Tremarctos consists of 2 species,
Tremarctos floridanus and the extant spectacled bear,
T. ornatus. As was the case with Arctodus, the
ancestral genus to Tremarctos is believed to be
Plionarctos (Kurten 1966; Thenius 1959, 1976).
Fossils of the North American spectacled bear have
been found most often in Florida and only rarely
elsewhere (Kurten and Anderson 1980). T. floridanus
seems to have been a slow-moving, heavily built,
medium-sized bear with powerful forelimbs. Kurten
(1966) suggested that T. floridanus filled a niche similar
to that of the European cave bear, Ursus spelaeus, as a
powerful, almost exclusively vegetarian bear. Reasons
for the extinction of T. floridanus in the last 8,000
years are unclear, although competition with U.
americanus has been suggested. These species, or at
least early versions, co-existed in North America for
about 3 million years, so if competition with U.
americanus was the cause of extinction, an additional
change, such as climate, must have precipitated it.
SUBFAMILY URSINAE
The subfamily Ursinae has been divided into many
different phylogenetic groups in the past. Until
recently, 5 genera, Melursus (sloth bear), Helarctos
(sun bear), Thalarctos (polar bear), Selenarctos (Asiatic
black bear), and Ursus (brown bear and American black
bear) were recognized. Molecular and cytological
methods (O'Brien et al. 1985, Goldman et al. 1989)
plus successful crossing between several of the species
in captivity (C. Servheen, U.S. Fish and Wildlife
Service, pers. commun.) suggests that these bears are
congeneric.
The evolution of Ursinae over the past 5 million
years is well recorded in fossils of Europe. Early
Ursinae likely evolved from Ursavus of the Miocene,
perhaps through the genus Protursus of the midMiocene
(Thenius 1959, Crusafont and Kurten 1976;
Fig. 5). Climatic conditions in Europe during the late
Miocene were dry, and savannahs and deserts were
common. Such conditions were poor for bears, and
their fossils are scarce until the Pliocene began, 5-6
MYBP.
Black Bears
Next to Protursus, the earliest member of the
subfamily Ursinae is believed to be Ursus minimus,
which has been found in many locations in Europe
(Kurten 1968). U. minimus was a small, primitive
species, that generally increased in size during its
existence. It appears that U. minimus or a species
similar to it radiated through Asia and was in North
America at least by the early Blancan, perhaps 3.5
MYBP (Kurten and Anderson 1980). In North
America, this species is called U. abstrusus, but may be
conspecific with U. minimus. This small early ursid
likely gave rise to the Asiatic and American black
bears. The timing of the divergence of U. americanus
estimated from fossils is similar to the 4.4 MYBP
derived through 2-dimensional electrophoresis
(Goldman et al. 1989) and the 3.8 MYBP estimated
from mitochondrial DNA divergence (Shields and
Kocher 1991).
U. thibetanus ranged into Europe during in the midPleistocene
with remains being found in many countries
(Kurten 1968). Why it was extirpated from Europe is
unknown, but competition with the largely herbivorous
cave bears may have been a factor. In North America,
black bears are by far the most common fossil bears of
the Pleistocene and have been found across most of the
continent. As was the case with many species, latePleistocene
black bears were much larger than they are
today. Behavioral and morphological characteristics
imposed on black bears by other ursids have been
discussed by Herrero (1972), Stirling and Derocher
(1990), and McLellan (1993).
Cave Bears
The small, primitive U. minimus gave rise to
U. etruscus, which was also initially small but
continued the trend towards a larger body size. This
species radiated across Eurasia. In Europe, it gave rise
to the cave bears and in Asia it was ancestral to brown
bears.
Caves are a good environment for fossilization.
Bears, with large, stocky bones, are especially well
preserved. Thus cave bears, which often died in caves,
have one of the best fossil records. The evolutionary
lineage is so complete that delineating species has been
difficult (Kurten 1968). In addition, the tens of
thousands of individuals represented (an estimated
30,000-50,000 in 1 cave) has enabled the typical study
of phylogeny and morphology plus studies of age and
sex class structure, individual variation, and mortality
rates.
Thenius (1959) recognizes 2 species of cave bear: U.
deningeri and U. spelaeus, the giant cave bear of
Europe. Kurten (1968) identifies another species, U.
savini, between the etruscan bear and U. deningeri.
Based on fossils, U. spelaeus appears to have been a
large, stocky, mostly herbivorous bear. It had a
relatively small geographic distribution, being found
only in Europe and into the southwestern corner of
Russia and the Ukraine. Such a small distribution
suggests a dietary specialization or perhaps a
dependence on caves or mountainous country where
caves are found (Kurten 1976). Remains of U.
spelaeus have not been found where caves are
uncommon.
Many possible causes for the extinction of cave bears
have been proposed. Kurten (1958) suggested that
rapidly increasing numbers of humans may have settled
caves during summer dispersal periods and thus
excluded bears returning to hibernate in the early
winter. Extant brown and black bears often hibernate
in caves much too small for human occupation, so if
this hypothesis is correct, then cave bears would have
had different requirements than the extant species, or
small caves were rare in their range. Competition with
increasing numbers of humans and brown bears for
caves and other resources plus climate change appear to
be likely factors.
Brown Bears
The brown bear is believed to have evolved from
U. etruscus in Asia. The oldest fossils were found in
China from about 0.5 MYBP (Kurten 1968) and there
has been a continuous record of U. arctos in Asia since
then. U. arctos entered Europe about 0.25 MYBP and
North Africa shortly after. Pleistocene remains of
U. arctos are common in Great Britain and they may
have contributed to the extirpation of the cave bear
there.
U. arctos apparently entered Alaska about 100,000
YBP but did not move south until the late Wisconsin,
about 13,000 YBP. Kurten and Anderson (1980)
suggest the possibility of 2 independent migrations;
narrow-skulled bears from northern Siberia through
central Alaska to the rest of the continent becoming
U. a. horribilis, and a southern migration of broadskulled
bears from Kamchatka to the Alaskan peninsula
becoming U. a. middendorffi. Fossils of brown bears
in Ontario, Ohio, Kentucky (Guilday 1968), and
Labrador (Spiess and Cox 1976) indicate they were
once found much farther east than historical records
show. Guilday (1968) suggested that immediately after
the glacial retreat, a relatively boreal, parkland
coniferous forest spread across the central and southern
portions of the continent and with it, several western
species, including brown bears.
Polar Bear
The Polar bear is a recent offshoot of U. arctos.
Indications of a recent divergence include the rarity of
fully fossilized polar bear remains (Kurten 1964)
whereas subfossils are common, and that these species
produce fertile hybrids in captivity (Kowalska 1965).
Mitochondrial DNA divergence (Shields and Kocher
1991) and 2-dimensional electrophoresis (Goldman et
al. 1989) also suggest a recent split. Polar bears are
repeating the trend that was seen with Agriotherium and
Arctodus by becoming carnivores; this time specializing
on marine mammals. The apparent morphological and
behavioral differences between polar bears and brown
bears indicate that polar bears are rapidly evolving as
they exploit a new niche.
Sun and Sloth Bears
The fossil records of south Asian bears, the sun
(Ursus malayanus) and sloth bears (U. ursinus) are
poor, and their origins more speculative than for other
species (Kurten 1966). U. malayanus is first found in
the late Pliocene and U. ursinus in the Pleistocene.
Thenius (1959) thought they separated from the other
Ursinae even before Protursus, whereas Hendey (1972)
speculated that the split was after Protursus but before
U. minimus. Electrophoretic analysis indicates a more
recent split, not significantly different from that of other
extant members of the subfamily except the polar bear
(Goldman et al. 1989). Recent analyses of
mitochondrial DNA suggest that the 6 ursine species
originated sequentially during the past 6 million years,
beginning with U. ursinus and ending with the polar
bear (Waits et al. unpubl. data, submitted). It is
becoming increasingly evident that U. ursinus,
malayanus, thibetanus, americanus, and etruscus all
branched from the primitive U. minimus or U.
abstrusus that radiated through Eurasia near the
Miocene/Pliocene boundary and into North America
shortly after. The great morphological differences
between U. malayanus, U. ursinus, and other bears is
likely due to recent adaptive change as the south Asian
bears exploited new niches.
SUBFAMILY AILUROPODIDAE
The phylogeny of the giant panda has been disputed
since 1870, when Milne-Edwards placed Ailuropoda
into the family Procyonidae, while David had called it
an ursid the previous year (O'Brien et al. 1985). Some
recent authors (Tagle et al. 1986) linked Ailuropoda
closer to the lesser panda (Ailurus fulgens) than to the
bears, but there has been an overwhelming number of
papers placing Ailuropoda not close, but closer to bears
than to the lesser panda (Kurten 1985, O'Brien et al.
1985, Mayr 1986, Ramsay and Dunbrack 1987,
Goldman et al. 1989). These papers were based on 6
independent molecular and genetic measures, fossil
evidence, and reproductive characteristics. These
recent reports plus the comparative anatomical work of
Davis (1964), earlier protein evolution work of Sarich
(1973), and the synthesis of Thenius (1979) indicate the
giant panda should be placed into its own subfamily,
Ailuropodidae, of the family Ursidae.
The earliest evidence of Ailuropoda was during the
late Pliocene, about 3 MYBP (Schaller et al. 1985).
Wang (1974) divides Ailuropoda into 2 species: A.
microta was a smaller, primitive species that became
extinct during the mid-Pleistocene. A. melanoleuca was
once larger than it is today and then ranged south of the
Yangtze river at least to Burma. The decrease in the
panda's range has been attributed to climatic changes
during the Pleistocene and, like almost all extant
species, man's activities during the postglacial (Schaller
et al. 1985)
The fossil record leading to Ailuropoda is poor and
evolutionary links speculative at best. Matthew and
Granger (1923), Davis (1964), Hendey (1972), and
Wolff (1978) believed the panda evolved from Indarctos
of the subfamily Agriotheriinae. Thenius (1979)
however, identified a possible ancestor, Agriarctos,
from the late Miocene of Hungary, and suggested it was
a descendant of Ursavus from the mid-Miocene (Fig.
6). After re-evaluation, Hendey (1980) concluded that
Ursavus depereti was more likely to be the ancestor of
the pandas than was Indarctos. The divergence
between bears and the giant panda has been estimated
to be 18-22 MYBP (Goldman et al. 1989), and thus,
separation from an early Ursavus, either via Indarctos
or not, is possible. The panda has a specialized niche
and, like other species that diverged from the
omnivorous trend, likely went through a period of rapid
evolution, which accounts for their morphological and
behavioral differences from other bears.
A REVIEW OF BEAR EVOLUTION
BRUCE MCLELLAN, Forest Sciences Research Branch, Revelstoke Forest District, RPO#3, Box 9158, Revelstoke, BC VOE 3KO
DAVID C. REINER, U.S. Fish and Wildlife Service, NS 312, University of Montana, Missoula, MT
Members of the bear family, Ursidae, currently
inhabit North America, Europe, Asia, and South
America. Using generic names suggested by Hall
(1981), Nowak and Paradiso (1983), Goldman et al.
(1989), and Wayne et al. (1989), species found in Asia
include the brown bear (Ursus arctos), Asiatic black
bear (U. thibetanus), sun bear (U. malayanus), sloth
bear (U. ursinus), and polar bear (U. maritimus). The
taxonomic status of the sixth Asian member, the giant
panda (Ailuropoda melanoleuca), remains in question,
although most evidence suggests that it belongs to
Ursidae. Brown bears and polar bears are found in
Europe and these 2 species plus the American black
bear (Ursus americanus) inhabit North America. The
spectacled bear (Tremarctos ornatus) is the only
member of Ursidae in South America.
Ursidae are placed in the order Carnivora but, except
for the largely carnivorous polar bear, bears are
omnivorous, feeding mostly on plant material, insects,
fish, and mammals. They are generally large, stocky,
and powerful animals. All bears are plantigrade,
walking on their entire foot. The radius and ulna and
the tibia and fibula are separate, which enables bears to
rotate their limbs, improving their ability to dig and
manipulate food, and facilitating climbing by some
species. Bears' teeth reflect their omnivorous habits by
lacking the carnassials common in other mammalian
carnivores and having flattened molars adapted to
crushing and grinding vegetation. Bears' premolars are
reduced in size and utility, creating a gap between
incisors and molars similar to that found in many
herbivorous mammals.
Bears are a young family, evolving from early canids
during the late Oligocene and early Miocene, about 20-
25 million years before present (MYBP). So recent is
this divergence that some taxonomists believe that
canids and ursids should be considered as one family
and dividing them is due to "custom of more than a
century" (Simpson 1945). The majority of students,
however, have separated these 2 families but frequently
disagree on where the line between canids and ursids,
and many other taxonomic boundaries, should be
drawn. Recently, the families Ursidae and Otariidae
have been placed in the superfamily Ursoidea. These
2 families have been joined with members of the
Canoidea superfamily, Canidae, Procyonidae,
Mustelidae, and Phocidae, into the suborder Caniformia
(Wozencraft 1989).
The purpose of this paper is not to create another
view of bear evolution and resulting taxonomy. Rather,
it is to summarize for the interest of biologists who
work with extant species the large body of knowledge
that has grown over the past century but is scattered in
journals and papers that many of us rarely encounter.
SUBFAMILIES OF URSIDAE
Although 2 major contributors to bear taxonomy,
Simpson (1945) and Erdbrink (1953), did not favor
subfamily divisions as suggested by Kragavlich (1926),
most systematicians divide the bear family into 3
(Kragavlich 1926, Kurten 1966) or 4 (Pilgrim 1932,
Thenius 1959) subfamilies without including
Ailuropodidae, the subfamily that includes the giant
panda. These authors generally disagree over the
inclusion of the subfamily Hemicyoninae, or dog-like-bears
(or bear-like-dogs), with either the canids or
ursids. Hendey (1980) splits the bears into 5
subfamilies and 7 tribes; his groups include different
genera than other students. We will include the giant
panda as a bear and discuss 5 subfamilies (Fig. 1):
(1) Hemicyoninae, (2) Agriotheriinae, (3) Tremarctinae,
(4) Ursinae, and (5) Ailuropodidae.
SUBFAMILY HEMICYONINAE
It is believed that bears evolved from the canid line
during the late Oligocene and early Miocene (Kurten
1966). The change from canids to ursids left a fossil
record relatively rich with intermediate genera and this
has led to various opinions on where to differentiate the
2 families. Frick (1926) separated some intermediate
forms from both the canids and ursids by creating the
family Hemicyoninae, in which he included the genera
Hemicyon, Dinocyon, Hyaenarctos, and Ursavus (since
Matthew [1929] Hyaenarctos is considered to be
Agriotherium). Frick (1926) specifically refrained from
including this new family in the Ursidae, although
Mitchell and Tedford (1973) thought that he had
presented enough evidence that suggested it did belong
with bears. Pilgrim (1932) placed Frick's family,
Hemicyoninae, plus earlier genera, Amphicyon and
Cephalogale, into Ursidae. This general division
between canids and ursids was favored by many authors
(Thenius 1959, Hendey 1972, Mitchell and Tedford
1973) although Kurten (1966) placed this group with
the canids and considers the genus Ursavus to be the
first bears.
Several students (Erdbrink 1953, Kurten 1966,
Mitchell and Tedford 1973, Thenius 1979) suggested
that the evolutionary line between the canid subfamily
Amphicynodontinae and the ursid subfamily
Hemicyoninae was through the genera Cephalogale and
Ursavus (Fig. 2).
Members of Hemicyoninae were relatively small
during their early history with Cephalogale being about
the size of a raccoon. As was often the trend with
Ursidae, however, they increased in size with time and
some became the size of the largest modem bears.
Hemicyoninae were largely carnivorous, but it appears
that Cephalogale became increasing omnivorous, which
is why it is considered to be the ancestor of all ursids.
Cephalogale occurred in Eurasia from the late
Oligocene and North America from the early Miocene
(Tedford et al. 1987). The much larger Hemicyon was
confined to Eurasia during its early history but became
very successful and spread to North America in the
mid-Miocene (Hendey 1972, Tedford et al. 1987).
Scott (1937) believed the bear-dogs were the dominant
canid type in North America during the late Miocene
and Pliocene. The extinction of Hemicyoninae has
been related to the radiation of a more advanced
subfamily of bears, Agriotheriinae (Hendey 1972,
Kurten 1971), or possibly due to competition with large
felids (Lydekker 1883, Frick 1926).
SUBFAMILY AGRIOTHERIINAE
The ursid subfamily Agriotheriinae as described by
Thenius (1959) and Kurten (1966) include the genera
Ursavus, Indarctos, and Agriotherium. The genus
Ursavus, which is believed to have evolved in Europe
from Cephalogale, appears to have given rise to its own
subfamily Agriotheriinae plus the subfamilies
Tremarctinae, Ursinae, and Ailuropodidae (Fig. 3).
Ursavus elmensis, which was about the size of a fox, is
thought to be the most primitive species of this genus
(Crusafont and Kurten 1976). It existed during the
early Miocene, when the climate in Europe was
relatively stable and sub-tropical.
From the small, primitive U. elmensis, 2 larger
species appear to have evolved, U. primaevus and U.
brevirhinus (Crusafont and Kurten 1976). Crusafont
and Kurten believe that these 2 species were sympatric
in Europe for a relatively long period and therefore
deserve specific distinction, which Stromer (1940) and
Thenius (1949, cited from Crusafont and Kurten 1976)
did not give them. From U. primaevus, Crusafont and
Kurten (1976) suggest that 2 larger species, U.
ehrenbergi and U. depereti evolved. Thenius (1949)
mentioned that U. ehrenbergi may have been close to
the line leading to the subfamily Ursinae, though
Crusafont and Kurten (1976) propose that the ursine
bears probably arose from a line closer to the more
primitive U. brevirhinus. Members of the genera
Ursavus were found in Eurasia for over 10 million
years. It appears that Ursavus, perhaps the debated
species U. pawniensis, was also in North America from
the early to mid-Miocene (Tedford et al. 1987).
Hendey (1980) suggests the genus Ursavus should be
their own subfamily, Ursavinae.
The transition from the genus Ursavus to Indarctos
during the early Pliocene appears to follow an increase
in body size (Hendey 1972). Crusafont and Kurten
(1976) proposed that the relatively lightly built
Indarctos vireti was the most primitive species of this
genus. The various species of Indarctos generally
continued to increase in size and spread from Europe
and Asia to North America, where the remains of I.
oregonensis, a very large bear, was discovered in
Oregon (Merriam et al. 1916) and Nebraska (Shultz and
Martin 1975).
For many years there has been disagreement on the
relationships of the third genus of the subfamily,
Agriotherium. Recently, Thenius (1959) and Hendey
(1972 and 1980) conclude that Agriotherium evolved
from Indarctos and may even have been congeneric
(Hendey 1980). Dalquest (1986), however, believes
that the 2 are not closely related. The long-standing
confusion over the evolutionary direction of the 2
species may be due to Agriotherium being more
carnivorous than Indarctos, which is against the general
trend of the Ursidae (Hendey 1980).
Remains of Agriotherium have been found in many
parts of the world including Europe, Iran, India, South
Africa, and North America (Hendey 1972). In North
America, its known range extended from California to
Florida and from Nebraska to southern Mexico
(Dalquest 1986). In North America Agriotherium
became larger than any extant species of bear (Scott
1937, Shultz and Martin 1975).
The extinction of Agriotherium due to competition
with early Ursinae, as was a probable cause of
extinction for other genera of Agriotheriinae, appears
less plausible because Agriotherium was thought to be
largely carnivorous. Competition with other carnivores
may have been involved (Hendey 1972). Many
carnivores had difficulties during the general extinction
of the late Hemphillian near the Miocene/Pliocene
boundary, when 60 mammalian genera disappeared
from North America (Martin 1984). The last known
representative of this subfamily, A. insigne, disappeared
after the Villafranchian in Europe (Kurten 1968) as the
climate began gradual cooling and warming oscillations,
foreshadowing the ice age.
SUBFAMILY TREMARCTINAE
The subfamily Tremarctinae includes the genera
Plionarctos, Arctodus, and Tremarctos (Thenius 1959,
Kurten 1966). There is 1 extant species, Tremarctos
ornatus of the South American Andes. Although the
fossil evidence leading to Tremarctinae is poor,
paleontologists suggest that Ursavus is likely the
ancestral genus (Thenius 1959, 1976; Kurten 1966).
Cytological and molecular methods indicate that
T. ornatus diverged from the genus Ursus 10.5-15
MYBP (O'Brien et al. 1985, Goldman et al. 1989),
when Ursavus sp. were common in Eurasia and
apparently present in North America. Although there
are several morphological and biochemical differences
between Tremarctinae and Ursidae, including a
different number of chromosomes (2n is 74 in Ursus
and 52 in Tremarctos), the fact that T. ornatus and U.
thibetanus have crossed in captivity questions placing
Tremarctos in its own subfamily (Mondolfi 1983).
The earliest Tremarctinae is Plionarctos and was
found in the upper Pliocene of California (Frick 1926)
although earlier roots in Asia are suspected (Kurten and
Anderson 1980). This genus is likely the ancestor of
the 2 other genera of the subfamily, Tremarctos and
Arctodus (Fig. 4). These 2 genera made their first
appearance in the Pleistocene of North and South
America (Kurten 1966).
The early history of both Arctodus and Tremarctos is
poorly recorded in the fossil record. Early students
such as Merriam et al. (1916) and Frick (1926) and
more recently Erdbrink (1953), believed Arctodus
evolved from a line close to Indarctos. Thenius (1959)
and Kurten (1966), however, believe that Arctodus was
a Tremarctinae.
Of the 5 species of Arctodus, A. pristinus,and A.
simus were in North America and A. bonariensis, A.
pamparus, and A. brasiliensis were found in South
America. Kurten (1967) suggested A. pristinus may
represent a relatively primitive form. It was smaller,
more lightly built, and probably less carnivorous than
the other species and so far has only been found in the
southeastern portion of North America.
More is known of A. simus, the great short-faced
bear, than other Arctoid bears, because the fossil record
is extensive in area, covering most of North America,
and relatively complete. A. simus was a very large
bear, with some individuals weighing at least 650 kg
(Emslie and Czaplewski 1985). This bear had long legs
and stood about 2 m at the shoulder, which suggests an
adaptation for fast movement. Kurten (1967) thought
that A. simus was not truly a cursorial predator, but
may have been capable of bursts of speed exceeding
those of U. arctos. Kurten (1967) suggested that its
short but wide jaws demonstrated convergence with the
great cats and indicated that it was largely carnivorous.
According to Kurten's interpretation, A. simus was by
far the most powerful predator during the Pleistocene
and possibly preyed on contemporary species of bison,
deer, horse, and ground sloth. Stirling and Derocher
(1990) and McLellan (1993) suggested that co-existence
with A. simus for or over 1 million years inflicted
significant constraints on the evolution of Ursus
americanus.
Emslie and Czaplewski (1985) disagree with the
conclusions of Kurten (1967). Based on characteristics
of the skull, body size, and relative lengths of distal and
proximal limb segments, Emslie and Czaplewski (1985)
suggest that A. simus was largely herbivorous.
A. simus disappeared at the end of the Wisconsin
glaciation, perhaps due to competition with Ursus arctos
(Kurten and Anderson 1974). These 2 bears, however,
apparently co-existed in Beringia for about 100,000
years, so if competition was the leading factor, other
conditions must have changed. The close proximity in
time between the disappearance of A. simus and the late
Pleistocene extinction, when 57 of the 79 species of
large mammal ( > 45 kg) in North America disappeared
(Marten 1984), suggests a correlation, particularly if A.
simus was largely carnivorous and much of its prey
disappeared. The peak of this extinction was about
11,000 YBP (Martin 1984) whereas the last A. simus
remains to be dated were more than 1,000 years older.
This date, however, was obtained on an Equus bone
found at the deepest level of a cave (Kurten and
Anderson 1980) and the bear was likely younger,
perhaps near the 10,000 years ago suggested by
Harrington (1973). Two competing hypotheses
explaining the late Pleistocene extinction, which
perhaps led to the extinction of A. simus, are overkill
by Paleoindians of the Clovis culture, and climatic
change that included a very dry period and a resulting
reduction of habitat diversity.
Relatively little is known of Arctodus in South
America because few fossils have been found. A.
brasiliensis was the smallest of the 3 species and, as it
more closely resembled the North American species,
may have been an intermediate. A. pamparus was also
relatively small, whereas A. bonariensis was very large,
rivaling A. simus in size (Kurten 1967). A. bonariensis
had large canines and carnassials but short posterior
molars, suggesting a carnivorous diet. Based on the
structure of their molars, Kurten (1967) suggested the
possibility of a mollusk-eating specialization for the
other South American species.
The genus Tremarctos consists of 2 species,
Tremarctos floridanus and the extant spectacled bear,
T. ornatus. As was the case with Arctodus, the
ancestral genus to Tremarctos is believed to be
Plionarctos (Kurten 1966; Thenius 1959, 1976).
Fossils of the North American spectacled bear have
been found most often in Florida and only rarely
elsewhere (Kurten and Anderson 1980). T. floridanus
seems to have been a slow-moving, heavily built,
medium-sized bear with powerful forelimbs. Kurten
(1966) suggested that T. floridanus filled a niche similar
to that of the European cave bear, Ursus spelaeus, as a
powerful, almost exclusively vegetarian bear. Reasons
for the extinction of T. floridanus in the last 8,000
years are unclear, although competition with U.
americanus has been suggested. These species, or at
least early versions, co-existed in North America for
about 3 million years, so if competition with U.
americanus was the cause of extinction, an additional
change, such as climate, must have precipitated it.
SUBFAMILY URSINAE
The subfamily Ursinae has been divided into many
different phylogenetic groups in the past. Until
recently, 5 genera, Melursus (sloth bear), Helarctos
(sun bear), Thalarctos (polar bear), Selenarctos (Asiatic
black bear), and Ursus (brown bear and American black
bear) were recognized. Molecular and cytological
methods (O'Brien et al. 1985, Goldman et al. 1989)
plus successful crossing between several of the species
in captivity (C. Servheen, U.S. Fish and Wildlife
Service, pers. commun.) suggests that these bears are
congeneric.
The evolution of Ursinae over the past 5 million
years is well recorded in fossils of Europe. Early
Ursinae likely evolved from Ursavus of the Miocene,
perhaps through the genus Protursus of the midMiocene
(Thenius 1959, Crusafont and Kurten 1976;
Fig. 5). Climatic conditions in Europe during the late
Miocene were dry, and savannahs and deserts were
common. Such conditions were poor for bears, and
their fossils are scarce until the Pliocene began, 5-6
MYBP.
Black Bears
Next to Protursus, the earliest member of the
subfamily Ursinae is believed to be Ursus minimus,
which has been found in many locations in Europe
(Kurten 1968). U. minimus was a small, primitive
species, that generally increased in size during its
existence. It appears that U. minimus or a species
similar to it radiated through Asia and was in North
America at least by the early Blancan, perhaps 3.5
MYBP (Kurten and Anderson 1980). In North
America, this species is called U. abstrusus, but may be
conspecific with U. minimus. This small early ursid
likely gave rise to the Asiatic and American black
bears. The timing of the divergence of U. americanus
estimated from fossils is similar to the 4.4 MYBP
derived through 2-dimensional electrophoresis
(Goldman et al. 1989) and the 3.8 MYBP estimated
from mitochondrial DNA divergence (Shields and
Kocher 1991).
U. thibetanus ranged into Europe during in the midPleistocene
with remains being found in many countries
(Kurten 1968). Why it was extirpated from Europe is
unknown, but competition with the largely herbivorous
cave bears may have been a factor. In North America,
black bears are by far the most common fossil bears of
the Pleistocene and have been found across most of the
continent. As was the case with many species, latePleistocene
black bears were much larger than they are
today. Behavioral and morphological characteristics
imposed on black bears by other ursids have been
discussed by Herrero (1972), Stirling and Derocher
(1990), and McLellan (1993).
Cave Bears
The small, primitive U. minimus gave rise to
U. etruscus, which was also initially small but
continued the trend towards a larger body size. This
species radiated across Eurasia. In Europe, it gave rise
to the cave bears and in Asia it was ancestral to brown
bears.
Caves are a good environment for fossilization.
Bears, with large, stocky bones, are especially well
preserved. Thus cave bears, which often died in caves,
have one of the best fossil records. The evolutionary
lineage is so complete that delineating species has been
difficult (Kurten 1968). In addition, the tens of
thousands of individuals represented (an estimated
30,000-50,000 in 1 cave) has enabled the typical study
of phylogeny and morphology plus studies of age and
sex class structure, individual variation, and mortality
rates.
Thenius (1959) recognizes 2 species of cave bear: U.
deningeri and U. spelaeus, the giant cave bear of
Europe. Kurten (1968) identifies another species, U.
savini, between the etruscan bear and U. deningeri.
Based on fossils, U. spelaeus appears to have been a
large, stocky, mostly herbivorous bear. It had a
relatively small geographic distribution, being found
only in Europe and into the southwestern corner of
Russia and the Ukraine. Such a small distribution
suggests a dietary specialization or perhaps a
dependence on caves or mountainous country where
caves are found (Kurten 1976). Remains of U.
spelaeus have not been found where caves are
uncommon.
Many possible causes for the extinction of cave bears
have been proposed. Kurten (1958) suggested that
rapidly increasing numbers of humans may have settled
caves during summer dispersal periods and thus
excluded bears returning to hibernate in the early
winter. Extant brown and black bears often hibernate
in caves much too small for human occupation, so if
this hypothesis is correct, then cave bears would have
had different requirements than the extant species, or
small caves were rare in their range. Competition with
increasing numbers of humans and brown bears for
caves and other resources plus climate change appear to
be likely factors.
Brown Bears
The brown bear is believed to have evolved from
U. etruscus in Asia. The oldest fossils were found in
China from about 0.5 MYBP (Kurten 1968) and there
has been a continuous record of U. arctos in Asia since
then. U. arctos entered Europe about 0.25 MYBP and
North Africa shortly after. Pleistocene remains of
U. arctos are common in Great Britain and they may
have contributed to the extirpation of the cave bear
there.
U. arctos apparently entered Alaska about 100,000
YBP but did not move south until the late Wisconsin,
about 13,000 YBP. Kurten and Anderson (1980)
suggest the possibility of 2 independent migrations;
narrow-skulled bears from northern Siberia through
central Alaska to the rest of the continent becoming
U. a. horribilis, and a southern migration of broadskulled
bears from Kamchatka to the Alaskan peninsula
becoming U. a. middendorffi. Fossils of brown bears
in Ontario, Ohio, Kentucky (Guilday 1968), and
Labrador (Spiess and Cox 1976) indicate they were
once found much farther east than historical records
show. Guilday (1968) suggested that immediately after
the glacial retreat, a relatively boreal, parkland
coniferous forest spread across the central and southern
portions of the continent and with it, several western
species, including brown bears.
Polar Bear
The Polar bear is a recent offshoot of U. arctos.
Indications of a recent divergence include the rarity of
fully fossilized polar bear remains (Kurten 1964)
whereas subfossils are common, and that these species
produce fertile hybrids in captivity (Kowalska 1965).
Mitochondrial DNA divergence (Shields and Kocher
1991) and 2-dimensional electrophoresis (Goldman et
al. 1989) also suggest a recent split. Polar bears are
repeating the trend that was seen with Agriotherium and
Arctodus by becoming carnivores; this time specializing
on marine mammals. The apparent morphological and
behavioral differences between polar bears and brown
bears indicate that polar bears are rapidly evolving as
they exploit a new niche.
Sun and Sloth Bears
The fossil records of south Asian bears, the sun
(Ursus malayanus) and sloth bears (U. ursinus) are
poor, and their origins more speculative than for other
species (Kurten 1966). U. malayanus is first found in
the late Pliocene and U. ursinus in the Pleistocene.
Thenius (1959) thought they separated from the other
Ursinae even before Protursus, whereas Hendey (1972)
speculated that the split was after Protursus but before
U. minimus. Electrophoretic analysis indicates a more
recent split, not significantly different from that of other
extant members of the subfamily except the polar bear
(Goldman et al. 1989). Recent analyses of
mitochondrial DNA suggest that the 6 ursine species
originated sequentially during the past 6 million years,
beginning with U. ursinus and ending with the polar
bear (Waits et al. unpubl. data, submitted). It is
becoming increasingly evident that U. ursinus,
malayanus, thibetanus, americanus, and etruscus all
branched from the primitive U. minimus or U.
abstrusus that radiated through Eurasia near the
Miocene/Pliocene boundary and into North America
shortly after. The great morphological differences
between U. malayanus, U. ursinus, and other bears is
likely due to recent adaptive change as the south Asian
bears exploited new niches.
SUBFAMILY AILUROPODIDAE
The phylogeny of the giant panda has been disputed
since 1870, when Milne-Edwards placed Ailuropoda
into the family Procyonidae, while David had called it
an ursid the previous year (O'Brien et al. 1985). Some
recent authors (Tagle et al. 1986) linked Ailuropoda
closer to the lesser panda (Ailurus fulgens) than to the
bears, but there has been an overwhelming number of
papers placing Ailuropoda not close, but closer to bears
than to the lesser panda (Kurten 1985, O'Brien et al.
1985, Mayr 1986, Ramsay and Dunbrack 1987,
Goldman et al. 1989). These papers were based on 6
independent molecular and genetic measures, fossil
evidence, and reproductive characteristics. These
recent reports plus the comparative anatomical work of
Davis (1964), earlier protein evolution work of Sarich
(1973), and the synthesis of Thenius (1979) indicate the
giant panda should be placed into its own subfamily,
Ailuropodidae, of the family Ursidae.
The earliest evidence of Ailuropoda was during the
late Pliocene, about 3 MYBP (Schaller et al. 1985).
Wang (1974) divides Ailuropoda into 2 species: A.
microta was a smaller, primitive species that became
extinct during the mid-Pleistocene. A. melanoleuca was
once larger than it is today and then ranged south of the
Yangtze river at least to Burma. The decrease in the
panda's range has been attributed to climatic changes
during the Pleistocene and, like almost all extant
species, man's activities during the postglacial (Schaller
et al. 1985)
The fossil record leading to Ailuropoda is poor and
evolutionary links speculative at best. Matthew and
Granger (1923), Davis (1964), Hendey (1972), and
Wolff (1978) believed the panda evolved from Indarctos
of the subfamily Agriotheriinae. Thenius (1979)
however, identified a possible ancestor, Agriarctos,
from the late Miocene of Hungary, and suggested it was
a descendant of Ursavus from the mid-Miocene (Fig.
6). After re-evaluation, Hendey (1980) concluded that
Ursavus depereti was more likely to be the ancestor of
the pandas than was Indarctos. The divergence
between bears and the giant panda has been estimated
to be 18-22 MYBP (Goldman et al. 1989), and thus,
separation from an early Ursavus, either via Indarctos
or not, is possible. The panda has a specialized niche
and, like other species that diverged from the
omnivorous trend, likely went through a period of rapid
evolution, which accounts for their morphological and
behavioral differences from other bears.