nates, and infants were present (Arnold 1990b). Interest-
ingly, less-related subordinates lost less mass when hiber-
nating in groups with infants. Because the other animals in-
crease burrow temperatures, these subordinates profit from
the increased heat but apparently do not increase metabo-
lism to contribute to burrow warming and alloparental
care. Two essential points should be emphasized: social
thermoregulation and alloparental care require the pres-
ence of subordinate adults, and alloparental care is strongly
biased to favor close kin.
Although golden marmots live in groups containing up
to seven adults and hibernate in groups, total group size
does not influence overwinter survival of juveniles. For non-
juveniles, total group size, including juveniles, does not in-
fluence the probability of survival. There is a tendency for
nonjuvenile survival to decrease as the number of nonjuve-
niles increased. For juveniles, when only parents and litter-
mate sibs were present, there is less mortality than in groups
containing other animals. When juveniles are excluded
from the count, overwinter survival of juveniles decreased
when hibernating in groups with more nonjuveniles (Blum-
stein and Arnold 1998). Thus, I find no compelling evi-
dence for social themoregulation, and some indication that
group hibernation may be costly.
Social thermoregulation
Social thermoregulation refers to the active contribution of
heat to other individuals during hibernation at a cost in in-
creased mass loss in the donor. It requires synchrony in the
arousal-torpor patterns, which reduces mass loss (Ruf and
Arnold 2000). The presence of juveniles can decrease over-
all synchrony and offset the reduction in mass loss provided
by the increased insulation of group huddling. Decreased
synchrony probably explains high mass loss and mortality
when only juveniles and the two territorial adults hibernate
together, and the mass loss of related subordinate adults
who provide the alloparental care. Juvenile yellow-bellied
marmots hibernating in groups had higher rates of mass
loss than animals hibernating singly; torpor cycles in the
grouped young were asynchronous (Armitage and Woods
2003). Furthermore, there was no difference in overwinter
survivorship of yellow-bellied marmot young hibernating
in groups ranging from one to eight individuals. Although
these young probably hibernate with their mother (Lenihan
and Van Vuren 1996), there is no evidence for benefits from
joint hibernation.
Probably all species of marmots, except M. monax,hi-
bernate in groups (I use the term group hibernationto indi-
cate that members of a social group share a hibernaculum
and have the potential to jointly benefit from huddling).
Group hibernation raises the question of its evolution to so-
cial hibernation and of the importance of social hibernation
in the evolution of marmot sociality. Social hibernation ap-
parently requires the presence of subordinate adults; this
social group is present only in species that have extended
families (table 30.1). Although delayed dispersal was inter-
preted to occur so that subordinate adults could provide
alloparental care (Arnold 1990a, 1990b), this interpretation
seems unlikely. Alloparental care would have to occur be-
fore delayed dispersal; in other words, the nondispersers
would have to “anticipate” future social thermoregulation.
It seems more likely that delayed dispersal occurred to in-
crease the probability that subordinates would become
more competitive for a territorial position. By remaining
in the family reproduction is delayed and a direct fitness
cost is incurred. However, by hibernating with other family
members, the subordinates have an opportunity to recoup
some of the fitness loss by gaining indirect fitness through
alloparental care directed to close kin. The territorials ben-
efit from delayed dispersal. The fitness of the territorials is
zero unless their offspring reproduce. By accepting delayed
dispersal, the territorial dominants increase the probabil-
ity of having grand-offspring while increasing their fitness
through their increased survival and that of the next gener-
ation of their offspring through social hibernation. In effect,
delayed dispersal incorporates subordinate adults into the
social group and makes possible the evolution of social
thermoregulation from group hibernation.
Group hibernation probably evolved as a consequence of
retaining infants at home for their first hibernation. A good
hibernaculum may be a limited resource (Holmes 1984a),
and may be used over many generations (Lenti Boero 2001;
Armitage 2003b). Thus survival of young was likely greater
if they hibernated in a proven hibernaculum independent
of any possible benefits from huddling. Group or social hi-
bernation likely is more important in those species that are
relatively small and have a long hibernation period. For ex-
ample, the alpine marmot at the onset of hibernation is rel-
atively small and hibernates for about 6.5 months; thus so-
cial thermoregulation characterizes this species. By contrast,
the larger woodchuck hibernates for only about 4.5 months
and relies solely on its fat stores during solitary hiberna-
tion (Armitage et al. 2000). Although woodchucks at more
northern latitudes hibernate about a month longer under
more harsh winter conditions, they survive by using a larger
percentage of their mass (Ferron 1996). Two of the largest
marmots, the Olympic and hoary, lose the largest amount
of mass during a hibernation of about 7.5 months (Armi-
tage and Blumstein 2002). This large mass loss may be a
consequence of a long hibernation in a harsh environment;
possibly the adults provide heat to their young and thus
incur a mass loss (Armitage 1999a). Virtually nothing is
known about hibernation energetics outside the woodchuck
and the alpine and yellow-bellied marmots. ConsiderableEvolution of Sociality in Marmots: It Begins with Hibernation 363