Natal dispersal is assumed to be costly because of the risk of mortality, y
et rarely are movement patterns and survival of dispersers observed directl
y. We determined the fates and dispersal distances of 150 radio-collared ju
venile arctic ground squirrels from 1993 to 1995 at Kluane, Yukon Territory
, Canada (61 degrees N, 138 degrees W). We tested the hypothesis that dispe
rsal has a high mortality cost, and we also attempted to distinguish among
three hypotheses to explain natal dispersal: competition for mates, competi
tion for resources, and inbreeding avoidance. Juveniles were radio-collared
at emergence from the natal burrow on five 9-ha grids nested within larger
(1 km(2)) experimental manipulations: two controls, a predator exclosure,
a food-supplemented grid, and a predator exclosure + food grid. In all year
s and on all areas, dispersing juveniles were more likely to die than philo
patric squirrels, and the risk of mortality increased with distance from th
e natal burrow for both sexes. Overall, survival of philopatric squirrels w
as 73%, whereas survival of dispersing squirrels ranged from a maximum of 4
0% to a minimum of 25%. Juvenile females were strongly philopatric independ
ent of population density, except on the predator exclosure + food grid in
1995, where population density was extremely high and resources other than
food were probably limiting. Resource competition may explain patterns of p
hilopatry and dispersal in female arctic ground squirrels. Juvenile males m
oved farther from their natal site than females and more of them died. Male
s also had a strong tendency to disperse that was independent of food avail
ability or population density, which suggests that male arctic ground squir
rels ultimately may disperse to avoid either inbreeding with female relativ
es or intrasexual competition for mates.