The New Zealand fur seal (Arctocephalus forsteri) appears to show regular s
hore attendance in the form of seasonal oscillations. This phenomenon shoul
d be quantified to properly interpret counts of fur seals >1 year old (i.e.
, non-pups). Here we test the predictability of peaks in the annual shore-a
ttendance oscillation on Otago Peninsula using an autoregressive sine model
and >2 years of intensive survey data. We predicted that the peak in fur s
eal numbers ashore would lie between 14 January - 4 April (1996) and 8 Janu
ary - 2 April (1997), although this low predictability is undesirable when
attempts are made to monitor population trends. Estimating population size
from counts of non-pups also requires knowledge of the rate of turnover of
individuals. Therefore, we tested the hypothesis that ragged animals from o
ther colonies are immigrants to Otago Peninsula. With sightings on Otago Pe
ninsula of fur seals tagged elsewhere in New Zealand, we used a Monte Carlo
approach to simulate the expected frequency of single and multiple sightin
gs of individuals. We found that the observed frequency of multiple sightin
gs was significantly less than predicted by the model (P < 0.0001), indicat
ing that tagged animals were transients. We also discovered that the sex ra
tio of tagged animals varied with breeding colony of origin (G(1) = 52.07,
P < 0.0001), suggesting that the impetus for emigration differs among colon
ies. We concur with the view that counting pups is the only way to estimate
the relative abundance of New Zealand fur seals. In addition, we showed th
at counts of non-pups cannot be used to estimate population size because an
unknown proportion of individuals is transient. However, counting of pups
does not address the issue of estimating relative abundance for locations w
ith large numbers of nonbreeding individuals and few or no breeders. With f
ew or no pups it is impossible to estimate relative abundance using counts
of pups.