Predicting extinction probabilities for populations of various sizes has be
en a primary focus of conservation biology. Berger (1990) presented an empi
rically based extinction model for mountain sheep (Ovis canadensis) populat
ions in five southwestern states that predicted disappearance within 50 yea
rs of all populations estimated to number 50 sheep or fewer, but essentiall
y no loss in that time period of populations estimated at over 100. The maj
ority of the 122 populations he used in his analysis were from California,
but his analysis did not use many of the historical size estimates for thes
e populations. I tested Berger's (1990) model using the complete data set f
rom California and found - contrary to his results - that, for all size cla
sses of population estimates, at least 61% of the populations persisted for
50 years. Also, two predictions from Berger's model were not consistent wi
th the data from California: (1) 10 populations have increased from estimat
es of 50 or fewer animals to over 100, whereas the Berger model predicted t
hat these populations would only decline to extinction: and (2) of 27 extan
t populations with long enough records, 85% were estimated at least 50 year
s ago to be 50 individuals or fewer and should therefore be extinct by now.
Berger's model has now failed tests in three states and therefore does not
support the strong population size effect on extinction probability that i
t first appeared to provide, and it may serve conservation poorly through m
isdirected effort if it is used as the basis for setting policies or taking
actions.