Temporal variability in population density has been shown to increase
with increasing length of a study. This has serious implications for e
cological studies and suggests that obtaining an objective measure of
temporal variability may not be possible. To test this hypothesis more
generally, I calculated inter-annual variability, using the standard
deviation of the logarithm of annual mean population density (STD(log
D)), for 70 populations of phytoplankton, zooplankton and fish which h
ave been sampled over 10 to 51 consecutive yr in lakes around the worl
d. STD(log D) fluctuates most widely during the first 3-6 yr of a stud
y, but these changes do not necessarily represent long-term trends in
temporal variability. Median changes in STD(log D) over time are small
(<< 0.l unit, even over a logarithmic time scale) and decrease as mor
e years are included in the analysis. The inter-annual variability of
most freshwater populations appears to reach an asymptotic value. Seve
ral populations, however, show an alternation of plateaux and of sudde
n increases in STD(log D), suggesting that the asymptotic values found
in most populations may be temporary and may not extrapolate beyond t
he temporal scale of this study (10-50 yr). Overall, the increases in
inter-annual variability are expected to be small(<0.l units, on avera
ge, per 25 yr). The pattern and the magnitude of these changes did not
differ significantly between algae, zooplankton and fish, and were si
milar to those reported for birds and mammals. Increases in temporal v
ariability may only be important for questions addressed over long per
iods of time.