Long term population fluctuations of the fossorial form of the water v
ole, Arvicola terrestris, can be reconstructed on the basis of the rew
ards given to vole trappers (''tail-turn statistics''). Nine time seri
es of at least 20 yr of continuous data recorded during the 20th centu
ry in Switzerland have been analysed using time series analysis and te
sted for direct and delayed density dependence. Seven of these data se
ts showed cyclic changes in abundance with statistically significant m
axima of the periodogram for periods ranging between 5 and 7 yr. Their
partial autocorrelation functions suggested the presence of both dire
ct or delayed density dependence in the data. This was confirmed by re
gression analysis using the nonlinear approach advocated by Turchin. I
n contrast, no simple periodicity could be found in the two remaining
time series, nor indication for direct or delayed density dependence i
n their partial autocorrelation functions. However, direct density dep
endence was also statistically demonstrated in these two data sets. In
most cases, the analyses were complicated by the presence of trends i
n the data. The detection of density-dependent regulatory effects was
obscured by linear and nonlinear trends in one and two cases, respecti
vely. The data support the hypothesis that time-delayed and nonlinear
mechanisms of population regulation operate in fossorial populations o
f A, terrestris. Furthermore, the data are consistent with regulatory
mechanisms involving predator-prey interactions, while the causation o
f the unusually long cycle is still an open question. Finally, the pre
sence of trends in these time series suggests that water vole populati
ons track some long-term climatic changes in the environment.