Large dam reservoirs are probably long-period oscillators of fish diversity.

Long-term fish population time series obtained annually in a backwater (upstream) and a tailwater (downstream) site of the large lowland Jeziorsko Reservoir, Poland, were analysed and compared to detect impoundment influence on the fluctuation of fish population density and on the fluctuation of a fish diversity measure. The partial-rate correlation function (PRCF) of all constantly occurring fish species displayed negative first-order feedback of past density on present density, both upstream and downstream. Weak and irregular cyclic patterns were observed in the autocorrelation function (ACF) of populations of several species upstream and of two species downstream. Most fish species exhibited no significant concordant variations between the two sites, but there were no data supporting different regulation. Probably, different exogenous (environmental) factors control fishes at the two sites. The ACFs of the time series of the exponential of Shannon entropy (N(1) diversity measure) obtained at the Jeziorsko tailwater site, and in the main body of the smaller Římov Reservoir, Czech Republic, revealed strong fluctuation patterns. The fluctuation was well fitted by a sine-wave model of a 9 year period at the Římov, although it co-occurred with a positive linear trend. A sine-wave model also fitted well with the trend at the Jeziorsko tailwater site, where, however, only one 16·5 year period occurred; hence, it is indicative of cyclicity. The results indicate that dam reservoirs may be oscillators of fish diversity and that part of the literature controversy over the effect of a reservoir on fish populations may be due to a too short sampling period: it is shown that both increase and decrease in diversity may be observed on the basis of samples selected from one diversity time series obtained at the tailwater.