TREATMENT EFFECTS IN A STREAM FISH ENCLOSURE EXPERIMENT - INFLUENCE OF PREDATION RATE AND PREY MOVEMENTS

Field experiments involving manipulations of predator or competitor de nsities within a small portion of the habitat are one of the most impo rtant tools for studying biotic interactions. A crucial assumption mad e in such experiments is that the treatment effects are unbiased estim ates of the true effects. In this paper we examine how prey movements influence bias in predation experiments with enclosures that allow pre y to migrate to and from the surrounding unmanipulated habitat. A math ematical model of a habitat with spatially variable predation intensit y was used to analyse how prey movements affect the relationship betwe en treatment effects in predator caging experiments and the effects of predators on the prey populations. Here we define the population effe ct as the effect on prey population size that would be seen if the den sity of predators were manipulated in the entire habitat used by the p rey population. The model predicts that treatment effects in small sca le experiments decrease with prey movement rates, and that population effects increase with prey movement rates. We manipulated the density of benthic feeding sculpins (Cottus gobio) in enclosures for three mon ths during autumn and early winter. Treatment effects were pronounced. Total density of invertebrates as well as the densities of the five m ost abundant taxa were reduced by more than 50% in cages containing fi sh. Treatment effects on different taxa were, as predicted, negatively correlated with an index of per capita drift rate. Furthermore, an in dex of predation rate per prey, used as an indicator of population eff ects, was positively correlated with the drift rate index. This means that the relationship between the effects found in the experiments and the effects on the prey populations was negative rather than positive as is usually assumed. We conclude that prey movements can cause the effects found in small scale experiments to be severely biased estimat es of the population effects of predation on prey populations.