Interactions among size-structured populations in a whole-lake experiment:size- and scale-dependent processes

Interactions in sits-structured populations are characterized by a mixture of predatory and competitive interactions dependent on the size of the indi vidual organism. We analyzed this mixture of size-structured interactions f or a species constellation consisting of perch (Perca fluviatilis) and roac h (Rutilus rutilus) in a replicated whole lake experiment over 4 yr. Roach are preyed upon by large perch, but at the same time compete with small per ch for zooplankton. Predictions regarding the effects of roach on perch per formance and resource dynamics in the whole-lake experiment were based on r esults from previous pond and enclosure experiments carried out over short (months) time periods. Variables measured in the whole-lake experiment incl uded both individual level parameters (diet and growth of perch) as well as population level parameters (mortality, population numbers and size struct ures of perch, abundance and biomass of zooplankton and benthic macroinvert ebrates). The population size of perch greater than or equal to 2 yr old de creased to less than 10% of the pre-treatment levels in treatment lakes aft er the introduction of roach and remained low over the whole study period. The mortality in perch one-year-old and older increased with size. Larger p erch had a higher growth decrease than smaller perch, but also the individu al growth rate of young-of-the-year (YOY) perch was lower in treatment lake s than in control lakes. YOY perch were almost excluded during the winter a nd spring following the roach introduction. The few perch of this year clas s that survived had the following years a higher growth than the correspond ing year class in control lakes. This could be related to an increased avai lability of predator-sensitive macroinvertebrates in treatment lakes. Perch in treatment lakes fed to a larger extent on macroinvertebrates and less o n pelagic zooplankton than perch in control lakes. As expected, zooplankton abundances decreased in treatment lakes the year following the roach intro duction, but thereafter zooplankton abundances were higher in treatment lak es due to the absence of perch recruitment in these lakes. No incidence of perch predation on roach was recorded during the study period. We attribute this to the spatial scale dependent escape ability and habitat use of vuln erable size classes of roach. YOY perch had a stronger impact on between-ye ar variation in zooplankton than YOY roach and this can be related to the h igher foraging (handling) capacity of the former and the restricted habitat use of the latter. Individual level parameters (diet, growth) had a higher statistical power t han population level parameters (abundance, biomass, mortality), and small- scale experiments successfully predicted a higher proportion (81%) of the i ndividual level parameters. However, the lower success in predicting popula tion level responses (38%) was directly connected to a failure to predict i ndividual level parameters (large perch diet and growth). Our results sugge st that the small spatial scale was the major limitation on the predictive ability of the small-scale experiments.