Experimental manipulation of forest ecosystems: lessons from large roof experiments

Environmental impacts on forest ecosystems can be studied by manipulating energy, water, and element input or by changing the internal element cycling. In practice, the intended manipulations in a complex ecosystem such as a forest are followed by unintended manipulations of other factors that may cause artifacts in the experiment. The character and extent of such unintended changes were assessed in five major roof manipulation studies in coniferous forests in Europe. In all five cases the roofs were placed beneath the canopy 2–5 m above the ground and designed to study the response to reduced N and S deposition and effects of drought. Photosynthetic light was reduced 15–50% below the roofs and might have contributed to an observed decrease in forest floor moss cover. Soil temperature differences were up to ± 0.5°C, colder than outside during summer and warmer during winter. Climatic differences were least at the smallest roofs. The sprinkling system was the most critical component in the experimental design. The sprinklers could not reproduce the temporal and spatial variability of natural rain; event size and rain intensity increased, and the number of rain events decreased. It proved particularly difficult to reproduce small rain events. The stemflow proportion of the water input was increased by sprinkling. Observed decreases of litter decomposition and mineralisation under some of the roofs were probably caused by a reduced moisture content of the surface litter due to the differences from natural rain. Exclusion of throughfall by the roof disturbed the internal cycle of nutrients leached from the canopy (Ca, K, Mg) or present in suspended material (N, P, Mg). The circulation of these elements had to be restored by addition or recycling of suspended matter. The unintended changes probably delayed the soil response to reduced acidity input and accelerated the decline of nitrate leaching in response to reduced N input.