EVALUATING ALTERNATIVE EXPLANATIONS IN ECOSYSTEM EXPERIMENTS

Unreplicated ecosystem experiments can be analyzed by diverse statisti cal methods. Most of these methods focus on the null hypothesis that t here is no response of a given ecosystem to a manipulation. We suggest that it is often more productive to compare diverse alternative expla nations (models) for the observations. An example is presented using w hole-lake experiments. When a single experimental lake was examined, w e could not detect effects of phosphorus (P) input rate, dissolved org anic carbon (DOC), and grazing on chlorophyll. When three experimental lakes with contrasting DOC and food webs were subjected to the same s chedule of P input manipulations, all three impacts and their interact ions were measurable. Focus on multiple alternatives has important imp lications for design of ecosystem experiments. If a limited number of experimental ecosystems are available, it may be more informative to m anipulate each ecosystem differently to test alternatives, rather than attempt to replicate the experiment.