Long-term studies and reproducibility: Lessons from whole-lake experiments

An important tenet of science is establishing the reproducibility of findings. While long-term studies may seem ill-suited to this goal, here we provide an example of reproducible results from repeated nutrient additions to a lake. We added nitrogen and phosphorus to Peter Lake in 9 yr of a 33-yr study. For seven of these nine additions, phytoplankton biomass, as measured by seasonal mean chlorophyll a, increased in proportion to the rate of nutrient loading. Additionally, for these seven additions, similar nutrient loading rates resulted in mean Chl a concentrations within a roughly twofold range-an outcome within expectation given uncontrolled sources of variation in a whole-lake manipulation. However, for two of the nine nutrient additions, Chl a concentrations were well below expected concentrations. The low chlorophyll responses co-occurred with years having the highest water color (absorbance of light at 440 nm). The number of years of nutrient additions was too limited to strongly test the influence of color at the scale of seasonal mean values. We, therefore, tested for the effect of phosphorus load and color, on Chl a using time series models of weekly data. At the weekly scale, there was a strong negative effect of color on chlorophyll concentration. Overall, the repeated nutrient additions provided a confirmation of existing models at the whole-lake scale and demonstrated an interesting exception to these models. Including repeated manipulations as part of long-term studies is an important way to test generalizations and to identify unexpected outcomes that raise new questions.