ABSTRACT
A recent paper by Beretta-Blanco and Carrasco-Letelier (2021) claims that agricultural eutrophication is not one of the main causes for cyanobacterial blooms in rivers and artificial reservoirs. By combining rivers of markedly different hydrological characteristics e.g., presence/absence and number of dams, river discharge and geological setting, the study speculates about the role of nutrients for modulating phytoplankton chlorophyll-a. Here, we identified serious flaws, from erratic and inaccurate data manipulation. The study did not define how erroneous original dataset values were treated, how the variables below the detection/quantification limit were numerically introduced, lack of mandatory variables for river studies such as flow and rainfall, arbitrary removal of pH > 7.5 values (which were not outliers), and finally how extreme values of other environmental variables were included. In addition, we identified conceptual and procedural mistakes such as biased construction/evaluation of model prediction capability. The study trained the model using pooled data from a short restricted lotic section of the (large) Uruguay River and from both lotic and reservoir domains of the Negro River, but then tested predictability within the (small) Cuareim River. Besides these methodological considerations, the article shows misinterpretations of the statistical correlation of cause and effect neglecting basic limnological knowledge of the ecology of harmful algal blooms (HABs) and international research on land use effects on freshwater quality. The argument that pH is a predictor variable for HABs neglects overwhelming basic paradigms of carbon fluxes and change in pH because of primary productivity. As a result, the article introduces the notion that HABs formation are not related to agricultural land use and water residence time and generate a great risk for the management of surface waterbodies. This reply also emphasizes the need for good practices of open data management, especially for public databases in view of external reproducibility.