Comment on “Forest and floods: A new paradigm sheds light on age‐old controversies” by Younes Alila et al.

[1] The paper by Alila et al. [2009, hereafter referred to as AKSH] presents a technique for analyzing altered peak flow frequencies after logging. The paper suggests that the established method of chronologically pairing peak flows by corresponding hydrologic input at control and treated watersheds is inappropriate, leading to irrelevant research hypotheses and impeding scientific progress. In general, we agree that analyses of changes in flood frequency are useful for evaluating the effects of watershed disturbance, and that simple regression models often provide inadequate descriptions of posttreatment peak flow responses. However, the proposed method and accompanying discussion have several problems that undercut the strength of the paper’s conclusions: [2] 1. The recovery adjustment used by the method augments the effect the analysis is attempting to detect. [3] 2. Even in the absence of impacts, the frequency distribution of observed peaks is expected to have greater variance than that of predicted peaks, thereby introducing an artificial shift when comparing upper quantiles of observed and expected frequency distributions. [4] 3. A more appropriate analysis of uncertainty is needed if the utility of the method is to be validly assessed. [5] 4. Frequency pairing does not overcome the problem of low power in testing for changes in very large events prior to forest regrowth. [6] 5. The relative merits of alternative statistical approaches are mischaracterized.