Reanalysis ignores pertinent data, includes inappropriate observations, and disregards realities of applied ecology: Response to Huso and Dalthorp (2023)

Abstract

Abstract We recently demonstrated efficacy of automated curtailment of wind turbines in reducing fatalities of eagles at a study site in Wyoming, USA. Huso and Dalthorp criticize our work, asserting that there are several ‘major errors’ that render our previous work as providing ‘no meaningful support’ for automated curtailment. As we show here, our data do indeed provide support for the efficacy of automated curtailment. The purported major errors they identify include: having only 1 year of post‐treatment data, having only one control and one treatment site, using the control site for inference, choosing an arbitrary demarcation date between the ‘before’ and ‘after’ periods, analysing the number, rather than the rate, of fatalities, and not including data they consider pertinent. These claims are unsupported because all result from misreading our study, ignoring realities of causal inference in field ecology, or are matters of preference. Most importantly, we did not fail to include pertinent data and we provided sufficient criteria in the manuscript to explain inclusion of carcasses in our study. As stated in our previous study, the eagle carcasses we did not include were found either outside of search areas or incidentally, and thus did not meet criteria for inclusion. Further, Huso and Dalthorp present a standard for spatial and temporal replication that is inconsistent with their own recent work and many other studies in this field. Finally, their reanalysis of our data ignores the control site while including inappropriate data and thus is not suitable for inference. We appreciate the unusually thorough critique of our work provided by Huso and Dalthorp. Despite the major errors in their argument, this discussion provides a platform for further evaluation of our original work. Synthesis and applications. We show that incorporating control site data and using our a priori demarcation of before and after periods into Huso and Dalthorp's reanalysis returns meaningful support for automated curtailment. We also direct the reader to an updated analysis that uses components of the approach Huso and Dalthorp suggest, along with additional data and a number of other analytical improvements, to validate and strengthen our original conclusions.