Data transformations cause altered edaphic-climatic controls and reduced predictability on soil carbon decomposition rates

Abstract

Data transformation of the reference soil organic matter (SOM) decomposition rates (k_ref), often derived as turnover times or in alternative formats, is commonly used to develop ecological models for projecting the persistence of SOM. However, the effects of reciprocal or logarithmic transformation of k_ref on model performance and edaphic-climatic patterns remain uncertain. Here, we convert published k_ref values into reciprocal or logarithmic formats and establish machine learning models between the transformed k_ref and edaphic-climatic predictors. We show that models trained with the transformed k_ref exhibit a 11.6%−68.4% reduction in performance upon re-conversion to k_ref compared to those trained with the original k_ref. The variable importance analysis identifies distinct key predictors governing the original k_ref and its transformed counterparts. This suggests that data transformation alters the relative significance of predictors without necessarily improving k_ref prediction performance. Consequently, our study underscores the importance of directly focusing on the original values rather than alternative representations when dissecting a given variable's patterns and mechanisms in ecological modeling.