Improving minimum detectable differences in the assessment of soil organic matter change in short‐term field experiments

Abstract

AbstractThe demand for information on cropping system impact on soil organic matter (SOM) calls for efforts to improve the utilization of short‐term field experiments (e.g., to evaluate the parameterization of cropping systems in models). Those approaches have coped with the problem of determining small SOM changes within a large background mass. Thus, objectives of this survey are (1) the improvement of the minimum detectable difference (MDD) in SOM in the hudycrop short‐term field experiment by methods of sampling design and data treatment, (2) the verification to what extend the hudycrop short‐term field experiment allows for the determination of management induced effects on SOM, and (3) the investigation to what extent the obtained results may be suitable to evaluate the parameterization of a SOM balance model. The design of the hudycrop is suitable for excluding outliers plotwise. The estimation of plot means can be improved by the sampling design. Instead of determining a single plot mean in a mixed sampling procedure, the design provides multiple values for each plot, allowing for the identification of extreme values before calculating plot means. In consequence, minimum detectable differences decrease by a factor of 0.53 for soil organic C (SOC) and 0.63 for soil total N (STN) masses, allowing for detection of changes in the magnitude of 3.7 and 2.6% of background SOC and STN levels, respectively. Differences between treatments, however, are significant with corrected values (after outlier exclusion) for the crop production systems with the highest impact (potatoes and mulched red clover). Determining outliers based on Student's t‐test gives the lowest MDD and is therefore considered to be the most suitable method in this case. Correlations between apparent changes and SOM balances according to the HU‐MOD–2 model, used in this survey, indicate that the experimental design, in principal, is suitable for the evaluation of the parameterization of crop production systems in models. Still, an improved precision in SOM change detection is necessary. Reasonable options for that purpose are discussed in the paper.