The dilemma of analytical method changes for soil organic carbon in long‐term experiments

Abstract

AbstractLong‐term experiments (LTEs) have provided data to modellers and agronomists to investigate changes and dynamics of soil organic carbon (SOC) under different cropping systems. As treatment changes have occurred due to agricultural advancements, so too have analytical soil methods. This may lead to method bias over time, which could affect the robust interpretation of data and conclusions drawn. This study aims to quantify differences in SOC due to changes in dry combustion methods over time, using soil samples of a LTE established in 1963 that focuses on mineral and organic fertilizer management in the temperate zone of Northeast Germany. For this purpose, 1059 soil samples, collected between 1976 and 2008, have been analysed twice, once with their historical laboratory method right after sampling, and a second time in 2016 when all samples were analysed using the same elementary analyser. In 9 of 11 soil sampling campaigns, a pairedt‐test provided evidence for significant differences in the historical SOC values when compared with the re‐analysed concentrations of the same LTE sample. In the sampling years 1988 and 2004, the historical analysis obtained about 0.9 g kg−1lower SOC compared with the re‐analysed one. For 1990 and 1998, this difference was about 0.4 g kg−1. Correction factors, an approach often used to correct for different analytical techniques, could only be applied for 5 of 11 sampling campaigns to account for constant and proportional systematic method error. For this particular LTE, the interpretation of SOC changes due to agronomic management (here fertilization) deviates depending on the analytical method used, which may weaken the explanatory power of the historical data. We demonstrate that analytical method changes over time present one of many challenges in the interpretation of time series data of SOC dynamics. Therefore, LTE site managers need to ensure providing all necessary protocols and data in order to retrace method changes and if necessary recalculate SOC.HighlightsA total of 1059 LTE soil samples taken between 1976 and 2008 were re‐analysed for SOC in 2016Several methodological changes for SOC determination led to significant different SOC concentration in the same sampleInterpretation and time series of LTE soil data suffer from consideration of analytical method changes and poor documentation of the sameSoil archive establishment, thorough method protocols and diligent proficiency testing after soil method changes ameliorate the dilemma