Pedotransfer functions for prediction of soil organic carbon content for Chernozems Haplic and Calcic: A Case Study from the Left-Bank Forest- Steppe of Ukraine

The aim of the present research is to simulate and evaluate pedotransfer functions (PTF) for predicting soil organic carbon in Chernozem Haplic and Chernozem Calcic of the Left-Bank Forest-Steppe of Ukraine using data from agrochemical soil certification. Soil samples were collected from the upper arable horizon (0–20 cm) within the Left Bank Upland Province of the Forest-Steppe zone for Chernozem Haplic and Chernozem Calcic. According to the Ukrainian national soil classification system, the studied soils are classified as Typical Chernozems and Ordinary Chernozems. In addition to standard agrochemical analyses made by Poltava branch of State Institution «Institute for Soil Protection of Ukraine», in the Kyiv laboratory of this Institution determined the total carbon (TC), total inorganic carbon (TIC), and total organic carbon (TOC) contents. The Carbon content was determined using an analyzer Multi EA 4000 analyzer (Jena) via combustion. The total carbon released was automatically measured using a non-dispersive infrared detector (NDIR). The TOC content was determined as: TOC = TC - TIC (g/kg of soil). The descriptive statistical analysis and correlation and regression analyses were carried out using Excel; to assess the accuracy of the models and the agreement between the predicted and measured values of TOC (RMSE, NRMSE, MAE, R² and MAD). The statistically significant of the differences between the groups of predicted values of the TOC content was assessed using the t-test: Two-Sample Assuming Unequal Variances. The Humus and TOC content practically follow a normal distribution, which allows us to correctly estimate confidence intervals and significance levels; apply parametric statistical tests. The Humus content did not appear to be a key factor driving the TOC content. Taking into account the wide variability in TOC-to-Humus ratios observed in our study (ranging from 0.495 to 0.92), the use of a conventional conversion factor (e.g., 0.58) to estimate TOC content from Humus content may lead to substantial inaccuracies. The predicted values of TOC content using the PTFs, can be considered statistically interchangeable with the observed ones under the tested conditions, thereby confirming the adequacy of the models for practical application in TOC estimation. Given the absence of bias and the acceptable levels of prediction error (NRMSE < 10% for all models), the tested PTFs may be reliably used for TOC assessment in monitoring, land-use planning, and carbon stock estimation. Our vision of the practical use of modeled PTFs is as follow: if no exact information on the taxonomic classification of soils is available, it is better making use of PTF3 (TOC (g/kg) = 11.0+2.394*Humus - 0.049*P + 0.065*К); if the soil type is determined clearly, then for Chernozems Haplic – PTF5 (TOC (g/kg) = 32.1 + 1.21*Humus + 0.0017N + 0.088K - 0.085P - 2.71*pH), and for Chernozems Calcic – PTF7 (TOC (g/kg) = 15.4 + 1.81*Humus).