Compositional analysis of the relationships between the organic matter content and chemical and physical properties of soil

Soil organic matter (SOM) plays a crucial role in soil fertility, carbon sequestration, and ecosystem sustainability, making its accurate analysis essential for environmental and agricultural management. However, studying the relationships between soil organic matter content (SOMC) and its influencing factors remains challenging due to the compositional nature of soil constituents. This study addresses key methodological challenges in analyzing the relationships between SOMC and soil texture, chemical composition, and bulk density using compositional data analysis. Specifically, we solve methodological issues related to integrating compositional and non-compositional variables in regression modeling and apply, for the first time, compositional data analysis to a mix of compositions, including the SOMC composition. The study explores the multivariate dependencies of the log-ratio coordinates—transformations that map compositional data from the constrained simplex space to real space—of major chemical elements in the soil and their relationship to log-ratio coordinates of SOMC. To appropriately account for the compositional nature of both the chemical element composition and soil texture, compositional data analysis methods are employed. Additionally, since outliers are common in soil data, all estimations are carried out using robust estimation methods. The application focuses on topsoil in the canton of Zurich (Switzerland), providing new insights into these relationships. Some findings contrast with previous studies that did not adopt a compositional approach, revealing, for example, a weak positive association between calcium and SOMC, a positive effect of phosphorus, and a decreasing dominance of organic matter in soil texture with increasing bulk density. Furthermore, free and open-source software has been extended to enable linear regression modeling that integrates both compositional and non-compositional explanatory variables, offering a practical solution to these methodological challenges in soil science. • Compositional data analysis improves the understanding of soil organic matter dynamics by appropriately handling the relative nature of soil constituents. • Software for a compositional regression framework that integrates both compositional and non-compositional explanatory variables for more accurate modeling is provided. • Robust statistical methods mitigate the influence of outliers, ensuring more reliable interpretations of soil composition data. • Phosphorus positively influences SOMC, while calcium and SOMC exhibit a strong correlation, highlighting key chemical interactions in soil.