A multivariate analysis of the influence of soil conditions on wheat yields

Rapid global population growth has led to food demand out-pacing supply, increasing food insecurity. In light of this, the United Nations declared ending hunger as its Second Sustainable Development Goal in 2015. Environmental changes from pollution and climate change further threaten food security. In this study, machine learning regression models were employed to analyze the impact of fertilizer use, soil organic carbon displacement due to erosion, surface water conductivity, soil pH, and organic carbon content on wheat yields. Permutation feature importance scores were used to evaluate the importance of each variable. Through the application of a ridgeCV stacking regression model, we identified that soil pH was the most influential variable in determining wheat yields. In the regressions tested, all p-values were below 0.005, so the null hypothesis could be rejected. The study’s results suggest that regular pH monitoring, building organic carbon-boosting infrastructure, using drainage systems, and employing sustainable agricultural practices are necessary to maximize agricultural production efficiency. Soil pH could further be protected through directed future policymaking. To expand upon the research conducted in this study, investigating the effects of climate change on properties affecting soil quality could yield valuable information on strategies to promote sustainable agriculture. Furthermore, analyzing crops other than wheat would extend the study’s results to different agricultural practices, such as tropical agriculture.