Soybean (Glycine max (L.) Merr.) is a species of relatively little economic importance in Central and Eastern Europe, including Poland. Due to its popularity for the production of soybean oil, livestock feed, and human food, soybeans are a widely cultivated agricultural crop in the world. The aim of the presented research is to determine the most important agronomic and environmental variables in soybean production in Central and Eastern Europe. This work used a dataset from the Polish Post-Registration Variety Testing System in multi-environmental trials from the years 2012-2021. Variables classified for crop management included doses of mineral fertilizers (N, P, and K) and herbicides, sowing, and the type of previous crops. The environment was also included in the analysis through soil and weather characteristics using climatic water balance (CWB). The analysis was performed using multiple linear regression models and regression trees. It found that the variability of the soybean yield depended mainly on water available to plants and physical soil properties. This means that environmental variables have a stronger effect in comparison to crop management variables. The effect of the nutrients applied in the fields was relatively weak and only important in the case of phosphorus. Other variables which characterize crop management (including sowing date, previous crop, and plant protection using pesticides) have a weak effect on grain yield and yield-related traits variability. As there are not many studies on soybean cultivation in Poland, this work might be used as an introduction to research on soybean management in a hemiboreal climate.
Keywords: multivariate analysis; regression trees; yield variability.