Extreme-Temperature Indices and Seasonal Precipitation Deficits Characterize Soybean Yield Variability in Eastern Croatia: The 2024–2025 Climatically Stressful Seasons in Context (2020–2025)

Recent hydrothermal extremes threaten soybean productivity in rainfed systems of Southeastern Europe, but single-location field datasets require cautious interpretation. This study evaluated cultivar-level soybean yield variability in large-scale rainfed field trials near Osijek, eastern Croatia, during 2020–2025 and compared the climatic context of the adverse 2024 and 2025 seasons. Grain yield, adjusted to 13% moisture, was analyzed across years and maturity groups (MGs) using ANOVA, Fisher’s LSD mean separation, ANCOVA/linear trend analysis, and exploratory climate–yield correlations. Daily station data for 2024–2025 were used to calculate extreme-temperature indices. ANOVA showed a highly significant year effect on yield (p < 0.001), whereas MG and Year × MG were not significant. Fisher’s LSD separated 2023 and 2021 as the highest-yielding years, 2024 as intermediate, and 2022 and 2025 as the lowest-yield group. ANCOVA indicated a significant negative common temporal trend (−0.155 t ha−1 year−1; p < 0.001). Yield was positively associated with June–August precipitation (Pearson r = 0.885; n = 6). The 2024 season showed a stronger heat-related signal, whereas 2025 showed a stronger precipitation-deficit signal. The results should be interpreted within the limits of a site-specific, unbalanced field-trial dataset.