Heat stress variation exerts a more pronounced effect than vapor pressure deficit on maize yield in Northeast China

Northeast China (NEC) stands as one of the dominant maize-growing regions in China, playing a pivotal role in national food security. Maize is highly susceptible to inter-annual variations in both mean and extreme climate conditions. Identifying the most relevant climate variables that constrain maize yield is crucial for designing tailored management strategies for rainfed maize production in vulnerable environments. Here we compiled databases encompassing long-term observed maize phenology, yield, and weather data from agro-meteorological experimental stations in NEC spanning the period 1981–2022 for analysis. Using boosted regression tree (BRT) models, we explored the impacts of mean climate anomalies, temperature-related extreme anomalies, and temperature-moisture compound stresses (TMCS) anomalies on maize yield during different maize growing periods. Results showed that the BRT model explained a large proportion of variability in maize yield, with cross-validation correlations ranging from 0.78 to 0.88, indicating that BRT method was robust for exploring the impact of climate change on maize yield. Anomalies in heat growing degree days (HDD) during the maize growing period, particularly in the reproductive phase, were identified as the most significant climatic factor negatively affecting maize yield in NEC, followed by vapor pressure deficit (VPD) anomalies. HDD anomalies were strongly correlated with VPD anomalies, indicating that extremely high temperatures and high VPD occurred simultaneously and were common during the maize growing period in NEC. The impact of TMCS anomalies on maize yields were significantly more pronounced in Heilongjiang (19.5%) compared to Jilin (8.5%) and Liaoning (9.4%). These findings highlight the critical role of heat stress variations in affecting crop yield, and provide crucial insights for the development of province-specific and effective management strategies.