Thermal-Time-Based Prediction of Maize Kernel Moisture Dynamics and Its Association with Kernel Filling for Mechanical Kernel Harvesting

High kernel moisture content at harvest severely restricts mechanical kernel harvesting in China, necessitating an in-depth understanding of the dynamic coordination between kernel filling and dehydration, alongside robust and practical prediction models. Here, we systematically quantified the temporal coordination between kernel filling and dehydration rates, and developed a thermal-time-based LogisticPower model (M=901+(Ta)b) using two independent field trials (multi-ecological network from 2023 to 2025 and multi-sowing date from 2024 to 2025) conducted in the Hexi Corridor, a representative sub-region of the Northwest irrigated maize area in China. Our results showed that kernel dehydration rates at Pre-PM (1.35∼1.85%/day) and Post-PM (0.30∼0.75%/day) were primarily determined by genotype, though modulated by precipitation as an environmental covariate rather than a model input. While a simple positive correlation existed between filling and dehydration rates (r = 0.312), partial correlation controlling for filling duration revealed a negative association (r = −0.218), indicating that kernel filling duration may confound the simple correlation; a formal mediation analysis was not performed, so this reflects a statistical association rather than an established causal mediating effect. Some cultivars showed the co-occurrence of relatively rapid filling and rapid dehydration under the tested conditions, but this association requires further validation before being treated as a breeding target. The LogisticPower model exhibited excellent predictive performance across environments (CCC: 0.921∼0.961; RMSE: 3.45∼4.58%; R2: 0.932∼0.958). The average kernel moisture content at physiological maturity in this sub-region was 33.5%. Critically, the thermal time from sowing to 25% kernel moisture content was validated as a relatively stable index with high cross-environmental robustness under well-watered and well-fertilized conditions for cultivar screening, providing crucial support for breeding high-yielding cultivars adapted to mechanical kernel harvesting.