An Assessment of the Relationship Between 1000‑Kernel Weight Size and Ten Years of Historical 1000‑Kernel Weight Data from 26 Doubled Haploid Maize (Zea mays L.) Lines

Abstract Prediction of maize 1000‑kernel weight, defined as the estimation of potential harvest level in a given growing season, is of critical importance for both farmers and the consumer-oriented agri‑food sector. It also represents a key component of the cereal production process. The study tested the hypothesis that temporal trends represented by the year of observation can be used to predict 1000-kernel weight and that doubled‑haploid (DH) maize lines differ in the predictability of this trait. The plant material consisted of 26 DH maize lines developed by crossing two flint‑kernel cultivars. The lines were grown in the northern part of the Lower Silesia Voivodeship in Poland over a ten‑year period (2013–2022). To predict the trait values, linear, logarithmic, polynomial (second‑ and third‑degree), and power regressions were evaluated. The results demonstrated that the coefficients of determination obtained for the examined models were significantly correlated for all model pairs. In three cases (Lin–Log, Lin–Pow, and Log–Pow), a perfect association was observed, with correlation coefficients equal to 1.00. For line KN26, the overall best model fits were obtained across all five regression types. Therefore, this line may be recommended for further breeding programmes as the most predictable with respect to 1000‑kernel weight potential.