Random regression models for multi-environment, multi-time data from crop breeding selection trials

Context In order to identify best crop genotypes for recommendation breeders, and ultimately use in breeding, evaluation is usually conducted field trials across a range of environments, known as multi-environment trials. Increasingly, many breeding traits are measured over time, example with high-throughput phenotyping at different growth stages annual crops or repeated harvests perennial crops. Aims This study aims provide an efficient, accurate approach modelling genotype response time accounting non-genetic sources variation such spatial temporal correlation. Methods Because the aim selection, genetic effects fitted random effects, so based on regression, which linear non-linear models used model responses. A method fitting regression outlined situation, using underlying cubic smoothing splines mean trend time. illustrated six wheat experiments, data grain-filling thermal Key results The correlates providing predicted responses while incorporating correlation between observations. Conclusions provides robust predictions by simultaneously under various situations including those have measurement times where within not same times. facilitates investigation into environment interaction (G × E) both environments. Implications presented potential increase accuracy trials, other than observed, give greater understanding G E interaction, hence improving selection environments repeated-measures traits.