Predictive models for phenotyping and classification of wheat cultivar viability

Given the global importance of wheat cultivation in the agricultural landscape, the practicality of the tetrazolium test in post-harvest management of seed lots, and the significant increase in the use of technologies in agriculture, this work aimed to evaluate predictive models for high-efficiency phenotyping and classification of the viability of different commercial wheat cultivars using the tetrazolium test. The predictive models proved accurate in estimating the viability of the 50 seed lots, reaching over 90% viable seeds, depending on the wheat cultivar. The recommended tetrazolium salt solution concentrations were 0.125% for the BRS 264 cultivar, 0.1% for MGS Brilhante and BRS 404, and 0.075% for TBIO DUQUE and BRS 394, with a pre-conditioning period of 9 hours, regardless of the cultivar. Correlations between the percentage of viable seeds obtained by the tetrazolium test and physiological vigor variables were statistically significant, serving to validate each chosen predictive model, as well as to rank the seed lots by cultivar. Through computational phenotyping of over 10,000 seed images, and subsequently, individual digital analyses of the respective embryonic tissues, the cultivars BRS 264, TBIO DUQUE, and BRS 404 were classified into four viability classes, and MGS Brilhante and BRS 394 into three classes. Therefore, predictive models, specific to each cultivar, associated with the tetrazolium test and image analysis resources, represent significant advances for decision-making regarding the implementation or post-harvest management of wheat crops, especially cultivars planted under different climates and regions. • Predictive models reliably estimate wheat seed viability. • Computational phenotyping of 10,000+ images enables rapid seed classification. • Tetrazolium combined with digital analysis allows precise post-harvest decisions. • Wheat cultivars ranked based on vigor and specific viability classes. • Optimized cultivar-specific protocols ensure reliable evaluation across regions.