Context. Morphometric digital analysis of plant seeds enables taxonomic discrimination of species based on morpho-colorimetric traits, and may be used to classify genotypes of wheat (Triticum aestivum L.).

Aims. This study was focused on the isolation and classification of cultivars and landraces of Iranian wheat based on morpho-colorimetric traits, and the prediction of yield and seedling vigour based on these traits.

Methods. In total, 133 wheat genotypes (91 native landraces and 42 cultivars) were evaluated by alpha lattice design in two crop years (2018–19 and 2019–20) under rainfed and conditions. After seed harvesting, 40 morpho-colorimetric traits of wheat seeds were measured by imaging. Seed colour, morphometric seed, seed vigour and yield were also assessed.

Key results. Using linear discriminant analysis based on morpho-colorimetric traits, wheat cultivars and landraces were separated with high validation percentage (90% in well-watered and 98.6% in rainfed conditions). Morpho-colorimetric traits L, Whiteness index, Chroma, a, Feret and Rectang were found to be the most discriminant variables in the rainfed field. In analysis based on seed colour according to descriptors of the International Union for the Protection of New Varieties of Plants and International Board for Plant Genetic Resources, wheat genotypes were classified into four groups with high accuracy by using linear discriminant analysis. Specifically, 97.3% could be identified as yellow and 99.7% as medium-red wheat groups.

Conclusions. Our observations suggest that seed digital analysis is an affordable and valuable approach for evaluating phenotypic variety among a large number of wheat genotypes. Morphometric analysis of cultivars and native populations can provide an effective step in classifying genotypes and predicting yield and seedling vigour.

Implications. Morphometric databases will help plant breeders when selecting genotypes in breeding programs.