Application of non-destructive MARVIN and NIR seed screening of pre-breeding material: A case study on common bean ( <i>Phaseolus vulgaris</i> ), white lupin ( <i>Lupinus albus</i> ) and Andean lupin ( <i>L</i> . <i>mutabilis</i> )

Grain legumes are important for sustainable agriculture due to their agronomic and nutritional properties. In this study, a rapid, high-throughput, non-destructive seed screening was performed on 835 lines of Phaseolus vulgaris , Lupinus albus and L . mutabilis grown under a sub-alpine climate. The lines were analysed for seed size characteristics (thousand seed weight ‐ TSW, width and length) using a MARVIN analyser and for protein and moisture content using near-infrared transmittance spectroscopy. The results showed a strong correlation between TSW and seed size in all species, while protein content varied inversely with moisture content, especially in the Lupinus species. Regression analysis identified moisture content as a key factor for protein content, which explained up to 86.2% of the variance in L . mutabilis . Principal component analysis showed that trait variation was more strongly influenced by genetic and other environmental factors than by geographical origin. Lines selected based on normalised rank (NR) and weighted trait index (WTI) methods were characterised by balanced trait profiles, such as high TSW and protein content with optimal moisture contents. These results underline the importance of integrating non-destructive screening methods into pre-breeding programs. In this way, breeders can efficiently select legume seeds with superior agronomic and nutritional traits, such as optimal seed size, high protein content and adequate moisture. This approach not only improves the selection process, but also ensures the development of legume varieties that meet both agricultural and nutritional standards.