Harnessing faba bean MAGIC populations for enhanced protein content, yield, and agronomic performance in diverse environments

Faba bean (<i>Vicia faba</i> L.) is an increasingly important source of plant-based protein; however, improving both grain yield and protein content remains a major breeding challenge. This study aimed to identify superior haplotypes combining these traits using a large multi-parent advanced generation inter-cross (MAGIC) population. A total of 2,431 MAGIC lines were developed through two rounds of eight-way F1 inter-crossing among 28 parental lines and evaluated during the 2023/2024 growing season at Terbol, Lebanon, and Marchouch, Morocco, using an augmented design with four repeated checks. Phenological and yield-related traits were recorded, and protein content was quantified using both the Kjeldahl method and Near-Infrared Spectroscopy (NIRS; FOSS DS2500). The grain-based calibration model developed using modified partial least squares regression demonstrated high predictive accuracy (R² = 0.983), with cross-validation and training/test split confirming its reliability (R² = 0.82). Flowering and maturity periods were longer at Terbol, where significantly higher numbers of pods and seeds per plant, grain yield, and biological yield were also observed. Protein content ranged from 12.0% to 30.4% and showed a strong positive correlation between locations (r = 0.6, <i>p</i> < 0.001), indicating high repeatability across environments. Notably, protein content was not correlated with grain yield, highlighting the potential for simultaneous improvement of both traits. Ten transgressive lines exhibiting at least 25% higher protein content and grain yield than the parental checks were identified across sites. These findings demonstrate the effectiveness of MAGIC populations for capturing genetic variability and accelerating the development of high-yielding, protein-rich faba bean cultivars, providing valuable resources for future breeding programs.