Harnessing Faba Bean MAGIC Populations for Enhanced Protein Content, Yield, and Agronomic Performance in Diverse Environments

Lynn Abou Khater¹Fouad Maalouf^1*Cassandra Walker²Outmane Bouhlal³Khawla Aloui³Rind Balech⁴Kaoutar El Mahmoudi⁵Mohammed Ibriz⁵Francis Chuks Ogbonnaya⁶Shiv Kumar³

• ¹International Center for Agricultural Research in the Dry Areas (Lebanon), Verdun, Lebanon
• ²AgriBio Centre For AgriBioscience, Bundoora, Australia
• ³International Center for Agricultural Research in the Dry Areas Morocco, Rabat, Morocco
• ⁴International Center for Agricultural Research in the Dry Areas, Beirut, Lebanon
• ⁵Universite Ibn Tofail Faculte des Sciences, Kenitra, Morocco
• ⁶Australian Government Grains Research and Development Corporation, Barton, Australia

Faba bean (Vicia faba L.) has gained prominence as a key source of plant-based proteins owing to its high protein concentration, and substantial yield potential. To identify superior haplotypes for grain yield and protein content, 2,431 MAGIC lines were developed through two rounds of eight-way F1 inter-crossing of 28 crosses. Field trials were conducted using augmented design with four repeated checks at two locations, Terbol in Lebanon and Marchouch in Morocco during 2023/2024. Observations were recorded on phenological and yield-related traits. Protein content was analyzed using both the traditional Kjeldahl method and Near-Infrared Spectroscopy (NIRS) using FOSS DS2500 system. Calibration model was developed using Modified Partial Least Squares (MPLS) regression for grain samples. The grain-based calibration model showed strong predictive accuracy with an R² of 0.983. Cross-validation confirmed the model's reliability, with an additional training/test split yielding an R² of 0.82. Flowering and maturity periods were longer at Terbol compared to Marchouch. Significantly higher values for the number of pods and seeds per plant, grain yield, and biological yield were recorded at Terbol. Protein content ranged from 12.0% to 30.4% with strong correlation between locations (r = 0.6, p < 0.001), demonstrating high repeatability of protein estimates at both locations. Notably, protein content and grain yield did not show any correlation, indicating potential for simultaneous selection for both the traits. Based on the performance at both sites, the top 10 lines showing transgressive segregation with at least 25% higher protein content and grain yield than the parental checks were identified.