The rapidly growing global population, combined with climate change, is placing unprecedented pressure on food production systems, raising major concerns related to food security, greenhouse gas emissions, and environmental sustainability. In response, diversification toward alternative plant-based protein sources has become increasingly important. Global demand for plant-based protein is rising rapidly, driven by population growth and shifting consumer preferences toward healthier and more sustainably sourced diets. The global plant protein market, valued at USD 29.4 billion in 2020, is projected to increase more than fivefold by 2030, potentially accounting for 7.7% of the total global protein market (Elkin, 2021). Legumes are central to this transition due to their low environmental footprint and high nutritional value, and they remain a primary source of dietary protein in many regions, particularly South Asia and the Middle East.

Among grain legumes, faba bean (Vicia faba L.) has emerged as a promising crop, containing two to three times more protein than cereals and higher protein levels than many other legumes, including peas, lentils, beans, and chickpeas (Chávez-Murillo et al., 2018; Rahate et al., 2021). Faba bean seeds typically contain 24–33% protein, along with carbohydrates, dietary fiber, minerals, bioactive compounds, and essential amino acids suitable for human and monogastric animal nutrition (Crépon et al., 2010; Koivunen et al., 2016; Valente et al., 2019; Khazaei et al., 2019). However, antinutritional compounds such as tannins, vicine, and convicine may limit utilization, particularly in non-ruminant diets. These compounds can be substantially reduced or eliminated through targeted breeding and processing approaches (Cabrera and Martín, 1986; Luo and Xie, 2013; Khazaei et al., 2017; Sharan et al., 2021).

Despite recent increases in global faba bean production, supply remains insufficient to meet growing demand. In 2024, global imports reached approximately 1 million tons of dry seeds and 0.06 million tons of green seeds (FAOSTAT, 2026). Closing this demand–supply gap requires accelerating yield gains and narrowing yield gaps. Globally, average dry faba bean yield increased from less than 1 t ha⁻¹ in 1961 to approximately 2.2 t ha⁻¹ in 2024, corresponding to an annual increase of about 1.28% (FAOSTAT, 2026). This yield improvement has also resulted in increased protein yield, rising from approximately 250 kg ha⁻¹ to nearly 550 kg ha⁻¹, underscoring the crop’s growing contribution to global plant protein production.

Improving grain yield and protein concentration simultaneously remains a key objective of faba bean breeding programs. Grain yield is a complex trait strongly influenced by environmental conditions and genotype × environment (G×E) interactions. Previous studies examining the relationship between grain yield and protein content in faba bean have reported inconsistent results, ranging from no clear association (Picard, 1977) to significant negative (Lizarazo et al., 2015; Barłóg et al., 2019) or positive correlations (El-Sherbeeny and Robertson, 1992; Skovbjerg et al., 2020). These contrasting findings highlight the complex and context-dependent nature of protein–yield relationships, emphasizing the need for advanced populations and multi-environment evaluations.

Biparental populations have been widely used in faba bean to identify QTLs associated with biotic and abiotic stress tolerance and key agronomic traits (Román et al., 2003; Arbaoui et al., 2008; Cruz-Izquierdo et al., 2012; Catt et al., 2017). However, their limited genetic diversity and restricted recombination reduce mapping resolution and statistical power for dissecting complex, polygenic traits such as grain yield and protein content. These limitations can be effectively addressed using Multi-parent Advanced Generation Inter-Cross (MAGIC) populations, which are developed by intercrossing multiple genetically diverse founder lines, resulting in high levels of recombination and a broad genetic base (Cavanagh et al., 2008; Huang et al., 2015; Scott et al., 2020). MAGIC populations have proven particularly powerful for resolving the genetic architecture of complex traits and for identifying loci with small to moderate effects across diverse environments. In faba bean, MAGIC populations have been successfully used to dissect traits including frost tolerance, flowering time, seed size, disease resistance, and yield-related traits (Sallam and Martsch, 2015; Khazaei et al., 2018; Skovbjerg et al., 2023). Notably, an eight-parent MAGIC population developed at ICARDA integrates diverse genetic backgrounds spanning the paucijuga, minor, and equina types, with substantial variation in protein content, seed size, grain yield, and phenological traits (Maalouf et al., 2019). This population provides a robust platform for simultaneously evaluating grain yield, protein concentration, and protein yield, as well as their interactions across environments, thereby supporting the development of high-performing faba bean cultivars for sustainable plant-based protein production.

This study exploits one of the largest faba bean MAGIC populations reported to date, developed through two cycles of eight-way intercrossing, enabling high recombination and broad genetic diversity. Unlike previous faba bean MAGIC studies that primarily focused on yield components or genetic mapping under limited conditions, this work integrates multi-environment field evaluation with high-throughput protein phenotyping to jointly dissect grain yield and protein content. The objectives of the present study were to evaluate this MAGIC population for grain yield and protein content, assess the feasibility of selecting lines combining high yield and protein content, and examine the relationships between protein content and other key agronomic traits.

This study evaluated protein content and a range of agronomic traits in MAGIC lines derived from an eight parental MAGIC population grown at two distinct locations. Significant differences among recombinant inbred lines (RILs) were expected, given that these lines were developed from the combination of eight genetically diverse parents representing different botanical groups of faba bean. Also, MAGIC populations are known for their high genetic diversity, which allows for a wide range of phenotypic traits to be expressed (Jiang and Xu, 2024).

Faba bean MAGIC lines flowered and matured earlier at Marchouch than at Terbol, due to higher temperatures and terminal drought conditions, which accelerated growth and development as part of a drought escape response. These stress conditions also negatively impacted growth, leading to reduced plant height and yield traits at Marchouch. The observed reductions are likely linked to impaired reproductive development, limited water uptake, and decreased photosynthetic efficiency under heat and drought conditions (Awasthi et al., 2014; Bishop et al., 2016; Abdelhaleim et al., 2022). Similar observations have been reported in faba bean (Maalouf et al., 2015; Siddiqui et al., 2015; Abou-Khater et al., 2022) and other legumes like chickpea (Awasthi et al., 2014), lentil (Sehgal et al., 2017; Choukri et al., 2020), and common bean (Beebe et al., 2013; Seidel et al., 2016).

The average protein content in MAGIC lines was slightly lower at Marchouch compared to Terbol. The lower protein content observed at Marchouch was likely due to impaired nitrogen fixation and reduced protein biosynthesis, which are typical responses to heat and drought that limit protein accumulation in seeds (Apel and Hirt, 2004; Kabbadj et al., 2017). However, due to the large number of accessions evaluated in this study, direct physiological measurements related to nitrogen fixation or assimilation were not feasible. Therefore, this interpretation should be regarded as a working hypothesis that requires future validation through targeted approaches such as on-site δ¹⁵N determination and post-anthesis nitrogen remobilization assessment. Fixed nitrogen supports both vegetative biomass accumulation and reproductive development, thereby influencing seed yield and grain protein content (Neugschwandtner et al., 2015). Similar reductions in seed protein content have been reported in lentil, chickpea and common bean under heat and combined heat and drought (Choukri et al., 2020; Behboudian et al., 2001; Ghanbari et al., 2015). While some studies have observed increases in certain stress related proteins under stress (Pareek et al., 1998), the overall reduction observed in our study supports the notion that environmental stress generally suppresses protein synthesis in faba bean. These differences are also likely influenced by genetic variation, as drought-tolerant varieties often continued to produce certain stress-related proteins, while susceptible ones showed greater reduction in protein content (Abdellatif et al., 2012; Belal et al., 2018; Hummel et al., 2018; Samineni et al., 2022). Protein content measured at both locations showed a moderately strong and statistically significant correlation. This indicates that, although the overall protein contents were lower at Marchouch, the ranking of the lines remained fairly consistent at both locations Therefore, genotypes that performed well at one location tended to perform well at the other, indicating reliable protein estimates across environments.

Understanding the relationship between protein content and other traits is crucial for developing breeding strategies that enhance protein content while maintaining yield and other important agronomic traits. Our findings highlight that environmental conditions influence how protein content correlates with other agronomic traits. Notably, the absence of a yield–protein trade-off observed in this study refers to the lack of a significant negative relationship between protein content and grain yield across environments, indicating that improvements in one trait do not necessarily reduce the other.

At Marchouch, protein content was not correlated with grain or biological yield. This suggests that both traits can be selected independently during breeding under similar environmental conditions, with improvements in one not necessarily limiting progress in the other. Previous studies have also reported an absence of correlation between protein content and seed yield in faba bean (Picard, 1977), as well as in lentil (Choukri et al., 2020) and in chickpea when evaluated under rainfed conditions in Southern Italy where the climate is similar to Marchouch (Sellami et al., 2021). On the other hand, positive correlations were observed between protein content and the number of pods and seeds per plant, indicating that greater pod and seed numbers are associated with high protein content. In contrast, other studies reported a negative correlation between these traits in many legume crops, such as lentil, pea and chickpea (Stoddard et al., 1993; Lizarazo et al., 2015; Gaur et al., 2016). Additionally, shorter plants with earlier flowering and maturity tended to have higher protein content, making them suitable targets for breeding in stress-prone environments. Although a negative correlation was found between protein content and seed size, the relationship was weak, suggesting that it may be possible to identify genotypes that combine both large seed size and high protein content. Lafiandra et al. (1981) also reported a negative correlation between faba bean protein content and seed size and Gaur et al. (2016) reported similar observations in chickpea.

In contrast, at Terbol, with its cooler and wetter climate, protein content showed positive correlations with plant height, grain yield, seeds per plant, and seeds per pod indicating that in favorable environments, it is possible to simultaneously improve yield and protein content. This might be attributed to efficient nitrogen fixation under favorable conditions, enabling plants to support these traits without the typical trade-off. Additionally, taller plants often have more efficient nutrient uptake, including nitrogen fixation, which contributes to protein synthesis. These findings are consistent with previous research (El-Sherbeeny and Robertson, 1992) that reported positive correlations between protein content and grain yield in faba bean and demonstrated that selecting for both yield and high protein content is achievable. Nevertheless, like Marchouch, protein content remained negatively correlated with pod number, seed size, and days to flowering, suggesting some consistent trade-offs at different environments. The absence of a yield–protein trade-off in this study likely reflects both genetic and physiological factors. Extensive recombination in the MAGIC population may have disrupted unfavorable linkages between loci controlling grain yield and protein content, allowing favorable alleles for both traits to be combined (Jiang and Xu, 2024). In addition, as a nitrogen-fixing legume, faba bean can support protein accumulation without directly competing with yield formation. Previous studies in faba bean and other legumes have similarly reported an absence of correlation between seed yield and protein content, attributed to the capacity for biological nitrogen fixation (Frimpong et al., 2009; Stoddard et al., 1993; Picard, 1977; El-Sherbeeny and Robertson, 1992). Cluster analysis revealed considerable genetic variability among the faba bean MAGIC lines, with nine clusters identified at Marchouch and seven at Terbol. These differences reflect both genotypic diversity and the influence of different environmental conditions. Furthermore, the observed variability among clusters also highlights the advantage of using multiparent lines for selection.

The ranking of clusters C7 and C1 at Terbol demonstrated that simultaneous improvement for protein content and grain yield is achievable, aligning with the positive and significant correlation observed between these two traits at this location. However, cluster C8 at Marchouch also showed strong performance for both traits, despite the absence of a correlation between protein content and grain yield in that environment. This likely reflects the use of multiparent lines in this study, as their broad genetic diversity increases the chance of identifying plants that combine multiple traits, like high yield and high protein content, making them a valuable resource for combined trait selection in breeding programs.

The PCA results visualized through biplot provide a comprehensive overview of the diversity among the MAGIC lines and their relationships with the studied traits. It also underscores the complexity of trait interactions in faba bean and their strong dependence on environmental conditions. At Marchouch, protein content contributed minimally to the total trait variation and showed a weak association with grain yield and its components as indicated by the nearly perpendicular vectors in the biplot. The observed trend in trait distribution confirms the absence of correlation between them in this environment.

In contrast, the PCA conducted at Terbol, a more favorable environment, revealed that although protein content had a minimal contribution to the total trait variation it showed stronger positive associations with the number of seeds and grain yield. While this association was not very strong, it suggests a higher likelihood of simultaneous improvements for protein content and grain yield in environments similar to Terbol. The closer alignment of protein content with traits like grain yield, seed number, and pod number supports the concept that under optimal conditions, plants can absorb and translocate nitrogen into the seed, improving both yield and seed quality.

Importantly, the lines exhibiting superior performance across multi-trait, multi-environment evaluations underscore the utility and robustness of MAGIC populations. The populations effectively combined diverse alleles to generate superior lines and enhance genetic gain. Their broad genetic diversity likely facilitates the identification of recombinant lines that integrate high protein content with favorable agronomic traits, even under environmental constraints. These lines represent valuable resources for breeding programs focused on the improvement of multiple traits and for investigating genotype-by-environment interactions underlying complex trait expressions. A total of 403 faba bean MAGIC lines exhibiting high protein content and grain yield relative to two Australian cultivars, Ascot and Icarus, were selected at each location. These lines will be re-evaluated in Australia in faba bean production areas. Re-clustering of the selected lines enhances the understanding of trait profiles, facilitates the identification of high-performing lines with shared agronomic characteristics, and enables assessment of the stability of trait-combinations across environments. This study also provides a valuable resource for targeted breeding efforts.

The selected lines exhibited a distinct clustering pattern compared to the full set, indicating differentiation in their trait profiles. At Marchouch, lines exhibited distinct trait profiles within clusters. MAGIC lines in C1 showed strong potential for maximizing seed yield. C2 was characterized by favorable performance in key yield components (BYPLT, NSPLT, NPPLT). C3 appeared to benefit from an extended vegetative period supporting biomass accumulation and seed filling; this group included tall lines with larger seeds and longer crop cycles. MAGIC lines in C4 combined high values for yield components with elevated biological yield per plant, suggesting the presence of highly productive recombinant lines carrying favorable alleles for both yield and biomass traits.

At Terbol, cluster analysis of the selected high-yielding and high-protein MAGIC lines revealed a distinct grouping pattern. C1 and C2 comprised of tall plants with large seeds, a higher number of pods, and extended growth periods, traits likely influenced by specific combinations of parental alleles. C3 is particularly valuable for selection and breeding programs targeting grain yield improvement in Terbol-type environments, as it includes MAGIC lines with high grain yield and yield components, potentially harboring novel combinations of yield-enhancing alleles generated through the MAGIC design. C4 contained lines with favorable yield, plant height, and pod number per plant, suggesting a balanced inheritance of alleles contributing to both biomass and yield-related traits under the environments assessed in this study.

The contrasting clustering patterns observed at Marchouch and Terbol highlight the significant influence of genotype-by-environment interactions on the performance of recombinant MAGIC lines. The diverse genetic backgrounds within the MAGIC population enable a wide spectrum of phenotypic responses across environments, highlighting the effectiveness of the intercrossing and selection strategy in generating novel and potentially superior genotypes.

The top performing 51 MAGIC lines across locations were selected for further analysis. Among these, 10 lines – 119548, 119501, 121678, 121451, 121613, 121189, 121678, 121876, 121843 and 121843 – consistently exhibited the highest protein content and grain yield. These lines emerged as the most promising, demonstrating stable performance across environments. Notably, they demonstrated transgressive segregation, outperforming both the parental lines and check cultivars, thereby highlighting the effectiveness of the MAGIC design in generating novel and favorable allele combinations and the unique value of this MAGIC population for protein-oriented faba bean breeding. Similar patterns of transgressive segregation have also been reported in eight-parent MAGIC populations of common bean and cowpea, supporting the broader applicability of this breeding strategy (Huynh et al., 2018; Diaz et al., 2020).

In conclusion, this study highlights the value of MAGIC populations in accelerating genetic gain through enhanced trait dissection and selection. Their broad genetic base, higher recombination rates, and increased phenotypic diversity facilitate the identification of superior lines that combine high protein content and grain yield. The identification of top-performing lines demonstrates the potential of MAGIC populations to simultaneously improve nutritional quality and yield stability, offering a powerful tool for breeding resilient cultivars, particularly under the environmental conditions evaluated in this study. Furthermore, the extensive genetic diversity and recombination within this population provide an excellent foundation for future genome-wide association studies (GWAS) and quantitative trait loci (QTL) mapping to dissect the genetic basis of yield, protein content, and other agronomic traits, thereby supporting targeted breeding strategies in faba bean.

As a next step, re-sequencing the top-performing lines using the published 13K faba bean SNP array could identify identity-by-descent (IBD) segments and generate marker-assisted selection (MAS) resources to accelerate breeding. In parallel, multi-location G×E trials in diverse environments will be conducted to evaluate the stability and adaptability of the selected lines.