Chickpea (Cicer arietinum L.) is a globally important food legume and a major source of protein in human diets (Baite et al., 2017). It contributes substantially to soil fertility through biological nitrogen fixation, making it an ideal rotational crop in sustainable cereal-based systems (Saraf et al., 1998).
       
Chickpea is cultivated under rainfed systems in semi-arid Mediterranean environments (Bouri et al., 2019). It requires relatively low water (300-400 mm), but its production is increasingly constrained by irregular rainfall, terminal drought during flowering and pod filling (Street et al., 2016) and high temperatures. In fact, terminal drought can potentially reduce yields by 58-95% compared to irrigated conditions (Farooq et al., 2017).
       
The semi-arid Algerian environments are broadly representative of Mediterranean rainfed systems (Djellouli, 2025), where water scarcity is increasingly critical from eastern to western regions (Alrteimei et al., 2022).
       
Developing varieties with drought tolerance and adaptation to variable rainfall is crucial to sustain productivity and food security under climate variability (Arriagada et al., 2022). Chickpea breeding pathways include: (i) pure (bred) varieties, genetically homogeneous; (ii) participatory selection varieties, co-developed with farmers to enhance local adaptation and adoption; and (iii) heterogeneous local varieties, maintained by farmers and characterized by high intra-population diversity and resilience under marginal rainfed conditions (Raggi et al., 2019).
       
Expanding the genetic base enhances chickpea resilience to drought and heat stress while improving productivity (Kharrat et al., 2023). Nevertheless, some self-pollinating homogeneous varieties retain notable tolerance to abiotic stresses (Panigrahi et al., 2024). In the context of increasing climatic variability and a limited genetic pool (Varshney et al., 2024), it is therefore important to determine not only the best-adapted genotypes but also the most appropriate type of classical selection to adoptto maximize productivity for long-term resilience.
       
Understanding genotype × environment is critical for guiding varietal selection toward yield stability and resilience under variable semi-arid conditions. While multi-year trials are ideal, evaluating varieties across contrasting environments within a single year can provide robust insights into stability and adaptation, as demonstrated in recent studies (Khan et al., 2024).
       
To capture these interactions, this study employs Additive Main Effects and Multiplicative Interaction (AMMI) and Genotype plus Genotype × Environment (GGE) biplot analyses, integrating robust statistical modeling with visual interpretation (Laxuman et al., 2025; Shekhawat et al., 2024; Annicchiarico, 2020; Zobel et al., 1988).
       
The specific objectives of this study are to: (i) evaluate the performance of chickpea varieties differing by breeding pathway across six contrasting Algerian environments; (ii) quantify G×E interactions for grain yield and stability; and (iii) identify varieties most stable and adapted for rainfed semi-arid conditions, provid guidance for sustainable varietal deployment.

Genetic material
 
Ten varieties of chickpea were obtained through different selection methods; their pedigree, origins and supplying institutes are listed in Table 1. They correspond to local varieties (LV), participatory selected varieties (PS) and introduced varieties (IV) selected by ICARDA.


 
Experimental sites and environmental variability
 
Six sites, favorable for chickpea cultivation, with contrasting conditions (climatic and edaphic) were selected for the study and are illustrated in Fig 1 (S1 to S6).


       
Multiple Factor Analysis  was carried out using climatic data (Algerian National Office of Meteorology) for twenty-three years (2000 to 2022) and soil analysis (2021-2022). This enabled us to confirm the environmental variability of the sites along the axes 1-2-3.
       
For each site, we assigned a code to the corresponding environment by identifying its most distinctive characteristics (Table 2).


 
Trial setup
 
The experiments were conducted during the 2021-2022 season on a farmer’s field. The climatic and edaphic characteristics of each experiment by site are presented in Table 2.
       
Climatic data for 2021-2022 (Fig 2) indicate higher or stable rainfall from March to May (early spring) compared to the 23-year average, but reduced precipitation from December to February (winter) and after May. This atypical winter deficit contrasts with the usual spring water stress in these semi-arid regions and aligns with recent reports of declining winter rainfall, shorter wet seasons and greater variability in northern Algeria (Bessaoud and Montaigne, 2009; Meddi et al., 2010; Djellouli and Laborde, 2022). Mean temperatures remained close to the historical average.


 
Experimental design and recorded traits
 
The experiment was laid out as a completely randomized block design with three replicates. Elementary plots measured 6 m × 1.8 m and consisted of four rows sown at 20 seeds m^-1 (150 kg ha^-1) with a thousand-seed weight of 340 g. The crop cycle was between December and June for all sites except site 1 (sowing in February).
       
Six agronomic traits were selected based on their relevance to the yield, plant architecture and adaptation to environmental constraints (Singh et al., 2020): grain yield, plant height, number of primary branches, stem length to the first pod, 100-seed weight and the harvest index.
 
Statistical analysis
 
Analysis of variance (ANOVA) was conducted to assess yield differences among varieties within each environment, aiming to test the variety effect and to classify the ten cultivars individually or according to their breeding pathways. In addition, a factorial ANOVA was applied to several agronomic traits measured in the six environments, considering the main effects of the selection pathway, environment and their interaction. Analyses were carried out using RStudio software (version 2023.12.0+369) Multiple comparisons of the means were performed using Tukey’s test.
       
Statistical analyses of the Additive Main Effects and Multiplicative Interaction (AMMI) Zobel et al., (1988) and genotypic main effect + genotype-by-environment interaction (GGE) biplot models (Ponsiva et al., 2024) were performed using the “metan” package in RStudio software version 12.0. AMMI models were applied to study the G×E interaction, adaptation and stability of varieties across the six environments.
               
According to Gauch et al., (2008), the display of nominal variety yields (AMMI-1) clearly describes the winning varieties and adaptive responses.

Grain yield performance across environments
 
ANOVA for grain yield revealed significant differences (1.01 t·ha^-1 to 2.18 t·ha^-1) among varieties across most environments, except for E1 (Table 3).


       
Participatory selection varieties showed the most consistent yields across environments, performing best under moderately productive conditions. Local varieties were well adapted to favorable environments, particularly under adequate spring rainfall and exhibited moderate stability. In contrast, introduced varieties displayed the highest variability, occasionally achieving high yields but often underperforming, indicating greater sensitivity to environmental conditions.
 
Effects of breeding pathways, environment and their interaction on morpho-agronomic traits
 
The environment had a highly significant effect on all traits, except Number of primary branches (Table 4), highlighting the major role of local conditions in shaping phenotypic expression. The pathway × environment interaction was especially pronounced for 100-grain weight. 


       
The results indicate that the selection method influences the genetic response of the lines depending on the environment, emphasizing the importance of evaluating varieties under diverse conditions to fully capture their potential.

AMMI and GGE analyses and multivariate exploration of yield stability and genotype × environment interaction
 
Analysis of variance (Table 5) revealed significant effects of genotype and environment, as well as a highly significant G×E interaction. The first principal component (PC1) explained 60.3% of the interaction variance, reflecting a clear environmental gradient and well-structured response.


       
The AMMI analysis identified G6-PS, G8-IV and G2-LV as the most productive varieties. G8-IV, G4-PS and G7-LV exhibited the highest stability values (Table 6). Considering both high yield and low instability, the introduced variety G8-IV emerged as the best-performing and most broadly adapted genotype, followed by G4-PS and G2-LV.


       
Regarding environments, E4 was clearly separated along the negative side of PC1, reflecting specific conditions highlighting the performance of certain varieties (Fig 3). E2, E5 and E6 were positioned on the positive side of PC1, representing moderately differentiated environments favorable to other genotypic profiles. E1 and E3, near the center of PC1, generated fewer G×E interactions and represented average environmental conditions.


       
The nominal stability plot (Fig 4) provided complementary insight: varieties with steep slopes (G6-PS, G10-IV, G5-PS) exhibited strong specific interactions, whereas G8-IV, G2-LV and G4-PS displayed more regular responses indicative of stability.


       
Fig 4 also highlighted significant environmental differences. E4 generated pronounced variation among varieties, reflecting strong specific interactions, while E1 and E3 elicited more homogeneous responses, identifying them as reference environments. E2, E5 and E6 showed intermediate structuring, differentiating certain varieties while maintaining overall consistency.
       
Finally, the GGE biplot (Fig 5) illustrated environmental structure. E5, with a long vector, was the most discriminating environment.


       
The evaluation of performance and stability of chickpea genotypes derived from different selection methods under contrasting environments is essential for identifying varieties with high resilience potential. AMMI and GGE biplot analyses proved effective in characterizing genotype × environment (G×E) interactions, facilitating the selection of varieties that combine high productivity and stability, key traits for adaptation to semi-arid conditions with increasingly irregular and limited rainfall. Environmental factors accounted for the largest portion of variation (over 25% of total variation), followed by G×E interactions and genotypic differences, a pattern reported in recent studies (Istanbuli et al., 2025; Eskezia et al., 2025). This confirms the predominant influence of the environment on chickpea grain yield, with GEI playing a secondary but significant role. These findings align with Din et al., (2024), who emphasized the contribution of both environmental and G×E effects to the stability and adaptation of Desichickpeagenotypes.
       
The contrasting performance of both homogeneous and heterogeneous varieties across environments suggests that stability is not solely determined by genetic background. Local varieties, often heterogeneous and co-adapted, continue to evolve under local conditions, whereas introduced varieties, developed through intensive breeding in foreign environments, display distinct morphological and agronomic traits.
       
The analysis across six experimental sites confirmed significant genetic variability and underscored the importance of G×E interactions in chickpea agronomic performance (Jagan et al., 2023; Ligarreto-Moreno and Pimentel-Ladino, 2022). Environments E1, E4 and E5 were the most discriminating, but only E4 and E5 displayed high productive potential, supporting the relevance of current chickpea zoning and multi-location breeding strategies (Pérez et al., 2005). This highlights the need to consider agro-climatic specificity rather than generalizing performance across apparently similar zones (Srivastava et al., 2024).
       
Contrary to commonassumptions, varieties introduced through international programs such as ICARDA are not inherently less plastic than local varieties. For instance, G8 (FLIP 93-93C, Ghab 4), selected in the WANA region, exhibited both high yield and stability, confirming its adaptation to Algeria’s diverse semi-arid conditions. This observation agreeswith Meddi et al., (2010); Annicchiarico (2017) and Dhuria and Babbar (2019), demonstrating that varieties developed for Mediterranean arid zones can combine high yield potential with broad adaptation. Similarly, the locally participatory-selected variety G4 displayed substantial plasticity across environments, challenging the traditional dichotomy between plastic local varieties and specialized improved varieties.
       
Selecting genotypes such as G8-IV, G4-PS and G2-LV, which perform stably across most environments, provides a strong foundation for breeding broadly adapted varieties. These genotypes represent valuable resources for addressing climate change challenges, including drought, salinity and increasing precipitation variability (Bessah et al., 2023; Djellouli and Laborde, 2022; Khoshro and Maleki, 2025).
       
Joint AMMI and GGE analyses revealed that no single site encompassed the full range of constraints and opportunities in semi-arid zones. Environments E5 and E4 proved strategic for selection, offering a balanced compromise between experimental efficiency and representativeness by capturing both general and specific adaptations of chickpea varieties (Yan and Rajcan, 2002). Other sites showing redundancy or low discriminatory power could be excluded in an optimized experimental design (Gauch and Zobel, 1997).

Promising candidates, G8-IV, G4-PS and G2-LV, appear to be for direct use by farmers or breeding stocks in selection programmes. Their contrasting profiles, derived from different breeding methods, underscore the importance of diversifying the genetic sources for chickpea breeding. Multi-environment analysis showed that the E5 and E4 environments are particularly informative for breeding. These elements reinforce the relevance of flexible agro-ecological zoning for chickpeas in Algeria, enabling breeding to be oriented towards greater resilience of farming systems in the face of climatic variability.

This study was carried out in collaboration between Algeria’s National Institute of Agronomic Research (INRAA) and the Algerian Interprofessional Office of Cereals and Dry Vegetables (OAIC).We sincerely thank Mrs. Dalila Boukecha (INRAA) for her technical assistance and guidance and the farmers who provided their land for the field trials.
 
Disclaimers
 
The views and conclusions expressed in this article are solely those of the authors and do not necessarily represent the views of their affiliated institutions. The authors are responsible for the accuracy and completeness of the information provided, but do not accept any liability for any direct or indirect losses resulting from the use of this content.
 
Informed consent
 
All the experimental procedures and handling techniques were approved by the National Higher School of Agronomy (ENSA), Laboratory of Integrated Improvement of Plant Production.

The authors declare no conflicts of interest regarding the publication of this article. No funding or sponsorship influenced the conduct of the study, data collection, analysis, decision to publish, or manuscript preparation.