The common bean (Phaseolus vulgaris L.), is one of the most important food crops worldwide due to its widespread consumption and high nutritional value. Its grains contain approximately 50-60% carbohydrates and 15-25% proteins, in addition to significant amounts of vitamins and minerals (Cid-Gallegos et al., 2023). These characteristics make common bean a key component of food security, particularly in developing countries, where it represents an accessible source of nutrients and a strategic crop for income generation among small- and medium-scale farmers (Lamz-Piedra et al., 2023; Curay et al., 2024). Furthermore, common bean contributes to the sustainability of agricultural systems through its ability to fix atmospheric nitrogen via symbiosis with soil bacteria, thereby reducing dependence on synthetic fertilizers (Karavidas et al., 2022).
       
Genetic resources of P. vulgaris exhibit broad genetic diversity, mainly structured into the Mesoamerican and Andean gene pools, reflecting their evolutionary differentiation and high potential for breeding programs (Long et al., 2020; Delfini et al., 2021). In Peru, several studies have demonstrated that common bean germplasm possesses considerable phenotypic variability, representing an important basis for the selection of promising materials (Aybar-Peve et al., 2025a). in this sense, the use of multivariate analyses has proven effective for identifying patterns of variation and relationships among traits in common bean, highlighting the contribution of vegetative growth and reproductive traits to diversity structuring (Long et al., 2020). Complementarily, phenotypic diversity indices allow quantification of the variability present within germplasm collections, facilitating the identification of accessions with high agronomic potential (Aydin and Baloch, 2019; Ventura-Neyra et al., 2026). However, despite advances in the characterization of common bean in Peru, information regarding the phenotypic diversity of the accession collection maintained in the Germplasm Bank of the National Institute of Agrarian Innovation (INIA) remains limited, restricting the efficient utilization of these genetic resources in breeding programs. In this context, the present study aimed to characterize the phenotypic diversity of 66 common bean accessions from the INIA Germplasm Bank through the application of diversity indices and multivariate analyses, in order to identify materials with potential for breeding programs and contribute to the conservation and sustainable use of plant genetic resources.

The study was conducted at plot 7 of the Cercado Grande field at the Chincha Agricultural Experimental Station (EEA Chincha) of INIA, located in the district of Alto Larán, Chincha Province, Ica Department, Peru (13°27′29.58″S, 76°08′01.10″W; 61 m a.s.l.), during the period from May to October 2024. During this period, the average monthly temperature was 17.3°C, with maximum and minimum values of 21.8°C in October and 12.7°C in September, respectively. The average monthly relative humidity was 86.9%, according to data obtained from the National Service of Meteorology and Hydrology of Peru (SENAMHI, 2026).
       
Prior to the conduct of the experiment, a physicochemical characterization of the soil was performed. The results showed a soil pH of 8.3 and an electrical conductivity of 16.2 mS m^-1. Equivalent calcium carbonate content was 3.09% and organic matter content was 1.10%. Available phosphorus reached 398.29 mg kg^-1, whereas total nitrogen content was 0.06%. Soil texture was classified as loam. Regarding exchangeable bases, concentrations of Ca²⁺ [9.89 cmol(+) kg^-1], Mg²⁺ [2.42 cmol(+) kg^-1], Na⁺  [4.27 cmol(+) kg^-1] and K⁺  [0.57 cmol(+) kg^-1] were determined.
       
A total of 66 common bean accessions, representing part of the genetic diversity in the INIA Germplasm Bank, were evaluated. The regeneration sites of each accession, as well as their respective identification codes within the germplasm bank, are presented in Table 1.


       
To evaluate phenotypic diversity, 35 agro-morphological traits (18 qualitative and 17 quantitative) (Table 2) were analyzed according to the descriptors established by the International Center for Tropical Agriculture (CIAT) (Muñoz et al., 1993).


       
The accessions were arranged systematically in the experimental field, consisting of 66 subplots corresponding to experimental units, each representing a distinct accession. In each experimental unit, observations were recorded from five plants for each evaluated trait, except for DE, DF, DPM, DH and HSW, whose values were recorded at the experimental unit level.
       
Each experimental unit covered an area of 17 m² and consisted of four rows of 5 m length, with 0.85 m spacing between rows and 0.30 m between plants, resulting in a total experimental area of 3,231 m² including border effects. Three seeds were sown per hill and one row was left as separation between accessions.
       
All experimental units received the same agronomic management, including phytosanitary control of pests such as Bemisia tabaci, Crocidosema aporema and Empoasca kraemeri using imidacloprid (350 g L^-1), emamectin benzoate (50 g kg^-1) and alpha-cypermethrin (100 g L^-1) at doses of 0.6 mL L^-1, 0.5 g L^-1 and 0.75 mL L^-1, respectively. Furrow irrigation was used, with a total of seven applications during the growing season. Soil fertilization consisted of 80 kg ha^-1 of N, 100 kg ha^-1 of P₂O₅ and 50 kg ha^-1 of K₂O.
       
Hierarchical clustering analysis was performed using Gower distance as the dissimilarity measure and the Ward.D2 agglomerative method and the goodness of fit was evaluated through the cophenetic correlation coefficient. Quantitative traits were compared among the clusters obtained from hierarchical clustering through analysis of variance (ANOVA), followed by Tukey’s honestly significant difference (HSD) test. Subsequently, factorial analysis of mixed data (FAMD) was applied to the dataset. Additionally, principal component analysis (PCA) and multiple correspondence analysis (MCA) were performed to support the interpretation of quantitative and qualitative trait contributions identified in the FAMD. For qualitative traits, chi-square (χ²) tests were conducted and Shannon diversity (H′) and Pielou evenness (J′) indices were calculated. All analyses were performed using R version 4.5.1 (R Core Team, 2025) and R-Studio version 2026.1.0.392 (Posit Team, 2026), considering a significance level of α = 0.05.

Cluster analysis
 
Hierarchical clustering analysis based on Gower distance grouped the 66 accessions into four distinct clusters (Fig 1). The cophenetic correlation coefficient indicated a moderate fit (CCC = 0.63) between the original distance matrix and the generated dendrogram, suggesting an acceptable representation of the dissimilarity structure among the accessions. Higher values (CCC = 0.86) have been reported in previous studies that evaluated a lower number of quantitative and qualitative traits (de Paula et al., 2024), as well as fewer accessions and variables (Camargo-Cobeñas et al., 2026). This variation may be associated with the larger number of evaluated traits, the size of the germplasm collection and the environmental variability considered in the present study.


       
Tukey’s mean comparison analysis revealed significant differences among clusters for the evaluated quantitative traits (Table 3), reflecting broad phenotypic variability among accessions. Cluster I exhibited significantly higher mean values for MSL (176.54), PLL (7.53), PLW (8.34), CLL (12.06), PW (1.19) and DH (132.80), indicating predominance of traits associated with vegetative growth, larger leaf size and late maturity. Traits related to plant architecture, vegetative growth and phenology have been identified as important sources of differentiation in common bean due to their association with adaptation processes and genetic divergence (Long et al., 2020; González et al., 2021).


       
Cluster IV showed superior performance for yield-related traits, particularly NPP (21.12) and NSP (5.57), whereas Cluster II presented the lowest mean values for these variables (NPP = 10.67; NSP = 4.42), indicating lower expression of reproductive traits. Previous studies have highlighted these traits as key yield components and important selection criteria in common bean breeding programs (Long et al., 2020; Sundharaiya et al., 2023). These findings suggest that accessions from Clusters I and IV may constitute valuable parental sources for breeding programs aimed at improving vegetative vigor and yield potential, respectively. In contrast, Cluster III displayed intermediate phenotypic performance for most evaluated traits, indicating moderate agronomic differentiation relative to the remaining groups.
 
Factorial analysis of mixed data (FAMD)
 
The simultaneous integration of quantitative and qualitative traits is essential for understanding phenotypic variation in common bean and its usefulness in breeding programs (Scarano et al., 2014; Córdova-Sinarahua et al., 2026). In this context, the FAMD explained 18.2% of the total variation in the first two dimensions (Fig 2), revealing differentiation and grouping of the accessions into four clusters. Cluster III exhibited lower intragroup variability, whereas Cluster II showed greater dispersion despite being composed of fewer accessions. Recent studies in common bean reported 35.8% of explained variance in the first dimensions (Sinkovič et al., 2025), which may be associated with differences in the number of evaluated accessions and traits among studies.


       
Dimension 1 was mainly associated with traits related to yield (HSW), vegetative growth (MSL), leaf morphology (PLW, PLL and CLL) and seed traits (PSC and SSC) (Table 4), suggesting a phenotypic gradient associated with plant growth, organ size and seed variability. Similar results were reported by Sinkovič et al. (2025), who identified a high contribution of traits associated with seed weight, size and coloration in the differentiation of common bean accessions. Likewise, Nogueira et al., (2021) highlighted the importance of seed traits in breeding programs, whereas Ibrahim et al., (2020) indicated that seed color constitutes a relevant attribute for breeders, farmers and consumers. Dimension 2 showed predominance of variables related to yield (NPP and NSP), vegetative pigmentation (HC and VCL) and reproductive morphology (SPC, PSC, SS and PW), reflecting substantial variability in reproductive and pigmentation-related traits among accessions. Similar findings were reported by Ventura-Neyra et al. (2026) in lima bean accessions.


       
On the other hand, Dimension 3 was mainly determined by phenological variables (DF and DH), in addition to pod-related (PCM) and seed-related traits (PSC and SSC). Ventura-Neyra et al. (2026) and Machado et al., (2022) reported that traits associated with earliness contribute significantly to differentiation among accessions and represent important attributes for adaptation to different environments. Likewise, Sinkovič et al. (2025) reported that seed traits are determinant factors in the multivariate differentiation of common bean accessions.
 
Principal component analysis (PCA)
 
The variance explained by the first two principal components (48.7%; Fig 3) indicated an adequate representation of the phenotypic diversity among the evaluated accessions. Similar values have been reported in common bean and other legume species, where the first two dimensions explained between 24% and 70% of the total variability (Long et al., 2020; Dadther-Huaman et al., 2024).


       
PC1 showed predominance of variables related to yield and leaf morphology, particularly HSW, NPP, PLW, PLL and CLL, indicating an association between vegetative growth, organ size and productivity. Similar results were described by Aybar-Peve et al. (2025b) and Camargo-Cobeñas et al. (2026), who reported that yield-related traits constitute key factors in the phenotypic differentiation of legume accessions. PC2 was mainly determined by phenological variables such as DF, DPM and DH, coinciding with previous studies that identified phenology as one of the main components of variation in common bean (Meza-Vázquez et al., 2015; Long et al., 2020).
       
Positive correlations among HSW, PLW and CLL evidenced an association between vegetative growth and seed weight, whereas NPP and NSP showed an opposite pattern relative to leaf morphology and seed weight variables, suggesting possible physiological compensation mechanisms between the number of reproductive structures and individual seed size. Similar patterns have been reported in several legume species (Manson et al., 2025; Camargo-Cobeñas et al., 2026).
       
The distribution of accessions revealed differentiation among clusters. Cluster I was mainly associated with vegetative growth (MSL), seed weight (HSW) and leaf morphology traits (CLL and PLW), whereas Clusters III and IV were mainly related to NPP and NSP. In contrast, Cluster II exhibited greater dispersion, reflecting high variability among its accessions and the absence of clearly defined associations with the evaluated variables. Similar results were reported by Aybar-Peve et al. (2025a), who indicated that genotypes with greater dispersion tend to exhibit higher phenotypic diversity. This variability may be influenced by differences in plant architecture, genotype × environment interaction and adaptation strategies (Maqueira-López et al., 2021).
 
Multiple correspondence analysis (MCA)
 
The variation explained by the first two MCA dimensions (15.4%; Fig 4) indicated an adequate representation of the phenotypic diversity of the 18 evaluated qualitative traits, a common pattern in multiple correspondence analyses due to the high number of categories and the discrete nature of morphological variables (Santa Cruz-Padilla et al., 2025; Aybar-Peve et al., 2025b).


       
Cluster III exhibited lower dispersion and greater phenotypic uniformity, whereas clusters I and IV showed partial overlap due to similarities in lilac and purple floral pigmentation, as well as cream and purple seed coloration. In contrast, Cluster II presented greater dispersion within the factorial plane, reflecting high phenotypic diversity despite being composed of a smaller number of accessions.
       
The categories with the highest contribution were mainly associated with seed traits (SSC_gr and PSC_grw), flower traits (SPC_pn and SPC_w) and pod traits (PCM_gy). Similar results were reported by Aybar-Peve et al. (2025b), who identified a high contribution of categories associated with pod and seed coloration in Andean common bean accessions. Likewise, Carvalho et al., (2016) indicated that seed coloration constitutes a trait of commercial and cultural importance, whereas Zhu et al., (2017) reported that this variability depends on epistatic interactions related to flavonoid and anthocyanin biosynthesis.
 
Diversity indices
 
The high diversity observed for PSC (H′ = 1.925), GH (H′ = 1.702), SSC (H′ = 1.546), PCM (H′ = 1.443) and SPC (H′ = 1.407) (Table 5) revealed substantial phenotypic variability within the evaluated collection. Likewise, PSC (J′ = 0.876) and GH (J′ = 0.819) exhibited high evenness values, reflecting a relatively homogeneous distribution of their categories. Similar results were reported by Long et al., (2020); Savić et al. (2020); Kouam et al., (2023) and Patel et al., (2025), who documented high phenotypic diversity in common bean and other legume accessions, mainly in seed- and flower-related traits.


       
The higher diversity values observed for PSC, SSC, PCM and SPC are consistent with the findings of Savić et al. (2020), who identified high diversity in seed and flower traits, indicating that these descriptors constitute important sources of phenotypic differentiation in common bean. Likewise, Romero-Astudillo et al. (2024) reported that the high diversity recorded in Phaseolus vulgaris L. may be associated with its broad geographic distribution and historical selection processes.
       
Most variables showed significant association with the clustering structure, particularly PSC (p<0.001), SPC (p<0.001), SSC (p<0.001), HC (p<0.001), VCL (p<0.001), WC (p<0.001) and PCM (p<0.001), indicating high discriminatory capacity among clusters. Similar to the findings reported by Ventura-Neyra et al. (2026), internal diversity measured through H′ and J′ does not necessarily guarantee high discriminatory power among groups, since some descriptors may exhibit high diversity but low statistical significance in the χ² test. In contrast, variables combining high diversity and significant association with the clustering structure constitute more efficient descriptors for phenotypic differentiation.
       
At the cluster level, Cluster II exhibited higher diversity values for traits mainly associated with seed, flower and pod characteristics, particularly SSC (H′ = 1.889), PCM (H′ = 1.677), SPC (H′ = 1.311) and SS (H′ = 1.369), in addition to high evenness values (J′ > 0.7), indicating balanced category frequencies. These results are consistent with the greater dispersion previously observed in MCA and FAMD, reflecting high phenotypic heterogeneity among its accessions. In contrast, cluster IV exhibited higher diversity in variables related to vegetative growth and pigmentation, including GH (H′ = 1.992), VCL (H′ = 1.205), PCH (H′ = 1.266), HC (H′ = 0.898) and PP (H′ = 0.859), indicating substantial morphological variability among its accessions. Conversely, Cluster III showed lower diversity levels and greater phenotypic uniformity, a pattern similar to that reported by Ventura-Neyra et al. (2026); Dadther-Huaman et al. (2024) and Espinoza de Arenas et al. (2022).

Diversity indices and multivariate analyses revealed high variability among the four clusters, confirming the wide phenotypic diversity present in the 66 Peruvian common bean germplasm accessions. The FAMD identified hundred-seed weight, primary leaf width and primary seed color as the main contributors to overall variability, whereas Cluster II exhibited the highest phenotypic dispersion and trait diversity. In contrast, Cluster IV showed greater agronomic potential, as it presented higher yield-related components.
       
MCA and PCA analyses showed a positive association between hundred-seed weight and leaf traits, whereas the number of pods per plant and seeds per pod were negatively associated with morphological variables and seed weight. Likewise, primary seed color and growth habit were the most discriminating qualitative traits, evidencing high diversity and evenness among accessions.
       
Overall, the results demonstrate a broad genetic base within the evaluated germplasm, identifying key traits for the selection of genotypes with potential use in breeding programs and for the sustainable conservation of common bean genetic resources.

The authors acknowledge the Directorate of Genetic Resources and Biotechnology (DRGB) of the National Institute of Agrarian Innovation (INIA), Peru, for providing the accessions used in this research.
 
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.

The authors declare that there is no conflict of interest regarding the publication of this manuscript.