Origin of the Ñuña accessions

The Ñuña samples were collected from plots belonging to producers located in eight districts within the regions of La Libertad, Cajamarca, and Ancash in Peru, as detailed in Table 1 and illustrated in Figure 1. Most of the Ñuña germplasm collected in northern Peru was from the Cajamarca region (88 accessions), followed by the La Libertad region with 31 accessions and the Ancash region with 3 accessions (Table 1).

Location of the study

The study covered the period from December 2019 to July 2020 and was carried out at the Cochamarca Experimental Annex of the Baños del Inca Agricultural Experimental Station, Cajamarca (7.2756 S, 78.2186 W, 2,820masl, Figure 1). The study area exhibits a climatic classification corresponding to the tropical low montane dry forest (bs-MBT) category, determined according to the methodology of (Holdridge, 1947). Throughout the research period, mean, minimum and maximum temperatures were 14.3°C, 7.6°C and 21°C respectively, with a rainfall of 117.4mm (SENHAMI, 2020).

Soil chemical properties

A composite sample was extracted from the experimental area using an auger at a 0 to 30cm depth. This sample was treated at the Baños del Inca - INIA Soil, Water and Foliar Laboratory, where it underwent a process of air drying, followed by grinding and sieving through a 2mm mesh. Subsequently, organic matter was determined using the (Walkley & Black, 1934) method, while phosphorus (P) was evaluated according to the (Olsen, Cole, Watanabe, & Dean, 1954) protocol, and potassium (K) was determined using the silver thiourea method. Soil analysis revealed a pH of 6.5, with a 1% organic matter content, 3.82ppm P and 295ppm K.

Soil preparation

The experimental area covered 0.16 hectares, where soil preparation was carried out to a depth of 30cm using a disc plough coupled to an agricultural tractor (New Holland, 110 HP, Model: TS6.110). To correct soil fertility, amendments such as island guano (1,450kg/ha), diammonium phosphate (150kg/ha) and potassium chloride (100kg/ha) were applied. Subsequently, plots of 6m² (6m × 1m) were demarcated, resulting in a total of 122 plots of one row each, corresponding to the 122 Ñuña accessions. Planting was carried out in December 2019, with an arrangement of plants at distances of 0.5m between plants and 1.0m between rows, totalling 12 plants per row, equivalent to 20,000 plants per hectare.

Cultivation treatments

Manual weeding was carried out 45 days after planting (DAP). Subsequently, trellising was carried out at 50 DAP by installing 2.5m high posts at the ends of the plots. On these posts, a galvanized wire (N° 16) was stretched along 6m, which supported the plants. A phytosanitary control against boring larvae (Lepidoptera: Noctuidae) was also implemented, using alphacypermethrin (25ml/20L of water).

Determination of variables

Evaluations were carried out from the beginning of flowering until harvest, recording data from 10 plants per accession. Descriptors adapted from the International Plant Genetic Resources Institute for Phaseolus vulgaris (IPGRI, 2001) were studied (Table 2). The Royal Horticultural Society Colour Chart (RHS, 2001) was used to assign the colours of flowers, pods and seeds. The study included consideration of 12 qualitative descriptors (Table 2) and 9 quantitative descriptors (Table 3).

Statistical analysis

Characterization information was subjected to descriptive and multivariate statistical analysis. Multivariate statistics included multiple correspondence analysis (MCA) for qualitative traits and principal component analysis (PCA) for quantitative traits. In both cases, decision trees were constructed using hierarchical and phylogenetic dendrograms. The Euclidean distance and the Ward.D2 (Ward, 1963) method were used as a similarity measure to carry for grouping between accessions. For the clusters of the quantitative traits, a comparison of means was carried out using Tukey's test (p ˂ 0.05). The analyses were carried out with the packages Factoextra (Kassambara & Mundt, 2020) and FactorMiner (Lê, Josse, & Husson, 2008) for MCA and PCA. The dendrograms were elaborated with the cluster (Maechler, Rousseeuw, Struyf, Hubert, & Hornik, 2021) and circlize (Gu, Gu, Eils, Schlesner, & Brors, 2014) packages, while the visualization of the results was performed with ggplot2 (Wickham, 2016). Comparison of means was run with the AgroR package (Shimizu, Marubayashi, & Goncalves, 2023). All analyses were performed using RStudio statistical software (R Core Team, 2023).

Multiple correspondence analysis of qualitative characters

The qualitative characters evaluated were subjected to MCA, the results of which are presented in Figure 2. A marked association was observed between several characters, such as darker colour of seeds (DCS) and lighter colour of seeds (LCS) which presented the highest contributions to the variability of the 122 accessions of Ñuña. Seed coat pattern (SCP), flower wing colour (WIC) also associated and had similar contributions to the clustering of the accessions. There was a joint association between standard colour (STC), dry pod colour (DPC) and colour of immature pods (CIP). This pattern suggests that certain characteristics share discernible similarities, while others show a less prominent relationship. These findings provide further insight into the interrelationships between the qualitative variables assessed.

Hierarchical analysis of qualitative characteristics

The analysis of the phylogenetic hierarchical tree of the 122 accessions of Ñuña is shown in Figure 3. It is possible to observe the formation of four morphological clusters, grouped according to their most similar characters. Cluster I represents 37% of the accessions (45 accessions), which showed similarity in seven characters associated with different morphological stages: darker colour of seeds, between white-tinged black, grey-brown and greyish purple-tinged white to brown; lighter colour of seeds between white tinged purple, yellow, greyish orange, brown, purple to absent; seed coat pattern between mottled, spotted, around the hilum; standard colour between green yellow and yellow green pigmented; wing colour between white, white pigmented violet to violet blue; colour of immature pods between green, yellowish green and green pigmented; and dry pod colour between yellow and orange. Cluster II represents 4% of the accessions (five accessions), which showed similarity in nine characters associated with different phenological stages: darker colour of seeds and lighter colour of seeds dyed purple; striped seed coat pattern; purple colour of immature pods; pod curvature curved; seed shine matt; oval seed shape; intermediate leaf persistence and absent vein in seeds. Cluster III represents 7% of the accessions (nine accessions), which showed similarity in eight characters associated with different phenological stages: in the presence of a dark colour stripe moving from the hilum towards the top of the grain, the standard colour between yellow-green pigmented with purple to purple violet; the darker colour of seeds between purple with white lateral stripe and violet blue with orange-white lateral stripe; lighter colour of seeds between greyish orange with white lateral stripe to greyish purple with white lateral stripe; long-stripes seed coat pattern; wing colour between purple to purple violet; colour of immature pods between green and yellowish green; pod curvature between slightly curved to curved; dry pod colour between yellow and orange. Cluster IV represents 52% of the accessions (63 accessions), which showed similarity in seven characters associated with different phenological stages: darker colour of seeds between greyish orange, greyish red, grey brown, purple, violet-blue and black; lighter colour of seeds between greyish orange, brown grey, greyish purple, purple, purple or greyish red and green yellow; seed coat pattern between absent, striped, to pattern around the hilum; standard colour between purple, purple violet and yellow-green pigmented with purple violet; wing colour between white to purple violet; colour of immature pods between yellowish green and green pigmented and dry pod colour between yellow, orange and purple.

Principal component analysis (PCA) of quantitative traits

The PCA of quantitative traits explained 71.6% of the total variability and is presented in Figure 4. In this analysis, the accessions did not show differentiation of their quantitative traits according to their origin. Leaflet length and leaflet width showed a high positive correlation, as indicated by the direction of the arrows in Figure 4. The same, but to a lesser degree, occured for the characters: immature pod length, pod width, pod beak length, hundred seed weight, seed length, seed width and seed height, respectively.

Hierarchical analysis of quantitative characteristics

The hierarchical cluster analysis based on the quantitative traits (Figure 5) revealed the formation of four clusters, established according to their most similar characteristics. Table 4 shows the mean values of the characters of Clusters A, B, C and D, highlighting significant differences (p ˂ 0.05) between clusters. Cluster A, representing 22% of the accessions (27 in total), was characterized by lower values for immature pod length, pod width, pod beak length, hundred seed weight, seed length, seed width and seed height compared to the rest of the clusters (Table 4). Cluster B, comprising 16% of the accessions (20 in total), was significantly superior to the other clusters showing higher values for immature pod length, pod width, pod beak length, hundred seed weight, seed length, seed width and seed height. However, Cluster B had lower values for leaflet length and leaflet width compared to the other clusters. Cluster C, comprising 30% of the accessions (37 in total), showed higher mean values than Cluster A, but lower than Clusters B and D for the character's immature pod length, pod beak length, hundred seed weight, seed length, seed width and seed height. For width of immature pod trait, Cluster C had a higher mean than Clusters A and D, but lower than Cluster B. Cluster C had a lower mean for the leaflet length trait, than the rest of the clusters. For the leaflet width character, the mean of Cluster C was lower than Clusters A and D, but higher than the mean of Cluster B. Finally, Cluster D, comprising 31% of the accessions (38 in total), was characterized by higher values than Clusters A and C for immature pod length, pod beak length, hundred seed weight, seed length, seed width and seed height. As for pod width, its mean value was higher than that of Cluster A, but lower than the means of Clusters B and C. Cluster D stood out with higher values for leaflet length and leaflet width compared to Clusters A, B and C. There was a marked diversity in the distances between clusters, particularly between Cluster A and Clusters C and D. This suggests that the accessions in these clusters show significant morphological differences.

Qualitative characterization

Qualitative analysis revealed that characters associated with seed colour showed the greatest morphological variability, followed by seed coat pattern and, to a lesser degree, flower-related characters (Figure 2). This indicates that seed-related characters are fundamental for the identification and differentiation of Ñuña bean accessions, establishing associations between seed morphology and the phenotypic diversity observed among them. This result is consistent with previous studies by (Martirena-Ramírez et al., 2017) and (Espinosa-Pérez et al., 2015), who highlighted that flower colour and seed colour are essential to assessing phenotypic variability of common bean. Similarly, (Morales-Morales et al., 2019) have determined the same for cowpea (Vigna unguiculata L. Walp). (Murga-Orrillo, Gonzales, López, & A, 2024) affirmed that in cowpea seed colours mark the preferences of producers and consumers since the dark colour of the testa presents greater antioxidant activity. From this perspective, a generalized tendency in bean research to use morphological characters of the seed as primary indicators of genetic variability is confirmed, underlining its importance in the classification and differentiation of accessions.

The phylogenetic tree presented in Figure 3 showed that the qualitative characters related to seed, flower and pod are valid phenotypic traits to discriminate morphologically the different accessions of Ñuña; these characteristics, which vary among accessions, could be related to their areas of origin. In addition, it highlighted that there were no duplicate accessions in the studied collection at the phenotypic level. These results are valid for genetic diversity conservation since they indicate that the 122 accessions could represent unique genetic resources, which should be studied in detail to determine the variations of these accessions due to environmental and anthropogenic factors. Also, (Cruz-Balarezo et al., 2009) conducted clustering analyses on 24 accessions of Ñuña, which determined that there was no duplication of germplasm, even in cases where the accessions shared similar characters. On the other hand, in common bean, (Vásquez et al., 2024) identified four distinct morphological groups in 58 accessions with promising characteristics for breeding programmes. These studies highlighted the importance of characterizing Ñuña bean accessions for conservation and breeding initiatives.

Quantitative characterization

Regarding the spatial distribution of the 122 Ñuña accessions, based on quantitative traits, no defined clustering pattern for geographic origin was identified (Figure 4). This could be due, in part, to similar climatic conditions in the Andean region or common agricultural practices (Figure 1). (Mishra, Sharma, Singh, & Yadav, 2010) have shown that genetic diversity among bean genotypes did not show a direct relationship between clustering patterns and geographic origin. Consequently, the maintenance of Ñuña bean genetic characteristics from different geographic origins is crucial to preserving its biodiversity and genetic improvement. The PCA presented in Figure 4 shows strong positive correlations between leaf, immature pod and seed traits, demonstrating that these traits are mutually dependent. This could be related to shared developmental processes or the expression of common genes that regulate the size and structure of the mentioned plant organs. In common bean, (Lescay-Batista, Ramírez, & Rodríguez, 2017) and (García-Fernández et al., 2023) found a close association between pod characters and seed weight. Also, in lima bean (Phaseolus lunatus L.), a strong correlation was found between yield and stem length, number of seeds per pod, hundred-seed weight, primary leaf length and width, pod weight and pod length (Akande & Balogun, 2007; López-Alcocer et al., 2016).

The four clusters identified in Figure 5 were evaluated through a mean comparison test (p < 0.05), as detailed in Table 4. Cluster B presents significant differences, with higher means in the traits related to productivity, compared to the means of Clusters A, C and D. The accessions of Cluster B will be valuable in genetic improvement programmes because they could provide higher yields. Also, (Pesantes-Vera & Soto, 2013) in Ñuña, and (Kinhoégbè et al., 2020) in pigeonpea (Cajanus cajan (L.) Huth), used characters related to productive yield in the selection of promising individuals. However, it is essential to conduct further studies on the 122 accessions of Ñuña to identify relevant traits, such as resistance to abiotic and biotic factors, as well as nutritional aspects, in order to enrich future breeding programmes.