The research was performed at the experimental farm of the Department of Genetics and Plant Breeding, Bangladesh Agricultural University (BAU), Mymensingh, from August 2021 to March 2022. The experimental site had sandy loam soil with a pH ranging from 6.5 to 6.7. The region is characterized by a tropical climate with wet summers and dry winters. While the early stages of the experimental period experienced moderate rainfall and temperature levels, the study was specifically designed to simulate and analyze the impact of drought conditions on the plants. To ensure drought stress, controlled irrigation and water deficit treatments were employed during the critical growth stages of the plants, replicating the conditions needed to evaluate drought responses effectively.

The experiment involved thirty genotypes of the country bean, Lablab purpureus (L.), as detailed in Table 1 . The seeds for these genotypes were obtained from the Field Laboratory of the Department of Genetics and Plant Breeding at Bangladesh Agricultural University in Mymensingh. The genotypes selected for the study were those commonly found in local rural communities. The genotypes chosen for this research are widely cultivated varieties, esteemed for their high productivity by local farmers.

The experiment followed a Randomized Complete Block Design (RCBD) with three replications. Seeds were sown on August 29, 2021. The plot for the experiment was prepared by digging pits approximately one week prior to sowing. Pits were spaced 2.0 meters apart between rows and within rows, and each pit was dug to a depth of 1.2 feet. Three to four seeds were sown in each pit. The seeds were planted at a depth of 2 to 3 cm into the soil.

The seeds germinated within 4 - 6 days after sowing. After approximately one-month, extra seedlings were thinned out, leaving only one healthy plant per pit. Each young plant was supported with a single stick in its pit. Weeding was conducted twice: first at 25 days after sowing and again at 50 days after sowing. The harvesting of green edible pods started in the second week of January 2022 and continued until March 2022. The timing of pod harvesting varied among the genotypes, depending on their individual maturity stages.

Data were collected on several parameters including plant height, number and length of primary, secondary, and tertiary branches, as well as the number of flowers.

Multivariate analysis was conducted using XLSTAT software. PCA was used to identify groupings and determine the axes and attributes that significantly contributed to variance, based on the similarity matrix. Scatter plots were created using the first two principal components, as these accounted for the greatest variance. Agglomerative Hierarchical Clustering (AHC) was performed using Ward’s minimal variance method and squared Euclidean distances to measure dissimilarity [10] [11] . Descriptive statistics like range, mean, standard deviation, and coefficient of variation (CV%) for various quantitative traits were calculated. A one-sample variance test was performed to evaluate the null hypothesis and determine the significance of the observed variation. Additionally, Pearson’s correlation matrix was utilized to analyze the relationships between the variables [10] . RStudio software was utilized to compute the genotypic and phenotypic coefficients of variation (GCV and PCV), environmental coefficient of variation (ECV), broad-sense heritability, genetic advance (GA), and genetic advance as a percentage of mean. Established methods were applied in RStudio to calculate these statistical parameters [12] - [14] .

Protein content in the dry pods was measured from ten randomly chosen plants. On December 5, fresh pods were harvested from ten distinct genotypes, as detailed in Table 2 . The seeds from each pod were oven-dried at 65˚C for 5 days to lower their moisture content. The crude protein content was determined using the Kjeldahl method, with the results multiplied by the 6.25 conversion factor and expressed as a percentage (%) based on AOAC standards [15] .

The DNA was isolated from actively growing young fresh leaves (14 days old seedling) by DNeasy Plant Mini Kit method. A UV spectrophotometer was used to measure DNA concentration by reading the absorbance at wavelengths of 260 nm and 280 nm. The concentration was determined using the equation: Purity = absorbance at 260 nm/absorbance at 280 nm. The result of purity concentrations was recorded in this study ( Table 3 ). DNA quality was checked by running the genomic DNA sample on 1.5% Agarose gel. Two set of SSR primers were used in the polymerase chain reaction (PCR). The names of the SSR primer were ROCII and SG6. The gene specific primers were designed by NCBI primer BLAST using country bean genome. PCR was an important step in molecular analysis. PCR cocktail was prepared for DNA marker and then 9 µl of the PCR cocktail was taken in the separate PCR tubes and 1 µl DNA extract of each sample was added in it ( Table 4 ). The PCR tubes with the PCR reaction mixture were sealed and placed in a thermos cycler and the PCR reaction was started immediately ( Table 5 ). For confirmation of the target gene, agarose gel electrophoresis was done after completing the PCR. The presence and absence of DNA band was considered for identifying the presence or absence of gene of interest.

Significant variations have been observed among the Lablab germplasms in traits of plant height, number of primary branches, primary branch length, number of secondary branches, secondary branch length, number of tertiary branches, tertiary branch length and number of flowers. Plant height ranged from 151 to 240 cm with the mean value of 180.63 cm, SD and CV are found to be 24.29 cm and 13.45%, respectively. DS20 exhibits the tallest plant height, while DS13 is recorded the lowest height. The average number of primary branches is 13.33 cm, with the ranged from 5 to 20 cm, SD and CV are found to be 3.84 cm and 28.83%, respectively.

DS4 and DS9 are observed to be the highest number of primary branches, while DS10 exhibits the lowest. The average length of primary branches ranged from 77 to 167 cm, with a mean of 121.80 cm, a coefficient of variation (CV) is 26.46%, and a standard deviation (SD) is 32.23 cm. DS5 is recorded the highest primary branch, whereas DS24 is the shortest.

The number of secondary branches varied between 2 and 12 cm, with an average of 5.40 cm. The standard deviation (SD) was 2.50 cm, and the coefficient of variation (CV) was 46.29%. DS20 exhibited the highest number of secondary branches, whereas DS2 had the fewest. Secondary branch lengths ranged from 17 to 148 cm, with an average length of 77.67 cm. The standard deviation was 32.44 cm, and the coefficient of variation was 41.77%. DS24 had the longest secondary branches, whereas DS3 had the shortest. The number of tertiary branches averaged 2.40 cm, ranging from 1 to 4 cm. The standard deviation was 0.93 cm, and the coefficient of variation was 38.84%. The genotypes DS4, DS8, and DS14 had the highest number of primary branches, whereas DS7, DS10, DS13, DS17, DS23, and DS29 had the lowest. Tertiary branch lengths ranged from 6 to 30 cm, with an average length of 13.90 cm. The standard deviation was 5.26 cm, and the coefficient of variation was 37.85%. DS20 had the longest tertiary branches, whereas DS7 had the shortest. The number of flowers averaged 2.47 cm, ranging from 1 to 6 cm. The standard deviation was 1.36 cm, and the coefficient of variation was 55.05%. DS25 had the highest number of flowers, whereas DS1, DS3, DS8, DS9, DS11, DS14, DS19, and DS28 had the fewest ( Table 6 ). The observed variations stem from the different genotypes, offering substantial potential for yield enhancement. High coefficients of variation (CV > 20%) in traits like the number and length of branches and the number of flowers indicate considerable variability, which influences plant yield. The high CV values highlight traits with substantial variability, which can be harnessed for further enhancement. The extensive diversity in traits offers ample opportunities to select genotypes suited to specific breeding program needs [16] .

Traits having more than >20.00% higher CV value are highlighted with orange colour. * = significance at alpha 0.05 of one-sample variance z-test.

Successful crop improvement programs rely on a breeder’s capacity to identify and harness genetic variability and highly heritable traits, once environmental influences on phenotypic variation are accounted for [17] . Understanding both the phenotypic (PCV) and genotypic coefficients of variation (GCV) is crucial for breeders to estimate heritability and predict the potential for genetic advance (GA) through trait selection.

In this study, the phenotypic and genotypic coefficients of variation, as well as heritability, were estimated for eight traits within the collected germplasms. The phenotypic coefficient of variation (PCV) exceeded the genotypic coefficient of variation (GCV) for all traits ( Table 7 ). This suggests that the observed variation is influenced by both genetic factors and environmental effects on these traits. Traits such as primary branch length (26.11, 23.65), number of secondary branches (42.23, 26.90), secondary branch length (42.13, 42.10), number of tertiary branches (37.99, 27.54), and tertiary branch length (35.79, 33.53) exhibited high phenotypic and genotypic coefficients of variation.

It suggests that there are both genetic diversity and environmental influence contributing to the trait’s variability. Moderate-high PCV (26.85) and moderate GCV (16.07) are recorded for number of primary branches. The low GCV of plant height (7.79) and number of flowers (8.12) suggests that influence of the environment on the expression of this characters ( Table 7 ). Hence, relying solely on the phenotypic expression of this trait for selection would be ineffective in enhancing crop improvement.

The quantitative morphological traits are plant height (PH), number of primary branches (NPB), primary branch length (PBL), number of secondary branches (NSB), secondary branch length (SBL), number of tertiary branches (NTB), tertiary branch length (TBL) and number of flowers (NF).

GV: Genotypic Variance, PV: Phenotypic Variance, EV: Environmental Variance, GCV: Genotypic Coefficient of Variance, PCV: Phenotypic Coefficient of Variance, ECV: Environmental Coefficient of Variance, GA: Genetic Advance, GA (% mean): Genetic Advance as percentage of mean.

High heritability combined with a high genetic advance as a percentage of the mean suggests that additive gene effects may be influencing this trait, which can be advantageous for plant breeders during selection. High heritability coupled with high genetic advance as percent of mean has been observed in following characters, primary branch length (82.06, 44.14), secondary branch length (99.85, 86.66), tertiary branch length (87.78, 64.72) ( Table 7 ). High values for heritability and genetic advance as a percent of the mean for seed yield indicate that this trait is likely influenced by additive gene effects, making selection for improvement highly effective. Selected varieties of cowpea exhibited a considerable degree of heritability in relation to yield [18] . High heritability combined with moderate genetic advance could be attributed to moderate phenotypic standard deviation values. Since heritability is high for these traits and the selection differential remains constant [19] . Here, moderate heritability, and genetic advance as percent of the mean has been observed in plant height (42.22, 10.43), number of primary branches (35.84, 19.83), number of secondary branches (40.59, 35.31), number of tertiary branches (52.55, 41.12).

Genotypic correlations for various trait combinations in the Lablab germplasms are displayed in Table 8 . Several quantitative traits exhibit significant correlations, either positive or negative. Plant height (PH) shows a strong positive and significant correlation with the number of primary branches (NPB). It also shows positive but non-significant correlations with the number of secondary branches (NSB), the number of tertiary branches (NTB), and the length of tertiary branches (TBL). Conversely, plant height (PH) has negative and non-significant correlations with primary branch length (PBL), secondary branch length (SBL), and the number of flowers (NF). This pattern suggests that selecting for taller plants (PH) may effectively enhance the number of primary branches in Lablab germplasms.

**Indicate highly significant results (P < 0.05 and P < 0.02); *Indicate significant results (P < 0.05 and P < 0.02). The quantitative morphological traits are plant height (PH), number of primary branches (NPB), primary branch length (PBL), number of secondary branches (NSB), secondary branch length (SBL), number of tertiary branches (NTB), tertiary branch length (TBL) and number of flowers (NF).

The number of primary branches (NPB) exhibits a strong positive and significant correlation with the number of tertiary branches (NTB). It shows a significant negative correlation with the number of flowers (NF) and a significant positive correlation with the number of secondary branches (NSB). Additionally, NPB has positive but non-significant correlations with primary branch length (PBL), while it shows negative and non-significant correlations with secondary branch length (SBL) and tertiary branch length (TBL).

Primary branch length (PBL) demonstrates positive but non-significant correlations with the number of secondary branches (NSB), the number of tertiary branches (NTB), and tertiary branch length (TBL). In contrast, it has negative and non-significant correlations with secondary branch length (SBL) and the number of flowers (NF).

The number of secondary branches (NSB) shows a strong positive and significant correlation with tertiary branch length (TBL). It also exhibits a positive and significant correlation with secondary branch length (SBL) and the number of tertiary branches (NTB). NSB has a negative and non-significant correlation with the number of flowers (NF).

Secondary branch length (SBL) has strong positive and significant correlations with tertiary branch length (TBL) and the number of flowers (NF). It shows a positive but non-significant correlation with the number of tertiary branches (NTB). The number of tertiary branches (NTB) has negative and non-significant correlations with both tertiary branch length (TBL) and the number of flowers (NF). Lastly, tertiary branch length (TBL) exhibits a positive but non-significant correlation with the number of flowers (NF).

Principal Component Analysis (PCA) was conducted on 8 quantitative morphological traits to identify the main factors or components that significantly influence the overall indicators. The eigenvalues, the percentage of contribution to variability, and the cumulative contribution rates of the first three principal components (F1, F2, F3) are depicted in Figure 1 . The PCA results show that these three components collectively explain approximately 75% of the total variation in quantitative traits. These components were selected for further analysis ( Table 9 , Figure 1 ). The first principal component (F1) explains the largest portion of the variance at 31.18%, followed by the second component (F2) with 27.08%, and the third component (F3) which accounts for 14.79% of the total morphological variation ( Table 9 ).

The Principal Component Analysis (PCA) of the quantitative traits shows that the plant height (PH), number of primary branches (NPB), primary branch length (PBL), number of tertiary branches (NTB), number of secondary branches (NSB), and tertiary branch length (TBL) are significantly represented in the first principal component (F1). This component has positive contributions from all variables except for secondary branch length and number of flowers, which have negligible contributions. The second principal component (F2) is positively influenced by the number of secondary branches, secondary branch length, number of tertiary branches, tertiary branch length, and number of flowers, but negatively influenced by plant height, number of primary branches, and primary branch length. The third principal component (F3) shows positive contributions from primary branch length, the number of secondary branches, and tertiary branch length ( Table 9 ).

Together, the first two components (F1 and F2) account for 58.27% of the total variation in quantitative morphological traits. Figure 2 illustrates the contributions and directions of the various traits within these principal components. It also presents the genotypic distribution of Lablab germplasms. Genotypes located close to each other in the plot show similarities across the eight assessed variables, indicating a lack of significant diversity as they cluster within the same region. Notably, genotypes such as DS25, DS20, DS5, DS12, DS10, DS13, and DS23 are positioned farther from the others along F1 and F2, suggesting that crosses between these genotypes might yield better segregants.

PCA is a valuable tool for understanding the relationships between variables, simplifying data collection, and interpreting complex datasets [20] . The analysis of the eight quantitative traits reveals that the first three components contribute 73.06% of the total variation, aligning with previous findings where the first six components of the Lablab germplasms core collection explained 73.74% of the total variation [21] . Similarly, in the field pea crop, the first three principal components account for 75% of the total variation [22] .

The agglomerative hierarchical clustering (AHC) analysis was performed using the Ward method. The resulting dendrogram, shown in Figure 3 , illustrates the grouping of the 30 Lablab germplasms based on their quantitative morphological traits. These germplasms were categorized into 3 distinct clusters. Specifically, 12, 6, and 12 germplasms were grouped into Clusters I, II, and III, respectively ( Table 10 , Figure 3 ). Clusters I and III are the largest, each containing 12 genotypes.

Further analysis of the mean performance for different quantitative traits within each cluster, as compared to the overall mean of the 30 germplasms, is summarized in Table 11 . Cluster I, which includes 12 germplasms, is characterized by higher mean values for the number of flowers compared to Cluster III and the overall mean for the 30 germplasms ( Table 11 ). Cluster II, which consists of 6 germplasms, exhibits higher values for primary branch length, the number of secondary branches, secondary branch length, tertiary branch length, and the number of flowers. Cluster III, also comprising 12 germplasms, shows higher mean values for traits such as plant height, the number of primary branches, and the number of tertiary branches compared to the other clusters and the overall mean for the 30 germplasms evaluated ( Table 11 ).

Cluster analysis organizes objects based on high levels of similarity, which is beneficial for identifying diverse allele sets in future heterotic breeding programs [23] . None of the clusters encompassed all the desired traits for breeding purposes; rather, desirable traits were distributed across different clusters. These traits significantly contribute to the genetic diversity observed in country bean germplasms [24] . The dendrogram indicates that germplasms with similar morphological characteristics are often grouped within the same cluster. This observation aligns with findings in Soybean and other crops [25] - [30] . Crossbreeding between highly divergent clusters is recommended to develop genotypes with preferred traits, aiding breeders in selecting parent lines for hybrid varieties and leveraging heterosis in subsequent generations.

The diversity of flower colors among the Lablab germplasms is illustrated in Figure 4 . The flower colors, specifically the wing and petal hues, exhibit six types of variation. The predominant colors are white, purple, and light purple. The Lablab germplasms studied show six distinct flower color variations: white, light purple, light violet, pink, purple, and very light violet. These colorful flower traits are important for attracting pollinators, which play a crucial role in cross-pollination [31] [32] . Additionally, the pigments responsible for these colors are part of a diverse group of compounds that provide defensive benefits to plants, including protection against UV radiation, diseases, and insect pests [33] .

The protein content across different Lablab germplasms ranges from 33.60% to 45.67%. The highest protein content is observed in genotype DS28, with 45.67%, while the lowest is found in DS3, at 33.60% ( Figure 5 ). The dry seeds of country beans contain five protein fractions: globulin-1 (phaseolin), globulin-2, albumin, prolamine, and alkali-soluble proteins [34] . The protein content in country beans can vary due to environmental factors and the interaction between the plant’s genetics and its growing location [35] . Furthermore, the protein content is slightly altered when beans are cooked [36] .

Following the DNA extraction from leaf samples, agarose gel electrophoresis was performed to verify the presence of DNA in the samples. DNA was successfully detected in all thirty samples ( Figure 6(a) , Figure 6(b) ). The primers ROCII and SG6 effectively identified 2 out of the 30 genotypes, producing distinct bands at approximately 595 bp, which confirmed the presence of the targeted gene ( Figure 6(c) , Figure 6(d) ). The distinct banding patterns for all thirty genotypes are detailed in Table 12 . The ROCII and SG6 markers are associated with the bc3 gene. These SSR markers, linked to the bc3 gene, are utilized to identify genomic regions that influence dehydrogenase responsiveness. This information can be applied in marker-assisted selection to enhance the genetic traits of the country bean crop, specifically selecting lines with alleles for DREB responsive traits.