Unraveling genetic variability and diversity for the enhancement of yield and yield contributing traits in finger millet (Eleusine coracana L.) under semi arid climatic niches.

Keywords

Finger millet; Genetic variability; Heritability; Correlation analysis; Path analysis; Clustering; PCA analysis

Introduction

Finger millet (Eleusine coracana L.) is an ancient cereal crop belonging to the family Poaceae, widely cultivated in semi-arid and tropical regions of Africa and South Asia. Known for its resilience to drought and poor soil fertility, finger millet plays a vital role in ensuring food and nutritional security among smallholder farmers. The crop is particularly valued for its rich nutritional profile. It contains calcium (344-364 mg per 100 g), iron (4.6 mg per 100 g), magnesium and Phosphorus, zinc (2.5 mg per 100 g), boosts immunity and wound healing. High dietary fiber (11.2 g per 100 g of grain), protein and amino acids (7.2 g protein per 100 g), rich in tryptophan and Gluten free grain. These attributes make finger millet an important candidate in addressing malnutrition, lifestyle related disorders, and the growing demand for functional foods [1 -3 ]. Despite its nutritional and agronomic advantages, finger millet remains underutilized and often referred to as a "neglected and orphan crop." Limited research attention, low productivity, and inadequate value chain development have hindered its wider adaptation. However, recent advances in crop improvement, genomics, and food processing technologies have renewed interest in finger millet as a climate resilient and health promoting cereal. In the context of global challenges such as climate change, hidden hunger, and the need for sustainable agriculture, finger millet offers unique opportunities for diversification of food systems and development of nutritionally enriched products [4 ].

 

Despite its agronomic potential and nutritional richness, genetic improvement in finger millet remains limited. The available germplasm exhibits considerable variability, which needs systematic evaluation to identify superior genotypes for yield enhancement. Understanding the genetic diversity and interrelationships among yield and yield contributing traits is vital for designing efficient breeding strategies. This study sought to evaluate variability, heritability, and genetic progress, analyze the correlations and direct impacts among traits and investigate the genetic diversity of finger millet germplasm through multivariate analyses focused on yield enhancement.

Materials and Methods

An experiment was conducted during the rabi season of 2020–2021 at the Regional Agricultural Research Station, Palem (16°30′18″ N latitude, 78°19′ E longitude, and 458 m above mean sea level), PJTAU, Hyderabad, Telangana, India. The crop received 475 mm of rainfall during the growing period. The experimental material comprised thirty-one finger millet (Eleusine coracana L.) germplasm accessions, including two check varieties, GPU-67 and Maruthi (Table 1). The trial was laid out in a Randomized Complete Block Design (RCBD) with three replications. Each plot consisted of three rows of 3 m length, with inter- and intra-row spacing of 45 × 15 cm. The recommended dose of fertilizers was applied at sowing, and seeds were hand-sown. Thinning was carried out 21 days after sowing. Observations were recorded on five randomly selected plants per plot for yield related traits, including days to 50% flowering, plant population per plot, plant height (cm), number of productive tillers per plant, number of fingers per ear, finger length (cm), pedicel length (cm), days to maturity, 1000 seed weight (g), grain yield (kg/ha), and fodder yield (kg/ha). Data were analyzed for genetic parameters following Burton and Johnson et al. [5], correlation analysis as per Snedecor and Cochran [6], and path coefficient analysis according to Dewey and Lu [7]. Principal component analysis (PCA) was performed to identify major traits contributing to total variation, while Mahalanobis D2 statistics were used to assess genetic divergence and group genotypes into distinct clusters based on yield and yield component traits. All statistical analyses were carried out using INDOSTAT and R software packages.

Table 1. The finger millet genotypes utilized in the present study.

Results and Discussion

Based on the observations recorded for various yield and yield-contributing traits, the results of the present investigation were systematically analyzed and are presented under the following sections:

• Analysis of Variance (ANOVA)

• Genetic variability, heritability, and genetic advance

• Character association

• Path coefficient analysis

• Principal Component Analysis (PCA)

• Genetic divergence analysis (Mahalanobis D2 Statistics)

Analysis of Variance (ANOVA)

The Analysis of Variance (ANOVA) conducted on 33 finger millet genotypes (31 accessions and two checks) revealed significant differences across all eleven yield and yield contributing traits evaluated (Table 2). The observed variation among genotypes confirmed the presence of considerable genetic diversity, indicating strong potential for identifying superior genotypes for quantitative traits in breeding programs [8].

Table 2. Analysis of variance for yield and yield components among 33 genotypes of Finger millet (Eluesine coracana).

Genetic variability, heritability, and genetic advance

The nature and extent of variability in a trait of interest play a crucial role in the success of genetic improvement. For effective plant breeding, it is essential to recognize not only the presence of genetic variability but also its magnitude within the cropâ??s gene pool [9]. Such variability provides the foundation for improving specific traits, particularly yield, which remains a primary target in breeding programs [10].

The evaluation of 33 finger millet germplasm accessions for 11 yield and yield contributing traits revealed considerable variability in performance. Descriptive statistical measures including mean, range, standard deviation, standard error, and Coefficient of Variation (CV%) were computed to assess trait distribution [11]. In addition, broad sense heritability (%) and genetic advance as a percentage of the mean (at the 5% level) were estimated to provide insights into the genetic parameters underlying trait variability (Table 3).

The Genotypic Coefficients of Variation (GCV) for all traits studied were lower than the corresponding Phenotypic Coefficients of Variation (PCV), indicating the influence of environmental factors in modifying trait expression at the genotypic level [12]. High PCV coupled with high GCV was observed for 1000 grain weight (15.03 and 15.03), grain yield (22.5 and 14.8), and fodder yield (25.49 and 22.07), reflecting substantial inherent variability in these traits [13]. Estimates of heritability and genetic advance provide a basis for predicting genetic gain, thereby guiding the selection of superior individuals. High broad-sense heritability combined with high genetic advance as a percentage of the mean was recorded for plant population per plot (84 and 20.52), fodder yield (75 and 39.35), and grain yield (43 and 20.06) [14 -17]. Overall, traits exhibiting high heritability along with high genetic advance offer greater scope for direct selection, whereas those with moderate estimates suggest moderate potential for genetic improvement.

Table 3. Mean performance and genetic parameters among 33 genotypes of Finger millet (Eluesine coracana).

Character association analysis

Genotypic correlations were generally higher than phenotypic correlations, indicating that the observed associations are primarily genetic in origin. The differences can be attributed to the masking or modifying effects of the environment on genetic relationships among traits [18,19].

Grain yield in finger millet is a complex trait influenced by multiple yield contributing factors. Correlation analysis revealed significant positive associations of grain yield with plant population per plot (0.065, 0.053), number of productive tillers per plot (0.463, 0.019), finger length (0.132, â??0.065), and fodder yield (0.648, 0.382) at both genotypic and phenotypic levels, highlighting their importance in yield improvement (Table 4). These findings suggest that selecting for traits with strong positive correlations can enhance grain yield in finger millet [20-22].

Table 4. Genotypic correlation (upper) and phenotypic correlation (lower) of yield attributes among 33 genotypes of Finger millet (Eluesine coracana).

Path coefficient analysis

For path coefficient analysis, grain yield was considered the dependent variable, while the remaining traits were treated as independent variables. Both phenotypic and genotypic path analyses were performed to partition the correlations into direct and indirect effects [23,24]. A path coefficient is defined as the proportion of the standard deviation of a dependent variable (Y) resulting from variation in an independent variable (X), essentially representing a standardized partial regression coefficient. This technique enables the separation of correlation coefficients into measures of direct and indirect effects, thereby providing valuable insights for indirect selection in genetic improvement of yield. In the present investigation, genotypic correlation coefficients were further decomposed into direct and indirect effects using path analysis. Grain yield per plant was considered the resultant (dependent) variable, while the traits significantly correlated with grain yield were treated as causal (independent) variables [25,26].

Path coefficient analysis (Tables 5 and 6) revealed that fodder yield (0.4109), days to maturity (0.1421), number of productive tillers per plot (0.0429), and plant population per plot (0.0387) exerted the highest direct positive effects on grain yield at the phenotypic level. At the genotypic level, days to 50% flowering (2.5739), fodder yield (2.3138), pedicel length (2.0943), and number of fingers (1.1817) exhibited strong direct positive effects on grain yield. These findings highlight the critical role of these traits in influencing yield performance and emphasize their potential utility in selection strategies for finger millet improvement (Figures 1-4).

Overall, the combined results of correlation and path coefficient analyses indicated that plant population per plot, number of productive tillers per plot, and fodder yield are the most reliable selection criteria for enhancing grain yield in finger millet.

Table 5. Direct and indirect effects (phenotypic) of yield attributes on yield among 33 genotypes of Finger millet (Eluesine coracana).

Table 6. Direct and indirect effects (genotypic) of yield attributes on yield among 33 genotypes of Finger millet (Eluesine coracana).

GMR: https://www.geneticsmr.org/articles-images-2025/gmr-path-192198-g001.png" alt="Image" />

 

Figure 1. Genotypic path diagram showing cause-effect relationship in finger millet (Eluesine coracana).

GMR: https://www.geneticsmr.org/articles-images-2025/gmr-path-192198-g002.png" alt="Image" />

 

Figure 2. Shaded correlation matrix in finger millet (Eluesine coracana) at genotypic level.

GMR: https://www.geneticsmr.org/articles-images-2025/gmr-path-192198-g003.png" alt="Image" />

 

Figure 3. Phenotypic path diagram showing cause-effect relationship in finger millet (Eluesine coracana).

GMR: https://www.geneticsmr.org/articles-images-2025/gmr-path-192198-g004.png" alt="Image" />

 

Figure 4. Shaded correlation matrix in finger millet (Eluesine coracana) at phenotypic level.

Principal Component Analysis (PCA)

Principal Component Analysis (PCA) revealed that the first three Principal Components (PCs) accounted for 57.91% of the total variability among the 33 finger millet genotypes (Table 7). PC1 contributed the maximum variability (22.76%), followed by PC2 (22.05%) and PC3 (13.09%). Since the eigen values of these three PCs exceeded one (6.3), they were retained for further analysis. Among the eleven traits considered, grain yield (0.424) had the highest loading in PC1, followed by fodder yield (0.359), days to maturity (0.285), and number of productive tillers per plant (0.143). In PC2, days to 50% flowering (0.503), days to maturity (0.459), 1000 seed weight (0.259), and number of productive tillers per plant (0.207) showed the highest loadings. PC3 was primarily explained by finger length (0.556) and 1000 seed weight (0.067). Overall, traits such as days to maturity, 1000 seed weight, number of productive tillers per plant, fodder yield, and grain yield emerged as key contributors to variability, underscoring their significance in differentiating finger millet genotypes (Figures 5 and 6) [27-32].

Table 7. Eigenvalues, proportion of variance, and cumulative proportion of variance explained by principal components in finger millet.

GMR: https://www.geneticsmr.org/articles-images-2025/gmr-path-192198-g005.png" alt="Image" />

 

Figure 5. 3D plot to explain first two principal components towards genetic divergence.

GMR: https://www.geneticsmr.org/articles-images-2025/gmr-path-192198-g006.png" alt="Image" />

 

Figure 6. Contribution of principal components towards total divergence.

Genetic divergence analysis (Mahalanobis D2 statistics)

To support the findings, the clustering pattern was examined using Tocherâ??s method. Genetic diversity analysis is crucial for plant breeders, as it facilitates the identification of divergent genotypes that can be exploited for heterosis. Mahalanobisâ?? D2 analysis is a widely used tool for classifying genotypes into distinct clusters. In the present study, eight clusters were identified (Table 8). Among them, Cluster II contained eleven genotypes, Cluster IV seven genotypes, Cluster III six genotypes, and Cluster I five genotypes, while Clusters V, VI, VII, and VIII each comprised a single genotype. Cluster means (Table 9) for the eleven traits studied revealed considerable variation: Days to 50% flowering ranged from 63 (Cluster VIII) to 74.67 (Cluster VI); plant population per plot from 56 (Cluster VI) to 88 (Cluster VII); plant height from 56.57 (Cluster V) to 73.85 (Cluster III); number of productive tillers per plant from 6.09 (Cluster II) to 7.67 (Cluster VI); number of fingers from 5.67 (Cluster VII) to 7.71 (Cluster IV); finger length from 5.33 (Cluster VI)to 7.67 (Cluster V); pedicel length from 14 (Cluster VIII) to 20.67 (Cluster VI); days to maturity from 93 (ClusterVIII)to 109.67 (Cluster VI); 1000-seed weight from 1.60 g (Cluster VIII) to 2.75 g (Cluster III); grain yield from1981.41 kg/ha (Cluster IV) to 2546.33 kg/ha (Cluster V); and fodder yield from 2858.45 kg/ha (Cluster IV) to 3581.80kg/ha (Cluster III). These results clearly demonstrate the existence of substantial variability among clusters, highlightingthe potential for selecting divergent genotypes to enhance yield and related traits in finger millet [33-39].

Table 8. Clustering of finger millet germplasm accessions based on yield and yield component characters.

Table 9. Mean of eight clusters (Tocher method) of finger millet germplasm accessions for yield and yield components.

Cluster analysis was performed to assess genetic diversity among 33 finger millet genotypes using yield and yield component traits. Based on 11 yield and yield related traits, the genotypes were grouped into eight distinct clusters. Inter cluster distance analysis revealed maximum divergence between Clusters III and VIII (469.44), followed by Clusters III and VI (396.43), and Clusters V and VIII (336.13), suggesting these as promising parental combinations for hybridization (Table 10) [40-47]. The intra cluster distance was minimum (0.00) for Clusters V, VI, VII, and VIII, each comprising a single genotype [48,49]. Greater improvement in yield and its related traits can be expected by crossing genotypes from Clusters III and VIII, as well as Clusters III and VI, owing to their higher genetic divergence Figures 7-10.

Table 10. Intra and inter-cluster distance based on yield and yield components in finger millet.

GMR: https://www.geneticsmr.org/articles-images-2025/gmr-path-192198-g007.png" alt="Image" />

 

Figure 7. Cluster diagram for 33 genotypes of finger millet (Eluesine coracana).

GMR: https://www.geneticsmr.org/articles-images-2025/gmr-path-192198-g008.png" alt="Image" />

 

Figure 8. Hierarchial cluster denrogram illustrating genetic relationships among 33 finger millet (Eluesine coracana) genotypes based on eleven yield and yield component traits.

GMR: https://www.geneticsmr.org/articles-images-2025/gmr-path-192198-g009.png" alt="Image" />

 

Figure 9. Identification of superior finger millet (Eluesine coracana) genotypes based on combined multivariate analysis.

GMR: https://www.geneticsmr.org/articles-images-2025/gmr-path-192198-g010.png" alt="Image" />

 

Figure 10. PCA biplot based on eleven yield and yield component traits of 33 finger millet (Eluesine coracana) genotypes.

Conclusion

The present study involving 33 finger millet genotypes revealed substantial genetic variability, which can be effectively utilized in future breeding programs. Analysis of variance across all agronomic traits confirmed the presence of considerable variability among the genotypes under investigation. For path coefficient analysis, grain yield was considered the dependent variable, while the remaining traits were treated as independent variables. The residual effects, with standardized partial regression values of (1–1.0132) at the genotypic level and 0.880 at the phenotypic level, indicate the contribution of factors beyond those included in the study. Consequently, emphasis should be placed on traits such as number of fingers, finger length, pedicel length, and 1000-grain weight when selecting for yield improvement in finger millet.

Genetic diversity plays a pivotal role in the exploitation of heterosis, as hybrids derived from genotypes belonging to highly divergent clusters are generally more effective than those from less diverse groups. Results from PCA and cluster analysis suggested the utilization of genotypes VR-1131, VR-1147, VR-1150, GPU-67 and VR-1174 in hybridization programs to develop superior advanced lines through transgressive segregation for yield and related traits. Among the traits studied, 1000-seed weight contributed the maximum proportion to genetic divergence (82.95%), followed by plant population per plot (7.39%), underscoring their importance in differentiating genotypes and guiding selection strategies.

Future Scope

The thirteen genotypes exhibited considerable variability across several quantitative traits, indicating strong potential for grain yield improvement through simple phenotypic selection. Furthermore, these genotypes can serve as valuable parents in future crossing programs, where systematic selection will enable the isolation of superior segregants.

Competing Interests

Authors have declared that no competing interests exist.

Declerations To Be Made Regarding Ethical Issues

None

Author's Contributions

JR contributed to the introduction, conducted experiments, and participated in writing the results and discussion sections. SK responsible for conceptualization, supervision, preparation of the materials and methods section, and drafting the conclusions. SR, HK and SR was involved in data processing and statistical analysis. All authors read and approved the final manuscript.

Acknowledgment

The study was supported by Millets scheme, RARS, Palem from Professor Jayashankar Telangana State Agricultural University, Hyderabad.