Genetic Variability, Character Association and Path Analysis for Yield and Yield Components in Foxtail Millet [Setaria italica (L.) P. Beauv.] Accessions under Multi Environments in Foothills of Nagaland

Foxtail millet [Setaria italica (L.) P. Beauv.] is a self-pollinating, C4 cereal crop with a rich history of cultivation dating back to 5000-6000 BC along the Yellow River in China (Li and Wu 1996). The Food and Agriculture Organization of the United Nations and International Crops Research Institute for the Semi-Arid Tropics projected that the global foxtail millet production amounted about 6 million ton in the year 2023 wherein India alone contributed more than 50% of the total production. In India, foxtail millet is cultivated on about 0.8 lakh hectares area with 0.6 lakh tonnes production in Andhra Pradesh, Karnataka, Telangana, Rajasthan, Maharashtra, Tamil Nadu and north eastern states (Hariprasanna, 2023 and Pine, 2020). The International Year of Millets 2023, United Nations initiative, aims to raise awareness about the significance of millets as a nutritious and sustainable food source, while promoting their cultivation, consumption and trade. Improving millet grain yield is crucial for enhancing food security, nutrition and economic stability, especially in developing countries; leveraging diverse germplasm, conventional molecular breeding approaches, participatory breeding, policy support and extension services can develop high-yielding, resilient varieties, promoting adoption, maximizing benefits and contributing to biodiversity, climate resilience and poverty reduction, thus fulfilling the goals of the International Year of Millets.
       
Grain yield is a complex trait influenced by both genetic and environmental factors. Selection only for yield is ineffective as yield is complex trait (Madhavilatha et al., 2022). Therefore, more effective approach is to made for selection of traits that contribute to yield (Munirathnam et al., 2006). Phenotypic variance allows breeders to assess the overall variability in yield and yield related traits. On other hand, genotypic variance is quantifying the genetic contribution to trait variability. Heritability tells us how much of a trait variation is due to genes (Schmidt et al., 2019). Correlation analyses and the interplay between traits are crucial for understanding how improvements in one trait can lead to simultaneous changes in others. Path coefficient analysis is a valuable statistical method used to separate correlation coefficients into direct and indirect effects (Schmidt et al., 2019).
       
Foxtail millet is an important staple crop used extensively for food and fodder. Foxtail millet has one of the largest collections of cultivated as well as wild-type germplasm rich with phenotypic variations and hence provides prospects for identifying elite and novel variants suitable to varied agroclimatic conditions. Growing foxtail millet in diverse environments of Nagaland helps assess genotypic adaptability and performance under varied conditions. This approach identifies stable, high-yielding genotypes, informing breeding programs to develop resilient varieties suitable for the regions. Hence, present study was conducted in 30 foxtail millet genotypes to characterization and assessment of variability in multi-environments in foothills of Nagaland to identify better genotypes to be incorporated in crop improvement programs is timely and useful for foxtail millet breeders.

Experiment location
 
The investigation was carried out during July 2022 to May 2023 for four different dates of sowing (Table 1). Each sowing date was chosen to create varying environmental conditions, including different temperatures and moisture levels throughout the crop growth stages. Two environments maintained under rainfed condition and the remaining two environments are maintained under irrigated condition. The experiment was conducted at the Research Farm of the Department of Genetics and Plant Breeding, School of Agricultural Sciences, Nagaland University, Medziphema, India.
 

 
Plant materials
 
Thirty genotypes of Foxtail millet including one check variety were collected from Indian Institute of Millets Research (IIMR), Hyderabad to assess genetic variability across different environments. List of 30 genotypes were presented in Table 2.
 

 
Experimental design and intercultural practices
       
The experiment used was randomized complete block design (RCBD) with three replications across different environments. Each of the three replications had 30 plots (1m×1m) spaced 10cm apart with plants and rows 10 cm and 22.5 cm apart, respectively. The total plot size was 30 m×5 m accommodating 90 beds. Recommended agricultural practices were followed throughout the experiment.
 
Data collection
 
A total of fourteen quantitative characters of foxtail millet were taken for experimentation. These characteristics were chosen based on descriptions and guidelines provided by PPV and FR in 2001 (DUS). For each characteristic, data were gathered from five randomly sampled plants from each genotype and in each replication. The quantitative data encompassed various traits, including days to 50% flowering (DF), days to maturity (DM), plant height (PH) (cm), panicle length (PL) (cm), flag leaf length (FL) (cm), flag leaf width (FW) (cm), peduncle length (PDL) (cm), total tiller numbers per plant (NT), panicle width (PW) (cm), biological yield (BY) (g), harvest index (HI) (%), test weight (TW) (g), fodder yield per plant (FY) (g) and grain yield per plant (GY) (g). Days to 50% flowering and days to maturity data were collected on plot basis.
 
Statistical analysis
 
The analysis of variance was conducted using the OPSTAT open-source software to assess the pooled data. Burton (1952) formulas were employed to calculate the phenotypic coefficient of variation (PCV) and genotypic coefficient of variation (GCV). Broad-sense heritability (h²) was determined using the Burton and De Vane (1953) formula. Genetic gain was assessed utilizing formulas proposed by Lush (1949). The OPSTAST software was utilized for computing genotypic and phenotypic correlation coefficients in line with Kashiani and Salehs (2010) approach. Additionally, path coefficient analysis was employed to partition genotypic and phenotypic correlations into direct and indirect effects (Dewey and Lu, 1959).

Analysis of variance
 
The pooled analysis of variance (ANOVA) was used to examine the interactions between different genotypes and environments. Table 3, presents the results of the pooled ANOVA for all genotypes across various environments, focusing on yield and its components. There were significant variations observed among the different environments (E), genotypes (G) and the interaction between genotypes and environments (G×E). In fact, all the variables in present study showed highly significant differences (at 5%) in terms of the environment, genotype and genotype-environment interaction. These significant differences suggest that there is a substantial amount of genetic variation among the evaluated genotypes. Comparable findings are presented in studies conducted by Zhang et al., (2023) on foxtail millet.
 

 
Variability analysis
 
Genetic and environmental factors contribute to variation within populations. While genetic variability is heritable across generations, distinguishing between heritable and non-heritable traits poses challenges for breeders during the selection process. Therefore, before initiating a thoughtful breeding effort, breeders need to differentiate between traits that are heritable and those that are not. Table 4, displays the estimated phenotypic coefficient of variation (PCV) and genotypic coefficient of variation (GCV) for all characters. Several traits displayed moderate variability (10-20%), suggesting moderate fluctuations in their measurements. Traits such as "plant height" (GCV: 10.06, PCV: 11.09), "peduncle length" (GCV: 11.87, PCV: 13.16), "panicle width" (GCV: 13.90, PCV: 15.69), "grain yield per plant" (GCV: 13.56, PCV: 15.33), "biological yield" (GCV: 14.85, PCV: 16.05), "flag leaf width" (GCV: 15.35, PCV: 17.60), "fodder yield per plant" (GCV: 16.80, PCV: 17.93) and "panicle length" (GCV: 17.85, PCV: 19.50) may exhibit slight variations but generally remain within an acceptable range. Similar studies reported by Ayesha et al., (2019). Genetic advance and heritability are crucial in crop improvement. In the present study, high heritability traits like test weight, fodder yield, grain yield and biological yield indicate a strong genetic influence on crop productivity. Other traits like plant height, panicle length and days to 50% flowering also show high heritability, making them ideal for selection in breeding programs. Conversely, traits with medium heritability, like harvest index, are more affected by environmental factors.
 

       
Genetic advance measures the potential extent a population can progress through selection. While heritability alone may not consistently indicate substantial genetic gains, it becomes significant when coupled with a high genetic advance. In the present study, traits such as fodder yield per plant, panicle length, biological yield, flag leaf width, peduncle length, panicle width and grain yield per plant show high heritability coupled with high genetic advance, indicating that they are strongly influenced by genetic factors and can be improved through traditional breeding methods. These traits predominantly exhibit additive gene action. Traits like plant height, test weight, number of basal tillers and flag leaf length exhibit high heritability coupled with moderate genetic advance its implies both additive and non-additive gene actions. This suggests that genetic improvement can be achieved through traditional breeding methods, as well as by harnessing non-additive gene interactions. On the other hand, traits such as days to 50% flowering and days to maturity have high heritability but low genetic advance. This suggests that their improvement through selection and breeding might be limited. This could be due to the involvement of non-additive gene actions, where gene interactions play a larger role than individual genes. The medium heritability and low genetic advance observed in traits like harvest index indicate that their expression is strongly influenced by environmental factors and involves non-additive gene action. These complex traits require specialized breeding strategies and alternative approaches to achieve significant improvement. Similar study was also reported by Patel et al., (2018).
 
Correlation analysis of the traits
 
Table 5 presents the correlation coefficients (both genotypic and phenotypic) for the 14 yield attributes in the combined analysis. The correlation coefficients, specifically Pearson’s correlation coefficient (r-value), help identify relationships between independent variables.
 

       
In present study, grain yield per plant was positively and significantly associated with various traits. These included days to 50% flowering (r_g: 0.232*, r_p: 0.238**), days to maturity (r_g: 0.276**, r_p: 0.257**), plant height (r_g: 0.331**, r_p: 0.312**), panicle length (r_g: 0.513**, r_p: 0.356**), flag leaf length (r_g: 0.190*, r_p: 0.297**), peduncle length (r_g: 0.278**, r_p: 0.236**), biological yield (r_g: 0.924**, r_p: 0.889**) and fodder yield per plant (r_g: 0.868**, r_p: 0.756**). These associations were observed at both the genotypic and phenotypic levels. The number of basal tillers (r_g: 0.022NS, r_p: 0.225*) and panicle width (r_g: 0.131^NS, r_p: 0.218*) also showed positive and significant associations, but only at the phenotypic level. Similar result was also reported by Ayesha et al., (2019).
 
Direct and indirect effects of yield traits on grain yield per plant
 
By using path coefficient analysis, it was found that each component played a dual role - not only did it directly impact grain yield, but it also had an indirect effect on other component characters. This is a crucial finding because such nuances were not visible through traditional correlation analysis. The study highlights the importance of taking a more comprehensive and multi-dimensional approach to understanding the complex relationship between yield and its components traits (Amarnath et al., 2018).
       
The results of the path analysis are presented in Table 6. Wright (1921) distinguished between direct and indirect effects by assigning correlations to evaluate the cause-and-effect relationship more precisely. The current study found a significant correlation among various yield and yield contributing components. These variables have both direct and indirect effects on the grain yield per plant and its contributing traits due to their interrelation.
 

       
Our path analysis results indicate that biological yield has the greatest direct effect on grain yield per plant (r_g=2.093, r_p=1.956), followed by harvest index (r_g=0.0915, r_p=0.0987), flag leaf width (r_g=0.0150, r_p=0.0013) and number of base tillers (r_g=0.0056, r_p=0.0053) at both the genotypic and phenotypic levels. At the genotypic level, days to flowering (r_g=0.0329) and peduncle length (r_g=0.0036) show a positive direct effect. Panicle length shows a positive direct effect at the phenotypic level on grain yield per plant. These characteristics can be used to develop an effective selection index for improving the yield of foxtail millet. Lenka and Mishra (1973) observed a similar classification trend for path coefficients, where values greater than 1 were considered very high, 0.3-1 were high, 0.2-0.29 were moderate, 0.1-0.19 were low and 0.00-0.09 were negligible. In our study, a residual effect of (r_g=0.0001, r_p=0.0139) indicates that the causative features explained approximately 99.98% of the variability in grain yield per plant, leaving only 0.02% unexplored. Similar report was given by Sapkal et al., (2019).

In the present study, traits such as fodder yield per plant, panicle length, biological yield, flag leaf width, peduncle length, panicle width and grain yield per plant show high heritability coupled with high genetic advance indicating that they are strongly influenced by genetic factors and can be improved through traditional breeding methods. Characters like panicle length and biological yield showed positive correlation with grain yield and also exhibited direct positive effect on grain yield. Therefore, present study suggests that while selection emphasis should be given on panicle length and biological yield.

All authors declare that they have no conflict of interest.