GENETIC DIVERSITY, DROUGHT TOLERANCE AND GENOTYPE x ENVIRONMENT INTERACTION IN FINGER MILLET [Eleusine coracana (L.) Gaertn.] IN ETHIOPIA

Abstract

In Ethiopia, finger millet [Eleusine coracana (L.) Gaertn.] is an important indigenous, orphan or underutilized crop which has got little research and development attention by regional and national agricultural research systems compared to other cereal crops. As a result, very limited information is available on genetic diversity, drought tolerance and adaptability of lowland collections of finger millet genotypes. Therefore, the present study was designed to assess the genetic diversity, identify drought tolerant, stable and high yielding genotypes using morpho-agronomic traits and their performance across different locations. Morphological diversity experiment was conducted in 2016/17 main and off-season at Adet Agricultural Research Center and Koga Irrigation Site, respectively. The drought screening experiment was done at Werer Agricultural Research Center in 2018 off-season. The G x E experiment was conducted at Rama, Assosa, Arsi Negele, Maystebri, Pawe and Sirinka in the 2018 main cropping season. Two hundred and twenty five finger millet genotypes were evaluated in 15x15 simple lattice design with two replications using 11 qualitative and 15 quantitative traits. Moreover, the grain yield data collected from moisture stressed and non-stressed conditions were evaluated using 13 drought tolerance indices and subjected to correlation, PCA and cluster analysis. Thirty six genotypes were evaluated over six environments in the 2018 main cropping season using 6x6 triple lattice design and ten morpho-agronomic traits were collected on plant and plot basis. The result of qualitative traits indicated that the diversity within (0.37-0.98) altitude classes of collection is higher than between (0.02-0.63) altitude classes of collection. The highest diversity was found in altitude class ranging from 751-1250 m.a.s.l. The highest diversity indices were recorded for plant pigmentations (0.99), pericarp persistence (0.82) and growth habit (0.74), showing relatively even distribution of traits within altitude classes. The analysis of variance revealed highly significant differences among the genotypes for all 15 morpho-agronomic traits, except for number of fingers per earhead. High heritability coupled with high genetic advance was observed for culm diameter, number of earheads per plant, earhead width and grain yield per plant, indicating that these traits are regulated by additive gene action. Hence, these traits can be improved through selection. The first six PCs, with eigenvalues greater than one, contributed about 65% of the total variation. The genotypes were grouped into five clusters comprising 13 to 64 genotypes. The correlation analysis showed that xxii xxii grain yield was highly significantly and positively correlated with plant height, culm diameter, finger length, earhead width, harvest index and thousand seed weight both at phenotypic and genotypic levels. As per the path analysis, traits such as harvest index, grain filling period, aboveground biomass, days to heading, plant height, culm diameter and number of total tillers per plant both at genotypic and phenotypic levels showed positive significant correlation and positive direct effect on grain yield. This, indicated that selection of high yielding genotypes for grain yield via these traits would be effective. The results of drought screening experiment revealed that genotypes varied significantly for all of the traits measured under moisture-stress and non-stress conditions, except number of fingers per earhead. The mean grain yield in moisture stressed condition was 1887.2 kg ha-1 and 2652.9 kg ha-1 in non-stress condition with an average reduction of 29%. Acc203257, Acc203259, Acc203326, Acc203398, Acc203401, Acc203414, Acc203423, Acc203445, Acc203253, Acc203429, Acc203480 and Acc203496 were found to be high yielders and performed high to moderate for most of the other traits under both moisture conditions. Based on the results of drought tolerance indices using ranking, cluster, PCA and biplot analysis, Acc203399, Acc203401, Acc203414, Acc203423 and Acc203446 were high yielders and drought tolerant finger millet genotypes. Positive and significant associations of grain yield with SSI, YI, GMP, MP and HM under both moisture conditions were observed, suggested that these indices would be appropriate indices for screening high yielding and drought tolerant genotypes. The combined analysis of variance showed significant differences among environments, genotype x environment interactions (GEIs) and genotypes for all the measured morpho-agronomic traits, except plant height, which were not significantly affected by the main effect of environment. AMMI analysis of variance exhibited that mean squares of the environments, genotypes and GEIs were highly significant for grain yield. The largest proportion of the total variation was explained by the environments (42.38%), followed by the GEIs (41.84%) and the genotypes (11.49%). The first five interaction principal component axis (IPCAs) were highly significant and explained about 100% of the variation due to GEI. The first two IPCA axis (IPCA1 and IPCA2) explained about 67.3% of the GEI variation; therefore, AMMI model with the first two IPCA was the best predictive model. The two principal components of the GGE biplot explained about 60% of the total GEI variation, where PC1 and PC2 accounted for 31.34 and 29.16%, respectively. GGE biplot identified G35 as an ideal genotype and E2 (Assosa) as ideal test environment and E6 (Sirinka) as less representative environment to discriminate genotypes. The ‘which-won-where’ analysis of the GGE biplot grouped the six environment into four mega environments (MEs) with different “winning” genotypes. AMMI and GGE biplot, ASV, YSI, Wricle’s and Shukla stability analyses identified (G11) (Acc203336), G20 (Acc203413) and G27 (Acc203445) as stable and high yielder, whereas G16 (Acc203397) was the least stable genotype. Overall, the present study indicated the existence of a wide genetic diversity among finger millet genotypes for yield and yield-related traits, which could be used as a source of variability for the future finger millet breeding program.