Multi-Locational Evaluation of Medium-Staple Cotton Genotypes for Seed-Cotton Yield under the Middleveld Agro-Ecological Zones of Zimbabwe

superior . experiments in a Randomized Block conducted on ten genotypes test genotypes and to evaluate cotton yield performance, stability and adaptability by Analysis of Variance and Genotype and Genotype by Environment Interaction (GGE)


Background
Cotton (Gosspyium Hirsutum) is predominantly a smallholder crop and represents a crucial source of income for millions of farmers and their families in more than 20 countries across all regions of Sub-Saharan Africa (Travella 2017). Despite its economic potential, cotton variety development in Zimbabwe has been very low leaving farmers to continuously cultivate obsolete varieties. Currently cotton production in Zimbabwe is very low, with national average seed cotton yield of 650kg ha -1 (Fig. 1) The cotton genotype selection and recommendations by breeders has been slowed down due to the effects of genotype by environment interaction (GEI). This complicates the identi cation of superior cotton genotypes, thus making selection and recommendation of new genotypes for different environments di cult and expensive (Mare et al. 2017;Mukoyi et al. 2018). Some multi-locational eld experiments from 2012 to 2019 growing season were conducted aiming to evaluate relative response of cotton genotypes across different environments and identify varieties with good adaptation and stability (Baa and Safo-Kantanka 2008) through use of GGE biplot analysis. Multilocational Yield Trials (MYT) are important to evaluate the relationship between genotypes and environments for selected traits by use of a genotype by genotype by environment (GGE) biplot that allows visual assessment of genotype by environment interaction (GEI) pattern (Yan et al. 2000;Yan and Hunt 2001). GGE is the most recent approach proposed by Yan et al. 2000, and has shown extensive usefulness and a more comprehensive tool in quantitative genetics and plant breeding (Yan et al. 2001;Yan and Rajcan 2002). This tool for analysis of GEI is increasingly being used in GEI studies in plant breeding research (Butran et al. 2004). The objectives of this study were (i) to identify the genotype and environmental components that are associated with the GxE interaction across the diverse environments, looking at percentage source of variation and joint effects of G and GE as partitioned by Principal Components (PC) in the total sum of squares, (ii) to identify the ideal genotype(s) based on high yielding potential and stability across test-locations, (iii) to identify mega, representative and ideal environment for testing genotypes and (iv) to identify which variety won where of the given test-locations.

Study Sites
The General description of the sites encompassing longitude, altitude acetra, is given in Table   3.

Results
An across seasons and environments, general combined analysis of variance (ANOVA) was conducted and the results indicated that variance on the measured yield and yield-related traits was due to the presence of genotype by environment interaction (GEI) (P<0.001) except for boll weights. The highest percentage of variation was explained by E/G/GE (60.34%) while G/E+GE together explained the rest of the variation (<40%) ( Table 1). Joint effects of G and GE were partitioned using the GGE biplot analysis explaining total of 59.08% (PC1 = 36.96% and PC2 =22.12%) of the GGE sum of squares ( Table 1). The effect of GxE interaction on the parameters invited the need for further analysis using the GGE biplot analysis to be able to identify genotypes which are stable and adaptable. Overall seed cotton yield mean for the candidates was 1663kg ha -1 , whilst candidate recorded 1755kg ha -1 ( Table 2) which was 5% and 5.5% yield gain over checks CRI-MS1 and CRI-MS2 respectively (Fig. 2).

Genotype Stability Analysis (GEI) for total seed cotton yield for cotton genotypes across seasons and environments
Which-won-where and mega-environments (ME) The GGE scatter plot (Fig. 3) showed dissected pentagon into sectors with winning genotypes located at the vertex of the polygon. The biplot revealed that candidate 917-05-7 (G5) and TN96-05-9 were the winning genotypes in six environments (Chisumbanje Exp, projected perpendicular line to the environmental axis whilst candidate TN96-05-9 was more stable and above average in terms of yield performance (Fig. 4). Candidate 912-05-1 was moderately yielding thus above average and very stable. So candidates 917-05-7, TN96-05-9 and 912-05-1 are selected as good varieties which are high yielding and stable compared to the check varieties CRI-MS1 and SZ9314 which were around average yield performance and highly unstable.

Ideal Genotype and environment
The GGE analysis positioned the candidate genotype 917-05-7 in first concentric ring (Fig.   5), identifying it as the ideal genotype. This also reveals that the genotype is high yielding and moderately stable compared to check varieties which positioned in the 11 th concentric ring thus low yielding and unstable. Some good varieties closer to the ideal genotypes were shown, and these included TN96-05-9, 912-05-1 and GN 96 (b)-05-8. The biplot displayed Umguza as the most ideal environment (Fig. 5)    Ranking biplot showing the high yielding and stable test genotypes