Variety adaptability and yield stability
Analysis of variance indicated that specific genotype and environment combinations were significantly (P < 0.01) different (Table 1). The AMMI analysis divided the main effects of treatments into genotype, environment and G × E interactions (Table 1). There were significant (P < 0.01) differences among the components. Environmental effects accounted for 37.13% of the total variation while the genotypes (G) and the G × E interaction accounted for 23.68% and 23.54%, respectively. Anuradha et al., (2017) observed comparable significant results while evaluating 36 soybean genotypes in three environments. The first and second interaction principal component axis (IPCA) explained 18.76% (IPCA1-9.8%; IPCA2-8.96%) of the interaction. The IPCA1 score indicated that about 66% of the genotypes were highly interactive (Table 4). The most interactive genotypes were DPSB 3 (-28.8) and Nyala (17.23). The least interactive genotype based on first IPCA was DPSB 8 (0.49). Genotype by environment interaction makes it difficult for breeders to make genotype selections and variety recommendations Cucolotto et al., (2007). Prediction of performance or selection of elite genotypes based on seed yield is better achieved by use of multi-environment trials Yan et al., (2010). The high yielding genotypes were Nyala, SBH 7/1/1, SBH 1/12/19 and DPSB 8 [Table 4 and Fig 1(a)]. The highest seed yielding genotype (Nyala) was unstable Fig 1(b). Low yielding genotypes (DPSB 3) had high IPCA1 scores (Table 4) and DPSB 19 in Fig 1 (b) denoting that either high yielding or low yielding genotypes maybe unstable. Selection of genotypes based on seed yield may result in selection of unstable genotypes. However, genotypes identified as seed yield stable were medium yielding [Table 4 and Fig 1(b)] with yields 1.8-21.8% higher than the mean yield of 1267 Kg ha-1.
@table4
Analysis by AMMI determined the best four high yielding genotypes per environment [Table 2 (a)]. The check genotype (EAI 3600) was placed second in two out of the five environments. The best yielding genotype (Nyala) outperformed the check by 16.4%.
The GGE biplot showed genotypes that were stable in yield and adaptable [Fig 1(b) and (c)]. The first principal component is represented on the x-axis and the higher the value along its length on Fig 1(a), the higher the genotype yield. The first principal component accounted for 52.25% of the variation in stability and the two principal components explained 71.8% of the variation. The 2
nd principal component is represented on the y-axis and the further from the average environment axis (AEA) line the genotype is placed, the lower the stability in yield on Fig 1(b).
The stability analysis showed that genotype SBH 4/4/4 as most stable in seed yield while genotype 931/5/34 had longest vector suggesting that it was the least stable [Fig 1(b)]. The GGE biplot indicated two potential main-environments [Fig 1(c)]. The polygon view of the GGE biplot had five sectors that identified wining genotypes for the four sectors [Fig 1(c)].
Genotype adaptability and stability based on oil content
Analysis of variance indicated significant (P < 0.01) genotype differences for their seed oil content across environments (Table 3). The effects due to genotypes, environments and G × E interactions were significant (P < 0.01) among the components. The environment component contributed 52.17% to the total sum of squares while the genotypes contributed 36.57% and the G × E interaction contributed 5.26%. The IPCA1 explained 2.68% of the variation while IPCA 2 explained 1.71%. The mean oil content was highest at Nakuru west (223 g kg-1) and least at Njoro II (179 g kg-1) [Table 2 (b)]. Genotypes SBH 4/4/4, 931/5/34 and Gazelle were the most stable genotypes for oil content based on AMMI analysis and genotypes SBH 4/4/4, Gazelle and SBH 7/1/1 based on GGE biplot analysis [Table 4; Fig 2 (a)]. The choice of stable varieties by both methods of analysis were similar but there were differences in the ranking. Tubic et al., (2011) while evaluating thirteen soybean genotypes observed small differences in oil content stability. Genotype by environment interactions observed in AMMI analysis could be attributed to weather and edaphic factors as was suggested by Tukamuhabwa et al., (2012).
All the genotypes were relatively stable for oil content as they had absolute IPCA1 scores of < 1 (Table 4). Genotypes DPSB 3, DPSB 19, DPSB 8, SBH 6/6/6/2 and SBH 3/8/4/1 were stable but had oil contents lower than the overall mean of 194.1 g kg
-1.
The GGE biplot analysis explained 94.71% of the genotypic variation in stability for oil content [Fig 2 (a)]. The most stable genotype was DPSB 3 while SBH 931/5/34 was the least stable for its oil content. The polygon view GGE biplot for oil content indicated the best or worst genotypes in each environment and groups of environments [Fig 2 (b)]. The best genotypes were located at the vertex of the polygon. Genotype 931/5/34 was the best in oil production in all environments. There were two main-environments that were delineated in the GGE biplot [Fig 2 (b)].
Genotype adaptability and stability based on protein content
Analysis of variance for protein content indicated significant (P < 0.01) genotype and environment interactions (Table 3). Genotypes, environments and genotype by environment interactions accounted for 9.9%, 69.85% and 7% of the variation, respectively. This implied that specific environments were better placed for production of soybean with higher protein content.
Fehr (2003) analyzed protein content and found that genotype × environment interaction had no significant effects on soybean protein components. The first and second interaction principal components accounted for 5.89% of the interaction (IPCA1-3.21%; IPCA2-2.68%). High mean protein content was recorded on genotype DPSB 19 (402.8 g kg
-1) and least on genotype 931/5/34 [(352.1 g kg
-1) (Table 4)]. Interactive scores (Table 4) indicated genotype SBH 7/1/1(1.11) had low stability while DPSB 3 (0.02) had high stability.
Ramana and Satyanarayana (2006) identified a genotype that was stable for seed yield plant
-1, oil and protein content. Studies have indicated that environmental factors have significant effects on protein and oil content in soybean (
Ramana and Satyanarayana, 2006;
Tubic et al., 2011). Environments fitted into one main-environment where DPSB 19 was the winning genotype [Fig 3 (b)].
Stability of important soybean yield components. Stability was highest on EAI 3600 on pods plant-1 and least on DPSB 3 [Fig 4 (a)]. The principal components explained 79.68% of the variation in stability among genotypes. Genotype EAI 3600 had highest stability based on 100-seed weight [Fig 4.4 (b)]. The first principal component explained 80.63% of the variation while the second principal component explained 12.09% of the variation in stability of genotypes based on seed weight.
The principal components accounted for 82.66% of the variation in stability of the genotypes based on number of seeds pod
-1. Genotype SBH 6/6/6/2 was highly stable for seeds pod
-1 while DPSB 3 was unstable for the trait [Fig 4.4 (c)]. In soybean, important yield components are pods plant
-1, seed size and seeds pod
-1 (
Khodadad, 2012). Genotype Gazelle was considered stable by GGE biplot analysis but the genotype had average stability for other yield components. This implies that there are more factors that contribute to stability of genotypes than the contribution by important yield components.
Sujay et al., (2012) had observed that a genotype stable for one trait, may not be stable for other traits as each trait is governed by different sets of genes and influence of environment on the cumulative expression of different sets of genes vary considerably as indicated by variation in stability of genotypes for seed yield [Fig 1 (b)].
Yothasiri and Somawang (2006) observed that genotypes were above average in stability for 100-seed weight. However, stability in number of seeds pod
-1 and number of pods plant
-1 was below and above average among the genotypes.