Comparison of shifted multiplicative model, rank correlation, and biplot analysis for clustering winter wheat production environments

Abstract

Categorization of locations with similar environments helps breeders to efficiently utilize resources and effectively target germplasm. This study was conducted to determine the relationship among winter wheat (Triticum aestivum L.) yield testing locations in South Dakota. Yield trial data containing 14 locations and 38 genotypes from 8 year were analyzed for crossover genotype (G) 9 environment (E) interactions according to the AzzaliniCox test. G 9 E was significant (P \ 0.05) and contributed a small proportion of variation over the total phenotypic variation. This suggested that for efficient resource utilization, locations should be clustered. The data were further analyzed using the Shifted Multiplicative Model (SHMM), Spearman's rank correlation and GGE biplot to group testing locations based on yield. SHMM analysis revealed four major cluster groups in which the first and third had three locations, with four locations in each of the second and fourth groups. Spearman rank correlations between locations within groups were significant and positive. GGE biplot analysis revealed two major mega-environments of winter wheat testing locations in South Dakota. Oelrichs was the best testing location and XH1888 was the highest yielding genotype. SHMM, rank correlation and GGE biplot analyses showed that the locations of Martin and Winner in the second group and Highmore, Oelrichs and Wall in the third group were similar. This indicated that the number of testing locations could be reduced without much loss of grain yield information. GGE biplot provided additional information on the performance of entries and locations. SHMM clustered locations with reduced cross-over interaction of genotype 9 location. The combined methods used in this study provided valuable information on categorization of locations with similar environments for efficient resource allocation. This information should facilitate efficient targeting of breeding and testing efforts, especially in large breeding programs.

Identifying Mega‐Environments and Essential Test Locations for Pearl Millet Cultivar Selection in India

“…The vectors representing locations in SD-scaled biplot should be of equal or similar length if the biplot adequately displays the patterns in the data. The rank correlations among the test locations of a particular mega-environment were used to find the degree of closeness among them (Malla et al, 2010). Nine of the B zone locations along with one A zone location formed the third megaenvironment (ME3) ( Table 3; Fig.…”

Structuring an Efficient Organic Wheat Breeding Program

“…While there are clear elevation, moisture, and temperature trends across Nebraska [18], specific weather events prevent the locations from clustering along these trends. The clusters Peterson [16] described were based on 30 years of data. Hence aposteri clustering of environments is superior to apriori clustering of environments ( [18], Table 2).…”

“…Hence aposteri clustering of environments is superior to apriori clustering of environments ( [18], Table 2). Also, the genotypes used to estimate the environmental similarity will provide different results between nurseries (e.g., the clusters developed by Peterson [16] changed with the nurseries [genotypes] that he used). The elite nursery contains lines with multiple years of testing and on average should have higher and more stable grain yields than lines in the advanced trial where the lines are in their first year of replicated testing.…”

“…Other agronomic performance (e.g., grain yield) and end-use quality (milling and baking) traits are poorly measured in early generations, are subject to large environmental effects, have large G × E, and require larger quantities of seed for testing. Hence for most agronomic and end-use quality traits, they are measured in later generations in replicated trials grown in the target set of environments [16]. Furthermore, the relative weights for selection based upon these traits will vary between production systems and their markets.…”

Algorithmic models of seed yield and its components in smooth bromegrass (Bromus inermis L.) via large sample size under field conditions

“…Molecular marker-assisted selection has long been recognised as a useful tool in plant breeding programs ( Van et al 2010;Kalia et al 2011), and it has been extensively researched (Mahmood et al 2005;Prohens 2009;Yan et al 2009;An et al 2010;Herrmann et al 2010;Hristov et al 2010;Landjeva et al 2010;Malla et al 2010). However, as the competition for the finite crop plant germplasm resources (Bishaw and Turner 2008) has increased, it has triggered a search for natural or semi-natural plant germplasm resources, such as forage plants, that possess important germplasm or genes of resistance to rigorous biotic and abiotic stresses for crop improvement (Humphreys et al 2007;Abreu and Munne-Bosch 2009;Tamura et al 2009).…”