Statistical analysis and field evaluation of the type 2 modified augmented design (MAD) in phenotyping of flax (Linum usitatissimum) germplasms in multiple environments

摘要

The type 2 modified augmented design (MAD) was used to phenotype seed yield, oil content and fatty acid compositions in a collection of 120 flax genotypes at two locations during three years. All six experiments had the same design, in which whole plots were arranged in 10 rows and 10 columns and each whole plot was split into five paralleled rectangular subplots with a control subplot in the centre of each whole plot. Two additional subplot controls were allocated at random in each of five randomly selected whole plots. Relative efficiency (RE) of adjusted versus unadjusted observations was evaluated for all six experiments. The RE was redefined as a ratio of pooled variance within both plot and subplot controls of the unadjusted values to that of the adjusted values. Two adjustment methods based on the row and column effect of plot controls (M1) and the regression of the test plots on the plot control (M3), were assessed to adjust for soil heterogeneity. The analysis of variance (ANOVA) results revealed that either M1 or M3 alone failed to sufficiently eliminate effects due to both additive and non-additive soil variation across the field. A combined method (M1+M3) appeared to be the most effective in most cases. The redefined RE can be used as an indicator of adjustment efficiency. A joint analysis of 120 flax genotypes over four environments showed that seed yield was significantly affected by environments and had significant interaction of genotype × environment. High yield mean and low coefficients of variation over multiple environments compared with a control cultivar are indicators of a stable and high yielding genotype, whereas oil and linolenic acid content were relatively stable traits. The automated statistical analysis of MAD with the corrected ANOVA and improved observation adjustment was implemented with SAS software and Perl scripts, which are freely available at http://probes.pw.usda.gov/bioinformatics_tools/MADPipeline/index.html.