In this study, the use of different multivariate approaches to classify
rapeseed genotypes based on quantitative traits has been presented. Tree
regression analysis, PCA analysis and two-way cluster analysis were applied
in order todescribe and understand the extent of genetic variability in
spring rapeseed genotype by trait data. The traits which highly influenced
seed and oil yield in rapeseed were successfully identified by the tree
regression analysis. Principal predictor for both response variables was
number of pods per plant (NP). NP and 1000 seed weight could help in the
selection of high yielding genotypes. High values for both traits and oil
content could lead to high oil yielding genotypes. These traits may serve as
indirect selection criteria and can lead to improvement of seed and oil yield
in rapeseed. Quantitative traits that explained most of the variability in
the studied germplasm were classified using principal component analysis. In
this data set, five PCs were identified, out of which the first three PCs
explained 63% of the total variance. It helped in facilitating the choice of
variables based on which the genotypes? clustering could be performed. The
two-way cluster analysissimultaneously clustered genotypes and quantitative
traits. The final number of clusters was determined using bootstrapping
technique. This approach provided clear overview on the variability of the
analyzed genotypes. The genotypes that have similar performance regarding the
traits included in this study can be easily detected on the heatmap.
Genotypes grouped in the clusters 1 and 8 had high values for seed and oil
yield, and relatively short vegetative growth duration period and those in
cluster 9, combined moderate to low values for vegetative growth duration and
moderate to high seed and oil yield. These genotypes should be further
exploited and implemented in the rapeseed breeding program. The combined
application of these multivariate methods can assist in deciding how, and
based on which traits to select the genotypes, especially in early
generations, at the beginning of a breeding program.