Title : Multivariate analysis of conserved rice germplasm in Ghana
Abstract:
Rice (Oryza sativa) landraces are important for agriculture and crop improvement. However, the lack of characterization data on rice genetic resources in Ghana limits their use in production and crop improvement purposes. Agro-morphological characterisation was conducted on one hundred and thirty (130) rice accessions at the Genbank in Ghana. Rice seedlings were transplanted at a 20 x 20 cm distance on a plot size of 2m x 1m, with 0.75m in-between plot distance. Trait data were collected for 15 traits following the Bioversity International Descriptor list. Diversity indices, Analysis of variance (ANOVA), Principal components, and Trait diagnostic Biplot were established in GenStats and Ward's Minimum Variance Cluster Analysis performed using DARwin Statistical packages.
ANOVA revealed highly significant differences among the genotypes for all the characters under this investigation (p≤.001) except flag leaf width which was significant (p≤ 0.023). Diagnostic Biplots indicate that the major contributing traits to the observed variation are spikelet fertility, grain weight, and panicle length which showed an SDI of 0.4892. 0.492, and 0.4891. The first five principal components cumulatively explained 74.98% of the variation observed. The PC1 explained 31.15 % of the variation and was negatively associated with all traits except spikelet fertility. The main traits associated with PC1 include leaf blade length (0.36), yield per plot (0.337), culm length (0.334), and flag length (0.323). The PC2 explained 15.44% of the variation and correlates mostly with spikelet per panicle (0.46), and filled grain (0.387). Ward's Minimum Variance Cluster Analysis produced five clusters. The first cluster consisted of 26 accessions was more associated with grain weight. The second cluster consisted of 39 accessions and was associated with spikelet fertility. The third cluster consisted of 24 accessions and was associated with awn length and panicle size. Cluster four consisted of 21 accessions and was distinguished by the presence of ligules and high panicle length. The fifth cluster consisting of 20 accessions is characterized by high yield, high panicle number, and high culm number. The narrow distance within the clusters suggests possible duplicates in the collection. Based on the genetic relatedness, a core collection of 25 accessions could be assembled as a representative sample for further evaluation. These clusters could form the basis of parent selection for improving rice for the associated traits.
