Abstract

Tall fescue (Festuca arundinacea Schreb.) is a primary cool-season grass species that is widely used as a cold-season forage and turfgrass throughout the temperate regions of the world. The key seed yield components, namely fertile tillers m^-2 (Y₁),spikelets fertile tiller ^-1 (Y₂), florets spikelet ^-1 (Y₃), seed number spikelet ^-1 (Y₄), seed weight (Y₅), and the seed yield (Z) of tall fescue were determined in field experiments from 2003 to 2005. The experiments produced a large sample for analysis. The correlations among Y₁ to Y₅ and their direct and indirect effects on Z were investigated. All of the direct effects of the Y₁, Y₃, Y₄ and Y₅ components on the seed yield were significantly positive. However, the effect of Y₂ was not significant. In decreasing order, the contributions of the five components to seed yield are Y₁ >Y₄ >Y₃ >Y₅ >Y₂. Y₄ and Y₅ were not significantly correlated with Z. However, the components Y₁, Y₂ and Y₃ were positively correlated with Z in all the three experimental years and the intercorrelations among the components Y₁, Y₂and Y₃ were significant. Ridge regression analysis was used to derive a steady algorithmic model that related Z to the five components; Y₁ to Y₅. This model can estimate Z precisely from the values of these components. Furthermore, an approach based on the exponents of the algorithmic model could be applied to the selection for high seed yield via direct selection for large Y₂, Y₃ and Y₅ values in a breeding program for tall fescue.

Key words: Modelling, seed yield, components, tall fescue, path and ridge analyses, large sample.