Submitted to: Grass and Forage Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/1/2007
Publication Date: 2/8/2008
Citation: Conaghan, P., Casler, M.D., Mcgilloway, D.A., O'Kiely, P., Dowley, L. 2008. Genotype _ environment interactions for forage yield of Lolium perenne L. sward plots in Ireland. Grass and Forage Science. 63:107-120. Interpretive Summary: Testing new forage varieties is an expensive and time-consuming exercise, requiring a balance between the number of locations, harvest years, and sowing years that provide the greatest information for the least cost. We used a historical data set of perennial ryegrass forage yields from six locations to identify an optimal resource allocation for the Irish National Perennial Ryegrass Test. Based on the structure of this data set, we recommend a system that is based on 2 (or 3) locations, 3 (or 2) sowing years, and 2 harvest years per trial. This information will be of value to agronomists, breeders, seed dealers, and farmers who rely on the results of the Irish National Perennial Ryegrass Test to choose varieties for forage production.
Technical Abstract: Lolium perenne L. (perennial ryegrass) is by far the most widely sown grass species in Ireland. Genotype by environment (G by E) interactions are a frequent occurrence in forage yield evaluations. The objectives were to determine (i) the nature and relative magnitudes of the pertinent G by E interaction variance components for forage dry matter yield of perennial ryegrass sward plots in Ireland and (ii) the optimal allocation of replicates, locations, and years in a testing program. Sixteen perennial ryegrass cultivars were sown at six locations throughout Ireland between 2000 and 2004. Plots from each sowing were harvested for 2 consecutive years under a simulated mixed grazing and conservation management program. The largest component of the G by E variance was generally genotype by location by year emphasizing the need for evaluation of genotypes across locations and years in order to adequately characterize genotypes for differences in yield. Relative differences among genotypes from year to year and location to location were due mainly to changes in genotype rankings. Weather was estimated to have a greater effect on annual variation than age-of-stand. The optimum allocation of resources for a testing program was estimated at four replicates per location, and either two locations and 3 sowing years or three locations and 2 sowing years with 2 harvest years for each sowing year. The most appropriate option depends on the relative importance of time vs. financial resource