hrvatski

Komponente varijance analizirane mješovitim modelom u zajedničkim pokusima kukuruza dvaju hrvatskih instituta

Bc Institute Zagreb and Agricultural Institute Osijek perform joint pre-registration maize yield trials in several locations of Croatia for more than a decade. Th e trials include the newest and the highest yielding hybrid combinations of each maize breeder from the institutes changing their genotypes every year. Trials are balanced within a year, but usual combined analysis of variance across several years is not possible since trials are inherently severely unbalanced. Objective of this study was to estimate variance components for grain yield of the trials planted from 2012 to 2015 in three locations of Osijek, Rugvica and Šašinovec using the mixed model and method of restricted maximum likelihood (REML). Median, the fi rst and the third quartile of all experiments were 10.1, 8.6 i 12.3 t/ha respectively, while the mean was 10.3 t/ha. Th e structure of variance components was similar to those obtained in offi cial registration variety trials by Croatian Institute for seed and seeedlings, whereby the genotypic variance was about 10% of the total variation. Although the percentage seems small, it is equivalent to the genotypic component of variance for grain estimated in offi cial variety trials of Western Europe. Our results indicate that the multi-year pre-registration maize yield trials of the two institutes is a worthwhile test for hybrids that are planned be included into the offi cial variety trials of the Republic of Croatia.

komponente varijance, kukuruz, mješoviti model, pretkomisijski pokusi, prinos

nije evidentirano