ESTIMATION OF GENETIC-PARAMETERS FOR THE DAM LINES LARGE WHITE AND LANDRACE CONSIDERING DIFFERENT ENVIRONMENTS

Variance components were estimated for fattening performance traits me asured under seven different testing conditions (five nucleus farms an d two test stations). Data on 98969 performance tested Large White and Landrace pigs were used. Variance components were estimated for these two dam lines within each nucleus farm considering common environment as random effect. The estimated heritabilities of daily gain varied a mong nucleus farms between h(2) = 0.26 to 0.49 for Landrace and h(2) = 0.28 to 0.41 for Large White. For backfat thickness the variation was h(2) = 0.25 to 0.53 (Landrace) and h(2) = 0.32 to 0.51 (Large White). The estimated genetic correlations between live weight daily gain and backfat thickness were low and varied from r(g) = 0.01 to r(g) = 0.36 for Landrace and r(g) = -0.04-0.16 for Large White pigs. In a multiva riate approach without considering common environmental effects varian ce components were estimated for live weight daily gain, daily gain on test, backfat thickness and feed intake separately for each of two te st stations. In comparison of the heritabilities estimated for each te st stations showed that heritabilities for backfat thickness were gave hardly any differences between test station or dam line. Heritability estimates for the other traits showed smal differences between dam li nes but varied extremely between test stations with h(2) = 0.15-0.35 f or averaged daily gain on test, h(2) = 0.21-0.34 for live weight daily gain and h(2) = 0.24-0.35 for feed intake. Estimated genetic correlat ions between live weight daily gain and daily gain on test varied from r(g) = 0.74 to 0.77. Daily gain on test and feed intake showed for bo th test stations a genetic correlation of r(2) = 0.49. Hardly any gene tic correlation was observed between daily gain and backfat thickness (r(g) = -0.09-0.08). Based on the large variation of heritabilities in different environments, it was conclused, that selection across farms is only efficient when there is accounted for these heterogeneous var iances in estimation of breeding values.