7th World Congress on Genetics Applied to Livestock Production, August 19-23, 2002, Montpellier, France HETEROGENEOUS GENETIC VARIANCES OF DAIRY COW PERFORMANCE IN SICILIAN ENVIRONMENTS E. Raffrenato 1 , R.W. Blake 2 , P.A. Oltenacu 2 and J. Carvalheira 3 1 Consorzio Ricerca Filiera Lattiero-Casearia, 97100 Ragusa, Italy 2 Department of Animal Science, Cornell University, Ithaca, NY 14853, USA 3 Universidade do Porto, Vairão, Portugal INTRODUCTION Alternative environmental definitions have been used to study the potential interaction of genotype and environment (GEI). Studies using herd mean and phenotypic variance of milk yield to differentiate herd environments showed important heterogeneity in genetic and residual variances across geographic regions. Specific objectives of this study were to estimate the genetic and residual variances for yields of milk, fat and protein, and somatic cell score of Friesian and Brown Swiss cows in contrasting Sicilian herd environments. We were especially interested in the potential sacrifices from genotype-environmental interaction in low opportunity herds. MATERIALS AND METHODS Final data sets of mature equivalent yields of milk, fat and protein and somatic cell counts (SCC) comprised 8897 305-day lactation records of daughters of 825 Friesian sires and 1143 records of daughters of 220 Brown Swiss sires. A weighted somatic cell score (WSCS) per lactation was calculated by adjusting SCC with test day milk yields (Kennedy et al., 1982). Three criteria were used to define alternative herd environments. The phenotypic within herd- year standard deviation (HYSD) for 305-day ME yield was used to group herds into alternative environments: low HYSD <1330 kg and high HYSD >1370 kg for Friesian and, low HYSD <950 kg and high HYSD >990 kg for Brown Swiss. A second criterion grouped herds by the incidence of abnormal lactation in primiparous and/or pluriparous cows. Abnormal lactation was defined as undetectable ascent to peak daily yield testing the sign of the b parameter of the gamma function using a type I error of 0.05. The herd management criterion comprised an unweighted subset of 17 milk-enhancing practices obtained by field survey to stratify farms into high and low opportunity environments. The distance of Jaccard was calculated among all farms (using the Distance Macro in SAS; Kuo, 1997). The Lance-Williams (1967) flexible-beta method was used to form high and low opportunity herd clusters. The variance-covariance components were estimated using a multiple-trait linear mixed sire model for each breed. All traits were analyzed for the average herd and for alternative high and low opportunity environments. A two-trait linear mixed sire model was used to estimate the genetic covariance and correlation between high and low opportunity environments. Session 18. Genotype by environment interaction Communication N° 18-03