Multivariate diallel analysis of ethanol withdrawal symptoms in mice

Mice from five highly inbred strains were crossed in a complete diallel design. Five hundred and fifty-five offspring were given a liquid diet containing ethanol as their sole food source for 9 days in an attempt to induce dependence. At 6 hr postwithdrawal on day 10, the mice were given a battery of physiological and behavioral measures designed to assess the symptoms of withdrawal. Littermate data were aggregated separately by sex and the aggregate litter mean data was employed in a multivariate extension of Hayman's diallel analysis. Hypothesis sums of squares and cross-products were tested against pooled-within-cell sums of squares and cross-products using Wilk's lambda as a criterion. For each sex, a highly significant additive effect was found, with some evidence for dominance effects as well, suggesting the raw materials are present for successful genetic selection. There was some evidence for dominance effects as well, but no evidence for maternal and other reciprocal effects was found for the set of seven measures. Composites derived from the additive and general dominance hypotheses correlation matrices and error correlation matrix were compared with a composite index currently being used in a multivariate selection study of ethanol dependence. Total ethanol consumption during the exposure period was positively correlated with low dependence between strains, and negatively correlated with measures of low dependence within cells.