The precision of estimates of realized genetic parameters obtained from multiple-trait selection experiments is considered. Statistical optimality theory is used in conjunction with Monte-Carlo techniques to investigate the relationship between index weights and the variability of the estimates. Results indicate that arbitrary choice of index weights is ill advised, and that with some prior knowledge (or good guesses) about the magnitudes of the genetic parameters, index weights can be identified that yield estimates with predictable variation.

Two experiments were conducted to examine seasonal changes in circulating LH concentrations in ovariectomized heifers. In experiment 1, four Holstein heifers were ovariectomized in April 1977 during middiestrus. Blood samples were collected daily for 30 days surrounding each equinox and solstice for one year to examine changes in plasma LH levels at the time of seasonal photoperiod changes. The LH concentrations were highest during the winter solstice period and lowest during the summer solstice period. In addition, samples taken at two-week intervals indicated a distinct LH profile with maximal LH concentrations during November–April and minimal concentrations during May–October. In experiment 2, eight Holstein heifers were ovariectomized in June–July, 1979 and given an estradiol or a control implant in October. A distinct LH profile for the interval extending from January, 1980 to February, 1981 was found in the heifers that were not treated with estradiol. Concentrations were maximal during December–April and minimal during May–November. The LH profile followed a similar pattern in the estradiol-treated heifers; however, the overall profile was at a higher level. These data indicate that underlying seasonal reproductive mechanisms are present in cattle even though the species ovulates and breeds throughout the year.

Atchley, W. R. (Dept. Entomology and Dept. Genetics, Univ. Wisconsin, Madison, Wisconsin 53706); E. V. Nordheim (Dept. Statistics and Dept. Forestry, Univ. Wisconsin, Madison, Wisconsin 53706); F. C. Gunsett1 (Dept. Meat and Animal Science, Univ. Wisconsin, Madison, Wisconsin 53706); and P. L. Crump (Dept. Veterinary Science, Univ. Wisconsin, Madison, Wisconsin 53706) 1982. Geometric and probabilistic aspects of statistical distance functions. Syst. ZooL, 31:445–460.—Some geometric and probabilistic aspects are examined for three commonly used statistical distance functions. It is argued that the distance between centroids can be viewed as a probabilistic function reflecting both the geometric position of each taxon and/or the variance and correlation between the variables. A Monte Carlo simulation study is carried out to assess the effects of sample size, different variances and intercorrelations on the variability or “precision” of the distance estimates using the Euclidean distance, Pearson's Coefficient of Racial Likeness, and the Mahalanobis distance. All three distance estimators are shown to be biased and consistently overestimate the theoretical distance. The effects of these results on studies in systematic biology are discussed. [Distance statistics; Euclidean distance; Pearson's Coefficient of Racial Likeness; Mahalanobis distance; numerical taxonomy; multivariate geometry.]

Log-linear and generalized least-squares approaches to analysis of categorical data are briefly introduced. Two data sets are analyzed to illustrate the methods of analysis.

The genetic change from multiple-trait selection experiments can be equated to the regression of genotype on phenotype. This gives rise to a method of obtaining estimates of additive genetic variances and covariances. The method requires the use of selection weights, derived by means of the index-in-retrospect, to provide invariant solutions. Solution variance estimates obtained from Monte Carlo simulation do not agree with variance estimates from ordinary least squares methods. This indicates that the errors are distributed with some structure V. A form of V is proposed which utilizes knowledge of the errors. Monte Carlo variance estimates from generalized least squares (GLS) methods agree closely with the average variance estimates from GLS when the proposed V is used. Use of an estimated V, derived after the initial estimation procedure, is shown to provide adequate information on the variance of the estimates.

Selection for feed conversion substantially influenced growth and gross feed efficiency of mice. Realized heritabilities and genic correlation for increased gain on fixed feed intake (FF) and decreased feed intake on a constant gain (FG) were estimated to be .56, .73 and -.93, respectively. The genic correlations between FF and 56-day weight and between FG and 56-day weight were estimated to be .67 and -.95, respectively. The relative efficiency in changing FF or FG by selecting directly for 56-day weight was found to be .7 in either case. When efficiency was defined as the incorporation of biomass independent of maintenance, no difference was found between selection treatments. Differences in mature weight and feed intake approached significance with selected lines having higher means than controls. The absence of a correlated response in the ability to use energy for growth indicates that selection for feed to gain ratio changes maintenance requirements but not requirements for growth.

A technique of visual assessment of cattle for reproductive efficiency, described by Professor Jan Bonsma of South Africa, was evaluated in two well-managed large herds of 12 to 34 Brahman cross heifers and cows located in the dry tropics of north Australia. Individual lifetime performance records were available for all animals. Experienced cattlemen carried out the assessments. Higher scores were previously claimed to indicate higher fertility.The technique had high repeatability (0.7) and was quickly learned by the assessors. Scores from visual assessment had no useful predictive value for either heifer or cow fertility or for growth rate up to 27 mo of age, although 2.5-yr-old heifers which were scored as subfertile matured into 4% smaller cows than heifers which had scored higher. Scores decreased as fatness increased (P < 0.05).Some biases in visual assessment occurred. Lactating cows scored higher than nonlactating cows (P < 0.05), independently of their reproductive record. Red and grey cows scored higher than brindle and black/brown cows (P < 0.05). Bonsma scores were not influenced by the percentage of Brahman in the genotype. Significant, but apparently random, age effects on scores also occurred.It was concluded that the visual assessment criteria described by Bonsma were of no practical value in assessing potential productivity of breeding animals in well-managed Brahman cross cattle in the dry tropics.