Growth results of 43 ostrich males and females were used to compile growth parameters of a flock that is representative of Oudtshoorn birds, using the Gompertz model. Growth results comprised 19 recordings of body weight for each individual commencing from 1-day-old to day 520. Individual fits of data to the growth model were exceptionally good, resulting in low CV values of 0·02 and 0·03 for the mean values of all estimates of mature live weight for male and females respectively. The estimated mean mature body weights for the flock were 119·2 kg for males and 122·3 kg for females. Rates of maturing were 9100 and 8500 for males and females respectively which corresponded to ages of 180·83 and 199·2 days at which maximum gain in weight were achieved. None of these parameter estimates differed significantly between sexes. Results in the present study suggested substantial adjustments to growth parameters for Oudtshoorn ostriches published by du Preez, Jarvis, Capatos and de Kock, 1992.

A method to obtain linear selection indexes that maximize objectives that involve average profit in one or more generations within a planning horizon based on non-linear profit functions, was derived through application of optimal control theory. The method involves simultaneous optimization of indexes for each generation in the planning horizon. Optimum linear indexes were found to be conform indexes derived from selection index theory, with economic values equal to a weighted sum of partial derivatives of the profit function at the trait means which result in each generation of the planning horizon. Numerical procedures to derive optimum indexes are presented. Methods and properties of alternative strategies for selection witli non-linear profit functions are illustrated for selection on egg weight and rate of lay in poultry. In the example, the additional benefit of selection indexes that maximize cumulative net present value of profit over a planning horizon of10 years was small relative to use of traditional selection procedures. Optimal indexes were also derived with a derivative-free non-linear programming optimizer, with identical results. The latter method also allows incorporation of additional constraints.

Possible extensions of the optimal control methodology to address other problems related to optimization of selection over multiple generations are discussed.

Estimation of genetic parameters for litter size in Danish Landrace and Yorkshire pigs