Assumptions underlying the use of economic weights: are they valid in breeding for eucalypt kraft pulp?

An existing breeding objective cost function relating standing volume at harvest, basic density, pulp yield and stem form to the cost of producing unbleached eucalypt kraft pulp was used to examine assumptions underlying the use of economic weights in prediction of gain. The cost function showed a non-linear relationship between total pulp-cost and the levels of all traits in the breeding objective cost function, resulting in a decline in the true economic weight at higher trait values. The objective cost function also showed interdependence between traits: the economic weight for one trait was dependent upon the value of other traits. These effects resulted in predicted gains being overestimated, with errors being greater at high selection intensities: at a selection intensity of 1% expected gain was over-estimated by 25%. It is suggested that the use of economic weights for prediction of gain may be inappropriate where exploitable variation is great and selection pressure high. An alternative method of gain determination is proposed: predicted values of selection traits are used to generate values for objective traits which are substituted directly into the breeding objective cost function thus avoiding errors due to non-linearity and interdependence.