The objective of this thesis was to investigate the uncertainty in the input parameters needed to calculate selection index values for the New Zealand radiata pine breeding program. A novel methodology was developed to determine which input parameters are the most important determinants of selection index value. It was found that age-age correlations related to the objective trait tree volume were the most important parameters but paradoxically there is little information available on these parameters. New estimates are provided for ten of the 20 age-age genetic correlations currently required to compute the radiata pine selection index in New Zealand, using information on phenotypic correlations.

Distributions of the values of economic weights for the five objective traits currently used in the radiata pine breeding programme (tree volume, sweep, branch size, wood density, and wood stiffness) were estimated. These estimates were derived using a bio-economic model that involved a partial regression that was run on 111 quarterly New Zealand domestic log prices series from Sept-1994 to Jun-2020 to provide market variation and hence variation in the value of economic weights.

Robust selection (an application of risk analysis) was implemented using the estimated probability distributions of economic weights, where the selection of economic weights was constrained by correlations between the economic weights. These correlations were calculated from the data produced by the bioeconomic model.

As a result, it was found that the trees selected by robust selection were similar than those selected by deterministic selection. The main reason for this result was the lower-than-expected variability in economic weights obtained by the bio-economic model. This means that the uncertainty in future log prices is not relevant because it does not change the decision of which trees must be selected, across the range of economic weight produced by the range of historic log prices.