IMPROVING SELECTION ACCURACY IN CLONAL ASSESSMENT TRIALS BY ACCOUNTING FOR SITE VARIABILITY

RESEARCHERS ROUTINELY USE ‘blocking’ as a technique to reduce the natural variation within field experiments. This enables treatments to be compared without the confounding effects of differences in environmental factors such as soil type. However, it is often difficult to determine the locality of appropriate blocks within a trial when the natural variation at the site is unknown or difficult to measure. Examples are chemical and physical properties of soil. This is particularly an issue in a sugarcane breeding program, where large land areas are often required to assess the performance of >2000 clones. The assumption of homogeneity of the site within a replicate or block may not always be valid. This assumption violation could lead to inefficient selection. Apparent electrical conductivity (ECa) is an easily measured parameter that can be used to identify differences in soil properties. In this study we demonstrated that ECa could be used in variety trials to improve the selection accuracy. In two clonal assessment trials, each site was grouped into four zones based on ECa measured for 0–90 cm of soil. The difference in average cane yield between high and low performing zones was up to 32 tonnes per hectare. If the soil variation was ignored in clonal evaluation, selection would be biased to high performing zones and clones planted in poor soil zones would be less likely to be selected.