Analysis of sugarcane yield productivity trends in the wet tropics at a district level

Data on tonnes of sugarcane per hectare (TCH) are now available to assess varietal improvement over time on a district basis. This study was conducted to develop a methodology for regular analysis of productivity data for mill districts to assess productivity trends over time, adjusted for variety and variety x environment interaction effects. In a 1996 Sugar Research and Development Corporation Report, Leslie and Wilson investigated district level data using a single varietal indexing method. This methodology was found to be subjective in the selection and calibration of the reference varieties. Restricted maximum likelihood (REML) methodology, a more rigorous analysis for district level data, was used to overcome this subjectivity. Best linear unbiased predictors (BLUPs) for random effects and best linear unbiased estimates (BLUEs) for fixed effects were computed for the mixed model analysis. Two analyses were conducted using combined district data and separate district data. Varieties were ranked according to their BLUP values, and district trends in TCH were plotted for the period 1958-1995. In the combined analysis, the BLUPs for TCH for each variety have shown little improvement since the introduction of Q99 in 1976. Separate district analyses indicated more recent advances in Babinda and Tully, but not Mulgrave. In general, the district production estimates from the separate district analyses were similar to those from the combined district analysis. Overall production of TCH in each district has increased over time, in part because a greater proportion of the area in the regions has been allocated better yielding varieties.