Negligible interplot competition of cane yield in four-row plots

Interplot competition confounds genotype performance and bias selections in small single-row plots. In the final stage of sugarcane selection, four-row wide by ten-meter-long plots are used on the basis that larger plots experience less interplot competition. However, only the middle two rows are evaluated in these trials and using borderless plots would be more resource efficient. The extent of interplot competition for cane yield in four-row plots was investigated in experiments in the Burdekin and Herbert regions under irrigated and rainfed conditions, respectively, with each experiment containing two replicates of 90 genotypes. Differences in cane yield between the middle two rows (TCHm), representing non-competitive cane yield, and the outer two rows (TCHo), subject to competition from nearby plots, were investigated. Competition was modelled using autoregressive models including genetic competition (GC) and genetic-residual competition models (GEC). Goodness of fit relative to a reference spatial model (AR1) and the correlation between estimated BLUPs were used to assess the effectiveness of the competition models. In both experiments, the interaction between genotype and weigh row had no effect on cane yield. There was a strong correlation between BLUPs of TCHm and TCHo in Burdekin (r = 0.74) and Herbert (r = 0.87). Genetic correlation between TCHm and TCHo was high (rg = 0.99) in both experiments. Relative to TCHm, TCHo had inflated genetic and error variances; however, TCHo had higher broad-sense heritability than TCHm. The GC and GEC models did not improve accuracy when compared to the AR1 model. These indicate that interplot competition was modest for cane yield in four-row plots, even for TCHo. Relative selection efficiency of TCHo was equal to or greater than 1 in both experiments suggesting that unguarded plots are worth considering.