Cluster-based method for a practical leaf area measurement in cassava field

Total leaf area (LA) is typically employed to infer growth and productivity of crop plants. Measurement of LA in field experiment is often infeasible because of too large number of leaves samples, especially for perennial crop plants such as cassava. A cluster-based method was developed, herein, to resolve the massive numbers of sample in a large canopy plant. The method used leaves clusters to represent sets of leaves with comparable LA, to reduce the number of measurements yet maintaining the population size to entire plant samples. In this regard, a clustering board was developed and optimized to size distribution of 189 cassava palmate leaves that allowed a precise leaves classification of the samples. The accuracy of the cluster-based LA measurement (M1: clustered leaves with Image-J, M2: clustered leaves with LA meter) was demonstrated by contrasting with the non-clustered measurement (M3: all leaves with LA meter) using nine field-grown cassava plants. The results showed that the cluster-based method gave comparable measured LA to the current gold standard method, while reducing the number of measurements by 90-98 percent of the total sample. In summary, we proposed that the cluster-based method is a promising technique for a practical measurement of LA in large-scale experiment.