In rice production, the measurement and estimation of growth indicators are indispensable for making decisions on topdressing, estimating protein content in grain for classification collection, determinant analysis of yield and quality, etc. We proposed an estimation method for dry matter weight of rice plants per unit field area (PDW), tiller number per hill (NT), nitrogen content in leaf blade per dry matter weight (N
dw), nitrogen content in leaf blade per unit leaf area (N
a), and nitrogen uptake of rice plants per unit field area (NU), based on plant area index (PAI) and chlorophyll meter readings (SPAD readings). We evaluated predictive accuracy of an estimation equation for each growth indicator and clarified the factors affecting its accuracy. Mean relative error of cross validation (MRECV) for the estimation equation of PDW was 23.3 % for overall growth period by heading. For the predictive accuracy of each growth period, MRECV in the maximum tiller number period was 48.3 %, whereas it decreased for the late growth periods. MRECV in the middle ripening period was 5.7 %. The MRECV for the estimation equation of NT was 9.5 % for the overall growth period by heading, whereas it was 16.2 % in the maximum tiller number period, which was larger than that of the other growth periods. Variations in explanatory variables in the estimation equations of PDW and NT were large in the late growth periods. Thus, the contribution of data observed during these periods to the estimation equations was large, resulting in low predictive accuracy in the maximum tiller number period. The MRECVs for the estimation equations of N
dw and N
a were 15.3 % and 13.4 %, respectively. The predictive accuracy was high for the equation of N
a because differences in a linear relationship between N
a and SPAD readings were small among the growth periods. The small differences were attributable to the fact that the N
a indicator changed with leaf thickness, similar to that observed for SPAD readings. The MRECVs for the estimation equation of NU were 9.8 % for overall growth period by heading and 4.0 % for the middle ripening period. The predictive accuracy of NU mainly depended on the predictive accuracy of nitrogen content in rice plant per unit plant area based on SPAD readings.
View full abstract