A simple tillering model for irrigated japonica rice based on measured relative SPAD for lower reaches of Yangtze River delta, China

We suggested using Relative SPAD of the 3rd and 4th leaf (from the apex) of the main stem (RSPAD), which contains information of both LNC and LAI as the sole input variable to predict tillering rate, and can be measured without destructive sampling. We studied the relationship between Tillering Rate Relative to Tillerable Tiller (TRTT) and LAI and RSPAD from a wide range of plant densities and N inputs. And, based on the "synchronously emerging characteristics" of tillers and the quantitative relationship between TRTT and RSPAD, a simple and applicable model for predicting rice tillering dynamics during tiller-increasing period (before jointing stage) have been presented. Moreover, the model was also validated using wide range of plant densities and N inputs data in two filed experiments. Results showed that TRTT decreased with the increase of LAI. The relationship between TRTT and LAI was affected by N input level, but not by planting density. At a certain LAI, the higher N input level resulted in a greater TRTT. The N input level had a significant effect on critical LAI (LAIc), which was greater in high N input level than lower level. Tillering stopped at LAI of 2.5 to 4.0 depending on N input level. Nonetheless, planting density had little effect on LAIc. TRTT increased with the increase of RSPAD. Neither N input level nor planting density had significant effect on the relationship between TRTT and RSPAD. The coefficient of determination (r2) and the root mean squared error (RMSE) between the predicted and observed tiller number were ranged from 0.939 to 0.964 and 24.56 to 30.72, respectively. These values of r2 and RMSE indicates that the model can predict tillering dynamics during tiller-increasing period (before jointing stage) of japonica rice. Although genotype specific calibrations are needed, the fact that one variety may be used for many years in a specific area, and production conditions in a specific area show slight inter-annual changes, and production practices are rather standardized implies a general applicability of our model.