Fertilisation strategy and ground sensor measurements to optimise rice yield

Nitrogen (N) fertilisation is the main agronomic practice that affects rice yield and quality; similarly, its mismanagement can affect both economic and environmental aspects of crop production. Therefore, it is highly important to direct N fertilisation during the critical growth stages of rice development using vegetation indices (VIs). To this end, a two-year experiment was conducted in 2014 and 2015 in Castello d’Agogna (PV), northwest Italy. The study had three aims: i) establish the best N fertilisation management in temperate rice cropping systems, in terms of total N supply and splitting, to maximise crop yield and N apparent recovery (AR); ii) evaluate the capability of crop N status indicators (CNSIs) measured at panicle initiation stage (PI) to determine grain yield; iii) derive Nfertiliser_rate_at_PI=f(CNSI) from a field trial to attain specific yield goals. Results obtained for Centauro variety suggested that to maximise yield while avoiding AR reduction, a low dose of about 50 kg N ha−1 should be supplied during early growth, then increased at PI. In addition, the final topdressing fertilisation can compensate for any previous stage supply deficiency and can be determined from VI measurements. Findings also identified the normalised difference red edge (NDRE) index as the best VI to determine rice N status in specific agro-environmental conditions. SPAD and NDVI values measured with Rapid Scan can be used to determine N fertilisation at PI, although such measurements require correction through Sufficiency Indices (SIs) calculated as the ratio between VI measurements and VI values of a well-N fertilised plot. The trial also demonstrated that plots supplied with N amounts of 140 kg N ha−1 (pre-sowing and tillering stages combined) can serve as reference plots for SI calculation that allows to consider the effect of weather and soil variability on VI measurements. A notable exception to this finding was NDVI measured with GreenSeeker, which showed limited ability to assess rice N status under study environmental conditions. Indeed, both VI and the derived SI were influenced by seasonal and soil fertility conditions. Finally, a specific statistical method to derive calibration functions for variable rate application fertiliser spreaders from a suitable experiment was defined. These functions will establish the N amount to be supplied at PI related to the CNSI measure. For each CNSI, a specific slope of the calibration function is determined while the intercept is varied depending on the grain yield goal. The higher the acceptable reduction relative to the maximum obtainable yield, the lower the N supply required at PI.