LAUSR

Abstract Rice (Oryza sativa L.) in the tropics is often produced in small landholdings, which can vary in farming practices that influence optimal management of fertilizer. A web-based decision support tool named Rice Crop Manager (RCM) was developed for the Philippines to handle variability in optimal fertilization by calculating field-specific nutrient management. RCM adapted principles developed earlier with site-specific nutrient management (SSNM). We explain the selection and calibration of calculations used by RCM to determine fertilizer N, P, and K requirements for irrigated and rainfed rice. A financial analysis will be reported in other manuscripts evaluating the performance of RCM. A yield-gain approach calculated nutrient rate to achieve a target yield, as a direct function of the estimated yield gain with added nutrient. Yield gains with added N, P, and K were estimated from 282 nutrient omission plot technique (NOPT) trials conducted across the Philippines. Each NOPT trial had a full fertilization plot and nutrient omission plots without added N (0-N), without added P (0-P), and without added K (0-K). Differences in yield between full fertilization and nutrient omission plots provided an estimate of yield gain with added nutrient as a function of a target yield with full fertilization. RCM calculated a unique target yield for each field in each growing season based on field-specific crop management and historical yields, which RCM collected through a pre-season interview of a farmer. RCM calculated fertilizer N rates with a yield-gain approach using mean yield gain estimated for a target yield from countrywide NOPT trials, a target agronomic efficiency of fertilizer N, and a field-specific target yield within the range of 3–9 Mg ha–1. RCM calculated fertilizer K rates with a combined yield-gain + nutrient-balance approach, which considered K inputs from crop residues and irrigation water. We conducted 299 measurements of K in irrigation water across the Philippines to develop a countrywide estimate of K input from irrigation water of 21 kg K ha–1 in the dry season and 8 kg K ha–1 in the wet season. RCM calculated fertilizer P rates using a nutrient-balance approach, which estimated field-specific P inputs from crop residue. It avoided depletion of soil P through use of higher P rates than a yield-gain approach. The algorithms and calibrations for RCM in the Philippines form a basis that can be improved as new information becomes available. The algorithms and procedures for calibration can also be used to develop and enhance nutrient management decision support tools for rice in other countries.