Re-evaluating the precision of yield targets set by MOET app v.2.0

ABSTRACT

This paper presents the evaluation results of the yield target setting precision of the revised MOET App (v.2.0), following the inclusion of the correction factors generated from rice biomass correlations between MOET and nutrient omission plot setups obtained from 2017 to 2018. The project started with trainings on MOET kit and MOET App use for the seed production personnel across PhilRice stations in Nueva Ecija, Negros, Bukidnon and Agusan in 2018 DS. Included in the trainings were the establishment of MOET kit tests and generation of variety- and site-specific recommendations via the MOET App for several nationally or regionally recommended varieties (NSIC Rc 122, 160, 216, 218, 222, 238, 286, 300, 358, 402, 436, 440, 442, 480, PSB Rc18 and PSB Rc82) that each PhilRice station intended to produce in the succeeding 4 cropping periods from 2019 to 2020. Relative yield advantages and economic benefits from using the MOET App fertilizer recommendations over PhilRice' current fertilizer management in seed production per station were monitored every cropping while the precision evaluation of yield target setting was done after the last cropping of 2020 WS. In 2019, relative yield advantages averaged 0.43t ha-1 in DS and 0.25t ha-1 in WS. In 2020 DS, an average relative yield advantage of 0.63t ha-1 was obtained across stations and 0.93t ha-1 in 2020 WS in Nueva Ecija only due to travel restrictions brought about by the COVID-19 pandemic. Economic benefits of using the MOET App showed an average of 0.50t ha-1 and 0.65t ha-1 yield increase over the seed production units' fertilizer management in DS and WS, respectively. While savings in fertilizer cost were better realized during the WS at an average of Php 4,126.34 ha-1 season-1 across stations. Results of the precision evaluation of the yield target showed marked improvements with a 95.24% probability of achieving 17% higher grain yields than the target set by MOET App v.2.0. However, the overall normalized Root Mean Square Error (nRMSE) of 38.14% exceeded the range for a fairly acceptable fit with the model due to large gaps between target and actual yields obtained from DS field trials.