A methodology for estimating phenological parameters of rice cultivars utilizing data from common variety trials

Abstract

Crop phenology models play a pivotal role in predicting yields under climate change. Cultivar-specific model parameters are essential for accurate prediction, but their estimation generally requires elaborate and laborious experiments, and such parameter sets have therefore been available only for a small number of cultivars. We propose methodology for estimating phenological parameters, combining a stochastic parameter estimation method (genetic algorithm) with the use of a database comprising 30 years of records from variety trials conducted at experimental stations across Japan. Optimal parameter sets were selected based on the results of cross-validation tests. This methodology allowed us to estimate phenological parameters objectively. We estimated phenological parameters for the 10 leading cultivars currently planted in Japan, and showed that these parameters reflect the cultivars' sensitivity to temperature and/or photoperiod. The proposed methodology can be used to provide quantitative evaluations of the environmental responses of rice cultivars, without relying on elaborate and laborious experiments, and substantially improves the efficiency of phenological trait phenotyping.

Large-scale evaluation of the effects of adaptation to climate change by shifting transplanting date on rice production and quality in Japan

“…Although some cultivars had relatively large mean bias and RMSE values, such as Kirara 397 4.70 and 6.02 and Aichinokaori 3.40 and 4.99 , their percentage of the observed duration from transplanting to heading was less than 10 , and they were within 7 days, which is the interval of shifting transplanting date in this study. The RMSEs calculated in this study are comparable to the results of Fukui et al 2015 , which ranged from 2.73 to 4.90 days average, 3.4 days and were calculated from another observed dataset. The biases tended to be negative, as was the case in the study by Fukui et al 2017 , in which new phenological models using water temperature as input instead of air temperature during the early growing period were validated using the CSCS data, and the biases were estimated to be negative even if the RMSEs were small.…”

“…However, that formula neglected the decrease of DVR at extremely high temperatures with the aim of simplifying the function Horie and Nakagawa, 1990 , detracting from the confidence in the simulation at high temperatures. Considering the objectives of this study, we adopted an alternative formula given by Nakagawa et al 2005 , in which the decrease in DVR at extremely high temperatures is represented by a β-function Yin et al, 1997 , as introduced in the latest version of the H / H model Yoshida et al, 2015 estimated parameters of the DVR formula for 15 major rice cultivars, which accounted for >80 of the entire rice cultivation area in Japan in 2005MAFF, 2006b , by using data collected during field experiments conducted across Japan and a stochastic genetic algorithm for optimization of parameters Fukui et al, 2015 . This methodology enabled to estimate parameters objectively without biases caused by the regional inhomogeneous distribution in the sample data.…”

“…Because Fukui et al 2015 andYoshida et al 2015 validated the performance of the H / H model for the estimation of rice phenology and yield, respectively, we performed incremental validation procedures to complement the insufficient parts of those studies. To evaluate the validity of applying the H / H model nationwide, we compared the simulated heading date and yield with observed data for each cultivar at a sufficient number of points across Japan where the statistics of rice productivity are available.…”

“…To assess the impact of projected CO 2 and temperature change on estimated yield in the mid and late 21st century on the basis of standard transplanting date no adaptation , we averaged yields for the periods 1981 -2000 "baseline period" , 2031 -2050 "mid period" , and 2081 -2100 "late period" in each cell. We aggregated the yields to the nationwide production levels in the mid and late periods, all of which are expressed as the percentage of that of the baseline Table 3 in Fukui et al (2015). Values in italic (last five cultivars) denote that the RMSE was based on the sampling datasets that were used for parameterization, oweing to insufficient samples.…”

“…As the impact assessment model, we used the Hasegawa -Horie model hereafter, "H / H model" , a process-based rice simulation model for estimating rice phenological development and yield developed by Hasegawa and Horie 1997 and recently modified by Fukui et al 2015 andYoshida et al 2015 . As for the quality of rice, unfortunately there is no appropriate model for quantitatively estimating the quality degradation associated with climatic variability at a national scale. Therefore, we adopted a heat stress index, which is related to rice quality degradation due to high temperature during the early reproductive stage; measured in heat degree-days, the index reflects the deviation in daily average air temperature from a base of 26 C during the 20 days after the heading date of rice Nagahata et al, 2006;Ishigooka et al, 2011 .…”