Premium
Estimating Genetic Coefficients for the CSM‐CERES‐Maize Model in North Carolina Environments
Author(s) -
Yang Zhengyu,
Wilkerson Gail G.,
Buol Gregory S.,
Bowman Daryl T.,
Heiniger Ronnie W.
Publication year - 2009
Publication title -
agronomy journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.752
H-Index - 131
eISSN - 1435-0645
pISSN - 0002-1962
DOI - 10.2134/agronj2008.0234x
Subject(s) - calibration , anthesis , grain yield , mean squared error , environmental science , mathematics , yield (engineering) , statistics , simulation modeling , agronomy , biology , cultivar , materials science , mathematical economics , metallurgy
CSM‐CERES‐Maize has been extensively used worldwide to simulate corn growth and grain production, but has not been evaluated for use in North Carolina. The objectives of this study were to calibrate CSM‐CERES‐Maize soil parameters and genetic coefficients using official variety trial data, evaluate model performance in North Carolina, and determine the suitability of the fitting technique using variety trial data for model calibration. The study used yield data for 53 maize genotypes collected from multiple locations over a 10‐yr period. A stepwise calibration procedure was utilized: (i) two genetic coefficients which determine anthesis and physiological maturity dates were adjusted based on growing degree day requirements for each hybrid; and (ii) plant available soil water and rooting profile were adjusted iteratively with two other genetic coefficients affecting yield. Cross validation was used to evaluate the suitability of this approach for estimating soil and genetic coefficients. The root mean squared errors of prediction (RMSEPs) were similar to fitting errors. Results indicate that CSM‐CERES‐Maize can be used in North Carolina to simulate corn growth under nonlimiting N conditions and variety trial data can be used for estimating genetic coefficients. Hybrid average simulated yields matched measured yields well across a wide range of environments, and simulated hybrid yield rankings were in close agreement with rankings based on measured yields. Data from several site‐years could not be used in fitting genetic coefficients due to large root mean squared errors. In some cases, this could be attributed to a weather event, such as a late‐season hurricane.