Calibration of a crop growth model in APSIM for 15 publicly available corn hybrids in North America

Abstract

Application of crop growth models (CGMs) in plant breeding is limited by the large number of candidate cultivars that breeders work with and the large number of CGM parameters that affect cultivar performance. The objectives of this study were to (1) calibrate 15 publicly available maize hybrids in Agricultural Production Systems sIMulator and quantify prediction accuracy in modeling physiological trait differences (yield, biomass, phenology, etc.) among genotypes; (2) better understand minimum phenotypic data requirements for CGM cultivar calibration to inform breeding efforts; and (3) quantify simulated genotype by environment interactions (G × E) across years for five traits. We calibrated hybrids with two years of multi‐trait, temporal field measurements. The R2 of simulated versus observed phenotypes was 0.89 for grain yield and over 0.80 for half of all other simulated traits. Phenology parameters accounted for nearly half of the variability in grain yield. Average (across traits) normalized root mean square error was reduced from 35% to 30% with calibration based on phenological measurements and was reduced to 20% with inclusion of physiological and nitrogen‐related measurements such as radiation use efficiency and grain nitrogen. Long‐term simulations demonstrated distinct G × E among the hybrids which accounted for 2%–29% of the total genetic variation across traits. Parameter values derived in this work will provide insight regarding important physiological traits for further phenotyping, selection, and understanding of G × E. These calibrations are for publicly available hybrids, which are currently lacking.

From a point to a range of optimum estimates for maize plant density and nitrogen rate recommendations

“…Others in the literature have used crop modeling data to calculate net revenue response to plant densities and N rates (Jin et al., 2016; McNunn et al., 2019). Crop models have documented capacity to simulate grain yield response to N rate and plant density (Baum et al., 2023), but they are limited by the need to develop unique hybrid parameter values for the model (Winn et al., 2023). Combining the use of field experiments with crop modeling will enable the creation of a large database (with more years and more management factors) to develop future multifactor decision support tools for optimizing corn productivity in the United States and abroad.…”

Leveraging Data from the Genomes to Fields Initiative to Investigate G×E in Maize in North America

“…Consequently, the EC data we provide is defined at the year-location level. Phenotypic prediction accuracy could be further improved by optimizing crop model parameters for specific cultivar-year-location combinations (Winn et al 2023); however, as stated, if one does so, ECs will also be affected by genotypic effects making the interpretation of genomic models such as (3) difficult.…”

Accelerated leaf appearance and flowering in maize after four decades of commercial breeding

“…Typical applications of the empirical models are for prediction of the sixth leaf stage to apply in‐season fertilizer and anthesis date to apply fungicides (Abendroth et al., 2011). Furthermore, the changes in LAR parameters with the YOR have important implications for crop modeling as the rate of crop development governs several other processes in the model including root front velocity, dry matter partitioning, and target nitrogen concentrations (Battisti et al., 2017; Yakoub et al., 2017; Archontoulis et al., 2020; Winn et al., 2022). Therefore, for temporal applications of crop models, that is, explain historical crop yield increase in the US Corn Belt, cultivar parameters should account for the changes in LAR.…”

“…The overlap between the end of the vegetative phase and the beginning of the reproductive phase, as opposed to a clear transition, is currently not included in many maize crop growth models (Kimball et al, 2019). Our findings provide strong evidence that this needs to be improved because, around the transition period from the vegetative to the reproductive phase, crop models determine final kernel numbers, which has a strong impact on grain yield prediction (Schussler & Westgate, 1991;Winn et al, 2022). The AgMaize model (Tollenaar et al, 2018) is among the few to explicitly simulate ASI, whereas other maize models simulate a flag leaf (Soufizadeh et al, 2018).…”