Session 1.7: Wheat Model Intercomparison
For a complete list of all of the workshop abstracts click here (PDF).
If you are presenting a poster, please view the Poster Specifications here
Session 1.7: Oral Presentation
Title: Improved modelling of wheat processes through inter-comparison of multiple models
Authors: Enli Wang1*, P. Martre2,3*, S. Asseng4, F. Ewert5, Z. Zhao1, A. Maiorano2,3,
and the AgMIP-Wheat team
Abstract: While the multi-model ensemble modeling approach is useful to quantify prediction uncertainty in crop simulations, the approach itself does not necessarily lead to improvement in process understanding. We extend the model inter-comparison to investigate how the uncertainties in simulation results arise from process-level algorithms and parameterization in the models. We systematically compared 29 physiologically based wheat models in terms of how the key temperature-responsive physiological processes are simulated. We categorized the temperature response equations in the models into four types based on their shapes. To demonstrate the impact of the different temperature equations on simulated phenology, total above ground biomass and grain yield, we implemented the four types of temperature responses in the APSIM and SiriusQuality models and tested the modified models against the USDA ‘Hot Serial Cereal’ (HSC) field experiment. In addition, the uncertainty in simulation results from the two models caused by various temperature response functions was compared to those generated from the 29 models. Our analysis revealed contrasting temperature response functions for the same physiological process among different models. The range of simulated grain yield caused by variations of temperature response functions in APSIM and SiriusQuality was on average 52% and 64% of the uncertainty of the whole ensemble of 29 models, respectively. We further developed improved general temperature response functions for key developmental and growth processes of wheat. Implementation of these temperature functions in the APSIM and SiriusQuality model led to improved simulations of wheat yield against the HSC data across a wide temperature range.
Session 1.7: Oral Presentation
Title: Prediction of the rate of development and anthesis date using QTL-based parameters of an ecophysiological model for durum wheat
Authors: Pierre Martre1, R. Motzo2, G. Sanna2, A. M. Mastrangelo3, P. De Vita3, and
F. Giunta2 1 UMR LEPSE, INRA, Monptellier SupAgro, France; 2 Unit of Agronomy, Field Crops and Genetics, Department of Agriculture, University of Sassari, Italy; 3 CREA Cereal Research Centre, Italy
Abstract: Breeding for a fine tuning of crop development to target population of environments is an avenue for future increases in grain yield and adaptation to climate change. The use of ecophysiological modeling has been proposed to get insights into how genotype-by environment interactions come about. However, models cannot account for the genetic basis of differences in response to the environment unless model parameters are linked with easily measurable physiological traits and known QTL or genes. Here, vernalized and non-vernalized plants of a population of recombinant inbreed lines (RILs) of durum wheat were grown under long and short days. Measured final leaf number and anthesis date data were used to calibrate five genotypic parameters of the SiriusQuality wheat model. A QTL analysis of the genotypic parameters was performed. Several QTL co-localized with previously identified QTL for earliness per se, cold requirement, and photoperiodic sensitivity. The performance of the model using either the original parameters or the QTL based parameters was validated in three field experiments for the whole population of RILs and in more than nine year/sowing/location combinations for the two parents, which were not used in the QTL analysis. The QTL-based model predicted the final leaf number, the date of flag leaf ligule appearance, and the anthesis data for the validation data set with an error of 0.5 leaves, 4.2 days and 2.5 days, respectively. These information will allow simulating the impact of genetic recombination on crop development under new environmental conditions and will help breeders identifying genetic makers to fine tune the development of new cultivars to target environments.
Session 1.7: Oral Presentation
Title: Designing Wheat ideotypes for a changing climate
Authors: Mikhail A. Semenov and P. Stratonovitch
Computational and Systems Biology Department, Rothamsted Research, Harpenden, UK
Abstract: Global warming is predicted to increase adverse weather events that are considered a major threat for wheat production in Europe. To meet increasing demands for wheat, new wheat cultivars with improved yield potential and resilience to climate change will be required, putting severe pressure on breeders who must select for uncertain future. Crop
modelling is a powerful tool to identify key traits for improvement and to quantify potential threats to wheat. Moreover, wheat ideotypes, optimised for a wide range of future climates and target environments, can be designed and tested in silico using a wheat simulation model. We used Sirius, a process-based model for wheat, to estimate yield potential of wheat ideotypes optimized for future climatic conditions in Europe as predicted by global climate models from the CMIP5 ensemble. Substantial increase in yield potential could be achieved through optimal phenology and extending grain filling and thereby improve resource capture and partitioning. However, the model predicted an increase in frequency of heat stress around flowering. Controlled environment experiments showed the detrimental effects of heat and drought at booting and flowering on grain numbers and potential grain size. The use of early maturing cultivars in areas of Europe with hotter and drier summers helps to escape from excessive heat and drought stress during the reproductive period, but results in lower yields. The refined Sirius wheat model, that incorporates responses to heat and drought stress around flowering, showed that yield potential and yield stability would be substantially affected for wheat ideotypes sensitive to these stresses. Therefore, to increase yield potential and resilience to climate change, increased tolerance to heat and drought stress during reproductive development should remain a priority for the genetic improvement of wheat.
Session 1.7: Oral Presentation
Title: Model improvements reduce the
uncertainty of wheat crop model ensembles under heat stress
Authors: Andrea Maiorano1, P. Martre1, S. Asseng2, F. Ewert3, D. Wallach4, and the AgMIP Wheat Team 1 INRA-SupAGRO, France, 2 University of Florida, USA, 3 University of Bonn, Germany, 4 INRA, France
Abstract: Model improvements can reduce uncertainty of climate impact assessments and as a consequence reduce the number of models required for an acceptable level of simulation uncertainty. Here 15 wheat crop models were improved for the simulation of heat stress impacts and the effect on multi-model ensemble (MME) performances and
predictive skills were investigated. The models were improved through re-parameterization or by the incorporation or modification of heat stress effects on phenological development and/or growth processes. Field data from the USDA ‘Hot Serial Cereal’ (HSC) experiment and the ‘International Heat Stress Genotype Experiment’ (IHSGE) coordinated by CIMMYT were used to calibrate and evaluate the improved models, respectively. The results show that model improvements decreased the variation of simulated grain yields on average by 26% in the independent evaluation dataset for crops grown in mean seasonal temperatures > 24°C. The mean squared error for grain yield of the model population decreased by 37%. The prediction skills of the model population increased by 47% due to a 26% reduction in the model population uncertainty range. The latter improvement was mostly due to a decrease in MME variance. The number of models required for MME impact assessments was halved, from 15 with the unimproved original models to 8 with the improved models. We conclude that model improvements using field-based experimental datasets can increase the simulation and predictive skills of MME and reduce the number of models required for practical impact assessments.