Wheat blast, caused by the fungus Magnaporthe oryzae Triticum pathotype (MoT), is a serious disease capable of causing severe losses, especially during warm and humid weather conditions. Although the pathogen attacks all aboveground parts, infection of the wheat spikes is of major concern. In this work we developed and evaluated a prediction model based on the analysis of historical epidemics and weather series in the northern Paraná state, Brazil (Apucarana, Maringá and Londrina) and available epidemiological knowledge. The disease and weather datasets (hourly scale) examined encompassed the 2001-2012 period. A specific database management application (agroDb) helped to visualize and identify patterns in weather variables during two major outbreaks (2004 and 2009). Specifically, uncommonly humid and warm weather for most locations during a 60-day period preceding wheat heading during years of major outbreaks were considered key drivers of inoculum build up and airborne spores from regional inoculum sources in the surroundings. An inoculum potential (IP) and a spore cloud (SPOR) variable were estimated from models adapted from literature to predict inoculum build-up and availability. A day favoring infection (DFI) was conditioned to rules relating temperature and relative humidity for the day derived from the epidemic analysis. Successful daily infection (INF) during a DFI was conditioned to IP > 30 and SPOR >0.4. To test the model, a wheat model simulated heading date for 10 planting dates, spaced 5 days apart, within a year, totaling 320 simulations. The model described well epidemic and non-epidemics conditions for the historical dataset, and was able to correctly predict epidemic (2015) and non-epidemic (2016) years not analyzed to build the model. An interactive risk-mapping tool that collects real-time weather data was developed for the target area to warn potential outbreaks.The system can be adapted to other regions where the disease is endemic or to asses the epidemic potential in regions where the disease is not present.

Dynamic crop simulation models are tools that predict plant phenotype grown in specific environments for genotypes using genotype-specific parameters (GSPs), often referred to as “genetic coefficients.” These GSPs are estimated using phenotypic observations and may not represent “true” genetic information. Instead, estimating GSPs requires experiments to measure phenotypic responses when new cultivars are released. The goal of this study was to evaluate a new approach that incorporates a dynamic gene-based module for simulating time-to-flowering for common bean (Phaseolus vulgaris L.) into an existing dynamic crop model. A multi-environment study that included 187 recombinant inbred lines (RILs) from a bi-parental bean family was conducted in 2011 and 2012 to measure the effects of quantitative trait loci (QTL), environment (E), and QTL×E interactions across five sites. A dynamic mixed linear model was modified in this study to create a dynamic module that was then integrated into the CSM-CROPGRO-Drybean model. This new hybrid crop model, with the gene-based flowering module replacing the original flowering component, requires allelic makeup of each genotype that is simulated and daily E data. The hybrid model was compared to the original CSM model using the same E data and previously estimated GSPs to simulate time-to-flower. The integrated gene-based module simulated days of first flower agreed closely with observed values (root mean square error of 2.73 days and model efficiency of 0.90) across the five locations and 187 genotypes. The hybrid model with its gene-based module also described most of the G, E and G×E effects on time-to-flower and was able to predict final yield and other outputs simulated by the original CSM. These results provide the first evidence that dynamic crop simulation models can be transformed into gene-based models by replacing an existing process module with a gene-based module for simulating the same process.

No abstract

Dynamic crop simulation models are tools that predict plant phenotype grown in specific environments for genotypes using genotype-specific parameters (GSPs), often referred to as "genetic coefficients." These GSPs are estimated using phenotypic observations and may not represent "true" genetic information. Instead, estimating GSPs requires experiments to measure phenotypic responses when new cultivars are released. The goal of this study was to evaluate a new approach that incorporates a dynamic gene-based module for simulating time-to-flowering for common bean (Phaseolus vulgaris L.) into an existing dynamic crop model. A multi-environment study conducted in 2011 and 2012 included 187 recombinant inbred lines (RILs) from a bi-parental bean family to measure the effects of quantitative trait loci (QTL), environment (E), and QTLxE interactions across five sites. The dynamic mixed linear model from Vallejos et al. (2020) was modified in this study to create a dynamic module that was then integrated into the CSM-CROPGRO-Drybean model. This new hybrid crop model, with the gene-based flowering module replacing the original flowering component, requires allelic makeup of each genotype being simulated and daily E data. The hybrid model was compared to the original CSM model using the same E data and previously estimated GSPs to simulate time-to-flower. The integrated gene-based module simulated days of first flower agreed closely with observed values (root mean square error of 2.73 days and model efficiency of 0.90) across the five locations and 187 genotypes. The hybrid model with its gene-based module also described most of the G, E and GxE effects on time-to-flower and was able to predict final yield and other outputs simulated by the original CSM. These results provide the first evidence that dynamic crop simulation models can be transformed into gene-based models by replacing an existing process module with a gene-based module for simulating the same process.