Genotype and environment impacts on Canada western spring wheat bread-making quality and development of weather-based prediction models

A study was conducted to quantify weather conditions at specific growth stages of Canadian Western Spring wheat (Triticum aestivum) and relate those growing conditions to variations in wheat grade and quality characteristics and to develop pre-harvest prediction models for wheat quality using weather input data. The Canada Western Red Spring (CWRS) genotypes AC Barrie, Superb, Elsa, Neepawa, Canada Prairie Spring-White genotype (CPS-white) Vista and Canada Western White Spring (CWWS) genotype Snowbird were grown in five locations across the Canadian prairies during the 2003 and 2004 growing seasons, which provided a wide range growing conditions. The experimental layout at each location was a randomized complete block design with three replicates. Intensive weather data was collected during the growing season at each location and used to calculate accumulated heat stress, useful heat, moisture demand, moisture supply, moisture use and moisture stress variables for numerous crop development stages. Crop development was observed on a regular basis at each location in order to partition the growing season into several development stages. Grain samples from each plot were subjected to full visual analysis and official grading by the Canadian Grain Commission and were milled into flour using a Buhler Experimental flour mill at the University of Manitoba. Flour samples underwent an extensive analysis of flour, dough, and bread making quality. ANOVA indicated that genotype, environment and their interactions had significant effects on most quality parameters tested. Environmental contribution to wheat quality variance was considerably larger (62 to 89%) than the variance contribution of either genotype (2 to 26%) or GxE interaction (2 to 16%). Regression analysis was completed in order to determine relationships between growing season weather and wheat quality.;The development periods of planting to jointing and anthesis to soft dough were the stages most frequently exhibiting the highest correlation to wheat quality indicating weather needs to be monitored during the entire growing season to accurately predict quality. The level of variance in wheat quality explained by weather variables was improved when more detailed phenological stages were considered.;Grain quality forecast models were validated using 2005 weather and crop data. Prediction models developed from the 2003 and 2004 data required modification in order to accurately and consistently predict the grain properties in 2005. Generally, the best predictive models were developed by using data from a group of genotypes which responded similarly to the environment. Yield was predicted to within 120 to 530 kg/ha, on average, between the three sites using the modified model. The standard error of prediction (SEP) for yield improved from 927 using the original model to 288 using the modified model. Test weight was forecast to within 2.2 to 3.0 kg/hL using the modified model and the original SEP of 6.15 improved to 1.46 using a modified equation. TKW was predicted between 0.4 and 3 g at each location using the modified regression equation. The original TKW model had an SEP value of 13.19, which improved to 0.91 using the best modified model. Protein content results were more varied, with protein content in Regina predicted to within 0.6%, while at the other two test sites, predicted grain protein content was more than 1.5% from the actual. SEP results reflected protein content variability as SEP values did not improve using modified models.;Using the weather and crop development stage information, significant regression equations with high regression coefficients were developed for most quality parameters using just a single independent weather variable. Moisture related variables explained the majority of the variation for all the grain properties except yield as well as for most of the flour properties. The farinograph measured dough parameters, except Farinograph stability, were driven by water related variables and the mixograph measured dough properties by useful heat variables and water stress variables. The bread properties were found to be best predicted using useful heat and heat stress variables. Multiple regression equations with even higher R2 values were developed using three complex weather variables, leading to the opportunity to predict wheat quality 2-5 weeks prior to harvest. R2 values ranged from 0.29 to 0.95, with the grain and dough properties producing the strongest forecast models. For 13 of the 27 quality properties tested, R 2 values were above 0.80. Equally strong prediction models were developed utilizing basic weather variables which could be obtained from weather stations monitoring only daily maximum and minimum air temperature and precipitation. R2 values for these models ranged from 0.22 to 0.95.