Gene-based systems approach to simulate soybean growth and development and application to ideotype design in target environments

Crop yields must increase to satisfy an increasing food demand. Plant breeding and improved crop management will constitute the backbone for breaking productivity constraints. Rapid advances in molecular biology promise to radically change plant genetic improvement. However, we need methods to bridge the gap between genes and crop performance, to predict crop responses to environmental conditions and management, and to design predictable phenotypes.; Crop models, software programs that imitate plant growth and development, have the potential to become powerful genetic engineering tools. Paradoxically, model parameters that characterize genotypic differences are phenotypic in nature. If these can become functions of loci, we can establish a bridge between genetics, crop biology, and crop and environmental management. This dissertation develops, tests, and demonstrates an approach to tailor a crop model to the genetic makeup of the crop for ideotype design for target environments.; Using soybean as a model organism, a set of E loci that control reproductive development was studied using 48 near-isogenic lines. New functions were assigned to the E5 locus and other E loci that control reproductive duration and pod number determination. These experimental results were used to develop linear models to predict crop model parameters from E loci alleles bridging the gap between genetics and integrated crop physiology. For the first time, this kind of approach was tested for its ability to predict growth and development in commercial varieties. The gene-based model predicted 75% of the variance in the time to maturity and 54% of the yield variance in variety trials conducted in Illinois. Gene-based approaches can thus reduce or replace expensive and time-consuming experimentation for model parameterization. A phenotype reverse-engineering method was implemented by coupling the gene-based model to a simulated annealing optimization algorithm. The new method was used to design ideotypes for target environments in Argentina. The coupled model found ideotypes yielding at least 40% more than actual varieties grown in the region. Although more research is needed to fully parameterize the soybean model, it was shown that there is great potential for decreasing model parameterization requirements, and for designing ideotype for food production systems.