Integrating High Throughput Phenotyping Genomic Selection And Spatial Analysis For Plant Breeding And Management

Recent advances in high-throughput phenotyping, genomics, and precision agriculture have provided plant breeders and farmers with a wealth of information on the growth and development of crop plants. Methods for effectively leveraging these data resources are needed in order to drive genetic gain in breeding programs and to increase efficiency in farming systems. Three novel approaches for the development and management of high yielding, adapted crop varieties are presented. First, aerial hyperspectral reflectance phenotypes of bread wheat (Triticum aestivum L.) were used to develop relationship matrices for the prediction of grain yield within and across environments with genomic selection. Multi-kernel models combining marker/pedigree information with hyperspectral reflectance phenotypes gave the highest accuracies overall; however, improvements in accuracy over single-kernel marker- and pedigree-based models were reduced when correcting for days to heading. Second, aerial phenotypes collected on small, unreplicated plots representing the seed limited stage of wheat breeding programs were evaluated for their potential use as selection criteria for improving grain yield. The aerial phenotypes were shown to be heritable and positively correlated with grain yield measurements evaluated in replicated yield trials. Results also suggest that selection on aerial phenotypes at the seed-limited stage would cause a directional response in phenology due to confounding of those traits. Lastly, on-farm trials were conducted in collaboration with the New York Corn and Soybean Growers Association to identify optimal planting rates for corn (Zea mays L.) and soybean (Glycine max L.) given the underlying spatial variability of the soil and topographical characteristics of the fields. A random forest regression-based approach was created to develop variable rate planting designs for maximizing yields.