Tassel Morphology in Zea mays: Novel Phenotyping Tools and Signatures of Selection

摘要

The male inflorescence of maize (Zea mays L.) is a branched structure atop the mature plant that is also known as the tassel. Tassel size and shape plays an important role in pollen quantity and dispersal, which can impact the reproductive success of an individual. Though the impact of tassel morphology on maize reproduction is generally accepted, less is known about genetic control of quantitative variation for tassel morphology. This thesis contains research projects which provide tools for genetic studies of tassels and which uncover evidence for selection by breeders on certain tassel morphologies. First, an image-based phenotyping tool for maize tassels was developed. Automated software measures two-dimensional images taken with a standard digital camera and produces estimates of tassel length, branch number, and weight that have correlations between 0.66 and 0.89 with manual measurements of the same traits. In addition to estimating traits that are traditionally measured by hand, the software also quantifies tassel characteristics that cannot easily be measured manually, including complexity, curvature, and compactness. Second, we used simulations to evaluate how different measures of an underlying phenotype, such as image-based and manual quantification, affect the outcome of genome-wide association studies (GWAS). We simulated traits controlled by 10, 100, or 1,000 causative variants that had differing heritability between 0.1 and 0.9. Using the area under the receiver operating curve, we quantified the success of GWAS in identifying the causative variants. Results show that higher number of causative variants and lower heritability decrease ability to detect true associations by GWAS. We also find that the difference in GWAS results between two traits of differing heritability is greater when the number of causative variants is high. Finally, we used phenotypic data generated by image analysis in conjunction with manual measurements to search for signals of selection for tassel size. We use GWAS to identify genomic regions associated with tassel traits in a diverse association panel and a nested association mapping panel. Those regions have enriched selection signals when comparing maize lines developed in the 1930s with more highly selected lines commercially developed in the late 1990s.