Applications of Linear Mixed Models for Screening of Transgenic Crops

The first requirement of crop improvement through genetic modification is the discovery of efficacious transgenes. Screening experiments with transgenic plants are a direct approach to this end. This thesis describes investigations that aimed at the improvement of these experiments and of the analysis of data they provide. The solutions to this aimneed to consider the constraints and specificities of transgenic plant material and the requirements of high throughput, of minimized costs andof short time lines. The statistical models need to accommodate the specific structure of a population of plants obtained after introducing a foreign gene in a plant genome. The analysis needs to allow inference about commercial potential of the transgene, despite being based on experiments with only non-commercializable plant material. A general introduction to screening of transgenes with the goal to improve crop yield is provided in order to explain the constraints and peculiarities of this application area. In the first steps of this process, phenotypes ofplants that received a transgene are observed and measured in experiments set up for that purpose. The analysis of these phenotyping experiments for transgenics differs from the analysis as it is usually applied in similar screening experiments in the context of traditional breeding. There are additional complexity and sources of variation and the inferencehas different aims. Transgenes are inserted in the original plant genome but cannot be directed to a specific spot in the genome. Expression levels of the transgene differ considerably among insertions, because somemay have been inserted in a non-transcribed part of the genome (and hence have no effect) or because they have disrupted a functional endogenous gene. Agronomically most interesting insertions may have an intermediate level of transgene expression. Therefore, observations need to be made on many insertion events and studied how their effect on the phenotypediffers. All off-spring originating from an insertion forms anevent. An elite event is a successful insertion of an active transgene.Many insertions need to be screened to detect such an elite event. Thisis expensive, hence optimization of the process is needed. The screening of events in a commercial crop is usually preceded by the screening ofgenes in an easily transformable test plant. This first phase of the research mainly aims at the discovery of which genes have a phenotypic effect. When an effec