Genetic analysis and evaluation of tall fescue for low maintenance applications

Tall fescue [Festuca arundinacea (Schreb.)] is a cool-season turfgrass species that has shown great potential for low-maintenance turfgrass applications. This dissertation sought to explore and demonstrate the applications of statistical models in the genetic analysis and cultivar development of tall fescue. Specific attention was placed on two low-maintenance traits that the Rutgers turfgrass breeding program has been focusing on, improvement of drought tolerance using the rainout shelter and resistance to red thread disease caused by Laetisaria fuciformis (Berk.) Burds. Rainout shelters have been widely used in the breeding of tall fescue for improved drought tolerance. Persistence of green coloration of leaves during drought is one of the crucial traits with noticeable variations for selection. In this project, we studied two consecutive generations of tall fescue evaluated in rainout shelter trials with different experimental designs. Bayesian mixed linear models were applied to collected datasets. Variance components, narrow-sense heritability (h^2), and prediction accuracy of estimated breeding value (EBV) were estimated. The theoretical foundations of genetic analysis and application to the breeding of tall fescue were also discussed. We first reported h^2 for green persistence of tall fescue in rainout shelter selection. Mean heritability from parental generation was 0.18 with a 95% highest density interval (HDI) of (0.04,0.51), while that from progeny generation was 0.08 with a 95% HDI of (0.01,0.23). Though significantly greater than zero, both estimates were low, indicating a large proportion of non-genetic variance. Given the heritability estimate and experimental design in the progeny generation, the prediction accuracy for EBV with different selection methods was calculated. Selection methods ranked by mean prediction accuracy from the highest to the lowest are parental selection > family selection > mass selection. Given the heritability estimations, the theoretical prediction accuracy for these selection methods was also calculated. Extra attention was paid to stratified mass selection. Under the stratified mass selection method, our work demonstrated the application of best linear unbiased prediction(BLUP), A-BLUP, and G-BLUP in the breeding tall fescue, and illustrated how prediction accuracy could be further improved by increasing the number of blocks or/and the implementation of A-BLUP and G-BLUP. Data analysis in the studies has so far focused on continuous data. However, it is not uncommon to see discrete data in the breeding of tall fescue. The last chapter of this dissertation provided an example of an analysis of binary disease incidence data. The study investigated binary red thread disease incidence in tall fescue populations evaluated in two locations over multiple years, highlighted the importance of specific selection effort against red thread disease in tall fescue, and estimated heritability of disease incidence with two different experimental designs. Narrow-sense heritability (0.52,0.74 and 0.48) estimated from different experimental designs are all in the range of moderate to high, supporting the idea that additive genetic variance accounts for large phenotypic variance in red thread incidence in tall fescue populations. Hence, disease incidence of red thread in tall fescue can be effectively reduced through selection and breeding. This is also the first report documenting the efficacy of family selection in reducing red thread incidence in tall fescue.Ph.D.Includes bibliographical reference