Fig. 2. Simulation comparison of a marker-assisted (MAS) and a phenotypic (PS) selection strategy following the methods developed in Cooper and Podlich (2002). The four diagrams plot the difference between the population mean performances achieved for a quantitative trait at cycle 5 by the MAS and PS recurrent selection strategies for a large number of putative genetic architectures of a quantitative trait [normalized difference in response (MAS-PS)] against the complexity of the trait measured as an autocorrelation on a performance landscape from walks in genetic space (at the extremes the autocorrelation 1 represents the more simple additive genetic models and the autocorrelation 0 represents the more complex genetic models of the architecture of the trait). In the four subfigures, H = broad sense heritability in the base population; the percentage of QTL identified represents the percentage of the total QTL used in the MAS strategy. The large symbols represent the grand mean of the normalized difference in response between MAS and PS and the bars represent the standard deviations of the individual estimates of the normalized difference between MAS and PS.