Effects of stripe rust on the evolution of genetically diverse wheat populations.

Eighteen populations, composed of four wheat (Triticum aestivum) varieties that were originally mixed together at equal frequencies, were grown for one-to-three generations at two locations. In addition, pure stands of the four varieties were grown in each year. Populations were either exposed to two stripe rust (Puccinia striiformis) races, protected from stripe rust, or exposed to alternating years of diseased and disease-free conditions. Regression of the logit of a variety's frequency versus generation number was used to calculate the relative fitness of each variety in each population. These analyses suggest that the relative fitnesses of the wheat varieties were affected by disease and geographic location and were constant over time. However, frequency-changes of varieties in the mixtures were negatively correlated with their planting frequencies (0.0001 < P < 0.085 in 14 out of 16 cases), suggesting that fitnesses were frequency-dependent in both the presence and absence of disease. We hypothesize that failure to detect frequency-dependence of fitness in the logit analyses was due to a limited number of generations and a limited range of initial variety frequencies. This is supported by data from longer-term studies in the literature that provide evidence for frequency-dependence of fitness in plant mixtures. Analyses of currently available field data suggest that stable equilibria may be a more likely outcome for mixtures of varieties that are more closely related and/or more uniformly adapted to the environment in which they are grown.