Linear models to estimate relative fitness: a reply.

Sun and Zeng (6) proposed a linear model to estimate the relative fitness of plant pathogens in cultivar mixtures. The linear model was considered “wrong” by Ostergard and Shaw, supported critically by a proof of numerical validation (5). Because Sun and Zeng’s model is the first to estimate relative fitness of pathogens in a cultivar mixture for practical use, it is worthwhile to examine Ostergard and Shaw’s critique carefully. We have found that Ostergard and Shaw’s numerical approach to evaluate pathogen fitness is biologically and mathematically sound. However, they have incorrectly applied their approach to evaluate Sun and Zeng’s model, which resulted in an erroneous conclusion. This letter shows that Ostergard and Shaw improperly cross-validated two systems that are defined over different ranges of values, a flaw not uncommon in plant pathology literature of quantitative analysis. We also propose an alternative method to calculate overall fitness value in a cultivar mixture. On Ostergard and Shaw’s validation. Sun and Zeng (6) described relative fitness by using a simple system of two pathogen genotypes (they called races), 1 and 2. The relative fitness, f, is calculated using frequencies of the genotypes at the beginning of a season (p10 and p20) and at the end of the season (p1 and p2) as