The second method used to assess mutation accumulation in the various samples was through AFLP analysis. Here, the experiment simply didn't work but the authors analyzed their failed datasets and produced a number of meaningless figures and invalid conclusions. How do I know that the experiment didn't work? Because there was ZERO relationship between a culture's genotype at the start of the experiment and it's genotype at the end of the experiment - EVEN FOR THE CONTROLS! So, in the previous SSR experiment, the authors told us that the controls showed ZERO variation at the SSR loci, yet by AFLP analysis, they were COMPLETELY DIFFERENT! In other words, the genotype of every isolate analyzed changed to being completely unrecognizable simply by culturing for 12 generations on complete medium. Clearly, this experiment simply failed causing the AFLP profiles to essentially appear random. This is clearly evident in the neighbor joining trees shown in Figure 5, because no two isolates look remotely similar (i.e. huge intervening branch lengths). It appears that the bifurcation in the NJ tree is an artifact, due to certain isolates producing more AFLP fragments in the experiment. As a result, all the principle component studies, etc. are meaningless. True conclusion: data are insufficient to test if sublethal fungicides affect mutational rates in Sclerotinia.