Homoplasy and distribution of AFLP fragments: an analysis in silico of the genome of different species.

We carried out an in silico analysis of the complete genome sequences of 14 species, including eukaryotes, prokaryotes, and archaea, to investigate the proportion of amplified fragment length polymorphism bands that are homoplasious for the different species, as well as the distribution of fragment lengths. We investigated several possible reasons for the disagreement, previously observed in Arabidopsis thaliana, between the observed fragment length distribution and the null random sequence distribution, which occurs in the direction of a deficit of fragments of small length and an excess of those of large length with respect to the null distribution. We made the following findings: 1) The positive relationship previously found between the percentage of homoplasy and genome size is a direct consequence of the number of observed bands and the GC content. For the same number of observed bands, the percentage of homoplasy is independent of the genome size of the species. 2) The disagreement between the observed fragment length distribution and the null random sequence distribution observed in A. thaliana is a phenomenon that also occurs in other species. 3) This disagreement is due neither to the structure of the genomes in isochores nor the possible impact of indels in reducing the number of restriction sites, two hypotheses discussed in the literature. 4) Nor is the disagreement eliminated by using restriction enzymes with balanced motifs. 5) The discrepancy seems to be caused, rather, by the nonrandom distribution of restriction enzyme motifs.