Abstract
Genetic variation is the driving force of evolution and as such is of central interest for biologists. However, inadequate discrimination of errors from true genetic variation could lead to incorrect estimates of gene copy number, population genetic parameters, phylogenetic relationships and the deposition of gene and protein sequences in databases that are not actually present in any organism. Misincorporation errors in multi-template PCR cloning methods, still commonly used for obtaining novel gene sequences in non-model species, are difficult to detect, as no previous information may be available about the number of expected copies of genes belonging to multi-gene families. However, studies employing these techniques rarely describe in any great detail how errors arising in the amplification process were detected and accounted for. Here, we estimated the rate of base misincorporation of a widely-used PCR-cloning method, using a single copy mitochondrial gene from a single individual to minimise variation in the template DNA, as 1.62×10(-3) errors per site, or 9.26×10(-5) per site per duplication. The distribution of errors among sequences closely matched that predicted by a binomial distribution function. The empirically estimated error rate was applied to data, obtained using the same methods, from the Phospholipase A(2) toxin family from the pitviper Ovophis monticola. The distribution of differences detected closely matched the expected distribution of errors and we conclude that, when undertaking gene discovery or assessment of genetic diversity using this error-prone method, it will be informative to empirically determine the rate of base misincorporation.