BOOTSTRAP ANALYSIS OF PHYLOGENETIC TREES DERIVED FROM DNA HYBRIDIZATION DISTANCES

Genetic distances measured by DNA-DNA hybridization experiments are subject to inherent levels of measurement imprecision that may limit the resolution of estimated phylogenetic trees. This type of error is reflected by the variance of replicate pairwise comparisons of OTUs within cells of the distance matrix. We propose an application of the bootstrap method to gauge the effects of such imprecision in phylogenetic reconstruction. Pseudoreplicate distance matrices are generated by resampling the original data with replacement, and phylogeny estimates are made for each pseudoreplicate matrix using a least-squares pairwise algorithm; the process is repeated many times to produce a distribution of possible trees. The frequency of occurrence of specific clades in this distribution measures the resolving power of the original measured distances. This procedure does not assay the effects of inaccuracy (bias) in measured distances, but, in conjunction with a jackknife approach for resampling OTUs, provides a means for quantitative assessment of the robustness of trees estimated from DNA hybridization distances. © 1990 Society of Systematic Zoology.