How Much Data are Needed to Resolve a Difficult Phylogeny? Case Study in Lamiales

Reconstructing phylogeny is a crucial target of contemporary biology, now commonly approached through computerized analysis of genetic sequence data. In angiosperms, despite recent progress at the ordinal level, many relationships between families remain unclear. Here we take a case study from Lamiales, an angiosperm order in which interfamilial relationships have so far proved particularly problematic. We examine the effect of changing one factor-the quantity of sequence data analyzed-on phylogeny reconstruction in this group. We use simulation to estimate a priori the sequence data that would be needed to resolve an accurate, supported phylogeny of Lamiales. We investigate the effect of increasing the length of sequence data analyzed, the rate of substitution in the sequences used, and of combining gene partitions. This method could be a valuable technique for planning systematic investigations in other problematic groups. Our results suggest that increasing sequence length is a better way to improve support, resolution, and accuracy than employing sequences with a faster substitution rate. Indeed, the latter may in some cases have detrimental effects on phylogeny reconstruction. Further molecular sequencing-of at least 10,000 bp-should result in a fully resolved and supported phylogeny of Lamiales, but at present the problematic aspects of this tree model remain.