A recent simulation study argued for accumulating many more characters from slow evolving markers, and recommends 10,000-20,000 characters for Lamiales [ 40].
We recognize, however, that some contigs may contain paralogs, especially for slow-evolving gene families as discussed in other reports on plant EST clustering [ 29, 30].
In light of this information, the molecular timescale estimates we have presented that seem most reasonable for the butterflies are the lower bounds of the 95% confidence intervals for the slow evolving data set (Table 3).
The best fit branch model for each region of ago-1 and for dcr-1 was the two-ratio model, with one ratio of non-synonymous to synonymous rates (ω=dN/dS) for the fast evolving copies (− 1b copies) and a different ratio for the slow evolving copies (ago-1a or dcr-1/1a) copies (Table 3).
In the presence of misleading signal derived from homoplasy or biases at fast evolving sites, we might expect support for correct basal splits to increase as signal from slower evolving sites begins to predominate.
In the "two-ratio" model for ago-1, the ago-1b sequences were set as fast evolving copies and the ago-1a sequences as slow evolving copies.
For dcr-1, the dcr-1b sequences were set as fast evolving copies and the rest of aphid dcr-1 sequences as slow evolving copies (including the dcr-1a copies of Acyrthosiphon).
We compared the performance of both methods for fast versus slow evolving gene families (see Materials and Methods).
Another way to identify orthologous regions is to compare non-coding regions among species via similarity (e.g., BLAST) searches, but this approach is also likely to enrich for slow-evolving regions under selective constraint.
Slow evolving genes are sometimes more informative for phylogeny construction, since fast evolving genes might cause problems such as long branch attraction [ 28, 32].
