Fig. 6.

SWSC-EN split UCEs into meaningful partitions. Comparisons between the AICc scores of empirical data sets (circle) and 25 permuted data sets (triangles) of partitioning schemes derived from the SWSC-EN and optimized in PartitionFinder 2. For each of the randomizations, the sites within each UCE were shuffled and the new alignments were used with the original data block definitions from the un-shuffled data as input to PartitionFinder 2. Each panel includes a data set and 25 AICc scores of partitioning schemes optimized in PartitionFinder 2. The AICc scores suggest that the SWSC-EN method improves model-fit and parameter estimates not because it searches a larger space of possible partitioning schemes, but primarily because splitting each UCE into three data blocks is uncovering biological differences in rates and/or patterns of molecular evolution within each UCE.