Table 5 Results of posterior predictive tests indicating the ability of different approaches to reduce compositional bias in mitochondrial amino acid data sets.
Approach | Remaining taxa | Z score | p value | Number of taxa with significantly deviating amino acid composition |
|---|---|---|---|---|
Original data set | 49 | 8.657 | 0.000 | 40 |
Exclusion of the 10 taxa with the most strongly differing amino acid composition | 39 | 7.308 | 0.000 | 32 |
Recoding using 9 minmax chi-squared bins | 49 | 8.690 | 0.003 | 38 |
Recoding using 6 minmax chi-squared bins | 49 | 7.196 | 0.005 | 21 |
Recoding using Dayhoff groups | 49 | 11.285 | 0.000 | 30 |