a  Based on alignment with reconstructed transcripts (e-value < 1e-5).
b  Extracted from Morais do Amaral et al. 2012.