Results of the tests for substitution saturation.

Proportions of invariant sites have been taken from maximum likelihood estimators, since DAMBE crashed during their computation: COI-5P, 1 and 2 codon positions = 0.283; COI-5P, 3 codon positions = 0.000; 5′-partial LSU rDNA, Chroomonas clade = 0.652; 5′-partial nuclear LSU rDNA, genus Cryptomonas = 0.627; 5′-partial LSU rDNA, combined data sets, including Rhodomonas sp. strain CCAP 978/13; Rhodomonas sp. strain M1480 (acc. no. AM396399) and Storeatula sp. strain CCMP1868 (acc. no. FJ973366) = 0.594. , index of substitution saturation as inferred from the mean value of sitewise entropies () divided by global entropy of nucleotides (); , saturation threshold of for symmetrical trees; , saturation threshold for asymmetrical trees. If and difference between and is significant (), saturation probably is low in the data set. If and the difference between and is significant, the data set is useless for phylogenetic analyses. If and the difference between and is not significant, the quality of the data set for phylogenetic analyses is poor. Resampling of OTUs is required for data sets larger than 32 taxa, because no critical values for larger data sets were available from simulations [67][66]. For the 5′-partial LSU rDNA data sets, the values for resampling of 32 taxa are shown.