Table 1. Comparison of bootstrap percentages of phylogenetic estimates and data set manipulations for selected taxonomic groups.^a .

		Edafopoda	Altocrust-acea	Vericrust-acea	Multicrust-acea	Miracrust-acea	Xenocarida
phylogenetic estimates
GTR+I+G	degen1	92	93	86	100	94	93
JTT	20AA	64	20	24	52	16	55
	21AA	89	66	64	90	66	87
	Δ	+25	+46	+40	+38	+50	+32
GTR	20AA	59	16	13	37	10	39
	21AA	80	58	40	74	45	70
	Δ	+21	+41	+27	+37	+35	+31
ECM	20AA	70	6	15	43	8	27
	21AA	87	52	54	86	42	64
	Δ	+17	+46	+39	+43	+34	36
mimicking amino acids with nucleotides
Ser1 to Ser2	degen1	54	57	40	84	42	59
	Δ	−38	−36	−46	−16	−52	−34
Ser2 to Ser1	degen1	66	63	47	90	41	58
	Δ	−26	−30	−39	−10	−53	−35
identifying important serines
no Ser1, no Ser2	degen1	59	32	26	53	27	52
	Δ	−33	−61	−60	−47	−67	−41
no co-Ser1, no co-Ser2	degen1	52	51	47	75	38	68
	Δ	−40	−42	−39	−25	−56	−25
no co-Ser1	degen1	60	55	41	75	42	69
	Δ	−32	−38	−45	−25	−52	−24
no co-Ser2	degen1	59	52	53	92	33	61
	Δ	−33	−41	−33	−8	−61	−32
no non-co-Ser1, no non-co-Ser2	degen1	94	88	78	100	91	91
	Δ	+2	−5	−8	0	−3	−2
no non-co-Ser1	degen1	94	90	82	100	93	92
	Δ	+2	−3	−4	0	−1	−1
no non-co-Ser2	degen1	92	90	84	100	93	92
	Δ	0	−3	−2	0	−1	−1
separating serine codon clusters artificially in amino acid data
co-Ser1 to Phe	20AA	94	59	59	89	65	84
	Δ	+29	+39	+35	+37	+49	+29
co-Ser1 to Trp	20AA	95	69	66	97	64	84
	Δ	+30	+49	+42	+45	+48	+29
co-Ser1 to Tyr	20AA	95	65	70	95	67	83
	Δ	+31	+45	+46	+43	+51	+28
co-Ser2 to Phe	20AA	94	52	56	97	78	91
	Δ	+30	+33	+32	+45	+62	+36
co-Ser2 to Trp	20AA	95	67	69	98	80	92
	Δ	+31	+47	+46	+46	+64	+37
co-Ser2 to Tyr	20AA	93	62	62	97	73	90
	Δ	+29	+42	+38	+45	+57	+35

^aMaximum likelihood bootstrap results of phylogenetic estimates ( = uppermost cluster of bootstrap values) for all nodes are in Table S1. Bootstrap results of manipulated matrices (mimicking amino acids with nucleotides, indentifying important serines, separating serine codon clusters artificially in amino acid data) are in Table S2. GTR+I+G degen1: Empirical nucleotide rate matrix estimated from the degen1-encoded data matrix according to a general time reversible model with rate heterogeneity estimated by a gamma function plus invariant sites. JTT 21AA & 20AA: Amino acid rate matrix of Jones, Taylor, and Thornton [12] with or without separate rate estimates for Ser1 and Ser2 (21AA and 20AA, respectively). GTR 20AA & 21AA: Empirical amino acid rate matrix estimated from the actual data according to a general time reversible model. ECM 20AA & 21AA: Amino acid rate matrix estimated from the PANDIT-based Empirical Codon Model [16]. Δ: For amino acid analyses, the difference in bootstrap percentages of the 21AA result minus the 20AA result. For nucleotide analyses, the difference in bootstrap percentages of the altered data set minus the unaltered data set (GTR+I+G). Differences >15% are bold. Ser1 to Ser2, Ser2 to Ser1: Ser1 codons (TCN) were artificially changed to Ser2 (AGY) from the degen1 data matrix, or vice versa. No Ser1, no Ser2: Ser1 and Ser2 codons were deleted (changed to NNN) from the degen1 data matrix. No (non-)co-Ser1, no (non-)co-Ser2: (Non-)ComminglingSer1 and Ser2 codons were both separately and together deleted from the degen1 data matrix. Co-Ser1 (Ser2) to Phe, Trp, Tyr: Co-Ser1 (Ser2) residues were changed to Phe, Trp, or Try residues, respectively.