Figures and data in Are interphylum spiralian relationships resolvable?

Figures
Tables
Additional files

7 figures, 12 tables and 2 additional files

Figures

Figure 1

Download asset Open asset

There are 15 unrooted trees linking the 5 major spiralian phyla and each has been supported in at least one publication (Supplementary file 1).

Tree numbering throughout this study refers to the trees in this figure. For all 105 rooted trees, the rooting schemes follow that depicted in unrooted T1.

Figure 2

Download asset Open asset

Branches linking spiralian phyla are shorter than both the short, disputed deuterostome branch (Kapli et al., 2021; Serra Silva et al., 2025) and the short branch leading to Lophotrochozoa (Telford et al., 2015).

(A and B) are data from SERR. (C and D) are data from MARL. Compared to the site-homogeneous model (**A, C**), internal branches (orange) become shorter under the site-heterogeneous model while the long branches leading to Platyhelminthes (blue) and Mollusca (dark green) become even longer (**B, D**). Circles on the y-axis indicate the median branch lengths of Deuterostomia (D), Ecdysozoa (E), Lophotrochozoa (L), and Protostomia (P). Triangles correspond to the median spiralian internal branches (Int), values given in Appendix 2—table 1. Scales are identical on either side of the y-axis breaks.

Figure 3

Download asset Open asset

Discordance-aware summary tree analyses show extensive gene-tree conflict within Spiralia.

For SERR (A), the Astral analysis recovers all spiralian phyla as monophyletic but no internal Spiralia branch with quartet scores (QS) over 0.50. The exception is the branch separating Platyhelminthes + outgroups from the remaining Spiralia phyla; however, Platyhelminthes are recovered within paraphyletic Gnathifera. For MARL (B), the Astral analysis yields very similar results, except that Annelida is not monophyletic. Branch-lengths are in coalescent units and monophyletic phyla have been collapsed, with the QS (Q1/Q2/Q3) for the branch subtending each phylum listed underneath the phylum names. Detailed QS can be found in the Astral.zip folder in the Dryad data repository (https://doi.org/10.5061/dryad.280gb5n3j).

Figure 4

Download asset Open asset

Relative median resampling estimated log-likelihood (RELL) pseudo-bootstraps (**A–D**) and randomised taxon-jackknives (**E–H**) log-likelihood ranks for all rooted trees, see Appendix 2—table 3 and Appendix 2—table 5 for all ranks.

Highest- and lowest-ranked topologies labelled for all analyses, along with every seventh topology from highest- to lowest-ranked. Topologies ranked by their difference in log-likelihood from the mean log-likelihood of each RELL or jackknife distribution. The RELL analyses of both datasets under the site-homogeneous LG + F + G4 model (**A, B**) occupy the widest ranges of log-likelihoods, which is consistent with the statistically significant differences found by the Kruskal–Wallis (KW) test. The RELL analyses under the site-heterogeneous EDM + F + G4 model (**C, D**) occupy a much narrower range of log-likelihoods. In practical terms, the wider the log-likelihood range, the more dissimilar scores are and the more we would expect to recover significantly different scores between topologies. The jackknife analyses show a similar trend, wider range of median log-likelihood scores under LG + F + G4 (**E, F**) than under EDM + F + G4 (**G, H**); however, the log-likelihood ranges of the jackknives under LG + F + G4 are very similar to that occupied by RELL under EDM + F + G4, consistent with the lack of statistically significant differences between topologies in the jackknife analyses.

Figure 5

Download asset Open asset

With simulated data inadequate models incorrectly support a platyhelminth root.

Using data simulated on an unresolved Spiralia tree, we ranked rooted spiralian topologies by difference to mean log-likelihood score. Highest- and lowest-ranked topologies are labelled for all analyses, along with every seventh-topology from highest- to lowest-ranked. For datasets simulated under EDM (complex) but scored under LG (simple) the 15 topologies with Spiralia rooted on Platyhelminthes are highest ranked (**A, B**). The plots for the ‘correctly’ specified models (**C, D**) show the expected approximately equal log-likelihood scores for all 105 rooted topologies. See Appendix 2—table 7 for complete topology rankings.

Figure 6 with 1 supplement

Download asset Open asset

Most fixed-taxon resampling estimated log-likelihood (RELL) bootstrap and randomised taxon-jackknife replicates, under both site-homogeneous (LG) and site-heterogeneous (EDM) models, recover unrooted T4 as the best-scoring topology.

Heights of bars correspond to the number of replicates where each topology had the highest log-likelihood score.

Figure 6—figure supplement 1

Download asset Open asset

Figure 7

Download asset Open asset

With simulated data, misspecified models recover unrooted T4 within the set of highest-scoring topologies.

Using data simulated on an unresolved Spiralia tree we ranked unrooted spiralian topologies by difference to mean log-likelihood score. For datasets simulated under EDM (complex) but scored under LG (simple), three topologies (uT4, uT5, and uT14) are within the highest-ranked topologies for both empirical datasets (**A, B**). The plots for the ‘correctly’ specified models (**C, D**) show the expected approximately equal log-likelihood scores for all 15 unrooted topologies.

Tables

Key resources table

Reagent type (species) or resource	Designation	Source or reference	Additional information
Other	Alignments, phylogenetic analyses, simulations, and ancillary scripts	DRYAD; https://doi.org/10.5061/dryad.280gb5n3j	Please see Materials and methods for data processing details
Software and algorithms	IQ-TREE2 (v.2.3.5)	Minh et al., 2020; http://www.iqtree.org/
Software and algorithms	EDCluster	Schrempf et al., 2020; Schrempf, 2020; https://github.com/dschrempf/EDCluster
Software and algorithms	ASTRAL-III (v.5.1.3)	Zhang et al., 2017a; Zhang et al., 2017b; https://github.com/smirarab/ASTRAL
Software and algorithms	MAFFT (v.7.455)	Katoh and Standley, 2013; https://mafft.cbrc.jp/alignment/software/
Software and algorithms	trimAl (v.1.4.rev15)	Capella-Gutiérrez et al., 2009; https://vicfero.github.io/trimal/
Software and algorithms	Amino acid diversity tests	Giacomelli et al., 2022; Giacomelli et al., 2025; Giacomelli, 2026; https://github.com/mgiacom/tardigrades_catpmsf
Software and algorithms	Additional data processing and analysis scripts	DRYAD; https://doi.org/10.5061/dryad.280gb5n3j

Appendix 2—table 1

Median branch lengths for the 105 rooted topologies.

Phylum-level branch lengths shown in Figure 2.

Branch	SERR + LG	SERR + EDM	MARL + LG	MARL + EDM
Deuterostomia	0.03139	0.02812	0.02305	0.01782
Ecdysozoa	0.04762	0.06257	0.03835	0.03941
Lophotrochozoa	0.02440	0.02607	0.02382	0.02613
Protostomia	0.06420	0.09325	0.04852	0.06641
Internal	0.01616	0.01393	0.01316	0.009462

Appendix 2—table 2

Fixed-taxon RELL pseudo-bootstraps rankings for 105 rooted topologies.

Topology	SERR + LG	SERR + EDM	MARL + LG	MARL + EDM
1	40	34	26	12
2	5	3	2	8
3	38	35	15	7
4	73	77	46	27
5	71	57	38	21
6	92	81	77	77
7	101	91	61	80
8	56	66	80	24
9	35	31	48	6
10	27	22	40	11
11	41	42	56	62
12	19	21	51	55
13	30	24	66	57
14	2	2	13	40
15	72	70	68	23
16	50	36	39	5
17	33	23	34	9
18	4	1	6	10
19	43	25	42	19
20	77	54	70	71
21	64	45	82	67
22	39	38	37	4
23	22	18	22	1
24	17	12	20	2
25	21	16	32	18
26	18	13	12	15
27	45	26	35	49
28	6	4	5	29
29	57	62	53	33
30	46	47	52	37
31	34	28	25	14
32	23	19	14	3
33	20	14	8	13
34	55	33	29	51
35	10	5	3	30
36	97	96	69	59
37	90	86	71	61
38	75	60	33	31
39	78	68	31	52
40	48	40	10	20
41	49	39	21	26
42	7	6	1	22
43	100	92	95	72
44	89	78	101	79
45	70	53	49	43
46	13	8	9	28
47	74	51	59	53
48	93	69	65	68
49	67	46	41	32
50	82	102	99	99
51	81	100	85	103
52	47	88	67	87
53	29	49	45	44
54	16	32	58	83
55	26	43	72	88
56	1	7	19	66
57	102	104	73	86
58	103	105	60	94
59	66	95	44	69
60	52	75	50	70
61	36	55	16	34
62	37	52	28	42
63	3	9	4	35
64	98	103	96	101
65	96	101	86	104
66	60	90	64	90
67	8	20	11	65
68	54	89	63	84
69	51	63	55	46
70	59	76	90	91
71	65	64	74	41
72	11	11	17	25
73	63	61	79	45
74	58	44	54	17
75	84	59	75	38
76	99	82	105	98
77	105	93	102	102
78	88	83	94	74
79	14	17	27	58
80	85	79	100	82
81	91	80	103	93
82	94	71	97	76
83	86	65	78	56
84	104	87	98	96
85	87	98	89	95
86	15	27	24	78
87	83	97	92	100
88	95	99	88	105
89	80	94	87	92
90	69	72	81	54
91	79	84	104	97
92	25	29	30	16
93	31	41	57	47
94	32	56	36	39
95	42	74	43	48
96	24	30	18	36
97	62	67	47	85
98	12	15	7	63
99	53	50	62	50
100	76	73	84	89
101	61	58	83	64
102	68	85	93	75
103	28	37	76	73
104	44	48	91	81
105	9	10	23	60

Appendix 2—table 3

Fixed-taxon RELL pseudo-bootstraps best-ranked topology frequencies for 105 rooted topologies.

Omitted topologies were never recovered as the best-ranked tree.

Topology	SERR + LG	SERR + EDM	MARL + LG	MARL + EDM
2	8.4	11.96	27.62	0.43
3	0	0	0	0.59
14	8.09	15.74	0	0
18	11.23	51.78	1.72	0.34
22	0	0	0	0.04
23	0	0	0	57.2
24	0	0.28	0	36.52
26	0	0.1	0.01	0.05
28	0.41	3.8	4.72	0
29	0	0	0	0.01
31	0	0	0	0.67
32	0	0	0	3.53
33	0	0.04	0.36	0.5
35	0.26	6.15	13.15	0
38	0	0	0	0.01
39	0	0	0	0
40	0	0	0	0.01
42	0.34	5.26	50.33	0.05
46	0.01	1.49	0	0.02
56	63.53	2.23	0	0
61	0	0	0	0.02
63	7.65	0.54	2.03	0
67	0.03	0	0	0
72	0	0.07	0	0
92	0	0	0	0.01
98	0	0	0.06	0
105	0.05	0.5	0	0

Appendix 2—table 4

Randomised taxon-jackknife rankings for 105 rooted topologies.

Topology	SERR + LG	SERR + EDM	MARL + LG	MARL + EDM
1	39	18	36	19
2	7	15	3	24
3	22	14	6	10
4	63	64	30	13
5	54	13	28	17
6	80	54	75	83
7	100	88	65	72
8	58	45	85	11
9	33	12	52	6
10	20	2	35	4
11	48	41	50	53
12	15	7	45	50
13	34	16	77	69
14	4	9	19	70
15	77	58	71	12
16	52	11	48	7
17	24	4	26	5
18	2	3	1	14
19	35	5	43	20
20	91	61	79	76
21	67	27	91	77
22	55	33	38	3
23	38	6	22	2
24	19	1	11	1
25	27	8	31	23
26	23	10	9	16
27	56	37	47	60
28	9	24	12	54
29	72	59	63	26
30	61	40	64	40
31	43	22	20	9
32	31	31	15	8
33	36	65	10	18
34	75	87	44	67
35	13	80	13	58
36	93	81	69	46
37	79	63	78	63
38	64	55	25	25
39	92	92	27	42
40	30	56	7	22
41	42	46	32	28
42	8	42	4	37
43	96	76	96	71
44	90	70	102	90
45	60	38	58	44
46	11	26	16	52
47	68	47	67	57
48	101	93	72	78
49	69	53	57	30
50	82	100	99	85
51	83	104	86	94
52	29	69	54	56
53	21	21	39	32
54	12	17	49	66
55	28	30	76	82
56	1	20	17	84
57	95	101	60	61
58	99	105	55	75
59	46	77	23	38
60	37	50	42	55
61	16	39	5	33
62	26	34	24	43
63	3	32	2	50
64	97	98	92	87
65	102	102	88	95
66	44	71	46	68
67	6	60	8	65
68	45	68	51	64
69	50	35	56	39
70	49	43	90	89
71	70	75	70	34
72	10	57	21	41
73	66	79	61	45
74	53	25	53	21
75	71	23	62	36
76	98	62	104	105
77	105	89	100	102
78	87	84	98	74
79	14	78	41	91
80	81	86	93	93
81	84	74	103	104
82	88	66	95	79
83	86	72	80	49
84	104	95	97	92
85	89	97	87	96
86	17	94	34	100
87	85	99	82	101
88	103	103	84	103
89	73	83	73	81
90	76	44	81	48
91	78	52	105	99
92	40	19	33	15
93	47	28	66	47
94	41	49	29	29
95	62	85	40	31
96	32	73	14	27
97	74	96	59	88
98	18	90	18	86
99	57	67	68	35
100	94	91	89	73
101	59	51	83	59
102	65	82	94	62
103	25	29	74	80
104	51	48	101	98
105	5	36	37	97

Appendix 2—table 5

Randomised taxon-jackknife best-ranked topology frequencies for 105 rooted topologies.

Omitted topologies were never recovered as the best-ranked tree.

Topology	SERR + LG	SERR + EDM	MARL + LG	MARL + EDM
2	15	0	0	0
8	0	0	0	5
9	0	0	0	0
10	0	10	0	5
12	0	5	0	0
14	5	5	0	0
17	0	5	0	0
18	30	5	25	0
23	0	5	0	30
24	0	40	0	55
26	0	5	5	0
31	0	5	0	0
32	0	5	0	5
33	0	0	15	0
40	0	0	10	0
42	0	0	5	0
54	10	0	0	0
56	30	0	0	0
61	0	0	10	0
63	10	0	20	0
67	0	0	10	0
92	0	5	0	0
103	0	5	0	0

Appendix 2—table 6

Simulations on rooted Spiralia star-tree rankings for 105 rooted topologies.

For columns SERR + EDM and MARL +EDM, proportions are based on 10 alignments.

Topology	SERR + LG EDM sim	SERR + EDM	MARL + LG EDM sim	MARL + EDM
1	24	9	24	71
2	4	2	4	7
3	21	41	18	34
4	57	44	37	44
5	54	89	36	46
6	69	103	70	47
7	81	91	56	61
8	95	86	105	93
9	64	90	101	78
10	49	46	93	80
11	51	61	86	87
12	23	19	58	56
13	26	53	73	86
14	6	5	14	8
15	94	99	81	99
16	68	57	77	92
17	47	95	55	102
18	10	31	9	11
19	48	97	48	100
20	73	77	88	98
21	62	73	95	96
22	103	81	74	21
23	78	87	68	28
24	55	8	49	24
25	53	22	50	9
26	37	3	21	32
27	39	42	33	101
28	11	47	6	16
29	102	29	65	23
30	90	58	66	20
31	50	71	29	25
32	46	70	32	35
33	34	50	17	48
34	38	79	23	105
35	8	48	2	19
36	87	65	57	42
37	67	26	62	27
38	41	83	27	50
39	43	76	30	55
40	19	68	16	37
41	22	40	22	82
42	3	7	1	1
43	96	62	98	75
44	79	45	99	60
45	42	78	39	79
46	9	60	7	13
47	44	98	45	90
48	61	94	63	103
49	52	80	61	76
50	91	34	104	64
51	85	67	103	72
52	31	20	84	67
53	30	35	78	77
54	17	10	53	69
55	20	39	67	65
56	2	4	13	12
57	80	36	71	51
58	74	75	64	59
59	29	16	43	54
60	28	24	44	41
61	16	37	19	45
62	18	13	26	81
63	1	1	3	2
64	99	28	87	95
65	98	84	85	97
66	32	12	54	83
67	5	11	10	18
68	33	82	47	94
69	35	43	83	88
70	36	30	92	85
71	60	100	35	84
72	13	105	8	3
73	56	21	25	43
74	66	14	51	53
75	82	23	46	49
76	100	69	91	29
77	105	74	82	70
78	63	88	42	74
79	14	63	12	6
80	59	59	34	36
81	76	25	80	15
82	84	64	76	39
83	92	52	90	58
84	104	85	94	68
85	72	96	38	89
86	15	101	11	5
87	70	56	28	40
88	101	51	59	63
89	89	15	52	38
90	86	38	79	33
91	93	33	89	30
92	75	72	60	14
93	83	17	72	10
94	77	32	40	17
95	97	49	41	26
96	40	18	20	52
97	45	93	31	104
98	12	66	5	22
99	65	92	100	73
100	88	102	102	91
101	58	104	96	66
102	71	55	97	57
103	25	27	69	31
104	27	54	75	62
105	7	6	15	4

Appendix 2—table 7

Simulations on rooted Spiralia star-tree best-ranked topology frequencies for 105 rooted topologies.

For columns SERR + EDM and MARL + EDM, proportions are based on 10 alignments. Omitted topologies were never recovered as the best-ranked tree.

Topology	SERR + LG EDM sim	SERR + EDM	MARL + LG EDM sim	MARL + EDM
10	0	0	0	20
18	0	20	0	0
23	0	20	0	0
28	0	0	15	0
33	0	0	1	0
35	0	0	84	0
49	2	0	0	0
56	98	0	0	0
60	0	20	0	0
72	0	0	0	20
85	0	0	0	20
86	0	0	0	20
89	0	20	0	0
91	0	20	0	0
98	0	0	0	20

Appendix 2—table 8

Fixed-taxon RELL pseudo-bootstraps best-ranked topology frequencies for 15 unrooted topologies.

Topology	SERR + LG	SERR + EDM	MARL + LG	MARL + EDM
1	0	0	0	0
2	19	2	0	3
3	0	0	0	0
4	81	98	60	96
5	0	0	35	1
6	0	0	5	0
7	0	0	0	0
8	0	0	0	0
9	0	0	0	0
10	0	0	0	0
11	0	0	0	0
12	0	0	0	0
13	0	0	0	0
14	0	0	0	0
15	0	0	0	0

Appendix 2—table 9

Compositional diversity model adequacy tests on the fixed-taxon data matrices.

Data simulated under the EDM + G4 model are closest in amino acid compositional diversity to the empirical data. Bold values correspond to the simulated data that most closely resemble the empirical data.

Model	Amino acid diversity		Z-score
Model	Empirical data	Simulated data	Z-score
MARL
LG + G4	5.1592	5.4859 ± 0.005859	55.7581
EDM + G4	5.1592	5.3082 ± 0.005771	25.8182
SERR
LG + G4	6.2950	6.5769 ± 0.007286	38.6836
EDM + G4	6.2950	6.1735 ± 0.006021	–20.1752

Appendix 2—table 10

Randomised taxon-jackknife best-ranked topology frequencies for 15 unrooted topologies.

Topology	SERR + LG	SERR + EDM	MARL + LG	MARL + EDM
1	2	0	4	0
2	23	11	0	2
3	0	0	0	2
4	38	61	43	92
5	0	6	26	1
6	0	0	12	0
7	0	0	0	0
8	30	0	0	1
9	5	1	15	1
10	0	0	0	0
11	0	0	0	0
12	0	0	0	0
13	0	0	0	0
14	2	20	0	1
15	0	1	0	0

Appendix 2—table 11

Simulations on 5-taxon Spiralia star-tree best-ranked topology frequencies for 15 unrooted topologies.

For columns SERR + EDM and MARL + EDM, proportions are based on 10 alignments.

Topology	SERR + LG EDM sim	SERR + EDM	MARL +LG EDM sim	MARL + EDM
1	0	0	1	0
2	73	20	0	0
3	0	0	0	0
4	0	0	0	0
5	0	40	0	80
6	0	0	99	20
7	0	0	0	0
8	22	0	0	0
9	0	0	0	0
10	0	0	0	0
11	0	0	0	0
12	0	20	0	0
13	0	0	0	0
14	0	0	0	0
15	5	20	0	0

Additional files

Supplementary file 1 Literature review data with published spiralian topologies.: https://cdn.elifesciences.org/articles/110607/elife-110607-supp1-v1.xlsx
Download elife-110607-supp1-v1.xlsx
MDAR checklist: https://cdn.elifesciences.org/articles/110607/elife-110607-mdarchecklist1-v1.docx
Download elife-110607-mdarchecklist1-v1.docx

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Mendeley

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

Ana Serra Silva
Maximilian J Telford

(2026)

Are interphylum spiralian relationships resolvable?

eLife 15:RP110607.

https://doi.org/10.7554/eLife.110607.3

Figures

There are 15 unrooted trees linking the 5 major spiralian phyla and each has been supported in at least one publication (Supplementary file 1).

Branches linking spiralian phyla are shorter than both the short, disputed deuterostome branch (Kapli et al., 2021; Serra Silva et al., 2025) and the short branch leading to Lophotrochozoa (Telford et al., 2015).

Discordance-aware summary tree analyses show extensive gene-tree conflict within Spiralia.

Relative median resampling estimated log-likelihood (RELL) pseudo-bootstraps (A–D) and randomised taxon-jackknives (E–H) log-likelihood ranks for all rooted trees, see Appendix 2—table 3 and Appendix 2—table 5 for all ranks.

With simulated data inadequate models incorrectly support a platyhelminth root.

Most fixed-taxon resampling estimated log-likelihood (RELL) bootstrap and randomised taxon-jackknife replicates, under both site-homogeneous (LG) and site-heterogeneous (EDM) models, recover unrooted T4 as the best-scoring topology.

Relative median resampling estimated log-likelihood (RELL) pseudo-bootstraps (A–D) and randomised taxon-jackknives (E–H) log-likelihood ranks for all unrooted trees, see Appendix 2—table 8 and Appendix 2—table 10 for all ranks.

With simulated data, misspecified models recover unrooted T4 within the set of highest-scoring topologies.

Tables

Median branch lengths for the 105 rooted topologies.

Fixed-taxon RELL pseudo-bootstraps rankings for 105 rooted topologies.

Fixed-taxon RELL pseudo-bootstraps best-ranked topology frequencies for 105 rooted topologies.

Randomised taxon-jackknife rankings for 105 rooted topologies.

Randomised taxon-jackknife best-ranked topology frequencies for 105 rooted topologies.

Simulations on rooted Spiralia star-tree rankings for 105 rooted topologies.

Simulations on rooted Spiralia star-tree best-ranked topology frequencies for 105 rooted topologies.

Fixed-taxon RELL pseudo-bootstraps best-ranked topology frequencies for 15 unrooted topologies.

Compositional diversity model adequacy tests on the fixed-taxon data matrices.

Randomised taxon-jackknife best-ranked topology frequencies for 15 unrooted topologies.

Simulations on 5-taxon Spiralia star-tree best-ranked topology frequencies for 15 unrooted topologies.

Additional files

Supplementary file 1

MDAR checklist

Download links

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

Be the first to read new articles from eLife

Share this article

Cite this article

There are 15 unrooted trees linking the 5 major spiralian phyla and each has been supported in at least one publication (Supplementary file 1).

Branches linking spiralian phyla are shorter than both the short, disputed deuterostome branch (Kapli et al., 2021; Serra Silva et al., 2025) and the short branch leading to Lophotrochozoa (Telford et al., 2015).

Discordance-aware summary tree analyses show extensive gene-tree conflict within Spiralia.

Relative median resampling estimated log-likelihood (RELL) pseudo-bootstraps (A–D) and randomised taxon-jackknives (E–H) log-likelihood ranks for all rooted trees, see Appendix 2—table 3 and Appendix 2—table 5 for all ranks.

With simulated data inadequate models incorrectly support a platyhelminth root.

Most fixed-taxon resampling estimated log-likelihood (RELL) bootstrap and randomised taxon-jackknife replicates, under both site-homogeneous (LG) and site-heterogeneous (EDM) models, recover unrooted T4 as the best-scoring topology.

Relative median resampling estimated log-likelihood (RELL) pseudo-bootstraps (A–D) and randomised taxon-jackknives (E–H) log-likelihood ranks for all unrooted trees, see Appendix 2—table 8 and Appendix 2—table 10 for all ranks.

With simulated data, misspecified models recover unrooted T4 within the set of highest-scoring topologies.

Median branch lengths for the 105 rooted topologies.

Fixed-taxon RELL pseudo-bootstraps rankings for 105 rooted topologies.

Fixed-taxon RELL pseudo-bootstraps best-ranked topology frequencies for 105 rooted topologies.

Randomised taxon-jackknife rankings for 105 rooted topologies.

Randomised taxon-jackknife best-ranked topology frequencies for 105 rooted topologies.

Simulations on rooted Spiralia star-tree rankings for 105 rooted topologies.

Simulations on rooted Spiralia star-tree best-ranked topology frequencies for 105 rooted topologies.

Fixed-taxon RELL pseudo-bootstraps best-ranked topology frequencies for 15 unrooted topologies.

Compositional diversity model adequacy tests on the fixed-taxon data matrices.

Randomised taxon-jackknife best-ranked topology frequencies for 15 unrooted topologies.

Simulations on 5-taxon Spiralia star-tree best-ranked topology frequencies for 15 unrooted topologies.

Supplementary file 1

MDAR checklist

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)