Simulation

Parameter set

Tree height

Ntaxa

Nchar

Simulation software

No.


I

A

0.7

30

1000

SeqGen

I1


A

0.5

30

1000

SeqGen

I2


A

0.3

30

1000

SeqGen

I3


A

0.1

30

1000

SeqGen

I4


B

0.7

30

1000

SeqGen

I5


B

0.5

30

1000

SeqGen

I6


B

0.3

30

1000

SeqGen

I7


B

0.1

30

1000

SeqGen

I8

II

A

0.5

22

1000

SeqGen

II1


A

0.5

50

1000

SeqGen

II2

III

A

0.5

22

1000

Mesquite

III1

IV

A

Nonclock

22

1000

SeqGen

IV1

V

A

0.5

30

300

SeqGen

V1


A

0.5

30

2000

SenGen

V2

 One hundred replicates were performed for each set of conditions. The models consisted of JC (0) [61], K80 (1) [62], SYM (5) [63], F81 (3) [64], HKY (4) [65, 66], GTR (8) [67], JC + I (1), K80 + I (2), SYM + I (6), F81 + I (4), HKY + I (4), GTR + I (9), JC + Γ (1), K80 + Γ (2), SYM + Γ (6), F81 + Γ (4), HKY + Γ (4), GTR + Γ (9), JC + I + Γ (2), K80 + I + Γ (3), SYM + I + Γ (7), F81 + I + Γ (5), HKY + I + Γ (5) and GTR + I + Γ (10), where 'I' represents the proportion of invariable sites, 'Γ' represents the discrete gamma distribution with four rate categories, and number in parentheses is the number of free parameters of each model. One classification of the 24 models was to put them into four categories: base (JC, K80, etc.), base + I (JC + I, K80 + I, etc.), base + Γ (JC + Γ, K80 + Γ, etc.) and base + I + Γ (JC + I + Γ, K80 + I + Γ, etc.). The other was that models with the same number of free parameters were grouped together, resulting in a total of 11 categories. In addition, we called every four models having the same parameters in the substitutionrate matrix as baselike models (e.g., the four models of SYM, SYM + I, SYM + Γ, SYM + I + Γ were called as SYMlike models).