Skip to main content

Table 2 Accuracy of model selection for 100 simulations based on two ‘difficult’ empirical datasets

From: Does the choice of nucleotide substitution models matter topologically?

  

TTR

RF

BS

TMR

Dataset 1

AIC

0.10

0.176

0.057

0.34

 

AICc-S

0.09

0.179

0.057

0.40

 

AICc-M

0.10

0.176

0.057

0.34

 

BIC-S

0.09

0.176

0.055

0.51

 

BIC-M

0.10

0.174

0.054

0.40

 

GTR+ Γ

0.09

0.173

0.059

0.00

Dataset 2

AIC

0.35

0.036

0.001

0.49

 

AICc-S

0.35

0.036

0.001

0.48

 

AICc-M

0.35

0.036

0.001

0.49

 

BIC-S

0.35

0.036

0.001

0.14

 

BIC-M

0.35

0.036

0.001

0.02

 

GTR+ Γ

0.35

0.037

0.001

0.00

  1. TTR = fraction of recovery of true topology, RF = average relative Robinson-Foulds distance to the true tree over 100 simulation replicates, BS = average branch score difference to the true tree over 100 simulation replicates, TMR = fraction of true, generating model recovery