Table 4.
Clade model likelihood ratio tests comparing Physalis Clade A (subtree) and its subtending branch to all other S-RNasesa
| Model | ||
|---|---|---|
| 1) Global dN/dS rate (whole tree) | ||
| Shared Parameters | lnL | AIC |
| dNdS Clade A = dNdS Alla | -17167.25 | 34716.51 |
| Global dN/dS = 0.90; CI = (0.87, 0.94) | ||
| 2) Separating Branch Versus Two Clades | ||
| Shared Parameters | lnL | AIC |
| dNdSdNdS Shared Clades A + Alla = 0.90; CI = (0.86, 0.94) |
-17166.63 | 34717.27 |
| Phys dNdS Branch A = 5.18; CI = (1.64, 10.49) | ||
| LRT p-value vs the single rate model = 0.266 | ||
| 3) Clade A + Branch vs Alla | ||
| Shared Parameters | lnL | AIC |
| dNdS Branch A = Clade A = 2.19; CI = (2.03, 2.36) | -17118.27 | 34620.54* |
| dNdS Clade S = 0.70; CI = (0.73, 0.77) | ||
| LRT p-value vs the single rate model < 0.001 | ||
| 4) Clade A (subtree) vs Branch + Alla | ||
| Shared Parameters | lnL | AIC |
| Phys dNdS Clade A = 2.04; CI = (2.20, 2.37) | -17118.80 | 34621.60 |
| Phys dNdS Branch A = dNdS; Clade Alla = 0.70; CI = (0.73, 0.77) |
||
| LRT p-value vs the single rate model < 0.001 | ||
| 5) Clade A, Solanum, and Branch | ||
| Shared Parameters | lnL | AIC |
| Phys dNdS Clade A = 2.04; CI = (2.20, 2.37) | -17118.25 | 34622.51 |
| dNdS Clade Alla 0.70; CI = (0.73, 0.77) | ||
| Phys dNdS Branch A = 2.96; CI = (0.83, 6.17) | ||
| LRT p-value vs the single rate model < 0.001 | ||
a) Background branches include all Solanum alleles and Physalis alleles outside of sub-Clade A
Alternative models (2-5) were tested using likelihood ratio tests (LRTs) against the null model 1 where dN and dS rates are shared among all branches on the phylogeny. Significance of p ≤ 0.05 was determined using χ2 with degrees of freedom (DF) equal to the number different parameters between models.
*Model 3 showing a common dN/dS for Clade A AND the subtending branch has the best fit AIC score sharing a distinct dN/dS.