To compare our model with existing methods and also to evaluate the effect of neglecting variation in infectivity, a restricted version of the model assuming variation only in susceptibility (hereafter denoted the alternative model) was also fitted to the simulated infection data.
One model attributes the variation to differences in the frictional coefficients of rocks, assuming that the pore-fluid pressure field is uniform everywhere; the other model assumes variation in the pore-fluid pressure field with a constant friction coefficient (e.g., Zoback 1992; Rivera and Kanamori 2002).
In brief, the branch-specific model assumes variation among branches, but not among sites; the site-specific model assumes variation of selective pressure among sites, but not among branches.
Random-site models M0, M1a, M2a, M3, M7, M8 and M8a were fitted to our data; these models assume variation in ω among sites but not among lineages.
Codon substitution models [ 71] M1a M2a, M3a, M7, M7 and M8 were applied to the alignments and these models assume variation in ω among sites.
Additionally, random effect model assuming the variation between trials was applied.
Morphological characters were treated with the standard discrete model assuming gamma-shaped rate variation and variable coding bias.
The random effect model, unlike the fixed effect model, assumed the variation across countries to be random and uncorrelated with the predictor or independent variables included in the model.
A random-effects model assumes that variation across units (municipalities in this case) is random (as if drawn from a sample) and of substantive interest.
We ran M ODELTEST V.3.7[ 50] on this extended dataset and selected a GTR+R+I model, assuming a gamma-shaped rate variation and invariant sites.
This model assumes that NEE variations on interannual time scales can be represented as a linear perturbation of temperature anomalies, with the magnitude of the response proportional to the gross ecosystem fluxes in a given region.
