Neighbor-joining (NJ) analyses were performed in PAUP*4.0b10 using the BIONJ algorithm with the HKY85 distance option or with maximum likelihood distance settings as determined by Modeltest 3.7 [ 27, 29, 30].
Prior to each test, the respective genome was removed from the reference phylogeny to avoid circularity, and MLTreeMap placements were made using either Maximum Parsimony or Maximum Likelihood (all other settings were identical; bootstrapping was not used).
Unambiguously aligned sequences were analysed using the distance, parsimony, and maximum likelihood approaches with default settings.
The same alignment was used as the basis for trees constructed by maximum parsimony (using MEGA version 2.1 with default settings) and maximum likelihood (using PHYLIP version 3.6 [ 39] with default settings).
Maximum likelihood analysis was performed under the likelihood settings suggested for the given dataset by the result of the modeltest using the heuristic search option with Tree Bisection Reconnection (TBR) branch swapping and 100 random sequence addition replicates.
An intra-specific phylogeny was produced using maximum likelihood (ML) methods and default settings in the program RAXML 7.2.7 [ 51].
During the calculation of the individual maps from the six populations the locus order within chromosomes and estimation of recombination frequencies were established employing the provided maximum likelihood algorithm with modified calculation settings.
The first issue was that where data for a key population are very sparse, which is a common occurrence in concentrated epidemic settings, the maximum likelihood estimate could vary substantially with the addition of just a single data point.
Using default settings, 9 maximum likelihood models of the evolution of gene expression were compared according to the Akaike Information Criterion (AIC), and the model with the minimum AIC value was chosen as the best fitting model of evolution of gene expression.
