(A) Total number of genes in conserved micro-synteny regions using three different definitions of conserved synteny blocks: ‘relaxed’: only one syntenic homologous gene on either side; ‘current’: the definition used throughout our manuscript, which is more stringent than the ‘relaxed’ one; ‘stringent’: three syntenic homologous genes on either side, or two syntenic homologues if the −3, +3 neighbours have no homologues at all (see Materials and methods for details). The five most distant species in the fly dataset are shown on a separate, logarithmic scale for visual purposes, since the ‘stringent’ definition greatly reduces the number of identified conserved regions. Notice that this is true as well in other cases, such as for ‘Vpol’ and ‘Tbla’ in the yeast dataset. (B) Estimated ‘proportions explained by divergence’ for the pairwise (left) and the phylogeny-based (right) approaches, using three different definitions of conserved synteny blocks (explained in A). Note that the ‘relaxed’ definition introduces some known false positives (non-homologous genes which due to rearrangements are placed ‘opposite’ each other), and it is for this reason that the current definition was eventually preferred (see Materials and methods for more details). As expected from the presence of these false positives the results show a limited increase but are generally similar: pairwise overall mean 23% (relaxed) vs. 20.6% (current), phylogeny-based overall mean 33.7% (relaxed) vs. 30% (current). The ‘stringent’ definition becomes problematic in many cases, especially in the comparison of distant species, as the number of genes for which the synteny criterion is satisfied becomes too small (as shown in A). A comparison including only species with >200 conserved micro-synteny regions (i.e. excluding ‘Vpol’, ‘Tbla’, ‘Tpha’, ‘Ecym’, ‘Agam’, ‘Aaeg’, ‘Bmor’, ‘Tcas’, ‘Amel’, ‘Drer’) shows that in the pairwise case the mean drops from 16% in the current one to 10.9% in the stringent one. Applying the same cut-off in the phylogeny-based case (meaning that we exclude the four more distant phylostrata of the fly dataset), we get an average of 14% for stringent compared to 27.4% for current.