Figure 4

Probability of detecting HGT for compositional and phylogenetic methods by taking into account the time since the event and the phylogenetic distance between donor and receptor. The shaded surfaces are these combinations of time and the phylogenetic distance for which a HGT can be detected by each method. As composition ameliorates faster than phylogenies, old events can only be detected using phylogenetic methods, and only when they involve distant taxa. Compositional methods usually detect more events since HGT seems to occur preferentially between close taxa, where the resolution of phylogenetic methods is low. A putative event can be identified by both methods when both surfaces overlap; that is, when it involves relatively distant taxa and happened recently. The exact shape of the surfaces indicating the capacity of the methods is as yet unknown. The green points indicate examples of different HGT events: 1) Old events between closely related taxa. We will observe normal composition (by amelioration), and congruent trees (not enough resolution since species are close). The event will be undetectable by both methods; 2) Old events between distant taxa. Normal composition, incongruent trees. Detectable by phylogenetic methods; 3) Recent events between close taxa. Atypical composition, congruent trees. Detectable by compositional methods; 4) Recent events between distant taxa. Atypical composition, incongruent trees. Detectable by both methods.