(A) OG count vs. the number of Cyanidiales species contained in an OG (=OG size). Only genes from the sequenced genomes were considered (13 species). A total of 60 OGs are exclusive to the Galdieria lineage (11 species), 23 OGs are exclusive to the Cyanidioschyzon lineage (two species), and 13 OGs are shared by both lineages. A total of 46/96 HGT events seem to be affected by later gene erosion/partial fixation. (B) OG-wise PID between HGT candidates vs. their potential non-eukaryotic donors. Point size represents the number of sequenced species contained in each OG. Because only two genomes of Cyanidioschyzon were sequenced, the maximum point size for this lineage is 2. The whiskers span minimum and maximum shared PID of each OG. The PID within Cyanidiales HGTs vs. PID between Cyanidiales HGTs and their potential non-eukaryotic donors is positively correlated (Kendall's tau coefficient, p=0.000747), showing evolutionary constraints that are gene function dependent, rather than time-dependent. (C) Density curve of average PID towards potential non-eukaryotic donors. The area under each curve is equal to 1. The average PID of HGT candidates found in both lineages (‘ancient HGT’, left dotted line) is ~5% lower than the average PID of HGT candidates exclusive to Galdieria or Cyandioschyzon (‘recent HGT’, right dotted lines). This difference is not significant (pairwise Wilcoxon rank-sum test, Benjamini-Hochberg, p>0.05). (D) Presence/Absence pattern (green/white) of Cyanidiales species in HGT OGs. Some patterns strictly follow the branching structure of the species tree. They represent either recent HGTs that affect a monophyletic subset of the Galdieria lineage, or are the last eukaryotic remnants of an ancient gene that was eroded through differential loss. In other cases, the presence/absence pattern of Galdieria species is random and conflicts with the Galdieria lineage phylogeny. HGT would assume either multiple independent acquisitions of the same HGT candidate, or a partial fixation of the HGT candidate in the lineage, while still allowing for gene erosion. According to DL, these are the last existing paralogs of an ancient gene, whose erosion within the eukaryotic kingdom is nearly complete.