*2.1. Data Mining*

To evaluate the GOX phylogeny, we considered a broad spectrum of phyla, including non-photosynthetic species. The well-characterized At-GOX2 from *Arabidopsis thaliana* (At3g14415) [30] was used in BLASTP [33] searches against defined taxonomic groups (e.g., cyanobacteria or proteobacteria) to find related GOX-like proteins. Proteins from 5 to 14 species of each taxonomic group were selected, which showed the best BLAST hits. In the case of underrepresented taxonomic groups, all of the sequences that could be identified as putative GOX proteins were used for the alignment (Table S1). In addition to the sequences in the databases, we determined the cDNA sequence of the putative GOX from the streptophyte green alga *Spirogyra pratensis* (sequence is shown in Figure S2; accession number AVP27295.1). Furthermore, the obviously miss-annotated GOX sequence of *Cyanophora paradoxa* was corrected (see below). The complete gene with a corrected exon/intron structure and the complete protein coding sequence is shown in Supplementary Material 1 (Figure S3; accession number AVP27296.1).
