Finally, we aimed to learn about functional impact of inferred subs from protein structure. We showed that changes during transitions are clustered round the binding pocket. Consistent with our other results, divergent changes are closer to the binding pocket in Stegnosperma

We also tested for evidence of adaptive convergence using a suite of different methods. Their consensus shows an array of sites with inferred convergent and divergent subs, supporting adaptive convergence for high activity

However, drift can cause repeated changes to occur by chance. We tested if we saw more changes in the same site than expected. There was significantly more convergences between two transitions, but significantly more divergences with a third, in Stegnosperma.

We know that convergent evolution can reuse the same genetic variation, but sometimes the same trait can be produced by different changes. We compiled all the inferred overlapping amino acid substitutions between our transitions and observed a mosaic of convergent, divergent and unique change.

So, we aimed to test our hypothesis of multiple transitions and infer the key changes in the protein leading to high activity. To do this, we used phylogenetic models to infer ancestral sequences and expressed them in yeast - and our results conclusively support 3 transitions to high activity!

In Sheehan et al. 2020 doi.org/10.1111/nph.... we reconstructed 4 origins of betalains and showed that a key biosynthetic enyzme, DODA (L-DOPA->betalamic acid) had a pattern of high activity following repeated gene duplication indicating multiple transitions⇒multiple betalain origins.

Firstly some background: betalains are red and yellow pigments that in plants occur only in Caryophyllales🌵they never co-occur with anthocyanins, but some lineages in Caryophyllales still produce anthocyanins instead. E.g. carnations produce anthocyanins, but beetroot is coloured by betalains.