Four of the five clades of the Sinningieae are either all tuberous or all non-tuberous. The only clade in which there is a mixture of tuberous and non-tuberous species is the Sinningia clade.
Here is the tree for the Sinningia clade derived from the data in Perret et al..
N____ Vanhoutteas | _____| | | | T____ S. hirsuta _____| | | | | | | T____ S. gigantifolia and S. cochlearis | | | | |_____| | | | N____ V. fruticulosa ____| | | | T________________ S. lindleyi | ________| | | | | ________________ Speciosa/guttata group | | | | T____| | | | |________________ S. tuberosa Clade _______| Father | | N______________________________ Paliavanas T = branch with all plants tuberous N = branch without tubers
The 64-dollar question is this: did Clade Father have a tuber?
In other words, did Sinningia speciosa and Sinningia cochlearis invent tubers on their own (because Clade Father didn't have one), or did the paliavanas and vanhoutteas in this clade lose the tubers they inherited (from a tuberous Clade Father)?
The T marks in the diagram show where tubers would have to have been invented, if Clade Father were tuberless. The N marks in the diagram show where tubers would have to have been lost, if Clade Father had a tuber.
One would assume that it would be easier to lose a now superfluous tuber than to invent one.
It's a lot easier to turn a soufflé into a quiche than vice versa.
T's should be more improbable that N's. Given that, one would expect that the most likely evolution would be the one which minimized T's. Given the choice between a route which had some number of T's and another which had an equal number of N's, the N route would be the more probable.
In the diagram above, there are four T's and three N's. On the face of it, then, it would appear more likely that Clade Father had a tuber, and that the three N branches lost it.
(The updated tree in Perret et al. has fewer implied T's and N's than the original paper's tree. The fact that a couple puzzling features of the original tree have disappeared in the new one gives one more confidence in the results.)
There is also a hidden N or T in the above diagram. Let's look back at the entire Sinningieae tree:
________T____ Dircaea clade | | | ___T___ Corytholoma clade __W___ | | | |_____| | |___?___ Sinningia clade ______ | | | | |___________N_________ "Free calyx lobes vanhouttea" clade Original ____| sinn | | |______________N______________ Thamnoligeria clade
If the branch marked by the question mark is to be an N, then either the Dircaea and Corytholoma clades must have invented tubers independently (W = N), or the Sinningia clade lost its inherited tuber (W = T) and then re-invented it and re-lost it in the Paliavanas branch.
This is pretty much the best case for the multiple tuber invention viewpoint. The lost/invented/lost scenario isn't unprecedented in the plant kingdom. Actively evolving families like the Asteraceae (aka Compositae) and the Orchidaceae seem to have gone through a bunch of it.
Even so, I still like the single-invention line best.
It might be possible to clarify the issue by showing the presence or absence of fundamental differences in structure or development between tubers on different T branches.
Isn't this fun?