Baobab Phylogeny:

Some colleagues thought that a project seeking intra-specific variation in boab DNA was unlikely to be fruitful when other DNA studies showed that this species of baobab branched off from the others millions of years ago!

There are two different aspects of baobab phylogeny that are relevant here;-

1. The first is the sister relationships between pairs of baobab species that are revealed by DNA similarities. These are incontrovertible and show, for example, that the African and Australian species are closer to each other than either is to any of the other six species. They are so close that the DNA sequences used (genomic DNA from ITS) are not capable of separating them or of providing a reliable date for their separation. Similarly, the two Madagascar species that have colourless flowers and experience promiscuous pollination (Adansonia suarezensis, Adansonia grandidieri) are paired by DNA sequence similarities to the exclusion of the 4 Madagascar species with brightly-coloured flowers  and hawkmoth pollination.

2. The second aspect is the overall shape of the phylogenetic tree that reveals how the sister pairs defined in (1) have branched off in time. This aspect is much more difficult to define because of the rooting problem which bedevils many modern attempts at phylogenetic reconstruction. To take one of the many examples, there are presently two, completely incompatible views of the evolution of passerines, the most speciose group of birds. Are they a very ancient group that has much time to evolve their present diversity? Or are they a “crown” group that is recently evolved and therefore must be very rapidly evolving to have produced so many species in a short time? There is not time to go into all the reasons why I think that the ancient origin of passerines is correct, but in the present context it is enough to point out that choosing the wrong, root, or “outgroup” will invert the tree and allow for either of the two topologies. Say for example, that one chose an outgroup, which should represent the common ancestor, which was actually a member of the ingroup that have evolved after the passerines. In phylogenetic analysis, the chosen outgroup is forced to occupy the base of the tree and will force related taxa toward the base also. If the outgroup is actually part of the ingroup it will share more substitutions with the more advanced members, which will be dragged toward the base of the tree!  The more primitive members of the group will have less in common with this wrong “outgroup”, and will end up occupying positions at the top of the tree rather than their appropriate location around the base.
    This is the problem with Baum et al’s (1998) phylogenetic tree, which I think is inverted by the use of an inappropriate outgroup. There is no doubt that Bombax is closely related to Adansonia, with which it shares many characters, including pollen and floral features. The problem is that Bombax is part of the ingroup, instead of representing the ancestors of Adansonia. This is evident in a phylogeny from Baum’s own laboratory, based on the matK and other chloroplast sequences,  that shows Bombax and congeners emerging out of the clade that contains Adansonia, instead of belonging to the clade that gives rise to Adansonia, as required.

Fig 1:
The widely-accepted phylogeny of baobabs from the work of David Baum and collaborators. The sister pairings (e.g. the close relationship between the African and Australian species, the close pairing of the two Malagasy species with colourless flowers and promiscuous pollination) are well supported, but the order of branching on the tree has probably been inverted by the inappropriate choice of an outgroup which is derived, rather than primitive, with respect to the baobabs. This tree conflicts with most characters (floral, polyploid, biogeographic) but the conflicts are easily resolved if the tree is rooted using a taxon like Sterculia or Brachychiton that is unquestionably ancestral to Adansonia, rather than Adansonia being ancestral to it, such as Bombax.


    Choosing an appropriate outgroup (Sterculia and Brachychiton would be good choices, based upon the matK tree) would solve many conflicts of which Baum et al (1998) were aware in their study.  For example, the placement of the African species at the base of the (upside down) tree is hard to reconcile with its numerous derived characters, such as tetraploidy. Polyploidy only rarely, if ever, reverts, so one has to explain the loss of tetraploidy in seven species if the African species is placed in a primitive position in the tree. Similarly, the four Madgascar species with colourful flowers and hawkmoth pollination are considered by Baum to be primitive within baobabs, consistent with the primitive flower of A. perrieri, but this is inconsistent with the (upside down) tree. A change in the root will move these colourful Malagasy species to the base of the phylogeny, where they will also be consistent with the commonsense view, not to be gainsaid, that baobab origins lie in Madagascar, which is the home of six of the eight species and their greatest diversity.



Fig 2:
A chloroplast phylogeny of Malvales and relatives that shows that Bombax (along with Pachira and Pseudobombax) are questionable choices as outgroups for Adansonia. Bombax seems to be derived from Adansonia, rather than the reverse, making it a poor choice of outgroup. Forcing a derived taxon to be the root will distort the phylogeny by dragging the derived taxa along with it toward the base, leaving the primitive taxa to occupy the crown. This could explain why the phylogeny shown in Fig. 1 is inverted, with the taxa having known primitive characters on the crown, and clearly derived taxa at the base.


Fig. 3:
An alternative phylogeny of baobabs that is based upon Sterculia/Brachychiton as an outgroup. Note that the sister pairings are unchanged from those in the phylogeny (Fig 1) where Bombax was used as outgroup, but that the order of branchings is now inverted. The four colourful, longitube (referring to the fact that only the very long-tongued hawk moth can get access to the nectar of these four Malagasy taxa) species are now at the base of the tree where  one would expect them. It seems likely that the highly specialised evolution of baobabs has been driven by hawkmoths on Madagascar, which lacked bees until recently. Brightly-coloured nocturnal flowers are uniquely connected to the unique nocturnal colour vision of hawkmoths, which has also probably been responsible for the evolution of the baobab’s enormous flowers, beacons for hawkmoths searching for flowers at night. A. perrieri also has primitive flowers compared with all other baobabs, so its position at the base of the tree in this phylogeny is also appropriate. The (inverted) topology of this phylogeny therefore is consonant with the floral evolution of baobabs in a way that the “upside-down” phylogeny constructed using the Bombax outgroup is not (Fig. 1).
This alternative phylogeny also places the African species in the crown of the tree, as befits the derived characters it has, such as polyploidy, promiscuous pollination and colourless flowers.



This interpretation of the available phylogenetic information leaves open many lines of enquiry about baobabs that might have seemed closed by previous study:- for example:-
1.    If the African and Australian species actually occupy a more apical position on the phylogeny, more work with the appropriate genes could establish when they split.
2.    Since it has been possible to etablish the time of polyploidy in some species such as Glycine, the recent polyploid event in the African species could be defined using the same methods.
3.    A more appropriate outgroup (such as Brachychiton) could be used to define the relationships of the 8 baobab species and to check if my interpretation of the rooting problem has the outcome that I have sketched above.
4.    Haplotype analysis of the Australian species could define both the time and place of its origin in Australia.
5.    The site of origin of the Australian species within Africa could be defined by haplotype analysis of the African species.