Divergence of bacilli and the purple bacteria

arlin at ac.dal.ca arlin at ac.dal.ca
Mon Mar 7 18:46:46 EST 1994

In article <1994Mar1.181935.21441 at dal1>, aroger at ac.dal.ca relays comments
from Peter Gogarten:

>[ . . .]  The deepest branching eubacterium for which the ATPase
> catalytic subunit sequence is known is Thermotoga maritima.  This sequence
> groups in significant distance to the root placed by the noncatalytic
> subunits (it is closer to E.coli than to the root). The known
> archaebacterial type ATPases subunits (Halobacterium, Methanococci,
> Methanosarcina, Sulfolobus and Thermoplasma) group either as a mono or a
> paraphyletic group (depending on the algorithms used, but in either case
> all of the archaebacteria group far away from the root given by the
> non-catalytic subunits.  The same is true for the elongation
> factor data (Karl Heinz Schleifer, Tech. Univ. Munich, pers.
> communication).  The root placed by the ancient gene duplications of ATPase
> subunits and elongation factors is well outside the known
> groups of eubacteria.
> Peter Gogarten
> (I'm posting his email to me...A. Roger)
I don't see how the general conclusion about "known groups of eubacteria"
follows from the work described.  I thought Andrew was questioning the
inter-phylum relationships on the 16S rRNA tree.  This is a reasonable
skepticism because the links on the tree are short and bootstraps or other
indicators of certainty have not been done (I've not seen a 16S rRNA tree
with bootstraps, but if anyone has seen one, please speak up!).
Furthermore, 16S rRNA trees in which the high-GC sequences of thermotogales
appear as "outgroups" to other eubacteria are (again, in my limited
experience) ones that are "rooted" with a few high-GC sequences from
archaebacteria or eukaryotes, rather than with a complete set (or a random
subset) of other sequences.

If the branching order of the 16S rRNA tree is not correct, then perhaps
the node closest to {eukaryotes, archaebacteria} is not {thermotogales,
other eubacteria} but {XXXX, other eubacteria} where XXXX might be
planctomycetes, cyanobacteria, high-GC gram-positives or any of a dozen
other eubacterial phyla.  Questioning the rRNA tree means questioning such
matters.  If the root of duplicated protein trees falls between the nodes
{eukaryotes, archaebacteria} and {thermotoga, {E.coli, B.subtilis}} this is
not evidence that eubacteria are a clade or that Thermotoga is an outgroup
to any eubacteria other than E.coli and B.subtilis.  Instead, eubacterial
phylum XXXX might be an outgroup to all life.

In short, if one doesn't use representatives of all eubacterial phyla in a
protein tree, then one is relying on the (questionable) 16S rRNA tree to
inform one's judgment about which groups to include.  Such pruned protein
trees will certainly be valuable in clarifying the relationships of the
phyla *that are included in them,* but they leave open the possibility that
"known groups of eubacteria" are paraphyletic, or if holophyletic, that
thermophiles are not outgroups to other eubacteria.  And these issues are
important for resolving questions about the history of thermophily and the
status of the three 'urkingdoms.'


More information about the Mol-evol mailing list