Names, ranks, and monophyletic groups

[Bengt Oxelman]

Torsten Eriksson <Torsten.Eriksson at BOTAN.SU.SE> writes:
>   A very interesting suggestion on this topic was made by de Queiroz and
>Gauthier [for example in TREE. 9: 27-31 (1994). "Toward a phylogenetic
>system of biological nomenclature"]. They suggest that ranks should be
>abandoned, and I agree, since ranks obviously have NO meaning unless they
>are also monophyletic groups.

RANKS cannot be monphyletic in any sense, but taxa can. If we insist in
making hierarchical classifications, ranks are obligatory. Of course, they
need not be Linnean and there is a conflict between Linnean and
phylogenetic hierarchies (i.e. they are in opposite directions, Soren
Lovtrup wrote about this conflict over twenty years ago).
If we throw away Linnean ranks, the relative inclusiveness of Ranunculales
and Ranunculus are impossible to tell, without having a tree including all
the types for the names.

>   For example, let's assume a unique species name vesca(XLS8976P) is a
>species which is included in a monophyletic group "Fragaria" which is
>included in a monophyletic group "Rosoideae", one could name it:
>Rosoideae Fragaria vesca(XLS8976P)  ...or...
>Rosoideae vesca(XLS8976P) ...or...
>Fragaria vesca(XLS8976P) ...or...
>vesca(XLS8976P)
>   To use just the binary "Fragaria vesca" would not be exact but
>unproblematical within small monophyletic groups, meaning the species
>"vesca" within the more inclusive monophyletic group "Fragaria".

As someone (Norman Johnson?) already noted, Fragaria vesca L. is already a
unique number when converted to bits in a computer. It is not illegal to
say  within the present system 'Rosoideae Fragaria vesca L.' (which in fact
is the way Floras, checklists etc. works implicitly). One nice thing about
binomials is that if you only knew Fragaria ananassa and heard the name
Fragaria vesca, you will also receive the information that these two taxa
are more related to each other than to any other taxon you have heard of.
This may of course be true or false, but that is the case for any
nomenclatural system. If one finds the monophyly of Fragaria uncertain, or
if the recogition of it would necissitate a huge split of e.g. Potentilla,
I cannot see why the name of a higher rank (Rosoideae) would be preferable.
Possiible, yes, but what, exactly, is gained?

>When it comes to types, I think that they are necessary for species (as
>used above) for practical purposes, but groups of species should be
>monophyletic groups - and a monophyletic group can never have ONE type and
>still have unequivocal borders. To have TWO "type species" as suggested by
>de Queiroz and Gauthier would solve that problem.

I acknowledge that the de Queiroz and Gauthier (QG) system makes naming of
clades in a tree unarbitrary, provided that both type taxa are included in
the analysis. BUT, when it comes to taxonomy and nomenclature, very often
(always?) our systematic (cladistic or whatever) analyses are incomplete.
If an analysis is made on four taxa of five, how should the fifth taxon be
unarbitrarily classified? Say that the four taxa (ABCD) are very well
characterized (say by ten non-homoplastic synapomorphies). A taxonomist
describe it as a taxon T and typify it according to the QG system with A
and D. If the fifth taxon (E) is found to be sister to ABCD, T cannot be
include E even if it differed only in lacking one of the ten synapomorphies
joining ABCD. Let's take a real example. Angiosperms are characterized by a
large number of synapomorphies. Suggest that someone typified them using
the rbcL tree of Chase et al. Ceratophyllum and some Asterid taxon would
probably be selected. Looking at another tree with Ceratophyllum nested
with say monocots (as in some rDNA trees), all clades occurring more
'basal' (like the Nymphaeales) would have to be excluded from the
angiosperms. Desirable?
I would argue that the major strength of the Linnean system is that
arbitrary decsions on inclusiveness CAN be made. It is possible to name
only 'monophyletic' taxa, and choose to formally rank only those taxa that
are well corroborated. The QG system may be relevant for naming clades on
particular trees, but the clades can be inspected on the tree itself, and
it is unclear what information the names are bearers of if one does not
have the tree they are based on at hand.

Bengt Oxelman
Dept. of Systematic Botany, University of Goteborg
Carl Skottsbergs Gata 22, S-413 19 Goteborg
Sweden

Phone: +46 31 7732669
Fax: +46 31 7732677
Internet e-mail: bengt.oxelman at systbot.gu.se