To subscribers on the ANNELIDA list,
The latest issue of Zoologica Scripta contains a 'Point of View'
by Greg Rouse (1997. Rearticulating with extra assumptions: a response to
Eibye-Jacobsen and Nielsen. -- Zoologica Scripta 26: 61-66). This paper is
in answer to Eibye-Jacobsen & Nielsen (1997. The rearticulation of
annelids. -- Zoologica Scripta 25: 275-282), which in turn was a comment on
Rouse & Fauchald (1995. The articulation of annelids. -- Zoologica Scripta
24: 269-301).
Claus Nielsen and I do not feel that it would be helpful to
publish a response to the response to our comment. However, there are some
important remarks that we would like to make available to the readers of
ANNELIDA. If you have not read all three papers, the following discussion
might be difficult to follow.
Greg has written an eloquent paper that appears to answer all
the matters raised by Eibye- Jacobsen & Nielsen. The problem is that it
never really gets to the heart of the matter. A short introduction is
necessary to explain this conclusion.
The most important issues discussed by Eibye-Jacobsen & Nielsen
were:
1) Rouse & Fauchald (hereafter R & F) included Pogonophora and
Vestimentifera in their cladistic analysis, alongside Polychaeta and
Clitellata. Eibye-Jacobsen & Nielsen (hereafter E-J & N) suggested that
this very probably made Polychaeta paraphyletic and that a second analysis,
including Pogonophora and Vestimentifera in Polychaeta, was required.
2) R & F included Echiura in their analysis. E-J & N stated
that this might also render Polychaeta paraphyletic, and that a
supplementary analysis would be useful (see also McHugh, D. 1997.
Molecular evidence that echiurans and pogonophorans are derived
annelids. -- Proceedings of the National Academy of Sciences 94:
8006-8009).
3) R & F ran only one analysis, with all characters given equal
weight; the 18 resultant trees were summarized as a strict consensus
tree. E-J argued that the 18 trees should have been analyzed further
using successive character weighting.
4) R & F gave equal credence to all 18 of the trees that
resulted from their analysis. E-J & N suggested that trees containing soft
polytomies could be removed, providing that the remaining set of trees
contains compatible, more highly resolved solutions that have no zero-
length branches.
In addition to these points E-J & N suggested some alternative
interpretations of the characters used by R & F.
To summarize, E-J & N argued that uncertainties of the type
mentioned above in 1) and 2) could not be represented in just one data
matrix and that the issues in 3) and 4) suggest that the information
present in this matrix is not presented adequately in a consensus tree
based only on the trees resulting from the initial analysis. The entire
point of E-J & N was to argue that R & F had considered NO alternatives.
The title of Greg's paper highlights our disagreement. He
called it "Rearticulating with extra assumptions". Please note that R & F
certainly used assumptions in constructing their data matrix: for example,
keeping Pogonophora and Vestimentifera out of Polychaeta IS an assumption,
and one that is probably wrong. "Extra assumptions" implies that R & F
somehow used a "clean" matrix, with no assumptions. What E-J & N were
arguing for were not EXTRA assumptions, just OTHER ones. We felt strongly
that given the small size of R & F's matrix and the importance of the
entire subject, it was reasonable to expect that alternative
interpretations would be explored through supplementary analyses, rather
than just being mentioned in the text.
Finally, we would like to comment on some specific points in
Greg's paper.
Greg uses Euhirudinea to demonstrate that the addition of a
paraphyletic taxon (in this case Clitellata) can lead to the opposite
result of that shown by E-J & N, i.e., incorrectly INCREASED resolution.
The point is well taken, but this is certainly not a desirable result
either! Including paraphyletic taxa in an analysis is in the best of cases
neutral and otherwise a bad thing. We fail to understand why one would
ever knowingly include such a group. Alternatively, when the balance of
evidence is tipping heavily towards the conclusion that pogonophorans,
including vestimentiferans, are polychaetes, then running only one
analysis, one that ignores this evidence, is not enough. I (E-J) was
personally surprised to see the pogonophorans treated this way because in
October of 1994 I attended a lecture given by Kristian Fauchald in which he
stated that this group was probably closely related to the Terebellida.
Close enough to the Sabellida for me and certainly a very different opinion
of the group than that expressed in R & F.
The use of "Dinophilidae" by E-J & N in order to demonstrate the
problems that may arise from the use of paraphyletic taxa in a cladistic
analysis was criticized by Greg as being a "carefully selected case".
Dinophilus is, as Greg states, an aberrant polychaete (dorvilleid) group.
Our point was that pogonophorans (including vestimentiferans) are probably
also aberrant polychaetes. As Greg writes, "analyses that seek resolution
at one hierarchical level may be inappropriate at another level". That is
why E-J & N argued that R & F should have carried out a second analysis,
including these groups among the polychaetes. Surely neither van der Land,
Norrevang, Southward, Ivanov, Gardiner nor the recently deceased Jones
would have regarded this as "contemptuous".
Greg implies that E-J & N would a priori include Gnathostomulida
and Lobatocerebridae among the Polychaeta. We spoke directly to the
case of Lobatocerebridae and agreed with R & F in their treatment of
that group. E-J & N did not mention gnathostomulids at all, for the
simple reason that they do not agree with each other on the position of
that group.
The point E-J & N made about the molluscan cuticle was that
although there are areas of the epithelium that are covered by a
chitinous cuticle, usually more or less calcified (this character is an
autapomorphy for the Mollusca), there are also "naked" areas where the
cuticle has an ultrastructure resembling that of both annelids, sipunculans
and most non-chordate deuterostomes. We know of no investigation of these
parts of the molluscan cuticle which has searched specifically for
collagen. On the other hand, there are also very few studies of other
organisms where the collagenous nature of the fibres between the microvilli
has been positively demonstrated.
Greg defends the use of a consensus tree to summarize the
information in R & F's original 18 trees. Four out of five cladists
will agree that a consensus tree is not worth much and that the
information content is higher in any one of the trees used to construct the
consensus tree.
Greg's rejection of E-J & N's arguments for using successive
character weighting is more semantic than substantial. He is, however,
correct in pointing out that the procedure does in rare cases lead to an
increase in the total number of most parsimonious trees (but not in the
case at hand). The method has not been perfected yet, but rests on the
sound logic that all characters are not created equal (i.e., we are not
infallible in determining homologies). Claus and I regard successive
character weighting as a necessary part of any complete cladistic analysis.
Greg argues that "the detailed analysis of all trees in R & F was a better
way of dealing with the search results". We will have to agree to disagree
on that.
Regarding soft and hard polytomies, Greg writes that "the choice
of resolving polytomies may be an appropriate measure provided that
there is actually information available that the polytomies are indeed
'soft' or 'hard'." He goes on to note that there is no such information in
the case of the taxa treated by R & F. This is a strange statement.
Without time travel (or altering the data matrix) we will never have
information to decide whether any polytomy is soft or hard. The question
is more fundamental: does the hypothesis that a polytomy at the phylum
level can be hard (i.e., the result of a multiple speciation event) have
ANY credence? We don't believe it does. Again, we must agree to disagree.
Greg also suggests that it was unfair of E-J & N to criticize R & F on the
basis of a paper that had not yet been published (namely, Coddington &
Scharff. 1996. Problems with 'soft' polytomies. -- Cladistics 12:
139-145). This would be true if it were the entire story, but the option
to filter out soft polytomies has existed in the PAUP program since 1993
and is explained in the manual (Maddison & Maddison. 1992. MacClade ver.
3.01. Sinauer Associates, Sunderland, U.K.). Coddington & Scharff just
highlighted the problem.
In the end, each reader will have to form their own opinions in
the matter. We argued that the subject was so important that it
required a more complete analysis than that provided by R & F.
Danny Eibye-Jacobsen
Zoological Museum
Universitetsparken 15
DK-2100 Copenhagen O
Denmark
tel: (+45) 35 32 11 15
fax: (+45) 35 32 10 10
e-mail: dejacobsen at zmuc.ku.dk
-- ANNELIDA LIST
Discuss = <annelida at net.bio.net> = talk to all members
Server = <biosci-server at net.bio.net> = un/subscribes
Archives = http://www.bio.net:80/hypermail/ANNELIDA/
Resources = http://biodiversity.uno.edu/~worms/annelid.html
--
