Incongruence between molecular trees

Harvey E. Ballard, Jr. hballard at STUDENTS.WISC.EDU
Wed May 7 16:25:47 CDT 1997

Incongruence in different molecular phylogenies has in the majority of
cases been between those constructed from organellar and nuclear data.  For
plants, this has particularly been true of chloroplast vs. nuclear
ribosomal phylogenies.  Such incongruence in the placement of particular
taxa has routinely been ascribed to chloroplast capture or as general
evidence of hybridization, since chloroplasts are in most plants maternally
inherited (in the bulk of the remainder, paternally inherited) whereas
nuclear genes are biparentally inherited.

One must be sure to look CLOSELY at instances of incongruence between gene
trees from the same fundamental genomic data (different types of nuclear
gene regions, for instance), to make sure one ignores clades that are
poorly supported and ought best be considered as polytomous, and that taxa
sampled in both (or all) data sets are being treated.  Different taxa
included for different genomic data sets will greatly alter phylogenetic
interpretations for two or more data sets, and inferences from different
data sets with different sampling of taxa have at BEST dubious validity, in
my opinion.  This point and the one I just made concerning
node/branch/clade support is the case for several recent publications
comparing phylogenies from different gene regions of nuclear DNA.
Conclusions were made in these papers about "significant" incongruence, but
when I reduced to polytomies those branches in the different phylogenies
with excessively low bootstrap support and deleted taxa that were not
common to both, the "incongruence" virtually disappeared.

As Dr. Iltis here at UW-Madison is fond of saying, one should always turn
one's "crap detector" WAY up when scrutinizing the literature

As for "total evidence" approaches, I am always loathe to use them.  The
molecular and macromorphological data sets I have analyzed cladistically
diverge wildly, and the latter shows less than 30% resolution.  The ITS
phylogeny is completely congruent, however, with base chromosome numbers
and patterns of natural hybridization.  When one maps macromorphological
characters, many of which could well be considered traits comprising a few
recurrent adaptive syndromes of pollinator specificity, fruit dispersal
mechanism, leaf and stem architecture, and growth habit, one finds
overlapping and divergent patterns for all of these repeated over and over
throughout the genus.  It's not surprising that I get virtually no
resolution from the macromorphological data set and relationships in "good"
clades that make no sense in the context of chromosome numbers or
hybridization or micromorphological characters, either.  My interpretation
is that many larger, more diverse groups (genera on up) will prove to have
recurrent patterns of convergence in macromorphological features that will
actually CONFOUND phylogenetic relationships based on other data, and
combining such characters with data sets taken from features that are
arguably under less substantial selective pressures will inevitably provide
misleading interpretations of relationships.  Anyone questioning such an
argument need only spend a little time making phylogenetic studies of an
oceanic island group and putative continental outgroups for confirmation of
this.  I would argue that similar patterns, although perhaps less striking,
are present in many larger groups of angiosperms but have not yet been
detected because few studies have on the diversity of larger groups.

Harvey Ballard

Harvey E. Ballard, Jr., Ph.D.
Postdoctoral Researcher, USDA, ARS
Horticulture Department, University of Wisconsin
1575 Linden Drive, Madison WI 53706
phone: (608) 262-0159
fax: (608) 262-4743
Honorary Fellow, University of Wisconsin Herbarium
Botany Department, UW
132 Birge, 430 Lincoln Drive, Madison WI 53706
phone: (608) 262-2792
fax: (608) 262-7509

Assistant Professor
Department of Environmental and Plant Biology
Porter Hall
Ohio University
Athens OH 45701
Phone: To be announced
Fax: (614) 593-1130
Email: A mystery as yet

More information about the Taxacom mailing list