[Taxacom] Taxonomy in Europe in the 21st century

[Karl Magnacca]

On May 8, 2008, Doug Yanega wrote:
>>If a DNA barcode project on midges in Norway reveals that there
>>are 100 genetic clusters that we are unable to put a name on and
>>possibly represents new species, is that not a decent argument to fund
>>taxonomy on this particular group?
>
> Actually, if those 100 clusters have no morphological characters to
> back them up, that's a decent argument that DNA barcoding does not
> work for identifying species in midges, and should be treated with
> utmost circumspection, if not abandoned altogether.

But that's not what he was saying - the point is that if there is evidence
of extensive genetic segregation, it provides justification for taxonomic
(i.e. morphological) investigation of the group.  Maybe there are
genuinely cryptic species based on characters that no one paid attention
to before, or that were considered to be variable; or maybe it really is
just all one species.  But without taking at least a basic look at the
genes like this, it's just one widespread fly that no one cares about.

> Gael Lancelot similarly wrote:
>>The members of the group felt that making identification more
>>effective through barcoding[...]
>
> 90% of the species already described are insects, as are 90% of the
> *undescribed* species. There's very little evidence that mtDNA
> barcoding - in and of itself - EVER works reliably for identifying
> insect species; what work *has* been done in this regard is not very
> encouraging, and it looks like the technique has only limited
> utility, as an *adjunct* to traditional taxonomic delineation of
> species.

I think the problem here (and perhaps above as well) is that you're using
different definitions of "identifying"; it sounds like you (Doug) are
talking about characterizing/describing species, whereas my impression was
that Gael was talking about identifying *specimens* (correct me if I'm
wrong).  It's unfortunate that the former was pushed so strongly in the
early hype over barcoding, because even though it seems to have been
widely recognized by now among the molecular people that it isn't very
good without reciprocal feedback from morphological taxonomy, the idea
persists in a lot of people's minds, particularly taxonomists.

I've been working on a project to look at the efficacy of barcoding for
several months now, and reviewing the literature on it has been
instructive in seeing authors' biases.  On one side you have people
claiming to find "cryptic species" which are never fully characterized; on
the other you have claims that misidentification is rampant, based on
sequences pulled from GenBank with no attempt to verify that they were
correctly identified in the first place.  It's also somewhat amusing, but
rather disturbing, to find papers with abstracts that are stridently pro-
or anti-barcoding, and to then read through and find that in fact they
come to almost exactly the same conclusions: that it works reasonably well
when you have a group of well-characterized species, and can be useful as
an alert to the possible presence of cryptic species, but there are so
many instances of both multiple species with identical sequences, and
multiple haplotypes within a species, that traditional taxonomy is very
much needed.

Karl
=====================
Karl Magnacca
Postdoctoral Researcher
Department of Zoology
Trinity College, Dublin 2
Ireland