[Taxacom] FW: DNA only descriptions

[Kim van der Linde]

Doug, Sergio, Thomas and Jason,

I hope you guys understood that I was to a degree playing advocate of 
the devil. What we need is discussion about ho0w to integrate the 
various ways of unraveling species boundaries, evolutionary 
relationships etc.

Yes, a single COI sequence won't be enough in some/many cases. 
Obviously. But it is a start. Gut bacteria, endosymbionts, and what ever 
more can mess up the signature.

Crytic species, yes, I agree, ideally we would like to have the 
reciprocal crossing schemes showing incompatibility. I would like to see 
those also for morphologically closely related species, because the same 
issues arise there.

I think any absolutism won't work. But my experience is that molecular 
phylogenies often have a substantially higher degree of resolution than 
cladistic phylogenies. And that is logical, because often, it is based 
on a lot more data than cladistic studies. So, I think we should use all 
possible sources and weigh them appropriately on a case by case basis.

Anyway, that is where I stand in this discussion.

Kim

On 8/17/2010 2:18 PM, Doug Yanega wrote:
> To follow up on Sergio and Thomas' comments:
>
> Kim van der Linde wrote:
>
>> Well, DNA barcoding is exactly developed for doing just that, with the
>> added advantage that you do not need to know all your species.
>>
>> For insects, we only need one leg (even from a living individual) to
>> make an ID, without even knowing what family those individuals are in.
>
> If you are talking about using multiple genes, including ribosomal
> and/or nuclear, then this might be true. If you are talking strictly
> about COI barcodes, then it is demonstrably not true. It is known
> that a single species can have more than one COI barcode, and that
> multiple species can have identical COI barcodes. In some cases,
> there are extra copies of the COI gene elsewhere in the cell, meaning
> the same *individual* can yield more than one barcode. In practice,
> COI sequences do not *always* map one-to-one with species boundaries,
> therefore the claims about accuracy of ID are overinflated (and
> under-tested; how many putative cryptic COI-based species have been
> subjected to attempts to interbreed, to confirm their reproductive
> isolation?). We might not be having this discussion if the gene
> selected as the "default option" for barcoding had not - as in COI's
> case - been linked to the mitochondrion, and therefore prone to all
> sorts of non-Mendelian artifacts. Ever hear of Wolbachia?
> Mitochondrial sweeps? Hybrid introgression? For a tool that is used
> so often, by so many, to have so many situations where it can be
> expected to fail, is not very comforting if that's the only tool in
> the proverbial kit.
>
> I do not have any real issue with research that utilizes data from
> multiple genes - I suspect that many of us "naysayers" would admit to
> this - but the barcoding *concept* (that we can get by with a single
> gene for *every* life form on the planet) is not realistic,
> especially in the case of COI, and investing a lot of time and energy
> in that sort of thing to the exclusion of other approaches can
> undermine the function of the systematics community. To return to the
> original point of this thread, it is certainly possible for a taxon
> to be recognizable and diagnosable solely by virtue of its genome,
> and to be named on such a basis - but I would argue VERY strongly
> against it were it to be based on a single gene, especially a
> mitochondrial gene. One piece of evidence does not automatically
> constitute proof.
>
> Sincerely,

-- 
http://www.kimvdlinde.com