David's caveat about sequencing shortcomings is valid, but the reason that science is so successful is that it is self-correcting. Errors perpetrated by one scientist are discovered and rectified by subsequent workers, and improved understanding results.

Here is a recent example: In a molecular genetics-based interpretation of species-level phylogeny, Duda, Kohn & Palumbi (2001, Biological Journal of the Linnean Society of London, 73:391) inadvertently included an erroneous sequence (in Fig. 2) that resulted in a clade consisting of Conus furvus and C. litteratus. In a broader study of morphological, behavioral and ecological as well as molecular characters of the former species, Espino et al. (2008, Nautilus, 122:143) showed that all other character sets pointed to a closer relationship with molluscivorous species than with vermivores such as C. litteratus. And their restudy sequencing the same genes showed that we had initially reported an erroneous sequence.

Of course as is always true in science, this is likely not the last word.

If anyone would like pdfs of these papers, I'll be glad to send them.

Alan




On Thu, 16 Jul 2009, David Campbell wrote:

>> John brings up an interesting point: just how reliable are all these new DNA tests at determine species, subspecies, etc.?<
>
> Relying entirely on DNA provides greater opportunity to be totally,
> wildly wrong due to contamination, a bad sample, etc.  There are
> nematode and insect sequences in the database identified as molluscan,
> for example.  I've gotten an insect sequence, a bacterial sequence,
> and a trematode sequence myself when trying to amplify snail DNA.  The
> trematode no doubt was parasitizing the snail, whereas the insect was
> a booklouse, a tiny dot that eats paper and is common in the
> collections building.  The bacterium was probably involved in decay,
> since that specimen was a dead-collected endangered species.
>
> There are also plenty of mixups and misidentifications for things
> closer to the right name.
>
> Some sort of actual sample is needed for the analyses.  This can be a
> problem if it's not easy to poke the organism you want to identify
> (What kind of cone shell is that?  Just pick it up and poke it with
> your barcoder and see if its sting is fatal or not.), not to mention
> the ease of getting bacteria or other things living on the organism
> instead of the animal itself.
>
> Although the gene may work well in cowries, it is giving strange
> results with several other mollusks.  Pseudogenes (i.e., old copies of
> the gene that have been incorporated into the genome somewhere, not
> functioning as genes but available to be amplified in your analysis)
> have been shown to be a serious problem in several crustaceans, and I
> strongly suspect they're in the mollusk data already.
>
> The biggest problem, however, is that there needs to be a thorough,
> reliable data base to compare the sequences generated by the barcoder
> to.  This won't happen unless there's support for people to work on it
> and do it right-in conjunction with careful morphological work to
> identify the species.  Try searching for jobs in invertebrate
> systematics and you'll have an idea as to whether the necessary data
> base will be in place in ten years.
>
> --
> Dr. David Campbell
> 425 Scientific Collections
> University of Alabama
> "I think of my happy condition, surrounded by acres of clams"
>
> ----------------------------------------------------------------------
> [log in to unmask] - a forum for informal discussions on molluscs
> To leave this list, click on the following web link:
> http://listserv.uga.edu/cgi-bin/wa?SUBED1=conch-l&A=1
> Type your email address and name in the appropriate box and
> click leave the list.
> ----------------------------------------------------------------------
>

----------------------------------------------------------------------
[log in to unmask] - a forum for informal discussions on molluscs
To leave this list, click on the following web link:
http://listserv.uga.edu/cgi-bin/wa?SUBED1=conch-l&A=1
Type your email address and name in the appropriate box and
click leave the list.
----------------------------------------------------------------------