Assorted musings on barcoding

Speaking of Guelph (see prev post), Paul Hebert, the Canadian barcode god (or so I'm told), gave a talk today about...well, barcoding life. Overall, it was a good talk, and I can see why he's in charge of a bunch of stuff -- his talks are quite convincing. At least he's convinced me not to automatically fear/resent barcoding. Not that anyone asked for my opinion on these matters, but this is my little corner of the internet so I'll ramble some thoughts about it. (yes, ramble. You've been warned.)

I have a few traditional taxonomist friends who have painted a pretty negative image of it for me; there's even a mycology grad student who does quite a bit of barcoding herself and has come to be quite skeptical of it. Furthermore, being fascinating by the biology of things rather than their mere existence, all those billions of species don't exist as far as I'm concerned until they've been actually studied. And no, 600bp of a conserved gene sequence does not qualify.

The value of being able to identify something in the field without having to look at the 'length of the spine on the 3rd tarsus', to [badly] paraphrase Hebert, is clearly there. His dream of a handheld barcoding device perhaps could be especially useful to people who actually do fieldwork (wait, there's biology outside the lab? WTF?), although how you can 'read' a DNA sequence sufficiently quickly still eludes me. Don't you still have to wait for crap to get amplified? Would be awesome if they have some way of doing really quick (and cheap) PCR, that could come in quite handy even for the model organism crowd. I don't necessarily enjoy waiting 2h to genotype some crap (oh, and another hour before that to extract gDNA). But I digress. This barcoding technology stuff does sound awesome.

However, I often have some skepticism towards awesome-sounding revolutionary technology -- too often people get carried away in the hype, and start attempting to use machinery to replace human intellect. That can be dangerous. Some tasks appear more mechanical than they really are, and it seems taxonomic identification and description is definitely among them. Most reasonable barcoding proponents don't seem to dismiss traditional taxonomy, but there is the fear that public opinion and funding agencies might not feel the same way. And as much as we love to reduce everything within our reach to digitised strings of characters, we still have to interact with the physical world.

Hopefully, the barcoding 'revolution' (that word is automatically associated with TC-S for me now, for some reason...) would be used in combination with the insight, knowledge and intelligence (ie talent) of traditional taxonomists, rather than attempting to replace them. Molecular phylogeny is substantially more reliable than anything morphology-based, as anyone who's wandered around in the protist kingdom would know*. But that doesn't mean morphological evolution is now to be ignored - on the contrary, molecular biology has made it much more exciting now that you could see how the organisms are related, and then investigate their morphology.

Another thing that annoys me is this whole 'species counting' business. You know, when someone spits out "There are 10 000 species of blah in blah", and the media picks up on it and we end up with "OMG, there are 20 000[sic] species of blah in [incorrect] blah!!!one! We are doomed/saved/awesome/cured of cancer[pick one]!" Seriously, who cares? Does it really matter whether there's 10 000 reproductively isolated clusters of something or 15 000 or 100 of them? Does counting them reveal anything profound? The majority of life on earth doesn't particularly care much about reproductive isolation! And among the few sorry exceptions where reproductive isolation becomes somewhat important (metazoa mostly; plants turn out to be a little more promiscuous), all species-counting can say is perhaps something interesting about...evolutionary dynamics of reproductive isolation. Furthermore, this ignores intraspecific variation, which is important and seemingly ignored by many evolutionary ecologists. And then we wonder why people get confused about where variation are. They see diversity as being composed of heavily discrete units, which may perhaps be confusing when trying to understand evolution itself.

That's not to say variation isn't important - it is, and it's definitely fascinating. But it does not do it much justice to simplify everything down to 'species', which then become treated as solid units of variation, rather than a subset thereof. It's far more complex than a mere absense of interbreeding.

Barcoding may be quite interesting for exploring how asexual vs. sexual (and facultatively sexual) organisms tend to cluster, how this clustering differs among various phyla, habitats, etc. Hebert did mention plans to pursue some of those directions, and it will be interesting to see what turns up.

Lastly, they seem to be using a single gene for barcoding. I think that's might even be what 'barcoding' would mean, strictly speaking. This is worrisome. Single gene phylogenies are CREEPY and SCARY o_O. I wouldn't touch one with a 10 foot pole anymore. Of course they picked something that [they think] is extremely conserved and constant (yet flexible enough to allow for variation at their desired resolution) - a commonly-shared mitochondrial gene (by the sound of it, Cox1 perhaps). They say it's conserved and reliable. Awesome - that's what they said about SSU rDNA. How did that turn out?

Bits of freaking FUNGI ended up as basal to all Eukaryotes, along with diplomonads, parabasalia and archamoebae, which have little to do with each other. Oh, and they still print this tree (which was quite a breakthrough at the time, to be fair; but things have changed in the last couple of decades...) Why? Some lineages experience faster rates of evolution than others, and this leads to much greater sequence divergence from its neighbours, which often screws things up in the alignment algorithms, resulting in Long Branch Attraction - things that diverged a shitload relative to everything else tend to cluster together. Some normally stable genes under certain circumstances can go haywire. Single gene trees a fucking dangerous. So are multigene trees, but we haven't got much else going for us...

Basically, the problems that plague SSU alignments can also fuck with barcoding. Especially once you wander outside the small comfy familiar home we call Metazoa...

Of course, everyone knows that, but I wouldn't be surprised if people suddenly started worshipping (and publishing) single gene trees based on barcode sequences. Faith in the scientific community seems to be inversely proportional to experience with it...

So yeah, it's exciting, but we must also be cautious. I guess that would be my [rather unhelpful] answer to pretty much everything...

Do you guys have any thoughts?

*Morphological classification led to almost all non-photosynthetic basal stramenopiles/heterokonts being mistaken for something considerably different: Labyrinthulids, oomycetes and Blastocystis were considered to be fungi, opalinids were 'ciliates', actinophryids were 'heliozoa', and there's even a genus called Pseudobodo (a bicosoecid) that was mistaken for...bodonids! Molecular phylogeny continues to churn out one surprise after another!