My background on the edge of mycology gives me a slightly different perspective (that might be wrong!). For the majority of mushrooms classical morphology and ICBN nomenclature is still a principle driver. That classical view is increasingly informed by sequence data, both by barcodes for identification, and by multi-gene sequences for phylogenetic reconstruction.

However, fungi are a big group. If you focus on cryptic micro-fungi of economic importance then you get a different picture. Here you find a different approach has emerged due to limited morphology, and/or the limitation of gross morphology as an indicator of pathogenicity.

For example, the current global epidemic of wheat stem rust, which is formally defined as a race of Puccinia graminis, is now more universally, but informally known as ‘Ug99′ (for Uganda 1999), or its informal synonym ‘Pgt TTKS’ (for Puccinia graminis f.sp. tritici race TTKS [I don't know what TTKS stands for]). Just Google Ug99 and you’ll find many of the pages won’t even refer to the formal name at all.

My view is that Ug99, and its synonym Pgt TTKS, are ‘effectively’ barcode names. If a strain has the resistance markers for Ug99 then it is Ug99. In addition, all the valuable information around the world is linked to the namestring Ug99.

Plant pathologists working on fungi, bacteria and viruses have adopted this approach for a long time now. In the pre-molecular era the plant pathologists used ICBN acceptable designations such as ‘Puccinia graminis formae specialis tritici’ to indicate physiologically distinct races without morphological distinction (c.f. pathovar in bacteriology). Now we commonly see informal name strings, with implied barcoding, such as ‘Ug99′ or ‘Pgt TTKS’ to pin down the concept.

Also, formal virus names such as ‘Influenza A subtype H5N1’ have embedded ‘barcoding’ – with implications for the viral species concept definition.

How different are these extant working practices to your proposal for inverting the notion of identity so it’s is based on a ‘barcode’ identifier rather than a mental construct?

I know even less about the domain of microbial ecology but I do see the enormous number of unassigned fungal environmental sequences in GenBank. It can only be a matter of time before humans start inventing namestrings for clusters of such sequences (which are effectively defined by barcodes) that are remote from anything with a Latin binomial. Maybe it already has started and I haven’t noticed. They’ll do that because they will want to link those barcode defined namestrings to associated data.

The issue I see is that systematsist’s traditional working practice compounds two fuzzy sets. The fuzzy set of living things ‘out there in the real world’, and the fuzzy set of namestrings we’ve applied to the fuzzy set of living things. Your inversion removes the fuzziness but I don’t see how it solves the inevitable problem of linking old with new.

The challenge for biodiversity informatics is to provide the universal linking/disambiguating mechanism for all these namestrings, their associated defined species concepts (fuzzy or not), and all the associated data.

Rich wrote:
“The goal, in my mind, has never been to get a computer to do the work of a taxonomist; it has been to make the job of the taxonomist easier and more efficient”

I think I agree with you Rich to a degree, as this is probably the easiest, quickest, most advantageous use of IT tools and processes. BUT imagine how useful it would be to be able to get a computer to identify specimens, suggest new taxon names/concepts (my “Organism Concepts”, Roger’s “Barcode Taxa”), map relationships between organism concepts, pick up on discrepancies, etc, etc!

I agree that our needs are evolving; and perhaps they are evolving faster and more substantially now than they ever have before. This is why I am not opposed to alternative strategies (I’m not a Phylocode-basher or a Barcode-basher). They are certainly worth pursuing, in my opinion; as long as they are fulfilling a real need that is not currently being met by the existing system.

However, I just don’t see these alternate approaches to organizing and naming biodiversity as “replacements” to the existing system, because the existing system continues to fulfill certain needs as well.

I worry about your sunk losses argument i.e. just because we have been doing it this way for so long does not mean it is the best way to do it in the future.

The other worry I have is with the definition of what works. Neither you or I are the best people to say whether taxonomy works or not. Really we need to test such an argument with users of the system. This assumes we are not doing taxonomy just as an end in itself – which may be justified.

My rationalization of this is that taxonomy (and associated nomenclature) has existed as it has for the past 250 years because it serves a useful purpose for humans. As such, the primary client of taxonomy is, has been, and should continue to be optimized for humans. Biodiversity Informatics should not, in my mind, be about automating the intellectual process; but rather it should be about maximizing the efficiency of information access by humans.

You wrote: “The notion that we can scale taxonomy by increased use of information technology is erroneous, at least in part, if there will always be a bottleneck of overworked taxonomic experts.”

If we want taxonomy to continue to fulfill the role it has fulfilled for two and a half centuries, then I think we are a long, long way from the time when electronic intelligence will be able to supplement, let alone replace, the intellectual process of taxonomy. However, the bottleneck can be opened rather dramatically through the application of technology and biodiversity informatics. Having observed my PhD advisor engage in the art and science of taxonomy (he is the most prolific ichthyological taxonomist in the past 100 years or so), I noted early on how poor was the ratio of time he spent implementing his expertise, vs. time he spent manually typing and formatting diagnoses, synonymies, keys, bibliographies and tables of character data for manuscripts; looking up particular publications in his office library; going to a larger library to access some publication that he didn’t have his own copy of, etc. All told. he probably spent 15-20% of his time doing stuff that capitalized off his taxonomic expertise, and the rest of his time conducting tasks that are tedious and slow for a human, but incredibly east and fast for a computer. In fact, it was watching him work that led me down the path of biodiversity informatics in the first place — I knew that computer technology could streamline the process dramatically. And I still believe it can!

I disagree that taxonomy — as it now exists — cannot fulfill the needs of tasks like climate modelling; because I think such tasks do not require the level of precision of taxon-concept comparison that you seem to be striving for. By itself, nomenclature can allow automated reasoning to the resolution of primary type specimens, at least.

There is no denying that DNA Barcoding (at least as a concept) has real merit and value. I would say the same thing about the Phylocode approach to nomenclature. The important thing to remember is that where they have their strongest merit and value does not exactly coincide with the reasons why Linnean nomenclature has been so successful over the past 250 years. In other words, a system such as what you describe will certainly allow for certain things that traditional nomenclature is not optimally designed for. But it will fail to fulfill certain things that the traditional system has been (and continues to be) extremely effective for.

Addressing the first sentence of your last paragraph, I think that the purpose of taxonomy is to produce a system that people can actually use to *communicate with each other*. This involves much more than simply hanging data from. And even when data are hung on names, they still fulfill most of the need, most of the time, at the level of precision that most people care about (this includes things like climate modeling).

This is not to say that the approach you describe does not have merit. For its purposes, I think it has merit (see also Godfray, 2002, Nature 417:17-19; http://www.nature.com/nature/journal/v417/n6884/full/417017a.html — and the basic tenets of these ideas have been around quite a bit longer). Just don’t think of it as a replacement to traditional taxonomy.

I am taking a slightly naive approach to barcoding and assuming they are reproducible and reliable all the time.

My remark on not needing museums at all was a bit of showmanship perhaps. We would need specimen repositories.

So is it your contention that Barcodes used for identification need not be associated with specimens, nor have any repeatability – that the concept of type specimens is somehow unnecessary with barcoding? The folks who are actively building a repository of “barcode types” and those supplying specimens for that project would strongly disagree.