How to Recognise a Species

On a couple of previous occasions, I've discussed the use of the Biological Species Concept (BSC) versus the Phylogenetic Species Concept* (PSC) - see posts here and here, the first of which defines the two terms if you're not familiar with them. One question that someone, somewhere will always bring up whenever I broach the subject is whether use of the PSC would require the division of humans into separate species, so this post is partially in response to that. From a more general perspective, I thought that some of you might be interested simply in knowing how a taxonomist recognises if they have a new species.

*The more I cover this topic, though, the less I like the term "phylogenetic species concept", because the PSC isn't determined solely, or even primarily, by phylogeny. I prefer the term "diagnostic species concept". Real, honest-to-goodness phylogenetic species concepts, where species are defined by holophyly, have been proposed before (e.g. Donoghue, 1985) and I think the term should be reserved for them. Such concepts have never been particularly popular, though, because frankly they're shit.

The first point, that I think I've said before but can never be stressed enough, is that whatever species concept you are using, a species represents a hypothesis. Sadly, specimens do not come with convenient labels saying to what species they belong. If they did, I'd probably be out of a job (not that I won't soon be out of a job - that thesis deadline looks like it's getting closer every day*). A taxonomist presented with an array of specimens will divide them up into what appear to be distinct species, but that division will always be subject to further revision. Intermediate specimens may be found that blur the boundaries between previously distinct morphologies. Two different "species" may hatch out of eggs laid by a single parent. That sort of thing. No taxonomic revision can ever be final.

*Oh, wait...


Two species of the planktic nudibranch Glaucus (G. atlanticus on the left and G. marginatus on the right). Photo by Gary Cobb (yes, I've used this one before - good, isn't it?).

What most taxonomists are looking for when distinguishing species are discrete, discontinuous characters. Whatever your favoured species concept, ultimately what you are trying to identify are populations that are genetically isolated from others. To give an example, Megalopsalis sp. A is found in Western Australia (I'm calling it "sp. A" here because 'tain't published yet). It differs from other species of Megalopsalis because the femur of the pedipalp (a little leg-like appendage in front of the legs that's used for feeding, visible in the photo below of Protolophus by Robert Pearson) has a row of spines running along the top (dorsal) edge. Over in Victoria and South Australia lives Megalopsalis hoggi, a closely related species. Specimens of M. hoggi never have spines on top of the pedipalp, but they may or may not have spines underneath it (ventrally).

You may be wondering why I regard presence or absence of dorsal spines as a character distinguishing species, but not presence or absence of ventral spines. After all, both are equally distinct. The difference is a matter of geography. In the specimens that I've seen of Megalopsalis sp. A, all of them have dorsal spines on the pedipalp, and I've not seen that character in Megalopsalis specimens outside the range of species A. Specimens with dorsal spines are geographically discontinuous from specimens without, indicating that they represent a distinct population. However, in the case of M. hoggi, specimens with ventral spines on the pedipalp are found in the same locality as specimens without ventral spines. The two are not discontinuous, suggesting that in this case that they are not genetically isolated.



It is possible that the ventrally-spined and ventrally-spineless individuals do represent separate species, but with overlapping distributions (and indeed, western specimens of M. hoggi tend to have spines while eastern specimens tend not to). If so, then study of other character complexes (such as population genetics) may reinforce their distinction. It is also possible that future collections may uncover overlapping variation in the Western Australian population that I've called "species A", and it may not be so distinct after all. But the recognition of two species, one variable and one not, is the most consistent with the information I have on hand so far.

Which leads up to my second point - pretty much any character could potentially distinguish a species, but pretty much any character could also potentially not distinguish a species. The ultimately important factor isn't the characters themselves - it's what they say about the gene flow leading to their current distribution. A single widespread species may encompass more variation within its distribution than a whole complex of closely-related species with restricted ranges. The difference is that character distributions will overlap in the former case but not in the latter.

So to get back to the question of whether use of the PSC would lead to recognition of separate species - personally, not having studied human variation to any great extent*, I don't actually know (really, it's exactly the same question as whether human races have any biological validity, just phrased using different terminology). I suspect that it wouldn't - while humans do vary considerably over their extraordinarily wide range, most of that variation is probably clinal rather than falling into discrete populations. It doesn't reflect underlying barriers to gene flow. This is especially the case in the modern world, where increased transportation and migration is breaking down what geographical barriers there may once have been.

*As organisms go, primates are mostly fairly dull.

As an unrelated postscript, in both of the earlier posts I've referred to a "tendency" of using the PSC to lead to recognition of more species than the BSC. Which got me thinking - are there any situations where the PSC would recognise less species than the BSC? Some possibilities that occured to me were situations involving polyploidy, or cases where interfertility was determined by the action of a single gene such as chirality reversals in gastropods. In these cases, diploid vs. polyploid or dextral vs. sinistral individuals might not be directly interfertile (which would mean they belonged to distinct biological species), but mutation from one state to the other might mean that there was still a reasonably steady gene flow going on (so they would not be distinct phylogenetic species). Any thoughts?

REFERENCES

Donoghue, M. J. (1985). A critique of the biological species concept and recommendations for a phylogenetic alternative. Bryologist 88: 172-181.

Keeping an Eye on Inflation

Neofelis diardi, the "new" clouded leopard species (actually first described in 1823, but later sunk into synonymy until recently resurrected) that has become something of a poster child for arguments on the worth of recent vertebrate species splits. Photo from Tet Zoo.

Normally, there'd be a Taxon of the Week post here on a Monday, but due to various circumstances that won't be happening today. Instead, here's something I've been sitting on for a few days now:

Sangster, G. (in press) Increasing numbers of bird species result from taxonomic progress, not taxonomic inflation. Proceedings of the Royal Society of London Series B.

Despite the widespread (and, I should note, entirely valid) complaints about the decline of taxonomy as a field of research, a person unfamiliar with the dynamics of the field might possibly be forgiven if, on a cursory glance, they saw it as stronger than ever. The number of new taxa, particularly species, being described every day has not appreciably slowed down - indeed, new species are probably debuting faster than ever. In many cases, study of a "species" once believed widespread has lead to the recognition of multiple species, each occupying a different part of the originally-recognised taxon's "range". This latter pattern is particularly noticeable in studies on vertebrates. Often, the new "species" were previously recognised as "subspecies" before their promotion. This has led to the accusation that such cases are examples of taxonomic inflation.


Lesser bird of paradise, Paradisaea minor. Birds of paradise have become one of the prime examples of conflict between the Biological and Phylogenetic Species Concepts. Most authors claiming to follow the BSC have recognised about forty species of bird of paradise (and that total is possibly oversplit, considering the high interfertility of the recognised species), but applications of the PSC to the family have suggested more than ninety species. Photo by stboed.

For those of you unfamiliar with the term, the term "taxonomic inflation" refers to situations where names, but not the content, of taxa change as a result of their elevation in rank. When the term is used, it usually carries the disparaging implication that this is change simply for the sake of change without any underlying gain in information, and hence unnecessary at best and a misleading waste of everyone's time at worst*. Taxonomic inflation has also been cited as a problem with the use of the Phylogenetic Species Concept (PSC) rather than the Biological Species Concept (BSC) (see here for an earlier post of mine on the subject). Many authors have complained that the rise in recognised species is bad news for conservation, as it increases the number of required conservation targets.

*Personally, I (and others before me) would argue that taxonomic inflation is an inevitable side-effect of our current use of a rank-based classification. As I complained last week, an unfortunate side-effect of the pinning of our ranking system on certain primary ranks leads to the belief (whether concious or subconcious) that those ranks should be used for the most "significant" taxa. Now, imagine that I'm a researcher spending five years working on a particular family of organisms. As a result of my research, I find that my "family" of interest renders another "family" paraphyletic. Assuming that I prefer a classification recognising all-monophyletic taxa, I have two options - either I can subsume my "family" of interest into the other family, or I can divide up the other "family" in order to maintain the distinctiveness of my research family. It all depends on my perception of the relative significance of the taxa - and what do you think that might be, considering into what I've been investing the last five years of my life?


Variation in appearance between different populations of giraffe, Giraffa spp. Usually recognised as a single species, G. camelopardalis, recent authors have suggested giraffes should be divided between a number of species. Image from The Barcode Blog.

The new paper by Sangster responds to the claim that the increase in recognised vertebrate species is due solely to the increased popularity of the PSC, and does not reflect any net increase in our taxonomic understanding (invertebrate taxonomy has been less affected by this debate, because invertebrate systematists have, for the most part, been less inclined to recognise "subspecies" [with the notable exception of lepidopterists]). According to Sangster, if this claim is true, it leads to four predictions: (1) the increase in recognised species would have not begun until the introduction of the PSC in the early 1980s*; (2) most taxonomic changes would be based on reinterpretations of old data rather than on collection of new data; (3) most taxonomic splits would be based on specifically PSC-related criteria such as diagnosability and reciprocal monophyly to each other and other taxa, as opposed to less specifically PSC-related criteria such as degree of difference ("too different" to be the same species, or "too similar" to be different species), differences in adaptive zone (e.g. lowland vs montane taxa) or reproductive isolation; and (4) new taxonomic splits would be biased towards members of charismatic groups (in which there may be more of a vested interest in raising their conservation profile). To test these predictions, Sangster surveyed taxonomic proposals affecting taxon rank published in seven major ornithological journals (such as The Auk and Ibis) between 1950 and 2007 - whether proposals recommended a split (subspecies becoming species) or lump (species becoming subspecies), and what were the criteria cited as support for the proposal.

*The date is inexact because a number of variants on the PSC were proposed by different authors.



The results of Sangster's survey showed that the increase in the percentage of taxonomic proposals recommending splits rather than lumps (shown in the graph above from the paper) had been continuous over the time period surveyed, and the rate of increase had not significantly changed in the 1980s (indeed, there had been a slight downturn in the 2000s, though probably not a significant one). 76.4% of taxonomic proposals over the period covered had been supported by new data rather than based on reinterpretation of old data, and proposals for splits were significantly more likely to be based on new data than proposals for lumps (84.6% vs 64.2%). Prior to the introduction of the PSC, 71.7% of proposed splits were based on new data - post-PSC, 93.9% of them were. Only 10.3% of proposed splits overall were based solely on reinterpretation of old data using PSC-related criteria.

The most common criterion for a proposal was diagnosability, followed by reproductive isolation. The least commonly used criteria related to reciprocal monophyly. All criteria were more likely to lead to splits than lumps, though "degree of difference" had the smallest difference in propensity. Proposals based only on PSC-related criteria were more likely to lead to a split than those based only on non-PSC criteria, with the latter more likely to propose a lump than a split. There was no correlation between the increase of species in a family and the charisma of that family, but there was a correlation between the number of splits and the number of polytypic species (i.e. species divided into subspecies) in that family.


The two storm petrel species breeding on the Azores, of which Oceanodroma monteiroi was only described last year in Bolton et al. (2008), from which comes this figure. Though very similar, the two species breed at different times of the year.

Sangster concludes that the accusation that recent increases in species number are based solely on reinterpretation of old data is grossly unfounded. Proposed changes aren't taxonomic "inflation", they're taxonomic "progress". However, as much as I personally like his conclusion, it must be admitted that Sangster is potentially setting up a strawman here. Taxonomic research does not exist in a vacuum, and an author is not very likely to go about revising the taxonomy of a group unless they are working on that group already. The increase in the proportion of taxonomic changes supported by new data might indeed reflect changes in researcher practice - or it may reflect the journals becoming more discerning about what manuscripts they will accept for publication (still, for the reader it may not really matter whether the increased rigour is being driven by the researcher or the publisher).

On the other hand, Sangster brings up the very important point that many of these "new" species aren't really new at all. During the early 1900s, vertebrate taxonomy went through a period of significant lumping. Sangster cites the point that while Sharpe recognised 18,939 species of bird in 1909, Mayr & Amadon recognised only 8590 in 1951 - less than half Sharpe's total. The justification for this lumping was often unclear (they were in a time period when a researcher's authority was generally taken for granted, rather than their being required to show their working), and many current taxonomists feel that in many cases the lumping went too far (for a concrete example, see Darren Naish's post from a few years back on babirusas). Many proposed splits are arguably correcting the excesses of the past.

Finally, on a personal level, I wouldn't particularly care even if the species increase was based on a change in species concept, because as I've explained before (see the first post linked to at the top of this one), I'm a much greater fan of the PSC than of the impractical-to-test, everyone-says-they-follow-it-but-pretty-much-no-one-actually-does mess that is the Biological Species Concept. The conservation and PR argument cited above - that the PSC somehow leads to there being "too many" species - completely fails to impress me, because I don't think that scientific investigation should be directly influenced by political concerns. Once you have the information, then you can work out what to do about it, but changing your working information to what you want it to be first is just not on.

Poor Taxonomic Practice takes some F****ing Liberties!

Bortolus, A. 2008. Error cascades in the biological sciences: the unwanted consequences of using bad taxonomy in ecology. Ambio 37 (2): 114-118.

As you may have guessed from the title of this post, I'm kind of pissed. I got forwarded a copy of this paper earlier today, and was horrified by the results presented in it. Bortolus presents the results of a paper survey of 80 ecological papers published from 2005 to 2007 in ten top-ranked peer-reviewed ecology journals. The subjects of this survey were community ecology studies dealing with more than one species of organism, and for each study Bortolus took note of the sources for the identifications of the taxa surveyed. I hardly need point out how significant such identifications would be for the papers in question - after all, if the subjects of an ecological study are not reliably identified then the usefulness and significance of the entire study comes into question. So you would think that it would be pretty important for the authors to demonstrate the firmness of their identifications, no? Well, let me show you the graph of the results:



Note that the numbers do not necessarily add up to 100%. A single paper might haved used more than one form of identification. I also disagree slightly with Bortolus' treatment of "grey" (not or not significantly peer-reviewed) literature - lumping the citation of field guides in with unpublished theses does somewhat overlook that excellent field guides do exist for some taxa (most notably birds), and even amateurs without formal taxonomic experience may through these field guides hold a significant amount of expertise in identifying the taxa covered (also, of course, field guides are widely available and consultable while theses are not). Nevertheless, let me point out the part of the results that has incensed me the most. It's the second bar from the left. Fully a third of the papers surveyed gave no authority whatsoever for the identity of the species surveyed! It's not as if the bar was set particularly high for citing an authority, either. Acknowledging the input of a taxonomist who identified specimens (or having a taxonomist as one of the authors) counted. Noting that one of the authors of the paper had personally identified the taxa counted, or the use of a cited identification key, counted. All it would have taken in most cases would have been a couple of sentences. And yet, it seems that this is a couple of sentences too many in a lot of cases.

Of course, the availability of voucher specimens would do a lot to alleviate these concerns. I've written before about the importance of voucher specimens that allow other researchers to check specimen identifications. so how many of the papers surveyed noted the availability of voucher specimens?

2.5%.

Yes, you read right. Two point fricking five percent. What is more, those papers that did record voucher specimens were invariably among those that acknowledged the input of a professional taxonomist in some way, and so were probably not among the more suspect class in the first place.

As is de rigeur in this kind of discussion, Bortolus gives us examples of the kind of errors that can result from poor taxonomic practice. Some errors are fairly innocuous, but others can have fairly serious consequences. There's the case of a restoration programme in San Francisco that included the propagation and replanting of the native littoral grass Spartina foliosa. Unfortunately, the actual specimens propagated were not S. foliosa, but the non-native S. densiflora, and as a result of the programme S. densiflora was transformed from a minor adventive to a significant invasive. There's the case of three morphologically similar but ecologically distinct Mytilus mussel species, the decline of one of which due to the invasive spread of another went unnoticed for many years. There's the case of a failure to distinguish Anopheles mosquito species in Burma resulting in the long-term targeting of a non-malaria-carrying species by control programmes instead of the actual malaria vector. Bad taxonomy does matter.

Some of my readers may protest that it seems a little harsh to expect identification sources from authors who may be very familiar with the species involved and hence might be trusted to identify them reliably. To which I reply, bullshit. One of the main principles of science is that it is not enough to simply assert that one "just knows". And besides, is it really asking that much? Biomolecular studies, for instance, are expected to include full details of methods used, and it would not be considered unusual for an author to cite Saiki et al. (1988), no matter how many thousands of PCR reactions that author might have conducted during their career. Also, there are significant ethical considerations involved. In failing to cite or acknowledge their identification sources, authors are doing a significant disservice to the taxonomists on whose work their own relies. It is exactly this assumption that the reliability of taxon identifications can be taken for granted that has resulted in the undervaluing and loss of taxonomic expertise in many parts of the world. Is a passing mention in the acknowledgements really that much to ask?

The Importance of Vouchers: Even Molecular Workers Need Herbaria

Those of you who have seen the Monty Python movie The Meaning of Life may recall a scene near the beginning where a woman is brought into a hospital to have a baby, only to have attention to her condition overshadowed by the attention lavished by the hospital staff on their expensive equipment and machines going "ping". Part of the joke, of course, is that a lot of their flashy machinery is completely unnecessary under normal circumstances, considering that women have been successfully having babies without it for as long as there have been women to have babies. The other part is that while we all see the ridiculousness of the situation in the movie, we still all fall for the flashy expensive option in real life. If we go to hospital ourselves, we still damn want the machine that goes "ping".

The proposed closure of the Utrecht Herbarium is actually just another example of a general trend of recent years towards the downsizing and reduction in prestige of natural history collections everywhere. In these days of biochemical assays and gene sequencers, many people, unfortunately including many university and museum directors, seem to find specimen collections just a little antiquarian and hokey. Surely modern genomic and proteomic techniques have rendered such things obsolete? No, they haven't. Molecular investigations don't remove the need for physical specimen storage. Indeed, they make it even more important than before.

Yesterday, I wrote on the importance of conserving type specimens. Type specimens are a special type of voucher specimen, and I thought I'd enlarge today on the significance of voucher specimens in general. A voucher is a representative specimen of the organisms used in a study, such as the specimen used as the source of DNA for a molecular study or a specimen collected as part of an ecological survey. Like type specimens, voucher specimens allow for the confirmation of the identity of the species referred to in the study. They therefore provide a backup against misleading results due to such things as misidentification, changing species concepts, etc.

Apparently one of the projects that has been going on at the Utrecht Herbarium involves the identification of native medicinal plants used in Suriname, so I'll take that as a hypothetical example. Imagine that a researcher working on one of these medicinal plants, call it the "big-leaved bungleflower", manages to extract an active compound from it that he dubs "ifeelfantasticin", and which shows a great deal of promise in medical tests. The problem is that no-one else is able to extract that compound from the big-leaved bungleflower, despite much effort. Despair seems inevitable, until someone decides to check the voucher specimens that the original researcher deposited from his original study. At that point, it becomes clear that the original researcher was not using specimens of big-leaved bungleflower as he had thought, but actually the similar purple-leaved bungleflower. Focus switches to the latter species, ifeelfantasticin is recovered once more, and medical history is made.

Without the voucher specimens, the error would have never been corrected so easily. Griffiths and Bates (2002) report on a real life example of voucher specimens allowing the explanation of unusual results, when a molecular study on New World vultures was seemingly unable to distinguish between greater (Cathartes melambrotus) and lesser (C. burrovianus) yellow-headed vultures. Re-examination of the original specimens indicated that some had been misidentified, including one specimen that had been deposited in a collection before the two species were recognised as distinct.

Unfortunately, all too many researchers still fail to deposit vouchers for studies (Agerer et al., 2000; Wheeler, 2003). Dennis (1960, as quoted in Agerer et al., 2000) commented that "records that cannot be verified are mere waste paper". The need for proper natural history collections remains as strong now as ever.

REFERENCES

Agerer, R., J. Ammirati, P. Blanz, R. Courtecuisse, D. E. Desjardin, W. Gams, N. Hallenberg, R. Halling, D. L. Hawksworth, E. Horak, R. P. Korf, G. M. Mueller, F. Oberwinkler, G. Rambold, R. C. Summerbell, D. Triebel & R. Watling. 2000. Open letter to the scientific community of mycologists: “Always deposit vouchers”. Mycorrhiza 10 (2): 95-97.

Griffiths, C. S., & J. M. Bates. 2002. Morphology, genetics and the value of voucher specimens: an example with Cathartes vultures. Journal of Raptor Research 36 (3): 183-187.

Wheeler, T. A. 2003. The role of voucher specimens in validating faunistic and ecological research. Biological Survey of Canada (Terrestrial Arthropods) Document series no. 9. Ottawa.

The Significance of Type Specimens, and More on Utrecht

Last week I brought up the topic of the Utrecht Herbarium, which faces imminent closure as of June 1. Since then I have been in touch with Renske Ek, who started the petition against the closure. The closure of the herbarium is primarily a cost-cutting measure, and has been under consideration for some time, though it was not announced officially until the 26th of March. The most significant question you might be asking, though, is what is to become of the collection after the closure?

Let me remind you that, as well as over 800,000 specimens, the Utrecht Herbarium holds over 10,000 type specimens. What, some of you may ask, is a type specimen? When a new species is described, the author will select one of his/her specimens (or sometimes, a collection of specimens) of that species to act as the 'type'. That specimen then becomes the definitive example of that species. From that point, if the question ever arises, "What species does "species name X" refer to?", the answer essentially is, "The species that "type specimen X" belongs to". The type specimen also allows the identification of the species if the original published description turns out to be inadequate somehow. Imagine that a taxonomist working on ants describes a new species. He describes the new species' colour, ornamentation of the body, shape of its head, and calls it "species 1". Many years later, another ant taxonomist discovers that the area where the earlier taxo got his specimens from is actually home to two very similar species. One is the earlier author's "species 1", the other is a new "species 2". The problem is, the two species are identical in colour, body ornamentation and shape of the head. They can be distinguished by the shape of the antennae, but the earlier taxonomist, not realising that the shape of the antennae was going to turn out to be significant, never included it in his original description. How is the later taxonomist to work out which of the two species he has on hand is "species 1"? That is where the type specimen comes into play - even if the original description didn't include the appearance of the antenna, the later taxonomist can always look at the type specimen and fill in the missing details.

The type specimen means that a species name is not just a vague philosophical concept, but is tied to a real tangible object. As a result, it is critical that any type specimen is preserved and remains accessible to future researchers. If a collection is locked away from visitors, the specimens are unable to serve their purpose. Also, a locked collection does not simply sit intact until such a time as it is ever opened. Biological collections require constant maintenance and monitoring. After all, these are pieces of dead plant and animal matter we're talking about - their natural tendency is to decay. Alcohol and other preservatives can evaporate, dried insects or plants may be eaten by beetles, and when the collection finally gets opened up again, all that is left is unrecognisable dust and mush. The collected specimens, and type specimens in particular, have failed in their purpose.

Unfortunately, at present there seems to be no clear plan for what is to happen to the Utrecht herbarium collection after closure. Plans to transfer it to the Universiteit Leiden appear to have been shelved due to lack of funding. It seems to have been hoped that the Nederlands Centrum voor Biodiversiteit (Dutch Centre for Biodiversity) currently being put together would offer a permanent home for the collection, but it is unlikely that the Centre will be up and running before the herbarium closes.

Whatever is to become of the Utrecht Herbarium itself, it is imperative that the collection, particularly the type specimens, remains curated and accessible somewhere. Failure to ensure this would be an unforgivable tragedy.

Postscript: David Shorthouse of iSpiders has also commented on the situation.

What are the Bare Necessities?

Valdecasas, A. G., D. Williams & Q. D. Wheeler. 2007. ‘Integrative taxonomy’ then and now: a response to Dayrat (2005). Biological Journal of the Linnean Society 93 (1): 211-216.

Before I start this, I feel I should perhaps slap a 'Parental Advisory' warning on it. The link to this article turned up in a Table of Contents e-mail alert a moment ago, and it's gotten me feeling a little steamed up. Not by what was said by Valdecasas et al. - I agree with almost everything they had to say - but with the article they were rebutting.

The question behind both articles, ultimately, is the current Taxonomy Crisis - essentially, the fact that there are just too many undescribed species and not enough work being done to identify them. I've written about it before (or just click on the "principles of biodiversity" label attached to this post), so I'll refrain from explaining in detail again. Dayrat (2005) argues that in order to escape this crisis, we need a significant rehaul in how we do taxonomy. However, as Valdecasas et al. (2007) point out, some of his suggested 'solutions' would probably end up doing more harm in the long run than good.

Dayrat's call for a more 'integrative taxonomy' translates into the call for taxonomists to work more closely with molecular ecologists and population biologists in establishing species boundaries. In this, he is presenting us with a bit of a truism. It is certainly true that new species should be established on the basis of as much data as possible, and I don't believe any working taxonomist would argue with this. However, this should be an enabling, not a limiting, factor. As much as we would all love to have all that wonderful data when conducting our species revisions, the simple fact is that sometimes (probably even more often than not) we can't count on it. In such cases, surely it is better to go ahead with what data you have available, rather than allow your work to languish indefinitely while you wait for extra data that may never show. Valdecasas et al. provide a quote from Bonde (1977) that rather summarises my feelings on the subject: "An important aspect of any species definition whether in neontology or palaeontology is that any statement that particular individuals (or fragmentary specimens) belong to a certain species is an hypothesis (not a fact)" (emphasis my own).

Dayrat complains that the 'over-abundance of redundant species names' (synonyms and nomina dubia) is a major impediment to taxonomy (a nomen dubium, for anyone not in the know, is a taxonomic name for which not enough data is available to firmly establish to what species, etc. it originally referred). His implication is that to much time is being wasted on sorting out these 'redundant names' that should be spent on describing new species. It is true that the ICZN currently does not officially allow for a potential nomen dubium to be set aside for a better characterised later name, but there is a very good reason for that - namely, that the recognition that a name is a nomen dubium is entirely dependent on context, and to a certain degree on author preference. There is no objective standard to what constitutes 'enough data'. What is sufficient in one case may not be in another. Besides, Dayrat is overlooking the historical context of many, if not most, nomina dubia. While they may not be regarded as identifiable now, at the time they were described many such taxa were different from all that had been known to that date, and the original authors can hardly be blamed for failing to predict that their 'unique' new specimen would turn out to be not so unique. I find it a little ironic that Dayrat complains of the "typological approach" of taxonomists, yet speaks favourably of DNA barcoding, a more typological method than almost anything commited by any morphologist (well, except Carl-Friedrich Roewer).

For instance, it is a general rule of thumb in vertebrate palaeontology that taxa based on reptile teeth are not diagnostic, because teeth do not vary enough between species of reptiles. Nevertheless, Troodon formosus (shown at top in a reconstruction from here) was originally based on a tooth, and yet it can be identified as a species with other specimens because there happens to be only one species known from the type locality with that kind of teeth. Of course, the possibility always exists that another species will turn up in the same locality with the same sort of teeth, in which case the identity of the original Troodon becomes uncertain. But I feel that there's little point in playing such 'what if' games - I refer the reader back to the Bonde quote above.

Let's move on to the details of Dayrat's recommendations:

"No new species names should be created in a given group unless a recent taxonomic revision has dealt with the totality of the names available for the group." Valdecasas et al.'s primary response to this is "define recent". Many groups have not been substantially revised for a great many years (for instance, the last complete review of many harvestman families was probably Roewer's Die Weberknechte der Erde in 1923). Still, many of these ancient revisions are still considered very reliable due to the thoroughness of the original author. Even when there isn't a good revision available, this may not be an impediment to taxonomic work. Revising an entire group is a major task, and making it the minimum expectation will simply cause researchers to shun poorly studied groups, leaving them to languish in their taxonomic pits. The taxonomic crisis would then worsen rather than improve.

As a point of contrast, I have commented before on the systematic quagmire of South American harvestmen. For a number of years now, a colony of South American arachnologists have slowly been chipping away at this heaping mess, sorting out issues where they could. As a result, parts of the picture are slowly coming into view, and there is hope for more unravelling to come (e.g. Kury, 2003). It is true that some findings have been published as separate papers that could have arguably been included in the same publication (an artefact, I believe, of an academic accreditation system that values total number of publications to the individual quality of said publications), but overall the value of their work is, I think, unarguable.

"No new species names should be created if the infra- and interspecific character variation has not been thoroughly addressed." This is directly connected to the following recommendation: "No new species names should be created based on fewer than a certain number of specimens (a number which specialists of each group could agree upon), and never with a single specimen." Again - how many is enough? Is a species based on a single, very distinctive specimen necessarily less reliable than one based on a number of specimens but very similar to another, already named specimen? Valdecasas et al. argue the cases of fossil or endangered species, where the researcher may not be able to count on obtaining further specimens. If you have a distinctive and probably new taxon, surely it is better to bring it to attention rather than, again, letting it languish? Which brings us to:

"A set of specimens differing in some regard from existing species can be described with the abbreviation 'sp.' (for 'species') and not with a real species name regulated by the codes of nomenclature." Technically true, but not so hot in practice. Dayrat is overlooking that 'sp.' is often used in species lists to indicate specimens that can be identified to a genus but not to one of the species within a genus - usually, with the implication that the specimens cannot be reliably identified. Also, Dayrat is overlooking the power that names have when catching people's memory. Things named 'sp.' slip into obscurity, while taxa with actual names hold their place. Witness how even when a taxon becomes widely known (say, as a conservation target) before it gets officially described, the need to label it with a tag is inescapable. The Mahoenui giant weta was appearing as 'Deinacrida sp. Mahoenui' long before Deinacrida mahoenui was officially published. And because there are no regulations governing the use of such informal tags, they cause more confusion if allowed to persist without a suitable official replacement than otherwise.

"Ideally, names should only be created for species that are supported by broad biological evidence (morphology, genealogical concordance, ecology, behaviour, etc.)." Again, this is true, but how broad? As Valdecasas et al. point out, there is no upper limit to how much data could potentially be collected. Again, surely it's better to highlight the fact that an interesting new taxon potentially exists than to allow it to remain hidden?

"No new species names should be created if type specimens deposited in a museum collection are preserved in a way that prevents any further molecular study." Also: "All neotypes designated from now on should be preserved in a way that allows DNA extractions and sequencing." Aaaaaaaaaaaargh! NO! Dayrat is overlooking that retaining specimens for molecular analysis often renders them unusable for other forms of analysis. If I was collecting arachnid specimens, I would have to decide on collection whether to put it into 70% alcohol or 100% alcohol for preservation. In 70% alcohol, the specimen's morphology remains preserved, but DNA degrades fairly rapidly because of the presence of water. However, if I was to put the specimen in 100% alcohol, the absence of water dessiccates all membranes, rendering the specimen brittle and immobile, unusable for morphological analysis. The problem is even worse for entirely soft-bodied animals, which may shrivel in 100% alcohol to unidentifiable lumps. In the case of arachnids, the researcher may get around the problem by preserving most of the specimen in 70% alcohol, but remove a couple of legs and put them in 100%. But what would s/he do if s/he was working on rotifers, or some other minute organism? The appropriate means of preservation should be dictated by the requirements for identifying the particular taxon, not by any theoretical standard.

Ultimately, the problem with Dayrat's suggestions is that he is confusing the position of the donkey and the dray. Ultimately, the recognition of a new species is a hypothesis based on a collection of data to be tested by further data, not a data point in itself. If a set of objective 'minimum requirements' is imposed, it will probably have the negative effect of discouraging research and publication, and the taxonomy crisis will worsen rather than improve. 'Minimum requirements' may sound good in theory, but practical considerations speak otherwise. As Valdecasas et al. note, "In any case, experience demonstrates that it is far more detrimental to be a ‘lumper’ than a ‘splitter’, in contrast to Dayrat's assertion. If variation previously assigned to two species turns out to be more economically assigned to one, synonymy of subsequently identified specimens will easily solve the problem. When the same name has been given to specimens that exhibit enough variation to include several species, it is considerably more difficult to recover in subsequent work which identifications correspond to which species unless there is a repository of all specimens; and that is not always the case, as it is very common in much ecological work to refrain from preserving the specimens upon which the identifications were based, leaving no option for further rigorous identification." The taxonomy crisis will not be resolved by imposing outside standards (which incompetent workers will ignore anyway), but by knuckling down and describing species.

REFERENCES

Bonde N. 1977. Cladistic classification as applied to vertebrates. In Major Patterns in Vertebrate Evolution (M. K. Hecht, P. C. Goody, & B. M. Hecht, eds.) pp. 741-804. Plenum Press: New York.

Dayrat, B. 2005. Towards integrative taxonomy. Biological Journal of the Linnean Society 85 (3): 407-415.

Kury, A. B. 2003. Annotated catalogue of the Laniatores of the New World (Arachida, Opiliones). Revista Ibérica de Aracnología, special monographic volume 1: 1-337.

The Phylogenetic Species Concept: is there such a thing as too much?

A recent editorial in the open-access journal PLoS Biology (you can tell it's online because of the acronym title ;-P ) comments on the recent widespread adoption of the Phylogenetic Species Concept as opposed to the Biological Species Concept, and the resulting 'taxonomic inflation' in the number of recognised species around the world. The authors feel that this is a Bad Thing, because increased number of species by definition means more endangered taxa, which means less effective conservation. This is not the first time this point has been made, and I feel the need to put my own oar in on the subject. To do this, I'm going to have to address the two arguments that are really at stake here - I'm going to call them the idealistic and the pragmatic sides of the argument.

The idealistic point is that we should not be using the Phylogenetic Species Concept in recognising species. I wrote a page on Palaeos a while back explaining some species concepts, and I'll refer you there if you want a more detailed description. For now, it'll suffice to say that the 'Biological Species Concept' (BSC) defines a species as a cluster of individuals 'potentially interbreeding to produce viable, fertile offspring', while the 'Phylogenetic Species Concept' (PSC) defines a species as the 'smallest diagnosable population of organisms sharing a distinct pattern of descent'. The latter concept has also been referred to as the LITU (Least Inclusive Taxonomic Unit) by authors who do not necessarily feel that an LITU is equivalent to a 'species'.

One side-effect of using the PSC rather than the BSC is that the latter is indeed generally predisposed to recognise more species. Under the PSC, there is really no place for the concept of 'subspecies' because, after all, if a population can be distinguished as a subspecies then it fulfils the requirements for a species. For vertebrates in particular the subspecies rank has a long history of use, and so application of the PSC results in all those subspecies becoming recognised as species. [Invertebrate workers have, on the whole, been less inclined to recognise subspecies (with the notable exception of lepidopterologists), so this is less of an issue there. I'm not familiar enough with the situation in plants to comment.] Also, there is no theoretical lower limit on how fine distinctions between 'species' can be, so long as those distinctions are accurate.

The BSC, however, has its own problems. Despite being the most commonly quoted species concept, it is usually recognised more in the breach than the practice. The requirements for actually testing whether populations can interbreed soon become prohibitive, and the BSC is difficult to apply in the case of populations that are separated geographically (to ask a strawman question, are two individuals 'potentially interbreeding' if they're never going to meet?) Recognising species using the BSC, therefore, generally depends on the individual author's own subjective judgements about whether or not two populations are conspecific.

In my opinion, the balance weighs overall in favour of the Phylogenetic Species Concept. In practice, the PSC is more objective and falsifiable than the BSC. As an extreme example of taxonomic inflation, the authors of the PLoS Biology editorial refer to Merriam's 1918 classification of North American brown bears into 82 species, and quote George Gaylord Simpson's critique that the distinguishing characters of these species were so finely defined that "On such a system twin bear cubs could be of different species". There are two replies to this: (1) why not recognise 82 species if there are 82 distinguishable populations?, and (2) Simpson's complaint indicates that Merriam's species would not be valid under the PSC, because the PSC requires that a species be a population (the 'distinct pattern of descent' referred to above). If a character varies within a population to the extent that it overlaps with other populations, it is not usable in distinguishing them as species.

In regards to the pragmatic side of the argument, the PLoS authors have more of a point. As the number of species supposedly requiring conservation increases, funds to conserve them become more stretched and the effectiveness of conservation goes down. However, the practicalities deriving from species recognition should be divorced from species recognition itself. Populations of organisms exist in nature independently of whether or not humanity chooses to recognise them, and the role of science is to try and identify what is, not what should be. It is true that as the number of recognised species increases, it becomes increasingly evident that we will not be able to conserve them all. Nevertheless, the appropriate response is probably to try and work out how to direct funds in such a way that the maximum amount of biodiversity is saved. This will probably require some sort of prioritising of conservation targets rather than attempting to save everything, and thinking in terms of ecosystems and taxon groups rather than individual species. It is still better that we keep a clearer view of what is out there than masking diversity within expansive categories.