May 11, 2008

The perils of morphology and taxonomy

Now that im actually free from the clutches of undergraduatedom, i've got a few things i've really been wanting to say for the long spitfire comment thread on 25th April. I'd only dare comment on this as i've been on a bloody year long project reconstructing the phylogeny (family tree) of a whole family of 30 flies primarily based on morphology, and i've found out that using morphological characters isn't as straightforward as it seems.

To elaborate on Ria’s point, our definition of ‘species’ is exists only as a human construct, an arbitrary term for a certain population of organisms ‘delimited’ by features which we define. Traditionally, these features are derived from morphology – i.e., physical features which we can see, such as spikes or bumps, or even color. The best kind of feature (morphological character) would be one that is discreet: They can clearly be defined as one state or another, usually as ‘absent’ or ‘present’, or ‘3 spines’, ‘5 spines’.

Unfortunately, a majority of characters are not discreet, but rather, continuous: there is a spectrum of states with no clear distinction between the states. These can be like the ‘hairiness’ or length of a body part. Even seemingly discreet characters can become continuous – should a medium sized protrusion be considered a large bump or a short spike? Such characters can also be at the mercy of other factors. Some color characters observed in the field cannot be strictly considered as discreet – color is the result of pigments generated by the organism, itself subject to a whole range of other factors. Diet, or the lack of, can contribute to a dull color due to malnutrition. Otherwise, the organism may be feeding on a food source that may make it allocate certain pigments – like us humans turning slightly orange when we drink far too much carrot juice. The environment itself plays a huge role; look at a fair person who sees no sun compared with one who spends all his time in the sun – are these two people different species then?

Another important issue is called ‘homoplasy’, where characters may have different evolutionary origins but arrive at the same character state. For example, both sharks and dolphins have dorsal fins, but the genetic makeup and structural origins are totally different.

In the case of the Cyrene urchin, it was alleged to be a Diadema savignyi due to its shorter spines, while the orange periproct suggested it to be otherwise. In this case, these two characters alone are probably by no means adequate for an absolute confirmation. In fact, even morphological characters alone may not be able to solve this issue to a 100% conclusion. Historical context, genetic data, reproductive isolation may have to be invoked. Each however has their own problems. It may be hard to reconstruct the urchin’s historical and geographical context, since there are few prior records in this region and the movement of larvae is hard to determine as well.

Genetic data often comes with problems of its own: It is highly subjective to different analytical methods, and even then, genetic evidence for species difference exists as a predictor – as long as genetic sequences are over x% different, they are considered different species. This x% is arbitrary and may be different for different groups of animals. The sampling of DNA data is also usually a minute percentage of an organism’s entire genetic data; usually 1-2 genes out of a total of thousands of genes are used in a genetic analysis.

As for reproductive isolation, different species are determined by not being able to interbreed with each other. This is not a satisfactory solution either, as both tigers and lions can interbreed – are they the same species then? Same goes for urchins – the papers i’ve read suggest that a number of species can interbreed.

As you can see, plenty of ways exist for us to define what a species is, and none of them are failsafe methods. Even amongst academic circles, there is intense debate on species delimitations, which methods are better, and in fact, at least five camps who define species in five different ways. I wont even try to go into cryptic species complexes! Thus, it is very dangerous for us to just use just one or two characters from a type of data (morphological) to absolutely ascertain a claim. In exaggerating, it is akin to feeling the tail of an elephant and claiming it to be a thin snake.


One more thing. I did read through the paper on the taxonomy of the Diadema urchins, and found a few odd things. For one, the authors utilized a large number of continuous characters (usu lengths, percentages that were in a range), often which overlapped each other. As mentioned earlier such characters should be avoided in a rigorous morphological study. Secondly, while i'm not familiar with the form of analysis they performed on their data, i do know that cluster analysis does not take into account evolutionary history – it basically compares similarities between species, does not trace the evolution of that character itself. I am surprised that they didn’t use other analysis methods as parsimony, likelihood etc, which would take into account such things. Hence the paper while extensive in its morphological scope, does not seem to provide a robust analysis. This, of course, is just my view, i may be missing out on some of their arguments, or maybe their focus is elsewhere. Anyone with a better idea please enlighten me. Then again i am a systematist by nature; i just feel it would make more sense if it included other analytical methods.

I also did a check on the paper – it's from the journal Zoosystema, of impact factor (IF) 0.714. This means that an average it has an average of 0.714 citations per paper per year. It isnt very high at all. Of course, IFs exists as a proxy gauge for the academic world to approximate its importance, and may be prone to some statistical problems, but it still exists as a useful gauge. More info can be found at wiki. Anyway, the paper itself is only cited twice since 2006, but of course this is dependent on the field of interest as well.

No comments: