CLADISTIC THEORY

For many years, naturalists constructed family trees of plants, animals and lower forms by seemingly rather arbitrary whims of biologists.

Eventually, some came to see this as rather "Unscientific", and tried to construct a more objective system based on automatic processing of data.

Their approaches are based on the notion that closely related forms (eg animals) show more similar features than less closely related ones. Of course, this is the basic idea behind any reconstruction of lines of evolution, but as usual, the devil lies in the details.

Cladistic theory starts with the "Clade" - a group of forms (eg. creatures) sharing a particular characteristic. For example, once upon a time, when life was simple, all creatures sharing the feature "feathered" were placed in the "Bird" clade.

Another clade was all creatures whose eggs or developing young are held within a membrane called the "amnion". This clade is the amniotes, and includes "reptiles", birds, dinos, mammals etc but not the artists once known as "amphibians". (Also "reptiles" may not represent exactly the animals you had expected).

[picture, with "have amnion2; "have feathers"]

Note at this point, the first flaw in the theory: How do you decide what feature(s) to choose when defining your clade? This may be a trivial choice when the group is well defined. In a straightforward situation, all methods tend to give the same results, so there is no problem. But what when things get more complicated?

Complicated? In what way?

Well, lineages of animals have a way of evolving a feature (or character), then removing it, and then re-evolving it again, in a way they have often had to be spoken to about. What's more, closely related parallel lineages can simultaneously do the 'now you see it- now you don't' trick, sometimes in and sometimes out of phase with each other.

It turns out that often we cannot find any one feature that appears at one point in the tree and stays - and if we can, we're not always happy to rely on it, partly because we're not sure we trust it and partly because for a complex tree you need a lot of characters. Unfortunately these tend to disagree with each other.

[picture of clade with a lot of "on/off" stuff] This tends to happen with dinosaurs - we have a very few fossils, separated by long time spans. Over long periods of time, such character disappearances and reappearances (or reversals) are all too frequent.

In such cases, how do we decide which characters to rely on?

Enhancements to the basic cladistic theory make their appearance here but the unavoidable fact remains - even cladistics makes arbitrary choices about which characters to base the tree on.

Ok, what tricks does cladistics have up its sleeve when we have to rely on several characters which all suffer from reversals?

The first is "Occam's razor" - briefly, always give the simplest explanation most credence. (This principle is more often known as "Parsimony".) This can be a useful principle, but like any razor in use in the middle ages, if you're not going to use it properly, it's best not to play with it at all! Professor Bondi once told me (alright, along with a few 100 others) that when he went to work on radar in the war he was told he only needed to know Ohm's law - but he had to kow it Really Well! It's a good job those radar workers knew Ohm's law better than some people know Occam's razor!...

...because it is a guiding principle (any number of violations in nature can be provided) and NOT to be treated as an invariant truth. (It's much more useful for discovering laws of physics than evolved biological details.) In any case, problems can usually be posed where two razors are in opposition. Besides which, supporters of cladistics are surprisingly willing to chuck out any razor as soon as it stops cutting their way - example?: creatures predicted by cladistics not to be descended from birds are found to have feathers. What do we then do with the simplest idea that "all feathered creatures are birds" or even "all feathered creatures are descended from the earliest known"? "Oh - er - no problem - we have other razors somewhere in our theory we like better!"

How do cladists misuse their razors?

First, by assuming the family tree with the minimum number of changes is the right one. [[###name they use for this principle?]] Obviously, for sparsely populated trees spanning few fossils but lots of time, the tree we come up with, with the smallest number of reversals is likely to be wrong. Cladists appreciate this (though not as often as they should) and they try to deal with it:

They somehow estimate the "most likely number of reversals", then choose a tree close to this number. Unfortunately, even if this tree did have a high relative likelihood, its absolute probability is actually quite small, and anyway, meaningless. Probabilistic methods work best when a range of probabilities is employed, not a particular point along the curve, which usually has a very small absolute probability.

[picture of curve with small but "best" vertical slat of proability]

But anyway, how reliable is this "most likely number of reversals" in the first place? If we're talking about random changes in DNA or protein, there is some basis for estimating likely numbers, but when it comes to shapes and sizes of bones, the "most likely number of changes" is unknowable.

Is the result of choosing a tree with the "most likely number of reversals" meaningless? How could this be tested? That's hard to do on entirely fossil lineages. We could try it on family trees where:

1) We know the right tree - eg very detailed sequences of ancestral forms are known with no gaps, over the hundred million years or so comparable to our dinosaur problem.

2) The characters are related to shapes and sizes of bones, not molecular data, and it would be nice if the creatures concerned were vertebrates, not eg bacteria.

Not surprisingly, despite a few efforts, the validity of the cladistics used on dinosaurs is far from proven.

As a result, arguments pro and con are largely theoretical.

(CIRCULARITIES ########)

Can we learn anything from what we might call the "sociology of cladistic support"? It doesn't take long to notice that while cladistics is very popular in the litereature, in the pub it drops several popularity points. A cynic might say (and I most certainly do!) that cladistics is fashionable, and confers an aura of scientific respectability. For the career minded scientific paper writer, being cladistic is as important as being prolific. Morals do not operate at certain levels of the subconscious!

I'm not going to do a massive survey on great names that are and aren't in the cladistics camp, but there is one I might mention. (For this, a "Homologue" or "Homology" is a similar character shared by two or more creatures. In a recent letter to "Nature"...######

It would not be beyond the wit of Strangetruther to provide references to a few guides to cladistics, but since no cladist ever references theories in the BCF family, we'll give them a taste of their own medicine. After all, cladistics is just another unproven highly speculative theory with no sound theoretical basis!