Your attitude to numerical taxonomy

This is you about the cladistics as I understand it? There is something in this, but in general-complete nonsense, IMHO, after all, the signs are unequal. And there really are a lot of such works. And there are Russians.

So far, the machine cannot adequately assess the weight of traits, their actual significance in evolutionary terms. In this regard, it turns out either nonsense, or a lot of options that also do not fully reflect the real picture. Fashion for computer modeling, you know, and it, modeling, unfortunately, is still pseudoscientific. Maybe someday...

"Objective" clustering based on subjectively selected features? IMHO absurd.

Unfortunately, too often classifications (and even in biology in general) use the system not because it is correct, but because it is convenient and gives accurate results. But accurate doesn't mean correct.
To be honest, I've been very disappointed in taxonomy lately.

In my opinion, there are too many empirical conclusions.

I'm not familiar with the method, but I believe that any method has the right to exist. True, will determine the time.
Scientists should not constantly agree with each other, on the contrary, in disputes (including methodological ones) they will probably determine the truth.

To be honest, I was surprised by the results of the vote. It would be like asking a survey "Do you use a mouse when working with a computer" and most people would say "Mice are not yet perfect enough to be useful" or "This is the first time I've heard of such a device..."
Numerical methods of taxonomy have become deeply ingrained in science over the past 40 years, and only Russian biologists are out of step. This is very disappointing, especially since such famous Russian scientists as A. A. Lyubishchev (entomologist), E. S. Smirnov and N. A. Zarenkov, as well as the equally famous W. Hennig (also an entomologist!) are at the origin of these methods.
In addition, we should not forget that numerical systematics is not limited to the construction of trees by computers, but includes many different methods, including the so-called "phenetic" methods (in fact, methods of multidimensional statistics). Finally, there is still a large amount of "manual", creative work: choosing an external group, choosing a method, compiling a matrix of features (the most important thing!), analyzing and testing the results.
And how can I vote in such a survey if I think that numerical methods are fast and convenient, and at the same time are equivalent to other methods?

In my opinion, there are too many empirical conclusions.

Numerical methods are designed to avoid empiricism.

E. S. Smirnov is also an entomologist.

"Numerical methods are designed to avoid empiricism. "
Undoubtedly, the numerical method is not speculative, but it should also be adequate. I will repeat again that the exact result does not mean the correct one.

I agree with the Great Helmsman that all flowers should grow. As some expert in numerical modeling, I will defend the method. You need to soberly evaluate the pros and cons of different approaches, and not paint everything in black and white.

Do not confuse cladism (including Hennig's) with numerical taxonomy. And the use of morphometry in taxonomy (Lyubishchev) is also not quite a subject. As for the formalized approach to taxonomy, IMHO, there is a big problem here. What do we want to measure? Some abstract "similarity"? Time since the branch of the evolutionary tree branched? Something else? But if you consistently apply a certain principle , the entire system of living things, especially taxonomic ranks, will be distorted so that you will not know for yourself.

What is the difference between the taxonomy developed by a scientist and that obtained on a computer? That's right, full information. If the numerical approach is bad precisely because of the one - sidedness of classification based on a small number of attributes, then this part of it needs to be improved. How? "that's another question. After all, no one is talking about the hegemony of one approach - "uniqueness" is lobbied by supporters of different schools that receive funding. Let's look at the test system in education, which is often criticized for its imperfection, but which excludes subjective assessment of the result. Such a system works, for example, in the United States, but it is quite closely intertwined with essays, individual projects, and so on.

This post was edited by Nilson - 06.06.2006 09: 02

>What is the difference between the taxonomy developed by a scientist and that
obtained on a computer? That's right, full information.

Not just fullness. It also provides an estimate of the taxonomic weight of attributes. It is clear that a person who does not use formalized methods, weight is assessed subjectively. But will the criteria laid down in a certain algorithm also be better by a person? Another question is what to do with correlating features? Especially if we don't know if pleiotropism is here or not.

2 PVOzerski: if a subject is not a cladism or a phenetic taxonomy, then what is a subject?

If a subject is not a cladism or a phenetic taxonomy, then what is a subject?

The cladism in its original form (in the sense of Hennig) was devoid of a numerical component. So the Subject is more likely not about numerical systematics, but about numerical methods in taxonomy.

"Let's look at the system of tests in education, which is often criticized for imperfection, but which excludes subjective assessment of the result. "
This is a painful question for me. I firmly believe that testing is bullshit and originated from the laziness and incompetence of teachers. Without going into details, I will say that this is an objective assessment of subjective quality. That is, the same as the numbering system. I do not want to say that this method is completely absurd, but it should not be abused: you should not turn systematics into a formal system based on subjectively selected criteria (in my opinion, there is no way to take them all into account in reality) - it is better to forget about this science altogether.

IMHO, it is better not to talk about newfangled trends in education-firstly, because this issue is painful not only for one resident of Israel, but also for a huge number of residents of Russia, and secondly, because it is completely off-site. And as for the subject - again, IMHO: if we switch to these rails (i.e., to strictly formalized principles), this step should be accompanied by a reform of the Linnean-scale taxonomy.

This step should be accompanied by a reform of the Linnean-scale taxonomy.

It has been brewing for a long time. And not at all because of the introduction of numerical methods in taxonomy. After all, the appearance of Darwin's teachings did not affect taxonomy in any way. The evolutionary theory explained the system of organisms, showed that the reason for similarity is kinship, but in the principles of taxonomy there were no significant changes, and in vain. This can be explained by the dominance of traditional and, one might say, reactionary ideas in taxonomy. After all, the very doctrine of Darwin was initially accepted with hostility and even now is by no means universally recognized.

So applying the theory of evolution to taxonomy, it turns out that there are no taxa at all: evolution does not make the leaps necessary to distinguish them. There are only branches that diverged strongly and not very much. In my opinion, this is what we should proceed from. And why they diverged is already a question of the number of mutations and the rate of their accumulation.

As for attempts to reform taxonomy, that is, a huge literature on PhyloCode, do you need references? The reform is underway
2 Chromocenter: Who told you that evolution doesn't make leaps?

"Who told you that evolution doesn't make leaps? "
Yes, as if Darwin... (Nature does not make leaps) True, his teaching in its pure form is somewhat outdated, (the synthetic theory was corrected, and Darwin could not know anything about genes), but still, how can it be leaps: "The first bird hatched from a reptile egg" (c) so, what is it? "The legend is fresh, but it is hard to believe." (Griboyedov)
Why is it that I'm drawn to quotes?

<To be honest, I was surprised by the results of the vote. It would be like asking a survey "Do you use a mouse when working with a computer" and most people would say "Mice are not yet perfect enough to be useful" or "This is the first time I've heard of such a device...">

The voting results are predictable and perfectly adequate. Unlike your amazement. And no need to exaggerate.

< only Russian biologists are out of step.>

Just don't generalize. Taxonomy, now that we're talking about it, is one of the few areas where "our biologists" are more or less in step. Have you ever read "funny" Western articles and monographs, where a "taxonomist" who has never seen an insect under a microscope in his life, but knows how to confidently load matrices into a computer, builds a phylogeny and talks about the group system? A sad sight, I must say.

Numerical methods are one of many in biology. As well as cladistic analysis, in particular. Moreover, far from the most perfect, sorry-what is, is, And EVEN Western taxonomists recognize it (only quietly admit that grant-givers do not hear And it is impossible to categorically talk about these methods in the abstract. A specialist can benefit greatly from applying even imperfect numerical methods. A non-specialist trying to generalize will do something stupid with ANY methods.

Thus, applying the theory of evolution to taxonomy, it turns out that there are no taxa at all: evolution does not make the leaps necessary to distinguish them.>

There is only one completely objective taxon - this is a species. All other taxa are more or less conditional. Evolution goes at least in leaps and bounds from species to species.

This post was edited by Salix - 06/20/2006 03: 13

I exaggerated it on purpose, because the described situation got me. It is good that you recognize the right of numerical methods to exist along with others - in this respect, our positions seem to coincide. But, since these methods are equivalent to the rest (and, by the way, what are these _ostal_?), then they should be used in the work - and this is not observed

"Evolution goes at least in leaps and bounds from species to species."
It is possible in some cases, but it is not true in all cases. How does one species form another as a result of a jump? I can imagine a mutation in some morphogen that seriously changes the structure of an animal. But such mutations are unlikely, that is, such mutants will survive, I think, in quite rare cases: after all, in addition to one gene, others are needed to occupy a new niche. And this is possible only when several variants of development are possible in the prynype and the gene plays the role of a"switch". However, this is all not quite taxonomy.
P. S. By the way, phylogenies built only on one gene (most often ribosomal RNA) cause me a lot of skepticism.

< It is good that you recognize the right of numerical methods to exist along with others - in this respect, our positions seem to coincide. But, since these methods are equivalent to the rest (and, by the way, what are these _stal_?), then they should be used in the work - and this is just not observed >

I agree that all methods have the right to exist, and that an active struggle for the purity of science from some methods will not lead to anything good. But it's just as bad, in my opinion, to fight FOR some methods. Here I mean imposing - it does not matter administratively, by public opinion (populism), financially, etc. If people DO NOT use numerical methods, this is their full right. If they don't use these methods to compete, and their scientific results are worse, that's just their problem.

All methods are not equal. Objectively, subjectively, and by availability (costs, equipment, and availability of the necessary related specialists) - it doesn't matter. This is the high art of the researcher, that he decides for himself which methods to use in a particular case, and which ones not. Therefore, there is no need to talk about" it is necessary " to use; -) Long live Freedom! :-)

< "Evolution proceeds at least in leaps and bounds from species to species." It is possible in some cases, but it is not true in all cases. >

A leap - it can last hundreds, thousands of years. It is called a jump only in comparison with a longer "plateau" when the signs of b. m.are constant. Let's say that 5% (I take the figure from the top of my head) of "species "is currently in a"jump" state. This does not mean that referring to these 5% can challenge the existence of the species. "Individual cases", that is, exceptions, are always there and they cause a lot of headaches :-)

< How does one species form another as a result of a jump? >

I'm not ready to talk about speciation in detail - it's not my specialty... You can create a separate topic on this topic.

This post was edited by Salix - 22.06.2006 17: 07

Again, I agree with you. In Russian taxonomy, IMHO, it is often the imposition of so-called classical methods such as "when you look through 1000 herbarium sheets, everything will become clear to you with this group" or "build a comparative table". The researcher must have a choice, and the choice implies the availability of information. Do you know many practical manuals on numerical methods of taxonomy in Russian?

In Russian taxonomy, IMHO, it is often the imposition of so-called classical methods such as "when you look through 1000 herbarium sheets, everything will become clear to you with this group" or "build a comparative table". The researcher must have a choice, and the choice implies the availability of information.

There must be a choice, of course. But you must agree, what kind of specialist are you if you didn't" look through 1000 herbarium sheets " to get an idea of the group, but quickly loaded the signs found on a dozen sheets into the computer? So you can scan these sheets in the computer and let it choose the signs itself.

Well, I'm not saying it shouldn't be done. The point is that such methods are imposed, that is, it is proposed to use only them, and as an alternative to the rest. Speaking about herbarium-tabular approaches, I meant that in this case they are missing out not on cladistics, but on population-statistical alternatives.
<So you can scan these sheets in the computer and let it choose the signs itself>
You will probably laugh (but in vain ) But I did ( http://www.blackwell-synergy.com/doi/abs/1...12.2005.00468.x ), with the exception that the "comp" did not do anything "by itself".

"Horse racing - it can last hundreds, thousands of years."
Ah, I just took it literally - a jump means a race. If the jump is a driving selection (roughly speaking), then yes.
"You will probably laugh (but in vain ) But I did ( http://www.blackwell-synergy.com/doi/abs/1...12.2005.00468.x ), with the exception that the "comp" did not do anything "by itself". "
No, well, I think that no one is going to argue with the processing itself-the topic is just about those cases when the computer "does it itself" (And the person publishes it).

In Russian taxonomy, IMHO, it is often the imposition of so-called classical methods such as "when you look through 1000 herbarium sheets, everything will become clear to you with this group" or "build a comparative table".

These are the reactionary concepts in taxonomy. In this case, it usually does not even reach the comparison tables.

  
a "taxonomist" who has never seen an insect under a microscope in his life, but knows how to confidently load matrices into a computer, builds a phylogeny and talks about the group system.

And where does such a would-be taxonomist get a matrix for processing it by numerical methods? If he builds it himself, then he has to look through the microscope in any way, and if he takes signs from other people's works - so these signs were found by taxonomists who also looked through the microscope (I do not take cases of speculation). Of course, a good matrix is a necessary but not sufficient component when constructing a phylogeny by numerical methods, but only a good taxonomist can make a good matrix. Therefore, I consider a well-executed numerical work to be no worse than the traditional one in a systematic way.

Still, what criteria are used to select features for the matrix? How objective is this?
"These are the reactionary concepts in taxonomy. In this case, it usually does not even come to comparison tables."
Of course, it is bad to deny "from the threshold", but it is also not very good to abuse innovations. (Nothing personal)

< And where does such a would-be taxonomist get a matrix for processing it by numerical methods? If he builds it himself, then he has to look through the microscope in any way, and if he takes signs from other people's works - so these signs were found by taxonomists who also looked through the microscope >

The whole curiosity is that if he builds the matrix himself, that is, first "looks through 1000 herbarium sheets", then the matrix often becomes especially unnecessary for him - all these signs "hang" in his head, all sorts of thoughts about the group system are born in his head, etc. herbarium sheets" did not appear-this is a sure sign that you need to give up science and retrain as an agronomist.

If he blindly takes signs from other people's works, there are also a lot of "buts". First, not all authors should be trusted implicitly. Where is the confidence that ALL authors correctly identified the material? Where is the confidence that ALL authors correctly saw and understood the signs? Second, one author indicates some features, and another author indicates others. Usually there are a lot of spaces. To build a complete matrix, you still have to look a lot yourself. Third, the value of attributes. For example, when analyzing some taxa, the A attribute may work perfectly, but when analyzing others, it may not work. To understand this, you need to look at the material a lot and determine it yourself. How can you understand the value of attributes by looking at other people's work (even if the work is done by a highly qualified specialist), where the value of attributes is often not discussed? Fourth, computer systems are built on the principle of "rationality". I would argue that evolution is an exceptionally rational process.

I can give you an example of a mindless "computer" approach. There is a certain group of insects. Some species have a spikelet on the scutellum, while others do not have a spikelet. This would seem to be a very convenient feature. This attribute is often included in the matrix: "there is a spike" or "there is no spike". In fact, the spike is different from the spike. 1) The spike appears in one branch, then disappears in a number of species, and the reduction can be traced very clearly in a number of related species. 2) In another branch, a similar (but not quite) spike appears in several species as a result of radial divergence, but it is not typical for the group as a whole. Anyone who has looked through the "1000 herbarium sheets" comes to this understanding. But no, there are sometimes " computer specialists "who suddenly notice spikes that" these reactionary classical taxonomists " did not see, and build their own original crazy systems.

From what I have written, it may seem that I am an ardent opponent of numerical methods, cladism, etc. This is not entirely true. Each method has its own place in the study. And it is not necessary to categorically oppose these methods to each other.

If such thoughts did not appear after "viewing 1000 herbarium sheets", this is a sure sign that you need to give up science and retrain as an agronomist.

Not agronomists, but managers.

The whole funny thing is that if he builds the matrix himself, that is, first "looks through 1000 herbarium sheets", then the matrix often becomes especially unnecessary for him - all these signs "hang" in his head, all sorts of thoughts about the group system are born in his head, etc.

Numerical methods ensure reproducibility of results and at least visible objectivity of conclusions. The conclusions of a classical taxonomist may well turn out to be better than those obtained by numerical methods, but they are practically not reproducible, and the specific mechanism for obtaining them dies along with the classical taxonomist.
In addition, the matrix is the most visual form of data representation.
The example with spikes is a classic example of incorrect matrix composition, and numerical methods have absolutely nothing to do with it.

Listen - here's what I don't understand in any way: why in our time build such things as a system and phylogeny based on morphology? There is after all DNA-build on multiple alignments. Here, of course, there are also a lot of problems, but at least one of the main shortcomings: which features to add to the matrix and which ones not, practically disappears. They were based on morphology when they didn't know anything about DNA, but after all, the raw material of evolution - that is, mutations-occurs there, and morphology is already secondary, with all the consequences that follow from this?

< In Russian taxonomy, IMHO, it is often the imposition of so-called classical methods such as " when you look through 1000 herbarium sheets >

To be honest, I didn't really understand what "imposing"means. When I entered graduate school, I was taught to look at "herbarium sheets" and make comparative tables. This is called "School". If I had come to another institute, perhaps they would have taught me how to work differently, and this is normal. Indeed, the Russian school of taxonomy is based on viewing "herbarium sheets". I don't see anything wrong with that. I also haven't seen anyone seriously discouraged from using "advanced non-hyperbaric" methods. Of course, if a graduate student starts teaching the supervisor "how to do science correctly" using computers and cladistics, it looks a little strange. And it usually ends with the" imposition " of old classical methods on the graduate student. That's why a graduate student exists to study first and then study again, not to perform.