Evolution vs Baraminology: the empirical showdown.

Post Reply New Topic

This thread is for discussion of the test I proposed to iamnotaparakeet regarding baraminology (the prevailing view among the more serious creationists) and evolutionary biology (the prevailing view in most of the academic community). Specifically:

So, the first task is to define the parameters of our experiment. I can perform the analyses using standard phylogenetics software packages such as MrBayes, GARLI, Mesquite, Clustal, and Mafft (and PAUP* if I can get access to it). Most of these programs are freely available, so I can walk any interested person through the use of them if you are interested in verifying my results. Be warned that MrBayes requires insane amounts of computing time to get any reasonable results, but I believe that any relatively modern home computer should be able to handle the others.

Now, we must decide on a group of organisms to study. We want a sample of organisms which includes some species within the same "kind" and some outgroup organisms that baraminologists would unambiguously classify as being a different kind. I would like to leave this choice up to iamnotaparakeet to ensure that the selection is a fair test of baraminology's predictions, but for the sake of computational considerations I would request that all organisms chosen be predicted to be relatively close together according to the evolutionary standpoint, eg no comparing mollusks to monkeys. Such a comparison could be made, but to do it well would require much larger datasets and several months of time on a supercomputer. Given your interest in parakeets and related species, perhaps that could be an avenue we might explore.

We also need to agree beforehand on what will constitute a disconnect in the phylogenetic data that can be regarded as separating two hypothetical "kinds." My first attempt at a definition would be that molecular genetic homology would not likely exist between two different kinds, and if it did it would not be linked to evolutionary relationships, meaning we would have no reason to expect consistent results across different genes or different techniques. So, if we examine (for instance) 20 unlinked nuclear genes and almost all produce the same proposed evolutionary relationships, and continue to do so under different phylogenetic tests (the main three are maximum parsimony, maximum likelihood, and Bayesian posterior probability) then we must conclude that there is some genome-wide link between the species being analyzed and they are related, contradicting the prediction of baraminology that different kinds are unrelated. If, on the other hand, the different genes and different tests give different or inconsistent results, then phylogenetics has failed, undermining a key prediction of evolutionary biology that we should be able to ascertain relationships among species from their genetic sequences.

I've always wanted to know how closely related Nymphicus hollandicus (a cockatiel) is to Melopsittacus undulatus (budgies), and I consider these birds to be of the same kind due to their morphological similarities and nearly the same hearing ranges. Baraminologists who have PhD's haven't done anything regarding these though. For creatures of class Aves which are not of the Parrot kind, I would think that the various forms of Passer (sparrows, finches, canaries, et al) would work alright, and if more morphological and sensor dissimilarity is needed for contrast, then members of Anatidae (ducks, geese, swans) would also work for them not being of the same kind.



I don't know what such a disconnect would appear as. To an extent, in Michael Behe's Edge of Evolution it goes into the rates of change along with generational periods. Perhaps a reference to the number of generations necessary for transmutation [from mutual ancestor, as axiom] to occur with reference to observed ranges of generational time periods would give an idea as to whether such transformation could fit into the time period provided by radiometric dating, (assuming secular age of earth for argument sake)?

So we've got cockatiels, parakeets, sparrows, and some form of duck-duck-goose. Sounds good to me. Those are several different classes so they shouldn't be controversial as separate kinds from a baraminology perspective, but they are still probably close enough that the phylogenomic data shouldn't be too messy (if evolution is correct). I'll check a few proposed phylogenies for Aves and see if anything else looks appropriate in addition to these choices.


Basically, after running the phylogenetics software I will get proposed evolutionary trees, or phylogenies. Now, I can construct a phylogeny based on individual gene trees (one for gene A, one for gene B, etc) or for a whole large chunk of the genome at once. If the two kinds are truly unrelated, there is no reason to expect genetic homology to be the same across the entire genome*, so baraminology should predict that my phylogenies based on different genes would often look radically different from each other, and if I construct a phylogeny using two different methods I shouldn't generally expect to get similar results. Evolution, in contrast, predicts that gene trees should usually be similar to or the same as species trees, so if I construct two phylogenies using different genes I should generally expect the results to be comparable.


Hm... that would be a much different project, and broader in scope. You seemed in previous posts to emphasize the importance of here-and-now repeatable empiricism, so I wanted to go for a line of evidence that would be as uncontroversial as possible. We have the gene sequences which were observed and recorded recently (and could hypothetically be repeated, though this would be expensive) and we have software which can analyze those sequences in an objective, repeatable, quantitative manner.

I have several other commitments in the next week, so I am not likely to start serious work until probably Thursday or so. In the meantime, feel free to think it over and offer any additional suggestions as to methodology, sample choices, appropriate predictions for the two hypotheses, etc.

*In order to sidestep the "common design vs common descent" argument, I am not taking the mere existence of homology as evidence for or against baraminology here. I would argue that it should be regarded as evidence against, but that is a separate issue. Here I only wish to address the nature of any existing homology.

Okay, if you need to find other discrete kinds within the "King Phillip Cried Out For Goodness Sake" hierarchy, it is usually approximated by the family level unless other families of the same order basically appear the same in which case it would be more likely that it would be approximated by the order rather than the family level.



How radically different? I would think it would be of interest to determine the quantity of genes in disjunction between species of the same family/order versus species of a different family/order. The conjunction would be of all the genes in common, necessary for the processing of similar nutritional sources and other biochemical processes. Within the disjunction set of genes would be where the disconnect would probably be located.



I'll think it over as to how I think the disconnect should be defined, but I might as well throw dice at present until I research the topic a bit more.

This is usually kind of an eyeball thing when you're looking at a tree. I'll try to pull up some examples soon and give a more concrete criterion, but it is usually pretty obvious if phylogenies are in serious disagreement. Some slight variation in phylogenies is normal and expected, but radical shifts would be problematic.


Well, we won't be looking at whole genomes in this manner, as that is way too much data for this. Phylogenetic analyses look at certain genes that are found in analogous form in all species being examined and looks at the extent to which those genes differ from each other. For example, we would find a gene that codes for a similar protein in parakeets, sparrows, etc and then compare the sequence of that gene in each species to see how similar or dissimilar it is between them.

But you've given me an interesting idea in comparing the matches at various taxonomic levels. The number of genes that produce matching phylogenies within a family compared to the number that do so within an order or class (depending on where the baramin line is drawn) seems like a reasonable measure to report. I'll have to read up on some of the statistical methods, since in some cases they can be subject to artifacts which can distort the results. There are ways to mitigate these effects, and I'll want to make sure I'm applying the methods properly.

Am I allowed to post? Tough.

I applaud you both for putting it on the line, so to speak. Lately WP has been getting boring, but this really excites me. You are both very honest people and interested in truth.

I was a little rusty on what constitutes a control in this, but it seems a little more clear having read through both of your posts. Can you clear it more for the dumb crowd(me)?

The concept doesn't really apply because we're not performing an actual wet experiment. We have two predictions that will be either supported or rejected by the data we find.

I mean, you are going to show the keet some graphs to which he might say "Those so called minor differences are substantial." I understand from reading that you are going to come up with some example charts but i dont understand how you will say "different enough". So I guess control is the wrong term but I am not sure how you will agree on a watershed point.

At least we'll learn something out of it. Nobody is going to change their views overnight.

I look forward to the results.

I don't think anyone will, I mean, I believe for any creationist, this experiment, or any for that matter, could not mean much, if anything. First, iamnotaparakeet constantly talks about 'evolutionists' having their presupposed belief, and the science being biased towards favoring that presupposed belief. And secondly, in this experiment, which it seems the prediction is simulated, well I'd have to ask Orwell about that to be sure, but if I consult random creationists on this, they might as well dismiss the "simulated evidence" as a whole. However, I'm ignorant of the issue and importance of these type of experiments, so they would be as well I pressume.

We can learn something out of it, perhaps, I pressume that it will be about wether phylogenetics empirically supports Evolution and how it supports it, and that it makes even more sense.

Sorry Orwell, for a negative or pessimist remark in this thread, which probably comes out of place, I will edit it later if that seems better.

Well, assuming that the results support evolution, which I think they will, I don't think that this type of simulation would sway many other creationists at all. However, at least keet has agreed to Orwell's challenge which suggests differently in this case. I don't necessarily think he would be non-objective at all. I think he just chooses to believe in creationism based on ideology and chooses his sources in that same regard too.

Keet, just to point out. At this point it doesn't matter what I think the results will be. In the end we just have to wait and see what they are, they will be independent of what you I think. I'm looking forward to seeing the results because I think they'll be interesting nonetheless. This is one thread I'll be following up on closely.

Orwell, what do you think the definition for a "wall" should be? I doubt it would be something really defined in modern academics since the current paradigm presupposes common ancestry, but since you do have knowledge of genetics what would be your best guess as to how it would appear? If it were a matter of just organic chemistry, I could say something about reaction equilibria and functional groups, but you would have more of an idea of what the patterns in the genome would look like if such a disconnect exists.

I'll give another go at explaining what I meant; perhaps I was unclear earlier.

So the phylogenomic analysis will produce results in the form of proposed evolutionary trees. I can produce a phylogeny from any given gene or any given combination of genes. If two species are unrelated, genetic homology has no reason to be consistent from one gene to the next, so I should get radically different evolutionary trees when I sample different genes. If evolution is correct, the resulting evolutionary trees should be more or less consistent across the whole genome, no matter which gene I choose to sample. Essentially, a "wall" would be where phylogenetics fails to give internally consistent results.

I'm a creationist (at the atomic level) and evolutionist (as in evolution of species). My religious beliefs hold the importance of science and religion, and we don't take the Adam and Eve story literally. So on a personal note, I only see a good outcome out of the experiment.

Then you're not a creationist in the sense that we're referring to here. When I talk about creationists, I mean people who interpret Genesis literally. I don't call people creationists if they don't believe in a literal interpretation of the Bible, including the Adam and Eve story.

OK, so I'm going to be pinning down the exact species and exact genes we'll be looking at fairly soon. In the meantime, now is as good a time as any to describe the first step in the methodology.

The phylogenetics software that we'll be using will look at the genetic sequence (the A, T, C, G) and compare them across species to see how similar or different they are.
However, once we have our sequenced genes, we can't just throw them into the phylogenetics software. First, they have to be aligned. Let me explain why with an example:

Sequence 1: ACCAGTCATTGTACAACCTGGTAT...
Sequence 2: ACCAGCATTGTACAACCTGGTAT...

Now, as you can see these two strands are basically alike. There is only one visible different between them, but if you look closely that one difference leads to a frameshift error which would lead a naïve analysis of the plain sequence to vastly overestimate the actual differences between the two sequences. So we will "align" the genes to look like this:

Sequence 1: ACCAGTCATTGTACAACCTGGTAT...
Sequence 2: ACCAG–CATTGTACAACCTGGTAT...

With the dash indicating a gap that will be accounted for as a difference in the phylogenetic analyses. Now, this is very much a toy example, and the real-life ones will not always be as clear and obvious. Sequence alignment is a tricky business, and I will be employing two software packages (clustalx and mafft, google 'em if you want to play) to help me with the task.