Some molecular evidence for human evolution

[Rumraket (Mikkel R.)]

(Sites without numbers don’t have anything to say about this particular question.)

I don’t understand why? They contradict the (african apes)(gibbon+orangutan) as far as I can see. Though they just add to the number of hypotheses that contradict each other, still leaving the canonical relationship the most well supported one. For example I can see four rows that support the relationship (human+gorilla)(gibbon+orangutan+chimpanzee).

[John_Harshman]

They contradict the (african apes)(gibbon+orangutan) as far as I can see.

Those sites don’t actually contradict that tree, as they require one change on a tree that includes African apes as a clade as well as a tree that doesn’t. In other words, they’re only relevant to relationships within African apes. Picture a tree with one internal branch, separating African apes from gibbon and orangutan. Only sites that require two changes on that tree contradict the tree.

[T_aquaticus]

Out of curiosity, is incomplete lineage sorting a significant source of noise in these mitDNA comparisons? I would think that being a single copy genome along with a higher rate of sequence divergence would reduce ILS compared to genomic DNA, but I could be completely wrong here.

[Timothy_Horton]

The design condition is the designs are staggered in time which agrees with the fossil record.

Wow. So the Designer has been coming by every few thousand years and creating new species de novo for the last 3.5 billion years?

When and how did you come up with that gem?

[John_Harshman]

is incomplete lineage sorting a significant source of noise in these mitDNA comparisons?

No. It’s a source of exactly zero noise. Since the mitochondrial genome doesn’t recombine and is inherited at a unit, every site in the genome tracks the same phylogeny. Incomplete lineage sorting still might happen, but it would result in a single phylogeny that doesn’t match the organismal phylogeny. Further, there should be less a chance of this than for autosomal loci, since the probability of ILS depends on the effective population size, and mt genomes, to a first approximation, have a quarter of the effective population size of autosomes.

[Rumraket (Mikkel R.)]

Those sites don’t actually contradict that tree, as they require one change on a tree that includes African apes as a clade as well as a tree that doesn’t. Picture a tree with one internal branch, separating African apes from gibbon and orangutan. Only sites that require two changes on that tree contradict the tree.

Ahh I get it now. It is possible to make trees with african ape-clades that still imply only one change. Like the top tree here. But I was thinking the first row only implied the second tree.

I also let myself be confused by knowledge of the canonical phylogeny with humans and chimps most closely related, which would imply two changes.

[John_Harshman]

Exactly.

[T_aquaticus]

No. It’s a source of exactly zero noise. Since the mitochondrial genome doesn’t recombine and is inherited at a unit, every site in the genome tracks the same phylogeny. Incomplete lineage sorting still might happen, but it would result in a single phylogeny that doesn’t match the organismal phylogeny.

Thanks, that makes a lot of sense. I suspected this was the case, but I wasn’t absolutely sure how intraspecies mitDNA variation played out without recombination.

[scd]

If they were separately created, why would they share any DNA sequences?

because they share many traits in common.

[T_aquaticus]

because they share many traits in common.

You have the cart in front of the horse. They share traits because they share DNA. If you used completely different DNA then they could have completely different traits. Why would a designer need to share traits between separately created species?

We also have the flip side. You could have drastically different DNA and still have the same traits. For example, you could rewrite the anti-codons on tRNA’s which would allow you to have very different DNA sequences that produce the exact same proteins. Talkorigins has a great section on this concept:

One common objection is the assertion that anatomy is not independent of biochemistry, and thus anatomically similar organisms are likely to be similar biochemically (e.g. in their molecular sequences) simply for functional reasons. According to this argument, then, we should expect phylogenies based on molecular sequences to be similar to phylogenies based on morphology even if organisms are not related by common descent. This argument is very wrong. There is no known biological reason, besides common descent, to suppose that similar morphologies must have similar biochemistry. Though this logic may seem quite reasonable initially, all of molecular biology refutes this “common sense” correlation. In general, similar DNA and biochemistry give similar morphology and function, but the converse is not true—similar morphology and function is not necessarily the result of similar DNA or biochemistry. The reason is easily understood once explained; many very different DNA sequences or biochemical structures can result in the same functions and the same morphologies (see predictions 4.1 and 4.2 for a detailed explanation).

As a close analogy, consider computer programs. Netscape works essentially the same on a Macintosh, an IBM, or a Unix machine, but the binary code for each program is quite different. Computer programs that perform the same functions can be written in most any computer language—Basic, Fortran, C, C++, Java, Pascal, etc. and identical programs can be compiled into binary code many different ways. Furthermore, even using the same computer language, there are many different ways to write any specific computer program, even using the same algorithms and subroutines. In the end, there is no reason to suspect that similar computer programs are written with similar code, based solely on the function of the program. This is the reason why software companies keep their source code secret, but they don’t care that competitors can use their programs—it is essentially impossible to deduce the program code from the function and operation of the software. The same conclusion applies to biological organisms, for very similar reasons.

To reiterate, although similar genotypes (e.g. molecular sequences) often give similar phenotypes (e.g. morphological characters), similar phenotypes are not necessarily the result of similar genotypes. Thus, it is entirely possible that phylogenetic trees constructed from genotypic data could be radically different from phylogenetic trees constructed from phenotypic data. In fact, in the absence of common descent or any other reason to suppose that these two types of trees should be similar, the most likely result by far is that they will be radically different. This is precisely why it is possible to falsify the macroevolutionary prediction that independently derived phylogenies should be similar.
29+ Evidences for Macroevolution: Part 1

[NLENTS (Nathan H. Lents)]

have you done this type of analysis with other sequences?

I haven’t, but anyone is free to try it.

I have done this with the sequences from the L-GULO pseudogene and turned it into a classroom activity (that I just did with my students yesterday actually!). I picked the gene and the sequences pretty much at random. I used human, chimp, gorilla, orangutan, and macaque. It’s only a ~100bp stretch but it shows the human-chimp sister grouping very clearly, and the fact that macaca is the most divergent. It doesn’t resolve the gorilla and orang phylogeny, but that’s what exercise #2 is for (I switch to protein sequences for that).

The point here is that the evidence for common ancestry is very, very easy to find. I understand the concerns above that it doesn’t clarify things like selection, design, and so forth, but, at least in my mind, DNA sequences alone make common ancestry of all living things on earth about as scientifically certain as humanly possible. (And we have a lot more evidence than just DNA sequences.)

[John_Harshman]

Did you happen to try a chi-square test?

[NLENTS (Nathan H. Lents)]

I did not, because this is a course for non-SCI majors, so I keep it as math-free as possible, just counting basically. This way, they fully wrap their brains around what is happening and they do the analysis fully themselves, not even a calculator. Their eyes really open wide when they see how easy it is to see evidence for common ancestry. People always think that scientific data is just beyond them, totally inaccessible, and lots of it really isn’t!

[Timothy_Horton]

The design condition is the designs are staggered in time which agrees with the fossil record.

Wow. So the Designer has been coming by every few thousand years and creating new species de novo for the last 3.5 billion years?

When and how did you come up with that gem?

Hey Bill, are you going to explain and provide the supporting evidence for your latest assertion?

[Giltil (Gilbert Thill)]

Interesting.

One thing that strikes me is that this sequence of 76 nucleotides is very dense in mutations compared to all the other 694 positions. Why is that the case?
Moreover, this same sequence seems to display an abnormally high number of cases of homoplasy, doesn’t it?
IOW, I am wondering what is going on with this particular sequence?

[evograd (Blogging Graduate Student)]

One thing that strikes me is that this sequence of 76 nucleotides is very dense in mutations compared to all the other 694 positions. Why is that the case?

Because this sequence is made up of all the sites in those genes that have relevant mutations, it’s not an actual 76bp sequence within the 694bp. Read the OP again.

Here is a set of DNA sequences. They come from two mitochondrial genes, ND4 and ND5. If you put them together, they total 694 nucleotides. But most of those nucleotides either are identical among all the species here, or they differ in only one species. Those are uninformative about relationships, so I have removed them, leaving 76 nucleotides that make some claim. I’ll let you look at them for a while.

[Giltil (Gilbert Thill)]

Thanks.

[Giltil (Gilbert Thill)]

The sites I have marked with numbers from 1-6 contradict this relationship. (Sites without numbers don’t have anything to say about this particular question.) We expect a certain amount of this because sometimes the same mutation will happen twice in different lineages; we call that homoplasy.

Ok, we expect a certain amount of cases of homoplasy but do we expect such a high level of cases, in this case 22 out of 694 positions? This level seems extraordinary high to me, doesn’t it?