It does not address in this narrow point the issue of functional DNA. It just points out that a major piece of evidence for evolution was wrong. Junk DNA is functional. I don't want to rehash their failed argument because I never thought it was right. They'd find some mutation in junk DNA that they argued (not even correctly) was present in a group creatures they would then say had a common ancestor based on shared genetic sequences, and since it was not functional, it could not be related to similar features via design but could only be explained via Darwinism.
I still don't see wherein the problem lies. Mutations or mutational patterns that are shared between lineages are evidence of shared ancestry whether the gene is functional or not. Discovering that a gene previously thought to be nonfunctional is in fact functional has absolutely no effect on this, and if you look over phylogenetic papers based on purportedly nonfunctional DNA alignments and just excise the word "non-functional", the phylogenetic trees will remain the same. The phylogenetic methods are the same regardless of what gene region you use -- or even if you use randomly generated, genuinely nonsense, DNA (1) -- even though
e.g. the specific model of sequence evolution used in a Bayesian analysis may differ depending on the gene and/or the position in a gene.
As an example, let's look at 12S sequences in Phthirapteran lice, the group I work with. Let us assume that we believe 12S is non-functional. We have primers for this region, and sequence a whole lot from different genera of lice, and compare their primary and secondary structure. Apart from point mutations, we find certain patterns. For instance, compared with
Columbicola columba, the 47 loop of both
Halipeurus pelagicus and
Naubates harrisoni are bloated and supports a secondary loop on the 5' side (Page
et al., 2002). This suggests that
H. pelagicus and
N. harrisoni are closely related, but that
C. columbae is not closely related to these two. We thus propose that these two species should belong in one group, whereas
C. columbae should likely belong to another group.
Comparing this with morphology, we see that there is some sense in this. Eichler (1963) placed
Halipeurus and
Naubates in his Pseudonirmidae, whereas
Columbicola is placed in Esthiopteridae. While Eichler is often sparse in his detailing of why he formed certain groups, this makes sense when comparing such structures as male genitalia, chaetotaxy, gross morphology of abdominal and thoracical tergites, and internal structures of the head. Comparing host relationships, we find that
Columbicola is a parasite on pigeons and doves, whereas the other two are parasites on seabirds (Procellariiformes) (Price
et al., 2003).
So let's look at other genes and on the actual genetic analysis of the 12S alignment. Fortunately, Page
et al. (2002) provides such a comparison, which shows that for the alignment of 12S (which we assumed was nonfunctional),
Naubates and
Halipeurus are sister groups, and
Columbicola is placed almost in the other end of the tree. They are placed more closely together (and
Columbicola is paraphyletic) in a tree based on the secondary structure (discounting the actual sequences), but both the mitochondrial COI gene and the nuclear EF-1a gene place the two seabird lice together, and separate from the pigeon louse. Cruickshank
et al. (2001) shows the same pattern, using a larger data set, regardless of whether parsimony, minimum evolution or maximum likelihood is used. Smith
et al. (2004) show the same, again, with a different data set, and this result is stable whether you use only morphology, morphology and DNA, only DNA, or only protein-coding DNA, as well as stable under both maximum likelihood and Bayesian inference. Smith (2001) showed the same using only morphology.
Let us then assume that we find that 12S is actually functional. Where, more precisely, does this affect our conclusion? I argue that it does not make any difference whatsoever, and that your claim that "a major piece of evidence for evolution is wrong" is spurious, at best. Moving a given DNA region from the category "no known function" to "function known" does at no stage change the theory or practice of phylogenetic inference and reconstruction. The only difference would be in the perception of what level of resolution you would expect to get in your phylogeny based on a given gene region, as a protein coding region would be assumed to be more conservative than a structural region, which in turn would be assumed to be more conservative than a non-functional region. The only practical difference this would make is in the choice of gene regions used to look at the relationships within a given taxon,
i.e., if I want to study within-genus relationships, I may first try some less conservative genes to sequence, as I would want to work with something that can give me resolution and structure at that level.
Now, I believe you are arguing, in the last section of the above quote, that if gene is functional, similarities between different lineages could very well be because of design rather than evolution. The reason the two seabird lice above have similar 47 loops in their 12S is because they live in a different environment from the pigeon louse, and thus this gene, if we assume its expression has a tangible effect on its survival, could have been designed to cope with these circumstances rather than those under which a pigeon louse lives. For instance, if we assume that 12S in some way gives the lice a higher tolerance to salt in the feathers they eat, it would make sense that lice on seabirds have this extended loop, while pigeon lice do not, whether they are related or not. Its existence would then not necessarily show a common ancestry, but just a similar feature for coping with the same problems of high salt content of the host plumage. This could then be either a result of design or of convergent evolution. Due to the complexity of molecular biology, which ensures that we can never be quite sure what a certain gene region is used for, this is at least superficially a reasonable argument.
However, this does not adequately explain the common patterns of random mutations and indels in a given gene. Why, for instance, do the ectoparasites of birds so often show phylogenies significantly similar to those of their hosts(
e.g., Paterson
et al., 2000; Page
et al., 2004; Hughes
et al., 2007)? Are the differences between the environment on two given closely related host taxa so different that there is a need for designed differences in their parasites? If so, why do these differences, when analysed phylogenetically, so often form nested hierarchies which are concordant with the nested hierarchies obtained when performing the same kind of analysis on their hosts? And why do these designed differences so often occur mostly in the third positions in gene coding regions, positions which are known to be subject to a certain amount of redundancy in the translation codes?
And if these differences are so large between closely related bird hosts that it warrants seemingly random changes in the design of their DNA, why is this not consistent over different kinds of lice on the same birds? Why can wing lice spread by phoresy (2), through shared nest holes, and from prey to predator species and establish new populations on novel hosts in some cases (documented from a large variety of bird louse groups) if their genetic design is based on their normal host? Instinctively -- and this is most likely a false dichotomy -- the DNA would either be designed for a specific host (and then lateral spread to novel hosts would never be viable), or it would be designed for a large amount of hosts or a generic host (and then there is no reason for co-phylogenetic patterns to occur).
The fact remains that these differences in DNA sequence do occur, whether we know that a specific region is functional or not. At the very most, finding a function for what was previously thought to be "junk DNA" changes the taxonomic level on which we expect to get resolution in a phylogenetic reconstruction, but it does in no way render previous phylogenetic reconstructions based on the assumption that the gene region was nonfunctional wrong, as both structural, coding, "junk" and all other kinds of DNA will at least under normal circumstances show the same kind of pattern, though perhaps at different levels due to the speed with which these changes get fixed in the population.
ETA:
To clarify, this is what I am getting at:
Incidentally, that undermines the claims that we can track phylogeny by seeing common errors in junk DNA.
Naturally, if there is no such thing as "junk DNA", we cannot track phylogeny by seeing common errors in it. However, this does not in any way mean that we cannot track phylogeny in non-junk DNA by seeing common errors in it. And if what was previously thought to be "junk DNA" is shown to be functional, any phylogeny based on that DNA sequence would simply switch from one category to the other, but this would not invalidate any phylogenies based on the gene region when it was still believed to be "junk DNA".
---
References:
- Cruickshank, Johnson, Smith, Adams, Clayton, Page (2001). Mol. Phyl. Evol. 19, 202-215.
- Eichler (1963). Mallophaga. In: H. G. Bronn (Ed), Klassen und Ordnungen des Tierreichs, Leipzig.
- Hughes, Kennedy, Johnson, Palma, Page (2007). Syst. Biol. 56, 232-251.
- Page, Cruickshank, Johnson (2002). Insect Mol. Biol. 11, 361-369
- Page, Cruickshank, Dickens, Furness, Kennedy, Palma, Smith (2004). Mol. Phyl. Evol. 30, 633-652.
- Paterson, Wallis, Wallis, Gray (2000). Syst. Biol. 49, 383-399.
- Price, Hellenthal, Palma, Johnson, Clayton (2003). The Chewing Lice. Illinois Natural History Survey Special Publication 24
- Smith (2001). Zool. J. Linn. Soc. 132, 81-144.
- Smith, Page, Johnson (2004). Zool. Scr. 33, 239-259
---
(1) In the sense that it does not code for anything as it is not taken from a known extant or extinct organism. This is regularly done to test phylogenetic theory and for method development.
(2) "Hitch-hiking" on Hippoboscid flies.