Ok, if you are willing to get a little technical (but not painfully so)...
OK , are there some papers that demonstrate the the kinds of mutations used in supporting NDT are random (presumably WRT the fitness of the organism I suppose) and not subject to "non random" influences?
Biologist don't believe that mutations are not subject to non-random processes. There are clear non-random patterns in mutation rate and mutation type.
These patterns, for example, may depend on the position of a nucleotide on a chromosome, or its position in the nuclear genome versus mitochondrial genome [1]; the sequence itself (some combinations of nucleotides such as tandem repeats (i.e. ATATATATATATAT) are subject to higher mutation rates) [2]; sex ratio[3]; generation length, body size and metabolic rate [4]; taxonomic group [5]; the local richness of DNA in C and G nucleotides [6];
Are their papers that show that "convergent evolution" is unlikely to have produced the heirarchies we see.
There are a variety of tests we can use to detect convergent evolution in DNA sequences [i.e., 7]. For morphological characters, we will usually just infer convergent evolution on the basis of the phylogeny [8].
How do we know that the phylogenetic tree itself doesn't result from convergent evolution? First, we are increasingly using multiple genes and morphological characters to create our hierarchical phylogenetic trees. This means that if one or two genes are similar across species by virtue of convergent evolution, we still have many genes in our dataset that are not. It seems rather unlikely that lots of genes would all be convergent and all point at exactly the same incorrect tree. Rather, if convergence is a random process, we would expect to find different genes suggesting different shaped trees, and that would be a signal that our genetic markers aren't good enough.
Secondly, when we build a phylogenetic tree we use probabalistic models of nucleotide evolution that permit us to calculate statistical confidence limits on each node in the hierarchy. Our models account for the possibility that multiple mutations might occur at a single site (i.e. a nucleotide might mutate once, then mutate back to its original form) and they include the possibility that convergent evolution occurs between two sequences that is not indicative of their true phylogenetic relationships. When you see a tree published in a journal, you will often notice little numbers printed alongside each node. These are usually some measure of our confidence that a node in our hierarchy is a "real" node given the data that we used in generating the tree and given the statistical model that we incorporated into the analysis. There are various measures of confidence that we use but one popular method is the bootstrap [9].
Hope this helps! Of the references cited here, reference 4 is probably the easiest to understand, and the most interesting for the layperson.
Best wishes,
Mick
[1]
Just a moment...
[2]
http://www.oup.co.uk/isbn/0-19-850407-1
[3]
Shortened link
[4]
http://mbe.oupjournals.org/cgi/content/full/19/3/302
[5]
Missed this one: AdminJar
[6]
http://mbe.oupjournals.org/cgi/content/short/msi043v1
[7]
http://mbe.oupjournals.org/cgi/content/abstract/14/5/527
[8]
Mesquite
[9]
http://virgil.ruc.dk/kurser/Sekvens/bootstr.htm
This message has been edited by mick, 04-24-2005 12:29 PM
This message has been edited by AdminJar, 04-24-2005 11:31 AM
This message has been edited by AdminJar, 04-24-2005 11:32 AM
This message has been edited by mick, 04-24-2005 01:16 PM