For evolutionists the revolution in DNA technology has been a major advance. The reason is that the very nature of DNA allows it to be used as a "document" of evolutionary history: comparisons of the DNA sequences of various genes between different organisms can tell us a lot about the relationships of organisms that cannot be correctly inferred from morphology. One definite problem is that the DNA itself is a scattered and fragmentary "document" of history and we have to beware of the effects of changes in the genome that can bias our picture of organismal evolution.

Two general approaches to molecular evolution are to 1) use DNA to study the evolution of organisms (such as population structure, geographic variation and systematics) and to 2) to use different organisms to study the evolution of DNA. To the hard-core molecular evolutionist of the latter type, organisms are just another source of DNA. Our general goal in all this is to infer process from pattern and this applies to the processes of organismal evolution deduced from patterns of DNA variation, and processes of molecular evolution inferred from the patterns of variation in the DNA itself. An important issue is that there are processes of DNA change within the genome that can alter the picture we infer about both organismal and DNA evolution: the genome is fluid and some of the very processes that make genomes "fluid" are of great interest to evolutionary biologists. Thus molecular evolution might be called the "natural history of DNA". The points that follow are some interesting observations interspersed with some basic concepts.

Some important background: DNA has many different roles in terms of function. Most of our DNA does not code for proteins (more below) and thus is quite a different type of character/trait than DNA that does code for protein. In eukaryotes, genes are frequently broken up into exons (expressed) and introns (spliced out of the RNA before becoming a true messenger RNA). The genes also have regulatory sequences that indicate when and where to transcribe the DNA into RNA for protein synthesis. The genetic code is the information system for translating the sequence of RNA into the sequence of amino acids. Within this triplet code some of the nucleotide positions are silent or synonymous because any nucleotide in that position will do (see table 2.1 pg. 25). This "universal" code is not completely universal because the mitochondrial genome uses some of the codons in different ways (e.g., some termination codons in the universal code specify amino acids in the mitochondrial code). Thus even the genetic code can evolve.

Some important theoretical background: we want to develop a picture of what happens to a new mutant in a population, lets say a single nucleotide change a one position in the DNA. This is the starting point for molecular evolution. If the new mutant is governed by genetic drift, its fate should be quite different than another nucleotide mutation that is governed by selection (see below). To describe molecular evolution Kimura formulated the Neutral theory of molecular evolution which is remarkably simple. If:

u = mutation rate / gene / generation, N = population size, then the number of new mutations occurring per generation in a population = 2Nu (2 because we are considering diploid organisms).

Now, when a new mutation occurs in a population its initial frequency = 1/2N because it is the one new variant out of a total of 2N genes in the population. This is also its probability of fixation because the probability of you reaching into a barrel of 2N marbles and getting the one new marble is 1/2N. Thus taking these two values (the number of new mutations/generation and the probability of fixation), the rate of substitution, K is just their product: 2Nu 1/2N = u (substitution means that the new mutant goes to fixation in the population and substitutes the original nucleotide or gene). Kimura got very famous for this simple bit of algebra. It tells us that the neutral rate of molecular evolution is equal to the neutral mutation rate. Now a question arises: what is the distribution of the types of mutations? Are most neutral? Are some deleterious?; beneficial? (see figure 7.1, pg. 153 and box 7.2, pg. 176-177). One consequence of the neutral theory is that genes with different mutation rates will have different rates of evolution.

The rate of evolution of a gene or mutation that is under selection will be very different. Similarly different genes with different functions, or different parts of a gene with different functions will have different rates of evolution. Thus, different regions of DNA with different functional constraints will evolve at different rates (see table 7.1, pg. 156; table 7.7, pg. 183).

One prediction of the neutral theory is that silent (synonymous) sites in protein coding regions will evolve faster than replacement (nonsynonymous) sites (due to different functional constraints). This provides a null hypothesis about DNA evolution. Most sequences fit this neutral model; however, the histocompatibility loci appear to deviate from a neutral model in that there are more nonsynonymous substitutions than synonymous substitutions. This holds only for the antigen binding region; the rest of the molecule is consistent with neutral expectations. (see figure 7.10, pg. 185).

Another prediction of the neutral theory is that amount of sequence divergence will be correlated with the level of heterozygosity; heterozygosity is measured as 2pq for a two allele situation or (2pq+2pr+2qr) for a three allele situation, or 1-xi2 for i alleles; see figure 5.2 pg. 96 for a two allele view). Loci with high heterozygosity should evolve at a faster rate under the assumption that these loci have a higher rate of neutral mutation (thus more variation within species and more substitution between species; se Kimura's proof above). In general, loci fit this relationship, however, balancing selection at a locus will introduce more heterozygosity then expected. What about purifying selection?

Different rates of substitution have also been observed in different lineages of organisms: For the human - chimp divergence, the rate = 1.3 x10-9 substitutions/nucleotide site/year

for the human - Old World monkey split, the rate = 2.2x10-9 substitution/site/year

for the mouse - rat split (=rodents), the rate = 7.9x10-9 substitutions/site/year. Thus, rodents appear to have a faster rate of molecular evolution. It has been argued that a shorter generation time in rodents accounts for the faster rate of evolution, the so called generation time effect. (see table 7.5, pg. 179). There are examples where short generation species have slower rates of evolution; the point is that rates differ, the cause(s) of these rate differences have not been unambiguously identified.

Most discussions of the rates of DNA evolution have been with respect to the molecular clock hypothesis which states that there is a positive linear relationship between time since two species diverged and amount of genetic divergence (e.g., DNA sequence difference) between those species. These observations stated above indicate that there is not one molecular clock but probably many molecular clocks that "tick" at different rates.

Lets say we identify a reliable molecular clock (e.g., number of amino acid substitutions in the cytochrome C gene), we can use this to date, or corroborate, evolutionary events of interest (e.g., the divergence times for species that do not have good fossil data). For example: we know that there are KXY substitutions between species X and Y and we know that they diverged T years ago (from fossil data). Thus the rate of molecular evolution is r = KXY/2T. The denominator has a 2 in it because there are two paths of evolution on which the divergence can accumulate (ancestor to X and ancestor to Y). Now lets say we obtain the sequence of cytochrome C from species A, B and C (see figure below). From these data we count up the number of substitutions for all three pairwise comparisons (the K's). We are given the date of divergence between A and C (=T1) and we want to date the divergence of A and B (T2 = ?) but don't have any fossil data. If one assumes that the rate of evolution is the same in species A, B and C as it was measured to be between species X and Y, we can use the amount of sequence divergence between species A, B and C to estimates their dates of divergence (measured in millions of years before present, MYBP). See box 7.1, pg. 172.

Repetitive DNA Studies of many organisms has revealed that a large proportion of eukaryotic genomes consists of repetitive DNA. Some of this is short localized repeats: in the kangaroo rat the sequence (AAG) is repeated 2.4 billion times, the sequence (TTAGGG) is repeated 2.2 billion times and the sequence (ACACAGCGGG) is repeated 1.2 billion times. What it does is unclear. Sequences like this have been called junk DNA. Note that junk is stuff you don't throw away because it might be useful some day; garbage is stuff you don't want so you throw it away. These sequences might have some function we don't know about so they have been called junk DNA. The fact that such sequences seem to accumulate in genomes has lead to the notion that repetitive DNA is selfish DNA, since the sequence makes additional copies of itself within the genome decoupled from the reproduction rate of the host (i.e., the kangaroo rat).

Another form of repetitive DNA are transposable elements. These are sequences of DNA that generally code for certain proteins and have the ability to move around the genome in a process called transposition. There are quite a number of different types of such elements (we will not review them all). The point is that they (like other repetitive DNA) are governed by intragenomic dynamics as well as organismal population dynamics. An example is the P element in Drosophila melanogaster. There are strains of flies that have P elements (P strains) and strains that do not (M strains). When a P male is crossed to an M female the P elements enter the genome of the offspring and jump around causing mutations (this is what we mean by a fluid genome). A curious observation about P elements is that strains of flies collected from natural populations before 1950 do not have P elements whereas flies collected from the wild after that do have them. A variety of observations indicate that P elements invaded D. melanogaster recently. The best evidence is that D. melanogaster's close relative do not have P elements, but a more distantly related fly D. willistoni does have them and they differ by only a few nucleotides over 2900 base pairs of DNA. These data suggest that P elements in D. melanogaster are the result of a horizontal gene transfer (horizontal as opposed to "vertical" as one inherits DNA from ones parents or ancestor "above"). Thus, not only can DNA move around the fluid genome, but if DNA from one species can enter the gene pool of another without the species fusing into one, one has to be very aware of what DNA sequence one is using to determine phylogenies, etc.

If multiple copies of a DNA sequence are present in a genome we can think of each sequence as a single "species" evolving on its own "line of descent" because each repeat will be mutated at random. Thus if we had the complete DNA sequence of all the repeated P elements within a genome, we would find that they are not identical and thus we could build a cladogram of these elements much like we can build a cladogram of birds. When this sort of analysis was done on different kinds of repeated elements (many copies of ribosomal DNA, for example) it was found that the copies showed almost no variation. This observation suggested that all the repeats of this family (ribosomal DNA family) were evolving in concert, i.e., together. This pattern of homogeneity of repeats is called concerted evolution (figure 10.5, pg. 262). The process(es) that generate this pattern could be unequal crossing over (see figure 10.2 & 10.6, pg. 258, 263) or gene conversion. Gene conversion is when the sequence of one region of DNA is used as a "template" to "correct" or modify the sequence of another region of DNA. We do not need to go into the molecular details of these processes, but that DNA can evolve in concert with other sequences in the genome again indicates that intragenomic dynamics can influence the pattern of DNA variation we see within and between species.

Gene duplication is a minimalist version of repetitive DNA (figures 10.1-10.3, pgs. 257-259). Many genes in the genome are duplicated and when this happens one of the copies may be "freed" from constraints and evolve a new function. The best understood case of this phenomenon is the evolution of the globin genes myoglobin, a-hemoglobin, ß-hemoglobin. The existence of duplicated genes forces us to recognize different kinds of homology because there are two ways to have a common ancestor: by gene duplication and by speciation. When two genes share a common ancestor due to a duplication event we call them paralogous (a-hemoglobin and ß-hemoglobin in you are paralogous as are the a-hemoglobin in you and the ß-hemoglobin in chimps). When two genes share a common ancestor due to a speciation event we call them orthologous (a-hemoglobin in you and a-hemoglobin in chimps). Obviously when constructing a cladogram from molecular data one should use orthologous genes if one wants to build a tree of organisms.

A further example of the fluid genome is exon shuffling This is the pattern observed when exons or functional domains of genes are shuffled together to form new or modified genes. Some genes have very distinct domains that have clear relationships to other domains in very different genes. It is thought that these domains have been moved around the genome by transposition or illegitimate recombination events in evolution, accidentally forming new associations that happen to have novel functions. Wally Gilbert proposed the idea of exon shuffling and argued that such a phenomenon might accelerate evolution by creating new material for adaptive evolution.

One of the more interesting observations to ponder in molecular evolution is the C-value paradox. The C value of a species is the Characteristic or Constant amount of DNA in a haploid genome of that species. If we look at the diversity of organisms from viruses to humans we see a clear trend in biological complexity. If we compare the C values across this range of organisms (assuming viruses are "organisms") some of the less complex organisms have much more DNA than the more complex organisms (see tables 10.2, 10.3, pgs. 259-260). This presents a paradox. If DNA codes for proteins that give us form and function, what is a lowly alga doing with all that DNA? We just don't know. Presumable most of it is not "functional" in the sense of coding for proteins and RNAs. Much of it may be "junk DNA", but maybe this "junk" helps in aligning chromosomes properly during mitosis an meiosis.

You now see why we referred to this topic as the natural history of DNA: these are "stories" we know about how DNA "behaves" in evolution. The general point is that the intragenomic dynamics of DNA and the intergenomic (Åinterorganism) dynamics can be radically different, but the patterns of variation we see in the current day are the summed effects of both processes. This means we have to know some things about a piece of DNA before we can use it as an evolutionary tool.