Both ideas initially controversial but now {thesis/antithesis/synthesis} accepted that much (most) change is neutral with selection usually conservative, rarely disruptive and that much (most for sure at DNA level) polymorphism is also neutral

Best known tenet is rate of fixation of neutral alleles

Consider a population of genes equivalent but distinguishable doubling from N to 2N each generation with N surviving to double again. P for survival of each is 1/2Ne where 2Ne is the # gametes per generation. If neutral mutations occur at a rate mn per gamete per generation, then the # of new alleles per generation is 2Nemn and the net rate of fixation

kn = (2Nemn)/2Ne = mn; thus the net rate of fixation is independent of population size, but inversely related to the generation time

Similarly it can be shown that the time required to fix a neutral allele is 4Ne generations.

Fitch &_Markowitz (1970) - fitted a Poisson distribution to the variation at codons in a protein. They found that a two site (3 parameter) model was sufficient: invariant sites and covarions (for concommitantly varying codons). The positions of the invariant vs covarion sites could change over time but for a given protein family the number of covarions was essentially constant. Some proteins (e.g., histones) had few covarions; others (e.g., fibrinopeptides) had many.

Shoemaker & Fitch (1989) - extended the argument to DNA - covariotides

The picture (in my mind)

Proteins have regions that are critical for function and almost invariant; others that are there for class (polar, hydrophobic, etc) with limited variation; others that have little function except bridging = (most) covarions

DNA similarly has relatively fixed regions encoding proteins in invariant regions (1st 2 codon positions only), promoter or enhancer regions, splice consensus sequences, etc vs loose regions like intergenic and within introns). Certain sequences like transposons, mobil elements, integrated retroviruses and pseudogenes are particularly dispensable thus most free to vary.

RNA viruses both retro like HIV and non like polio lack proofreading systems for their replication; they are extremely rapid in their rate of change

How to reconcile the >= 3 categories model implicit in the picture with the two categories model for covarions & covariotides?

Need only to argue that there are two categories at a given time, but that over eons most nucleotide positions shuttle among the categories while the fastest evolving are always in the variable state (unless deleted, but even then . . . ) and the slowest are always in the invariant state.

Can debate relative contributions of selection and nonadaptive evolution ad infinitum but how to test? Kreitman has supplied a test based on comparing polymorphism to divergence. Eanes (1993) tested but comments (Hudson (1993)) help interpret.

If compare two present day species in same gene at amino acid residue level for extent of both divergence and of polymorphism, the two observations should be related according to neutral theory as both depend on mn above. Moreover can compare for both replacement changes and silent changes. Latter should mostly be neutral while former may have a fraction of adaptive change.

expect polymorphism proportional to 4Nm for neutral mutations

expect divergence proportional to 4NmT where T is the time in 2N generations since divergence

Compared D. melanogaster vs D. simulans by sequencing the G6pd gene for multiple feral individuals (32 & 12, resp.) of both species. Divergence for replacement positions was in excess relative to polymorphism:

21 is ~10x too high for neutralism

Most likely explanation: periodic selection
other possibilities: other forms of selection like frequency dependent ones but acting differently in the two species or variation in mn over time.

They rule out hitchhiking as Silent/Replacement ought to show comparable effects

This means selection is certainly a player but G6pd is clearly a critical function (hexose shunt ® ribose) and selection on it is well established in humans so neutrality is not driven from the field

MHC also supports selection (no 1º ref, but reviewed by Clayton & Gee (1993))
Background:

Genetic-extremely polymorphic in all species examined except a few through recent bottlenecks like cheetah and California sea lion

Function-transport peptide fragments to cell surface of T cells with class I = intracellular products & II = endocytosed

Allele limited-each individual has 2 haplotypes

New findings

human HLA: DR3 Drw18 haplotype unique to Africa;
B*4002 links Japanese to Amazonian Amerinds via Siberian populations:
New alleles in Amerinds despite recent origin;
Includes Havasupai (live on wall of gorge of Grand Canyon tributary) where most are heterozygotes despite inbreeding;
DP alleles best candidates for race/ethnic markers;

Common and pygmy chimp share A3/A1/A11 types with us;
gorilla and chimp share DRB1/DRB6 with us, but each species also has new alleles over 25 million yr;
rhesus cells can present M. leprae to human T cells when share DRB1*03;
some DRB1, 3 & 5 shared with monkeys for 8 37 millon yr

pseudogene DRB6 inactive in platyrrhines but active in humans and catarrhines L reactivated 23 million yr ago; MMTV-like promoter in humans suggests how retrovirus did it

can tree simians and prosimians via MHC data

Cichlid fish in Lake Malawi have extensive MHC polymorphism despite diverging from ancestors in 2 million yr and very little mitochondrial DNA variation

Send mail to Michael Garrick with questions or comments about this web site.
Last modified: September 03, 1998