Saturday, December 18, 2010

Evolution slows down in a crowd

A couple of papers show how small population size allows evolution to flower, by limiting natural selection.

One of the curious questions in biology is why multicellular animals have shown such rapid evolution, while microbes seem a good deal more static. This is not to say that bacteria can't evolve very rapidly in response to selection, or have not evolved an unending variety of metabolic capacities, but that they rarely change dramatically, such as in their physical forms. The Cambrian explosion is a classic example of relatively rapid animal evolution, and even if recent work tells us that it was a good deal slower than "explosion" would imply, animals seem to change over time at a relatively dizzying pace.

This question, biased and subjective as it perhaps is, has many answers, but a crucial one laid out in a few papers by Michael Lynch is that large population size slows evolution dramatically by reducing the chance that slightly deleterious mutations will ever become fixed in a population. Such mutations are the junk out of which new functions arise- the gene duplicates, introns, transposons, superfluous regulatory regions, etc. that pepper our genomes as large eukaryotes. Even entire genome duplications are known to have happened. This argument has significant implications for other evolutionary questions as well, like the origin of introns, early human evolution, and the nature of punctuated equilibria.

The empirical fact is that microorganisms are much less forgiving of "fat" in their genomes- they have little or no "junk" DNA. They have fewer genes, which are typically positioned cheek-by-jowel in close arrays, often even driving several different protein products from a single promoter. A frequent argument for this state revolves around their physical size- human cells are roughly about 50,000 times the volume of a bacterial cell, and yeast cells are about 50 times larger. So for a given genome of physical DNA, the metabolic cost of that DNA is much higher for smaller organisms, for whom a genome typically constitutes 5% of dry weight (compared to ~0.5% for human cells). But Lynch, an evolutionary biologist at Indiana University, has a different answer. Indeed, one might ask which came first- the small size, or the small genome?

The papers are from 2001, 2002, and 2007, of which the first two are math-heavy papers modelling how realistic mutations like novel gene duplications (paper 1) or introns (paper 2) propagate in populations of various size. The third is a more general and perhaps accessible jeremiad against making natural selection the be-all and end-all mechanism of evolutionary change.
"Most biologists are so convinced that all aspects of biodiversity arise from adaptive processes that virtually no attention is given to the null hypothesis of neutral evolution, despite the availability of methods to do so. Such religious adherence to the adaptationist paradigm has been criticized as being devoid of intellectual merit, although the field of molecular evolution has long been obsessed with potential for the ‘‘nearly neutral’’ accumulation of very slightly deleterious mutations. The condition for near-neutrality is fulfilled when the ratio of the powers of selection and drift is substantially <1, i.e., |2Ngs| << 1."  .. where Ng is the population size (N) of the given gene (g) or allele, and s is the relative selective advantage of the given gene or allele.

Insertions like gene duplications and introns are assumed to be, on average, slightly deleterious. A new intron may make the gene slightly slower to be expressed. A duplicate copy may double the gene's dosage, wasting some of the protein product or altering a regulated setting slightly. Typically, such alterations have near-neutral effects. And the issue for evolutionary innovation is why and for how long such mutated material (whether we call it junk DNA or gold-in-the-making) can be kept around in the genome.
"For example, the rate of duplication of entire genes is ~1% per gene per million years..."
For a population of one, any mutation is immediately fixed (i.e. reaches 100% of the population)- no math needs to be done. This is the situation of species on the brink of extinction, and it is dire, because bad mutations far outnumber good ones, and an inability to cleanse the population of bad mutations is fatal. Conversely, for an infinite population, no mutation will ever reach fixation- there isn't enough time. What happens in between? As population sizes go up, it becomes harder and takes longer for random drift to bring a mutant allele to fixation. Indeed the overall probability for any completely neutral allele is 1/2N, where N is the population size. The more time it takes to reach fixation, the more time natural selection has to weed out a deleterious allele, and the smaller "fitness" cost that can be detected and selected against.
Genetic drift of a neutral gene starting at 50% of the population, for populations of different sizes. Percent of population (Y axis) graphed against generations (X). Fixation is 100%, and complete loss of the gene is 0%.
This wiki site graph shows that with increasing population size, time to fixation by genetic drift increases dramatically. For truly neutral alleles, this relationship doesn't make a gross difference, because the number of new mutations arising is also proportional to population size, so overall, the neutral mutation rate of any population is similar, regardless of size. But for slightly deleterious alleles, the story is very different. The longer such an allele exists in a population, competing with its normal counterparts, the less likely it is to survive.

Successful organisms by definition grow to large populations. Bacteria of even rare species can easily attain populations of ten billion or beyond. A single drop of pond water can contain millions. In the case of humans, such a population is far beyond the natural carrying capacity and indeed is rendering the biosphere serverely degraded and unsustainable as we speak. Macro-organisms simply can't reach these kinds of populations- there are orders of magnitude differences determined intrinsically by the size of the individual.

If one combines these facts with the proposition that gene duplications, introns, and similar junk are the ingredients of novel function, diversity, and speciation, (as they are assumed to be by most biologists), then it is clear that natural selection can be too good. At very large population sizes, every scrap of deleterious DNA is excised and the organism streamlined to perfection. That process eliminates the slop and junk that forms the workshop of evolutionary innovation, ironically enough.


To put this visually, the selective space facing organisms is frequently portrayed as a two-dimensional landscape, with peaks of high fitness, separated by valleys of lower fitness. A major problem is, of course, how to transit from one local peak to another that may actually be higher. If an organism is stuck on a low local peak, and if natural selection is set on "high" by virtue of the organism's high population size, it may never be able to sample other locations in the landscape. Of course, if it has large population size, then it is already successful, and can well rest on its laurels.

This has obvious implications for, among others, the problem of punctuated equilibrium, which has been, to some at least, a rather unsettling finding of paleontology. After long study of the fossil record, paleontologists see many species staying quite consistent in visible respects for long periods of time, only to be suddenly replaced by other species, which then remain stable for long periods as well. What is observed is long-term equilibrium, sometimes accompanied by change, but more often punctuated by rapid revolution.

Punctuated equilibrium is hard to square with the usual gradualist view of natural selection at the largest scales- that species change all the time, and would be expected to optimize and adapt to their environments continuously. But here, we learn that large populations are strongly inhibited from changing, by their inability to accumulate even the least unfavorable alleles. Sure, in their continual race against diseases and simliar frequently changing threats, they are adaptively optimizing all the time. But transiting to truly novel ecological niches or developmental paths is essentially impossible for such large populations, if it requires accumulating any kind of temporarily deleterious genetic material.

Conversely, small populations accumulate junk like this all the time. Small populations can happen at range edges, on islands, or even sympatrically by small alterations of mating behavior, gamete chemistry, and other means. Of course most don't succeed to make much of this isolation. But those who do may eventually gain some new form of selective advantage and return to displace their ancestors, accounting for the observed nature of the fossil record.

Human evolution is a good example. For most of our history, human populations were vanishingly small. That is why fossils have been extremely hard to find, despite very high interest.  What fossils we have indicate that the last 5 million years represent a constant diversification and extinction of species with small populations displaced by other species carrying various innovations in a halting, punctuated sort of process. The punctuations are more frequent and the changes more apparent, since there were never any widespread, huge populations of proto-humans. So we are mostly looking at what in other paleontological contexts would be the punctuation part of punctuated equilibrium.

So, natural selection is not only not the only mechanism of evolution, it postively retards evolutionary innovation if carried to extremes. That, among many other properties, is the secret of the flourishing diversity of large metazoan eukaryotes.