Optimal population size allows maximum predictability of evolution

Jan 09, 2013

(Phys.org)—Evolution in very large populations of plants, animals or fungi can be predicted far less easily than one would expect. This has been shown by research at the Institute for Theoretical Physics of the University of Cologne and the Laboratory of Genetics at Wageningen University (published in the January 9 issue of PNAS, the Proceedings of the National Academy of Sciences). The researchers combined genetic information from a fungus with computer models which simulate the course of evolution. They concluded that, while evolution in small populations is known to be unpredictable, this is also the case for very large populations. There is an optimal population size for predicting evolutionary outcomes in every situation.

Evolution is exciting – you never know exactly what the result will be. Over time, mutations resulting in improved fitness have a good chance of surviving in a population. Mutations which decrease fitness, on the other hand, are much less likely to ultimately remain in a significant part of the population and thus influence the development of the species. But there are no safe bets, even for .

Researchers would like to know whether - and if so how - evolution can be predicted. For instance, one can, in retrospect, ask whether it could have been predicted that giraffes with long necks would eventually develop on earth through evolution. Researchers consider the question of the of evolution either theoretically, using computer models, or practically, via research in nature or in the laboratory. In the laboratory, the predictability of evolution can, for example, be examined by monitoring how often fast-evolving evolve along the same route. The more often the route and end result are the same, the more predictable evolution can be considered to be.

The Dutch-German team merged the two techniques, combining a laboratory approach with computer models. The Wageningen group used its laboratory to provide knowledge about the fitness profile of a fungus. All 256 possible combinations of eight mutations in various places in the DNA of the fungus were charted. The scientists calculated the effect on the growth of the fungus for each combination of mutations. The growth rate is an important component of the fitness of the fungus. Mutations with a positive effect on growth rates therefore have a greater chance of remaining definitively established after a large number of new generations. The likelihood is not one hundred percent, however.

In small populations, it is likely that each positive mutation that occurs accidentally will ultimately contribute to the evolution of the organism. The larger the population, the more positive mutations will arise, and the greater the chance will be that the best possible positive mutation is among them. And, as individuals with this mutation will be the fittest, it will repeatedly win the competition and contribute to evolution when the experiment is repeated. This means that, the larger a population, the more predictable its evolution should become. But this turns out to not quite be the case.

The German group processed the fitness data regarding the mutations of the Wageningen mould in a computer. They discovered that the evolutionary outcome is actually more difficult to predict in very large populations. This phenomenon is caused by the fact that they contain more individuals with not one but two or more mutations. Since the number of beneficial mutation combinations is much higher than the number of favourable mutations themselves, however, the predictability of evolution again declines in very large populations. In other words, evolution can be most easily predicted for populations that are not too small and not too big.

Explore further: Molecular gate that could keep cancer cells locked up

More information: www.pnas.org/content/110/2/571.abstract

Related Stories

Interacting mutations promote diversity

Jun 28, 2012

Genetic diversity arises through the interplay of mutation, selection and genetic drift. In most scientific models, mutants have a fitness value which remains constant throughout. Based on this value, they compete with other ...

Mutations: When benefits level off

Jun 08, 2011

Beneficial mutations within a bacterial population accumulate during evolution, but performance tends to reach a plateau. Consequently, theoretical evolutionary models need to take into account a "braking effect" in expected ...

Recommended for you

Invasive lionfish likely safe to eat after all

3 hours ago

Scientists have learned that recent fears of invasive lionfish causing fish poisoning may be unfounded. If so, current efforts to control lionfish by fishing derbies and targeted fisheries may remain the ...

Molecular gate that could keep cancer cells locked up

12 hours ago

In a study published today in Genes & Development, Dr Christian Speck from the MRC Clinical Sciences Centre's DNA Replication group, in collaboration with Brookhaven National Laboratory (BNL), New York, ...

User comments : 10

Adjust slider to filter visible comments by rank

Display comments: newest first

Whydening Gyre
1 / 5 (3) Jan 09, 2013
So... where does the human race fit on this curve?
Tausch
3.7 / 5 (3) Jan 09, 2013
Humans undergo life saving treatment.
Some treatments include genetic engineering.
Reducing the number of mutations decreasing fitness actively.
Genetic engineering modifies, deletes or adds to a human genome.

In this case evolution is dictated and predicted independent of population size.

Pretty soon we will not fit this curve.

Whydening Gyre
2.3 / 5 (3) Jan 09, 2013
Humans undergo life saving treatment.
Some treatments include genetic engineering.
Reducing the number of mutations decreasing fitness actively.
Genetic engineering modifies, deletes or adds to a human genome.

In this case evolution is dictated and predicted independent of population size.

Pretty soon we will not fit this curve.


A product of change becomes the ELEMENT of change... Bravo !!
Torbjorn_Larsson_OM
5 / 5 (3) Jan 09, 2013
This is pathway predictability of course. The curve is itself an example of generic predictability over all population scales.

@WG: Humans are way to the right. Many selective sweps are seen.

@Tausch: This result, and earlier ones on the effectiveness on selective sweeps and their ubiquitness in modern humans as population size increased has shown that humans evolves faster than ever before (as predicted).

Variation is not the only mechanism, see the article, and genetic engineering will only be another kind of (artificial) selection. It is said that human is the self-domesticated animal, has bred itself for sociability, and new methods will help that process.
Tausch
3 / 5 (2) Jan 09, 2013
Genetic engineers are selection bias free.

For disgruntled readers replace the period with a question mark in the above statement.
RobertKarlStonjek
5 / 5 (4) Jan 09, 2013
"Researchers would like to know whether - and if so how - evolution can be predicted. For instance, one can, in retrospect, ask whether it could have been predicted that giraffes with long necks would eventually develop on earth through evolution."

No. But you can identify the niche's that might develop and the kind of adaptations that could exploit those niches. For instance we may identify numerous niches that exist today and the kind of adaptations that could exploit those niches eg
locus breed to plague proportions but no predator exploits their number;
mosquitoes proliferate around stagnant water, a feed for a small insect predator but none exist;
etc
Torbjorn_Larsson_OM
not rated yet Jan 10, 2013
@Taush: The selection _is_ a bias. In any case, these things are evolution as we know it (but by new methods of artificial selection).
Tausch
3 / 5 (2) Jan 11, 2013
@TL
We now call artificial selection selective breeding.
http://en.wikiped...election

First paragraph.
Whydening Gyre
1 / 5 (2) Jan 12, 2013
@TL
We now call artificial selection selective breeding.
http://en.wikiped...election

First paragraph.


A change in definition due to relative magnitude and locality of POV - cool...
Tausch
1 / 5 (1) Mar 05, 2013
Life has evolution. Uncontested consensus.
If you conjecture planets have finite time spans, then all life as defined has finite time spans.

We call or label that 'predictability'.
All prediction is statistically based.
One planet with life.
One data point.
Statistically meaningless.
Conclusion?
An outlier to which statistics can make no prediction.