Weakly deleterious mutations and low rates of recombination limit the impact of natural selection on bacterial genomes
Morgan N. Price and Adam P. Arkin, Lawrence Berkeley National Lab
Abstract
Free-living bacteria are usually thought to have high effective population sizes, so that tiny selective differences
can drive their evolution.
However, because recombination is infrequent,
"background selection" against slightly-deleterious alleles should reduce the effective population size
(Ne) by orders of magnitude.
For example, for a well-mixed population with 1012 individuals
and a typical level of homologous recombination
(r/m=3, or nucleotides change due to
recombination at three times the mutation rate), we predict
that Ne < 107.
An argument for high Ne in bacteria has been the high genetic diversity within many bacterial "species,"
but this diversity may be due to population structure:
diversity across subpopulations can be far higher
than diversity within a subpopulation, which makes it difficult to estimate Ne correctly.
Given an estimate of Ne, standard population genetics models imply that selection should be sufficient to drive
evolution if Ne * s > 1, where s is the selection coefficient. We show that this remains
approximately correct if background selection is occuring or when population structure is present.
Overall, we predict that even for free-living bacteria with enormous populations, natural selection is only a significant
force if s is above 10-7 or so.
Downloads
The R images are up to 1 GB each (4 GB total).
- R code -- both simulations and theoretical calculations
- Steps for reproducing simulations
- R images from simulations of background selection with clonal evolution:
- N = 106 to 109, 106 sites with weakly-deleterious alleles (sback = 0.001), reversible mutation rate of 2*10-7, and a neutral focal site
- Similarly with N = 107 and a range of selection coefficients at the focal site
- R images from detailed small simulations of background selection with local recombination (5,000 individuals, 1,000 sites, sback = 0.02, U = 0.4)
- noR: R=0
- hi1: R=0.4,rSites=1
- hi10: R=0.4,rSites=10
- hi50: R=0.4,rSites=50
- lo1: R=0.1,rSites=1
- lo10: R=0.1,rSites=10
- lo50: R=0.1,rSites=50
- lo500: R=0.1,rSites=500
- R images from fast approximate simulations of background selection with local recombination:
- Matching the small simulations above (R=0.1 or 0.4, 1 to 500 sites):
- fast.noR R=0
- fast.lo1 R=0.1,rSize=1,rVar=0.0485
- fast.lo10 R=0.1,rSize=10,rVar=0.487
- fast.lo50 R=0.1,rSize=50,rVar=2.280
- fast.lo500 R=0.1,rSize=500,rVar=19.8
- fast.hi1 R=0.4,rSize=1,rVar=0.0492
- fast.hi10 R=0.4,rSize=10,rVar=0.469
- fast.hi50 R=0.4,rSize=50,rVar=1.955
- Large simulations (107 individuals, 106 sites, sback = 0.001, μ=2*10-7, 300 sites exchanged, Vr ≤ 9). (For N=1e8 and N=1e9 with R < 0.9, these are tab-delimited files, not images.)
- c6A: N = 1e6, R = 0
- c6B: N = 1e6, R = 0.001
- c6C_c: N = 1e6, R = 0.003
- c6D_b: N = 1e6, R = 0.01
- c6E: N = 1e6, R = 0.03
- c6F: N = 1e6, R = 0.1
- c6G: N = 1e6, R = 0.3
- w6H: N = 1e6, R = 0.9
- c7A_a: N = 1e7, R = 0
- c7B_c: N = 1e7, R = 0.001
- c7C: N = 1e7, R = 0.003
- c7D_f: N = 1e7, R = 0.01
- c7E: N = 1e7, R = 0.03
- c7F: N = 1e7, R = 0.1
- c7G: N = 1e7, R = 0.3
- w7H: N = 1e7, R = 0.9
- b8A: N = 1e8, R = 0
- b8B: N = 1e8, R = 0.001
- b8C: N = 1e8, R = 0.003
- b8D: N = 1e8, R = 0.01
- b8E: N = 1e8, R = 0.03
- b8F: N = 1e8, R = 0.1
- b8G: N = 1e8, R = 0.3
- w8H: N = 1e8, R = 0.9
- b9A: N = 1e9, R = 0
- b9B: N = 1e9, R = 0.001
- b9C: N = 1e9, R = 0.003
- b9D: N = 1e9, R = 0.01
- b9E: N = 1e9, R = 0.03
- b9F: N = 1e9, R = 0.1
- b9G: N = 1e9, R = 0.3
- w9H: N = 1e9, R = 0.9
- R images from simulations of population structure:
- 100 subpopulations of 100 individuals each, a reversible mutation rate of 10-6, and varying migration rates
- sims5: s = 0 or s = -5/nTot
- sims6: s = -50/nTot
- sims7: biased mutation (3x faster towards unfavorable allele), s = 0 or s = -5/nTot
- sims8: intermediate selection (s = -20/nTot or -10/nTot) with 0.01 or 0.1 migrants per generation
- R images from "rescaled" simulations with no subpopulation structure and reduced Ne: