2349Soft Sweeps
origin prior to the environmental change) can occur Third, while hard sweeps from single mutations pro-
duce the strongest footprint for strongly selected allelesonly for adaptations from the standing genetic variation.
They are frequent even for a very low mutation rate with short fixation times, the possibility of fixation of
multiple alleles leads to an opposite trend: Soft sweeps⌰
u
→ 0 if the allele has a high relative selective advantage
R
␣
ⲏ 4; see Figure 5. The sweep pattern depends on with weaker footprints are more frequent for high ␣
b
.
Since the increase is only logarithmic, this trend is notthe strength of deleterious selection that the allele has
experienced in the old environment. For R
␣
⬎ 1, we very strong. Nevertheless, it could be visible for nucleo-
tides that are tightly linked to the selected allele in regionsexpect a weaker footprint with a narrower sweep region
than predicted for a hard sweep with the same selective of low recombination or in sufficiently small windows
around the selection target. A genome-wide study ofadvantage h␣
b
. We predict, however, that differences in
the sweep patterns are visible only for a minimum R
␣
the small-scale reduction of heterozygosity in narrow
windows of 200 bp around replacement or silent fixa-of 20–100. For ␣
d
⫽ 0, where the probability of multiple
fixations and the resulting effect on the sweep pattern tions has recently been performed for D. simulans by
Kern et al. (2002). We note that their counterintuitiveare strongest, this has been studied in a recent publica-
tion by Innan and Kim (2004). Using computer simula- finding of a sweep signature for preferred codon substi-
tutions, but not for replacement substitutions, matchestions, these authors indeed find much weaker selective
footprints if the alleles are taken from the standing our prediction of a stronger sweep signal for weakly
selected alleles close to the selection center. However,genetic variation. Since their minimum value of R
␣
is
1000, their results fit our predictions. a quantitative analysis of soft sweeps that also accounts
for other factors like population substructure is neededWe can summarize our results on soft sweeps in three
observations. First, evidence of a soft sweep does not before any conclusions can be drawn.
result in an easy criterion to distinguish adaptive substi-
We thank Sylvain Mousset and Wolfgang Stephan for fruitful discus-
tutions from the standing variation and recurrent new
sions and John Parsch for helpful comments on the manuscript. The
careful comments by Sally Otto and an anonymous reviewer led to
mutations. For a large parameter space we will not be
many clarifying changes. We also thank Pieter van Beek for help with
able to detect any difference between these adaptive
the computer simulations. This work was supported by an Emmy
scenarios. This confirms the conclusion of Orr and
Noether grant from the Deutsche Forschungsgemeinschaft to J.H.
Betancourt (2001), although partly for different rea-
sons. For high ⌰
u
ⲏ 0.1, soft sweeps are frequent in
both cases; for low ⌰
u
and R
␣
ⱗ 20 they either are rare
LITERATURE CITED
in both cases or do not lead to significant differences
Barton, N. H., 1995 Linkage and the limits to natural selection.
in the selective footprints. For a range of “interesting”
Genetics 140: 821–841.
substitutions, namely alleles with a large effect but a low
Barton, N. H., 1998 The effect of hitch-hiking on neutral genealo-
gies. Genet. Res. 72: 123–133.mutation rate, however, the linked nucleotide pattern
Catania, F., M. O. Kauer, P. J. Daborn, J. L. Yen, R. H. Ffrench-
could be informative.
Constant et al., 2004 World-wide survey of an Accord insertion
Second, soft sweeps are frequent in a limited but rele-
and its association with DDT resistance in Drosophila melanogaster.
Mol. Ecol. 13: 2491–2504.vant parameter space. We expect soft sweeps with charac-
Ewens, W. J., 2004 Mathematical Population Genetics, Ed. 2. Springer,
teristic patterns on the selective footprints for high ⌰
u
,
Berlin.
i.e., either if the population size is large or if the allelic
Falconer, D. S., and T. F. C. Mackay, 1996 Introduction to Quantita-
tive Genetics. Addison Wesley Longman, Harlow, Essex, UK.mutation rate is high, such as at mutational hotspots
Fisher, R. A., 1930 The Genetical Theory of Natural Selection. Oxford
or if the adaptation corresponds to a loss-of-function
University Press, Oxford.
mutation of the gene. We also expect soft sweeps for
Haldane, J. B. S., 1927 A mathematical theory of natural and artifi-
cial selection. Part V: selection and mutation. Proc. Camb. Philos.large adaptations with h␣
b
Ⰷ h⬘␣
d
(thus R
␣
Ⰷ 1) from
Soc. 23: 838–844.
the standing variation, even if the mutation rate is small.
Hansen, T. F., C. Pelabon, W. S. Armbruster and M. L. Carlson,
The effect of a soft sweep in this last case is a reduction
2003 Evolvability and genetic constraint in Dalechampia blos-
soms: components of variance and measures of evolvability. J.in the width of the sweep region relative to a hard sweep.
Evol. Biol. 16: 754–765.
A possible candidate for a soft sweep of this type is the
Houle, D., 1992 Comparing evolvability and variability of quantita-
evolution of DDT resistance in non-African populations
tive traits. Genetics 130: 195–204.
Innan, H., and Y. Kim, 2004 Pattern of polymorphism after strongof D. melanogaster. In recent studies of nucleotide and
artificial selection in a domestication event. Proc. Natl. Acad. Sci.
microsatellite variability in the region around an Accord
USA 101: 10667–10672.
insertion that is associated with DDT resistance, Schlenke
Kacser, H., and J. A. Burns, 1981 The molecular basis of domi-
nance. Genetics 97: 6639–6666.
and Begun (2004) and Catania et al. (2004) found evi-
Kaplan, N. L., R. R. Hudson and C. H. Langley, 1989 The “hitch-
dence for a selective sweep. The width of the sweep region,
hiking effect” revisited. Genetics 123: 887–899.
however, was much narrower in D. melanogaster than ex-
Keightley, P. D., 1996 A metabolic basis for dominance and recessi-
vity. Genetics 143: 621–625.pected under putatively very strong selection (Catania et
Kern, A. D., C. D. Jones and D. J. Begun, 2002 Genomic effects
al. 2004) and, as observed, for the “same” adaptation (with
of nucleotide substitutions in Drosophila simulans. Genetics 162:
a Doc insertion) in D. simulans (Schlenke and Begun
1753–1761.
Kim, Y., and W. Stephan, 2000 Joint effects of genetic hitchhiking2004).