Download Experimental studies of ploidy evolution in yeast

FEMS Microbiology Letters 233 (2004) 187–192
www.fems-microbiology.org
MiniReview
Experimental studies of ploidy evolution in yeast
Clifford Zeyl
*
Department of Biology, Wake Forest University, P.O. Box 7325, Winston-Salem, NC 27109, USA
Received 22 December 2003; received in revised form 5 February 2004; accepted 6 February 2004
First published online 25 February 2004
Abstract
Variation in the prominence of haploidy and diploidy is a striking feature of eukaryote life cycles that has not been explained
from an evolutionary point of view. The ease with which ploidy and other variables of population genetics may be manipulated in
yeast make Saccharomyces cerevisiae an excellent subject for experiments on the fitness effects of ploidy. Several hypotheses have
been advanced to explain the emphasis on diploidy in plants and animals, and yeast experiments have been particularly informative
for a few. Evidence suggests that diploids may enjoy an immediate advantage over haploids in masking harmful mutations, avoiding
the fitness cost such mutations impose on haploids. A convincing longer-term advantage for diploidy has proven elusive, and
different evolutionary explanations for the origin and for the subsequent maintenance of diploidy may be required.
Ó 2004 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved.
Keywords: Haploid; Diploid; Evolution; Selection; Yeast; Mutation
1. Studying ploidy evolution in yeast
We have two copies of each chromosome, and none
of us really knows, from an evolutionary perspective,
why. The immediate explanation, of course, is that we
received a copy from each parent, illustrating one obstacle to understanding the evolution of ploidy: it is
often inter-related with the evolution of sex, making it
difficult to separate two of the biggest remaining questions in evolutionary biology. The fact that plants and
animals are diploid for most of their life cycles encourages the assumption that diploidy is an improvement on
haploidy. But there is actually a good deal of diversity in
life cycles among eukaryotes, with haploidy particularly
prominent among algae and fungi. It has proven surprisingly difficult to identify an evolutionary advantage
to being diploid that is broadly applicable across many
groups of organisms.
Fungi display a diversity of life cycles, with and
without sex, and with varying emphasis on haploidy and
*
Tel.: +336-758-4292; fax: +336-758-6008.
E-mail address: [email protected] (C. Zeyl).
diploidy. As an experimental model, yeast offers not
only the opportunity to control ploidy independently of
sex, but also many advantages of genetic and genomic
knowledge and techniques. Particularly important is the
ease with which diploid yeast strains may be constructed
from haploids as doubled-haploids, furnishing pairs of
strains that are identical except for ploidy [1]. Doubled
haploids are homozygous except at the mating-type locus, while heterozygous diploids may be produced by
crossing unrelated haploids, allowing experiments that
distinguish between theories proposing an immediate
physiological advantage to diploidy, longer-term evolutionary advantages, and those based on heterozygosity. Immediate effects of ploidy changes on fitness can
therefore be examined in a rigorous, reproducible way.
On a longer timescale, yeast also permit the study of
evolution in huge populations and over thousands of
generations, thanks to their short generation time. The
evolutionary effects of ploidy can therefore be observed
directly over a period of months or years, as experimental populations adapt to their laboratory environments. Experimental populations can be established as
replicate clones of a single genotype, so that populations
are initially not only identical but genetically homoge-
0378-1097/$22.00 Ó 2004 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.femsle.2004.02.007
188
C. Zeyl / FEMS Microbiology Letters 233 (2004) 187–192
neous. All adaptation that is subsequently observed
therefore comes from the selection of new mutations, clarifying the effects of ploidy on the process of
adaptation.
Yeast populations are typically maintained in the lab
for evolutionary study in one of two ways: in chemostats, or by serial transfer. Chemostats maintain a constant environment by supplying fresh growth medium
and removing old medium and cells at equal and constant rates. Chemostat populations adapt to an environment in which nutrients are present at constant
levels, one or more of them at levels that limit population density [2]. In serial transfer, or batch culture, a
random sample of each population is taken at regular
intervals (typically daily) and transferred to a culture
vessel containing fresh medium. Serially transferred
populations therefore experience an environment that
fluctuates regularly between feast and famine [3]. Small
yeast populations can also be maintained by serial
transfer of individual colonies grown on agar plates.
What have yeast experiments so far contributed to
our understanding of the evolutionary significance of
ploidy? This review summarizes those contributions,
and illustrates that the picture is far from simple. The
experiments require some context, so in addition to
discussing the yeast experiments to date, I will summarize theoretical work on the evolution of ploidy.
2. Physiological effects of ploidy
Ploidy is a fundamental genetic trait with major
genetic and genomic implications, so it is not surprising
that most of the theories regarding ploidy evolution are
genetic. But changes in ploidy also introduce potentially significant cytological and physiological changes,
which could provide immediate advantages to one
ploidy or the other without requiring any long-term
evolutionary processes. Under some culture conditions,
haploid yeast cells are smaller than diploids, and
therefore have a larger surface area to volume ratio [4].
Haploids could therefore have a competitive advantage
in nutrient-poor conditions, if growth rates are limited
by nutrient transport rates, while diploids may have the
advantage in rich media [5,6]. However, there is little
evidence in support of either suggestion. The general
yeast lore that diploids outgrow haploids in rich medium is not supported by published experiments, which
do not show any consistent growth difference between
ploidies [1,6]. Haploid advantages have been reported
in chemostat cultures in which the limiting nutrient,
organic phosphate, was present at very low concentrations, but not in other nutrient-limited media [6] and
not in serially transferred cultures in nutrient-poor
media [1].
3. Masking harmful mutations
Since neither ploidy has consistently been shown to
experience any immediate physiological advantage, attention turns to genetic hypotheses for the evolution of
diploidy. Here too is a theory that offers immediate
benefit: A simple observation of basic genetics is that
harmful mutations can be partly or completely masked
in heterozygotes if those mutations are partly or completely recessive, while haploids will fully express all
harmful mutations. Harmful mutations usually exist at
low frequencies as the result of a balance between selection removing mutations and mutation supplying new
ones, so they will be present mostly in heterozygotes,
lending theoretical credibility to this theory. An immediate diploid benefit of masking has been observed in
populations experimentally mutagenized with EMS [7]
although the benefit was short-lived.
A more empirical question is whether harmful mutations typically are recessive enough for diploids to be
shielded from their fitness effects. Quantitative and
population geneticists quantify dominance using a coefficient for which values from 0 to 1 represent completely recessive to completely dominant alleles,
respectively. If on average this dominance coefficient h is
less than 0.5, diploids will have an immediate fitness
advantage over haploids. For Drosophila [8] and other
organisms [9] this condition appears to be met. What
about yeast?
Yeast are ideally suited to a type of experiment that
can shed a great deal of light on the nature of random
mutations: mutation accumulation (MA) experiments.
Such experiments address a major challenge for geneticists. It is difficult to determine directly the attributes of
harmful mutations (such as how frequently they occur,
or on average how dominant they are, and what their
fitness costs are) because mutations are very rare. The
usual trick of artificially generating abundant mutations
using mutagens such as EMS is unsatisfactory because
the properties such as dominance and fitness cost of
chemically induced mutations may be different from
those of the all-natural, spontaneous kind. MA experiments are attempts to gather and examine as many
random, spontaneous mutations as possible while minimizing the effects of selection, which would otherwise
bias the sample towards less harmful mutations, and less
dominant ones if the experiment was performed with
diploids. To accomplish this, the experimenter establishes a series of initially identical MA lines from an
ancestral genotype, and propagates those lines by
transferring some minimum number of cells at regular
intervals (Fig. 1). A typical way to do this is to transfer
single colonies from one agar plate to the next, effectively bottlenecking that population to the single cell
that founded the randomly chosen colony. Bottlenecking the number of cells that are transferred ensures that
C. Zeyl / FEMS Microbiology Letters 233 (2004) 187–192
Serial transfer of random single colonies –
mutation accumulation (MA)
Fitness assays and
estimation of
parameters (fitness
cost, dominance, rate
of occurrence) of
spontaneous mutation
ancestor
replicate MA lines
Fig. 1. Schematic illustration of a mutation accumulation (MA) experiment. Multiple replicate MA lines are simultaneously established
from a single ancestor and propagated through a series of population
bottlenecks (often a single colony, representing a bottleneck of the
single cell that founded it). Colonies are chosen randomly. Since there
is no selection, other than that against lethal mutations, random mutations (represented as gray dots on linear chromosomes) accumulate.
Following MA, the fitness of each line is compared to the fitness of the
ancestor. From the rate of decline in average fitness of the lines, and
the increase in the variance in fitness among lines, parameters such
as the mutation rate and the average effect of each mutation can be
estimated.
the effect of selection will be minimized, as the genotypes
making up the next generation are determined by random chance rather than by competitive ability or any
other measure of fitness. As the replicate lines are
propagated, random mutations gradually accumulate,
and after numerous bottlenecks and a sufficiently large
number of generations (usually hundreds or thousands,
for yeast), the experimenter is rewarded with a collection
of random mutations distributed among the MA lines.
The properties of those random mutations can then be
examined using fitness assays and statistical analyses.
Korona [10] used such an experiment to compare the
fitness effects of random mutations in haploids and
diploids. First, MA was performed in haploids, using a
genetic ploy to accelerate mutation accumulation: the
DNA mismatch repair system was inactivated by deletion of the MSH2 gene (which in addition to increasing
the speed of MA may also alter the spectrum of mutations that accumulate). After random mutations had
accumulated, ploidy was manipulated. Using crosses
between and within MA lines, diploids that were heterozygous or homozygous for the accumulated mutations were constructed, and the fitnesses of all genotypes
were assayed.
Fitness values for heterozygotes showed no particular
relationship to the fitness values of the haploids used to
construct them, and gave a somewhat unclear picture of
the nature of the random mutations and of the interactions among them. Dominance and selection coefficients were negatively correlated – in other words,
mutations with the most severe effects were also the
most recessive, as in Drosophila [8]. The data also suggested epistasis, or interactions between mutations at
189
different loci in their fitness effects. With these complications, Korona’s estimate of h ¼ 0:08, while suggesting
that harmful mutations are recessive enough to provide
diploids a large advantage in masking them, must be
taken with caution. An additional complication is that
the fitness effects of these mutations may be much
greater in harsh environments, such as under thermal
stress [11] than in the usual benign lab environment.
Even if mutations are frequent and recessive enough
to give diploids an immediate advantage by masking
those mutations, there is a longer-term problem with
this hypothesis. Diploids have twice as many genes that
can mutate, and the masking effect that initially spares
diploids from the fitness cost of mutations then allows
those mutations to accumulate, hidden from natural
selection by that same heterozygosity. Eventually, when
mutation and selection have reached a balance, a lineage
of diploids will be less fit than a lineage of haploids [12].
Attempts have been made to rescue the masking hypothesis by allowing co-existing haploids and diploids to
interbreed [13] or by assuming high mutation rates and a
pattern of selection in which fitness is an all-or-none
outcome of the number of mutations a genotype carries
[14]. Although this pattern of ‘‘truncation selection’’
may apply in some populations that are controlled by an
upper limit on their density, there is no indication that
truncation selection is typical, so these rescue attempts
are unconvincing as explanations for a trait as widespread as diploidy.
4. Genetic variation in diploids
Because the theories described to date all offer immediate advantages for diploidy, they are all candidates
for explaining the evolutionary origin of diploidy, and
its initial spread. What remains to be discussed is a variety of theories that provide longer-term advantages.
Natural selection, the reproductive advantage of some
types over their competitors, has no predictive capacity,
so a trait can be favored only if it provides an immediate
advantage. For this reason it is important to distinguish
between the long-term effects of a trait such as diploidy,
and explanations for its evolution. This means that even
though, as discussed above, the long-term effect of
masking deleterious mutations is to reduce diploid fitness, the immediate benefit of masking may be enough
to allow diploidy to evolve, despite its eventual cost of a
greater load of harmful mutations [12], especially because any subsequent reversion to haploidy will reduce
fitness by fully expressing the accumulated harmful
mutations. This blindness of selection to the future poses
a problem for the remaining theories of the evolution of
diploidy discussed below, since their hypothesized advantages generally would not have appeared for many
generations [6].
190
C. Zeyl / FEMS Microbiology Letters 233 (2004) 187–192
Perhaps the most familiar and intuitive of the genetic
hypotheses for diploid advantage focuses on the greater
potential for genetic variation in diploid populations.
There are two ways that diploids might carry more
variation. Most simply, diploids can be heterozygous.
At both the individual and population levels, heterozygosity may store variation and permit a faster or greater
response to changes in the selective environment. It is a
common assumption in conservation biology that a
more heterozygous population faces a brighter future
because of greater adaptability, but the correlation between heterozygosity and fitness is generally weak and
inconsistent [15] and no such relationship has been
documented for yeast.
A second way that diploids may be more variable
stems from a contrast between the ways that haploids
and diploids produce gametes. Diplonts (organisms that
are diploid for most of their life cycles) produce genetically diverse gametes by meiosis. Haplonts produce
gametes during the haploid phase of their life cycles, so
their gametes are generated by mitosis and are identical
to each other. The hypothesized advantage of diploidy
here is that it may help to diversify your entries in the
contest that is the next generation. This would be a
particular advantage when competition among gametes
is intense [16]. This idea has received little experimental
study. It is better suited to the life cycles of plants and
invertebrates than to yeast, because there is intense
competition among the numerous pollen grains or
sperm of many species, while each yeast individual
produces only one gamete.
It may seem obvious that the ability to vary would be
favored by natural selection, but in fact it is anything
but obvious, and in many cases it is probably not true.
The ability to vary genetically (because of, for example,
diploidy) is a trait of entire populations or species, but
what selection acts on are the traits of individuals. The
ability to generate a more diverse population over many
generations cannot be selected unless it is a byproduct of
selection for individuals whose highly fit genetic strategy
involves producing variable offspring. The genetic basis
for that strategy could then spread throughout the
population. However, as in the quest for an evolutionary explanation for sex, it has proven difficult to devise a
plausible theory that could provide a consistent advantage for producing more variable offspring [16]. So even
if diploidy does increase variability in the short term, it
remains to be convincingly demonstrated that this
would be an evolutionary advantage.
5. Heterozygote advantage
As noted above, there is no evidence that yeast populations with high levels of heterozygosity respond more
quickly to selection. A different possibility is that, rather
than leading to long-term adaptability, heterozygosity is
itself the advantage. Here too, there is apparently no
evidence that yeast with high levels of heterozygosity are
generally fitter than haploids. But the MAT locus may
be an exception, providing one example of heterosis –
heterozygote advantage. The MAT locus determines the
mating type of a haploid yeast, expressing whichever of
two alleles has most recently transposed into it to produce MATa or MATa strains [17]. Only strains of opposite mating type will mate, so unless produced by
some laboratory artifice, diploids are heterozygous at
the MAT locus. MAT heterozygosity is required for the
ability to undergo meiosis and sporulation, and also
regulates the expression of many other genes [16]. It is
difficult to disentangle these roles in attempting to explain fitness effects of MAT heterozygosity, and at least
two of these roles come together in experiments by
Birdsell and Wills [18]. Their primary aim was to look
for a fitness advantage of sex, and their observations of
the benefits of the meiosis and recombination made
possible by MAT heterozygosity suggest that they found
it. However, in several but not all strains, MAT heterozygosity itself provided a fitness advantage even when
no meiosis or recombination had occurred. This result
effectively separates MAT heterozygosity from sex and
indicates heterosis at the MAT locus on some genetic
backgrounds. So far this appears to be a yeast idiosyncrasy, not a general phenomenon in diploid organisms
[15], so in the search for a broad explanation of diploidy,
it is an interesting anecdote.
6. Rates of adaptation in haploids and diploids
The final theory of diploid advantage to be considered here compares rates of adaptation in haploids and
diploids. Which ploidy will accumulate adaptive mutations most quickly? The most obvious effect of ploidy on
adaptation is that diploids have twice as many genes
that can mutate to superior alleles, and thus would appear to have the advantage. But diploids also pay the
price of masking recessive adaptive mutations, missing
out on their beneficial effects. Selection for those
adaptive mutations is therefore weakened in diploid
populations.
A yeast chemostat experiment comparing rates of
diploid and haploid adaptation [2] appeared to show
that the first effect is more important: adaptive mutations swept through diploid populations more often
than they did in haploids. As one of the first and very
few experimental studies of the evolutionary effects of
ploidy differences, and one that showed a clear diploid
advantage, this paper has come under intense scrutiny
[19,20]. Analysis of the speed with which adaptive mutations swept through the experimental populations re-
C. Zeyl / FEMS Microbiology Letters 233 (2004) 187–192
vealed an important discrepancy. Sweeps occurred
about every 65 generations in haploid populations and
every 50 generations in diploids. The average fitness
advantage resulting from an adaptive mutation was
10%. But with that average fitness gain per mutation, it
should have taken an average of over 400 generations
for a mutation to sweep through a population [20]. The
apparent speed with which adaptive mutations were
selected is incompatible with the mathematics of selection. The most likely explanation is that populations
were polymorphic, not fixed for new mutations: rather
than sweeping quickly through entire populations,
multiple adaptive mutations co-existed and evolving
populations were genetically variable [20]. The implications of this experiment for ploidy evolution remain
unclear.
Orr and Otto [21] advanced the field with a purely
theoretical study that investigated which of the two
conflicting effects of diploidy was more important, more
frequent adaptive mutations or more efficient selection.
The answer depends somewhat on the dominance coefficient h of adaptive mutations, and more directly on
population size. Unless adaptive mutations are nearly
dominant (h is well above 0.5), the selection of adaptive
mutations is seriously impeded in diploids. In addition,
large population sizes wipe out the diploid advantage of
producing more adaptive mutations, because if a population is large enough, one or more adaptive mutations
are produced each generation anyway, regardless of
ploidy. Under those conditions, the rate of adaptation is
not limited by the supply of new mutations, but by how
quickly selection can spread them throughout the population. Therefore, haploids should adapt faster when
population sizes are large, but diploids may have the
advantage with small population sizes. Fig. 2 illustrates
the theorized interplay of dominance of mutations and
population size in determining whether adaptation is
faster in haploids or diploids. A recent yeast experiment
using serially transferred evolving populations supported this prediction [3]. Haploids adapted more
quickly than diploids at large population sizes, but
with much smaller population sizes no difference was
detected.
Experiments that could more directly test theories of
ploidy evolution would pit haploids and diploids in direct competition with each other. Either chemostat or
serial transfer experiments could be started with equal
mixtures of otherwise haploid and diploid genotypes.
Assuming the haploids are all heterothallic (HO
knockout) and of the same mating type, both ploidies
could be stably maintained long enough to observe
whether one consistently displaced the other, under
varying conditions as warranted by the theory being
tested. One challenge to long-running experiments
would be the tendency for asexual yeast cultures to
change ploidies or become aneuploid [3,7].
191
Fig. 2. Rate of adaptation of diploids relative to haploid adaptation
rate as a function of population size (N ) for values of the dominance
coefficient (h) ranging from nearly recessive (h ¼ 0:1) to nearly dominant (h ¼ 0:9). The function giving relative rate of fitness increase in
diploids is ðh2 ÞðlnðN e1=ð2NvÞ 1ÞÞ= lnðN ðe1=ð4NvÞ 1ÞÞ [21]. The rate of
adaptive mutation v is assumed to be 108 per genome per generation;
the product of adaptive mutation rate and population size determines
the availability of adaptive mutations so any equivalent products of
mutation rate and population size would give the same results. Mutations are assumed to be selected in the heterozygous state in diploids.
Solid horizontal line indicates the ratio of 1 corresponding to equal
haploid and diploid fitness. Hatched region indicates parameter space
corresponding to diploid advantage. Note log scale of x-axis.
7. Conclusions
Reviewing our current understanding of the relationships between haploidy, diploidy and fitness leads
inevitably to the conclusion that we still do not have a
convincing, broadly applicable explanation for the
prominence of diploidy. It is worth noting that the hypotheses outlined above are not mutually exclusive, and
a complete understanding of ploidy evolution may require some combination of these hypotheses. Masking
of harmful mutations may offer an immediate advantage
of diploidy. Before the diploids pay the resulting longterm cost of a greater load of mutations, some other
advantage may take effect, such as faster adaptation if
populations are small enough and adaptive mutations
dominant enough. The evolutionary explanations for
the origin and for the long-term success of diploidy need
not be the same.
For the purpose of adaptive flexibility, the duplication of a few key genes in a haploid may be as good as
diploidy. Unlike diploidy per se, gene duplication has
proven beyond doubt to be adaptive in several yeast
experiments. Adaptation to low concentrations of organic phosphates [22,23], to antimycin A [24], and to
limiting glucose [25] has involved chromosome rearrangements and gene duplications. The relatively high
rate of gene duplication in evolving laboratory yeast
populations mirrors a recent result of DNA sequence
comparisons among nine eukaryote genomes [26] duplicated genes arise at unexpectedly high rates. Most are
then silenced at high rates.
With both experimental and comparative methods
yielding progress in the study of gene duplication, we
192
C. Zeyl / FEMS Microbiology Letters 233 (2004) 187–192
have at least begun to understand the evolution of
changes in gene copy number. Further advances in the
study of genome dynamics may provide new insight into
ploidy evolution as well.
References
[1] Mable, B.K. (2001) Ploidy evolution in the yeast Saccharomyces
cerevisiae: a test of the nutrient limitation hypothesis. J. Evol. Biol.
14, 157–170.
[2] Paquin, C. and Adams, J. (1983) Frequency of fixation of adaptive
mutations is higher in evolving diploid than haploid yeast
populations. Nature 302, 495–500.
[3] Zeyl, C., Vanderford, T. and Carter, M. (2003) An evolutionary
advantage of haploidy in large yeast populations. Science 299,
555–558.
[4] Weiss, R.L., Kukora, J.R. and Adams, J. (1975) The relationship
between enzyme activity, cell geometry, and fitness in Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA. 72, 794–798.
[5] Lewis Jr., W.M. (1985) Nutrient scarcity as an evolutionary cause
of haploidy. Am. Nat. 125, 692–701.
[6] Adams, J. and Hansche, P.E. (1974) Population studies in
microorganisms. I. Evolution of diploidy in Saccharomyces
cerevisiae. Genetics 76, 327–338.
[7] Mable, B.K. and Otto, S.P. (2001) Masking and purging
mutations following EMS treatment in haploid, diploid and
tetraploid yeast (Saccharomyces cerevisiae). Genet. Res. 77, 9–26.
[8] Crow, J.F. and Simmons, M.J. (1983) The mutation load in
Drosophila. In: The Genetics and Biology of Drosophila (Ashburner, M., Carson, H.L. and Thompson, J.N., Eds.), vol. 3c, pp.
1–35. Academic Press, London.
[9] Lynch, M., Blanchard, J., Houle, D., Kibota, T., Schultz, S.,
Vassilieva, L. and Willis, J. (1999) Perspective: spontaneous
deleterious mutation. Evolution 53, 645–663.
[10] Korona, R. (1999) Unpredictable fitness transitions between
haploid and diploid strains of genetically loaded yeast Saccharomyces cerevisiae. Genetics 171, 77–85.
[11] Szafraniec, K., Borts, R.H. and Korona, R. (2001) Environmental
stress and mutation load in diploid strains of the yeast Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 98, 1107–1112.
[12] Crow, J.F. and Kimura, M. (1965) Evolution in sexual and
asexual populations. Am. Nat. 99, 439–450.
[13] Perrot, V., Richerd, S. and Valero, M. (1991) Transition from
haploidy to diploidy. Nature 351, 315–317.
[14] Kondrashov, A.S. and Crow, J.F. (1991) Haploidy or diploidy:
which is better. Nature 351, 314–315.
[15] David, P. (1998) Heterozygosity-fitness correlations: new perspectives on old problems. Heredity 80, 531–537.
[16] Bell, G. (1982) The Masterpiece of Nature: The Evolution and
Genetics of Sexuality. University of California Press, Berkeley,
CA. 635pp.
[17] Herskowitz, I., Rine, J. and Strathern, J. (1992) In: The Molecular
and Cellular Biology of the Yeast Saccharomyces cerevisiae: Gene
Expression (Jones, E.W., Pringle, J.R. and Broach, J.R., Eds.),
vol. 2, pp. 583–656. Cold Spring Harbor, Plainview.
[18] Birdsell, J. and Wills, C. (1996) Significant competition advantage
conferred by meiosis and syngamy in the yeast Saccharomyces
cerevisiae. Proc. Natl Acad. Sci. USA 93, 908–912.
[19] Dykhuizen, D.E. (1990) Experimental studies of natural selection
in bacteria. Annu. Rev. Ecol. Syst. 21, 373–398.
[20] Otto, S.P. (1994) The role of deleterious and beneficial mutations
in the evolution of ploidy levels. In: Theories for the Evolution of
Haploid–diploid Life Cycles (Kirkpatrick, M., Ed.). American
Mathematical Society, Providence, RI.
[21] Orr, H.A. and Otto, S.P. (1994) Does diploidy increase the rate of
adaptation? Genetics 136, 1475–1480.
[22] Hansche, P.E. (1975) Gene duplications as a mechanism of genetic
adaptation in Saccharomyces cerevisiae. Genetics 79, 661–
674.
[23] Adams, J., Puskas-Rosza, S., Simlar, J. and Wilke, C.M. (1992)
Adaptation and major chromosomal changes in populations
Saccharomyces cerevisiae. Curr. Genet. 22, 13–19.
[24] Dorsey, M., Peterson, C., Bray, K. and Paquin, C.E. (1992)
Spontaneous amplification of the ADH4 gene in Saccharomyces
cerevisiae. Genetics 132, 943–950.
[25] Dunham, M.J., Badrane, H., Ferea, T., Adams, J., Brown, P.O.,
Rosenzweig, F. and Botstein, D. (2002) Characteristic genome
rearrangements in experimental evolution of Saccharomyces
cerevisiae. Proc. Natl. Acad. Sci. USA 99, 16144–16149.
[26] Lynch, M. and Conery, J.S. (2000) The evolutionary fate and
consequences of duplicate genes. Science 290, 1151–
1155.