SEARCH

SEARCH BY CITATION

Keywords:

  • conservation biology;
  • deleterious mutation rate;
  • extinction risk;
  • mildly deleterious mutations;
  • mutational load

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. References
  8. Appendix

Abstract We have analysed the effect of 288 generations of mutation accumulation (MA) on chromosome II competitive fitness in 21 full-sib lines of Drosophila melanogaster and in a large control population, all derived from the same isogenic base. The rate of mean log-fitness decline and that of increase of the between-line variance were consistent with a low rate (λ ≈ 0.03 per gamete and generation), and moderate average fitness effect [E(s) ≈ 0.1] of deleterious mutation. Subsequently, crosses were made between pairs of MA lines, and these were maintained with effective size on the order of a few tens. In these crosses, MA recombinant chromosomes quickly recovered to about the average fitness level of control chromosomes. Thus, deleterious mutations responsible for the fitness decline were efficiently selected against in relatively small populations, confirming that their effects were larger than a few percent.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. References
  8. Appendix

In large populations, the average reduction in fitness as a result of segregating deleterious mutations (relative to the fitness of an ideal genotype carrying none of these mutations, Crow & Simmons, 1983) mainly depends on the deleterious mutation rate (λ). However, because of partial recessiveness of deleterious effects, the main short-term threat for a large population is concealed in the heterozygous condition, and will only become manifest because of increased homozygosity after bottlenecks. This is usually measured as the rate of inbreeding depression, which depends on the rate, effect and degree of heterozygous expression of deleterious mutations. On the other hand, for populations maintained at small sizes during long periods, the main threat might arise from fixation of new deleterious mutations (Lynch et al., 1995). The fixation probability depends on the relationship between the homozygous deleterious effect (s) and the effective population size (Ne) and only mutations with small deleterious effects (say s < 5/Ne) can go to fixation. Thus, for small endangered populations, the threat from deleterious fixation will mainly depend on the rate of mildly deleterious mutation, typically those with an effect of a few per cent.

Several mechanisms have been invoked to explain how natural populations cope with the deleterious mutation threat. Sex and recombination could have evolved as systems helping to purify the genome from deleterious mutations (Kondrashov, 1988). Thus, for small populations with large λ, small amounts of sex and recombination can regenerate good genotypes from the limited number of available genotypes loaded with deleterious mutations, thus stopping the Muller's Ratchet. Furthermore, in large populations with synergistic epistasis, purifying selection is more efficient with sexual than with asexual reproduction. However, high deleterious mutation rates (about a new mutant per genome and generation) are required for improved purifying selection to compensate, in the long term, the twofold cost of anisogamic sexual reproduction. Conditions for the evolution of obligate sexuality are even more restrictive, requiring both large mutation rates and substantial deleterious effects.

Thus, the properties of deleterious mutation are relevant to many evolutionary and conservation issues. This is particularly true for mildly deleterious mutations, which have effects that are small enough to drift in small populations, but large enough to cause an appreciable fitness decline or a substantial disadvantage of asexual reproduction.

In Drosophila melanogaster, a number of experiments have been performed to obtain information on λ and s from the effect of mutation accumulation (MA) on several fitness components, particularly preadult viability. However, no general consensus has been attained (see reviews by García-Dorado et al., 1999; Keightley & Eyre-Walker, 1999; Lynch et al., 1999). Summarizing, in classical experiments (e.g. Mukai et al., 1972) large rates of mean viability decline (ΔM ≈ 0.01 per generation) were ascribed to MA, resulting in a high estimate of the mutation rate (λ > 0.3 per gamete and generation) with mild deleterious effect [average deleterious homozygotic effect E(s) <0.03]. On the other hand, more recent experiments and reanalysis of older ones detected smaller rates of decline (ΔM ≈ 0.002), giving low rates of viability deleterious mutation (λ ≈ 0.01) with moderate average effect [E(s) ≈ 0.1]. The later results question the relevance of mechanisms that could help populations to cope with the deleterious mutation load as, for example, synergistic epistasis. They also challenge the role of deleterious mutation in determining the evolutionary advantage of obligate sex and anisogamy. Finally, they limit the impact of mutation on the extinction risk of endangered populations and therefore have important consequences on conservation strategies.

Thus, experimental evidence helping to clarify the properties of deleterious spontaneous mutation is needed. The main interest is, of course, on the mutational effects on reproductive fitness but, for Drosophila, as far as we know, this has only been attempted by Houle et al. (1992). However, these authors detected that their control line had suffered external contamination (Houle et al., 1994), which may have biased the estimate of the fitness decline of the MA lines and therefore those of λ and E(s). Nonetheless, using the minimum distance (MD) method unconditional to the observed fitness decline gave ΔM ≈ 0.008, λ ≈ 0.03, E(s) ≈ 0.26 for relative fitness (García-Dorado et al., 1998). These rate and effects are somewhat larger than those obtained for viability in recent results and reanalysis (see above). This is to be expected, as some mutations can reduce fitness without reducing viability, and mutations reducing viability may also have deleterious effects on other fitness components.

We investigate the effect on competitive fitness of mutations accumulated at chromosome II during 288 generations in 21 randomly chosen surviving MA lines of the Fernández & López-Fanjul (1996) experiment, and in a control population that had been synchronously maintained with large effective population size. For these lines, ΔM ≈ 0.002, λ ≈ 0.015, E(s) ≈ 0.10 have been reported at generation 105 for noncompetitive viability (García-Dorado et al., 1998), and ΔM ≈ 0.002, λ ≈ 0.03, E(s) ≈ 0.08 for competitive viability at generation 250 (Chavarrías et al., 2001; chromosome II results adjusted for the whole haploid genome).

Material and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. References
  8. Appendix

MA lines and control

Initially, 200 MA lines and a control population were derived from a completely isogenic stock. The MA lines were thereafter maintained by brother–sister mating (details can be found in Fernández & López-Fanjul, 1996). The control was maintained in half-pint bottles with circular mating, bottle i in generation t contributing about 100 potential breeders, 50 to bottle i and 50 to bottle i + 1 of generation t + 1 (eight bottles were used up to generation 200 and 25 bottles thereafter). Even assuming that the effective size per bottle is as low as 20, as the number of breeders contributed per bottle and generation is roughly constant, the effective population size (Ne) of the control will be larger than about 300 up to generation 200 and larger than about 1000 thereafter.

By generation 255, 93 MA lines survived, but only a random sample of 30 were further maintained. Of these, 21 lines survived at generation 288 and were used in the present experiment.

Experiment 1. Competitive fitness assay

At generation 288, 10 males from each MA line were sampled and allowed to mate in a bottle to 10 virgin females from a balancer Cy/L2 stock (see Chavarrías et al., 2001), where one chromosome II carried the SM5 inversion inhibiting crossing-over and the Cy recessive lethal for curly wings, and the other one carried the dominant L2 mutation for lobe eye. Simultaneously, 230 virgin males were sampled from the control population and crosses were made in 23 bottles, each with 10 control males and 10 Cy/L2 females. From the progeny of each cross, males and virgin females not showing the lobe eye were placed into two bottles, that were maintained as two independent replicates during 10 generations by transfer of about 50 adults each generation. At generations 6 and 10, the numbers of Cy and wild-type offspring were recorded for each bottle. In both MA and control crosses, the average wild-type frequency did not change from generation 6–10 (see below). Thus, it was assumed that they had attained an overdominant equilibrium by generation 6. For each replicate cross, data from generations 6 and 10 were used to estimate the fitness of the wild chromosome II homozygotes (w/w) relative to that of Cy/w heterozygotes, given by (see Appendix)

  • image

where q is the observed frequency of Cy chromosomes after a selective equilibrium has been attained (i.e. half the frequency of curly type individuals) and v is the viability of w/w homozygotes relative to that of Cy/w heterozygotes. However, as the viability of the MA lines was not evaluated at generation 288, exact fitness values cannot be estimated. Thus, we computed the log-fitness of w/w relative to Cy/w assuming that selection against (w/w) homozygotes occurred at the viability level only, i.e. as

  • image

On the other hand, the homozygous mean viability (v) of MA chromosomes II relative to that of Cy heterozygotes had declined by 22% after 250 MA generations (Chavarrías et al., 2001). Thus, by generation 288 it is expected to have declined by 22% × (288/250) = 25%. Therefore, we estimated the approximate fitness (ω) assuming constant v = 0.75 for the MA chromosomes, and v = 1 for the control ones. In a few instances (six bottles), ω was negative and was assigned zero value. Therefore, this fitness measure was not log-transformed, but was expressed relative to the corresponding control mean.

Both W and ω data were analysed using two-way anova, main factors being generation and cross, with two replicates per cell. The mutational rate of increase in variance ΔV was estimated as the between-cross component of variance αb2 divided by the forward cumulated inbreeding coefficient (ΔV = Σb2/281). The rate of mean decline was estimated as ΔM = Dw/281, where Dw is the difference between the mean values of control and the MA crosses, both averaged over crosses and generations. Bateman–Mukai bound estimates (BM hereafter; see Mukai et al., 1972) for the deleterious mutation rate (λ per gamete and generation) and the average homozygous deleterious effect [E(s)], were also computed as λ > ΔM2V, E(s) < ΔV/ΔM. All ΔM, ΔV and λ were adjusted for the whole genome using a 2.5 factor.

Because of the relatively small number of lines assayed, we did not use the MD approach (García-Dorado, 1997) to analyse the present data.

Experiment 2. Assessment of the efficiency of natural selection

To assess the efficiency of natural selection to purge accumulated deleterious mutation in relatively small populations, 20 MA crosses (after eliminating a randomly chosen one from the 21 described above) were randomly paired to form 10 MA ‘hybrids’. For each pair, generation 11 offspring from replicate 1 of an MA cross was mixed with that of replicate 1 from the other cross, and the total was equally divided into two bottles. Identical procedure was followed with offspring of replicates 2. Thus, each hybrid was replicated, twice. Each hybrid replicate was maintained during 12 generations in two bottles as follows: each generation, the progenies obtained from the two bottles of the same replicate were mixed, and about 50 adults from this group were transferred to each of the two new bottles. Thus, on the average, each bottle contributed about 50 potential breeders to the next generation. Identical procedure was followed with a sample of 14 control crosses, producing seven control hybrids, which were replicated twice, each replicate being maintained in two bottles. The effective population size of each hybrid line is unknown and should be smaller than the corresponding number of breeders (∼100), although the contribution of potential breeders to the next generation is regulated separately by the resources available in each bottle. Thus, we consider that the effective population size of our hybrid lines [Ne (hybrids) ≈ 40] is about one order of magnitude lower than that of the control population during the MA experiment.

The wild-type frequency (pw) was assayed periodically to test for the increase in relative fitness of the wild homozygotes, because of the selection on recombinant chromosomes II. After a new equilibrium was approached (hybrid generations 6 and 12), log-fitness was estimated as in experiment 1.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. References
  8. Appendix

Experiment 1

The average frequency of wild-type offspring is given in Table 1 for the MA and control crosses at generations 6 and 10. As this frequency stabilized by generation 6, both fitness assays were used to estimate mutational parameters.

Table 1.  Mean wild-type frequency at the fitness assay (experiment 1).
 MA crossesControl crosses
Generation 60.367 ± 0.0230.574 ± 0.015
Generation 100.360 ± 0.0140.592 ± 0.018

The rates of decline in mean and of increase in variance (both highly significant, P < 10−3), together with the corresponding BM estimates of mutational parameters, are given in Table 2M, ΔV and haploid λ adjusted for the whole genome by a factor 2.5). These are qualitatively similar for W and ω fitness measurements. They are in agreement with previous viability estimates from the same MA lines and from the MD reanalysis of other data sets [ΔM ≈ 0.002, λ ≈ 0.01, E(s) ≈ 0.1, García-Dorado et al., 1999). They are also in agreement with MD estimates for competitive fitness obtained from Houle's et al., 1992] data [ΔM ≈ 0.008, λ ≈ 0.03, E(s) ≈ 0.26, García-Dorado et al., 1998], although the mean deleterious effect (and therefore the mean fitness decline) was smaller in our lines.

Table 2.  Fitness mutational parameters estimated in experiment 1.
TraitΔM*ΔV × 103*λ (BM) *E ( s ) (BM)
  • *

    Results adjusted for the whole genome.

  • Mutational estimates for log-fitness assuming selection at the viability level ( W ).

  • Mutational estimates for fitness (relative to the control) assuming relative viability 0.75 for the wild MA homozygotes (

  • ω

    ).

W0.0030 ± 0.00030.25 ± 0.090.037 ± 0.0180.083 ± 0.031
ω0.0044 ± 0.00050.62 ± 0.240.031 ± 0.0140.141 ± 0.057

Experiment 2

In the first generation after crossing, the mean wild-type frequency in the control hybrids was similar to that observed during experiment 1, but that of the MA hybrids significantly increased. Thus, the difference in wild-type frequency between the MA and control hybrids (Dp) was 0.177 ± 0.039, i.e. 81% that of the first experiment. This should be ascribed to a larger viability of the heterozygotes carrying wild chromosomes from different MA lines, because of the partial recessivity of the accumulated deleterious mutations. Figure 1 shows the subsequent evolution during experiment 2 of the wild-type frequency in both the MA and control hybrids and of the difference Dp, together with the corresponding standard errors. Up to generation 6, Dp remained roughly constant, suggesting that an apparent new equilibrium frequency was soon attained. At that moment, the log-fitness difference between control and MA hybrids was Dw = 0.20, i.e. 59% of the value for nonhybrid MA chromosomes estimated in experiment 1 (Dw = 0.34).

image

Figure 1. Wild-type frequency in both the MA (white circles) and control (black circles) hybrid crosses (experiment 2), and difference ( Dp ) between them (triangles), are plotted against generation number. The vertical bars above and below each point represent one standard error.

Download figure to PowerPoint

After generation 6, however, Dp consistently declined approaching a new plateau, and by generation 12 it was Dp = 0.039 ± 0.060 (nonsignificantly different from zero). This reduction should be ascribed to an increase in the fitness of wild-type MA chromosomes because of selection favouring recombinants relatively free of accumulated deleterious mutations. Thus, at generation 12, we obtain Dw = 0.038, i.e. 20% of the value estimated at hybrid generation 6 (Dw = 0.19), and 11% of that estimated in experiment 1 for nonhybrid MA chromosomes (Dw = 0.34). The stability of Dp during the first six hybrid generations is not unexpected, as it will take a time for favourable recombinants to occur and to reach substantial frequency subsequently.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. References
  8. Appendix

In experiment 1, we have found that mutations impairing competitive fitness occur at a low rate and have, on the average, moderate deleterious effect. Our BM estimates are in good agreement with previous estimates for viability obtained in the same MA lines, both in benign and competitive conditions (Chavarrías et al., 2001). They are also in good agreement with MD estimates for other viability and fitness Drosophila MA data (García-Dorado et al., 1999). All these estimates were obtained assuming no epistatic interactions, which is supported by the observed linear increase of the between-line variance through the MA experiment (García-Dorado et al., 2000). Considering that severe deleterious mutations should have been eliminated because of selection, and as long as few deleterious mutations have been accumulated per line, multiplicative gene action will be indistinguishable from additive effects. For the long-term data (Chavarrías et al., 2001 and this experiment), where according to our estimates several deleterious mutations are expected to have accumulated per line, estimates were obtained in the log scale to avoid possible bias from the assumption of between-loci additivity.

It should be borne in mind that our experimental procedure does not allow to detect the fraction of the fitness decline caused by mutations accumulating both in the MA lines and in the control population. Thus, some fitness decay could be masked. However, although some mildly deleterious mutations (typically those with an effect of a few per cent) could have accumulated in the control, most will have been eliminated by natural selection. Thus, from diffusion theory (Kimura, 1962), the fixation probability of a new mutation with deleterious effect as small as s = 0.02 will be only 7% that of a neutral one if Ne = 100, and it will dramatically decrease with increasing s or Ne. However, those mildly deleterious mutations will accumulate in the full-sib MA lines, where their fixation probability (for s = 0.02) is 96% that of a neutral one. Thus, selection against mildly deleterious mutations should be inefficient in these lines. In fact, even selection against moderately deleterious mutations should be quite inefficient, as our estimates [s > 0.05 and E(s) ≈ 0.1] suggest that most MA fitness decline can be attributed to this class of mutations. Furthermore, over the MA period considered, the rate of line extinction has increased only slightly, so that between-line selection is not expected to substantially bias our estimates (Chavarrías et al., 2001). Thus, the fitness decline because of mildly deleterious mutations will be much larger in the MA lines than in the control and therefore only a small fraction of such decline would be masked in our estimates. Simulation results (Caballero et al., 2002) show that, if mildly deleterious mutations occur at a rate large enough to cause substantial decline in the control, they would also produce a too dramatic decline in the MA lines, inconsistent with our experimental results. On the other hand, favourable mutations, could have occurred in the control, that had been maintained in conditions apparently similar to those of the fitness assay. Such favourable mutations would increase the fitness average estimated in the control assay, so that our low estimates of ΔM and λ could overestimate [and E(s) could underestimate] the real values. However, the control population did not show appreciable change for egg-to-adult viability between generations 100 and 200 (Caballero et al., 2002), so that there is no indication for any bias.

In Houle's et al., (1992) experiment, chromosome II had been maintained for 40 generations in heterozygous condition, thus being more efficiently sheltered from natural selection than those in our MA lines. This could explain why our BM estimate of the mean deleterious effect and the rate of log-fitness decline (Table 2) are smaller than the MD estimates based in Houle's et al. data for a similar measure of competitive fitness [E(s)=0.26, ΔM = 0.008; García-Dorado et al., 1998].

In experiment 2, the effective population size of the hybrid crosses (∼40) is about one order of magnitude smaller than that of the control population during the MA experiment. Therefore, only previously accumulated mutations with, roughly, s > 0.05 would be efficiently selected against in those hybrid crosses. Despite of this, the MA recombinant chromosomes quickly recovered about the same fitness level of the control chromosomes. Thus, the fitness decline in the MA lines should be mainly ascribed to mutations that were eliminated in the hybrid crosses, i.e. to mutations with s > 0.05. This means that mildly deleterious mutations do not contribute substantially to the fitness decline observed in the MA lines. This result suggests that chromosome II substantially similar to the original one have been reconstructed in the hybrid crosses through recombination, thus breaking most blocks of deleterious mutations that might have been generated during MA. This could very well occur in only 12 generations since, following our estimates (λ ≈ 0.035/2.5 per chromosome II), only about four deleterious mutations are expected to have accumulated per chromosome II after 288 MA generations.

Spontaneous mutations with very small deleterious effects (say s < 5 × 10−4; see Davies et al., 1999) can pass unnoticed in MA experiments but still be evolutionarily constrained. However, their fixation will only have substantial impact on population fitness in the very long-term, and their relevance to the evolution of sex is limited as they induce very small rates of genome degradation (Kondrashov, 1988). The results all indicate that the mutational decline of competitive fitness in MA lines is due to moderate to severely deleterious mutations occurring at low but considerable rate. These will not drift to fixation even in quite small populations (Ne of the order of a few tens). Nevertheless, because of their partial recessive gene action (García-Dorado & Caballero, 2000), these mutations can be found segregating in large populations, causing important inbreeding depression after a bottleneck. These results do not support the view that mildly deleterious mutation is a main cause of fitness erosion.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. References
  8. Appendix
  • Caballero, A., Cusi, E.,García, C. & García-Dorado, A. 2002. Accumulation of deleterious mutations: further D. melanogaster estimates and a simulation of the effects of selection. Evolution, in press.
  • Chavarrías, D., López Fanjul, C. & García-Dorado, A. 2001. The rate of mutation and the homozygous and heterozygous effects for competitive viability: a long-term experiment with Drosophila melanogaster. Genetics 158: 681693.
  • Crow, J.F. & Simmons, M.J. 1983. The mutation load in Drosophila. In: The Genetics and Biology of Drosophila (M.Ashburner, H. L.Carson & J. N.Thomson, eds), vol. 3c, pp. 1–35 . Academic Press, London.
  • Davies, E.K., Peters, A.D. & Keightley, P.D. 1999. High frequency of cryptic deleterious mutations in Caenorhabditis elegans. Science 285: 17481751.
  • Fernández, J. & López-Fanjul, C. 1996. Spontaneous mutational variances and covariances for fitness-related traits in Drosophila melanogaster. Genetics 143: 829837.
  • García-Dorado, A. 1997. The rate and effects distribution of viability mutations in Drosophila melanogaster. Evolution 51: 11301139.
  • García-Dorado, A. & Caballero, A. 2000. On the average degree of dominance of deleterious spontaneous mutations. Genetics 155: 19912001.
  • García-Dorado, A., Fernández, J. & López-Fanjul, C. 2000. Temporal uniformity of the spontaneous mutational variance of quantitative traits in Drosophila melanogaster. Genet. Res. 75: 4751.DOI: 10.1017/S0016672399004267
  • García-Dorado, A., López-Fanjul, C. & Caballero, A. 1999. Properties of spontaneous mutations affecting quantitative traits. Genet. Res. 74: 341350.DOI: 10.1017/S0016672399004206
  • García-Dorado, A., Monedero, J.L. & López-Fanjul, C. 1998. The mutation rate and the distribution of mutational effects of viability and fitness in Drosophila melanogaster. Genetica 102/103: 255265.
  • Houle, D., Hoffmaster, D.K., Charlesworth, B. & Assimacopoulos, S. 1992. The genomic mutation rate for fitness in Drosophila. Nature 359: 5860.
  • Houle, D., Hughes, K.A., Hoffsmaster, D.K., Ihara, J., Assimacopoulos, S., Canada, D. & Charlesworth, B. 1994. The effects of spontaneous mutation on quantitative traits. I. Variances and covariances of life history traits. Genetics 138: 773785.
  • Keightley, P.D. & Eyre-Walker, A. 1999. Terumi Mukai and the riddle of deleterious mutation rates. Genetics 153: 515523.
  • Kimura, M. 1962. On the probability of fixation of mutant genes in a population. Genetics 47: 713719.
  • Kondrashov, A. 1988. Deleterious mutation and the evolution of sexual reproduction. Nature 336: 435440.
  • Lynch, M., Blanchard, J., Houle, D., Kibota, T., Schultz, S., Vassilieva, L. & Willis, J. 1999. Spontaneous deleterious mutation. Evolution 53: 645663.
  • Lynch, M., Conery, J. & Bürger, R. 1995. Mutation accumulation and the extinction of small populations. Am. Nat. 146: 489518.
  • Mukai, T., Chigusa, S.I., Mettler, L.E. & Crow, J.F. 1972. Mutation rate and dominance of genes affecting viability in Drosophila melanogaster. Genetics 72: 333355.

Received 23 November 2001;accepted 22 March 2002

Appendix

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. References
  8. Appendix

Let ω, 1 and 0 (where ω < 1) be the fitness of genotypes w/w, Cy/w, Cy/Cy (wild, curly and unobserved lethal phenotypes), respectively. At equilibrium, the frequency (q̂) of the Cy chromosome each generation before any selection has occurred (or after all selection occurred) is

  • image(1)

Assume that, when we score phenotypes, selection on viability has eliminated all the Cy/Cy lethal homozygous, but that only a fraction of the natural selection against w/w has occurred, because of their viability up to that scoring stage being v times that of the Cy/w heterozygote (v < 1). Then, the observed (q) frequency of Cy chromosomes at scoring should be

  • image

Substituting q̂ from eqn 1 gives

  • image
  • so that

  • image

For the particular case v = ω (i.e. all selection at the viability level before scoring) gives

  • image

as expected, because in this case q̂=q.