• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of geneticsGeneticsCurrent IssueInformation for AuthorsEditorial BoardSubscribeSubmit a Manuscript
Genetics. Nov 2006; 174(3): 1387–1395.
PMCID: PMC1667051

Cumulative Effects of Spontaneous Mutations for Fitness in Caenorhabditis: Role of Genotype, Environment and Stress


It is often assumed that the mutation rate is an evolutionarily optimized property of a taxon. The relevant mutation rate is for mutations that affect fitness, U, but the strength of selection on the mutation rate depends on the average effect of a mutation. Determination of U is complicated by the possibility that mutational effects depend on the particular environmental context in which the organism exists. It has been suggested that the effects of deleterious mutations are typically magnified in stressful environments, but most studies confound genotype with environment, so it is unclear to what extent environmental specificity of mutations is specific to a particular starting genotype. We report a study designed to separate effects of species, genotype, and environment on the degradation of fitness resulting from new mutations. Mutations accumulated for >200 generations at 20° in two strains of two species of nematodes that differ in thermal sensitivity. Caenorhabditis briggsae and C. elegans have similar demography at 20°, but C. elegans suffers markedly reduced fitness at 25°. We find little evidence that mutational properties differ depending on environmental conditions and mutational correlations between environments are close to those expected if effects were identical in both environments.

Keywords: mutation accumulation (MA), mutation rate, mutational variance, mutational correlation, nematode

THE importance of deleterious mutations to the evolutionary process is well appreciated (Morgan 1903; Haldane 1927; Fisher 1930; Sturtevant 1937; Kondrashov 1988), and much effort has been expended in understanding the processes by which new mutations arise and their effects on the phenotype and on fitness (reviewed by Simmons and Crow 1977; Drake et al. 1998; Keightley and Eyre-Walker 1999; Lynch et al. 1999; Houle and Kondrashov 2006). Drake, especially, has emphasized the remarkable consistency of the per-genome mutation rate across very broad taxonomic categories, but has also noted that there is considerable variation within those broad taxa (e.g., Drake et al. 1998). The idea that certain mutational properties vary between related species and even within species is venerable (Sturtevant 1937 and references therein), but there is as yet nothing approaching a comprehensive understanding of the variation in mutational properties—rate, distribution of effects, environmental sensitivity, molecular spectrum—of any species or group of closely related species, with the possible exception of the bacterium Escherichia coli (Matic et al. 1997; Sniegowski et al. 1997; Bjedov et al. 2003).

An intriguing but almost completely unsubstantiated possibility (but see thel 1987; Bjedov et al. 2003) is that mutation rates are themselves an evolutionarily optimized property (Fisher 1930; Sturtevant 1937; Kimura 1960, 1967; Leigh 1970, 1973; Kondrashov 1995a,b; Dawson 1998). Because the vast majority of mutations with observable effects are deleterious, it is generally accepted that direct selection (almost) always favors a reduction in the mutation rate, with an optimal mutation rate of zero, at least in sexual taxa (e.g., Leigh 1973; Drake et al. 1998; Sniegowski et al. 2000). The fact that mutations occur is attributed to a “cost of fidelity” (Kimura 1967), whereby increasing the fidelity of DNA replication imposes a physiological cost to the organism. Decreasing the mutation rate below some lower bound costs more physiologically than it is worth genetically, so the optimum (nonzero) mutation rate represents the equilibrium between direct selection to reduce the number of deleterious mutations and indirect selection to minimize the cellular resources devoted to replication fidelity.

The evolution of mutation rates cannot be uncoupled from the mutational effects, because the strength of selection on a modifier of the mutation rate depends on the average effect of a mutation in the population (Kondrashov 1995a,b). In principle, two groups with the same mutation rate but different average effects could experience different degrees of selection to modify the mutation rate and thus evolve different rates, or vice versa. To complicate matters further, mutational effects themselves can vary depending on the environmental context; for example, there is some evidence that deleterious effects are magnified in harsh environments relative to their effects in benign environments (Kondrashov and Houle 1994; Shabalina et al. 1997; Szafraniec et al. 2001).

The degree to which mutational effects are limited to particular environmental contexts has an important implication. If most deleterious mutations are uniformly deleterious, they will be efficiently removed by natural selection. However, if there is a large class of mutations with context-specific effects, mutations that are neutral in the present environment but deleterious in another environment will accumulate at the neutral rate, causing a buildup of “hidden” genetic load. Then, when the environment changes the previously hidden load may be expressed with potentially disastrous evolutionary consequences, in the form of a mutational meltdown leading to extinction (Lynch et al. 1995).

As part of an ongoing effort to characterize and understand taxonomic variation in the mutational process, we initiated a mutation accumulation (MA) experiment using several species of Rhabditid nematodes, in which spontaneous mutations were allowed to accumulate under relaxed selection. Mutations were accumulated at 20°, using standard methods of Caenorhabditis elegans husbandry (Wood 1988) for >200 generations. Initial results after 200 generations of MA showed that two strains of C. briggsae declined in fitness significantly faster (~0.2%/generation) than did two strains of C. elegans (~0.1%/generation) under the same conditions in which the mutations were allowed to accumulate (Baer et al. 2005). To begin to characterize the relative importance of starting genotype and environmental context in determining mutational properties, we reassayed fitness of the same set of MA lines in a different experimental environment, in this case high temperature (25°). High temperatures (>25°) are “stressful” for C. elegans in the sense that survivorship and fecundity are substantially reduced relative to cooler temperatures, whereas fitness in C. briggsae is not reduced until temperatures reach ~28°–29° (M. Ailion, unpublished data; also see below). Thus, our experimental protocol allows us to separate effects of different environments (“context”) from the effects of environmental stress sensu strictu.


Nematode strains:

Justification for choice of strains is given in Baer et al. (2005); we provide a brief summary here. Two strains of C. briggsae (HK104, PB800) and C. elegans (N2, PB306) were used in this study. Both species are androdioecious hermaphrodites; hermaphrodites can outcross only to males, not to other hermaphrodites (Wood 1988). Generation time of both species at 20° is ~3.5 days, and fecundity is similar in both species. Collection information on all strains is available from the Caenorhabditis Genetics Center (http://www.cbs.umn.edu/CGC/CGChomepage.htm).

Mutation accumulation:

MA protocols have been outlined in detail elsewhere (Vassilieva and Lynch 1999; Baer et al. 2005). The principle is simple: many replicate lines of a highly inbred stock population are allowed to evolve in the relative absence of natural selection, thereby allowing deleterious mutations to accumulate. Descendant populations are then compared to the unmutated ancestral control stock. If the average effect of new mutations is nonzero, the mean phenotype will change over time. Since different lines accumulate different numbers of mutations, the variance among lines will increase over time, even if the average mutational effect is zero. In sexual diploids, the change in the mean phenotype is the product of the gametic mutation rate, U/2, where U is the diploid genomic mutation rate, and the average homozygous effect of a mutation, 2equation M1 (Lynch and Walsh 1998, p. 341). With certain assumptions, the per-generation change in mean phenotype (Rm) and the per-generation change in the among-line component of variance (Vb) can be employed to calculate U and equation M2.

We began by inbreeding each strain for six generations by self-fertilization of a single hermaphrodite. Populations were allowed to expand to large size and worms were frozen using standard methods (Wood 1988). Frozen worms were thawed and allowed to reexpand to large population size, at which time 100 replicate lines from each strain were started from single juvenile hermaphrodites. All lines were kept at 20° and propagated by transferring a single L4 worm at 4-day intervals. At every generation, the prior two generations of each line were kept as backups. If a worm failed to reproduce it was replaced with a single worm from the preceding generation. If a worm was slow to reproduce it was held over for the 4-day interval without going to backup. Lines that were held over thus had fewer generations of reproduction than those that were not held over. We refer to Gmax as (total time in days)/4, the maximum number of generations a line could have been through, and Gmin as the total number of single-worm transfers a line experienced. Because backups were (usually) taken from the same generation as the dead or missing worm, going to backup did not subtract from the true number of generations.

The probability of fixation of a new mutation is a function of its selection coefficient s and the effective population size Ne; mutations with a selection coefficient s < 1/4Ne are expected to accumulate at the neutral rate (Kimura 1962). With single hermaphrodites, Ne = 1, so mutations that reduce fitness by < ~25% will be effectively neutral.

Fitness assays:

Fitness was assayed three times, at Gmax = 100 and 200 at 20° and at Gmax = 220 at 25°. Details of the 20° fitness assays are presented in Baer et al. (2005). Briefly, all surviving MA lines were frozen at Gmax = 100(200, 220). Each assay was done in two blocks; each block contained (approximately) 34 MA lines and 20 control lines from each strain, except for block 2 of the G100 assay, which included 15 control and 32 MA lines. The two blocks in each assay contained different, nonoverlapping sets of MA lines. MA lines were thawed along with ancestral controls; 20 worms were picked from the thawed sample of each control and used to establish replicate control lines. From each of the 34 MA and 20 control lines, five replicates were started from a single haphazardly chosen L4 hermaphrodite and transferred by single-individual descent for three generations (P1–P3). Plates were assigned a random number and after the first generation were identified only by the random number and handled in random numerical order. If a worm failed to reproduce at the P1 or P2 generation, we started the plate again at the previous generation. A single newly hatched (L1) offspring (labeled R1) was collected from each P3 parent and put on a fresh plate. On the third day of an R1 individual's life it was removed and placed on a fresh seeded plate and transferred to a new plate daily for the next 3 days. The plate from which the parental (R1) worm was removed was incubated overnight at 20° to allow eggs to hatch and then stored at 4°. In most cases, reproduction after the third day of reproduction was negligible. Upon completion of the assay, plates were stained with 0.075% toluidine blue and worms were counted under a dissecting microscope at 20× magnification.

The 25° assay was essentially the same as the 20° assays, with two exceptions: (1) 40 MA lines were included in each assay block, and (2) the timing of transfers differed to accommodate the faster development at the higher temperature. Single adult (not L4) hermaphrodites were transferred at 3-day intervals (P1–P3). The focal R1 worm was established from the P3 parent and transferred on the second day of its life and daily for the next 2 days (R2 and R3). The plate from which the R1 parent was removed was incubated overnight at 25° and then stored at 4°. Reproduction after R3 was negligible.

Data analysis:

Replicates that made it to the R1 generation were scored as “present” in the assay. Individuals present in the R1 that produced offspring “survived.” Total fitness (equation M3) was calculated as the lifetime reproduction of all worms present at R1. Productivity was calculated as the lifetime reproduction of all worms that survived. Relative fitness (equation M4, Charlesworth 1994) was calculated from line means following Keightley et al. (2000); results for equation M5 were very similar to those for equation M6 (supplemental Table 6 at http://www.genetics.org/supplemental/).

The study is predicated on the assumption that the two species find the two thermal environments stressful to differing degrees. We compared control means of each strain separately in the two temperatures by restricted maximum likelihood using PROC MIXED in SAS v. 9.1 with the restricted maximum-likelihood (REML) option. The full model is: equation M7 = temperature + block(temperature) + line[block(temperature)]. Temperature is a fixed effect, and block and line (block) are random effects. Among-line and error components of variance were allowed to vary among blocks. Results are presented in supplemental Table 5 at http://www.genetics.org/supplemental/.

Differences among groups in the change in mean phenotype:

Our primary interest is in the differences among groups in the rate of change of mean phenotype due to the accumulation of new mutations, not differences in means per se. There are two usual ways to express the rate of change in mean phenotype: the per-generation change in mean phenotype Rm = (equation M8, where zt represents the trait mean in generation t, and ΔM (= Rm/z0), the percentage of change per generation scaled to the control mean (e.g., Vassilieva et al. 2000). Although it is common to use Rm because hypothesis tests are straightforward (i.e., linear regression), comparisons of Rm among groups are biologically meaningful only if the starting (control) mean is the same in the groups being compared. To see why this is so, recall that the change in mean phenotype over one generation as a result of mutation equation M9, where U is the genomic mutation rate and equation M10 is the average effect of a new mutation (Lynch and Walsh 1998, p. 341). Comparisons of equation M11 are meaningful only when scaled as a fraction of the mean. To illustrate, consider the hypothetical effects of mutations on body size in two taxa with different starting means, mice and elephants. Suppose that the cumulative ravages of deleterious mutation cause a decrement of 1 g of mass per generation in each species. Rm is the same in both species, but it would be unwise to conclude that the cumulative effects of deleterious mutation are the same in each species. An obvious possible remedy for the scaling effect is to log transform the data, so that linear regression represents proportional change. Unfortunately, the usual methods of analyses (ANOVA, REML) are sensitive to departures from normality and log transformation renders our data nonnormal. To circumvent these complications, we apply a bootstrap protocol to construct empirical 95% confidence limits on ΔM. Lines (control and MA) are resampled with replacement from each assay block and a pseudovalue of the block mean is calculated. Pseudovalues of Rm and ΔM are then calculated for each bootstrap replicate (n = 1000). The upper and lower 2.5% of the pseudodistribution establish the upper and lower 95% confidence limits (CLs) for each assay block (Efron and Tibshirani 1993).

Comparisons between strains and between temperatures are done by extension of the bootstrap approach. A bootstrap pseudoreplicate of the full data set is generated as above, and the statistics of interest are calculated for each block and averaged over the set of relevant blocks (e.g., over all 20° assay blocks); this protocol accounts for variation among lines within each block and differences between blocks. This procedure is repeated 1000 times, generating a distribution of among-block averages, and 95% CLs are determined as before.

Differences among groups in the mutational variance:

The per-generation input of genetic variation from new mutation, VM, is one-half the among-line component of variance divided by the number of generations, Vb = Vamong line/t, where t is number of generations of mutation accumulation (Lynch and Walsh 1998, p. 330). This calculation is predicated on the assumption that the among-line component of variance in the ancestral control is zero. Comparisons of mutational variances among groups are complicated by scaling effects, because the variance typically scales with the mean. Mutational variability is often reported either as mutational heritability equation M12 = VM/VE, where VE is the environmental (error) variance, or as a mutational coefficient of variation equation M13 (Houle et al. 1996). Both of these measures of mutational variability have potentially serious limitations when used in a comparative MA context. Mutational heritability depends on VE, so differences among groups in environmental variance, for whatever reason, can potentially provide a misleading picture of the variation actually due to new mutations (Houle 1992; Houle et al. 1996). Mutational coefficients of variation do not depend on VE and do account for scaling effects. However, the CVM does not account for differences among groups in the change in the mean over time, so comparisons of CVM may become misleading over time if the change in the mean differs among groups. A potential solution is to scale the mutational variance to the MA mean rather than to the control mean [“opportunity for selection” (Crow 1958)], but calculation of standard errors using samples from different generations is not straightforward because the MA mean is expected to change over time. We restrict hypothesis tests between groups to assays from the same generation, although we treat generation 200 (20°) and generation 220 (25°) as effectively the same.

Comparisons of mutational variability between groups were assessed by the same bootstrap protocol as described above for changes in the mean; data were resampled as above and variance components were calculated for each pseudosample using PROC VARCOMP in SAS v. 9.0 with the REML option.

Mutational covariance between environments:

The genetic component of covariance between the two thermal environments at the G200/G220 assays was determined using the “normal effects” model of Shaw et al. (2000); see Baer et al. (2005) for a detailed description. Confidence limits were generated using a parametric bootstrap. After the maximum-likelihood estimate (MLE) was estimated as described in Baer et al. (2005), 1000 parameter sets were simulated from a multivariate normal distribution using the MLE as mean and the information matrix as variance. A genetic correlation was calculated for each of the parameter sets and likelihood 1.96 less than the MLE was taken as the confidence limits. Values >1 were truncated to 1. In this analysis, the genetic correlation due to environmental differences is confounded with the genetic correlation due to the mutations common to the G200 and the G220 lines. This latter correlation is expected to be 0.9 because the lines share 90% of their evolutionary history.

Rate and average effect of new mutations:

The diploid genomic mutation rate, UMIN, and the average effect of a new mutation, equation M14MAX, were calculated for each group using the “Bateman–Mukai” method (Lynch and Walsh 1998, pp. 341–343). A lower-bound estimate of the genomic mutation rate is UMIN ≈ 2(Rm)2/Vb, and an upper-bound estimate of the average effect of a new mutation is equation M15MAXVb/2Rm, which we divide by the control mean phenotype z0 to express the average effect as a proportional reduction in mean phenotype. Homozygotes are expected to have mean fitness reduced by 2equation M16 and heterozygotes by 2equation M17h, where h is the average dominance and h = 0.5 denotes additivity.


Treatment (20° and 25°) means and averages over all blocks and genetic correlations across environments for total fitness (equation M18) are presented in Table 1. Results for productivity are given in Table 2. Summary statistics for each assay block are presented in supplemental Tables 3 (equation M19) and 4 (productivity) at http://www.genetics.org/supplemental/. Data from blocks 1–4 (20°) were originally reported in Baer et al. (2005); blocks 5 and 6 (25°) are new data and represent an additional 195,872 worms counted.

Summary statistics for total fitness (equation M30) averaged over block means
Summary statistics for productivity averaged over block means

Change in the mean:

This study was designed to characterize the relative importance of genotype (both within and between species) and environmental context in determining the cumulative effects of spontaneous mutations. In particular, we were motivated by the finding of Shabalina et al. (1997) that the percentage of per-generation mutational decay (ΔM) in the number of surviving offspring per female Drosophila was tenfold higher under competitive, highly stressful conditions than under noncompetitive, presumably benign conditions. If the conditions under which such experiments are usually performed are more benign that the natural environment of the organism and if the effects of deleterious mutations are magnified in stressful environments, the genomic mutation rate for fitness may be much higher than the available data would lead us to believe. Our results do not bear out that supposition. In no case did ΔM differ significantly within a strain between assay temperatures (Tables 1 and and2).2). There is a nonsignificant trend for the PB800 strain of C. briggsae and the PB306 strain of C. elegans to decline more slowly at 25° and for the N2 strain of C. elegans to decline more rapidly at 25°. Presumably, a larger experiment would allow us to detect a relatively subtle effect. However, the magnitude of the observed difference—no more than 1.5-fold in each case—is of the same degree as the variation among blocks in the 20° treatment (supplemental Tables 3 and 4 at http://www.genetics.org/supplemental/). Perhaps more importantly, there is no consistent trend for the decline in fitness to be greater in stressful environments. Of the C. elegans, N2 declined slightly faster at 25° (stressful) than at 20° (benign), but PB306 declined somewhat faster at 20° (benign) than at 25° (stressful). Of the C. briggsae, PB800 declined somewhat faster at 20° than at 25°, but HK104 declined at essentially the same rate at each temperature. We conclude that there is at best weak evidence for context-dependent mutational effects, but there is no evidence for a consistent influence of environmental stress on the cumulative effects of new mutations.

An obvious possibility is that the temperature regime we proposed as stressful for C. elegans was not actually stressful or stressful enough to make a difference. Temperature did affect fitness differently in the two species (supplemental Table 5 at http://www.genetics.org/supplemental/). Control fitness of the two strains of C. briggsae was similar between the two temperature treatments whereas equation M20 of the N2 strain of C. elegans was reduced by ~50% (P < 0.05) and that of the PB306 strain by twofold (P < 0.0001) in the 25° treatment relative to the 20° treatment (supplemental Tables 3 and 4 at http://www.genetics.org/supplemental/). The high-temperature environment clearly imposes a fitness cost on C. elegans that it does not impose on C. briggsae, consistent with previous reports (M. Ailion, unpublished data), although the temperature sensitivity itself appears to differ between strains of C. elegans. Vassilieva et al. (2000) found that intrinsic rate of increase (r) of MA lines of the N2 strain declined somewhat faster at 15° than at 20°, but that survivorship and productivity did not.

Attempts to relate mutational effects to environmental stress have produced inconsistent results (Kondrashov and Houle 1994; Fernández and López-Fanjul 1997; Shabalina et al. 1997; Szafraniec et al. 2001; Fry and Heinsohn 2002; Chang and Shaw 2003; Estes et al. 2005; Kavanaugh and Shaw 2005). The importance of environmental effects is most closely associated with competitive fitness in Drosophila melanogaster (e.g., Shabalina et al. 1997, but see Keightley et al. 1998), although at least one study found no effect of competitive vs. noncompetitive conditions on the change in mean fitness (Fry and Heinsohn 2002). It would certainly not be surprising if relatively small phenotypic effects translate into profound effects on competitive fitness—a small difference in footspeed between lions may result in a qualitative difference in fitness, depending on which lion gets the gazelle. However, to our knowledge this study represents the first attempt to examine multiple starting genotypes and species in more than one environment (Kondrashov and Houle 1994 examined two MA lines from the same starting D. melanogaster genotype in multiple environments).

Change in the variance:

Most importantly, the results depend critically on how the variance is scaled (Tables 1 and and2).2). For example, Vb, the per-generation change in the among-line component of variance, is much lower in HK104 than in the other three strains. Further, Vb for equation M21 is greatly reduced at 25° relative to 20° in C. elegans but not in C. briggsae. The mutational heritability, equation M22, is much lower at 25° than at 20°. However, when the variance is scaled by either the control mean (CVM,Control) or MA mean (CVM,MA) the discrepancy between the variance at 20° and 25° is substantially reduced, more so for total fitness (Table 1) than for productivity (Table 2). When scaled by the control mean and pooled over all six blocks, the four strains behave quite similarly, with N2 and HK104 having somewhat lower mutational variance than PB306 or PB800. However, when the variance is scaled by the MA mean and pooled over generations 200 and 220, ignoring the generation 100 data, the variances of the two strains in each species are still very similar, but C. briggsae has modestly greater variance for equation M23 (~1.6-fold) than C. elegans.

The fact that the mutational heritabilities differ substantially between 20° and 25° but the CVM's do not strongly suggests that the environmental component of variance is larger at 25°. The potentially misleading properties of heritability have been clearly articulated by Houle (1992; Houle et al. 1996) and our results add credence to the idea that genetic signal may be swamped by environmental noise. Nevertheless, the considerable difference in environmental variance between the two temperatures is interesting in its own right and demands explanation. The magnitude of the difference in equation M24 between temperatures is greater for C. elegans than for C. briggsae, which suggests that random environmental effects may be magnified in a stressful environment. We believe the most likely reason for the difference between temperatures is that the faster generation time at 25° magnifies small demographic differences between replicates.

Given the variation in the estimates of the variance, what is to be learned? For within-generation comparisons, we believe the MA-mean-scaled variance, CVM,MA, is probably the most meaningful of the various estimators; Crow (1958) derived the relationship between the scaled variance and the opportunity for response to natural selection. Scaled in that way, the rate of accumulation of mutational variation in fitness of C. briggsae is certainly no less than that of C. elegans and may be somewhat greater, consistent with the respective changes in the mean.

One additional point must be noted: the previous arguments depend on the assumption that the among-line variance in the controls is zero. For the N2, PB306, and PB800 strains, among-line variance in the controls did not differ significantly from zero in any block (P > 0.1 in all blocks). In HK104, however, the among-line variance of the controls approached significance (0.05 < P < 0.1) in three of the six blocks (supplemental Table 5 at http://www.genetics.org/supplemental/). Genotyping of >40 microsatellite loci revealed no ancestral heterozygotes in the HK104 controls (N. Phillips, C. Baer and A. Custer, unpublished data), so the among-line component is unlikely to be a result of (much) residual genetic variation. Nevertheless, the estimates of VM in HK104 may be overestimates; possible explanations are discussed below.

Rate and average effect of mutation:

Estimates of UMIN and equation M25MAX are inherently noisier than those of the change in the mean and variance because the variable is a function of quantities (mean and variance) that are themselves estimated with error. Moreover, we cannot calculate the within-block bootstrap distribution of the Bateman–Mukai estimators because the REML estimate of the among-line variance is zero in some pseudoreplicates, in which case UMIN is inestimable. We simply report point estimates of UMIN and equation M26MAX derived from block means of Rm and Vb and among-block standard errors and do not attempt formal hypothesis tests. There are two sources of downward bias in estimates of UMIN: a statistical bias due to the Bateman–Mukai assumption of equal effects (Lynch and Walsh 1998, pp. 341–343) and, probably much more importantly, an experimental inability to detect the effects of the presumably large class of mutations with very slightly deleterious effects (Davies et al. 1999; Denver et al. 2004). The inability to detect small effects is inherent to any MA experiment and can be overcome only with extremely large sample sizes, not by more sophisticated analytical methods. However, our primary interest is in comparative biology and there is no reason to expect that the bias should differ systematically between taxa or between experimental treatments. Interestingly, our estimates of UMIN (~0.01–0.02/generation) are qualitatively very similar to the values of U that Cutter and Payseur (2003) arrived at from synonymous substitution rates, although a more direct method of estimating U by sequencing random nuclear loci in MA lines gave an estimate of U ≈ 0.48 in the N2 strain of C. elegans (Denver et al. 2004). The genomic mutation rate of the HK104 strain is substantially higher than that of the other three strains and that result does not depend on assay temperature. Averaged over all blocks, the rank order among strains of UMIN is the same as that of ΔM. Given the inherent difficulty of estimating U, that result suggests that ΔM may be a useful surrogate for the genomic mutation rate, in the sense that a difference between groups in ΔM may reflect a proportional difference in U, a point that was made implicitly by Shabalina et al. (1997).

Mutational correlation between environments:

If all mutations have identical effects at 20° (assayed at Gmax = 200) and 25° (assayed at Gmax = 220), the expected genetic correlation between treatments (rM20,25) is 0.9. In three of the four strains (N2, PB306, and PB800), the point estimate of rM for both equation M27 and productivity is very close to 0.9 (Tables 1 and and2),2), reinforcing the conclusion that cumulative effects of new mutations are by and large not context dependent, at least in the context of assay temperature. This result is not trivial; temperature-sensitive mutations are common in C. elegans (Riddle et al. 1997). In HK104, rM for equation M28 is large and positive (0.64) but smaller than in the other strains; however, rM for productivity is near zero. Since equation M29 is simply productivity weighted by survivorship, we conclude that mutations that affect survivorship in HK104 are not context dependent but that mutations that affect fecundity may be. There are two reasons to think that the lack of genetic correlation in HK104 is not a result of qualitatively different pleiotropy. First, the within-line sample size in HK104 was consistently about half that of the other strains, due to lower survivorship, so the experimental error variance is greater, consistent with the low values of mutational heritability (note the generally wider confidence limits in HK104 than in the other strains). Second, the among-line variance of HK104 controls was consistently greater than that in the other strains, which suggests greater sensitivity of HK104 to cross-generation demographic effects. Nevertheless, the behavior of the HK104 strain is sufficiently different from that of the other strains that we suspect that a different mutational process (transposable elements?) contributes to the decay in fitness in that strain.

Positive genetic correlations in MA experiments are not solely a result of pleiotropy. Keightley et al. (2000) demonstrated that the MA process inherently leads to different numbers of mutant alleles in different MA lines, which can lead to genetic correlations even in the absence of correlated effects of new mutations, i.e., pleiotropy in the usual sense. Several previous studies have reported mutational covariances across environments and mutational correlations are often (Kondrashov and Houle 1994; Fry and Heinsohn 2002) but by no means always (Fernández and López-Fanjul 1997; Estes et al. 2005) large and positive. Tellingly, pairs of traits that have no obvious functional relationship tend to have lower mutational correlations than those that do (see Estes et al. 2005). “Spurious” mutational correlations arising due to different numbers of mutations in different MA lines appear to be a modest problem in most cases.

Summary and conclusions:

The mutational properties of four strains of self-fertile Caenorhabditis do not differ substantially depending on the assay temperature, and any subtle differences that there may be cannot consistently be attributed to the effects of environmental stress per se. At least over the range of environmental conditions we have examined, mutational properties seem to be relatively consistent properties of individual genotypes: the HK104 strain consistently declines more rapidly than the other genotypes, and PB306 consistently declines more slowly than the two C. briggsae. To the extent that one can generalize from two genotypes per species, the conclusion that the cumulative effects of new mutations are greater in C. briggsae than in C. elegans remains. The available phylogenetic evidence suggests that C. briggsae evolved self-compatibility from a gonochoristic (dioecious) ancestor more recently than C. elegans (Kiontke et al. 2004; M.-A. Felix, unpublished data). The strength of selection to reduce mutation rate is much stronger in selfers than in outcrossers (Kondrashov 1995b; Drake et al. 1998), which leads to the intriguing (but clearly speculative) notion that the apparent lower mutation rate in C. elegans may be an adaptation to self-fertilization. Adaptation to self-fertilization would also explain the well-documented much lower phenotypic estimates of U in selfing Rhabditids (Keightley and Caballero 1997; Vassilieva and Lynch 1999; Vassilieva et al. 2000; Ajie et al. 2005; Baer et al. 2005) than in D. melanogaster (Drake et al. 1998; Keightley and Eyre-Walker 1999; Lynch et al. 1999). Interestingly, the standing molecular and quantitative variation in C. briggsae is greater than that in C. elegans (Graustein et al. 2002; Jovelin et al. 2003; C. F. Baer and M. Salomon, unpublished data), consistent with a higher mutation rate in C. briggsae (Kimura 1983; Lynch and Hill 1986). Future research should focus on elucidating the extent of genetic variation in the mutational process, as well as the factors underlying the variation.


This project was begun while C.F.B. was a postdoc in the lab of M. Lynch at Indiana University. Thanks are due to the Indiana University worm crew and especially C. Steding for 200 generations of MA work. D. Appel, W. Bour, R. Klinger, and J. Rosenbloom counted worms. D. Denver, S. Estes, J. Fry, L. Higgins, P. Keightley, R. Shaw, M. Wayne, and the anonymous reviewers provided helpful advice and/or comments. Frank Shaw made substantial contributions to the calculation of mutational correlations. C.F.B. offers special thanks to Mike Lynch for his guidance and inspiration. Stocks were provided by the Caenorhabditis Genetics Center at the University of Minnesota. This work was supported by National Institutes of Health (NIH)/National Research Service Award postdoctoral fellowship 1 F32 GM20887-01 and start-up funds from the University of Florida to C.F.B. and by NIH grant RO1-GM36827 to M. Lynch.


  • Ajie, B. C., S. Estes, M. Lynch and P. C. Phillips, 2005. Behavioral degradation under mutation accumulation in Caenorhabditis elegans. Genetics 170: 655–660. [PMC free article] [PubMed]
  • Baer, C. F., F. Shaw, C. Steding, M. Baurngartner, A. Hawkins et al., 2005. Comparative evolutionary genetics of spontaneous mutations affecting fitness in rhabditid nematodes. Proc. Natl. Acad. Sci. USA 102: 5785–5790. [PMC free article] [PubMed]
  • Bjedov, I., O. Tenaillon, B. Gerard, V. Souza, E. Denamur et al., 2003. Stress-induced mutagenesis in bacteria. Science 300: 1404–1409. [PubMed]
  • Chang, S. M., and R. G. Shaw, 2003. The contribution of spontaneous mutation to variation in environmental response of Arabidopsis thaliana: responses to nutrients. Evolution 57: 984–994. [PubMed]
  • Charlesworth, B., 1994. Evolution in Age-Structured Populations, Ed. 2. Cambridge University Press, Cambridge, UK.
  • Crow, J. F., 1958. Some possibilities for measuring selection intensities in man. Hum. Biol. 30: 1–13. [PubMed]
  • Cutter, A. D., and B. A. Payseur, 2003. Rates of deleterious mutation and the evolution of sex in Caenorhabditis. J. Evol. Biol. 16: 812–822. [PubMed]
  • Davies, E. K., A. D. Peters and P. D. Keightley, 1999. High frequency of cryptic deleterious mutations in Caenorhabditis elegans. Science 285: 1748–1751. [PubMed]
  • Dawson, K. J., 1998. Evolutionarily stable mutation rates. J. Theor. Biol. 194: 143–157. [PubMed]
  • Denver, D. R., K. Morris, M. Lynch and W. K. Thomas, 2004. High mutation rate and predominance of insertions in the Caenorhabditis elegans nuclear genome. Nature 430: 679–682. [PubMed]
  • Drake, J. W., B. Charlesworth, D. Charlesworth and J. F. Crow, 1998. Rates of spontaneous mutation. Genetics 148: 1667–1686. [PMC free article] [PubMed]
  • Efron, B., and R. J. Tibshirani, 1993. An Introduction to the Bootstrap. Chapman & Hall, New York.
  • Estes, S., B. C. Ajie, M. Lynch and P. C. Phillips, 2005. Spontaneous mutational correlations for life-history, morphological and behavioral characters in Caenorhabditis elegans. Genetics 170: 645–653. [PMC free article] [PubMed]
  • Fernández, J., and C. López-Fanjul, 1997. Spontaneous mutational genotype-environment interaction for fitness-related traits in Drosophila melanogaster. Evolution 51: 856–864.
  • Fisher, R. A., 1930. The Genetical Theory of Natural Selection. Clarendon Press, Oxford.
  • Fry, J. D., and S. L. Heinsohn, 2002. Environment dependence of mutational parameters for viability in Drosophila melanogaster. Genetics 161: 1155–1167. [PMC free article] [PubMed]
  • Graustein, A., J. M. Gaspar, J. R. Walters and M. F. Palopoli, 2002. Levels of DNA polymorphism vary with mating system in the nematode genus Caenorhabditis. Genetics 161: 99–107. [PMC free article] [PubMed]
  • Haldane, J. B. S., 1927. A mathematical theory of natural and artificial selection. Part V. Selection and mutation. Proc. Camb. Philos. Soc. 23: 838–844.
  • Houle, D., 1992. Comparing evolvability and variability of quantitative traits. Genetics 130: 195–204. [PMC free article] [PubMed]
  • Houle, D., and A. S. Kondrashov, 2006. Mutation, pp. 32–48 in Evolutionary Genetics, edited by C. Fox and J. Wolf. Oxford University Press, Oxford.
  • Houle, D., B. Morikawa and M. Lynch, 1996. Comparing mutational variabilities. Genetics 143: 1467–1483. [PMC free article] [PubMed]
  • Jovelin, R., B. C. Ajie and P. C. Phillips, 2003. Molecular evolution and quantitative variation for chemosensory behaviour in the nematode genus Caenorhabditis. Mol. Ecol. 12: 1325–1337. [PubMed]
  • Kavanaugh, C. M., and R. G. Shaw, 2005. The contribution of spontaneous mutation to variation in environmental responses of Arabidopsis thaliana: responses to light. Evolution 59: 266–275. [PubMed]
  • Keightley, P. D., and A. Caballero, 1997. Genomic mutation rates for lifetime reproductive output in Caenorhabditis elegans. Proc. Natl. Acad. Sci. USA 94: 3823–3827. [PMC free article] [PubMed]
  • Keightley, P. D., and A. Eyre-Walker, 1999. Terumi Mukai and the riddle of deleterious mutation rates. Genetics 153: 515–523. [PMC free article] [PubMed]
  • Keightley, P. D., A. Caballero and A. García-Dorado, 1998. Surviving under mutation pressure. Curr. Biol. 8: R235–R237. [PubMed]
  • Keightley, P. D., E. K. Davies, A. D. Peters and R. G. Shaw, 2000. Properties of ethylmethane sulfonate-induced mutations affecting life-history traits in Caenorhabditis elegans and inferences about bivariate distributions of mutation effects. Genetics 156: 143–154. [PMC free article] [PubMed]
  • Kimura, M., 1960. Optimum mutation rate and degree of dominance as determined by the principle of minimum genetic load. J. Genet. 57: 21–34.
  • Kimura, M., 1962. On probability of fixation of mutant genes in a population. Genetics 47: 713–719. [PMC free article] [PubMed]
  • Kimura, M., 1967. On the evolutionary adjustment of spontaneous mutation rates. Genet. Res. 9: 23–34.
  • Kimura, M., 1983. The Neutral Theory of Molecular Evolution. Cambridge University Press, Cambridge, UK.
  • Kiontke, K., N. P. Gavin, Y. Raynes, C. Roehrig, F. Piano et al., 2004. Caenorhabditis phylogeny predicts convergence of hermaphroditism and extensive intron loss. Proc. Natl. Acad. Sci. USA 101: 9003–9008. [PMC free article] [PubMed]
  • Kondrashov, A. S., 1988. Deleterious mutations and the evolution of sexual reproduction. Nature 336: 435–440. [PubMed]
  • Kondrashov, A. S., 1995. a Dynamics of unconditionally deleterious mutations: Gaussian approximation and soft selection. Genet. Res. 65: 113–121. [PubMed]
  • Kondrashov, A. S., 1995. b Modifiers of mutation-selection balance: general approach and the evolution of mutation rates. Genet. Res. 66: 53–69.
  • Kondrashov, A. S., and D. Houle, 1994. Genotype-environment interactions and the estimation of the genomic mutation rate in Drosophila melanogaster. Proc. R. Soc. Lond. Ser. B 258: 221–227. [PubMed]
  • Leigh, E. G., 1970. Natural selection and mutability. Am. Nat. 104: 301–305.
  • Leigh, E. G., 1973. The evolution of mutation rates. Genetics 73(Suppl.): 1–18. [PubMed]
  • Lynch, M., and W. G. Hill, 1986. Phenotypic evolution by neutral mutation. Evolution 40: 915–935.
  • Lynch, M., and J. B. Walsh, 1998. Genetics and Analysis of Quantitative Traits. Sinauer Press, Sunderland, MA.
  • Lynch, M., J. Conery and R. Bürger, 1995. Mutation accumulation and the extinction of small populations. Am. Nat. 146: 489–518.
  • Lynch, M., J. Blanchard, D. Houle, T. Kibota, S. Schultz et al., 1999. Perspective: spontaneous deleterious mutation. Evolution 53: 645–663.
  • Matic, I., M. Radman, F. Taddei, B. Picard, C. Doit et al., 1997. Highly variable mutation rates in commensal and pathogenic Escherichia coli. Science 277: 1833–1834. [PubMed]
  • Morgan, T. H., 1903. Evolution and Adaptation. MacMillan, New York.
  • Nöthel, H., 1987. Adaptation of Drosophila melanogaster populations to high mutation pressure: evolutionary adjustment of mutation rates. Proc. Natl. Acad. Sci. USA 84: 1045–1048. [PMC free article] [PubMed]
  • Riddle, D. L., T. Blumenthal, B. J. Meyer and J. R. Priess (Editors), 1997. C. elegans II. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
  • Shabalina, S. A., L. Y. Yampolsky and A. S. Kondrashov, 1997. Rapid decline of fitness in panmictic populations of Drosophila melanogaster maintained under relaxed natural selection. Proc. Natl. Acad. Sci. USA 94: 13034–13039. [PMC free article] [PubMed]
  • Shaw, R. G., D. L. Byers and E. Darmo, 2000. Spontaneous mutational effects on reproductive traits of Arabidopsis thaliana. Genetics 155: 369–378. [PMC free article] [PubMed]
  • Simmons, M. J., and J. F. Crow, 1977. Mutations affecting fitness in Drosophila populations. Annu. Rev. Genet. 11: 49–78. [PubMed]
  • Sniegowski, P. D., P. J. Gerrish and R. E. Lenski, 1997. Evolution of high mutation rates in experimental populations of E. coli. Nature 387: 703–705. [PubMed]
  • Sniegowski, P. D., P. J. Gerrish, T. Johnson and A. Shaver, 2000. The evolution of mutation rates: separating causes from consequences. BioEssays 22: 1057–1066. [PubMed]
  • Sturtevant, A. H., 1937. Essays on evolution. I. On the effect of selection on mutation rate. Q. Rev. Biol. 12: 464–467.
  • Szafraniec, K., R. H. Borts and R. Korona, 2001. Environmental stress and mutational load in diploid strains of the yeast Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 98: 1107–1112. [PMC free article] [PubMed]
  • Vassilieva, L., and M. Lynch, 1999. The fitness effects of spontaneous mutation in Caenorhabditis elegans. Genetics 151: 119–129. [PMC free article] [PubMed]
  • Vassilieva, L. L., A. M. Hook and M. Lynch, 2000. The fitness effects of spontaneous mutations in Caenorhabditis elegans. Evolution 54: 1234–1246. [PubMed]
  • Wood, W. B. (Editor), 1988. The Nematode Caenorhabditis elegans. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

Articles from Genetics are provided here courtesy of Genetics Society of America
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...