Skip to main content

Are mutations usually deleterious? A perspective on the fitness effects of mutation accumulation


All adaptive alleles in existence today began as mutations, but a common view in ecology, evolution, and genetics is that non-neutral mutations are much more likely to be deleterious than beneficial and will be removed by purifying selection. By dramatically limiting the effectiveness of selection in experimental mutation accumulation lines, multiple studies have shown that new mutations cause a detectable reduction in mean fitness. However, a number of exceptions to this pattern have now been observed in multiple species, including in highly replicated, intensive analyses. We briefly review these cases and discuss possible explanations for the inconsistent fitness outcomes of mutation accumulation experiments. We propose that variation in the outcomes of these studies is of interest and understanding the underlying causes of these diverse results will help shed light on fundamental questions about the evolutionary role of mutations.


Mutations can be good or bad. While many genetic changes will disrupt organism function in ways that are detrimental, mutation is ultimately the source of all adaptive variation as well. To understand the impact of the mutation process on individuals and populations, and to predict how the mutation rate itself will evolve, we need to understand how often beneficial versus deleterious mutations occur. A simplified prediction is that beneficial mutations should eventually be depleted in populations adapting to a stable environment, as they become fixed by positive selection. In a perfectly adapted genotype, we would expect mutations to be either deleterious or neutral. In practice, interactions between genes, environment, and the ever-changing landscape of adaptation makes predicting the distribution of fitness effects (DFE) of mutations in natural populations far less straightforward. Nevertheless, there has been intense interest in understanding the DFE, due to its relevance for both predicting trajectories of adaptation and managing the conservation of threatened species (Lande 1994; Eyre-Walker and Keightley 2007; Halligan and Keightley 2009).

One way to study the effects of spontaneous mutations is with a mutation accumulation (MA) approach. In a typical MA experiment, a given genotype is replicated into many “lines”, which are each repeatedly bottlenecked for many generations. Bottlenecking ensures that the effective population size (N e) in each MA line is very low. As a result, most new mutations that arise have a probability of fixation approximately equal to that of a neutral mutation (Keightley and Caballero 1997; Lynch and Walsh 1998). In other words, the fate of mutant alleles in populations with small N e is due primarily to genetic drift, rather than natural selection, unless their effects on fitness are large.

Some simple equations are useful for understanding the results of MA (reviewed in Halligan and Keightley 2009). Following MA, the expected number of mutations in a line is Ut, where U is the mutation rate per genome and t is the number of generations of MA. If we measure some trait (typically a life history trait), the expected trait value for MA lines is z t = z 0 + UtE[s], where z 0 is the pre-MA (control) trait value, and E[s] is the average effect of a mutation on the trait (or E[hs] in heterozygous diploids, where h is the dominance coefficient). The expected genetic variance among MA lines is UtE[s 2]. We can describe the rate of change in the mean per generation as ΔM = (z tz 0)/t = UE[s]. Similarly, the rate of change in genetic variance is ΔV = UE[s 2], as there is presumed to be no genetic variance prior to MA. Modelling approaches have been applied that use MA data to estimate specific values for the mutation rate and complex distributions of mutational effects (e.g., Keightley 1994; García-Dorado 1997; Shaw et al. 2002; Böndel et al. 2019; Böndel et al. 2021), but because U cannot be negative, the sign of ΔM must always reflect the sign of E[s]. While approaches to modelling the DFE vary, we consider the sign of ΔM a “first order” metric of mutational effects that can always be compared among MA studies.

In many MA experiments, mean fitness declines (negative ΔM using the formulation above), meaning E[s] < 0. Thus, a decline in fitness components under MA is evidence that mutations have deleterious effects on average. Many, but not all MA experiments show this pattern; in some cases, there is little or no fitness decline (ΔM ≅ 0), even as significant genetic variance accrues among MA lines (ΔV > 0). We think that the fact that this most basic measurement of mutational effects is not consistent deserves explanation, even as approaches to estimating the full DFE become more advanced.

One explanation for ΔM ≅ 0 that can probably be dismissed is that no mutations have accumulated, regardless of fitness effect (U total = 0). Many MA studies now involve genome sequencing, confirming the presence of mutations. In general, MA experiments have been designed with a sufficient number of lines and generations that at least a modest number of mutations is expected, based on the genome size and likely mutation rate of the study organism. Another hypothesis is that mutations could be largely neutral, or do not affect the traits in question. However, increases in genetic variance under MA (ΔV > 0) indicate that at least some of the mutations must have measurable effects under the relevant assay conditions––the MA studies where ΔM ≅ 0 (see below) generally detect significant ΔV. Significant ΔV also implies sufficient statistical power to detect modest changes in mean fitness. Therefore, the most likely explanation for ΔM ≅ 0 is that the average effect of mutations on the trait is close to zero (E[s] ≅ 0), implying that beneficial and deleterious mutations have similar net impact on the trait. This runs counter to the idea that most mutations are deleterious and, in our view, this variation in outcomes lacks a consensus explanation.

In this perspective we summarize cases of MA experiments that show no evidence of fitness decline and discuss possible general explanations. We propose that inconsistency in the outcomes of MA experiments is an interesting pattern in its own right, because it may stem from genuine differences in mutation or “fitness landscapes” among genotypes, species, and environments.

Studies with ΔM ≅ 0

While many of the earliest studies using the MA method were conducted in the model fruit fly Drosophila melanogaster, none of these studies found results other than a fitness decline (Mukai 1964; Fry et al. 1999; Halligan and Keightley 2009). It should be noted that there are methodological difficulties with measuring MA line fitness in fruit flies: unlike many of the other organisms that have been studied under MA, in flies it has usually been impractical to effectively cryo-preserve the MA “ancestor” for later fitness comparisons (but see Pletcher et al. 1998). Instead, large populations are maintained as controls, which could potentially adapt and cause fitness decline to be over-estimated. However, deleterious mutations will also appear and segregate in control populations, creating a bias in the opposite direction. Attempts to account for evolution in controls or avoid the need for them have still concludes that fitness decline has occurred (Fry 2004; Sharp and Agrawal 2018). The results of MA in Drosophila are therefore quite consistent, but this consistency does not appear to hold in other organisms.

MA studies conducted in other animals––namely the model roundworm Caenorhabditis elegans and crustacean Daphnia pulex––also generally show significant declines in fitness, but with exceptions. In an early MA experiment, it was found that there was no significant decrease in two fitness metrics, reproductive output and lifespan, despite significant increases in variance (Keightley and Caballero 1997). The authors attributed this to a “deleterious mutation rate at least 100 times smaller than previously assumed” based on the framework of the previous D. melanogaster studies. Later MA studies in C. elegans found declines in fitness traits following MA, but two studies conducted by one group found no decline in lifespan following MA (Vassileva et al. 2000) and even noted one case of significantly improved lifespan after MA (Vassileva and Lynch 1999). In D. pulex, an MA study involving multiple genetic backgrounds found that while most showed a decline in fitness comparable to the results of other D. pulex MA experiments, a few genetic backgrounds actually outcompeted their ancestors in a competitive fitness assay (Schaak et al. 2014).

It is among plants that the most notable cases of MA studies without significant fitness decline can be found. Indeed, a debate around the notion that mutations are predominantly deleterious was ignited by a study in the model plant Arabidopsis thaliana (Shaw et al. 2000, 2002; Keightley and Lynch 2003) examined at three life history traits: number of seeds per fruit, fruit number, and reproductive mass after 17 generations of MA. No significant decline was noted for any of the traits despite a significant increase in among-line variance for all traits. A prior study in the same organism (Schultz et al. 1999) with more lines but fewer generations of MA detected a decline in fitness of about the same magnitude as in this later study, but it was found to be statistically significant although variance was not reported. From their data, Shaw et al. (2000) concluded that about half of all mutations that occurred in their study were beneficial. Later studies of these lines extended MA to further generations (Rutter et al. 2010, 2012) and in a variety of environmental conditions ranging from field plantings across geographic ranges to greenhouse studies and over different seasons (Roles et al. 2016; Rutter et al. 2018). Across this wide range of conditions, the results consistently showed little or no fitness decline, and evidence for substantial accumulation of beneficial mutations. Other studies in A. thaliana shows that mutations are dependent on environmental context by transplanting MA lines from vast gradients along the species’ range (Weng et al. 2021). In another plant study, 300 lines of the wild radish, Raphanus raphanistrum underwent 9 or 10 generations of MA in the field or the greenhouse, respectively (Roles and Connor 2008). No significant decline in fitness was detected, but the authors attributed this partially to the relatively few MA generations and possible selection pressure on the seeds during seed choice. In the same paper the authors found a decline in fitness compared with ancestors in less permissive growing conditions. MacKenzie et al. (2005) found no evidence of changes in the mean or variance of A. thaliana MA lines subjected to MA along with UV-B radiation, a common mutagen in plant MA studies.

Finally, there has also been mixed evidence of fitness decline in MA experiments with microbes. While “microbes” represent a far larger and less continuous slice of the tree of life than animals or land plants, we believe it is fitting to discuss them as a group because of their common laboratory culture and assay methodologies. The organisms most commonly studied under MA methods are yeasts, particularly the model Saccharomyces cerevisiae. In S. cerevisiae there have been experiments where mean fitness has declined following MA (e.g., Wloch et al. 2001; Joseph and Hall 2004; Dickinson 2008) but a series of experiments showed a much stronger signature of beneficial mutations than expected (Joseph and Hall 2004; Hall et al. 2008; Hall and Joseph 2010). These studies found that a large fraction of MA lines did not show significant deviation from ancestral fitness in traits like diploid growth rate, sporulation efficiency and haploid growth rate, with some lines showing significantly improved growth. A later MA study of comparable size and duration was conducted on both haploids and diploids from a single genetic background, finding significant decline in fitness in diploids but not haploids (Sharp et al. 2018). Other yeast studies have found a similar lack of fitness declines or improvements in a large number of their MA lines (Zeyl and DeVisser 2001). An MA study in green alga Chlamydomonas reinhardtii from multiple genetic backgrounds (Kraemer et al. 2017) found that, following ~ 1000 generations of MA, four genetic backgrounds showed fitness decline and one did not. This was suggested to be because said background was already very slow growing prior to MA, perhaps being evidence of poor adaptation to the laboratory and assay conditions (Kraemer et al. 2017; Bondel et al. 2021). Later, recombinant lines were generated by crossing six MA lines with their ancestors, generating 1526 recombinant lines, as a way of inferring the DFE; of the mutations with a fitness effect of 1% or more, one sixth were found to be beneficial (Bondel et al. 2019). Later modeling of the DFE of new mutations in C. reinhardtii MA lines has suggested a highly leptokurtic distribution and “approximately equal proportions” of growth rate increasing and growth rate reducing mutations (Böndel et al. 2021). Among bacteria, multiple studies have found significant fitness decline under MA in Escherichia coli (Kibota and Lynch 1996; Loewe et al. 2003; Funchain et al. 2000; Trindade et al. 2010), as well as Pseudomonas aeruginosa (Heilbron et al. 2014). The results of an MA experiment with Burkholderia cenocepacia were more mixed, with some lines showing significantly increased fitness in one environment (Dillon and Cooper 2016). In summary, MA in microbes mostly results in fitness decline, but exceptions can be found in multiple species.

Potential explanations

As shown above there are clearly many cases where the MA approach seems to result in a lack of fitness decline, and even fitness improvements. In discussing potential explanations, we should first consider limitations of the MA method itself. Theoretically, a new mutant allele will have about the same probability of fixation as that of a neutral allele when its effect on fitness is smaller than the inverse of the effective population size (Lynch and Walsh 1998; Lynch et al. 1998). In practice, this means that if N e is small (e.g., due to population bottlenecks) a mutation would have to be very impactful on fitness to be subject to any effective positive or negative selection (but note that most new alleles will be lost by chance even when selection is very effective). While the true distribution of the fitness effects of mutations is not known with much certainty, analyses often point to a highly skewed distribution, where weak effects are common and strong effects are rare (Eyre-Walker and Keightley 2007; Fig. 1). Under these circumstances, where most mutations have small effects, the MA strategy can be expected to be successful.

Fig. 1
figure 1

Hypothetical distribution of mutational effects. We often expect deleterious mutations (negative fitness effects) to appear more frequently than beneficial mutations (positive fitness effects). Available evidence suggests both types of mutations may have skewed distributions, where weak effects are common and strong effects are rare. Under MA, the effective population size (N e) will determine which mutations will behave neutrally. As N e increases, a larger fraction of mutations will be subject to effective selection, i.e., deleterious (beneficial) mutations would be less (more) likely to fix than the neutral expectation

Nevertheless, mutations with large fitness effects do exist, and experiments that combine sequencing with fitness assays have found that mutations with large effects can significantly skew the overall fitness of an MA line even if they are very rare (Schultz and Lynch 1997; Heilbron et al. 2014). Whether mutations of large effect will be acted upon by selection depends on how effectively N e can be reduced, which varies among study systems. In most MA studies on animals, only a single individual or pair is allowed to contribute to the next generation, ensuring that N e is small. In plants, single seeds can be used to propagate each MA line, but because plants do not have a segregated set of germlines cells, they are subject to clonal selection in the meristem Klekowski and Kazarinova-Fukshansky 1984; Otto and Orive 1995; Otto and Hastings 1998; Schoen and Schultz 2019; Cruzan et al. 2022), which may prevent some harmful mutations from being passed to the next generation. Similarly, more gene expression takes place in the male gametes of plants relative to animals, potentially exposing recessive mutations to selection (Mulachy et al. 1996; Otto et al. 2015). MA studies that include sequencing have some opportunity to address these potential sources of bias by examining the distribution of mutations within genes and across the genome (e.g., Zhu et al. 2014; Sharp and Agrawal 2016; Sharp et al. 2018; Liu and Zhang 2019; Weng et al. 2019).

Microbes suffer from a similar problem of clonal selection. Because microbial generation times are short, MA typically involves streaking to single colonies repeatedly. Multiple generations of growth between bottlenecks means that N e is not as low as it is in animal systems, and so selection could influence the fixation of new mutations. Statistical methods have been developed to account for selection when estimating the DFE in microbial MA experiments, greatly reducing the inferred frequency of beneficial mutations from both simulations and empirical datasets (Mahilkar et al. 2021; Wahl and Agashe 2022). These findings suggest that at least some of the variation in outcomes among microbial MA experiments is likely the result of differences in practical aspects of these experiments, such as transfer times and colony sizes.

One method to experimentally address how differences in N e affect the outcomes of MA is to conduct the procedure using a range of effective population sizes (e.g., Estes et al. 2004; Katju et al. 2014; Katju et al. 2018; Luijckx et al. 2018). These experiments have shown that as N e increases the likelihood of seeing a significant fitness decline decreases. For example, Katju et al. (2014) studied hermaphrodite nematode MA lines with various bottleneck sizes. Lines with N e = 1 showed significantly decreased productivity and survivorship after 409 generations, but lines with N e = 10 and N e = 100 did not. All population size treatments showed significant increases in among-line variance for both traits, with the exception of survivorship in the N e = 100 treatment. The simplest interpretation of these results is that deleterious alleles with moderate effects behaved neutrally in the smallest populations resulting in the observed decline of mean fitness, but selection counteracted this in the larger populations. Increased genetic variance in the larger populations implies that some mutations became fixed and produced changes in the fitness of those lines. This reinforces the idea that even with some level of selection acting, deleterious mutations are still able to reach fixation although at a lower rate (Schultz and Lynch 1997). The influence of population size on the rates of fixation for deleterious versus beneficial mutations is hard to predict, but these studies find little or no evidence of fitness improvement under large population sizes, suggesting beneficial mutations were mild or rare.

The above explanations center on selection as a source of bias during MA, inflating the impact of rare beneficial mutations and reducing the impact of deleterious mutations. An alternative explanation for the variation in fitness decline among MA experiments is that substantial beneficial genetic variation is available, but only in some circumstances. In particular, it is worth considering that different genotypes can have different histories of adaptation to any particular environment––genotypes that are initially well-adapted to the testing conditions for fitness may be more likely to show fitness decline under MA than genotypes that are poorly adapted to those conditions (Orr 2006; Stearns and Fenster 2016). A useful metaphor for this prediction is the “fitness landscape”, where a well-adapted genotype is located at or near the peak and less well adapted genotypes are farther from the peak (Fig. 2). In a genotype at the peak of such a landscape, a step in any direction would represent a descent: a movement towards a less fit genotype; consequently, mutations in well-adapted lines will tend to decrease fitness. By contrast, in genotypes further from the peak we might expect beneficial mutations to be more common, though the aggregate effect of multiple mutations is more difficult to predict, e.g., in the context of Fisher’s Geometric Model (Martin and Lenormand 2006; 2008; Tenaillon 2014; Martin and Lenormand 2015). Fitness landscapes in the wild may be complex and variable, perhaps creating many opportunities for beneficial mutations. Nevertheless, we can simply consider whether any given genotype has had an opportunity to adapt to given fitness assay conditions prior to MA.

Fig. 2
figure 2

The fitness landscape concept. Lighter shades indicate higher fitness. (a) In a well-adapted genotype (+ symbol) we expect the vast majority of mutations to be deleterious (red arrows). (b) In a poorly adapted genotype, the same genetic changes are more likely to be beneficial (green arrows)

While having huge practical benefits, seed storage and cryopreservation may limit the opportunities of laboratory organisms to become well-adapted to laboratory assay conditions. As noted above, Drosophila are rarely cryo-preserved, and outbred strains have typically been maintained in relatively large laboratory populations for many generations. We might therefore expect strong adaptation to fitness assay conditions in this system, particularly in terms of early-life traits (e.g., Sgro and Partridge 2000; 2001). In general, experimental evolution studies maintaining large populations show that various model organisms can improve their fitness under standard lab conditions (e.g., Frankham and Loebel 1992; Lenski et al. 1991; Knoppel et al. 2018; Johnson et al. 2021). Strains used in MA experiments that were previously maintained in small populations or under very different conditions might be expected to have access to mutations that improve fitness in a novel assay environment because of their lesser degree of adaptation (Bondel et al. 2021). Additionally, model organisms are often genetically modified, e.g., to control reproduction or create selectable markers, and may have had little opportunity for the evolution of compensatory changes.

There are a few direct tests of how adaptive history influences the fitness effects of spontaneous mutations, along with several types of indirect evidence. Stearns and Fenster (2016) performed EMS mutagenesis on A. thaliana strains and showed a negative correlation between founder fitness and the relative fitness of the post-treatment lines. Although mutagenesis creates a different spectrum of mutation than spontaneous mutation, this serves as support for the idea that the “adaptedness” of any MA line founder can influence the fitness consequences of MA. If mutations are more likely to be beneficial in less evolutionarily optimized genotypes, then we would expect an eventual reduction in the rate of fitness decline under MA, as the effects of new beneficial and deleterious mutations become balanced. Silander et al. (2017) observed just such an effect in a bacteriophage where the decline in fitness plateaued. Work that is conscious of the effect of genetic diversity on the fitness effects of mutations has increased, including studies on the fitness effects of MA in different genetic backgrounds (e.g., Schaak et al. 2014; Kraemer et al. 2016), which demonstrate that these effects are variable. An MA study of haploid and diploid S. cerevisiae from the same genetic background (Sharp et al. 2018) found that mean fitness declined in diploid lines but not haploid lines, contrary to the expectation that haploids should suffer more due to recessive deleterious mutations. It is possible that the particular genetic background used in this experiment was more adapted to growth in the diploid form, but not the haploid form. In nature this species spends most of its time in the diploid life stage and laboratory experiments with haploid populations have previously found evidence that diploidy is beneficial (e.g., Gerstein and Otto 2011; Voordeckers et al. 2015; Venkataram et al. 2016; Fisher et al. 2018; reviewed in Gerstein and Sharp 2021). The mutational spectrum differed between haploid and diploid MA lines in the Sharp et al. (2018) study, which somewhat confounds the comparison of their mean fitness. Additionally, selection in the haploid form against recessive alleles and the lack of masking of lethal alleles may have contributed to the result that was observed––in other studies of S. cerevisiae, sporulated MA lines showed no decline in fitness if recessive lethals were ignored (Joseph and Hall 2004; Hall et al. 2008). However, the haploid lines from Sharp et al. (2018) accumulated more mutations per base pair than diploids, and there were no indications of differential selection on mutations in haploids and diploids based on the molecular data.

More directly, a study in D. melanogaster disrupted 36 genes in strains that had previously adapted to two different stressful media environments (Wang et al. 2013). Each mutant strain was then tested in both the environment in which they adapted and in the alternative environment. The mean and variance of mutational effects depended on the assay environment, but not on adaptedness to the testing environment. A similar test used natural isolates of A. thaliana from across its native range. Weng et al. (2021) performed MA for 7–10 generations and then measured multiple traits at each natural site. MA lines showed greater reduction in fitness at “away” sites to which they had not been adapted than in their more familiar “home” sites.

At face value these studies do not seem to support the idea that diverse levels of prior adaptation can explain the variation in fitness decline observed in MA experiments. Instead, these experiments seem to reveal complex patterns, perhaps stemming from interactions between standing variation, new mutations, and environmental stresses. These studies also recapitulate other findings in their respective model organisms: little evidence for abundant beneficial mutations in D. melanogaster and a higher propensity towards it in A. thaliana. Given the relatively small set of mutations interrogated in the case of Wang et al. (2013), and the complex patterns of adaptation in the case of Weng et al. (2020), we suggest that it would be premature to discard the adaptedness hypothesis, and that more work on this question would be valuable.

Future directions

It remains an open question whether one can truly predict the fitness outcomes of any given MA experiment. Many experiments show a decline in fitness, but this outcome is far from uniform and the reasons for that are still being actively investigated. To fully understand the range of outcomes of MA experiments we recommend three strategies.

First, it is important to consider the limitations and assumptions underlying MA studies in general and how those factors may come into play depending on the biological realities of the study organism. After the first studies suggested that fitness does not necessarily decline after MA (Vassilieva and Lynch 1999; Shaw et al. 2000, 2002) some suggested that this could be due to the traits measured being under stabilizing selection rather than directional selection (Keightley and Lynch 2003; Shaw et al. 2003). In such a case, any deviation from the mean, regardless of sign, would actually correspond to a reduction in fitness. However, this suggestion has not been shown to be particularly relevant to the organisms they were directed at, A. thaliana and C. elegans, and the fitness metrics used in those studies continue to be used as fitness proxies. However, it is certainly possible for traits in a given organism to be under real or apparent stabilizing selection, and experimenters should carefully consider the life history of each study organism. Further, while MA is arguably the best available strategy for studying natural mutations, the resulting spectrum of mutations observed may still be influenced by selection to some extent, including within-individual selection (Otto and Orive 1995; Wei et al. 2019). We can think of within-individual selection as biasing the inheritance of new alleles in some organisms and not others, with the potential to contribute to variation in the outcomes of MA. Studies looking at mutation spectra following MA studies, however, have often found no evidence for somatic selection at least in A. thaliana (Monroe et al. 2022). An analogous challenge arises for studies of microorganisms, with colonies representing small populations with some opportunity for selection. It will be interesting to compare the effects of MA using the traditional approach versus the use of microfluidics to separate cells upon division (Kobel et al. 2010), maintaining truly small populations. Researchers should continue to develop and apply statistical methods to correct for potential biases due to selection (Mahilkar et al. 2021; Wahl and Agashe 2022).

Second, we would like to encourage experimenters to consider the specific adaptive history and genetic background of the founder lines and how that might affect the results and interpretation of MA. Microbes in particular spend much of the time in stasis, with little opportunity to adapt to the assay conditions to which they will later be subjected. While it may not always be feasible to give organisms time to adapt to their assay conditions, it remains important that results are interpreted in the light of this. More experiments that are designed to directly test the idea that adaptedness can impact the fitness effects of mutations would be particularly valuable. If beneficial mutations are indeed more likely for poorly adapted genotypes, MA should produce a less severe decline in fitness. Work like that of Silander et al. (2017) which sought to characterize the fitness effects of mutations in extremely unfit phenotypes, while difficult, could be performed in other organisms to perhaps get a finer understanding of how the fitness landscape behaves far away from the fitness optimum.

Additionally, organisms may have specific properties that affect the results we see from MA. In S. cerevisiae, for example, losses in fitness may be due to phenotypes created by aneuploidy (Joseph and Hall 2004; Hall et al. 2008; Sharp et al. 2018), a type of mutation we wouldn’t expect to accumulate in many other model organisms; species likely differ in both the overall spectrum of mutation types and in the fraction of these mutations that are effectively neutral under MA. Another possibility is that the DFE itself could evolve to reduce the rate of mutation to deleterious alleles. Recent work in A. thaliana using data from MA lines, polymorphism and divergence finds that mutations are biased against areas under purifying selection and toward those under positive selection (Weng et al. 2019; Monroe et al. 2022). This astounding result lines up with previous work in the system showing epigenetic regulation of DNA repair mechanisms to preferentially protect coding regions from mutation (Belfield et al. 2018) and provides a convincing explanation for why studies of A. thaliana tend to find little decay in mean fitness under MA. Biases in DNA repair activity and mutation have also been observed in tumors and human cell lines (Frigola et al. 2017; Supek and Lehner 2017; Huang et al. 2018), but the presence of such effects in organisms like bacteria is debated (Martincorena et al. 2012, Chen and Zhang 2013). Further studies combining molecular information on the (epi)genomics of DNA replication and repair coupled with fitness measures, along with theoretical model development, would help to establish the role of selection in determining the rate of deleterious mutation, and the causes of variation in these patterns among species. There also remains a need for further development of theoretical models addressing the DFE under non-equilibrium scenarios, particularly when environments fluctuate (e.g., Mustonen and Lassig 2009, Bataillon and Baley 2014).

Lastly, while MA can be informative about the aggregate effect of mutations on fitness, it provides much less insight into the impact of any specific mutation. To combat this, it is important to combine MA techniques with genomic sequencing and other methods that can allow a more fine-tuned analysis of individual mutations. The combination of fitness measures and genome sequencing for the same sets of MA lines, sometimes combined with crossing methods, is a promising way to understand genotype-phenotype connections (e.g., Shaw and Chang 2006; Rutter et al. 2012; Kraemer et al. 2017; Bondel et al. 2019; Sandell and Sharp 2021; Bondel et al. 2021). Mutation accumulation is a useful technique that has been applied for decades, and continues to be valuable in combination with other inference methods; these novel approaches hold promise for understanding the complex and sometimes unpredictable impacts of spontaneous mutations on fitness.

Availability of data and material:

Not applicable.

Code Availability

Not applicable.


  • Bataillon T, Bailey SF (2014) Effects of new mutations on fitness: insights from models and data. Ann N Y Acad Sci 1320:76–92

    PubMed  PubMed Central  Article  Google Scholar 

  • Belfield EJ, Ding ZJ, Jamieson FJ, Visscher AM, Zheng SJ, Mithani A, Harberd NP (2018) DNA mismatch repair preferentially protects genes from mutation. Genome Res 28(1):66–74

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Böndel KB, Kraemer SA, Samuels T, McClean D, Lachapelle J, Ness RW, … Keightley PD(2019) Inferring the distribution of fitness effects of spontaneous mutations in Chlamydomonas reinhardtii.PLoS biology, 17(6), e3000192

  • Böndel KB, Samuels T, Craig RJ, Ness RW, Colegrave N, Keightley PD (2021) The distribution of fitness effects of spontaneous mutations in Chlamydomonas reinhardtii inferred using frequency changes under experimental evolution. bioRxiv. Chen, X. & Zhang, J. (2013) No gene-specific optimization of mutation rate in Escherichia coli. Mol. Biol. Evol. 30(7) 1559–62

  • Cruzan MB, Streisfeld MA, Schwoch JA(2022) Fitness effects of somatic mutationsaccumulating during vegetative growth. BioRxiv, 392175.

  • Dillon MM, Cooper VS (2016) The fitness effects of spontaneous mutations nearly unseen by selection in a bacterium with multiple chromosomes. Genetics 204(3):1225–1238

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Dickinson WJ (2008) Synergistic fitness interactions and a high frequency of beneficial

  • changes among mutations accumulated under relaxed selection in Saccharomyces

  • cerevisiae.Genetics, 178(3),1571–1578

  • Estes S, Phillips PC, Denver DR, Thomas WK, Lynch M (2004) Mutation accumulation in populations of varying size: the distribution of mutational effects for fitness correlates in Caenorhabditis elegans. Genetics 166(3):1269–1279

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Eyre-Walker A, Keightley PD (2007) The distribution of fitness effects of new mutations. Nat Rev Genet 8(8):610–618

    CAS  PubMed  Article  Google Scholar 

  • Fisher KJ, Buskirk SW, Vignogna RC, Marad DA, Lang GI(2018) Adaptive genome duplication affects patterns of molecular evolution in Saccharomyces cerevisiae.PLoS genetics, 14(5), e1007396

  • Frankham R, Loebel DA (1992) Modeling problems in conservation genetics using captive Drosophila populations: rapid genetic adaptation to captivity. Zoo Biol 11(5):333–342

    Article  Google Scholar 

  • Frigola J, Sabarinathan R, Mularoni L, Muinos F, Gonzalez-Perez A, Lopez-Bigas N (2017) Reduced mutation rate in exons due to differential mismatch repair. Nat Genet 49:1684–1692

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Fry JD (2004) On the rate and linearity of viability declines in Drosophila mutation- accumulation experiments: genomic mutation rates and synergistic epistasis revisited. Genetics 166(2):797–806

    PubMed  PubMed Central  Article  Google Scholar 

  • Fry JD, Keightley PD, Heinsohn SL, Nuzhdin SV (1999) New estimates of the rates and effects of mildly deleterious mutation in Drosophila melanogaster. Proceedings of the National Academy of Sciences, 96(2), 574–579

  • Funchain P, Yeung A, Lee Stewart J, Lin R, Slupska MM, Miller JH (2000) The consequences of growth of a mutator strain of Escherichia coli as measured by loss of function among multiple gene targets and loss of fitness. Genetics 154(3):959–970

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • García-Dorado A (1997) The rate and effects distribution of viability mutation in

  • Gerstein AC, Otto SP(2011) Cryptic fitness advantage: diploids invade haploid populations despite lacking any apparent advantage as measured by standard fitness assays. PloS one, 6(12), e26599

  • Gerstein AC, Sharp NP (2021) The population genetics of ploidy change in unicellular fungi. FEMS microbiology reviews

  • Hall DW, Joseph SB (2010) A high frequency of beneficial mutations across multiple fitness components in Saccharomyces cerevisiae. Genetics 185(4):1397–1409

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Hall DW, Mahmoudizad R, Hurd AW, Joseph SB (2008) Spontaneous mutations in diploid Saccharomyces cerevisiae : another thousand cell generations. Genetics research, 90(3),229–241

  • Halligan DL, Keightley PD (2009) Spontaneous mutation accumulation studies in evolutionary genetics. Annu Rev Ecol Evol Syst 40:151–172

    Article  Google Scholar 

  • Heilbron K, Toll-Riera M, Kojadinovic M, MacLean RC (2014) Fitness is strongly influenced by rare mutations of large effect in a microbial mutation accumulation experiment. Genetics 197(3):981–990

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Huang Y, Gu L, Li G-M (2018) H3K36me3-mediated mismatch repair preferentially protects actively transcribed genes from mutation. J. Biol. Chem. 293(20): 7811–7823. Johnson, M. S., Gopalakrishnan, S., Goyal, J., Dillingham, M. E., Bakerlee, C. W., Humphrey

  • Joseph SB, Hall DW (2004) Spontaneous mutations in diploid Saccharomyces cerevisiae: more beneficial than expected. Genetics 168(4):1817–1825

    PubMed  PubMed Central  Article  Google Scholar 

  • Katju V, Packard LB, Bu L, Keightley PD, Bergthorsson U (2014) Fitness decline in spontaneous mutation accumulation lines of Caenorhabditis elegans with varyin effective population sizes. Evolution 69(1):104–116

    PubMed  Article  Google Scholar 

  • Katju V, Packard LB, Keightley PD (2018) Fitness decline under osmotic stress in Caenorhabditis elegans populations subjected to spontaneous mutation accumulation at varying population sizes. Evolution 72(4):1000–1008

    CAS  PubMed  Article  Google Scholar 

  • Keightley PD (1994) The distribution of mutation effects on viability in Drosophila melanogaster. Genetics 138(4):1315–1322

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Keightley PD, Caballero A (1997) Genomic mutation rates for lifetime reproductive output and lifespan in Caenorhabditis elegans. Proceedings of the National Academy of Sciences, 94(8), 3823–3827

  • Keightley PD, Lynch M (2003) Toward a realistic model of mutations affecting fitness. Evolution 57(3):683–685

    PubMed  Article  Google Scholar 

  • Kibota TT, Lynch M (1996) Estimate of the genomic mutation rate deleterious to overall fitness in E. coll. Nature 381(6584):694–696

    CAS  PubMed  Article  Google Scholar 

  • Klekowski EJ Jr, Kazarinova-Fukshansky N (1984) Shoot apical meristems and mutations: fixation of selectively neutral cell genotypes and selective loss of disadvantageous cell genotypes. Bioscience 34(3):180–181

    Article  Google Scholar 

  • Knöppel A, Knopp M, Albrecht LM, Lundin E, Lustig U, Näsvall J, Andersson DI (2018) Genetic adaptation to growth under laboratory conditions in Escherichia coli and Salmonella enterica. Front Microbiol 9:756

    PubMed  PubMed Central  Article  Google Scholar 

  • Kobel S, Valero A, Latt J, Renaud P, Lutolf M (2010) Optimization of microfluidic single cell trapping for long-term on-chip culture. Lab Chip 10(7):857–863

    CAS  PubMed  Article  Google Scholar 

  • Kraemer SA, Morgan AD, Ness RW, Keightley PD, Colegrave N (2016) Fitness effects of new mutations in Chlamydomonas reinhardtii across two stress gradients. J Evol Biol 29(3):583–593

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Lande R (1994) Risk of population extinction from fixation of new deleterious mutations. Evolution 48(5):1460–1469

    PubMed  Article  Google Scholar 

  • Lenski RE, Rose MR, Simpson SC, Tadler SC (1991) Long-term experimental evolution in Escherichia coli. I. Adaptation and divergence during 2,000 generations. Am Nat 138(6):1315–1341

    Article  Google Scholar 

  • Liu H, Zhang J (2019) Yeast spontaneous mutation rate and spectrum vary with environment. Curr Biol 29(10):1584–1591

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Loewe L, Textor V, Scherer S (2003) High deleterious genomic mutation rate in stationary phase of Escherichia coli. Science 302(5650):1558–1560

    CAS  PubMed  Article  Google Scholar 

  • Luijckx P, Ho EK, Stanić A, Agrawal AF (2018) Mutation accumulation in populations of varying size: large effect mutations cause most mutational decline in the rotifer Brachionus calyciflorus under UV-C radiation. J Evol Biol 31(6):924–932

    CAS  PubMed  Article  Google Scholar 

  • Lynch M, Latta L, Hicks J, Giorgianni M (1998) Mutation, selection, and the maintenance of life-history variation in a natural population. Evolution 52(3):727–733

    PubMed  Article  Google Scholar 

  • Lynch M, Walsh B, Saadé JL, Le FE, Bureau QH, Schoen DJ (1998) (2005). Genomic mutation in lines of Arabidopsis thaliana exposed to ultraviolet-B radiation. Genetics, 171(2), 715–723

  • Mahilkar A, Kemkar S, Saini S (2021) Selection in a growing bacterial/yeast colony biases results of mutation accumulation experiments. bioRxiv. Martin, G., & Lenormand, T. (2006). The fitness effect of mutations across environments: a survey in light of fitness landscape models. Evolution, 60(12), 2413–2427

  • Martin G, Lenormand T (2008) The distribution of beneficial and fixed mutation fitness effects close to an optimum. Genetics 179(2):907–916

    PubMed  PubMed Central  Article  Google Scholar 

  • Martin G, Lenormand T (2015) The fitness effect of mutations across environments: Fisher’s geometrical model with multiple optima. Evolution 69(6):1433–1447

    PubMed  Article  Google Scholar 

  • Martincorena I, Seshasayee ASN, Luscombe NM (2012) Evidence of non-random mutation rates suggests an evolutionary risk management strategy. Nature 485, 95–98. Monroe, J., Srikant, T., Carbonell-Bejerano, P., Becker, C., Lensink, M., Exposito-Alonso, M., … Weigel, D. (2022). Mutation bias reflects natural selection in Arabidopsis thaliana. Nature, 1–5

  • Mukai T, Mustonen V, Lassig M (1964) The genetic structure of natural populations of Drosophila melanogaster. I. Spontaneous mutation rate of polygenes controlling viability. Genetics, 50(1), 1. Mulcahy, D. L., Sari-Gorla, M., & Mulcahy, G. B. (1996). Pollen selection—past, present and future. Sexual Plant Reproduction, 9(6), 353–356. Mustonen V, Lassig M (2009) From fitness landscapes to seascapes: non-equilibrium dynamics of selection and adaptation. Trends Genet. 25(3), 111–119

  • Orr HA (2006) The distribution of fitness effects among beneficial mutations in Fisher’s geometric model of adaptation. J Theor Biol 238(2):279–285

    PubMed  Article  Google Scholar 

  • Otto SP, Hastings IM (1998) Mutation and selection within the individual. Genetica 102:507–524

    PubMed  Article  Google Scholar 

  • Otto SP, Orive ME (1995) Evolutionary consequences of mutation and selection within an individual. Genetics 141(3):1173–1187

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Otto SP, Scott MF, Immler S (2015) Evolution of haploid selection in predominantly diploid organisms. Proceedings of the National Academy of Sciences, 112(52), 15952- 15957

  • Pletcher SD, Houle D, Curtsinger JW (1998) Age-specific properties of spontaneous mutations affecting mortality in Drosophila melanogaster. Genetics 148(1):287–303

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Roles AJ, Conner JK (2008) Fitness effects of mutation accumulation in a natural outbred population of wild radish (Raphanus raphanistrum): comparison of field and greenhouse environments. Evolution: Int J Org Evol 62(5):1066–1075

    Article  Google Scholar 

  • Roles AJ, Rutter MT, Dworkin I, Fenster CB, Conner JK (2016) Field measurements of genotype by environment interaction for fitness caused by spontaneous mutations in Arabidopsis thaliana. Evolution 70(5):1039–1050

    CAS  PubMed  Article  Google Scholar 

  • Rutter MT, Shaw FH, Fenster CB (2010) Spontaneous mutation parameters for Arabidopsis thaliana measured in the wild. Evolution: Int J Org Evol 64(6):1825–1835

    Article  Google Scholar 

  • Rutter MT, Roles A, Conner JK, Shaw RG, Shaw FH, Schneeberger K, Fenster CB (2012) Fitness of Arabidopsis thaliana mutation accumulation lines whose spontaneous mutations are known. Evolution: Int J Org Evol 66(7):2335–2339

    Article  Google Scholar 

  • Rutter MT, Roles AJ, Fenster CB (2018) Quantifying natural seasonal variation in

  • mutation parameters with mutation accumulation lines.Ecology and evolution, 8(11),5575–5585

  • Sandell L, Sharp NP(2021) Submitted. Fitness effects of mutations: An assessment of PROVEAN predictions using mutation accumulation data.

  • Schaack, S., Allen, D. E., Latta IV, L. C., Morgan, K. K., &amp; Lynch, M. (2013). The effect of spontaneous mutations on competitive ability.Journal of evolutionary biology, 26(2),451–456

  • Schoen DJ, Schultz ST (2019) Somatic mutation and evolution in plants. Annu Rev Ecol Evol Syst 50:49–73

    Article  Google Scholar 

  • Schultz ST, Lynch M (1997) Mutation and extinction: the role of variable mutational effects, synergistic epistasis, beneficial mutations, and degree of outcrossing. Evolution 51(5):1363–1371

    PubMed  Article  Google Scholar 

  • Schultz ST, Lynch M, Willis JH(1999) Spontaneous deleterious mutation in Arabidopsis thaliana. Proceedings of the National Academy of Sciences, 96(20), 11393–11398

  • Sgrò CM, Partridge L (2000) Evolutionary responses of the life history of wild-caught Drosophila melanogaster to two standard methods of laboratory culture. Am Nat 156(4):341–353

    Article  Google Scholar 

  • Sgrò CM, Partridge L (2001) Laboratory adaptation of life history in Drosophila. Am Nat 158(6):657–658

    PubMed  Article  Google Scholar 

  • Sharp NP, Agrawal AF(2016) Low genetic quality alters key dimensions of the mutational spectrum.PLoS biology, 14(3), e1002419

  • Sharp NP, Agrawal AF(2018) An experimental test of the mutation-selection balance model for the maintenance of genetic variance in fitness components. Proceedings of the Royal Society B, 285(1890), 20181864

  • Sharp NP, Sandell L, James CG, Otto SP(2018) The genome-wide rate and spectrum of spontaneous mutations differ between haploid and diploid yeast. Proceedings of the National Academy of Sciences, 115(22), E5046-E5055

  • Shaw RG, Byers DL, Darmo E (2000) Spontaneous mutational effects on reproductive traits of Arabidopsis thaliana. Genetics 155(1):369–378

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Shaw RG, Chang SM (2006) Gene action of new mutations in Arabidopsis thaliana. Genetics 172(3):1855–1865

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Shaw FH, Geyer CJ, Shaw RG (2002) A comprehensive model of mutations affecting fitness and inferences for Arabidopsis thaliana. Evolution 56(3):453–463

    PubMed  Article  Google Scholar 

  • Shaw RG, Shaw FH, Geyer C (2003) What fraction of mutations reduces fitness? A reply to Keightley and Lynch. Evolution 57(3):686–689

    Google Scholar 

  • Silander OK, Tenaillon O, Chao L(2007) Understanding the evolutionary fate of finite populations: the dynamics of mutational effects.PLoS biology, 5(4), e94

  • Stearns FW, Fenster CB (2016) Fisher’s geometric model predicts the effects of random mutations when tested in the wild. Evolution 70(2):495–501

    CAS  PubMed  Article  Google Scholar 

  • Supek F, Lehner B (2017) Clustered Mutation Signatures Reveal that Error-Prone DNA Repair Targets Mutations to Active Genes. Cell 170(3):534–547

    CAS  PubMed  Article  Google Scholar 

  • Tenaillon O (2014) The utility of Fisher’s geometric model in evolutionary genetics. Annu Rev Ecol Evol Syst 45:179–201

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Trindade S, Perfeito L, Gordo I (2010) Rate and effects of spontaneous mutations that affect fitness in mutator Escherichia coli. Philosophical Trans Royal Soc B: Biol Sci 365(1544):1177–1186

    Article  Google Scholar 

  • Vassilieva LL, Lynch M (1999) The rate of spontaneous mutation for life-history traits in Caenorhabditis elegans. Genetics 151(1):119–129

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Vassilieva LL, Hook AM, Lynch M (2000) The fitness effects of spontaneous mutations in Caenorhabditis elegans. Evolution 54(4):1234–1246

    CAS  PubMed  Article  Google Scholar 

  • Venkataram S, Dunn B, Li Y, Agarwala A, Chang J, Ebel ER, Petrov DA (2016) Development of a comprehensive genotype-to-fitness map of adaptation-driving mutations in yeast. Cell 166(6):1585–1596

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Voordeckers K, Kominek J, Das A, Espinosa-Cantu A, De Maeyer D, Arslan A, … Verstrepen KJ(2015) Adaptation to high ethanol reveals complex evolutionary pathways.PLoS genetics, 11(11), e1005635

  • Wang AD, Sharp NP, Agrawal AF (2013) Sensitivity of the distribution of mutational fitness effects to environment, genetic background, and adaptedness: a case study with Drosophila. Evolution 68(3):840–853

    PubMed  Article  Google Scholar 

  • Wahl LM, Agashe D(2022) Selection bias in mutation accumulation. Evolution. Wei, W., Tuna, S., Keogh, M. J., Smith, K. R., Aitman, T. J., Beales, P. L., … Chinnery, P. F. (2019). Germline selection shapes human mitochondrial DNA diversity. Science, 364(6442), eaau6520

  • Weng ML, Ågren J, Imbert E, Nottebrock H, Rutter MT, Fenster CB (2021) Fitness effects of mutation in natural populations of Arabidopsis thaliana reveal a complex influence of local adaptation. Evolution 75(2):330–348

    CAS  PubMed  Article  Google Scholar 

  • Weng ML, Becker C, Hildebrandt J, Neumann M, Rutter MT, Shaw RG, Fenster CB (2019) Fine-grained analysis of spontaneous mutation spectrum and frequency in Arabidopsis thaliana. Genetics 211(2):703–714

    CAS  PubMed  Article  Google Scholar 

  • Wloch DM, Szafraniec K, Borts RH, Korona R (2001) Direct estimate of the mutation rate and the distribution of fitness effects in the yeast Saccharomyces cerevisiae. Genetics 159(2):441–452

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Zeyl C, DeVisser JAG (2001) Estimates of the rate and distribution of fitness effects of spontaneous mutation in Saccharomyces cerevisiae. Genetics 157(1):53–61

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  • Zhu YO, Siegal ML, Hall DW, Petrov DA(2014) Precise estimates of mutation rate and spectrum in yeast. Proceedings of the National Academy of Sciences, 111(22), E2310-E2318

Download references


This work was supported by NIH T32 Predoctoral Training Program in Genetics to KB.

Author information

Authors and Affiliations



KB, RHM and NPS wrote the manuscript.

Corresponding author

Correspondence to Kevin Bao.

Ethics declarations

Conflicts of interest/Competing interests:

The authors declare no conflict of interest.

Ethics approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

All authors have seen and approved the final manuscript.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Bao, K., Melde, R.H. & Sharp, N.P. Are mutations usually deleterious? A perspective on the fitness effects of mutation accumulation. Evol Ecol (2022).

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • DOI:


  • Distribution of fitness effects
  • Life history traits
  • Beneficial mutations
  • Fitness landscape