Population genetic theories predict a negative relationship between effective population size (N_e) and the fraction of deleterious mutations (Kimura and Ohta, 1978; Kimura, 1983). According to this, populations with small N_e accumulate more deleterious mutations than those with large N_e. This is because the effect of genetic drift is high amongst small populations, and hence the purifying selection is inefficient in purging deleterious mutations. Initially, accumulation of deleterious mutations was assumed to occur only in the asexual population, which was called as “Muller’s ratchet” (Muller, 1964; Felsenstein, 1974). Later, Lynch et al. (1995) showed the possibility of accumulation of harmful mutations in sexual populations, particularly when the population is small. The proportion of deleterious mutations has been estimated in a number of species, including humans, chimpanzees, mice, cows, chickens, fruit flies, and plants (Eyre-Walker and Keightley, 1999; Schultz et al., 1999; Li and Saunders, 2005; Rubin et al., 2010; Bourneuf et al., 2017; Sohail, et al., 2017). These studies showed that amongst mammals, humans have the highest proportion of deleterious mutations. This is because the ancestral effective population size of humans is known to be the smallest of the mammals examined so far (Keightley and Eyre-Walker, 2000). Comparisons involving different populations within a species also showed similar patterns. For example, previous studies reported that European Americans have a significantly higher proportion of deleterious mutations than African Americans (Lohmueller et al., 2008; Fu et al., 2014; Subramanian, 2016). This observation could be explained based on the population bottleneck that occurred in Europeans when they migrated out of Africa. This was further confirmed by another large-scale study that showed a positive correlation between deleterious mutation load and the distance from Africa (Henn et al., 2016). Human populations experienced serial population bottlenecks as they migrated out of Africa, and populations that are furthest from Africa might have experienced more bottlenecks than those proximal to Africa. Therefore, the latter harbour more deleterious mutations than the former (Henn et al., 2016).

The effects of bottlenecks were also observed in populations that migrated much more recently. For instance, French Canadians have more deleterious mutations than those who live in France or Non-French Canadians (Peischl et al., 2018). Similarly, people living on Islands such as Greenland were also found to have more deleterious mutations than their mainland counterparts (Pedersen et al., 2017). In both cases, severe bottlenecks occurred due to the limited number of individuals who formed the founders of these populations. Hence the reduction in the population size caused the accumulation of higher proportions of deleterious mutations in French Canadians and Greenlanders than in their mainland counterparts.

During domestication, only a small subset of wild animals was selected, and hence the effective population size of domesticated animals is significantly reduced compared to their wild counterparts (Moyers et al., 2018). Due to this reason, the proportion of deleterious mutations is expected to be higher in domesticated animals than in their wild progenitors. A number of studies provided empirical evidence for this prediction. For example, an earlier study showed that domesticated dog breeds have a much higher deleterious mutation load than wild wolves (Marsden et al., 2016). Similar results were reported by comparing the exomes of domesticated and wild yaks (Xie et al., 2018). Using 432 animals belonging to 54 worldwide cattle breeds, a previous study found a negative correlation between heterozygosity and deleterious mutational load, which suggests that breeds that have low genetic variation have higher deleterious mutational load than those with high genetic variation (Subramanian, 2021). Since heterozygosity is the product of N_e and mutation rate (μ), and the mutation rate is largely similar across breeds, the above relationship implies a negative correlation between N_e and the mutation load.

Apart from domestication, the difference in the natural population sizes also cause a variation in the mutational loads. For instance, the island population of kakapo (a parrot species endemic to New Zealand) was found to carry a much higher proportion of deleterious variants than the mainland population (Foster et al., 2021). Similarly, the deleterious mutational load of mammoths that lived on Wrangel island was much higher than that of those who lived on the mainland (Rogers and Slatkin, 2017). Since the population size of island species are smaller than that of mainland ones, the accumulation of deleterious mutations was the result of high genetic drift modulating the mutations of the former. Similarly, killifish populations in drylands harboured more deleterious variants than those from the wetlands (Willemsen et al., 2020). This is because the former is known to have a smaller population size than the latter. Furthermore, a study on Alpine ibex found evidence for the accumulation of mildly deleterious SNVs but the purging of highly deleterious SNVs during population bottleneck (Grossen et al., 2020).

Almost all of the previous studies used the ratio of nonsynonymous-to-synonymous diversities (dN/dS) as the proxy for the deleterious mutation loads. Their findings clearly demonstrated low and high mutation loads in populations with large and small N_e. However, genetic admixture occurs quite often between populations with distinctly different N_e. For instance, a previous study showed the fitness of Neanderthals was significantly less than that of humans because the former has smaller N_e than the latter (Harris and Nielsen, 2016). Furthermore, this study revealed that non-African populations that admixed with Neanderthals (non-Africans) have a 0.5% reduction in fitness (Harris and Nielsen, 2016). However, it is important to understand how the mutational load of the admixed progenies derived from two populations (or subspecies) with widely different N_e (and, in turn, mutation loads) is shaped by evolution. To investigate this, we obtained the whole genome data of Mus musculus castaneus and Mus musculus musculus and their hybrid or admixed populations. We specifically selected these because the subspecies M. m. musculus and M. m. castaneus have population sizes of 60,000–120,000 and 200,000–400,000, respectively (Salcedo et al., 2007; Geraldes et al., 2008), and previous studies have found a much higher mutational load in the former compared to that observed in the latter (Phifer-Rixey et al., 2012; Subramanian, 2018). We estimated the dN/dS ratios and the proportion of deleterious nonsynonymous SNVs (nSNVs) for the admixed mouse population and compared them with those of their parental populations. We also examined the correlation between the level of admixture and deleterious mutational loads.

Population genetic theories predict a higher proportion of deleterious mutations in a population with a small N_e compared to that with a large N_e (Crow and Kimura, 1970; Phifer-Rixey et al., 2012; Marsden et al., 2016; Subramanian, 2021). Many previous studies provided empirical evidence for this prediction using genome data from humans and other vertebrates (Lu et al., 2006; Mezmouk and Ross-Ibarra, 2014; Renaut and Rieseberg, 2015; Kono et al., 2016; Marsden et al., 2016; Pedersen et al., 2017; Ramu et al., 2017; Makino et al., 2018; Peischl et al., 2018; Xie et al., 2018; Bosse et al., 2019; Robinson et al., 2019; Dussex et al., 2021; Subramanian, 2021). Using the dN/dS ratio as the measure to quantify deleterious mutation load, these studies consistently showed that the populations with small N_e had a higher ratio than those with large N_e. However, the mutation load of the populations that derive from genetic admixture between two parental populations with distinctly different N_e is not known. In the present study, using the dN/dS ratio and the proportion of deleterious nSNVs as proxies to quantify deleterious mutation load in mouse populations that result from the admixture between the subspecies M. m. castaneus and M. m. musculus. We report two main findings. First, the mean dN/dS ratio and the proportion of deleterious nSNVs estimated for the admixed mouse population was intermediate between those of observed for their parental populations. Second, these measures positively correlated with the fraction of M. m. musculus ancestry (or negatively correlated with the fraction of M. m. castaneus ancestry).

The results of this study can be explained by the fact that the N_e of M. m. castaneus (200,000–400,000) is much higher than that of M. m. musculus (60,000–120,000; Salcedo et al., 2007; Geraldes et al., 2008). Since deleterious mutation load negatively correlates N_e (Kimura and Ohta, 1978; Kimura, 1983), this load is expected to be higher in M. m. musculus compared to that in M. m. castaneus. Our results exactly showed this, as the dN/dS ratio (Figure 2A) and the proportion of deleterious nSNVs (Figure 3) were found to be higher for the former than those for the latter. On the other hand, the deleterious mutation load for the population that derived from the admixture of these two subspecies is expected to be less than that of M. m. musculus as they have a fraction of M. m. castaneus ancestry, which will tend to reduce the overall load of the admixed individuals. Similarly, the deleterious mutation load of the admixed population is expected to be higher than that of M. m. castaneus as they have a fraction of M. m. musculus ancestry, which will tend to increase the overall load of the hybrids. Due to this reason, the dN/dS ratio (Figure 2A) and the proportion of deleterious nSNVs (Figure 3) observed for admixed genomes were higher than those of M. m. castaneus and less than those of M. m. musculus population. The above prediction was further confirmed by the positive correlation between the proportion of M. m. musculus ancestry and the dN/dS ratio (Figure 2B) or the proportion of deleterious nSNVs (Figure 4). Note that if we used the M. m. castaneus ancestry we found a negative correlation with the same strength and same level of statistical significance.

Previous studies on dogs, cows, and yak used deleterious heterozygous and homozygous nSNVs counts to compare mutational loads between populations with different N_e (Marsden et al., 2016; Xie et al., 2018; Subramanian, 2021). However, populations with small N_e are expected to have a higher number of deleterious as well as neutral (or benign) homozygous nSNVs than those with large N_e. Therefore, the above studies observed a high number of deleterious and neutral homozygous SNVs in the genomes of the former and low in the latter. Since the deleterious and neutral homozygous SNVs show similar patterns, the use of SNV counts to measure the mutational load is not accurate. Due to this reason, we used the proportion of deleterious homozygous nSNVs, which was calculated by dividing the number of deleterious homozygous nSNVs by the number of all (benign + deleterious) homozygous nSNVs (Figures 3B, 4B) and this proportion removes the bias expected in using the nSNV counts. A similar measure was used for heterozygous deleterious nSNVs as well (Figures 3C, 4C).

Although we observed the ratio of nonsynonymous and synonymous diversities in the admixed genomes it is important to examine the loads expected in these based on the allele frequencies of their parental populations. Although the observed nonsynonymous and synonymous diversities were significantly smaller than those predicted, their ratios were not. This is because the rate of reduction in the observed estimates of both nonsynonymous and synonymous sites was almost the same (~11%). This result and the similarity between the predicted and observed proportion of deleterious nSNVs suggest that the level of admixture more likely determines the mutational loads of the admixed progenies.

The distribution of fitness effects of nonsynonymous nSNVs revealed a higher proportion of mildly deleterious SNVs and a lower proportion of moderately and highly deleterious nSNVs in M. m. musculus than M. m. castaneus. This suggests that some of the highly deleterious nSNVs that were removed from M. m. castaneus population were present or segregating in the M. m. musculus population. This is because these nSNVs were deleterious for the former but became or behaved as mildly deleterious in the latter population. The potential reason is that the presence or removal of the nSNVs is determined by the product of effective population size (N_e) and selection coefficient (s). Therefore, for a large population, the N_es will be larger than that of a small population for the same vale of s. Hence, the fitness consequences of nSNVs in small populations will be relatively less deleterious than that in large populations. Therefore, some of the nSNVs that were highly deleterious for M. m. castaneus were mildly deleterious for M. m. musculus. The genetic Admixture between M. m. castaneus and M. m. musculus results in the loss of some of the moderately deleterious nSNVs whilst gaining or accumulating some of the mildly deleterious ones.

Although many previous studies estimated the mutation load in pure or unadmixed populations, how these loads are modulated by the proportion of ancestry in admixed genomes was not clearly demonstrated before. Hence, the analyses performed here make this study novel and unique. Due to this new approach, the results of this study have opened up new dimensions in understanding the diversity and mutational load of admixed populations in comparison to their ancestral/parental populations. Eventually, this approach will help to recognise the role of gene flow between populations, particularly in exchanging harmful mutations.

Using the whole genome data, this study revealed the patterns of deleterious mutation load in a mouse population, which derived from genetic admixture between two distinct populations belonging to the subspecies M. m. castaneus and M. m. domesticus. The results showed the dN/dS ratio and the proportion of deleterious nSNVs of admixed populations were intermediate compared to their parents. This could be observed only if the parental populations had significant variation in their mutational load, which is driven by their effective population sizes. Our study also revealed a significant positive correlation between the proportion of admixture and the dN/dS ratio or the proportion of deleterious nSNVs in admixed individuals. This suggests the role of admixture in shaping the deleterious mutation loads. Whilst this was observed in mice, it is highly likely that similar patterns are expected in other vertebrates as well. Therefore, it is important to consider the status and the level of genetic admixture of the populations whilst estimating the mutation loads.

Publicly available datasets were analysed in this study. This data can be found at: https://www.ddbj.nig.ac.jp/bioproject/ accession: PRJDB11027.

SS conceived the idea, designed and supervised the research, and wrote the manuscript with inputs from other authors. UR performed research and data analysis, and contributed to writing the manuscript. AE supervised the research and contributed to writing and editing the manuscript. All authors contributed to the article and approved the submitted version.

The research was supported by the University of the Sunshine Coast DVCR grant awarded to SS.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.