Natural radioactivity is the main natural source of exposure to ionizing radiations on Earth. Natural radioactivity is generated by cosmic radiation or by radionuclides released from the bedrock. While levels of cosmic radiation fluctuates over time due to cosmic events such as supernovae or solar flares, bedrock radioactivity remained mainly stable until 2 billion years ago, when it began to slowly decrease (Karam and Leslie, 2005). Bedrock radioactivity depends on the nature of the rocks which extensively varies spatially (e.g. Ielsch et al., 2017). While few extremely naturally radioactive sites such as the India Kerala and Iranian Ramsar region have been monitored for their impact on the human mutation rate (Forster et al., 2002; Masoomi et al., 2006) or on plant physiology (Saghirzadeh et al., 2008), the influence of regional variation in baseline natural radioactivity on the evolution of biodiversity is still unknown (Møller and Mousseau, 2013).

Radioactivity can impact species’ molecular evolution by modulating the rate at which different types of mutation appear and accumulate. Ionizing radiations damage DNA by breaking the DNA sugar-phosphate backbone (Hoeijmakers, 2001; van Gent et al., 2001). Alternatively, ionizing radiations can trigger the formation of reactive oxygen species (ROS) directly inside cells through the radiolysis of water (Wallace, 1998; Ward, 1988). ROS being also mutagenic (Barja, 2002), ionizing radiations are direct and indirect mutagens. The most deleterious and studied mutations generated by radioactivity are DNA double-strand breaks (DSBs). Two repair systems are able to manage double-strand breaks: homologous recombination (HR) and non-homologous end-joining (NHEJ). HR uses an homologous sequence to revert the damage, but is only used at specific cell cycles when an undamaged sister chromatid is available (Karran, 2000). Otherwise, NHEJ is preferentially used. As NHEJ does not use a template-strand to reconstruct the missing genetic information, it only restores the continuity of the DNA molecule to the price of frequent deletions and chromosomal exchanges. Rothkamm et al., 2001 showed that 50% of the double-strand breaks were misrejoined after a strong irradiation (80 Gy). Numerous studies demonstrate that an exposure to radiation produces chromosomal abnormalities (Dikomey et al., 2000; Loucas et al., 2004; Hande et al., 2005), deletions (Jostes et al., 1994; Huo et al., 2001; Adewoye et al., 2015; Allegrucci et al., 2015), and point mutations (Huo et al., 2001; Winegar et al., 1994; Forster et al., 2002; Barber et al., 2002). Radiation-induced chromosomal abnormalities and deletions have been thoroughly studied because of their frequent deleterious impact. The impact of radiation on point mutations has received less attention.

While the mutational impact of exposure to high doses of radioactivity is well characterized (Dubrova et al., 1996; Ziegler et al., 1993), the exposure to low doses of radioactivity is poorly known. Some authors (Tubiana et al., 2006) propose that DNA repair and apoptosis may completely counteract the effect of ionizing radiations for doses below 0.1 Gy, suggesting that low doses have no biological impact. Indeed, in vitro experiments on mammalian cells show that repair systems are more efficient to repair DSBs at low dose than at higher dose (Boucher et al., 2004, 0.05 vs 3.5 Gy/min). However, exposure to an even lower dose of ionizing radiation (less than 0.1 cGy/min) increases the number of mutants in mammals (Vilenchik and Knudson, 2000; Hooker et al., 2004) suggesting that repair systems are not activated at very low radiation doses. When repair systems are not activated and DNA damage accumulates, cells tend to die by apoptosis, preventing the transmission of radioactivity-induced mutations to the next cell generation (Rothkamm et al., 2001; Collis et al., 2004). Unrepaired DSBs are the main cause of radioactivity-induced cell apoptosis during mitosis (Krueger et al., 2007). As point mutations will not lead to cell death like DSBs, this type of radioactivity-induced mutations could stay unrepaired and be transmitted to the next generation. Exposure to low doses of radioactivity also induces an adaptive response: exposed cells are resistant to a following higher dose of radiation (see Rigaud and Moustacchi, 1996). An early or chronic exposure to low doses of radiation may strengthen the antioxidant defense and reduce the sensitivity to radioactivity-induced ROS. While this effect is well demonstrated for preventing DSBs, its efficiency to reduce point mutations is less certain (Rigaud et al., 1995). Moreover, the adaptive response seems to only protect the nuclear genome while the mitochondrial one may not benefit from it (Jarrett and Boulton, 2005).

The short-term impact of high doses of radiation is indisputable, but the long-term impact of exposure to low doses is unknown. Some studies found an increase in the number of mutations in the offspring of exposed people (Dubrova et al., 1996; Forster et al., 2002), while others studies found the opposite (Satoh et al., 1996; Czeizel et al., 1991), leaving open the question of the transmission of mutations generated by low doses of radioactivity. This lack of knowledge is likely contributed to by three factors. First, studies are mainly focused on the health effect of low dose of radiation and not on their long-term mutational impact (e.g. Beir, 2006; Tubiana et al., 2006). Second, most of the literature focuses on unrepaired double-strand breaks which are highly deleterious and are de facto not transmitted to the next generation. Third, studying the long-term mutational impact of natural radioactivity is challenging because it raises a number of methodological difficulties. On the one hand, experimentally exposing multiple generations of multicellular organisms to low doses of radiation would require years of experimentation and complex experimental controls. On the other hand, the main obstacles to in naturae studies are the organisms’ mobility, which prevents certainty that a population was exposed to the same natural radioactivity for many generations, and confounding factors such as ultraviolet radiation from the sun.

Here, we overcome these difficulties by coupling in situ radioactivity characterizations with the distinctive bio-ecological characteristics of subterranean waterlice within a phylogenetic comparative framework. Subterranean waterlice are never exposed to UV radiation, live in contrasted bedrock set-ups and have very limited dispersal capacity (Eme et al., 2018), allowing us to make the assumption that different species have persisted in different but nearly constant radioactive habitats for numerous generations. To test the impact of radioactivity on the transmission of point mutations, we estimated the long-term rate of neutral mutations in the nuclear and in the mitochondrial genome independently.

In order to build a robust and powerful comparative design aimed for testing the influence of natural radioactivity on the mutation rate, we first prospected for closely related subterranean species living in contrasted radioactive set-ups. Using the map of bedrock uranium content in France (Ielsch et al., 2017), we prospected areas with low and high radioactivity. From this large survey (58 sites with waterlice), we selected 14 sites with contrasted levels of α radioactivity which were inhabited by closely related groundwater waterlice species. We paid special attention that α radioactivity differed at least by a factor of three between two habitats, each containing a closely related species (Figure 1, Supplementary file 1). On average low level of radioactivity was around 0.357 Bq/g of dry sediment and high level around 1.259 Bq/g of dry sediment. Based on transcriptome sequencing and de novo assembly, we used a phylogenetic approach to estimate nuclear and mitochondrial substitution rates (i.e. the rate of mutations which are fixed). While experimental approaches allow to measure the impact of radioactivity on somatic mutations or on the transmission of mutations across few generations, this phylogenetic approach allows us to measure the impact of natural radiation on the germinal mutation rate over the course of a species history.

Figure 1 with 1 supplement see all

Download asset Open asset

Using the 14 selected species and locations, we tested whether there was a significant positive relationship between natural radioactivity and the long-term mutation rate. The latter was estimated using the synonymous substitution rate ($d_{\mathrm{S}}$) calculated on the terminal branches of the phylogenetic tree tracing the history of these 14 species. The $d_{\mathrm{S}}$ is the rate at which silent mutations accumulate in protein coding genes and, when calculated using many different loci and in the absence of a strong synonymous codon usage bias (see methods), is an estimator of the average mutation rate across a species’ genome (Kimura, 1983). To be comparable across species, these $d_{\mathrm{S}}$ has to be divided by the time over which it is measured. To achieve that, we used the software CoEvol (Lartillot and Poujol, 2011) which models directly a synonymous substitution rate relative to the root age ($d_{\mathrm{S}}$/ra). We computed $d_{\mathrm{S}}$/ra using 769 one-to-one nuclear and 13 mitochondrial orthologous protein-coding genes shared by all 14 species. At each sampling site, we measured the global α radioactivity and the activity of all radio-elements in the sediment. The analysis of the composition in radionuclides at each site reveals that two sites (BRETEMIN and BOREON) show a disruption of the secular equilibrium in the U-238 chain. This suggests that nearby industrial activities (e.g. lead mines) have modified the natural radioactivity at these two sites. As these industrial activities are very recent (since 1950), their impact on the substitution rate, which is measured on a much longer time scale, is unlikely. We therefore did not use these two sites to test the correlation between d_S and any site-specific radioactivity measurement (however, see next paragraph for a regional measurement). The d_S/ra is positively correlated with the α radioactivity in the nuclear genome as well as in the mitochondrial genome (Table 1; Figure 2). A linear model predicts a d_S/ra increase of 31.8% in the nuclear genome and 56.5% in the mitochondrial genome between species living in low (on average 0.357 Bq/g of dry sediment) and high radioactivity (on average 1.259 Bq/g of dry sediment). We also modeled the biologically effective dose of radioactivity received by each species (Received dose in µGy/h, Figure 1). This measure takes into account the transfer coefficient from environment to biota and the radio-toxicity of each radio-element for a crustacean model (ERICA tool V1.2.1 Brown et al., 2016). Again, we found positive correlations between the d_S/ra and the received dose of radioactivity (Table 1). These results are robust to the presence of influential cases and to variation in the model of evolution used to perform the test (see Methods, pGLS, p. values < 0.05).

Figure 2

Download asset Open asset

As previously explained, the measured radioactivity at two sites overestimates the radioactivity level to which the organisms have been exposed for many generations because it is influenced by recent human activities. Moreover, while most species collected in highly radioactive habitats were from metamorphic or igneous formations, two species were from sedimentary formations (BOREON and GROTTAZE). Contrary to metamorphic and igneous formations, radioactivity in sedimentary formations is often observed in restricted localities (Ielsch et al., 2017) and can show large variations at the meter scale. A single radiation measurement may not therefore accurately represent the average radiation that a species was exposed to. To account for this variability as well as to include the two human-impacted sites into the regression analysis, we calculated the areal proportion of metamorphic and igneous rock within a 15 km radius around each site (later called λ15, Figure 1). This proportion was used as a proxy for the long-term regional radioactive exposure because the average linear distribution range of a groundwater crustacean is 30 km (Eme et al., 2018). We found a positive and stronger correlation between the d_S/ra and λ15 in both genomes (Table 1, Figure 2, n = 14 species). The linear model predicts that the nuclear and mitochondrial d_S/ra of a species living in a metamorphic formation (>50% of metamorphic and igneous rocks) are on average 34.4% and 61.3% higher, respectively, than those of a species living in a sedimentary formation (<50% of metamorphic or igneous rocks).

As bedrock radioactivity is positively correlated with the mutation rate, the underlying question is whether radioactivity also modifies the mutational spectrum, that is, the specific types of mutations that tend to occur. To address this question, we reconstructed the mutational spectrum of 6 independent pairs of species, each composed of two species located in low and high bedrock radiation set-ups, respectively, with a minimum of a 3X increase in the received dose of radioactivity between the two species (Figure 1). Briefly, we first estimated species polymorphism across a set of 2490 one-to-one orthologous genes by sequencing transcriptomes for eight individuals per species. After ancestral sequence reconstruction, we then identified mutations that occurred in each species and computed the relative proportion of each type of mutation (from A to T, A to C, …), pooling together complementary mutations (e.g. p(${\displaystyle C\to A}$) + p(${\displaystyle G\to T}$) = p(${\displaystyle C:G\to A:T}$)). Two types of dependencies are present in testing mutational spectrum variation in a comparative data-set: (i) dependency among mutations – if the proportion of one mutation increases, the proportions of the other mutations will decrease – and (ii) the phylogenetic inertia, two closely related species have more chance to display more similar mutational spectrum. To take into account these two types of dependencies, we first used a forward selection approach as described in Harris and Pritchard, 2017 to pull out mutations at different frequencies across habitats (see Materials and methods). Only the ${\displaystyle C:G\to A:T}$ mutation was significantly more frequent in radioactive habitats (Table 2, Figure 3). Second, we checked that these results were not induced by the phylogenetic structure of the data-set by using a pGLS regression between the proportion of each type of mutation and different radioactivity proxies. Again, we found a strong positive correlation between the proportion of ${\displaystyle C:G\to A:T}$ mutations and bedrock radioactivity (Table 2, Figure 3). Selection is unlikely to be responsible for the increase of ${\displaystyle C:G\to A:T}$ as this correlation is also observed when the data set is limited to mutations found at the third, usually redundant, codon position (Figure 3—figure supplement 1, Supplementary file 2). A weak negative trend is observed for ${\displaystyle A:T\to T:A}$ mutations; however, this trend disappears when measured on third codon position.

Figure 3 with 1 supplement see all

Download asset Open asset

Variations in the frequency of genetic variants among populations can originate from variation in the mutagenic environment or from biases in the fixation of mutations that can be due to demographic factors (Mathieson and Reich, 2017), natural selection (Boussau et al., 2008), or biased gene conversion (Duret and Galtier, 2009). While the impact of fixation biases on sequence evolution have been relatively well described, the long term impact of the mutagenic environment is less well-known. The most extensive demonstration of environmentally induced changes in the mutation spectrum comes from the study of carcinogens (Seo et al., 2000). Many carcinogens induce a specific mutational signature, for instance, UVs increase the ${\displaystyle CC\to TT}$ mutations whereas estrogen treatments increase the frequency of ${\displaystyle A:T\to G:C}$ mutations (Flibotte et al., 2010; Nik-Zainal et al., 2015). Here, we found an increase of the proportion of ${\displaystyle G\to T}$ mutations (and complementary ${\displaystyle C\to A}$) which is a mutation that increases in a variety of contexts and is not specific to a mutagen in particular. The occurrence of this mutation increases after exposure to carbon black (Jacobsen et al., 2011), to tobacco smoke (Hollstein et al., 1991; Hainaut and Pfeifer, 2001), or to polycyclic aromatic hydrocarbons (Nik-Zainal et al., 2015). Oxidative stress is the likely source of this mutation (Wood et al., 1990; Besaratinia et al., 2004; Jacobsen et al., 2011). Indeed, the characteristic damage linked to oxidative stress is the formation of 8-hydroxyguanine (Shigenaga et al., 1989) which leads to ${\displaystyle G\to T}$ mutations (Shibutani et al., 1991; Cheng et al., 1992). Thus, the mutational spectrum modification observed in radioactive environments is suggestive of an increase of the overall oxidative stress in these environments. Similarly, high artificial doses of ionizing radiation were found to increase oxidative damage (Einor et al., 2016; Haghdoost et al., 2006). As oxidative stress triggers more 8-hydroxyguanine formation in the mitochondrial genome than in the nuclear genome (Richter et al., 1988), this could explain the twice as strong effect observed in this study in the mitochondrial compartment.

Radioactive environments can cause mutations in two intertwined ways. First, the ionisation of molecules in the cells can directly affect the DNA structure by breaking the sugar phosphate backbone or can affect DNA indirectly through the radiolysis of water which decomposes the H₂O molecules and create free radicals (Desouky et al., 2015). These free radicals can damage DNA molecules and create mutations. Second, radioactive decay chains also generate heavy metals (lead, polonium, etc) which are toxic for cells and also cause oxidative stress (Quinlan et al., 1988; Pinto et al., 2003). Due to the physicochemical association between radioactivity and heavy metals, in naturae correlative approaches alone cannot discriminate between the toxicity of heavy metals and the direct toxicity of radioactive rays.

While metamorphic and igneous rocks are more naturally radioactive than sedimentary rocks (Ielsch et al., 2017), they also display characteristic compositional and structural features which may also control species life history traits (Cornu et al., 2013; Eme et al., 2015). Here, we consider whether these characteristic features – specifically their poor calcium content and reduced habitat size and permeability – may confound the effect of radioactivity on the mutation rate. Metamorphic and igneous rocks are calcium poor relative to sedimentary rocks. In isopods, the cuticle is made of calcium carbonate which is acquired by isopods from their environments (Greenaway, 1985). A lower calcium availability may slow down the growth of isopod species. However, a lower growth rate, if it slows down the generation time, is expected to decrease the mutation rate, while we observed the opposite. Because metamorphic and igneous rocks have smaller size and less permeable habitats, they could support smaller effective isopod population sizes. The d_S is an estimator of the mutation rate that is not directly impacted by the effective population size (Kimura, 1983). However, as effective population size modulates the efficacy of natural selection, Lynch, 2010 proposed that species with small effective population size should evolve higher mutation rate as a result of the accumulation of weakly deleterious mutations in genes involved in repair efficiency and replication fidelity. Efficacy of selection can be estimated by computing a transcriptome-wide ratio d_N/d_S. Here, we found no correlation between radioactivity and d_N/d_S (pGLS, p.value = 0.3069) suggesting this effect is not at work in this data-set. Altogether, the influence of confounding factors that would drive the observed mutational variations is unlikely.

If oxidative stress is causing the increase of mutation rate, the radiolysis of water alone hardly explains the much higher impact of radioactivity on the mitochondrial mutation rate compared to the nuclear rate. As there is no reason to argue that radiolysis would not evenly occur within cells, it should impact both genomes similarly. However, differences between the two genomes may explain why the mitochondrial genome is more sensitive to radioactivity. First, the mitochondrial genome lacks some repair systems. For example, in a different yet analogous context, UV damage accumulates in the mitochondrial genome while they are repaired in the nuclear genome (Clayton et al., 1974). Second, the two genomes have very different organizations: while the nuclear genome is compacted into chromatin, the mitochondrial one is organized into nucleoids (Chen and Butow, 2005). While the role of the mitochondrial nucleoids is unclear, the chromatin structure and histone proteins protect the nuclear genome against radiation-induced strand breaks (Ljungman, 1991) and oxidative damage (Ljungman and Hanawalt, 1992). Third, direct radioactivity damage to the mitochondria may increase the activity of the mitochondrial respiratory chain and indirectly increase the production of ROS and DNA damage (Yamamori et al., 2012; Kam and Banati, 2013).

While life history traits are known to be central in controlling the mutation rate in most taxa (Nabholz et al., 2008; Bromham et al., 1996; Nikolaev et al., 2007; Smith and Donoghue, 2008), our results suggest that natural variation of radioactivity can have a comparable effect. Indeed, we found a minimum increase of around 30% percent of the nuclear mutation rate (60% in mitochondria) for species of waterlice living in the more naturally radioactive habitats made of igneous and metamorphic rocks. This increase is of the same magnitude as that observed when waterlice species evolve a 5-fold increase in generation time, a key life history trait controlling mutation rate in waterlice (Saclier et al., 2018) and organisms in general (Thomas et al., 2010; Weller and Wu, 2015; Nabholz et al., 2008). However, the influence of life history traits on the mutation rate varies widely among groups (Allio et al., 2017). Thus, the relative influence of radioactivity compared to life history traits could be different in groups like vertebrates, potentialy because of different gametogenesis (Saclier et al., 2018). Moreover, as groundwater waterlice ingest sediments (Francois et al., 2016), they are internally exposed to radioactivity, which may cause more mutations than through external exposure only (Sawada, 2007). The influence of environmental radioactivity on the mutation rate should therefore be explored across a wider range of organisms with contrasted diets and biologies.

Although the literature on the effect of low doses of radiation is far from being conclusive it often suggests a negligible biological impact (Tubiana et al., 2006; Tubiana et al., 2009) and only a handful of isolated studies support an impact of natural radioactivity on the mutation rate. For instance, a higher mutation rate was observed in the human mitochondrial genome in the Kerala region (Forster et al., 2002) and in satellite sequences of crickets inhabiting cave with high radon concentration (Allegrucci et al., 2015). In this study, by combining a large number of genes with the characteristics of the subterranean waterlice, namely the absence of UV confounding effect and limited dispersal, within a statistically powerful comparative framework allowing to work on large time scales and with numerous replicates, we found that a mild variation (≃ 3.5X) in natural bedrock radioactivity substantially alters the mutation rate, in particular the mitochondrial one. One key aspect that remains to be described is the shape of the relationship between the dose of radioactivity and the mutation rate: this study invalidates a model where low doses have no impact but falls short in differentiating between, for example, a linear and a hypersentivity (U-shaped) dependency model. Altogether, while the universality of this finding warrants corroborative studies in other taxa, it suggests that the influence of natural radioactivity on the evolution of biodiversity has been overlooked.

16S sequences have been deposited on the European Nucleotide Archive and are available under the accession numbers from LR214526 to LR214880 (https://www.ebi.ac.uk/ena/data/view/LR214526-LR214880). Alignments and the list of genes used to compute synonymous substitution rate have been deposited on Zenodo (https://zenodo.org/deposit/2563829). Transcriptome reads have been deposited on the European Nucleotide Archive and are available under accession numbers from LR536601 to LR536626 in the study ID PRJEB14193 (https://www.ebi.ac.uk/ena/data/search?query=PRJEB14193). Number of reads and data used for correlations, namely measures of radionuclides and mutations counts have been deposited on Zenodo (https://doi.org/10.5281/zenodo.4071754).

1. 1. Saclier N
   2. Chardon P
   3. Malard F
   4. Konecny-Dupré L
   5. Eme D
   6. Bellec A
   7. Breton V
   8. Duret L
   9. Lefébure T
   10. Douady CJ

   (2020) Zenodo

   Data used in the article.

   https://doi.org/10.5281/zenodo.4071754
2. 1. Saclier N
   2. Chardon P
   3. Malard F
   4. Konecny-Dupré L
   5. Eme D
   6. Bellec A
   7. Breton V
   8. Duret L
   9. Lefébure T
   10. Douady CJ

   (2020) Zenodo

   Bedrock radioactivity influences the rate and spectrum of mutation - Orthologous genes.

   https://doi.org/10.5281/zenodo.2563829
3. 1. Saclier N
   2. Chardon P
   3. Malard F
   4. Konecny-Dupré L
   5. Eme D
   6. Bellec A
   7. Breton V
   8. Duret L
   9. Lefébure T
   10. Douady CJ

   (2019) ENA

   ID PRJEB30668. Aselloidea isopods Sanger sequencing.

   https://www.ebi.ac.uk/ena/browser/view/LR214526-LR214880
4. 1. Saclier N
   2. Chardon P
   3. Malard F
   4. Konecny-Dupré L
   5. Eme D
   6. Bellec A
   7. Breton V
   8. Duret L
   9. Lefébure T
   10. Douady CJ

   (2019) ENA

   ID PRJEB14193. Aselloidea isopods transcriptomes sequencing and denovo assembly.

   https://www.ebi.ac.uk/ena/data/search?query=PRJEB14193
5. 1. Lefébure T
   2. Saclier N

   (2019) EBI European Nucleotide Archive

   ID LR214526-LR214880. 16S sequences.

   https://www.ebi.ac.uk/ena/data/view/LR214526-LR214880
6. 1. Lefébure T
   2. Saclier N

   (2019) EBI European Nucleotide Archive

   ID LR536601-LR536626. Transcriptome reads.

   https://www.ebi.ac.uk/ena/data/view/LR536601-LR536626

1. 1. Adewoye AB
   2. Lindsay SJ
   3. Dubrova YE
   4. Hurles ME

   (2015) The genome-wide effects of ionizing radiation on mutation induction in the mammalian germline

   Nature Communications 6:6684.

   https://doi.org/10.1038/ncomms7684

   • PubMed
   • Google Scholar
2. 1. Allegrucci G
   2. Sbordoni V
   3. Cesaroni D

   (2015) Is radon emission in caves causing deletions in satellite DNA sequences of cave-dwelling crickets?

   PLOS ONE 10:e0122456.

   https://doi.org/10.1371/journal.pone.0122456

   • PubMed
   • Google Scholar
3. 1. Allio R
   2. Donega S
   3. Galtier N
   4. Nabholz B

   (2017) Large variation in the ratio of mitochondrial to nuclear mutation rate across animals: implications for genetic diversity and the use of mitochondrial DNA as a molecular marker

   Molecular Biology and Evolution 34:2762–2772.

   https://doi.org/10.1093/molbev/msx197

   • PubMed
   • Google Scholar
4. 1. Altschul SF
   2. Gish W
   3. Miller W
   4. Myers EW
   5. Lipman DJ

   (1990) Basic local alignment search tool

   Journal of Molecular Biology 215:403–410.

   https://doi.org/10.1016/S0022-2836(05)80360-2

   • PubMed
   • Google Scholar
5. 1. Aronesty E

   (2013) Comparison of sequencing utility programs

   The Open Bioinformatics Journal 7:1–8.

   https://doi.org/10.2174/1875036201307010001

   • Google Scholar
6. 1. Barber R
   2. Plumb MA
   3. Boulton E
   4. Roux I
   5. Dubrova YE

   (2002) Elevated mutation rates in the germ line of first- and second-generation offspring of irradiated male mice

   PNAS 99:6877–6882.

   https://doi.org/10.1073/pnas.102015399

   • PubMed
   • Google Scholar
7. 1. Barja G

   (2002) Endogenous oxidative stress: relationship to aging, longevity and caloric restriction

   Ageing Research Reviews 1:397–411.

   https://doi.org/10.1016/S1568-1637(02)00008-9

   • PubMed
   • Google Scholar
8. Book

   1. Beir VII

   (2006) Research Council of the National Academies of USA. Health Risks From Exposure to Low Levels of Ionizing Radiation: Beir VII Phase 2

   National Academies Press.

   https://doi.org/10.17226/11340

   • Google Scholar
9. 1. Bernt M
   2. Donath A
   3. Jühling F
   4. Externbrink F
   5. Florentz C
   6. Fritzsch G
   7. Pütz J
   8. Middendorf M
   9. Stadler PF

   (2013) MITOS: improved de novo metazoan mitochondrial genome annotation

   Molecular Phylogenetics and Evolution 69:313–319.

   https://doi.org/10.1016/j.ympev.2012.08.023

   • PubMed
   • Google Scholar
10. 1. Besaratinia A
    2. Synold TW
    3. Xi B
    4. Pfeifer GP

    (2004) G-to-T transversions and small tandem base deletions are the hallmark of mutations induced by ultraviolet a radiation in mammalian cells

    Biochemistry 43:8169–8177.

    https://doi.org/10.1021/bi049761v

    • PubMed
    • Google Scholar
11. 1. Bou C
    2. Rouch R

    (1967)

    Un nouveau champ de recherches sur la faune aquatique souterraine

    Comptes Rendus de L'Academie des Sciences 265:369–370.

    • Google Scholar
12. 1. Boucher D
    2. Hindo J
    3. Averbeck D

    (2004) Increased repair of gamma-induced DNA double-strand breaks at lower dose-rate in CHO cells

    Canadian Journal of Physiology and Pharmacology 82:125–132.

    https://doi.org/10.1139/y04-006

    • PubMed
    • Google Scholar
13. 1. Boussau B
    2. Blanquart S
    3. Necsulea A
    4. Lartillot N
    5. Gouy M

    (2008) Parallel adaptations to high temperatures in the Archaean eon

    Nature 456:942–945.

    https://doi.org/10.1038/nature07393

    • PubMed
    • Google Scholar
14. 1. Bromham L
    2. Rambaut A
    3. Harvey PH

    (1996) Determinants of rate variation in mammalian DNA sequence evolution

    Journal of Molecular Evolution 43:610–621.

    https://doi.org/10.1007/BF02202109

    • PubMed
    • Google Scholar
15. 1. Brown JE
    2. Alfonso B
    3. Avila R
    4. Beresford NA
    5. Copplestone D
    6. Hosseini A

    (2016) A new version of the ERICA tool to facilitate impact assessments of radioactivity on wild plants and animals

    Journal of Environmental Radioactivity 153:141–148.

    https://doi.org/10.1016/j.jenvrad.2015.12.011

    • PubMed
    • Google Scholar
16. 1. Calvignac S
    2. Konecny L
    3. Malard F
    4. Douady CJ

    (2011) Preventing the pollution of mitochondrial datasets with nuclear mitochondrial paralogs (numts)

    Mitochondrion 11:246–254.

    https://doi.org/10.1016/j.mito.2010.10.004

    • PubMed
    • Google Scholar
17. 1. Castresana J

    (2000) Selection of conserved blocks from multiple alignments for their use in phylogenetic analysis

    Molecular Biology and Evolution 17:540–552.

    https://doi.org/10.1093/oxfordjournals.molbev.a026334

    • PubMed
    • Google Scholar
18. 1. Chen XJ
    2. Butow RA

    (2005) The organization and inheritance of the mitochondrial genome

    Nature Reviews Genetics 6:815–825.

    https://doi.org/10.1038/nrg1708

    • PubMed
    • Google Scholar
19. 1. Cheng KC
    2. Cahill DS
    3. Kasai H
    4. Nishimura S
    5. Loeb LA

    (1992)

    8-Hydroxyguanine, an abundant form of oxidative DNA damage, causes G----T and A----C substitutions

    The Journal of Biological Chemistry 267:166–172.

    • PubMed
    • Google Scholar
20. 1. Clayton DA
    2. Doda JN
    3. Friedberg EC

    (1974) The absence of a pyrimidine dimer repair mechanism in mammalian mitochondria

    PNAS 71:2777–2781.

    https://doi.org/10.1073/pnas.71.7.2777

    • PubMed
    • Google Scholar
21. 1. Collis SJ
    2. Schwaninger JM
    3. Ntambi AJ
    4. Keller TW
    5. Nelson WG
    6. Dillehay LE
    7. Deweese TL

    (2004) Evasion of early cellular response mechanisms following low level radiation-induced DNA damage

    Journal of Biological Chemistry 279:49624–49632.

    https://doi.org/10.1074/jbc.M409600200

    • PubMed
    • Google Scholar
22. 1. Cornu J-F
    2. Eme D
    3. Malard F

    (2013) The distribution of groundwater habitats in Europe

    Hydrogeology Journal 21:949–960.

    https://doi.org/10.1007/s10040-013-0984-1

    • Google Scholar
23. 1. Czeizel AE
    2. Elek C
    3. Susánszky E

    (1991) The evaluation of the germinal mutagenic impact of chernobyl radiological contamination in Hungary

    Mutagenesis 6:285–288.

    https://doi.org/10.1093/mutage/6.4.285

    • PubMed
    • Google Scholar
24. 1. Desouky O
    2. Ding N
    3. Zhou G

    (2015) Targeted and non-targeted effects of ionizing radiation

    Journal of Radiation Research and Applied Sciences 8:247–254.

    https://doi.org/10.1016/j.jrras.2015.03.003

    • Google Scholar
25. 1. Dikomey E
    2. Brammer I
    3. Johansen J
    4. Bentzen SM
    5. Overgaard J

    (2000) Relationship between DNA double-strand breaks, cell killing, and fibrosis studied in confluent skin fibroblasts derived from breast Cancer patients

    International Journal of Radiation Oncology*Biology*Physics 46:481–490.

    https://doi.org/10.1016/S0360-3016(99)00335-1

    • PubMed
    • Google Scholar
26. 1. Dubrova YE
    2. Nesterov VN
    3. Krouchinsky NG
    4. Ostapenko VA
    5. Neumann R
    6. Neil DL
    7. Jeffreys AJ

    (1996) Human minisatellite mutation rate after the chernobyl accident

    Nature 380:683–686.

    https://doi.org/10.1038/380683a0

    • PubMed
    • Google Scholar
27. 1. Duret L
    2. Galtier N

    (2009) Biased gene conversion and the evolution of mammalian genomic landscapes

    Annual Review of Genomics and Human Genetics 10:285–311.

    https://doi.org/10.1146/annurev-genom-082908-150001

    • PubMed
    • Google Scholar
28. 1. Dutheil J
    2. Boussau B

    (2008) Non-homogeneous models of sequence evolution in the bio++ suite of libraries and programs

    BMC Evolutionary Biology 8:255.

    https://doi.org/10.1186/1471-2148-8-255

    • PubMed
    • Google Scholar
29. 1. Einor D
    2. Bonisoli-Alquati A
    3. Costantini D
    4. Mousseau TA
    5. Møller AP

    (2016) Ionizing radiation, antioxidant response and oxidative damage: a meta-analysis

    Science of the Total Environment 548-549:463–471.

    https://doi.org/10.1016/j.scitotenv.2016.01.027

    • Google Scholar
30. 1. Eme D
    2. Zagmajster M
    3. Fišer C
    4. Galassi D
    5. Marmonier P
    6. Stoch F
    7. Cornu J-F
    8. Oberdorff T
    9. Malard F

    (2015) Multi-causality and spatial non-stationarity in the determinants of groundwater crustacean diversity in Europe

    Ecography 38:531–540.

    https://doi.org/10.1111/ecog.01092

    • Google Scholar
31. 1. Eme D
    2. Zagmajster M
    3. Delić T
    4. Fišer C
    5. Flot J-F
    6. Konecny-Dupré L
    7. Pálsson S
    8. Stoch F
    9. Zakšek V
    10. Douady CJ
    11. Malard F

    (2018) Do cryptic species matter in Macroecology? sequencing european groundwater crustaceans yields smaller ranges but does not challenge biodiversity determinants

    Ecography 41:424–436.

    https://doi.org/10.1111/ecog.02683

    • Google Scholar
32. Book

    1. Felsenstein J

    (2004)

    Inferring Phylogenies

    Sinauer Associates, Inc.

    • Google Scholar
33. 1. Flibotte S
    2. Edgley ML
    3. Chaudhry I
    4. Taylor J
    5. Neil SE
    6. Rogula A
    7. Zapf R
    8. Hirst M
    9. Butterfield Y
    10. Jones SJ
    11. Marra MA
    12. Barstead RJ
    13. Moerman DG

    (2010) Whole-genome profiling of mutagenesis in Caenorhabditis elegans

    Genetics 185:431–441.

    https://doi.org/10.1534/genetics.110.116616

    • PubMed
    • Google Scholar
34. 1. Forster L
    2. Forster P
    3. Lutz-Bonengel S
    4. Willkomm H
    5. Brinkmann B

    (2002) Natural radioactivity and human mitochondrial DNA mutations

    PNAS 99:13950–13954.

    https://doi.org/10.1073/pnas.202400499

    • PubMed
    • Google Scholar
35. 1. Francois CM
    2. Mermillod‐Blondin F
    3. Malard F
    4. Fourel F
    5. Lécuyer C
    6. Douady CJ
    7. Simon L

    (2016) Trophic ecology of groundwater species reveals specialization in a low‐productivity environment

    Functional Ecology 30:262–273.

    https://doi.org/10.1111/1365-2435.12484

    • Google Scholar
36. 1. Gayral P
    2. Melo-Ferreira J
    3. Glémin S
    4. Bierne N
    5. Carneiro M
    6. Nabholz B
    7. Lourenco JM
    8. Alves PC
    9. Ballenghien M
    10. Faivre N
    11. Belkhir K
    12. Cahais V
    13. Loire E
    14. Bernard A
    15. Galtier N

    (2013) Reference-free population genomics from next-generation transcriptome data and the vertebrate-invertebrate gap

    PLOS Genetics 9:e1003457.

    https://doi.org/10.1371/journal.pgen.1003457

    • PubMed
    • Google Scholar
37. 1. Grabherr MG
    2. Haas BJ
    3. Yassour M
    4. Levin JZ
    5. Thompson DA
    6. Amit I
    7. Adiconis X
    8. Fan L
    9. Raychowdhury R
    10. Zeng Q
    11. Chen Z
    12. Mauceli E
    13. Hacohen N
    14. Gnirke A
    15. Rhind N
    16. di Palma F
    17. Birren BW
    18. Nusbaum C
    19. Lindblad-Toh K
    20. Friedman N
    21. Regev A

    (2011) Full-length transcriptome assembly from RNA-Seq data without a reference genome

    Nature Biotechnology 29:644–652.

    https://doi.org/10.1038/nbt.1883

    • PubMed
    • Google Scholar
38. 1. Greenaway P

    (1985) Calcium balance and moulting in the crustacea

    Biological Reviews 60:425–454.

    https://doi.org/10.1111/j.1469-185X.1985.tb00424.x

    • Google Scholar
39. 1. Guéguen L
    2. Duret L

    (2018) Unbiased estimate of synonymous and nonsynonymous substitution rates with nonstationary base composition

    Molecular Biology and Evolution 35:734–742.

    https://doi.org/10.1093/molbev/msx308

    • PubMed
    • Google Scholar
40. 1. Guindon S
    2. Dufayard JF
    3. Lefort V
    4. Anisimova M
    5. Hordijk W
    6. Gascuel O

    (2010) New algorithms and methods to estimate maximum-likelihood phylogenies: assessing the performance of PhyML 3.0

    Systematic Biology 59:307–321.

    https://doi.org/10.1093/sysbio/syq010

    • PubMed
    • Google Scholar
41. 1. Haghdoost S
    2. Sjölander L
    3. Czene S
    4. Harms-Ringdahl M

    (2006) The nucleotide pool is a significant target for oxidative stress

    Free Radical Biology and Medicine 41:620–626.

    https://doi.org/10.1016/j.freeradbiomed.2006.05.003

    • PubMed
    • Google Scholar
42. 1. Hahn C
    2. Bachmann L
    3. Chevreux B

    (2013) Reconstructing mitochondrial genomes directly from genomic next-generation sequencing reads--a baiting and iterative mapping approach

    Nucleic Acids Research 41:e129.

    https://doi.org/10.1093/nar/gkt371

    • PubMed
    • Google Scholar
43. 1. Hainaut P
    2. Pfeifer GP

    (2001) Patterns of p53 G-->T transversions in lung cancers reflect the primary mutagenic signature of DNA-damage by tobacco smoke

    Carcinogenesis 22:367–374.

    https://doi.org/10.1093/carcin/22.3.367

    • PubMed
    • Google Scholar
44. 1. Hande MP
    2. Azizova TV
    3. Burak LE
    4. Khokhryakov VF
    5. Geard CR
    6. Brenner DJ

    (2005) Complex chromosome aberrations persist in individuals many years after occupational exposure to densely ionizing radiation: an mFISH study

    Genes, Chromosomes and Cancer 44:1–9.

    https://doi.org/10.1002/gcc.20217

    • Google Scholar
45. 1. Harris K
    2. Pritchard JK

    (2017) Rapid evolution of the human mutation spectrum

    eLife 6:e24284.

    https://doi.org/10.7554/eLife.24284

    • PubMed
    • Google Scholar
46. 1. Hoeijmakers JH

    (2001) Genome maintenance mechanisms for preventing Cancer

    Nature 411:366–374.

    https://doi.org/10.1038/35077232

    • PubMed
    • Google Scholar
47. 1. Hollstein M
    2. Sidransky D
    3. Vogelstein B
    4. Harris CC

    (1991) p53 mutations in human cancers

    Science 253:49–53.

    https://doi.org/10.1126/science.1905840

    • PubMed
    • Google Scholar
48. 1. Hooker AM
    2. Bhat M
    3. Day TK
    4. Lane JM
    5. Swinburne SJ
    6. Morley AA
    7. Sykes PJ

    (2004) The linear no-threshold model does not hold for low-dose ionizing radiation

    Radiation Research 162:447–452.

    https://doi.org/10.1667/RR3228

    • PubMed
    • Google Scholar
49. 1. Huo L
    2. Nagasawa H
    3. Little JB

    (2001) HPRT mutants induced in bystander cells by very low fluences of alpha particles result primarily from point mutations

    Radiation Research 156:521–525.

    https://doi.org/10.1667/0033-7587(2001)156[0521:HMIIBC]2.0.CO;2

    • PubMed
    • Google Scholar
50. 1. Ielsch G
    2. Cuney M
    3. Buscail F
    4. Rossi F
    5. Leon A
    6. Cushing ME

    (2017) Estimation and mapping of uranium content of geological units in France

    Journal of Environmental Radioactivity 166:210–219.

    https://doi.org/10.1016/j.jenvrad.2016.05.022

    • PubMed
    • Google Scholar
51. 1. Jacobsen NR
    2. White PA
    3. Gingerich J
    4. Møller P
    5. Saber AT
    6. Douglas GR
    7. Vogel U
    8. Wallin H

    (2011) Mutation spectrum in FE1-MUTA(TM) Mouse lung epithelial cells exposed to nanoparticulate carbon black

    Environmental and Molecular Mutagenesis 52:331–337.

    https://doi.org/10.1002/em.20629

    • PubMed
    • Google Scholar
52. 1. Jarrett SG
    2. Boulton ME

    (2005) Antioxidant up-regulation and increased nuclear DNA protection play key roles in adaptation to oxidative stress in epithelial cells

    Free Radical Biology and Medicine 38:1382–1391.

    https://doi.org/10.1016/j.freeradbiomed.2005.02.003

    • PubMed
    • Google Scholar
53. 1. Jostes RF
    2. Fleck EW
    3. Morgan TL
    4. Stiegler GL
    5. Cross FT

    (1994) Southern blot and polymerase chain reaction exon analyses of HPRT- mutations induced by radon and radon progeny

    Radiation Research 137:371–379.

    https://doi.org/10.2307/3578712

    • PubMed
    • Google Scholar
54. 1. Kam WW
    2. Banati RB

    (2013) Effects of ionizing radiation on mitochondria

    Free Radical Biology & Medicine 65:607–619.

    https://doi.org/10.1016/j.freeradbiomed.2013.07.024

    • PubMed
    • Google Scholar
55. Book

    1. Karam PA
    2. Leslie SA

    (2005) Changes in terrestrial natural radiation levels over the history of life

    In: Karam P. A, editors. Radioactivity in the Environment. Elsevier. pp. 107–117.

    https://doi.org/10.1016/S1569-4860(04)07011-1

    • Google Scholar
56. 1. Karran P

    (2000) DNA double strand break repair in mammalian cells

    Current Opinion in Genetics & Development 10:144–150.

    https://doi.org/10.1016/S0959-437X(00)00069-1

    • PubMed
    • Google Scholar
57. Book

    1. Kimura M

    (1983) The Neutral Theory of Molecular Evolution

    Cambridge University Press.

    https://doi.org/10.1002/ajpa.1330650413

    • Google Scholar
58. 1. Krueger SA
    2. Joiner MC
    3. Weinfeld M
    4. Piasentin E
    5. Marples B

    (2007) Role of apoptosis in low-dose hyper-radiosensitivity

    Radiation Research 167:260–267.

    https://doi.org/10.1667/RR0776.1

    • PubMed
    • Google Scholar
59. 1. Lartillot N
    2. Poujol R

    (2011) A phylogenetic model for investigating correlated evolution of substitution rates and continuous phenotypic characters

    Molecular Biology and Evolution 28:729–744.

    https://doi.org/10.1093/molbev/msq244

    • PubMed
    • Google Scholar
60. 1. Li H
    2. Handsaker B
    3. Wysoker A
    4. Fennell T
    5. Ruan J
    6. Homer N
    7. Marth G
    8. Abecasis G
    9. Durbin R
    10. 1000 Genome Project Data Processing Subgroup

    (2009) The sequence alignment/Map format and SAMtools

    Bioinformatics 25:2078–2079.

    https://doi.org/10.1093/bioinformatics/btp352

    • PubMed
    • Google Scholar
61. 1. Li H
    2. Durbin R

    (2009) Fast and accurate short read alignment with Burrows-Wheeler transform

    Bioinformatics 25:1754–1760.

    https://doi.org/10.1093/bioinformatics/btp324

    • PubMed
    • Google Scholar
62. 1. Ljungman M

    (1991) The influence of chromatin structure on the frequency of radiation-induced DNA strand breaks: a study using nuclear and nucleoid monolayers

    Radiation Research 126:58–64.

    https://doi.org/10.2307/3578171

    • PubMed
    • Google Scholar
63. 1. Ljungman M
    2. Hanawalt PC

    (1992) Efficient protection against oxidative DNA damage in chromatin

    Molecular Carcinogenesis 5:264–269.

    https://doi.org/10.1002/mc.2940050406

    • PubMed
    • Google Scholar
64. 1. Loucas BD
    2. Eberle R
    3. Bailey SM
    4. Cornforth MN

    (2004) Influence of dose rate on the induction of simple and complex chromosome exchanges by gamma rays

    Radiation Research 162:339–349.

    https://doi.org/10.1667/RR3245

    • PubMed
    • Google Scholar
65. 1. Löytynoja A
    2. Goldman N

    (2008) Phylogeny-aware gap placement prevents errors in sequence alignment and evolutionary analysis

    Science 320:1632–1635.

    https://doi.org/10.1126/science.1158395

    • PubMed
    • Google Scholar
66. 1. Lynch M

    (2010) Evolution of the mutation rate

    Trends in Genetics 26:345–352.

    https://doi.org/10.1016/j.tig.2010.05.003

    • PubMed
    • Google Scholar
67. 1. Martins EP
    2. Hansen TF

    (1997) Phylogenies and the comparative method: a general approach to incorporating phylogenetic information into the analysis of interspecific data

    The American Naturalist 149:646–667.

    https://doi.org/10.1086/286013

    • Google Scholar
68. 1. Masoomi JR
    2. Mohammadi S
    3. Amini M
    4. Ghiassi-Nejad M

    (2006) High background radiation Areas of ramsar in Iran: evaluation of DNA damage by alkaline single cell gel electrophoresis (SCGE)

    Journal of Environmental Radioactivity 86:176–186.

    https://doi.org/10.1016/j.jenvrad.2005.08.005

    • PubMed
    • Google Scholar
69. 1. Mathieson I
    2. Reich D

    (2017) Differences in the rare variant spectrum among human populations

    PLOS Genetics 13:e1006581.

    https://doi.org/10.1371/journal.pgen.1006581

    • PubMed
    • Google Scholar
70. 1. Miele V
    2. Penel S
    3. Duret L

    (2011) Ultra-fast sequence clustering from similarity networks with SiLiX

    BMC Bioinformatics 12:116.

    https://doi.org/10.1186/1471-2105-12-116

    • PubMed
    • Google Scholar
71. 1. Møller AP
    2. Mousseau TA

    (2013) The effects of natural variation in background radioactivity on humans, animals and other organisms

    Biological Reviews 88:226–254.

    https://doi.org/10.1111/j.1469-185X.2012.00249.x

    • PubMed
    • Google Scholar
72. 1. Morvan C
    2. Malard F
    3. Paradis E
    4. Lefébure T
    5. Konecny-Dupré L
    6. Douady CJ

    (2013) Timetree of aselloidea reveals species diversification dynamics in groundwater

    Systematic Biology 62:512–522.

    https://doi.org/10.1093/sysbio/syt015

    • PubMed
    • Google Scholar
73. 1. Muse SV
    2. Gaut BS

    (1994) A likelihood approach for comparing synonymous and nonsynonymous nucleotide substitution rates, with application to the chloroplast genome

    Molecular Biology and Evolution 11:715–724.

    https://doi.org/10.1093/oxfordjournals.molbev.a040152

    • PubMed
    • Google Scholar
74. 1. Nabholz B
    2. Glémin S
    3. Galtier N

    (2008) Strong variations of mitochondrial mutation rate across mammals--the longevity hypothesis

    Molecular Biology and Evolution 25:120–130.

    https://doi.org/10.1093/molbev/msm248

    • PubMed
    • Google Scholar
75. 1. Nik-Zainal S
    2. Kucab JE
    3. Morganella S
    4. Glodzik D
    5. Alexandrov LB
    6. Arlt VM
    7. Weninger A
    8. Hollstein M
    9. Stratton MR
    10. Phillips DH

    (2015) The genome as a record of environmental exposure

    Mutagenesis 30:763–770.

    https://doi.org/10.1093/mutage/gev073

    • PubMed
    • Google Scholar
76. 1. Nikolaev SI
    2. Montoya-Burgos JI
    3. Popadin K
    4. Parand L
    5. Margulies EH
    6. Antonarakis SE
    7. National Institutes of Health Intramural Sequencing Center Comparative Sequencing Program

    (2007) Life-history traits drive the evolutionary rates of mammalian coding and noncoding genomic elements

    PNAS 104:20443–20448.

    https://doi.org/10.1073/pnas.0705658104

    • PubMed
    • Google Scholar
77. 1. Noutahi E
    2. Semeria M
    3. Lafond M
    4. Seguin J
    5. Boussau B
    6. Guéguen L
    7. El-Mabrouk N
    8. Tannier E

    (2016) Efficient gene tree correction guided by genome evolution

    PLOS ONE 11:e0159559.

    https://doi.org/10.1371/journal.pone.0159559

    • PubMed
    • Google Scholar
78. 1. Paradis E
    2. Claude J
    3. Strimmer K

    (2004) APE: analyses of phylogenetics and evolution in R language

    Bioinformatics 20:289–290.

    https://doi.org/10.1093/bioinformatics/btg412

    • PubMed
    • Google Scholar
79. 1. Pinheiro J
    2. Bates D
    3. DebRoy S
    4. Sarkar D
    5. R Core Team

    (2007) Linear and nonlinear mixed effects models

    R Package Version 3:1–89.

    https://doi.org/10.4148/2475-7772.1273

    • Google Scholar
80. 1. Pinto E
    2. Sigaud-kutner TCS
    3. Leitao MAS
    4. Okamoto OK
    5. Morse D
    6. Colepicolo P

    (2003) Heavy metal-induced oxidative stress in algae1

    Journal of Phycology 39:1008–1018.

    https://doi.org/10.1111/j.0022-3646.2003.02-193.x

    • Google Scholar
81. 1. Quinlan GJ
    2. Halliwell B
    3. Moorhouse CP
    4. Gutteridge JM

    (1988) Action of lead(II) and aluminium (III) ions on iron-stimulated lipid peroxidation in liposomes, erythrocytes and rat liver microsomal fractions

    Biochimica Et Biophysica Acta (BBA) - Lipids and Lipid Metabolism 962:196–200.

    https://doi.org/10.1016/0005-2760(88)90159-2

    • PubMed
    • Google Scholar
82. 1. Quinn G
    2. Keough M

    (2002) Multiple testing

    Experimental Design and Data Analysis for Biologists 1:48–50.

    https://doi.org/10.1017/cbo9780511806384

    • Google Scholar
83. Software

    1. R Development Core Team

    (2020) R: A Language and Environment for Statistical Computing

    R Foundation for Statistical Computing, Vienna, Austria.

    https://www.R-project.org/
84. 1. Richter C
    2. Park JW
    3. Ames BN

    (1988) Normal oxidative damage to mitochondrial and nuclear DNA is extensive

    PNAS 85:6465–6467.

    https://doi.org/10.1073/pnas.85.17.6465

    • PubMed
    • Google Scholar
85. 1. Rigaud O
    2. Laquerbe A
    3. Moustacchi E

    (1995) DNA sequence analysis of HPRT- mutants induced in human lymphoblastoid cells adapted to ionizing radiation

    Radiation Research 144:181–189.

    https://doi.org/10.2307/3579257

    • PubMed
    • Google Scholar
86. 1. Rigaud O
    2. Moustacchi E

    (1996) Radioadaptation for gene mutation and the possible molecular mechanisms of the adaptive response

    Mutation Research/Fundamental and Molecular Mechanisms of Mutagenesis 358:127–134.

    https://doi.org/10.1016/S0027-5107(96)00113-3

    • PubMed
    • Google Scholar
87. 1. Rothkamm K
    2. Kühne M
    3. Jeggo PA
    4. Löbrich M

    (2001)

    Radiation-Induced genomic rearrangements formed by nonhomologous End-Joining of DNA Double-Strand breaks

    Cancer Research 61:3886–3893.

    • PubMed
    • Google Scholar
88. 1. Saclier N
    2. François CM
    3. Konecny-Dupré L
    4. Lartillot N
    5. Guéguen L
    6. Duret L
    7. Malard F
    8. Douady CJ
    9. Lefébure T

    (2018) Life history traits impact the nuclear rate of substitution but not the mitochondrial rate in isopods

    Molecular Biology and Evolution 35:2900–2912.

    https://doi.org/10.1093/molbev/msy184

    • PubMed
    • Google Scholar
89. 1. Saghirzadeh M
    2. Gharaati MR
    3. Mohammadi S
    4. Ghiassi-Nejad M

    (2008) Evaluation of DNA damage in the root cells of Allium cepa seeds growing in soil of high background radiation Areas of Ramsar-Iran

    Journal of Environmental Radioactivity 99:1698–1702.

    https://doi.org/10.1016/j.jenvrad.2008.03.013

    • PubMed
    • Google Scholar
90. 1. Satoh C
    2. Takahashi N
    3. Asakawa J
    4. Kodaira M
    5. Kuick R
    6. Hanash SM
    7. Neel JV

    (1996) Genetic analysis of children of atomic bomb survivors

    Environmental Health Perspectives 104 Suppl 3:511–519.

    https://doi.org/10.1289/ehp.96104s3511

    • PubMed
    • Google Scholar
91. 1. Sawada S

    (2007) Cover-up of the effects of internal exposure by residual radiation from the atomic bombing of hiroshima and Nagasaki

    Medicine, Conflict and Survival 23:58–74.

    https://doi.org/10.1080/13623690601084617

    • PubMed
    • Google Scholar
92. 1. Seo KY
    2. Jelinsky SA
    3. Loechler EL

    (2000) Factors that influence the mutagenic patterns of DNA adducts from chemical carcinogens

    Mutation Research/Reviews in Mutation Research 463:215–246.

    https://doi.org/10.1016/S1383-5742(00)00047-8

    • PubMed
    • Google Scholar
93. 1. Shibutani S
    2. Takeshita M
    3. Grollman AP

    (1991) Insertion of specific bases during DNA synthesis past the oxidation-damaged base 8-oxodG

    Nature 349:431–434.

    https://doi.org/10.1038/349431a0

    • PubMed
    • Google Scholar
94. 1. Shigenaga MK
    2. Gimeno CJ
    3. Ames BN

    (1989) Urinary 8-hydroxy-2'-deoxyguanosine as a biological marker of in vivo oxidative DNA damage

    PNAS 86:9697–9701.

    https://doi.org/10.1073/pnas.86.24.9697

    • PubMed
    • Google Scholar
95. 1. Smith SA
    2. Donoghue MJ

    (2008) Rates of molecular evolution are linked to life history in flowering plants

    Science 322:86–89.

    https://doi.org/10.1126/science.1163197

    • PubMed
    • Google Scholar
96. 1. Thomas JA
    2. Welch JJ
    3. Lanfear R
    4. Bromham L

    (2010) A generation time effect on the rate of molecular evolution in invertebrates

    Molecular Biology and Evolution 27:1173–1180.

    https://doi.org/10.1093/molbev/msq009

    • PubMed
    • Google Scholar
97. 1. Tubiana M
    2. Aurengo A
    3. Averbeck D
    4. Masse R

    (2006) The debate on the use of linear no threshold for assessing the effects of low doses

    Journal of Radiological Protection 26:317–324.

    https://doi.org/10.1088/0952-4746/26/3/N01

    • PubMed
    • Google Scholar
98. 1. Tubiana M
    2. Feinendegen LE
    3. Yang C
    4. Kaminski JM

    (2009) The linear no-threshold relationship is inconsistent with radiation biologic and experimental data

    Radiology 251:13–22.

    https://doi.org/10.1148/radiol.2511080671

    • PubMed
    • Google Scholar
99. 1. van Gent DC
    2. Hoeijmakers JH
    3. Kanaar R

    (2001) Chromosomal stability and the DNA double-stranded break connection

    Nature Reviews. Genetics 2:196–206.

    https://doi.org/10.1038/35056049

    • PubMed
    • Google Scholar
100. 1. Vilenchik MM
     2. Knudson AG

     (2000) Inverse radiation dose-rate effects on somatic and germ-line mutations and DNA damage rates

     PNAS 97:5381–5386.

     https://doi.org/10.1073/pnas.090099497

     • PubMed
     • Google Scholar
101. 1. Wallace SS

     (1998) Enzymatic processing of radiation-induced free radical damage in DNA

     Radiation Research 150:S60.

     https://doi.org/10.2307/3579809

     • PubMed
     • Google Scholar
102. Book

     1. Ward JF

     (1988) DNA Damage Produced by Ionizing Radiation in Mammalian Cells: Identities, Mechanisms of Formation, and Reparability

     In: Cohn W. E, Moldave K, editors. Progress in Nucleic Acid Research and Molecular Biology. Academic Press. pp. 95–125.

     https://doi.org/10.1016/s0079-6603(08)60611-x

     • Google Scholar
103. 1. Weller C
     2. Wu M

     (2015) A generation-time effect on the rate of molecular evolution in Bacteria

     Evolution 69:643–652.

     https://doi.org/10.1111/evo.12597

     • Google Scholar
104. 1. Winegar RA
     2. Lutze LH
     3. Hamer JD
     4. O'Loughlin KG
     5. Mirsalis JC

     (1994) Radiation-induced point mutations, deletions and micronuclei in lacI transgenic mice

     Mutation Research/Fundamental and Molecular Mechanisms of Mutagenesis 307:479–487.

     https://doi.org/10.1016/0027-5107(94)90258-5

     • PubMed
     • Google Scholar
105. 1. Wood ML
     2. Dizdaroglu M
     3. Gajewski E
     4. Essigmann JM

     (1990) Mechanistic studies of ionizing radiation and oxidative mutagenesis: genetic effects of a single 8-hydroxyguanine (7-hydro-8-oxoguanine) residue inserted at a unique site in a viral genome

     Biochemistry 29:7024–7032.

     https://doi.org/10.1021/bi00482a011

     • PubMed
     • Google Scholar
106. 1. Wright F

     (1990) The 'effective number of codons' used in a gene

     Gene 87:23–29.

     https://doi.org/10.1016/0378-1119(90)90491-9

     • PubMed
     • Google Scholar
107. 1. Yamamori T
     2. Yasui H
     3. Yamazumi M
     4. Wada Y
     5. Nakamura Y
     6. Nakamura H
     7. Inanami O

     (2012) Ionizing radiation induces mitochondrial reactive oxygen species production accompanied by upregulation of mitochondrial electron transport chain function and mitochondrial content under control of the cell cycle checkpoint

     Free Radical Biology and Medicine 53:260–270.

     https://doi.org/10.1016/j.freeradbiomed.2012.04.033

     • PubMed
     • Google Scholar
108. 1. Yang Z

     (2007) PAML 4: phylogenetic analysis by maximum likelihood

     Molecular Biology and Evolution 24:1586–1591.

     https://doi.org/10.1093/molbev/msm088

     • PubMed
     • Google Scholar
109. 1. Ziegler A
     2. Leffell DJ
     3. Kunala S
     4. Sharma HW
     5. Gailani M
     6. Simon JA
     7. Halperin AJ
     8. Baden HP
     9. Shapiro PE
     10. Bale AE

     (1993) Mutation hotspots due to sunlight in the p53 gene of nonmelanoma skin cancers

     PNAS 90:4216–4220.

     https://doi.org/10.1073/pnas.90.9.4216

     • PubMed
     • Google Scholar

Acceptance summary:

We still know little about how variation in natural levels of radioactivity impact germline mutation rates. By contrasting substitution rates across 14 waterlice species, the study reveals an increase in mutation rates in more radioactive environments; in particular of G>T mutations, consistent with oxidative stress. Thus, this comparative approach identifies the impact of an external mutagen on de novo mutation rates.

Decision letter after peer review:

Thank you for submitting your article "Bedrock radioactivity influences the rate and spectrum of mutation" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Detlef Weigel as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Shamil R Sunyaev (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

As you can see from the individual reviews below, the reviewers all appreciated the study design and found the results convincing and interesting. Nonetheless, they raised a number of concerns that will need to be addressed in revisions. The most serious set of concerns pertain to the interpretation of the findings. More discussion is needed of possible confounders to the association of bedrock radioactivity levels and substitution rates. Moreover, the assumption that the environmental conditions remain constant in a lineage seems tenuous, which casts doubt on the quantitative estimates provided. This point too merits further discussion. Finally, the interpretation in terms of ROS seems excessively strong. The reviewers also make a number of useful suggestions about statistical analyses, notably in terms of the analysis of the mutation spectrum. In terms of presentation, it was also felt that reviews of the field in the Introduction and Discussion could be improved substantially.

Reviewer #1:

I really liked the idea of comparing synonymous substitution rates and types in closely related species that experience different levels of bedrock radioactivity. I am convinced that there seems to be an association there, and the mutation spectrum results are also interesting, but it was hard to evaluate rival hypotheses for the association without knowing more about the species that live in environments with lower vs higher levels of radioactivity. I was also curious how diverged they are at the nucleotide level. Given that the authors have the data to look at these questions, it would be helpful to know how diversity levels and the allele frequency spectrum differs between different species compare for instance.

I also didn't quite know what to make of the quantitative estimates when I expect the level of bedrock activity likely changed quite a bit over the o(Ne) generations of the species. Could the authors comment on that?

For the mutation spectrum analysis, I think the authors should use a forward variable selection approach, such as the one employed in Harris and Pritchard, 2017.

In terms of presentation, I found the Introduction somewhat unhelpful, in that it lumps references for all sorts of ionizing radiation and UV, on both point mutations and microsatellites. For this reader at least, it would be helpful to overview what types have been studied, and what is known specifically for the type examined here. Also this study should be discussed: https://www.ncbi.nlm.nih.gov/pubmed/25809527. On a related but more minor note, I was unconvinced by the argument that this question cannot be studied experimentally and I don't think the argument is needed to motivate their approach.

In turn, the discussion of life history trait effects on mutation rates in metazoans was both too strong and oddly referenced. As examples, Martin and Palumbi, 1993 is actually an argument for consideration the effect of metabolic rates; Saclier et al., 2018 is only in isopods, when the claim is made for metazoans etc… Similar to my comment on the Introduction, I think a more systematic discussion of the literature is needed.

Finally, I thought the authors could do more to link their findings to studies in other organisms, in particular humans, where there are a number of studies of mutation patterns in populations living in diverse environments (e.g., https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1006581).

Reviewer #2:

This manuscript is proposing an attractive model suggesting that natural radioactivity (even at a low level) affects the mutation rate in waterlice. The authors collected unique biological material in a number of areas with variable radioactivity levels. The data and the proposed model are certainly of great interest. However, I find that the manuscript would benefit from a few additional statistical analyses of the data and from softening the discussion about ROS.

1) The analysis makes a major assumption that the habitat and the corresponding level of radiation remains unchanged after speciation. This assumption warrants a discussion. Overall, I agree that a violation of this assumption would result in a conservative estimate and main results should not be affected.

2) I think that the study lacks a test for robustness of pGLS.

3) The manuscript should discuss potential covariates and confounders. Is it possible that variables other than radioactivity level are responsible for the signal?

4) The conclusions of the manuscript are based on Figure 2 showing non-ulotrametric trees and ratios of synonymous branch lengths. It would be great to show the data for "polymorphic" variants (just terminal branches).

5) In human cells, radiation primarily induces deletions associated with the microhomology repair. If this is also the case in the waterlice system, the authors should be able to find a highly specific and strong signal on top of the results about point mutations. This would dramatically strengthen the conclusions.

6) I do not find that the shift in mutation spectra is a sufficiently strong evidence in favor of ROS. This either requires a much stronger argument, or the discussion about the role of oxidative damage should be softened.

Reviewer #3:

In this study, Saclier et al. explore the effects of natural radioactivity on the germline mutation rate and spectra of subterranean waterlice, aiming to address the question of whether radioactive habitats have long-term evolutionary effects on the genomes of species which inhabit them. The use of this organism to test such a fundamental question in evolutionary biology is very clever, as waterlice are not exposed to confounding effects of UV radiation and have naturally limited dispersal. I found the paper well-written and a fun topic to read, and the arguments that natural radioactivity can have a tangible impact on genome evolution were compelling.

I have a few concerns and suggestions about the regression models and statistical analyses.

Figure 2: This would be easier to follow if points were labeled/colored/shaped to indicate the corresponding species, site, and bedrock formation type.

Table 1: The authors use independent regression models to test for the associations between mutation rate and each three variables of interest: α radioactivity, Received Dose, and λ-15. Why did the authors perform three separate simple regressions rather than a multiple regression using all three of these variables as covariates? Given that λ-15 and Received Dose appear to be stronger predictors of mutation rate than α radioactivity, does this mean the statistically significant association of α radioactivity goes away when adjusting for the other two variables? I expect these three variables to be highly collinear and too many covariates could easily result in overfitting with such a small sample, but the claims drawn from this table would benefit from a more nuanced statistical analysis.

Also, were there any additional environmental covariates that may have been measured when samples were collected or integrated from other data sources (e.g., latitude, distance to nearest industrial activities, etc.)? If possible, it would be interesting to see if other environmental factors can also explain the variation in mutation rates, but I recognize that such data may not be available, and even if it were, this analysis may not be feasible with a small sample.

Table 2: Performing 3 separate regressions for each of 6 mutation types results in some multiple testing issues that must be addressed, at a conservative Bonferroni-corrected α value of.05/18=.0028, only the C:G>A>T mutation class is statistically significant across all 3 explanatory variables, but the A:T>T:A class is not. This multiple testing burden could perhaps be alleviated by using multiple regression as suggested above. Further, the response variables of the 6 regressions are proportions that add to 1, so these are not independent statistical tests, even though they are analyzed and presented as such. Is there a different statistical model that can be used that takes into account the interdependence of the 6 components of the mutation spectrum?

How well-correlated are the nuclear and mitochondrial dS/ra values across samples? A priori, I expect them to be strongly correlated, but it would be interesting and straightforward to investigate if radioactivity levels also affected relative differences in nuclear and mitochondrial mutation rates, e.g., perhaps radioactivity increases nuclear mutation rates in a linear fashion, but mitochondrial mutation rates in a non-linear fashion.

Reviewer #1:

I really liked the idea of comparing synonymous substitution rates and types in closely related species that experience different levels of bedrock radioactivity. I am convinced that there seems to be an association there, and the mutation spectrum results are also interesting, but it was hard to evaluate rival hypotheses for the association without knowing more about the species that live in environments with lower vs higher levels of radioactivity.

One of the most interesting feature of subterranean species belonging to the Asellidae family is their similarities in terms of morphology, lifestyle and life history traits. This high uniformity is likely due to a low rate of phenotypic evolution and a high level of convergence imposed by the subterranean lifestyle. As a result, distinguishing different species requires a high level of expertise and some species cannot be distinguished without molecular tools (Morvan et al., 2013). The birth of the Asellidae family is estimated at -350 My (Morvan et al., 2013) and divergence times between species are comparable to what is observed between species of mammals with very different biologies. These characteristics allow us to compare species that are divergent enough to compute an accurate rate of molecular evolution but which conserved very similar traits. We have included a paragraph in the Materials and methods describing these aspects of the biological model.

However, even if no phenotypic differences were observed for species living in radioactive habitat, we agree that radioactive environments could potentially influence life history traits, growth rate or population size. We have now included a paragraph to discuss these possible confounding factors in the manuscript.

I was also curious how diverged they are at the nucleotide level.

In the manuscript we used the estimation taken from the software CoEvol which gives a dS divided by a relative time (that we noted dS/ra). In the Supplementary file 1, we reported classical measures of the dS (as computed by mapNH and CodeML) which is the measure of the synonymous divergence. In the Figure 1—figure supplement 1, we added branch length values which represent the average nucleotide divergence for all the positions. On average, we found a 4,2% nuclear synonymous divergence between two closely related species (39% for the mitochondrial genome).

Given that the authors have the data to look at these questions, it would be helpful to know how diversity levels and the allele frequency spectrum differs between different species compare for instance.

Levels of genetic diversity (πS or πN) are significantly negatively correlated with the different estimates of natural radioactivity (pGLS p.value < 0.05). This observation is however not informative regarding the question of the mutational impact of natural radioactivity, because nucleotide diversity depends not only on the mutation rate but also on the effective population size (πS=4*Ne*μ). We therefore did not mention this point in the manuscript. We however included estimates of ⲡN and ⲡS in the Supplementary file 1, so that interested readers can have this data.

I also didn't quite know what to make of the quantitative estimates when I expect the level of bedrock activity likely changed quite a bit over the o(Ne) generations of the species. Could the authors comment on that?

We agree this point should be discussed. We included a new paragraph in the Materials and methods justifying this assumption. Natural radioactivity is unlikely to have changed in the time-frame of this study, but species exposure has probably changed due to dispersal. However, population movements should blur a signal, not generate one.

For the mutation spectrum analysis, I think the authors should use a forward variable selection approach, such as the one employed in Harris and Pritchard, 2017.

We agree that performing 6*3 pGLS tests does not take into account the dependency between the different proportions of mutation. At the same time, the test proposed by Harris and Pritchard, 2017, does not take into account phylogenetic inertia. Integrating the phylogenetic signal is important because this is one of the main factors explaining the mutational spectrum variations. To take into account these two types of non-independency in the mutational spectrum analysis and reconcile the two approaches, we first performed the test proposed by Harris and Pritchard, 2017, on the mutation counts as advised by the reviewer. This test gives only one mutation significantly linked to radioactivity : C:G -> A:T. We then performed a pGLS test where p.values were corrected with Holm’s method to adjust for the multiple tests and to control the false discovery rate. In the manuscript we reported the results of this statistical procedure in Table 2 and in the text, as well as in the Materials and methods.

In terms of presentation, I found the Introduction somewhat unhelpful, in that it lumps references for all sorts of ionizing radiation and UV, on both point mutations and microsatellites. For this reader at least, it would be helpful to overview what types have been studied, and what is known specifically for the type examined here. Also this study should be discussed: https://www.ncbi.nlm.nih.gov/pubmed/25809527. On a related but more minor note, I was unconvinced by the argument that this question cannot be studied experimentally and I don't think the argument is needed to motivate their approach.

We agree that a longer Introduction will be helpful and have included 3 new paragraphs or sections of paragraphs to the Introduction. A new paragraph summarizes the different types of mutations generated by radioactivity. We distinguished point mutations from deletions and chromosomal rearrangements, while acknowledging that the current knowledge comes from studies of radiation doses much higher than natural radioactivity. We added the article pointed by the reviewer which was completely relevant in this part as the authors showed that radioactivity generated clustered mutations and deletions. This new introductory paragraph concludes that point mutations have been poorly studied compared to double-strand-breaks or deletions.

We then added a part explaining more extensively what we know about the impact of very low doses of radioactivity such as in the case of natural radioactivity, and which factors could be at play to explain the difficulty to characterize a dose-response relationship.

Finally, we agree that the fact that the long term effect of low doses of radiation is challenging to test experimentally is not needed to motivate our approach. Our goal here is to highlight that a phylogenetic approach has never been used before to tackle that challenging question and can be a complementary approach. We have re-formulated this idea.

In turn, the discussion of life history trait effects on mutation rates in metazoans was both too strong and oddly referenced. As examples, Martin and Palumbi, 1993 is actually an argument for consideration the effect of metabolic rates; Saclier et al., 2018 is only in isopods, when the claim is made for metazoans etc… Similar to my comment on the Introduction, I think a more systematic discussion of the literature is needed.

We agree that this paragraph could be read as a too strong and too general about the relative influence of radioactivity compared to life history traits. As the literature do not allow to extend our results to metazoans, we modified the section to make it clear that this is an observation that may only be valid for isopods and needs to be investigated in other groups.

Finally, I thought the authors could do more to link their findings to studies in other organisms, in particular humans, where there are a number of studies of mutation patterns in populations living in diverse environments (e.g., https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1006581).

We agree that the mutation spectrum was too rapidly discussed. We have included a new paragraph where we discuss the factors known to modify the mutational spectrum in diverse organisms (mostly human and mice). We also enriched this discussion with studies about other carcinogens known to also modify the mutational spectrum.

Reviewer #2:

This manuscript is proposing an attractive model suggesting that natural radioactivity (even at a low level) affects the mutation rate in waterlice. The authors collected unique biological material in a number of areas with variable radioactivity levels. The data and the proposed model are certainly of great interest. However, I find that the manuscript would benefit from a few additional statistical analyses of the data and from softening the discussion about ROS.

1) The analysis makes a major assumption that the habitat and the corresponding level of radiation remains unchanged after speciation. This assumption warrants a discussion. Overall, I agree that a violation of this assumption would result in a conservative estimate and main results should not be affected.

This is a point that has also been raised by reviewer #1, please refer to the answer to his fourth comment.

2) I think that the study lacks a test for robustness of pGLS.

Indeed, pGLS makes some assumptions for which robustness should be tested. pGLS supposes normality of residuals, homoscedasticity, absence of influential cases, an evolutionary process (here a brownian motion) which is homogenous across the branches of the phylogenetic tree. Normality was checked by plotting residuals of models. Homoscedasticity was checked by plotting the residuals against the fitted values of models. Absence of influential cases was tested by a jackknife method consisting in removing one by one each species of the data and redoing the test each time the pGLS test p-value was below 0.05. Robustness to the evolutionary model was tested by doing the pGLS with the four main evolutionary models (Blomberg, Martins, Pagel and Brownian). P.values remained below 0.05 whatever the model used. These additional tests have been added in the Statistical analyses part. Testing robustness to the homogeneity of the model of evolution across the tree is more complicated. A test proposed by Mazel et al., 2016, Ecology incorporates shifts in the parameters of the model of evolution along the tree. However, the function behind this test needs a minimum of 64 taxa to be powerful enough in detecting shifts (Eastman et al., 2011, Evolution). Despite this, we performed Mazel’s test but detected no shift. For the sake of clarity and to keep a simple and straightforward presentation of the results, we have not included this latest additional result.

3) The manuscript should discuss potential covariates and confounders. Is it possible that variables other than radioactivity level are responsible for the signal?

This is a major point that has also been raised by the reviewer #1, please refer to the answer to his question n°1.

4) The conclusions of the manuscript are based on Figure 2 showing non-ulotrametric trees and ratios of synonymous branch lengths. It would be great to show the data for "polymorphic" variants (just terminal branches).

This is a point which converge with the question asked by the reviewer #1, please refer to the answer to his question n°2.

5) In human cells, radiation primarily induces deletions associated with the microhomology repair. If this is also the case in the waterlice system, the authors should be able to find a highly specific and strong signal on top of the results about point mutations. This would dramatically strengthen the conclusions.

Indeed, the major effect of radioactivity on DNA is to generate deletions (of note, we precise this point in the Introduction). However, the dataset used in this study was generated using transcriptome sequencing, not genome sequencing, and it is therefore limited to protein coding sequences. It is very unlikely to find deletions in protein coding sequences as they are often deleterious and removed by natural selection. Studying the impact of radioactivity on the deletion rate should be done on the non-coding part of the genome, which we don’t have access to for the time being.

Counting the number of deletion polymorphisms in the coding sequences is nevertheless possible. We used Freebayes to count the number of polymorphic deletions in the 2490 1-to-1 orthologous genes. We find no correlation between the level of radioactivity and the number of deletions, whatever the measure used (e.g. pGLS, p.value=0.7034 for λ15). But again this is not a valid demonstration that natural radioactivity has no influence on the deletion rate.

6) I do not find that the shift in mutation spectra is a sufficiently strong evidence in favor of ROS. This either requires a much stronger argument, or the discussion about the role of oxidative damage should be softened.

We agree and have included a completely new paragraph to discuss the evidence supporting a potential influence of the ROS. We have softened our conclusion by saying that the modification of the mutational spectrum suggests an impact of radioactivity mediated by the formation of ROS.

Reviewer #3:

In this study, Saclier et al. explore the effects of natural radioactivity on the germline mutation rate and spectra of subterranean waterlice, aiming to address the question of whether radioactive habitats have long-term evolutionary effects on the genomes of species which inhabit them. The use of this organism to test such a fundamental question in evolutionary biology is very clever, as waterlice are not exposed to confounding effects of UV radiation and have naturally limited dispersal. I found the paper well-written and a fun topic to read, and the arguments that natural radioactivity can have a tangible impact on genome evolution were compelling.

I have a few concerns and suggestions about the regression models and statistical analyses.

Figure 2: This would be easier to follow if points were labeled/colored/shaped to indicate the corresponding species, site, and bedrock formation type.

The Figure 2 has been modified following these recommendations.

Table 1: The authors use independent regression models to test for the associations between mutation rate and each three variables of interest: α radioactivity, Received Dose, and λ-15. Why did the authors perform three separate simple regressions rather than a multiple regression using all three of these variables as covariates? Given that λ-15 and Received Dose appear to be stronger predictors of mutation rate than α radioactivity, does this mean the statistically significant association of α radioactivity goes away when adjusting for the other two variables? I expect these three variables to be highly collinear and too many covariates could easily result in overfitting with such a small sample, but the claims drawn from this table would benefit from a more nuanced statistical analysis.

We chose to do three separate tests for three main reasons. First, the λ15 measure allows us to include the two sites nearby mines (showing a disequilibrium in the uranium's chains). Doing a single test with the three measures means we have to remove these two locations. Second, as pointed by the reviewer, as we have only 14 points, a test with 3 predictors will lack power to detect an effect. Finally, and maybe most importantly, the three measures of radioactivity are redundant. Absence of collinearity among predictors is required for the validity of a linear model. Conclusions about the impact of multiple predictors that are collinear are unreliable in the linear model framework (Quinn and Keough, 2002). α radioactivity is a part of the total radioactivity which is measured to compute the received dose of radioactivity, thus the two measures are expected to be collinear (R² = 0.82). λ15 is also collinear with the two other measures (R² = 0.61 and 0.77, with α radio and Received dose, respectively). As a result, if we do a single pGLS with the synonymous substitution rate against the three measures of radioactivity, none of the three factors remain significant (p.values = 0.29, 0.76 and 0.08 for α radioactivity, Received dose and λ15, respectively). Absence of statistical significance in such a test is the result of collinearity and falsely suggests an absence of effect. For the sake of clarity, we have not included this test in the manuscript, however we added a sentence in the caption of Table 1 justifying the absence of such test because of collinearity.

Also, were there any additional environmental covariates that may have been measured when samples were collected or integrated from other data sources (e.g., latitude, distance to nearest industrial activities, etc.)? If possible, it would be interesting to see if other environmental factors can also explain the variation in mutation rates, but I recognize that such data may not be available, and even if it were, this analysis may not be feasible with a small sample.

We agree that a more comprehensive description of other factors, in particular potentially mutagenic factors, would be valuable. Unfortunately, the already complex sampling strategy of pairs of contrasted species associated with a detailed description of the natural radioactivity (in situ characterisation and received dose modelisation) was already a challenge to put in place in the field and we could not reasonably complexify it even more.

We have modified the Discussion to integrate a description of other potential confounding factors. Of particular interest are heavy metals whose concentration is correlated with radioactivity as they are a sub-product of the decay chains. To disentangle the impact of radioactivity and heavy metal, it would require a lot of observations, which does not correspond to the sampling strategy of this study.

Table 2: Performing 3 separate regressions for each of 6 mutation types results in some multiple testing issues that must be addressed, at a conservative Bonferroni-corrected α value of.05/18=.0028, only the C:G>A>T mutation class is statistically significant across all 3 explanatory variables, but the A:T>T:A class is not. This multiple testing burden could perhaps be alleviated by using multiple regression as suggested above. Further, the response variables of the 6 regressions are proportions that add to 1, so these are not independent statistical tests, even though they are analyzed and presented as such. Is there a different statistical model that can be used that takes into account the interdependence of the 6 components of the mutation spectrum?

As suggested by the reviewer #1, we performed the test proposed by Harris and Pritchard, 2017, that we coupled with a pGLS where p.values have been corrected for multiple hypothesis testing following Holm’s method. Please refer to the answer to question #5 of the Reviewer #1.

How well-correlated are the nuclear and mitochondrial dS/ra values across samples? A priori, I expect them to be strongly correlated, but it would be interesting and straightforward to investigate if radioactivity levels also affected relative differences in nuclear and mitochondrial mutation rates, e.g., perhaps radioactivity increases nuclear mutation rates in a linear fashion, but mitochondrial mutation rates in a non-linear fashion.

Indeed, the mitochondrial and the nuclear dS are well correlated (PGLS, p.value = 0.0065, Cox-Snell pseudoR² = 0.41). However the rate ratio [mitochondrial dS/nuclear dS] is not linearly correlated with the level of radioactivity (p.values > 0.05) even if a positive trend is found with α radioactivity (p.value=0.09). A graphical evaluation suggests that for exposure to low levels of natural radioactivity (<3 µSv/h), the ratio seems to increase linearly, suggesting a faster acceleration of the mitochondrial rate, but for higher doses this ratio seems to drop. However, with only 14 points, this pattern cannot be accurately described. One complication is the saturation of the mitochondrial rate. As the mitochondrial rate is about 10X times higher than the nuclear one, an important rate acceleration will tend to be less well estimated in the mitochondrial genome compared to the nuclear one. When working on rate ratios, this bias alone could generate a pattern. A much bigger dataset and methods to compensate for saturation are therefore necessary to characterize the relative differences in nuclear and mitochondrial mutation rates in response to natural radioactivity. For the sake of clarity, we have not included these new analyses.

Author details

1. Nathanaëlle Saclier

   Univ Lyon, Université Claude Bernard Lyon 1, CNRS UMR 5023, ENTPE, Laboratoire d’Ecologie des Hydrosystèmes Naturels et Anthropisés, Villeurbanne, France

   Present address

   ISEM, CNRS, Univ. Montpellier, IRD, EPHE, Montpellier, France

   Contribution

   Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing - original draft, Project administration, Did the field work, defined orthologous gene families, built the phylogeny, computed substitution rate, performed statistical analyses and computed the mutational spectrum

   For correspondence

   nathanaelle.saclier@gmail.com

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-1522-9644

2. Patrick Chardon

   LPC, Université Clermont Auvergne, CNRS/IN2P3 UMR6533, Clermont-Ferrand, France

   Contribution

   Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Writing - review and editing, Did the field work, performed all radioactivity measurements and the received dose estimation

   Competing interests

   No competing interests declared

3. Florian Malard

   Univ Lyon, Université Claude Bernard Lyon 1, CNRS UMR 5023, ENTPE, Laboratoire d’Ecologie des Hydrosystèmes Naturels et Anthropisés, Villeurbanne, France

   Contribution

   Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Methodology, Project administration, Writing - review and editing, Did the field work, dissected and identified morphologically all sampled species

   Competing interests

   No competing interests declared

4. Lara Konecny-Dupré

   Univ Lyon, Université Claude Bernard Lyon 1, CNRS UMR 5023, ENTPE, Laboratoire d’Ecologie des Hydrosystèmes Naturels et Anthropisés, Villeurbanne, France

   Contribution

   Data curation, Formal analysis, Supervision, Writing - review and editing, Extracted DNA and RNA from all samples, made PCR and migrations and prepared library for sequencing

   Competing interests

   No competing interests declared

5. David Eme

   Univ Lyon, Université Claude Bernard Lyon 1, CNRS UMR 5023, ENTPE, Laboratoire d’Ecologie des Hydrosystèmes Naturels et Anthropisés, Villeurbanne, France

   Present address

   IFREMER, unité Ecologie et Modeles pour l’Halieutique, Nantes, France

   Contribution

   Data curation, Validation, Investigation, Writing - review and editing, Did the field work, dissected and identified morphologically all sampled species

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-8790-0412

6. Arnaud Bellec

   1. Univ Lyon, Université Claude Bernard Lyon 1, CNRS UMR 5023, ENTPE, Laboratoire d’Ecologie des Hydrosystèmes Naturels et Anthropisés, Villeurbanne, France
   2. Univ Lyon, Université Jean Moulin Lyon 3, CNRS UMR 5600 Environnement Ville Société, Lyon, France

   Contribution

   Formal analysis, Writing - review and editing, Extracted the outcrop cover of low-radioactivity sedimentary rocks and high-radioactivity metamorphic and igneous rocks in a radius of 15 km around the sampling, and made the map of uranium for Figure 1

   Competing interests

   No competing interests declared

7. Vincent Breton

   LPC, Université Clermont Auvergne, CNRS/IN2P3 UMR6533, Clermont-Ferrand, France

   Contribution

   Validation, Writing - review and editing

   Competing interests

   No competing interests declared

8. Laurent Duret

   Univ Lyon, Université Claude Bernard Lyon 1, CNRS UMR 5558, Laboratoire de Biométrie et Biologie Evolutive, Villeurbanne, France

   Contribution

   Validation, Methodology, Writing - review and editing, Supervised the computation of the mutational spectrum

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2836-3463

9. Tristan Lefebure

   Univ Lyon, Université Claude Bernard Lyon 1, CNRS UMR 5023, ENTPE, Laboratoire d’Ecologie des Hydrosystèmes Naturels et Anthropisés, Villeurbanne, France

   Contribution

   Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Writing - original draft, Project administration, Writing - review and editing, Did the field work, assembled transcriptomes and defined ORFs and orthologous gene families

   For correspondence

   tristan.lefebure@univ-lyon1.fr

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-3923-8166

10. Christophe J Douady

    1. Univ Lyon, Université Claude Bernard Lyon 1, CNRS UMR 5023, ENTPE, Laboratoire d’Ecologie des Hydrosystèmes Naturels et Anthropisés, Villeurbanne, France
    2. Institut Universitaire de France, Paris, France

    Contribution

    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Methodology, Writing - original draft, Project administration, Writing - review and editing, Did the field work

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-4503-8040

• 1,439

  Page views

• 120

  Downloads

• 3

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.