Genes shielded, repeats exposed: mutation bias in the midge Chironomus riparius

Abstract

Mutation is the fundamental source of genetic variation, yet growing evidence shows that mutations are not uniformly distributed across genomes but are shaped by genomic architecture, DNA-repair dynamics, and environmental conditions. Here, we investigate fine-scale determinants of mutation distribution in the non-biting midge Chironomus riparius, an ecologically important freshwater insect widely used in ecotoxicology. We integrated mutation data from five independent studies, including spontaneous mutation-accumulation experiments and multigenerational exposure assays involving cadmium, benzo[a]pyrene, tyre and road wear particles, and varying generational time. In total, we analysed 420 single-nucleotide mutations mapped to the chromosome-scale C. riparius reference genome. Using a Bayesian modelling framework, we tested whether mutation density is (i) randomly distributed, (ii) non-randomly distributed, or predicted by (iii) distance to telomeres and centromeres, (iv) proximity to genes, or (v) distance to repetitive elements. Models were compared using leave-one-out cross-validation (LOO-CV). We also quantified the proportion of mutations in exons and evaluated the synonymous vs. non-synonymous spectrum using BayesFactor in R. The best-supported model incorporated non-linear effects of genomic position and distance to genes, identifying proximity to coding regions as the dominant predictor of mutation rate. Mutation density increased with distance from genes, indicating strong protection of genic regions. A model including repetitive elements showed nearly equivalent support, suggesting that functional and structural features jointly shape mutational landscapes. Only 9.8% of mutations occurred in exons despite exons representing 22.85% of callable sites, demonstrating marked depletion of exonic mutations. Among exonic mutations, 70.7% were non-synonymous--statistically indistinguishable from the neutral expectation (75%). These findings show that mutation processes in C. riparius are strongly structured by genome architecture, with implications for evolutionary genomics, ecotoxicology, and population-genetic inference.