Temperature dependence of spontaneous mutation rates

Abstract

Mutation is the source of genetic variation and the fundament of evolution. Temperature has long been suggested to have a direct impact on realized spontaneous mutation rates. If mutation rates vary in response to environmental conditions, such as the variation of the ambient temperature through space and time, they should no longer be described as species-specific constants. By combining mutation accumulation with whole-genome sequencing in a multicellular organism, we provide empirical support to reject the null hypothesis of a constant, temperature-independent mutation rate. Instead, mutation rates depended on temperature in a U-shaped manner with increasing rates toward both temperature extremes. This relation has important implications for mutation-dependent processes in molecular evolution, processes shaping the evolution of mutation rates, and even the evolution of biodiversity as such.

Figure 1.

Experimental setup of mutation accumulation at different constant temperatures. The succession of circles indicates how many generations (F1–F5) could be passed under the different temperature conditions. Circles are labeled with the total number of MALs that survived until the respective generation. Mean generation time (GT) is decreasing with increasing temperature as indicated in the detail graph. Number of generational passages (GP = MALs × generation) sums up to a total of 205 (excluding GPs of mutator lines).

Figure 2.

Spontaneous mutation rates per haploid genome per generation in relation to temperature of the MAL experiment. Top: Mutation rate of the total sum of de novo mutations. Bottom: Separate mutation rates of single nucleotide mutations (SNMs) and single nucleotide indels (SNIs). Nonlinear fit as second-order polynomial regression; R² values are given to describe the goodness of model fit. Corresponding test statistics are listed in Supplemental Data 2.

Figure 3.

Heat map of detected SNM types per experimental temperature. Transitions in rows 1 and 2, transversions in rows 3 to 6. Different mutation types explain 43% (F = 7.52, P = 0.0002) and temperatures 28% (F = 4.94, P = 0.0028) of the total variance.

Figure 4.

Expected variation of mutation rate calculated per generation for a natural C. riparius field population (left y-axis) with regard to the annual variation in water temperature (Hasselbach in Hesse, Germany) averaged for each generation period (right y-axis). Due to the temperature-dependence of generation times, C. riparius can pass eight generations per year (x-axis). The mean effective µ is 5.29-fold higher than µ measured under laboratory conditions (indicated by arrow) (Oppold and Pfenninger 2017).