Evolution of anticipatory effects mediated by epigenetic changes

Abstract

Anticipatory effects mediated by epigenetic changes occur when parents modify the phenotype of their offspring by making epigenetic changes in their gametes, guided by information from an environmental cue. To investigate when do anticipatory effects mediated by epigenetic changes evolve in a fluctuating environment, I use an individual-based simulation model with explicit genetic architecture. The model allows for the population to respond to environmental changes by evolving plasticity, bet hedging, or by tracking the environment with genetic adaptation, in addition to the evolution of anticipatory effects. The results show that anticipatory effects evolve when the environmental cue provides reliable information about the environment and the environment changes at intermediate rates, provided that fitness costs of anticipatory effects are rather low. Moreover, evolution of anticipatory effects is quite robust to different genetic architectures when reliability of the environmental cue is high. Anticipatory effects always give smaller fitness benefits than within-generation plasticity, suggesting a possible reason for generally small observed anticipatory effects in empirical studies.

Keywords: between-generation plasticity; genetic architecture; intergenerational effects; phenotypic plasticity; transgenerational plasticity.

Fig 1.

Events in the simulation model. Each individual lives for five timesteps. In the first timestep, individuals can use epigenetic changes inherited from their parents to modify their phenotype and pay a fitness cost of k_e for this action. Epigenetic modifications are reset at the end of this step. In the next timestep, individuals can use developmental plasticity to adjust their phenotype based on an observed environmental cue, with a fitness cost of k_d. In Timesteps 3–5, individuals can use reversible plasticity to further adjust their phenotype based on the observed cues, with a fitness of cost of k_a for each adjustment. After Timestep 5, any potential epigenetic modifications are set based on an observed environmental cue from Timestep 5. Then, fitness is calculated for each individual, individuals produce offspring in proportion to their fitness, population regulation happens, and next generation is formed. Then, new mutations are produced in those individuals and the cycle starts again.

Fig 2.

Simulation results for and . For each small panel, the environmental cue is on the x-axis and phenotype on the y-axis. Each reaction norm is the population mean for one replicate simulation, and 100 replicates were performed for each combination of P and R. Red color shows reaction norms, where the population evolved reversible plasticity ( or , and ). Orange color shows reaction norms where reversible and anticipatory effects evolved (in addition ), blue color indicates that only developmental plasticity evolved ( or , and ), purple color indicates that both developmental plasticity and anticipatory effects evolved (in addition ), green color indicates that increased developmental variation evolved (), and black color indicates that there was no major change in reaction norm slope (). Note the logarithmic scale for R.

Fig 3.

Heat map for mean probability of epigenetic modification from 100 replicate simulations with each combination of P and R. Costs of plasticity were and .

Fig 4.

Simulation results with different costs for phenotypic plasticity and anticipatory effects. For each parameter combination of P and R, 100 replicate simulations are represented as colored tiles, with colors corresponding to different strategies evolved in that replicate. Costs are (A) , (B) and , (C) and .

Fig 5.

Sensitivity of results to genetic architecture. Hundred replicate simulations were run for each combination of R and P, where parameter values for the random environmental effects, number of loci, number of chromosomes, mutation rate, and mutational effects were drawn randomly. Costs of plasticity were and . Colored tiles show which strategy evolved in the population or if it went extinct.

Fig 6.

Effects of number of loci (A–C) and standard deviation of mutational effects (D–F). Hundred replicate simulations were run for each combination of R and P, but only those values of R were used where I had previously observed the evolution of plasticity when plasticity had a cost. Cost of plasticity were and . Colored tiles show which strategy evolved in the population. (A) , (B) , (C) , (D) , (E) , (F) .