Gene flow between populations adapting to differing local environmental conditions might be costly because individuals can disperse to habitats where their survival is low or because they can reproduce with locally maladapted individuals. The amount by which the mean relative population fitness is kept below one creates an opportunity for modifiers of the genetic architecture to spread due to selection. Prior work that separately considered modifiers changing dispersal, recombination rates, or altering dominance or epistasis, has typically focused on the direction of selection rather than its absolute magnitude. We here develop methods to determine the strength of selection on modifiers of the genetic architecture, including modifiers of the dispersal rate, in populations that have previously evolved local adaptation. We consider scenarios with up to five loci contributing to local adaptation and derive a new model for the deterministic spread of modifiers. We find that selection for modifiers of epistasis and dominance is stronger than selection for decreased recombination, and that selection for partial reductions in recombination are extremely weak, regardless of the number of loci contributing to local adaptation. The spread of modifiers that reduce dispersal depends on the number of loci, epistasis and extent of local adaptation in the ancestral population. We identify a novel effect, that modifiers of dominance are more strongly selected when they are unlinked to the locus that they modify. These findings help explain population differentiation and reproductive isolation and provide a benchmark to compare selection on modifiers under finite population sizes and demographic stochasticity.

Just as phenotypic plasticity can evolve when developing individuals get informational cues about their future adult environment, deterministic maternal effects, where offspring trait values depend on the maternal environment, can evolve when mothers gain reliable information about the environments their offspring will face. Randomizing maternal effects (a type of diversifying bet hedging), where offspring trait values are randomized, can evolve by natural selection even when information about future environments is unavailable. We investigate selection on both randomizing and deterministic maternal effects in environments that show correlated fluctuations between two environmental states. We compare the strength of selection for deterministic and randomizing maternal effects and explicitly consider maternal fitness costs of producing offspring with different phenotypes. Only a small set of environmental parameters allow randomizing maternal effects to outcompete deterministic maternal effects; not only must there be little or no information available about future environments, but the frequency of each environment must fall within a narrow range. By contrast, deterministic maternal effects can always invade an ancestral state lacking a maternal effect even if the amount of environmental information available is low. The long-term outcome may involve offspring trait value randomization but only if trait values first evolve to cause extreme differences in environment-specific fitness. Overall, deterministic maternal effects are more likely to evolve by natural selection than randomizing maternal effects.

All organisms live in temporally fluctuating environments. Theory predicts that the evolution of deterministic maternal effects (i.e., anticipatory maternal effects or transgenerational phenotypic plasticity) underlies adaptation to environments that fluctuate in a predictably alternating fashion over maternal-offspring generations. In contrast, randomizing maternal effects (i.e., diversifying and conservative bet-hedging), are expected to evolve in response to unpredictably fluctuating environments. Although maternal effects are common, evidence for their adaptive significance is equivocal since they can easily evolve as a correlated response to maternal selection and may or may not increase the future fitness of offspring. Using the hermaphroditic nematode Caenorhabditis elegans, we here show that the experimental evolution of maternal glycogen provisioning underlies adaptation to a fluctuating normoxia-anoxia hatching environment by increasing embryo survival under anoxia. In strictly alternating environments, we found that hermaphrodites evolved the ability to increase embryo glycogen provisioning when they experienced normoxia and to decrease embryo glycogen provisioning when they experienced anoxia. At odds with existing theory, however, populations facing irregularly fluctuating normoxia-anoxia hatching environments failed to evolve randomizing maternal effects. Instead, adaptation in these populations may have occurred through the evolution of fitness effects that percolate over multiple generations, as they maintained considerably high expected growth rates during experimental evolution despite evolving reduced fecundity and reduced embryo survival under one or two generations of anoxia. We develop theoretical models that explain why adaptation to a wide range of patterns of environmental fluctuations hinges on the existence of deterministic maternal effects, and that such deterministic maternal effects are more likely to contribute to adaptation than randomizing maternal effects.

Partial selfing, whereby self- and cross- fertilization occur in populations at intermediate frequencies, is generally thought to be evolutionarily unstable. Yet, it is found in natural populations. This could be explained if populations with partial selfing are able to reduce genetic loads and the possibility for inbreeding depression while keeping genetic diversity that may be important for future adaptation. To address this hypothesis, we compare the experimental evolution of Caenorhabditis elegans populations under partial selfing, exclusive selfing or predominant outcrossing, while they adapt to osmotically challenging conditions. We find that the ancestral genetic load, as measured by the risk of extinction upon inbreeding by selfing, is maintained as long as outcrossing is the main reproductive mode, but becomes reduced otherwise. Analysis of genome-wide single-nucleotide polymorphisms (SNPs) during experimental evolution and among the inbred lines that survived enforced inbreeding indicates that populations with predominant outcrossing or partial selfing maintained more genetic diversity than expected with neutrality or purifying selection. We discuss the conditions under which this could be explained by the presence of recessive deleterious alleles and/or overdominant loci. Taken together, our observations suggest that populations evolving under partial selfing can gain some of the benefits of eliminating unlinked deleterious recessive alleles and also the benefits of maintaining genetic diversity at partially dominant or overdominant loci that become associated due to variance of inbreeding levels.

Evolutionary responses to environmental change depend on the time available for adaptation before environmental degradation leads to extinction. Explicit tests of this relationship are limited to microbes where adaptation usually depends on the sequential fixation of de novo mutations, excluding standing variation for genotype-by-environment fitness interactions that should be key for most natural species. For natural species evolving from standing genetic variation, adaptation at slower rates of environmental change may be impeded since the best genotypes at the most extreme environments can be lost during evolution due to genetic drift or founder effects. To address this hypothesis, we perform experimental evolution with self-fertilizing populations of the nematode Caenorhabditis elegans and develop an inference model to describe natural selection on extant genotypes under environmental change. Under a sudden environmental change, we find that selection rapidly increases the frequency of genotypes with high fitness in the most extreme environment. In contrast, under a gradual environmental change selection first favors genotypes that are worse at the most extreme environment. We demonstrate with a second set of evolution experiments that, as a consequence of slower environmental change and thus longer periods to reach the most extreme environments, genetic drift and founder effects can lead to the loss of the most beneficial genotypes. We further find that maintenance of standing genetic variation can retard the fixation of the best genotypes in the most extreme environment because of interference between them. Taken together, these results show that slower environmental change can hamper adaptation from standing genetic variation and they support theoretical models indicating that standing variation for genotype-by-environment fitness interactions critically alters the pace and outcome of adaptation under environmental change.