Institut des Sciences de l'Evolution, UMR 5554
Can we predict adaptation trajectories on simple fitness landscapes?
Author(s): Martin, G
The dynamics of adaptation to a new environment is inherently complex, even in the simplest situations such as encountered in experimental evolution). Over the past decades, several empirical studies have measured the long-term dynamics of adaptation in different model species (mostly micro-organisms). Yet, existing theory does not, to our knowledge, provide quantitative predictions to which such trajectories could be compared. Indeed, the speed with which a population adapts (i.e. the speed of the mean fitness increase) depends on the rate, fitness effects, and fate of beneficial mutations. While tremendous progress has been made in modelling these processes in steady state regimes, the key parameters are still difficult to measure, and more important, they vary over time in observed adaptive trajectories, because of background dependence effects (epistasis). The net result is that populations do not simply adapt linearly over time (i.e. at some steady state rate of increase), and the form of this non-linearity is not predicted by any widely accepted model. Our goal here will be to show that some tools exist that provide testable predictions in this context, and check on a few examples if these predictions are accurate. We will first present some old and new results on Fisher’s model and how they allow to predict the change in rate and effect of beneficial mutations over adaptive trajectories, from empirically measurable deleterious mutation effects and rate. Then we will discuss several alternative tools that can be used to model adaptation trajectories in these types of landscapes, accounting for non - stationary distributions of mutation effects and rates. We will illustrate the use of this approach on some empirical trajectories in model species.
Empirical fitness landscapes reveals a limited number of accessible adaptive pathways for an RNA virus
Author(s): Elena, SF, Lalic, J, Franke, J, DelaIglesia, F
RNA viruses are the main source of emerging infectious diseases owed to the evolutionary potential bestow by their fast replication, large population sizes and high mutation and recombination rates. However, an equally important parameter, which is usually neglected, is the topography of the fitness landscape, that is, how many fitness maxima exist and how well connected they are, which determines the number of accessible evolutionary pathways. To address this question, we have reconstructed a fitness landscape describing the adaptation of Tobacco etch potyvirus to a new host, Arabidopsis thaliana. Two fitness traits were measured for most of the genotypes in the landscape, infectivity and virus accumulation. We found prevailing epistatic effects between mutations in the early steps of adaptation, while independent effects became more common at latter stages. Results suggest that the landscape was rather smooth, with a high number of potential neutral paths and a single fitness peak.
Department of Biology and Biochemistry
Epistasis and evolvability in experimentally evolved populations of Escherichia coli
Author(s): Cooper, TF
Epistatic interactions between mutations play a prominent role in many evolutionary theories. Many studies have found that epistasis is widespread, but direct analyses of epistasis can be technically difficult and has not generally considered beneficial mutations. We analyse the effects of epistasis on fitness in a set of genotypes including all combinations of the first five beneficial mutations to fix in an experimental population of Escherichia coli. We show that epistasis depends strongly on the fitness effects of the combined mutations – the larger the expected benefit, the more negative the effect of epistasis on fitness. A similar pattern of interactions is also seen among a set of seven mutations that substituted in a population that was selected in an environment containing a combination of two sugars. Epistasis thus tended to follow a simple relationship of diminishing returns with genotype fitness. This observation supports a model that predicts negative epistasis explains a decelerating rate of adaptation as populations approach a fitness peak. Preliminary experiments are consistent with this prediction, finding that the effective size, but not rate, of beneficial mutations declines across a series of replay populations started from genotypes of progressively higher fitness.
Institute for Theoretical Physics
Epistatic constraints and evolutionary predictability on empirical fitness landscapes
Author(s): Krug, J
The adaptive dynamics of an asexual population in the space of genotypes is constrained by epistatic interactions between mutations at different genetic loci. Recent empirical studies have shown that this strongly reduces the number of mutational pathways that are accessible under conditions of strong selection and weak mutation (SSWM). In the talk I will describe statistical models for fitness landscapes that quantify evolutionary accessibility under different assumptions on the amount of epistasis as well as on the underlying genetic architecture, and show how these models can be used to classify and interpret empirical data sets. I then discuss the impact of epistatic constraints on the predictability of evolutionary trajectories in asexuals, with particular emphasis on the role of population size. With increasing population size clonal interference implies a preference for mutational steps of large effect, which leads to an increase in predictability beyond the expectation under SSWM dynamics. However, a further increase of population size reduces predictability by opening up new pathways that involve the crossing of fitness valleys by multiple mutations. This nonmonotonic pattern of evolutionary predictability is found in large-scale simulations on an empirical fitness landscape, and argued to be observable in experiments that monitor the variability of fitness trajectories among replicate populations.
Evading genetic drift: an experimental test of the probability of fixation of new genetic variants
Author(s): Chelo, IM, Nédli, J, Gordo, I, Teotónio, H
Progression through an adaptive fitness landscape requires the appearance and fixation of beneficial mutations. But fixation of a beneficial variant is not a necessary outcome, as deduced by J.B.S. Haldane in 1927, instead its probability can be given by twice the fitness effect: Pfix= 2s. This is based in the reasoning that even beneficial variants are lost with high probability by genetic drift. In this work, we performed invasion experiments with inbred lines of Caenorhabditis elegans in well-defined demographic conditions to experimentally demonstrate the determinant role of drift in the initial dynamics of new beneficial alleles. We provide the evidence that extinction rates (Pext) decrease with the initial numbers of beneficial variants, as expected. We also show that the extinction of a deleterious variant, when at low frequency, is higher than that of a beneficial variant thus establishing that classical population genetics theory can accurately predict the fate of low frequency variants. Remarkably though we also find that, when at high frequency the fate of these variants is distinct from their low frequency dynamics, which results not in their ultimate fixation or loss, but on their maintenance. Our data confirm one of the key results of population genetics theory and highlights the complex nature of adaptation, where polymorphism can be maintained or lost depending on population structure.
Institute of Ecology and Evolution
Evolutionary consequences of female promiscuity
Author(s): Timmermeyer, N, Michiels, NK, Phillips, PC
To cope with stressful conditions like parasite attacks or temperature changes, females can increase their offspring diversity and fitness by mating multiply with several males. This can result in male harm but can be advantageous in fluctuating environments. Populations with promiscuous females therefore are supposed to be more stable than populations with monogamous females. We study the gonochoristic nematode Caenorhabditis remanei, which is mating multiply and affected by male harm. Using experimental evolution, we were able to show that females, in contrast to males, were able to adapt to sex ratio manipulations. Additionally, offspring of promiscuous females had a higher fitness than monogamous females under the influence of the microparasite Bacillus thuringiensis. Even though a powerful tool to investigate fitness effects, manipulating sex ratios in nematodes in scientifically meaningful numbers is very work intensive if done by hand. To increase the population size and replicate number and therefore decrease the effects of drift, we are using differently labeled males and females and an automated sorting device. Worms are loaded onto a polymeric silicon chip (PDMS) and screened for fluorescent makers, resulting in separated males and females, which can be combined in different sex ratios depending on the treatment. These populations will be tested under stressful conditions to investigate whether populations with promiscuous females are more resistant to stressful changes in their environment compared to the monogamous control. Additionally, individual females will be compared to analyze within and between population fitness.
Evolving in an unpredictable world - the E. coli story
Author(s): Karve, S, Daniel, S, Chavan, Y, Anand, A, Dey, S
Effects of temporally fluctuating environments on the fitness of populations are less explored compared to the effects of directional selection regimes. Existing studies on temporally fluctuating environments employ a narrow selection regime i.e. mostly one environmental parameter fluctuating predictably between two limit points and the fitness measurements happen in the environmental backgrounds similar to that of selection. Obviously such studies fail to predict the fitness outcomes in the complex and/or novel, unpredictable environments. We select replicate microbial populations under randomly fluctuating complex, stressful environmental regime. When fitness proxies of these selected lines are compared with the control populations grown in benign environments, under multiple novel environmental backgrounds, selected lines display ‘Generalist’ properties. We further characterize these Generalists by fitness proxy measurements at different time intervals during the growth in the novel environments. We see that acclimation is beneficial in both selected and control populations but consistently more advantageous in the populations with the history of randomly fluctuating environment. Characterization of the Generalists on the mechanistic level, shows that the commonly evoked explanations of evolution of hypermutators or modified permeability or carry over plasticity are not sufficient to explain observed Generalist phenomenon.
Experimental evolution for growth rate and its implications for infection success, co-infection dynamics, and virulence in a trypanosome parasite of bumblebees
Author(s): Marxer, M
Host-parasite interactions and their outcomes are strongly affected by several factors such as host and parasite genotypes and environmental conditions. Selection on basic growth properties in parasites may have far reaching consequences for numerous parasite traits, infection outcome and importantly the consequences for host fitness. It is know that strains of the trypanosome parasite of bumblebees Crithidia bombi have widely varying growth rates when cultured in vitro. We aim to experimentally evolve this parasite in vitro selecting for fast and slow growing sub-lines. This will enable us to investigate the costs, benefits and fitness trade-offs related to parasite growth rate by subsequently measuring in vivo infection profiles, transmission, and competitive ability under co-infection. To our knowledge, it would be the first time ecological trypanosome isolates have been experimentally evolved in vitro. These results will help define the fitness consequences for the observed natural variation in C. bombi growth, and will also inform important aspects of host-parasite evolution including the evolution of virulence.
Department of Ecology and Evolutionary Biology
Experimental evolution of an RNA virus in mixed host environments
Author(s): Morley, VJ, Usme Ciro, J, Wasik, BR, Turner, PE
Spatially heterogeneous environments pose unique challenges for evolving asexual populations. One possibility is that selection will lead to high genetic variance, where multiple subpopulations adapt to specialize on a subset of the available habitats. In contrast, spatial heterogeneity may lead to evolution of generalism, where the population is dominated by a single genotype with a broad niche. However, the evolution of populations in spatially heterogeneous environments has been rarely examined outside of theory, and to date there have been no experiments that test how rapidly evolving RNA virus populations adapt in the face of this challenge. Here, we allowed vesicular stomatitis virus (VSV) to evolve in replicated environments containing different mixtures of two host types in laboratory tissue culture: cancer-derived HeLa cells and non-cancerous BHK cells. After 25 passages (100 generations), the fitness (growth) of evolved viruses was assayed on each host type, relative to the common ancestor. We observed a correlated response to selection when VSV was evolved on pure cultures of either host (100% BHK or 100% HeLa). However, in mixed environments this correlated response was broken; populations evolved in spatially heterogeneous environments consistently improved on HeLa cells, even when HeLa cells were rare, but were much more variable in their improvement on BHK cells. Current work examines whole-genomics of evolved populations, to determine how simple versus spatially-complex environments affect genetic variation in VSV populations, and to compare/contrast trajectories of phenotypic and genetic change within and among treatments. This study demonstrates that simple versus mixed environments can pose fundamentally different challenges for adapting populations, and the need for theory that addresses how environmental complexity may influence adaptive trajectories.
Experimental evolution of heavy metal tolerance in changing environments
Author(s): Gorter, FA, Aarts, MMG, Zwaan, BJ, De Visser, JGM
In most long-term laboratory evolution experiments, organisms are exposed to a constant selection regime that initially causes a large reduction in fitness. However, the ecological relevance of this treatment may be questioned: under natural circumstances, environmental variables likely vary with time. We were interested in how the rate of directional environmental change affects the evolution of heavy metal tolerance in Saccharomyces cerevisiae. To this end, we grew replicate lines of yeast for 500 generations in the presence of (i) a constant high concentration of Cd, Ni or Zn or (ii) gradually increasing concentrations of these metals. We anticipated that these contrasting selection regimes would result in different adaptive dynamics and evolutionary endpoints, as the shape of the fitness landscape changes as a function of metal concentration. More specifically, we propose the following alternative scenarios: 1) the most resistant genotype is most fit at all metal concentrations, but strength of selection is proportional to concentration 2) the optimal genotype changes with concentration, such that the optimal genotype at intermediate concentrations will confer an intermediate level of tolerance. These scenarios predict that a gradual increase of metal concentration (as opposed to a constant high concentration) causes mutations of large effect to be fixed at later time points (scenario 1), only mutations of small or intermediate effect to be fixed (scenario 2) and, if the fitness landscape is rugged, evolutionary endpoints to be fitter and more diverse (both scenarios). Here, we present results from competition assays that were used to determine the relative fitness of evolved and ancestral isolates and thus differentiate between the alternative hypotheses. Although evolutionary dynamics differed between the treatments, evolutionary endpoints had a similar fitness, reflecting a smooth fitness landscape that changes as a function of metal concentration.