The time is ripe for developing evolutionary biology into a more predictive science. Progress has been made in predicting short-term evolution under carefully controlled circumstances in prokaryotes. However, societal challenges such as climate change, environmental degradation and pesticide resistance require prediction of short-term evolutionary responses in complex multicellular eukaryotes. This project will bring together theoretical, empirical scientists as well as those working on digital and robotic evolution to explore avenues for predicting evolution from first principles.
The project is a joint effort of a large number of research groups in the Netherlands and Belgium, contributing experimental work as well as theoretical expertise to the project. Principal investigators are Meike Wortel and Ken Kraaijeveld, based at the Astrid Groot group at the University of Amsterdam.
The project ‘Predicting evolution’ will be complemented by a few relatively small projects, that together should supply information and context to the main project. The following projects have been selected and will all be executed in 2019.
Judith Risse (NIOO): Automated phenotyping of great tit bills
Predicting evolution requires understanding a detailed understanding of the evolutionary processes involved. These processes are governed by the complex interplay between individuals, populations and environment. Understanding these in turn requires large volumes of detailed data on genotype, phenotype and environment. Whereas acquisition of genotypic and environmental data on a large scales is already commonplace and relatively simple, phenotyping is still a labor intensive and largely manual task. In this project we aim to explore methods to automate phenotyping. The bill is the most important tool of a bird and it is used for feeding and preening, nest building, and to get a mate.
Bill size is a variable polygenic trait with fitness consequences. Aim of our ongoing research is to understand the variation in bill length in Dutch great tit (Parus major) populations as well as the underlying molecular processes for selection to act on. This requires accurate phenotypes for a large number of birds. Using a dedicated camera system in a portable box we photograph bills in a standardised way. In collaboration with track32.nl we are developing a software tool to facilitate and automate the measurement of bill length and depth using image recognition and machine learning techniques.
Ken Kraaijeveld (UvA), Jeroen van Zon (AMOLF), Meike Wortel (UvA): Predicting evolution from detailed knowledge of the biological system
Is it possible to predict the course of evolution? Evolution happens because some individuals are more fit than others. Fit individuals will tend to leave more offspring than unfit individuals and these offspring will inherit many of the fitness-enhancing characteristics of their parents. If we know what makes an individual fit or unfit in a given environment, it should be possible to predict how fitness improves during the course of evolution. In this project, we will start with a very unfit mutant of the nematode C. elegans and let it evolve. We know exactly why the nematodes are unfit and we already know a mutations that improves fitness (from artificial mutant screens). Will evolution ‘find’ this same solution? Or will there be other solutions to the same problem? The answers to these questions will tell us whether we are able to predict evolution in such a well-studied organism, with the knowledge we have at the moment.
Dré Kampfraath (VU), M. Bosse (WUR & VU), Ken Kraaijeveld (UvA), J. Ellers (VU): Predicting gene decay underlying loss of male reproductive traits under asexuality
Evolutionary theory predicts that traits that are no longer under selection will decay either through mutation accumulation or negative selection. When species shift to asexual reproduction male sexual traits become redundant and are therefore predicted to decay. Recently, we established that male sexual phenotypic traits indeed have decayed in the asexual lines of the springtail Folsomia candida. The degradation of their phenotypic traits proceeded at variable rates within the different lines, suggesting mutation accumulation to be the main evolutionary process underlying this deterioration. Our goal is to link this phenotypic degradation to the presumed genotypic degradation. Mutation accumulation proceeds by drift and neutral processes, genes decay in this manner are therefore expected to have an equal ratio of synonymous and non-synonymous mutations. On the other hand, genes under negative selection will have a higher proportion of non-synonymous mutations or get lost from the genome entirely. By analysing the genomes of sexual and asexual F. candida strains we can test whether these predictions hold for the genes underlying the degradation of phenotypic traits.
Meike Wortel (UvA), Arjan de Visser (WUR), Paulien Hogeweg (UU), Jan Kammenga (WUR), Ken Kraaijeveld (UvA): Predicting evolution in a biotic environment - Experimental evolution with nematode predators and bacterial prey
Species evolve, and become better adapted to their environment. When this environment consists of other species, these might evolve in response, resulting in co-evolution. Co-evolution in predator-prey systems has been studied extensively, however, it is largely unknown whether co-evolution makes evolution more or less repeatable. We hypothesize that two mechanisms affect the repeatability of prey evolution in opposite ways: the increased selection pressure resulting from a co-evolving predator will increase repeatability, while random variation in the evolution of the predator will decrease repeatability.
We use an experimental system of a nematode predator and bacterial prey. Both organisms can be frozen and revived, which allows us to design an experiment where either both species may evolve (co-evolution) or only the prey evolves (no co-evolution). We will test whether co-evolution increases or decreases repeatability of prey evolution, by comparing the change in phenotype and genotype of different replicas for both treatments. We will complement this experimental approach with a computational approach using evolution of digital organisms, namely Virtual Microbes, to which we will add a simplified predator. We will use the computational approach to decouple the two mechanisms described above, and test which of the two hypothesized opposite effects has larger impact. Moreover, we will use the computational approach to estimate how general our results are and to inform experiments by simulating different conditions and generating new hypotheses.