Date of Award

Spring 2022

Document Type


Degree Name

Doctor of Philosophy (PhD)


Ecology and Evolutionary Biology

First Advisor

Sanchez, Alvaro


All organisms construct the shared environment in which they live. This is especially notable in micro-organisms as they secrete and uptake a diverse range of metabolites depending on their genotype and environment. Over the past few decades, systems biologists have developed computational tools to predict the nutrient uptake and secretions of microbes across environments, using metabolic networks inferred from whole-genome sequencing. These tools provide an opportunity to quantify eco-evolutionary dynamics at the genomic level, by combining genome-scale metabolic models mapping genotype to phenotype, with consumer-resource models predicting population dynamics from phenotype. By leveraging these new computational approaches and combining them with experiments using microbial communities in synthetic environments, this dissertation will quantify the impact of metabolite production and consumption on the evolutionary and ecological dynamics of multi-species microbial communities. In Chapter 1, I present a published paper in which I address how niche construction quantitatively determines evolutionary trajectories by deforming the fitness landscape of evolving populations. The chapter uses a combination of genome-scale metabolic modelling and experiments to systematically quantify the deformability of the E.coli metabolic fitness landscape. It shows that the effects of niche construction are quantitatively modest at short genomic scales but accumulate over longer evolutionary trajectories. These results suggest that fitness landscapes can predict evolution over short mutational distances, but that niche construction hampers predictability in the long term. In Chapter 2 I present a published paper in which I ask whether communities assembling in the same metabolic environment show similar ecological interactions. This chapters leverages previously published 16s rRNA sequencing data from an experiment in which complex-microbial communities were allowed to self assemble in laboratory environments containing a single limiting resource. I benchmark a newly developed statistical tool, Dissimilarity-Overlap Analysis, and use it to determine whether interaction parameters are similar across communities assembled in the same metabolic environment. I find a negative relationship between dissimilarity and overlap which is what we expect if interactions are strongly convergent. However, even in replicate, identical habitats, two different communities may contain the same set of taxa at different abundances in equilibrium. The formation of alternative states in community assembly is strongly associated with the presence of specific taxa suggesting that some taxa may differ in the niches they construct and occupy even across replicate abiotic conditions. In Chapter 3 I present a published paper which asks how different components of the environment interact to collectively determine the taxonomic composition of microbial communities. This paper tests whether the composition of communities assembled in a pair of carbon sources could be predicted from those assembled in each single carbon source alone. This paper develops a null-additive model and show that it can explain a high variation of the relative abundance of families in communities assembled in pairs of carbon sources. Deviation from this additive model reveal a characteristic pattern with sugars 'dominating' organic acids. Using consumer-resource modelling, I show that nutrient dominance can be explained by experimentally validated asymmetries in the family level specialisation on different resource types. Quantifying the asymmetric effect of metabolites on community composition is a key step towards engineering microbial communities by modulating nutrient composition. In Chapter 4, I present a draft manuscript in which I ask whether one can predict the composition of microbial communities assembling in different metabolic environments. I first use a combination of enrichment experiments, metabolomics and phenotypic assays to show that the predictability of community assembly depends on the phylogenetic distribution of quantitative metabolic traits selected for by different environments. This includes traits determining both the ability to exploit the supplied resource and the ability to grow on the constructed niches. I find that similarities in community composition across environments reflect correlations in conserved metabolic traits, which are predictable using metabolic models.Finally I show how one can use metabolic models to quantitatively predict the effect of novel environmental perturbations on microbial communities. The work presented herein illustrates how genome-scale models can be combined with analytical models of population dynamics to develop quantitative and predictive eco-evolutionary theory. Whilst focusing on microbial communities, the concepts developed are applicable to other cellular populations as well as to macro-organism engaging in niche-constructing activities.By quantifying the effects of niche construction in an explicit manner, the work I have presented moves beyond semantic arguments and descriptive studies towards a predictive and mechanistic understanding of eco-evolutionary dynamics.