The impact of the genetic architecture of dispersal and local adaptation on predicting range expansions

Abstract

Predicting range expansions and biological invasions is a central challenge in global change research. While already non-trivial in a purely ecological context, this task becomes truly challenging when rapid evolution comes into play. Importantly, range expansions and concurrent evolutionary change happen on similar timescales. Therefore, one must take the speed of evolution, that is, evolvability, into account, as well as potential evolutionary changes in the speed of evolution itself. This can be done mechanistically by considering the genetic architecture of traits of interest or their genotype-to-phenotype map. Realistically, a genotype-to-phenotype map can have multiple genotypes encoding the same phenotype, or the same genotype can produce different phenotypes depending on the environment (plasticity). One candidate model which has these properties is Wagner's gene-regulatory network (GRN) model. Here, I develop individual-based metapopulation models of range expansion into (i) a homogeneous environment and (ii) an abiotic gradient. For a homogeneous environment, I study the impact of the assumed genetic architecture of constitutive and plastic (specifically, density-dependent) dispersal on the speed of range expansions. I compare the resulting range expansion speeds for a model in which constitutive dispersal is encoded by a GRN to existing models of dispersal evolution that assume a single locus of large effect or multiple loci of small effect. Further, in the case of density-dependent dispersal, I compare the plastic response and the consequent range expansion speeds of the GRN model, which takes local population density as an input, to a classical reaction norm approach. I find that the GRN models for both constitutive dispersal and density-dependent dispersal predict faster range expansion speeds than the alternative genetic architectures and the reaction norm model, respectively, depending on the strength of selection on dispersal. I then extend the model to include separate GRNs for local adaptation to an external environmental gradient and constitutive dispersal. I show that this model leads to accelerated and, overall, faster range expansions than a model in which one locus each encodes dispersal and local adaptation, resulting from an increasing rate of adaptation due to increased sensitivity to mutation in the local adaptation GRN. My results highlight the importance of considering the genetic architecture of traits, especially under conditions of rapid ecological change.