Modelling dispersal and ecological competition in a statistical phylogeographic framework
Louis Ranjard, David Welch, Marie Paturel and Stéphane Guindon, Department of Statistics
Developing a new model for reconstructing colonisation history of habitats
Competition between organisms influences the processes governing the colonisation of new habitats. As a consequence, species or populations arriving first at a suitable location may prevent secondary colonisation. While adaptation to environmental variables (e.g., temperature, altitude, etc.) is essential, the presence or absence of certain species at a particular location often depends on whether or not competing species co-occur. Despite the potential of competition to maintain populations in isolation, past quantitative analyses of evolution have largely ignored that factor because of technical difficulties.
We have developed a model that integrates competition along with dispersal into a Bayesian phylogeographic framework. The proposed model takes genetic sequences and their spatial coordinates as input in order to date dispersal events and estimate dispersal and competition parameters.
Simulating landscape colonisation
Our statistical modelling on Pan uses the R software and high throughput is obtained by using parallel processing. In the first step to get an accurate understanding of how our approach performs, an extensive test on the NeSI Pan cluster was done by (i) performing simulations of organisms colonising space and competing while evolving and (ii) estimating the competitive and dispersal parameters from the genetic diversity generated by these simulations (Figure 1).
Figure 1: Results from simulations shows that our approach can successfully detect the influence of competition on the history of colonisation of space inferred from molecular sequences. Each dot on the graph corresponds to a different simulation instance performed on the Pan cluster.
In the second step, Pan was used to analyse a dataset of Hawaiian endemic species of the Banza genus called Hawaiian katydids –grasshopper-like insects. The genetic relationships among these organisms show evidence for island radiation with different species inhabiting the Hawaiian archipelago.
The phylogenetic tree built from mitochondrial sequences exhibits a striking “progression rule” pattern whereby each island appears to have been colonised only once. No obvious environmental feature varying across islands explains the observed geographical distribution of these organisms. Therefore, adaptation does not seem to be the factor causing the absence of secondary colonisaton.
We fitted our model to a rooted phylogenetic tree with node heights expressed in calendar time units (Figure 2). Using Pan we could demonstrate that competition can be detected with high sensitivity and specificity from the analysis of genetic variations in space.
In the next step, we are planning on extending our model to integrate extra environmental parameters that may have affected the colonisation of space. The NeSI Pan cluster will be used similarly to investigate the characteristics of the new model and how it can be applied to any biogeographical datasets for which genetic data is available.
