University of Glasgow
Biodiversity loss has reached critical levels in the past decade, just as computational techniques and available data sources have put us in a position to begin to quantify this reduction. The Global Biodiversity Information Facility (GBIF) holds hundreds of millions of records of plant species and where they are found. The European Centre for Medium-range Weather Forecasting (ECMWF) has reconstructed the climate of the whole earth since 1900, allowing us to understand the environment in which each plant was discovered. Other researchers, who have constructed a supertree of over thirty thousand of these plant species, give us the opportunity to investigate the evolutionary history of climate preferences among related species. Drawing these datasets together is a huge computational challenge, one exacerbated by our interest in simulating the potential changes these species will undergo in the face of sustained climate change. The Julia language has provided us with the opportunity to work with these huge datasets (JuliaDB.jl, GDAL.jl), and make spatial simulations based upon endangered plant species on a global scale, which would be intractable in languages commonly used in the life sciences, like R, Python etc. We have been building this platform for almost two years now, along with several other components that come with it (especially Diversity.jl, a package for the measurement and partitioning of biodiversity). We will talk about the results we have generated in understanding the evolutionary relationships between species across the whole kingdom of flowering plants, how our simulations work to predict responses of plants species to climate change, and how well we are likely to be able to detect these changes using existing biodiversity metrics. We have also attempted to base our work on Julia “ecosystems” like JuliaStats, JuliaArrays, EcoJulia, BioJulia, JuliaPlots and JuliaGeo, as well as many other individual packages (e.g. JuliaDB.jl, Unitful.jl), and we will discuss the advantages to and difficulties of using and integrating across such systems.
