Uncertainty in species diversity mapping: unravelling a long lasting theme

Many geospatial tools have been advocated in spatial ecology and biogeography to estimate biodiversity and its changes over space and time. Such information is essential in designing effective strategies for biodiversity conservation and management. Remote sensing is one of the most powerful approaches to identify biodiversity hotspots and predict changes in species composition in reduced time and costs. This is because, with respect to field-based methods, it allows to derive complete spatial coverages of the Earth surface under study in a short period of time. Furthermore, remote sensing provides repeated coverages of field sites, thus making studies of temporal changes in biodiversity possible. Thus far, species diversity estimates from remote sensing have rarely taken into account uncertainty in an explicit manner. On the contrary, the spatial distribution of uncertainty should explicitly be shown on maps to avoid ignoring overall accuracy or model errors. In this talk I will discuss, from a conceptual point of view, the potential of remote sensing in estimating biodiversity using various diversity indices. I will also face challenges in the representation of uncertainty methods mainly based on Bayesian logistic regression coupled with simulation-based Monte Carlo techniques and Cartograms applied to European and worldwide datasets for explicitly mapping uncertainty in the distribution of species diversity in a Free and Open Source environment.