Uncertainty surfaces and maps of ignorance: the possibility of spatially estimating dark diversity

In ecology, a number of studies have dealt with the prediction of species
\ndiversity over space and its changes over time based on a set of predictors
\nrelated to environmental variability, productivity, spatial constraints, and
\nclimate drivers. However, the observed diversity is a portion of the actual
\npool which is strictly related to abiotic conditions and evolutionary history
\nof species in the pool. In this study we aim to explicitly show uncertainty
\nof the prediction of species distribution at a global scale. This is in line
\nwith the “dark diversity” concept extended to a global spatial scale. We
\nwill not deal with problems in the detectability of species but with hidden
\npatterns in the probability of their distribution.
\nThus far, species distribution estimates based on field data sampling do
\nnot represent reality in a deterministic sense and are only estimates of
\npotential presence. Therefore, the use of “maps of ignorance” representing
\nthe bias or the uncertainty deriving from species distribution modeling,
\nalong with predictive maps, is strongly encouraged.
\nUncertainty can derive from a number of input data sources, such as the
\ndefinition or identification of a certain species, as well as location-based
\nerrors. The spatial distribution of uncertainty should explicitly be shown
\non maps to avoid ignoring overall accuracy or model errors.
\nWe propose methods mainly based on Bayesian logistic regression coupled
\nwith simulation-based Monte Carlo techniques and Cartograms applied to
\nEuropean and worldwide datasets for explicitly mapping uncertainty in
\nthe distribution of species in a Free and Open Source environment.