Estimating a fundamental niche is assumed to require physiological experiments, which are seldom performed in more than one dimension. However, the data that is widely available are occurrence points, a subset of the fundamental niche. Our work focuses on finding an approximation to the fundamental niche from occurrences and partial physiological data. We do this by: a) postulating a simple shape for the physiologically defined niche (Hutchinson’s fundamental niche), b) using prior information from experiments for the estimation of the parameters that represent the niche, and c) considering how the actual observables are conditioned on the set of existing environmental conditions. We provide a probabilistic (i.e. Bayesian) statement of the problem and the posterior distribution for estimating the parameters of fundamental niches. We created some examples of virtual species for which the resulting estimates were compared to its theoretical fundamental niche. Finally, we illustrate the application of our method through two case studies.
Results/Conclusions
A Bayesian model is proposed to estimate fundamental niches from occurrences and physiological data. The analysis of data from virtual species helped us describing conditions that may allow us to obtain better estimates for the fundamental niche. For these examples, the estimated fundamental niche coincided with the theoretical fundamental niche in up to 75%, when the occurrence points of the species (in environmental space) were placed in a dense region of the existing environmental space. Therefore, existing environments strongly affect samples of observations and the quality of the estimation. Using real species adds a number of decisions and complications about whether physiological details matter, and in what ways. However, it is only using real data that the validity and usefulness of the insights derived from virtual species can be assessed.