|Saturday, February 17|
|CS20 Understanding Populations||
Sat, Feb 17, 9:15 AM - 10:45 AM
Approaches and Techniques for Estimating the Total Number of Species in a Population, with Emphasis on Application to Mineral Species (303540)
Keywords: statistical mineralogy, Bayesian statistics, species estimation, Markov chain Monte Carlo methods
In statistical mineralogy, extensive data resources are employed on mineral species to model the frequency distribution of Earth’s mineral kingdom to facilitate prediction of Earth’s total, but as yet undiscovered, mineralogical diversity. The challenge of estimating the number of biological species in an ecological population has been studied for decades since pioneering work by Fisher et al. (1943). Ecology is concerned with estimation of the sizes of type-rich populations. In linguistics, one uses models that estimate an author’s vocabulary size and idiosyncrasies. One approach for species estimation is to use parametric models based on a mixed-Poisson distribution fitted to the observed frequencies, where several choices exists for the abundance distribution. The main focus here will be to estimate the number of species in a Bayesian framework for the mixed-Poisson likelihood using MCMC methods. The implementation of the code in the statistical software package, R, will be demonstrated and open to audience questions. The presentation is intended across a wide areas of disciplines needing advanced applied statistics to estimate the size of a population.