VAR-models for Predicting Biodiversity

Abstract

The aim of this work is to introduce a methodology for monitoring diversity of a biological population by using the tool of diversity profiles. In particular, we develop a theory for studying the diversity profile along the temporal axis. We propose forecasting techniques for diversity profile by using the VAR model on multivariate time series of abundance vector and ARMA models both on univariate time series of abundances and on the aggregated series of diversity profiles. A Monte Carlo simulation is performed in order to test the goodness of forecasts obtained applying the three different methods proposed.