Hierarchical and spatially aggregated plant cover data

Most plant species are spatially aggregated and here the importance of taking the spatial variation into account when analyzing plant cover data is demonstrated in a general stochastic model where both the within-site and the among-site spatial variation of species cover data are pararrieterized. Using a generalised binomial distribution (or Polya-Eggenberger distribution), where the among-site variation in mean cover is modeled by a zero-inflated beta distribution, it is possible to adequately analyze hierarchical plant cover data and link the estimates to the underlying ecological processes. The model is demonstrated in a case-study of pin-point cover data of Erica tetralix from 1148 wet heathland plots at 84 Danish sites, and it is shown that both parameter estimates and the conclusions of hypotheses testing critically depend on the correct modeling of the observed spatial variation. Finally, statistical power simulations of plant cover measurements are presented, which will be useful for planning ecological experiments and monitoring programs. (C) 2013 Elsevier B.V. All rights reserved.