A matter of dispersal: REVEALSinR introduces state-of-the-art dispersal models to quantitative vegetation reconstruction

Abstract

The REVEALS model is applied in quantitative vegetation reconstruction to translate pollen percentage data from large lakes and peatlands into regional vegetation composition. The model was first presented in 2007 and has gained increasing attention. It is a core element of the Landcover 6k initiative within the PAGES project. The REVEALS model has two critical components: the pollen dispersal model and pollen productivity estimates (PPEs). To study the consequences of model settings, we implemented REVEALS in R. We use a state-of-the-art Lagrangian stochastic dispersal model (LSM) and compare model outcomes with calculations based on a conventional Gaussian plume dispersal model (GPM). In the LSM turbulence causes pollen fall speed to have little effect on the dispersal pattern whereas fall speed is a major factor in the GPM. Dispersal models are also used to derive PPEs. The unrealistic GPM produces PPEs that do not describe actual pollen productivity, but rather serve as a basin specific correction factor. A test with pollen and vegetation data from NE Germany shows that REVEALS performs best when applied with the LSM. REVEALS applications with the GPM can produce realistic results, but only if unrealistic PPEs are used. We discuss the derivation of PPEs and further REVEALS applications. Our REVEALS implementation is freely available as the ‘REVEALSinR’ function within the R package DISQOVER. REVEALSinR offers an environment for experimentation and analysing model sensitivities. We encourage further experiments and welcome comments on our tool.