Mapping Ancient Forests: Bayesian Inference for Spatio-temporal Trends in Forest Composition Using the Fossil Pollen Proxy Record.

Ecologists use the relative abundance of fossil pollen in sediments to estimate how tree species abundances change over space and time. To predict historical forest composition and quantify the available information, we build a Bayesian hierarchical model of forest composition in central New England, USA, based on pollen in a network of ponds. The critical relationships between abundances of taxa in the pollen record and abundances as actual vegetation are estimated for the modern and colonial periods, for which both pollen and direct vegetation data are available, based on a latent multivariate spatial process representing forest composition. For time periods in the past with only pollen data, we use the estimated model parameters to constrain predictions about the latent spatio-temporal process conditional on the pollen data. We develop an innovative graphical assessment of feature significance to help to infer which spatial patterns are reliably estimated. The model allows us to estimate the spatial distribution and relative abundances of tree species over the last 2500 years, with an assessment of uncertainty, and to draw inference about how these patterns have changed over time. Cross-validation suggests that our feature significance approach can reliably indicate certain large-scale spatial features for many taxa, but that features on scales smaller than 50 km are difficult to distinguish, as are large-scale features for some taxa. We also use the model to quantitatively investigate ecological hypotheses, including covariate effects on taxa abundances and questions about pollen dispersal characteristics. The critical advantages of our modeling approach over current ecological analyses are the explicit spatio-temporal representation, quantification of abundance on the scale of trees rather than pollen, and uncertainty characterization.