Pollen-based reconstruction of regional vegetation and land cover in the Northern Hemisphere

Data on past vegetation compositions is crucial not only for understanding past vegetation dynamics and environmental interactions but also for predicting potential future vegetation trajectories and thus their feedback on climate and society. Pollen records from sediment cores provide temporally resolved data on pollen frequencies. These allow for inferences of taxa presence, but biased pollen deposition due to taxa-specific pollen productivity and dispersal prohibit direct inference of taxa abundance.

The model for Regional Estimates of Vegetation Abundance from Large Sites (REVEALS) corrects for these taxa-specific parameters and produces more realistic regional vegetation abundances. Previously applied in many regions such as North America, Southern Sweden, Norway, and more recently the entirety of Europe, REVEALS has performed well in providing estimates for large vegetation units as well as individual taxa. 

With this data set, we present reconstructed past regional vegetation for more than 2200 sites across the Northern Hemisphere. The REVEALS model was applied by using a harmonized pollen dataset for the entire Northern Hemisphere, taxa-specific pollen productivity estimates, pollen fall speeds, as well as pollen dispersal models. First validations show an improved fit of the reconstructed vegetation with remotely sensed tree cover compared to pure pollen percentages. For validation tree cover datasets from the CONSENSUS global 1-km land cover product were used. The pollen source areas were defined to include 80% of the area from which deposited pollen originated.

This first-time reconstruction for the entire Northern Hemisphere will allow for detailed analysis of vegetation dynamics and trajectories ultimately improving our understanding of climate-vegetation interactions, and may even act as input and validation for other vegetation and climate models and proxies.