Continental‐scale modern pollen‐climate data repositories are a primary data source for palaeoclimate reconstructions. However, these repositories can contain artefacts, such as records from different depositional environments and replicate records, that can influence the observed pollen‐climate relationships as well as the palaeoclimate reconstructions derived from these relationships. Because the depositional environment determines the relative strength of the local and regional pollen signals, combining data from different depositional environments results in pollen abundances that can be influenced by the local pollen signal. Replicate records in pollen‐climate datasets can skew pollen‐climate relationships by causing an overweighting of pollen abundances in a climate space. In this paper, we address the issues related to these artefacts as we define the methods used to create a research dataset from the North American Modern Pollen Database (NAMPD) (Whitmore et al., 2005). The research dataset we present consists of 2 613 records in eastern North America, of which 70.9% are lacustrine sites. The research dataset encompasses the majority of the temperature and Summer precipitation ranges of the NAMPD's climatic range, and 40% of its Winter precipitation range. We present our analyses of the research dataset and demonstrate that it improves upon regional pollen‐climate relationship modelling of eastern North American taxa.
Read full abstract