Pollen-based biome prediction dataset for palaeoenvironmental reconstruction

Harmonised pollen dataset used to train and evaluate machine-learning models for biome prediction in palaeoenvironmental reconstruction. The dataset contains 41,831 records from 547 site identifiers, geographic and chronological metadata, three biome target variables and pollen taxon counts. Source databases, schema, label definitions and reuse requirements are documented in the accompanying files. The data were retrieved from Neotoma: Williams, J.W., Grimm, E.G., Blois, J., Charles, D.F., Davis, E., Goring, S.J., Graham, R., Smith, A.J., Anderson, M., Arroyo-Cabrales, J., Ashworth, A.C., Betancourt, J.L., Bills, B.W., Booth, R.K., Buckland, P., Curry, B., Giesecke, T., Hausmann, S., Jackson, S.T., Latorre, C., Nichols, J., Purdum, T., Roth, R.E., Stryker, M., Takahara, H., 2018. The Neotoma Paleoecology Database: A multi-proxy, international community-curated data resource. Quaternary Research 89, 156-177.