Aims and location The potential of pollen records in quantitative climate reconstructions has been widely debated but seldom tested. Our aim is to develop a pollen–climate transfer function for northern Europe and test its performance and inference power by numerical cross-validation with modern climate data. Annual mean temperature (Tann) was assessed as the critical climatic variable because Tann has a distinct south–north gradient (5.5 to −4.7 °C) in the study region with a corresponding zonal vegetation gradient from the hemiboreal zone in the south to the northern boreal zone in the north.
Methods We collected 137 pollen surface samples from small- to medium size lakes from southern Estonia to northern Finland. The transfer function for Tann was developed with weighted averaging partial least squares (WA-PLS) regression. All 102 terrestrial pollen and spore types were included in the calculation sum and all 137 surface samples and all 102 taxa were included in the transfer function. The performance of the WA-PLS transfer function was evaluated by leave-one-out cross-validation.
Results A cross-validated root mean square error of prediction (RMSEP) of our model is 0.89 °C and the coefficient of determination (r2) between the observed meteorological Tann values and those predicted by the model in leave-one-out cross-validation is 0.88. The RMSEP as a percentage of the gradient length of Tann is 8.8%. These figures indicate high performance statistics for our transfer function compared with other inference models. This is probably because of standardization of our surface-sampling and pollen-analytical procedures, careful selection of the surface sample sites with consideration of the relevant pollen source area, the simple patterns of vegetation zones and climate in the study area, and the mostly natural floristic composition of the forests in northern Europe. However, we also demonstrate the limitations of our model in reliably detecting fine-scale climatic variability.
Main conclusions The study shows the strong influence of Tann on modern pollen composition and demonstrates the potential of pollen data for long-term climate reconstructions in northern Europe. It also provides evidence against simple interpretations of fine-scale variations in a single climate reconstruction. In particular, our results highlight the importance of careful study design and implementation in the construction of pollen–climate transfer functions.