Theory of quantitative reconstruction of vegetation I: pollen from large sites REVEALS regional vegetation composition

Abstract

Quantitative reconstruction of past vegetation is one of the primary goals in Quaternary palynology and palaeoecology but still remains difficult. This paper proposes a model, REVEALS, that estimates regional vegetation composition using pollen from ‘large lakes’ that have small site-to-site variations of pollen assemblages even if vegetation is highly heterogeneous. Once these data have been used to quantify regional vegetation composition within 10 4 -10 5 km 2, background pollen, one of the parameters crucial for vegetation reconstruction, can be estimated for smaller-sized sites, and incorporated into the Landscape Reconstruction Algorithm (LRA), a multistep framework for quantitative reconstruction of vegetation in smaller areas (≤ 10 4 ha). Simulations using the POLLSCAPE modelling show that REVEALS can provide accurate estimates of regional vegetation composition in various landscapes and under different atmospheric conditions. If pollen assemblages from lakes that are much smaller than ‘large lakes’ are used, estimates of regional vegetation at individual sites could be significantly different from the expected values, and their site-to-site variation could be large. However, when pollen data from multiple lakes ≥ 100-500 ha in size are available, REVEALS can provide accurate estimates of the regional vegetation with relatively small standard errors. Quantitative reconstruction of regional landscape and vegetation change will be critical for testing some of the controversial hypotheses and concepts in global change and conservation research, such as the impacts of agricultural activities on global climate over the last 8000 years and the open-woodland hypothesis in northern Europe in the early Holocene.