Pollen-landscape relationships in modern analogues of ancient cultural landscapes in southern Sweden ? a first step towards quantification of vegetation openness in the past

Abstract

This study aims to analyse how vegetation, and in particular the degree of openness of the landscape, is reflected in pollen assemblages from surface sediment in lakes. Modern analogues of ancient cultural landscapes in southern Sweden were selected. Surface sediments from 22 small lakes (0.5–20 ha) located mainly in the forest region of southern Sweden were collected and analysed for pollen in order to enlarge and complement an earlier data set of 13 lakes collected in the open, agricultural region of southernmost Sweden. The composition of the landscape surrounding the lakes was mapped within 1000-m and 500-m radii around the lakes using Colour InfraRed (CIR) aerial photographs. The pollen and landscape data were analysed using numerical ordination techniques. The results show that, despite the large variation of landscape openness, the variation in non-arboreal pollen (NAP) is low between the sites which was not the case for the 13 lakes of the previous study. It is hypothesised that this may be due to differences in the major characteristics of the two regions in which the sites were selected, i.e. mainly treeless and intensively farmed in the previous study and mainly forested in the present investigation. The difference in background pollen appears to play a decisive role on the relative representation of NAP. This implies that the background pollen should be estimated before NAP percentages can be used for quantitative reconstruction of past landscape openness. In the 22 lakes studied, Gramineae, Cerealia (excludingSecale),Filipendula andSalix are positively correlated to cultivated land within both radii, and with open land (tree cover not exceeding 20%) within the 1000-m radius.Quercus andFagus have some positive correlation with deciduous orest within 1000-m radius. We conclude that the landscape units cultivated land, open land and deciduous forest within 1000-m radius are reasonably well reflected in the pollen assemblages and could be predicted within this area.