Estimating source area of pollen and pollen productivity in the cultural landscapes of southern Sweden - developing a palynological tool for quantifying past plant cover

Abstract: Fossil pollen records retrieved from peat and lake sediments have great potential for quantifying past plant cover. This thesis is a contribution to the development of a palynological interpretation tool for reconstructing past cultural landscapes in terms of plant abundance and distribution. The aims of this thesis are (1) to explore pollen-vegetation relationships in the ancient, traditional cultural landscape of southern Sweden, (2) to test the potentials and limits of using mechanistic models of pollen dispersal and deposition to reconstruct past cultural landscapes from fossil pollen assemblages, (3) to estimate pollen productivity for the most common plant taxa characteristic of the past cultural landscapes of southern Sweden, and (4) to define and estimate the spatial scale of the cultural landscape reflected by pollen assemblages in moss polsters and small lakes. Data collection included sampling moss polsters and lake sediments for pollen and making vegetation surveys. The compilation of vegetation data was performed using GIS software. Pollen/vegetation relationships were explored using numerical techniques (PCA, RDA, Monte Carlo permutation test). Pollen productivity was estimated using Extended R-Value (ERV) -models. These models imply that vegetation is distance-weighted, which required the development of an appropriate design of vegetation survey in the field. The spatial scale of the landscape reflected by pollen assemblages is defined as the area beyond which the correlation of pollen and vegetation does not improve, and is named "relevant source area of pollen" (RSAP). RSAP was estimated using the maximum likelihood method. For hypothesis testing, a computer simulation model (POLLSCAPE) developed for forest vegetation was used to simulate pollen dispersal and deposition in cultural landscapes. The empirical pollen/vegetation data shows a major difference between the two contrasting vegetation regions of southern Sweden in terms of the relationship between NAP (herb pollen) percentages and open-land percentage cover. The relationship between NAP % and open-land % cover was also simulated using POLLSCAPE. The simulated and empirical relationships are very similar, which demonstrates that POLLSCAPE can be used for semi-open and open landscapes. Pollen productivity estimates (PPE) were obtained for 25 herb and tree taxa. RSAP for small lakes and moss polsters was estimated to be 800-1000 m and 400 m, respectively. Empirical data and simulations show that random site selection is required in order to obtain reliable estimates of RSAP, and that the spatial resolution of the vegetation survey close to the pollen sample, i.e. the size distribution of the vegetation mosaic, plays a critical role in RSAP estimates. Major implications of the results are (1) the cover of open land is likely to be strongly underestimated on the basis of NAP percentages alone in a mosaic semi-open landscape, (2) provided that past and present landscapes have a similar vegetation structure, a quantitative reconstruction of the past landscape would be possible within 1000 m radius around small lakes and 400 m radius around soil profiles, (3) the availability of PPE for critical herb taxa and the demonstration that POLLSCAPE may be used to study open and semi-open landscapes provide a reliable tool for proceeding with the study of pollen dispersal and deposition in cultural landscapes.