Relative pollen productivity and fall speed estimates for southern African savanna taxa

Understanding the characteristics of pollen dispersal and deposition of different plant taxa is crucial to accurately reconstructing past landscapes using fossil pollen data. Quantitative reconstruction of past landscapes from pollen data using the Prentice-Sugita approach requires estimates of fall speed and relative pollen productivity for all taxa modelled. This study presents estimates of pollen productivity and fall speeds for key southern African savanna taxa, providing a basis for the improved interpretation of fossil pollen records from this extensive and heterogeneous biome. The work was carried out in 5 steps. (1) Modern pollen assemblages from 34 surface sediment samples were analysed. (2) Vegetation around each sampling site was surveyed in concentric circles to a radius of 50 m, and data from existing park surveys were analysed to extend the survey distance to 5 km. (3) Fall speeds for the main pollen taxa were estimated using Stoke's Law of particle settling velocity. (4) Vegetation data were weighted using three different distance-weightings, one incorporating the different particle fall speeds. (5) Extended R-Value analysis was carried out on the pollen and distance-weighted plant abundance datasets using HUMPOL software to estimate relevant source area and relative pollen productivity for the main pollen taxa present. Results showed the Relevant Source Area of Pollen surrounding the sites to be 600-900 m radius, and Poaceae/Cyperaceae were found to be twice as productive (PPE 2.03) as the arboreal taxa analysed (PPE 0.50-0.99). The problems encountered in calculating pollen productivity estimates in savanna environments are discussed and improvements for future studies are suggested.