The use of rarefaction analysis for estimating palynological richness from Quaternary pollen-analytical data

Reconstructing temporal changes in diversity from pollen assemblages is potentially important both palaeoecologically and ecologically because community diversity may, in part, result from historical processes. The use of diversity indices such as Shannon's information index or Simpson's index is not appropriate with pollen percentage data because such indices consider both the numbers of different taxa and their relative frequencies or representation. The latter aspect in pollen data is inevitably influenced by inherent differences in pollen production and dispersal. The total number of taxa present in a sample is a robust and useful measure of palynological richness if. and only if, all the pollen counts are standardized to a fixed number of grains. Rarefaction analysis implements such a standardization and provides minimum variance unbiased estimates of the expected number of taxa (I) in a random sample of n individuals taken from a larger collection of N individuals containing T taxa. The underlying mathematical theory of rarefaction analysis and its important biological and palaeoecological assumptions are discussed. The use of rarefaction analysis is illustrated with three data-sets: Crose Mere. central England (0-c, 12 500 BP); Abernethy Forest, eastern Scotland (5500-12 100 BP); and three sites (Lochs Ashik. Cleat, Meodal) on the Isle of Skye. western Scotland, all covering the last 10 500 years. Palynological richness, as estimated by rarefaction analysis, is high in the protocratic phase (c. 9500-12 500 BP). low in the mesocratic phase (c. 5500-9500 BP). low in the oligocratic phase (0-c. 5500 BP), and high in the Homo sapiens phase (0-c. 5000 BP) of the Holocene. Although factors such as local site characteristics and pollen production, dispersal. and input may influence temporal changes in richness, changes in palynological richness are interpreted as reflecting predominantly the changing floristic richness of the vegetation types in the pollen-source area of a lake and the changing mosaic structure of the landscape through time. Intermediate levels of disturbance, either natural in the protocratic phase or anthropogenic in the Homo sapiens phase appear to be important in maximizing richness at the landscape scale by preventing the dominance of any single component but insufficient to cause extinction of components at the landscape scale.