Challenges in estimating past plant diversity from fossil pollen data: statistical assessment, problems, and possible solutions

Abstract

ABSTRACTFossil pollen data from sediment cores may be used as a measure for past plant diversity. According to the theory of probability, palynological richness is positively related to the pollen count. In a low pollen count, only common taxa are detected, whereas rare taxa are only detected by chance. The detection of all pollen taxa requires a very high pollen count, which is time‐consuming. In regular palynological investigations, the detected richness in pollen spectra varies with the pollen count. Rarefaction analysis estimates palynological richness in an exactly equal‐sum count for all samples, so that comparison between samples is meaningful. However, the over‐representation of some taxa suppresses the detection probability of rare taxa; low total pollen abundance in a sample enhances the detection probability of rare taxa and long‐distance transported pollen grains. These factors bias the observed palynological richness and distort comparisons. Palynological richness in a pollen count proportional to its pollen influx may be one proxy for reconstructing diversity trends through time. The use of this proxy overcomes most problems encountered in rarefaction analysis, but is constrained by inaccuracy in estimating pollen influx due to the imprecise time control of sediment cores. Estimating palynological richness by mathematical methods may be another way of reconstructing pollen diversity. Pollen data tend to reflect diversity on a regional scale. Sites from small basins have the advantage of recording diversity at both local and regional scales, if the detection of each taxon is independent. By associating one site from a large basin with a series of sites from very small basins (e.g. forest‐hollows), information about both regional and local diversity may be obtained. Entomophilous pollen taxa may have to be measured using a different strategy than anemophilous taxa.