Modern pollen–plant richness and diversity relationships exist along a vegetational gradient in southern Norway

Abstract

The relationships between modern pollen and floristic plant richness, diversity and evenness are assessed using pollen assemblages and associated vegetation data from 52 lakes along an elevational and vegetational gradient in the Setesdal valley of south-central Norway. Various data transformations were applied to minimise bias in the vegetation and pollen datasets. Plant species were transformed to their pollen or spore equivalents to reduce taxonomic biases. Pollen counts were transformed using Andersen’s general pollen-representation values for northern European trees and shrubs and the Regional Estimates of Vegetation Abundance from Large Sites (REVEALS) model with pollen-productivity estimates (PPEs) appropriate for Setesdal to minimise pollen-representation bias. Pollen count-size bias (before or after transformation) was eliminated by rarefaction analysis based on bootstrap resampling. Richness and diversity were quantified using Hill numbers ( N0, N1, N2), and evenness was estimated as the ratios of N0, N1 and N2. Diversity partitioning was used to estimate β diversity. The strongest correlations between pollen and plant richness and diversity are with pollen counts transformed using Andersen’s representation values and rarefied to a common count size and with plants transformed to their pollen equivalents. However, if sites from the low-alpine zone are excluded where there are high values of far-transported tree pollen, the richness and diversity relationships are also statistically significant for untransformed pollen data and plants transformed into their pollen equivalents. The effects of data transformation on diversity partitioning and estimates of β diversity are explored. We demonstrate that there are statistically significant positive relationships between pollen and plant richness and diversity along the entire elevational gradient after transforming the datasets to minimise biases due to taxonomic differences, differential pollen representation, and pollen-count size, and similar significant positive relationships along the forested parts of the gradient (nemoral, boreonemoral, southern boreal, middle boreal) after transforming the datasets to minimise biases due to taxonomic differences and pollen-count size.