Reliability of pollen ratios for environmental reconstructions on the Tibetan Plateau

Aim Pollen ratios are widely used to gain palaeovegetation and palaeoclimatic information from fossil pollen spectra, although their applicability has seldom been tested with modern pollen data. I used a data set of 113 lake‐surface sediments from the eastern Tibetan Plateau to test the reliability of several pollen ratios. Location The lake‐surface pollen spectra cover a wide range of vegetation types (temperate desert, temperate steppe, alpine desert, alpine steppe, high‐alpine meadow, sub‐alpine shrub, coniferous and mixed forest) and climatic conditions (mean July temperature, T_July: 4.0–17.4°C; mean annual precipitation, P_ann: 104–670 mm). Methods Lake‐surface sediments were analysed palynologically, and several pollen ratios were calculated. These ratios were interpreted with respect to vegetation and climatic conditions. Results The arboreal pollen sum (AP) was highest in samples from forested areas and was significantly correlated with P_ann (r² = 0.44). In non‐forested areas, samples from large lakes and from lakes surrounded by sparse vegetation had increased AP values, suggesting that AP is a useful vegetation density indicator. Artemisia/Chenopodiaceae (A/C) ratios were lowest in desert areas and were positively correlated with P_ann (r² = 0.25). The aridity pollen index was inappropriate for inferring (palaeo‐)climatic information from samples on the eastern Tibetan Plateau as it had no significant correlation with the environmental factors. Artemisia/Cyperaceae (A/Cy) ratios had a significant correlation with T_July (r² = 0.23), but only a weak correlation with P_ann, which indicates that the A/Cy ratio is applicable as a temperature indicator. Furthermore, it is a valuable tool for the differentiation of high‐alpine meadow from steppe vegetation. Main conclusions AP sum, A/C ratio and A/Cy ratio are useful tools for qualitative and semi‐quantitative palaeoenvironmental reconstruction on the Tibetan Plateau; however, the results obtained should not be interpreted quantitatively.