Modern pollen assemblages from human-influenced vegetation in northwestern China and their relationship with vegetation and climate

Modern pollen spectra can improve the interpretation of fossil pollen records used to reconstruct past vegetation, climate and human impacts. It is important, therefore, to carefully examine the relationships between modern pollen spectra, vegetation, climate and human activity. Here, we present the results of an analysis of the pollen spectra of 143 surface pollen samples from farmland, wasteland, desert, steppe/meadow, forest and river valley along a transect from Lanzhou to Urumqi, in northwestern China. The modern pollen assemblages are mainly composed of Amaranthaceae, Artemisia, Poaceae, Asteraceae, Ephedra and Nitraria. The results indicate that in general the surface pollen assemblages of different vegetation types reliably represent the modern vegetation in terms of the composition of the main taxa and the dominant types. Farmland is dominated by cereal-type (≥ 15%) and Amaranthaceae (≥ 20%), while the pollen assemblages of wasteland (i.e. the vegetation immediately surrounding farmland) are mainly composed of Amaranthaceae (= 25%), Artemisia (≥ 20%), Poaceae (≥ 10%), Asteraceae (≥ 5%) and Cyperaceae (≥ 5%). Amaranthaceae (≥ 45%) and Ephedra (= 10%) are the most important taxa in desert, and Cyperaceae (≥ 35%) and Thalictrum (≥ 2%) are the dominant pollen types in steppe/meadow. Forest and river valley samples are characterized by high frequencies of Picea (≥ 10%) and Cyperaceae (≥ 20%). Both constrained and partial canonical ordination techniques (RDA and partial RDA) of the main pollen types and environmental variables show that the modern pollen spectra are primarily controlled by mean annual precipitation (MAP). Cyperaceae, Thalictrum and Brassicaceae are positively correlated with MAP and negatively correlated with mean July temperature (T July), while the representation of certain other types, such as Amaranthaceae, Ephedra and Nitraria, is negatively correlated with MAP and positively correlated with TJuly. The Human Influence Index (HII) is significantly correlated with cereal-type pollen, and it can also differentiate human-influenced and natural vegetation. Our results provide a basis for improving the interpretation of fossil pollen records from arid northwestern China and similar regions.