Investigation of factors affecting surface pollen assemblages in the Balikun Basin, central Asia: Implications for palaeoenvironmental reconstructions

Abstract To interpret pollen assemblages in terms of past vegetation, it is important to consider the source of pollen for any given lake profile containing sufficient pollen, and the way it arrives at its preservation site. It is of value to know the proportion of the pollen assemblage derived from the lake catchment and that which was airborne. This has been a long-neglected issue for lakes from arid areas of central Asia. In this study, we used two modified Cour (M-Cour) traps in different biomes (steppe and desert) to collect airborne pollen in the enclosed Balikun Basin during the non-flowering season (late autumn to early winter). The pollen assemblages from the traps are similar with large proportions of herbaceous and (sub-)shrub taxa such as Artemisia and Chenopodioideae. This implies a shared long-distance transport component, in addition to pollen derived from the local vegetation. These xeromorphic species are ubiquitous in the inland desert and/or steppe areas beyond the lake basin, and their pollen is likely to be carried by the wind into lake sediments from a broad area, leading to high percentages of Artemisia and Chenopodioideae pollen in lake surface-sediments. Pollen types from upstream areas (e.g. Picea, Cyperaceae, Leguminosae) are barely found in samples from the river mouth and lake surface-sediments. Thus, we speculate that the transport of pollen by inflow streams makes little contribution to the overall pollen budget in Balikun Lake. This is probably related to the low transport capacity of the surface runoff into the lake, as precipitation is low and there is limited meltwater from snow and glaciers in the surrounding mountains. Pollen assemblages from the near-shore lakebed, which have a high percentage of Chenopodioideae pollen (74.8%), represent only the very local environment, reflecting the expansion of saline-alkali habitats rather than desert. Sedimentary archives from the near-shore lakebed will thus overestimate the desert Chenopodioideae pollen component and hence cause unreliable vegetation or climate reconstructions. We also found the local vegetation composition and pollen depositional processes could significantly bias the ratio of Artemisia/Chenopodioideae (A/C), reducing its effectiveness as an index to distinguish steppe from desert. We suggest using A/C ratios from lake surface-sediments which have a large airborne pollen source area in arid areas, as a better reference for down-core fossil pollen.