Assessment and calibration of representational bias in soil phytolith assemblages in Northeast China and its implications for paleovegetation reconstruction.

The assessment and calibration of representational bias in modern soil phytolith assemblages provide the basis for improving interpretation of fossil phytolith assemblages. We studied soil phytolith representation by comparing phytoliths from living plant communities with those from paired surface soils, representing 39 plant communities in Northeast China. Together with the use of representation indices, the 34 and 30 soil morphotypes observed in forest and grassland samples, respectively, were both classified into the following four groups: “Associated types” were similarly represented in soils and in the corresponding species inventory data; “Over-represented types” and “Under-represented types” were respectively over- and under-represented in soils compared to the inventory data; and, in the case of “Special types,” the relationship with the parent plants was unclear. In addition, the diagnostic types exhibited different degrees of representation, while the most common morphotypes were equally represented between grassland samples and forest samples. On this basis, a comparison between the original and corrected soil phytolith indices of the additional 29 soil samples was conducted. The soil phytoliths frequencies corrected by R-values differed between plots with differing plant compositions, and were moderately consistent with actual plant richness in the plot inventory data. We therefore confirmed that R-values are a promising means of correcting soil phytoliths for representational bias in temperate regions. The corrected soil phytoliths can be used to reliably reflect vegetation variability. Overall, our study provides an improved understanding of soil phytolith representation and offers a potential method for improving the accuracy of paleovegetation reconstruction. [ABSTRACT FROM AUTHOR]