Relationships between soil phytolith assemblages and vegetation of different steppe types of the central and eastern Inner Mongolia Plateau, China

Background and aims Phytolith analysis is an effective tool for steppe paleovegetation reconstruction, but many questions remain regarding quantitative relationships between surface soil phytolith assemblages and vegetation cover. Methods Phytolith analysis was conducted on 86 surface soil samples from the central and eastern Inner Mongolia Plateau to examine phytolith assemblages under meadow steppe, typical steppe, and desert steppe in this temperate Eurasian steppe region. This paper determined the most indicative phytoliths, developed discrimination criteria for these three steppe types, and built models of quantitative relationships between phytolith assemblages and vegetation cover using the Random Forest (RF) algorithm and the Weighted Average Partial Least Squared (WA-PLS) method which can be used for regional paleovegetation reconstruction. Results (1) The RONDEL FLAT phytolith morphotype indicates meadow steppe; the TRAPEZOID morphotype indicates typical steppe; and the RONDEL ECHINATE, BILOBATE Stipa, and RONDEL CARINATE morphotypes indicate desert steppe. (2) Different steppe types can be distinguished by percentages of RONDEL ECHINATE, ACUTE BULBOSUS, and TRAPEZOID: RONDEL ECHINATE ≥ 2.9% indicates desert steppe; RONDEL ECHINATE < 2.9%, ACUTE BULBOSUS < 7.0%, and TRAPEZOID < 23.2% indicate meadow steppe. (3) While the results from the RF and WA-PLS models are slightly different (R² = 0.85 and R² = 0.82, respectively), both models adequately reflect the quantitative relationships between surface phytoliths and vegetation in the study area. Conclusion Surface soil phytolith assemblages can distinguish meadow steppe, typical steppe, and desert steppe; soil phytolith analysis is a useful tool for quantitative paleovegetation reconstruction in temperate steppe regions.