LAUSR

Introduction For future vegetation projections and conservation planning in grassland ecosystems, accurate estimates of past plant cover changes in grassland composition and their responses to the various driving factors are essential. This study quantitatively reconstructs the past regional plant cover in the Songnen grasslands (northeastern China) and explores the relative importance of climate, fire, and human activity on vegetation dynamics. Methods For this purpose, the Regional Estimates of Vegetation Abundance from Large Sites (REVEALS) model is applied to three pollen records from two areas, two in the center of the Songnen grasslands and one located in an area marginal to the grasslands. Results Results from the most reliable REVEALS scenarios show that from the mid-Holocene, steppe (mean cover 40.6%) and dry steppe (mean cover 54.2%) alternately dominated the central part of the Songnen grasslands while the marginal grasslands were mainly characterized by alternating broadleaved forests (mean cover 26.3%), coniferous forests (mean cover 41.9%) and dry steppes (mean cover 30.1%). Discussion By comparing the plant cover results with previous published regional climate, fire and human activity records, the results show that long term vegetation dynamics were mainly driven by East Asia Summer Monsoon (EASM) and the related precipitation variations, but was also affected by fire frequency and human activity. Moreover, vegetation evolution was sensitive to abrupt cooling events including the 4.2 ka BP and stacked ice-rafted debris (IRD) events; the change from steppe to dry steppe, for example, was driven by these abrupt climate changes. Fire events can alter the original vegetation stability allowing the vegetation to respond rapidly to climate changes while human activity merely has limited influence on vegetation changes.