Phytolith reference study for identifying vegetation changes in the forest−grassland region of northeast China. (18th September 2017)

• Record Type: Journal Article
• Authors: Gao, Guizai; Jie, Dongmei; Wang, Yong; Liu, Lidan; Liu, Hongyan; Li, Dehui; Li, Nannan; Shi, Jichen; Leng, Chengcheng
• Is Part Of: Boreas. Volume 47:Number 2(2018)
• Page Start: 481
• Page End: 497
• Publication Date: 2017-09-18
• Subjects: Geology, Stratigraphic -- Quaternary -- Periodicals; 551.79
• DOI: 10.1111/bor.12280
• Languages: English
• ISSNs: 0300-9483

Abstract

Abstract : To provide a basis for tracing changes in vegetation and tree cover density, we studied the phytoliths of 129 common temperate plant species, and extracted the phytoliths from 75 surface soil samples from sites in grassland, forest−grassland ecotone and forest habitats in northeast China. From the analysis of shapes and morphological parameters of the plant samples, we developed a reference data set of herbaceous and woody phytoliths, and subsequently identified 21 herbaceous and 13 woody phytolith types in the surface soil samples. To test the reliability of soil phytolith analysis for distinguishing forest, grassland and the forest−grassland ecotone, we used principal components analysis (PCA) and discriminant analysis (DA) to summarize the soil phytolith assemblage characteristics of the different ecosystems. The results show that the grassland and forest samples are characterized by abundant herbaceous and woody phytoliths, respectively; and that forest−grassland ecotone habitats are characterized by low abundances of blocky polyhedral, multifaceted epidermal and sclereid phytoliths. In general, the surface soil phytolith assemblages can reliably differentiate samples from forest, grassland and the forest−grassland ecotone, with up to 92% of the samples classified correctly. We also tested the reliability of phytolith indices (W/G (1), W/G (2), W/G (3)) for discriminating different vegetation types in our study area, and found that W/G (2) was the most reliable index and corresponded well with the species inventory data. The W/G values for grassland ranged from 0 to 0.3, from 0.3 to 0.6 for the forest−grassland ecotone, and exceeded 0.6 for forest. We conclude that our study provides reliable analogues for phytolith assemblages from palaeoecological contexts, which can be used to reconstruct shifts in forest−grassland ecotones and vegetation succession in temperate areas.