内蒙古典型草原禾本科植硅体形态

运用地层中植硅体组合解释过去草原植被及气候变化的关键之一，是要了解研究区现代植硅体形态及表土植硅体组合与现代植被的关系。文中研究内蒙古典型草原禾本科植物根、茎、叶、芒以及种子等不同部位的植硅体，对其中的12种主要禾本科植物叶表皮短细胞硅酸体进行分类及统计。研究表明：内蒙古典型草原禾本科叶表皮短细胞硅酸体可分为8种特殊形态类型。C3植物早熟禾亚科的叶表皮短细胞硅酸体形态多样。几乎所有早熟禾亚科都能产生圆型硅酸体，以贝加尔针茅(85．5％)、大针茅(89．7％)、克氏针茅(90％)以及芨芨草(96．6％)中的圆型硅酸体含量最丰富。针茅哑铃型主要见于针茅植物叶表皮短细胞中，克氏针茅的针茅哑铃型含量相对较高。羊草中未见针茅哑铃型硅酸体。浴草、披缄草叶表皮短细胞硅酸体以齿型为主，分别含87．3％和57．2％，齿型在硬质早熟禾中也占一定比例。沙生冰草中的脊圆型占优势，含74．4％。C3植物早熟禾亚科的叶表皮短细胞产生的截锥型硅酸体含量较少。C4植物虎尾草亚科中的糙隐子草叶表皮短细胞硅酸体以黍哑铃型、简单哑铃型、鞍型为主；黍亚科狗尾草则以黍哑铃型占优势(82．9％)。

DIAGNOSTIC PHYTOLITHS FROM NEI MONGOL GRASSLAND

The use of modern phytolith assemblages for the interpretation of fossil phytoliths shows phytolith analysis is an effective method for understanding the grassland vegetation and climate changes. This paper presents firstly a phytolith reference collection and classification system for Nei Mongol grassland, 23 species of grasses were collected from the studied area, and separated into different leaf, glume, awn and so on for diagnostic Poaceae short cell phytolith investigations. 8 phytolith morphotypes were identified. The phytolith morphotypes from the selected species demonstrate that all Pooideae (C 3) subfamily grasses produce rondel form, especially Stipa baicalens ( 85.5% ), S. grandis (89.7%), S. krylovii (90%) and Aneurolepidium chinese (69.1%). Stipa bilobate form is a useful diagnostic type for differentiating the Stipa dominated grassland. Of the short cell phytoliths produced by Agropyron desertorum , about 74.4% of them are keeled rondel. The increase of percentage of the keeled rondel in the soil assemblages would aid in understanding the grassland desertification processes. Cleistogenes squarrosa of Chloridoideae subfamily (C 4) is dominated by panicoid lobate, simple bilobate and saddle forms, and Setaria viridis of Panicoideae subfamily (C 4) is mainly distinguished by panicoid lobate. The percentage of simple bilobate in the soil assemblages would be helpful for understanding the phytolith assemblages produced by Cleistogenes squarrosa dominated grasslands.