基于准噶尔盆地西部46个样方的植被调查和表土花粉资料，估算了19种主要花粉类型的联合指数（A）、低代表性指数（U）、超代表性指数（O）、代表性系数（R）、区内无植被出现时某花粉类型百分比的平均值（M_a）、区内有植被出现时某花粉类型百分比的平均值（M_p）、样方花粉组合与植物群落间相似系数（CC），定量描述花粉对植被的指示性。结果表明：第1组包括藜科，A值为1.0，O、U值均为0，R值为12.5，M_p远大于M_a，对植被指示意义明显；第2组包括蒿属和麻黄属，A值在0.1~0.4之间，O值在0.6~0.9之间，U值均为0，R值大于18.5，M_p与M_a相差3.6%~12.1%，对植被具较明显指示意义；第3组包括白刺属、百合科、唇形科、豆科、禾本科、菊科、蓼科、毛茛科、蔷薇科、伞形科、莎草科、十字花科、石竹科、旋花科、荨麻科和紫草科，A值小于0.6，U值在0.2~0.9之间，O值在0.1~1.0之间，R值小于4.0，M_p与M_a相差小于4.9%，对植被指示意义不明。研究结果可为利用化石花粉资料准确恢复古植被提供参考依据。

Pollen records have been widely employed as a significant proxy to reconstruct paleovegetation distribution as well as its spatiotemporal evolution worldwide. Due to the marked influence of many factors, such as differences among pollen productivity, transportation mode, biological characteristics of plants, climatic condition, and sedimentary environment, the associated relationship between pollen and vegetation has been shown to be fairly complicated in different regions. As a result, the information regarding the plant community reflected by the pollen assemblage is significantly different from that of the real situation. Therefore, it is clearly necessary to conduct a large number of studies to deeply explore the quantitative relationship between pollen assemblages and vegetation community. Based on vegetation survey and surface pollen data collected from 46 surface soil samples in the western Junggar Basin of Xinjiang, China, this study used various approaches to quantitatively estimate several pollen-based indices representative of vegetation, including the association index (A), underrepresentation index (U), overrepresentation index (O), representative coefficient (R), mean pollen percentage in the absence of vegetation (M_a), mean pollen percentage in the presence of vegetation (M_p), and coefficient of similarity between pollen assemblage and plant community (CC) for 19 major pollen types in our study region. The results of the analysis show that all the 19 pollen taxa can be divided into three groups. The first group includes Chenopodiaceae with A value of 1.0, O and U values of 0, and R value of 12.5. The value of M_p is much higher than M_a, therefore indicating that Chenopodiaceae pollen is over-representative for the corresponding vegetation. The second group consists of Artemisia and Ephedra, in which A values range from 0.1 to 0.4, O values vary from 0.6 to 0.9, U values are 0, and R values are greater than 18.5. Moreover, the differences in values between M_p and M_a range from 3.6% to 12.1%, implying a strong association between pollen distribution and their parent plants. The third group includes various pollen types of Nitraria, Liliaceae, Labiatae, Fabaceae, Poaceae, Asteraceae, Polygonaceae, Ranunculaceae, Rosaceae, Umbelliferae, Brassicaceae, Caryophyllaceae, Convolvulaceae, Urticaceae, and Boraginaceae, with A values of less than 0.6, U values between 0.2 and 0.9, O values between 0.1 and 1.0, and R values of less than 4.0. Moreover, the differences in values between M_p and M_a less than 4.9%, therefore showing that these pollen types are under-representative for the related vegetation. These results provide a basis to improve the reliability of pollen-based vegetation reconstruction.