homogeneous vegetation cover. Grasslands of the same type should have similar feed value and be treated with similar range management practices. Finally, the "subtype" is distinguished by the presence of one dominant plant species and a characteristic variety of subdominants.
Ren's classification system is based on the proposition that zonal grassland distribution is affected primarily by bioclimatic conditions, namely, heat and moisture. Therefore, grasslands can be identified by an index that represents a combination of these two factors. Ren et al. (1965) define the quantity of heat as the annual accumulated temperature above zero (Σø) and propose eight grades of heat quantity from frigid (Σø < 1100) to tropical (Σø > 8000). Their moisture index is K, where K = r/0.1*Σø and r is annual precipitation in millimeters, yielding six grades of moisture from extremely arid (K < 0.28) to damp (K > 1.82). By using this classification system, the world's grasslands can be divided into 48 classes, 38 of which have been identified in China and 27 studied in Gansu Province (Hu et al., 1978). Using annual precipitation (r) as the abscissa and annual accumulated temperature (Σø) as the ordinate, Ren and colleagues (1980) have developed a key chart that indicates the "classes" of China's grasslands. Given the annual precipitation and annual accumulated temperature, any grassland can be assigned a location on this chart. For example, the Yongfeng grassland in Tianzhu County has Σø = 1331°C and r = 441 mm, placing it in zone 42 on the chart, a "cold-temperate, humid" grassland, or alpine meadow.
This key chart shows fundamental soil and vegetation characteristics and can be used to estimate the primary biomass of grasslands (Ren, unpublished data, 1980). It also facilitates the study of relationships between classes and prediction of how the development of a particular grassland might be affected by climatic change. Ren's classification system has been used to study animal distribution and ecology, and ecological amplitudes of Xinjiang merino and Tan sheep have been located on the key chart (Xia, 1983; Hu et al., 1984).
Lu et al. (1984) compared Ren's moisture index with Bailey's moisture model and Holdrige's potential evapotranspiration ratio and found that they are highly correlated and simple to compute. An application of these three models in Gansu Province has produced similar results and shown that they can be used to classify natural landscapes. Using the moisture index, Ren et al. (1984) divided the arid regions of northwest China into three categories: semiarid, arid, and superarid, based on K values of 0.85–1.18, 0.28–0.83, and <0.28, respectively.
Using fuzzy mathematical expression, Ge and Chen (1984) derived a binary index for each grassland. These indices reflect the water-temperature features for range sites and describe the sites in the fuzzy spectrum of rangeland ecosystems. Thus, the typical characteristics of each "class" and the relationship between the "classes" can be determined quantitatively and calculated easily. Fuzzy mathematical expression has made Ren's classification system more applicable.