A rubber planation of the state farm in Hainan Island, China is traditionally managed with unified soil practices (e.g. fertilization, cultivation). This inevitably results in an inefficient use of resources since it ignores soil heterogeneity (e.g. variability in soil fertility and environmental conditions) of the rubber plantations. Soil management zones can be used to overcome the limitations above of the uniform soil management practices. However, studies on soil management zones are mainly carried out at the field scale. Additionally, data of some soil properties used to delineate soil management zones are difficult or expensive to acquire. This study selected easily available environmental variables and aimed to evaluate their validity in delineating soil management zones of rubber plantation at regional scale. Four types of environmental variables, including terrain attributes (elevation, slope, and aspect), parent materials, climate factors (precipitation and temperature) and vegetation index (normalized difference vegetation index), were selected as data source, and principal component analysis as well as fuzzy-C means clustering algorithm were applied to delineate soil management zones for a rubber plantation with an area of approximately 26000 ha. Two indices, fuzzy performance index (FPI), and normalized classification entropy (NCE) were used as criterion to determine the optimal number of soil management zones. Results showed that the optimal number of soil management zones for the rubber plantation was three. To test the validity of the soil management zones, 486 soil samples were collected and analyzed for 12 soil properties including pH, soil organic matter (SOM), total nitrogen (TN), available phosphorus (AP), available potassium (AK), exchangeable calcium (Ca), exchangeable magnesium (Mg), available sulfur (S), available copper (Cu), available ferrum (Fe), available manganese (Mn), and available zinc (Zn). One-way analysis of variance was employed to test the difference in the soil properties and environmental variables among the three soil management zones. Statistically significant differences in selected soil properties (except Zn) and environmental variables were found among the three management zones. In addition, the mean coefficients of variation (C.V.) of the soil properties and the environmental variables in the three management zones were much lower than that obtained before the management zones were applied to the rubber plantation. The results above verified that easily available environmental variables could be used to delineate soil management zones for rubber plantation at the regional scale. Further, the management practices corresponding to the characteristics of each zone should be adopted to improve the soil management efficiency of the rubber plantation.%为了提高大尺度范围内橡胶园土壤管理的针对性，以海南省国营八一农场橡胶园为研究对象，以地形因子（海拔、坡度和坡向）、成土母质、气候变量（平均降雨量和平均气温）和植被指数为数据源，对橡胶园进行土壤管理分区。利用模糊C均值聚类法进行分区，以模糊性能指数（fuzzy performance index，FPI）和归一化分类熵（normalized classification entropy，NCE）作为判断最佳分区数的标准，并通过单因素方差分析和分区前后土壤属性以及环境变量变异系数对比对分区结果进行评价和验证。研究结果表明，橡胶园管理分区的最佳数目为3个。不同管理分区之间土壤属性（pH 值、有机质、全氮、有效磷、速效钾、有效硫、交换性钙、交换性镁、有效铜、有效铁和有效锰）和环境变量（高程、坡度、降雨量、平均温和归一化植被指数）的差异性都达到了极显著水平（P＜0.01），同时，3个分区中土壤属性和环境变量变异系数的均值比分区前明显下降。这就验证了在大尺度范围内，利用较易获取的多源环境变量进行橡胶园土壤管理分区是可行的，同时依据不同分区的特点制定了相应的土壤管理措施，提高了大范围区域内橡胶园土壤管理的针对性。
展开▼
