Objectives Our study focused on understanding the spatial variation of soil fertility in subtropical hilly region with complex terrain, and aimed to approach an effective way for regional nutrient management in hilly regions.

Methods A case study was carried out in a typical watershed, Jinjing Town, Changsha County, Hunan Province, which belongs to subtropical hilly region. A total of 946 soil samples were collected within the study area of 112 km² in 2009. The tested soil fertility indicators included soil organic carbon (SOC), total nitrogen (TN), total phosphorus (TP), available nitrogen (AN), available phosphorus (AP), and pH. Geographical statistics and fuzzy c-means clustering algorithm were used to analyze the spatial heterogeneity of soil fertility. Principal component analysis was employed to divide the watershed into different fertility zones. The results were validated using significance test of difference and coefficient of variation.

Results Except for soil pH, SOC, TN, TP, AN and AP exhibited moderate to strong degree of spatial variations, with the coefficient of variation (CV) ranging from 36% to 125%. The study area was divided into three soil fertility zones (MZ1, MZ2 and MZ3) based on principal component analysis and fuzzy c-means clustering algorithm. The CV values of soil fertility indices after zoning was distinctively reduced, the upmost and least reduction were in the CV of pH and AP, with the zoning CV values of 6% and96%, respectively. Soil fertility indices tended to be homogeneous within a zone, and significantly heterogeneous between zones (P<0.01). The later rice yields and total yields of early and later rice were significantly higher in zone MZ1 than in MZ2 and MZ3. In the three zones, soil pH were 4.12, 4.04 and 4.00, respectively, all of which were at acidic levels. SOC contents were at a high level (15.15, 14.38, and 12.24 g/kg). TN contents were at a high level (1.56, 1.48, and 1.34 g/kg). TP contents were at high to very high level (0.86, 0.69, and 0.60 g/kg). AN contents were at a low level (41.08, 35.33, and 26.16 mg/kg). AP contents were at a middle or low level (8.63, 4.46, and 3.39 mg/kg).

Conclusions Soil fertility exhibites significantly spatial variations in the subtropical hilly region due to the complex topography. Zoning based on the soil fertility is an effective way to eliminate the heterogeneity and propose regional nutrient management strategies. Within the scope of this study, the MZ1, MZ2, and MZ3 zones should focus on ameliorating soil acidification, enhancing the utilization efficiency of fertilizer nitrogen, and avoid the over-application of chemical nitrogen fertilizers. For MZ1 zones, a moderate reduction in the application of inorganic phosphorus fertilizers was recommended, with particular attention to preventing excessive phosphorus fertilization during critical growth stages. MZ2 and MZ3 zones, should consider appropriate application of biological enzyme-activated phosphate fertilizer of increasing application of organic fertilizers to enhance the utilization efficiency of phosphorus by crops.