Abstract:
Heilonggang basin is one of the major grain production areas in Hebei Province, but its cultivated land production was strongly constrained by poor nutrients supply and limited water resources. In this study, a method of classification and regression trees model (CART) proposed by Breiman in 1984 was used to study the variation characteristics of cultivated land productivity of Xiong County at landscape and county scales. This nonparametric statistical model belonging to the classification data excavation algorithms is widely used in many fields because of its simplicity and high precision in comparison with conventional analysis methods. Meanwhile, the evaluation of the impacts of natural and management factors on the variability of cultivated land productivity to reveal its major controlling factors can enrich the study contents of cultivated land productivity and the methods for quantitative research on influence factors. Results showed that: 1) at county scale, the cultivated land productivity in Xiong County varied from 4.75 to 6.95 t/hm2, where the southwest and east region had high productivity and the southeast and middle west had low productivity. At landscape scale, the productivity rank of the cultivated lands was: the loam above and clay below slope (6.25 t/hm2)>the light loam homogeneous slope (6.02 t/hm2)>the clay loamy homogeneous bottom land (5.89 t/hm2)>the sandy above and loam below hillock (5.67 t/hm2)>the sandy loam homogeneous sustained hillock (5.39 t/hm2). 2) Management factors had a greater impact on the spatial variation of the cultivated land productivity than natural attributes of cultivated land factors at the county level, where the main controlling factors in an order of irrigation, nitrogen application amount, rapidly-available nitrogen and organic matter contributed 51% of the productivity. Among different landscapes, natural attributes of the cultivated land were the main controlling factors, where the total contribution of cation exchange capacity, content of organic matter, and rapidly-available potassium and nitrogen accounted for 55% of the productivity. It was found that limiting factors for the light loam homogeneous slope were irrigation assurance rate and nitrogen application amount; for the loam above and clay below slope they were cation exchange capacity and nitrogen application amount; for the sandy loam homogeneous sustained hillock with poor water retention capacity it was irrigation; for the sandy above and loam below hillock and the clay loamy homogeneous bottom land they were fertilization rates of potassium and nitrogen. Overall, the cultivated land productivity was significantly correlated with the factors of cultivated land management and natural attributes, especially cultivated land management factors. Cultivated land management factors including the economic water use, rational N fertilization (< 240.34 kg/hm2), and strengthening fertility management of cultivated land should be adjusted to maintain the high cultivated land productivity. This study is very useful in improving the quality and productivity and ensuring the sustainable utilization of cultivated land.