Vertical patterns and controlling factors of soil nitrogen in deep profiles on the Loess Plateau of China

Abstract

Soil nitrogen is highly variable both in horizontal and vertical directions and plays an important role in environmental quality. To explore the vertical distributions of soil total nitrogen (STN), nitrate nitrogen (NO₃^–-N) and ammonium nitrogen (NH₄⁺-N) in deep vadose zone soil, and to verify the notion that soil N is highly affected by land use and soil type and that is depth-dependent, we collected soil and fresh root samples to a maximum depth of 21 m at 25 sites across China’s Loess Plateau (CLP). STN and NO₃^–-N concentrations were significantly higher under orchard (0.41 g kg⁻¹ and 40.98 mg kg⁻¹, respectively) than under farmland, forest and shrub-grassland (p < 0.001), while NH₄⁺-N (7.32 mg kg⁻¹) was significantly lower under orchard than forest (8.32 mg kg⁻¹, p < 0.05). In silt clay loam soil, STN and NO₃^–-N concentrations were significantly higher (0.39 and 26.16 mg kg⁻¹) than in silt loam soil (0.22 and 1.89 mg kg⁻¹). Soil N differed greatly (especially in the upper four meters) but generally decreased with soil depth (SampD). Random forest model (RFM) and multiple linear regression analyses showed that the most important variables for STN were LogSOC, LogSampD, clay, pH, soil inorganic carbon, soil water content and LogAltitude; for LogNO₃⁻-N were pH, LogSampD, land use, LogSOC, and soil water content; and for LogNH₄⁺-N were pH, clay, LogAltitude, slope gradient and LogSOC. RFM had comparable good performance in predicting LogSTN and LogNO₃^–-N (r > 0.915, p < 0.001), followed by LogNH₄⁺-N (r = 0.653, p < 0.001). The amount, distribution and variation of soil N were co-controlled by land use, soil type, topography, soil properties and considered soil depth. Understanding this information is essential in estimating soil N budget, developing N cycling models and improving sustainable N management on the CLP and similar regions around the world.