Four-dimensional modelling reveals decline in cropland soil pH during last four decades in China’s Mollisols region

Abstract

The degradation of fertile Mollisols due to unsustainable management practices poses serious threats to climate change mitigation and food security. Yet, the lack of four-dimensional (4D) dynamic information (i.e., space, depth, and time) on cropland soil pH hinders sustainable soil management. To fill this knowledge gap, over 17,000 soil pH samples were first collected from the Mollisols region in Northeast China. Then, an automatic machine learning model coupled with space-for-time substitution (AutoMLst) was developed for mapping the 4D dynamics of cropland soil pH during 1980–2023. Results showed that AutoMLst performed well in 4D modelling of cropland soil pH, with a coefficient of determination of 0.88. The topsoil (0–30 cm) pH significantly declined from 6.83 in 1980 to 6.43 in 2023 in Northeast China, with an average decline rate of 0.0038 units yr−1 (0.0014–0.0063 units yr−1). The pH declines in the deeper soil layers (30–60 and 60–100 cm) were slight and statistically insignificant. The excessive use of chemical nitrogen (N) fertilizers and N deposition jointly contributed to the decline of cropland soil pH, but the impact of N deposition has increased over time. Although implementing China’s zero-growth policy for chemical fertilizer application will increase soil pH under the shared socioeconomic pathway (SSP) 1–2.6 and 5–8.5 scenarios, the current decline in cropland soil pH should not be overlooked to ensure the health of Mollisols. These findings suggest that the sustainable management of Mollisols resources requires strict monitoring of soil pH dynamics to mitigate potential soil acidification risks.

Keywords

• Automatic machine learning
• Four-dimensional modelling
• Mollisols
• Soil pH
• Climate change