Restoration of degraded cropland in the Pearl River Delta, China, through integrated agronomic measures: a modeling approach

Introduction Guangdong Province is an important grain and agricultural product production area in China, but part of the cultivated land in the Pearl River Delta has long been faced with complex degradation problems such as soil acidification, heavy metal pollution and lack of available phosphorus. These factors have posed continuous pressure on the sustainable development of regional agriculture and the restoration of farmland ecological functions. Methods Based on the multi-source data of 10 typical degraded farmland samples in the Pearl River Delta, this study constructed an analysis framework combining random forest and response surface method. Firstly, the random forest model was used to identify the key variables closely related to the soil function restoration index; Then box - Behnken design and response surface methodology were used to evaluate the comprehensive effects of biochar dosage, lime dosage and legume crop configuration on soil physical and chemical properties, CD bioavailability and microbial diversity. Results The results showed that available phosphorus and pH were the two variables most closely related to soil functional recovery index. Compared with the traditional remediation treatment, the biochar lime combined treatment was related to the increase of soil porosity, soil organic carbon and the proportion of residual CD, indicating that it had better comprehensive performance in improving soil environmental conditions and promoting CD stabilization. Among the tested legume crops, Astragalus sinicus is relatively superior in nitrogen fixation ability, microbial diversity, available phosphorus improvement, soil organic carbon increase and CD bioavailability reduction. In general, the combination of biochar 8 T/ha, lime 3 T/HA and Astragalus sinicus achieved a relatively balanced result among soil function recovery index, CD bioavailability control and input cost. Discussion This study combined random forest key factor identification, response surface parameter optimization and field validation, and provided a more specific decision-making basis for the zoning restoration of acidified degraded farmland and the selection of comprehensive agronomic measures in the Pearl River Delta. The results are helpful to optimize the restoration scheme of degraded farmland from the perspective of comprehensive trade-off among soil function restoration, heavy metal stabilization and cost constraint.