Global Spatiotemporal Patterns and Uncertainty of Soil C:P and N:P Ratios in the 0–30 cm and 30–100 cm Soil Layers

Description

This study compiled 4,302 observations of the soil carbon-to-phosphorus ratio (CPR) and 5,884 observations of the soil nitrogen-to-phosphorus ratio (NPR) for the 0–30 cm soil layer, as well as 2,610 CPR observations and 3,544 NPR observations for the 30–100 cm soil layer. In addition, 25 environmental covariates closely associated with soil CPR and NPR were integrated from five categories, including climate, vegetation, topography, soil properties, and human activities.To achieve reliable spatial mapping of global soil CPR and NPR, four modeling strategies were designed in this study: a global one-step strategy, a global two-step strategy, a climate-zone-based one-step strategy, and a climate-zone-based two-step strategy. By systematically evaluating the predictive performance of these four strategies, the climate-zone-based one-step strategy was identified as the optimal approach for mapping topsoil CPR and NPR, whereas the global one-step strategy was determined to be the optimal approach for mapping subsoil CPR and NPR.Based on these optimal modeling strategies, quantile random forest models were developed to predict soil CPR and NPR in the 0–30 cm and 30–100 cm soil layers. The independent validation results showed that, for the 0–30 cm layer, the climate-zone-based one-step strategy achieved the best predictive performance, with R² values of 0.51 and 0.52 for CPR and NPR, respectively. For the 30–100 cm layer, the global one-step strategy yielded higher predictive accuracy, with R² values reaching 0.53 for both indicators.Using the final models, this study mapped the spatial distribution and prediction uncertainty of global soil CPR and NPR in the 0–30 cm and 30–100 cm soil layers at a 1 km spatial resolution. Comparative analysis indicated that the 1 km global soil CPR and NPR datasets generated in this study achieved higher predictive accuracy than existing products, providing valuable data support for assessing global soil nutrient limitation patterns, biogeochemical cycles, and ecosystem functions.

Institutions

• Xinjiang University

  Xinjiang, Ürümqi
• Chinese Academy of Sciences

  Beijing, Beijing

Categories

Funders

• National Natural Science Foundation of China

  Beijing

  Grant ID: 42501070

Licence