Assessment of spatial uncertainty for delineating optimal soil sampling sites in rubber tree management using sequential indicator simulation

• We used sequential indicator simulation (SIS) to access soil sampling sites.
• Multiple sequential indicator simulation was used to deal with multiple variables.
• Sampling sites were in the shrub and ruderal zones in rubber tree plantations.
• SIS is potentially an effective way to determine optimal sites for soil sampling.

Where and how to sample soils in highly variable tree plantations are important to soil testing and nutrient recommendation for rubber trees (Hevea brasiliensis). The objectives of the study were to determine optimal sampling sites representing means of key soil nutrients at micro scale and to delineate probability maps for optimal soil sampling sites for nutrient management planning for rubber trees. The study was conducted in a rubber tree plantation in the tropical island of Hainan, China in 2011. A total of 168 soil samples (0–0.2 m in the soil depth) were collected in a 1 m × 0.5 m grid in equivalent rectangles. The air-dried soil samples were then analyzed for total nitrogen (TN) and soil organic matter (SOM) variables. Using the sequential indicator simulation (SIS) we discovered that sampling sites were with high probability for both soil TN and SOM within 10% relative standard deviation above and below the means. The high probability regions of uncertainties were near the rubber rhizome neck areas and in the shrub and ruderal zone in the high locations, where were no-tillage zones and at the specific non-cultivated land between the adjacent rubber planting strips with natural vegetation growth. The spatial variability in TN and SOM variables could be attributed to the combined effects of topographic micro-features and tree management practices. It was concluded that SIS method could be useful for effective determination of optimal sites for soil sampling for spatial uncertainty and nutrient management in rubber tree plantations.