Optimal soil-sampling design for rubber tree management based on fuzzy clustering

Abstract

Farming management practices related to nutrient recommendation for rubber tree plantations have been a challenge for scientists, farm managers and local producers. Specific caves and building contour ledges to prevent nutrient losses through soil erosion often cause spatial variation of topsoil nutrients in such plantations of rubber trees (Hevea brasiliensis). The design of soil-sampling schemes to test chemical properties of the soil is critical for successful nutrient recommendation for rubber trees. Our objectives were to characterize the spatia variability of soil pH, macronutrient NPK and organic matter in rubber plantations and to evaluate the rationality of soil sampling schemes in rubber plantations for tree nutrient management. The study was conducted in an area of 84m2 consistent of nine rubber trees and soil samples (0–0.2m depth) were taken from 168 grid points with a dimension of 1m×0.5m. Concentrations of total nitrogen, organic matter, available phosphorus, available potassium and pH levels were determined for each soil sample. Based on their spatial variability patterns, the analyzed variables were divided into several homogeneous zones through fuzzy cluster algorithm. The number of subzones was determined using fuzzy performance index and normalized classification entropy to optimize the classification algorithm. The classification results showed that there were three optimal sampling zones for the soil chemical properties. The analysis of variance indicated that chemical properties were significantly different between the delineated zones. The delineated management zones could be used as a reference for making soil-sampling scheme in the rubber plantation. The results of this study have the implication in optimization of soil sampling planning for soil testing for nutrient recommendation. Fuzzy cluster algorithms could classify soil chemical properties into three practical zones by reducing intrazone variability, which would provide with useful information for making effective soil-sampling schemes in rubber tree plantations.