Spatial analysis for management zone delineation in a humid tropic cocoa plantation

摘要

Identifying spatio-temporal patterns of key soil properties could ensure efficient management and input use in agricultural fields with possible increase in yields. A multi-variate geostatistical approach was used to characterize the spatio-temporal variability of the key soil variables to determine management zones in a cocoa field (5.81 ha). One hundred and twenty soil samples were collected. Additionally, a total of nine apparent electrical conductivity (ECa) sampling campaigns at shallow, ECas (0-0.75 m) and deep, ECad (0.75-1.5 m) were conducted with a DUALEM-1S EC meter at the International Cocoa GeneBank, Trinidad between 2009 and 2010. ECad and ECas gave the strongest linear correlation with clay-silt content (r = 0.67 and r = 0.78, respectively) and soil solution electrical conductivity (ECe), ECe (r = 0.76 and r = 0.60, respectively). Multiple linear regressions indicated that clay-silt content and ECe dominated the signal surface response of both ECad and ECas accounting for 66.7 and 63.2 % of ECa variability, respectively. Spearman's rank correlation coefficients (r(s)) ranged between 0.89 and 0.97 for ECad and 0.81 and 0.95 for ECas signifying strong temporal stability. Since ECas covers the depth where cocoa feeder roots concentrate, ECas of the wettest month surveyed (August 2009) was used as secondary data in cokriging to improve the spatial and temporal estimation of clay-silt content and ECe. Cokriged data was subjected to fuzzy cluster classification using the Management Zone Analyst software. Two was determined to be the optimum number of management zones. This zone delineation potentially facilitates cost-effective, environmentally friendly and energy efficient management of the field.