Variability of soil properties in a tropical Alfisol used for shifting cultivation

Many lands in the tropics are abandoned after few seasons of shifting cultivation due to loss of soil fertility. There is a great need for characterizing these soils for reforestation or revegetation using conservation farming systems. This study investigate the variability of soil pH, electrical conductivity (EC), clay content, organic C and available P in a tropical Alfisol and its spatial dependence subsequent to shifting cultivation. These parameters were measured in a total of 132 samples obtained from a 45 by 45 m area according to a pre-determined sampling grid. The coefficient of variation for the parameters ranged from 7% to 44%. The variability was assessed by conventional statistical methods and semi-variograms for spatial dependence using geostatistics. All the soil properties except soil pH exhibited spatial dependence at the distances measured. However, EC showed less spatial dependency when compared with clay, organic matter and available P contents. The spatial auto-correlation was strongest for soil organic C, less for clay and EC and least for available P. EC showed a spatial correlation to a highest distance while the organic C showed spatial correlation to a lowest distance. All properties that showed a spatial dependence exhibited high nugget variance. The parameters obtained from the semi-variograms were used to interpolate the values at unmeasured locations using block kriging. It showed that EC is not a limiting factor for crop production in these areas used for shifting cultivation. The organic C levels were low other than in few pockets showing the need in increasing the organic matter contents before cropping again. Available P contents were also limiting where cereals, soybean, maize and grasses which has a low P demand may be included in any conservation farming systems. The field variability obtained by block kriging is useful in applying soil amendments in pockets rather than to the total field to optimize resource use. This study shows the potential of geostatistics in agronomic applications.