Geostatistical analysis of soil properties in a secondary tropical dry forest, St. Lucia, West Indies

Abstract

Spatial variability of soil properties directly influences forest growth. However, spatial variation in soil properties has not been studied within tropical dry forests. As such, it is unclear whether soil properties, like moisture and N availability, display spatial variation at scales similar to that of other ecosystems. To gain insight into this variation, we established a 56 × 56 m sampling grid in tropical dry forest on the Caribbean island of St. Lucia. Samples collected at 4-m intervals were analyzed for forest floor mass, soil texture, pH, organic C, net N mineralization, net nitrification and available P. Geostatistical procedures were used to determine spatial autocorrelation of the aforementioned properties and processes. Semivariogram parameters were used in a block kriging procedure to produce spatial maps of soil properties. At the scale of our study, most soil properties exhibited spatial autocorrelation at distances of 24 m or less. Varying degrees of similarity were found between patterns of forest floor mass, organic C, net N mineralization, net nitrification and available P. No similarity was found between soil texture or pH and other properties. Fine-scale spatial patterns of net N mineralization and net nitrification are likely driven by overstory litter inputs, rather than variation in soil texture and water availability.