CREATING A LANDSCAPE-SCALE MODEL OF TROPICAL SOIL EVOLUTION ON THE PACIFIC COAST OF COSTA RICA

Well-studied chronosequences of terrace soils along the Pacific coast of Costa Rica provide an opportunity to develop and test a spatial model designed to predict tropical soil composition as a function of age and annual rainfall. Soil mapping in particular can be applied to a host of questions relating to landscape evolution, soil chemistry and mineralogy, nutrient availability and water resources. In tropical regions, available soil data may be antiquated, often dating back more than thirty years, and the “kaolinite-dominated, nutrient-depleted” paradigm only describes about 40 % of tropical soils. This study synthesizes research conducted in four sub-regions of Costa Rica’s Pacific coast to create a soil model along the entire coast (spanning 400 km, 2 degrees of latitude and 1600 to 4800 mm/yr mean annual precipitation). Soils were analyzed by XRD, XRF, ICP-MS, TEM, FTIR, and pH to determine bulk mineral and chemical composition, clay mineralogy, cation exchange capacity, acidity and C:N values. These data were used to create soil weathering classifications applicable across climatic sub-regions.

Across a climatic trend of relatively dry (1600 mm/yr) to wet (4800mm/yr), the four sub-regions differ appreciably in weathering rates. Soils in tropical wet forest environments (4800 mm/yr) evolve from an initial smectite-dominated, nutrient-rich composition to a nutrient-depleted, kaolinite-rich Oxisol assemblage in half the time required for soils in a drier tropical forest (1600 to 2200 mm/yr) to do the same (and these drier soils never become as nutrient depleted as the 4800 mm/yr climate zone). To model the spatial variability of soil evolution, this study employs a cokriged interpolation model to weigh climatic, temporal and topographic data with soil pH, CEC, geochemistry and mineralogy. In a region dependent on agricultural exports and characterized by active tectonic uplift, soil data and accurate methods for acquiring such data contribute to fundamental knowledge of the land, thus influencing current land use and future planning. Furthermore, a spatial model of this type can be applied to questions of terrace correlation in interpreting landscape evolution over Holocene to Pleistocene time scales.