Spatial variation of carbon and nutrients stocks in Amazonian Dark Earth

Amazonian Dark Earths (ADE) are anthropic soils that are enriched in carbon (C) and several nutrients, particularly calcium (Ca) and phosphorus (P), when compared to adjacent soils from the Amazon basin. Studies on ADE empower the understanding of complex pre-Columbian cultural development in the Amazon and may also provide insights for future sustainable agricultural practices in the tropics. ADE are highly variable in size, depth and soil physico-chemical characteristics. Nonetheless, the differentiation between ADE and the adjacent soils is not standardized and is commonly done based on visual field observations. In this regard, the pretic horizon has been recently proposed as an attempt to classify ADE systematically. Spatial modelling techniques can be of great use to study the structure of the spatial variation of soil properties in highly variable areas. Here, we predicted the carbon and nutrients stocks in ADE by applying spatial modelling techniques using an environmental covariate (i.e. expected anthropic enrichment gradient) in our model. In addition, we used the pretic horizon criteria to classify pretic and non-pretic areas and evaluate their relative contribution to the total stocks. In this study, we collected soil samples from five 20 -cm soil layers at n = 53 georeferenced points placed in a grid of about 10 to 60 m spacing in a study area located in Central Amazon (similar to 9.4 ha). Ceramic fragments were weighed and quantified. Samples were analysed for: Total C, Total Ca, Total P, Exchangeable Ca + Mg, Extractable P, soil pH, potential CEC (pH = 7.0) and the clay content. The use of the pretic horizon criteria allowed us to clearly distinguish two unambiguous areas with a sharp transition, rather than a smooth continuum, in contrast to previous studies in ADE. Depth- and profile-wise linear regression model parameters indicated a greater importance of the chosen environmental covariate (i.e. expected anthropic enrichment gradient) to explain the spatial variation of Total Ca and Total P stocks than Total C stocks. The overall Total Ca and Total P stocks were twice as large in the pretic area when compared to the non-pretic area.