Pedotransfer-based bulk density: An alternative approach to enable soil carbon accounting in grain regions of Brazil

Abstract

Soil bulk density (BD) is a critical variable in the estimation of carbon (C) stocks, which are central to agriculture-based C projects. However, field sampling of BD is often labor-intensive and costly. Pedotransfer functions (PTFs), which estimate BD from easily measured soil attributes, provide a promising alternative. This study evaluated 6941 observations from plots under grain production (GR) and 1246 from native vegetation (NV), sampled down to 1 m depth across 47 farms in the South-Central region of Brazil. Each observation included BD and the contents of sand, silt, clay, and C. Thirty-three PTFs from the international literature were tested, and four new ones (F34, F35, F36, F37) were developed. Function performance was assessed using the correlation coefficient (r), function efficiency (FE), mean error (ME), and root mean square error (RMSE). Two literature-based PTFs and the four developed in this study achieved the best performance, with RMSE ranging from 0.11 to 0.20 and ME from −0.01 to 0.09. Among them, F36 and F37 showed the strongest predictive ability, with FE values of 0.41 and 0.66 and r of 0.74 and 0.81, respectively. Carbon stocks estimated from BD predicted by the selected PTFs were strongly correlated with measured stocks (r > 0.90), demonstrating that these functions are reliable alternatives to direct BD measurements for estimating C stocks. For conservative application in C projects, function F37 is recommended for layers below 30 cm, while BD should continue to be directly measured in shallower layers, where management practices exert stronger influence.