Soil carbon measurement in clods and sieved samples in a Mediterranean Vertisol by Visible and Near-Infrared Reflectance Spectroscopy

Abstract

Measurement of the capacity of agricultural soils to sequester atmospheric CO 2 requires the quantification of soil C content. Near-Infrared Reflectance Spectroscopy (NIRS) may provide a fast, reliable and low-cost analytical tool for estimating soil C content. Soil particle size is one of the factors governing the reliability of the NIRS models obtained. This study compared the results obtained using Visible and NIRS (VNIRS) carbon-measurement models developed from soil clods (mean size of 24.5 mm) with those obtained from ground samples sieved to 2 mm. In both cases, the total C, inorganic C and organic C of 205 soil samples were quantified. Samples were taken in 2006 (at three different depths: 0–30, 30–60 and 60–90 cm) and 2007 (from the 0–90 cm profile as a homogenous mixture of the 0–30, 30–60 and 60–90 cm horizons) from a long-term experiment started in 1986 on a typical Mediterranean rainfed Vertisol in southern Spain. Ground and sieved samples provided more accurate predictions (higher R 2 and lower SECV) than soil clods. The VNIRS calibrations developed for organic C showed lower R 2, greater CV, and lower RER and RPD values than for total C and inorganic C. This could be explained by the error associated with the conventional determination of organic C. There were no large differences between the standard errors of prediction (SEP) in the validation group, standard error of cross-validation from the VNIRS equations (SECV) and standard error of laboratory (SEL), indicating satisfactory results with respect to the prediction capacity of the VNIRS models. Considering the difference in particle size and time and effort required to grind and sieve the samples (especially for Vertisols, which are very hard when dry), analyzing soil C with VNIRS using soil clods may be a viable option. In order to improve the results obtained, further research is needed to reduce the error associated with the conventional determination of organic C and adopt different calibration strategies.