Comparison of three carbon determination methods on naturally occurring substrates and the implication for the quantification of 'soil carbon'

Abstract

Accounting for carbon (C) in soil will require a degree of precision sufficient to permit an assessment of any trend through time. Soil can contain many chemically and physically diverse forms of organic and inorganic carbon, some of which might not meet certain definitions of ‘soil carbon’. In an attempt to assess how measurements of these diverse forms of C might vary with analytical method, we measured the C concentration of 26 substrates by three methods commonly used for soil C (Walkley–Black, Heanes, and Leco). The Heanes and Leco methods were essentially equivalent in their capture of organic C, but the Leco method captured almost all of the inorganic C (carbonates, graphite). The Heanes and Walkley–Black methods did not measure carbonates but did measure 92% and 9%, respectively, of the C in graphite. All three of the common soil test procedures measured some proportion of the charcoal and of the other burnt materials. The proportion of common organic substrates (not the carbonates, graphite, or soil) that was C by weight ranged from ~10% to 90% based on the Heanes and Leco data. The proportion of the organic fraction of those same substrates, as measured by loss-on-ignition, that was C by weight ranged from 42% to 100%. The relationship between Walkley–Black C and total C (by Heanes and Leco) showed that Walkley–Black C was a variable proportion of total C for the 26 substrates. Finally, the well-known, apparent artefact in the Cr-acid methods was investigated: dichromate digestion should contain at least 7–10 mg C in the sample or over-recovery of C might be reported. Our observation that common soil C procedures readily measure C in plant roots and shoots, and in burnt stubble, means that there will likely be intra-annual variation in soil C, because avoidance of these fresh residues is difficult. Such apparent intra-annual variation in soil C will make the detection of long-term trends problematic.