Based on this analysis, to detect a rate of SOC change of 1 Mg ha− 1 a− 1 (which was the rate observed) to 150 cm, if one were to follow the suggestion of Kravchenko and Robertson (2011) and consider each depth layer individually (and thus satisfying the β = 0.2 constraint for each depth layer) would require 1448 paired sampling plots (2896 individual samples). Considered cumulatively, the same analysis would require 78 paired plots (12 individual samples each: 936 individual samples, Table 6) to maintain β = 0.2.
The Kravchenko and Robertson (2011) method for calculating SOC stocks offers an improvement on the more traditional cumulative method insofar as it reduces the likelihood of masking of treatment effects at individual depth increments by other layers. But far from increasing power and reducing the sampling requirement, the high ratio of standard deviation to the mean (CV, Table 6) at individual depth increments requires more sampling (n = 2896 individual samples) to maintain an acceptable type II error rate across the whole depth profile (Fig. 8). This does not mean that researchers should not use the method, but must be prepared to sample much more intensively to avoid lower power problems at the field scale.