Comments on “No‐Tillage and Soil‐Profile Carbon Sequestration: An On‐Farm Assessment”

1 conducted a study to “…determine: (i) changes in SOC [soil organic carbon] within the topsoil due to conversion to NT [no tillage] farming, and (ii) the depth distribution (0–60 cm) of SOC in NT soils compared with PT [plow tillage] and forest soils.” The authors pointed out the difficulties in interpreting the results of the farm-survey approach undertaken, especially regarding the difference in cropping history of each field site, difference in current crop management system since enrolled in a particular tillage system, difference in type of tillage implements, difference in fertilizer use, and difference in crop residue returned. All of these concerns are real, but the greater concern was the sampling approach, which should have tempered the strength of conclusions. Only one field was sampled within a major land resource area (MLRA), and therefore, conventional statistical analysis should not have been used to assess the effects of management within a MLRA. A more appropriate choice of analysis should have been to use the 11 MLRA sampling locations as replicates for the three management systems (plow tillage, no tillage, and woodlot). Because multiple fields of a management system within a MLRA were not sampled, then the only valid comparison was of management systems across MLRAs. How management affects SOC within a MLRA (>1 Mha) should not have been based on three cores (40 cm2) within a single field. The data in Fig. 1 and 2 revealed striking differences between management systems, especially below 30 cm. For example, soil bulk density under woodlots from MLRA 140, 147, 139B, and 139C was much lower than under cropped soils below 30 cm, indicative of a vastly different landscape setting and/or soil type. Soil organic C was also much greater under woodlots at those depths than under cropped soils. By sampling only one field for each management system within a MLRA, the probability was high that the real effect of management on SOC storage would not be distinguishable from intrinsic variability; however without sufficient data, this distinction between sources of variation cannot be known. The calculation of SOC to a cumulative depth of 60 cm in Fig. 4 is not consistent with the data for bulk density and SOC reported in Fig. 1 and 2. Values in Fig. 4 are 4.5 ± 0.8-fold greater than values calculated from data in Fig. 1 and 2. This putative error in summation to a depth of 0 to 60 cm appears to have been perpetuated in Table 2 when reporting estimates for differences in SOC. Changes in SOC of >|2| Mg ha−1 yr−1 are phenomenal—these high values occurred in 8 of 11 cases. It is rare to see such large magnitudes of difference in the literature. The reported C loss rate of −5.2 ± 1.9 Mg ha−1 yr−1 in the four cases that were negative would be greater in magnitude than C accumulation rates in above-ground tree growth from temperate ecosystems. For example, fast-growing pine forests in Georgia accumulate C in aboveground biomass at a rate of ∼4 Mg ha−1 yr−1 (www.gacarbon.org). When considering the MLRA sampling as replicates for the effect of the three management systems, then the amount of C stored in soil would be as presented in Table 1 here. These data suggest that SOC was greater under NT than under PT in the surface 10 cm—consistent with the point made in the original paper. A point of contention, however, is the interpretation of what happened below this surface depth. Soil organic C at 10- to 30-cm depth was not numerically or statistically different between tillage systems, even though this would be the zone where residue C was likely preferentially deposited by inversion tillage operations. Soil organic C at 30- to 50-cm depth was not statistically different between tillage systems, but this was the zone that negated surface SOC accumulation with NT (at least from the experimental evidence). Statistical significance could not be declared between tillage systems, because of greater random variation with depth (coefficient of variation was 42% at 30–50 cm versus 19% at 0–10 cm). Technically, the only significant difference between tillage systems occurred at 0- to 10-cm depth; no difference between tillage systems occurred at 30- to 50-cm depth. Only because of greater random variation did the tillage effect on SOC disappear, not because of greater SOC under PT than under NT at lower depths. In a news release posted by the Soil Science Society of America Journal, the authors are quoted as saying “future studies in SOC sequestration must be done by analyzing the soil profile to about 2-m depth rather than the surface layer only.” This recommendation goes beyond the methods and scope of evaluation presented in this report. This recommendation should be reconsidered in light of the increasing random variation observed with depth. In fact, in a field study with repeated sampling in time in Georgia, for every additional 0.3 m of soil sampled below the plow layer (i.e., 30 cm), the change in SOC would have had to be 0.6 Mg ha−1 greater to maintain statistical significance between forage management means (2). These data indicated that significant SOC sequestration with time could have only been detected with a difference of 3.2 Mg ha−1 at a depth of 0 to 0.3 m and only with a difference of 5.3 Mg ha−1 at a depth of 0 to 1.5 m. The implication from this study was that a much more rigorous sampling approach is needed [e.g., (i) repeated sampling with time, (ii) stratified sampling to account for known landscape variations, and (iii) numerous cores to collect to properly capture a true mean] to overcome natural variations in soil at lower depths. Reasonable conclusions from the study of 1 should have been limited to: (i) SOC storage was greater under NT than under PT only in the surface 10 cm on farmers' fields in the eastern Corn Belt (KY, OH, and PA) and (ii) greater random variation with increasing soil depth limited the possibility to declare differences in SOC and N storage between tillage systems.