Statistical assessment of a paired-site approach for verification of carbon and nitrogen sequestration on Wisconsin conservation reserve program land

ABSTRACT:

The threat of global climate change has provoked policy-makers to consider plausible strategies to slow the accumulation of greenhouse gases—especially carbon dioxide (CO₂)—in the atmosphere. One such idea involves the sequestration of atmospheric carbon (C) in degraded agricultural soils as part of the Conservation Reserve Program (CRP). While the potential for significant C sequestration in CRP grassland ecosystems has been demonstrated, the paired-site sampling approach traditionally used to quantify soil C changes has not been evaluated with robust statistical analysis. In this study, 14 paired CRP (> 8 years old) and cropland sites in Dane County, Wisconsin, were used to assess whether a paired-site sampling design could detect statistically significant differences (ANOVA) in mean soil organic C and total nitrogen (N) storage. We compared 0 to 10 cm (0 to 3.9 in) bulk density and sampled soils (0 to 5 cm, 5 to 10 cm, and 10 to 25 cm [0 to 2 in, 2 to 3.9 in, and 3.9 to 9.8 in]) for textural differences and chemical analysis of organic matter (OM), soil organic C (SOC), total N, and pH. The CRP contributed to reducing soil bulk density by 13% (p < 0.0001) and increased SOC and OM storage (kg m⁻² [lb ft⁻²]) by 13% to 17% in the 0 to 5 cm (2 in) layer (p = 0.1). We tested the statistical power associated with ANOVA for measured soil properties and calculated minimum detectable differences (MDD). We concluded that 40 to 65 paired sites and soil sampling in 5 cm (2 in) increments near the surface were needed to achieve an 80% confidence level (α = 0.05; β = 0.20) in soil C and N sequestration rates. Because soil C and total N storage was highly variable among these sites (CVs > 20%), only a 23% to 29% change in existing total organic C and N pools could be reliably detected. While C and N sequestration (247 kg C ha⁻¹ yr⁻¹ and 17 kg N ha⁻¹ yr⁻¹ [220 lb C ac⁻¹ and 15 lb N ac⁻¹]) may be occurring and confined to the surface 5 cm (2 in) as part of the Wisconsin CRP, our sampling design did not statistically support the desired 80% confidence level. We conclude that usage of statistical power analysis is essential to insure a high level of confidence in soil C and N sequestration rates that are quantified using paired plots.