Emissions trading may make it possible to achieve reductions in net greenhouse gas emissions for far less cost than without trading. Storing carbon in soils can help offset greenhouse gas emissions while providing environmental benefits such increasing site productivity, speeding water infiltration, and maintaining soil flora and fauna diversity. Including soil carbon in greenhouse gas emission trading markets requires reliable and cost effective soil carbon measurements. At the plot scale, the precision of current soil sampling methods is poorly documented, thus our ability to reliably detect the amounts of carbon likely to be sequestered by a few years of improved management is uncertain. Soil baseline measurements are being installed, and the market for soil carbon is poised to become large. However, without quantified knowledge of the precision of our measurements we run the risk that the sampling designs we install now will fail to detect much of the change in soil carbon occurring between now and remeasurement in 5-10 years.
The Project
We propose to quantify the precision of state-of-the-art soil carbon sampling methods, at the plot scale. Calculating quantities of soil carbon requires destructive sampling of small samples and returning after some duration to re-sample. Soil carbon can vary substantially within a small, seemingly uniform, area due to the spatial variability of factors that influence soil carbon, such as microclimate, soil texture, and plant productivity. This variability makes quantifying small changes in soil carbon stocks difficult. Imprecision in measurements at each depth increases the imprecision of the final calculation. One way to minimize the uncertainty in soil carbon measurements due to spatial variability is to intensively sample a small plot and then re-sample at slightly different undisturbed locations within the same plot in the future. This method is designed to incorporate variations in soil C occurring at horizontal distances of a few centimeters to a couple meters, but remaining variability is eliminated. We will quantify the compounded effects of measurement imprecision and soil variability by intensively sampling small plots, and immediately re-sampling the same plots, quantifying the differences between initial samples and resampling. Repeating this process many times will allow us to determine the smallest changes in soil carbon that can be observed with this type of field measurement.
There are at least two additional methods that may be incorporated to maximize the ability to measure and interpret differences in soil carbon. First, a portion of the variability in soil carbon, and changes in soil carbon over time, may be explained by other soil characteristics. We propose to measure soil moisture, soil texture, and bulk density in addition to soil carbon for a subset of the sample in order to eliminate some of the measured variation in soil carbon. Second, it may be that a portion of the soil carbon is less variable than the total pool of soil carbon. A likely candidate is particulate organic matter (POM) which can constitute a substantial portion of the total carbon and changes more rapidly in response to management changes. We will measure POM for a subset of the samples to determine whether the variability of POM is substantially different from that of the entire soil C pool.
We will compare soils managed differently, to assess whether land management changes the soil in ways that change the precision of sample measurements. To assess the effects of soil variability on the precision of measurements, we will compare measurements made in different vegetation types, with different management histories. We will contrast measurements in from two different climatic regions, moist, Pacific northwest forest and southeastern US agricultural lands. If funding allows, and frozen ground does not prevent sampling within available time, we will also compare grassland soil and cultivated fields in the interior US. Candidate locations for the interior US site are Wisconsin, eastern Oregon, and Colorado. In the western forest, we will compare undisturbed old-growth to third-growth commercial forest that has undergone site preparation after logging. In the southeastern US, we will compare one site that has been cultivated for several decades with a native deciduous forest.
Our work is designed to apply to leading soil sampling methods. In particular, we will build on sampling protocols developed at the Natural Resource Ecology Laboratory at Colorado State University. On each landuse/treatment, four plots (2 x 5 m) will be established. Six cores (6 cm dia.) will be taken from each plot to a depth of 30 cm. Each plot will be resampled to mimic future sample collection (Fig. 1). Cores will be separated into three depths (0-10, 10-20, & 20-30cm) for immediate use in the analysis of how spatial variability affects estimation efficiency. However, at the initial sampling time only we will collect cores from the 30-50 and 50-100 cm depth intervals and determine C contents on bulked (within plot) samples, as part of the total soil C inventory measurement for each plot. The carbon content of each sample from each depth will be analyzed for total soil carbon using a dry combustion analyzer at the Natural Resources Ecology Laboratory at Colorado State University. If samples are collected in two climatic regions, the total number of samples combusted will be: 6 cores x 3 depths x 4 replicates x 2 replicate intervals x 2 treatment x 2 regions = 576 samples. An additional 32 analyses will be performed on the deeper (>30 cm) samples taken as part of the C inventory (4 plot replicates X 2 depths X 2 treatments X 2 regions).