Soil C Accumulation in a White Oak CO2-Enrichment Experiment via Enhanced Root Production (2024)

Keywords: Biogeochemistry; Carbon dioxide fertilization; Soil carbon storage; Atmospheric carbon dioxide; Global warming

1. Introduction

Currently, atmospheric CO₂ levels are increasing at a rate one-third slower than expected, based on the known sources and sinks (Dixon et al. 1994; Field 2001; Pacala et al. 2001; Schimel et al. 2001). Major sources of atmospheric CO₂ include fossil fuel combustion, cement production, and deforestation. Sinks of atmospheric CO₂ include uptake by the ocean and the terrestrial biosphere. A variety of studies indicate that the terrestrial biosphere may be the location of the “missing sink.” Using measurements of stable carbon isotopes in the atmosphere, Ciais et al. (Ciais et al.1995) have shown that the terrestrial biosphere removed about half of fossil fuel emissions in 1992 and 1993, or about 3.5 Gt C yr⁻¹ from the atmosphere. Using atmospheric oxygen measurements, Keeling et al. (Keeling et al. 1996) have shown that the terrestrial biosphere sequestered about 2.0 ± 0.9 Gt C yr⁻¹ (about one-third of fossil fuel input) from 1991 to 1994.

Because trees and soil exchange large amounts of carbon with the atmosphere on an annual basis, both are likely locations for storing the “missing” carbon (Harrison et al. 1993). Caspersen et al. (Caspersen et al. 2000) have reported that the aboveground net ecosystem production due to growth enhancement (e.g., CO₂ fertilization) was only 2.0% ± 4.4% in five states in the eastern United States throughout the twentieth century. At least for one region, Caspersen et al.'s results suggest that the “missing carbon” is not being stored in trees; instead, the missing carbon may be soil bound. Radiocarbon measurements collected worldwide suggest that soil carbon does have the potential to significantly alter atmospheric carbon dioxide levels (Harrison 1996), but do not identify a particular mechanism.

This study suggests that the mechanism for storing carbon in soil is CO₂ fertilization and investigates the extent to which CO₂ fertilization could be increasing the amount of carbon stored in soil in a white oak system. Carbon dioxide fertilization is the increase in plant growth in response to elevated atmospheric CO₂ levels (Strain and Cure 1985; Wullschleger et al. 1995). Bazzaz and Fajer (Bazzaz and Fajer 1992) and Field (Field 1997) have suggested that increased plant growth may result from increased water-use efficiency and decreased photorespiration under elevated CO₂. Norby et al. (Norby et al. 1999) found a 25% increase in stem production/leaf area in trees grown in 12 open-top chamber studies. The increase in vegetation, in turn, may increase the flux of carbon to soil.

To see if additional carbon is stored in soil due to CO₂ fertilization, we measured soil carbon inventories in five soil cores collected from ambient chambers and six soil cores collected from elevated chambers (Figure 1). Figures 2a–c shows a hypothetical scenario for how soil carbon inventories may evolve over time in a CO₂ fertilization experiment. At the start of the experiment, the histograms of carbon inventory between the elevated and ambient chambers will be similar (Figure 2a). After the vegetation has been exposed to elevated carbon dioxide levels, the carbon inventories will begin to shift away from ambient inventories and migrate to higher carbon inventories (toward the right in the histograms; Figures 2b,c). In the actual experiment, the variation is likely to be greater than what is shown in Figure 2, with jumps reflecting the uneven nature of carbon input to soil. Sporadic root growth, root exudation, root death, and decomposition cause variations in soil carbon inventories.

2. Measuring changes in soil carbon inventories due to CO₂ fertilization

Carbon dioxide fertilization may be increasing the amount of carbon stored in soil (Harrison et al. 1993). Just as a savings account balance will increase over time if the amount of money deposited exceeds the amount withdrawn, so too will the soil carbon inventory increase if the flux into soil exceeds the flux out, despite rapid turnover. Such an increase may be due to an increased flux of carbon to the soil that is not matched by an equivalent flux out over a relevant time frame. Processes that add carbon to soil include input from fine root decomposition, litter decomposition, root exudation, and dissolved organic carbon in throughfall. A portion of this organic material undergoes microbial transformations and becomes soil organic material (Zak et al. 1993; Gleixner et al. 2002; Kiem and Kogel-Knabner 2003). A decrease in the rate of soil organic matter decomposition could also increase the inventory of soil organic carbon. For example, researchers have found that elevated carbon dioxide levels may slow the rate of soil organic matter oxidation (Hu et al. 2001). Hence, the observed increases in soil carbon accumulation could be due to increased soil carbon input and decreased soil organic carbon oxidation. Soil carbon's response to CO₂ fertilization has proven very difficult to measure due to soil heterogeneity and the 12–25-yr turnover time of soil carbon (Harrison et al. 1995; Hungate et al. 1996; Hungate et al. 1997; Van Kessel et al. 2000; Schlesinger and Lichter 2001).

In this study, improved soil sampling and soil preparation procedures enabled us to measure increases in soil carbon inventories (Mahoney and Harrison 2003). Statistical analysis suggests that our results are robust.

3. Experimental setting

This CO₂ fertilization experiment was conducted at the Global Change Field Research Site at Oak Ridge National Laboratory, Oak Ridge, Tennessee (Figure 1). Kelly et al. (Kelly et al. 1969) have described the land-use history and soil characteristics of the Global Change Field Research Site. Farmers tilled the land until 1942. The land lay fallow until 1956, when it was planted with Loblolly pine. In 1964, the site was clear-cut, a bulldozer removed the stumps, and a “bush and bog” disc broke up the soil, cut the roots, and sliced the vegetation mat. A furrow plow, disc harrow, and drag were used on the soil before it was seeded with Festuca and treated with a 10–10–10 fertilizer. Since 1964, the land has received no further treatment, until the CO₂-enrichment experiments began in 1989.

This study was part of a broader study that explored the effects of elevated CO₂ levels on white oak seedlings (Quercus alba L.; Norby et al. 1986; Norby et al. 1995). This experiment started with seedlings planted in chambers in 1989, which grew into saplings under open-canopy conditions. White oaks were exposed to three different levels of carbon dioxide concentrations: ambient (control), ambient + 150 ppm, and ambient + 300 ppm. The ambient atmospheric carbon dioxide concentrations for this site can be approximated using Mauna Loa Volcano (Hawaii) data: in 1988, values were about 353 ppm; in 1992, they were about 356 ppm. The experiment lasted for four growing seasons in six open-top chambers (Figure 1). The chambers were 3.0 m in diameter and 2.4 m tall. When the trees were harvested, the tree height averaged 1.33 m in the ambient chambers, 1.55 m in the (ambient + 150) chambers, and 2.23 m in the (ambient + 300) chambers. Leaves were not allowed to decompose and form soil organic material. The fallen leaves were collected and analyzed for litter chemistry. The actual soil carbon response to CO₂ fertilization may be higher when leaves are allowed to contribute to soil organic material.

4. Prior results from the white oak experiment

In 1992, after four growing seasons, Norby et al. (Norby et al. 1995) found that whole plant mass was 135% higher for the (ambient + 300 ppm) treatment than the control (p < 0.03); lateral woody roots were the most responsive and showed a 294% increase (over the control) in dry biomass (ambient + 300 ppm; p < 0.06); and root density was 141% higher for the (ambient + 300 ppm) trees than the control (p < 0.19). The leaf area index (LAI) was 1.9 m² m⁻² in the ambient chambers, 2.3 m² m⁻² in the (ambient + 150 ppm) chambers, and 3.6 m² m⁻² in the (ambient + 300 ppm) chambers (Norby et al. 1995). Once the canopy had closed and maximum LAI was achieved, differences between elevated and ambient LAI diminished.

5. Experimental methods

Soil was collected in January 1994 from chambers where white oak trees had been growing under ambient and elevated carbon dioxide levels for four growing seasons (Figure 1). The elevated chambers had been subjected to (ambient + 300 ppm) CO₂ levels. We did not expect the combination of only 150-ppm elevated CO₂ and 4 yr of exposure to be long enough to significantly elevate soil carbon levels for chambers 1 and 6, so we did not sample those chambers.

Five soil cores were collected from ambient chambers and six soil cores were collected from elevated (ambient + 300) chambers. These 30-cm-deep soil cores were split into four sections by depth: 0–5, 5–10, 10–20, and 20–30 cm. The soil was collected using a volumetric soil probe (diameter = 5 cm) so that bulk density could be calculated for each core. The coring device could sample to a depth of 35 cm.

Cores that showed signs of soil compression or soil loss were discarded. At the time of sampling, if the soil in the coring device was not even with the surface of the surrounding soil, it was discarded. If a piece of core was missing, the entire core was also discarded. Only one core was discarded.

Our sample processing inhibited microbial activity, isolated soil organic material bound to minerals, and hom*ogenized the sample to attain high levels of precision and accuracy in subsequent measurements (Mahoney and Harrison 2003). The collected soil was dried to a constant weight at 100°C. This process removed sufficient moisture to inhibit microbial oxidation of soil organic carbon and stabilized the soil for storage. After the soil was dried, it was sifted through 2-mm and 30-μm sieves. Visible rocks, charcoal, litter, and root fragments were removed by inspection. Any remaining litter and charcoal were removed by flotation. It was essential that the litter and root fragments were removed from the mineral soil, because litter and root fragments have a turnover time much shorter than the turnover time of active soil carbon (Harrison et al.1993). Similarly, charcoal has a turnover time of thousands of years and must be removed to get accurate turnover times for soil organic material (Trumbore 1988; Telles et al. 2003). After delittering, the soil was hand ground with a mortar and pestle, and dried again at 100°C to a constant weight. We detected no carbonates in this soil.

Soil samples were analyzed for organic carbon using a Carlo Erba NC2100 analyzer at Boston College. The procedure is described by Verardo et al. (Verardo et al. 1990). The precision for measuring the amount of carbon in a sample is ± 0.3%. Soil carbon inventories were calculated by multiplying the bulk density by the mass C in the sample, divided by the mass of sample. The bulk density was calculated by dividing the dry weight of the soil by the volume of the soil. Clay content was measured using a sedigraph (Micromeritics 2001). We wet sieved the soil with a 63-μm sieve to isolate the sand fraction (Folk 1968), and clay content of the remaining soil was determined (Micromeritics 2001).

6. Results

Table 1 lists the soil carbon inventories. The average inventory of soil carbon was higher in the elevated chambers than in the ambient chambers for all depth intervals, although this difference was statistically significant only in the top 5 cm. When considering samples pooled from all depths, the soil beneath trees exposed to elevated CO₂ levels had an average of 14% more carbon than soil beneath trees exposed to ambient CO₂ levels. These differences were found to be statistically significant (see sections 7.1. and 7.2.). For this study, we have considered results statistically significant for p values of < 0.05. Figures 3 and 4 show that the carbon inventories in the elevated chambers were, on average, higher than the carbon inventories in the ambient chambers. Figure 4 shows that soil carbon variability for the elevated chambers was greater than soil carbon variability for the ambient chambers, while Figure 2 explains the greater variability.

The greatest percent differences in carbon inventory occurred at the shallowest depth interval (Table 1). At this shallow depth, 0–5 cm, the elevated soil had about 30% more carbon than the ambient soil (p = 0.03; t test). In the 5–10-cm depth interval, the soil collected from the elevated chambers had about 9% more carbon than the soil collected from the ambient chambers (p = 0.21; t test). In the 10–20-cm interval, the elevated soil had about 12% more carbon than the ambient soil (p = 0.07; t test), while at the 20–30-cm interval, the difference was about 10% (p = 0.07; t test).

The clay content averaged 36.7% ± 3.4% for the ambient chambers and 35.5% ± 3.8% for the elevated chambers (Figure 5). There is not a strong relationship between clay content and soil carbon concentrations in the ambient chambers (R² = 0.16; dotted line) or the elevated chambers (R² = 0.16; solid line).

The bulk density measurements for the soil collected from the ambient and elevated chambers were similar. Bulk density averaged 1.04 ± 0.13 g cm⁻³ for the ambient chambers and 0.97 ± 0.15 g cm⁻³ for the elevated chambers. Figure 6 shows the relationship between the percent soil carbon and bulk density. We believe the additional influx of carbon in the elevated chambers decreased the soil bulk density, because the density of organic material is less than the density of rock and clay particles. Bulk density increased with increasing depth.

7. Discussion