Carbon sequestration in a surface flow constructed wetland after 12 years of swine wastewater treatment

North Carolina ranks second in swine production after Iowa and the industry generates a high amount of waste. Swine producers currently flush the waste from swine houses into an anaerobic lagoon and spray the lagoon wastewater onto agricultural fields. Field applications have resulted in excessive soil nutrient accumulations, especially phosphorus (P), which can potentially run off, or leach below the root zone, and pollute surface waters or groundwater. In most situations, producers lack the acreage to apply manure nutrients to crops at recommended nutrient rates to avoid high nutrient accumulations. To find a solution to this problem, we evaluated the efficiency of an artificial wetland that treated wastewater from an anaerobic lagoon that received effluent from an indoor production facility (Reddy et al. 2001). Results showed 75% N, 45% P, and 50% chemical oxygen demand removal (Reddy et al. 2001; Poach et al. 2004). We concluded that crop recommended application rates could be achieved without soil nutrient accumulations when small to marginal production operations use these wetlands.

During the last five decades, atmospheric levels of CO₂ and CH₄ have increased and are believed to be contributing to climatic changes. This notion led to an interest in research that searches for ways to increase C sequestration in terrestrial systems to help reduce atmospheric levels. It is not known how much carbon can be sequestered in constructed wetland systems on a long-term basis. The cycling process of C in wetlands is complex. A few studies have focused on components of the wetland carbon cycle and key relationships have been developed. For example, biomass input versus resulting carbon sequestration rates have been reported (Whiting & Chanton 2001; Euliss et al. 2006; Kayranli et al. 2010). These studies have discovered that the amount of C sequestered depends on the rate and concentration of both exogenous C deposition (wastewaters) and endogenous C inputs (plant detritus and microphyte biomass). It also depends on the rate of biomass decomposition and of carbon emission. Also, the efficiency of wetlands is largely dependent on the amount of organic matter present. In addition, components within the wetland carbon cycle may serve as a C sequestration offset (Rosso & Stenstrom 2008); e.g., the slow organic matter decomposition in wetlands is offset by the release of CO₂ from the living vegetation (Whiting & Chanton 2001; Kayranli et al. 2010).

The dynamics of C in wetlands varies greatly, being time dependent and affected by hydrology (Whiting & Chanton 2001; Kayranli et al. 2010), biomass input, and amount and concentration of wastewater (Euliss et al. 2006; Villamar et al. 2015). Carbon sequestration is evident by the amount of carbon compounds that wetlands accumulate and the type of C fraction. The complex soil organic carbon (SOC) pool is divided into labile and recalcitrant pools. Microbes readily use the labile pool while microbial consumption from the recalcitrant pool is very slow. The labile pool resident time could be months to decades, whereas the resident time for the recalcitrant pool could be hundreds of years (Cheng et al. 2007). Resident time depends on plant biomass productivity, rate of organic matter decomposition, climatic conditions, and input of sources other than plant material. Studies have found the amount of recalcitrant SOC to be less than the amount of labile SOC (Melillo et al. 2002). Fang et al. (2005) found about equal amounts of each, while Leifeld & Fuhrer (2005) found greater amounts of recalcitrant SOC.

Wetlands have aerobic and anaerobic conditions that may facilitate C accumulation. Therefore, understanding how wetlands operate as a carbon sink and the types of carbon fractions that dominate the wetland ecosystem will help improve wetland construction design and management for optimal carbon sequestration and wetland efficiency. The objectives of this study were to: (1) determine plant biomass decomposition rates, (2) estimate C sequestration potential, and (3) determine the distribution of C among the labile, intermediate, and recalcitrant fractions in the detritus layer and in the soil, of a constructed wetland that treated swine wastewater for 12 years.

In March 1996, the marsh sections were planted with a mix of cattails (Typha latifolia L.) and American bulrushes (Schoenoplectus americanus). To establish equilibrium in the wetland ecosystem, the cells were continuously loaded with wastewater having a nitrogen equivalent rate of 5 to 6 kg N ha⁻¹ day⁻¹ from March 1996 through April 1998 (Reddy et al. 2001). Loading consisted of pumping wastewater from either the primary or the secondary lagoon into an 80,000-l storage tank located uphill from the cells, and transported into cells by gravitational flow through an underground piping system. Over the years, the cattails predominated over the bulrushes and eventually the marsh areas had only cattails.

A study that began in May 1998 treated cells with swine wastewater at different daily N loading rates. Experimental methods and results are described in Reddy et al. (2001) and Poach et al. (2004). The varied loading rates were applied until April 2010, constituting a 12-year period of wastewater treatment. Loading each year was discontinued during mid-December through mid-April because of frequent freezing conditions and cattails being dormant during that period. In April 2010, the soil and detritus samples were taken for the C sequestration study.

An auger having a diameter of 4.4 cm was used to collect soil samples from within each of the two marsh areas and from the pond area. Four samples collected at random within each of the three sampling areas were composited into one sample in each area (marsh inlet, pond, and marsh outlet). Sampling depths were the upper 4 cm layer of detritus, and soil depths of 0 to 6 cm and 6 to 12 cm. The wet detritus slush and soil samples were placed in polythene bags, transported in an ice chest to the laboratory, and stored in the refrigerator at 4 °C until being processed for chemical analysis. Plant debris was removed from all soil samples and then air-dried and crushed to pass through a 2 mm sieve. For bulk density, an Ehland sampler was used to collect undisturbed soil cores having a diameter of 7.5 cm. Three cores were collected from the marsh and the pond areas.

Soil pH was determined using an Orion 3 Star pH meter and a 2:1 water:soil mix. Total carbon (TC) and total nitrogen (TN) were determined using a CHN analyzer (Perkin Elmer 2400 Series II model). Ammonium (NH₄) and nitrate (NO₃), extracted using a 2.0 M KCl solution, and orthophosphate (PO₄), were measured with a flow injection analyzer (Lachat Instruments QuikChem 8500 Series 2). Microbial biomass carbon (MBC) was determined using the procedure described by Vance et al. (1987). In general, field-moist soil that was chloroform (CHCl₃)-fumigated or un-fumigated was shaken for 30 minutes as a mix of 1:5 soil to 0.5 M K₂SO₄ extract solution. Fumigated soil was placed in a separate canning jar for microbial CO₂ production. Soil MBC was calculated as the quantity of CO₂ evolved following fumigation divided by an efficiency factor of 0.41 (Franzluebbers et al. 1999).

Cattail shoots were removed from a 0.5 m² area chosen at random within inlet and outlet marsh areas and weighed. Moisture content from subsamples was determined by weighing, drying in a forced-air oven at 65 °C for 48 hours, and reweighing. The remaining fresh plant material was cut into lengths of 5 cm, and 100 g of this material (25 g dry weight equivalent) was placed inside nylon 1 mm mesh bags of size 30 × 30 cm. In the inlet and outlet marsh areas, three mesh bags were buried under the 4 cm detritus layer that had formed over the years. Mesh bag placement was on December 15, 2012. One bag was removed from each marsh area on April 4, July 17, and September 11, 2013. Immediately after mesh bag removal, contents were oven dried at 65 °C for 48 h and weighed. A subsample of the dried material was ashed at 500 °C in a muffle furnace to determine the soil content and adjust each sample to an organic matter content by weight (Aber et al. 1990).

Soil pH values ranged from 6.7 to 7.7 and no significant differences were found between soil depths in marsh or pond areas (Table 1). The soil bulk density was higher in the pond area than in the marsh and it was due to the lack of vegetation in the pond area. The amount of NH₄⁺, NO₃⁻, and PO₄⁻ in the marsh areas was three to four times higher in the upper soil depth than in the lower depths, whereas, in the pond area, the nutrient concentrations did not differ much among the soil depths. The pond area was 75 cm deep and therefore anaerobic conditions prevailed in this area. Hence, nitrification did not occur as much as in the marsh area.

Cattails are known to contain approximately 28.7% cellulose, 23.4% hemicellulose, and 10.1% lignin (Suda et al. 2009). One of our previous studies showed that the plant decomposition function is supported by the abundance of heterotrophic bacteria (Dong & Reddy 2010); however, fungi that aid in the decomposition process may also prevail in this wetland.

Wastewater was the major carbon source in the pond while in the marsh area C originated from the wastewater and the decomposing cattail biomass. Also, we reiterate that each year wetland loading was done during a 7-month period due to freezing conditions and absence of cattail activity during the December through April months.

Soil TC, TN, and MBC results are shown in Table 2. Differences in TC occurred between soil depths and it was higher in the marsh area than in the pond area. Contents of TN and MBC were also higher in the marsh area but differences were not as large as with TC. In general, averaged across depths, TC content accumulated over 12 years was 21 Mg ha⁻¹ (i.e., 1.75 Mg ha⁻¹ yr⁻¹) in the marsh area and 12.1 Mg ha⁻¹ (i.e., 1 Mg ha⁻¹ yr⁻¹) in the pond area. TN contents in the marsh and pond areas were 2.0 Mg ha⁻¹ and 1.7 Mg ha⁻¹, respectively, while the concentrations of MBC were 579.8 mg kg⁻¹ and 445.7 mg kg⁻¹. The MBC was less than 2% of the TC in both wetland areas. Bernal & Mitsch (2008) reported a C accumulation of 2.42 Mg ha⁻¹ yr⁻¹ in created riverine wetlands. Also, Mitsch et al. (2012) reported a stable C accumulation of 2.19 and 2.69 Mg C ha⁻¹ yr⁻¹ for planted and non-planted wetlands, respectively. These values are consistent with ours, which were 1.75 Mg ha⁻¹ yr⁻¹ in marsh area and 1 Mg ha⁻¹ yr⁻¹ in the pond area, respectively, based on accumulations per year.

These C and N contents constitute end result estimates following a highly dynamic sequence of events during the 12-year period, involving major N cycle processes including immobilization and mineralization (Aber et al. 1990; Hunt et al. 2003), nitrification (Dong & Reddy 2010), denitrification (Hunt et al. 2006) and ammonia volatilization (Poach et al. 2004). The larger contents of these soil parameters: TC, TN, and MBC, in the marsh area were likely due to the cattail biomass added to the wetlands after they senesced in November each year.

Plant dry matter measurements (Table 3) showed that C content was slightly higher in the cattails from the wetland outlet than the inlet area. The wetlands are facilitated with gravity flow and the differences may be due to the bottom slope that could have caused a greater accumulation of water and nutrients in the outlet, resulting in a higher total nutrient accumulation in the harvested biomass. Maucieri et al. (2014) have also reported similar observations. Further, high levels of nutrients entering the wetland system denote higher plant biomass resulting in higher nutrient accumulations in the vegetation (Lee et al. 2014). Averaging the data from inlet and outlet, cattails were observed to accumulate 6.1 Mg of TN and 0.25 Mg of TC ha⁻¹ yr⁻¹ (Table 3). Maucieri et al. (2014) reported a dry matter organic carbon content of 3.9 Mg ha⁻¹ in the aerial parts of common reed (Phragmites australis (Cav.)).

Swine wastewater contains many organic compounds that contribute to the formation of stable carbon structures. These include carbohydrates, aliphatic and aromatic hydrocarbons, proteins, N-acetyl amino sugars, lipids, and fulvic acid (Leenheer & Rostad 2004). The higher C content of the upper depth, and the lower C content in the pond area than in the marsh area, indicate that both swine wastewater and plant decomposing biomass and roots are significant C contributing sources. It is likely that C accumulation in lower depths is more related to plant root turnover. It is a known fact that high plant biomass, reducing soil conditions, and slow decomposition rates make wetlands effective C sinks. Clearly, a constructed wetland with a 7-month yearly loading period like the one used in this study will have a lower C accumulation than a continuous plant-based operated wetland. Hossler & Bouchard (2010) stated that poorly developed soil properties could be limiting soil C accumulation in constructed wetlands with variable hydrology. However, our study shows that when vegetated marsh areas are included, plant decomposition occurs at a fairly rapid rate, significantly adding to the stable carbon pool. The carbon-rich detritus layer, which is a combination of plant litter and the organic matter received from the wastewater settling on the wetland surface, contributes to this enhanced carbon accumulation in the marsh areas.

SOC dynamics assessments require temporal chemical fractionation of SOM constituents. Our fractionation assesses the SOM status after 12 years of wetland function. This fractionation provides knowledge of the potential of this system to sequester carbon in stable forms. The organic matter fractionation of the 4 cm detritus layer is illustrated in Table 4.

The soil humin C content was two-fold lower than the NaOH extracted C. The SOC extracted with CW, HW, HCl-HF, and sodium pyruvate (Na₄P₂O₄) was less than 500 mg kg⁻¹ and represents less than 10% of total C. Humic and fulvic acids extracted with NaOH were 16 to 20 times higher than humic and fulvic acids extracted with sodium pyruvate. Evidently, most SOM was at a high level of decomposition and C was mostly present as humin C. Marsh area Eh values ranged from −143 to +250 mV during the wastewater loading period, and during the non-loading winter period Eh values were <300 mV; i.e., redox potential was within the range where organic matter decomposition occurs.

In general, the lower content of soil C compared to the marsh area was in large part due to the absence of vegetation in the pond area. Therefore, most of the C that accumulated in the pond area soil was from the swine wastewater, a source of organic matter that lacks compounds containing significant lignin content. The pond C extracted with NaOH likely originated from fulvic acid, aliphatic hydrocarbons, aromatic hydrocarbons, proteins, and lipids, which are common in swine wastewater (Leenheer & Rostad 2004).

In conclusion, the labile C decomposes readily and humic substances accumulate in these wetlands. Our previous studies have shown that these MPM wetlands are efficient in removing nutrients, but less efficient than continuous marsh wetlands. We can also conclude based on this study, that continuous marsh wetlands would accumulate more stable forms of carbon than MPM wetlands. Dissolved organic C and MBC are important indicators of an active C pool and, although they constitute a small part of total C, the turnover rates are important to the cycling of nutrients in wetland soil. Being high in humin content, the detritus layer in continuous marsh wetlands can be used as a soil amendment for improving soil quality.

This work was financially supported by the Evans-Allen/USDA-NIFA Grant, NCX-259-5-11-130-1.