Environ. Sci. Technol. 2008, 42, 5992–5997
Restored Wetlands as a Source of Disinfection Byproduct Precursors F R A N C I S C O J . D ´I A Z , * , † A L E X T . C H O W , † ANTHONY T. O′GEEN,† RANDY A. DAHLGREN,† AND PO-KEUNG WONG‡ Department of Land, Air, and Water Resources, University of California, Davis, California 95616, and Department of Biology, The Chinese University of Hong Kong, Shatin, NT, Hong Kong SAR, China
Received March 18, 2008. Revised manuscript received May 22, 2008. Accepted May 28, 2008.
The effects of a restored wetland system in the Sacramento Valley, California on the production of dissolved organic carbon (DOC) and nitrogen (DON) and the formation potential of common disinfection byproducts (DBPs: trihalomethanes, haloacetonitriles,andchloralhydrate)wereexamined.Additionally, the effects of photodegradation and microbial degradation on dissolved organic matter properties and reactivity with respect to DBP formation potential (DBP-FP) were evaluated. The wetlands increased DOC and DON concentrations by a factor of 2.2 and 1.9 times, respectively, but had little influence on the DOC and DON quality as compared to their source waters. The increase in DOC and DON concentrations increased the formation potential of all DBP species by >100%. Solar radiation and microbial degradation reduced the trihalomethane formation potential by 24 and 10%, respectively, during a 14 day incubation. In contrast, the chloral hydrate formation potential was increased by 22% after phototreatment. Results indicate that current flood-pulse management practices with a 2-3 week residence time could lead to wetlands acting as a source of DBP precursors. Enhanced DBP-FP is especially important as these wetlands contribute to a watershed that is a drinking water source for more than 23 million people.
1. Introduction Increases in dissolved organic matter (DOM) concentrations over the past two decades have been reported in natural water bodies across Europe and North America (1, 2). Although these trends are the combined result of many different processes (3), it is generally accepted that the quality and quantity of DOM can be related to land use with modifications caused by soil type, vegetation, hydrology, and climate (4). Among the different sources of DOM within watersheds, wetlands are recognized as one of the most important, contributing a disproportionate amount of DOM relative to their land surface area (5). Wetlands are highly productive and tend to increase DOM concentrations and change its chemical characteristics through internal processing and loading (6). As a result, some studies have found a positive correlation between downstream DOM concentration, expressed by dissolved organic carbon (DOC) and * Corresponding author phone: (530)752-0630; e-mail: fjdiaz@ ucdavis.edu. † University of California. ‡ The Chinese University of Hong Kong. 5992
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 16, 2008
dissolved organic nitrogen (DON) levels, and the extent of wetland area (7). Creation of wetlands on agricultural fields that were historically natural wetlands has recently gained popularity in the Central Valley of California. While wetlands provide several valuable ecological services, the creation of wildlife habitat is often the primary goal of most newly created wetlands. A total of 83 000 ha of managed wetlands exist in the Central Valley, and this number is likely to increase substantially in the future (8). Thus, the biogeochemical processes controlling DOM concentrations in surface water, relative to previous land use, will potentially be affected. In addition to potential ecological services, a number of environmental concerns are associated with wetlands, including the formation of disinfection byproduct (DBP) precursors and the methylation of mercury (9). An elevated DOM concentration is a primary water quality concern in the Central Valley of California as 23 million people receive a portion of their drinking water from the Sacramento-San Joaquin River system. The formation of DBPs, such as trihalomethanes (THMs), haloacetonitriles (HANs), and chloral hydrate (CH), during chlorination could have human-health consequences as some of these compounds are suspected to be mutagens, carcinogens, or developmental toxicants if ingested over extended periods of time (10). In response to those potential health hazards, DBP concentrations are regulated in drinking water by the WHO (11), European Union (12), and U.S. EPA (13). Particularly, the Stage 2 Disinfectants and Disinfection Byproduct Rule (Stage 2 DBP rule) was promulgated in the U.S., establishing 80 µg L-1 THMs following chlorination as the maximum contamination level (MCL) for drinking water (13). Additionally, the proposedguidelineforDOCexportedfromtheSacramentoSan Joaquin River Delta as a drinking water source is 0.05).
FIGURE 2. Relationship of DOC and UVA254 and W-2 (n ) 45 samples per location).
nm
from CD, PD, W-1,
In addition to increased DOC concentrations, the characteristics of DOC were slightly different among water bodies, and these characteristics could be affected by plant leaching of DOM and degradation of DOC by sunlight (29). SUVA, a surrogate of DOC aromaticity, was significantly lower in the PD than in CD (2.19 vs 2.45 L mg-1 m-1) (Table 2). The lower SUVA may result from photodegradation as the pond has no vegetative shading and turbidity is low due to sedimentation. SUVA increased again in both wetlands with respect to the PD, possibly due to the production of aromatic compounds derived from wetland plants. Among different surface water sources in the Sacramento-San Joaquin watershed, SUVA ranged from 1.8 to 10.7 L mg-1 m-1 (7, 30), so the differences in SUVA among the four bodies (CD, PD, W-1, and W-2) within this wetland system are relatively minor. DOC was plotted against UVA254 to illustrate the changes in both quantity and quality of DOC (Figure 2). Regression lines for the different locations showed a very strong relationship between DOC and UVA254; however, small differences in slopes (range of 34.4-39.7) were detected. Overall, if all the data were considered together, a strong correlation exists (r2 > 0.92), suggesting that the composition of DOC was similar among sites. Thus, we conclude that the increase in chlorine demand and ∆UVA254 is due primarily to changes in DOC concentration rather than DOC quality. 3.3. DBP Formation. In this study, all waters exceeded 100 µg L-1 THMs following chlorination (Figure 3). Waters from CD and PD were similar in terms of DBP formation and had the lowest THM-FP (mean ∼211 µg L-1). In contrast, mean THM-FP increased to 576 and 459 µg L-1 in the outflows from W-1 and W-2, respectively. In some instances, THM
FIGURE 3. Seasonal THM formation potential from CD, PD, W-1, and W-2 (n ) 24). Plots show the 5th, 25th, 50th, 75th, and 95th percentiles; isolated circles represent extreme values. levels exceeded 800 µg L-1. In addition to THMs, other DBP species, including HANs and CH, were significantly higher in wetland drainage as compared to CD and PD waters (Table 2). CD and PD waters have similar HAN-FP and CH-FP, with mean values of 18 and 14 µg L-1, respectively. After passing through the wetlands, HAN-FP and CH-FP more than doubled (39 and 38 µg L-1, respectively). The BIF in THMs and dibromoacetonitrile/dichloroacetonitrile ratios were low, suggesting that the effect of bromide is minimal in this system. Notably, strong correlations were found between THM-FP, HAN-FP, and CH-FP (Figure 4), suggesting that the formation of HANs and CHs increased linearly with THM formation (17, 31). For every 100 molecules of THMs formed during chlorination, there were ∼8 molecules of HANs and 6 molecules of CH produced. The reactivity of DOC, expressed as specific THM-FP (STHM-FP; THM-FP normalized by DOC concentration), was slightly higher in wetland drainage as compared to CD input waters. STHM-FP increased from 31 µg mg-1 in CD to ∼33-36 µg mg-1 in the two wetlands (Table 2). In particular, W-1 had the largest vegetation coverage and also the highest SUVA and specific DBP-FP, suggesting that vegetation could be a source of DOC and DBP precursors. However, as discussed previously for SUVA, the changes in STHM-FP were relatively minor as compared to the wide range of STHM-FP in the Sacramento-San Joaquin watershed (Delta wetland ) 71 µg mg-1 and Delta drainage ) 24 µg mg-1) and other water bodies (Tomhannock Reservoir, New York is 52 µg mg-1, and Colorado River, Nevada is 14 µg mg-1) (20, 30). Again, the increase of DOC concentration, rather than a change in DOC VOL. 42, NO. 16, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5995
FIGURE 4. Relationship of THM-FP with HAN-FP and CH-FP from CD, PD, W-1, and W-2 (n ) 24).
FIGURE 6. Changes in (a) UV absorbance and (b) DBP-FP due to phototreatment and microbial treatment after 14 days (mean ( SE, n ) 8).
FIGURE 5. Relationship of (a) THM-FP and DOC and (b) HAN-FP and DON from CD, PD, W-1, and W-2 (n ) 24). quality, appears to be the main concern with wetland drainage waters in this study. Both DOC and DON are considered to be precursors in forming THMs, HANs, and CH (16, 17, 32). Regardless of the sources of DOC and DON, linear relationships were found between THM-FP, HAN-FP, and CH-FP to DOC and DON (Figure 5). Slopes of the regression lines in Figure 5a are equivalent to STHM-FP. Considering W-1 as an example, the slope is 3.3 µmol of THMs mmol of C–1, indicating that ∼3 in 1000 C atoms formed THM molecules during chlorination. Similarly, the slope of the regression for HAN-FP versus DON (Figure 5b) represents the reactivity of DON in 5996
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 16, 2008
forming HANs. The reactivity of DON in W-1 is 45 µmol of HANs mmol of N–1, suggesting that 45 in 1000 N atoms formed HANs during chlorination. Particularly, dichloroacetonitrile (DCAN) is unstable, and it can continue to react with chlorine to form THMs and other DBPs (32). Therefore, the increased DON and DOC in the wetland effluents will likely increase the formation of both N-DBPs and regulated THMs. 3.4. Microbial and Photodegradation of DBP Precursors. DBP precursors produced in wetlands could be different from the precursors in downstream water facilities due to alteration during long conveyance times. Independent microbial degradation and photodegradation experiments were conducted to evaluate potential changes in DBP precursors in the natural environment. There was more than a 25% decrease in UVA254 after 14 days of exposure to sunlight, whereas there was no significant change in the 14 day incubation with E. coli in the dark (Figure 6a). The UVA254 decrease in the photodegradation study followed a first-order kinetic response with a rate constant of -0.034 day-1. Furthermore, we did not detect significant changes in DOC concentration in either the microbial degradation or photodegradation experiments. Thus, there was an ∼25% decrease in SUVA after sun exposure. Results are consistent with previous studies that showed that solar radiation was more effective at converting aromatic carbon into CO2 and other soluble organic compounds than microbial mineralization (29, 33). The high molecular weight DOC, which usually comprises the chromophoric DOC fraction, is relatively recalcitrant to bacterial utilization (34, 35). Thus, UVA254 did not significantly change, even though some organic carbon was utilized by the microorganisms.
A significant reduction of THM-FP was observed in both microbial degradation and photodegradation experiments (Figure 6b). There was a 24% decrease in THM-FP in water exposed to sunlight. The reduction was comparable to the changes in SUVA. In spite of no differences in UVA254 and DOC concentration, water incubated with E. coli decreased THM-FP by ∼10%. Results suggested that at least 70% of DBP precursors was relatively recalcitrant to natural degradation. Furthermore, there was a significant increase in CH-FP after phototreatment, suggesting that its precursors were produced during exposure. Although the relative increase in CH was >29%, the actual increase was
