Environ. Sci. Technol. 2009, 43, 5647–5651
Inputs of Fossil Carbon from Wastewater Treatment Plants to U.S. Rivers and Oceans D A V I D R . G R I F F I T H , * ,†,‡ R E B E C C A T . B A R N E S , †,§ A N D PETER A. RAYMOND† School of Forestry and Environmental Studies, Yale University, New Haven, Connecticut 06511, MIT/WHOI Joint Program in Oceanography, Woods Hole, Massachusetts 02543, and U.S. Geological Survey, Boulder, Colorado 80303
Received February 9, 2009. Revised manuscript received May 4, 2009. Accepted May 6, 2009.
Every day more than 500 million cubic meters of treated wastewater are discharged into rivers, estuaries, and oceans, an amount slightly less than the average flow of the Danube River. Typically, wastewaters have high organic carbon (OC) concentrations and represent a large fraction of total river flow and a higher fraction of river OC in densely populated watersheds. Here, we report the first direct measurements of radiocarbon (14C) in municipal wastewater treatment plant (WWTP) effluent. The radiocarbon ages of particulate and dissolved organic carbon (POC and DOC) in effluent are old and relatively uniform across a range of WWTPs in New York and Connecticut. Wastewater DOC has a mean radiocarbon age of 1630 ( 500 years B.P. and a mean δ13C of -26.0 ( 1‰. Mass balance calculations indicate that 25% of wastewater DOC is fossil carbon, which is likely derived from petroleumbased household products such as detergents and pharmaceuticals. These findings warrant reevaluation of the “apparent age” of riverine DOC, the total flux of petroleum carbon to U.S. oceans, and OC source assignments in waters impacted by sewage.
ages), which is attributed to inputs of preaged soil organic matter, resuspended sediments, and black carbon derived from fossil fuel combustion (10, 11). In contrast, riverine DOC is typically 14C-enriched (modern radiocarbon ages), reflecting inputs of recently fixed carbon from aquatic and terrestrial primary production (3, 11-15). The fact that some rivers and estuaries exhibit 14C-depleted DOC (2, 11, 12, 14) suggests long residence times, preferential removal of young DOC, and/or inputs of preaged DOC (16, 17). Unfortunately, preaged DOC sources like soil organic matter and petroleumderived compounds are not well constrained. The chemical composition of wastewater treatment plant (WWTP) effluent is determined by a combination of inputs (residential, industrial, stormwater) and treatment methods. Both factors are generally constant for a single WWTP but may vary between WWTPs. Thus, it is unsurprising that effluent DOC concentrations span a large range (2-33 mg L-1) (4, 18), but remain relatively stable for any one WWTP. Finally, wastewater is produced at a continuous rate, and therefore wastewater DOC could play an increasingly dominant role in the carbon dynamics of receiving rivers at times of low river flow (5). Toxins in WWTP effluent have been well characterized for obvious ecological and human health reasons. There is also interest in identifying wastewater-specific markers and measuring the lability of wastewater OC once it is released to natural waters (19). Molecular markers of wastewater include natural products like coprostanol and synthetic detergent compounds such as linear alkylbenzenes, nonylphenols, and fluorescent whitening agents among many others (19). Wastewater dissolved organic matter contains a much larger “hydrophilic” fraction than natural dissolved organic matter, and this may influence its utilization in aquatic environments (18, 20). Remarkably, little is known about the carbon isotopic content of wastewater. While the 13 C signature of wastewater POC and sludge (δ13C ∼ -23 to -24‰) (21-23) has been used to track physical and trophic transport mechanisms, wastewater DOC isotopes remain largely uncharacterized. The present study was undertaken to address this need.
Experimental Section Introduction Rivers export 0.21 Pg (1 Pg ) 1015 g) of dissolved organic carbon (DOC) to the oceans each year (1), yet significant gaps exist in our understanding of DOC dynamics in rivers and estuaries and how riverine DOC impacts coastal oceans. Riverine DOC characteristics are determined by organic carbon (OC) sources and in situ transformations. Major source pools include soil organic matter, plant and algal biomass, and anthropogenic compounds such as petroleum hydrocarbons (2-6). A combination of biodegradation, photochemistry, and sorption to suspended solids and bed sediments alters DOC composition in rivers and estuaries (3, 7-9). Carbon isotopes (13C and 14C) have been widely used to describe sources, sinks, and residence times of OC in natural waters. Generally, riverine particulate organic carbon (POC) tends to be 14C-depleted (decadal to millennial radiocarbon * Corresponding author phone: 508-289-3396; fax: 508-457-2164; e-mail:
[email protected]. † Yale University. ‡ MIT/WHOI Joint Program in Oceanography. § U.S. Geological Survey. 10.1021/es9004043 CCC: $40.75
Published on Web 05/29/2009
2009 American Chemical Society
The stable and radiocarbon isotopes of DOC and POC were measured in the effluent of 12 WWTPs within the Hudson and Connecticut River watersheds (Table 2). These WWTPs are broadly representative of U.S. WWTPs because they span a wide range of sizes (0.04 - 11.65 m3 s-1) and process predominantly residential wastewater using biologically activated sludge, the most widely used treatment method (24, 25) (Table 2). In addition, six of these WWTPs were sampled during three different months (December, February, and June) to identify seasonal trends. Effluent samples were collected in acid washed (10% hydrochloric acid) 1 L polycarbonate bottles at the effluent access pipe within each WWTP. This water was filtered through a precombusted 47 mm diameter quartz filter (Whatman QMA; 1 µm) to separate the dissolved and particulate organic carbon fractions. The filtered water used for DOC measurements was collected in duplicate acid washed 125 mL polycarbonate bottles. POC filters were wrapped in combusted aluminum foil and placed in individual bags. All samples were processed with clean 14C collection techniques (acid washed or combusted equipment) and stored on ice during transit to the laboratory. DOC samples were then acidified to pH ∼ 2.5 with 1 mL VOL. 43, NO. 15, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5647
of 60% phosphoric acid, and frozen until analysis. POC filters were stored frozen, then acidified with concentrated (100%) sulfurous acid (26) and dried under a stream of ultra high purity N2 prior to analysis. Organic carbon was converted to CO2 by high-energy UV irradiation (2400 W; DOC) or high-temperature combustion (850 °C; POC), and subsequently purified on a gas extraction line at Yale University (27, 28). DOC and POC concentrations were determined by measuring CO2 pressure with a calibrated Baratron absolute pressure gauge. Samples were then catalytically converted to graphite and analyzed for 14C and 13C at the National Ocean Sciences Accelerator Mass Spectrometry (NOSAMS) Facility (Connecticut samples) or the NSF-Arizona AMS Laboratory (Hudson samples). Radiocarbon ages are reported according to accepted conventions (29, 30). The analytical precision for these 14C and 13C analyses is currently ∼ (3-7‰ (31) and ∼0.1‰, respectively. An isotope mass balance (eqs 1-3) was used to calculate the relative contributions of petroleum (fP), C3 plants (fC3), and C4 plants (fC4) to “average” wastewater DOC and POC (isotope values: δ13CWW and FmWW). FmWW ) fP(FmP) + fC3(FmC3) + fC4(FmC4)
(1)
δ13CWW ) fP(δ13CP) + fC3(δ13CC3) + fC4(δ13CC4)
(2)
1 ) fP + fC3 + fC4
(3)
These calculations assume average source pool isotope values for petroleum (δ13CP ) -28‰, FmP ) 0), C3 plants (δ13CC3 ) -27‰, FmC3 ) 1.09), and C4 plants (δ13CC4 ) -12‰, FmC4 ) 1.09). Here, δ13C and Fm are defined as
δ13C )
[
13
Rsample
13
Fm )
Rstandard
]
- 1 × 1000
14
Rnorm / 14Rmodern
(4)
(5)
where R is 13C/12C or 14C/12C, 14Rnorm is the sample ratio normalized to a δ13C of -25‰, and 14Rmodern is defined as 0.95(14Rstandard oxalic acid). The sensitivity of our results to a range of end-member δ13C assignments is presented in Figure 2.
Results and Discussion Table 1 summarizes the concentration and isotopes of OC in wastewater effluent. Overall, mean δ13C values for DOC (-26.0‰) and POC (-23.3‰) varied by 1‰ (SD) or less, despite differences in WWTP size, effluent OC concentration, and season (Figure 1; Table 1; Table 2). The δ13C value for wastewater POC reported here is consistent with prior studies (21-23). Mean DOC and POC radiocarbon ages were 1630 and 780 years B.P. respectively (Figure 1; Table 1). At the time of sampling, all WWTPs used secondary treatment via biologically activated sludge. WWTP-11 also used activated carbon, which is reflected in the anomalously old radiocarbon ages of effluent POC (Table 1). We speculate that the modern 14 C POC values found in WWTP-1 may be due to small amounts of biomedical 14C tracer waste (Table 1). Millennial OC ages suggest that municipal wastewaters contain significant quantities of fossil carbon. Any product that is made from petroleum and reaches WWTPs through sewers and storm drains is a potential source of fossil carbon, including motor oil, tire rubber, pharmaceuticals, food additives, personal care products, and detergents (surfactants). Effluent was sampled during dry conditions, so it is
TABLE 1. 13C and 14C in WWTP Effluent DOC and POCa WWTP
date sampled
DOC (mg L-1)
δ13C DOC (‰)
Fm DOC
DO14C age (y B.P.)
POC (mg L-1)
δ13C POC (‰)
Fm POC
PO14C age (y B.P.)
WWTP-1 WWTP-2 WWTP-3 WWTP-4 WWTP-5 WWTP-6 WWTP-7 WWTP-7 WWTP-7 WWTP-8 WWTP-8 WWTP-8 WWTP-9 WWTP-9 WWTP-9 WWTP-10 WWTP-10 WWTP-10 WWTP-11 WWTP-11 WWTP-11 WWTP-12 WWTP-12 WWTP-12 averagec std deviationc
8/1/06 8/1/06 8/1/06 8/14/06 8/14/06 8/14/06 12/8/05 2/21/06 6/6/06 12/5/05 2/21/06 6/6/06 12/7/05 2/23/06 6/9/06 12/5/05 2/22/06 6/6/06 12/7/05 2/23/06 6/9/06 12/7/06 2/21/06 6/8/06
nd 13.31 16.43 10.19 10.79 6.59 9.81 6.47 6.86 4.14 3.25 4.10 5.23 5.91 4.38 4.35 0.88 4.55 11.71 17.50 7.88 4.35 6.89 6.27 8.7 4
nd -25.4 -26.8 -27.3 -27.9 -25.5 -24.4 nd -24.8 -26.9 nd -26.1 -25.9 nd nd -25.8 nd nd -25.5 nd -24.8 -25.5 nd -25.0 -26.0 1
nd nd 0.843 0.808 0.851 0.866 0.858 nd 0.810 1.009 nd 0.801 0.838 nd 0.749 0.757 nd nd 0.755 nd 0.719 0.814 nd 0.773 0.819 0.05
nd nd 1364 1707 1290 1153 1230 nd 1690 -72b nd 1790 1420 nd 2320 2230 nd nd 2260 nd 2650 1660 nd 2070 1630 500
5.91 5.32 6.93 4.58 19.70 1.80 2.19 2.08 2.13 1.53 1.54 1.25 1.70 3.16 1.96 1.50 1.95 6.70 1.68 1.39 7.35 2.68 0.70 1.79 4.9 5
-23.4 -22.7 -22.3 -23.0 -23.4 -23.4 -23.2 -23.5 -23.5 -23.1 nd -23.5 -23.0 -22.7 -24.0 -22.8 -23.2 -22.9 -24.2 -24.2 -24.6 -26.1 nd nd -23.3 0.5
1.602 0.904 0.875 nd 0.924 1.006 nd 0.945 nd nd 0.936 0.908 nd nd 0.851 nd 0.931 nd nd 0.575 0.347 nd 0.938 0.948 0.94 0.3
-3788b 811 1073 nd 634 -50b nd 454 nd nd 535 775 nd nd 1297 nd 575 nd nd 4441 8509 nd 515 435 780 2000
a Plants 1-6 are located in the lower Hudson River watershed. All others are located within the Connecticut River watershed. For definitions of “Fm” and “δ13C” see the Experimental section. Radiocarbon (14C) age is defined as -8033 × ln(Fm) and is expressed in years before present (y B.P.) where “present” is defined as AD 1950. Additional WWTP characteristics are presented in Table 2. nd ) No data. b Negative radiocarbon ages are typically reported as “>modern”. c WWTPs weighted equally.
5648
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 15, 2009
FIGURE 1. DOC and POC isotopes in WWTP effluent (squares), plotted with C3, C4, and petroleum source pools (circles). Error bars are ( 1 standard deviation.
TABLE 2. Individual WWTP Characteristicsa,b WWTP
basin
flowc (m3 s-1)
dominant source
treatment process
activated carbon?
WWTP-1 WWTP-2 WWTP-3 WWTP-4 WWTP-5 WWTP-6 WWTP-7 WWTP-8 WWTP-9 WWTP-10 WWTP-11 WWTP-12
HR HR HR HR HR HR CR CR CR CR CR CR
0.39 1.05 11.65 0.22 0.18 3.99 0.04 0.26 0.35 0.13 0.22 0.04
residential residential residential residential residential residential residential residential residential residential residential residential
BAS BAS BAS BAS BAS BAS BAS BAS BAS BAS BAS BAS
N N N N N N N N N N Y N
a
All WWTPs were sampled during dry conditions to avoid combined sewage overflows and groundwater seepage sources. b HR ) Hudson River; CR ) Connecticut River; BAS ) biologically activated sludge. c 2005 average.
unlikely that storm drains were a significant source of motor oils and tire rubber (32). Furthermore, hydrophobic compounds (hydrocarbons and oils) tend to sorb onto settling particles, which are removed as sludge (33). Finally, although a large percentage of pharmaceutical and personal care compounds survive wastewater treatment, their mass flux into the municipal waste stream is small compared to that of surfactants (34, 35). Currently, total surfactant production is 22 Tg y-1 (1Tg ) 1012 g) (36). And if the average surfactant is 60% carbon by weight, then ∼ 13 Tg of surfactant OC is produced each year, or about 6% of the global riverine DOC flux (1). Global estimates indicate that 50% of these surfactants, including soaps, are made entirely from “modern” animal and plant products, whereas the other half is synthesized primarily from petrochemical precursors (37, 38). Radiocarbon analyses of two common detergents (“Cascade” and “Sparkleen” powder) used in the U.S. indicate that they contain ∼95% petroleum carbon (average Fm ) 0.0513). Calculating the exact flux of petrochemical surfactant carbon through WWTPs requires significantly more data about local surfactant usage and composition, and degradation efficiencies within sewers and treatment plants. Nonetheless, surfactants seem to be a likely source of fossil carbon to the municipal waste stream. Modern surfactants are designed so that primary biodegradation (i.e., loss of surface activity) is rapid in sewers and WWTPs (35, 36). Yet, surfactant biodegradation me-
FIGURE 2. Contributions of C3 and C4 plants and petroleum to wastewater DOC and POC. Error bars indicate how results change when each source pool is assigned a range of δ13C values (C3 plants: -23‰ to -34‰; C4 plants: -9‰ to -17‰; petroleum: -18‰ to -34‰). The contribution of petroleum to wastewater DOC and POC is determined solely by radiocarbon (FmP ) 0), and thus is insensitive to the petroleum δ13C assignment. tabolites are often much more resistant to biodegradation (35). Surfactant-derived carbon that enters a WWTP can be sorbed onto particles and removed as sludge, fully oxidized within the treatment plant and released as CO2 or CH4 gas, incorporated into bacterial biomass, or discharged as effluent (35). Our 14C results and the large input of petrochemical surfactants to WWTPs lead us to hypothesize that surfactantderived fossil carbon is largely responsible for the observed radiocarbon age of OC in WWTP effluent. If true, then the observed older age of DOC could reflect preferential partitioning of surfactant-derived fossil carbon into the dissolved phase (as soluble metabolites or bacterial exudates). This interpretation is consistent with several studies showing that wastewater DOC contains numerous small hydrophilic compounds and surfactant metabolites (18, 20, 35, 39). A recent study found that a widely used family of surfactants (alkylphenol ethoxylates) was present at similar concentrations in U.S. WWTP influent despite differences in season and geographic region (40). This has intriguing implications for the uniform isotopic signature of WWTPs studied here, and the possibility that our results might apply to larger geographic regions. An isotope mass balance calculation (see eqs 1-3) suggests that 25% of WWTP effluent DOC derives from petroleum sources, 67% from C3 plants (e.g., wheat and soy), and 8% from C4 plants (e.g., corn and sugar cane; Figures 1 and 2). In contrast, effluent POC reflects a higher percentage of C4 plants (24%) and a lower percentage of petroleum sources (14%; Figures 1 and 2), which could be related to larger contributions from human feces and the prevalence of corn in the U.S. diet. To account for natural variability in source pool δ13C and small isotope effects (∼1‰) (41-44) related to heterotrophic respiration of OC during wastewater treatment, we repeated our mass balance calculations using a range of source pool δ13C values (C3 plants: -23‰ to -34‰; C4 plants: -9‰ to -17‰; petroleum: -18‰ to -34‰). These sensitivity tests affected the contributions of C3 and C4 plants but not of petroleum, which is determined solely by radiocarbon (Figure 2). Except for a few outliers, there is a consistent offset between DOC and POC radiocarbon ages (Table 1; Figure 1). And because WWTP effluent, on average, contains 3× more DOC than POC (Table 1), our mass balance calculations suggest that the dissolved phase (
