Environ. Sci. Technol. 2007, 41, 5795-5802
Distributions of Pharmaceuticals in an Urban Estuary during both Dryand Wet-Weather Conditions MARK J. BENOTTI† AND BRUCE J. BROWNAWELL* Marine Sciences Research Center, Stony Brook University, Stony Brook, New York 11794-5000
Pharmaceuticals and selected major human metabolites are ubiquitous in Jamaica Bay, a wastewater-impacted estuary at concentrations in the low ng/L to low µg/L range. Concentrations throughout the bay are often consistent with conservative behavior during dry-weather conditions, as evidenced by nearly linear concentration-salinity relationships. Deviation from conservative behavior is noted for some pharmaceuticals and attributed to microbial degradation. Caffeine, cotinine, nicotine, and paraxanthine were detected with the greatest analytical signal, although evidence is presented for in situ removal, especially for nicotine and caffeine. There is little evidence for significant removal of carbamazepine and sulfamethoxazole, suggesting they are more conservative and useful wastewater tracers. Immediately following heavy precipitation, which induced a combined sewer overflow (CSO) event, the concentrations of all compounds but acetaminophen and nicotine decreased or disappeared. This observation is consistent with a simple model illustrating the effect of precipitation has on pharmaceutical concentration in the wastewater stream, given the balance between dilution from rain and the bypass of treatment.
Introduction The widespread occurrence of pharmaceuticals in surface waters has been well documented in areas that are influenced by inputs from municipal wastewater systems (1, 2). Presently, more is known regarding the occurrence of pharmaceuticals in wastewaters and surface waters than about their fate and transport. Many compounds appear to be relatively persistent in the aquatic environment, as indicated by either inefficient removal in biologically treated wastewaters (3), lack of significant attenuation in waters far removed from their likely sources (4), or their detection in slow-moving groundwaters (5). For example, wastewater-derived carbamazepine has been shown to survive sewage treatment (3, 6) and behave conservatively after introduction into groundwater (7, 8). Similarly, caffeine has been shown to act as a useful longterm tracer of municipal wastewater into a lacustrine system (9). Pharmaceuticals typically have properties favoring their use as wastewater tracers. Their physiochemical properties suggest that they are relatively water-soluble and nonvolatile. Additionally, their low natural background levels, combined * Corresponding author phone: (631)632-8658; fax: (631)632-3072; e-mail:
[email protected]. † Present location: U.S. Geological Survey, 2045 Route 112, Building 4, Coram, NY 11727. 10.1021/es0629965 CCC: $37.00 Published on Web 07/13/2007
2007 American Chemical Society
with ever decreasing detection limits, allow for detection even when wastewater is substantially diluted. There is a relative paucity of information on the occurrence of pharmaceuticals in estuarine or marine systems despite the fact that many municipalities discharge wastewater into these environments. Isolated reports investigating pharmaceutical presence in estuaries and marine systems include the presence of 14 pharmaceuticals in five UK estuaries at ng/L concentrations (10). Wiegel et al. (11) developed highvolume extraction methods to determine neutral and acidic pharmaceuticals from seawater, and reported clofibric acid, caffeine, and N,N-diethyl m-toluamide (DEET) concentrations of 0.1-18.6 ng/L in the North Sea. Subsequently, they suggested that caffeine, ibuprofen, and an ibuprofen metabolite served as only qualitative indicators of wastewater discharges to Tromsø Sound (12). On the basis of relationships of caffeine with salinity and proximity to known discharges, it was concluded that caffeine distributions within Boston Harbor were not consistent with conservative mixing, whereas outside the harbor, caffeine concentrations decreased in proportion to increasing salinity (13). Ideally, a conservative wastewater tracer would be employed in a system that is at steady state. Large combined sewer overflow (CSO) events represent perturbations in contaminant inputs that could disrupt steady-state conditions and confound the use of any wastewater tracer. Approximately 46 million people in the United States (mostly in the Northeast, Great Lakes Basin, and Pacific Northwest) are served by CSO systems in which stormwater runoff combines with municipal sewage. Moderately heavy rain or snowmelt can overload the capacity of a wastewater treatment plant (WWTP), leading to discharge of untreated wastewater. In this study, we have characterized the occurrence of a suite of high volume pharmaceuticals, in Jamaica Bay, NY, a sewage-impacted estuary in New York City (Figure 1), during both dry weather and immediately following a heavy precipitation event. One of the study goals was to learn more about the potential of using some of these compounds as wastewater tracers. Data are also presented on the effluent concentrations and removal efficiencies of these analytes at a representative WWTP. A simple model illustrating the effect of a CSO event on the freshwater concentrations of pharmaceuticals entering the bay is also presented. We employed an extraction and high performance liquid chromatography coupled with mass spectrometry (LC-MS) method for a suite of high-volume pharmaceuticals developed by the U.S. Geological Survey (2, 14) with a slightly modified analyte list. Numerous studies have applied LCMS approaches to detect pharmaceuticals in aqueous samples, and most have relied upon single quadrupole and triple quadrupole mass analyzers. Unique to this study was the use of a time-of-flight mass-spectrometer (ToF-MS), which offers some advantages in LC-MS analysis of polar organic compounds (increased resolution, accurate mass capability, and full spectral sensitivity) when compared to quadrupole-based mass spectrometry (15).
Methods Study Site. Jamaica Bay (Figure 1) served as the study site. It is a highly sewage-impacted estuary (16) that has proven to be a suitable site for environmental fate studies of other wastewater contaminants in urban estuaries (17-19). The large urbanized watershed (200 km2) houses 2.05 million residents of the New York City boroughs of Brooklyn and Queens (US Census estimate, 2003). Municipal wastewater VOL. 41, NO. 16, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5795
FIGURE 1. Map of Jamaica Bay, NY, showing sampling sites and four largest WWTPs (/). Sites sampled in October 2004, under “dry” conditions (D01-D24), are designated by those in the main body of the bay (b), Hendrix Creek (0), and Thurston Basin (4). Sites sampled in October 2005 immediately following a heavy rain event (×; W01-W07) are shown. Also listed are average and maximum WWTP flows in millions of gallons per day (MGD). and stormwater are combined in a CSO system, and four municipal WWTPs deliver an average of 261 million gallons per day (MGD) of treated wastewater to the bay (20). The other major source of freshwater is subsurface groundwater discharge that delivers ∼30 MGD (21). Thus, under dry weather conditions, treated wastewater accounts for the largest input of freshwater, and salinity is a reliable tracer of municipal wastewater. Despite a relatively shallow average depth (5 m), mixing inside the bay can be slow: the hydraulic residence time has been estimated as 35 days (16). The northeastern portion of the bay is deeper (up to 20 m) because of historic dredging and receives direct discharge of sewage effluent. The slow mixing of these waters is indicated by lower salinities and much higher wastewater-derived alkylphenol ethoxylate metabolite concentrations (18). During periods of heavy rain, CSOs discharge untreated wastewater from outfalls concentrated along the northern shore of the bay. Shellfishing throughout Jamaica Bay is closed, in addition to selected areas being designated unacceptable for swimming and/or fishing, largely due to large wastewater and CSO inputs of pathogen indicators (22). Sample Collection. Most of the results presented here are for samples collected during two sampling trips characterized by extremely different hydraulic conditions. Twentyfive 1 L surface water samples were collected from a small vessel at 24 sites under dry weather conditions on October 9-10, 2004 (Figure 1). In the preceding 10 days, precipitation had totaled 0.46 cm at the local national weather station at John F. Kennedy Airport (www.ncdc.noaa.gov). More limited sampling of Jamaica Bay surface waters was conducted under relatively dry weather conditions on two other occasions (see Supporting Information and Benotti, 2006 (23)). Seven 1 L surface water samples were also collected near the end of a 5796
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 16, 2007
large storm and corresponding CSO event from areas accessible from shorelines or bridges, on October 13, 2005 (Figure 1). A total of 22.8 cm of rain had fallen in the 10 days leading up to sampling, including 10.8 cm during the prior 24 h (www.ncdc.noaa.gov), inducing a large CSO event. All samples were collected 1-2 m below surface with a stainless steel Kemmerer sampling bottle (Wildco Instruments, Saginaw, MI) equipped with Teflon end seals, transferred to combusted amber glass bottles, and stored on ice for transport to the laboratory. Water temperature and salinity were measured at the time of sampling using a YSI 6600 Multi Parameter Water Quality Monitor (YSI Inc., Yellow Springs, OH). Influent and effluent grab samples were collected from the 26th Ward WWTP on February 8, 2005 at 1100, 1400, and 1700. As shown in Figure 1, the 26th Ward WWTP contributes a significant fraction of the wastewater discharge to Jamaica Bay. The removal efficiency of biological oxygen demand on the date of sampling was 95.4%, which was not different from the long term, seasonally independent, daily average for this plant (93.8 ( 3.2%) (22). All four WWTPs discharging into the bay employ activated sludge treatment and serve both residential and industrial areas of New York City (20). Thus, wastewater pharmaceutical concentrations in the 26th plant should be representative of the three other WWTPs. Sample Extraction and Analysis by LC-ToF-MS. Bay and wastewater samples were filtered and extracted within 48 h, according to guidelines set forth by the USGS for sampling of pharmaceuticals and organic wastewater contaminants (24). Vacuum filtration employed pre-combusted 1.2 µm glass-fiber filters (Whatman, GF/C), after which a surrogate standard was added (100 ng each of 13C3-caffeine and d3codeine). Following filtration, two aliquots of effluent (100
TABLE 1. WWTP Concentrations,a Removal Efficiencies, and Data Derived from Salinity Mixing Plots compound
influent conc. (ng/L)
effluent conc. (ng/L)
removal efficiency (%)
acetaminophen antipyrine caffeine carbamazepine cimetidine codeine cotinine diltiazem fenofibrate fluoxetine hydrocodone ketoprofen metformin nicotine nifedipine paraxanthine ranitidine salbutamol sulfamethoxazole trimethoprim warfarin
61 000 ( 19 000 29 ( 30 42 000 ( 6300 100 ( 78 14 ( 7.3 170 ( 61 7800 ( 550 57 ( 13 nd 600 ( 280 70 ( 31 1000 ( 1300 26 000 ( 17 000 17 000 ( 12 000 nd 55 000 ( 34 000 330 ( 260 13 000 ( 4000 360 ( 210 300 ( 100 170 ( 130
860 ( 710 13 ( 9.5 15 200 ( 4400 65 ( 15 12 ( 6.9 170 ( 40 4000 ( 240 53 ( 25 nd 560 ( 250 8.6 ( 3.5 nd 11 000 ( 7100 2100 ( 1700 nd 25 000 ( 14 000 62 ( 24 8100 ( 3400 140 ( 94 120 ( 71 nd
99 54 64 37 13 3.0 49 7.7 na 7.5 88 >99 59 87 na 54 81 36 62 60 >99
pred. effluent conc. (ng/L)
r2
MDL (ng/L)
dynamic range
370
0.43 1.48 579 nd in surface waters 2400 0.57 11.5 1330 120 0.70 0.98 66.6 35 0.47 0.106 112 110 0.86 0.412 407 1300 0.51 3.18 1260 33 0.77 0.105 499 nd in surface waters isobaric interference (see text) 28 0.83 0.235 36.6 nd in surface waters only detected in three samples 43 0.57 1.04 2060 only detected in three samples 5200 0.60 4.93 5060 nd in surface waters isobaric interference (see text) 110 0.67 1.42 97.9 190 0.81 1.58 76.3 nd in surface waters
a Measured influent and effluent concentrations are mean values ((standard deviation) from samples collected over 6 h at the 26th Ward WWTP: Influent samples include one taken at 1100, two at 1400, and two at 1700 (total n ) 5); effluent samples include two samples each taken at 1100, 1400, and 1700 (total n ) 6). See text for a description of predicted effluent concentrations, MDL, and dynamic range. nd, not detected. na, not applicable.
and 1000 mL) and influent (10 and 100 mL) were extracted and analyzed separately to ensure that measured concentrations fell within the dynamic range of LC-ToF-MS analysis. For SPE extraction, samples were loaded at a rate of ∼15 mL/min onto SPE cartridges (Oasis HLB), which had been preconditioned with methanol and DI water. They were eluted sequentially with methanol and methanol containing trifluoroacetic acid (0.1%): eluents were concentrated under a gentle stream of N2 in a 35 °C bath, reconstituted to 1.0 mL to approximate initial LC mobile phase composition, filtered using 0.45 µm syringe filters (Millipore Millex-HV), and stored at 2 °C. Internal standard (d6-phencycline) was added (100 ng/mL concentration), and 10 µL of sample was injected. The LC operation (Waters 2695), positive ionization electrospray parameters (ZSPRAY), and ToF-MS (MicroMass LCT) conditions are fully described elsewhere (15) and in the Supporting Information. Raw data files were processed using the software’s all file accurate mass measure (afamm) process, and resultant data files were used for quantification. On selected samples, more precise accurate mass estimates were determined for each compound detected using the raw data files (15). Pharmaceutical concentrations were calculated from relative response to surrogate 13C3-caffeine (d3-codeine for codeine and hydrocodone), but not explicitly corrected for recoveries. Method detection limits (MDLs) were calculated from dry-weather bay samples and represent concentrations producing a signal-to-noise ratio of 3. They ranged from 0.105 (diltiazem) to 11.5 ng/L (caffeine; Table 1). The precision of the analytical method was generally
