Food, beverage, and feedstock processing facility wastewater: A unique and underappreciated source of contaminants to U.S. streams

Abstract

Process wastewaters from food, beverage, and feedstock facilities, although regulated, are an under-investigated environmental contaminant source. Food process wastewaters (FPWW) from 23 facilities in 17 U.S. states were sampled and documented a plethora of chemical and microbial contaminants. Of 576 analyzed organics, 184 (32%) were detected at least once, with concentrations as large as 143 μg L⁻¹ (6:2 fluorotelomer sulfonic acid) and as many as 47 detected in a single FPWW sample. Cumulative per/polyfluoroalkyl substance concentrations up to 185 μg L⁻¹ and large pesticide transformation product concentrations (e.g., methomyl oxime, 40 μg L⁻¹, clothianidin TMG, 2.02 μg L⁻¹) were observed. Despite 48% of FPWW receiving disinfection prior to discharge, bacteria resistant to third-generation antibiotics were found in each facility type and multiple bacterial groups were detected in all samples, including total coliforms. Exposure-activity ratios and toxicity quotients exceeded 1.0 in 13 and 22% of samples, respectively, indicating potential biological effects and toxicity to vertebrates and invertebrates associated with the discharge of FPWW. Organic contaminant profiles of FPWW differed from previously reported contaminant profiles of municipal effluent and urban storm water, indicating FPWW is another important source of chemical and microbial contaminant mixtures discharged to receiving surface waters.

Graphical Abstract

Introduction

Wastewater generated from food, beverage, and feedstock production is treated via a variety of primary, secondary, and tertiary processes. Given the range in products of these facilities (e.g., fruit and vegetable processing, meat processing, dairy, oil and fat processing), the wastewater generated is a potential source of complex mixtures of environmental contaminants to receiving surface waters. In the United States, food process wastewaters (FPWW) may be distributed to municipal wastewater treatment facilities (MWWTF) but many FPWW discharge directly to surface water under National Pollutant Discharge Elimination System (NPDES) permits. FPWW are currently monitored for basic parameters and are generally characterized by high biochemical oxygen demand (up to 5,000 mg L⁻¹), chemical oxygen demand (up to 15,000 mg L⁻¹), total suspended solids (up to 6,000 mg L⁻¹), and nutrients (e.g., various nitrogen and phosphorus species; up to 900 mg L⁻¹), at levels often exceeding those of MWWTF discharges.^1–5 The quality and quantity of FPWW varies depending on the type of food processing and on daily fluctuations in facility process volume.^1,6

Little research has been conducted in the United States to date on organic chemicals in FPWW, however, a few studies in Europe detected several pesticides in vegetable and fruit wastewater and pharmaceuticals and personal care products in retail chicken, ground beef, and milk.^7–9 Anerobic bacteria has been isolated from food wastewater in Asia and pathogenic and antibiotic-resistant Escherichia coli (E. coli) have been isolated in cattle and veal and poultry slaughterhouse wastewater in Europe.^10–15

MWWTF and urban stormwater (StW) discharges are widely considered to be important environmental sources of a variety of legacy and contaminants of emerging concern, including pharmaceuticals, personal care products, per/polyfluoroalkyl substances (PFAS), pesticides, hormones, and bacteria and antibiotic resistant bacteria, with well-documented implications for stream and aquatic health.^16–24 Comparable understanding of contaminant exposures from FPWW is currently lacking and critical, as more than 5,000 food, beverage, and feedstock NPDES permitted facilities exist across the United States.^25,26

To address this knowledge gap, the U.S. Geological Survey (USGS) and U.S. Environmental Protection Agency (USEPA) conducted a nationally distributed synoptic assessment of contaminant exposures from 23 FPWW in 17 states in the continental United States. FPWW was analyzed for 576 organics, 32 inorganics, and 15 bacterial groups. Potential aquatic health concerns were explored using cumulative exposure-activity ratios (Σ_EAR)²⁷ associated with in vitro bioactivity and cumulative toxicity quotients (Σ_TQ) associated with aquatic life benchmarks for fish, invertebrates, and aquatic plants. This paper focuses on characterizing chemical and bacterial growth exposure and potential effects of FPWW.

Materials and Methods

Site Selection

Twenty-three sites were selected from 17 U.S. states and a variety of food manufacturing North American Industry Classification System (NAICS) codes,²⁸ translating to seven facility types (Figure 1, Table 1). All FPWW except one was treated at the respective facilities before discharge (Table 1). Nine of the 23 facilities discharged to MWWTF. Twenty-one facilities used biological treatment (15 tertiary, 6 secondary), one used primary treatment only, and one did not treat prior to discharge to surface water; 11 incorporated ultraviolet (UV), chlorination, or steam disinfection prior to discharge (Table 1). Nine facilities combined FPWW with sanitary wastewater from the facility before treatment (Table 1).

Figure 1.

National distribution of the 17 states (shaded) from which 23 process wastewater samples were collected in 2018. States that are crosshatched each contained two samples.

Table 1.

Select Ancillary Information for Investigated Food, Beverage, and Feedstock Processing Facilities.¹

Short name	NAICS code	NAICS description	Average MGD	Sample collection	Combined with FPWW sanitary?	Treatment level	Biological treatment used	Disinfectant/sterilizer used
BVRG-1	311213	Malt Manufacturing	1.50	Final effluent	No	Tertiary	sequencing batch reactor	NA
BVRG-2	312140	Distilleries	0.91	D/S from outfall	No	Secondary	sequencing batch reactor	NA
BVRG-3	312120	Breweries	0.23	Final effluent	No	Not defined	NA	steam
DAIRY-1	311513	Cheese Manufacturing	0.50	Final effluent	No	Tertiary	activated sludge	ultraviolet
DAIRY-2	311511	Fluid Milk Manufacturing	0.15	Before mixing with non-contact cooling water	Yes	Tertiary	activated sludge; clarifier	chlorination
DAIRY-3	311513	Cheese Manufacturing	0.40	Outfall submerged in stream	No	Tertiary	aeration tank; clarifier	NA
ETHNL-1	325193	Ethyl Alcohol Manufacturing	0.22	Final effluent	No	NA	NA	NA
FRTVG-1	311421	Fruit and Vegetable Canning	0.95	Final effluent	No	Secondary	surface impoundments; oxidation ponds	NA
FRTVG-2	424480	Fresh Fruit and Vegetable Merchant Wholesalers	0.01	Downstream of outfall	No	Primary	NA	NA
FRTVG-3	311421	Fruit and Vegetable Canning	1.56	Final effluent	No	Tertiary	activated sludge; clarifier	chlorination
FRTVG-4	311313	Beet Sugar Manufacturing	5.00	Final effluent	No	Tertiary	activated sludge	NA
FRTVG-5	311421	Fruit and Vegetable Canning	1.04	Final effluent	No	Tertiary	aeration lagoons; clarifier	NA
FRTVG-6	311421	Fruit and Vegetable Canning	0.30	Final effluent	Yes	Tertiary	sequencing batch reactor	ultraviolet
MEAT-1	311615	Poultry Processing	1.22	Final effluent	Yes	Tertiary	activated sludge; clarifier	ultraviolet
MEAT-2	311615	Poultry Processing	1.30	Final effluent	No	Tertiary	activated sludge	chlorination
MEAT-3	311612	Meat Processed from Carcasses	3.00	Final effluent	Yes	Secondary	aeration lagoons	NA
MEAT-4	311611	Animal (except Poultry) Slaughtering	1.83	Final effluent	Yes	Tertiary	sequencing batch reactor	chlorination
MEAT-5	311611	Animal (except Poultry) Slaughtering	2.00	Final effluent	No	Tertiary	activated sludge	ultraviolet or chlorination
MEAT-6	311615	Poultry Processing	1.10	Final effluent	Yes	Tertiary	activated sludge	ultraviolet
MEAT-7	311119	Other Animal Food Processing	0.32	Before mixing with MWWTF influent	Yes	Secondary	activated sludge	NA
SEAFD-1	311710	Seafood Product Preparation and Packaging	0.26	Downstream of outfall	Yes	Secondary	aeration lagoons	NA
SEAFD-2	311712	Fresh and Frozen Seafood Processing	0.04	Before mixing with MWWTF influent	No	Secondary	chlorine	chlorination
SOYOIL-1	311224	Soybean and Other Oilseed Processing	0.18	Final effluent	Yes	Tertiary	aeration lagoons	NA

¹NAICS, North American Industry Classification System²⁸; NA, not available; D/S, downstream; MWWTF, municipal wastewater treatment facility; FPWW, food process wastewaters; MGD, million gallons per day

Sample Collection

FPWW samples were collected once at each of the 23 facilities from July to November 2018. Sampling protocols and all necessary supplies (e.g., bottles, preservatives) were provided to sampling staff to ensure consistency in collection methods. FPWW grab samples were collected from a flume, pipe, or as it entered the receiving water by USGS staff using established trace-level protocols.²⁹ In total, there were 17 final effluent samples, 3 surface water samples collected immediately downstream of the effluent pipe, 2 effluent samples collected before mixing with MWWTF influent, and 1 effluent sample collected before mixing with facility non-contact cooling water (Table 1). FPWW samples were immediately chilled at 4 °C and shipped on ice the same day as sample collection to USGS laboratories for analysis. Complete sampling details for this study are provided elsewhere.³⁰

Analytical Methods

FPWW samples were analyzed by USGS laboratories using 10 target organic (576 unique analytes), 13 inorganic (32 unique analytes), and 18 microbial (15 bacterial groups) methods (Table S1, S2).³⁰ Analyses conducted at USGS laboratories included the Organic Geochemistry Research Laboratory (OGRL) in Lawrence, Kansas (liquid chromatography-tandem mass spectrometry (LC-MS/MS) antibiotics;³¹ LC-MS/MS glyphosate, glufosinate, and aminomethylphosphonic acid (AMPA);³² LC-MS/MS ionophores and mectins;³³ LC-MS/MS steroid hormones, hormone conjugates, phytoestrogens, progestins, and mycotoxins),³⁴ the National Water Quality Laboratory (NWQL) in Denver, Colorado (gas chromatography mass spectrometry (GC/MS) volatile organic compounds (VOCs);³⁵ LC-MS/MS pesticides and pesticide transformation products (TPs);³⁶ neonicotinoids;³⁷ LC-MS/MS human-use pharmaceuticals, pharmaceutical metabolites, polar organic compounds;³⁸ LC-MS/MS PFAS;^39,40 and nutrients^41–44), the Redox Chemistry Laboratory (RCL) in Boulder, Colorado (trace elements^44–47), and the Michigan Bacteriology Research Laboratory (MIBaRL) in Lansing, Michigan (bacterial growth^30,48). Method information is available in Tables S1−S2 and in the companion data release.³⁰

Data handling, Quality Assurance, Statistics, and ∑_EAR, Σ_TQ analysis

Quality assurance/quality control included two field blanks and laboratory blanks, spikes, surrogates, and replicates.³⁰ Bacterial growth and inorganic analytes above the reporting limit were not detected in the field blanks. Five organic analytes (dichloromethane, 0.43 and 0.26 μg L⁻¹; trichloromethane, 0.12 and 0.07 μg L⁻¹; ethyl acetate, 0.7 μg L⁻¹; hexamethylenetetramine, 0.052 μg L⁻¹; 1,4-dichlorobenzene, 0.024 μg L⁻¹) were detected in at least one field blank (Table S8);³⁰ corresponding field results were censored at two times the maximum blank detection level (e.g. dichloromethane, 0.86 μg L⁻¹), affecting 37 chemical results. For hexamethylenetetramine, the analyzing laboratory issued an additional censoring value of 0.150 μg L⁻¹.⁴⁹ The overall median surrogate recovery for applicable methods (pesticide, pharmaceutical, VOC, PFAS) was 96% (Table S9).³⁰ Quantitative (≥ limit of quantification, LOQ) and semi-quantitative (< LOQ, ≥ method detection limit, MDL) results were considered detections.^50–52 Laboratory reported estimated values were used as reported for statistical analyses. Complete site information, additional method details, and complete chemical and microbial results, including quality assurance/quality control and inorganic results (not reported), can be found in the accompanying data release.³⁰

All statistical analyses were conducted in R statistical software.⁵³ Principal component analysis (PCA; factoextra version 1.0.7)⁵⁴ was used to explore correlation patterns of organic concentration results from the FPWW and MWWTF effluent and StW results from a parallel study using a similar set of target analytes. Relations were considered significant when p < 0.05. Plots were constructed in R statistical software using ggplot2 (version 3.3.2),⁵⁵ cowplot (version 1.0.0),⁵⁶ and gridExtra (version 2.3)⁵⁷ graphics packages. Exposure-activity ratios (EARs) and toxicity quotients (TQs) were computed and plotted in R using toxEval (version 1.2.0; ToxCast^™ database version 3.2).⁵⁸

Potential adverse organic chemical-contaminant effects were assessed by comparison to the USEPA Office of Pesticide Programs (OPP) Aquatic Life Benchmarks (ALBs).⁵⁹ Effects of mixed organic contaminants were estimated using the toxEval R package.⁵⁸ Detected organic chemicals were compared to the ALB to explore potential pesticide-exposure effects based on in vivo exposures. To estimate mixture effects, individual TQs of different chemicals were summed by each taxa and chronic or acute effects to provide a Σ_TQ by facility. Detected organic chemicals were compared with ToxCast high-throughput effects data to determine potential biological effects based on in vitro exposures. Individual EARs (ratio of detected concentration to activity concentration at the cutoff (ACC) from ToxCast)⁶⁰ were summed to provide Σ_EAR by facility. The Σ_EAR provides a relative assessment across sites because the same chemicals were monitored in each sample. Maximum EARs for each chemical across all sites (EAR_chem) were also calculated to prioritize detected chemicals based on their potential for biological effects (Table S10–S12). Individual EAR_chem ≥ 1 reflect exposure concentrations shown to trigger the observed biological activity in an in vitro test system whereas an EAR < 1 indicates a lower probability of biological activity.^61,62 An EAR ≥ 0.001 has been used as a level of potential concern and was used as a precautionary effects-screening threshold.^63–65 However, an EAR or Σ_EAR should not be interpreted in the same manner as a TQ based on ALBs (in vivo toxicity studies). Rather, EARs are intended as a prioritization tool that takes both the concentrations of the individual contaminants detected and their relative potencies in relation to effects on specific biological pathways into context. Not all these pathway responses would necessarily be considered adverse. Collectively, consideration of such ratios has been viewed as a relative ranking tool for comparing among sites or chemicals or as a lower bound (conservative) estimate of in vivo adverse effect levels.⁶⁶

Results and Discussion

Overall, 184 organic contaminants were detected at least once with concentrations up to 143 μg L⁻¹ and as many as 47 detected in a single FPWW sample (median = 22); 48% were pesticides or pesticide TPs, 18% VOCs, 14% pharmaceuticals, 9% PFAS, 6% hormones/ phytoestrogens/mycotoxins, and 5% antibiotics (Table S4). Of the largest 20 detected concentrations (i.e., > 3.50 μg L⁻¹), 12 were VOCs, followed by four pesticides or pesticide TPs, three PFAS, and one pharmaceutical. Eleven of the 12 largest VOC detections were detected in either MEAT-5 or SEAFD-2, two of the four pesticide detections were in MEAT-2, all three PFAS detections were in SOYOIL-1, and the largest pharmaceutical detection was in SEAFD-1 (Table S4). Of the 23 most frequently detected organics (i.e., ≥ 30%), 12 were pesticides or pesticide TPs, followed by five VOCs, three pharmaceuticals, two PFAS, and one hormone. Food Processing Wastewater Contaminant Comparison to Municipal Wastewater Effluent and Urban Stormwater

It was not unexpected that FPWW contained a broad suite of organics as MWWTF effluent^{16–21,67,68} and StW²⁴ have been previously documented to contain a plethora of organics, including pharmaceuticals, hormones, pesticides, VOCs, PFAS, and personal care products. For instance, 108 and 121 organics were detected in two MWWTF effluent (EFF) and three StW samples collected in 2018 and 2019 from a study in Oklahoma⁶⁹ with a similar set of target organic analytes, respectively.

When organic contaminant concentrations for FPWW were compared to EFF and StW,⁶⁹ the FPWW profiles clustered together, apart from both the EFF and StW (Figure 2A; Tables S4, S16). Two hundred and twenty-seven organic contaminants were analyzed in FPWW, EFF, and StW by the same analytical methods and were detected at least once. With six principal components (PCs), 60.3% of the total matrix variation could be explained (Table S5A).³⁰ The EFF and StW were influenced by PC1 (top six contributions were pharmaceuticals) and PC2 (top six contributions included five herbicides and one pharmaceutical), respectively, whereas FPWW clusters near the origin of the PCs, indicating average properties (Figures 2A, S1A). Previous studies have demonstrated that MWWTF effluent is not the only source of in-stream contamination.^24,64,70 Our characterization of FPWW, which yielded a different contaminant profile than other known sources, indicates FPWW as another important source of organic chemical mixtures to streams in the United States.

Figure 2.

Principal components analysis (PCA) of organic compound concentrations of food process wastewater (FPWW), municipal wastewater treatment facility effluent (EFF), and municipal stormwater (StW) (A); PCA of FPWW (B). FPWW legend definitions can be found in Table 1.

PCA was used to compare organic contaminant profiles among FPWW samples (Figure 2B). With six PCs, 59.3% of the total matrix variation could be explained (Table S5B).³⁰ PCA demonstrated that the majority of FPWW samples clustered together, with the exception of SEAFD-2 and SOYOIL-1. SEAFD-2, sampled before co-mingling with MWWTF influent, is mostly driven by large concentrations of disinfection byproducts and VOCs, whereas SOYOIL-1 is driven by large concentrations of PFAS and AMPA (Figure S1B; Table S4).

Food Process Wastewater as Source of Per- and Polyfluoroalkyl Substances

Seventeen PFAS compounds were detected in FPWW, ranging from 1 to 15 detected per facility, and at least one PFAS compound was detected in 15 PFWW samples (65% of facilities). Perfluorooctanoic acid (PFOA) and perfluorobutanesulfonic acid (PFBA) were most commonly detected (35% of facilities; Table S4).³⁰ The presence of perfluorooctane sulfonate (PFOS), PFOA, and other PFAS in the environment originates from their industrial and consumer use and subsequent release into the environment along with indirect sources.⁷¹ Many of the products produced by facilities in this study originate from the production of livestock and crops. Research has documented multiple agricultural pathways for PFAS exposure, including municipal biosolids application, municipal wastewater irrigation, livestock waste, fertilizers, and pesticide containers.^72–77 In addition to the possible raw food material source, industrial uses of PFAS could contribute to the detections in FPWW; the U.S. Food & Drug Administration lists gaskets, O-rings, and other parts used in food processing equipment, processing aids to reduce build-up on manufacturing equipment, and grease-proofing agents in paper and paperboard food packaging as authorized uses of PFAS in food contact applications.⁷⁸ The variation in both the number of PFAS detections and concentrations is supported by the variety of industry types sampled. To our knowledge, this is the first documentation of FPWW as an environmental source of PFAS.

The PFAS profile varied greatly among the FPWW samples, with cumulative PFAS (Σ_PFAS) concentrations ranging from less than detection to 185 μg L⁻¹ (median 0.002 μg L⁻¹; SOYOIL-1; Figure S2). SOYOIL-1 had the most PFAS detected (15) and the largest Σ_PFAS, due to the large detections of 6:2 fluorotelomer sulfonic acid (6:2 FTS; 143 μg L⁻¹) and perfluoropentanoic acid (PFPeA; 27.3 μg L⁻¹).³⁰ In addition to the largest Σ_PFAS, all but two of the PFAS detected in this study were found in SOYOIL-1. For the 15 compounds detected in SOYOIL-1, all of them had the largest concentration in the study. PFBA, perfluorohexanoic acid (PFHxA), and perfluoroheptanoic acid (PFHpA) also exceeded 1 μg L⁻¹ (Table S4). For comparison, the EFF samples ranged from Σ_PFAS 0.040 to 0.104 μg L⁻¹ and StW ranged from Σ_PFAS 0.015 to 0.095 μg L⁻¹ and fall within ranges detected by a study of 19 MWWTF in Australia (0.0093 to 0.520 μg L⁻¹).^69,79 SOYOIL-1 Σ_PFAS was three orders of magnitude larger than both the Australian MWWTF and the other FPWW in this study. Interestingly, the compound 6:2 FTS was found to be an important contributor to PFAS discharges from Australian WWTPs, was strongly associated with the proportion of industrial or commercial discharges to the influent, and did not significantly decrease between influent and final effluent.⁷⁹ The slow biotransformation of 6:2 FTS has previously been documented and is attributed to a microbial aerobic de-sulfonation rate-limiting step in MWWTF activated sludge.⁸⁰

SOYOIL-1 PFAS concentrations were also compared to other known PFAS dischargers. PFAS manufacturing facility wastewater was sampled in 2005 and analyzed for 13 PFAS; sample concentration data available ranged from less than detection to 104 μg L⁻¹, with Σ_PFAS 291 μg L⁻¹,⁸¹ similar to detections at SOYOIL-1. Large PFAS concentrations have also been detected in groundwater monitoring wells near Department of Defense military installations (PFOA+PFOS); off-base wells had concentrations ranging from 0.075 to 19.0 μg L⁻¹, whereas on-base wells had concentrations ranging from 0.074 to 10,970 μg L⁻¹.⁸² While these wells vary widely in concentrations, many are similar to concentrations observed in SOYOIL-1 FPWW.

Food Process Wastewater as Source of Pesticide and Pesticide Transformation Products

Ninety pesticide compounds (61 herbicides and herbicide TP, 17 insecticides and insecticide TP, 12 fungicides and fungicide TP) were detected (4.3 to 70% detection frequency, DF). Methomyl oxime, a TP of the insecticide methomyl, was the largest pesticide concentration observed in the study (40 μg L⁻¹; 4.3% DF) and was found in MEAT-2, a poultry processor (Tables 1, S4). This detection of methomyl oxime did not have a corresponding methomyl parent compound detection, demonstrating why evaluation of TPs is crucial in environmental research. Methomyl is used for chemical control of flies in poultry facilities⁸³ and is not commonly found in streams and rivers.⁸⁴ Methomyl can degrade in the presence of Cl⁻, but the most effective degradation is by microbial digestion,^85–87 both of which are used by MEAT-2 (Table 1). Therefore, this poultry processing facility could have received large concentrations of methomyl from the poultry facilities that was degraded into TP during food processing and/or treatment. While toxicity data is lacking for the TP methomyl oxime,⁸⁴ methomyl is moderately to highly toxic to fish and very highly toxic to aquatic invertebrates at concentrations as low as 8.8 μg L⁻¹.⁸⁷

Metolachlor sulfonic acid (SA), a metolachlor TP, had the third largest pesticide concentration observed (2.96 μg L⁻¹; 43% DF) and was found in MEAT-4, an animal (except poultry) slaughtering facility (Tables 1, S4). Meat, specifically meat processing facilities, are not typically thought of as a source of pesticides and this warrants further investigation. Metolachlor is a broad-spectrum herbicide primarily applied to corn, soybeans, and sorghum. Corn is the primary U.S. grain feed, accounting for > 95% of feed grain production and use.⁸⁸ In addition to MEAT-4, a relatively large concentration of metolachlor SA was observed in ETHNL-1 (1.71 μg L⁻¹) indicating corn (and corn as animal feed) as the primary source of metolachlor SA in FPWW.

AMPA, a TP of glyphosate (a widely used herbicide that controls broadleaf weeds and grasses), had the second and fourth largest pesticide concentration observed (14 μg L⁻¹, SOYOIL-1; 2.7 μg L⁻¹, FRTVG-5) and was the most detected pesticide in the study (70%; Table S4). Glyphosate and AMPA have been detected in beer and other food products such as oat-based breakfast cereal and honey^89–92 and are commonly present in feed used for livestock^93,94. AMPA was detected in FPWW at a variety of facilities that were directly and indirectly related to crops (e.g., ETHNL, SOYOIL, FRTVG, BVRG); however, AMPA was also detected in DAIRY and MEAT FPWW, indicating animal feed as an additional potential source of glyphosate and AMPA (Table S4). Glyphosate was detected in 43% of samples, and 100% of the glyphosate detections were in the presence of a corresponding AMPA detection.

The fifth largest pesticide concentration detected was a clothianidin TP, clothianidin TMG (2.02 μg L⁻¹), in wastewater from FRTVG-4, a beet sugar manufacturing facility (Tables 1, S4). This sample also had a relatively large concentration of the parent compound clothianidin (0.269 μg L⁻¹); however, it was 10-times smaller than the TP concentration. Neonicotinoid treated beet sugar seeds are typically grown in Europe and the United States and as the seeds germinate, the active substance is systematically taken up by the roots and distributed to the plant;^95–100 as such, the likely source of clothianidin to the facility. There is currently a lack of data on TMG occurrence in water resources; however, the parent compound clothianidin has been frequently detected in the environment.^37,101,102 Clothianidin has been reported to transform during chlorination of drinking water;¹⁰³ however, chlorination was not part of the wastewater treatment system for FRTVG-4. Clothianidin was previously found to persist during wastewater treatment and is episodically discharged from wastewater effluent.^104,105

Bacterial Growth Ubiquitous Across Food Process Wastewater Facility Categories

Wastewater disinfection reduces the risk of human exposure to pathogenic microorganisms, therefore disinfection procedures such as UV or chlorination are often used at MWWTF prior to discharge into surface water.¹⁰⁶ Eleven out of the 23 FPWW sampled (48 %) received disinfection or sterilization (chlorination, UV, or steam) prior to discharge and nine of these samples were final effluent (Table 1). Even with nearly half of the FPWW receiving disinfection prior to sampling, multiple bacterial groups were detected in all samples, including gram-positive and gram-negative bacteria and total coliforms (Figure 3, Tables S6, S7). The high counts of bacteria indicate high residuals of bacteria are possible even after disinfection (bacterial counts were not correlated with facility disinfection; Table S6). SEAFD was the only facility type to have at least one detection in all tested bacterial growth categories, both SEAFD-1 and SEAFD-2 FPWW had high counts of Total Coliforms and E. coli, and SEAFD-2 had high counts of extended spectrum beta-lactamase (ESBL) resistant E. coli.

Figure 3.

Detected bacterial growth (MPN 100mL⁻¹ or CFU 100mL⁻¹) by facility type. Triangles indicate growth exceeded the assay limit; circles are bacteria counts. Campylobacter spp., Vibrio spp., and Mycobacterium spp. are presence/absence data shown as percent of facilities in each type. Total coliforms, E. coli, ESBL-resistant total coliforms, ESBL-resistant E. coli, Carbapenemase-producing total coliforms, and Carbapenemase-producing E. coli shown in MPN 100mL⁻¹. Staphylococci, oxacillin-resistant staphylococci, and lactobacilli shown in CFU 100mL⁻¹. MPN, MPN, most probable number; CFU, colony-forming unit; E. coli, Escherichia coli; ESBL, extended spectrum Beta-lactamase; <, less than. *No data available for BVRG-2 Mycobacterium spp. or DAIRY-3 Campylobacter spp.

The ubiquitous detection of oxacillin resistant staphylococci and ESBL resistant total coliforms (except SOYOIL-1) indicates the presence of antibiotic resistant bacteria growth in the FPWW released from every facility type sampled. The detection of antibiotic resistant bacteria, including ESBL resistant and carbapenemase-producing bacteria (subgroup of carbapenemase-resistant organisms), is of public health concern as this indicates resistance to several antibiotic classes such as third generation cephalosporins, carbapenems, and penicillins, which are among the most commonly prescribed antibiotics in the United States.¹⁰⁷

E. coli, a bacterium with potential pathogenicity, was found in every facility type sampled and ESBL resistant E. coli was detected in BVRG, MEAT, and SEAFD facility categories. E. coli is often part of state water-quality standards under the Clean Water Act (CWA). To put FPWW in context, 2021 recreational water-quality criteria (RWQC) for E. coli range from 100 to 410 colony-forming unit (CFU) 100 mL⁻¹,¹⁰⁸ whereas FPWW ranged from < 1 to > 2,000 CFU 100 mL⁻¹ (median 129 CFU 100 mL⁻¹; Table S6). While a direct comparison cannot be made between FPWW and these recreational criteria, we can postulate that many of the small receiving waters in this study may exceed the RWQC.

Campylobacter spp., the etiologic agent of Campylobacteriosis, is a leading cause of bacterial foodborne diarrheal disease and while most commonly found in chickens, has also been reported in other animal products including shellfish;¹⁰⁹ Campylobacter was detected in every facility type sampled (Figure 3). Staphylococci and oxacillin resistant staphylococci were found in every facility type and at levels exceeding other bacteria in this study. The presence of staphylococci can be widespread, indicating improper sanitation, but could also be spread by employees in the processing plant.¹¹⁰ Lactobacilli was detected at high counts across all facility categories sampled. Lactobacilli is used as a probiotic and in food fermentation and preservation, commonly in dairy products, fermented vegetables, sausages, and silage.¹¹¹ The ubiquitous detection of lactobacilli, coupled with the frequent use in food processing, indicates that lactobacilli may be a possible tracer of FPWW.

Vibrio spp. is commonly found in marine systems and aquaculture;¹¹² however, Vibrio spp. was detected in every facility type sampled (Figure 3). Vibrio spp. presence is not commonly studied in non-coastal areas of the United States; however, a recent study isolated Vibrio cholerae from lakes and rivers in northwest Ohio, with several isolates showing resistance to antibiotics.¹¹³ Vibrio organisms have been isolated in 75 to 100% of MWWTF effluents sampled in South Africa, Italy, and France.^7,114,115 Species of Mycobacterium spp., such as M. avium, are highly resistant to chlorine and ozone-based disinfection.^116,117 While only 6 of the 23 facilities used chlorine-based disinfection (Table 1), all samples had growth positive for Mycobacterium spp. (except BVRG-2, no available data; Figure 3).

Potential Organic Contaminant Effects: Exposure Activity Ratios and Toxicity Quotients

The detected organic chemicals were compared with ToxCast high-throughput bioactivity data to determine relative potential to elicit biological effects to vertebrates, particularly for contaminants for which conventional ALBs are not available. EAR_chem results are summarized in Figure S3 and Tables S12–S13.³⁰ Out of the 184 organic chemicals detected in FPWW, data were available for 118 chemicals (64%), and 110 chemicals (60%) had a least one active hit call (dose-response meets criteria; Table S14)¹¹⁸ in ToxCast at the time of access (5/05/2021). Butyraldehyde/butanal, chlorodibromomethane, and perfluorohexanoic acid all had EAR_chem > 1.0 in at least one sample, indicating an exposure previously shown to modulate bioactivity in vitro (Figure S3). Fifty-three organic chemicals had an EAR_chem > 0.001, the precautionary effects screening threshold,⁶⁴ at least once at one or more facilities. Facilities SEAFD-2, MEAT-5, and SOYOIL-1 had Σ_EAR > 1.0 (Figure 4A). SOYOIL-1 had the largest Σ_EAR of 54 and the third largest number of chemicals for which EARs could be calculated (23; Figure 4A). All 23 FPWW facility samples had at least one EAR_chem above the 0.001 effects threshold, indicating potential for biological effects to vertebrates at each facility during the study period. Similar results were reported in a Great Lakes waters study that documented Σ_EAR as high as 58, and four watersheds with at least one sampling location Σ_EAR > 2, all in proximity to MWWTFs.²⁷ Both SOYOIL-1 and MEAT-5 (Σ_EAR = 54 and 8.7, respectively) discharge directly to their receiving water, and the contribution of FPWW may be large enough to elicit biological effects in vertebrates depending on hydrologic conditions and bioavailability.

Figure 4.

Individual (boxes) and cumulative (open triangles) target organic exposure-activity ratio (EAR) by facility (A), Individual (boxes) and cumulative (open triangles) fish and invertebrate aquatic life benchmark (ALBs) toxicity quotient (TQ) by facility (B). Boxes, centerlines, and whiskers indicate interquartile range, median, and 5th and 95th percentiles, respectively.

Detected pesticides were also compared to pesticide ALBs including fish (acute and chronic), invertebrates (acute and chronic), nonvascular plants (acute), and vascular plants (acute) to explore potential pesticide-exposure effects in FPWW (Figures 4B, S4; Table S15).⁵⁹ An individual TQ ≥ 1 indicates an exceedance of toxicity criterion and a potential concern for adverse biological effects;⁶³ a TQ ≥ 0.1 is used as a medium risk effects-screening level.⁶⁴ ALB for fish and invertebrates were available for 41 and 52, respectively, of the 90 pesticides detected (Table S15). Several chemicals had at least one TQ > 1, including carbendazim (acute and chronic fish ALB, chronic invertebrate ALB), imidacloprid (acute and chronic invertebrate ALB), and clothianidin (chronic invertebrate ALB). No chemicals exceeded a TQ of 1 for vascular and nonvascular plant ALBs (Figure S4). MEAT-2 facility had the largest Σ_TQ (79, chronic invertebrates ALBs), with carbendazim exceeding a TQ of 1 for both acute and chronic fish ALBs (3.64, 18.4, respectively) and chronic invertebrate ALBs (5.87). Imidacloprid in MEAT-2 also exceeded a TQ of 1 for both acute and chronic invertebrate ALBs (1.29, 49.7, respectively; Figure 4B, Table S15). In addition, Σ_TQ for chronic invertebrate ALBs exceeded 1 at ETHNL-1, FRTVG-2, FRTVG-4, MEAT-1, and SEAFD-1. Thirty-nine percent of FPWW facility samples (9 of 23) had individual or Σ_TQ ≥ 0.1, at least one individual TQ exceeded 1 in 22% of samples (5 of 23), and Σ_TQ exceeded 1 in 30% of samples (7 of 23), indicating that chronic exposures to approximately one-third of FPWW sampled would potentially cause toxic effects to aquatic invertebrates and fish in receiving waters.

Implications for the health of the aquatic environment

This national-scale study provides the most comprehensive characterization of organic chemical mixtures, biological activity, and microbial contaminants conducted in the United States to date, as related to wastewater generated in food, beverage, and feedstock processing facilities. While many of the organic contaminants detected in FPWW have been previously documented in MWWTF discharges, FPWW profiles of chemical and microbial contaminants differed from other important environmental sources of contaminants (i.e., MWWTF, StW), indicating FPWW as a currently underappreciated source of chemical mixtures to surface waters. In addition, FPWW was a source of both large PFAS and pesticide TP concentrations, which was previously undocumented. The effects from complex contaminants mixtures are poorly understood; however, approximating cumulative exposures using additive assumptions can assess possible risk to the aquatic environment. Many EAR and TQ ≥ 1 indicate potential biological effects and toxicity associated with the discharge from FPWW. Bacteria were documented to be ubiquitously detected during this study, including those with resistance to several types of third generation cephalosporins. While this study was an initial synoptic reconnaissance of FPWW, results indicate FPWW is an important environmental source of a plethora of organic contaminants and bacteria, in a variety of food processing facility categories across the United States with potential ecological health implications. This study underscores the need to further assess the effects of FPWW discharges to the receiving waters.

Synopsis.

Food process wastewater contained a plethora of contaminants and a different contaminant profile than other known sources such as municipal wastewater and urban stormwater and thus, is an underappreciated source of contaminants to U.S. streams.