Introduction

Contaminants of emerging concern (CECs) are either compounds newly introduced into the environment or those that may have been in the environment for years but are just now able to be sampled with current laboratory analytical methods. Natural compounds, such as hormones, and man-made compounds, such as pharmaceuticals, could be considered CECs. They include many categories of compounds, including hormones, pharmaceuticals and personal care products, various compounds found in wastewater, flame retardants, per- and polyfluoroalkyl substances (PFAS), and pesticides. In 2013, Pennsylvania Department of Environmental Protection (DEP) initiated a study of CECs in flowing surface waters using passive water samplers. Samples have been collected at many sites across the state and is ongoing. This summary encompasses results from 2013 through 2017. A detailed report and data are located on DEP’s website at https://www.dep.pa.gov/Business/Water/CleanWater/WaterQuality/Pages/CECs.aspx at “Passive Sampler Analyses 2013 - 2017” and “Passive Sampler Data 2013 - 2017”.

Methods Summary

Data Collection Protocols

Polar organic chemical integrative samplers (POCIS) and semi-permeable membrane devices (SPMDs) were utilized to collect data for several CECs throughout Pennsylvania rivers and streams according to DEP’s Passive Water Chemistry Data Collection Protocol (Williams 2017). POCIS are used to collect hydrophilic compounds (likely to mix with or dissolve in water). SPMDs are used to collect lipophilic compounds (combine with or dissolve in fats/lipids). Passive samplers were typically deployed instream for about 30 days then retrieved and sent to various labs for analysis. The samplers are composed of filters or membranes attached to carriers. The carriers are placed in metal canisters to protect them while out in the water. Once retrieved, POCIS filters and SPMD membranes are processed into small ampules to be analyzed for various compounds.

CEC data was collected at 68 sites throughout Pennsylvania from 2013 through 2017. Typically, sites sampled in any given year were sampled over multiple seasons to allow for seasonal comparisons. Some sites were designated as core sites and were sampled over multiple years; others were sampled only one year. Use the interactive map below to view each passive sampler site and information about that site.

Explanatory Variable Analyses

Correlation analyses were completed to determine if any relationships existed between percent passive sampler compounds detected and various explanatory variables. The variables evaluated are in the table below.

Explanatory variables included in correlation analyses
Category	Parameter	Details
Land Use	% LowDev	% Developed - Open + % Developed - Low Intensity
	% MedHighDev	% Developed - Medium Intensity + % Developed - High Intensity
	% Hay	% Hay/Pasture
	% Crops	% Cultivated Crops
	% Forested	% Evergreen Forest + % Mixed Forest + % Deciduous Forest
Other	Drainage Area	Area drained to sample site (square miles), large (>1000 mi2) or small (<1000 mi2)
	Flow	Calculated as percent of average flow from last 30 days to long-term monthly median flow

Results

Common Compounds

From 2013 through 2017, 395 compounds were tested from various CEC groups. Although many chemicals were detected, 70.4% of final environmental passive sampler results were non-detect. There were 283 compounds detected at least once during this time, 22 of which were detected in more than 90% of the samples in which they were analyzed (see table below).

Compounds Detected >90% of Time, 2013 - 2017
Compound	Description
Hormone
Androstenedione	natural & artificial steroid hormone
Estrone	weak, natural form of estrogen
Pesticide
Atrazine	herbicide
Metolachlor	herbicide
N,N-Diethyl-meta-toluamide (DEET)	insecticide
Polycyclic Aromatic Hydrocarbon (PAH)
Benzo(b&j)fluoranthenes	formed from variety of combustion sources
Benzo[e]pyrene	formed from variety of combustion sources
Chrysene	formed from variety of combustion sources
Fluoranthene	formed from variety of combustion sources
Pyrene	formed from variety of combustion sources
Pharmaceutical
Carbamazepine	anticonvulsant /analgesic drug
Desvenlafaxine	treats depression
Erythromycin-H2O	antibiotic
Fexofenadine	antihistamine
Fluconazole	anti-fungal
Lidocaine	pain reliever
Metoprolol	treats heart failure, high blood pressure
Nicotine	in cigarettes, supplements
Tramadol	used to treat pain
Triamterene	water pill
Venlafaxine	SSNRI; treats depression, anxiety; Effexor
Wastewater
Methyl-1h-benzotriazole	deicing fluid

PAHs were among the most frequently detected compounds (fluoranthene, chrysene, benzo(b&j)fluoranthenes, and pyrene), with several detected in more than 97% of the samples in which they were analyzed. Carbamazepine was the most detected pharmaceutical, detected in 156 of 160 samples in which it was analyzed. Estrone was the most common hormone (156 detections of 161 samples). Atrazine was the most common pesticide (141 detections of 148 samples) followed by metolachlor (89 detections of 94 samples). The polybrominated diphenyl ether (PBDE) compound with the highest number of detections was PBDE-47, with 143 detections of 160 samples. PCBs (considered as their own CEC category) had 16 detections of 603 total analyses, all of which were at very low levels. Of seven PCBs tested for, only Arochlor-1254 and Arochlor-1248 were detected (5 and 11 times, respectively).

Explanatory Variable Analyses

An analysis was performed to determine if watershed size explained any CEC data variability. Sites were grouped based on the size of the watershed upstream of the sample location. Watersheds were organized into two groups, those >1000 square miles (large drainage areas) and those <1000 square miles (small drainage areas). The results of this analysis indicated that low correlations existed in large drainage areas when percent detected compounds were compared to abiotic explanatory variables, including land cover. In small drainage areas, percent detected compounds increased as both percent low-intensity development and percent medium/high-intensity development increased. Percent compounds detected showed a slight negative correlation with percent forested land cover, indicating that percent compounds detected decreased with increased percent forested.

Next, percent detections were compared across the following compound categories: pharmaceuticals, pesticides, hormones, PAHs, PBDEs, and wastewater compounds. Since there were very few detections within the semi-volatile and PCB categories, these were not analyzed separately. These correlation analyses are a statistical way to show if variables may be related. The results indicated that in small drainage areas:

Percent pharmaceutical detections were most strongly correlated with development land cover categories, meaning that, out of the land use categories analyzed, percent pharmaceutical detections were most strongly related to the percent developed land cover in an area.
Percent pesticide detections were most strongly correlated with percent low-intensity developed land cover, percent high development, and percent forested, but not as strongly with percent hay/pasture or percent crops.
Percent hormone detections were not strongly correlated with any explanatory variable; the highest correlation was with percent hay/pasture land cover.
Percent PAH detections were most strongly correlated with percent low-intensity developed land cover and percent medium/high-intensity developed land cover and was negatively correlated with percent forested land cover, meaning that PAH detections decreased with increasing percent forested land cover.
Percent PBDEs were most strongly correlated with percent forested land cover than percent low-intensity developed land use and percent medium/high-intensity developed land use.
Percent wastewater compounds were most strongly correlated with percent medium/high intensity developed land, then percent low-intensity developed land.

View the percent detections of different compound categories in each sample by navigating the interactive map below. Note: Dots located at a site represent the general area of sampling; locations of dots are jiggled at a site so they do not overlap and are visible.

View the percent detections of all compounds tested in each sample by navigating the interactive map below.

Local-Scale Patterns

Smaller subsets of the main dataset were investigated here.

The effect of point-source sewage treatment plant (STP) discharges on the percent of CEC compounds detected was investigated by comparing sample results from sites upstream and downstream of two STPs. Samples were collected upstream and downstream of the Lancaster Sewage Treatment Plant located on the Conestoga River and the Hollidaysburg Borough Sewage Authority located on the Frankstown Branch Juniata River. Sites downstream of the STPs generally had higher percentages of compounds detected than at the upstream sites.

Sites were categorized into main channel sites that were at the center of waterbodies versus near-shore sites along the banks, and large river sites versus tributary sites. Differences between main river channel sites versus near-shore sites were investigated. Near-shore sites tended to have higher numbers of compounds detected than main channel sites. In addition, differences between larger river sites versus tributaries for all compounds was investigated. The two groups were not significantly different; however, pesticides and PBDEs did tend to have higher percent detections in tributaries than large rivers.

Seasonal Patterns

Samples were categorized into spring, fall, or winter collection timeframes. There was no significant difference observed with percent compounds detected among seasons. There was a difference detected for semi-volatile compounds, with most, although few, detections occurring in spring. For pesticides, spring tended to have more percent detections than winter and fall. For PAHs, winter tended to have more percent detections than spring and fall.

Discussion

The passive sampler studies implemented by DEP show the extent and patterns associated with CECs in flowing surface waters of Pennsylvania. Roughly 72% of the CEC compounds tested were detected at some point during the study; however, most individual passive sampler results were non-detect. This suggests that CECs do persist in Pennsylvania’s surface waters, but typically at low levels and they are only detected sporadically. The most commonly detected compounds were PAHs, which are often found in road runoff, plasticizers, wastewater, and fertilizers. In this study, pesticides as a group were detected less frequently than most other categories of CECs. This follows the idea that newer pesticides likely break down faster.

Developed land or urban area appears to be a variable influencing the occurrence of many compounds in smaller streams, with developed urban areas having higher percent compounds detected. Percent forested land cover in smaller streams generally had an inverse relationship with percent compounds detected (i.e., smaller streams with highly forested watersheds had fewer CEC detections). These results are logical considering most CEC compounds measured are anthropogenic and would be expected to increase with increased human population density.

In this study, the percent CEC compounds detected were higher downstream of STPs surveyed than upstream; this was particularly true for pharmaceuticals. Although results are somewhat limited in scope, this suggests that traditional wastewater treatment does not remove all pharmaceutical compounds prior to discharge. Nearshore locations also tended to have more compounds detected than main channel locations, suggesting that compounds may accumulate along banks or may be higher in those locations due to runoff being a more predominant nearshore influence. While streamflow appeared to be an influence on detections across years and sampling events, it was not strongly correlated with percent compounds detected overall.

Ongoing DEP passive sampling data collection efforts will continue to document the presence of CECs throughout Pennsylvania surface waters. In addition to the compounds described in this summary, per- and polyfluoroalkyl substances (PFAS) are currently being monitored with passive samplers in Pennsylvania streams. While passive samplers have, to date, not been used for the assessment of protected uses in Pennsylvania, the increasing amount of data collected indicates that the development of assessment methods using passive sampler data is very likely, especially for contaminants that are harmful to human health or aquatic life at very low concentrations. Passive samplers also create the opportunity to measure contamination over longer periods of time and capture acute pollution events that discrete data collection efforts cannot always robustly or accurately characterize. Given their sensitivity, passive samplers are valuable tools for detecting low-level emerging contaminants in Pennsylvania’s streams and rivers.

Literature Cited

Williams, A. 2017. Passive water chemistry data collection protocol. Chapter 4.5 in Water Quality Monitoring Protocols for Streams and Rivers (2021). Pennsylvania Department of Environmental Protection. Harrisburg, Pennsylvania.