Double dip: Sampling contaminants of emerging concern two ways improves accuracy

UNIVERSITY PARK, Pa. — Two are better than one when it comes to sampling methods for tracking the location and timing of toxic chemical contamination in waterways, according to researchers at Penn State. The team tested two typical sampling approaches and found that both are needed to accurately assess both the variety and concentration of contaminants of emerging concern — pharmaceuticals, personal care products, livestock manure-borne hormones, veterinary antibiotics and pesticides — in a watershed.

The team published their findings in the Journal of Natural Resources and Agricultural Ecosystems.

“Even at trace levels, contaminants of emerging concern in surface waters can threaten aquatic organisms and human health,” said study senior author and team leader Heather Preisendanz, professor of agricultural and biological engineering in the College of Agricultural Sciences. “This study addresses uncertainty about the occurrence, fate and transport of environmentally relevant concentrations of contaminants. We wanted to determine the best way to detect trace amounts of chemicals in streams, so we can determine where and when those chemicals are entering a watershed.”

The researchers compared the results of grab sampling — collecting bottles of water at specific moments — to the results of sampling with polar organic chemical integrative samplers (POCIS) — devices left in the stream for days or weeks. Grab sampling, known as active sampling, reveals exactly what is in the stream at the instant the sample is taken — think of it like taking a snapshot, Preisendanz explained. In contrast, the POCIS slowly absorbs certain chemicals over time. That passive sampling method shows the average exposure during that period, more like a time-lapse composite image.

The team has monitored five sites in the 24-square-mile Halfmoon Creek watershed located in an agricultural area in central Pennsylvania since 2021. For this study, they collected both types of samples from the five sites every two weeks during the 2023 and 2024 crop-growing seasons, May through September. The five sites were arranged in a nested watershed design, meaning that they that were placed at key locations to help researchers understand how contaminants varied across different sections and tributaries in the watershed.

The nested design allowed the researchers to better understand how well each sampling method was able to capture spatial and temporal — space and time — patterns of where contaminants occur and how concentrations change over time. It also helped them learn how the long-term averages measured by POCIS compared with the instantaneous measurements from grab samples.

In addition, monitoring multiple locations helped identify “hotspots.” Because the researchers sampled five sites instead of just one at the watershed outlet, they could identify parts of the watershed where contaminant levels were consistently higher. These hotspots may indicate nearby pollution sources, such as agricultural fields or wastewater inputs, Preisendanz said.

While each method captured contaminants — most commonly herbicides Atrazine and Simazine, insecticide Clothianidin and the stimulant caffeine, with each appearing in at least 68% of samples collected by both methods — each method also missed significant information.

The POCIS passive samples found chemicals more often than grab samples, likely because if a chemical appears only briefly — for example, after rainfall or pesticide application — a grab sample might miss it. POCIS keeps collecting for weeks, Preisendanz pointed out, making it more likely to detect that chemical if it was present in the stream over that time.

POCIS was also better able to capture the pollution source differences across sites. When the team compared the overall mixtures of chemicals at each site, POCIS showed clearer differences among locations than grab samples did. This suggests, the researchers said, that passive sampling may be better for comparing long-term contamination patterns across a watershed.

However, grab samples showed bigger seasonal changes with concentrations that changed more dramatically over the season. This makes sense, Preisendanz noted, because they measure only one moment in time. If sampling happens right after a storm or pesticide application, concentrations may be much higher than a week later. Although POCIS may be able to detect that the chemical was present, it would be unable to capture the rapid change in concentration from low to high.

The study showed that there isn’t one perfect sampling method, Preisendanz said.

“We found that the most complete picture of water contamination comes from combining one-time water samples with long-term passive samplers at multiple sites, because each method captures different aspects of how contaminants move through streams,” she said. “Grab samples are good for measuring short-term spikes and seasonal changes, while POCIS are good for detecting chemicals that may be missed by grab samples and for estimating average exposure over time.”

As new conservation practices are adopted across the Halfmoon Creek watershed and the larger Chesapeake Bay watershed, Preisendanz added, these sampling methods could help to characterize the co-benefits of these practices for mitigating pesticides and emerging contaminants, in addition to reducing sediment and nutrient loads.

Henry Kibuye, a doctoral degree candidate in the Department of Agricultural and Biological Engineering, was first author on the study. Tyler Groh, assistant research professor in the Department of Ecosystem Science and Management and watershed management extension specialist, and Tameria Veith, agricultural engineer with the U.S. Department of Agriculture’s Agricultural Research Service, contributed to the research.

This research was funded by the U.S. Department of Agriculture (USDA) under grant number 2023-67019-39708 and the USDA’s National Institute of Food and Agriculture Federal Appropriations under project numbers PEN04985, PEN04870 and PEN04874 and accession numbers 7008079, 7005711 and 7005702. This content is solely the responsibility of the authors and does not necessarily reflect the views of the funders.