Upstate New York Pesticides in Water 2021-2025

Summary of results

Results are available through Year 3 (2023) of the Cornell Soil and Water Lab’s (SWL) five-year Memorandum of Understanding with New York State Department of Environmental Conservation (NYSDEC) to monitor potential mobilization of pesticides residues into groundwater under normal label-compliant use in upstate New York.

Two key principles have undergirded SWL’s ongoing research program for NYSDEC: 1) a primary emphasis on sites with greater than average vulnerability due to soil, aquifer, and topographic characteristics in combination with known or presumed pesticide use, and 2) maintaining strict landowner and location confidentiality to encourage participation.

Once or twice during 2022 (year 2), and twice during 2023 (year 3) and 2024 (year 4), Cornell personnel sampled 21 categorical groundwater sites which are a primary focus of the program. These sites were selected based on falling into one of eight land-use categories: golf courses, sod farms, other managed turfgrass, greenhouses, outdoor nurseries, fruit/vegetable farms, vineyards, and utility/railroad rights-of-way (ROWs). Each site has one to five sampling locations including landowner domestic or irrigation wells, shallow Cornell-installed monitor wells, and in a few cases groundwater-fed ponds. Sampling locations were classified as within (located within a pesticide use area), downgradient (generally downhill from pesticide use areas), or upgradient (generally uphill of the landowner pesticide use areas, more likely reflecting neighbors’ pesticide use). Total samples in 2022 (our sampling ramp-up year) and 2023 (full sampling of the completed site roster) reached 204, representing 75% of our cumulative sample load to date. 2024 sampling is almost complete as of early 2025.

Beginning in 2023, complementing the categorical sites are 33 long-term groundwater sites sampled once per year. These sites are intended for long-term, lower-intensity, tap-based sampling of the landowner’s well (typically one per site). The sites were selected for apparent vulnerability based on topography and potential nearby pesticide use. Some of these sites were first sampled in the initial years of our county-based sampling dating back to 2003, providing valuable past data. The long-term site roster was completed and fully sampled in 2023, reaching 37 samples collected (14% of cumulative total).

The 2021-2025 project also includes a small sampling effort at four lakes (Chautauqua, Canadarago, Upper Little York and Waccabuc), yielding a total two-year lake sample count of 30 (11% of the total).

Analysis by the NYSDEC laboratory of these groundwater and surface-water samples for nearly 90 pesticides and pesticide degradation products provided over 23,000 test results through the 2023 samples. Of the tests performed on groundwater samples (categorical and long-term sites), 99.1% were non-detects (Figure 1). Of the detections, 0.41% were very low level (0.01-0.1 micrograms per liter (µg/L)), 0.35% were low level (0.1-1.0 µg/L), and only 0.07% reached a moderately low level (>1 µg/L).

Popular agricultural herbicides and their breakdown products (DPs) were the most common detections (Figure 2).

There were no exceedances of any pesticide groundwater standards among these detections, continuing the pattern of findings of our sampling program since its inception around 2000. Detections were dominated by pesticide breakdown products.

In contrast, the lake samples had greater detection frequencies but at similarly low levels. These detections were again dominated by herbicides, with detections of metolachlor OA (degradation product) detected in 37% of lake samples, atrazine in 30%, metolachlor ESA (degradation product) in 16%, and 2,4-D in 13%. S-metolachlor and insecticide imidacloprid were detected in two lake samples, and fungicide mefentrifluconizole in one.

The 2023 annual report and these web pages contain the program’s first efforts to interpret (==> more detail) observed groundwater detection patterns in terms of site (soil, aquifer, land use) and pesticide characteristics. Although, as expected, most detections occurred with pesticides having the weakest soil adsorption (sticking) strength (as reflected in their KocKoc​ property¹), a surprising number of pesticide detections occurred for moderately and strongly adsorbed pesticides. Similarly, detections occurred for analytes across the entire range of half-lives (representing relative persistence in the environment), although most were for analytes with longer half-lives.

Land use categories (of categorical sites) had widely-ranging detection frequencies, although these were somewhat associated with the prevalent soil types common to several land uses. Examination of detection patterns for sampling locations for categorical sites showed those within the application areas unsurprisingly had the greatest detection frequencies, while only the breakdown products having the weakest adsorption (smallest KocKoc​ properties) were found in downgradient wells.

Relationships between detections and presumed site vulnerability (based on site soil textures or drainage class) also yielded unexpected results. Soils like Long Island’s found in glacial outwash, river alluvium, and glacial era lakeshore deposits had about the same frequency of groundwater detections of active ingredients as soils derived from muckland that presumably would adsorb pesticides tightly instead of letting them leach downwards with percolating water. Sites having soils that tended to have preferential flowpaths were relatively more likely to have detections, particularly of breakdown products.

Given the size, scope and complexity of the database of detections and site/pesticide characteristics, the team invested heavily in improved approaches for analysis and synthesis. First was the development and initial publication² of the Theoretical Groundwater Ubiquity Score (TGUS), which is an improvement of the long-used Groundwater Ubiquity Score (GUS)³. Both approaches use the ratio of degradation to sorption parameters, t_1/2 ⁴ and K_oc​, to predict leaching potential and yield a higher leaching index for longer half life or lower sorption. The TGUS approach adds site properties and analytical detection limits for assessing the likelihood of detectable leaching within a specified period of leaching vulnerability, as well as testing which characteristics contribute most strongly to detectable leaching. The TGUS method predicted most of the detections correctly. Finally, the team is supporting development of AI-based machine learning (here termed Interactive Machine Learning Detection Assessment, IMLDA) to help reveal which site and pesticide characteristics can best predict detectable leaching. Development of all three assessment approaches is ongoing, as is the project’s sampling data accumulation.

Footnotes

1. A measure of how strongly a chemical adsorbs (sticks) to soil organic carbon compounds. Higher means more tightly adsorbing. Units are mililiters per gram of organic carbon. Sorption requires something to adsorb to, thus a K_oc value is only important when there is some organic matter present where soil moisture is carrying the pesticide molecules AND when the sorbing molecules can actually touch the organic carbon compound surfaces. K_oc​ is an index of how mobile the chemical is in the environment in the presence of sorbing organic materials.↩︎
2. Tammo S. Steenhuis, Naaran Brindt, Steven Pacenka, Brian K. Richards, J.-Yves Parlange, Bahareh Hassanpour. 2024. A theoretical underpinning of the pesticide Groundwater Ubiquity Score (GUS). Journal of Hydrology and Hydromechanics, 72(3), 349-361. URL: https://doi.org/10.2478/johh-2024-0016 .↩︎
3. Gustafson, D.I. 1989. Groundwater ubiquity score: A simple method for assessing pesticide leachability. Environmental Toxicology and Chemistry, 8(4), 339–357. URL: https://doi.org/10.1002/etc.5620080411.↩︎
4. Half-life: Time it takes for half of the original amount of a chemical to be decomposed chemically or biochemically to other compounds. Units are days. More days means that the chemical persists longer.↩︎