Floods Control How Pesticides Move into Riverbank Soils

Every year, rivers swollen by rain and snowmelt spill over their banks and wash across the vegetated strips of land that border them. Scientists have long understood that these floods can carry sediment, nutrients and even toxic pollutants into floodplain soils, but a new study provides some of the most controlled evidence yet that ordinary, recurring floods—not just catastrophic ones—systematically deliver agricultural pesticides from streams into the root-zone soil of riparian areas. The research, conducted in a state-of-the-art mesocosm facility in southwestern Germany, shows that both the frequency and the duration of flooding events drive the accumulation of pesticide mixtures in these ecologically vital transition zones, a finding with troubling implications as climate change makes floods more frequent and more intense.

The study, led by Franziska Fiolka of the iES Landau Institute for Environmental Sciences at RPTU Kaiserslautern-Landau and published in the Archives of Environmental Contamination and Toxicology, set out to answer a deceptively simple question: how much pesticide travels from a stream into the adjacent land when the stream floods, and does more flooding mean more contamination? While field studies had previously documented elevated contaminant levels in flood-prone soils, disentangling the effects of flood frequency and flood duration from the many confounding variables of natural landscapes—variable pesticide inputs, differing soil types, changing hydrology—has been extremely difficult. The team solved this problem by building the flooding themselves.

The experiments took place at the Riparian Stream Mesocosm facility in Landau, Germany, a setup consisting of sixteen spatially independent aquatic-terrestrial units, each just over fifteen meters long. Every unit contains a narrow artificial stream approximately 0.71 meters wide, flanked by a 3.70-meter-wide riparian area planted with grasses. Crucially, each stream is fed independently with water pumped directly from the nearby River Queich, a fourth-order stream whose catchment includes viticulture, agriculture and urban land uses. This design meant that the pesticide load entering the mesocosms reflected the genuine chemical signature of an agriculturally influenced river rather than artificially spiked water.

Between May and September 2023—covering the main pesticide application season in the surrounding farmland—the researchers simulated four flooding events, spaced thirty-eight days apart. Flooding was triggered by raising the drainpipe at the outlet of each stream, which raised the water level until sixty percent of the downstream riparian area was submerged, covering its full width with water reaching a maximum depth of about nine centimeters, all while maintaining stream flow. Four treatments were tested in a randomized block design: no flooding as a control, and flooding lasting three, seven or fourteen days. Under normal flow, the streams ran at roughly 4.2 liters per second with a flow velocity of 0.2 meters per second; during floods, velocity dropped to about 0.1 meters per second, and the mean arrival time of water roughly doubled, clear evidence that the floodwater was exchanging with water sitting in the inundated riparian zone rather than simply passing through.

Sampling was deliberately tied to plant roots. Twenty-four hours after the end of each flooding event, the researchers uprooted an individual of a common grass, Elymus spp., from the sampling point in each flume and shook the adhering soil—representing the top twenty centimeters of root-zone soil—into collection bags. Weekly grab samples of river water established the baseline pesticide load in the incoming flood water. Back in the laboratory, soil samples were freeze-dried at minus 57 degrees Celsius, sieved to two millimeters, spiked with an isotopically labelled internal standard and extracted using an acetonitrile-based protocol capable of quantifying 96 insecticides, fungicides and herbicides. Water samples were analyzed for 70 pesticides by direct injection. All measurements were performed with high-performance liquid chromatography coupled to a triple quadrupole mass spectrometer via electrospray ionization, with analyte identification requiring both qualifier-quantifier peak area ratios and retention times to match calibration standards within strict tolerances.

The results were striking. After four repeated flooding events, the researchers detected six pesticides in the riparian root-zone soil that were absent from unflooded control soils: four fungicides, one herbicide and one insecticide. Every one of these compounds had been measured in the flood water supplied by the River Queich, confirming the aquatic origin of the contamination. Overall, 31 pesticides were detected in the river water over the season, with the fungicide Fluopyram appearing in 92 percent of water samples, followed by Isoproturon at 69 percent, S-metolachlor at 65 percent, and Metalaxyl, Metrafenone and Acetamiprid each at around 54 percent. Nineteen percent of the pesticides quantified in the river water were also present in the riparian soil by the end of the season.

The statistical analysis, built on a generalized linear mixed model with a Tweedie distribution and log-link function, revealed two clear trends. First, the total pesticide concentration of the six flood-transferred compounds in riparian root-zone soil tended to increase more than three-fold after four flooding events compared with a single event. Second, every additional day of flooding raised the total pesticide concentration—by a factor of 1.29 during the first flooding event and 1.14 by the fourth—indicating that longer inundation continuously loads soil with contaminants carried in the floodwater. The researchers attribute this to the extended contact time between moving flood water and soil, supported by their hydrological measurements showing water exchange within the flooded riparian area. Neither trend reached conventional statistical significance individually, but the consistent direction across treatments, together with agreement from earlier field studies, strengthens the case for a real cumulative effect.

Notably, the six pesticides transferred by flooding spanned an enormous range of chemical properties. Their water solubilities ranged from low values below 10 milligrams per liter for compounds like Azoxystrobin, Metrafenone and Boscalid, to moderate solubility for Spiroxamine and Isoproturon, to high solubility for the insecticide Acetamiprid. Similarly, the compounds varied widely in their organic carbon adsorption coefficients, vapor pressures and environmental half-lives. This means that no single physicochemical parameter reliably predicts whether a pesticide will be moved by floods—an important knowledge gap, because it prevents scientists and regulators from simply screening databases to identify flood-mobile compounds. The authors suggest that flood transfer instead depends on a combination of sorption behavior, solubility, mobility, sedimentation and biotic degradation processes, which may themselves shift under the anaerobic conditions that follow flooding. Acetamiprid, for instance, has a half-life ten times longer in dry soil than in wet conditions, while the herbicide Imazapyr degrades up to three times faster anaerobically.

The concentrations measured in the riparian soils were at least three orders of magnitude lower than those typically found in agricultural fields, and individual compound levels sat more than five orders of magnitude below the chronic no-observed-effect concentration for earthworms. Yet the researchers caution against dismissing these findings as ecologically trivial. The number of pesticide residues detected in the flood-exposed riparian soils was on the same order of magnitude as the medians reported by large-scale European surveys of agricultural topsoil, suggesting that flooding gradually assimilates the pesticide profile of riparian soils toward that of farmland, even if the absolute concentrations remain far lower. Moreover, meta-analyses have shown that pesticide mixtures can exert stronger negative effects on soil fauna communities than single substances, and a recent study in Nature found that pesticide residues alter taxonomic and functional biodiversity in soils. Chronic, low-dose mixture exposure across wide stretches of flood-prone riparian habitat—exposure that recurs season after season—may therefore matter for the invertebrates, microbes and food webs that depend on these soils.

The climate context sharpens the concern. Flood risk in European river basins is projected to increase by as much as 220 percent within this century under high-end climate scenarios, and summer floods—precisely the type that coincides with peak pesticide application—are expected to grow in magnitude and frequency. Because pesticide concentrations in flooded soils may decline within days after a single event but accumulate across repeated floods, the recurrent flooding typical of small and medium agricultural streams could turn riparian zones into long-term chemical sinks. Once contaminated, these soils may themselves become secondary sources, slowly releasing stored pesticides to plants, soil organisms and the wider floodplain. Related work by the same group has already shown that flood-borne pesticides can move from riparian soil through plants into aphids, opening a pathway into terrestrial food webs.

The study’s authors conclude that flooding deserves recognition as a genuinely ecologically relevant exposure pathway for pesticides—one that complements the better-known biotic transfer route via aquatic insects emerging from contaminated streams. For decades, riparian buffer zones have been promoted as filters that protect water bodies from agricultural runoff; this research reveals that the protection is not one-way. In a warming, wetter world with more frequent floods and intensifying agriculture, the same strips of vegetation celebrated as biodiversity hotspots may be quietly absorbing the chemical burden of the waters they border, with consequences for habitat quality and non-target organisms that current environmental risk assessment has barely begun to account for.