Study region: Brittany, France Study focus: After decades of rising agricultural nitrate pollution, several regions of Western Europe are entering a recovery phase driven by stringent nutrient-reduction policies. Brittany offers a natural experiment in large-scale mitigation, particularly in "green-tide" catchments affected by recurrent coastal algal blooms. Using 23-year time series (Oct 2000 - Sept 2023) of nitrate concentrations from 286 agricultural catchments, we assessed: i) the magnitude of nitrate recovery and ii) whether recovery has continued or slowed in recent years. We hypothesised that, due to more extensive measures, green-tide catchments would exhibit faster and more sustained recovery than control catchments. We further hypothesised that accounting for initial conditions and intrinsic catchment properties is essential for evaluating policy effectiveness. New hydrological insights for the region: Nitrate concentrations declined by 34% in control catchments and by 49% in green-tide catchments. Initial nitrate concentration, a proxy for pre-study N management effectiveness, explained 68% of the variance in recovery magnitude, indicating greater improvement potential where initial nitrate losses were highest. Green-tide catchments exhibited significantly stronger and more sustained declines than controls (ANCOVA, p < 0.05), although differences were modest after accounting for initial concentrations and intrinsic catchment properties. The latest programme of measures will be critical in determining whether nitrate in green-tide catchments will continue to decline or begin to plateau, as observed in control catchments.

The pervasive influence of climate change on the hydrological cycle raises concerns about its consequences for water quality, particularly for nutrients and organic matter concentrations in agricultural catchments. Here, we report on the water quality response of 126 agricultural catchments in western France to hydrological droughts that they experienced during the last two decades (2000–2023). We hypothesised that the direction of concentration change for nitrate, soluble reactive phosphorus (SRP) and dissolved organic carbon (DOC) during droughts could be predicted from the slopes of long-term concentration–discharge (c–Q) relationships. Our results showed a majority of positive nitrate-Q slopes, a majority of negative SRP-Q slopes and a majority of positive but weaker DOC-Q slopes. Consistent with our hypothesis, nitrate concentrations generally decreased during droughts (median of −14%), SRP increased (median of +24%), and DOC varied weakly (median of −2%). This was regardless of whether the drought occurred during the low-flow or high-flow seasons. Although the sign of the c–Q slope helps to reliably predict the direction of concentration changes during droughts, the magnitude of these changes depends on a combination of catchment and in-stream processes, which we identified using catchment geographic properties. We also observed a flushing effect during post-drought periods in the high-flow season in 19% of the catchments for nitrate and 80% of the catchments for DOC. Our analysis provides generalizable insights into the changes in water quality under droughts, which are expected to increase in frequency and severity in the future.

Lentic waters are biogeochemical reactors, producing and receiving carbon (C) originally fixed by the terrestrial and aquatic biosphere, which is then buried in sediments or respired back to the atmosphere in the forms of carbon dioxide (CO2) and one of the more potent greenhouse gas (GHG) methane (CH4). Additionally, lakes serve as archives of terrestrial and aquatic carbon processes within their sediments, enabling the reconstruction of historical changes spanning thousands of years. These changes encompass alterations in land cover, indicated by pollen records, soil carbon erosion and shifts in lake productivity resulting from changes in land use and climate. Both the burial of C in lakes and the emissions of GHGs are recognised as important components of Earth's climate system, yet they remain poorly understood and constrained due to inadequate quantities and qualities of observations. In the case of GHG emissions from lakes, observations are often sporadic, failing to capture the significant spatial and temporal variations in emissions across diverse lentic systems. To address this challenge, process-based models that incorporate the interconnected biogeochemical processes occurring within lakes and their watersheds would arguably be the best tool to extrapolate from site-level observations to regional and finally global scales, to quantify the anthropogenic impact on these fluxes and to reconstruct long-term shifts in emissions and burial due to changes in land cover and climate. However, the development and evaluation of such models is hampered by the lack of observations in sufficient quality. In this project, we bring together a unique consortium of specialists in aquatic ecology, biogeochemistry, palynology, sedimentology and modelling of terrestrial and aquatic biogeochemistry. This project will put forth a national programme of systematic, long-term observations of lake GHG and C cycling processes of unmet detail, consistency and quality. First, at 40 pilot sites spanning typological and environmental gradients, there will be a comprehensive data acquisition endeavour to evaluate biological processes and mesological factors influencing the sequestration or recycling of organic carbon. This effort will be complemented with a synthesis of existing data (WP1). Second, based on well-dated sediment records, which include both newly-acquired and synthesised existing data, variability of lake C burial and their climate and land-use controls will be reconstructed over the past 150 years (WP2). For 15 of these pilot sites, reconstruction will go back until the mid-Holocene (5,000 years BP), allowing us to shed light on the anthropogenic perturbation of the C cycle in this earlier part of human history, which is commonly excluded from this type of research due to lack of information. The activities of these first two WPs will result in an open-source national database, guaranteeing valorisation of our research far beyond this project. In WP3, we will use the land surface model (LSM) ORCHIDEE C-lateral to assess C cycling in the terrestrial biosphere and the mobilisation of biospheric C into lakes, which is possible due to an explicit representation of soil C leaching and erosion processes and a downscaling scheme permitting us to assess C exports from watersheds at sub-grid scale. While LSMs are used to assess evolution of biospheric C budgets from the beginning of the Industrial Period, we will use it to hindcast the evolution since the mid-Holocene, using lake sediment records for model validation. Moreover, we will develop a new process-based lake C model supported by the database established in WPs 1 and 2, which we will couple to ORCHIDEE C-lateral to simulate lake C burial and GHG emissions in response to climate and processes in the lake watershed. This model set-up will first be used to better constrain contemporary large-scale lake GHG emissions and to disentangle the anthropogenic perturbation of these fluxes from the natural background flux. These estimates will be revolutionary, as they will allow attributing part of lake GHG emissions to anthropogenic emissions for national GHG budget reporting. Then, these models will be emulated to reconstruct evolution of lake GHG budgets and C budgets of the whole lake watershed since the mid-Holocene. While simulations will first be performed at the scales of France and Europe, the development of international partnerships to implement observations from other biomes (WP4) will finally support simulations at the global scale.

The aim of this study was to assess positive or negative impacts of anaerobic digestion (AD) on water quality using a systemic approach. To this end, we used the agro-hydrological model Topography-based Nitrogen Transfer and Transformation (TNT2), a spatially explicit model that simulates nitrogen and water flows at the watershed scale on a daily time step. Four scenarios were constructed and analyzed: a baseline before the introduction of AD (S0), AD with adjusted fertilization (S1), AD with unadjusted fertilization (S2), and agroecological AD (S3). The results showed that, when spreading practices were similar and an equivalent amount of effective nitrogen was applied, digested pig slurry generally had a predicted amount of nitrate leaching similar to that of undigested pig slurry. In addition, replacing catch crops with energy cover crops had little impact on water quality. Scenario S3 was the most favorable one for water quality and biogas production, but not for soil organic nitrogen storage and food and feed production. This study’s strength is its systemic approach, which considered both environmental and agronomic aspects to assess the scenarios.

To better understand the seasonal variations in environmental conditions regulating dissolved organic matter (DOM) export in agricultural headwater catchments, we combined the monitoring of nitrate, iron, soluble phosphorus, and DOM concentration (as dissolved organic carbon; DOC) and composition (3D fluorescence) in soil and stream waters at regular intervals during 1 hydrological year. We installed 17 zero-tension lysimeters in organic-rich top soil horizons (15 cm below the surface) in the riparian area of a well-monitored agricultural catchment in French Brittany and collected them at a fortnightly frequency from October 2022 to June 2023. We observed a large increase in DOC concentrations in soil waters during the high-flow period linked to the establishment of Fe-reducing conditions and the subsequent release of DOM. We also noted that the timing and the spatial variability in Fe(II) biodissolution in soils was regulated by nitrate from agricultural origin and the heterogeneity of water flow paths at the hillslope scale. Contrary to our current understanding of DOM export in headwater catchments, these results lead us to consider the winter high-flow period as an active phase of both DOM production and export.

Landscape organized (or structured) heterogeneity influences hydrological and biogeochemical patterns across space and time. We developed landscape indices that describe the spatial configuration of nutrient sources and sinks as a function of their hydrological distance to the stream (lateral dimension) or to the outlet (longitudinal dimension) and their intersection with flow-accumulation areas. Using monthly nitrate, total phosphorus (TP), soluble reactive phosphorus (SRP) and daily discharge (Q) data from 221 rural catchments (1-300 km(2)) from 2010-2020, we observed higher variability in flow-weighted mean concentrations in smaller catchments than in larger ones. The variability in landscape configurations also decreased with increasing catchment size. A landscape configuration index, calculated as mean arable land use weighted by spatial data on hydrological distance and flow accumulation, improved prediction of TP and SRP, but not nitrate, compared to the unweighted mean arable land use. We conclude that landscape configuration influences phosphorus transfer more than nitrate transfer, and that flow-accumulation zones and riparian areas are critical source areas for TP and SRP, respectively. By contrast, landscape spatial configuration in the lateral (upslope-downslope) and longitudinal (upstream-downstream) dimensions did not have an identifiable influence on nutrients temporal dynamics. The indices developed in this study can help design landscapes that minimize diffuse phosphorus losses to streams and show that landscape management is not a first order control for nitrate losses.

Macroalgal blooms in coastal areas, which pose unique risks to the environment, citizens, stakeholders, and the economy, mainly are due to nitrate pollution in rivers. The Saint-Brieuc Bay has some of the largest algal blooms in Brittany (western France). It is fed by three catchments-Gouessant (420 km(2)), Gouet (256 km(2)), and Anse d'Yffiniac (130 km(2))-characterized by an oceanic climate and mixed-farming systems with high livestock density. The main objective of this study was to assess the ability of nitrogen (N) mitigation scenarios to reduce N emissions to coastal water using the spatially distributed agrohydrological model TNT2. Once TNT2 was calibrated and validated for the catchments, scenarios were simulated for them from 2008 through 2035: (a) reference (REF) scenario, (b) agricultural management practices (AMP) scenario (i.e., changes to mineral and organic N fertilization, catch crops, and crop rotations), (c) increasing percentages of agricultural land conversion (ALC) into unmanaged grassland from downhill to uphill, and (d) scenarios that combined AMP and ALC. Results showed that the AMP scenario could reduce N load in the bay by 31% vs. the REF scenario. The ALC scenario was much more effective when located downhill, with >50% of maximum effectiveness achieved with 10% of agricultural land converted. These results varied strongly among catchments. Because reducing coastal eutrophication requires a drastic decrease in N loads, we recommend combining AMP and ALC scenarios to achieve good environmental status. The study illustrates the TNT2 model's ability to simulate complex scenarios and guide mitigation policy.

Landscape organized (or structured) heterogeneity is often assumed to influence hydrological and biogeochemical patterns across space and time. In this study, we quantified landscape organized heterogeneity with two indices describing the spatial configuration of nitrogen sources or sinks regarding 1) their hydrological distance to the nearest stream (i.e. upslope/downslope heterogeneity: in the lateral dimension) and 2) their hydrological distance to the outlet in the river network (i.e. upstream/downstream heterogeneity: in the longitudinal dimension). The nitrogen sources considered are agricultural fields, defined from interpretation of satellite images, and the sinks are riparian wetland, defined from a topoclimatic index. Using public nitrate concentration and discharge data from 180 catchments in western France (5-150km²), we tested whether landscape organized heterogeneity influenced riverine nitrate concentration and dynamics. The metrics computed to characterize nitrate concentration and dynamics were the flow-weighted concentration (FWNO3), the slope of the log(C)-log(Q) relationship (slope b) and the ratio of the coefficients of variation of concentration and discharge (CVratio). Results showed a high positive correlation between slope b and the CVratio, but no correlation between the later and FWNO3. 43% of the catchment exhibited a positive b slope, indicating maximum nitrate during the winter high flow period and 17% exhibited a negative b slope, indicating maximum nitrate during the summer/fall low flow period; the remaining 40% exhibited a near-zero slope. Landscape organized heterogeneity was larger in the lateral dimension for both nitrogen source and sinks than in the longitudinal dimension. In the lateral dimension, nitrogen sources were primarily located upslope and nitrate sinks downslope. In the longitudinal dimension, no general trend was observed for nitrogen sources and nitrate sinks were rather located upstream. Heterogeneity in the lateral dimension was highly variable among catchments for the smaller catchments and less variable for the larger ones. Heterogeneity in the longitudinal dimension did not exhibit a visible relationship with catchment size. No relationship was found between indices of landscape heterogeneity and FWNO3, arguably because other primary factors (such as the nitrogen surplus or runoff) control most of the regional variability in FWNO3. We found non-linear relationships between our indices of nitrogen sink organization and the b-slope or the CVratio, both in the lateral and longitudinal dimensions. The catchments with a negative b-slope (maximum nitrate during low-flow season) had their wetlands located more upstream and/or more upslope than the average. The relationship with nitrogen sources were opposite by construction (agricultural fields are often located outside wetland areas) but less clear. Further work is ongoing to explore the influence of landscape spatial organization on phosphorus concentration and dynamics.