You are on page 1of 17

Journal of Hydrology: Regional Studies 45 (2023) 101283

Contents lists available at ScienceDirect

Journal of Hydrology: Regional Studies


journal homepage: www.elsevier.com/locate/ejrh

There is glory in prevention! Regional spatio-temporal


agrochemical runoff into aquatic ecosystems and its potential
mitigation using multifunctional buffers
Leonardo R. Ramírez, Ina Säumel *
The Integrative Research Institute on Transformation of Human-Environment Systems (IRI THESys), Humboldt-Universität zu Berlin, Unter den
Linden 6, Berlin 10099, Germany

A R T I C L E I N F O A B S T R A C T

Keywords: Study region: Uruguay, South America


Campos Study focus: Riparian ecosystems accumulate the impacts of land use change from extensively
Eutrophication used grasslands to industrial cash crops in their watersheds and transport a proportion of
Ecotoxicology
pollution downstream. We modelled and assessed land use changes within different zones along
Fertilizer
Land cover
Uruguayan freshwater ecosystems and estimated the potential risks of phosphorus and glyphosate
Landscape ecology runoff based on different land use scenarios.
Non-point-source pollution New hydrological insights for the region under study: We found that ongoing intensification of
Nutrient delivery ratio grasslands, as well as the conversion to cropland and timber plantations, exerted high pressure on
Retention aquatic ecosystems. Phosphorus runoff is expected to increase by more than 40% and glyphosate
Pesticides runoff by 500%. Excluding intensified land use within the first 100 m of streams can reduce
Temperate grassland pollutant discharge by ≈ 20%. We propose an expansion of pollution monitoring from large to
Uruguay
small streams. Results from modelling can contribute to decision making and to raising awareness
among stakeholders involved in land management and planning.

1. Introduction

Pristine riparian ecosystems along many rivers around the world are being converted to crops, silviculture or rangeland, variation
of watercourses by channelization and urban development with corresponding degradation of the remaining habitat functions, flow
regulation and flood plain clearing (Opperman et al., 2017). Thus, today’s ‘Riparia’ (Naiman et al., 2005) are open, co-constructed
socio-ecological systems at the crossroads of the biosphere, hydrosphere, lithosphere, atmosphere and anthroposphere (Dufour
et al., 2019)— to sum up a highly contested part of our rural landscapes due to its conservation value for biodiversity and because it is
an area that provides freshwater for productive activities. At rural sites, the pressure of land use change on aquatic ecosystems is
related to non-point-source pollution by agrochemicals such as fertilizers and pesticides, and by organic waste from livestock and
overall sediment erosion (Stafford et al., 1996; Ongley et al., 2010; Chalar et al., 2017; Doehring et al., 2020). In addition to the
well-researched and challenging macro-pollutants (nitrogen or phosphorus), there are also thousands of synthetic trace contaminants
in aquatic ecosystems (Schwarzenbach et al., 2010; Mateo-Sagasta et al., 2017). While the multibillion-dollar agrochemical industry
produces more than five million tons of pesticides per year, and continues to develop new commercial products with different active

* Corresponding author.
E-mail address: ina.saeumel@hu-berlin.de (I. Säumel).

https://doi.org/10.1016/j.ejrh.2022.101283
Received 30 June 2022; Received in revised form 27 November 2022; Accepted 29 November 2022
Available online 9 December 2022
2214-5818/© 2022 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

chemical ingredients (Schwarzenbach et al., 2006), a standardized monitoring of the effects on local ecosystems and human health
using in-situ studies is lacking (Schwarzenbach et al., 2006; Schreinemachers and Tipraqsa, 2012; Mateo-Sagasta et al., 2017).
The temperate grasslands of the Southern cone, along with their rivers and streams, are experiencing dramatic land use change
from traditionally extensive uses to intensive livestock, cash crops, timber or pulp production, predominately for the global market
(Alvarez et al., 2015). The productive sector of Uruguay often discusses the low productivity of Uruguayan grasslands as a driver of
land conversion (Modernel et al., 2016) and proposes new strategies to intensify production to conserve grasslands (Jaurena et al.,
2021), whereas others call for the recognition of the values of ‘old growth grasslands’ (Veldman et al., 2015).
However, the resulting impacts of this land use change, and further intensification of local ecosystems, remain largely unknown.
While some studies have explored the effects on terrestrial ecosystems such as riparian forests (Ramírez and Säumel, 2022) or
grasslands (e.g. Jaurena et al., 2016; Pañella et al., 2020), only a few have been published in Uruguay on water balance (Reichert et al.,
2017), water pollution and eutrophication (Chalar et al., 2017; Eguren et al., 2018; Gonçalves et al., 2020; Gorgoglione et al., 2020;
Beretta-Blanco and Carrasco-Letelier, 2021). These topics, hotly debated in the Uruguayan press and across different sectors (e.g.
Alonso et al., 2019), are attracting increasing awareness among local stakeholders, as are the constitutional right to water (Moshman,
2005; Taks, 2008) and the antagonistically functioning new irrigation law (Santos, 2020). Here, for the first time, we examine the
dimensions of land use change in the surroundings of aquatic ecosystems at a distance of up to 1 km from Uruguay’s fresh waters.
As Uruguay is among the upper-middle income countries that has experienced double-digit growth in the intensity of pesticide use
during the last decade (Mateo-Sagasta et al., 2017), and because it has limited infrastructure to track the effects of pesticides, a
precautionary approach is necessary (Schwarzenbach et al., 2006). In addition, the state of the art of monitoring and regulations has
led to a ‘compartmentalization’ of risks, which ultimately underestimates the transversal actual hazards that chemicals catalyze in
waterways for humans and other organisms (see discussion in Hendlin et al., 2020).
To narrow this gap, we focus on glyphosate [N-(phosphonomethyl) glycine] and its derivates, which comprise the most widely used
herbicide group in the world (Duke and Powles, 2008; Benbrook, 2016; Sabzevari and Hofman, 2022) and has multiple consequences
for soils and aquatic ecosystems (Gomes de Carvalho Marques et al., 2021), and on phosphorus, a fertilizer that is an important
contributor to the eutrophication of aquatic ecosystems (Elser et al., 2007).
In addition to a wide variety of measures to reduce erosion and the use of potential pollutants in agriculture and forestry (McDowell
et al., 2018), the implementation and management of buffer zones is a common low-cost approach to trap, eliminate or transform
pollutants flowing from terrestrial into aquatic ecosystems (Reddy et al., 1999; Viaud et al., 2004, Collins et al., 2009; Zhang et al.,
2010; Stutter et al., 2012). Beyond this, buffer zones can provide multiple benefits for riparian restoration, such as enhancing
hydro-morphological and thermal conditions and overall functioning and are no-regret management options (Mander et al., 2017; Feld
et al., 2018). However, the provision of habitat services is powerfully shaped by buffer management (Hille et al., 2018; Zak et al., 2019)
and human activities in the adjacent zones (Luke et al., 2019). Furthermore, cost calculations in rural Europe have demonstrated that
the implementation of buffer zones on a larger scale is mainly a question of setting policy priorities rather than a financial impossibility
(Jabłońska et al., 2020). However, for Uruguay, the legislation (Law No 14.859 of the Water Code) does not explicitly indicate buffer
zones for biodiversity conservation and to amortize the effects of economic activities near watercourses.
We aim to answer the following research questions: i) to what extent are different surrounding zones along aquatic ecosystems
affected by land use changes?; ii) what are the potential phosphorus and glyphosate run-offs assuming different land use scenarios?;
and iii) what trade-offs are typically possible between economic exploitation and the environmental impacts when establishing ri­
parian buffer zones? To our knowledge, this is the first study to estimate glyphosate and phosphorus runoffs related to land use change
along Uruguayan fresh waters. We assess different land use scenarios, i.e. ‘status quo’ or ‘business as usual’, assuming continuation of
the current management strategies, management of grassland intensification based on application of agrochemicals (Jaurena et al.,
2021) and the grassland conservation strategy with extensive use (Veldman et al., 2015).

2. Methods

2.1. Study area delimitation

We focus on areas around Uruguayan watercourses up to a distance of 1 km from their aquatic ecosystems. To identify the areas of
aquatic ecosystems, we used geospatial information from the Uruguayan government (MVOTMA, 2015) to delimit the study area
(referred to henceforth as ‘surrounding areas’). We combined geospatial information on the hydrological network and surface water.
We created two types of surrounding area: (i) from the border of the aquatic ecosystems to the surroundings using four different
distances (0–0.1, 0–0.25, 0–0.5 and 0–1 km); and (ii) by contour or section (0–0.1, 0–0.25, 0.25–0.5 and 0.5–1 km). The different
distances of the sections were selected because we are interested in exploring whether the change processes are unique to any one
section or whether there is a general pattern of land use change. Thus, in total we covered ≈ 120,800 km2, which represents 68% of the
terrestrial area of Uruguay and ≈ 12% of the temperate grasslands of South America.

2.2. Study area context

Uruguay is located in South America between latitudes 30◦ and 35◦ South and longitudes 53◦ and 58◦ West. It has a temperate
climate without dry seasons or hot summers (Peel et al., 2007). The average annual temperature is 17.6 ◦ C, and annual total pre­
cipitation is 1221 mm (Fick and Hijmans, 2017).
Uruguayan aquatic ecosystems have been largely modified by anthropogenic activities, such as regulation by the building of

2
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

numerous large dams, and have an eutrophic to hypereutrophic water status (see status report by Alonso et al., 2019; Fig. 1A–D for
land use changes).
Currently, 791 plant protection products have been approved, and over 1500 have been registered for use in Uruguay, 264 of which
are classified as prioritized pesticide products (MA 2020). Between 2015 and 2020, more than 85 kilotons of pesticides were imported
into Uruguay (OAN, 2020). The top three were: glyphosate (54 kilotons), 2,4-D (8 kilotons) and metalochlor (2 kilotons). Nearly 5
megatons of fertilizer were imported, 1.3 megatons of which were phosphorus-based (i.e. mono- or diammonium phosphate or triple
superphosphate; OAN, 2020). In 2019, phosphorus-containing fertilizer valued at U$S81 million was imported into Uruguay. As
systematic monitoring of use is lacking, no insights into the fate of these agrochemicals are available beyond these general data.

Fig. 1. Study area, schematic representation of buffer building, land use/cover maps used to model land use change and transport of phosphorus
and glyphosate (A–D) and Quickflow raster used in the NDR model (E–F).

3
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

2.3. Reclassification of land use/cover

As the original classification was based on Landsat images with a resolution of 30 × 30 m, the land use maps in vector format for
the years 2000 (Alvarez et al., 2015) and 2015 (MVOTMA, 2015) were converted to raster with a resolution of 30 × 30 m pixels. We
assessed land use change for both types of surrounding area for the years 2000 and 2015.
We aggregated and reclassified original land use/cover maps from 17 to 10 land use/cover types: (i) native forests; (ii) shrublands;
(iii) semi-natural grasslands, including those with dispersed palm stands; (iv) highly modified grasslands, including artificial and
improved grasslands; (v) bare soils, including quarries, stand pits and open pit mines; (vi) urban areas; (vii) timber plantations; (viii)
orchards; (ix) crops; and (x) aquatic ecosystems, including artificial waters, natural waters and flooded natural areas (Appendix 1:
Table S1). Modified grassland cover was extracted from the work of Petraglia et al. (2019) and incorporated into the land use/cover
maps of 2000 and 2015. This incorporation consisted of the reclassification of pixels belonging to original land use/cover maps to the
pixels from the work of Petraglia et al. (2019), classified as: (i) highly modified natural pastures, introduced pastures and old stubbles;
and (ii) areas including forage crops, sown grasslands, field and stubble with a high temporal variation, and generally associated with
raising dairy cattle. We only reclassified original pixels as modified grassland when the intersection occurred with cropland and timber
plantation land uses (Appendix 1: Fig. S1). As the raster from the work of Petraglia et al. (2019) has a resolution of 10 m, the
reclassification allowed identification of so-called ‘improved pastures’ (i.e. fertilized and/or with a high proportion of legumes
introduced to increase farm productivity; Berretta et al., 2000) originally classified as crops in the 2000 and 2015 land use/cover maps.
We were also able to identify grassland patches inside cropland and timber plantations which have been highly modified due to the
pressure of anthropic activities in their surroundings (Appendix 1: Fig. S1). Additionally, because there is no reported determination of
the accuracy of the classification for land use/cover maps (2000 and 2015) from Uruguayan institutions, we created a confusion matrix
from 569 points over the land use map for the year 2015. Thus, we estimated an overall accuracy of 66% (Appendix 1: Table S2). The
processing of geospatial information was carried out in QGIS v3.10.11-A (QGIS Development Team, 2020).

2.4. Transport of phosphorus and glyphosate

We used the nutrient delivery ratio (NDR) model (Sharp et al., 2020) to calculate the surface transport of phosphorus and
glyphosate from different terrestrial land use types to the watercourses using Integrated Valuation of Ecosystem Services and Tradeoffs
(InVest) software (Sharp et al., 2020). We modelled phosphorus and glyphosate discharge using the land use maps of the surrounding
areas for the years 2000 and 2015, with a width of 1 km from the aquatic ecosystem (Fig. 1A–D and Fig. 2).

Fig. 2. Flow chart of geospatial information, processing, analysis and NDR model simulation.

4
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

The NDR model requires data for agrochemical load per hectare and the efficiency of retention expected from a given land use type
(Sharp et al., 2020). These calibration parameters were based on a literature review, which is listed in the Appendices (see Table 1 and
Appendices S1–S3). The NDR model estimates the movement of chemical compounds across land based on mass balance (Sharp et al.,
2020). The model considers that the movement and retention efficiency of chemicals across the landscape depend on land use,
topography and runoff (Sharp et al., 2020; Appendix 2: Table S1). The topographic data were based on the Digital Elevation Model
(DEM) of Uruguay, and the runoff data were based on Quickflow calculated from the Seasonal Water Yield (SWY) model of the InVest
model (Sharp et al., 2020; see Appendix 2). Quickflow represents the potential runoff, considering soil types and land cover properties
that influence the balance between infiltration of rain and the runoff across the surface landscape (Sharp et al., 2020). We used
Quickflow from SWY because it considers the hydrological characteristics of the soils, land use cover, precipitation, rain events,
evapotranspiration, topography and the Crop/Vegetation coefficient for the calculation (Sharp et al., 2020). All values, geospatial
information and specific references are listed in Appendices S1, S2 and S3.
In this study, we evaluated the transport of chemicals across land surfaces and omitted subsurface transport. The surface nutrient
delivery ratio (NDRsurf) per pixel is given by:
( ( ) )− 1
IC0 − ICi
NDRsurf = NDR0,i 1 + exp
k

Where IC0 and k are calibration parameters. ICi is a topographic index calculated from the DEM. NDR0,i is the proportion of the
agrochemical that is not retained by downslope pixels. Thus, the total export for all study areas is given by:
Xexpi = loadsurf,i • NDRsurf,i

Xexp = Xexpi
i

where Xexpi is the nutrient export from each pixel to the aquatic ecosystem and Xexp_tot is the total nutrient exported for all study areas.
The NDR makes the following four assumptions: (i) application of an agrochemical is uniform depending on land use (only the
amount applied varies); (ii) the method of application is directly proportional to groundcover; (iii) the model does not consider the
effects of spray drift or subsurface movement of agrochemicals; and (iv) the point sources are not considered.

2.5. Model validation

The absence of independent in-situ data that would allow us to link the load of agrochemicals at a certain land use with the
respective surface runoff makes the validation of a classical operational model, i.e. measured versus modelled data, impossible.
However, our results are of heuristic value (Oreskes et al., 1994; Eker et al., 2018). We know comparatively little about the quantities
and frequencies of agrochemical application on the different land uses and their fate in the landscape. Thus, we explored and tested the
model using visualization techniques, evaluating the extreme-condition test (Rykiel, 1996) and sensitivity of the model in five
randomly selected sub-watersheds (Appendix 3: Fig. S1) that represent different gradients of Quickflow, topographic features (Ap­
pendix 3: Fig. S2) and land use dominances that have been identified for Uruguay (Ramírez and Säumel, 2022; Appendix 3: Table S1), i.
e. dominance of grassland, dominance of timber plantations with grassland, dominance of crops with modified and natural grasslands,
a mixture of crops and modified and natural grasslands or a mixture of grassland and timber plantations. We ran the models for each
sub-watershed with different loads of phosphorus and glyphosate ranging from no agrochemical load (0%) to doubling the input
(200%) compared with current use. Thus, we evaluated the performance of the model based on our knowledge (Oreskes et al., 1994;
Rykiel, 1996; Eker et al., 2018). The results and visualization are given in Appendix 3 (Fig. S3–S13).
We observed that the NDR model was sensitive to loads of agrochemicals and geophysical features of each sub-watershed

Table 1
Load and efficiency of retention of each land use with and without application of phosphorus and glyphosate used as biophysical table to parametrize
the nutrient delivery ratio (NDR) model in InVest software. Details of literature used in Table S2.
Phosphorus Glyphosate

With application Without application With application Without application

Load (kg/ha) Efficiency (%) Load (kg/ha) Efficiency (%) Load (kg/ha) Efficiency (%) Load (kg/ha) Efficiency (%)

Native forests 0.14 0.80 0.14 0.80 0 0.44 0 0.44


Shrublands 0.10 0.50 0.10 0.50 0 0.59 0 0.59
Grasslands 0.80 0.87 0 0.87 1.44 0.74 0 0.74
Modified grasslands 0.80 0.87 0 0.87 1.44 0.74 0 0.74
Bare soils 3 0.15 3 0.15 0 0.87 0 0.87
Urban 0.26 0.05 0.26 0.05 0 0.10 0 0.10
Timber plantations 0.01 0.75 0.01 0.75 2 0.44 2 0.44
Orchards 1.22 0.53 1.22 0.53 1.80 0.85 1.80 0.85
Crops 2.76 0.25 2.76 0.25 2.41 0.99 2.41 0.99
Aquatic ecosystems 0.19 0.05 0.19 0.05 0 0.90 0 0.90

5
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

(Appendix 3: Fig. S3–S13). In general, the model was sensitive to modification of parameters, geophysical context (climate and
topography) and variations of land use. The visualization of the output maps revealed that the model follows the generalizations of the
transport of pollutants and fits the process we want to evaluate, i.e. the modification of agrochemical loads and export due to land use
change based on different scenarios.
Additionally, to evaluate the model, we only found four water monitoring stations that meet the conditions to compare with our
model (Appendix 3: Table S2 and S3). The observed data from the water monitoring stations report values in u/L. The data simulated
by the NDR model report values in kg and also report the areas of the sub-watershed then we could calculate values in kg/ha.
Therefore, we transformed the observed data using the average annual flow of the reported gauged catchments in L/s km2 (DINAGUA,
2012). Thus, we calculated Kling-Gupta efficiency (KGE, Gupta et al., 2009), Nash-Sutcliffe efficiency (NSE, Nash and Sutcliffe, 1970)
and Root Mean Square Error (RMSE). We evaluated the above by standardizing (between 0 and 1) and unstandardized the data.
We found that for glyphosate and phosphorous, the unstandardized KGE and NSE data were negative (Appendix 3: Table S3). In
contrast, for the standardized data, KGE and NSE were positive for glyphosate (KGE = 0.47 and NSE = 0.03) and negative for
phosphorus (KGE = − 0.86 and NSE = − 2.96; Appendix 3: Table S3). For standardized data of RMSE, we found that glyphosate was
close to zero (RMSE = 0.01), and for phosphorus, it was greater than 1 (RMSE = 1.22). So for the standardized data, the RMSE for
glyphosate (RMSE = 0.38) and phosphorus (RMSE = 0.72) were between 0 and 1 (Appendix 3: Table S3).

2.6. Different management and land use scenarios

We analysed six different scenarios of phosphorus and glyphosate run-off from terrestrial surface to aquatic ecosystems depending
on the land use allocation and management of application of agrochemicals in the surrounding areas along the Uruguayan freshwater
ecosystems (Table 2). The scenarios comprise: (i) the status quo (SQ) scenario, where agrochemicals are applied to modified grass­
lands, timber plantations and crops as currently established; (ii) the status quo with buffer (SQB) scenario, where agrochemicals are
applied to modified grasslands, timber plantations and crops as currently established but, within a 100-m buffer around fresh waters,
productive land uses were transformed to natural grassland without application of agrochemicals; (iii) the grassland intensification
(GI) scenario, in which agrochemicals are applied on all (natural and modified) grasslands (Jaurena et al., 2021), timber plantations
and crops; (iv) the grassland intensification with buffer (GIB) scenario, in which agrochemicals are applied on all grasslands, timber
plantations and crops but, within a 100-m buffer around aquatic ecosystems, productive land uses were transformed to natural
grassland without application of agrochemicals; (v) the grasslands conservation (GC) scenario, in which all grasslands are managed
without application of agrochemicals (Veldman et al., 2015), and agrochemicals are only applied on timber plantations and crops as
currently established; and (vi) the grasslands conservation with buffer (GCB) scenario, in which all grasslands are managed without
application of agrochemicals, which are used only on timber plantations and crops as currently established but, within a 100-m buffer
around aquatic ecosystems, productive land uses were transformed to natural grassland without application of agrochemicals (e.g. in
Appendix S4: Fig. S1).

2.7. Data analysis

Considering the totality of the patches, we characterized land use change pattern by calculating area, mean area, standard devi­
ation of the mean and number of patches per land use/cover for the different buffer zones using Fragstat v.4.2 (McGarigal et al., 2012).
We calculated the net change of land use for both buffer types between 2000 and 2015 in square kilometres, and percentages for area,
mean area and number of patches. For the percentage of net change, we considered land use in the year 2000 as being equal to 100%.

Table 2
Description of scenarios used in the study (an example of a scenario with buffer is shown in Appendix S4: Fig. S1).
Scenario Code Description

Status quo SQ Fertilizer and/or pesticide are used on modified grasslands, timber plantations and crops as currently established.
Natural grassland is without application of agrochemicals.
Status quo with buffer SQB Fertilizer and/or pesticide are used on modified grasslands, timber plantations and crops as currently established.
Natural grassland is without application of agrochemicals. All productive land uses (timber, crops, orchards, modified
grassland) are transformed to natural grassland without application of agrochemicals in a buffer of 100 m from the
aquatic ecosystems.
Grassland intensification GI Fertilizer and/or pesticide are used on modified grasslands, timber plantations and crops as currently established.
Additionally, natural grassland is modified by application of chemicals in order to increase productivity (Jaurena et al.,
2021).
Grassland intensification with GIB Fertilizer and/or pesticide are used on modified grasslands, timber plantations and crops as currently established.
buffer Natural grassland is modified by application of chemicals in order to increase productivity (Jaurena et al., 2021). All
productive land uses (timber, crops, orchards, modified grasslands) are transformed to natural grassland without
application of agrochemicals in a buffer of 100 m from the aquatic ecosystems.
Grassland conservation GC Fertilizer and/or pesticide are used on timber plantations and crops as currently established. Natural grassland and
modified grasslands without application of agrochemicals.
Grassland conservation with GCB Fertilizer and/or pesticide are used on timber plantations and crops as currently established. Natural grassland and
buffer modified grasslands without application of agrochemicals. All productive land uses (timber, crops, orchards, modified
grasslands) are transformed to natural grassland without application of agrochemicals in a buffer of 100 m from the
aquatic ecosystems.

6
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

We also ran an intensity analysis using the OpenLand package (Exavier and Zeilhofer, 2021) in R (R Core Team, 2016) to evaluate the
change from a pixel belonging to specific land use/cover to other land use/cover, as well as to detect situations of no change between
2000 and 2015.
In order to explore the net change of the discharge of phosphorus and glyphosate from terrestrial to aquatic ecosystems, we focused
on relative rather than absolute values. We calculated the temporal net change between 2000 and 2015, exploring what the trajectory
of change would have been if the different scenarios had begun in the year 2000. As the SQ scenario most closely reflects the current
land use mode, we used it as a baseline to calculate the net change between scenarios. We therefore averaged the export of each
agrochemical between 2000 and 2015 according to each scenario. Finally, to identify geographic risk areas, we visually explored the
raster maps generated in the analysis of land use change and the outputs of phosphorus and glyphosate. The visualization was
generated by calculating the export of agrochemical per hectare and by sub-watershed. To compare between scenarios, we created a
legend with five ranges of agrochemical export using the method of natural break-Jenks in Quantum Geographic Information System
v.3.16.3-Hannover (QGIS Development Team, 2020). The SQ map legend was taken as a reference and extrapolated to the remaining
scenarios.
We also estimated costs of indemnification for the land owner if buffer areas were taken out of use by using costs per hectare based
on the CONEAT Index (MGAP, 2022; Cutiño, 2015; Lanzilotta and Rosas, 2019), which remains an important source for today’s land
taxation and management plans under the legal conservation regulations (Lanfranco and Sapriza, 2011), based on a detailed classi­
fication which considers soil type, texture, natural vegetation, altitude and geology.

3. Results

3.1. Land use change adjacent to fresh waters

Across all land uses, the mean patch size decreased by a third from 2000 to 2015, while the patch number increased by 40%
(Table 3). In 2015, natural grasslands were the dominant land use (62,314 km2; 52%), followed by crops (21,017 km2; 17%), modified
pastures (11,542 km2; 10%), timber plantations (7993 km2; 7%), native forests (7322 km2; 6%) and aquatic ecosystems (7169 km2;
6%; Appendix S5: Table S1).
Between 2000 and 2015, 17% of natural grasslands within 1 km of the freshwater ecosystems were transformed to crops, timber
plantations or highly modified grasslands (Fig. 3A–D, Appendix S5: Table S1). The net change of land use by different contour areas
showed similar patterns (Fig. 3A, E–G, Appendix S5: Table S2). In 2015, the first 100 m of the surrounding areas were mostly covered
by natural grasslands (43%), followed by aquatic ecosystems (19%), riparian forests (18%), crops (9%), modified pastures (5%) and
timber plantations (4%). The area covered by riparian forests remained constant, while shrubland decreased by 7% (Fig. 3A,
Appendix S5: Table S2).
Grassland also dominated in all contours of the study in 2000 and 2015, and was the land use with the highest total area loss
(14–19%) and the highest mean area loss for most contours (Fig. 3A–G, Appendix S5: Tables S1–S2). Crops were the land use with the
second largest area in the contour over 100 m, increasing by ≈ 40% compared with 2000 (Fig. 3E–G, Appendix S5: Table S2).
However, the net change in mean area decreased in all surrounding areas (Table 3, Appendix S5: Tables S1–S2). Timber plantations
increased by over 100% of total area for all surrounding areas compared with 2000, and there was also an increase in net change of the
mean area (Table 3, Fig. 3A–G, Appendix S5: Tables S1–S2). Modified grassland increased by ≈ 37% of total area for all surrounding
areas, and showed a slight decrease in mean area (Table 3, Appendix S5: Tables S1–S2). The area of shrublands decreased by 6%, as did
the mean area (Table 3, Fig. 3, Appendix S5: Tables S1–S2). Although the areas of orchards, native forest and bare soils remained
constant, the mean areas of these three land use types also decreased (Table 3, Fig. 3A–G, Appendix S5: Tables S1–S2). There was also
an increase in urban area, with a decrease in mean area within most of the surrounding areas evaluated (Table 3, Fig. 3A–G,
Appendix S5: Tables S1–S2).
The number of patches of crops and grassland increased by over 50% for the total study area compared with 2000, followed by

Table 3
Mean (X‾) and standard deviation (SD) of patch area (ha) and net change in patch area and percentage from 2000 to 2015 in the total study area
(buffer of 1 km). NP = number of patches.
2000 2015 Net change of area (mean) 2000 2015 Net change of number of patches

Land use X‾ ± SD (ha) X‾ ± SD (ha) X‾ ha (%) NP NP NP (%)

Native forests 65 ± 324 61 ± 313 –4 (–6) 11,295 12,032 737 (7)


Shrublands 38 ± 78 35 ± 67 –3 (–9) 4988 5086 98 (2)
Grasslands 345 ± 17,559 186 ± 8398 –159 (–46) 21,840 33,488 11,648 (53)
Modified grassland 3 ± 62 3 ± 60 0 (–1) 266,741 372,136 105,395 (40)
Bare soil 15 ± 70 15 ± 68 –1 (–5) 2592 2719 127 (5)
Urban 67 ± 454 60 ± 431 –6 (–10) 1432 1652 220 (15)
Timber plantations 11 ± 82 18 ± 166 7 (62) 34,900 44,479 9579 (27)
Orchards 23 ± 118 20 ± 104 –3 (–13) 1511 1766 255 (17)
Crops 18 ± 143 16 ± 148 –2 (–10) 85,465 132,568 47,103 (55)
Aquatic ecosystems 77 ± 1498 68 ± 1369 –9 (–12) 9311 10,612 1301 (14)
Total 27 ± 3920 20 ± 1969 –7 (–29) 440,075 616,538 176,463 (40)

7
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

Fig. 3. Sankey plot of intensity analysis and percentage of net change in different buffer zones along watercourses in Uruguay from 2000 (left side of
the Sankey plot) to 2015 (right side of the Sankey plot) in the total study area. Main land use types are riparian forests (F), native grasslands (G),
modified grasslands (GM), timber plantations (T), crops (C) and aquatic ecosystems (A). Other land use types are shrublands (S), bare soils (B),
urban areas (U) and orchards (O). The buffer zones 0–0.1, 0–0.25, 0–0.5 and 0–1 km correspond to 10%, 22%, 40% and 69%, respectively of the
total area of Uruguay. The buffer zones by contour 0–0.1, 0–0.25, 0.25–0.5 and 0.5–1 km correspond to 10%, 12%, 18% and 28%, respectively of
the total area of Uruguay.

modified grassland (40%), timber plantations (27%), orchards (17%), urban (15%), aquatic ecosystems (14%), native forests (7%),
bare soils (5%) and shrublands (2%) (Table 3). Grassland decreased slightly within the first 100 m of the surrounding areas, but
increased in the other areas, ranging between 11% and 53% (Table 3, Appendix S5: Tables S1–S2). Crops increased by over 49% in all
surrounding areas, and modified grassland increased by ≈ 40% (Appendix S5: Tables S1–S2). The number of patches covered by
timber plantations increased by over 80% in the first 100 m of the surrounding areas and, for all other areas, the increase ranged
between 36% and 61% compared with 2000 (Appendix S5: Tables S1–S2). The number of patches of orchard, aquatic ecosystems,
native forests and bare soils increased slightly (Appendix S5: Tables S1–S2). The number of patches covered by shrublands decreased
slightly (≈2%) in most of the surroundings of freshwater ecosystems (Appendix S5: Tables S1–S2).

3.2. Phosphorus and glyphosate runoff to watercourses

The calculated runoff of phosphorus to the aquatic ecosystems in the currently established land use and management mode
increased between 2000 and 2015 by 30%, and that for glyphosate by 80% (SQ scenario in Table 4). Runoff of phosphorus and

Table 4
Total export of phosphorus and glyphosate by year and net change (t and %) of export of phosphorus and glyphosate from 2000 to 2015. Temporal net
change is given by: net change (t)=total 2015–total 2000; net change (%)= (net change (t)/total 2000)× 100. Scenarios: status quo (SQ); status quo
with buffer (SQB); grassland intensification (GI); grassland intensification with buffer (GIB); grassland conservation (GC); grassland conservation
with buffer (GCB).
Phosphorus Glyphosate

2000 2015 Net change 2000 2015 Net change

Scenario (t) (t) (t) (%) (t) (t) (t) (%)

SQ 493 642 149 30 102 184 82 80


SQB 377 479 102 27 86 145 59 69
GI 747 862 115 15 946 898 –48 –5
GIB 637 707 70 11 955 891 –64 –7
GC 472 614 142 30 64 139 75 117
GCB 359 453 94 26 50 105 55 110

8
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

glyphosate changed according to different land use scenarios (Tables 4–5, Figs. 4 and 5), ranging between 406 tons of phosphorus and
78 tons of glyphosate per year (GCB scenario) to 805 tons of phosphorus and 922 tons of glyphosate per year (GI scenario; Table 5).
Setting the SQ scenario as baseline, our scenarios showed that the exclusion of highly productive land uses from a buffer zone of
100 m around rivers and streams can reduce the runoff of phosphorus and glyphosate into fresh waters by 25% and 19%, respectively
(Table 5; SQB versus SQ scenario). Considering only crops and timber plantations, but not modified grasslands, as fertilized and treated
with glyphosate (GC scenario), export of phosphorus to aquatic ecosystems is reduced by 4% and glyphosate runoff to aquatic eco­
systems is reduced to a third of that with the SQ scenario (Table 5). Finally, excluding any application of fertilizer or glyphosate within
100 m of rivers and streams can reduce the export of phosphorus into aquatic ecosystems by 29% and export of glyphosate by 46%
(Table 5; GCB versus SQ scenario). Following the GI scenario would increase the input of phosphorus by over 40%, and the runoff of
glyphosate into the aquatic ecosystem by more than 500% (Table 5; GI versus SQ scenario). The establishment of a strict buffer zone of
100 m along aquatic ecosystems reduces the export of phosphorus by a third compared with the GI scenario, and would result in an
increase of 18% compared with the current land use mode (GIB versus SQ scenario in Table 5). The runoff of glyphosate would not
change if the buffer zone were implemented (GIB versus GI scenario in Table 5).

3.3. Costs of exclusion of high load uses from buffer zones

The 100-m buffer zone from the aquatic ecosystems covers an area of ≈ 2167 km2, which is currently covered by 1510 km2 of
modified grasslands, 26 km2 of crops and 632 km2 of timber plantations (Appendix S5: Table S1). We calculated the mean and median
land prices per department (Table 6) according to figures from the Uruguayan Planning and Tax Office (Lanzilotta and Rosas, 2019)
and the maximal prices assuming best land quality depending on the main productive categories (i.e. agricultural–livestock field with
CONEAT Index 100: U$S2700/ha; good agricultural field with CONEAT Index 150: U$S4050/ha; and a forestry area with CONEAT
Index 50: U$S1350 /ha; see e.g. Cutiño, 2015). Thus, the nationalization of these areas would cost U$S775 million, equivalent to less
than 1.4% of the Uruguayan BIP for the year 2020. Renting the buffer zones for 10 years would cost US$34 million, less than 0.06% of
the Uruguayan BIP for the year 2020. The establishment of buffer zones could also be a stepwise process, starting with highly polluted
parts of the Uruguayan watercourses such as the Santa Lucía catchment, which provides drinking water for more than 60% of the
Uruguayan population (Alonso et al., 2021). The nationalization of these areas of the Canelones, San José and Florida departments
would cost U$S227 million (or U$S9 million for 10 years’ rent), equivalent to less than 0.4% (or 0.02%) of the Uruguayan BIP for the
year 2020.

4. Discussion

The use of modelling results as a basis for political decisions must be weighed carefully as they can suggest a certain truth that is not
implicit (Oreskes et al., 1994; Rykiel, 1996; Eker et al., 2018). Nevertheless, we provide heuristic evidence and shed light on a complex
system of agrochemical loads, retention and impacts in an intensively used agricultural landscape. Our models estimated a runoff of
over 560 tons of phosphorus, and of over 140 tons of glyphosate per year into the fresh waters of Uruguay (SQ scenario in Table 5). The
grassland intensification strategy (GI scenario; Jaurena et al., 2021) is expected to increase the discharge of phosphorus into the fresh
waters of Uruguay by 40% and of glyphosate by 500% (Table 5).

4.1. Major land use changes along freshwater ecosystems

The freshwater bodies of the west and southwest of Uruguay are strongly influenced by intensified crops, which also increased in
the central and eastern parts of the country between 2000 and 2015. The timber plantations in the northwest and west–central parts
are expanding in area, and new plantation hotspots increasingly influence watercourses in the south-eastern parts of Uruguay (Fig. 1,
Maps). In 2015, while natural grasslands still dominated the upper catchments of the rivers Cuarem, Arapey Grande and Queguay
Grande, the natural grasslands along the watercourses of the catchments of the rivers Yi, Cebollati and Olimar were replaced by
intensive land uses (Fig. 1A–B; Appendix S6: Fig. S1). For decision making, the legacies and dynamics of past land use change are
crucial to an understanding of long-term ecosystem behaviour (Davis et al., 2015). We demonstrate that land use intensifications,
which represent a common threat to almost all freshwater ecosystems worldwide (Brondízio et al., 2019; Davis et al., 2015), occurred

Table 5
Net change with respect to average for status quo (SQ) scenario. Net change with respect to SQ scenario is given by ((average between 2000 and
2015–average between 2000 and 2015 of SQ)/average between 2000 and 2015 of SQ)× 100). Scenarios: status quo (SQ); status quo with buffer
(SQB); grassland intensification (GI); grassland intensification with buffer (GIB); grassland conservation (GC); grassland conservation with buffer
(GCB).
Scenario of land use

SQ SQB GI GIB GC GCB

Phosphorus Average between 2000 and 2015 (t) 568 428 805 672 543 406
Net change (%) with respect to SQ scenario 0 –25 42 18 –4 –29
Glyphosate Average between 2000 and 2015 143 116 922 923 102 78
Net change (%) with respect to SQ scenario 0 –19 545 545 –29 –46

9
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

Fig. 4. Scenarios of total export of phosphorus (kg⋅ha-1⋅year-1) calculated by sub-watershed.

directly beside the aquatic ecosystems of rural Uruguay (Table 3, Fig. 3). A third of the land within a distance of 1 km of the aquatic
ecosystems experienced changes of use (Table 3), with natural grasslands being particularly increasingly fragmented and disappearing
due to their replacement by timber plantations, crops and modified pastures (Fig. 3). New discussions about intensification of the
grasslands of the Campos region have recently begun. These state that “native grassland is at the core” (Jaurena et al., 2021) and
involve related socio-ecological trade-offs between productivity, diversity and ecosystem integrity (Tittonell, 2021). The intensifi­
cation of grassland embraces a diverse portfolio, ranging from process- to input-based measures and system technologies (Jaurena
et al., 2021). In addition, the increasing number of patches of highly intensified land use (i.e. livestock on modified grasslands, crops,
timber; Fig. 3) along rivers and streams also affects the more natural riparian forests (Ramírez and Säumel, 2022).

10
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

Fig. 5. Scenarios of total export of glyphosate (kg⋅ha-1⋅year-1) calculated by sub-watershed.

4.2. Land use, topography and rainfall pattern determine input of pollutants

Although the intensified grassland modules may appear to more closely resemble native grasslands than crop or timber plantations,
trade-offs with regard to biodiversity have been identified (Jaurena et al., 2021). Widespread inappropriate management practices
concerning stocking rates, dairy effluents, timing and amount of fertilizer and application of pesticides have led to water pollution
originating from the runoff of modified grasslands and crops (Chalar et al., 2017) and of timber plantations (Little et al., 2015).
Glyphosate and/or phosphorus concentrations in streams correlate with intensified land uses (Kolpin et al., 2006; Mahler et al., 2017;
Chalar et al., 2017, Medalie et al., 2020).
The Uruguayan topography and the increase in the intensity of rainfall that occurs on travelling from south-west to north-east
determine areas with particularly high Quickflow values from the respective hills to the local catchments in the north (e.g. Cuchilla
de Belen) and east (e.g. Cuchilla de Heado, Cuchilla Grande or Cuchilla de las Averías; Fig. 1). These regions are among the areas with

11
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

Table 6
Estimated costs of indemnification to the landowner if buffer areas were put out of use by using costs per hectare based on the CONEAT Index (MGAP,
2022; Cutiño, 2015; Lanzilotta and Rosas, 2019). *Indicates departments that are part of the Santa Lucía catchment, which provides drinking water
for more than 60% of the Uruguayan population.
Department Area of the 100-m buffer per land use type (km2) Median Cost (US$ million) based on 10 years rent (US$ million)

Timber plantations Crops Modified grasslands USD/ha Median Mean Max

Artigas 1.3 0.0 59.2 843.74 8 17 32 1


Canelones* 23.6 10.9 129.1 2278.8 39 47 59 2
Cerro Largo 49.2 0.0 96.5 1056.66 20 28 58 2
Colonia 13.4 0.1 162.5 1830.93 55 103 101 5
Durazno 45.1 0.0 74.6 1024.5 16 26 45 2
Flores 13.3 0.0 57.5 1082.79 11 22 33 1
Florida* 27.3 0.1 93.2 2551.61 48 54 59 2
Lavalleja 69.6 0.2 44.0 1170.71 17 25 36 1
Maldonado 29.4 0.9 12.7 3806.15 23 35 14 1
Paysandú 37.1 2.9 99.0 1137.17 29 52 77 2
Rio Negro 45.3 0.2 91.4 2168.22 47 68 64 4
Rivera 86.9 0.0 35.5 1316.73 19 26 31 1
Rocha 34.7 0.8 76.3 947.29 15 30 48 1
Salto 4.7 4.2 83.4 659.2 9 21 48 1
San Jose* 9.8 1.7 122.0 2407.53 66 98 89 5
Soriano 24.1 0.0 150.4 1409.45 33 65 81 5
Tacuarembó 69.3 0.0 45.7 1285.81 18 24 35 1
Treinta y Tres 43.7 0.0 72.0 738.05 12 20 47 1
Total 631.6 25.6 1509.7 1539.74 482 755 961 34

the highest loads of phosphorus discharge into the local aquatic ecosystems, independent of scenario (Figs. 4 and 5). In these highly
sensitive areas, the influence of rainfall pattern, crop stages and different forms of management on pesticide and nutrient runoff should
be especially monitored, both frequently and at a local scale (Andrade et al., 2021).
As a result of the changing climate in Uruguay, heavy rainfall events (Barreiro, 2017) and induced peaks in pesticide concentrations
are expected to increase (Andrade et al., 2021). As those events are not covered by missing or low-frequency sampling, maximum
pesticide concentrations and fluxes may therefore be largely underestimated (Lefrancq et al., 2017 and see Appendix S6: Fig. S1).

4.3. Many scattered non-point pollution sources drive edge effects

The ongoing intensification of land use by silvi- and agriculture in rural Uruguay results in increasing patchiness of landscapes
along watercourses. Half of the areas covered by natural grasslands disappeared over a 15-year period, leaving many fragmented, small
grassland patches (Table 3). The numbers of patches covered by timber plantations and crops increased. The high number of scattered
crop and timber patches consistently reduces those parts of the natural habitats that are unaffected by agrochemicals. In Chinese
studies, fertilizer-driven trace metal pollution was positively correlated with land fragmentation, possibly as a result of higher
application of fertilizer to compensate for lower use efficiency in smaller crop areas (Ren et al., 2018).
In the surroundings of aquatic ecosystems, native vegetation has the potential to reduce and stop the transport of agrochemicals
towards aquatic ecosystems by biotic and abiotic mechanisms (Syversen and Bechmann, 2004; Weigelhofer et al., 2012; Lerch et al.,
2017). Firstly, plants can capture and store agrochemicals that depend on the growing season, type of vegetation and coverage per­
centage (Reddy et al., 1999; Lerch et al., 2017; Stutter et al., 2021). Secondly, plant cover reduces runoff velocity and increases the
potential for infiltration and adsorption (Lerch et al., 2017). Specifically, there is evidence from experimental evaluation that grassland
cover could diminish glyphosate runoff by ≈ 60% (Lerch et al., 2017) depending on the plant cover width, which acts as a buffer
(Syversen and Bechmann, 2004).

4.4. Low cost/high impact of buffer zones?

In our model, exclusion of highly productive land uses (crops, timber and highly managed grasslands, such as improved or artificial
grasslands) from a distance of only 100 m from aquatic ecosystems reduced the input of phosphorus and glyphosate by a third
(Table 5). Nationalization of these zones, with an area of ≈ 2167 km2, would cost U$S755 million, equivalent to less than 1.4% of the
Uruguayan BIP for the year 2020 (Table 6). Alternatively, implementation of a system of subsidies comprising ecological compensation
measures (Dworak et al., 2009), direct payments to farmers for ecosystem services provided by buffer zones (Veldman et al., 2015) or
the introduction of managed buffer zones could be integrated into (voluntary) certification procedures (Tröster and Hiete, 2018, but
see also Kill, 2016). Buffer zones integrated within agricultural landscapes have been tested as a novel edge-of-field approach within
riparian zones (Luke et al., 2019; Zak et al., 2019). Similarly, integration of buffer zones into grassland management approaches could
reduce trade-offs of grassland intensification (Jaurena et al., 2021) or an increase in cropland (Baeza and Paruelo, 2020).
In our scenarios, a considerable reduction of agrochemical runoff to aquatic ecosystems requires interventions at the scale of 100-m
buffers around aquatic ecosystems. If phosphorus and glyphosate are not applied within these zones, the runoff of both can be reduced
from the status quo by a third for phosphorus and by almost a half for glyphosate. Livestock should be excluded from all fluvial

12
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

channels, e.g. by fencing the littoral buffer riparian zone (Chalar et al., 2017). Our models also provide evidence that the expected
inputs of glyphosate resulting from grassland intensification (Jaurena et al., 2021) cannot be compensated for by the implementation
of buffers (Table 5).
Although the area of riparian forest has slightly increased and formerly isolated patches have merged with others (Ramírez and
Säumel, 2022), riparian forest fragmentation continues within a corridor from the aquatic ecosystems (Table 3). In general, these
riparian forests are the native vegetated buffer stripes along watercourses of the predominantly forestless temperate grasslands of
South America (Chebataroff, 1942). There is no need to plant buffer stripes; it is necessary only to restore the original riparian forests,
which will shape the pesticide concentrations detected in runoff water (Andrade et al., 2021). The riparian forests will also provide
multiple other benefits and ecosystem services (Sutherland et al., 2016).
In our modelling study, we created a 100-m buffer zone mainly due to the spatial resolution of the satellite imagery used to create
the land use maps (see Methods section). Although we observed a considerable reduction in agrochemical transport, other strategies
are essential to describe field evaluations. For example, the evidence and the broad literature have documented differences in the
effectiveness of reducing water pollution due to nutrient and agrochemical runoffs (Collins et al., 2009; Zhang et al., 2010; Haddaway
et al., 2018; Luke et al., 2019). There is evidence that there are no differences between C3 and C4 grass species if the buffer is
monospecific or diverse, but the buffer width is more critical than vegetation type (Syversen and Bechmann, 2004; Lerch et al., 2017).
Furthermore, a significant reduction in pesticides in buffer zones containing grasses has been documented compared with that con­
taining trees (e.g., Giaccio et al., 2016). Therefore, in the context of our study, a strategy for natural grassland conservation bordering
aquatic ecosystems would not require any significant investment, other than economic compensation to landowners.
In our modelling study, we created a buffer zone of 100 m mainly due to the spatial resolution of the satellite imagery used to create
the land use maps (see Methods section). Although we observed a considerable reduction in agrochemical transport, other strategies
are essential to describe from field evaluations. For example, the evidence and broad literature have documented differences in the
effectiveness of reducing water pollution due to nutrient and agrochemical runoff (Collins et al., 2009; Zhang et al., 2010; Haddaway
et al., 2018; Luke et al., 2019). There is evidence that there are no differences between C3 and C4 grasses species if the buffer is
monospecific or diverse, but the buffer width is more critical than vegetation type (Syversen and Bechmann, 2004; Lerch et al., 2017).
Furthermore, a significant reduction of pesticides in buffers with grasses has been documented over trees (e.g., Giaccio et al., 2016).
Therefore, in the context of our study, a strategy for natural grassland conservation bordering aquatic ecosystems would not require
any significant investment other than economic compensation to landowners.

4.5. There is glory in prevention!

The rivers of the Anthropocene have become depositories, with forms and processes drastically shaped by human impacts within
their catchment areas (Hein et al., 2021). In terms of input, mitigation should begin at the farm scale with implementation of best
management strategies with regard to the use of agrochemicals (Chalar et al., 2017), complemented at the landscape scale by riparian
buffers encouraged by territorial planning (Luke et al., 2019; Jabłońska et al., 2020). Clearly, to test and prove and evaluate the
effectiveness of conservation and restoration measures in the long term, the modelled impact of land use change on Uruguayan aquatic
ecosystems needs to be calibrated with a country-wide monitoring system that expands the existing monitoring stations along larger
rivers (Beretta-Blanco and Carrasco-Letelier, 2021), detecting pollutants beyond phosphorus and glyphosate. However, as tracing the
fate and impacts of yearly newly emerging agrochemicals in riparian zones is complex, and there will be time lags between the
implementation of mitigation measures and measurable impacts (Doehring et al., 2020), prevention is better than cure. Despite
knowledge about the negative consequences of the use of agrochemicals (Nogueira et al., 2021), and local communities bearing the
costs of monitoring and compensation for potential damage (Schwarzenbach et al., 2006; Schreinemachers and Tipraqsa, 2012;
Mateo-Sagasta et al., 2017), large corporations are generating profits for global shareholders. Political action by local governments is
needed to implement prevention and mitigation measures coordinated at a regional scale (Nogueira et al., 2021) and, as both con­
sumers and shareholders do not come from the local communities, pressure from the global citizenship is also required.

4.6. Model utility and uncertainties

While environmental modelling supports analyses, assessments and policy decisions concerning land use change and its impacts, its
dependence on data quality and availability is critical because monitoring networks for validating models are either lacking or are too
coarsely meshed (e.g. Redhead et al., 2018, Benez-Secanho and Dwivedi, 2019, Han et al., 2021). There has also been controversy in
the local scientific community about water pollution modelling and its origins (Beretta-Blanco and Carrasco-Letelier, 2021, and re­
plies). Although there is consensus regarding the applicability and good performance of the NDR model (Redhead et al., 2018, Yang
et al., 2019), some limitations need to be considered. For example, the NDR model is sensitive to the input of some parameters relating
to retention and the flow accumulation threshold, which modifies the outputs (Redhead et al., 2018; Han et al., 2021). In addition, the
spatial resolution of the geospatial information used could influence the results (Benez-Secanho and Dwivedi, 2019) and, as the
calibration parameters were obtained from literature, they may differ from the local context (Sharp et al., 2020). We have, however,
drawn on the best available data to estimate glyphosate and phosphorus discharge, and to identify areas with a higher likelihood of
increased export of agrochemicals. This leads to the findings that north-west Uruguay should be a priority area for new water-quality
monitoring stations, and that monitoring should be extended towards minor watercourses, particularly because only 19 of the 121
monitoring stations we were able to locate in Uruguay’s publicly available data contain data associated with glyphosate (Appendix S6:
Fig. S1).

13
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

Some considerations should be highlighted when evaluating the modeling performance using the KGE, NSE and RMSE coefficients.
Firstly, negative KGE and NSE results could be considered unsatisfactory (Knoben et al., 2019). Therefore, for non-standardized data,
the model fell. However, for standardized data, the modeling for glyphosate showed a good performance. These results are mainly due
to technical reasons and biophysical attributes not captured by the tool for assessing agrochemical runoff into aquatic ecosystems.
Firstly, we only found four water monitoring stations to intersect with hydric flow data to evaluate the model performance in the same
unit of measure (Appendix 3: Table S2). Thus, our access to reliable data was a critical limit to mathematical validation. Other studies
have raised the same limitation where the scarcity and access to data make the performance evaluation less satisfactory than expected
(e.g., Redhead et al., 2018, Anjinho et al., 2022). Secondly, the tool used (InVest software) does not consider other biophysical
processes acting inside and around aquatic ecosystems (Sharp et al., 2020), such as degradation of agrochemicals by microorganisms,
fixation by aquatics organisms (e.g., algae, plant, plankton) or organic matter (Hoffmann et al., 2009). Therefore, our model should be
considered as a guide to know visually the current state, the spatio-temporal change, and possible comparative impacts between
intensification and conservation of areas around aquatic ecosystems.
Although the NDR model is a discrete model and the export of agrochemicals represents the potential discharge after a rain event
prior to the application of agrochemicals, exploration of relative values of the outputs is suitable for exploring potential risks at a
national level. A regional approach is crucial for addressing water quality and management (Redhead et al. 2018, Nogueira et al, 2021;
Ouchi-Melo et al., 2021). These regional models help to sensitize local actors and policymakers, and to highlight the need for detailed,
spatially linked and publicly available pesticide and fertilizer use data (see example from the Canadian prairies; Malaj et al., 2020).

5. Conclusions

The dramatic extent of land use intensification leads to significant discharges of pollutants into local freshwater streams. Our model
provides evidence that a strategy to transform intensively used areas within the first 100 m of fresh waters to extensively used ones can
considerably reduce the pollution of aquatic ecosystems. However, the model and its results could be improved by the availability of
more precise and higher-resolution data on phosphorus and glyphosate inputs at the local level, e.g. with evaluations that consider
local landscape heterogeneity, such as soil properties, crop types and related data on the amount and temporal pattern of agrochemical
use than were available for our study. At the same time, our models and results provide a basis for participatory evaluation among
different stakeholders from local communities, farmers, the silvi- and agricultural industries, governmental institutions and NGOs. We
suggest that the water-quality monitoring network should be expanded to small freshwater streams, as well as to evaluating gov­
ernment investment in the creation of buffer zones to both dampen and decrease pollutant runoff. At the regional level, our results
show the relevance of collecting and accessing databases. Our study area, immersed in the temperate grasslands of South America, also
shares this biome with Argentina and Brazil, where the processes of land use change have similarities in terms of intensification by
monocultures. Therefore, there is a need to increase collaboration and methodological standardization and information flow so that
social actors interested in agrochemical runoff modeling can overcome current limitations. In the context of global change, prevention
is vastly superior to cure.

CRediT authorship contribution statement

The authors contributed equally to the study. I.S. received the funding grant for the project leading to this publication.

Declaration of Competing Interest

We have no conflicts of interest to declare.

Data Availability

Data will be made available on request.

Acknowledgements

We are very grateful to Vera Krause for the assistance with land use/cover change analysis, Serafina Bischoff for preparing and
running the SWY model and Sophia Reitzug for preparing and running the NDR model and for the design of the map visualization. The
study was funded by the German Federal Ministry of Education and Research (BMBF; 01LN1305A). Special thanks go to Amal
Chatterjee and Derek Collett for improving our English.

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.ejrh.2022.101283.

14
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

References

Alonso, J., Quintans, F., Bonilla, S., Arocena, R., Haakonsson, S., Aubriot, L., Goyenola, G., Muniz, P., Marrero, A., M, H., Venturini, N., AL, P., Iglesias, K., Conde, D.,
M, R., A, S., Eguren, G., Victoria, M., Tort, L.F., H, M., 2019. Water Quality in Uruguay: current status and challenges. pp. 592–631.
MA (Ministerio de Ambiente de Uruguay) (2020) [WWW Document]. Plaguicidas prioritarios para la gestión ambiental. URL 〈https://www.gub.uy/ministerio-
ambiente/comunicacion/publicaciones/plaguicidas-prioritarios-para-gestion-ambiental〉.
MGAP, 2022. Consulta de Índices y Grupos de Suelo CONEAT [WWW Document]. Minist. Ganad. Agric. Pesca. URL 〈https://www.gub.uy/ministerio-ganaderia-
agricultura-pesca/tramites-y-servicios/servicios/consulta-coneat〉 (accessed 1.24.22).
Duke, S.O., Powles, S.B., 2008. Glyphosate: a once-in-a-century herbicide. Pest Manag. Sci. 64, 319–325. https://doi.org/10.1002/ps.1518.
Dworak, T., Berglund, M., Grandmougin, B., Mattheiss, V., Holen, S.N., 2009. International review on payment schemes for wet buffer strips and other types of wet
zones along privately owned land. Ecologic Institute, Berlin/Vienna.
Eguren, G., Rivas-Rivera, N., García, C., Böcking, B., Bandeira, S., 2018. Water quality index for agricultural systems in Northwest Uruguay. Environ. Monit. Assess.
190, 710. https://doi.org/10.1007/s10661-018-7090-8.
Eker, S., Rovenskaya, E., Obersteiner, M., Langan, S., 2018. Practice and perspectives in the validation of resource management models. Nat. Commun. 9, 5359.
https://doi.org/10.1038/s41467-018-07811-9.
Elser, J.J., Bracken, M.E.S., Cleland, E.E., Gruner, D.S., Harpole, W.S., Hillebrand, H., Ngai, J.T., Seabloom, E.W., Shurin, J.B., Smith, J.E., 2007. Global analysis of
nitrogen and phosphorus limitation of primary producers in freshwater, marine and terrestrial ecosystems. Ecol. Lett. 10, 1135–1142. https://doi.org/10.1111/
j.1461-0248.2007.01113.x.
Exavier, R., Zeilhofer, P., 2021. OpenLand: software for quantitative analysis and visualization of land use and cover change. R. J. 12 https://doi.org/10.32614/RJ-
2021-021.
Feld, C.K., Fernandes, M.R., Ferreira, M.T., Hering, D., Ormerod, S.J., Venohr, M., Gutiérrez-Cánovas, C., 2018. Evaluating riparian solutions to multiple stressor
problems in river ecosystems - a conceptual study. Water Res. 139, 381–394. https://doi.org/10.1016/j.watres.2018.04.014.
Fick, S.E., Hijmans, R.J., 2017. WorldClim 2: new 1–km spatial resolution climate surfaces for global land areas. Int. J. Climatol. 37, 4302–4315. https://doi.org/
10.1002/joc.5086.
Giaccio, G.C.M., Laterra, P., Aparicio, V.C., Costa, J.L., 2016. Glyphosate retention in grassland riparian areas is reduced by the invasion of exotic trees. Int. J. Exp.
Bot. 85, 108–116.
Haddaway, N.R., Brown, C., Eales, J., Eggers, S., Josefsson, J., Kronvang, B., Randall, N.P., Uusi-Kämppä, J., 2018. The multifunctional roles of vegetated strips
around and within agricultural fields. Environ. Evid. 7, 14. https://doi.org/10.1186/s13750-018-0126-2.
Han, B., Reidy, A., Li, A., 2021. Modeling nutrient release with compiled data in a typical Midwest watershed. Ecol. Indic. 121, 107213 https://doi.org/10.1016/j.
ecolind.2020.107213.
Hein, T., Hauer, C., Schmid, M., Stöglehner, G., Stumpp, C., Ertl, T., Graf, W., Habersack, H., Haidvogl, G., Hood-Novotny, R., Laaha, G., Langergraber, G., Muhar, S.,
Schmid, E., Schmidt-Kloiber, A., Schmutz, S., Schulz, K., Weigelhofer, G., Winiwarter, V., Baldan, D., Canet-Marti, A., Eder, M., Flödl, P., Kearney, K., Ondiek, R.,
Pucher, B., Pucher, M., Simperler, L., Tschikof, M., Wang, C., 2021. The coupled socio-ecohydrological evolution of river systems: towards an integrative
perspective of river systems in the 21st century. Sci. Total Environ. 801, 149619 https://doi.org/10.1016/j.scitotenv.2021.149619.
Hendlin, Y.H., Arcuri, A., Lepenies, R., Hüesker, F., 2020. Like oil and water: the politics of (Not) assessing glyphosate concentrations in aquatic ecosystems. Eur. J.
Risk Regul. 11, 539–564. https://doi.org/10.1017/err.2020.65.
Hille, S., Andersen, D., Kronvang, B., Baattrup-Pedersen, A., 2018. Structural and functional characteristics of buffer strip vegetation in an agricultural landscape -
high potential for nutrient removal but low potential for plant biodiversity. Sci. Total Environ. 628–629, 805–814. https://doi.org/10.1016/j.
scitotenv.2018.02.117.
Hoffmann, C.C., Kjaergaard, C., Uusi-Kämppä, J., Hansen, H.C.B., Kronvang, B., 2009. Phosphorus retention in riparian buffers: review of their efficiency. J. Environ.
Qual. 38, 1942–1955. https://doi.org/10.2134/jeq2008.0087.
Jabłońska, E., Wiśniewska, M., Marcinkowski, P., Grygoruk, M., Walton, C.R., Zak, D., Hoffmann, C.C., Larsen, S.E., Trepel, M., Kotowski, W., 2020. Catchment-scale
analysis reveals high cost-effectiveness of wetland buffer zones as a remedy to non-point nutrient pollution in North-Eastern Poland. Water 12, 629. https://doi.
org/10.3390/w12030629.
Jaurena, M., Lezama, F., Salvo, L., Cardozo, G., Ayala, W., Terra, J., Nabinger, C., 2016. The dilemma of improving native grasslands by overseeding legumes:
production intensification or diversity conservation. Rangel. Ecol. Manag. 69, 35–42. https://doi.org/10.1016/j.rama.2015.10.006.
Jaurena, M., Durante, M., Devincenzi, T., Savian, J.V., Bendersky, D., Moojen, F.G., Pereira, M., Soca, P., Quadros, F.L.F., Pizzio, R., Nabinger, C., Carvalho, P.C.F.,
Lattanzi, F.A., 2021. Native grasslands at the core: a new paradigm of intensification for the campos of southern south america to increase economic and
environmental sustainability. Front. Sustain. Food Syst. 5, 547834 https://doi.org/10.3389/fsufs.2021.547834.
Kill, J., 2016. The role of voluntary certification in maintaining the ecologically unequal exchange of wood pulp: the Forest Stewardship Council’s certification of
industrial tree plantations in Brazil. J. Polit. Ecol. 23. https://doi.org/10.2458/v23i1.20247.
Knoben, W.J.M., Freer, J.E., Woods, R.A., 2019. Technical note: Inherent benchmark or not? Comparing Nash–Sutcliffe and Kling–Gupta efficiency scores. Hydrol
Earth Syst Sci 23, 4323–4331. https://doi.org/10.5194/hess-23-4323-2019.
Kolpin, D.W., Thurman, E.M., Lee, E.A., Meyer, M.T., Furlong, E.T., Glassmeyer, S.T., 2006. Urban contributions of glyphosate and its degradate AMPA to streams in
the United States. Sci. Total Environ. 354, 191–197. https://doi.org/10.1016/j.scitotenv.2005.01.028.
Lanfranco, B., Sapriza, G., 2011. El índice CONEAT como medida de productividad y valor de la tierra. https://doi.org/10.13140/2.1.3264.5926.
Lanzilotta, B., Rosas, E., 2019. Estimación del valor de los predios rurales y sus determinantes, Descentralización y desarrollo territorial. Oficina de Planeamiento y
Presupuesto, Montevideo, Uruguay.
Lefrancq, M., Jadas-Hécart, A., La Jeunesse, I., Landry, D., Payraudeau, S., 2017. High frequency monitoring of pesticides in runoff water to improve understanding of
their transport and environmental impacts. Sci. Total Environ. 587–588, 75–86. https://doi.org/10.1016/j.scitotenv.2017.02.022.
Lerch, R.N., Lin, C.H., Goyne, K.W., Kremer, R.J., Anderson, S.H., 2017. Vegetative buffer strips for reducing herbicide transport in runoff: effects of buffer width,
vegetation, and season. JAWRA: J. Am. Water Resour. Assoc. 53, 667–683. https://doi.org/10.1111/1752-1688.12526.
Little, C., Cuevas, J.G., Lara, A., Pino, M., Schoenholtz, S., 2015. Buffer effects of streamside native forests on water provision in watersheds dominated by exotic forest
plantations. Ecohydrology 8, 1205–1217. https://doi.org/10.1002/eco.1575.
Luke, S.H., Slade, E.M., Gray, C.L., Annammala, K.V., Drewer, J., Williamson, J., Agama, A.L., Ationg, M., Mitchell, S.L., Vairappan, C.S., Struebig, M.J., 2019.
Riparian buffers in tropical agriculture: scientific support, effectiveness and directions for policy. J. Appl. Ecol. 56, 85–92. https://doi.org/10.1111/1365-
2664.13280.
Mahler, B.J., Van Metre, P.C., Burley, T.E., Loftin, K.A., Meyer, M.T., Nowell, L.H., 2017. Similarities and differences in occurrence and temporal fluctuations in
glyphosate and atrazine in small Midwestern streams (USA) during the 2013 growing season. Sci. Total Environ. 579, 149–158. https://doi.org/10.1016/j.
scitotenv.2016.10.236.
Malaj, E., Liber, K., Morrissey, C.A., 2020. Spatial distribution of agricultural pesticide use and predicted wetland exposure in the Canadian Prairie Pothole Region.
Sci. Total Environ. 718, 134765 https://doi.org/10.1016/j.scitotenv.2019.134765.
Mander, Ü., Tournebize, J., Tonderski, K., Verhoeven, J.T.A., Mitsch, W.J., 2017. Planning and establishment principles for constructed wetlands and riparian buffer
zones in agricultural catchments. Ecol. Eng. 103, 296–300. https://doi.org/10.1016/j.ecoleng.2016.12.006.
Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models part I — a discussion of principles. J. Hydrol. 10, 282–290. https://doi.org/
10.1016/0022-1694(70)90255-6.
Nogueira, M.G., Perbiche-Neves, G., de Oliveira Naliato, D., Caglierani Casanova, S.M., Roberto Debastiani-Júnior, J., Espíndola, E.G., 2021. Limnology and water
quality in La Plata basin (South America) – spatial patterns and major stressors. Ecol. Indic. 120, 106968 https://doi.org/10.1016/j.ecolind.2020.106968.

15
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

Opperman, J.J., Moyle, P.B., Larsen, E.W., Florsheim, J.L., Manfree, A.D., 2017. Floodplains: processes and management for ecosystem services. Floodplains. Univ.
Calif. Press. https://doi.org/10.1525/9780520966321.
Oreskes, N., Shrader-Frechette, K., Belitz, K., 1994. Verification, validation, and confirmation of numerical models in the earth sciences. Science 263, 641–646.
https://doi.org/10.1126/science.263.5147.641.
Pañella, P.G., Cardozo, G., Cuadro, R., Reyno, R., Lezama, F., 2020. La fertilización fosforada disminuye la riqueza y aumenta el número de especies exóticas de
plantas en pastizales intersembrados con leguminosas. Ecol. Austral 30, 354–365. https://doi.org/10.25260/EA.20.30.3.0.1063.
Peel, M., Finlayson, B., Mcmahon, T., 2007. Updated world map of the koppen-geiger climate classification. Hydrol. Earth Syst. Sci. Discuss. 4. https://doi.org/
10.5194/hess-11-1633-2007.
R Core Team, 2016. R: The R Project for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria.
Alvarez, A., Blum, A., Gallego, F., 2015. Atlas de Cobertura del Suelo del Uruguay, 2015 (1st ed.). DINOT and FAO, Montevideo.
Andrade, V.S., Gutierrez, M.F., Regaldo, L., Paira, A.R., Repetti, M.R., Gagneten, A.M., 2021. Influence of rainfall and seasonal crop practices on nutrient and pesticide
runoff from soybean dominated agricultural areas in Pampean streams, Argentina. Sci. Total Environ. 788, 147676 https://doi.org/10.1016/j.
scitotenv.2021.147676.
Anjinho, P., da, S., Barbosa, M.A.G.A., Mauad, F.F., 2022. Evaluation of InVEST’s water ecosystem service models in a Brazilian subtropical basin. Water 14, 1559.
https://doi.org/10.3390/w14101559.
Mateo-Sagasta, J., Marjani, S., Turral, H., 2017. Water Pollution from Agriculture: A Global Review. Executive Summary, Global and Regional Assessments of
Wastewater and Water Quality. FAO and IWMI.
McDowell, R.W., Schallenberg, M., Larned, S., 2018. A strategy for optimizing catchment management actions to stressor-response relationships in freshwaters.
Ecosphere 9, e02482. https://doi.org/10.1002/ecs2.2482.
Moshman, R., 2005. The constitutional right to water in Uruguay. Sustain. Dev Law Policy 5, 65.
Ongley, E.D., Xiaolan, Z., Tao, Y., 2010. Current status of agricultural and rural non-point source Pollution assessment in China. Environ. Pollut. 158, 1159–1168.
https://doi.org/10.1016/j.envpol.2009.10.047.
Ouchi-Melo, L.S., do Amaral, B., Tavares, D.A., Bartozek, E.C.R., de Souza, J.E., Pereira, L.H.G., Auricchio, M.R., da Silva, M.X., Toyama, N.P., Nunes, P.H.,
Lambrecht, R.W., Peres, C.K., 2021. Brazilian vs. Paraguayan streams: Differences in water quality in a cross-border subtropical region. Limnologica 90, 125904.
https://doi.org/10.1016/j.limno.2021.125904.
QGIS Dvelopment Team, 2020. QGIS Geographic Information System. Open Source Geospatial Foundation.
DINAGUA (Dirección Nacional de Aguas). 2012. Ciclos anuales y estacionales de parámetros hidrólogicos (1980–2004). URL: 〈https://www.gub.uy/ministerio-
ambiente/sites/ministerio-ambiente/files/documentos/publicaciones/ciclos2011.pdf〉.
MVOTMA, 2015. Cobertura del Suelo 2015 [WWW Document]. URL 〈https://www.ambiente.gub.uy/geoservicios/〉 (accessed 10.5.21).
OAN, 2020. Plaguicidas prioritarios para la gestión ambiental [WWW Document]. Datos Abiertos - Minist. Ambiente - Obs. Ambient. Nac. URL 〈https://www.
ambiente.gub.uy/oan/datos-abiertos/〉 (accessed 6.12.21).
Baeza, S., Paruelo, J.M., 2020. Land use/land cover change (2000–2014) in the Rio de la Plata Grasslands: an analysis based on MODIS NDVI time series. Remote Sens
12, 381. https://doi.org/10.3390/rs12030381.
Barreiro, M., 2017. Interannual variability of extratropical transient wave activity and its influence on rainfall over Uruguay. Int. J. Climatol. 37, 4261–4274. https://
doi.org/10.1002/joc.5082.
Benbrook, C.M., 2016. Trends in glyphosate herbicide use in the United States and globally. Environ. Sci. Eur. 28, 3. https://doi.org/10.1186/s12302-016-0070-0.
Benez-Secanho, F.J., Dwivedi, P., 2019. Does quantification of ecosystem services depend upon scale (Resolution and Extent)? A case study using the InVEST nutrient
delivery ratio model in Georgia, United States. Environments 6, 52. https://doi.org/10.3390/environments6050052.
Beretta-Blanco, A., Carrasco-Letelier, L., 2021. Relevant factors in the eutrophication of the Uruguay River and the Río Negro. Sci. Total Environ. 761, 143299 https://
doi.org/10.1016/j.scitotenv.2020.143299.
Brondízio S., E., Settele, J., Díaz, S., Ngo T., H., 2019. In: Brondízio, E.S., et al. (Eds.), Global assessment report of the Intergovernmental Science-Policy Platform on
Biodiversity and Ecosystem Services. IPBES Secretariat, Bonn, Germany, 1144 pages. ISBN: 978-3-947851-20-1.
Chalar, G., Garcia-Pesenti, P., Silva-Pablo, M., Perdomo, C., Olivero, V., Arocena, R., 2017. Weighting the impacts to stream water quality in small basins devoted to
forage crops, dairy and beef cow production. Limnologica 65, 76–84. https://doi.org/10.1016/j.limno.2017.06.002.
Chebataroff, J., 1942. La Vegetación del Uruguay y sus Relaciones Fitogeográficas con la del Resto de la América del Sur. Rev. Geográfica Inst. Panam. Geogr. E
História 2, 49–90.
Collins, A.L., 2009. Mitigating diffuse water pollution from agriculture: riparian buffer strip performance with width. CAB Rev. Perspect. Agric. Vet. Sci. Nutr. Nat.
Resour. 4. https://doi.org/10.1079/PAVSNNR20094039.
Cutiño, J., 2015. Proyecciòn del precio de la tierra Campos en Uruguay 2021 / 2022 [WWW Document]. cuitinopropiedades.com.uy. URL 〈https://
cuitinopropiedades.com.uy/blog/proyeccion-precio-de-la-tierra-campos-en-uruguay-2020/2581〉 (accessed 1.24.22).
Davis, J., O’Grady, A.P., Dale, A., Arthington, A.H., Gell, P.A., Driver, P.D., Bond, N., Casanova, M., Finlayson, M., Watts, R.J., Capon, S.J., Nagelkerken, I.,
Tingley, R., Fry, B., Page, T.J., Specht, A., 2015. When trends intersect: the challenge of protecting freshwater ecosystems under multiple land use and
hydrological intensification scenarios. Sci. Total Environ. 534, 65–78. https://doi.org/10.1016/j.scitotenv.2015.03.127.
Doehring, K., Young, R.G., Robb, C., 2020. Demonstrating efficacy of rural land management actions to improve water quality - how can we quantify what actions
have been done? J. Environ. Manag. 270, 110475 https://doi.org/10.1016/j.jenvman.2020.110475.
Dufour, S., Rodríguez-González, P.M., Laslier, M., 2019. Tracing the scientific trajectory of riparian vegetation studies: main topics, approaches and needs in a globally
changing world. Sci. Total Environ. 653, 1168–1185. https://doi.org/10.1016/j.scitotenv.2018.10.383.
Gonçalves, C., Marins, A.T., do Amaral, A.M.B., Nunes, M.E.M., Müller, T.E., Severo, E., Feijó, A., Rodrigues, C.C.R., Zanella, R., Prestes, O.D., Clasen, B., Loro, V.L.,
2020. Ecological impacts of pesticides on Astyanax jacuhiensis (Characiformes: Characidae) from the Uruguay river, Brazil. Ecotoxicol. Environ. Saf. 205, 111314
https://doi.org/10.1016/j.ecoenv.2020.111314.
Gorgoglione, A., Gregorio, J., Ríos, A., Alonso, J., Chreties, C., Fossati, M., 2020. Influence of land use/land cover on surface-water quality of Santa Lucía River,
Uruguay. Sustainability 12, 4692. https://doi.org/10.3390/su12114692.
Gupta, H.V., Kling, H., Yilmaz, K.K., Martinez, G.F., 2009. Decomposition of the mean squared error and NSE performance criteria: implications for improving
hydrological modelling. J. Hydrol. 377, 80–91. https://doi.org/10.1016/j.jhydrol.2009.08.003.
Marques, J.G., de, C., Veríssimo, K.J., da, S., Fernandes, B.S., Ferreira, S.R., de, M., Montenegro, S.M.G.L., Motteran, F., 2021. Glyphosate: a review on the current
environmental impacts from a Brazilian perspective. Bull. Environ. Contam. Toxicol. 107, 385–397. https://doi.org/10.1007/s00128-021-03295-4.
McGarigal, K., Cushman, S.A., Ene, E., 2012. FRAGSTATS v4: Spatial Pattern Analysis Program for Categorical and Continuous Maps. University of Massachusetts,
Amherst.
Medalie, L., Baker, N.T., Shoda, M.E., Stone, W.W., Meyer, M.T., Stets, E.G., Wilson, M., 2020. Influence of land use and region on glyphosate and
aminomethylphosphonic acid in streams in the USA. Sci. Total Environ. 707, 136008 https://doi.org/10.1016/j.scitotenv.2019.136008.
Modernel, P., Rossing, W.A.H., Corbeels, M., Dogliotti, S., Picasso, V., Tittonell, P., 2016. Land use change and ecosystem service provision in Pampas and Campos
grasslands of southern South America. Environ. Res. Lett. 11, 113002 https://doi.org/10.1088/1748-9326/11/11/113002.
Naiman, R.J., Décamps, H., McClain, M.E., 2005. Riparia: Ecology, Conservation, and Management of Streamside Communities. Elsevier, Burlington, MA, p. 448.
Petraglia, C., Dell’Acqua, M., Pererira, G., Yussim, E. 2019. Mapa integrado de cobertura / uso del suelo del Uruguay, año 2018. In Anuario OPYPA (Oficina de
Programación y Política Agropecuaria). Accessed: 10.10.2021; 〈https://www.gub.uy/ministerio-ganaderia-agricultura-pesca/comunicacion/publicaciones/
mapa-integrado-coberturauso-del-suelo-del-uruguay-ano-2018〉.
Ramírez, L.R., Säumel, I., 2022. Beyond the boundaries: do spatio-temporal trajectories of land-use change and cross boundary effects shape the diversity of woody
species in Uruguayan native forests? Agric. Ecosyst. Environ. 323, 107646 https://doi.org/10.1016/j.agee.2021.107646.

16
L.R. Ramírez and I. Säumel Journal of Hydrology: Regional Studies 45 (2023) 101283

Reddy, K.R., Kadlec, R.H., Flaig, E., Gale, P.M., 1999. Phosphorus retention in streams and wetlands: a review. Crit. Rev. Environ. Sci. Technol. 29, 83–146. https://
doi.org/10.1080/10643389991259182.
Redhead, J.W., May, L., Oliver, T.H., Hamel, P., Sharp, R., Bullock, J.M., 2018. National scale evaluation of the InVEST nutrient retention model in the United
Kingdom. Sci. Total Environ. 610–611, 666–677. https://doi.org/10.1016/j.scitotenv.2017.08.092.
Reichert, J.M., Rodrigues, M.F., Peláez, J.J.Z., Lanza, R., Minella, J.P.G., Arnold, J.G., Cavalcante, R.B.L., 2017. Water balance in paired watersheds with eucalyptus
and degraded grassland in Pampa biome. Agric. . Meteorol. 237–238, 282–295. https://doi.org/10.1016/j.agrformet.2017.02.014.
Ren, S., Li, E., Deng, Q., He, H., Li, S., 2018. Analysis of the impact of rural households’ behaviors on heavy metal pollution of arable soil: taking lankao county as an
example. Sustainability 10, 4368. https://doi.org/10.3390/su10124368.
Rykiel, E.J., 1996. Testing ecological models: the meaning of validation. Ecol. Model. 90, 229–244. https://doi.org/10.1016/0304-3800(95)00152-2.
Sabzevari, S., Hofman, J., 2022. A worldwide review of currently used pesticides’ monitoring in agricultural soils. Sci. Total Environ. 812, 152344 https://doi.org/
10.1016/j.scitotenv.2021.152344.
Santos, C., 2020. La frontera hídrica del agronegocio en Uruguay: nuevas dinámicas de acumulación y despojo. Quid 16 Rev. Área Estud. Urbanos 12–34.
Schreinemachers, P., Tipraqsa, P., 2012. Agricultural pesticides and land use intensification in high, middle and low income countries. Food Policy 37, 616–626.
https://doi.org/10.1016/j.foodpol.2012.06.003.
Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B., Johnson, C.A., von Gunten, U., Wehrli, B., 2006. The challenge of micropollutants in aquatic systems.
Science 313, 1072–1077. https://doi.org/10.1126/science.1127291.
Schwarzenbach, R.P., Egli, T., Hofstetter, T.B., von Gunten, U., Wehrli, B., 2010. Global water pollution and human health. Annu. Rev. Environ. Resour. 35, 109–136.
https://doi.org/10.1146/annurev-environ-100809-125342.
Sharp, R., Douglass, J., Wolny, S., Arkema, K., Bernhardt, J., Bierbower, W., Chaumont, N., Denu, D., Fisher, D., Glowinski, K., Griffin, R., Guannel, G., Guerry, A.,
Johnson, J., Hamel, P., Kennedy, C., Kim, C.K., Lacayo, M., Lohnsdorf, E., Mandle, L., Rogers, L., Silver, J., Toft, J., Verutes, G., Vogl, A.L., Wood, S., Wyatt, K.,
2020. InVEST User Guide — InVEST Documentation [WWW Document]. URL 〈https://invest-userguide.readthedocs.io/en/latest/〉 (accessed 6.9.22).
Stafford, C., Leathers, M., Briggs, R., 1996. MR399: forestry-related nonpoint source pollution in maine: a literature review. Maine Agricultural and Forest Experiment
Station, Orono.
Stutter, M., Chardon, W.J., Kronvang, B., 2012. Riparian buffer strips as a multifunctional management tool in agricultural landscapes: introduction. J. Environ. Qual.
41, 297–303. https://doi.org/10.2134/jeq2011.0439.
Stutter, M., Costa, F.B., hUallacháin, D, Ó., 2021. The interactions of site-specific factors on riparian buffer effectiveness across multiple pollutants: a review. Sci. Total
Environ. 798, 149238 https://doi.org/10.1016/j.scitotenv.2021.149238.
Sutherland, I.J., Gergel, S.E., Bennett, E.M., 2016. Seeing the forest for its multiple ecosystem services: Indicators for cultural services in heterogeneous forests. Ecol.
Indic. 71, 123–133. https://doi.org/10.1016/j.ecolind.2016.06.037.
Syversen, N., Bechmann, M., 2004. Vegetative buffer zones as pesticide filters for simulated surface runoff. Ecol. Eng. 22, 175–184. https://doi.org/10.1016/j.
ecoleng.2004.05.002.
Taks, J., 2008. ‘El Agua es de Todos/Water for All’: water resources and development in Uruguay. Development 51, 17–22. https://doi.org/10.1057/palgrave.
development.1100464.
Tittonell, P., 2021. Beyond CO2: multiple ecosystem services from ecologically intensive grazing landscapes of South America. Front. Sustain. Food Syst. 5, 664103
https://doi.org/10.3389/fsufs.2021.664103.
Tröster, R., Hiete, M., 2018. Success of voluntary sustainability certification schemes – a comprehensive review. J. Clean. Prod. 196, 1034–1043. https://doi.org/
10.1016/j.jclepro.2018.05.240.
Veldman, J.W., Overbeck, G.E., Negreiros, D., Mahy, G., Le Stradic, S., Fernandes, G.W., Durigan, G., Buisson, E., Putz, F.E., Bond, W.J., 2015. Where tree planting and
forest expansion are bad for biodiversity and ecosystem services. BioScience 65, 1011–1018. https://doi.org/10.1093/biosci/biv118.
Viaud, V., Merot, P., Baudry, J., 2004. Hydrochemical buffer assessment in agricultural landscapes: from local to catchment scale. Environ. Manag. 34, 559–573.
https://doi.org/10.1007/s00267-004-0271-y.
Weigelhofer, G., Fuchsberger, J., Teufl, B., Welti, N., Hein, T., 2012. Effects of riparian forest buffers on in-stream nutrient retention in agricultural catchments.
J. Environ. Qual. 41, 373–379. https://doi.org/10.2134/jeq2010.0436.
Zak, D., Stutter, M., Jensen, H.S., Egemose, S., Carstensen, M.V., Audet, J., Strand, J.A., Feuerbach, P., Hoffmann, C.C., Christen, B., Hille, S., Knudsen, M., Stockan, J.,
Watson, H., Heckrath, G., Kronvang, B., 2019. An assessment of the multifunctionality of integrated buffer zones in northwestern Europe. J. Environ. Qual. 48,
362–375. https://doi.org/10.2134/jeq2018.05.0216.
Zhang, X., Liu, X., Zhang, M., Dahlgren, R.A., Eitzel, M., 2010. A review of vegetated buffers and a meta-analysis of their mitigation efficacy in reducing nonpoint
source pollution. J. Environ. Qual. 39, 76–84. https://doi.org/10.2134/jeq2008.0496TABLES.

17

You might also like