Content uploaded by Christian Schürings
Author content
All content in this area was uploaded by Christian Schürings on Dec 20, 2023
Content may be subject to copyright.
Agriculture, Ecosystems and Environment 361 (2024) 108818
Available online 22 November 2023
0167-8809/© 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Assessment of cultivation intensity can improve the correlative strength
between agriculture and the ecological status in rivers across Germany
Christian Schürings
a
,
*
, Daniel Hering
a
,
b
, Willem Kaijser
a
, Jochem Kail
a
a
Department of Aquatic Ecology, Faculty of Biology, University of Duisburg-Essen, Universit¨
atsstrasse 5, Essen D-45141, Germany
b
Centre for Water and Environmental Research, University of Duisburg-Essen, Universit¨
atsstrasse 5, Essen D-45141, Germany
ARTICLE INFO
Keywords:
Benthic Invertebrates
Cropland
Diatoms
Macrophytes
Nutrients
Pesticides
ABSTRACT
Agriculture has been identied as a main cause for more than 90% of Germany´s rivers still not meeting good
ecological status in 2021. While many large-scale studies observed a negative effect of catchment agricultural
land use on river biota, they rarely considered differences in cultivation intensities, although small-scale studies
highlight clear differences between the effects of agricultural crops. Here we used Germany-wide and spatially
explicit information on crop types to calculate agricultural intensity indices for nutrients and pesticides,
weighting different crop types based on average pesticide treatment and nutrient application rates. These indices
were then used as explanatory variables for the ecological status of n =7677 biological sampling sites. Pesticides
were more important than nutrient pollution for macroinvertebrates and macrophytes, while diatoms were more
sensitive to nutrients. Considering the most relevant intensity index (pesticide or nutrient) slightly increased the
correlative strength with ecological status, as compared to the correlation with agricultural land or cropland
cover by up to R
2
=0.14 for diatoms. Correlative strength of agricultural intensity indices was substantially
larger in small mountain and (pre)-alpine streams compared to lowland streams, with an R
2
up to 0.43 for
macroinvertebrates. These results not only conrm previous large-scale studies by demonstrating the detrimental
effects of present-day agriculture on river biota, but also shed light on the main pathways involved, particularly
highlighting the adverse impacts of agrochemicals. Consequently, to protect river biota, a shift to more sus-
tainable agricultural practices, like reducing pesticide application, is urgently required.
1. Introduction
Agricultural transition as targeted by the Convention on Biological
Diversity (CBD, 2022) is urgently required to protect biodiversity,
because current agricultural practices strongly impair both terrestrial
organisms such as mammals (Janova and Heroldova, 2016), birds (Rigal
et al., 2023) and ying insects (B¨
orschig et al., 2013) as well as river
biota (V¨
or¨
osmarty et al., 2010). There is strong empirical evidence for
the negative effect of agricultural land use on various riverine organisms
(Schürings et al., 2022), including macroinvertebrates, diatoms, mac-
rophytes and sh, notwithstanding large between-study heterogeneity.
Although positive effects of agricultural land use are observed in
exceptional cases (e.g. Townsend, et al., 2004; Niyogi et al., 2007), most
large-scale studies found negative impacts on biota with increasing
shares of agriculture adjacent to rivers (e.g. Turunen et al., 2016; Feld,
2013). Agriculture has even been identied as the key factor for fresh-
water deterioration (Wolfram et al., 2021) and consequently is one of
the main reasons for more than 90% of German rivers failing good
ecological status in 2021 (UBA, 2022).
These studies indicate that the share of agricultural land use in the
upstream catchment is already a good proxy for the agricultural stress
exerted on river biota. However, it does not reect differences in agri-
cultural cultivation intensities between crop types, referred to as agri-
cultural intensity in the following. Depending on agricultural intensity,
the effects on aquatic communities can strongly differ, as shown in many
small-scale studies (Weijters et al., 2009). In Germany, average maize
elds are fertilized with approximately twice the amount of nitrogen
and phosphorus compared to cereal elds, three times more than veg-
etables, and at least ten times more than permanent crops such as or-
chards and vineyards (Britz and Witzke, 2014). In contrast, permanent
crops such as orchards receive the highest pesticide treatment, which is
on average three times higher compared to vegetables, more than ve
times higher compared to cereals, and even 15-fold higher compared to
maize (Andert et al., 2015; Dachbrodt-Saaydeh et al., 2021).
* Corresponding author.
E-mail address: christian.schuerings@uni-due.de (C. Schürings).
Contents lists available at ScienceDirect
Agriculture, Ecosystems and Environment
journal homepage: www.elsevier.com/locate/agee
https://doi.org/10.1016/j.agee.2023.108818
Received 9 June 2023; Received in revised form 10 November 2023; Accepted 13 November 2023
Agriculture, Ecosystems and Environment 361 (2024) 108818
2
These differences in agricultural intensity are reected in the stress
exerted on rivers. Permanent crops such as orchards and vineyards have
been linked with high pesticide concentrations in rivers and river sedi-
ments (Schulz, 2001; Bermúdez-Couso et al., 2007). Corn farming
adjacent to rivers can cause massive ne sediment inux combined with
strong phosphorus loads (Secchi et al., 2011), while intensive livestock
farming in the river catchment is prone to nutrient inux following
heavy rainfall events (Mouri and Aisaki, 2015). These stressors are
known to impact river biota through many pathways:
Macroinvertebrates have been found to be strongly sensitive to ag-
rochemicals (Berger et al., 2016) and ne sediments eroded from agri-
cultural land, clogging interstitial spaces on the stream´s bottom and
covering lentic zones (Gieswein et al., 2019; Davis et al., 2022). For
sensitive macroinvertebrates, pesticides were even identied as the most
important stressor (Liess et al., 2021). While for both macrophytes and
diatoms negative effects of pesticides have also been observed (Debenest
et al., 2010; Ribeiro et al., 2019), macrophytes have been found mainly
depending on river morphology (Kaijser et al., 2022) and to some extent
also on nutrients (van Zuidam and Peeters, 2013, O’Hare et al., 2018),
whereas diatoms most strongly responded to eutrophication (Giorgio
et al., 2016; O’Hare et al., 2018). Therefore, organism groups should be
distinguished when investigating the effect of agricultural intensity
resulting from different crop types on river biota.
Furthermore, the relationships between agricultural intensity and
river biota are likely to differ between ecoregions because stronger ef-
fects in mountain compared to lowland streams were already observed
in studies using the sheer share of agriculture to quantify agricultural
stress (Feld, 2013, Li et al., 2018). Possible reasons may be truncated
gradients in lowlands yielding smaller effect sizes (Mack et al., 2022),
but also differences in land-use legacies. Since the mid-19th century,
European lowland rivers have been strongly degraded by adjacent
agriculture, in particular pesticides, nutrients, and morphological al-
terations (Feld, 2013), which likely still mirrored by the degree of
hydromorphological alterations (Greenwood et al., 2012). Moreover,
Leps et al. (2015) found that the effect of agricultural stress on river
biota depends on river size, with smaller associated effects for large
rivers compared to small streams. Hence, analysing stream types sepa-
rately is advisable.
These differences in agricultural intensity and biological effects
resulting from different crop types are likely responsible for a substantial
part of the between-study heterogeneity of large-scale studies, which
used the sheer share of catchment agricultural land use to quantify
agricultural stress (e.g. Turunen et al., 2016; Davis et al., 2022). Such
crop-specic differences have not yet been considered in large-scale
empirical studies, because respective data on crop types to assess agri-
cultural intensity are usually not available at large spatial scales. Only
recently, spatially explicit high-resolution maps on crop types were
published, such as the German-wide maps of Blickensd¨
orfer et al.
(2022). These maps allow for large-scale crop-specic analyses of agri-
cultural effects on river biota, to complement and upscale the existing
small-scale case studies on the effects of nutrients and agrochemicals,
which found differences in the organism groups’ responses to agricul-
tural stress.
Against this background, this large-scale study aimed at analysing
the impact of agricultural land use on river biota, considering agricul-
tural intensity by calculating pesticide and nutrient intensity indices
based on typical application rates for different crop-types and recently
published crop-type maps for Germany. More specically, we hypoth-
esized that: (1) Organism groups differ in their sensitivity to agricultural
stress, with macroinvertebrates being more sensitive to pesticides, while
macrophytes and diatoms are more sensitive to nutrients. (2) Consid-
ering agricultural intensity increases the correlative strength of the re-
lationships between freshwater biota and agriculture compared to the
sheer share of agriculture or cropland cover. (3) The effect of agricul-
tural intensity clearly differs between ecoregions, with the strongest
effects in mountain and (pre)-alpine streams.
2. Methods
2.1. Biological data
Biological data on n=7677 sampling sites were acquired from all
German federal states, except the Saarland, which are partly autono-
mous regions within the federal republic of Germany, with samples
taken between 2010 and 2019. The sampling sites were grouped into the
following ve groups of river types (Pottgiesser and Sommerh¨
auser,
2008) that were investigated separately (from here on referred to as
stream type groups). This was done to reect the main differences in
river size (streams vs. rivers) and between the three main ecoregions,
resulting in differences in environmental factors such as soil types, ow
velocity, or stream morphology, potentially inuencing the effect of
agricultural intensity on river biota. The following groups were
considered: Small (n=3065) and large (n=924) mountain streams,
small (n=2648) and large (n =630) lowland streams and (pre-)alpine
streams (n=410) (grouping of stream types see Table S1).
The biological samples were taken by the German federal agencies
using standardized methods for ecological status assessment according
to the EU Water Framework Directive in Germany. Macroinvertebrates
were sampled according to a multi-habitat sampling method (Haase
et al., 2004) and the species-level taxa lists were then processed by the
online tool PERLODES (https://www.gewaesser-bewertung-ber-
echnung.de/index.php/perlodes-online.html) to calculate a river-type
specic multimetric macroinvertebrate index (n=7677). In a similar
way, macrophytes and diatoms were sampled according to a
multi-habitat method (Schaumburg et al., 2012) and the species-level
taxa lists were then processed by the online tool PHYLIB
(https://www.gewaesser-bewertung-berechnung.de/index.php/phyli-
b-online.html) to calculate both, a multimetric macrophyte index
(n=2792 sampling sites) and diatom index (n=3307). The three bio-
logical metrics (macroinvertebrate index, macrophyte index, and
diatom index) assess the ecological status and reect the response of the
three organism groups to various stressors, i.e. pollution or hydro-
morphological alterations (B¨
ohmer et al., 2004). For sampling sites that
had been sampled multiple times, the sample closest to 2018 was chosen
to describe the biological conditions the year after 2017 from which the
land use data originate.
2.2. Catchment delineation
For each sampling site, terrestrial land use in the entire upstream
catchment was quantied. First, the drainage basins upstream of the
sampling sites were derived from a digital elevation model (DEM) with
10 m resolution using ArcView 3.3 and subsequently visually checked.
Second, the percentage cover of 23 different crop types was quantied
using ESRI ArcGIS Pro 2.9.0 and Spyder (Phyton 3.7). This was based on
land use data for 2017 derived through random forest classication of
Sentinel-2, Landsat 8 and Sentinel-1 data with a 10 m resolution by
Blickensd¨
orfer et al. (2022).
2.3. Agricultural land use
To quantify agricultural land use in the catchments, the 23 different
crop types distinguished by Blickensd¨
orfer et al. (2022) were grouped in
three different ways. First, all crop types were merged to calculate the
area in the catchments covered by any type of agricultural land use,
referred to as agriculture in the following, with an average catchment
coverage of 44.46% (SD =22.38). Second, the area covered by all crop
types except for grassland was calculated (referred to as cropland in the
following), given that grassland is the by far least intensive agricultural
land use type, and this distinction has already been made by other
large-scale studies with an average catchment coverage of 31.92% (SD =
23.36). Third, the area in the catchments covered by the different crop
types (including grassland) was weighted based on pesticide treatment
C. Schürings et al.
Agriculture, Ecosystems and Environment 361 (2024) 108818
3
(Dachbrodt-Saaydeh et al., 2021) and nutrient application rates, in
particular nitrogen and phosphorus (Britz and Witzke, 2014), referred to
as the pesticide-, nitrogen-, and phosphorus indices in the following. The
weights for the pesticide intensity index were derived by allocating the
23 crop types of Blickensd¨
orfer et al. (2022) to the 15 crop types
distinguished by Dachbrodt-Saaydeh et al. (2021), who assessed pesti-
cide treatment based on data from 90 farms distributed across Germany
for the year 2017, supplemented with the information on pesticide
treatment for legumes reported in Andert et al. (2015). For grassland, no
pesticide treatment was assumed. Subsequently, the percentage area
covered by the 23 crop types in the catchments was weighted based on
the average pesticide treatment (Table S2), ranging between 0 and 1,
resulting in values for the pesticide index. Similarly, the 23 crop types of
Blickensd¨
orfer et al. (2022) were allocated to the 16 crop types
(including grassland), for which the mean nutrient application rates
were derived from the CAPRI model (Britz and Witzke, 2014) for the
years 2011–2013, separately for nitrogen and phosphorous (used as
baseline for the CAPRI model). The crop-specic nutrient applications
used in the CAPRI model draw from European statistics, including the
Land Use / Cover Area Frame Statistical Survey (LUCAS), EUROSTAT,
and FAOSTAT. They are estimated based on the overall nutrient input
per region and area, crop-specic nutrient and phosphorous consump-
tion and nutrient loss from harvest, biological xation, and nutrient
transformation processes. Again, the percentage area covered by the 23
crop types in the catchments (including grassland) was weighted
accordingly (Table S2), resulting in values for the nitrogen index and the
phosphorus index. Forest land use was disregarded for all grouping, as
pesticide and nutrient application rates are generally not available
(Halbach et al., 2021) and observed concentrations are low, wherefore
forests are often regarded as refuges (Schneeweiss et al., 2022).
2.4. Statistical analysis
To test hypothesis 1, that organism groups differ in their sensitivity
to agricultural stressors (pesticides, nitrogen, and phosphorus), we used
random forest models as implemented in the R-package randomForest
(v4.7–1.1). This method was chosen because no prior decision on dis-
tribution is required and potential non-linear relations can be assessed.
We ran the models with ntree =1000 and default settings separately for
each combination of the three organism groups and ve stream groups,
resulting in a total of 15 models. The ecological status of the respective
organism group was used as response and the following variables were
used as predictors: the three agricultural intensity indices, river type
(Pottgiesser and Sommerh¨
auser, 2008) to account for i.a. altitude and
size of rivers, year of biological sampling to account for differences in
agricultural land use between years, federal states to account for dif-
ferences in environmental conditions and possible differences in sam-
pling, and the water body category, i.e. whether a sampling site was
identied as natural or heavily modied according to Article 4 of the EU
Water Framework Directive. Then we calculated the relative variable
importance of the three intensity indices to identify the intensity index
capturing the most variance relative to the other two indices for each of
the 15 models. Relative variable importance is a measure of how well
the intensity indices are predicting changes in the response and can
therefore be considered a proxy for the strength of the correlation. For
each of the 15 combinations of organism groups and stream groups (15
models), the relative importance of the three intensity indices sums up to
100% and the index with the highest relative variable importance was
selected for further analyses.
To investigate hypothesis 2 of higher correlative strength between
agriculture and freshwater biota, when agricultural intensity is consid-
ered, we tted Generalized Linear Mixed Models (GLMMs) with the
gamlss package in r (v5.2–0). A logit link was used in the GLMMs as well
as a beta distribution (BEINF) given that the ecological status used as
response can take values from zero to one. More precisely, zero-one
inated beta models were used to allow the ecological status to take
the exact value of zero and one. We built separate models for each
combination of the three organism groups as response and three
different ways to group the 23 crop types as predictors (percentage cover
of agriculture, cropland, and the intensity index with the highest relative
variable importance), called xed effects in GLMMs, resulting in nine
different models. Only one xed effect was used in each model to
adequately compare the different ways to group the crop types and
avoid co-correlation issues between the different ways of grouping the
crop types, wherefore we refrained from also using the intensity indices
with lower relative importance.
These nine models were further subdivided into models for each of
the ve groups of sampling sites (small and large mountain and lowland
streams, as well as (pre-)alpine streams), resulting in a total of 9 ×5 =45
models. In addition to the single xed effect per model, the following
random effects were included to account for general differences in the
ecological status resulting from these random variables: river types, year
of biological sampling, federal state, category (natural vs. heavily
modied water body). The xed variables (percentage cover of agri-
culture and cropland and the intensity indices) were square root trans-
formed to improve the models. For each of the 45 models, 70% of the
data were bootstrapped, and cross-validation was applied to the
remaining 30% for 1000 iterations, each to calculate a mean-pseudo-R
2
(from here on referred to as R
2
) for the xed effect, including condence
intervals. The R
2
of the xed effect was calculated as the squared cor-
relation between the tted response and the predicted response solely
based on the xed effects (the R
2
of the full models are shown in
Table S3. The models were checked visually for residual distribution
against predicted values and each variable, yielding centered averages
and symmetrical distributions.
To investigate hypothesis 3 of clear differences in agricultural effects
among ecoregions, the same models already built for testing hypothesis
2 were used, focusing on the difference between the stream type groups.
3. Results
3.1. Relative importance of agricultural intensity indices (pesticide,
nitrogen and phosphorus) differed between organism groups
Our rst hypothesis on differences in the sensitivity of the three or-
ganism groups to the three agricultural stressors (pesticides, nitrogen,
and phosphorus) was largely supported by the results of the random
forest models. Macroinvertebrates were indeed most sensitive to pesti-
cides (Table 1). In the ve random forest models for macroinvertebrates
in the different stream groups, the relative importance of the pesticide
index ranged between 39% and 49% and was always larger compared to
the relative importance of the nitrogen (29–34%) and phosphorus index
(22–30%). For both macrophytes and diatoms, the differences in relative
importance of the three agricultural intensity indices in each of the ve
models were smaller compared to the models for macroinvertebrates.
Other than expected, macrophytes were generally not most sensitive to
nutrients but to pesticides (Table 1). In four out of the ve models,
relative importance values for the pesticide index were larger (35–41%)
compared to the other two intensity indices, and the phosphorus index
was only largest in the model on the (pre)-alpine stream group (37%).
For diatoms, the relative importance of the phosphorus index (34–39%)
exceeded the importance of the nitrogen index (30–34%) and pesticide
index (30–33%) in all ve random forest models for the different stream
groups, in accordance with our expectations (Table 1). For the subse-
quent analyses to test hypotheses 2 and 3, the pesticide index was used
as a proxy for agricultural intensity for macroinvertebrates and macro-
phytes, except for macrophytes in the (pre)-alpine stream group, while
the phosphorus index was used for macropyhtes in the (pre-)alpine
stream group and for diatoms in all ve stream groups.
C. Schürings et al.
Agriculture, Ecosystems and Environment 361 (2024) 108818
4
3.2. Considering agricultural intensity slightly increased correlative
strength
As expected in hypothesis 2, the correlative strength of the re-
lationships between the ecological status and agriculture tended to be
higher when considering agricultural intensity compared to using the
sheer share of cropland or agricultural land, but differences were small
in most of the 45 models. In general, the agricultural intensity indices
and the percentage cover of cropland and agriculture were strongly
correlated (Table S4). This was especially true for the agricultural in-
tensity indices and the percentage cover of cropland, with Spearman
ρ
ranging between 0.77 and 0.98. Hence, the correlative strength only
slightly differed between the models, with the intensity indices and
percentage of cropland cover as xed effects, and most condence in-
tervals overlapped (Figs. 1–3).
The highest overall correlative strength and differences between the
agricultural intensity index and percentage cover of cropland and agri-
culture was found for macroinvertebrates’ ecological status. The
correlative strength R
2
of the pesticide intensity index as xed effect was
up to R
2
=0.43, capturing 0.06 more than cropland cover (R
2
=0.37)
and 0.17 more compared to the percentage cover of agriculture (R
2
=
0.26) as xed effect (Fig. 1). While the correlative strength of percentage
agricultural cover (including grassland) was clearly lower in some of the
ve stream groups, condence intervals of the percentage cover of
cropland and pesticide intensity index overlapped in all ve stream
groups.
For macrophytes, the correlative strength of the relationships be-
tween the ecological status and agriculture was generally lower, and R
2
was only up to 0.20 (Fig. 2). Again, the correlative strength tended to be
higher for the models with the pesticide or phosphorus intensity index as
xed effect (up to R
2
=0.20) compared to the models´percentage cover
of cropland (up to R
2
=0.16) or agriculture (up to R
2
=0.12) as pre-
dictor, but condence intervals strongly overlapped in all ve stream
groups.
For diatoms, the correlative strength was similarly low compared to
the macrophytes’ models, except for the (pre)-alpine stream group with
R
2
up to 0.40 (Fig. 3). However, in all models, the correlative strength of
the models with the phosphorus intensity index as xed effect tended to
be higher compared to the models with the sheer share of cropland and
agriculture as predictors, even though most condence intervals over-
lapped. At the same time, the correlative strength of the phosphorus
index of R
2
=0.14 was more than four times higher compared to the
models with percentage cover of cropland (R
2
=0.03) and agriculture
(R
2
=0.03) in one model. Other than for macroinvertebrates and mac-
rophytes, the correlative strength of the diatom models did not tend to
be higher for the percentage cropland compared to the percentage
agriculture.
Table 1
For each of the three organism groups, relative importance of the three different
agricultural intensity indices (pesticide, nitrogen, phosphorus) are given, sepa-
rately for each stream group. The largest relative importance value for each
model is given in bold. Relative importance was calculated based on random
forest models with the three intensity indices as xed effects and four additional
random factors (river type, year of biological sampling, federal state, category).
Macroinvertebrates
Mountain
small
Mountain
large
Lowland
small
Lowland
large
(Pre)-
alpine
Pesticide
index
49% 41% 39% 37% 41%
Nitrogen
index
29% 31% 33% 33% 29%
Phosphorus
index
22% 28% 28% 30% 30%
Macrophytes
Mountain
small
Mountain
large
Lowland
small
Lowland
large
(Pre)-
alpine
Pesticide
index
41% 38% 36% 35% 33%
Nitrogen
index
32% 31% 34% 35% 30%
Phosphorus
index
27% 31% 30% 30% 37%
Diatoms
Mountain
small
Mountain
large
Lowland
small
Lowland
large
(Pre)-
alpine
Pesticide
index
32% 32% 33% 33% 30%
Nitrogen
index
29% 30% 33% 33% 34%
Phosphorus
index
39% 38% 34% 34% 36%
Fig. 1. Zero one inated GLMMs with the macroinvertebrate’s multimetric index (ecological status) as response. For each stream group (MS =Mountain Small,
n=3065; ML =Mountain Large, n=924; LS =Lowland Small, n=2648; LL =Lowland Large, n=630; AL =(Pre-)alpine, n=410) the correlative strength (R
2
) of the
three ways of grouping crop types in the catchment as xed effect (percentage cover of agriculture and cropland, pesticide intensity index) is shown with 95%
condence intervals (based on bootstrapping with 1000 replications).
C. Schürings et al.
Agriculture, Ecosystems and Environment 361 (2024) 108818
5
3.3. The effect of agricultural intensity clearly differed between ecoregions
and stream types
As expected in hypothesis 3, the effect of agricultural intensity on the
ecological status clearly differed between ecoregions and stream types,
as reected in the ve stream type groups. For all three organism groups,
agricultural intensity had the strongest effects in (pre-)alpine and small
mountain streams. The variation in correlative strength among stream
type groups was most pronounced for macroinvertebrates, with a
correlative strength that was more than three times higher in small
mountain streams (R
2
=0.43) and (pre-)alpine streams (R
2
=0.40),
compared to large mountain streams with R
2
=0.12 (Fig. 1). In small-
and large lowland streams (R
2
=0.06 and 0.01), the correlative strength
was almost negligible. Regarding macrophytes, the correlative strength
was smaller in general, and the differences were less distinct between
small mountain streams (R
2
=0.20) and (pre)-alpine streams (R
2
=
0.18), compared to large mountain streams with R
2
=0.14 (Fig. 2).
Similar to macroinvertebrates, small- and large lowland streams showed
only minimal correlative strength (R
2
=0.02 and 0.03). For diatoms,
unlike macroinvertebrates and macrophytes, the correlative strength in
the lowlands was not negligible and much higher for the intensity index
compared to percentage cover of cropland and agriculture (Fig. 3). The
differences between mountain and lowland streams were relatively
small, ranging between R
2
=0.11 in small lowland streams and R
2
=
0.17 in small mountain streams. However, large differences were
observed when compared to (pre)-alpine streams (R
2
=0.40).
4. Discussion
This large-scale study aimed at analysing the effect of agricultural
Fig. 2. Zero one inated GLMMs with the macrophyte´s multimetric index (ecological status) as response. For each stream group (MS =Mountain Small, n=449; ML
=Mountain Large, n=924; LS =Lowland Small, n=677; LL =Lowland Large, n=186; AL =(Pre-)alpine, n=232) the correlative strength (R
2
) of the three ways of
grouping crop types in the catchment as xed effect (percentage cover of agriculture and cropland, pesticide Index for mountain- and lowland streams and Phosphor
Index for (pre)-alpine streams) is shown with 95% condence intervals (based on bootstrapping with 1000 replications).
Fig. 3. Zero one inated GLMMs with the diatom’s multimetric index (ecological status) as response. For each stream group (MS =Mountain Small, n=1556; ML =
Mountain Large, n=506; LS =Lowland Small, n=747; LL =Lowland Large, n=171; AL =(Pre-)alpine, n=327) the correlative strength (R
2
) of the three ways of
grouping crop types in the catchment as xed effect (percentage cover of agriculture and cropland, phosphorus intensity index) is shown with 95% condence
intervals (based on bootstrapping with 1000 replications).
C. Schürings et al.
Agriculture, Ecosystems and Environment 361 (2024) 108818
6
intensity (as reected by pesticides, nitrogen, and phosphorus) on three
riverine organism groups (macroinvertebrates, macrophytes, diatoms)
in ve different types of streams (stream type groups). In contrast to
many previous large-scale studies using the sheer share of cropland or
agriculture (including grassland), agricultural intensity was assessed by
calculating pesticide and nutrient intensity indices. The area of the up-
stream catchment of biological sampling sites covered by different crop
types in maps recently published for Germany (Blickensd¨
orfer et al.,
2022) was weighted based on typical pesticide and nutrient application
rates (Table S2) for the crop types (Britz and Witzke, 2014;
Dachbrodt-Saaydeh et al., 2021). Even though the application data are
affected by uncertainties caused by needed assumptions and upscaling,
yielding the large spatial coverage of this study, they still offer crucial
insight into crop-specic differences in farming practices.
The results might potentially be affected by a temporal mismatch
between the crop data for 2017 and the year of the biological sampling
(2010–2019) as well as years of fertilizer (2011–2013) and pesticide
inputs (2017). However, crop-specic fertilizer and pesticides inputs
only slightly varied in the last decade (Britz and Witzke, 2014;
Dachbrodt-Saaydeh et al., 2021). Although the biological sample closest
to 2018 was chosen to describe the biological conditions the year after
2017, the mean deviation of the sampling year from 2018 was nearly
three years. However, even though the crops grown on specic elds
change between years, the overall share of the individual crops at larger
spatial scales, i.e. within the catchments, tended to change only slightly
in the years 2017, 2018, and 2019 (Blickensd¨
orfer et al., 2022), and
analysis with data from 2018 gave similar results (Table S5). Still, the
year of biological sampling was used as a random factor in the GLMMs to
account for differences in the crops grown between years.
4.1. Relative importance of agricultural intensity indices on pesticide,
nitrogen and phosphorus differed between organism groups
Our analysis highlights clear differences in the responses of the
different organism groups to the agricultural stressors: nutrients and
pesticides (Hypothesis 1). The expectation of higher pesticide sensitivity
of macroinvertebrates was largely supported. The higher relative
importance of pesticides for macroinvertebrates compared to the two
nutrient intensity indices is consistent with recent ndings in literature.
While nutrients affect macroinvertebrates rather indirectly by
increasing primary production and temporarily reducing oxygen con-
centrations (Dodds, 2006), which favors competitive species and elim-
inates sensitive species (Weijters et al., 2009), direct pesticide effects on
macroinvertebrates have been shown in detail in many mesocosm
studies (Roessink et al., 2013; Morrissey et al., 2015). Also in the eld,
effects of pesticide groups such as pyrethroid (Wurzel et al., 2020) have
been observed with negative effects, particularly on sensitive aquatic
insect larvae (e.g. Ephemeroptera and Trichoptera), which are particu-
larly relevant for the assessment of the ecological status. Findings of
Wernecke et al. (2019) indicate that pesticide mixtures, as present in
real-life situations such as assessed in this study, are likely to have even
larger effects. On that note, Liess et al. (2021) identied pesticides as the
most relevant stressors for sensitive macroinvertebrates. Similarly,
Wolfram et al. (2021) reasoned that agriculture is the most important
stressor for macroinvertebrates, mainly based on pesticide pressure.
While we expected macrophytes to be most sensitive to nutrients
following van Zuidam and Peeters (2013) and O’Hare et al., (2018), this
was only true for the (pre-)alpine streams. The relative importance of the
pesticide intensity index was higher in the other stream groups, sug-
gesting higher pesticide sensitivity, likely caused by herbicides (e.g.
Mohr et al., 2007). However, macrophytes seem to recover rather
quickly after typical short-term pesticide peaks (King et al., 2016;
Wieczorek et al., 2017), and the differences in relative importance be-
tween the intensity indices in the models on macrophytes were small.
Several recent studies rather suggest that hydrological and morpholog-
ical factors as well as river management may be more important for
macrophyte occurrence than pesticides and nutrients (Baczyk et al.,
2018; Kaijser et al., 2022), which also explains the relatively low
explanatory power of our macrophyte models.
The results for diatoms, on the other hand, met our expectation of a
stronger response to nutrients compared to pesticides, as suggested by
the highest relative importance for the phosphorus index in all stream
types. This is in line with the observations of Hilton et al., (2006) and
O’Hare et al., (2018). Similar to macrophytes, diatoms also quickly
recover after eld typical pesticide peaks (Bighiu et al., 2020). However,
they appear to strongly react to nutrients, and tolerant taxa may benet,
while sensitive taxa are eliminated (Kelly et al., 2009; Giorgio et al.,
2016). The higher relative importance of the phosphorus index
compared to the nitrogen index could be associated with the phosphorus
limitation of diatoms (Bothwell and Kilroy, 2011). The relative impor-
tance particularly differs in the mountain streams, which could be
related to different pathways of entry. Other than nitrogen, phosphorus
mainly enters rivers bound to particles (Dorioz et al., 2006) and the
steeper slopes in mountainous streams may result in fewer particles
settling before reaching the rivers (Parkyn, 2004). Consequently, the
phosphorus index may also indicate the impact of ne sediment on di-
atoms (Jones et al., 2017).
4.2. Considering agricultural intensity slightly increased correlative
strength
Our expectation of an increase in the correlative strength of agri-
culture for the ecological status when cultivation intensity is considered
(Hypothesis 2) could be only partly supported by the results, as the
differences were relatively small, particularly between percentage
cropland and the intensity indices. Only for macroinvertebrates and, to
some extent, for macrophytes, the correlative strength strongly differed
between percentage share of cropland and agriculture, which is likely
because no pesticides are applied on grasslands (Riedo et al., 2022). As
diatoms appear more sensitive to nutrients compared to pesticides (see
above), the observed similar correlative strength between percentage
share of cropland and agriculture for diatoms is likely caused by nutrient
runoff into rivers caused by both fertilization of grasslands and nutrients
from livestock farming (Mouri and Aisaki, 2015). Our observation that
the correlative strength of the intensity indices were only slightly higher
in comparison to percentage share of cropland can be explained by
strong correlations between both indices (Table S4). This can be
explained by the differences in area cover between the individual crops.
Wheat, maize, rape seed, and barley account for more than 70% of the
cropland area in Germany (Blickensd¨
orfer et al., 2022), which are all
known for relatively high nutrient (Britz and Witzke, 2014) but only
medium pesticide application rates (Andert et al., 2015). Pesticide
intensive crops such as vineyards, orchards, hops, and vegetables, on the
other hand (Dachbrodt-Saaydeh et al., 2021), only cover less than ten
percent of the cropland. Consequently, those crop types less frequently
sum up to a percentage resulting in more severe stress than the ubiq-
uitous crop types already impose. Aside from the strong correlation with
cropland, the relatively low performance of the intensity indices may
also be linked to the underlying data, particularly the underlying crop
maps of Blickensd¨
orfer et al. (2022), which do not allow discrimination
between regional differences of agricultural practices such as organic
farming. While nutrient application rates do not tend to differ strongly
between conventional- and organic farming (Oelofse et al., 2010), no
pesticides are applied in the latter resulting in lower pesticide residues in
soils from organic farming (Geissen et al., 2021). Additionally, the soil
conditions (Dobbie and Smith, 2003), slope (Ekholm et al., 2000;
Cambien et al., 2020) and riparian vegetation (Palt et al., 2023) have
been shown to strongly inuence agricultural effects on river biota.
C. Schürings et al.
Agriculture, Ecosystems and Environment 361 (2024) 108818
7
4.3. The effect of agricultural intensity clearly differed between ecoregions
and stream types
As expected in hypothesis 3, the correlative strength between agri-
cultural land use and the biota response differed between ecoregions and
was higher in the small mountain streams and the (pre-)alpine region
than in large streams, particularly in the lowlands. This coincides with
the ndings of Li et al. (2018), who showed that catchment land use is
less important for large streams compared to low-order streams. This can
be partly explained by the river continuum concept (Vannote et al.,
1980), suggesting that species composition changes downstream from
the sources and becomes more homogeneous. Moreover, small streams
generally have smaller catchments, which are more likely to be domi-
nantly covered by individual crop types, resulting in larger land use
gradients, while in large catchments, differences in land use are more
likely to average out. The difference in gradients may partly explain the
difference in correlative strength between small and large streams
(Mack et al., 2022), showing strong relations between gradients and
effect sizes. While the correlative strength in the (pre)-alpine streams
increased for diatoms using the phosphorus index, no increase (even a
small decrease) was observed for the (pre-)alpine streams for the
pesticide index compared to cropland for macroinvertebrates. A poten-
tial explanation for this may be that agricultural crop types with similar
pesticide application rates are grown in the (pre)-alpine regions, so no
differences are found. In addition, there is a strong correlation between
the pesticide index and cropland of Spearman
ρ
=0.97 in the (pre)--
alpine region. Another explanation may be that (pre-)alpine regions are
home to very pesticide sensitive species, which strongly react to small
amounts of pesticides, so that the crop compositions do not matter too
much. This is also supported by preliminary analysis (results not shown)
using the SPEAR Index as a response measure, which was developed to
identify species at risk of being affected by pesticides (Liess et al., 2008).
The SPEAR Index showed the highest correlative strength of agriculture
for macroinvertebrates in (pre-)alpine streams, compared to other
ecoregions, where the ecological status was stronger correlated to the
land use intensity indices. The overall small correlative strength of
agriculture with the macroinvertebrates´and macrophytes´ecological
status in the lowlands and, to some extend, also in large mountain
streams is likely resulting from truncated gradients with fewer sampling
sites in good ecological status, compared to (pre-)alpine and small
mountain streams (Figures S1-S3) and can be explained by ndings on
the relation between gradient and effect size of Mack et al. (2022).
Hence, subsequent analysis by dividing the agricultural gradient using
recursive partitioning (Zeileis et al., 2008), a method recently used by
Palt et al. (2023) to successfully disentangle the effects of riparian
vegetation in agricultural catchments, showed much larger effects in the
rst part of the gradient with low agricultural intensity (Figures S4-S6).
Reasons for the truncated gradients in the lowland rivers may be the
land use past in the European lowlands, which degraded the rivers now
lacking sensitive species (Feld, 2013). Those river ecosystems may
already be stressed to such a degree that agricultural land use cannot
deteriorate the ecological state much further.
5. Conclusion
The present study clearly shows that the effect of agriculture on river
biota differs with agricultural intensity, environmental conditions, and
organisms. Its results advise against oversimplied conclusions drawn
based on single organism groups. On the contrary, the different organ-
ism groups strongly differ in their sensitivity to agricultural stressors.
While macroinvertebrates respond most strongly to pesticides, diatoms
appear to be more sensitive to nutrients. Macrophyte response is less
clear and likely depends on hydromorphology. Weighting agricultural
land use based on typical pesticide and nutrient application rates can
slightly improve the correlative strength, even though information on
regional practices such as organic farming is lacking on a large scale. The
correlative strength of agriculture is highest in the mountainous eco-
systems and (pre-)alpine regions and the effects are best captured at the
catchment scale, explaining up to 43% of the ecological status. Conse-
quently, the overall clear effect of agriculture on the ecological status
and the strong relation to pesticides for macroinvertebrates and to nu-
trients for diatoms shows that a transition to more sustainable agricul-
tural practices is urgently required to protect river biodiversity.
Declaration of Competing Interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Data availability
Data will be made available on request.
Acknowledgements
This work was nancially supported by a scholarship funding from
the German Federal Environmental Foundation (DBU) to Christian
Schürings, which is gratefully acknowledged. Willem Kaijser and Daniel
Hering were supported by the Collaborative Research Centre 1439
RESIST (Multilevel Response to Stressor Increase and Decrease in
Stream Ecosystems; www.sfb-resist.de) funded by the Deutsche For-
schungsgemeinschaft (DFG, German Research Foundation; CRC 1439/1,
project number: 426547801). We are grateful to the German federal
environmental departments, who provided the biological data and to
Lukas Blickensd¨
orfer et al. for the availability of the land use maps.
Appendix A. Supporting information
Supplementary data associated with this article can be found in the
online version at doi:10.1016/j.agee.2023.108818.
References
Andert, S., Bürger, J., Gerowitt, B., 2015. On-farm pesticide use in four Northern German
regions as inuenced by farm and production conditions. Crop Prot. 75, 1–10.
https://doi.org/10.1016/j.cropro.2015.05.002.
Baczyk, A., Wagner, M., Okruszko, T., Grygoruk, M., 2018. Inuence of technical
maintenance measures on ecological status of agricultural lowland rivers–Systematic
review and implications for river management. Sci. Total Environ. 627, 189–199.
https://doi.org/10.1016/j.scitotenv.2018.01.235.
Berger, E., Haase, P., Oetken, M., Sundermann, A., 2016. Field data reveal low critical
chemical concentrations for river benthic invertebrates. Sci. Total Environ. 544,
864–873. https://doi.org/10.1016/j.scitotenv.2015.12.006.
Bermúdez-Couso, A., Arias-Est´
evez, M., N´
ovoa-Mu˜
noz, J.C., L´
opez-Periago, E., Soto-
Gonz´
alez, B., Simal-G´
andara, J., 2007. Seasonal distributions of fungicides in soils
and sediments of a small river basin partially devoted to vineyards. Water Res. 41
(19), 4515–4525. https://doi.org/10.1016/j.watres.2007.06.029.
Bighiu, M.A., Gottschalk, S., Arrhenius, Å., Goedkoop, W., 2020. Pesticide mixtures cause
short-term, reversible effects on the function of autotrophic periphyton assemblages.
Environ. Toxicol. Chem. 39 (7), 1367–1374. https://doi.org/10.1002/etc.4722.
Blickensd¨
orfer, L., Schwieder, M., Pugmacher, D., Nendel, C., Erasmi, S., Hostert, P.,
2022. Mapping of crop types and crop sequences with combined time series of
Sentinel-1, Sentinel-2 and Landsat 8 data for Germany. Remote Sens. Environ. 269,
112831 https://doi.org/10.1016/j.rse.2021.112831.
B¨
ohmer, J., Rawer-Jost, C., Zenker, A., Meier, C., Feld, C.K., Biss, R., Hering, D., 2004.
Assessing streams in Germany with benthic invertebrates: development of a
multimetric invertebrate based assessment system. Limnologica 34 (4), 416–432.
https://doi.org/10.1016/S0075-9511(04)80010-0.
B¨
orschig, C., Klein, A.M., von Wehrden, H., Krauss, J., 2013. Traits of buttery
communities change from specialist to generalist characteristics with increasing
land-use intensity. Basic Appl. Ecol. 14 (7), 547–554. https://doi.org/10.1016/j.
baae.2013.09.002.
Bothwell, M.L., Kilroy, C., 2011. Phosphorus limitation of the freshwater benthic diatom
Didymosphenia geminata determined by the frequency of dividing cells. Freshw.
Biol. 56 (3), 565–578. https://doi.org/10.1111/j.1365-2427.2010.02524.x.
Britz, W., Witzke, P. 2014: CAPRI Model Documentation 2014. Available online at https
://www.capri-model.org/lib/exe/fetch.php?media=docs:capri_documentation.pdf.
Viewed 02 May 2023.
C. Schürings et al.
Agriculture, Ecosystems and Environment 361 (2024) 108818
8
Cambien, N., Gobeyn, S., Nolivos, I., Forio, M.A.E., Arias-Hidalgo, M., Dominguez-
Granda, L., Goethals, P.L., 2020. Using the soil and water assessment tool to simulate
the pesticide dynamics in the data scarce Guayas river basin, Ecuador. Water 12 (3),
696. https://doi.org/10.3390/w12030696.
CBD. 2022 Kunming-Montreal Global biodiversity framework. UN Environment
Programme, Report No. CBD/COP/15/L.25, 〈https://www.cbd.int/doc/c/e6d3
/cd1d/daf663719a03902a9b116c34/cop-15-l-25-en.pdf〉.
Dachbrodt-Saaydeh, S., Sellmann, J., Strassemeyer, J., Schwarz, J., Klocke, B., Krengel,
S., Kehlenbeck, H. 2021. Netz Vergleichsbetriebe Panzenschutz Jahresbericht 2017
Analyse der Ergebnisse der Jahre 2007-2017, Julius Kühn-Institut, Report No. 210.
Davis, N.G., Hodson, R., Matthaei, C.D., 2022. Long-term variability in deposited ne
sediment and macroinvertebrate communities across different land-use intensities in
a regional set of New Zealand rivers. NZ J. Mar. Freshwat Res. 56 (2), 191–212.
https://doi.org/10.1080/00288330.2021.1884097.
Debenest, T., Silvestre, J., Coste, M., Pinelli, E., 2010. Effects of pesticides on freshwater
diatoms. Rev. Environ. Contam. Toxicol. 87–103. https://doi.org/10.1007/978-1-
4419-1352-4_2.
Dobbie, K.E., Smith, K.A., 2003. Nitrous oxide emission factors for agricultural soils in
Great Britain: the impact of soil water-lled pore space and other controlling
variables. Glob. Change Biol. 9 (2), 204–218. https://doi.org/10.1046/j.1365-
2486.2003.00563.x.
Dodds, W.K., 2006. Eutrophication and trophic state in rivers and streams. Limnol.
Oceanogr. 51 (1part2), 671–680. https://doi.org/10.4319/lo.2006.51.1_part_
2.0671.
Dorioz, J.M., Wang, D., Poulenard, J., Trevisan, D., 2006. The effect of grass buffer strips
on phosphorus dynamics—a critical review and synthesis as a basis for application in
agricultural landscapes in France. Agri. Ecosyst. Environ. 117 (1), 4–21 https://doi.
org/10.1016/j.agee.2006.03.029.
Ekholm, P., Kallio, K., Salo, S., Pietil¨
ainen, O.P., Rekolainen, S., Laine, Y., Joukola, M.,
2000. Relationship between catchment characteristics and nutrient concentrations in
an agricultural river system. Water Res. 34 (15), 3709–3716. https://doi.org/
10.1016/S0043-1354(00)00126-3.
Feld, C.K., 2013. Response of three lotic assemblages to riparian and catchment-scale
land use: implications for designing catchment monitoring programmes. Freshw.
Biol. 58 (4), 715–729. https://doi.org/10.1111/fwb.12077.
Geissen, V., Silva, V., Lwanga, E.H., Beriot, N., Oostindie, K., Bin, Z., Pyne, E., Busink, S.,
Zomer, P., Mol, H., Ritsema, C.J., 2021. Cocktails of pesticide residues in
conventional and organic farming systems in Europe–Legacy of the past and turning
point for the future. Environ. Pollut. 278, 116827 https://doi.org/10.1016/j.
envpol.2021.116827.
Gieswein, A., Hering, D., Lorenz, A.W., 2019. Development and validation of a
macroinvertebrate-based biomonitoring tool to assess ne sediment impact in small
mountain streams. Sci. Total Environ. 652, 1290–1301. https://doi.org/10.1016/j.
scitotenv.2018.10.180.
Giorgio, A., De Bonis, S., Guida, M., 2016. Macroinvertebrate and diatom communities as
indicators for the biological assessment of river Picentino (Campania, Italy). Ecol.
Indic. 64, 85–91. https://doi.org/10.1016/j.ecolind.2015.12.001.
Greenwood, M.J., Harding, J.S., Niyogi, D.K., McIntosh, A.R., 2012. Improving the
effectiveness of riparian management for aquatic invertebrates in a degraded
agricultural landscape: stream size and land-use legacies. J. Appl. Ecol. 49 (1),
213–222. https://doi.org/10.1111/j.1365-2664.2011.02092.x.
Haase, P., Lohse, S., Pauls, S., Schindehütte, K., Sundermann, A., Rolauffs, P., Hering, D.,
2004. Assessing streams in Germany with benthic invertebrates: development of a
practical standardised protocol for macroinvertebrate sampling and sorting.
Limnologica 34 (4), 349–365. https://doi.org/10.1016/S0075-9511(04)80005-7.
Halbach, K., M¨
oder, M., Schrader, S., Liebmann, L., Sch¨
afer, R.B., Schneeweiss, A.,
Reemtsma, T., 2021. Small streams–large concentrations? Pesticide monitoring in
small agricultural streams in Germany during dry weather and rainfall. Water Res.
203, 117535 https://doi.org/10.1016/j.watres.2021.117535.
Hilton, J., O’Hare, M., Bowes, M.J., Jones, J.I., 2006. How green is my river? A new
paradigm of eutrophication in rivers. Sci. Total Environ. 365 (1-3), 66–83. https://
doi.org/10.1016/j.scitotenv.2006.02.055.
Janova, E., Heroldova, M., 2016. Response of small mammals to variable agricultural
landscapes in central Europe. Mamm. Biol. 81 (5), 488–493. https://doi.org/
10.1016/j.mambio.2016.06.004.
Jones, J.I., Douthwright, T.A., Arnold, A., Duerdoth, C.P., Murphy, J.F., Edwards, F.K.,
Pretty, J.L., 2017. Diatoms as indicators of ne sediment stress. Ecohydrology 10 (5),
e1832. https://doi.org/10.1002/eco.1832.
Kaijser, W., Hering, D., Lorenz, A.W., 2022. Reach hydromorphology: a crucial
environmental variable for the occurrence of riverine macrophytes. Hydrobiologia
849 (19), 4273–4285. https://doi.org/10.1007/s10750-022-04983-w.
Kelly, M., Bennett, C., Coste, M., Delgado, C., Delmas, F., Denys, L., Vilbaste, S., 2009.
A comparison of national approaches to setting ecological status boundaries in
phytobenthos assessment for the European Water Framework Directive: results of an
intercalibration exercise. Hydrobiologia 621, 169–182. https://doi.org/10.1007/
s10750-008-9641-4.
King, R.S., Brain, R.A., Back, J.A., Becker, C., Wright, M.V., Toteu Djomte, V.,
Chambliss, C.K., 2016. Effects of pulsed atrazine exposures on autotrophic
community structure, biomass, and production in eld-based stream mesocosms.
Environ. Toxicol. Chem. 35 (3), 660–675. https://doi.org/10.1002/etc.3213.
Leps, M., Tonkin, J.D., Dahm, V., Haase, P., Sundermann, A., 2015. Disentangling
environmental drivers of benthic invertebrate assemblages: the role of spatial scale
and riverscape heterogeneity in a multiple stressor environment. Sci. Total Environ.
536, 546–556. https://doi.org/10.1016/j.scitotenv.2015.07.083.
Li, S., Yang, W., Wang, L., Chen, K., Xu, S., Wang, B., 2018. Inuences of environmental
factors on macroinvertebrate assemblages: differences between mountain and
lowland ecoregions, Wei River, China. Environ. Monit. Assess. 190, 1–13. https://
doi.org/10.1007/s10661-018-6516-7.
Liess, M., Liebmann, L., Vormeier, P., Weisner, O., Altenburger, R., Borchardt, D.,
Brack, W., Chatzinotas, A., Escher, B., Foit, K., Gunold, R., Henz, S., Hitzfeld, K.L.,
Schmitt-Jansen, M., Kamjunke, N., Kaske, O., Knillmann, S., Krauss, M., Küster, E.,
Link, M., Lück, M., M¨
oder, M., Müller, A., Paschke, A., Sch¨
afer, R.B.,
Schneeweiss, A., Schreiner, A.C., Schulze, T., Schüürmann, G., Tümpling, Wv,
Weitere, M., Wogram, J., Reemtsma, T., 2021. Pesticides are the dominant stressors
for vulnerable insects in lowland streams. Water Res. 201, 117262 https://doi.org/
10.1016/j.watres.2021.117262.
Liess, M., Sch¨
afer, R.B., Schriever, C.A., 2008. The footprint of pesticide stress in
communities—species traits reveal community effects of toxicants. Sci. Total
Environ. 406 (3), 484–490. https://doi.org/10.1016/j.scitotenv.2008.05.054.
Mack, L., de la Hoz, C.F., Penk, M., Piggott, J., Crowe, T., Hering, D., Kaijser, W.,
Aroviita, J., Baer, J., Borja, A., Clark, D.E., Fern´
andez-Torquemada, Y., Kotta, J.,
Matthai, C.D., O´Beirn, F., Paerl, H.W., Sokolowski, A., Vilmi, A., Birk, S., 2022.
Perceived multiple stressor effects depend on sample size and stressor gradient
length. Water Res. 226, 119260 https://doi.org/10.1016/j.watres.2022.119260.
Mohr, S., Berghahn, R., Feibicke, M., Meinecke, S., Ottenstr¨
oer, T., Schmiedling, I.,
Schmidt, R., 2007. Effects of the herbicide metazachlor on macrophytes and
ecosystem function in freshwater pond and stream mesocosms. Aquat. Toxicol. 82
(2), 73–84. https://doi.org/10.1016/j.aquatox.2007.02.001.
Morrissey, C.A., Mineau, P., Devries, J.H., Sanchez-Bayo, F., Liess, M., Cavallaro, M.C.,
Liber, K., 2015. Neonicotinoid contamination of global surface waters and associated
risk to aquatic invertebrates: a review. Environ. Int. 74, 291–303. https://doi.org/
10.1016/j.envint.2014.10.024.
Mouri, G., Aisaki, N., 2015. Using land-use management policies to reduce the
environmental impacts of livestock farming. Ecol. Complex. 22, 169–177. https://
doi.org/10.1016/j.ecocom.2015.03.003.
Niyogi, D.K., Koren, M., Arbuckle, C.J., Townsend, C.R., 2007. Stream communities
along a catchment land-use gradient: subsidy-stress responses to pastoral
development. Environ. Manag. 39, 213–225. https://doi.org/10.1007/s00267-005-
0310-3.
O’Hare, M.T., Baattrup-Pedersen, A., Baumgarte, I., Freeman, A., Gunn, I.D., L´
az´
ar, A.N.,
Sinclair, R., Wade, A.J., Bowes, M.J., 2018. Responses of aquatic plants to
eutrophication in rivers: a revised conceptual model. Front. Plant Sci. 9, 451 https://
doi.org/10.3389/fpls.2018.00451.
Oelofse, M., Høgh-Jensen, H., Abreu, L.S., Almeida, G.F., El-Araby, A., Hui, Q.Y., de
Neergaard, A., 2010. A comparative study of farm nutrient budgets and nutrient
ows of certied organic and non-organic farms in China, Brazil and Egypt. Nutr.
Cycl. Agroecosystems 87, 455–470. https://doi.org/10.1007/s10705-010-9351-y.
Palt, M., Hering, D., Kail, J., 2023. Context-specic positive effects of woody riparian
vegetation on aquatic invertebrates in rural and urban landscapes. J. Appl. Ecol.
https://doi.org/10.1111/1365-2664.14386.
Parkyn, S., 2004. Review of riparian buffer zone effectiveness, Vol. 2005. Ministry of
Agriculture and Forestry,, Wellington, New Zealand.
Pottgiesser, T., Sommerh¨
auser, M., 2008. Erste Überarbeitung der Steckbriefe der deut-
schen Fließgew¨
assertypen. 〈http://www.wasserblick.net/servlet/is/18727/〉.
Ribeiro, V.H.V., Alencar, B.T.B., Dos Santos, N.M.C., da Costa, V.A.M., Dos Santos, J.B.,
Francino, D.M.T., Silva, D.V., 2019. Sensitivity of the macrophytes Pistia stratiotes
and Eichhornia crassipes to hexazinone and dissipation of this pesticide in aquatic
ecosystems. Ecotoxicol. Environ. Saf. 168, 177–183. https://doi.org/10.1016/j.
ecoenv.2018.10.021.
Riedo, J., Herzog, C., Banerjee, S., Fenner, K., Walder, F., Van der Heijden, M.G.,
Bucheli, T.D., 2022. Concerted evaluation of pesticides in soils of extensive grassland
sites and organic and conventional vegetable elds facilitates the identication of
major input processes. Environ. Sci. Technol. 56 (19), 13686–13695. https://doi.
org/10.1021/acs.est.2c02413.
Rigal, S., Dakos, V., Alonso, H., Aunin¸ˇ
s, A., Benk˝
o, Z., Brotons, L., Devictor, V., 2023.
Farmland practices are driving bird population decline across Europe. Proc. Natl.
Acad. Sci. 120 (21), e2216573120 https://doi.org/10.1073/pnas.2216573120.
Roessink, I., Merga, L.B., Zweers, H.J., Van den Brink, P.J., 2013. The neonicotinoid
imidacloprid shows high chronic toxicity to mayy nymphs. Environ. Toxicol. Chem.
32 (5), 1096–1100. https://doi.org/10.1002/etc.2201.
Schaumburg, J., Schranz, C., Stelzer, D., Vogel, A., Gutowski, A. 2012.
Verfahrensanleitung für ¨
okologische Bewertung von Fließgew¨
assern zur Umsetzung
der EU-Wasserrahmenrichtlinie: Makrophyten und Phytobenthos. Bayrisches
Landesamt für Umwelt. 〈https://gewaesser-bewertung.de/les/verfahrensanleitung
_fg.pdf〉. Viewed 02 May 2023.
Schneeweiss, A., Schreiner, V.C., Reemtsma, T., Liess, M., Sch¨
afer, R.B., 2022. Potential
propagation of agricultural pesticide exposure and effects to upstream sections in a
biosphere reserve. Sci. Total Environ. 836, 155688 https://doi.org/10.1016/j.
scitotenv.2022.155688.
Schulz, R., 2001. Rainfall-induced sediment and pesticide input from orchards into the
Lourens River, Western Cape, South Africa: importance of a single event. Water Res.
35 (8), 1869–1876. https://doi.org/10.1016/S0043-1354(00)00458-9.
Schürings, C., Feld, C.K., Kail, J., Hering, D., 2022. Effects of agricultural land use on
river biota: a meta-analysis. Environ. Sci. 34, 124 https://doi.org/10.1186/s12302-
022-00706-z.
Secchi, S., Gassman, P.W., Jha, M., Kurkalova, L., Kling, C.L., 2011. Potential water
quality changes due to corn expansion in the Upper Mississippi River Basin. Ecol.
Appl. 21 (4), 1068–1084. https://doi.org/10.1890/09-0619.1.
Townsend, C.R., Downes, B.J., Peacock, K., Arbuckle, C.J., 2004. Scale and the detection
of land-use effects on morphology, vegetation and macroinvertebrate communities
of grassland streams. Freshw. Biol. 49 (4), 448–462. https://doi.org/10.1111/
j.1365-2427.2004.01192.x.
C. Schürings et al.
Agriculture, Ecosystems and Environment 361 (2024) 108818
9
Turunen, J., Muotka, T., Vuori, K.M., Karjalainen, S.M., R¨
a¨
apysj¨
arvi, J., Sutela, T.,
Aroviita, J., 2016. Disentangling the responses of boreal stream assemblages to low
stressor levels of diffuse pollution and altered channel morphology. Sci. Total
Environ. 544, 954–962. https://doi.org/10.1016/j.scitotenv.2015.12.031.
UBA. 2022 Die Wasserrahmenrichtlinie Gew¨
asser in Deutschland 2021 Fortschritte und
Herausforderungen, German Environmental Agency, 〈https://www.umweltbundesa
mt.de/sites/default/les/medien/1410/publikationen/221010_uba_fb_wasserri
chtlinie_bf.pdf〉.
Vannote, R.L., Minshall, G.W., Cummins, K.W., Sedell, J.R., Cushing, C.E., 1980. The
river continuum concept. Can. J. Fish. Aquat. Sci. 37 (1), 130–137. https://doi.org/
10.1139/f80-017.
V¨
or¨
osmarty, C.J., McIntyre, P.B., Gessner, M.O., Dudgeon, D., Prusevich, A., Green, P.,
Glidden, S., Bunn, S.E., Sullivan, C.A., Liemann, C.R., Davies, P.M., 2010. Global
threats to human water security and river biodiversity. Nature 467 (7315), 555–561.
https://doi.org/10.1038/nature09440.
Weijters, M.J., Janse, J.H., Alkemade, R., Verhoeven, J.T.A., 2009. Quantifying the effect
of catchment land use and water nutrient concentrations on freshwater river and
stream biodiversity. Aquat. Conserv.: Mar. Freshw. Ecosyst. 19 (1), 104–112.
https://doi.org/10.1002/aqc.989.
Wernecke, A., Frommberger, M., Forster, R., Pistorius, J., 2019. Lethal effects of various
tank mixtures including insecticides, fungicides and fertilizers on honey bees under
laboratory, semi-eld and eld conditions. J. Consum. Prot. Food Saf. 14 (3),
239–249. https://doi.org/10.1007/s00003-019-01233-5.
Wieczorek, M.V., Bakanov, N., Lagadic, L., Bruns, E., Schulz, R., 2017. Response and
recovery of the macrophytes Elodea canadensis and Myriophyllum spicatum
following a pulse exposure to the herbicide iofensulfuron-sodium in outdoor stream
mesocosms. Environ. Toxicol. Chem. 36 (4), 1090–1100. https://doi.org/10.1002/
etc.3636.
Wolfram, J., Stehle, S., Bub, S., Petschick, L.L., Schulz, R., 2021. Water quality and
ecological risks in European surface waters–Monitoring improves while water
quality decreases. Environ. Int. 152, 106479 https://doi.org/10.1016/j.
envint.2021.106479.
Wurzel, S., Ford, M.A., Dority, D., Tronstad, L., 2020. Evaluating the impact of
Permethrin on non-target invertebrates in an urban stream. Hydrobiologia 847,
91–104. https://doi.org/10.1007/s10750-019-04074-3.
Zeileis, A., Hothorn, T., Hornik, K., 2008. Model-based recursive partitioning. J. Comput.
Graph. Stat. 17 (2), 492–514. https://doi.org/10.1198/106186008×319331.
van Zuidam, J.P., Peeters, E.T., 2013. Occurrence of macrophyte monocultures in
drainage ditches relates to phosphorus in both sediment and water. SpringerPlus 2
(1), 11. https://doi.org/10.1186/2193-1801-2-564.
C. Schürings et al.
