Impact assessment of agricultural driven stressors on benthic macroinvertebrates using simulated data

STOTEN-18205; No of Pages 11 Science of the Total Environment xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Impact assessment of agricultural driven stressors on benthic macroinvertebrates using simulated data K. Stefanidis ⁎, Y. Panagopoulos, M. Mimikou Laboratory of Hydrology and Water Resources Management, Department of Water Resources and Environmental Engineering, School of Civil Engineering, National Technical University of Athens (NTUA), 5, Iroon Politechniou Street, 15780 Zografou, Athens, Greece

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• We linked a biotic metric (ASPT) with the output of a process based model • We investigated the ASPT response to the application of three management scenarios • Reduced fertilization improved the nitrogen concentration and ASPT values • Reduced irrigation did not have any effect on water quality but saved water • The combination of both scenarios would be the best management practice in our case

a r t i c l e

i n f o

Article history: Received 30 April 2015 Received in revised form 4 August 2015 Accepted 4 August 2015 Available online xxxx Editor: D. Barcelo Keywords: ASPT Agricultural land use Management practices Process-based model Water quality Rivers Mediterranean

a b s t r a c t Agricultural land use poses a significant threat to the ecological integrity of rivers in Europe. Particularly in the Mediterranean, water abstraction and nutrient application are anthropogenic pressures that have a significant impact on aquatic habitats and biodiversity. In this article, we assessed the effects of agricultural management practices on benthic macroinvertebrates in a large river basin of central Greece using simulated data based on the application of SWAT (Soil Water Assessment Tool) model. Physicochemical and hydrological output variables of the model were used as predictors of the ASPT (Average Score Per Taxon) metric based on a correlated component regression analysis (CCR) built on empirical data. The estimation of ASPT was performed for the wet and dry seasons within a 20-year period for a total of 47 subbasins under the baseline conditions and after implementing three management scenarios that reduced: a) irrigation water applied to crops by 30%, b) chemical fertilization applied to crops by 30% and c) both irrigation and fertilization by 30%. The results revealed that application of the reduced irrigation resulted to a slight increase of the simulated dissolved inorganic nitrogen concentration (DIN), which in turn decreased the mean ASPT in 21 of the 47 subbasins implying a negative effect on the macroinvertebrate communities. On the contrary, the reduction of fertilization as well as the combined scenario decreased both the simulated DIN and phosphate concentration causing an increase of the mean ASPT for a total of 40 of the 47 subbasins. Based on these results, we suggest that the best management

⁎ Corresponding author. E-mail address: [email protected] (K. Stefanidis).

http://dx.doi.org/10.1016/j.scitotenv.2015.08.015 0048-9697/© 2015 Elsevier B.V. All rights reserved.

Please cite this article as: Stefanidis, K., et al., Impact assessment of agricultural driven stressors on benthic macroinvertebrates using simulated data, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.08.015

2

K. Stefanidis et al. / Science of the Total Environment xxx (2015) xxx–xxx

option is a combined practice of deficit irrigation and fertilization reduction since it improved water quality, increased ASPT values and saved a considerable amount of water. Overall, this work demonstrates a simple methodology that can efficiently assess the effects of agricultural management practices on biotic indicators. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Streams and rivers are among the most threatened ecosystems in Europe due to the combined effects of multiple pressures related to anthropogenic activities (Schinegger et al., 2012). Nutrient pollution, habitat fragmentation, bank erosion and alteration in hydrology are some of the most common problems that riverine systems are facing today (Milly et al., 2005; Ormerod et al., 2010; Vörösmarty et al., 2010). Particularly in the Mediterranean region, changes in hydromorphology along with increased nutrient loadings are known to heavily influence both the ecological integrity and the aquatic biodiversity of rivers (Gasith and Resh, 1999; Bonada and Resh, 2013; Hershkovitz and Gasith, 2013). Human activities such as water abstraction, flow regulation, water diversion and nutrient application pose a significant threat to the functionality and structure of river ecosystems. To prevent further deterioration and improve the ecological quality of water bodies effective mitigation and restoration actions are needed (Aguiar and Ferreira, 2005; Hooke, 2006). Implementing such actions is mandated by the Water Framework Directive (WFD) through the development of specific river basin management plans (RBMPs) for each River Basin District (WFD CIS, 2003). However, the fulfillment of such actions requires first to understand how the aquatic biota, and specifically the biological quality elements (BQEs, fish, macroinvertebrates, benthic diatoms and aquatic macrophytes) according to the WFD, respond to anthropogenic pressures and to further explore the interactions between ecosystem traits under various multiple pressures. Predictive models offer the potentiality to forecast changes in the ecology of aquatic ecosystems caused by numerous anthropogenic activities and as such they can act as helpful and effective tools in modern ecology for assessing and averting the impact of harmful effects (Carlisle and Meador, 2007; Jähnig et al., 2012; Maloney et al., 2012). However, modeling the complex interactions between biotic and abiotic elements such as hydrological and hydraulic traits is not an easy task and there is still a lack of models that can fully describe the links between the aquatic organisms and their environment (Einheuser et al., 2013). Several studies have proposed the use of habitat models that can link the local hydraulic conditions and the preference of organisms to these conditions. These models are used to predict the effects of pressures, e.g. changes in discharge, on the aquatic hydraulic habitat and therefore estimating the impact on the aquatic organisms. However, the majority of these models was mostly developed for using fish as a target organism group and therefore may not be accurate and effective for other organisms, namely the benthic macroinvertebrates. More recently, species distribution models (SDMs) have been applied in freshwater ecosystems in order to identify the responses of aquatic organisms under the pressures derived by the climate change. Most applications of these models at freshwater ecosystems are broad scale studies that do not address the implementation of management measures at local scale (Kuemmerlen et al., 2014). Moreover, the effectiveness of SDMs highly depends on the detail and quality of the data required such as the occurrence and abundance of specific species. An alternative approach that has been proposed to study these complex interactions is the use of simple regression models that can be easily developed and applied in cases where data are limited. There are numerous studies in the field of river ecology that have used effectively various regression techniques such as multiple linear regression analysis, general additive models, multivariate approaches, regression tree analysis, artificial neural networks and fuzzy logic methods (Clarke et al., 2003; Buffagni et al., 2006; Carlisle and Meador, 2007; Gabriels

et al., 2007; Weigel and Robertson, 2007; Pool et al., 2010; Marchini, 2011). More recently, Einheuser et al. (2013) applied various regression methods, including stepwise linear regression, partial least squares and fuzzy logic, in order to develop predictive models that estimate the biological integrity of a large river basin using as descriptors output variables of the Soil and Water Assessment Tool (SWAT) model (Arnold et al., 1998). Based on this approach, the purpose of this article is to develop robust regression formulae to be used as user-friendly statistical models, to calculate benthic macroinvertebrate metrics from in-stream hydrological and physicochemical data. Our approach is based on an integrated process-based modeling with the extensively applied SWAT river basin model (Gassman et al., 2014), and uses the physicochemical output variables of the model as river descriptors for predicting the responses of macroinvertebrate variables. We decided to use macroinvertebrate metrics as target response variables for two reasons. Benthic invertebrates are considered an ideal group of organisms for assessing the ecological quality of rivers because of their sensitivity to changes in properties of water quality and aquatic habitat (Feld and Hering, 2007; Skoulikidis et al., 2009). Combined with the fact that their sampling and identification methodology is relatively easy and cost effective (Clarke et al., 2003), benthic macroinvertebrates are commonly used as bioindicators to detect the impact of human pressures on river ecosystems. The other reason is practical and has to do with the data availability and the fact that we have at our disposal a solid empirical dataset provided by the Greek Ministry of Environment and previous published studies that have been carried out in the area of interest (Chatzinikolaou, 2007; Chatzinikolaou et al., 2010). The methodology is developed in a Greek agricultural catchment, where water scarcity and nitrate pollution are the major environmental threats, and the proposed approach is demonstrated both under the baseline conditions and the implementation of alternative management practices, with the purpose to provide predictions of benthic macroinvertebrate metrics under various management conditions. 2. Methods 2.1. Study area and process-based model setup The Pinios basin covers almost entirely the River Basin District (RBD) of Thessaly in Central Greece (see Fig. 1). The 10,600 km2 basin is an important agricultural area in Greece, with feed (corn, alfalfa) and industrial crop production (cotton). However, the crop growth periods are the dry periods (May to September) of the year and are accompanied with high temperatures, high evapotranspiration rates and dry soils (Vasiliades et al., 2011). Inevitably, these conditions result in irrigation cutbacks and overexploitation of both surface and groundwater resources with significant impacts on the natural water cycle and water availability, especially in the driest years. Intense agriculture and overfertilization in the region is also considered responsible for surface and groundwater quality degradation mainly in terms of nutrient concentration in waters (Bellos and Sawidis, 2005; Chatzinikolaou et al., 2010; Loukas, 2010). Recent environmental analyses in the area have shown that the likelihood of the Pinios basin's water bodies failing to achieve good ecological status is generally high (Ioannou et al., 2009). In order to discover favorable mitigation options that address these important environmental issues, a detailed Pinios basin modeling system has been constructed and tested with the use of SWAT 2009 (version 488) and ArcSWAT interface (Panagopoulos et al., 2012;

Please cite this article as: Stefanidis, K., et al., Impact assessment of agricultural driven stressors on benthic macroinvertebrates using simulated data, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.08.015

K. Stefanidis et al. / Science of the Total Environment xxx (2015) xxx–xxx

3

Fig. 1. The Pinios river basin in Central Greece. Key data layers are shown including, among others, subbasins as delineated in SWAT, irrigated areas and source of irrigation water as well as sample sites along the main river.

Panagopoulos et al., 2014a; Panagopoulos et al., 2014b). Pinios basin has been divided in 49 subbasins as shown in Fig. 1 and topography, crop allocation and soil features have been represented by appropriate layers leading to a schematization of 361 Hydrologic Response Units (HRUs), which represent land pieces with unique combinations of land use, management, slope and soil characteristics. As a distributed GIS model, SWAT allows the division of complex agricultural landscapes into homogenous land parcels where physical, biogeochemical and hydrochemical processes are simulated (Neitsch et al., 2009). Simulation of the hydrology is separated into the land and the routing phase of the hydrological cycle including surface runoff/infiltration, evapotranspiration, lateral flow, percolation, and return flow. Sediment yields are estimated with the Modified Universal Soil Loss Equation (MUSLE; Neitsch et al., 2009), while both nitrogen (N) and phosphorus (P) are divided in the soil into two parts, each associated with organic and inorganic N and P transport and transformations. Agricultural management practices such as planting, harvesting, tillage, irrigation, nutrient applications and other in-field conservation practices can be simulated with specific dates and by explicitly defining the appropriate management parameters for each HRU (Neitsch et al., 2009). The baseline situation with respect to water and crop management in the basin has been accurately represented with SWAT considering groundwater, reservoir and crop management practices representation (Fig. 1). The model has been successfully calibrated and validated based on monthly river flows for a long period with available data, while crop yield predictions have also been compared with measured ones on a mean annual basis to ensure that SWAT produces reasonable estimates (Panagopoulos et al., 2014b). As far as water quality calibration is concerned, simulated river loads have been compared with NO3–N observations and the correlation was successful for several river sites within the basin achieving high R2 values. Mean annual nitrate (NO3–N) concentrations ranged between 0.7–1.2 mg/L with the mean annual NO3–N losses from the total area of the catchment being 2 kg/ha and from cotton fields nearly 10 kg/ha (Panagopoulos et al., 2012). A detailed presentation of the model's hydrologic and water quality

calibration is beyond the scope of this article as these details have been already published (Panagopoulos et al., 2012; Panagopoulos et al., 2014a; Panagopoulos et al., 2014b) and can be easily found by those who are interested. The present study builds on the existing Pinios model setup and calibration and uses the simulated river flows, NO3–N and P concentrations at 47 of the 49 subbasin outlets to assess a biological indicator, crucial for the assessment of the ecological status of water. Outlets of subbasins No 48 and 49 (see Fig. 1) are excluded as river flows are zero there due to water works upstream that collect water throughout the year. 2.2. Field data used for the calculation of biological metrics For the purposes of our study we used a large dataset of macroinvertebrate community data comprised by data obtained from two different sources. The first source is based on previously published data collected from 80 sites along the stretch of river Pinios during autumn of 2002 (Chatzinikolaou, 2007; Chatzinikolaou et al., 2010). The second source is the Greek National Monitoring program which provided us with data collected from 30 sites in summer of 2012 and 32 sites in spring of 2013 (Fig. 1). Therefore the merged dataset contains semiquantitative community information at family level from a total of 142 samples collected from 101 different sites across the catchment of river Pinios. Eleven sites were common among the three different sampling periods and 30 were common between 2012 and 2013. This dataset also contains information for environmental data collected from the same sampling sites. These data include nutrient concentration in water (phosphate, nitrate, nitrite and ammonium) as well as BOD, surface dissolved oxygen (DO), saturation %, pH, conductivity, total suspended solids (TSS) and water discharge. The laboratory analysis was performed according to the APHA standard methods for the examination of water and wastewater as described in Chatzinikolaou et al. (2010). Based on the empirical data collected during the aforementioned samplings we calculated candidate metrics that we subsequently used

Please cite this article as: Stefanidis, K., et al., Impact assessment of agricultural driven stressors on benthic macroinvertebrates using simulated data, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.08.015

4

K. Stefanidis et al. / Science of the Total Environment xxx (2015) xxx–xxx

Table 1 Descriptive statistics for the data used in our analysis. Variable

Mean

Std. error Median

BMWP score 50.51 2.90 ASPT 4.65 0.10 EPT 3.84 0.27 1-GOLD 0.74 0.03 Log10(SelEPTD + 1) 0.39 0.07 Number of Taxa 10.39 0.47 Shannon diversity index 1.30 0.07 Evenness index 0.61 0.03 PO4-P (mg/L) 0.093 0.020 NH4–N (μg/L) 69.98 9.99 NO2–N (μg/L) 36.29 4.46 NO3–N (mg/L) 1.728 0.195 DIN (mg/L) 1.76 0.194 BOD (mg/L) 2.427 0.207 DO (mg/L) 8.41 0.18 Saturation (%) 89.89 2.06 Water Temperature 18.68 0.38 pH 7.90 0.04 Conductivity (mS/cm) 0.53 0.03 TSS (mg/L) 268.93 18.16 3 Discharge (m /s) 3.28 0.28

43.00 4.53 3.00 0.86 0.00 9.00 1.46 0.67 0.042 29.71 19.06 1.158 1.157 1.85 8.63 91.30 18.50 7.94 0.48 239.67 2.42

Min

Max

N

3.00 194.00 1.00 7.44 0.00 11.00 0.01 1.00 0.00 2.03 1.00 35.00 0.00 2.40 0.00 0.97 0.000 2.412 0.00 793.00 0.00 350 0.00 17.082 0.00 17.167 0.11 6.79 1.64 14.39 17.70 191.00 8.30 34.20 6.27 9.47 0.22 5.01 110.00 2656.00 0.01 8.68

142 142 142 80 80 142 80 80 142 120 128 140 140 80 142 142 142 142 142 142 102

Note: PO4-P: phosphate, NH4–N: ammonium, NO2-N: nitrite, NO3–N: nitrate, DIN: dissolved inorganic nitrogen, BOD: biochemical oxygen demand, DO: dissolved oxygen, TSS: total suspended solids.

as response variables to nutrient and hydrological parameters predicted by the hydrological model SWAT. Specifically, from the available data we calculated the BMWP (Biological Monitoring Working Party) and the ASPT (Average Score Per Taxon) score according to the BMWP system (Armitage et al., 1983). Most macroinvertebrate families have been assigned a score ranging from 1 to 10 depending to their perceived tolerance to organic pollution. The BMWP is the sum of the scores of the families present in the sample while ASPT is the average score. As such the BMWP score increases with sampling effort and is more susceptible to sampling efficiency whereas ASPT is independent of the sampling effort. Depending on the data availability we calculated additional metrics such as the number of Ephemeroptera, Plecoptera and Trichoptera families (EPT), the relative abundance of Gastropoda, Oligochaeta and Diptera families subtracted from 1 (1-GOLD), the Log transformed abundance of selected families of Ephemeroptera, Plecoptera, Trichoptera and Diptera (Log10(SelEPTD + 1), the total number of taxa, the

Shannon diversity index and the species evenness index. Summary statistics for the candidate metrics and the environmental parameters is presented in Table 1. 2.3. Data analysis The next step in our methodological approach was to run Spearman rank correlations to identify statistically significant relations between environmental variables and the candidate biological metrics. The results suggested several significant relations between biotic metrics such as ASPT, EPT, BMWP, and abiotic variables such as DIN, phosphate and discharge (Table 2). Specifically ASPT correlated with the most abiotic variables and presented the strongest linear regressions with DIN and phosphate concentration (Fig. 2). On the contrary BMWP and EPT had strong linear relationships only with discharge and weaker linear relationships with DIN and phosphate (Fig. 2). Based on these results we selected ASPT as the most suitable metric to develop a predictive model for the purposes of our analysis. For the development of predictive models we used correlated component regression (CCR), a new regression method (Magidson, 2013), which provides reliable predictions of a dependent variable even with collinear or near multicollinear data. CCR utilizes K correlated components, each a linear combination of the predictors, to predict an outcome variable. The analysis was performed using a step-down algorithm to eliminate the weak predictors. CCR was run using the XLSTAT software. The residuals did not violate assumptions of homoscedasticity and normality. In order to select the best suitable model, the results from each model were cross-validated using random subsets of data from the initial dataset. After performing multiple stepwise regression analyses the best model identified dissolved inorganic nitrogen (DIN), phosphate and discharge explaining the 31% (R2 = 0.31) of the total variance (Fig. 3). The equation that describes the model is given below: ASPT ¼ 6:262−0:468  LogDIN−0:18  LogPhosphate−0:11  LogDischarge:

2.4. Selection of scenarios/agricultural management practices To reduce high exploitation of water resources in dry regions deficit irrigation is considered to be a valuable and sustainable strategy (Geerts

Table 2 Spearman rank correlations among the biological metrics and selected environmental descriptors. ASPT BMWP ASPT No taxa EPT 1-GOLD BOD PO4-P Discharge DIN Saturation

Cor. coeff. Sig. Cor. coeff. Sig. Cor. coeff. Sig. Cor. coeff. Sig. Cor. coeff. Sig. Cor. coeff. Sig. Cor. coeff. sig. Cor. coeff. Sig. Cor. coeff. Sig. Cor. coeff. Sig.

.611⁎⁎ .000 1.000

No taxa .924⁎⁎ .000 .311⁎⁎ .000 1.000

EPT .856⁎⁎ .000 .617⁎⁎ .000 .732⁎⁎ .000 1.000

1-GOLD

BOD

PO4-P

Discharge

DIN

.175 .121 .575⁎⁎ .000 −.041 .719 .349⁎⁎ .001 1.000

−.140 .214 −.188 .095 −.077 .499 −.214 .057 −.168 .137 1.000

−.119 .158 −.414⁎⁎ .000 .028 .737 −.062 .462 −.298⁎⁎

−.332⁎⁎ .001 −.288⁎⁎

−.260⁎⁎ .002 −.408⁎⁎

.413⁎⁎ .000 .366⁎⁎

.438⁎⁎ .000 .447⁎⁎

.003 −.298⁎⁎ .002 −.326⁎⁎ .001 −.101 .375 .595⁎⁎

.000 −.157 .059 −.328⁎⁎ .000 −.416⁎⁎

.000 .328⁎⁎ .000 .525⁎⁎ .000 .489⁎⁎

.000 .334⁎⁎ .000 .526⁎⁎ .000 .429⁎⁎

.000 .504⁎⁎ .000 .409⁎⁎ .000 .469⁎⁎

.000 −.287⁎⁎ .010 −.136 .109 −.393⁎⁎

.000 −.324⁎⁎ .003 −.312⁎⁎ .000 −.458⁎⁎

.000 −.457⁎⁎ .000 1.000

.000 −.513⁎⁎ .000 .897⁎⁎ .000

.007 .505⁎⁎ .000 1.000

.000 .297⁎⁎ .003 1.000

.000 1.000

Saturation

DO

BMWP: Biological Monitoring Working Party score, ASPT: Average Score Per Taxon, EPT: number of Ephemeroptera, Plecoptera, Trichoptera families, 1-GOLD: 1- (relative abundance of Gastropoda, Oligochaeta and Diptera families), PO4-P: phosphate, DIN: Dissolved Inorganic Nitrogen. ⁎ Indicates p value ≤ 0.05. ⁎⁎ Indicates p value ≤ 0.001.

Please cite this article as: Stefanidis, K., et al., Impact assessment of agricultural driven stressors on benthic macroinvertebrates using simulated data, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.08.015

K. Stefanidis et al. / Science of the Total Environment xxx (2015) xxx–xxx

5

Fig. 2. Linear regression plots between macroinvertebrate metrics (BMWP, EPT, 1-GOLD, ASPT) and environmental variables (DIN, phosphate, discharge).

and Raes, 2009). Thus, as the first scenario in this study we applied a 30% reduction of irrigation water throughout the crop growth cycle, with all irrigation doses being uniformly reduced. It should be noted that this practice does not always result in a net 30% abstraction reduction. This level of reduction occurs only on those HRUs with significant water availability. In areas with no adequate water on the source, the level of reduction is inevitably smaller. The practice results in more water stress days for the crop but also in reduced runoff losses from the HRU as the reduced soil moisture contents cause lower Curve Number values in

SWAT, which are updated daily by the model, causing less surface runoff generation (Neitsch et al., 2009). Reduction of chemical fertilization to crops is the second practice representing the second extensive management scenario tested. This was implemented by reducing chemical fertilizers by 30% to all the four crops cultivated in the basin (the three irrigated crops mentioned above and wheat, which is growing in the winter period). Fertilization reduction is a common practice to limit N and P losses from land to waters as it reduces available quantities on the ground. In SWAT total fertilizations of N and P are reduced by changing the application rates. The third and final scenario tested in this paper is the combination of the previous two thus including a simultaneous reduction of irrigation water and fertilizers by 30% in the cropland of the entire basin. Reduction in crop yields is certainly expected when each one or both of these practices are implemented. However, it is beyond the scope of this paper to analyse economic or socio-economic consequences due to the implementation of these practices/scenarios in Pinios. What is of interest in the present paper is to feed the regression model with SWAT physicochemical outputs and demonstrate a useful connection with biological indicators. Irrigation and fertilization reduction are expected to influence water runoff and river flows in the basin as well as the amount of nutrients and other pollutants traveling along the rivers, and SWAT is used as the estimator of these changes at multiple locations. 2.5. Implementation of scenario analysis

Fig. 3. Scatter plot of observed vs predicted values of ASPT.

In order to assess the effects of applied management practices, the SWAT model was executed four times (baseline conditions and three management practices) and simulated the mean monthly values of DIN and phosphate in river concentration as well as discharge for a period of 20 years (from 1991 to 2010) for a total of 47 sites, each representing a different subbasin of river Pinios (see Fig. 1). Based on

Please cite this article as: Stefanidis, K., et al., Impact assessment of agricultural driven stressors on benthic macroinvertebrates using simulated data, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.08.015

6

K. Stefanidis et al. / Science of the Total Environment xxx (2015) xxx–xxx

Fig. 4. Mean annual DIN concentration at the 47 subbasin outlets of Pinios under the baseline conditions and under the simulation of the three management practices (scenarios).

the simulated data we calculated the ASPT values using the equation derived by the regression analysis described in the previous section. The statistical significance of the effect of management practices on the environmental descriptors and the ASPT was tested after performing a 2-way ANOVA followed by Tukey's HSD post hoc test to identify significant differences between all the possible pairs. As a second independent variable we used the season (wet or dry) in order to examine if there is a seasonal effect on the ASPT scores and the environmental descriptors. Seasonal mean values of the simulated data were calculated for the dry and wet period for each of the 20 years and for the total of 47 sites under the baseline conditions and the implementation of the management practices.

range of mean phosphate concentration extends from approximately as low as 6 μg/L to 78 μg/L while the mean DIN concentration ranges from 0.42 mg/L to 11.65 mg/L. The mean discharge is 10.98 m3/s. Table 3 presents the mean ASPT simulated for the 47 subbasins under baseline conditions and after the implementation of the management practices. From this table it appears that the calculated ASPT ranges from 4.63 (subbasin 35) to 6.35 (subbasin 13) among the 47 subbasins with a total mean value of 5.76. In Fig. 6 we can note how the spatial variation of the ASPT follows a reverse pattern compared to the one observed from Figs. 4 and 5.

3. Results

Table 4 summarizes the overall mean values of ASPT, DIN, phosphate and discharge and the significance level of difference (p value) among the three management practices. Specifically, it appears that after applying a 30% reduction on irrigation the discharge increases from 10.98 m3/s to 11.36 m3/s. This may appear a small increase but it actually means that the application of this practice can save annually approximately 12 × 106 m3 of water. The mean DIN concentration under the same scenario appears to increase by 0.231 mg/L (Table 4). Mean phosphate remained practically unaffected as it decreases by just 0.26 μg/L (Table 4). The reduction of the fertilization by 30% as expected has a significant effect on the DIN and phosphate concentration by decreasing the mean values from 2.765 mg/L and 0.031 mg/L to

3.1. Baseline conditions Our results showed that the overall mean simulated concentration of DIN and phosphate for the 47 subbasins under baseline conditions are 2.765 mg/L and 0.031 mg/L respectively. The spatial variation of the mean DIN and phosphate concentration among the 47 sites for the baseline conditions is shown in Figs. 4 and 5. From these figures it is apparent that DIN and phosphate follow a similar pattern for the majority of the sites. High DIN values coincide with high phosphate values which depict the impacts from nonpoint sources of nutrient pollution. The

3.2. Implementation of management practices/scenarios

Fig. 5. Mean annual phosphate concentration at the 47 subbasin outlets of Pinios under the baseline conditions and under the simulation of the three management practices (scenarios).

Please cite this article as: Stefanidis, K., et al., Impact assessment of agricultural driven stressors on benthic macroinvertebrates using simulated data, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.08.015

K. Stefanidis et al. / Science of the Total Environment xxx (2015) xxx–xxx Table 3 Mean values of the ASPT metric simulated in the outlets of 47 subbasins of Pinios for the baseline conditions and under the three management practices (standard error in brackets). Subbasin

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47

ASPT metric Baseline

−30% irrigation

−30% fertilization

Both

5.82 (0.22) 6.03 (0.03) 6.09 (0.02) 6.16 (0.02) 5.99 (0.02) 6.16 (0.02) 5.88 (0.01) 5.99 (0.16) 5.79 (0.02) 5.95 (0.15) 5.60 (0.21) 5.84 (0.01) 6.35 (0.06) 6.10 (0.02) 5.69 (0.02) 5.49 (0.20) 5.78 (0.01) 5.02 (0.27) 5.62 (0.02) 5.49 (0.15) 5.55 (0.02) 6.15 (0.08) 5.36 (0.25) 6.23 (0.03) 5.97 (0.02) 5.98 (0.01) 5.82 (0.02) 5.58 (0.14) 5.58 (0.15) 5.57 (0.14) 5.57 (0.02) 5.63 (0.02) 5.62 (0.02) 5.62 (0.02) 4.63 (0.31) 5.60 (0.02) 5.55 (0.30) 6.13 (0.03) 5.94 (0.16) 5.76 (0.02) 5.98 (0.16) 5.78 (0.02) 5.80 (0.02) 5.80 (0.02) 5.81 (0.02) 5.56 (0.02) 5.39 (0.25)

5.92 (0.22) 5.99 (0.04) 5.76 (0.03) 5.97 (0.03) 5.64 (0.03) 5.99 (0.03) 5.73 (0.03) 6.24 (0.17) 6.14 (0.04) 6.22 (0.16) 5.92 (0.22) 5.46 (0.03) 6.37 (0.05) 6.22 (0.04) 6.18 (0.03) 5.93 (0.22) 5.40 (0.03) 5.72 (0.31) 5.97 (0.16) 6.12 (0.16) 5.53 (0.21) 6.00 (0.04) 5.98 (0.22) 6.26 (0.03) 5.96 (0.03) 5.93 (0.03) 5.87 (0.03) 5.85 (0.15) 5.82 (0.15) 5.77 (0.15) 5.47 (0.04) 5.24 (0.03) 5.24 (0.03) 5.24 (0.03) 5.17 (0.35) 5.23 (0.03) 5.56 (0.30) 6.15 (0.03) 5.97 (0.16) 5.89 (0.04) 6.17 (0.16) 5.84 (0.04) 5.70 (0.03) 5.68 (0.03) 5.67 (0.03) 5.17 (0.03) 6.28 (0.03)

5.86 (0.22) 6.07 (0.03) 6.09 (0.02) 6.16 (0.02) 6.01 (0.02) 6.16 (0.02) 5.93 (0.01) 6.07 (0.16) 5.88 (0.02) 5.98 (0.15) 5.66 (0.21) 5.89 (0.01) 6.35 (0.06) 6.17 (0.02) 5.79 (0.02) 5.57 (0.21) 5.84 (0.01) 5.14 (0.28) 5.58 (0.15) 5.64 (0.15) 5.33 (0.20) 6.22 (0.07) 5.48 (0.25) 6.22 (0.03) 6.04 (0.02) 6.10 (0.01) 5.95 (0.02) 5.73 (0.15) 5.72 (0.15) 5.71 (0.15) 5.71 (0.02) 5.73 (0.02) 5.73 (0.02) 5.73 (0.02) 4.74 (0.32) 5.71 (0.02) 5.55 (0.30) 6.14 (0.03) 5.97 (0.16) 5.89 (0.02) 6.03 (0.16) 5.90 (0.02) 5.89 (0.02) 5.90 (0.02) 5.91 (0.02) 5.68 (0.02) 5.71 (0.15)

5.86 (0.22) 6.06 (0.03) 6.09 (0.02) 6.16 (0.02) 6.01 (0.02) 6.16 (0.02) 5.93 (0.01) 6.07 (0.16) 5.87 (0.02) 5.99 (0.15) 5.66 (0.21) 5.88 (0.01) 6.35 (0.06) 6.14 (0.02) 5.76 (0.02) 5.54 (0.21) 5.82 (0.01) 5.14 (0.28) 5.58 (0.15) 5.64 (0.15) 5.33 (0.20) 6.22 (0.07) 5.63 (0.21) 6.22 (0.03) 6.03 (0.02) 6.10 (0.01) 5.95 (0.02) 5.72 (0.15) 5.72 (0.15) 5.70 (0.15) 5.70 (0.02) 5.72 (0.02) 5.72 (0.02) 5.71 (0.02) 4.73 (0.32) 5.70 (0.02) 5.55 (0.30) 6.14 (0.03) 5.96 (0.16) 5.89 (0.02) 6.03 (0.16) 5.90 (0.02) 5.90 (0.02) 5.90 (0.02) 5.91 (0.02) 5.68 (0.02) 5.84 (0.03)

7

1.874 mg/L and 0.024 mg/L respectively. The decrease in nutrients results to an increase of mean overall ASPT to 5.835. Similar results are obtained after the implementation of the third management practice which is the combination of a 30% reduction in irrigation and fertilization. Actually, from the results of the post-hoc comparisons (Table 4) it appears that the combined effect of irrigation and fertilization reduction on the ASPT, DIN and phosphate did not significantly differ with the effect of the 30% fertilization reduction when applied solely. Table 3 presents some interesting findings regarding the response of ASPT to the application of the three management scenarios for each subbasin separately. First it is obvious that there is no practical difference between the effect of the fertilization reduction and the effect of the combination of both practices (irrigation and fertilization reduction). In both scenarios the mean ASPT increases in 40 out of the total 47 subbasins. On the contrary, the 30% irrigation reduction appears to cause a decrease of the mean ASPT in several occasions (21 subbasins). 3.3. Differences between wet and dry season The statistically significant differences for the ASPT, DIN, phosphate and discharge between the wet and dry season are presented in Table 5. ASPT present a significant variation between the two periods only under the implementation of the reduction of irrigation. DIN concentration decreased during the dry period while phosphate concentration appeared to increase significantly during the dry period regardless of the applied scenario. Discharge as expected increased during the wet season. A graphical representation of the results regarding the DIN and phosphate concentration between the wet and dry season can be shown in Figs. 7 and 8. 4. Discussion Our analysis is based on significant relationships between a biological indicator (ASPT) and abiotic variables such as nitrogen and phosphorus. Of course, this finding is not surprising as these relations imply an important effect of nutrient pollution, mainly nitrogen, on the occurrence of sensitive in pollution taxa and agree with results from other studies performed elsewhere in Europe and the rest of the world (Weigel and Robertson, 2007; Johnson and Hering, 2009; Dahm et al., 2013; Lakew and Moog, 2015; Villeneuve et al., 2015). There are other studies as well that have reported specific important relationships between nitrogen species (TN, NH4–N) and macroinvertebrate communities (Wang et al., 2007; Ashton et al., 2014) which agree with the identification of DIN as a key predictor of ASPT metric. These relationships apparently indicate an indirect effect of eutrophication on

Fig. 6. Mean annual value of ASPT at the 47 subbasin outlets of Pinios under the baseline conditions and under the simulation of the three management practices (scenarios).

Please cite this article as: Stefanidis, K., et al., Impact assessment of agricultural driven stressors on benthic macroinvertebrates using simulated data, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.08.015

8

K. Stefanidis et al. / Science of the Total Environment xxx (2015) xxx–xxx

Table 4 Mean values of ASPT, DIN, phosphate and discharge (std. error in brackets) for baseline conditions and management practices. Statistical significance is given according to ANOVA results. Values sharing the same superscripted letter are significantly different according to post-hoc comparisons.

ASPT DIN (mg/L) Phosphate (mg/L) Discharge (m3/s)

Baseline

−30% Irrigation

−30% Fertilization

Combination of both practices

Sig.

5.76 (0.019)abc 2.765ab (0.089) 0.031ab (6.9 × 10−4) 10.98 (0.52)

5.819 (0.019)a 2.996cd (0.096) 0.031cd (6.7 × 10−4) 11.36 (0.52)

5.835 (0.019)b 1.874ac (0.054) 0.024ac (4.9 × 10−4) 11.26 (0.53)

5.837 (0.019)c 1.908bd (0.054) 0.024bd (4.9 × 10−4) 11.26 (0.53)

p b 0.05 p b 0.001 p b 0.001 ns

F values in italics indicate non significant differences.

macroinvertebrate communities. Although the exact mechanisms behind the eutrophication effect are not fully understood yet there is enough evidence to show that nutrient excess can lead to oxygen depletion due to increase in consumption from algal and microbial growth. In addition, the decomposition of the organic matter leads further to increase of oxygen consumption which in turn favors more resistant taxa to anoxic conditions (Johnson and Hering, 2009; Dahm et al., 2013). Our results also suggest that ASPT scores are affected by the discharge implying that higher discharges will have an impact on macroinvertebrate communities. It was also shown that other metrics (BMWP, EPT) presented even stronger relationships with discharge. Several studies have also reported significant relationships between macroinvertebrate indicators and stream flow. For example Buffagni et al. (2010) and Dunbar et al. (2010) showed how Lotic Invertebrate index for Flow Evaluation (LIFE) responds to changes in flow and channel morphology. Stream flow influences the morphological characteristics of a stream, such as channel morphology and type of substratum, affecting indirectly the benthic aquatic communities. Therefore, the effect on macroinvertebrate communities is rather connected to the habitat heterogeneity that is affected in some extent by the flow (Cortes et al., 2002; Graeber et al., 2013; Waddle and Holmquist, 2013). High flow is possible to have an impact on macroinvertebrates by destabilizing the bed sediments, increasing the macroinvertebrate drift, removing food resources and benthic detritus and affecting the substrata that function as habitat for macroinvertebrates (Robinson et al., 2004; Suren and Jowett, 2006). Other studies have also reported significant differences in the macroinvertebrate densities between high flow and low flow periods due to a possible diluting effect on the benthic fauna (Barbero et al., 2013). Although the predictability of our empirical model is not very high, we should consider the fact that apart from the physicochemical and hydrological variables there are many other factors that influence the macroinvertebrate indicators, such as substrate and habitat quality. However, the collection of this amount of information is not always practical whereas the development of an appropriate modeling technique that implements these data requires a more complex process that is beyond the scope of our article. Besides, the linkage of a predictive model that uses variables such as habitat heterogeneity or channel features with a process based model such as SWAT is not feasible. SWAT cannot model features related to channel morphology and habitat structure. It would be possible however to use land use metrics to develop a predictive model that takes into account changes of land use, but this approach could be implemented in a future research. Regarding the simulated results, the concentrations of nitrogen and phosphorus Table 5 Statistically significant differences for the means of ASPT, DIN, phosphate and discharge among groups (2-Way ANOVA results for scenarios, seasons and scenarios × seasons). Scenarios

Seasons

Scenarios × seasons

1.018 (p = 0.313) 111.325 (p b 0.001) 352.991 (p b 0.001) 539.65 (p b 0.001)

1.595 (p = 0.188) 3.455 (p b 0.05) 21.756 (p b 0.001) 0.009 (p = 0.99)

F value ASPT DIN Phosphate Discharge

420.370 (p b 0.001) 57.973 (p b 0.001) 56.311 (p b 0.001) 0.156 (p = 0.926)

F values in bold indicate significant differences (p b 0.05 and 0.001) and F values in italics indicate non significant differences.

are comparable with observations made by Chatzinikolaou et al. (2010) who reported phosphate and DIN concentrations as high as 0.231 mg/L and 17.1 mg/L respectively in the main stretch of Pinios. In general it appears that there is a spatial variation of the nitrogen and phosphorus concentration which reflects the differences of land uses among the 47 subbasins. On this basis we were able to simulate the value of the ASPT for the outlets of 47 subbasins across the catchment of river Pinios under the baseline conditions and after the implementation of management practices. Under the baseline conditions our modeling approach resulted to realistic values that are representative of a wide range of nutrient concentration. The majority of the subbasins are characterized by ASPT values between 5 and 6 indicating probably moderate and good ecological conditions. From several studies that used ASPT as an indicator for assessing ecological status we know that usually values larger than 6 fall in a “high ecological status” (Armitage et al., 1983; Sandin and Hering, 2004; Birk and Hering, 2006) and we can assume that sites with values smaller than 5 reflect a strong influence of organic pollution. From the work of Chatzinikolaou and Lazaridou (2007) we know that almost half of the sites of Pinios that were assessed using macroinvertebrates, were characterized by moderate biological quality and only 11% of the sites (mostly mountainous site) met good biological quality. Regarding the effects of the applied management practices on the overall mean ASPT we have to note that the magnitude of change appear to be quite small and one could argue that practically the impact is negligible. However in examining specific cases (subbasins) there seems to be a significant change of the ASPT that actually could reflect an important ecological change. For example in sub-basin 47, mean ASPT increased from 5.39 in baseline conditions to 5.84 under the application of the combined management practice. Apparently the most impaired sites, meaning the ones with the highest degree of water quality degradation, would benefit more from a significant change of nutrient concentration, whereas sites that were characterized by low nutrient concentration would remain rather unaffected by a reduction of fertilization. Although in our case it is not feasible to characterize the ecological quality based solely on ASPT, we know from studies that assessed the ecological status of streams and rivers across Europe using the ASPT metric, that the boundaries between the different classes can be rather small, reflecting probably the high sensitivity of macroinvertebrate taxa in small changes of environmental conditions. For example, from a pan-European study made by Sandin and Hering (2004) the difference in class boundaries between high/good and good/moderate status was as low as 0.21 and 0.345 depending on the type of the stream. Therefore, in our case is not unlikely that a small increase of ASPT around 0.2–0.3 for some sub-basins can reflect a significant change of ecological status. Another key finding of our work appears to be that the reduction of irrigation by 30% resulted to an increase of the mean concentration of nitrogen by approximately 0.2 mg/L. A possible explanation for the increase of nitrogen levels is related to the reduction of the crop growth simulated by SWAT due to the reduced irrigation. The reduction of crop productivity will result to a consequent reduction of nutrient uptake from the soil which in turn increases the nitrogen accumulation. As a result of this process the excess amount of nitrogen in the soil will eventually end in surface and/or ground water explaining the predicted increase in concentration. Although the overall change is small,

Please cite this article as: Stefanidis, K., et al., Impact assessment of agricultural driven stressors on benthic macroinvertebrates using simulated data, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.08.015

K. Stefanidis et al. / Science of the Total Environment xxx (2015) xxx–xxx

9

Fig. 7. Mean concentration of DIN for wet and dry season among the baseline conditions and the 3 management practices.

in some sub-basins the increase of DIN concentration appears to reach almost 1 mg/L which can be responsible for a significant decrease of the ASPT (e.g. by 0.39 and 0.38 for sub-basins 46 and 12 respectively). Regarding the observed seasonal differences of phosphate and DIN concentrations, the results presented a contrasting pattern where DIN concentration is significantly higher in the wet season and phosphate is higher during the dry season regardless of the scenario. This pattern appears to be confirmed by similar findings in the literature. In general, higher concentration of nitrogen during the wet season reflects the effect of autumn and winter runoff of nitrate from agricultural land (Arazuzo et al., 2011; Miller et al., 2014). On the contrary a peak of phosphate during the summer and low concentration during winter and autumn is attributed to a dilution effect of the overall catchment phosphorus load (Miller et al., 2014). From a management point of view our results suggest that reduction of irrigation by 30% in some subbasins can have a negative effect for water quality, since it increases the nitrogen concentration, although it saves a considerable amount of water. On the contrary the results showed that there are very small differences between the effects of a 30% reduction of fertilization and a combined practice of deficit irrigation and reduced fertilization. Based on this outcome it appears that the best management practice would be to apply a simultaneous

reduction of fertilization and irrigation that improves water quality and consequently ASPT, while at the same time saves a considerable quantity of water. 5. Conclusion In this paper we presented a simple methodology that assesses the effects of agricultural management practices on a common biotic indicator such as ASPT. We showed that a process based model can be coupled with simple empirical predictive models and that is possible to explore the response of biological related metrics to stressors determined by agricultural activities such as irrigation and nutrient application. The results presented in this paper showed that reduced irrigation without a simultaneous fertilization reduction, although it saves considerable amounts of water, it may increase the nitrogen concentration in some subbasins, which in turn impacts the macroinvertebrate communities. Thus, it is important for management authorities to explore thoroughly all the possible effects caused by the implementation of management practices even if their aim is to restore ecosystem functions and services (e.g. saving water). Based on our results, we suggest that in our case the best management option would be a combined practice of deficit irrigation and fertilization reduction since it improves water quality, and

Fig. 8. Mean concentration of phosphate for wet and dry season among the baseline conditions and the 3 management practices.

Please cite this article as: Stefanidis, K., et al., Impact assessment of agricultural driven stressors on benthic macroinvertebrates using simulated data, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.08.015

10

K. Stefanidis et al. / Science of the Total Environment xxx (2015) xxx–xxx

potentially ecological quality as well, and saves a considerable amount of water that could mitigate water scarcity in the basin. To this end, our proposed methodology provides a simple tool for an effective assessment of management practices on ecological related metrics without being restricted by the limitation of complex and large datasets. The present approach can also be applied for other biological quality elements (e.g. fish and macrophytes) as long as there are available empirical data to form relationships between biotic indicators and environmental predictors. Although we acknowledge that other factors such as habitat heterogeneity, which are not possible to be modeled within the current modeling framework, are possible to play a decisive role on the biotic metrics, we should note that the improvement of model predictability would require more effort of field work, larger datasets and sophisticated modeling techniques, which is beyond the scope of this article. However, because there is a continuous need to improve our knowledge on the interactions between the aquatic biota and the environment, the development of new and more advanced integrated modeling frameworks could be the objective of future research studies. Acknowledgments This work was supported by the MARS project (Managing Aquatic ecosystems and water Resources under multiple Stress) funded under the 7th EU Framework Programme, Theme 6 (Environment including Climate Change), Contract No: 603378 (http://www.mars-project.eu). References Aguiar, F., Ferreira, M.T., 2005. Human-disturbed landscapes: effects on composition and integrity of riparian woody vegetation in the Tagus River basin, Portugal. Environ. Conserv. 32, 30–34. Arazuzo, M., Valladolid, M., Mariznez-Bastida, J.J., 2011. Spatio-temporal dynamics of nitrogen in river-alluvial aquifer systems affected by diffuse pollution from agricultural sources: Implications for the implementation of the Nitrates Directive. J. Hydrol. 411, 155–168. Armitage, P.D., Moss, D., Wright, J.F., Furse, M.T., 1983. The performance of a new water quality score system based on macroinvertebrates over a wide range of unpolluted running-water sites. Water Res. 17, 333–347. Arnold, J.G., Srinivasan, R., Muttiah, R.S., Williams, J.R., 1998. Large area hydrologic modelling and assessment part I: model development. J. Am. Water Resour. Assoc. 34, 73–89. Ashton, M.J., Morgan II, R.P., Stranko, S., 2014. Relations between macroinvertebrates, nutrients and water quality criteria in wadeable streams of Maryland, USA. Environ. Monit. Assess. 186, 1167–1182. Barbero, M.D., Oberto, A.M., Gualdoni, C.M., 2013. Spatial and temporal patterns of macroinvertebrates in drift and on substrate of a mountain stream (Cordoba, Cental Argentina). Acta Limnol. Bras. 25 (no4). Bellos, D., Sawidis, T., 2005. Chemical pollution monitoring of the River Pinios (ThessaliaGreece). J. Environ. Manag. 76, 282–292. Birk, S., Hering, D., 2006. Direct comparison of assessment methods using benthic macroinvertebrates: a contribution to the EU Water Framework Directive intercalibration exercise. Hydrobiologia 566, 401–415. Bonada, N., Resh, V.H., 2013. Mediterranean-climate streams and rivers: geographically separated but ecologically comparable freshwater systems. Hydrobiologia 719, 1–29. Buffagni, A., Erba, S., Cazzola, M., Murray-Bligh, J., Soszka, H., Genoni, P., 2006. The STAR common metrics approach to the WFD intercalibration process: full application for small, lowland rivers in three European countries. Hydrobiologia 566, 379–399. Buffagni, A., Erba, S., Armanini, D.G., 2010. The lentic–lotic character of Mediterranean rivers and its importance to aquatic invertebrate communities. Aquat. Sci. 72, 45–60. Carlisle, D.M., Meador, M.R., 2007. A biological assessment of streams in the eastern United States using a predictive model for macroinvertebrate assemblages. J. Am. Water Resour. Assoc. 43, 1194–1207. Chatzinikolaou, Y., 2007. Effect of management practices on the water quality and ecology of rivers in Greece. Pinios River as a study case, 2007 Doctorate thesis. Department of Biology, Aristotle University of Thessaloniki, Volume A: pp. 229 and volume B, p. 470. http://dx.doi.org/10.12681/eadd/20876. Chatzinikolaou, Y., Lazaridou, M., 2007. Identification of the self-purification stretches of the Pinios River, Central Greece Mediterranean. Mar. Sci. 8, 19–32. Chatzinikolaou, Y., Ioannou, A., Lazaridou, M., 2010. Intra-basin spatial approach on pollution load estimation in a large Mediterranean river. Desalination 250, 118–129. Clarke, R.T., Wright, J.F., Furse, M.T., 2003. RIVPACS models for predicting the expected macroinvertebrate fauna and assessing the ecological quality of rivers. Ecol. Model. 160, 219–233. Cortes, R.M.V., Ferreira, M.T., Oliveira, S.V., Oliveira, D., 2002. Macroinvertebrate community structure in a regulated river segment with different flow conditions. River Res. Appl. 18, 267–382.

Dahm, V., Hering, D., Nemitz, D., Graf, W., Schmidt-Kloiber, A., Leitner, P., Melcher, A., Feld, C.K., 2013. Effects of physico-chemistry, land use and hydromorphology on three riverine organism groups: a comparative analysis with monitoring data from Germany and Austria. Hydrobiologia 704, 389–415. Dunbar, M.J., Pedersen, M.L., Cadman, D., Extence, C., Waddingham, J., Chadd, R., Larsen, S.E., 2010. River discharge and local-scale physical habitat influence macroinvertebrate LIFE scores. Freshw. Biol. 55, 226–242. Einheuser, M.D., Nejadhashemi, A.P., Wang, L., Sowa, S.P., Woznicki, S.A., 2013. Linking biological integrity and watershed models to assess the impacts of historical land use and climate changes on stream health. Environ. Manag. 51, 1147–1163. Feld, C.K., Hering, D., 2007. Community structure or function: effects of environmental stress on benthic macroinvertebrates at different spatial scales. Freshw. Biol. 52, 1380–1399. Gabriels, W., Goethals, P.L.M., Dedecker, A.P., Lek, S., De Pauw, N., 2007. Analysis of macrobenthic communities in Flanders, Belgium, using a stepwise input variable selection procedure with artificial neural networks. Aquat. Ecol. 41, 427–441. Gasith, A., Resh, V.H., 1999. Streams in Mediterranean regions: abiotic influences and biotic responses to predictable seasonal events. Annu. Rev. Ecol. 30, 51–81. Gassman, P.W., Sadeghi, A.M., Srinivasan, R., 2014. Applications of the SWAT model special section: overview and insights. J. Environ. Qual. 43, 1–8. Geerts, S., Raes, D., 2009. Deficit irrigation as an on-farm strategy to maximize crop water productivity in dry areas. Agric. Water Manag. 96, 1275–1284. Graeber, D., Pusch, M.T., Lorenz, S., Brauns, M., 2013. Cascading effects of flow reduction on the benthic invertebrate community in a lowland river. Hydrobiologia 717, 147–159. Hershkovitz, Y., Gasith, A., 2013. Resistance, resilience, and community dynamics in mediterranean-climate streams. Hydrobiologia 719, 59–75. Hooke, J.M., 2006. Human impacts on fluvial systems in the Mediterranean region. Geomorphology 79, 311–335. Ioannou, A., Chatzinikolaou, Y., Lazaridou, M., 2009. A preliminary pressure-impact analysis applied in the Pinios river basin (Thessaly, Central Greece). Water Environ. J. 23, 200–209. Jähnig, S.C., Kuemmerlen, M., Kiesel, J., Domisch, S., Cai, Q., Schmalz, B., Fohrer, N., 2012. Modelling of riverine ecosystems by integrating models: conceptual approach, a case study and research agenda. J. Biogeogr. 39, 2253–2263. Johnson, R.K., Hering, D., 2009. Response of taxonomic groups in streams to gradients in resource and habitat characteristics. J. Appl. Ecol. 46, 175–186. Kuemmerlen, M., Schmalzc, B., Gusec, B., Caid, Q., Fohrerc, N., Jähnig, S.C., 2014. Integrating catchment properties in small scale species distribution models of stream macroinvertebrates. Ecol. Model. 277, 77–86. Lakew, A., Moog, O., 2015. A multimetric index based on benthic macroinvertebrates for assessing the ecological status of streams and rivers in central and southeast highlands of Ethiopia. Hydrobiologia http://dx.doi.org/10.1007/s10750-015-2189-1. Loukas, A., 2010. Surface water quantity and quality assessment in Pinios River, Thessaly, Greece. Desalination 250, 266–273. Magidson, J., 2013. Correlated component regression: re-thinking regression in the presence of near collinearity. Springer Proc. Math. Stat. 56, 65–78. Maloney, K.O., Schmid, M., Weller, D.E., 2012. Applying additive modelling and gradient boosting to assess the effects of watershed and reach characteristics on riverine assemblages. Methods Ecol. Evol. 3, 116–128. Marchini, A., 2011. Modelling ecological processes with fuzzy logic approaches. Model. Complex Ecol. Dyn. 2, 133–145. Miller, C., Magdalina, A., Willows, R.I., Bowman, A.W., Scott, E.M., Lee, D., Burgess, S., Pope, L., Pannullo, F., Haggarty, R., 2014. Spatiotemporal statistical modelling of long-term change in river nutrient concentrations in England & Wales. Sci. Total Environ. 466467, 914–923. Milly, P.C., Dunne, K.A., Vecchia, A.V., 2005. Global pattern of trends in streamflow and water availability in a changing climate. Nature 438, 347–350. Neitsch, S.L., Arnold, J.G., Kiniry, J.R., Williams, J.R., 2009. Soil and Water Assessment Tool – Theoretical Documentation—Version 2005. Blackland Research Center – ARS, TX, USA (http://twri.tamu.edu/reports/2011/tr406.pdf). Ormerod, S.J., Dobson, M., Hildrew, A.G., Townsend, C.R., 2010. Multiple stressors in freshwater ecosystems. Freshw. Biol. 55, 1–4. Panagopoulos, Y., Makropoulos, C., Mimikou, M., 2012. A multi-Objective Decision Support Tool for Rural Basin Management. International Environmental Modelling and Software Society (iEMSs), International Congress on Environmental Modelling and Software, Managing Resources of a Limited Planet: Pathways and Visions under Uncertainty, Leipzig, Germany (1–5 July 2012). Panagopoulos, Y., Makropoulos, C., Kossida, M., Mimikou, M., 2014a. Optimal implementation of irrigation practices: a cost effective desertification action plan for Pinios. J. Water Resour. Plan. Manag. 140, 05014005. Panagopoulos, Y., Makropoulos, C., Gkiokas, A., Kossida, M., Evangelou, E., Lourmas, G., Michas, S., Tsadilas, C., Papageorgiou, S., Perleros, V., Drakopoulou, S., Mimikou, M., 2014b. Assessing the cost-effectiveness of irrigation water management practices in water stressed agricultural catchments: the case of Pinios. Agric. Water Manag. 139, 31–42. Pool, T.K., Olden, J.D., Whittier, J.B., Paukert, C.P., 2010. Environmental drivers of fish functional diversity and composition in the Lower Colorado River Basin. Can. J. Fish. Aquat. Sci. 67, 1791–1807. Robinson, C.T., Aebischer, S., Uehlinger, U., 2004. Immediate and habitat-specific responses of macroinvertebrates to sequential, experimental floods. J. N. Am. Benthol. Soc. 23, 853–867. Sandin, L., Hering, D., 2004. Comparing macroinvertebrate indices to detect organic pollution across Europe: a contribution to the EC Water Framework Directive intercalibration. Hydrobiologia 516, 55–68.

Please cite this article as: Stefanidis, K., et al., Impact assessment of agricultural driven stressors on benthic macroinvertebrates using simulated data, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.08.015

K. Stefanidis et al. / Science of the Total Environment xxx (2015) xxx–xxx Schinegger, R., Trautwein, C., Melcher, A., Schmutz, S., 2012. Multiple human pressures and their spatial patterns in European running waters. Water Environ. J. 26, 261–273. Skoulikidis, N., Karaouzas, I., Gritzalis, K., 2009. Identifying key environmental variables structuring benthic fauna for establishing a biotic typology for Greek running waters. Limnologica 39, 56–66. Suren, A.M., Jowett, I.G., 2006. Effects of floods versus low flows on invertebrates in a New Zealand gravel-bed river. Freshw. Biol. 51, 2207–2227. Vasiliades, L., Loukas, A., Liberis, N., 2011. A water balance derived drought index for Pinios River Basin, Greece. Water Resour. Manag. 25, 1087–1101. Villeneuve, B., Souchon, Y., Ussegio-Polatera, P., Ferréol, M., Valette, L., 2015. Can we predict biological condition of stream ecosystems? A multi-stressors approach linking three biological indices to physico-chemistry, hydromorphology and land use. Ecol. Indic. 48, 88–98. Vörösmarty, C.J., McIntyre, P.B., Gessner, M.O., Dudgeon, D., Prusevich, A., Green, P., Glidden, S., Bunn, S.E., Sullivan, C.A., Reidy Liermann, C., Davies, P.M., 2010. Global threats to human water security and river biodiversity. Nature 467, 555–561.

11

Waddle, T.J., Holmquist, J.G., 2013. Macroinvertebrate response to flow changes in a subalpine stream: predictions from two-dimensional hydrodynamic models. River Res. Appl. 29, 366–379. Wang, L., Roberston, D.M., Garrison, S.F., 2007. Linkages between nutrients and assemblages of macroinvertebrates and fish in wadeable streams: implications to nutrient criteria development. Environ. Manag. 39, 194–212. Weigel, B.M., Robertson, D.M., 2007. Identifying biotic integrity and water chemistry relations in nonwadeable rivers of Wisconsin: toward the development of nutrient criteria. Environ. Manag. 40, 691–708. WFD CIS, 2003. Common Implementation Strategy for the Water Framework Directive (2000/60/EC). Guidance document no. 11Planning process.

Please cite this article as: Stefanidis, K., et al., Impact assessment of agricultural driven stressors on benthic macroinvertebrates using simulated data, Sci Total Environ (2015), http://dx.doi.org/10.1016/j.scitotenv.2015.08.015