Ecological Engineering 60 (2013) 299–308
Contents lists available at ScienceDirect
Ecological Engineering journal homepage: www.elsevier.com/locate/ecoleng
Regulation of diffuse pesticide pollution: Combining point source reduction and mitigation in stormwater wetland (Rouffach, France) Franc¸ois Destandau a,∗ , Gwenaël Imfeld b,1 , Anne Rozan a a Laboratory of Territorial Water and Environment Management (GESTE) UMR ENGEES-IRSTEA, BETA, UMR 7522 CNRS, ENGEES, 1 quai Koch, BP 61039, 67070 Strasbourg Cedex, France b Laboratory of Hydrology and Geochemistry of Strasbourg (LHyGeS) University of Strasbourg/ENGEES, UMR 7517 CNRS, 1 rue Blessig, 67084 Strasbourg, France
a r t i c l e
i n f o
Article history: Received 21 March 2013 Received in revised form 14 June 2013 Accepted 5 July 2013 Available online 22 August 2013 Keywords: Diffuse pollution Economic regulation Fungicides Mitigation Stormwater wetland
a b s t r a c t The economic crisis and increasingly stringent water quality requirements demand integrative approaches to reduce pesticide pollution in aquatic ecosystems and limit water treatment. In this study, environmental economists and wetland scientists joined forces to analyze the combination of different abatement measures to reduce pollution with more efficiency. Pesticide reduction directly at source (i.e., reduction of pesticide use), and combining pesticide source reduction with mitigation using a stormwater wetland to treat pesticide runoff are compared. The capacity of the buffer zone to reduce additional diffuse pollution with a given total abatement cost is evaluated, by placing emphasis on how the contribution of a buffer zone evolves according to the total cost. Fungicides were used as a representative class of synthetic pesticides widely used in vine growing, and more largely in conventional agriculture. Our results show that coupling reduction of pesticide source with the use of buffer zones collecting pesticide runoff can be economically advantageous. For a given total cost, the reduction of fungicide runoff is 90% greater when pesticide reduction at source is combined with pesticide mitigation by a stormwater wetland compared to the case of pesticide reduction at source only. However, the higher the total cost is, the more it is necessary to reduce pesticides at source and thus reduce pesticide mass transfer into aquatic systems. The results of this study is anticipated to be a starting point for considering cost and efficiency when combining different measures targeting pesticide mitigation in surface water, and in particular when using stormwater wetlands as a management practice. © 2013 Elsevier B.V. All rights reserved.
1. Introduction The European Water Framework Directive (WFD) (European Commission, 2000) aims at achieving good quality for water bodies by 2015. The European member states committed to it may request an exemption if the cost of reaching the quality objectives is deemed too expensive (see Hanley et al. (2006) for more details). Exemptions to the general objectives due to the poor ecological and chemical status of surface water involve about 30% of water bodies in Europe (European Commission, 2012). In particular, innovative solutions are required to reduce diffuse pesticide pollution
∗ Corresponding author. Tel.: +33 3 88 24 82 40; fax: +33 3 88 24 82 84. E-mail addresses:
[email protected] (F. Destandau),
[email protected] (G. Imfeld),
[email protected] (A. Rozan). 1 Tel.: +33 03 68 85 04 07. 0925-8574/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ecoleng.2013.07.030
without increasing the abatement cost, to reach the general objectives of the WFD and decrease the exemption rate for the next cycle in 2021. As long as pesticides are used, a certain portion of the pesticides used in agriculture and in urban areas can move from land to aquatic ecosystems during rainfall-runoff events (Poissant et al., 2008; Lefrancq et al., 2013). Thus complementary measures at plot and catchments scale, such as conservation tillage on cultivated surfaces and buffer zone implementation on specific areas are needed (Mitsch, 1992). Buffer zones such as stormwater wetlands can intercept and partly retain runoff-related contaminants in agricultural and urban catchment areas, thereby limiting the contamination of water bodies (Fournel et al., 2013; Grégoire et al., 2009; Hatvani et al., 2011; Ockenden et al., 2012). Stormwater wetlands, storm basins or detention ponds are engineered wetlands to temporarily store runoff and are specifically designed for flood control. In addition to their capacities to detain and dampen
300
F. Destandau et al. / Ecological Engineering 60 (2013) 299–308
storm-flow pulses, wetlands can also retain excess nutrients and solids that may pollute downstream waters (Holland et al., 2005). Recent studies have underscored the potential of wetland systems as a management practice targeting the removal of pesticides and water quality improvement (Grégoire et al., 2009; Ockenden et al., 2012). The temporary water storage in shallow pools of stormwater wetlands supports conditions suitable for the growth of wetland plants and bioremediation. Stormwater wetlands can capitalize intrinsic physical, chemical and biological detention as well as degradative processes useful for treating various organic chemicals (Imfeld et al., 2013), including pesticides (Grégoire et al., 2009; Stehle et al., 2011). In particular, recent studies have underscored the potential of stormwater wetlands as a management practice targeting pesticide attenuation and water quality improvement (Budd et al., 2009; Imfeld et al., 2013; Maillard et al., 2011). From the economic perspective, strategies to reduce fertilizer or pesticide inputs, including the use of buffer zones to improve water quality, have been the subject of numerous studies. Since the beginning of the 2000s, several studies have emphasized the role of wetlands to limit the problems of eutrophication in the Baltic Sea in response to the Helsinki Convention (Elofsson, 2010). A cost function (Byström, 1998; Söderqvist, 2002) was used to estimate the cost-effectiveness relation of wetland systems (Byström, 2000; Byström et al., 2000), and to design incitative policies for the use of buffer zones by farmers (Lindhal and Söderqvist, 2004). More recently, Crépin, 2005 and Heberling et al. (2010) studied incentives to create or restore wetlands through subsidies and contracts. Paulsen (2007) focused on the role of uncertainty in farmers’ decisions. Ribaudo et al. (2001) evaluated various options to reduce nutrients in the Mississippi River Basin, including the reduction of fertilizer use and the use of buffer wetlands for treating nutrient loads. For a large catchment area, the latter study shows that a policy based exclusively on wetlands could be more efficient beyond a particular level of total nitrogen reduction (reduction up to 26%). In contrast, measures to reduce pollution at source are more efficient below 26%. The optimal use of buffer zones treating contaminant runoff relies on the specific characteristics of the agricultural catchment (i.e., size, topography, land use). However, to the best of our knowledge, the relative advantage of combining measures to reduce source inputs and the use of a buffer zone treating contaminant fluxes before they reach aquatic ecosystems has not yet been studied. The objectives of this study were as follows: (i) To evaluate the pollution reduction achieved when pesticide reduction at the source is combined with pesticide mitigation using a buffer zone for a given total cost TC, i.e., the cost of pesticide input reduction for the farmers (upstream cost) plus the cost of building a stormwater wetland at the catchment outlet for the public authorities (downstream cost); (ii) To measure the relative advantage of using buffer zones by evaluating how the total cost is shared between upstream and downstream. This study is in line with the European Water Framework Directive which obliges member states to reach good ecological status and therefore select the most efficient cost measures. The case study is a 42.7 ha vineyard catchment area in Alsace (eastern France), including 28.9 ha of vine plants (Fig. 1). Conventional winegrowers generally apply several fungicides during a growing season to limit the occurrence of fungal diseases such as powdery mildew, downy mildew and botrytis. Since the transport of runoff-related fungicides from the vineyards represent a significant threat to drinking water resources, human health (Israeli et al., 1983) and aquatic ecosystems, the present study focuses
Fig. 1. Scheme of the vineyard catchment (Rouffach, Alsace, France; 47◦ 57 9 N, 07◦ 17 3 E).
on fungicides as a representative class of pesticides used worldwide. The buffer zone considered is a stormwater wetland located at the catchment’s outlet, primarily built to temporarily collect runoff water from the vineyard catchment. The potential of the stormwater wetland to mitigate runoff-related fungicides has been demonstrated previously (Maillard et al., 2011). The study is structured in two major sections. In Section 2, the theoretical optima are calculated, including (i) the levels of pesticides applications which optimize the objective function with an upstream action only, and (ii) the levels of pesticides application and size of wetland, which optimize the objective function when upstream and downstream actions are combined. Moreover the model functions based on field observations are estimated. In Section 3, the empirical optima are calculated, before comparing the pesticide reduction obtained with the optimal combination of upstream and downstream actions for a given total cost TC to the pesticide reduction obtained with an abatement cost TC upstream only. Finally, the evolution of the distribution of the total cost between upstream and downstream is evaluated, with respect to changes in the total cost, while combining upstream and downstream actions. 2. Materials and methods 2.1. Theoretical equilibria 2.1.1. Notations and hypotheses We considered a system consisting of a vineyard catchment area upstream of a river composed of n winegrowers using fungicides and a public authority in charge of pollution regulation called: the regulator. The regulator is introduced in order to reduce the runoff-related mass of fungicides transported from the vineyard catchment to aquatic ecosystems downstream. The following theoretical part is valid for any type of farm and any diffuse pollutant. To achieve this goal, the regulator can use two complementary measures to reduce the export of fungicides from the catchment, namely the reduction of fungicide use directly at source, and the building of a buffer zone at the catchment outlet to mitigate fungicide runoff. Hence, the regulator considers the economic effort to be undertaken at source (abatement cost for the winegrowers) and at the catchment outlet (cost of the buffer zone), in order to combine for a given total cost, measures for reducing fungicide runoff. fi is the quantity of fungicides applied by a winegrower i. f i is the maximum quantity applied without regulation.
F. Destandau et al. / Ecological Engineering 60 (2013) 299–308
When winegrowers reduce fungicide use, the risk of losing part of their income due to grapevine diseases such as powdery mildew, downy mildew and botrytis increases. Winegrowers are therefore subject to a cost K, which is a random variable expressing the loss of benefits that mainly fluctuates as a function of meteorological and climatic conditions. The winegrowers are assumed to be risk-neutral and to incorporate the expectation of benefit loss following reduced fungicide use into their profit maximization. This upstream abatement cost, difference between the benefits for a maximum yield (obtained with a maximum quantity of fungicides) and benefits with the quantity of fungicides used, is:
301
derivative bS is cancelled. These hypotheses will be confirmed by the field data. The properties of the mass of fungicides M are: ∂M/∂S = MS = −˛ bS −˛ bSS
n
n
fi < 0 and
∂2 M/∂S 2 = MSS =
i=1
fi > 0
i=1
∂M/∂fi = Mfi = ˛ [1 − b] > 0
i (fi − fi ) where ∂i /∂fi = ifi < 0 and ∂2 i /∂fi2 = i fi fi > 0.
When the quantity of fungicides reaches its maximum fi , the upstream abatement cost is cancelled: i (f i ) = 0. A winegrower particularly afraid of the risk of benefit loss, or ‘risk-adverse’ would have a cost K larger than the benefit loss. During the period of fungicide application, the main portion of fungicides applied generally reaches the target area, whereas another portion of fungicides is transferred from agricultural lands to the atmosphere (loss by drift), or to subsurface water (leaching). Surface runoff is a major mechanism responsible for fungicide transfer from land to aquatic ecosystems. In an agricultural catchment area, buffer zones such as wetlands can intercept and partially retain runoff-related contaminants, thus limiting the contamination of water bodies. In the present study, ˛ represents the transfer coefficient, i.e., the portion of fungicides transferred by runoff according to Destandau and Nafi (2010). Therefore, 1 − ˛ (where ˛ ∈ [0, 1]) includes the portion of fungicides applied that reached its target and was not transported downstream by surface runoff. The transfer coefficient differs for each winegrower as a function the vineyard’s location. Since fungicide pollution is diffuse, the regulator cannot evaluate the transfer coefficients at the scale of each vineyard. Hence the regulator considers that the transfer coefficient for each winegrower is the global coefficient ˛ when formulating their policy. The regulator is assumed to be able to build a buffer zone of size S, which has the capacity to partly mitigate runoff-related fungicides at the outlet of the vineyard catchment. The annual cost of a buffer zone depends on its surface area, C(S), where ∂C/∂S = CS > 0 and ∂2 C/∂S 2 = CSS ≥0. The portion of fungicides mitigated through the buffer zone is called b, where b ∈ [0, 1]. b is referred to as the mitigation coefficient. The mitigation coefficient will depend on the size of the buffer zone and the different non-observable biotic and abiotic processes of null expectation, such as biodegradation, photolytic degradation (via UV light), hydrolytic degradation and sorption to wetland plants and sediments. Therefore, in the case of a buffer zone, the mass of fungicides M that will reach the aquatic ecosystem is: n
M=˛
∂2 M/∂fi2 = Mfi fi = 0
2
∂ M/∂S∂fi = MSfi = ∂2 M/∂fi ∂S = Mfi S − ˛bS < 0 As previously specified, k(), ˛ and b are random variables for which an average value has been chosen in the model. In fact, the production strategies of the winegrowers and the regulator’s policies can only be annual or multi-annual. The production strategies cannot be modulated as a function of the daily variations of factors that influence these variables because they often cannot be observed. Therefore, the production strategies of the winegrowers and the policy of the regulator must rely on a specific quantile of these random variables. However, the selection of a specific quantile has no effect on the calculations, results or analysis. For a risk-adverse winegrower, the cost induced by the reduction of fungicide use exceeds the expectation of lost benefits. Indeed, the increasing probability that their harvest will be affected by fungal diseases is a prejudice that must be added to the loss of expected benefits. For a risk-averse winegrower, function k() will therefore be more convex than that of a risk-neutral winegrower, as in our model. Confronted with the variability of two mitigation rates, (1−˛) and b, the regulator’s attitudes depend on their objectives. If the regulator focuses on the average quality of the river, as in the present model, and if the distributions of probability are symmetric, water quality may improve in 50% of cases, whereas it decreases in 50% of the other cases. If the regulator wants mass M to be exceeded in at most 30%, 20% or 10% of cases, the regulator voluntarily selects pessimistic mitigation rates in their model (1−˛) and b (quantiles of an order lower than 50%). A static model is used in order to focus on how the economic effort should be distributed between pollution sources and buffer zones treating pesticide runoff. In the future, a dynamic model could be used to account for the inter-annual mass of pesticide use and the mobilization and exports of fungicides used the previous year.
2.1.2. Theoretical equilibrium without a buffer zone The objective function of the regulator will be the minimization of M under the constraint of a total given cost TC:
fi 1 − b(S, )
i=1
min M
For a regulator dealing with average pollution: n
M=˛
fi [1 − b(S)]
i=1
The function b(S) is assumed to be concave: ∂b/∂S = bS > 0 and ∂2 b/∂S 2 = bSS < 0 so that a maximum size S exists for which the
fi
n fi
i=1
subject to n i=1
i = TC
(1)
302
F. Destandau et al. / Ecological Engineering 60 (2013) 299–308
which gives the Lagrangian: L1 = M
n fi
+ 1
i=1
n
The program accepts a minimum if
∂2 L2 /∂fi2 > 0 and Therefore, if
i − TC
∂2 L2 /∂fi2 = Mfi fi + 2 fi fi > 0,
i=1
where 1 is the coefficient of the Lagrangian expressing the weight of the constraint. The conditions of the first order are: ∂L1 /∂fi = Mfi + 1 fi = 0 ⇔ 1 = −Mfi /kfi > 0
2
∂2 L2 /∂fi2 × ∂2 L2 /∂S 2 − (∂2 L2 /∂fi ∂S) > 0
∀i ∈ [1, n]
(2)
this condition is still verified. And if: ∂2 L2 /∂fi2 × ∂2 L2 /∂S 2 − (∂2 L2 /∂fi ∂S)
2
= [Mfi fi + 2 fi fi ] [MSS − 2 CSS ] − (MSfI )2 > 0
(6)
n
∂L1 /∂1 =
i − TC = 0
This condition must be validated based on experimental field data.
i=1
At equilibrium, coefficient 1 is equal to the marginal cost of removing a unit of pollution from an aquatic system. To obtain a minimum, the condition of the second order is: ∂2 L1 /∂fi2 = Mfi fi + 1 fi fi > 0
(3)
The functions fi are bounded by a maximum value enabling corner solutions. The equilibrium above only corresponds to interior solutions. 2.1.3. Theoretical equilibrium with a buffer zone When the regulator envisages the transfer of part of the abatement effort downstream by building a buffer zone, the objective function becomes: minM( fi
n
fi , S)
(4)
i=1
Subject to n
i + C(S) = TC,
i=1
gives
which
2
n
the
Lagrangian:
L2 = M
n
fi , S
+
i=1
i + C(S) − TC]
i=1
The conditions of the first order are: ∂LMfi + 2 fi 2 /∂fi = 0;
∂L2 /∂S = MS − 2 CS = 0;
i + C(S) − TC = 0
2.2.1. Description of the vineyard catchment Fig. 1 shows the 42.7 ha vineyard catchment, including 28.9 ha of vine plants, and its outlet where contaminated runoff converges (bottom left) (Rouffach, Alsace, France; 47◦ 57 9 N, 07◦ 17 3 E). The characteristics of the catchment area and the agricultural practices were previously described in Grégoire et al. (2010). Briefly, fungicide is the major group of pesticides applied during the wine-growing season. Fungicides were mainly applied from the middle of April (grapevine bud opening) until the end of June (fruit setting). Azoxystrobin, cymoxanil, cyprodinil, dimethomorphe, kresoxim-methyl, metalaxyl, pyrimethanil and tetraconazole were selected for the present study because of their widespread use and the high frequency of application and detection revealed in previous studies (Grégoire et al., 2010). During the wine-growing season, approximately 1.030 kg of fungicides were applied on the vineyard catchment by 28 winegrowers. 2.2.2. Upstream abatement cost (at pesticide source) The abatement cost for each of the 28 winegrowers i corresponds to the expected loss of profit resulting from a reduction of fungicide use. The adaption of the cost function corresponding to a reduction of fungicide use at source is based on Bazoche et al. (2009), and the Agricultural Statistics of the French Ministry of Agriculture (2008) and the French Chamber of Agriculture of the Upper-Rhine department (2008). The explanation of the following function (7) is the object of Appendix A. (f i − fi ) = (45.6 − fi ) = 8.44.(45.6 − fi )
n
∂L2 /∂2 =
2.2. Empirical application: Specification of functions
(5)
+ 38(45.6 − fi ) D
2
∀i ∈ [1, n = 28]
(7)
i=1
The first two conditions give: 2 = −Mfi /fi = −MS /CS > 0 This equilibrium indicates that in the case of an interior solution (since fi and S are bounded, we can obtain corner solutions), the marginal cost of removing a unit of pollution from an aquatic ecosystem by pollution reduction directly at the source or by building a buffer zone is the same. The Hessian matrix of the conditions of the second order is written as
2 ∂ L2 2 ∂fi 2 ∂ L2
∂S∂fi
Mf f + 2 f f MSfi ii i i = ∂2 L2 MSfi MSS − 2 CSS 2
∂2 L2 ∂fi ∂S ∂S
where fi = 16.88fi − 807.73 < 0 and fi fi = 16.88 > 0 Since the origin of the diffuse pollutants cannot be established precisely, the above cost function is considered to be the same for each winegrower. The cost function enables the regulator to study the behavior of an average winegrower and therefore, the policy effects for the whole catchment area. 2.2.3. Description and cost of the stormwater wetland Runoff-related fungicides are collected at the outlet of the vineyard catchment by a stormwater wetland. Surface runoff represents the main entry route of fungicides into the wetland. The present study is based on a previous study evaluating the removal of runoff-associated fungicides by the wetland during the period from 4 April to 29 September 2009 (Maillard et al., 2011). The stormwater wetland has a surface area of 319 m2 and a total volume of 1500 m3 . During the wine-growing season, approximately
F. Destandau et al. / Ecological Engineering 60 (2013) 299–308
303
100% 90%
Mitigation coefficient
80% 70% 60% 50% 40% 30% 20%
Fig. 2. Schematic of the stormwater wetland (Rouffach, Alsace, France). Sediment deposition zone (SDZ), gabion barrier (G) and the gravel filter (GF).
10% 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 520 540
0%
Wetland size [m2]
2 g of the selected runoff-associated fungicides entered the wetland. Considering the amount of fungicides applied, this represents a fungicide transfer coefficient of 2‰. The stormwater wetland (Fig. 2) is composed of a sediment deposition forebay (215 m2 ) that collects the suspended solids in runoff and a gravel filter (13 m long, 8 m wide and 0.6 m deep). The main purpose of installing an artificial gravel filter is to increase hydraulic retention time and thus enhance the fungicide removal capacity of the wetland. Water losses by vertical infiltration were negligible due to the clay liner on the wetland bed (Ks < 1010 m s−1 ), verified by the water balance. Sediments and vegetation in the forebay area were removed in February 2009. Sediment is removed every four years and is part of the local maintenance scheme of the stormwater detention system. The vegetation cover in the sediment deposition zone was mainly composed of Phragmites australis, Juncus effusus and Typha latifolia. The investment and maintenance costs of this buffer zone are evaluated on the basis of expert opinion (cf. Table B.1, Appendix B). The cost function (8) is obtained by annualizing the costs so that they can be compared to the annual costs at the point source. The discount rate used (4%) is the base rate recommended by the French General Strategic Committee for public investments (Commissariat Général du Plan, 2005). C(S) = 40.5S + 1386 euros
(8)
Fig. 3. Mitigation coefficient according to wetland size.
a runoff event. The pesticide mitigation coefficient as a function of the stormwater wetland size is provided in Fig. 3. b(S) = 10−6 [−3S 2 + 3300S] %
where S ∈ [0; 550 m2 ] The maximum surface area S = 550 m2 corresponds to the size at which the mitigation coefficient ceases to increase. bfi = 0 bS = 10−6 [−6S + 3300] > 0 bSS = −6 × 10−6 < 0 2.2.5. Mass of fungicides found in the river The mass of fungicides M found in the aquatic systems can be deduced from the mitigation rate calculated in Section 2.2.4: n
M=˛
n
fi [1 − b(S)] = 0.002
i=1
MS = −˛ bS
fi [1 − 10−6 (−3S 2 + 3300S)] g
i=1
n
(10)
n
fi = −2 × 10−9 (−6S + 3300)
i=1
where S is the size of the buffer zone and where CS = 40.5 and CSS = 0. MSS = −˛ bSS 2.2.4. Pesticide mitigation using a stormwater wetland: Definition of the mitigation coefficient The relationship between pesticide mitigation using the stormwater wetland and its surface was established previously (Maillard et al., 2011). The latter study showed that the seasonal load mitigation of runoff-related fungicides through the stormwater wetland is 75%. Further estimates of this relationship were based on recent and related studies that emphasize the relationship between wetland size and pesticide mitigation of the wetland (Lange et al., 2010; Stehle et al., 2011). The extension of the wetland’s surface area is expected to enhance the capacity of fungicide mitigation for an equivalent fungicide load, since the sorption surface (plant and wetland sediment) and the water surface, both subject to UV light and thus to fungicide photolysis, are expected to increase proportionally. By increasing the surface area of the wetland, its pesticide mitigation capacity is expected to tend toward at least 90% during the fungicide application period. Consequently, the mitigation capacity of the stormwater wetland reaches a limit as a portion of the fungicide mixture may persist and be transported further through the stormwater wetland within the time-scale of
(9)
fi < 0
i=1
n
fi = 12 × 10−9
i=1
n
fi > 0
i=1
Mfi = ˛ [1 − b] = 0.002 [1 − 10−6 (−3S 2 + 3300 S)] > 0 Mfi fi = 0 MSfi = Mfi S = −˛ bS = −2 × 10−9 (−6S + 3300) < 0 3. Results and discussion 3.1. Empirical results 3.1.1. Empirical equilibrium without a stormwater wetland As seen in Section 2.1.2 (Eq. (2)), the interior solutions verify the condition of the first order: 1 = −Mfi /kfi =
−0.002[1 − 10−6 (−3S 2 + 3300S)] >0 (16.88fi − 807.73)
(11)
304
F. Destandau et al. / Ecological Engineering 60 (2013) 299–308
100% 2
Relative reduction of M
M with stormwater wetland
M [g]
M without stormwater wetland
1
0 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 520 540
The condition of the second order (Eq. (3)) is 2
∂ L1 /∂fi2 = Mfi fi + 1 fi fi = 16.881 > 0
(12)
3.1.2. Empirical equilibrium with a stormwater wetland The conditions of the first order equation defined in Section 2.1.3 (Eq. (5)) are rewritten as: ∂L2 /∂fi = Mfi + 2 fi = 0.002[1 − 10−6 (−3S 2 + 3300 S)] + 2 (16.88fi − 807.73) = 0 28
∂L2 /∂S = MS − 2 CS = −2 × 10−9 (−6S + 3300)
fi − 40.52 = 0
i=1
i + C(S) − TC =
28
i=1
70% 60% 50% 40%
30% 20%
0% 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 520 540
Fig. 4. Mass of pollution for a given total cost (cost at point source and cost of the buffer zone) according to total cost in the case where a stormwater wetland is present and when it is not.
n
80%
10%
Total abatement cost [Keuros]
∂L2 /∂2 =
90%
(13)
2
8.44 (45.6 − fi ) + 38 (45.6 − fi )
i=1
+ 40.5 S + 1386 − TC = 0 As defined in Section 2.1.3 (Eq. (6)), the above equilibrium corresponds to a minimum if: [Mfi fi + 2 fi fi ][MSS − 2 CSS ] − (MSfI )2 > 0 ⇔ [Mfi fi + 2 fi fi ][MSS − 2 CSS ] − (MSfI )2 = 05671.68 ×10−9 2 fi − 4 × 10−18 (−6S + 3300)2 > 0
Total abatement cost [Keuros] Fig. 5. Relative reduction of fungicides with a stormwater wetland, according to total abatement cost.
The relative reduction (Fig. 5) represents the difference of the fungicide mass in runoff reaching the catchment’s outlet with and without a wetland, divided by the mass without a wetland (see Appendix C). The relative reduction of fungicides can reach 90%. 3.1.4. Relative advantage of the stormwater wetland for pesticide mitigation The size of the stormwater wetland varies little at equilibrium (Table C.1 in Appendix C) while the total cost varies. It varies from 540 to 548 m2 for a total cost ranging from D 30k to D 500k, which emphasizes the low range of required wetland area. This is a positive point since if the abatement cost can vary from one year to the other, the choice of the size of stormwater wetland, once built, cannot be easily modified. As can be seen in Fig. 6, the consequence is that the downstream cost, i.e., the financial effort made for the wetland, is stable regardless of the global effort desired. Better environmental policy results will be achieved by increasing efforts at source. Because of the stability of the downstream cost at equilibrium, the more the total cost increases, the greater the financial effort will be upstream (see Fig. 7 for illustration). At a time when the European Water Framework Directive asks member states to make efforts or run the risk of being penalized collectively, this result underlines the issue of effort redistribution between pesticide users and the public authorities. In a context in which local topography causes polluted runoff water to converge at the outlet of an urban or agricultural catchment, building a stormwater wetland is a less costly method 600 Upstream cost [Keuros]
4 × 10
(−6S + 3300) 5671.68fi
500
2
(14)
3.1.3. Optimal combination according to the total cost Figs. 4 and 5 serve as an aid for better understanding the result based on the simulations (which details are presented in Appendix C). Interior solutions are obtained from a budget of D 30k upwards. For lower budgets, the mass of fungicides is minimized with a buffer zone without any effort at source. Fig. 4 shows the reduction of the mass of fungicides with and without a stormwater wetland for a given total cost (cost at point source and cost of the buffer zone). The presence of a wetland enable reducing the mass of fungicides significantly faster than in the case where there is no wetland (Fig. 4).
Downstream cost [Keuros] 400 300
200 100 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 520 540
⇔ 2 >
−9
Total abatement cost [Keuros] Fig. 6. Upstream and downstream cost at equilibrium.
F. Destandau et al. / Ecological Engineering 60 (2013) 299–308
305
Third, the mitigation capacity of the wetland has a limit over which the mitigation ceases to increase, and then over which upstream action becomes more efficient. Ribaudo et al. (2001) studied different scenarios of nitrogen reduction, but apparently not high enough to reach this limit. 4. Conclusion
Fig. 7. Distribution of the total abatement cost between upstream and downstream.
that enables significantly reducing pesticide transport to aquatic ecosystems. According to our practical case, the combination of upstream/downstream equilibrium is obtained for a wetland size close to that of maximum efficiency, regardless of the total effort invested. However, this policy may not be sufficient to reach significant pollution reductions. Reaching the objective of improving water quality will require greater efforts for reducing pesticide use at source. The use of a treatment wetland is complementary to the financial effort made at source. The latter finally remains the main variable in the decision-making process that has to be modulated in line with the ambition of the environmental policy. 3.2. Discussion The intensity of pollution reduction by the means of a constructed wetland depends on the pollutant and the local context. Mitsch (1992) found a retention of phosphorus between 63% and 96%, Wu et al. (2011): 68% for total nitrogen, 67% for total phosphorus, 93% for ammonium, and Mustafa et al. (2009): 97.6% for biochemical oxygen demand, 99% for ammonia–nitrogen, 74% for nitrate–nitrogen, 91.8% for molybdate reactive phosphorus. In this study, the reduction of fungicide runoff is 90% greater when pesticide reduction at source is combined with pesticide mitigation by a stormwater wetland compared to the case of pesticide reduction at source only. However, the advantage of the stormwater wetland decreases when it is closer to the lower and upper limits of the total cost (Fig. 5). This is in agreement with the results obtained by Destandau and Nafi (2010), indicating that the spatial discrimination of pollution regulation instruments provides less benefits for both very lax and very severe standards. Nonetheless, in the present study, the advantage decreases very close to the bounds of the total cost since the costs at source are much higher than the cost of the wetland. The latter remains very advantageous regardless of the total cost, with a 90% reduction of the pollution mass. Ribaudo et al. (2001) showed that the effort at source is more efficient with a nitrogen reduction goal below 26%, whereas the use of a wetland is more efficient beyond 26%. In the contrary, the conclusion of this study is that the more the total abatement cost is, the greater the more the effort necessary upstream will be (Fig. 7). The explanation is multiple. First, Ribaudo et al. did not combine both options that are exclusives. Second, when the quality objectives are few stringent, the fixed cost of the constructed wetland is determining. For a large catchment and then a large constructed wetland, that is the case in their study but not for the present, the fixed cost of wetland could make the upstream action less costly.
This article shows that combining complementary measures for reducing diffuse agricultural pollution may provide ecologicallysound results with the potential for improving water quality, by limiting pesticide transport from land to aquatic ecosystems. In this study, we evaluated the combination of pesticide mitigation measures to reduce pesticide pollution at source by limiting pesticide use and using stormwater wetlands, whereby limiting transport of runoff-related pesticide to aquatic ecosystems at the catchment’s outlet. This system involves the implementation of a local policy that can only be efficient if the topography allows building a buffer zone for collecting pesticide runoff at the catchment’s outlet. Although generalizing the large-scale and exclusive use of wetlands would not be efficient, our results show that the use of a stormwater wetland represents a complementary measure for reducing the use of pesticides in the framework of an ambitious pesticide reduction policy. This study also shows that, for the same cost, pesticide transport from land to aquatic ecosystems can be reduced by up to 90% of the mass of runoff-related fungicides compared to actions undertaken at source only. This underscores that more ambitious ecological objectives can be reached while complying with cost requirements. The potential added-value of a buffer zone, such as a stormwater wetland, must also be considered in relation to the fact that a compound not emitted at source will always provide a far greater ecological guarantee than one that is partly or totally mitigated in the environment. Indeed, little is known at present of the fate and the ecotoxicological impacts of pesticide degradation products in wetlands. In addition, buffer zones can also provide other amenities and functions (biodiversity pool, combating erosion, runoff water storage), which were not mentioned here. Acknowledgments The authors are members of REALISE, the Network of Laboratories in Engineering and Science for the Environment in the Alsace Region (France; http://realise.u-strasbg.fr), from which support is gratefully acknowledged. This research has been partly funded by the PhytoRET project (C.21) of the European INTERREG IV program Upper Rhine. The authors wish to thank the Agricultural and Viticulture College of Rouffach, the anonymouns reviewers for their help in improving the present article, and William J. Mitsch for his precious editing assistance. Appendix A. Within the framework of the “Wine and Environment” project (Bazoche et al., 2009), Leroy and Soler estimated the reduction of average yield when winegrowers decreased the quantity of fungicides (Table A.1). Table A.1 Yields as a function of the number of fungicide applications. Yields (100 kg/ha)
Number of fungicide applications
99% 92.5% 82.5% 37.5%
8 6 4 0
306
F. Destandau et al. / Ecological Engineering 60 (2013) 299–308
Table A.2 Construction of the upstream cost function. Yield (%)
Number of applications
38 45 51 57 63 68 73 78 82 85 88 91 94 95 97 98 99
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5
7 7.5 8
Quantity of fungicides (g)
Sales (D )
Fungicide costs (D )
Other costs (D )
Benefit (D )
Treatment (g)
Upstream treatment cost (D )
0
13,248 15,711 18,025 20,189 22,204 24,069 25,784 27,350 28,766 30,032 31,149 32,116 32,934 33,602 34,120 34,489 34,708
0 24 48 71 95 119 143 167 191 214 238 262 286 310 334 357 381
17,107 17,314 17,509 17,690 17,860 18,016 18,160 18,292 18,411 18,517 18,611 18,692 18,761 18,817 18,861 18,892 18,910
−3860 −1627 469 2427 4249 5933 7481 8891 10,164 11,301 12,300 13,162 13,887 14,475 14,926 15,240 15,417
46 43 40 37 34 31 29 26 23 20 17 14 11 9 6 3 0
19,277 17,044 14,949 12,990 11,168 9484 7937 6526 5253 4117 3117 2255 1530 942 491 177 0
2.9 5.7 8.6 11.4 14.3 17.1 20.0 22.8 25.7 28.5 31.4 34.2 37.1 39.9 42.8 45.6
As in Fig. A.1, we use the following function for each winegrower:
20
Treatment Cost [Keuros]
18
y = 8.44 x2 + 38 x R2 = 1
16
(fi − fi ) = (45.6 − fi ) = 8.44.(45.6 − fi )
2
+ 38(45.6 − fi ) ∀i ∈ [1, n = 28]
14 12 10
Appendix B.
8
To calculate the opportunity cost, we based ourselves on the benefits withdrawn from the winegrowing activity in this catchment area, which we reduced to the scale of the wetland (Table B.1).
6 4
2
Appendix C.
0 0
5
10
15
20
25
30
35
40
45
50
Treatment [g]
The following results (Table C.1) were obtained from conditions of the first order using an Excel solver.
Fig. A.1. Upstream treatment (abatement) cost function.
Approximately 1030 g of fungicides were spread upstream over the Rouffach catchment area every year by 28 winegrowers assumed to spread the same amount. Thus, each winegrower spread 37 g of fungicides. The average yield of the winegrowers was 95% in 2008 (source: statistics of the French Ministry of Agriculture: http://agreste.maapar.lbn.fr), which corresponds on average to 6.5 applications of fungicide. We therefore assumed that each winegrower used 5.7 g of fungicide per application.Average turnover for wine production in the Haut-Rhin department (France) is D 293,000k for 9000 ha (source: statistics of the Chamber of Agriculture of the Haut-Rhin: http://www.haut-rhin.chambagri.fr). We therefore estimated that turnover was D 33,602 per winegrower on the 28.9 ha of our catchment area (i.e., a little more than 1 ha per winegrower). Consequently, a yield equal to 95% was equivalent to a turnover of D 33,602. To estimate the winegrowers’ profit, we deducted costs from turnover. According to expert opinion, fungicides cost each winegrower D 47.6 per application. The other costs represent an average of 56% of the turnover: 85% are fixed and 15% are proportional to yield. Thus, in our study, the upstream abatement cost is the difference between the benefits for a maximum yield (obtained with a maximum quantity of fungicides: 45.6 g per winegrower) and benefits with the quantity of fungicides used (see Table A.2).
-
“TC” is the Total Cost: Upstream and downstream cost (k D ). “fi ” is the quantity of fungicides (g) applied by i. “S” is the wetland surface area (m2 ) “lambda1 ‘ and “lambda2 ‘ correspond to −Mfi /fi and −MS /CS , respectively, defined in Section 3.3, multiplied by 1010 . These two values are the same for the interior solutions. We find interior solutions for TC ≥ D 30k. Below this, we obtain corner solutions with fi = f i .
- The “2nd order” corresponds to 4 × 10−9 (−6S + 3300)2 /567.68fi multiplied by 1010 . It should be recalled that, as shown in Section 4.2, our interior solutions are minimums if 2 > 4 × 10−9 (−6S + 3300)2 /567.68fi . This is verified for all our interior solutions. - “M with” expresses the mass obtained with the stormwater wetland, and “M without” the mass without the stormwater wetland. - “fi without” is the individual quantity of pollutant applied without the stormwater wetland. - “% M” is the relative reduction of M calculated as follows: (M without − M with)/M without. - “ups. cost” and “downs. cost” are the upstream cost and the downstream cost, respectively (k D ). - “upstream” and “downstream” are the share of the total cost devoted to abatement at source and to mitigation downstream, respectively.
F. Destandau et al. / Ecological Engineering 60 (2013) 299–308
307
Table B.1 Cost of a stormwater wetland built for mitigation (D ).. Fixed cost
Cost/square meter
Life expectancy
10
30
Discount rate 4% Fixed cost/year
Price of the land Opportunity cost Gravel barrier Gravel Planting Removal of sediments Plant management Study and emptying Earthwork Waterproofing Laying out of the stormwater wetland
5000
750
0.6 1.4 7.5 7.5
15 10 59 0.00575
5725 59 52.5 223
Variable cost/year
785
3 3 3 30 30 30 30
0.6 2.7 16 0.002
270 331
3.4 3 13 40.5
1386 Table C.1 Results of the simulation (Excel solver). TC
fi
S
– 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 390 400 410 420 430 440 450 460 470 480 490 500 510 520 530 540
46 46 46 42 39 37 35 34 32 31 30 29 28 27 26 25 24 23 22 21 20 20 19 18 18 17 16 15 15 14 14 13 12 12 11 11 10 9 9 8 8 7 7 6 6 5 5 4 4 3 3 2 2 1
0 213 460 545 546 547 547 547 547 547 548 548 548 548 548 548 548 548 548 548 548 548 548 548 548 547 547 547 547 547 547 547 547 547 547 547 547 546 546 546 546 546 545 545 545 544 544 543 542 541 540 538 535 530
lambda1
19,146 12,672 10,132 8685 7720 7019 6479 6048 5692 5393 5136 4913 4717 4542 4386 4244 4116 3998 3890 3791 3698 3612 3532 3457 3386 3320 3257 3198 3142 3089 3038 2990 2944 2900 2859 2818 2780 2743 2708 2673 2641 2609 2579 2550 2522 2495 2469 2445 2423 2404 2392
lambda2
19,146 12,672 10,132 8685 7720 7019 6479 6048 5692 5393 5136 4913 4717 4542 4386 4244 4116 3998 3890 3791 3698 3612 3532 3457 3386 3320 3257 3198 3142 3089 3038 2990 2944 2900 2859 2818 2780 2743 2708 2673 2641 2609 2579 2550 2522 2495 2469 2445 2423 2404 2392
2nd order
2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 3 3 3 4 5 6 7 8 10 13 16 22 29 41 60 92 155 292 661
M with
fi without
M without
%M
ups. cost
down. cost
upstream (%)
downstream (%)
1.11 0.30 0.22 0.20 0.19 0.18 0.17 0.17 0.16 0.15 0.15 0.14 0.14 0.13 0.13 0.12 0.12 0.11 0.11 0.11 0.10 0.10 0.09 0.09 0.09 0.08 0.08 0.08 0.07 0.07 0.07 0.06 0.06 0.06 0.06 0.05 0.05 0.05 0.04 0.04 0.04 0.04 0.03 0.03 0.03 0.02 0.02 0.02 0.02 0.01 0.01 0.01 0.01 0
46 41 38 36 35 33 32 30 29 28 27 26 25 24 23 23 22 21 20 19 19 18 17 17 16 15 15 14 13 13 12 12 11 10 10 9 9 8 8 7 7 6 6 5 5 4 4 3 3 2 2 1 1 0 0
2.29 2.15 2.04 1.94 1.86 1.78 1.71 1.64 1.58 1.52 1.46 1.41 1.36 1.31 1.26 1.22 1.17 1.13 1.09 1.05 1.01 0.97 0.93 0.89 0.85 0.82 0.78 0.75 0.71 0.68 0.65 0.62 0.58 0.55 0.52 0.49 0.46 0.43 0.40 0.37 0.34 0.32 0.29 0.26 0.23 0.21 0.18 0.15 0.13 0.10 0.08 0.05 0.02 0
52 86 89 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 89 89 89 89 89 89 89 89 89 89 88 88 88 87 87 86 85 83 80 69
0 0 7 16 26 36 46 56 66 76 86 96 106 116 126 136 146 156 166 176 186 196 206 216 226 236 246 256 266 276 286 296 306 316 326 336 346 356 367 377 387 397 407 417 427 437 447 457 467 477 487 497 507 540
10 20 23 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 23 23 23 23 23 23 23 23 23 23 23 23 23 23 0
0 0 22 41 53 61 66 71 74 76 79 80 82 83 84 85 86 87 88 88 89 89 90 90 91 91 91 92 92 92 92 93 93 93 93 93 94 94 94 94 94 94 95 95 95 95 95 95 95 95 95 96 96 100
100 100 78 59 47 39 34 29 26 24 21 20 18 17 16 15 14 13 12 12 11 11 10 10 9 9 9 8 8 8 8 7 7 7 7 7 6 6 6 6 6 6 5 5 5 5 5 5 5 5 5 4 4 0
308
F. Destandau et al. / Ecological Engineering 60 (2013) 299–308
Appendix D. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ecoleng. 2013.07.030. References Agricultural Statistics of the French Ministry of Agriculture, 2008. http://agreste.maapar.lbn.fr Bazoche, P., Cartolaro, P., Deliere, L., Del’Homme, B., Hernandez, F., Goutouly, J.-P., Léger, B., Leroy, P., Naud, O., Soler, L. G., Ugaglia, A., 2009. Vin et environnement: comment réduire les quantités de pesticides utilisés en viticulture? Synthèse du projet ANR Agriculture et Développement Durable (2005–2008). Budd, R., O’Geen, A., Goh, K.S., Bondarenko, S., Gan, J., 2009. Efficacy of constructed wetlands in pesticide removal from tailwaters in the Central Valley, California. Environ. Sci. Technol. 43, 2925–2930. Byström, O., 2000. The replacement value of wetlands in Sweden. Environ. Resour. Econ. 16, 347–362. Byström, O., Andersson, H., Gren, I.M., 2000. Economic criteria for using wetlands as nitrogen sinks under uncertainty. Ecol. Econ. 35, 35–45. Byström, O., 1998. The nitrogen abatement cost in wetlands. Ecol. Econ. 26, 321–331. Chamber of Agriculture of the Upper-Rhine department (France), 2008. http://www.haut-rhin.chambagri.fr Commissariat Général du Plan, 2005. Révision du taux d’actualisation des investissements publics. Rapport du groupe d’experts présidé par Daniel Lebègue, 21 janvier. Crépin, A.S., 2005. Incentives for wetland creation. J. Environ. Econ. Manage. 50, 598–616. Destandau, F., Nafi, A., 2010. What is the best distribution for pollution abatement efforts? Information for optimizing the WFD Programs of measures. Environ. Resour. Econ. 46 (3), 337–358. Elofsson, K., 2010. The costs of meeting the environmental objectives for the Baltic Sea: a review of the literature. AMBIO 39, 49–58. European Commission, 2000. Directive 2000/60/EC of the European Parliament and of the Council Establishing a Framework for Community Action in the Field of Water Policy. OJ L 327, 22.12.2000. European Commission, 2012. The Third Implementation Report–Assessment of the River Basin Management Plans. http://ec.europa.eu/environment/ water/water-framework/implrep2007 Fournel, J., Forquet, N., Molle, P., Grasmick, A., 2013. Modeling constructed wetlands with variably saturated vertical subsurface-flow for urban stormwater treatment. Ecol. Eng. 55, 1–8. Grégoire, C., Payraudeau, S., Domange, N., 2010. Use and fate of 17 pesticides applied on a vineyard catchment. Int. J. Environ. Anal. Chem. 90, 406–420. Grégoire, C., Elsaesser, D., Huguenot, D., Lange, J., Lebeau, T., Merli, A., Mose, R., Passeport, E., Payraudeau, S., Schutz, T., Schulz, R., Tapia-Padilla, G., Tournebize, J., Trevisan, M., Wanko, A., 2009. Mitigation of agricultural nonpoint-source pesticide pollution in artificial wetland ecosystems: an interdisciplinary approach in the EU ArtWET project. Environ. Chem. Lett. 7 (3), 205–231. Hanley, N., Colombo, S., Tinch, D., Black, A., Aftab, A., 2006. Estimating the benefits of water quality improvements under the Water Framework Directive: are benefits transferable? Eur. Rev. Agric. Econ. 33 (3), 391–413.
Hatvani, I.G., Kovács, J., Székely, K.I., Jakusch, P., Korponai, J., 2011. Analysis of long-term water quality changes in the Kis-Balaton Water Protection System with time series-, cluster analysis and Wilks’ lambda distribution. Ecol. Eng. 37, 629–635. Heberling, M.T., García, J.H., Thurston, H.W., 2010. Does encouraging the use of wetlands in water quality trading programs make economic sense? Ecol. Econ. 69, 1988–1994. Holland, J.F., Martin, J.F., Granata, T., Bouchard, V., Quigley, M., Brown, L., 2005. Analysis and modeling of suspended solids from high-frequency monitoring in a stormwater treatment wetland. Ecol. Eng. 24, 157–174. Imfeld, G., Lefrancq, M., Maillard, E., Payraudeau, S., 2013. Transport and attenuation of dissolved glyphosate and AMPA in a stormwater wetland. Chemosphere 90 (4), 1333–1339. Israeli, R., Sculsky, M., Tiberin, P., 1983. Acute intoxication due to exposure to maneb and zineb. Scand. J. Work, Environ. Health 9, 47–51. Lange, J., Schuetz, T., Grégoire, C., Elsässer, D., Schulz, R., Passeport, E., Tournebize, J., 2010. Multi-tracer experiments to characterize contaminant mitigation capacities for different types of artificial wetlands. Int. J. Environ. Anal. Chem. 91, 768–785. Lefrancq, M., Imfeld, G., Payraudeau, S., Millet, M., 2013. Kresoxim methyl deposition, drift and runoff in a vineyard catchment. Sci. Total Environ. 442, 503–508. Lindhal, T., Söderqvist, T., 2004. Building a catchment-based environmental programme: a stakeholder analysis of wetland creation in Scania, Sweden. Regional Environ. Change 4, 132–144. Maillard, E., Payraudeau, S., Faivre, E., Grégoire, C., Gangloff, S., Imfeld, G., 2011. Removal of pesticide mixtures in a stormwater wetland collecting runoff from a vineyard catchment. Sci. Total Environ. 409, 4317–4324. Mitsch, W.J., 1992. Landscape design and the role of created, restored and natural riparian wetlands in controlling non-point source pollution. Ecol. Eng. 1, 27–47. Mustafa, A., Scholz, M., Harrington, R., Carroll, P., 2009. Long-term performance of a representative integrated constructed wetland treating farmyard runoff. Ecol. Eng. 35, 779–790. Ockenden, M.C., Deasy, C., Quinton, J.N., Bailey, A.P., Surridge, B., Stoate, C., 2012. Evaluation of field wetlands for mitigation of diffuse pollution from agriculture: sediment retention cost and effectiveness. Environ. Sci. Policy 24, 110–119. Paulsen, S., 2007. Investment in Wetlands for Pollution Abatement Under Uncertainty, Topics on the Ecological Economics of Coastal Zones. Doctoral Dissertation. Dept. of Economics, SLU, Acta Universitatis agriculturae Sueciae. Poissant, L., Beauvais, C., Lafrance, P., Deblois, C., 2008. Pesticides in fluvial wetlands catchments under intensive agricultural activities. Sci. Total Environ. 404 (1), 182–195. Ribaudo, M., Heimlich, R., Claassen, R., Peters, M., 2001. Least-cost management of nonpoint source pollution: source reduction versus interception strategies for controlling nitrogen loss in the Mississippi Basin. Ecol. Econ. 37, 183–197. Söderqvist, T., 2002. Constructed wetlands as nitrogen sinks in southern Sweden: an empirical analysis of cost determinants. Ecol. Eng. 19, 161–173. Stehle, S., Elsaesser, D., Gregoire, C., Imfeld, G., Niehaus, E., Passeport, E., Payraudeau, S., Schäfer, R.B., Tournebize, J., Schulz, R., 2011. Pesticide risk mitigation by vegetated treatment systems: a meta-analysis. J. Environ. Qual. 40 (4), 1068–1080. Wu, H., Zhang, J., Li, P., Zhang, J., Xie, H., Zhang, B., 2011. Nutrient removal in constructed microcosm wetlands for treating polluted river water in northern China. Ecol. Eng. 37, 560–568.