- Email: [email protected]
A field study of the impact of different irrigation practices on herbicide leaching
Europ. J. Agronomy 32 (2010) 280–287
Contents lists available at ScienceDirect
European Journal of Agronomy journal homepage: www.elsevier.com/locate/eja
A field study of the impact of different irrigation practices on herbicide leaching Gabriella Fait a,∗ , Matteo Balderacchi a , Federico Ferrari a , Fabrizio Ungaro b , Ettore Capri a , Marco Trevisan a a b
Istituto di Chimica Agraria ed Ambientale, Università Cattolica del Sacro Cuore, via Emilia Parmense 84, 29100 Piacenza, Italy Consiglio Nazionale delle Ricerche, Istituto di Ricerca per la Protezione Idrogeologica, U.O.S. di Firenze Pedologia Applicata, Sesto F.no, Italy
a r t i c l e
i n f o
Article history: Received 12 November 2009 Received in revised form 29 January 2010 Accepted 1 February 2010 Keywords: Terbuthylazine Desethylterbuthylazine Maize Modelling MACRO
a b s t r a c t Agricultural practices, such as subsurface drainage, irrigation and tillage, may significantly affect pesticide leaching and, consequently, the risk of groundwater contamination. The aim of the present study was to investigate the impact of different irrigation systems on herbicide leaching to shallow groundwater through direct monitoring at the field scale in northern Italy over a 3-year period. Concentrations of the herbicide terbuthylazine (TBA) and its metabolite desethylterbuthylazine (DES) were monitored on 10 farms cropped with maize and irrigated by sprinkler, basin and border systems. Considering the results grouped according to the different irrigation systems, the mean TBA and DES concentrations was lower than the arbitrary non-health based legal limit of 0.1 g/L using sprinkler and border systems, while it was 0.19 and 0.30 g/L respectively for TBA and DES using basin systems. However, since many factors other than the irrigation systems can contribute to pesticide leaching and in a field study it is impossible to discriminate between all the different variables, the concentrations of both compounds were simulated with and without irrigation using the model MACRO 5.1 in order to gain a deeper understanding of the role of irrigation on leaching. First, the groundwater table depth, which was measured daily in all fields, was used to calibrate the model and thus achieve a good soil hydrology calibration. To assess the performance of the model the root mean squared error (RMSE) was used. RMSE ranged from 0.2 to 0.5 m, showing that a satisfactory hydrology calibration was obtained. Afterward, the solutes were modelled and the results showed that under non-irrigated conditions, concentrations of both compounds would be very low. These findings validate the hypothesis that careful selection of agricultural practices, such as the type of irrigation, can reduce pesticide leaching. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Pesticide occurrence in groundwater is controlled by many factors such as the intensity of use, soil properties and pesticide properties (Barbash et al., 2001). In addition certain agricultural practices, such as tillage, the use of subsurface drains and irrigation, may appreciably contribute to pesticide leaching and subsequently to groundwater contamination (Gilliom et al., 2007). In this study, attention was focused on irrigation. An adequate water supply is critical for plant growth, and various methods can be used to supply water to plants. These different irrigation techniques influence water flow patterns in the soil (Bandaranayake et al., 1998) and solute movement. The aim of this study was to investigate the impact of different irrigation systems on pesticide leaching to shallow groundwater by direct monitoring at the field scale over a 3-year period. Sprinkler, border and basin irrigation systems were considered.
∗ Corresponding author. Tel.: +39 0523 599 345; fax: +39 0523 599 217. E-mail address: [email protected] (G. Fait). 1161-0301/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.eja.2010.02.001
As water-quality monitoring shows that herbicides are the most frequently detected group of pesticides in ground and surface water (Carter, 2000), the herbicide terbuthylazine [N2 -tert-butyl6-chloro-N4 -ethyl-1,3,5-triazine-2,4-diamine] (TBA) was chosen. TBA belongs to the triazine group of herbicides, which is amongst the most frequently used group for selective weed control in several crops (Barra Caracciolo et al., 2005). In Italy TBA uses are now limited to maize and sorghum according to the implementation of the Environmental Stewardship Programme as agreed by the TBA manufactures. The major dealkylation product of TBA is desethylterbuthylazine (DES) (Dousset et al., 1997). Dealkylated products are generally more persistent and water soluble, and therefore pose a risk of groundwater contamination (Guzzella et al., 2003). Thus DES concentrations were also monitored. The focus of this study was to investigate the effect of irrigation on TBA and DES leaching to shallow groundwater in 10 farms located in the northern Italy selected to cover a range of different soil types and climatic conditions. Such field studies are helpful in order to better understand how agricultural practices affect pesticide leaching under real farm situations; however it is impossible to carry out these types of studies under identical conditions, so
G. Fait et al. / Europ. J. Agronomy 32 (2010) 280–287
the effect of irrigation systems on pesticide leaching could be confounded by other factors. For this reason, the solute transport model MACRO 5.1 (Larsbo and Jarvis, 2003) was used to simulate TBA and DES leaching under both irrigated and non-irrigated conditions. Measured parameters (i.e. soil, climatic, pedological and hydrological data) were used to set a scenario for each farm. First, the model was run using real irrigation data. If a good fit between simulated and measured concentration is attained, it is possible to investigate the effect of irrigation on leaching by running the model without the input of water coming from irrigation. In this way, the effect of the irrigation system can be isolated and then it is possible to examine further the effect of differing irrigation systems on pesticide leaching under different situations. 2. Materials and methods 2.1. Study design The 10 farms (Table 1) were located in the Po Valley of northern Italy (Fig. 1), which is an important agricultural area. The farms were cropped with maize and managed according to good agricultural practices. Two farms were irrigated by the basin system, two by border irrigation, one was not irrigated and the remaining farms were irrigated by sprinklers. On each farm, a field approximately 10 ha in size was monitored. Before starting the monitoring, each field was assessed for the absence of surface-water bodies, the uniform direction of groundwater flux, slope (which should not exceed 5%), and the absence of cracking soil, perched groundwater and tile drainage. The groundwater depth was shallow, but varied amongst the 10 fields (Table 1). Three piezometers were installed in each field, placed at the edge of the field in the direction of down-gradient groundwater flow in order to maximize interception of shallow groundwater flow leaving the field. The piezometers were made of PVC pipes, with a rubber plug inserted at the base to prevent backflow and a cast-iron manhole cover above each piezometer. A layer of bentonite pellets was placed outside and inside the manhole cover to act as a hydraulic insulating agent, thereby minimizing potential contamination from the percolation of preferential fluxes. Meteorological data (Table 2) were collected on each farm from a digital remote-controlled station, which recorded minimum and maximum temperature, precipitation and groundwater level on an hourly basis. On farms Ba1 and Ba2, the measurements came from the regional meteorological station closest to the farms, provided by the regional agency for the protection of the environment (ARPA Lombardia). 2.2. Irrigation practices and field management Irrigation is a common practice in the Po Valley in order to ensure maize production. Only in some areas in the Friuli region it is possible to obtain satisfactory yields without irrigation. The type of irrigation system depends upon the crop, landscape and local customs. The most recent Italian agricultural census (ISTAT, 2000) showed that sprinkler and border systems are the most common irrigation systems. Basin irrigation is widely used in the North of Italy for rice crop and then for the crops in rotation with rice, such as maize. Information on irrigation practices (Table 1), such as the number of applications and the volume of water applied, was provided by the farmers. All the fields were cropped with maize for the entire study period. TBA was applied pre-emergence annually as the formulation BoleroTZ® (Monsanto, containing 214 g ai/L) and was applied at 4 L/ha. Spray-bar calibration was performed prior
281
to each application, and each piezometer was protected with a plastic film during application in order to avoid any direct contamination. 2.3. Water sampling and analysis Groundwater was sampled by applying low negative pressure. The depth of the piezometer screen is reported in Table 1. Each piezometer was purged prior to sampling in order to remove standing water. The sample volume (approximately 1 L) was collected in glass bottles, being stored immediately in a portable freezer and then transferred to a refrigerator at 4 ◦ C prior to analysis. Samples were collected every 2 months from April 2005 to December 2007. The water table depth within each piezometer was measured at each sampling event. Less than 1 month elapsed between sample collection and analysis. Water samples were analysed measuring both TBA and DES concentrations according to the SPE–GC–MS method presented in Pichon (2000). The analytes were identified through: (i) comparison between the retention time of the samples and the analytical standards (purity > 99%), (ii) comparison of the mass spectra of the samples and the analytical standards, and (iii) comparison of the mass spectrum in the NIST library with that of the samples. Quantification was performed in single-ion monitoring mode. Concentrations of each analyte were calculated using an external calibration curve, the limit of quantification in the water samples being 5 ng/L. 2.4. Evaluation of the hydraulic transit times beneath the fields In order to assess the hydraulic transit times beneath each field, the horizontal movement of water was estimated using “Darcy’s Law”: q=K ×i where K is the hydraulic conductivity (m/s) and i is the hydraulic gradient, which is equal to the hydraulic head difference (H) between two points in the soil divided by the distance between those points (L) (m). Furthermore, the q value was divided by the porosity in order to take into account that only the interstitial spaces in the soil are able to conduct water. Based on estimated K values from the literature (Bear, 1972) and measured hydraulic heads and K values calculated using the pedotransfer function (PTF) proposed by Wösten et al. (1999) and site-specific data, lateral groundwater flow velocities were estimated using Darcy’s law to vary between 0.3 and 98.3 m/in 2 months, and so the sampling frequency (2 months) was considered to be adequate (Table 3). 2.5. Model description MACRO (version 5.1) is a comprehensive mechanistic onedimensional non-steady state model of water flow and solute transport in structured or macroporous soils (Larsbo and Jarvis, 2003). This model was chosen because such soils are typical in the Po Valley. The model accounts for macropore flow, with the soil porosity divided into two flow domains (macropores and micropores) each characterized by a flow rate and solute concentration, with the boundary between them defined by a fixed water potential associated with a saturated matrix water content and hydraulic conductivity (Stenemo and Jarvis, 2007). Each domain has its own pressure head and solute concentration. Richards’ equation and the convection–dispersion equation are used to model soil water flow and solute transport in the soil micropores, while in the macropores
282
G. Fait et al. / Europ. J. Agronomy 32 (2010) 280–287
Table 1 Field characteristics. Field
Field area (ha)
Topsoil texture class (USDA)
pH
Topsoil OC %
Groundwater table depth 2005–2007 (m)
Piezometer screen depth (m)
Irrigation system
Number of irrigations (per year)
Total amount of water (mm)
Clay loam
7.5
1.3
0.9–2.8
4.50
Sprinkler
3 (2005) 7 (2006) 7 (2007)
120 280 280
S1
8.5
S2
12.0
Sandy loam
5.7
1.1
0.9–1.8
4.90
Sprinkler
6 (2005) 5 (2006) 7 (2007)
120 110 126
S3
20.0
Loam
7.6
0.7
1.1–2.8
4.80
Sprinkler
1 (2005) 1 (2006) 3 (2007)
30 50 150
S4
3.6
Sandy loam
5.5
1.1
0.3–2.7
3.70
Sprinkler
1 (2005) 2 (2006) 2 (2007)
30 60 60
S5
12.0
Loam
7.9
0.9
0.5–1.8
4.20
Sprinkler
3 (2005) 5 (2006) no (2007)
90 125 –
Bo1
15.0
Sandy loam
5.4
1.0
5.3–6.7
8.00
Border
2 (2005) 3 (2006) 3 (2007)
120 180 180
Bo2
3.7
Sandy loam
7.2
2.6
4.4–5.4
6.50
Border
1 (2005) 2 (2006) 2 (2007)
60 120 120
Ba1
9.1
Sandy clay
7.2
1.2
3.8–5.8
6.30
Basin
4 (2005) 6 (2006) 6 (2007)
280 420 420
Ba2
12.3
Sandy loam
6.3
1.7
3.6–6.9
4.60
Basin
3 (2005) 5 (2006) 5 (2007)
210 350 350
NoIrr
10.5
Loam
7.8
0.9
0.7–3.9
5.20
Not irrigated
– – –
– – –
Fig. 1. Farms location. The farms are indicated by the black dots (S1–S5: sprinkler irrigation; Ba1, Ba2: basin irrigation; Bo1, Bo2: border irrigation; NoIrr: not irrigated).
G. Fait et al. / Europ. J. Agronomy 32 (2010) 280–287
283
Table 2 Average annual rainfall (mm) at each farm. Field
Location
Rainfall from April to December 2005 (mm)
Annual rainfall in 2006 (mm)
Annual rainfall in 2007 (mm)
S1 S2 S3 S4 S5 Bo1 Bo2 Ba1 Ba2 NoIrr
Lombardia Lombardia Veneto Veneto Friuli-Venezia Giulia Piemonte Piemonte Lombardia Lombardia Emilia Romagna
752 662 1060 1089 832 561 561 716 654 544
625 595 783 573 711 663 663 779 525 637
437 785 745 391 965 349 349 689 531 438
a modified kinematic wave approach is applied to calculate fluxes and solute transport is described with an advective flow solute, neglecting dispersion. The solute concentration in water flowing into the macropores is calculated assuming complete mixing in a shallow mixing depth at the soil surface. A modified form of the van Genuchten function is used to describe the water retention function. A Freundlich isotherm is used to describe pesticide sorption, and first-order kinetics is assumed for degradation, with a rate coefficient given as a function of soil temperature and moisture content. Water and solute exchange between the flow domains are calculated using approximate first-order expressions based on an effective diffusion path length. In MACRO there are four different options for the bottom boundary condition (Larsbo and Jarvis, 2003). In the present study groundwater table in the profile was chosen. 2.6. Model parameterization The required inputs for MACRO parameterization include: soil physical and hydraulic parameters, crop, irrigation, solute characteristics and weather data. The soil physical and hydraulic parameters govern retention and transport of water and chemicals, but are difficult to measure and so are often derived using PTFs, which relate these model parameters to more easily measured soil properties such as soil texture, organic
matter content and/or other routine measurements. In the present study the following properties were analysed on a soil core sampled from each field: texture, pH, organic carbon (OC), and horizon depth. Bulk density was not available, and so it had to be estimated, even if this kind of approach using a double pedotransfer is a last option to estimate hydrological data. Bulk density was calculated with the software SOILPAR 2.00 (Acutis and Donatelli, 2003) using the method described by Rawls and Brakensiek (1989). The parameters of the van Genuchten soil–water retention function were calculated using the PTFs developed by Ungaro et al. (2005) for Po Valley soils. The total saturated hydraulic conductivity was calculated using the PTFs developed by Wösten et al. (1999), as stated above. Information about the crop, such as harvesting and seeding periods, pesticide application and irrigation, was provided by the farmers. The other information required by the model in order to parameterize the crop, such as day of maximum leaf area/root depth, root distribution, canopy interception capacity, etc., came from the FOCUS (FOrum for Co-ordination of pesticide fate models and their Use) Groundwater Scenarios Workgroup (FOCUS, 2000). The FOCUS groundwater scenarios are a set of nine standard combinations of weather, soil and cropping data which collectively represent agriculture in the EU for the purposes of a Tier 1 EU-level assessment of leaching potential. In the present study information for maize crop for Piacenza scenario was taken into account.
Table 3 Estimates of horizontal water flow. Field
K for bottom layera (m/s)
Corresponding soil textural type for K for bottom layer (USDA limits)a
H (hydraulic head) (m)
S1
0.00013 0.0001
Sand Sand
0.2
270
1.2 0.9
S2
0.00034 0.0001
Sand Sand
0.8
420
9.6 2.8
S3
0.00225 0.01
Sandy silt–clay silt Gravel
0.8
270
98.3 437.7
S4
0.00084 0.0001
Sandy loam Sand
0.6
255
28.2 3.4
S5
0.00026 0.0001
Sandy–silty sandy Sand
0.05
1130
0.2 0.1
Bo1
0.00005 0.0001
Silty sandy–silty clay Sand
1.4
415
2.2 4.5
Bo2
0.00003 0.0001
Sand Sand
0.4
105
1.7 5.0
Ba1
0.00006 0.00001
Loam Sand
0.2
400
0.8 1.4
Ba2
0.00009 0.0001
Sandy loam Sand
0.1
290
0.3 0.3
NoIrr
0.00003 0.0001
Silty clay Sand
0.6
420
0.6 1.7
a
L (field length) (m)
q (m in 2 months)
The values coming from the literature are reported in normal type, while the values coming from the PTF and site-specific data are reported in bold type.
284
G. Fait et al. / Europ. J. Agronomy 32 (2010) 280–287
Table 4 Concentrations of TBA and DES (g/L). Field
S1 S2 S3 S4 S5 Bo1 Bo2 Ba1 Ba2 NoIrr
Median
90th
Mean
Maximum
TBA
DES
TBA
DES
TBA
DES
TBA
DES
<0.005 0.01 <0.005 <0.005 <0.005 <0.005 <0.005 0.15 0.04 <0.005
0.03 0.02 0.01 0.01 <0.005 <0.005 0.006 0.30 0.11 <0.005
0.02 0.03 0.17 0.02 <0.005 0.01 <0.005 0.99 0.23 <0.005
0.14 0.05 0.10 0.60 0.02 <0.005 0.01 1.41 0.34 <0.005
0.01 0.02 0.05 0.01 <0.005 <0.005 <0.005 0.29 0.08 0.01
0.07 0.02 0.04 0.12 0.01 <0.005 0.01 0.53 0.14 <0.005
0.06 0.08 0.38 0.03 <0.005 0.01 0.01 1.17 0.34 0.08
0.40 0.07 0.26 0.77 0.13 0.02 0.02 1.59 0.44 0.01
Simulation of the different irrigation systems employed diverse strategies. In MACRO irrigation is treated separately from natural precipitation. The user specifies irrigation dates and times, application amounts, duration of the event. The irrigation rate is calculated as the amount of water during the irrigation event divided by the irrigation duration. This ratio was calibrated in different ways according to the irrigation system and to the data provided by the farmers (Table 1): • for sprinkler irrigation, the irrigation rate was considered to be lower than the infiltration capacity, and so all the water reaching the soil was able to percolate, and then run off and ponded condition did not occur and • for basin and border irrigations, in order to simulate ponded conditions, the irrigation rate was considered to be higher than the infiltration capacity.
It should be underlined that: MACRO is a one-dimensional model, therefore it is possible to model only an average condition of irrigation in the field; the amount of irrigation water was considered to be that which infiltrates into the soil. Concerning the weather data, daily minimum and maximum temperature and precipitation were recorded during the monitoring study, as stated above. The other weather data (i.e. wind speed, relative humidity, solar radiation) required for the calculation of evapotranspiration – Penman–Monteith equation is used in MACRO – came from the regional meteorological stations placed closest to the farms. When solar radiation was not available, it was calculated with the software RADEST vs. 3.0 beta (Donatelli et al., 2003). Vapour pressure was calculated following the FAO approach (Allen et al., 1998). Information about TBA and DES characteristics were taken respectively from the FOOTPRINT Pesticide Properties DataBase (2006) and Huber (2002). The required information was the soil sorption coefficient normalised to organic carbon content Koc (mL g−1 ) (220 for TBA and 83 for DES), the time for 50% decline of the initial pesticide concentration DT50 (days) (77 for TBA and 43 for DES), and the exponent of the Freundlich adsorption isotherm (0.92 for TBA and 0.85 for DES). The estimated maximum formation fraction of the parent that degrades to the metabolite was 21%.
In order to have a reliable simulation of pesticide movement in the soil profile, it is important to reach previously a good simulation of water movement. The groundwater table depth, which was measured daily in all fields (except fields S2 and Ba2), was used to calibrate the model and thus achieve a good soil hydrology calibration. The calibration was done considering the parameters involved in the equation governing the water percolation in the model. As stated before, in this study groundwater table in the profile was chosen as bottom boundary condition. This condition occurs when the bottom soil layer is unsaturated, an upward directed hydraulic gradient exists and capillary rise into the profile is calculated assuming a zero pressure head at the base. When the bottom soil layer is saturated, the percolation qout (m/h) is calculated as a linear function of the water table height (H [m]): qout = BGRAD × H where H is the water table height, which is function of the soil thickness, and BGRAD = qconst (K/Ks). qconst (h−1 ) is an empirical parameter and K refers to the saturated hydraulic conductivity of either the macropore or micropore regions in the deepest horizon of the profile. At the saturation, when K = Ks, BGRAD is equal to qconst . The water lost from the soil profile by percolation is assumed to recharge the groundwater reservoir (Larsbo and Jarvis, 2003). The simulation period corresponded to the years monitored, from April 2005 to December 2007. This period was doubled with the aim to perform the model warm up. 3. Results 3.1. Monitoring results Before starting the monitoring, groundwater from each field was analysed for TBA and DES to check for any previous contamination; neither were detected at any site (results not shown). Concentrations of TBA and DES measured in groundwater samples between April 2005 and December 2007 (Table 4) were analysed by means of descriptive statistics considering the mean value of the three piezometers. The European Directive 2006/118/EC on the protection of groundwater against pollution deterioration sets the 0.1 g/L limit for pesticides in groundwater. It has to be underlined that usu-
Table 5 TBA and DES concentrations (g/L) grouped according to different irrigation practices. Median
No irrigation (1 field) Sprinkler (5 fields) Border (2 fields) Basin (2 fields)
90th
Mean
75th
TBA
DES
TBA
DES
TBA
DES
TBA
DES
<0.005 <0.005 <0.005 0.07
<0.005 <0.005 <0.005 0.10
<0.005 0.04 0.005 0.50
<0.005 0.09 0.01 0.83
<0.005 0.04 <0.005 0.19
<0.005 0.06 0.01 0.30
<0.005 0.005 <0.005 0.18
<0.005 0.03 0.01 0.33
G. Fait et al. / Europ. J. Agronomy 32 (2010) 280–287 Table 6 RMSE values for hydrology calibration. Field
RMSE (m)
S1 S2 S3 S4 S5 Bo1 Bo2 Ba1 Ba2 NoIrr
0.19 – 0.37 1.21 0.32 0.20 0.20 0.46 – 0.34
Table 7 RMSE values for TBA and DES simulations. Field
RMSE for TBA (g/L)
RMSE for DES (g/L)
S1 S2 S3 S4 S5 Bo1 Bo2 Ba1 Ba2 NoIrr
7.65 0.02 1.22 0.42 0.31 0.06 0.01 0.40 0.06 0.23
0.29 0.03 0.41 0.45 0.06 0.02 0.01 0.70 0.18 0.04
ally this limit is used as a threshold value, but it is an arbitrary non-health based value. Furthermore, the groundwater levels in the present study fluctuate too much to be considered suitable for drinking water abstraction. An annual mean would be a more appropriate endpoint as this water is not drinking water, however we chose to discuss the data considering the maximum concentrations occurred during the study as they represent the worst case. Data are showed in Table 4. On farms where sprinkler irrigation was used, concentrations of TBA and DES were often under the statutory limit of 0.1 g/L. On field S1, TBA was always under this limit, while the highest DES concentrations were detected in October 2005 and 2007. On field S2, TBA and DES concentrations never exceeded the 0.1 g/L threshold. On field S3, TBA and DES were seldom detected. The highest concentrations of TBA occurred in February 2006 and June 2006 and 2007; the highest concentrations of DES occurred in June 2005 and 2007. On field S4, the TBA concentrations never exceeded the 0.1 g/L threshold, while DES concentrations were higher than
285
the statutory limit in June and October 2005. Data of 2007 were not considered because the farmer did not protect the piezometers with plastic film during application, giving rise to a point risk from contamination. The concentrations of TBA on field S5 were in most cases undetectable, as occurred also for DES with the exception of June 2005. On the farms employing border irrigation, Bo1 and Bo2, TBA and DES concentrations were always under the statutory limit of 0.1 g/L. On the farm which was not irrigated, NoIrr, TBA and DES concentrations were also always lower than the legal limit. On the farms with basin irrigation, TBA and DES were often detected at concentrations higher than the statuatory limit. On field Ba1, the highest TBA and DES concentrations occurred in August 2005 and in October and December 2007. DES was detected at high concentration also in October 2005. On field Ba2 the highest TBA concentrations occurred in June, October and December 2007; the highest DES concentrations occurred in April, October and December 2005, and in October and December 2007. The highest concentrations of both parent compound and the metabolite occurred on sites Ba1 and Ba2, which were irrigated using basin systems (Table 5, wherein the concentrations of TBA and DES in groundwater are grouped according to the four irrigation practices).
3.2. Modelling results To assess the performance of the model, both graphic evaluations and a statistical index were used. For the hydrology calibration, the graphic evaluations were done comparing the groundwater table depth measured and simulated, while for the concentrations of the solutes in the soil profile, simulated concentrations at 1 m depth were taken into account, as suggested by the FOCUS (2000) and were compared with the measured concentrations in shallow groundwater. The depth at which groundwater was sampled was different in each field and corresponded to the depth between the water level in the profile and the screen depth of the piezometer (Table 1). The median concentration calculated for the simulated period using the monthly average concentrations was considered in order to make comparisons with the monitoring data. The root mean squared error (RMSE) was used as statistical index because it combines both bias and lack of precision:
RMSE =
n (Ei i=1
− Mi)
2
n
Fig. 2. Measured and simulated groundwater table depth (m) and rain (mm) in field S1.
286
G. Fait et al. / Europ. J. Agronomy 32 (2010) 280–287
Table 8 Monitored versus simulated concentrations (g/L). Field
S1 S2 S3 S4 S5 Bo1 Bo2 Ba1 Ba2 NoIrr
Monitored concentrations (g/L)
Simulated concentrations with irrigation (g/L)
Simulated concentrations without irrigation (g/L)
TBA
DES
TBA
DES
TBA
DES
<0.005 0.01 <0.005 <0.005 <0.005 <0.005 <0.005 0.15 0.04 <0.005
0.03 0.02 0.01 0.01 <0.005 <0.005 0.01 0.30 0.11 <0.005
0.91 <0.005 1.53 <0.005 0.22 <0.005 <0.005 0.31 <0.005 0.23
0.23 <0.005 0.34 0.02 0.05 <0.005 <0.005 0.06 0.03 0.03
0.29 <0.005 1.38 0.18 0.11 <0.005 <0.005 <0.005 <0.005 0.13
0.05 <0.005 0.29 <0.005 0.03 <0.005 <0.005 <0.005 <0.005 0.02
where Ei and Mi are respectively the estimated and the measured values, and n is the population size (Fox, 1981). The lowest limit of RMSE is 0 that means full adherence between model estimates and measures. RMSE has the same units as the quantity being estimated. The graphic evaluations (one example shown in Fig. 2) and the RMSE (Table 6), which ranged from 0.19 to 0.46 m, showed that the average groundwater level simulated was fitting, but the dynamics were poorly captured by the model. A satisfactory hydrology calibration was obtained in all fields, except for field S4. In fields S2 and Ba2 daily groundwater table depth measurements were not available, and then the simulated values were compared only graphically with the measurements carried out during each sampling event. In Fig. 2 it is also possible to observe that the process of hydrology calibration resulted in a better fit between measured and simulated groundwater table depth. Concerning the solutes modelling, RMSE values (Table 7) ranged from 0.01 to 0.42 g/L for TBA and from 0.01 to 0.70 g/L for DES showing that simulated concentrations for both compounds under irrigated conditions were in agreement with monitored concentrations, except for fields S1 and S3. In general simulated concentrations for both compounds under irrigated conditions were higher than those simulated under nonirrigated conditions (Table 8, one example shown in Fig. 3). 4. Discussion In recent years, there has been growing concern about the persistence, mobility and toxicity of the metabolites of triazine herbicides (Adams and Thurman, 1991; Squillace et al., 1993). In this study, concentrations of TBA and DES in groundwater were
generally low. DES concentrations were higher than those of TBA, this being in agreement with other studies which found pesticide metabolites to be detected more frequently, or at higher concentrations, than their parent compound (Gilliom et al., 2007; Kolpin et al., 1998, 2004). Higher concentrations of DES are also consistent with the greater mobility of this metabolite, as demonstrated in field studies (Guzzella et al., 2006, 2003). The months in which contamination of groundwater was most frequent were June, August, October and December. June and August are the months immediately following treatment and in which irrigation is usually performed. October and December are generally rainy months. Guzzella et al. (2003) have showed that dry soil facilitates the formation of dealkylated and hydroxylated derivatives of TBA, whereas rainfall events promote the leaching of DES, the most mobile of the TBA metabolites. This explains why the highest concentrations occurred in the aforementioned months. The most contaminated groundwater in the study was in those sites receiving basin irrigation, indicating that the extent of TBA and DES leaching depends on the type of irrigation. The results obtained from the monitoring study were further investigated through the use of modelling because there are many factors other than irrigation that can affect herbicide leaching to groundwater. Simulations were in agreement with the measured concentrations. This result was important in order to understand the effect of the different irrigation systems on herbicide leaching in the studied fields. When the model was run without the input of water coming from irrigation, the results suggested that concentrations would be lower by between a factor 0.29 g/L for sprinkler to 0.005 g/L for basin irrigation. This finding supports the hypothesis that different irrigation systems can influence leaching.
Fig. 3. TBA concentrations (g/L) in field Ba1 simulated with irrigation (black line), without irrigation (dotted line), and measured (dots).
G. Fait et al. / Europ. J. Agronomy 32 (2010) 280–287
5. Conclusions The present study showed that concentrations of TBA and its metabolite DES in groundwater were in general low. The highest concentrations were found on farms using basin irrigation, indicating that this irrigation system can influence the leaching of both TBA and DES. Furthermore, the basin irrigation requires the greatest amount of water which is a valuable resource and should be used in a manner that maximises crop productivity per litre. The use of this type of irrigation should be avoided in order to protect groundwater from TBA and DES contamination and to avoid water wastage or TBA should not be used when using basin irrigation. Border and sprinkler systems gave similar TBA and DES leaching. Concentrations were generally <0.1 g/L and these systems seem compatible with the use of TBA. Therefore, careful choice of irrigation systems can be very effective in reducing the risk of groundwater contamination. Mathematical models for pesticide fate and transport have been widely used in order to predict future conditions (Corwin et al., 1999). In the present study, such models were useful to test the initial hypothesis. Thus the coupling of experimental results and mathematical models should always be considered not only for making predictions but also for improving our understanding of the complex processes governing leaching. Acknowledgments The authors thank Richard H. Bromilow for proofreading the English manuscript, and Tommaso Ferrari and Lucio Botteri for technical assistance in samplings and analyses. References Acutis, M., Donatelli, M., 2003. SOILPAR 2.00: software to estimate soil hydrological parameters and functions. Eur. J. Agron. 18, 373–377. Adams, C.D., Thurman, E.M., 1991. Formation and transport of deethylatrazine in the soil and vadose zone. J. Environ. Qual. 20, 540–547. Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop Evapotranspiration. Guidelines for Computing Crop Water Requirements. FAO Irrigation and Drainage Paper 56. ARPA Agenzia regionale per la protezione dell’ambiente della Lombardia. Available at http://www.arpalombardia.it/meteo/dati/richiesta.asp (verified 13 October 2008). Bandaranayake, W.M., Butters, G.L., Hamdi, M., Prieksat, M., Ellsworth, T.R., 1998. Irrigation and tillage management effects on solute movement. Soil Till. Res. 46, 165–173. Barbash, J.E., Thelin, G.P., Kolpin, D.W., Gilliom, R.J., 2001. Major herbicides in ground water: results from the national water-quality assessment. J. Environ. Qual. 30, 831–845. Barra Caracciolo, A., Giuliano, G., Grenni, P., Cremisini, C., Ciccoli, R., Ubaldi, C., 2005. Effect of urea on degradation of terbuthylazine in soil. Environ. Toxicol. Chem. 24, 1035–1040.
287
Bear, J., 1972. Dynamics of Fluids in Porous Media. Elsevier, New York. Carter, A.D., 2000. Herbicide movement in soils: principles, pathways and processes. Weed Res. 40, 113–122. Corwin, D.L., Loague, K., Ellsworth, T.R., 1999. Advanced information technologies for assessing nonpoint source pollution in the vadose zone: conference overview. J. Environ. Qual. 28, 357–365. Donatelli, M., Bellocchi, G., Fontana, F., 2003. RadEst3.00: software to estimate daily radiation data from commonly available meteorological variables. Eur. J. Agron. 18, 363–367. Dousset, S., Mouvet, C., Schiavon, M., 1997. Degradation of [14 C]terbuthylazine and [14 C]atrazine in laboratory soil microcosms. Pest. Sci. 49, 9–16. FOCUS, 2000. FOCUS groundwater scenarios in the EU review of active substances. Report of the FOCUS Groundwater Scenarios Workgroup, EC Document Reference Sanco/321/2000 rev.2. FOOTPRINT, 2006. The FOOTPRINT Pesticide Properties DataBase. Database collated by the University of Hertfordshire as part of the EU-funded FOOTPRINT project (FP6-SSP-022704), Available at http://www.eu-footprint.org/ppdb.html (verified 28 November 2003). Fox, D.G., 1981. Judging air quality model performance: a summary of the AMS workshop on dispersion models performance. Bull. Am. Meteorol. Soc. 62, 599– 609. Gilliom, R.J., Barbash, J.E., Crawford, C.G., Hamilton, P.A., Martin, J.D., Nakagaki, N., Nowell, L.H., Scott, J.C., Stackelberg, P.E., Thelin, G.P., Wolock, D.M., et al., 2007. Pesticides in streams and ground water of the United States. In: Del Re, A.A.M. (Ed.), XIII Symposium Pesticide Chemistry—Environmental Fate and Ecological Effects of Pesticides. La Goliardica Pavese s.r.l, pp. 731–742. Guzzella, L., Rullo, S., Pozzoni, F., Giuliano, G., 2003. Studies on mobility and degradation pathways of terbuthylazine using lysimeters on a field scale. J. Environ. Qual. 32, 1089–1098. Guzzella, L., Pozzoni, F., Giuliano, G., 2006. Herbicide contamination of surficial groundwater in Northern Italy. Environ. Pollut. 142, 344–353. Huber, A., 2002. Leaching and behaviour of terbuthylazine (GS 13529) and its metabolites (GS 26379 and GS 23158) after use in Italy. Modelling study conducted with FOCUS PELMO 3.3.2, FOCUS PRZM 2.4.1 and FOCUS PEARL 1.1.1. Internal report no. Mod02AH24. ISTAT, 2000. 5th Agricultural Census, Available at http://www.census.istat.it (verified 13 October 2008). Kolpin, D.W., Thurman, E.M., Linhart, S.M., 1998. The environmental occurrence of herbicides: the importance of degradates in ground water. Arch. Environ. Contamin. Toxicol. 35, 385–390. Kolpin, D.W., Schnoebelen, D.J., Thurman, E.M., 2004. Degradates provide insight to spatial and temporal trends of herbicides in ground water. Ground Water 42, 601–608. Larsbo, M., Jarvis, N.J., 2003. MACRO 5.0. A Model of Water Flow and Solute Transport in Macroporous Soil. Technical Description. Rep Emergo 2003, 6. Swedish University of Agricultural Sciences, Uppsala, Sweden. Pichon, V., 2000. Solid-phase extraction for multiresidue analysis of organic contaminants in water. J. Chromatogr. A 885, 195–215. Rawls, W.J., Brakensiek, D.L., 1989. Estimation of soil water retention and hydraulic properties. In: Morel, S. (Ed.), Unsatured Flow in Hydrologic Modeling. Theory and Pratice. Kluwer Academic Publishers, pp. 275–300. Squillace, P.J., Thurman, E.M., Furlong, E.T., 1993. Groundwater as a nonpoint source of atrazine and deethylatrazine in a river during base flow conditions. Water Resour. Res. 29, 1719–1729. Stenemo, F., Jarvis, N., 2007. Accounting for uncertainty in pedotransfer functions in vulnerability assessments of pesticide leaching to groundwater. Pest. Manage. Sci. 63, 867–875. Ungaro, F., Calzolari, C., Busoni, E., 2005. Development of pedotransfer functions using a group method of data handling for the soil of the Pianura Padano-Veneta region of North Italy: water retention properties. Geoderma 124, 293–317. Wösten, J.H.M., Lilly, A., Nemes, A., Le Bas, C., 1999. Development and use of a database of hydraulic properties of European soils. Geoderma 90, 169–185.
Contents lists available at ScienceDirect
European Journal of Agronomy journal homepage: www.elsevier.com/locate/eja
A field study of the impact of different irrigation practices on herbicide leaching Gabriella Fait a,∗ , Matteo Balderacchi a , Federico Ferrari a , Fabrizio Ungaro b , Ettore Capri a , Marco Trevisan a a b
Istituto di Chimica Agraria ed Ambientale, Università Cattolica del Sacro Cuore, via Emilia Parmense 84, 29100 Piacenza, Italy Consiglio Nazionale delle Ricerche, Istituto di Ricerca per la Protezione Idrogeologica, U.O.S. di Firenze Pedologia Applicata, Sesto F.no, Italy
a r t i c l e
i n f o
Article history: Received 12 November 2009 Received in revised form 29 January 2010 Accepted 1 February 2010 Keywords: Terbuthylazine Desethylterbuthylazine Maize Modelling MACRO
a b s t r a c t Agricultural practices, such as subsurface drainage, irrigation and tillage, may significantly affect pesticide leaching and, consequently, the risk of groundwater contamination. The aim of the present study was to investigate the impact of different irrigation systems on herbicide leaching to shallow groundwater through direct monitoring at the field scale in northern Italy over a 3-year period. Concentrations of the herbicide terbuthylazine (TBA) and its metabolite desethylterbuthylazine (DES) were monitored on 10 farms cropped with maize and irrigated by sprinkler, basin and border systems. Considering the results grouped according to the different irrigation systems, the mean TBA and DES concentrations was lower than the arbitrary non-health based legal limit of 0.1 g/L using sprinkler and border systems, while it was 0.19 and 0.30 g/L respectively for TBA and DES using basin systems. However, since many factors other than the irrigation systems can contribute to pesticide leaching and in a field study it is impossible to discriminate between all the different variables, the concentrations of both compounds were simulated with and without irrigation using the model MACRO 5.1 in order to gain a deeper understanding of the role of irrigation on leaching. First, the groundwater table depth, which was measured daily in all fields, was used to calibrate the model and thus achieve a good soil hydrology calibration. To assess the performance of the model the root mean squared error (RMSE) was used. RMSE ranged from 0.2 to 0.5 m, showing that a satisfactory hydrology calibration was obtained. Afterward, the solutes were modelled and the results showed that under non-irrigated conditions, concentrations of both compounds would be very low. These findings validate the hypothesis that careful selection of agricultural practices, such as the type of irrigation, can reduce pesticide leaching. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Pesticide occurrence in groundwater is controlled by many factors such as the intensity of use, soil properties and pesticide properties (Barbash et al., 2001). In addition certain agricultural practices, such as tillage, the use of subsurface drains and irrigation, may appreciably contribute to pesticide leaching and subsequently to groundwater contamination (Gilliom et al., 2007). In this study, attention was focused on irrigation. An adequate water supply is critical for plant growth, and various methods can be used to supply water to plants. These different irrigation techniques influence water flow patterns in the soil (Bandaranayake et al., 1998) and solute movement. The aim of this study was to investigate the impact of different irrigation systems on pesticide leaching to shallow groundwater by direct monitoring at the field scale over a 3-year period. Sprinkler, border and basin irrigation systems were considered.
∗ Corresponding author. Tel.: +39 0523 599 345; fax: +39 0523 599 217. E-mail address: [email protected] (G. Fait). 1161-0301/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.eja.2010.02.001
As water-quality monitoring shows that herbicides are the most frequently detected group of pesticides in ground and surface water (Carter, 2000), the herbicide terbuthylazine [N2 -tert-butyl6-chloro-N4 -ethyl-1,3,5-triazine-2,4-diamine] (TBA) was chosen. TBA belongs to the triazine group of herbicides, which is amongst the most frequently used group for selective weed control in several crops (Barra Caracciolo et al., 2005). In Italy TBA uses are now limited to maize and sorghum according to the implementation of the Environmental Stewardship Programme as agreed by the TBA manufactures. The major dealkylation product of TBA is desethylterbuthylazine (DES) (Dousset et al., 1997). Dealkylated products are generally more persistent and water soluble, and therefore pose a risk of groundwater contamination (Guzzella et al., 2003). Thus DES concentrations were also monitored. The focus of this study was to investigate the effect of irrigation on TBA and DES leaching to shallow groundwater in 10 farms located in the northern Italy selected to cover a range of different soil types and climatic conditions. Such field studies are helpful in order to better understand how agricultural practices affect pesticide leaching under real farm situations; however it is impossible to carry out these types of studies under identical conditions, so
G. Fait et al. / Europ. J. Agronomy 32 (2010) 280–287
the effect of irrigation systems on pesticide leaching could be confounded by other factors. For this reason, the solute transport model MACRO 5.1 (Larsbo and Jarvis, 2003) was used to simulate TBA and DES leaching under both irrigated and non-irrigated conditions. Measured parameters (i.e. soil, climatic, pedological and hydrological data) were used to set a scenario for each farm. First, the model was run using real irrigation data. If a good fit between simulated and measured concentration is attained, it is possible to investigate the effect of irrigation on leaching by running the model without the input of water coming from irrigation. In this way, the effect of the irrigation system can be isolated and then it is possible to examine further the effect of differing irrigation systems on pesticide leaching under different situations. 2. Materials and methods 2.1. Study design The 10 farms (Table 1) were located in the Po Valley of northern Italy (Fig. 1), which is an important agricultural area. The farms were cropped with maize and managed according to good agricultural practices. Two farms were irrigated by the basin system, two by border irrigation, one was not irrigated and the remaining farms were irrigated by sprinklers. On each farm, a field approximately 10 ha in size was monitored. Before starting the monitoring, each field was assessed for the absence of surface-water bodies, the uniform direction of groundwater flux, slope (which should not exceed 5%), and the absence of cracking soil, perched groundwater and tile drainage. The groundwater depth was shallow, but varied amongst the 10 fields (Table 1). Three piezometers were installed in each field, placed at the edge of the field in the direction of down-gradient groundwater flow in order to maximize interception of shallow groundwater flow leaving the field. The piezometers were made of PVC pipes, with a rubber plug inserted at the base to prevent backflow and a cast-iron manhole cover above each piezometer. A layer of bentonite pellets was placed outside and inside the manhole cover to act as a hydraulic insulating agent, thereby minimizing potential contamination from the percolation of preferential fluxes. Meteorological data (Table 2) were collected on each farm from a digital remote-controlled station, which recorded minimum and maximum temperature, precipitation and groundwater level on an hourly basis. On farms Ba1 and Ba2, the measurements came from the regional meteorological station closest to the farms, provided by the regional agency for the protection of the environment (ARPA Lombardia). 2.2. Irrigation practices and field management Irrigation is a common practice in the Po Valley in order to ensure maize production. Only in some areas in the Friuli region it is possible to obtain satisfactory yields without irrigation. The type of irrigation system depends upon the crop, landscape and local customs. The most recent Italian agricultural census (ISTAT, 2000) showed that sprinkler and border systems are the most common irrigation systems. Basin irrigation is widely used in the North of Italy for rice crop and then for the crops in rotation with rice, such as maize. Information on irrigation practices (Table 1), such as the number of applications and the volume of water applied, was provided by the farmers. All the fields were cropped with maize for the entire study period. TBA was applied pre-emergence annually as the formulation BoleroTZ® (Monsanto, containing 214 g ai/L) and was applied at 4 L/ha. Spray-bar calibration was performed prior
281
to each application, and each piezometer was protected with a plastic film during application in order to avoid any direct contamination. 2.3. Water sampling and analysis Groundwater was sampled by applying low negative pressure. The depth of the piezometer screen is reported in Table 1. Each piezometer was purged prior to sampling in order to remove standing water. The sample volume (approximately 1 L) was collected in glass bottles, being stored immediately in a portable freezer and then transferred to a refrigerator at 4 ◦ C prior to analysis. Samples were collected every 2 months from April 2005 to December 2007. The water table depth within each piezometer was measured at each sampling event. Less than 1 month elapsed between sample collection and analysis. Water samples were analysed measuring both TBA and DES concentrations according to the SPE–GC–MS method presented in Pichon (2000). The analytes were identified through: (i) comparison between the retention time of the samples and the analytical standards (purity > 99%), (ii) comparison of the mass spectra of the samples and the analytical standards, and (iii) comparison of the mass spectrum in the NIST library with that of the samples. Quantification was performed in single-ion monitoring mode. Concentrations of each analyte were calculated using an external calibration curve, the limit of quantification in the water samples being 5 ng/L. 2.4. Evaluation of the hydraulic transit times beneath the fields In order to assess the hydraulic transit times beneath each field, the horizontal movement of water was estimated using “Darcy’s Law”: q=K ×i where K is the hydraulic conductivity (m/s) and i is the hydraulic gradient, which is equal to the hydraulic head difference (H) between two points in the soil divided by the distance between those points (L) (m). Furthermore, the q value was divided by the porosity in order to take into account that only the interstitial spaces in the soil are able to conduct water. Based on estimated K values from the literature (Bear, 1972) and measured hydraulic heads and K values calculated using the pedotransfer function (PTF) proposed by Wösten et al. (1999) and site-specific data, lateral groundwater flow velocities were estimated using Darcy’s law to vary between 0.3 and 98.3 m/in 2 months, and so the sampling frequency (2 months) was considered to be adequate (Table 3). 2.5. Model description MACRO (version 5.1) is a comprehensive mechanistic onedimensional non-steady state model of water flow and solute transport in structured or macroporous soils (Larsbo and Jarvis, 2003). This model was chosen because such soils are typical in the Po Valley. The model accounts for macropore flow, with the soil porosity divided into two flow domains (macropores and micropores) each characterized by a flow rate and solute concentration, with the boundary between them defined by a fixed water potential associated with a saturated matrix water content and hydraulic conductivity (Stenemo and Jarvis, 2007). Each domain has its own pressure head and solute concentration. Richards’ equation and the convection–dispersion equation are used to model soil water flow and solute transport in the soil micropores, while in the macropores
282
G. Fait et al. / Europ. J. Agronomy 32 (2010) 280–287
Table 1 Field characteristics. Field
Field area (ha)
Topsoil texture class (USDA)
pH
Topsoil OC %
Groundwater table depth 2005–2007 (m)
Piezometer screen depth (m)
Irrigation system
Number of irrigations (per year)
Total amount of water (mm)
Clay loam
7.5
1.3
0.9–2.8
4.50
Sprinkler
3 (2005) 7 (2006) 7 (2007)
120 280 280
S1
8.5
S2
12.0
Sandy loam
5.7
1.1
0.9–1.8
4.90
Sprinkler
6 (2005) 5 (2006) 7 (2007)
120 110 126
S3
20.0
Loam
7.6
0.7
1.1–2.8
4.80
Sprinkler
1 (2005) 1 (2006) 3 (2007)
30 50 150
S4
3.6
Sandy loam
5.5
1.1
0.3–2.7
3.70
Sprinkler
1 (2005) 2 (2006) 2 (2007)
30 60 60
S5
12.0
Loam
7.9
0.9
0.5–1.8
4.20
Sprinkler
3 (2005) 5 (2006) no (2007)
90 125 –
Bo1
15.0
Sandy loam
5.4
1.0
5.3–6.7
8.00
Border
2 (2005) 3 (2006) 3 (2007)
120 180 180
Bo2
3.7
Sandy loam
7.2
2.6
4.4–5.4
6.50
Border
1 (2005) 2 (2006) 2 (2007)
60 120 120
Ba1
9.1
Sandy clay
7.2
1.2
3.8–5.8
6.30
Basin
4 (2005) 6 (2006) 6 (2007)
280 420 420
Ba2
12.3
Sandy loam
6.3
1.7
3.6–6.9
4.60
Basin
3 (2005) 5 (2006) 5 (2007)
210 350 350
NoIrr
10.5
Loam
7.8
0.9
0.7–3.9
5.20
Not irrigated
– – –
– – –
Fig. 1. Farms location. The farms are indicated by the black dots (S1–S5: sprinkler irrigation; Ba1, Ba2: basin irrigation; Bo1, Bo2: border irrigation; NoIrr: not irrigated).
G. Fait et al. / Europ. J. Agronomy 32 (2010) 280–287
283
Table 2 Average annual rainfall (mm) at each farm. Field
Location
Rainfall from April to December 2005 (mm)
Annual rainfall in 2006 (mm)
Annual rainfall in 2007 (mm)
S1 S2 S3 S4 S5 Bo1 Bo2 Ba1 Ba2 NoIrr
Lombardia Lombardia Veneto Veneto Friuli-Venezia Giulia Piemonte Piemonte Lombardia Lombardia Emilia Romagna
752 662 1060 1089 832 561 561 716 654 544
625 595 783 573 711 663 663 779 525 637
437 785 745 391 965 349 349 689 531 438
a modified kinematic wave approach is applied to calculate fluxes and solute transport is described with an advective flow solute, neglecting dispersion. The solute concentration in water flowing into the macropores is calculated assuming complete mixing in a shallow mixing depth at the soil surface. A modified form of the van Genuchten function is used to describe the water retention function. A Freundlich isotherm is used to describe pesticide sorption, and first-order kinetics is assumed for degradation, with a rate coefficient given as a function of soil temperature and moisture content. Water and solute exchange between the flow domains are calculated using approximate first-order expressions based on an effective diffusion path length. In MACRO there are four different options for the bottom boundary condition (Larsbo and Jarvis, 2003). In the present study groundwater table in the profile was chosen. 2.6. Model parameterization The required inputs for MACRO parameterization include: soil physical and hydraulic parameters, crop, irrigation, solute characteristics and weather data. The soil physical and hydraulic parameters govern retention and transport of water and chemicals, but are difficult to measure and so are often derived using PTFs, which relate these model parameters to more easily measured soil properties such as soil texture, organic
matter content and/or other routine measurements. In the present study the following properties were analysed on a soil core sampled from each field: texture, pH, organic carbon (OC), and horizon depth. Bulk density was not available, and so it had to be estimated, even if this kind of approach using a double pedotransfer is a last option to estimate hydrological data. Bulk density was calculated with the software SOILPAR 2.00 (Acutis and Donatelli, 2003) using the method described by Rawls and Brakensiek (1989). The parameters of the van Genuchten soil–water retention function were calculated using the PTFs developed by Ungaro et al. (2005) for Po Valley soils. The total saturated hydraulic conductivity was calculated using the PTFs developed by Wösten et al. (1999), as stated above. Information about the crop, such as harvesting and seeding periods, pesticide application and irrigation, was provided by the farmers. The other information required by the model in order to parameterize the crop, such as day of maximum leaf area/root depth, root distribution, canopy interception capacity, etc., came from the FOCUS (FOrum for Co-ordination of pesticide fate models and their Use) Groundwater Scenarios Workgroup (FOCUS, 2000). The FOCUS groundwater scenarios are a set of nine standard combinations of weather, soil and cropping data which collectively represent agriculture in the EU for the purposes of a Tier 1 EU-level assessment of leaching potential. In the present study information for maize crop for Piacenza scenario was taken into account.
Table 3 Estimates of horizontal water flow. Field
K for bottom layera (m/s)
Corresponding soil textural type for K for bottom layer (USDA limits)a
H (hydraulic head) (m)
S1
0.00013 0.0001
Sand Sand
0.2
270
1.2 0.9
S2
0.00034 0.0001
Sand Sand
0.8
420
9.6 2.8
S3
0.00225 0.01
Sandy silt–clay silt Gravel
0.8
270
98.3 437.7
S4
0.00084 0.0001
Sandy loam Sand
0.6
255
28.2 3.4
S5
0.00026 0.0001
Sandy–silty sandy Sand
0.05
1130
0.2 0.1
Bo1
0.00005 0.0001
Silty sandy–silty clay Sand
1.4
415
2.2 4.5
Bo2
0.00003 0.0001
Sand Sand
0.4
105
1.7 5.0
Ba1
0.00006 0.00001
Loam Sand
0.2
400
0.8 1.4
Ba2
0.00009 0.0001
Sandy loam Sand
0.1
290
0.3 0.3
NoIrr
0.00003 0.0001
Silty clay Sand
0.6
420
0.6 1.7
a
L (field length) (m)
q (m in 2 months)
The values coming from the literature are reported in normal type, while the values coming from the PTF and site-specific data are reported in bold type.
284
G. Fait et al. / Europ. J. Agronomy 32 (2010) 280–287
Table 4 Concentrations of TBA and DES (g/L). Field
S1 S2 S3 S4 S5 Bo1 Bo2 Ba1 Ba2 NoIrr
Median
90th
Mean
Maximum
TBA
DES
TBA
DES
TBA
DES
TBA
DES
<0.005 0.01 <0.005 <0.005 <0.005 <0.005 <0.005 0.15 0.04 <0.005
0.03 0.02 0.01 0.01 <0.005 <0.005 0.006 0.30 0.11 <0.005
0.02 0.03 0.17 0.02 <0.005 0.01 <0.005 0.99 0.23 <0.005
0.14 0.05 0.10 0.60 0.02 <0.005 0.01 1.41 0.34 <0.005
0.01 0.02 0.05 0.01 <0.005 <0.005 <0.005 0.29 0.08 0.01
0.07 0.02 0.04 0.12 0.01 <0.005 0.01 0.53 0.14 <0.005
0.06 0.08 0.38 0.03 <0.005 0.01 0.01 1.17 0.34 0.08
0.40 0.07 0.26 0.77 0.13 0.02 0.02 1.59 0.44 0.01
Simulation of the different irrigation systems employed diverse strategies. In MACRO irrigation is treated separately from natural precipitation. The user specifies irrigation dates and times, application amounts, duration of the event. The irrigation rate is calculated as the amount of water during the irrigation event divided by the irrigation duration. This ratio was calibrated in different ways according to the irrigation system and to the data provided by the farmers (Table 1): • for sprinkler irrigation, the irrigation rate was considered to be lower than the infiltration capacity, and so all the water reaching the soil was able to percolate, and then run off and ponded condition did not occur and • for basin and border irrigations, in order to simulate ponded conditions, the irrigation rate was considered to be higher than the infiltration capacity.
It should be underlined that: MACRO is a one-dimensional model, therefore it is possible to model only an average condition of irrigation in the field; the amount of irrigation water was considered to be that which infiltrates into the soil. Concerning the weather data, daily minimum and maximum temperature and precipitation were recorded during the monitoring study, as stated above. The other weather data (i.e. wind speed, relative humidity, solar radiation) required for the calculation of evapotranspiration – Penman–Monteith equation is used in MACRO – came from the regional meteorological stations placed closest to the farms. When solar radiation was not available, it was calculated with the software RADEST vs. 3.0 beta (Donatelli et al., 2003). Vapour pressure was calculated following the FAO approach (Allen et al., 1998). Information about TBA and DES characteristics were taken respectively from the FOOTPRINT Pesticide Properties DataBase (2006) and Huber (2002). The required information was the soil sorption coefficient normalised to organic carbon content Koc (mL g−1 ) (220 for TBA and 83 for DES), the time for 50% decline of the initial pesticide concentration DT50 (days) (77 for TBA and 43 for DES), and the exponent of the Freundlich adsorption isotherm (0.92 for TBA and 0.85 for DES). The estimated maximum formation fraction of the parent that degrades to the metabolite was 21%.
In order to have a reliable simulation of pesticide movement in the soil profile, it is important to reach previously a good simulation of water movement. The groundwater table depth, which was measured daily in all fields (except fields S2 and Ba2), was used to calibrate the model and thus achieve a good soil hydrology calibration. The calibration was done considering the parameters involved in the equation governing the water percolation in the model. As stated before, in this study groundwater table in the profile was chosen as bottom boundary condition. This condition occurs when the bottom soil layer is unsaturated, an upward directed hydraulic gradient exists and capillary rise into the profile is calculated assuming a zero pressure head at the base. When the bottom soil layer is saturated, the percolation qout (m/h) is calculated as a linear function of the water table height (H [m]): qout = BGRAD × H where H is the water table height, which is function of the soil thickness, and BGRAD = qconst (K/Ks). qconst (h−1 ) is an empirical parameter and K refers to the saturated hydraulic conductivity of either the macropore or micropore regions in the deepest horizon of the profile. At the saturation, when K = Ks, BGRAD is equal to qconst . The water lost from the soil profile by percolation is assumed to recharge the groundwater reservoir (Larsbo and Jarvis, 2003). The simulation period corresponded to the years monitored, from April 2005 to December 2007. This period was doubled with the aim to perform the model warm up. 3. Results 3.1. Monitoring results Before starting the monitoring, groundwater from each field was analysed for TBA and DES to check for any previous contamination; neither were detected at any site (results not shown). Concentrations of TBA and DES measured in groundwater samples between April 2005 and December 2007 (Table 4) were analysed by means of descriptive statistics considering the mean value of the three piezometers. The European Directive 2006/118/EC on the protection of groundwater against pollution deterioration sets the 0.1 g/L limit for pesticides in groundwater. It has to be underlined that usu-
Table 5 TBA and DES concentrations (g/L) grouped according to different irrigation practices. Median
No irrigation (1 field) Sprinkler (5 fields) Border (2 fields) Basin (2 fields)
90th
Mean
75th
TBA
DES
TBA
DES
TBA
DES
TBA
DES
<0.005 <0.005 <0.005 0.07
<0.005 <0.005 <0.005 0.10
<0.005 0.04 0.005 0.50
<0.005 0.09 0.01 0.83
<0.005 0.04 <0.005 0.19
<0.005 0.06 0.01 0.30
<0.005 0.005 <0.005 0.18
<0.005 0.03 0.01 0.33
G. Fait et al. / Europ. J. Agronomy 32 (2010) 280–287 Table 6 RMSE values for hydrology calibration. Field
RMSE (m)
S1 S2 S3 S4 S5 Bo1 Bo2 Ba1 Ba2 NoIrr
0.19 – 0.37 1.21 0.32 0.20 0.20 0.46 – 0.34
Table 7 RMSE values for TBA and DES simulations. Field
RMSE for TBA (g/L)
RMSE for DES (g/L)
S1 S2 S3 S4 S5 Bo1 Bo2 Ba1 Ba2 NoIrr
7.65 0.02 1.22 0.42 0.31 0.06 0.01 0.40 0.06 0.23
0.29 0.03 0.41 0.45 0.06 0.02 0.01 0.70 0.18 0.04
ally this limit is used as a threshold value, but it is an arbitrary non-health based value. Furthermore, the groundwater levels in the present study fluctuate too much to be considered suitable for drinking water abstraction. An annual mean would be a more appropriate endpoint as this water is not drinking water, however we chose to discuss the data considering the maximum concentrations occurred during the study as they represent the worst case. Data are showed in Table 4. On farms where sprinkler irrigation was used, concentrations of TBA and DES were often under the statutory limit of 0.1 g/L. On field S1, TBA was always under this limit, while the highest DES concentrations were detected in October 2005 and 2007. On field S2, TBA and DES concentrations never exceeded the 0.1 g/L threshold. On field S3, TBA and DES were seldom detected. The highest concentrations of TBA occurred in February 2006 and June 2006 and 2007; the highest concentrations of DES occurred in June 2005 and 2007. On field S4, the TBA concentrations never exceeded the 0.1 g/L threshold, while DES concentrations were higher than
285
the statutory limit in June and October 2005. Data of 2007 were not considered because the farmer did not protect the piezometers with plastic film during application, giving rise to a point risk from contamination. The concentrations of TBA on field S5 were in most cases undetectable, as occurred also for DES with the exception of June 2005. On the farms employing border irrigation, Bo1 and Bo2, TBA and DES concentrations were always under the statutory limit of 0.1 g/L. On the farm which was not irrigated, NoIrr, TBA and DES concentrations were also always lower than the legal limit. On the farms with basin irrigation, TBA and DES were often detected at concentrations higher than the statuatory limit. On field Ba1, the highest TBA and DES concentrations occurred in August 2005 and in October and December 2007. DES was detected at high concentration also in October 2005. On field Ba2 the highest TBA concentrations occurred in June, October and December 2007; the highest DES concentrations occurred in April, October and December 2005, and in October and December 2007. The highest concentrations of both parent compound and the metabolite occurred on sites Ba1 and Ba2, which were irrigated using basin systems (Table 5, wherein the concentrations of TBA and DES in groundwater are grouped according to the four irrigation practices).
3.2. Modelling results To assess the performance of the model, both graphic evaluations and a statistical index were used. For the hydrology calibration, the graphic evaluations were done comparing the groundwater table depth measured and simulated, while for the concentrations of the solutes in the soil profile, simulated concentrations at 1 m depth were taken into account, as suggested by the FOCUS (2000) and were compared with the measured concentrations in shallow groundwater. The depth at which groundwater was sampled was different in each field and corresponded to the depth between the water level in the profile and the screen depth of the piezometer (Table 1). The median concentration calculated for the simulated period using the monthly average concentrations was considered in order to make comparisons with the monitoring data. The root mean squared error (RMSE) was used as statistical index because it combines both bias and lack of precision:
RMSE =
n (Ei i=1
− Mi)
2
n
Fig. 2. Measured and simulated groundwater table depth (m) and rain (mm) in field S1.
286
G. Fait et al. / Europ. J. Agronomy 32 (2010) 280–287
Table 8 Monitored versus simulated concentrations (g/L). Field
S1 S2 S3 S4 S5 Bo1 Bo2 Ba1 Ba2 NoIrr
Monitored concentrations (g/L)
Simulated concentrations with irrigation (g/L)
Simulated concentrations without irrigation (g/L)
TBA
DES
TBA
DES
TBA
DES
<0.005 0.01 <0.005 <0.005 <0.005 <0.005 <0.005 0.15 0.04 <0.005
0.03 0.02 0.01 0.01 <0.005 <0.005 0.01 0.30 0.11 <0.005
0.91 <0.005 1.53 <0.005 0.22 <0.005 <0.005 0.31 <0.005 0.23
0.23 <0.005 0.34 0.02 0.05 <0.005 <0.005 0.06 0.03 0.03
0.29 <0.005 1.38 0.18 0.11 <0.005 <0.005 <0.005 <0.005 0.13
0.05 <0.005 0.29 <0.005 0.03 <0.005 <0.005 <0.005 <0.005 0.02
where Ei and Mi are respectively the estimated and the measured values, and n is the population size (Fox, 1981). The lowest limit of RMSE is 0 that means full adherence between model estimates and measures. RMSE has the same units as the quantity being estimated. The graphic evaluations (one example shown in Fig. 2) and the RMSE (Table 6), which ranged from 0.19 to 0.46 m, showed that the average groundwater level simulated was fitting, but the dynamics were poorly captured by the model. A satisfactory hydrology calibration was obtained in all fields, except for field S4. In fields S2 and Ba2 daily groundwater table depth measurements were not available, and then the simulated values were compared only graphically with the measurements carried out during each sampling event. In Fig. 2 it is also possible to observe that the process of hydrology calibration resulted in a better fit between measured and simulated groundwater table depth. Concerning the solutes modelling, RMSE values (Table 7) ranged from 0.01 to 0.42 g/L for TBA and from 0.01 to 0.70 g/L for DES showing that simulated concentrations for both compounds under irrigated conditions were in agreement with monitored concentrations, except for fields S1 and S3. In general simulated concentrations for both compounds under irrigated conditions were higher than those simulated under nonirrigated conditions (Table 8, one example shown in Fig. 3). 4. Discussion In recent years, there has been growing concern about the persistence, mobility and toxicity of the metabolites of triazine herbicides (Adams and Thurman, 1991; Squillace et al., 1993). In this study, concentrations of TBA and DES in groundwater were
generally low. DES concentrations were higher than those of TBA, this being in agreement with other studies which found pesticide metabolites to be detected more frequently, or at higher concentrations, than their parent compound (Gilliom et al., 2007; Kolpin et al., 1998, 2004). Higher concentrations of DES are also consistent with the greater mobility of this metabolite, as demonstrated in field studies (Guzzella et al., 2006, 2003). The months in which contamination of groundwater was most frequent were June, August, October and December. June and August are the months immediately following treatment and in which irrigation is usually performed. October and December are generally rainy months. Guzzella et al. (2003) have showed that dry soil facilitates the formation of dealkylated and hydroxylated derivatives of TBA, whereas rainfall events promote the leaching of DES, the most mobile of the TBA metabolites. This explains why the highest concentrations occurred in the aforementioned months. The most contaminated groundwater in the study was in those sites receiving basin irrigation, indicating that the extent of TBA and DES leaching depends on the type of irrigation. The results obtained from the monitoring study were further investigated through the use of modelling because there are many factors other than irrigation that can affect herbicide leaching to groundwater. Simulations were in agreement with the measured concentrations. This result was important in order to understand the effect of the different irrigation systems on herbicide leaching in the studied fields. When the model was run without the input of water coming from irrigation, the results suggested that concentrations would be lower by between a factor 0.29 g/L for sprinkler to 0.005 g/L for basin irrigation. This finding supports the hypothesis that different irrigation systems can influence leaching.
Fig. 3. TBA concentrations (g/L) in field Ba1 simulated with irrigation (black line), without irrigation (dotted line), and measured (dots).
G. Fait et al. / Europ. J. Agronomy 32 (2010) 280–287
5. Conclusions The present study showed that concentrations of TBA and its metabolite DES in groundwater were in general low. The highest concentrations were found on farms using basin irrigation, indicating that this irrigation system can influence the leaching of both TBA and DES. Furthermore, the basin irrigation requires the greatest amount of water which is a valuable resource and should be used in a manner that maximises crop productivity per litre. The use of this type of irrigation should be avoided in order to protect groundwater from TBA and DES contamination and to avoid water wastage or TBA should not be used when using basin irrigation. Border and sprinkler systems gave similar TBA and DES leaching. Concentrations were generally <0.1 g/L and these systems seem compatible with the use of TBA. Therefore, careful choice of irrigation systems can be very effective in reducing the risk of groundwater contamination. Mathematical models for pesticide fate and transport have been widely used in order to predict future conditions (Corwin et al., 1999). In the present study, such models were useful to test the initial hypothesis. Thus the coupling of experimental results and mathematical models should always be considered not only for making predictions but also for improving our understanding of the complex processes governing leaching. Acknowledgments The authors thank Richard H. Bromilow for proofreading the English manuscript, and Tommaso Ferrari and Lucio Botteri for technical assistance in samplings and analyses. References Acutis, M., Donatelli, M., 2003. SOILPAR 2.00: software to estimate soil hydrological parameters and functions. Eur. J. Agron. 18, 373–377. Adams, C.D., Thurman, E.M., 1991. Formation and transport of deethylatrazine in the soil and vadose zone. J. Environ. Qual. 20, 540–547. Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop Evapotranspiration. Guidelines for Computing Crop Water Requirements. FAO Irrigation and Drainage Paper 56. ARPA Agenzia regionale per la protezione dell’ambiente della Lombardia. Available at http://www.arpalombardia.it/meteo/dati/richiesta.asp (verified 13 October 2008). Bandaranayake, W.M., Butters, G.L., Hamdi, M., Prieksat, M., Ellsworth, T.R., 1998. Irrigation and tillage management effects on solute movement. Soil Till. Res. 46, 165–173. Barbash, J.E., Thelin, G.P., Kolpin, D.W., Gilliom, R.J., 2001. Major herbicides in ground water: results from the national water-quality assessment. J. Environ. Qual. 30, 831–845. Barra Caracciolo, A., Giuliano, G., Grenni, P., Cremisini, C., Ciccoli, R., Ubaldi, C., 2005. Effect of urea on degradation of terbuthylazine in soil. Environ. Toxicol. Chem. 24, 1035–1040.
287
Bear, J., 1972. Dynamics of Fluids in Porous Media. Elsevier, New York. Carter, A.D., 2000. Herbicide movement in soils: principles, pathways and processes. Weed Res. 40, 113–122. Corwin, D.L., Loague, K., Ellsworth, T.R., 1999. Advanced information technologies for assessing nonpoint source pollution in the vadose zone: conference overview. J. Environ. Qual. 28, 357–365. Donatelli, M., Bellocchi, G., Fontana, F., 2003. RadEst3.00: software to estimate daily radiation data from commonly available meteorological variables. Eur. J. Agron. 18, 363–367. Dousset, S., Mouvet, C., Schiavon, M., 1997. Degradation of [14 C]terbuthylazine and [14 C]atrazine in laboratory soil microcosms. Pest. Sci. 49, 9–16. FOCUS, 2000. FOCUS groundwater scenarios in the EU review of active substances. Report of the FOCUS Groundwater Scenarios Workgroup, EC Document Reference Sanco/321/2000 rev.2. FOOTPRINT, 2006. The FOOTPRINT Pesticide Properties DataBase. Database collated by the University of Hertfordshire as part of the EU-funded FOOTPRINT project (FP6-SSP-022704), Available at http://www.eu-footprint.org/ppdb.html (verified 28 November 2003). Fox, D.G., 1981. Judging air quality model performance: a summary of the AMS workshop on dispersion models performance. Bull. Am. Meteorol. Soc. 62, 599– 609. Gilliom, R.J., Barbash, J.E., Crawford, C.G., Hamilton, P.A., Martin, J.D., Nakagaki, N., Nowell, L.H., Scott, J.C., Stackelberg, P.E., Thelin, G.P., Wolock, D.M., et al., 2007. Pesticides in streams and ground water of the United States. In: Del Re, A.A.M. (Ed.), XIII Symposium Pesticide Chemistry—Environmental Fate and Ecological Effects of Pesticides. La Goliardica Pavese s.r.l, pp. 731–742. Guzzella, L., Rullo, S., Pozzoni, F., Giuliano, G., 2003. Studies on mobility and degradation pathways of terbuthylazine using lysimeters on a field scale. J. Environ. Qual. 32, 1089–1098. Guzzella, L., Pozzoni, F., Giuliano, G., 2006. Herbicide contamination of surficial groundwater in Northern Italy. Environ. Pollut. 142, 344–353. Huber, A., 2002. Leaching and behaviour of terbuthylazine (GS 13529) and its metabolites (GS 26379 and GS 23158) after use in Italy. Modelling study conducted with FOCUS PELMO 3.3.2, FOCUS PRZM 2.4.1 and FOCUS PEARL 1.1.1. Internal report no. Mod02AH24. ISTAT, 2000. 5th Agricultural Census, Available at http://www.census.istat.it (verified 13 October 2008). Kolpin, D.W., Thurman, E.M., Linhart, S.M., 1998. The environmental occurrence of herbicides: the importance of degradates in ground water. Arch. Environ. Contamin. Toxicol. 35, 385–390. Kolpin, D.W., Schnoebelen, D.J., Thurman, E.M., 2004. Degradates provide insight to spatial and temporal trends of herbicides in ground water. Ground Water 42, 601–608. Larsbo, M., Jarvis, N.J., 2003. MACRO 5.0. A Model of Water Flow and Solute Transport in Macroporous Soil. Technical Description. Rep Emergo 2003, 6. Swedish University of Agricultural Sciences, Uppsala, Sweden. Pichon, V., 2000. Solid-phase extraction for multiresidue analysis of organic contaminants in water. J. Chromatogr. A 885, 195–215. Rawls, W.J., Brakensiek, D.L., 1989. Estimation of soil water retention and hydraulic properties. In: Morel, S. (Ed.), Unsatured Flow in Hydrologic Modeling. Theory and Pratice. Kluwer Academic Publishers, pp. 275–300. Squillace, P.J., Thurman, E.M., Furlong, E.T., 1993. Groundwater as a nonpoint source of atrazine and deethylatrazine in a river during base flow conditions. Water Resour. Res. 29, 1719–1729. Stenemo, F., Jarvis, N., 2007. Accounting for uncertainty in pedotransfer functions in vulnerability assessments of pesticide leaching to groundwater. Pest. Manage. Sci. 63, 867–875. Ungaro, F., Calzolari, C., Busoni, E., 2005. Development of pedotransfer functions using a group method of data handling for the soil of the Pianura Padano-Veneta region of North Italy: water retention properties. Geoderma 124, 293–317. Wösten, J.H.M., Lilly, A., Nemes, A., Le Bas, C., 1999. Development and use of a database of hydraulic properties of European soils. Geoderma 90, 169–185.