Chemosphere 74 (2009) 1044–1052
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
A bioaccumulation model for herbicides in Ulva rigida and Tapes philippinarum in Sacca di Goro lagoon (Northern Adriatic) Roberta Carafa a, Dimitar Marinov a, Sibylle Dueri a, Jan Wollgast a, Gianmarco Giordani c, Pierluigi Viaroli c, José-Manuel Zaldívar b,* a
European Commission, Joint Research Centre, Institute for Environment and Sustainability, Italy European Commission, Joint Research Centre, Institute for Health and Consumer Protection, Via Enrico Fermi 2749, Ispra 21027 (VA), Italy c Department of Environmental Sciences, Parma University, Parma, Italy b
a r t i c l e
i n f o
Article history: Received 19 August 2008 Received in revised form 29 October 2008 Accepted 31 October 2008 Available online 9 December 2008 Keywords: s-Triazines Ulva rigida Tapes philippinarum Sacca di Goro Bioaccumulation modelling
a b s t r a c t A bioaccumulation model to predict concentrations of s-triazine herbicides in the macroalgae Ulva rigida and in clams Tapes philippinarum has been implemented, calibrated and validated. The model uses input data from a 3D biogeochemical model that provides biomasses in the different compartments, i.e. phytoplankton, zooplankton and bacteria; and from a 3D fate model that provides the herbicides concentrations in the water column as well as in the sediments. Simulated data were compared with experimental data collected during a set of sampling campaigns carried out in 2004 and 2005 in the Sacca di Goro lagoon (Northern Adriatic). The model predicts correctly the concentrations of herbicides measured in Ulva rigida and reproduces with good agreement the values of concentration of herbicides found in clams. Furthermore, the simulated spatial and temporal dynamics in the biota compartment, following those of the water and sediments, are also in agreement with the experimental data. This integrated approach combining biogeochemical, fate and bioaccumulation models provide an overall assessment of the importance of the different environmental compartments and it can also support the testing of different management strategies to improve ecosystem state and functioning. Further research is necessary to elucidate the role and importance of the metabolism of these compounds by clams. Ó 2008 Elsevier Ltd. All rights reserved.
1. Introduction During the last decades coastal lagoons have been exposed to many anthropogenic pressures, e.g. urban, domestic, agricultural and/or industrial effluents, port use and management, aquaculture, fishing, etc., that are responsible for internal perturbations – pollution, sediment dredging, removal of indigenous species, changes in food web structure, etc. In particular, plant protection products occur in coastal lagoon water and sediments as a result of the impacts of upstream agricultural activities (Carafa et al., 2007). These contaminants have shown bioaccumulation tendency through the aquatic food web and toxic effects for aquatic biota. Ecotoxicological effects of s-triazines on algae (e.g. Nitschke et al., 1999), aquatic plants (e.g. Cedergreen and Streibig, 2005) and aquatic microorganisms in general (De Lorenzo et al., 2000) have been investigated. A complete review of most important studies on toxic effects of these contaminants can be found in Eisler (2000). A combined additive effect of s-triazines mixtures on algae has been described (Faust et al., 2001) and a low bioaccumulation potential of these contaminants has been noticed (Pérez-Ruzafa * Corresponding author. Tel.: +39 0332 789202; fax: +39 0332 785807. E-mail address:
[email protected] (J.-M. Zaldívar). 0045-6535/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.chemosphere.2008.10.058
et al., 2000). However, relatively few analytical field data are available for evaluating the toxic effects in marine fish (Ward and Ballantine, 1985) and in molluscs (e.g. Cheney et al., 1997; Boutet et al., 2004). Data of organic contaminants accumulated in the sediment and in the tissues of target species may provide an assessment of pollutant occurrence and distribution in aquatic ecosystems, acting as a time integrated measure (Pereira et al., 1996). Specifically, in coastal lagoon ecosystems, molluscs have been used as bioindicators of pollution because of their feeding behaviour and their scarce mobility, which make them particularly exposed to contamination both through water column and sediment, directly or after resuspension. Among molluscs, clams show a tolerance to pesticide, but less is known of their metabolic responses to contamination. Moreover, clams are farmed for human consumption and, if contaminated, may represent a potential risk for human health. The Manila clam (Tapes philippinarum) was introduced in the Italian lagoons in the 1980s and replaced almost completely the indigenous species Tapes decussatus, because of its higher growth and reproduction rates and greater resistance to pollution. T. philippinarum is a macrobenthonic filter-feeding species living in the uppermost oxidized layer of the sediment. About 30,000 tons of edible clams are
R. Carafa et al. / Chemosphere 74 (2009) 1044–1052
produced per year (Binelli and Provini, 2003), in particular the main producers of shellfish are Venice and Sacca di Goro lagoons. The Sacca di Goro lagoon is located in the northern part of the Po river delta and clam cultivation areas cover more than one third of the Sacca di Goro lagoon surface (Viaroli et al., 2006) with densities attaining about 2000–2500 adult individuals m2; such densities and sediment dredging due to harvesting activities, have a strong impact on the benthic system and in the biogeochemical cycles in the lagoon (Bartoli et al., 2001). Furthermore, seafood represents a significant means of contamination of human diet (Micheletti et al., 2004) and for this reason legal thresholds have been established in order to protect human health from a number of toxic compounds and complex mixtures of chemicals. A default maximum pesticide residue level in foodstuffs has been fixed at 0.01 mg kg1 (Regulation EC 396/2005). Along with field studies and monitoring activities, model tools are necessary to understand the fate and transport of contaminants and to assess their impacts on communities and ecosystems (Carafa et al., 2006). Modelling can also support the testing of different management strategies to improve ecosystem state (Marinov et al., 2007). In the management of hazardous chemicals the prediction of bioconcentration and bioaccumulation factors from water in aquatic organism has become a very important tool. The quantitative knowledge of uptake, metabolism, excretion and depuration processes of chemicals in the organisms is needed to predict the fate and bioaccumulation of contaminants along the food web (Moriarty and Walker, 1987). However, little information exists regarding the uptake dynamics of herbicides in clams (e.g. Uno et al., 1997; Nordone et al., 1998), the retention time, the metabolism pathways, the excretion rates, etc. All these processes are strictly related to specific physiological characteristics, feeding behaviour and metabolism of the aquatic organism and to the particular chemical–physical features of the compound (Pereira et al., 1996), for this reason it is difficult to make comparisons between different studies and to determine uptake and depuration constants. The objective of this work is to provide an evaluation tool for calculating the herbicide concentration in T. philippinarum edible clam and Ulva rigida seaweed from the contaminants values in the water column, specifically for the s-triazine family. For this reason a bioaccumulation model has been developed, implemented and validated. The model uses input data from a 3D biogeochemical model (Zaldívar et al., 2003; Marinov et al., 2008) that provides biomasses in the different compartments, i.e. phytoplankton, zooplankton and bacteria; and from a 3D fate model that provides the herbicides concentrations in the water column as well as in the sediments (Carafa et al., 2006). The bioaccumulation model is based on Thomann, 1989 for the clam module; and on Del Vento and Dachs (2002) for the phytoplankton/bacterial module. Samples of T. philippinarum and U. rigida were collected in five seasonal sampling campaigns in the Sacca di Goro lagoon during 2004–2005 (Carafa et al., 2007), and used to calibrate and validate the model. 2. Materials and methods 2.1. Study area Sacca di Goro is a shallow coastal lagoon with a surface area of 26 km2, an average depth of about 1.5 m and a volume of approximately 39 millions m3 as well as a time variable bathymetry and boundaries (Marinov et al., 2006). The lagoon is located in the southern part of the Po river delta (44.78–44.83 N and 12.25– 12.33 S) and its basic hydrodynamic forcing is linked with the tidal dynamics of the Northern Adriatic Sea. Sacca di Goro is connected to the sea by 2 mouths about 0.9 km wide (Fig. 1).
1045
The main fresh water input in Sacca di Goro lagoon comes from the Po di Volano canal (about 3.5 108 m3 y1) that flows directly into the lagoon and the Po di Goro deltaic branch, which has a similar discharge and whose inflow is controlled by sluices but periodically floods the lagoon. Additionally three irrigation canals, named Giralda, Bonello and Canal Bianco, with similar flows (2.0– 5.5 107 m3 y1), contribute to overall fresh water input. The Burana-Po di Volano watershed covers an area of 3000 km2. Approximately 800 km2 affect directly Sacca di Goro lagoon. About 80% of the watershed is dedicated to agriculture and present crop typologies include corn and wheat, rice, sugar beets, soybean and vegetables. The lagoon is one of the most important shellfish aquaculture systems in Italy. About 13 km2 of the aquatic surface are exploited for the Manila clam farms within the lagoon and support an annual production between 8000 and 15,000 tons of clams (second largest producer in Italy after Venice lagoon). The pesticides delivered by the Po plain affect coastal lagoon ecosystems in the Po River Delta. Herbicide contamination has been reported in freshwater, sediment (Baldi et al., 1991) and mussels (Baldi et al., 1994). Atrazine was found at high concentrations, which corresponded with its use as a pre-emergent herbicide for maize and in controlling annual broad-leaved weeds and annual grasses. Even though atrazine has been banned from Italy for more than 10 years, in two more recent studies (Carafa et al., 2007), atrazine and its metabolites are still detectable in water samples indicating a long persistence in the soil and in the aquatic environment. 2.2. Experimental data generation Sampling campaigns were performed in May, August and October 2004, February and April 2005. Water samples, sediment and U. rigida (when present) were collected in four sites into the lagoon, and in one reference site near to the mouth of communication with the sea (Carafa et al., 2007). Clams (Tapes philipinarum) samples were collected at two sites (S and V, see Fig. 1), located in the clam farms. To evaluate the load of contaminants coming from the watershed, water samples were collected along the main freshwater tributaries: Po di Goro, Po di Volano, Canal Bianco and Canal Giralda. For a detailed description of sampling campaign planning, sample collecting and handling as well as the analytical results see Carafa et al. (2007). 2.3. Model development Fig. 2 shows the integrated modelling approach using, as a framework, the hydrodynamic model COHERENS (Luyten et al., 1999) adapted to the Sacca di Goro lagoon (Marinov et al., 2006) in terms of terms of bathymetry, watershed inputs, exchanges with the Adriatic Sea and meteorological and oceanographic forcing (tides and waves). COHERES provides temperature, salinity and currents for the biogeochemical module of the lagoon (Zaldívar et al., 2003) and the contaminant fate module (Carafa et al., 2006). The biogeochemical module produces simulated data on nutrient cycles (in water column and in sediments), oxygen, organic matter and biota, including phytoplankton, zooplankton, bacteria, U. rigida and clams (Marinov et al., 2008), whereas the contaminant module gives concentration of contaminants in the water column and sediments. The biogeochemical model was assessed using long time series data from 1989 to 1999 (Zaldívar et al., 2003) in its zero-dimensional version and with a complete set of data from 1992 (Marinov et al., 2008) in its 3D version. The fate model was calibrated and validated using data from a series of experimental campaigns that measured the concentrations of plant protection products in the water column and in the
1046
R. Carafa et al. / Chemosphere 74 (2009) 1044–1052
Fig. 1. Sampling stations in Sacca di Goro lagoon (satellite image from Servizio Ambiente e Difesa del Territorio, Provincia di Ferrara, Rapporto Annuale 2004).
Fig. 2. Integrated modelling approach using COHERENS.
sediments (Carafa et al., 2007). These simulated results from the hydrodynamics, the biogeochemical and the fate modules are used as input data for the bioaccumulation model (Fig. 2). Even though in the original biogeochemical model six stage-classes of T. philippinarum clams based on size of organisms are present (Zaldívar et al., 2005), in this work only marketable individuals with ‘large’ size (40 mm, wet weight ww = 19 g) have been considered. This is done to compare simulated data with experimental concentrations since only marketable individuals were collected during the experimental campaigns. In addition, missing parameters for modelling the contaminant concentrations in seaweed and clams were obtained using experimental measurements in one station to calibrate the model’s parameters, whereas the concentrations in the other station have been used to validate the model results. The bioaccumulation model is able to calculate the values of the concentrations of s-triazines in phytoplankton, bacteria, seaweed and clams. In the clams compartment the direct uptake by filtrations as well as the indirect uptake by predation have been considered. The model is based manly on Thomann (1989) and Del Vento and Dachs (2002) for phytoplankton, bacteria and seaweed, and Solidoro et al. (2000) and Zaldívar et al. (2005) for clams.
2.3.1. Bioconcentration in phytoplankton, bacteria and seaweed Bioconcentration of contaminants by phytoplankton can be calculated assuming constant uptake and depuration rates and by modelling the water-phytoplankton exchange, as shown by Del Vento and Dachs (2002). The concentration of a chemical in phytoplankton, bacteria and seaweed (Ci, ng kg1) over time can be expressed as (Thomann, 1989):
dC i ¼ ku C w kd C i dt
ð1Þ
where Cw (ng m3) is the concentration of the chemical dissolved in water, ku (m3 kg1 h1) and kd (h1) are the uptake and depuration rates constants. Uptake and depuration constants can be parameterized as function of bioconcentration factors of the chemical, permeability (P, m h1) of the cell membrane and specific surface area of phytoplankton (Sp, m2 kg1) (Del Vento and Dachs, 2002):
Sp P BCF k u ¼ Sp P
kd ¼
ð2Þ
1047
R. Carafa et al. / Chemosphere 74 (2009) 1044–1052
In order to predict uptake and depuration rates it is necessary to know values for BCF and P. Since estimations of BCF and P exist only for a few number of compounds, these parameter has been calculated using empirical approximation (Del Vento and Dachs, 2002):
log BCF ¼ 1:085 log K ow 3:770 for log K ow < 6:4
ð3Þ
log P ¼ 1:340 log K ow 8:433 for log K ow < 6:4
ð4Þ
Bioconcentration of contaminants in U. rigida was calculated using the same equations applied for phytoplankton with a Sp assuming a disk shape with a radius of 65 cm and a thickness of 40 lm; the density of U. rigida (qUlva) was assumed equal to phytoplankton, i.e. 1025 kg m3. However, the concentrations calculated using the values obtained applying Eq. (2) were around five times lower than those measured experimentally. For these reasons the measured concentrations of station S were used to adjust the constants, whereas the concentrations in station V were used to validate the model results. Uptake and depuration constants of phytoplankton, bacteria and seaweed are listed in Table 1. 2.3.2. Bioaccumulation in clams The mass balance of contaminants in the tissue of clams can be defined as (adapted from Thomann, 1989):
dC i ¼ ku C w þ SF DIA C dia þ SF FLA C fla þ SF BAC C bac kd C i ke C i dt km C i kG C i
ð5Þ
where the first term indicates the uptake via water; the second, third and fourth terms indicate the uptake of contaminant by predation of diatoms, flagellates and bacteria (Sorokin and Giovanardi, 1995), respectively; whereas the other terms indicate losses of contaminants through depuration, excretion, metabolization and dilution effect of growth, respectively. 2.3.2.1. Bioconcentration and transfer efficiency in clams. Following Thomann (1989) the uptake rate for aquatic species can be expressed as
ku ¼ V E=wlp
ð6Þ 3
1
tion rate (QSF = 0.321) and ft(T) is the functional response of temperature for filtration:
fm ðTÞ ¼
T mv T T mv T ov
bv ðT mv T ov Þ
ebv ðTT ov Þ
ð9Þ
where the maximum temperature for filtration Tmv, the optimal temperature for filtration Tov and the coefficient of temperature filtration bv were set to 32 °C, 20.5 °C and 0.2 °C1, respectively, in line with the values proposed by Solidoro et al. (2000). The mean dry weight is calculated from the wet weight using the following allometric relationship
dw ¼ BSF wwPSF
ð10Þ
where BSF (0.0234) and PSF (1.26) are allometric coefficients of the dw/ww relation (Solidoro et al., 2000). The efficiency of the chemical uptake has been observed to be a function of Kow of the contaminant and of the weight of the organism. Konemann and van Leeuwen (1980) measured the uptake efficiency for a range of different fish weights and found an empirical correlation, represented by the following log linear equation for organisms of order
