Article pubs.acs.org/est
Modeling Uptake of Selected Pharmaceuticals and Personal Care Products into Food Crops from Biosolids-Amended Soil Ryan S. Prosser,*,† Stefan Trapp,‡ and Paul K. Sibley† †
School of Environmental Sciences, University of Guelph, Guelph, Ontario N1G 2W1, Canada Department of Environmental Engineering, Technical University of Denmark, Lyngby, 2800, Denmark
‡
S Supporting Information *
ABSTRACT: Biosolids contain a variety of pharmaceuticals and personal care products (PPCPs). Studies have observed the uptake of PPCPs into plants grown in biosolids-amended soils. This study examined the ability of Dynamic Plant Uptake (DPU) model and Biosolids-amended Soil Level IV (BASL4) model to predict the concentration of eight PPCPs in the tissue of plants grown in biosolids-amended soil under a number of exposure scenarios. Concentrations in edible tissue predicted by the models were compared to concentrations reported in the literature by calculating estimated human daily intake values for both sets of data and comparing them to an acceptable daily intake value. The equilibrium partitioning (EqP) portion of BASL4 overpredicted the concentrations of triclosan, triclocarban, and miconazole in root and shoot tissue by two to three orders of magnitude, while the dynamic carrot root (DCR) portion overpredicted by a single order of magnitude. DPU predicted concentrations of triclosan, triclocarban, miconazole, carbamazepine, and diphenhydramine in plant tissues that were within an order of magnitude of concentrations reported in the literature. The study also found that more empirical data are needed on the uptake of cimetidine, fluoxetine, and gemfibrozil, and other ionizable PPCPs, to confirm the utility of both models. All hazard quotient values calculated from literature data were below 1, with 95.7% of hazard quotient values being below 0.1, indicating that consumption of the chosen PPCPs in plant tissue poses de minimus risk to human health.
■
plant and translocated to the shoots of the plant. Wu et al.13 observed the uptake and translocation of carbamazepine, diphenhydramine, fluoxetine, triclosan and triclocarban in soybean (Glycine max) grown in biosolids-amended soils, and triclosan and triclocarban were detected in the seeds of the plants. While these studies show that plants grown in biosolidsamended field can take up PPCPs, there remains considerable uncertainty regarding what risk these residues pose to ecological and human receptors. For most PPCPs found in biosolids, there is dearth of data on uptake by and residues in plants grown in biosolidsamended soil. Therefore, when attempting to characterize the exposure of organisms (including humans) that may consume the plant tissue, for the purpose of chemical risk assessment, concentrations of PPCP residues in plant tissue would be estimated through the use of models. The challenge is choosing a model that will predict relevant tissue residues, and consequently provide an accurate characterization of exposure. Extreme over- or under-estimation of concentrations could
INTRODUCTION The land application of biosolids provides an important source of nutrients and organic matter to agricultural fields, resulting in improved crop yield and soil tilth and reduced fertilizer costs to farmers. The land application of biosolids allows for rural-urban nutrient cycling, which will become increasingly important for the sustainability of agriculture as a result of a growing population and the potential for nutrient scarcity in many regions.1 However, biosolids have been found to contain a variety of inorganic and organic contaminants.2−4 Many of these contaminants (e.g., metals, industrial pollutants) are monitored, and their concentration is a factor in determining whether biosolids are applied to agricultural land.4 Pharmaceuticals and personal care products (PPCPs) are classes of organic chemicals that are frequently found in biosolids but are not regularly monitored and therefore are not considered in the decision to amend soil with biosolids.3,5 A number of laboratory and field studies have observed that PPCPs can be taken up by plants that are grown in biosolidsamended soils.6−14 Holling et al.9 investigated the potential for Chinese cabbage (Brassica campestris) to accumulate carbamazepine, sulfamethoxazole, salbutamol, triclosan, and trimethoprim derived from biosolids-amended soil. Carbamazepine, salbutamol, and triclosan were taken up into the roots of the © 2014 American Chemical Society
Received: Revised: Accepted: Published: 11397
June 24, 2014 September 9, 2014 September 10, 2014 September 10, 2014 dx.doi.org/10.1021/es503067v | Environ. Sci. Technol. 2014, 48, 11397−11404
Environmental Science & Technology
Article
Table 1. PPCPs Chosen from USEPA Targeted Sewage Sludge Survey for Use in This Study
α
Ref 3. βRef 35. γRef 36. δRef 37. εRef 38. ζRef 39.
Partitioning and advection with water are the transport processes considered in the model to describe movement of chemical throughout the plant. A cell model is coupled with the dynamic plant uptake model.21−23 The cell model is used to calculate partition coefficients (e.g., cytosol to soil, cytosol to xylem, and xylem to root) that are then used in the dynamic plant uptake model to estimate transport into different tissues of the plant. The cell model was used for all of the PPCPs included in the current study. Biosolids-Amended Soil Level IV Model. The BASL4 model is a level IV fugacity model that has been specifically designed for assessing the fate of biosolids-derived chemicals in amended soil and concentrations in plant, invertebrate, and mammal receptors.15,24,25 The plant portion of BASL4 includes both an equilibrium partition model (EqP) and a dynamic carrot root model (DCR) that considers the rate of water uptake. EqP and DCR were used in this study to determine the concentration of residues in the root and shoot of a simulated leafy tuber plant. The equations of the model are described in detail in Hughes and Mackay,15 Webster and Mackay,25 and Hughes et al.24 Parameterization of Models. DPU Model. Environmental Parameters. Air temperature, soil temperature, precipitation, and relative humidity parameters were produced by calculating the arithmetic mean of daily average values collected at University of Guelph Elora Research Facility weather station (43 39N, 80 25W) from 2010 to 2012 (Supporting Information, SI, Tables S1 and S2).26 Estimated monthly potential evaporation parameters were determined using data collected from the Climate Prediction Center of the United States National Weather Service (SI Table S1).27 Soil and Chemical Parameters. A single soil type (i.e., loam) was used in the simulation; the physical characteristics for the soil are presented in SI Table S3. A combination of frequency of detection and concentration reported in a sewage sludge survey conducted by the United States Environmental Protection Agency (USEPA) was used to determine the PPCPs modeled in this exercise (Table 1).3 Antibiotics (i.e., ofloxacin, ciprofloxacin, azithromycin, tetracycline, and doxycycline) were not included in this study due to the molecules containing
have significant consequences for risk characterization. The objective of this study was to evaluate and compare two models, the Dynamic Plant Uptake (DPU) model and the Biosolids-amended Soil Level IV (BASL4) model, in terms of their ability to predict PPCP concentrations in food crops grown in biosolids-amended soils under a variety of scenarios.15,16 These two models were chosen because they were specifically designed and have been used to investigate the uptake of organic chemicals from biosolids-amended soil. The concentrations of PPCPs in plant tissue derived from model outputs were then compared to concentrations reported in the peer-reviewed literature. Furthermore, to evaluate the ability of the models to predict PPCP exposure and resulting hazard to human health, estimated daily intake (EDI) values determined from modeled and experimental data were then compared to allowable daily intake (ADI) values. Hughes and Mackay15 demonstrated that the BASL4 model could effectively model the fate of biosolids-derived triclosan in soil. However, the model overestimated biota-soil accumulation factor (BSAF) values for triclosan in earthworms, but accumulation of triclosan in plant tissue was not included in the study.15 In a study conducted by Trapp and Eggen,16 the DPU model was able to predict the pattern and level of uptake of three organophosphates, an insect repellant, and a plasticizer into barley plants grown in bisolids-amended soil. However, to our knowledge, an evaluation of the ability of the BASL4 model and DPU model to predict the uptake of biosolids-derived PPCPs into the tissue of plants has not been presented in the literature.
■
MATERIALS AND METHODS Models. Dynamic Plant Uptake Model. The DPU model is composed of a dynamic plant uptake model based on the multicascade approach coupled with a tipping buckets model that describes water and solute transport in soil.17−21 Detailed descriptions of the equations used in the dynamic plant uptake model and the water and solute transport in soil model are presented in Rein et al.19 and Legind et al.,18 respectively. 11398
dx.doi.org/10.1021/es503067v | Environ. Sci. Technol. 2014, 48, 11397−11404
Environmental Science & Technology
Article
an agricultural field in the province of Ontario and the upper range in most jurisdictions.5,32,33 Biosolids amendment was simulated to occur on April 15th for carrot, radish, and lettuce plants, and April 1st for wheat and soybean plants. This amendment date represents a worst-case PPCP exposure scenario for carrot, radish, and lettuce cultivation. According to OMAF BMPs, a 12-month preharvest waiting period needs to be applied when land applying biosolids in the cultivation of carrot, radish, or lettuce.5,32 Therefore, the model was also run with a simulated application event on August 4st of the preceding year for all plant species. BASL4 Model. Soil and Chemical Parameters. The simulated field was assigned an area of 1 ha. The physical and chemical properties of the soil simulated in this model are presented in SI Table S3. The bioturbation rate (cm/yr) was given a value of 0.3.15,34 The bioavailability factor was given a value of 1 to neglect sequestration and to simulate a high exposure scenario. The soil compartment of the model is divided into two layers for the purpose of addressing the surface application, injection, and/or ploughing of biosolids in the amendment process. If biosolids are applied to the surface, then the chemical contained within the biosolids is assumed to be evenly distributed in the first layer of soil. If the biosolids are applied by injection or ploughed into the soil, then the chemical is assumed to be evenly distributed in the two soil layers. The two soil layers were given a depth of 0.2 m and a diffusion distance of 0.1 m.15 The fractions of fast-degrading, slowdegrading, and nondegrading organic carbon were given values of 0.5, 0.2, and 0.3, respectively.15 The fractions of organic carbon are separated into two classes to account for the fasterdegrading organic matter contributed to the soil from the biosolids and the slower-degrading organic matter present in the soil before amendment.15 The half-lives of fast-degrading and slow-degrading organic carbon were give values of 60 and 360 days, respectively. Physical and chemical properties for PPCPs modeled using the BASL4 model are presented in SI Table S4. Partition coefficients (i.e., KOC, KOW) and water solubility of the neutral PPCP molecules were replaced with distribution ratios (i.e., DOC, DOW) and adjusted water solubility to address the speciation that occurs at the chosen soil pH (i.e., 7.8) (SI Table S4). A value of 1 was used for the mineral matter-water partition coefficient (KMW) parameter.25 Plant Parameters. The fraction of water and lipid in the leafy tuber root were set to 0.89 L/kg and 0.025 g/g, respectively, and the leaf was set to 0.70 L/kg and 0.01 g/g, respectively. The volume and water uptake rate of DCR was set at 0.0001 m3 and 1 L/day, respectively. These two parameters used in DCR could not be manually altered in BASL4. Biosolids Application Parameters. A 24-h time step was used for the simulation with a total simulation time of 1800 h, which corresponds to a 2.5-month time period. The 2.5-month time period represents the time required to grow a carrot plant to harvestable size. It was assumed that the initial concentration in the soil layers was 0 μg/g dw. A single application event applied biosolids to the surface at a rate of 5 and 22 t dw/ha at time 0 h and a ploughing event was simulated at time 24 h. Amendment rates of 5 and 22 t dw/ha represent a rate recommended by the OMAF NMAN3 software for carrot cultivation and a maximum allowable application rate.5,32 It was assumed that organic carbon in the biosolids was 50% fast degrading and 50% slow degrading.25 Human Health Risk Assessment: Modeled vs Literature Data. Peer-reviewed literature was searched for studies
multiple functional groups that ionize at soil-relevant pH values, which the models were not designed to address.3 The greatest concentration of the chosen PPCPs reported in the survey was used in the model simulations.3 Physical and chemical characteristics for the chosen PPCPs are presented in SI Table S4. The longest dissipation half-life under conditions reflecting an agricultural field cited in the literature was used for each chemical. If an experimentally determined dissipation halflife value was not available in the literature, then it was estimated using the Level III Fugacity Model in USEPA EPI Suite.28 The log KOW value for the ionic form of the PPCPs was estimated by ACD iLabs.29 For calculations of partitioning inside plant tissue with the cell model, a value of 1 was added (acids) or subtracted (bases) from the pKa value to correct for the depressed ionization in cellular membranes due to the lower dielectrical constant.23 The partition coefficients KOC and KAW of the neutral PPCPs molecule were replaced with distribution coefficients DOC and DAW to address the speciation that occurs at the chosen soil pH (i.e., 7.8) (SI Table S4). A default value of 1 × 10−12 d−1 was used for the degradation rate of the chemical in individual plant tissues.30 A value of 1.099 was used for the temperature correction constant.16 Plant Parameters. The model was used to simulate uptake into carrot (Daucus carota), radish (Raphanus sativus), lettuce (Lactuca sativa), spring wheat (Triticum aestivum), and soybean (Glycine max). The germination day for carrot, radish, and lettuce plants was set to May 1st and the germination day for wheat and soybean plants was set to April 15th. The predicted concentrations of PPCPs in carrot and lettuce tissues on July 1st and radish tissues on June 1st were tabulated. Predicted concentrations in wheat and soybean tissue on July 31st were tabulated. These dates for the five plant species were chosen as they represent a theoretical harvest event. Plant-species specific physical parameters (i.e., transpiration coefficient, lipid-octanol coefficient (b), specific surface area, initial and final mass, growth rate, water content, and lipid content) used for different portions (i.e., root, stem, leaf, and fruit) of plants are presented in SI Table S5. The same values used by Trapp31 for volume, surface area, inner pH, ionic strength, and electrical potential across membrane parameters for cytosol, vacuole, xylem, and phloem were used in the model (SI Table S6). The model also required the lipid, protein, and water content of cytoplasm, vacuole, phloem, and xylem. Lipid and water content parameters for specific plant tissues were used for the lipid and water content of the cytoplasm and vacuole (SI Table S5).31 Protein content for the specific plant tissue was used only for the protein content of cytoplasm. A value of 0 was used for the protein content of vacuole, phloem, and xylem and the lipid content of phloem and xylem. The water content of phloem and xylem were given a value of 1. Biosolids Application Parameters. The application rate of biosolids was determined using the NMAN3 software available from the Ontario Ministry of Agriculture and Food (OMAF).5,32 Dewatered anaerobically digested municipal biosolids matching those described in Prosser et al.11 were used in this simulation. The same conservative approach and assumptions used by Prosser et al.11 for determining biosolids application rate were used in this simulation. The application rate of biosolids for carrot, radish, lettuce, spring wheat, and soybean were 5.0, 2.8, 6.2, 4.9, and 6.7 t dry weight (dw)/ha. The simulation was also run using the maximum application rate of 22 t dw/ha for all crop species. This application rate represents the greatest application rate that could be applied to 11399
dx.doi.org/10.1021/es503067v | Environ. Sci. Technol. 2014, 48, 11397−11404
Environmental Science & Technology
Article
Figure 1. Hazard quotient values for adults based on residues of eight PPCPs in edible tissue of plants predicted by the BASL4 (DCR and EqP) and DPU model and reported in the literature.
screening values at Superfund sites (i.e., 0.085).41 The term m represents the average mass of a toddler or adult based on Statistics Canada surveys (i.e., toddler (1 to
