Environmental Toxicology and Chemistry, Vol. 28, No. 4, pp. 701–710, 2009 䉷 2009 SETAC Printed in the USA 0730-7268/09 $12.00 ⫹ .00 DEGRADATION AND ECOTOXICITY OF THE BIOMEDICAL DRUG ARTEMISININ IN SOIL KARINA K. JESSING,*† NINA CEDERGREEN,‡ JOHN JENSEN,§ and HANS C.B. HANSEN† †University of Copenhagen, Department of Basic Sciences and Environment, Thorvaldsensvej 40, 1870 Frederiksberg C, Denmark ‡University of Copenhagen, Department of Agricultural Sciences, Højbakkegård Allé 13, 2630 Taastrup, Denmark §Aarhus University, National Environmental Research Institute, Vejlsøvej 25, 8600 Silkeborg, Denmark ( Received 3 April 2008; Accepted 6 October 2008) Abstract—The plant Artemisia annua L. is cropped in many countries for production of the antimalarial drug artemisinin. Artemisinin is phytotoxic and has insecticidal activity. Large-scale cultivation of A. annua may cause transfer of artemisinin to soil and, hence, may affect both soil organisms and the aquatic environment if the compound is leachable. In the present study, a new method for extraction of artemisinin from soil was developed, and field concentrations and degradation kinetics of artemisinin in sandy and loamy soils were measured. The soil concentrations in a Danish A. annua field were up to 11.7 mg/kg. The degradation kinetics could be modeled as the sum of two first-order reactions, a fast initial degradation followed by a reaction that was 11- to 25-fold slower. It took at least 35 d before artemisinin could not be detected (⬍0.36 mg/kg) at 20⬚C, classifying artemisinin as being relatively persistent in the environment. Combined with its water solubility of 49.7 ⫾ 3.7 mg/L, this makes it potentially leachable. In soil, artemisinin repelled the earthworm (Eisenia fetida; the 10 and 50% effect concentrations [EC10s and EC50s, respectively] were 5.24 ⫾ 2.64 and 21.57 ⫾ 4.73 mg/kg, respectively) and inhibited growth of lettuce (Lactuca sativa L.; EC50, 2.48 mg/kg). Springtails (Folsomia candida) were not affected in the tested concentration range of 1 to 100 mg/kg. Artemisinin had toxicity to the freshwater algae (Pseudokirchneriella subcapitata; EC50, 0.24 ⫾ 0.01 mg/L) and duckweed (Lemna minor; EC50, 0.19 ⫾ 0.03 mg/L) similar to that of the commercial herbicide atrazine. Based on the presented data, the risks of adverse environmental effects because of cultivation of A. annua are high and comparable to those when using commercial pesticides. Keywords—Natural toxins Artemisia annua L. Qinghaosu Degradation kinetics Toxicity modium, the malaria parasite [7]. Artemisia annua is of Asian origin, but the plant is widely dispersed throughout the temperate region and has become naturalized in many European countries and the United States [8]. Artemisia annua is cropped at large scale in Asia and the Middle East for medicinal purposes, and in Africa, cultivation is in the establishing phase [9]. In addition, A. annua is cultivated for experimental purposes in The Netherlands, Switzerland, Finland, and Denmark [10]. At present, chemical synthesis or in vitro production of artemisinin is not economically feasible [11]. Thus, A. annua is a potential new medicinal crop for temperate areas. Artemisia annua will have to be cropped at large scale to satisfy the need for medicine, because 40% of the world population is threatened by malaria [8]. The mechanism of action of artemisinin in Plasmodium sp. includes activation and alkylation. The first step in the mechanism comprises reductive cleavage of the peroxide bond, facilitated by Fe(II), leading to formation of oxygen-centred radicals, which in turn can transform to carbon-centered radicals [12]. The proposed second step, alkylation, involves the formation of covalent adducts between the activated intermediate and specific parasite membrane-associated proteins [13]. It also has been suggested that other metal ions, such as Co(II), Cu(II), Ni(II), Ti(IV), and Mn(II), can perform scissoring of the endoperoxide bridge in the presence of an excess of cysteine, which acts as a hydrogen donor [14]. Artemisinin is stored in subcuticular, extracellular space in glandular trichomes [9] located on the surface of the leaves and stem [15] as well as on the corolla and on receptacles of the florets [16]. Production of artemisinin appears to peak with INTRODUCTION Many plant species produce secondary metabolites with the purpose of defending themselves against herbivores or pathogenic microbes [1]. In most cases, such substances are quite toxic toward both target and nontarget organisms in the environment, similar to what is seen for synthetic pesticides. Especially when monocultures of toxin-producing plants are cultivated, the soil and aqueous environment can be exposed to high concentrations of toxic substances. Introduction of toxin-producing plants to new environments therefore might cause severe effects, because no adaptation to the compounds in question has occurred among organisms in that environment. The situation becomes even worse if the introduced plant becomes invasive, as is seen for Partenium hysterophorus L., which produces the natural toxin parthenin, a sesquiterpene lactone [2]. Incorporation of such plants in soil can cause release of toxic substances to soil in concentrations that affect soil organisms, pests, and beneficial organisms alike [3,4]. The toxic substances also can be transferred to the soil with defoliation or leached off from the leaves by rain. Finally, decomposition of leaf material can cause release to soil [2]. Sweet wormwood (Artemisia annua L.) synthesizes and accumulates artemisinin [5], a sesquiterpene lactone with an endoperoxide bridge (Table 1) [6]. Artemisinin is available commercially as an antimalarial drug that is efficacious against drug-resistant strains of Plas* To whom correspondence may be addressed (jessing@life.ku.dk). Published on the Web 11/12/2008. 701 702 Environ. Toxicol. Chem. 28, 2009 Table 1. Chemical structure and selected chemical properties of artemisinina Property CAS no. IUPAC name Description 63968-64-9 (3R,5aS,6R,8aS,9R,12S,12aR)-octahydro3,6,9-trimethyl-3,12-epoxy-12Hpyrano[4,3-j]-1,2-benzodioxepin10(3H)-one Chemical structureb Molecular formula Molar mass Solubility in water Log KOW Log KOC Henry’s law constant: C15H22O5 282.2 g/mol 49.7 ⫾ 3.7 mg/Lc 2.90 L/kgd 2.51 L/kgd 4.92 ⫻ 10⫺9 atm·m3/mold a CAS ⫽ Chemistry Abstract Service; IUPAC ⫽ International Union of Pure and Applied Chemistry. b Liu et al. [6]. c Present study. d Calculated with EPIwin (Ver 3.12; U.S. Environmental Protection Agency, Washington, DC). flowering, and the content of artemisinin in the aboveground plant parts usually is in the range of 0.01 to 0.40% dry weight [16]. Some clones, however, can produce more than 2% artemisinin [7]. The total production of artemisinin by A. annua is approximately 7.5 kg/ha [10]. At least some of this artemisinin will be released to the soil either via dead plant material, leaching off leaves by rain, or incorporation of plant parts left over after harvest. To our knowledge, no investigations of artemisinin degradation in soil have been performed, and even though organic peroxides are widely used in the industry, the environmental fate of such compounds has hardly been investigated. The estimated log octanol–water partition coefficient (KOW) of artemisinin of 2.90 and log organic carbon partition coefficient (KOC) of 2.51 (Table 1) indicate that the compound is not strongly sorbed by partitioning into soil organic matter. Hence, leashing of artemisinin to surface water or groundwater cannot be excluded. Because artemisinin forms adducts with proteins in the malaria parasite, it is possible that similar reactions may occur with other proteins, including enzymes in soil. Artemisinin may show adverse effects in the soil environment, because artemisinin is toxic to both insects [17,18] and plants [19,20]. Duke et al. [19] revealed that artemisinin has selective phytotoxic properties, because it inhibited germination of lettuce (Lactuca sativa L.) and even A. annua at a concentration of 0.33 ␮M, but not that of redroot pigweed (Amaranthus retroflexus L.) and pitted morningglory (Ipomoea lacunosa L.) at the tested concentrations. To evaluate whether an increased commercial growth of A. annua with high artemisinin contents could pose a risk to the local environment, the environmental concentrations of artemisinin K.K. Jessing et al. must be estimated and compared to the concentrations that can cause an adverse effect in nontarget organisms. The aim of the present study was to determine the concentration and degradation rate of artemisinin in soil, to evaluate effects linked with exposure to artemisinin in terrestrial and freshwater organisms, and to provide a preliminary characterization of the risks associated with these effects. To do this, an extraction method for artemisinin from soil was developed, and concentrations were measured both under field conditions and under controlled laboratory conditions. To estimate toxicity to soil organisms, a chronic test on the soil insect Folsomia candida was conducted together with an avoidance test on the earthworm Eisenia fetida and a germination and growth test on lettuce. Because artemisinin has a potential for leaching to the aquatic environment, its toxicity toward the algae Pseudokirchneriella subcapitata and the floating plant Lemna minor also was measured. No data regarding water solubility of artemisinin were available. To be able to dose artemisinin at realistic bioavailable concentrations in the biological experiments, water solubility therefore also was experimentally determined. MATERIALS AND METHODS Chemicals Artemisinin (purity, 98%) was provided by Sigma-Aldrich (Brøndby, Denmark). Ethanol (purity, 96%) was provided by Kemetyl (Køge, Denmark). Methanol and acetonitrile, both of high-pressure liquid chromatography (HPLC) grade, were provided by Sigma-Aldrich. Sodium hydroxide (NaOH) was provided by J.T. Baker (Herlev, Denmark). Acetic acid (purity, ⬎99.8%), Na2HPO4, and NaH2PO4 were provided by Merck (Glostrup, Denmark). Gypsum (CaSO4 · 0.5H2O) was provided by Borup Kemi (Køge, Denmark). Charcoal wood powder (particle size, ⬍0.15 mm) was provided by Merck, and baker’s yeast was provided by V&S Distillers (Aalborg, Denmark). Quantification of artemisinin Determination of artemisinin was performed using the method developed by Zhao and Zeng [21]. Artemisinin was converted to the strongly ultraviolet-absorbing compound Q260 by a precolumn reaction. Dry soil extract was dissolved in 1 ml of 96% ethanol and treated with 4 ml of 0.2% (w/v) NaOH at 50⬚C for 30 min and cooled to room temperature. The solution was then acidified with 5 ml of 0.08 M acetic acid and filtered through a Millipore (Copenhagen, Denmark) filter (pore size, 0.45 ␮m) before determination as Q260 by HPLC. The HPLC used was an Agilent 1100 series (Agilent Technologies, Waldbronn, Germany). The samples were separated on a Supelco Discovery Bio C18 Bio wide-pore column (length, 25 cm; inner diameter, 4.6 mm; film thickness, 5 ␮m; Supelco Park, Bellefonte, PA, USA) fitted with a Supelco Discovery Bio wide-pore C18 guard column (length, 2 cm; inner diameter, 4.0 mm; film thickness, 5 ␮m). The mobile phase was a methanol/acetonitrile/0.9 mM Na 2HPO 4–3.6 mM NaH2PO4 buffer (pH 7.76) solution (45%/10%/45%, v/v/v), and the injection volume was 20 ␮l. The elution speed was 1 ml/min, and the detection wavelength was set at 260 nm. The limit of detection was determined to be 0.18 mg/L, or 0.36 mg/kg soil, calculated as being the lowest measurable signal plus threefold the standard deviation for 10 measurements of samples with a low artemisinin concentration. Because the majority of the samples in the present study were Artemisinin in soil Environ. Toxicol. Chem. 28, 2009 703 Table 2. Characterization of soils used in the study Textureb Soil Depth (cm) pHa Clay (%) Jyndevad, Denmark (sandy) Sjællands Odde, Denmark (clayey) Aarslev, Denmark (loamy) 0–25 0–30 0–20 6.9 7.2 6.5 5 19 14 Silt Sand (%) (%) Nc (%) Cd (%) 3 18 15 0.12 0.14 0.13 2.43 1.25 1.35 92 63 71 CaCO3e FeOXf AlOXf FeCBDg AlCBDg (%) (mg/kg) (mg/kg) (mg/kg) (mg/kg) NMh 0.4 NM 1,440 1,760 3,390 1,010 750 1,540 2,220 4,080 4,950 1,030 760 1,340 a pH measured in 0.01 M CaCl2. Particle sizes as determined by the hydrometer method and sieving: Clay, ⬍2 ␮m; silt, 2–20 ␮m; and sand, ⬎20 ␮m. c As determined by the Kjeldahl method. d As determined by dry combustion. e As determined using potentiometric titration. f Oxalate-extractable Fe and Al. g Citrate-bicarbonate-dithionite–extractable Fe and Al. h NM ⫽ not measured. b soil extracts and quantification was performed in dissolved soil filtrates, interferences of soil solutes were tested. This was done by comparing the slopes of standard curves produced from artemisinin dissolved in soil solution and in pure ethanol. Quantification was unaffected by soil matrix, because no interfering peaks or reactants were present and the slopes were identical. Determination of water solubility The water solubility of artemisinin was determined following the Organization for Economic Co-operation and Development Guideline for the Testing of Chemicals, Water Solubility [22]. A 10-ml Econo column with a reservoir of 250 ml was connected to a peristaltic pump (Alitea, Copenhagen, Denmark) equipped with polyvinyl chloride tubings (inner diameter, 1.85 mm). Artemisinin (70 mg dissolved in 100 ml of acetone) was coated on 14-g glass beads (diameter, 0.25–0.50 mm; Art. A 553, 1; Carl Roth, Karlsruhe, Germany) washed twice with acid and rinsed with double deionized water; acetone was removed using a rotary evaporator (25⬚C). The column was packed with the coated glass beads, and double distilled water was circulated through the column at a flow of 100 ml/h to replace the bed volume 10 times per hour. Samples of 1 ml were taken at 1, 2, 3, 4, and 5 h after the start and then at 24-h intervals. The artemisinin content was determined using the method described above. The experiment was continued until the artemisinin concentrations in five successive samples did not differ by more than 30%. Then, the flow was decreased to 50 ml/h, and samples were collected over the following 10 days. The criteria were that five successive samples at this flow did not differ by more than 30%. Because no significant difference was found between the measurements at the two flows, the stated water solubility is an average of all samples at both flow rates. Soil materials A sandy, a loamy, and a clayey Danish agricultural soil were used in the experiments. The sandy soil was from Jyndevad in the southwestern part of Denmark. It was developed on glacifluvial material and has been classified as a Humic Psammentic Dystrudept [23]. The loamy soil was from Aarslev in the middle of Denmark. It was developed on till plain from the late Weichelian glaciation and has been classified as a Typic Agrudalf [24]. The clayey soil was from Sjællands Odde in the northeastern part of Denmark. It was developed on calcareous clayey lodgement and melt-out till from the Weichse- lian glaciation and has been classified as a Typic Agriudoll [23]. Selected characteristics of the soils used are shown in Table 2. Soil material was sampled from the A horizons (the uppermost layer of the soil [depth, 0–30 cm]) and air-dried. Then, the soil was sieved (mesh size, 2 mm) and stored dry at 20⬚C. All soils have approximately neutral pH. The content of carbon in humic matter differs, but this content is lowest for the most clayey soil, which, however, has the lowest carbon to nitrogen ratio, reflecting its more fertile status. In addition, the loamy soil has the highest content of citrate-bicarbonatedithionite–extractable iron and aluminum and the lowest ratio between oxalate and citrate-bicarbonate-dithionite–extractable iron, reflecting that the iron oxides are more crystalline in this soil compared with those in the other soils. The loamy soil, however, still contains the highest amount of oxalate-extractable iron and aluminum. The sandy soil was chosen for use in all the ecotoxicity tests, because it was easier to obtain a homogenous mixing of artemisinin with this soil material—a prerequisite for reproducible soil ecotoxicity tests. In the degradation kinetic experiments and ecotoxicity tests, soils were spiked with known amounts of artemisinin. Spiking was performed using a modification of the method developed by Brinch et al. [25]. Five days before spiking, the soil was wet again to a gravimetric water content of 15 to 18% (corresponding to 60% of field capacity) and incubated in the dark at 20⬚C. A stock solution of artemisinin dissolved in acetone (5 mg/ml) was made. One-quarter of the soil amount needed in the experiment was mixed with 1 ml of acetonic stock solution per 10 g of soil. The mixed soil was stored under an exhaust device for 24 h, allowing the acetone to evaporate. Then, the soil was wet again to compensate for water loss and mixed thoroughly with the remaining three-quarters of soil. In the ecotoxicology tests, the stock solution was diluted by a factor of 2.5 to each soil start concentration, and the same mixing procedure was followed. In all soil experiments, the gravimetric water content was determined using the following procedure: Less than 20 g of wet soil was weighed before and after drying at 105⬚C for 24 h. The gravimetric water content was calculated as mass of water loss (g)/mass of dry soil (g). Soil sampling in A. annua field Soil samples were collected randomly in a Danish A. annua field situated at Aarselv in the middle of Denmark (Table 2, loamy soil). The A. annua seedlings were planted in the field at eight weeks old on May 15, 2007. Sampling dates were June 12th, July 9th, August 20th, September 20th, November 704 Environ. Toxicol. Chem. 28, 2009 K.K. Jessing et al. Table 3. Recoveries of artemisinin from spiked soils using different solvents, mass of soil, and extraction time Parameters Solvent Ethanol Ethanol Ethanol n-Hexane Ethanol Ethanol Methanol Ethanol Solvent volume (ml) Soil mass (g) Soil moisturea 25 25 25 25 25 25 25 25 10 10 10 10 5 5 5 5 Dry Dry Dry Dry Moist Moist Moist Moist Concn.b Extraction time (mg/kg dry (h) wt) 1 1 24 24 20 24 24 24 12.4 37.2 13.1 13.1 ⬃15 ⬃8 ⬃8 ⬃3 Recovery (%)c Clayey soil 42 50 91 69 92 79 29 82.6 ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ 1.1 11 14 2.7 2.6 2.9 28 1.67 Sandy soil Loamy soil —d — 80 ⫾ 6.5 69 ⫾ 3.7 87 ⫾ 9.4 81 ⫾ 15.3 60 ⫾ 10 87.8 ⫾ 1.91 — — — — — — — 71.4 ⫾ 4.70 a Moist ⫽ 18% gravimetric water content. Spiked concentration of artemisinin. c Standard deviations based on triplicates. d — ⫽ not determined. b 30th, and December 22nd. Three soil samples were collected in glass flasks from two depths (0–2 and 2–5 cm). Each sample was pooled from three different spots and then mixed. The samples were stored at soil temperature (10⬚C) during transportation (⬍1 d) and then stored at ⫺21⬚C until analysis. Extraction of artemisinin from soil To develop a reliable method for extraction of artemisinin from soil, different solvents, soil to solution ratios, and extraction times were tested. Recoveries from these experiments are listed in Table 3. The soils were spiked and mixed for a few minutes, and then the extraction was initiated. A soil to solution ratio of 0.2 gave a better recovery than a ratio of 0.4. Ethanol was a better extraction solvent than methanol and n-hexane. Extraction times longer than 20 to 24 h did not result in further improvements of recovery efficiency. The final extraction method (Table 3, last row) was performed using the following procedure: 5.00 g of moist (15–18% gravimetric water content) soil were weighed into a 50-ml, round-bottomed, centrifuge glass tube (Hounisen Laboriatorieudstyr, Risskov, Denmark). Then, 25 ml of 98% ethanol were added, and the tube was shaken laterally with 80 strokes per minute. After 20 to 24 h, the tube was centrifuged for 6 min at 1,360 g (Hettich Zentrifugen, Universal 30F, Andreas Hettich, Tuttlingen, Germany). Subsequently, the supernatant was filtered quantitatively through a 20- to 25-␮m Whatman 41 (Schleicher and Schuell, Keene, NH, USA) filter into 50-ml glass vials. The filter paper was rinsed once with a few milliliters of 98% ethanol, and the filtrate was evaporated to dryness under a stream of air. The dry extract was stored at ⫺21⬚C until analysis. Degradation kinetics in soil Degradation kinetic experiments were carried out for the sandy and loamy soils. The clay soil was excluded from this experiment, because it was very difficult to obtain a homogeneous mixture with artemisinin. After wetting 500 g of dry sandy soil and 250 g of dry loamy soil to moisture contents of 18% and then reactivation for 5 d at 20⬚C [25], the soils were spiked with artemisinin according to the procedure described above, and each soil was incubated in 500-ml polyethylene flasks at 21⬚C. The initial artemisinin contents were 20.00 and 20.58 mg/kg dry weight for sandy and loamy soils, respectively. Three separate 5-g samples were taken from each flask at the following times, which were anticipated to cover the period until full degradation, and with highest sampling density in the first part of the degradation: 8 and 30 min; 2.5, 4.5, 6.5, 8.5, 11.5, 14, 16, 18, 20, and 24 h; and 6, 11, 20, 29, 44, and 64 d. The soil samples were extracted immediately after sampling for determination of artemisinin. Springtail test Folsomia candida, a soil-dwelling Collembola that reproduces asexually, was supplied from the laboratory culture at the National Environmental Research Institute at Aarhus University (Silkeborg, Denmark). Animals were reared on a substrate of a water-saturated mixture of gypsum and charcoal wood powder (8:1). The animals were kept at 20 ⫾ 1⬚C with a 12:12-h light:dark photoperiod and were fed a diet of dry baker’s yeast. The Collembola test was performed following the procedure described by Wiles and Krogh [26]. One-weekold eggs were collected and allowed to hatch over 3 d to produce a synchronized culture. Nonhatched eggs were removed and disposed of, and the hatched juveniles were grown until maturity. For F. candida, animals from 16 to 19 d of age were used in the experiments. The animals were added to the soil 24 h after spiking. Seven concentrations of artemisinin and two controls, with and without acetone, each with four replicates, were tested. Nominal artemisinin concentrations were 0, 0.3, 1, 3, 7, 15, 30, and 100 mg/kg dry soil and were chosen from pilot studies of earthworm avoidance and knowledge of measured soil artemisinin concentrations [20]. A gravimetric water content of 18% was used. The moist and spiked soil was then transferred to the test containers (plastic cylinders with a 1-mm nylon mesh at the bottom; length 55 mm; inner diameter, 60 mm). Ten individuals of F. candida (age, 16–19 d) were added to each test container, and granulated dry yeast (15 mg) was added as food source on top of the soil. Test containers were kept at 20 ⫾ 1⬚C with a 12:12-h light:dark photoperiod. After two weeks, all containers were weighed, and lost water was added together with a new portion of 15 mg of granulated dry yeast. At the end of the 21-d exposure, animals were extracted out of the soil and into beakers using a controlled temperature gradient extractor [27]. The beakers were frozen, which fixed the springtails so that they could be counted manually under a light microscope. Artemisinin in soil Earthworm avoidance test Specimens of E. fetida with a weight of approximately 1 g/worm were obtained from Hvide Sande Ormeeksport (Hvide Sande, Denmark). The subchronic test was performed following the procedure described by U.S. Environmental Protection Agency [28]. To conduct the avoidance test, containers (6 ⫻ 11 ⫻ 17 cm) with two sections divided by a vertical plastic divider were used. Half of the container was filled with artemisinin-spiked soil (0.5 kg), and the other half was filled with control (nonspiked) soil. All soil material had a gravimetric water content of 17%, as described previously in Soil materials. After packing the test containers with soil, the divider was removed. Ten adult E. fetida worms were placed in the middle of the container at the top of the soil. The containers were covered by a tight, plastic lid with eight 2-mm holes for aeration and placed in the dark (20⬚C) for 48 h. After incubation, the divider was reinserted between the spiked and nonspiked soil, and the number of worms in each container section was counted. Six concentrations and two controls, with and without acetone, each with 10 replicates, were tested. The nominal concentrations were 0, 1.03, 2.56, 6.4, 16, 40, and 100 mg/kg dry soil and were chosen on the basis of preliminary experiments. Seedling emergence and seedling growth test This test was performed following the Organization for Economic Co-operation and Development guideline [29]. Six artemisinin concentrations and two controls, with and without acetone, each with four replicates, were tested. The same concentrations were prepared as were used in the earthworm test. To conduct the test, large glass Petri dishes (diameter, 15 cm) were used. Each dish contained 690 g of soil with a gravimetric water content of 15%. In each Petri dish, 10 lettuce seeds (Lord Nelson; Brødrene Nelsons Frø A/S, Bergen, Norway) were planted, and the dishes were placed in climate chambers at 16⬚C and a 16:8-h light:dark photoperiod. After one week, seedling emergence was observed in all dishes, and the number of seedlings was counted. The dishes were moved to the greenhouse (temperature: 13⬚C night, 18⬚C day; light conditions: Artificial light from 8:00 AM to 10:00 PM if the sunlight did not exceed 30 ␮mol/m2/ s of photosynthetic active radiation [PAR]). The dishes were moistened when needed to prevent soil drying. After 21 d, the lettuce plants were harvested without roots and dried in an oven at 60⬚C overnight, and the dry weight per dish was recorded. Algae test The acute algae test was conducted according to the procedure described by Mayer et al. [30]. Seven artemisinin concentrations, each with three replicates, and two controls, with and without acetone, were tested. The controls each had 10 replicates. Artemisinin concentrations were 0, 0.0078, 0.016, 0.031, 0.063, 0.125, 0.25, and 0.5 mg/L. The concentration range was chosen on the basis of preliminary tests. Artemisinin was dissolved in acetone, and 250 ␮l of this solution were added to 0.5 L of growth medium in a 1,000-ml beaker. The solution was stirred magnetically and acetone allowed to evaporate for 3 h. Algae of the species P. subcapitata were used in the experiment and obtained from a culture (NIVA-CHL) supplied by the Institute for Water Research in Oslo, Norway (NIVA). To initiate the experiment, algae were incubated with Environ. Toxicol. Chem. 28, 2009 705 4 ml of growth medium [31] in 20-ml glass vials at a density of approximately 10,000 algae/ml. The algae vials were closed by a lid with a 2-mm hole to ensure gas exchange and then placed on a shaking table (300 rpm) in holders ensuring continuous illumination from below with 80 ␮mol/m2/s of PAR. At 0 h and after 24 and 48 h, 400 ␮l of the algae suspension were removed from each vial and mixed in test tubes with 1.6 ml of acetone saturated with MgSO4 · 7H2O (12 g/L). The test tubes were sealed with a tight lid and placed in the dark for chlorophyll extraction. After 72 h, chlorophyll fluorescence was measured for all algae samples on a Turner Quantec digital filter fluorometer (model FM109520-33; Barnstead Thermolyne, IA, USA) using an excitation wavelength of 420 nm and an emission wavelength at 670 nm. The relative growth rate of the algae culture was calculated as the slope of the linear regression of the ln-transformed fluorescence data as a function of time. Lemna minor test The L. minor test closely followed the method developed by Cedergreen and Streibig [32]. For this purpose, 10-ml, sixwell, TC-test plates (CM-LAB Aps, Vordingborg, Denmark) were used. Seven artemisinin concentrations, each with three replicates, and two control treatments, with and without acetone, were tested. The controls each had six replicates. Concentrations were the same as those used for the algae test and were chosen on the basis of preliminary tests. Artemisinin was dissolved in acetone, and 250 ␮l of this solution were added to 0.5 L of growth medium prepared according to the method described by Maeng and Khudairi [33]. The solution was magnetically stirred and acetone allowed to evaporate for 3 h. The L. minor plants (type UTCC 490) were purchased from Judy Acreman, University of Toronto Culture Collection of Algae and Cyanobacteria (Toronto, ON, Canada), and were kept aseptically in Erlenmeyer flasks in growth medium at 24⬚C, pH 5, and a continuous photon flux density of 85 to 120 ␮mol/m2/ s of PAR. The experiment was initiated by transferring one duckweed frond to every well. The plants were photographed with a digital camera alongside a 1- ⫻ 1-cm, white plastic square, and the total frond surface area was determined by pixel counts using the computer program Coral Photo Paint 12 (Corel, Ottawa, ON, Canada). The plants were then placed in a growth chamber at 25⬚C and a continuous photon flux density of 85 to 120 ␮mol/m2/s of PAR. After 7 d, the plants were photographed again, and the frond surface area was determined. Relative growth rates were calculated according to (ln At ⫺ ln A0)/t, where At is the surface area at time t and A0 is the initial surface area. Statistics Degradation kinetics were modeled as the sum of two firstorder reaction terms using the least-squares method (SigmaPlot 2002 for Windows, Ver 8.0; Systat Software, Chicago, USA): [Art] ⫽ Ffast · exp(⫺k fast · t) ⫹ (100 ⫺ Ffast ) · exp(⫺k slow · t) (1) where [Art] is the relative amount of artemisinin (%) at time t (d), Ffast is the percentage of artemisinin being degraded by the fast process, and kfast (1/d) and kslow (1/d) are the first-order rate constants of the fast and slow processes, respectively. Half-lives were estimated by nonlinear regression using R 2.5.0 [34]. Modeling of dose–response curves was done using R 2.5.0 and the add-on package, Dose–Response Curves 706 Environ. Toxicol. Chem. 28, 2009 K.K. Jessing et al. Table 4. Degradation kinetic parameters and half-lives of artemisinin in sandy and loamy soil at 21 ⫾ 1⬚C and approximately 17% gravimetric moisture content Parametersa Soil Sandy Loamy Ffast (%) kfast (1/d) kslow (1/d) t1/2 (d)b 41.8 ⫾ 9.97 55.4 ⫾ 10.02 0.584 ⫾ 0.16 2.064 ⫾ 0.68 0.0535 ⫾ 0.012 0.0817 ⫾ 0.029 4.22 0.90 a Ffast ⫽ percentage of artemisin being degraded by the fast process; kfast ⫽ first-order rate constant for the fast process; kslow ⫽ first-order rate constant for the slow process; t1/2 ⫽ half-life. b Half-lives were estimated by nonlinear regression using R 2.5.0 [34]. Fig. 1. Degradation kinetics of artemisinin in sandy (䢇) and loamy (䡬) soils at 21 ⫾ 1⬚C and approximately 17% gravimetric moisture content shown as relative content of artemisinin versus time (d). The insert shows in detail data from the first 24 h. Data are given as the average ⫾ standard deviation (n ⫽ 3). The degradation model (Eqn. 1) fitted to data is shown with full curves. (http://www.bioassay.dk). The three parameter log-logistic dose–response model was used for all the ecotoxicity tests: y⫽ d 1 ⫹ (x /e) b (2) where y is response, x is the artemisinin concentration (mg/ kg or mg/L), d is the upper limit of the curve corresponding to the response of the untreated control, e (mg/kg) is the dose at which the value of d is reduced by 50% (i.e., the 50% effect concentration [EC50]), and b is proportional to the slope around EC50. The model was used with the assumption of normally distributed data for gradual responses (growth and biomass) and of binominally distributed data for binominal data (earthworm presence). The present study attempted to determine if a significant difference existed between controls with and without acetone. This was done using a two-tailed t test, with the hypothesis that the difference in means equals zero in all experiments except the earthworm avoidance test, in which a ␹2 test with pairwise comparisons was performed. sandy and loamy soil, respectively, whereas after 5 d, a total of 46.8 and 29.6% of the artemisinin had degraded in the sandy and loamy soil, respectively. Artemisinin was detectable (⬎0.36 mg/kg) for 60 and 35 d in the sandy and loamy soils, respectively. Comparison of the rate constants (kfast and kslow) (Table 4) showed that in the sandy soil, the fast process was 11-fold faster than the slow process. In the loamy soil, however, the ratio between the rates of the fast and slow reactions was 25. Approximately 42 and 55% of the artemisinin was participating in the fast process in the sandy and loamy soil, respectively. Hence, the fast reaction is determining the halflife of artemisinin in soil. Water solubility The aqueous solubility of artemisinin was 0.176 ⫾ 0.013 mM (49.7 ⫾ 3.7 mg/L) at 21 ⫾ 1⬚C. This result is close to the estimated water solubility of 0.184 mM (51.9 mg/L) at 25⬚C (EPISuite, Ver 3.12; U.S. Environmental Protection Agency, Washington, DC). At 1 h after initiation of the experiment, a solubility close to the final value already was measured, and the measurement was stable over 9 d. The rapid equilibration shows that artemisinin dissolves quickly in water. Artemisinin concentration in an A. annua field Measurement of artemisinin concentrations in an A. annua field revealed soil concentrations ranging from 0.16 to 11.7 mg/kg in the upper 2 cm and up to 0.5 mg/kg at a depth of 2 to 5 cm (Fig. 2). The concentration of artemisinin increased RESULTS Degradation kinetics Measurement of artemisinin degradation revealed a twophase decay process in both the sandy and loamy soils (Fig. 1). The shape of the degradation curves was almost identical in the two soils, despite the different soil properties. With the exception of a few sampling points, the variation between triplicates expressed as the coefficient of variation (CV) was small. The range of CV was 0.9 to 17.5% and 2.9 to 28.3% for the sandy and loamy soils, respectively. This indicates sufficient soil homogenization as well as stable extraction and quantification of artemisinin in the soil. The first, fast phase takes place within the first day, whereas the slower process continues for more than 30 d. The kinetics were modeled as the sum of two first-order reactions (Eqn. 1). The correlation coefficient (r2) of the fitting was 0.97 in sandy soil and 0.95 in the loamy soil. The degradation parameters for the two soils are listed in Table 4. Data in Figure 1 show that after 24 h, a total of 21.5 and 51.9% of the artemisinin had degraded in the Fig. 2. Artemisinin concentrations measured in Danish Artemisia annua L. field at a depth of 0 to 2 cm (䢇) and 2 to 5 cm (䡬) versus sampling time. Data are shown on a logarithmic scale as the average ⫾ standard deviation (n ⫽ 3). Artemisinin in soil Fig. 3. The proportion of earthworms (Eisenia fetida) present in the contaminated soil compartment from the earthworm avoidance test as a function of the nominal start artemisinin concentration. Data are given as the average ⫾ standard error (n⫽ 10 containers/dose, with 20 containers for control and 10 worms in each container). Full lines represent the curve fits according to Equation 2. The curve–fit parameters are listed in Table 5. over the growing season in the topsoil, whereas the concentration decreased at a depth of 2 to 5 cm after September. Ecotoxicological tests in soil Dose–response curves for the soil ecotoxicology tests with invertebrates are depicted in Figures 3 and 4 and for lettuce growth in Figure 5. No significant differences were found between controls with and without acetone. Hence, the controls are pooled in the figure, and in the analyses (springtail test: p ⫽ 0.27, lettuce test: p ⫽ 0.16; earthworm avoidance test: p ⬎ 0.17). Artemisinin was repellent to earthworms and inhibited growth of lettuce. The CVs of the control treatments were 29.2, 19.3, and 30.3% for the earthworm, springtail, and lettuce tests, respectively. The variation in the earthworm and lettuce Environ. Toxicol. Chem. 28, 2009 707 Fig. 5. Seedling growth test with lettuce (Lactuca sativa L.) with weight of biomass per dish given as a function of the nominal start artemisinin concentration. Data are given as the average ⫾ standard deviation (n ⫽ 4 for treatments and 8 for control). Full lines represent the curve fits according to Equation 2.The curve–fit parameters are listed in Table 5. tests was within the acceptable range [29,35]. Although a relatively large variation was observed, the springtail test fulfilled the international validation criteria, because more than 200 juveniles were produced in the controls and the CV was less than 20% [26]. In the earthworm avoidance test, the soil structure by the end of the experiment clearly showed very little or no earthworm activity in the contaminated site at the highest concentrations, whereas the uncontaminated compartment had the usual crumb structure, reflecting earthworm activity. Responses from both the earthworm avoidance test (Fig. 3) and the lettuce growth test (Fig. 5) followed a three-parameter loglogistic dose–response relationship (Eqn. 2). The parameters for the two tests are listed in Table 5. The springtail test (Fig. 4) did not show significant effects at the tested concentration range (analysis of variance [ANOVA], p ⫽ 0.10). A weak tendency toward a decreasing number of juveniles with increasing artemisinin content, however, seems to be present. In the seedling emergence test with lettuce, no effects on germination frequency were observable within the tested concentration range (ANOVA, p ⫽ 0.18). The germination frequency was, on average, 77% ⫾ 0.34% (data not showed). Measurement of seedling weight after 21 d of growth (Fig. 5) reflected both postgermination mortality and decreased growth with increasing artemisinin concentration. In Figures 3–5, the biological responses are depicted as a function of added artemisinin content. The artemisinin was added 24 h before organisms were added to the soil, however, and the degradation experiments demonstrated that a substantial part of the added artemisinin was degraded within the first 24 h (Fig. 1 and Table 4). Therefore, the effective soil concentrations calculated by use of Equation 1 have been used in the estimation of the effect concentrations presented in Table 5. Ecotoxicology tests in freshwater Fig. 4. Dose–response experiment results from springtail (Folsomia candida) test in sandy soil, with responses given as a function of the nominal start artemisinin concentration. Data are given as the average ⫾ standard deviation (n ⫽ 4 for treatments and 8 for control). Full lines represent the curve fits according to Equation 2. The curve–fit parameters are listed in Table 5. The dose–response curves from the ecotoxicology tests in water spiked with artemisinin are depicted in Figure 6. Artemisinin was toxic to algae and duckweed at the tested concentrations, and the growth of L. minor and algae data followed the logistic dose–response relationship with three parameters according to Equation 2 (Table 5). The CV of the controls was 708 Environ. Toxicol. Chem. 28, 2009 K.K. Jessing et al. Table 5. Dose–reponse parameters (Eqn. 2) and the 10 and 50% effect concentrations (EC10 and EC50, respectively) from the earthworm (Eisenia fetida) avoidance and lettuce (Lactuca sativa L.) growth tests in sandy soil at 20⬚C and 15 to 17% gravimetric moisture and from the duckweed (Lemna minor) and algae (Pseudokirchneriella subcapitata) tests at 25 and 22⬚Ca Parameters Test Earth worm avoidance Lettuce growth Algae Duckweed b 1.58 1.44 2.28 0.30 ⫾ ⫾ ⫾ ⫾ e (mg/kgb or mg/L) d 0.38 0.51 0.065 0.01 0.61 0.10 5.70 1.12 ⫾ ⫾ ⫾ ⫾ 0.047 0.007 4.23 0.27 27.62 2.48 0.24 0.19 ⫾ ⫾ ⫾ ⫾ 6.37 0.56 0.014 0.031 EC10 (mg/kgb or mg/L) 5.24 0.54 0.16 0.026 ⫾ ⫾ ⫾ ⫾ 2.64 0.32 0.05 0.02 EC50 (mg/kgb or mg/L) 21.56 2.48 0.24 0.19 ⫾ ⫾ ⫾ ⫾ 4.73 0.56 0.01 0.03 a Data are given with ⫾ standard deviation. d ⫽ upper limit of the curve corresponding to the response of the untreated control; e ⫽ dose at which the value of d is reduced by 50% (i.e., EC50); b ⫽ proportional to the slope around EC50. b Soil concentrations are corrected for degradation using Equation 1. 22.4% in the algae test and 9.7% in the L. minor test. These variations are within the range reported in the literature [32,36]. In the L. minor test, the controls with acetone did not differ significantly from the controls without acetone ( p ⫽ 0.50). In the algae test, however, a significant difference was found in responses from controls with and without acetone ( p ⫽ 0.01), resulting in less growth in the controls with acetone. A dose–response experiment with acetone therefore was made, but in that experiment, no significant difference was observed between the control and the eight tested acetone concentrations Fig. 6. Dose response curves from algae (Pseudokirchneriella subcapitata) and duckweed (Lemna minor) tests in freshwater: (a) Relative growth rates of the algae (n ⫽ 6 for treatments, 19 for pure controls, and 10 for control with acetone [open symbol]); (b) relative growth rates of duckweed (n ⫽ 3 for treatments and 6 for both controls with and without acetone). All data are given as the average ⫾ standard deviation. Full lines represent the curve fits according to Equation 2. The curve–fit parameters are listed in Table 5. of less than 0.5 ml/L (ANOVA, p ⫽ 0.59) or between the highest acetone concentration and the control treatment (ANOVA, p ⫽ 0.62). Hence, we believe the low control acetone data to be erratic (because they could not be repeated), and the algae dose–response relationship therefore is described excluding these data. DISCUSSION Degradation kinetics of artemisinin Despite the short half-lives of artemisinin 4.22 and 0.90 d in sandy and loamy soil, respectively, artemisinin was detectable (⬎0.36 mg/kg) for 60 and 35 d in the sandy and loamy soils, respectively, when the initial artemisinin concentration was 20 mg/kg dry weight. This was because of the two-phased degradation kinetics, in which the rate constants of the first, rapid-degradation phase were 11- and 25-fold faster than the rate constant of the following, slower-degradation phase. No significant difference was found between soils in the fraction of artemisinin undergoing the fast degradation, but the rate of the fast reaction was approximately fourfold higher in the loamy soil compared to the sandy soil (Table 4). Because the soil properties were similar, except for the higher content of clay and a higher content of extractable metal oxides (Feox and Alox) in the loamy soil, degradation enhanced by mineral surfaces may play a role during the fast-degradation phase. The Fe(II) can act as a catalyst in cleavage of the peroxide bridge [12]. Because both soils are aerobic, however, Fe(II) is not likely to be present in solution, and other metal cations at soil mineral surfaces (e.g., Fe(III) and Mn(II) species) could be the active catalysts. The activated artemisinin readily reacts irreversibly with biological macromolecules [13]. Hence, part of the fast degradation likely may be attributed to reaction of artemisinin with amino acids, proteins, amino sugars, enzymes, and dissolved humic matter in the soils. Also, sorption of artemisinin may affect the extent of the fast reaction. The rate of the slow-degradation phase was similar in the two soils (Table 4). It is anticipated that artemisinin, which has a log KOC of 2.51 and which lacks groups binding strongly to soil minerals, sorbs predominantly to soil organic matter. By use of the partitioning coefficient Kd value calculated from the KOC and the soil contents of organic matter, it appears that approximately 8% of the artemisinin will be present in soil solution. Hence, even if degradation caused by reaction with organic macromolecules in soil is the main route of degradation, sorption may slow these processes, which might explain the slow reaction. There may be other degradation routes as well, including microbial degradation. The low vapor pres- Artemisinin in soil sure of the compound (4.92 ⫻ 10⫺9 atm·m3/mol) excludes volatilization, and because the soil experiments were carried out in the dark, photolysis also can be excluded. Hydrolysis is not likely to take place, because artemisinin has a polycyclic structure, known to be inert to hydrolysis, and in addition, the experiments were carried out at pH values close to neutral. Compared with other natural toxins, artemisinin is quite persistent in soil. The sesquiterpene lactone toxin parthenin produced by P. hysterophorus L. had a half-life of 59 h under conditions similar to those in the artemisinin experiments, and biotic degradation was strongly indicated [2]. Parthenin is a sesquiterpene lactone similar to artemisinin but lacking the peroxide bridge, and obviously, other mechanisms control degradation of this compound. The hydrolysis of other natural toxins, such as cyanogenic glucosides and glucosinolates, and the following degradation of their toxic metabolites are within the range of a few days [37,38]. Soil persistence, however, is comparable to that of commercial pesticides with a lipophilicity similar to that of artemisinin, such as atrazine, terbuthylazine, or metholachlor (log KOW ⫽ 2.5–3.2), which have field half-lives in the range of 20 to 80 d, with the longest half-lives occurring under cool and dry conditions [39]. Also, the measured water solubility of artemisinin of approximately 50 mg/L parallels that of the commercial herbicides [39]. The combination of soil persistence and medium to high water solubility will categorize artemisinin as having a leaching potential. We measured artemisinin in field soil at concentrations up to 11.7 mg/kg. Whether this artemisinin occurs in the soil matrix or is present within dead plant material is not known. Because artemisinin is present in external glands on the leaves, however, most of the artemisinin likely will have been released to the soil matrix. Ecotoxicity and risk evaluation Artemisinin repelled earthworms strongly at realistic field concentrations (10% effect concentration [EC10], 5.24 mg/kg; EC50, 21.56 mg/kg). Chronic toxicity was not investigated, but even observation of avoidance is ecologically relevant: Earthworms are very important for soil quality. Springtails, on the other hand, were not affected in the tested concentration range. Because the literature reports insecticidal properties of A. annua extracts [17,18], we had expected springtails to be affected in the same concentration range affecting earthworm behavior. The insect cuticle of the springtails, however, might have protected them better against chemical exposure compared with the soft skin of earthworms. Also, the literature on insecticidal activity is derived from plant extracts of A. annua, which include compounds other than artemisinin [17]. In addition, the extracts were dosed either directly at the insects, giving EC50s of 2.5 to 4.1 ml A. annua extract/L, or on their food, giving EC50s of 150 kg/ha [17,18]. These are dose rates far above those of the present study. Care should be taken when comparing avoidance behavior and effects on reproduction and growth, but in the case of the effect of A. annua extracts on beetles, the effect concentrations for avoidance, reproduction, and growth were within the same order of magnitude [17]. Growth of lettuce was inhibited at relatively low concentrations (EC10, 0.54 mg/kg; EC50, 2.48 mg/kg). Germination of lettuce seed, however, was not affected by artemisinin, contrary to the data reported in Duke et al. [19]. Field experiments support this observation: A trial with weeds in maize and Environ. Toxicol. Chem. 28, 2009 709 potato showed that emergence of weeds was not significantly inhibited by incorporated A. annua dry plant material, whereas weed biomass was. For the maize crop, both the emergence and height growth were reduced in the artemisinin-rich, A. annua–treated plots [20]. The highest treatment in the field of 400 g dried plant material/m, corresponding to an EC50 to 80% effect concentration for the weed and maize growth, contained 120 mg artemisinin/kg soil (depth, 0–5 cm) at 10 d after incorporation of the plant material, decreasing to 30 mg/kg at 60 d after incorporation [20]. Hence, the field effect concentrations were remarkably higher than those found in the laboratory for lettuce. This large difference might stem from differences in the availability of artemisinin added directly as a chemical versus that added as incorporated in plant material, combined with differences between species and differences in growth conditions. Artemisinin contents measured in the topsoil in a Danish A. annua field ranged from 0.16 to 11.7 mg/ kg during a growing season. With the EC10 for earthworm avoidance and the EC10s and EC50s for lettuce being less than the measured field concentrations, cultivation of A. annua might affect soil quality in terms of earthworm abundance and activity as well as the vigor of crops grown directly after A. annua in monocultures. It is possible to perform an initial risk evaluation of artemisinin based on the knowledge obtained in the present study. We determined a predicted exposure concentration (PEC) for artemisinin in soil to be equal to the highest measured concentration of 11.7 mg/kg. This concentration will affect sensitive organisms in the soil environment. The scenario is realistic, as demonstrated in the field experiment carried out by Delabays et al. [20], in which incorporation of dried A. annua leaf material showed artemisinin concentrations greater than 30 mg/kg up to 60 d after incorporation. It is unknown to what extent intact artemisinin in plant material contributes to this measurement and how biological organisms, apart from the plants monitored, will respond to artemisinin bound to plant material. Considering that artemisinin is categorized as having a moderate leaching potential, it might leach into drains and pollute surface waters. As mentioned, the properties of artemisinin are comparable to those of the commercial pesticide atrazine, which is regularly detected in surface waters [40], when it comes to lipophilicity and persistence (log KOW ⫽ 2.61 and half-life of 16–77 d [39]). The aquatic toxicity tests for artemisinin showing EC50s of 0.19 mg/L (0.67 ␮M) in the duckweed test and 0.24 mg/L (0.85 ␮M) in the algae test also proved to be comparable to the aquatic toxicity of several commercial pesticides, including the triazines. The EC50 for atrazine to green algae is in the range of 0.043 to 0.138 mg/L in 96-h tests, whereas the EC50 for atrazine to duckweed is in the range of 0.080 to 0.102 mg/L [39,41]. If artemisinin was marketed as a commercial pesticide, predicted no-effect concentrations (PNECs) and the predicted exposure concentration (PEC) would be evaluated. Setting the PEC to be the highest measured field concentration in the present study of 11.7 mg/kg and the PNEC to the EC50 for the most sensitive species tested in soil, and including a safety factor of 1,000, would result in a hazard quotient (PEC/PNEC) of 500. The fact that a hazard quotient of much greater than one is obtained reflects a high environmental risk in plantations of A. annua. Hence, based on our present knowledge, crop rotation is necessary to maintain good soil quality. Further investigations are needed to identify exposure routes from 710 Environ. Toxicol. Chem. 28, 2009 plant to soil, however, and to clarify the fate and toxicity of artemisinin in the soil and water environments. K.K. Jessing et al. 21. REFERENCES 1. Taiz L, Zeiger E. 2002. Plant Physiology. Sinauer Associates, Sunderland, UK. 2. Belz RG, Laan Mvd, Reinhardt CF, Hurle K. 2007. Soil degradation of parthenin—Does it contradict a role in allelopathy of the invasive weed Parthenium hysterophorus L.? Proceedings, 14th European Weed Research Society Symposium, Hamar, Norway, June 18–21, p 166. 3. Viaene NM, Abawi GS. 1998. Management of Meloidogyne hapla on lettuce in organic soil with sudangrass as a cover crop. Plant Dis 82:945–952. 4. Gimsing AL, Kirkegaard JA. 2006. Glucosinolate and isothiocyanate concentration in soil following incorporation of Brassica biofumigants. Soil Biol Biochem 38:2255–2264. 5. Klayman DL. 1985. Qinghaosu (artemisinin)—An antimalarial drug from China. Science 228:1049–1055. 6. Liu JM, Ni MY, Fan JF, Tu YY, Wu ZH, Qu YL, Chou MS. 1979. Structure and reactions of arteannuin. Acta Chimica Sinica 37: 129–143. 7. Meshnick SR, Taylor TE, Kamchonwongpaisan S. 1996. Artemisinin and the antimalarial endoperoxides: From herbal remedy to targeted chemotherapy. Microbiol Rev 60:301–315. 8. Dhingra V, Rao KV, Narasu ML. 2000. Current status of artemisinin and its derivatives as antimalarial drugs. Life Sci 66:279– 300. 9. Ferreira JFS, Laughlin JC, Delabays N, Magalhães PM. 2005. Cultivation and genetics of Artemisia annua L. for increased production of the antimalarial artemisinin. Plant Genetic Resources 3:206–229. 10. Porskjaer Christensen L, Grevsen K, Jensen M. 2007. Pharmland—Illusion eller mulighed? Dansk Kemi 88:13–16. 11. Delabays N, Simonnet X, Gaudin M. 2001. The genetics of artemisinin content in Artemisia annua L. and the breeding of high yielding cultivars. Curr Med Chem 8:1795–1801. 12. Meshnick SR, Yang YZ, Lima V, Kuypers F, Kamchonwongpaisan S, Yuthavong Y. 1993. Iron-dependent free-radical generation from the antimalarial agent artemisinin (Qinghaosu). Antimicrob Agents Chemother 37:1108–1114. 13. Eckstein-Ludwig U, Webb RJ, van Goethem IDA, East JM, Lee AG, Kimura M, O’Neill PM, Bray PG, Ward SA, Krishna S. 2003. Artemisinins target the SERCA of Plasmodium falciparum. Nature 424:957–961. 14. Wu YK, Liu HH. 2003. Cleavage of qinghaosu (artemisinin) induced by noniron transition-metal ions in the presence of excess cysteine. Helv Chim Acta 86:3074–3080. 15. Duke MV, Paul RN, El Sohly HN, Sturtz G, Duke SO. 1994. Localization of artemisinin and artemisitene in foliar tissues of glanded and glandless biotypes of Artemisia annua L. Int J Plant Sci 155:365–372. 16. Janick J, Ferreira JFS. 1996. Distribution of artemisinin in Artemisia annua. In Janick J, ed, Progress in New Crops. American Society for Horticultural Science, Alexandria, VA, USA, pp 579– 584. 17. Tripathi AK, Prajapati V, Aggarwal KK, Khanuja SPS, Kumar S. 2000. Repellency and toxicity of oil from Artemisia annua to certain stored-product beetles. J Econ Entomol 93:43–47. 18. Maggi ME, Mangeaud A, Carpinella MC, Ferrayoli CG, Valladares GR, Palacios SM. 2005. Laboratory evaluation of Artemisia annua L. extract and artemisinin activity against Epilachna paenulata and Spodoptera eridania. J Chem Ecol 31:1527–1536. 19. Duke SO, Vaughn KC, Croom EM, ElSohly HN. 1987. Artemisinin, a constituent of annual wormwood (Artemisia annua), is a selective phytotoxin. Weed Sci 35:499–505. 20. Delabays N, Slacanin I, Bohren C. 2008. Herbicidal potential of 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. artemisinin and allelopathic properties of Artemisia annua L.: From the laboratory to the field. J Plant Dis Prot 21:317–322. Zhao SS, Zeng MY. 1985. Spektrometische Hochdruck-Flüssigkeits-Chromatographische (HPLC) Untersuchungen zur Analytik von Qinghaosu. Planta Med 51:233–237. Organisation for Economic Co-operation and Development. 1981. Testing of Chemicals, Water Solubility, Guideline 105. Paris, France. Soil Survey Staff. 1999. Soil Survey Staff, Keys to Soil Taxonomy. Pochahontas, Blacksburg, VA, USA. Dissing Nielsen J, Møberg JP. 1984. Klassificering af 5 jordprofiler fra forsøgsstationer i Danmark. Tidsskr Planteavl Beretning nr. 1706:155–167. Brinch UC, Ekelund F, Jacobsen CS. 2002. Method for spiking soil samples with organic compounds. Appl Environ Microbiol 68:1808–1816. Wiles JA, Krogh HP. 1998. Testing with the collembolans I. viridis, F. candida and F. fimetaria. In Løkke H, van Gestel, CAM, eds, Handbook of Soil Invertebrate Toxicity Tests. John Wiley, Chichester, UK, pp 131–156. Krogh PH, Johansen K, Holmstrup M. 1998. Automatic counting of collembolans for laboratory experiments. Appl Soil Ecol 7: 201–205. U.S. Environmental Protection Agency. Ecological effects test guidelines, OPPTS: 850.6200: Earthworm subchronic toxicity test. EPA 712-C-96-167. Final Report. Washington, DC. Organization for Economic Co-operation and Development. 2003. Testing of Chemicals Proposal for Updating Guideline 208, Terrestrial Plant Test: 208. Seeding Emergence and Seedling Growth Test. Paris, France. Mayer P, Cuhel R, Nyholm N. 1997. A simple in vitro fluorescence method for biomass measurements in algal growth inhibition tests. Water Res 31:2525–2531. International Organization for Standardization. 1989. Water quality—Freshwater algal growth inhibition test with Scenedesmus subspicatus or Selenastrum capricornutum. 8692. Geneva, Switzerland. Cedergreen N, Streibig JC. 2005. The toxicity of herbicides to nontarget aquatic plants and algae: Assessment of predictive factors and hazard. Pest Manag Sci 61:1152–1160. Maeng J, Khudairi AK. 1973. Studies on flowering mechanism in Lemna. 1. Amino acid changes during flower induction. Physiol Plant 28:264–270. R Development Core Team. 2004. R: A Language and Environment for Statistical Computering. R Foundation for Statistical Computering, Vienna, Austria. Reinecke AJ, Maboeta MS, Vermeulen LA, Reinecke SA. 2002. Assessment of lead nitrate and mancozeb toxicity in earthworms using the avoidance response. Bull Environ Contam Toxicol 68: 779–786. Cedergreen N, Kudsk P, Mathiassen SK, Sorensen H, Streibig JC. 2007. Reproducibility of binary-mixture toxicity studies. Environ Toxicol Chem 26:149–156. Gimsing AL, Sorensen JC, Tovgaard L, Jorgensen AMF, Hansen HCB. 2006. Degradation kinetics of glucosinolates in soil. Environ Toxicol Chem 25:2038–2044. Johansen H, Rasmussen LH, Olsen CE, Hansen HCB. 2007. Rate of hydrolysis and degradation of the cyanogenic glycoside, dhurrin, in soil. Chemosphere 67:259–266. Tomlin CDS. 2002. The e-Pesticide Manual. [2.2]. Surrey, British Crop Protection Council, Farnham, UK. Giddings JM, Anderson TA, Hall LW, Hosmer AJ, Kendall RJ, Richards RP, Solomon KR, Williams WM. 2005. Atrazine in North American Surface Waters: A Probalilistic Aquatic Ecological Risk Assessment. SETAC, Pensacola, FL, USA. Fairchild JF, Ruessler DS, Carlson AR. 1998. Comparative sensitivity of five species of macrophytes and six species of algae to atrazine, metribuzin, alachlor, and metolachlor. Environ Toxicol Chem 17:1830–1834.