Environ Sci Pollut Res DOI 10.1007/s11356-015-5093-3

RESEARCH ARTICLE

Design of an optimized biomixture for the degradation of carbofuran based on pesticide removal and toxicity reduction of the matrix Juan Salvador Chin-Pampillo 1 & Karla Ruiz-Hidalgo 1 & Mario Masís-Mora 1 & Elizabeth Carazo-Rojas 1 & Carlos E. Rodríguez-Rodríguez 1

Received: 8 April 2015 / Accepted: 16 July 2015 # Springer-Verlag Berlin Heidelberg 2015

Abstract Pesticide biopurification systems contain a biologically active matrix (biomixture) responsible for the accelerated elimination of pesticides in wastewaters derived from pest control in crop fields. Biomixtures have been typically prepared using the volumetric composition 50:25:25 (lignocellulosic substrate/humic component/soil); nonetheless, formal composition optimization has not been performed so far. Carbofuran is an insecticide/nematicide of high toxicity widely employed in developing countries. Therefore, the composition of a highly efficient biomixture (composed of coconut fiber, compost, and soil, FCS) for the removal of carbofuran was optimized by means of a central composite design and response surface methodology. The volumetric content of soil and the ratio coconut fiber/compost were used as the design variables. The performance of the biomixture was assayed by considering the elimination of carbofuran, the mineralization of 14C-carbofuran, and the residual toxicity of the matrix, as response variables. Based on the models, the optimal volumetric composition of the FCS biomixture consists of 45:13:42 (coconut fiber/compost/soil), which resulted in minimal residual toxicity and ∼99 % carbofuran elimination after 3 days. This optimized biomixture considerably differs from the standard 50:25:25 composition, which remarks the importance of assessing the performance of newly developed biomixtures during the design of biopurification systems.

Responsible editor: Gerald Thouand * Carlos E. Rodríguez-Rodríguez [email protected] 1

Centro de Investigación en Contaminación Ambiental (CICA), Universidad de Costa Rica, 2060 San José, Costa Rica

Keywords Biopurification system . Pesticides . Degradation . Coconut fiber . Optimization

Introduction Improper use of pesticides in agricultural activities results in a negative impact on surface water bodies (Chin-Pampillo et al. 2012). The point source release due to intentional or accidental spills, or incorrect disposal of wastewater from pesticide applications is one of the main causes of the arrival of pesticides to rivers and other water environmental compartments (Carter 2000; Wenneker et al. 2010). The use of biopurification systems (BPSs) arises as an ecofriendly approach for the treatment of pesticide-containing wastewaters derived from field pesticide application, mainly present in application equipment or in the preparation containers. The use of BPS, initially named as biobeds, is a practice implemented in the 1990s in Sweden and later disseminated to some countries of Europe and America. BPSs are low-cost devices used for the accelerated degradation of pesticides; they contain an active core called biomixture, composed by a mixture of solid substrates. The biomixture consists of a lignin-rich substrate, a humic component, and soil previously exposed to pesticides (Castillo et al. 2008; Torstensson and Castillo 1997). The driving force in the BPS is the biomixture activity of microorganisms mainly contributed by the soil, and the action of lignin-modifying enzymes produced by white rot fungi; these enzymes are capable of degrading lignocellulosic substrates and are widely known to be involved in the transformation of several organic pollutants (Asgher et al. 2008). The presence of lignin-rich substrates promotes the growth of these fungi, whereas the humic component helps to maintain favorable conditions of moisture for the development of microbial communities, plus retention

Environ Sci Pollut Res

of the pesticides while the biological activity takes place (Castillo et al. 2008). The original biomixture developed in Sweden was composed of straw as the lignocellulosic substrate, peat as the humic component, and local soil, all agricultural by-products or readily available materials in that country. Various adaptations of the biomixture have been elaborated with alternative materials in other latitudes according to local availability (Castillo et al. 2008; Chin-Pampillo et al. 2015; Fogg et al. 2004; Pussemier et al. 2004; Spliid et al. 2006). Peat has been usually replaced by compost of different origins (Coppola et al. 2007; Kravvariti et al. 2010), and the use of diverse lignocellulosic substrates such as bagasse, coconut chips, citrus peel, vine branches, coconut fiber, and rice husks has been implemented (Castillo and Torstensson 2007; Chin-Pampillo et al. 2015; Coppola et al. 2007; de Roffignac et al. 2008; de Wilde et al. 2009a, b; Karanasios et al. 2010a; Spanoghe et al. 2004; Urrutia et al. 2013; Vischetti et al. 2004). Generally, the performance of such variations in the composition of biomixtures is evaluated only by determining the removal or half-life of one or more pesticides (Castillo et al. 2008), without considering the toxicological changes in the matrix during the degradation process; the latter are essential to assay the environmental impact of the use of BPSs. Similarly, the incorporation of radiometric techniques to assess mineralization of the pesticides to determine the extent of complete removal of potentially toxic transformation products has been limited in these matrices. The volumetric composition of the originally proposed biomixture was 50 % lignocellulosic substrate, 25 % humic component, and 25 % soil (Torstensson and Castillo 1997). This composition has been traditionally used to date, with modifications related to the use of alternative materials. Similarly, Karanasios et al. (2012) optimized the use of BPS in terms of the moisture content during wastewater applications. Given that biomixture composition strongly affects the performance of biobeds (Karanasios et al. 2010b), optimization remains a necessary step in the design of improved BPS. Previous work has revealed that the biomixture made of coconut fiber (as the lignocellulosic substrate), garden compost (as the humic component), and top primed soil has favorable properties for the degradation of pesticides such as carbofuran (CFN, 2,2-dimethyl-2, 3-dihydro-1-benzofuran7-yl methylcarbamate) (Chin-Pampillo et al. 2015). This is a broad-spectrum insecticide from the family of carbamates which displays neurotoxic activity, and is primarily used in the control of nematodes. Due to its high toxicity and still wide use in developing countries (despite being banned in the USA and the EU), CFN was employed as the model pesticide in this work. This work aims to optimize the composition of a biomixture for the degradation of CFN using a central composite design (CCD), employing a multifactorial approach that

comprises not only the removal of the pesticide but also the mineralization rate and the ability of the biomixture to reduce the residual toxicity in the matrix.

Materials and methods Chemicals, radiolabeled pesticide standards, and formulates Commercial CFN (Furadan® 48SC, 48 g active ingredient (a.i.)/100 mL formulate) was purchased from a local market. Analytical standards CFN (2,2-dimethyl-2,3-dihydro-1benzofuran-7-yl methylcarbamate, >99 % purity), 3hydroxycarbofuran (99.5 %), and 3-ketocarbofuran (99.5 %) were obtained from Chemservice (West Chester, Pennsylvania, USA). Radiolabeled CFN [Ring-U-14C]-Carbofuran (14C-CFN; 2.89×109 Bq g−1; radiochemical purity 100 %; chemical purity 99.59 %) was obtained from Izotop (Institute of Isotopes Co., Budapest, Hungary). Carbendazim-d3 (surrogate standard, 99.0 %) and carbofuran-d4 (internal standard, 99.5 %) were purchased from Dr. Ehrenstorfer (Augsburg, Germany). Ultima Gold cocktail for liquid scintillation counting was purchased from Perkin Elmer (Waltham, MA, USA). Solvents and extraction chemicals are listed in RuizHidalgo et al. (2014).

Biomixture components and preparation Clay loam soil (sand 40 %, silt 27 %, clay 33 %; pH 5.9; C, 2.71 %, N, 0.29 %) was collected from the upper soil layer (0–20 cm) of a beet/onion field with history of CFN application, in Tierra Blanca, Cartago, Costa Rica (9° 55′ 34.17″ N; 83° 53′ 11.76″ W). Soil was air-dried and sieved through a 2-mm sieve. Garden compost (pH 7.3; C, 10.37 %, N, 0.76 %), employed as the humic component, was collected from a composting station located at Universidad de Costa Rica and sieved as described for the soil after air drying. Coconut fiber (CF; lignin content, 56.1 %) acquired from a local market was used as the lignocellulosic substrate. Biomixtures were prepared by mixing the lignocellulosic substrate, the compost, and the pesticide-primed soil at the volumetric ratios shown in Table 1 in order to obtain a total of nine biomixtures of different composition, according to the design variables described in the Experimental design, response surface methodology, and statistical analysis section. The biomixtures were moistened to approximately 75 % of the maximum water-holding capacity and aged at 25 °C during 1 month with weekly homogenization prior to use.

10 25 46.2

25 3.8

9 10 11

12 13

b

a

0 2

3.4 4 2

2 2 3.4 1.4 1.4

25 3.8

10 25 46.2

25 25 40 40 10

25 25 25

Soil (%)

0.0 64.1

69.5 60.0 35.9

50.0 50.0 46.4 35.0 52.5

50.0 50.0 50.0

Coconut fiber (%)

75.0 32.1

20.5 15.0 17.9

25.0 25.0 13.6 25.0 37.5

25.0 25.0 25.0

Compost (%)

Corresponding FCS biomixture composition

76.8±19.3 93.5±8.6

87.2±9.1 98.4±8.7 92.4±9.2

50.4±7.0 72.9±7.7 95.5±8.6 65.1±10.4 82.1±13.2

49.2±7.0 70.8±7.6 62.5±7.4

R1 CFN removal in 3 days (%)

Responses

Confidence limits (for EC50 values)

Toxicity values at 48 h (expressed as TU) were employed in the CCD analysis

NT non-toxic

25 25 40 40 10

4 5 6 7 8

2 2 2

Volumetric ratio CF/compost (%/%)

Soil (%v/v)

25 25 25

x2

x1

Actual factors

27.3 NT

47.9 37.4 38.3

NT 86.7 NT NT NT

NT NT NT

EC50 (%)

24 h

25.0–50.0 –

3.7 0

2.1 2.7 2.6

0 1.2 0 0 0

– 85.0–100 – – – 25.0–50.0 25.0–50.0 25.0–50.0

0 0 0

TU – – –

CLb

R2 residual toxicity after 8 days

CCD design matrix and response values during the removal of CFN in FCS biomixtures

1 2 3

Run

Table 1

50.0 NT

47.9 37.4 32.2

NT 86.4 NT NT NT

NT NT NT

EC50 (%)

48 ha

– – –

25.0–75.0 –

25.0–50.0 25.0–50.0 25.0–50.0

– 85.0–100 – – –

CLb

2.0 0

2.1 2.7 3.1

0 1.2 0 0 0

0 0 0

TU

0.0026±0.0001 0.0014±0.0001

0.0014±0.0001 0.00166±0.00006 0.00138±0.00006

0.0012±0.0001 0.0043±0.0004 0.0034±0.0002 0.0028±0.0002 0.0020±0.0001

0.0009±0.0001 0.0032±0.0004 0.0031±0.0004

k (day−1)

R3 mineralization

268 485

495 418 502

598 161 204 251 350

815 214 221

Half-life (days)

Environ Sci Pollut Res

Environ Sci Pollut Res

Experimental design, response surface methodology, and statistical analysis

residual toxicity after 8 days, and to maximize both CFN removal after 3 days and desirability.

A CCD methodology with two factors (k=2) was applied to optimize the composition of a biomixture composed of CF, compost, and soil, to remove CFN. The design variables or factors were the volumetric content of soil (%) (x1) and the volumetric ratio CF/compost (x2). The effect of these factors was observed on the following response variables: the percentage of CFN removal after 3 days (R1), the residual acute toxicity of the matrix after 8 days (R2), as evaluated in an immobilization Daphnia magna test, and the rate of 14CCFN mineralization over a period of 35 days (R3). CCD employs 2k factorial points representing all combinations of the codified values (±1), 2k axial points at a distance ±α from the origin, and at least three central points in the origin (encoded as 0,0). The factor levels were normalized and coded in the range (−α, +α). The α value corresponds to 1.414 (α=F1/4, where F=2k). Nine combinations of the design variables were evaluated; to determine the experimental uncertainty, the central point was performed by quintuplicate. This resulted in an experimental design that included 13 runs. The experimental design matrix is shown in Table 1 and includes actual values for the different combinations of factors x1 and x2. The CCD was centered in the point where x1 =25 % and x2 =2, which corresponds to the most commonly described volumetric composition of biomixtures employed in BPS, that is, lignocellulosic substrate/compost/soil at a ratio of 50:25:25 (Castillo et al. 2008). The volumetric content of soil, x1, ranged from 3.8 % to 46.2 %, considering that this is the main source of degrading microbiota. The ratio CF/compost, x2, was chosen in order to estimate the role of lignin degrading microbiota on CFN removal and ranged from 0 (which represents absence of lignocellulosic substrate) to 4. Each response variable can be fitted to a second-order polynomial model (k=2), according to the following equation:

Experimental procedures

y ¼ b0 þ b1 x1 þ b2 x2 þ b12 x1 x2 þ b11 x21 þ b22 x22

ð1Þ

The estimation of the model parameters (bi) and the statistical analysis were performed using the software Design Expert 9.0 (Stat-Ease Inc., Minneapolis, USA). The quality of the fit polynomial model was determined by the Fisher’s Ftest, model terms were evaluated by the P value with 95 % confidence level, and results were completely analyzed by analysis of variance (ANOVA) employing the same software. In addition, a variable called Bdesirability^ (which ranges from 0 to 1) that encompasses the simultaneous effects of the response variables was also determined. Optimization of the biomixture composition was conducted with response surface methodology by the analysis of contour plots and numerical solutions by the software, in order to minimize

Degradation experiments The degradation of CFN and the formation of its transformation products, 3-hydroxycarbofuran and 3-ketocarbofuran, were assayed by weighing 5 g of each biomixture into 50-mL polypropylene tubes. Each sample was spiked with commercial CFN (30 mg kg−1), manually homogenized, and incubated in the dark at (25±1)°C during 3 or 8 days. The remaining pesticide concentration was determined by sacrificing duplicate systems (quintuplicates for the central point of the experimental design) at times 0 h and 3 days, as described in the Analytical procedures section. Additional unitary systems (10 g each) obtained after 8 days of treatment (eight per biomixture) were combined as a composite sample to perform the acute toxicity test with D. magna. Mineralization studies The mineralization of 14C-CFN was determined through 14 CO2 production in biometric systems containing 14CO2 traps with 10 mL KOH (0.1 M) (Ruiz-Hidalgo et al. 2014). For each biomixture, 50 g was weighed into the biometric flask and spiked with commercial CFN (30 mg kg−1) and 14 C-CFN (50 Bq g−1). The systems were incubated in the dark at (25±1)°C during 35 days. The KOH in the flasks was withdrawn at selected times (two times weekly) and replaced with the same amount of fresh KOH. Activity of 14C from the 14 CO2 mineralized was analyzed in the KOH samples as described in the Analytical procedures section. Analytical procedures Extraction and analysis of CFN and transformation products Unitary samples (5 g) obtained from degradation experiments were extracted using 10 mL of water, 15 mL of acidified acetonitrile (formic acid 1 %v/v), 1.5 g of anhydrous sodium acetate, 6 g of anhydrous magnesium sulfate, and 1 g of sodium chloride. Samples were shaken manually during 2 min and then centrifuged at 4500 rpm during 7 min at 10 °C. Carbendazim-d4 was added as a surrogate standard. An aliquot (3 mL) of supernatant was cleaned adding 900 mg of anhydrous magnesium sulfate, 150 mg of Bondesil-PSA, and 75 mg of silica-C18 and then shaken during 1 min and centrifuged at 4500 rpm during 7 min at 10 °C. A 1.5-mL aliquot of supernatant was evaporated until dryness and then carbofuran-d 3 was added as internal standard and reconstituted to a final volume of 1.5 mL using acidified water

Environ Sci Pollut Res

(formic acid 1 %v/v). Sample extract was filtered through a 0.45-μm syringe Teflon filter (Millipore, Billerica, MA), before liquid chromatography–mass spectrometry (LC-MS) analysis. Recovery of extraction methodology in the matrix was 91 %. CFN and its transformation products were analyzed by LCMS/MS using ultra high performance liquid chromatography (UPLC 1290 Infinity LC, Agilent Technologies, CA, USA) coupled to a triple quadrupole mass spectrometer (6460, Agilent Technologies, CA, USA). Chromatographic separation was done at 40 °C by injecting 6-μL samples (2-μL loop) in a Poroshell 120 EC-C18 column (100×2.1 mm i.d., particle size 2.7 μm; Agilent Technologies, CA, USA), and using acidified water (formic acid 0.1 %v/v, A) and acidified methanol (formic acid 0.1 %v/v, B) as mobile phases. The mobile phase flow was 0.3 mL min−1 at the following conditions: 30 % B for 3 min, followed by a 15-min linear gradient to 100 % B, 4 min at 100 % B, and 0.1-min gradient back to 30 % B, followed by 5 min at initial conditions. Selected transitions for the analytes are shown in Table 2. The mass spectrometer employed a jet stream (electrospray) ionization source operating at the following conditions: gas temperature 300 °C, gas flow 7 L min−1, nebulizer 45 psi, sheath gas temperature 250 °C, sheath gas flow 11 L min−1, capillary voltage 3500 V (for positive and negative), nozzle voltage 500 V (for positive and negative), and heater MS1 and MS2 100 °C. Recovery of the method was 91 % for CFN, 98 % for 3-hydroxycarbofuran, and 95 % for 3-ketocarbofuran. Limit of detection (LOD) and limit of quantification (LOQ) were 13 and 26 μg kg−1 for CFN and 3-ketocarbofuran, and 16 and 32 μg kg−1 for 3-hydroxycarbofuran.

Beckman LS6000SC counter. The total cumulative 14CO2 activity evolved (mineralized) and the initially added 14C-CFN were used to calculate the percentage of 14C-pesticide mineralized. Pesticide mineralization was modeled according to a first-order model to determine mineralization rate constants. Toxicological analysis An acute test based on D. magna immobilization was employed to estimate the residual toxicity in the biomixtures. The test was conducted following the US EPA guidelines (EPA 2002) and some modifications as described by RuizHidalgo et al. (2014), using elutriates from the biomixtures at time 0 h (after CFN spiking) and after 8 days of treatment, as analytical matrix. Elutriates were obtained according to the protocol EPA-823-B-01-002 (EPA 2001); briefly, 40 mL distilled water were added to 10-g samples, and the mixture was mechanically shaken for 1 h and centrifuged (3500 rpm); the supernatant was used as elutriate. Tests were performed in triplicates in glass vials containing five daphnid neonates (less than 24 h) each, and 10-mL elutriate dilutions. EC50, the concentration producing 50 % of immobilization in the daphnids (bioassay endpoint, determined at 24 and 48 h), was determined using the binomial probability test on the TOXCALC—Toxicity Data Analysis Software from Tidepool Scientific Software. Toxicity results were expressed as toxicity units (TU), calculated according to the expression TU=(EC50)−1 ·100.

Results and discussion

Determination of 14CO2 from mineralization assays

Removal of CFN from the biomixtures: optimization

Scintillant liquid (10 mL) was added to 2-mL aliquots from the removed KOH solution samples, and the 14C activity from the trapped 14CO2 was measured by liquid scintillation using a

A previous screening of several lignocellulosic substrates, humic components (compost and peat), and one CFN-primed soil employed in the elaboration of biomixtures (at a

Table 2 Selected transitions and other parameters in the detection of CFN and its transformation products using the dynamic MRM method

Compound

Retention time (min)

3-Hydroxycarbofuran

3.60

3-Ketocarbofuran

6.10

Carbofuran

7.92

Carbendazim-d4

1.45

Carbofuran-d3

7.92

a

Transition

238→163 238→107 236→161 236→179 222→165 222→123 196→164 196→136 225→165 225→123

Q, quantification transition; q, qualifier transition

Fragmentor (V) 72 82 82 102 86

Collision energy (V)

Type of transitiona

9

Q

33 17 9 9 21 17 34 9 21

q Q q Q q Q q Q q

Environ Sci Pollut Res

volumetric proportion of 50:25:25) revealed the fastest CFN removal by the biomixture composed of CF-compost-soil (FCS) (Chin-Pampillo et al. 2015). This biomixture showed an estimated CFN half-life of around 2.5 days, and 79.6 % elimination after 4 days, a better performance than the other matrices. For this reason, and considering the lack of known reports on formal biomixture optimization, the FCS biomixture was selected for composition optimization. Taking into account that the half-life of CFN was 2.5 days in the FCS biomixture (50.25:25), its removal after 3 days was selected as the response variable to compare the performance of biomixtures of different composition. The removal capacity of such biomixtures after 3 days of treatment is shown in Table 1 and ranged from 49.2 % (% soil=25, CF/compost= 2) to 98.4 % (% soil=25, CF/compost=4). Response surface methodology was employed to analyze the correlation between the variables (% soil and CF/compost) and the CFN removal after 3 days (R1). The data was used to fit a second-order polynomial model, according to Eq. (2) which represents the regression in terms of the actual factors: R1 ¼ 120:21−3:39x1 −20:49x2 þ 0:066x21 þ 6:27x22

ð2Þ

The model was statistically analyzed to determine the accuracy of the fit (Table 3). ANOVA demonstrated that the model was significant, as revealed by the low probability value (Pmodel >F=0.0084), and the lack of fit (LOF, not significant, P=0.7804), which implies a significant relationship between design and response variables within the model. Moreover, the determination coefficient (R2 =0.7881) indicates an acceptable goodness in the fit, and the adequate precision (5.985, desired over 4) suggests an adequate signal over noise in the model. Table 4 shows the coefficient values of the fitted model as coded factors. These coded factor coefficients permit to estimate the relative impact of the factor on the response variable. The positive value of the coefficient related to the ratio CF/ compost (x2) indicates that this variable has a positive effect

Table 3 Analysis of variance (ANOVA) and other statistical parameters for the quadratic model employed to fit the removal of CFN after 3 days in the FCS biomixtures Source of variations

Sum of squares

df

Mean square

F-value

Probability P (>F)

Model Residual Lack of fit Pure error Total

2622.94 705.21 213.56 491.640 3328.14

4 8 4 4 12

655.73 88.15 53.39 122.910

7.44

0.0084

0.43

0.780

R2 =0.7881; adjusted R2 =0.6822; adequate precision=5.985

Table 4 Regression results from the CCD experiments for the modeling of response variables R1 (CFN removal after 3 days) and R2 (residual toxicity after 8 days) according to the selected quadratic model Model

Model term

Coefficient Standard F-value P value (coded factors) error

CFN removal Intercept 63.7 after 3 days x1 −1.28 9.59 x2 14.87 x12 6.27 x22

4.15 3.32 2.82 3.65 1.87

0.15 11.54 16.61 11.19

0.7099 0.0094 0.0036 0.0102

Residual Intercept 0.42 toxicity 0.13 x1 after 8 days 0.48 x2 −0.78 x1x2 0.46 x22

0.33 0.36 0.30 0.48 0.19

0.13 2.63 2.69 5.76

0.7274 0.1437 0.1399 0.0432

on the degradation extent in 3 days; this means that higher CF/compost ratios improve CFN removal. On the other hand, the effect of % soil (x1) is negative; however, this latter effect is much lower than that exerted by the ratio CF/compost (coefficients 9.59 vs −1.28). This finding suggests that, despite essential, the amount of soil used in the biomixtures plays a negligible role on the removal of CFN; this is in agreement with Sniegowski et al. (2012), who reported that concentrations as low as 0.5 % pesticide-primed soil may be efficient enough to provide the degrading population for proper biobed performance. The coefficients for the quadratic factors x1 and x2 showed a positive and significant effect on the removal; other interactions were not significant and, therefore, they were not included in the model. The response surface and the contour plots are the graphical representations of the regression equation (Fig. 1). As it can be observed on the figure, two regions in the contour maximize the removal of CFN, those corresponding to high CF/compost ratios (x2, around 3.4) and either low (10 %) or high (40 %) amounts of soil (x1). Optimization under the criterion of maximizing R1 in the Expert Design software was performed. According to the model, the maximum removal of CFN in 3 days takes place at conditions x1 =41.9 % and x2 =3.52, that is, in a biomixture containing CF/compost/soil (F/C/S) at a ratio of 45:13:42. Interestingly, lower R1 values were obtained near the central point, which corresponds to the typical ratio of 50:25:25, widely employed to prepare biomixtures since the appearance of biobeds in the 1990s. Two typical transformation products from CFN metabolism, 3-hydroxycarbofuran and 3-ketocarbofuran, were analyzed in all the biomixtures. Only 3-ketocarbofuran was detected at the time of spiking at a concentration of 13.2 μg kg−1; however, none of the products was detected after 3 days in any of the biomixtures. This finding suggests that regardless of the

Environ Sci Pollut Res

A

B

98.43 49.2

Removal after 3 d (%)

3.8

100

140

100

80

120 2.975

100 80

CF/compost

Removal after 3 d (%)

90

90

60 40

3.8

70

2.15

×

1.325

2.975

80

42

2.15

35.2

CF/compost

28.4

1.325

21.6 14.8 0.5

8

% soil

0.5 8

14.8

21.6

28.4

35.2

42

% soil

Fig. 1 Surface response for the removal of CFN after 3 days (R1) in coconut fiber/compost/soil biomixtures of different composition. Three-dimensional response surface curve (a); contour plot (b). Standard composition (50:25:25) is indicated with “×”

composition, the FCS biomixtures are efficient at eliminating these transformation products. In other words, such metabolites are removed at higher rates than they are produced in the biomixture, thus limiting their detection with the analytical approach employed. These results remark the suitability of FCS biomixtures with respect to other biomixtures that show some accumulation of transformation products from CFN (Chin-Pampillo et al. 2015). Mineralization of 14C-CFN The mineralization rate of 14C-CFN was determined in all the runs for a period of 35 days; the rate constants and estimated half-lives are shown in Table 1. Rate constants ranged from 0.0009 to 0.0043 days−1; however, those extreme values were obtained from replicates of the central point of the CCD. A correlation between removal (Removal of CFN from the biomixtures: optimization section) and mineralization was not found in the biomixtures; this can be explained given that mineralization evaluates not only the transformation of parental CFN but also the complete oxidation of the molecule to CO2 and water; particularly, with the radiolabeled CFN employed (labeled in the ring moiety), mineralization is determining the breakdown of the ring, while removal considers only the initial transformation (dissipation) of CFN. The results from mineralization assays (R3) could not be modeled in the Expert Design software; however, some interesting trends were observed. First, leaving out the central point, the higher mineralization rates were achieved in the biomixtures containing 40 % soil, regardless of the ratio CF/compost (1.4 vs 3.4) (Fig. 2). It is suggested that larger initial microbial populations contained in pesticide-primed soil (not necessarily CFN degraders) might accelerate the degradation of the metabolites released from CFN transformation (Sniegowski et al. 2011); nonetheless, run 11 (46.2 % soil, the highest) showed a low mineralization rate, pointing out that biomixture performance

is based not only on soil content but also on a balance between its components. On the other hand, the lower mineralization rates correlated the lowest removal values after 3 days; nonetheless, for higher removal values, there was no correlation with mineralization rates. This finding could be ascribed to the fact that CFN can be used as a C source by microbiota in the biomixtures, and therefore, an important fraction of the 14C was assimilated into microbial biomass and not trapped in the KOH as 14CO2; similarly, part of the 14C contained in the transformation products was probably retained through the adsorption of these molecules to organic matter in the matrix. Moreover, the much longer mineralization half-lives (compared to the degradation half-lives) suggest that the removal of CFN metabolites is the limiting step in the mineralization, and not the initial transformation of the pesticide. Despite that removal rates are much faster in the biomixtures compared to other systems,

Fig. 2 Mineralization of 14C-CFN represented as the percentage of 14 CO2 evolved from several biomixtures composed of coconut fiber/ compost/soil at different proportions. Biomixtures: run 6 (●); run 8 (○); run 11 (▼); run 12 (Δ)

Environ Sci Pollut Res

mineralization half-lives resemble those obtained in soils previously unexposed to CFN (Ou et al. 1982). Nonetheless, most of the mineralization rates were similar or higher than those obtained for CFN in other biomixtures containing peat or compost, and lignocellulosic substrates such as rice husk, cane bagasse, and woodchips (Chin-Pampillo et al. 2015). Similarly, the lack of correlation between the mineralization rates and the residual toxicity also indicates that mineralization is not indispensable to achieve complete removal of toxicity from the biomixtures, which is in the end the goal of the system. Toxicity reduction during CFN removal: optimization In order to estimate differences among biomixtures of different compositions, the end point to determine toxicity was set in 8 days because previous screening of biomixtures prepared with several lignocellulosic residues produced almost complete toxicity removal after 60 days (including a CF-based biomixture) (Chin-Pampillo et al. 2015). The FCS biomixtures showed a marked ability to reduce the toxicity in the matrix during CFN removal. Initial toxicity values of around 200 TU obtained from the immobilization test with D. magna were sharply decreased after 8 days of treatment to residual values below 4 TU. Moreover, no toxicity was obtained in 8 out of the 13 runs. Besides the degradation and mineralization processes, the overall toxicity decrease achieved in the matrices could have been also favored by some extent of adsorption of residual CFN and its transformation products to the organic matter in the biomixtures on the course of the degradation process or Baging^ of the pesticide in the matrix (Alexander 2000). The correlation between the design variables and the residual toxicity after 8 days (R2) was analyzed through response surface methodology. Toxicity values were so low that fitting a model was a difficult task; a second-order polynomial model partially fitted the data; the mathematical regression model for residual toxicity in the actual factors is given as follows in Eq. (3): R2 ¼ −1:44 þ 0:13x1 −0:43x2 −0:05x1 x2 þ 0:46x22

ð3Þ

According to ANOVA employed to estimate the accuracy of the fit, this is not a significant model (Pmodel >F=0.1289); however, the adequate precision (4.697; desired values over 4) indicates that the model can be used to navigate the data and, therefore, it is useful to see the trend. The coefficient values (as coded factors) of the fitted model are shown in Table 4. Analysis of coefficients indicates that individual linear factor x2 exerts a positive relative impact on the residual toxicity, that is, higher x2 values tend to leave some low residual toxicity in the matrix. Similarly, the relative effect of soil content (x1) is quite low, and as observed in the

response surface, it seems not to determine the extent of toxicity elimination. The use of higher amounts of soil might favor the toxicity removal, perhaps due to a largest initial degrading population that results in a shorter lag phase in the combined removal of toxic metabolites. On the other hand, the combined effect of x1 and x2 markedly exerts a desirable and large effect on the biomixture, as it results in the largest extent of toxicity reduction. From the other terms in the equation, only the quadratic effect of x2 was significant with a positive effect, which indicates that more residual toxicity remains in the biomixtures at higher values of x22, in agreement with the linear effect of x2. The response surface and the corresponding contour plot are shown in Fig. 3. The standard 50:25:25 biomixture succeeded at minimizing residual toxicity and therefore fulfilled desired behavior, contrary to optimization results for R1. The contour plot confirms that most of the surface results in very low residual toxicity values, as desired. Therefore, one single optimal combination of design variables was not selected to minimize residual toxicity. However, the area that combines % soil values from 8 % to 12 % and CF/compost ratios above 3 results in higher residual toxicity values; therefore, such combinations should be avoided in the design of biomixtures for CFN removal. Global optimization of the biomixture composition Simultaneous optimization was performed using the variable Bdesirability,^ which considers the models explaining the behavior of both response variables, R1 and R2. The criteria for optimization were: maximizing the CFN removal after 3 days and minimizing the residual toxicity after 8 days. The response surface and the contour plots for desirability are presented in Fig. 4. From the numerical solutions yielded by the software, the one with the highest desirability (0.889) was obtained for the values: % soil= 42.0 and CF/compost=3.45, which correspond to a volumetric proportion of 45:13:42 (F/C/S). The optimized composition is almost identical to that selected only taking into account the response R1 (Removal of CFN from the biomixtures: optimization section), and permitted to distinguish between the two optimum regions for R1: the region containing low values for x1 and high for x2 does not result in optimum desirability as it produces relatively higher residual toxicity values; in the meantime, the area that comprises high values for both x1 and x2 produces almost negligible residual toxicity. This latter area coincidently contains the optimum combination of x1 and x2 to maximize both desirability (as described above), and R1. The optimized composition markedly differs from the standard 50:25:25 biomixture and basically suggests that a larger amount of soil should be employed in biomixture preparation, by slightly reducing the amount of the lignocellulosic substrate and half-reducing the amount of compost to

Environ Sci Pollut Res 3.11

A

B

0

Residual toxicity after 8d (TU)

3.4

1

3

2.85

2

CF/compost

Residual toxicity after 8d (TU)

4

1 0 -1

2.3

0

1.75

3.4 2.85 2.3

CF/compost

1.75 1.2

8

14.8

28.4

21.6

35.2

1

42 1.2 8

% soil

14.8

21.6

28.4

35.2

42

% soil

Fig. 3 Surface response of the residual toxicity of the matrix after 8 days (R2) during the removal of CFN in coconut fiber/compost/soil biomixtures of different composition. Three-dimensional response surface curve (a); contour plot (b). Standard composition (50:25:25) is indicated with “×”

13 % as humic component. Most studies that employ compost as a substitute for peat use the same concentration of 25 % (Fogg et al. 2003; Karanasios et al. 2010a, b); nonetheless, Coppola et al. (2011) demonstrated higher removal of isoproturon and bentazone in biomixtures containing 50 % compost (compared to 87.5 %), which supports the finding that lower amounts of compost results in better performance of biomixtures. On the contrary, a biomixture made of soil and compost (75:25v/v) was more efficient to degrade chlorpyrifos than a straw/compost/soil biomixture (25:25:50) (Kravvariti et al. 2010). These contrasting results highlight the importance of optimizing biomixtures according to their intended application. Differences among the standard and the optimized compositions for carbofuran degradation relies in the fact that a larger amount of soil might contribute with larger populations at the moment of biomixture preparation; the effect of such

A

populations will probably translate into shorter lag phases for the processes of toxicity decrease and pesticide removal.

Conclusions Optimization of a FCS biomixture intended for the removal of CFN resulted in a volumetric composition of 45:13:42 (F/C/S). This composition differs from the standard 50:25:25 biomixture that has been mostly employed since biobeds appeared. As it happened for the treatment of CFN, the optimal removal of other pesticides in BPS might take place at different biomixture compositions; therefore, conditions should be evaluated in each case. Moreover, the optimization of biomixtures for the simultaneous removal of several pesticides is highly recommended, particularly taking into account the diverse pesticides employed in the application program for a specific crop. In this work the

B

1

Desirability

3.6

0 0.8

1

3

0.6 CF/compost

Desirability

0.8 0.4 0.2 0 3.6

0.6

0.6

2.4 0.8

1.8

3

42 33.5

2.4

CF/compost

1.2

25 1.8

16.5 1.2

8

% soil

8

Fig. 4 Surface response for the desirability (which combines the effect of the CFN removal and residual toxicity as response variables) during the removal of CFN in coconut fiber/compost/soil biomixtures of different

16.5

25

33.5

42

% soil

composition. Three-dimensional response surface curve (a); contour plot (b). Standard composition (50:25:25) is indicated with “×”

Environ Sci Pollut Res

optimization included not only the removal of CFN but also the elimination of toxicity within the biomixture during the treatment of the pesticide, thus providing a global assess of the environmental suitability of the matrix to be employed in BPS. To obtain more insights into potential residual toxicity of the optimized biomixture, the use of toxicity tests employing organisms from different levels of the trophic chain is highly recommended, particularly if the biomixture is intended for the simultaneous treatment of several pesticides. Acknowledgments This work was supported by Vicerrectoría de Investigación, Universidad de Costa Rica (projects 802-B2-046 and 802-B4-503), the Costa Rican Ministry of Science, Technology and Telecommunications, MICITT (project FI-093-13/802-B4-503), and the Joint FAO/IAEA project TC COS5/029.

References Alexander M (2000) Aging, bioavailability, and overestimation of risk from environmental pollutants. Environ Sci Technol 34:4259–4265 Asgher M, Bhatti HN, Ashraf M, Legge RL (2008) Recent developments in biodegradation of industrial pollutants by white rot fungi and their enzyme system. Biodegradation 19:771–783 Carter AD (2000) How pesticides get into water-and proposed reduction measures. Pestic Outlook 11:149–156 Castillo MP, Torstensson L (2007) Effect of biobed composition moisture, and temperature on the degradation of pesticides. J Agric Food Chem 55:5725–5733 Castillo MP, Torstensson L, Stenström J (2008) Biobeds for environmental protection from pesticide use - a review. J Agric Food Chem 56: 6206–6219 Chin-Pampillo JS, Ruiz-Hidalgo K, Aguilar-Mora P, Arias-Mora V, Masís-Mora M (2012) Quality of water from Quebrada Sanatorio (Tierra Blanca), in the county region of Cartago province and impacts on public health. O Mundo da Saúde 36:548–555 Chin-Pampillo JS, Ruiz-Hidalgo K, Masís-Mora M, Carazo-Rojas E, Rodríguez-Rodríguez CE (2015) Adaptation of biomixtures for carbofuran degradation in on-farm biopurification systems in tropical regions. Environ Sci Pollut Res. doi:10.1007/s11356-015-4130-6 Coppola L, Castillo MP, Monaci E, Vischetti C (2007) Adaptation of the biobed composition for chlorpyrifos degradation to southern Europe conditions. J Agric Food Chem 55:396–401 Coppola L, Castillo MP, Vischetti C (2011) Degradation of isoproturon and bentazone in peat- and compost-based biomixtures. Pest Manag Sci 67:107–113 de Roffignac L, Cattan P, Mailloux J, Herzog D, Le Bellec F (2008) Efficiency of a bagasse substrate in a biological bed system for the degradation of glyphosate, malathion and lambda-cyhalothrin under tropical climate conditions. Pest Manag Sci 64:1303–1313 de Wilde T, Mertens J, Simunek J, Sniegowski K, Ryckeboer J, Jaeken P, Springael D, Spanoghe P (2009a) Characterizing pesticide sorption and degradation in microscale biopurification systems using column displacement experiments. Environ Pollut 157:463–473 de Wilde T, Spanoghe P, Mertens J, Sniegowski K, Ryckeboer J, Jaeken P, Springael D (2009b) Characterizing pesticide sorption and degradation in macroscale biopurification systems using column displacement experiments. Environ Pollut 157:1373–1381 EPA (2001) Methods for collection, storage and manipulation of sediments for chemical and toxicological analyses: technical manual (EPA/823/B-01/002). Office of Water (4305), Washington DC

EPA (2002) Methods for measuring the acute toxicity of effluents and receiving waters to freshwater and marine organisms (EPA/821/R02/012). Office of Water (4303T), Washington DC Fogg P, Boxall ABA, Walker A, Jukes A (2003) Pesticide degradation in a Bbiobed^ composting substrate. Pest Manag Sci 59:527–537 Fogg P, Boxall ABA, Walker A, Jukes A (2004) Leaching of pesticides from biobeds: effect of biobed depth and water loading. J Agric Food Chem 52:6217–6227 Karanasios E, Tsiropoulos NG, Karpouzas DG, Ehaliotis C (2010a) Degradation and adsorption of pesticides in compost-based biomixtures as potential substrates for biobeds in southern Europe. J Agric Food Chem 58:9147–9156 Karanasios E, Tsiropoulos NG, Karpouzas DG, Menkissoglu-Spiroudi U (2010b) Novel biomixtures based on local Mediterranean lignocellulosic materials: evaluation for the use in biobed systems. Chemosphere 80:914–921 Karanasios E, Papadi-Psyllou A, Karpouzas DG, Tsiropoulos NG (2012) Optimization of biomixture composition and water management for maximum pesticide dissipation in peat-free biobeds. J Environ Qual 41:1787–1795 Kravvariti K, Tsiropoulos NG, Karpouzas DG (2010) Degradation and adsorption of terbuthylazine and chlorpyrifos in biobed biomixtures from composted cotton crop residues. Pest Manag Sci 66:1122– 1128 Ou LT, Gancarz DH, Wheeler WB, Rao PSC, Davidson JM (1982) Influence of soil temperature and soil moisture on degradation and metabolism of carbofuran in soils. J Environ Qual 11:293–298 Pussemier L, de Vleeschouwe C, Debongnie P (2004) Self-made biofilters for on-farm clean-up of pesticides wastes. Outlook Pest Manag 15:60–63 Ruiz-Hidalgo K, Chin-Pampillo JS, Masís-Mora M, Carazo E, Rodríguez-Rodríguez CE (2014) Degradation of carbofuran by Trametes versicolor in rice husk as a potential lignocellulosic substrate for biomixtures: from mineralization to toxicity reduction. Process Biochem 49:2266–2271 Sniegowski K, Bers K, van Goetem K, Ryckeboer J, Jaeken P, Spanoghe P, Springael D (2011) Improvement of pesticide mineralization in on-farm biopurification systems by bioaugmentation with pesticideprimed soil. FEMS Microbiol Ecol 76:64–73 Sniegowski K, Bers K, van Goetem K, Ryckeboer J, Jaeken P, Spanoghe P, Springael D (2012) Minimal pesticide-primed soil inoculum density to secure maximum pesticide degradation efficiency in on-farm biopurification systems. Chemosphere 88:1114–1118 Spanoghe P, Maes A, Steurbaut W (2004) Limitation of point source pesticide pollution: results of bioremediation system. Commun Agric Appl Biol Sci 69:719–732 Spliid NH, Helweg A, Heinrichson K (2006) Leaching and degradation of 21 pesticides in a full-scale model biobed. Chemosphere 65: 2223–2232 Torstensson L, Castillo MP (1997) Use of biobeds in Sweden to minimize environmental spillages from agricultural spraying equipment. Pestic Outlook 8:24–27 Urrutia C, Rubilar O, Tortella GR, Diez MC (2013) Degradation of pesticide mixture on modified matrix of a biopurification system with alternatives lignocellulosic wastes. Chemosphere 92:1361–1366 Vischetti C, Capri E, Trevisan M, Casucci C, Perucci P (2004) Biomassbed: a biological system to reduce pesticide point contamination at farm level. Chemosphere 55:823–828 Wenneker M, Beltman WHJ, de Werd HAE, van Zeeland MG, van der Lans A, van der Weide RY (2010) Quantifying point source entries of pesticides in surface waters. Asp Appl Biol 99:69–74

Design of an optimized biomixture for the degradation of carbofuran based on pesticide removal and toxicity reduction of the matrix.

Pesticide biopurification systems contain a biologically active matrix (biomixture) responsible for the accelerated elimination of pesticides in waste...
1KB Sizes 1 Downloads 9 Views