This article was downloaded by: [New York University] On: 08 May 2015, At: 10:46 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Food Additives & Contaminants: Part A Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tfac20
Non-destructive detection of pesticide residues in cucumber using visible/near-infrared spectroscopy a
b
c
d
a
Bahareh Jamshidi , Ezeddin Mohajerani , Jamshid Jamshidi , Saeid Minaei & Ahmad Sharifi a
Agricultural Engineering Research Institute, Agricultural Research Education and Extension Organization (AREEO), Karaj, Iran b
Laser and Plasma Research Institute, Shahid Beheshti University, Tehran, Iran
c
Zarif Mosavar Industrial Manufacturing Company, Isfahan, Iran
d
Agricultural Machinery Engineering Department, Faculty of Agriculture, Tarbiat Modares University, Tehran, Iran Accepted author version posted online: 19 Mar 2015.Published online: 14 Apr 2015.
Click for updates To cite this article: Bahareh Jamshidi, Ezeddin Mohajerani, Jamshid Jamshidi, Saeid Minaei & Ahmad Sharifi (2015) Non-destructive detection of pesticide residues in cucumber using visible/near-infrared spectroscopy, Food Additives & Contaminants: Part A, 32:6, 857-863, DOI: 10.1080/19440049.2015.1031192 To link to this article: http://dx.doi.org/10.1080/19440049.2015.1031192
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Food Additives & Contaminants: Part A, 2015 Vol. 32, No. 6, 857–863, http://dx.doi.org/10.1080/19440049.2015.1031192
Non-destructive detection of pesticide residues in cucumber using visible/near-infrared spectroscopy Bahareh Jamshidia*, Ezeddin Mohajeranib, Jamshid Jamshidic, Saeid Minaeid and Ahmad Sharifia a
Agricultural Engineering Research Institute, Agricultural Research Education and Extension Organization (AREEO), Karaj, Iran; Laser and Plasma Research Institute, Shahid Beheshti University, Tehran, Iran; cZarif Mosavar Industrial Manufacturing Company, Isfahan, Iran; dAgricultural Machinery Engineering Department, Faculty of Agriculture, Tarbiat Modares University, Tehran, Iran b
Downloaded by [New York University] at 10:46 08 May 2015
(Received 7 January 2015; accepted 14 March 2015) The feasibility of using visible/near-infrared (Vis/NIR) spectroscopy was assessed for non-destructive detection of diazinon residues in intact cucumbers. Vis/NIR spectra of diazinon solution and cucumber samples without and with different concentrations of diazinon residue were analysed at the range of 450–1000 nm. Partial least squares-discriminant analysis (PLS-DA) models were developed based on different spectral pre-processing techniques to classify cucumbers with contents of diazinon below and above the MRL as safe and unsafe samples, respectively. The best model was obtained using a first-derivative method with the lowest standard error of cross-validation (SECV = 0.366). Moreover, total percentages of correctly classified samples in calibration and prediction sets were 97.5% and 92.31%, respectively. It was concluded that Vis/NIR spectroscopy could be an appropriate, fast and non-destructive technology for safety control of intact cucumbers by the absence/presence of diazinon residues. Keywords: cucumber; diazinon; maximum residue level; near-infrared spectroscopy; non-destructive; pesticide residues
Introduction Fruit and vegetables are vital for a healthy diet because they contain vitamins and minerals. Pesticides such as fungicides, insecticides and herbicides are chemical treatments that are widely used during the growing stage of such agricultural crops to control plant pests and diseases, as well as to increase productivity. By using pesticides excessively, incorrectly or irrationally during production, ignoring the degradation period of pesticides and early harvesting, transferring to the market and selling fruit and vegetable products immediately or a few days after spraying, the pesticides will remain as chemical residues in these products. In most cases, failure to meet safety requirements when using these chemical treatments may mean that residues in fruit and vegetables exceed the MRLs established by current standards, thus indicating a potential risk to consumer health and safety (Sánchez et al. 2010; Peng et al. 2012). To ensure public health, it is essential to detect and control pesticide contamination of agricultural products (Saranwong & Kawano 2005). Typically, the determination of pesticides in fruit and vegetables involves a sample treatment using different techniques such as SPE, supercritical fluid extraction, microwave-assisted extraction and accelerated solvent extraction. Also, conventional techniques such as GC, HPLC, TLC, supercritical fluid chromatography, GC-MS, capillary electrophoresis, an enzyme inhibition method, an immunoassay method and a biosensor method are *Corresponding author. Email: [email protected] © 2015 Taylor & Francis
used to measure the concentration of pesticide residue (Juan-García et al. 2005; Li et al. 2013; Cho et al. 2014; Masiá et al. 2014; Mercader et al. 2014; Tao et al. 2014; Watanabe et al. 2015). However, these techniques are destructive, have a highly time-consuming sample preparation, are very expensive, and require well-trained personnel and an advanced laboratory for controlling individual products (Sánchez et al. 2010; Teye et al. 2013). Therefore, development of a non-destructive, fast, simple, low-cost, environmentally friendly and reliable detection technique of pesticide residue is imperative. Near-infrared (NIR) spectroscopy is one of the most promising non-destructive techniques that is flexible for both qualitative and quantitative analyses in chemistry, agriculture, medicine and other areas. It is rapid, less expensive than conventional methods and environmentally friendly, requires little or no sample preparation and can be used in processing lines (Sánchez et al. 2010; Xue et al. 2012). While visible/near-infrared (Vis/NIR) and NIR spectroscopy have been widely used for quality assessment and chemical compounds measurement of intact fruit and vegetables (Subedi & Walsh 2011; Jamshidi et al. 2012, 2014; Sánchez et al. 2012; Tiwari et al. 2013; Maniwara et al. 2014; Pan et al. 2015), few publications have addressed the use of this technique for the detection and determination of pesticide residues in such products. One of the first studies with this issue was carried out by Sarawong and Kawano (2005), who used a dry-extract
Downloaded by [New York University] at 10:46 08 May 2015
858
B. Jamshidi et al.
system for infrared analysis (DESIR) technique to detect dichlofluanid fungicide levels on the surface of tomato. Sánchez et al. (2010) assessed the feasibility of reflectance NIR spectroscopy to detect pesticide residues in peppers. Moreover, non-destructive detection of trichlorfon pesticide residues on the surface of longans was studied by Dai et al. (2010) based on a Vis/NIR spectroscopy method. Xue et al. (2012) used this technique for the determination of dichlorvos residues on the surface of navel oranges. Salguero-Chaparro et al. (2013) also assessed the viability of using NIR spectroscopy to detect diuron herbicide levels in intact olives. All these researches confirm the possibility and reasonability of detection of these pesticide residues using NIR spectroscopy. Cucumber is a popular vegetable in people’s diet and is cultivated and grown in most countries. According to the FAO, Iran is third in world ranking in terms of the production quantity of cucumbers (FAO 2012). In Iran, greenhouse production of cucumber is becoming more popular. While cucumber is susceptible to insect attacks, especially in greenhouse conditions, intensive use of pesticides is needed to control insects. Diazinon is a dangerous organophosphorus pesticide that is widely used to control insects and pests in greenhouses (Cengiz et al. 2006). Unfortunately, using diazinon excessively and incorrectly, even in the few hours before harvesting, has caused accumulation of pesticide residues in this product at levels higher than the MRL. This research aims to assess the feasibility of using a Vis/NIR spectroscopy technique combined with partial least squares regression-discriminant analysis (PLS-DA) for non-destructive detection of diazinon residues in cucumber that contain levels higher than the MRL (0.1 mg kg−1) established by the FAO/WHO Codex Alimentarius (FAO/WHO 2013).
length) were also measured by digital balance and Vernier caliper, respectively. Reference measurements were carried out 1 day after spectroscopy analysis. Therefore, cucumbers were frozen to be sent to the laboratory for reference data analysis.
Vis/NIR spectroscopy A USB2000 spectrometer (Oceanoptics Inc., Dunedin, FL, USA) with charge coupled device (CCD) detector and a tungsten halogen light source (LS-1; Oceanoptics) were used for Vis/NIR spectroscopy at a range of 450–1000 nm with 1.5 nm resolution. In order to identify the spectral regions at which diazinon absorbance took place and to determine the absorbance bands of the cucumber spectra that might correspond to diazinon, the pesticide solution was placed in a quartz cell and its transmittance spectrum (T) was acquired using two fibre optics of P400-2-Vis-NIR model (Oceanoptics) (Figure 1a). Then, the collected spectrum was converted to absorbance values (log 1/T). To collect the spectra of intact cucumbers including internal information of the sample, a fibre optic of P400-7-Vis-NIR model (Oceanoptics) was used for spectroscopy in interactance mode (Figure 1b). For each sample, interactance spectra at six positions along the cucumbers and around equatorial locations on the samples (Figure 1c) with five scans at each position were acquired by OOIBase32 software (Oceanoptics). Then, the mean spectrum was
Materials and methods Sample preparation A total of 120 cucumbers were harvested in greenhouses near Karaj city in Alborz province, Iran, and stored at 5°C until use. Some cucumbers with no treatment were used as a diazinon-free set. The commercial pesticide, diazinon (O,O-diethyl-O-(6-methyl-2-(1-methylethyl)-4pyrimidinyl)phosphorothioate), containing 60% diazinon (C12H21N2O3PS), was utilised to obtain a different pesticide residue content in other samples. The pesticide was diluted to 1/500 and sprayed on some of these cucumbers. Moreover, the remained samples were placed in the prepared solution for 1 h. All samples were stored at 5°C until Vis/NIR measurements. Prior to each measurement, they were left to reach equilibrium temperature in a laboratory environment. Morphological properties of each sample including weight and size (maximum diameter and
Figure 1. Spectra acquisition of diazinon solution (a) and intact cucumbers (b) in six positions on the samples (c).
Food Additives & Contaminants: Part A calculated from a total of 30 scans for each cucumber and converted to absorbance values (log 1/R).
Downloaded by [New York University] at 10:46 08 May 2015
Reference measurements After Vis/NIR spectroscopy, all cucumbers were sent to the Chemical Analysis Center at the Iranian Institute of R&D in Chemical Industries for diazinon measurement using a reference method. To this end, sample preparation was done based on British Standard BS EN 15662:2008 (British Standard 2008). Extracted sample (2 µl) was injected into the GC using an Agilent 7890A gas chromatograph (Agilent Technologies Inc., Santa Clara, CA, USA) fitted with a flame ionisation detector. It was equipped with an Agilent Technologies 7683B series injector, a split/splitless injector port operated in splitless mode and with a HP-5 capillary column (30 m × 0.32 mm i.d.), film thickness 0.25 µm. Helium was used as the carrier gas. Throughout the analysis, the injector and detector temperature were kept at 280 and 300°C, respectively. The column temperature was programmed from 50°C (held for 2 min) to 225°C at 10°C min−1 then to 300°C at 30°C min−1 (held for 2 min). The analyte (diazinon) was determined with a linearity range of 0.01–5 mg kg−1. Results were summarised as safe (all samples with or without diazinon concentrations ≤ 0.1 mg kg−1) or unsafe (all samples with diazinon concentrations > 0.1 mg kg−1) groups. Data analysis Before developing the classification models, outliers which are the samples containing interferences (sample contamination, measurement artefacts or temperature effects) having a negative influence on modelling (Heise & Winzen 2006) were detected and removed. To this end, the spectral variation of all the samples was analysed using PCA and the outlier samples were determined as the points outside the normal range of variability in the PCA scores plot (Nicolaï et al. 2007). Figure 2 shows the PCA scores plot for all the samples with the detected outliers. After removing the outliers (14 samples), all the
859
106 remained samples were arranged in order of their diazinon residue measured by the reference method of GC analysis. Then, after every three samples, one sample was selected for the external validation set. Therefore, 26 samples (approximately 25% of the samples), i.e. one out of every four samples in the overall set, were used as the external validation set. The remained 80 samples (approximately 75% of the samples) were selected for the calibration set. Partial least squares-discriminant analysis (PLS-DA) was performed for classification of the cucumbers in two groups defined as safe or unsafe samples. Briefly, PLS-DA is a supervised classification technique based on modelling the differences among several classes using a PLS calibration model to discriminate the unknown samples. In this approach, the PLS model relates the spectral variations (X) to the assigned dummy variables (Y) by maximising the covariance between two types of variable (Cen et al. 2007; Gu et al. 2012). When there are only two classes to discriminate, this model uses one response variable (Y) coding for each class as follows: −1 for membership of one class and +1 for membership of the other class. Each unknown sample takes a predicted value between −1 and +1 after predicting with the developed PLS model. A sample with predicted value close to −1 will belong to one class and if the predicted value is close to +1 the sample is identified as the other class. However, a value close to zero indicates that it is not easy to belong it to any of the classes, especially when the estimated deviation (uncertainty) around the predicted value includes zero. All PLS models were developed with full crossvalidation in the calibration set and investigating the maximum of 10 latent variables (LVs). Before modelling, multiplicative scatter correction (MSC) and standard normal variate (SNV) were used to correct both multiplicative and additive effects of the spectra. To increase the spectral resolution, first and second derivatives of the spectra (D1, D2) based on the Savitzky–Golay algorithm with five smoothing points and polynomial order of 2, were also performed. The precision of the developed PLS-DA models was evaluated based on having the lowest standard error of cross-validation (SECV) and the highest percentage of correctly classified samples for both calibration and external validation sets. All analyses were conducted using Unscrambler software X10.3 (CAMO Software, Oslo, Norway).
Results and discussion Statistics of the samples Figure 2. (colour online) PCA scores plot (PC1 × PC2 × PC3) with the detected outliers.
The statistics of morphological properties and diazinon contents of the samples are shown in Tables 1 and 2, respectively. According to Table 1, the weight, maximum
860
B. Jamshidi et al.
Table 1. Statistics of both calibration and prediction sets for morphological properties of the samples. Calibration set (N = 80)a
Weight (g) Diameter (mm) Length (mm)
Prediction set (N = 26)
Range
Mean
SDb
Range
Mean
SD
37.937–93.081 20.000–36.000 110.500–150.400
59.989 25.484 131.706
13.204 2.932 9.717
41.965–91.108 20.000–30.400 110.500–161.000
60.867 25.419 132.346
12.664 2.324 13.203
Notes: aN, number of the samples. b SD, standard deviation.
Table 2. Statistics of both calibration and prediction sets for diazinon contents in the samples.
Downloaded by [New York University] at 10:46 08 May 2015
Diazinon (mg kg−1)
Diazinon (mg kg−1)
N
Range
Mean
SD
Group
N
Range
Mean
SD
Calibration set
80
0.000–32.000
5.069
7.684
Prediction set
26
0.000–28.000
5.867
8.298
Safe Unsafe Safe Unsafe
16 64 5 21
0.000–0.070 0.650–32.000 0.000–0.100 0.200–28.000
0.008 6.335 0.020 7.260
0.021 8.118 0.450 8.689
equatorial diameter and length of the samples were in the range of 37.94–93.08 g, 20–36 mm and 110.5–161 mm, respectively. Therefore, the cucumbers were quite varied in terms of morphology. As can be seen in Table 2, there was a large variability in the diazinon concentrations of the samples ranging from 0 to 32 mg kg−1 including both safe (MRL ≤ 0.1 mg kg−1) and unsafe (MRL > 0.1 mg kg−1) cucumbers.
Spectra interpretation Figure 3 shows the chemical structure of diazinon, the Vis/ NIR absorbance spectrum (log 1/T) of diazinon solution and its first derivative in the ranges of 450–1000 nm. The curve had a decreasing trend in the visible region. A perceptible peak around 900 nm was found that could be due to the second overtone of O-H or the third overtone of C-H according to the distribution of overtones of organic bonds in NIR region (Cen & He 2007). After that there was an increasing trend up to 1000 nm because of the second overtone of O-H. According to the chemical structure of diazinon, the spectrum trend in NIR region could be more related to C-H absorbance. The first derivative of the spectrum also confirmed the absorbance peaks in NIR region. The absorbance spectra (log 1/R) of intact cucumbers with different pesticide concentrations and their first derivative are also presented in Figure 4. The absorbance in the visible region was mainly related to compounds responsible for cucumber colour. There was a strong peak around 680 nm due to the chlorophyll a content of
Figure 3. Absorbance (log 1/T) spectrum of diazinon solution and its first derivative.
Food Additives & Contaminants: Part A
861
NIR region, the increasing trend of the spectra could be due to the second overtone of O-H or the third overtone of C-H. Therefore, it was noted that diazinon measurements in the NIR region of the spectrum may be more related to C-H absorbance similar to that suggested by Saranwong and Kawano (2005) and Sánchez et al. (2010) for other pesticide residues.
Downloaded by [New York University] at 10:46 08 May 2015
PLS-DA analysis Table 3 shows the results of PLS-DA models developed with different pre-processing methods for discrimination of the cucumbers based on the presence/absence of diazinon residues. All the developed calibration models had an excellent ability to discriminate the samples into two groups of safe and unsafe cucumbers based on the presence of diazinon residues. After cross-validation in the calibration set, the total percentage of correctly classified samples was 97.5% for all models developed using the different pre-processing techniques. In calibration, there were 16 samples of safe group of which only two samples were misclassified. Therefore, the percentage of correctly classified samples of safe group was 87.5%. However, 100% of the samples of unsafe group in the calibration set (64 samples) were correctly classified. While all the models developed using various preprocessing techniques had similar predictive ability as reported by Sánchez et al. (2010), the best results were achieved using first derivative method with the lowest standard error of cross-validation (SECV = 0.366). The predicted versus reference values for this calibration model are presented in Figure 5. This plot shows how close to the ideal values of −1 and 1 (groups of safe and unsafe, respectively) are the predicted values. To investigate how the discriminant model will behave on the unknown samples and predict their class, external validation was performed on the samples of the prediction set not included in model development (26 samples). Results indicated that all PLS-DA models had very good
Figure 4. Absorbance (log 1/R) spectra of intact cucumbers with different pesticide concentrations and their first derivative.
the samples, similar to that described by Kavdir et al. (2007). According to the first derivative of the spectra, more details were also found in the visible region around 650 nm regarding the absorbance of chlorophyll b. In the
Table 3. Results of PLS-DA models for discrimination of safe and unsafe cucumbers based on the presence of diazinon residues. Calibration set
Prediction set
Correctly classified (%)
MSC SNV D1 D2 MSC + D1 MSC + D2 SNV + D1 SNV + D2
Correctly classified (%)
LVs
Total
Safe
Unsafe
SECV
Total
Safe
Unsafe
8 8 7 9 7 4 7 4
97.5 97.5 97.5 97.5 97.5 97.5 97.5 97.5
87.5 87.5 87.5 87.5 87.5 87.5 87.5 87.5
100 100 100 100 100 100 100 100
0.378 0.383 0.366 0.369 0.374 0.396 0.372 0.392
88.46 88.46 92.31 92.31 92.31 88.46 92.31 88.46
60 60 60 60 60 60 60 60
95.24 95.24 100 100 100 95.24 100 95.24
862
B. Jamshidi et al. 2 Unsafe
1.5
Predicted class
1 0.5 0 -1.5
-1
-0.5
0
0.5
1
1.5
-0.5 -1
destructively screening the cucumbers with a content of diazinon above/below the MRL (0.1 mg kg−1). This conclusion was in agreement with the results reported by Sánchez et al. (2010) and Salguero-Chaparro et al. (2013) who assessed the feasibility of reflectance NIR spectroscopy in the range of 1100–1650 nm to detect residues of some pesticide classes in intact peppers and in the range of 400–2500 nm to detect diuron herbicide levels in intact olives, respectively.
-1.5 Safe -2 Reference class
ability to classify the unknown samples. The total percentage of correctly classified samples of the prediction set using the developed models was above 88%. It was also 92.31% for the best calibration model developed based on the first derivative method. While all the samples of unsafe group were 100% correctly classified, the percentage of correctly classified samples of safe group was 60%. Figure 6 shows the predicted values of the unknown samples as the horizontal lines with the estimated uncertainties displayed as the vertical lines for the best calibration model. As it can be seen, all unsafe samples had predicted values close to 1 which assigns them to class ‘unsafe’. For safe group, while two samples were identified as the unsafe group because of having the predicted values close to 1, the remained samples had predicted values close to −1, which assigns them to class ‘safe’. Moreover, no sample had the predicted value close to zero and the deviations around the predicted value did not include zero. Therefore, the quality of the prediction for the unknown samples could be ensured. The results showed the potential of interactance Vis/ NIR spectroscopy in the range of 450–1000 nm for non-
Acknowledgements The authors are grateful to the Agricultural Engineering Research Institute and the Laser & Plasma Institute of Shahid Beheshti University for their suppport, and to the Chemical Analysis Center at the Iranian Institute of R&D in Chemical Industries for cooperation on reference measurements.
Disclosure statement No potential conflict of interest was reported by the authors.
Funding 2.5
This research was funded by the Iran National Science Foundation (INSF) [project number 91051000].
2 1.5 1
References
0.5 0 -0.5
Safe Safe Safe Safe Safe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe
Predicted value
Downloaded by [New York University] at 10:46 08 May 2015
Figure 5. Predicted versus reference values plot obtained using the calibration model developed based on a first-derivative preprocessing method.
Conclusions The feasibility of utilising Vis/NIR spectroscopy combined with PLS-DA was investigated for nondestructive detection of diazinon residues in intact cucumbers that contain levels below/above the MRL. The results confirmed the ability of this technology for discrimination of intact cucumbers by the presence/ absence of diazinon residues. Therefore, Vis/NIR spectroscopy could be used for fast, low-cost and non-destructive preliminary safety and healthy control of the cucumbers before or after harvesting. Nevertheless, further work should be considered to adapt the Vis/NIR spectroscopy to detect diazinon residues in other varieties of cucumber at different regions which have been contaminated directly by pesticide spraying in greenhouse and field.
-1 -1.5
Samples
Figure 6. Predicted values with estimated deviation for the unknown samples using the best calibration model.
British Standard (BS EN 15662). 2008. Foods of plant origin – Determination of pesticide residues using GC-MS and/or LC-MS/MS following acetonitrile extraction/partitioning and clean-up by dispersive SPE – QuEChERS-method. Cen H, He Y. 2007. Theory and application of near infrared reflectance spectroscopy in determination of food quality. Trends Food Sci Technol. 18:72–83. Cen H, He Y, Huang M. 2007. Combination and comparison of multivariate analysis for the identification of orange varieties using visible and near infrared reflectance spectroscopy. Eur Food Res Technol. 225:699–705.
Downloaded by [New York University] at 10:46 08 May 2015
Food Additives & Contaminants: Part A Cengiz M, Certel M, Gocmen H. 2006. Residue contents of DDVP (Dichlorvos) and diazinon applied on cucumbers grown in greenhouses and their reduction by duration of a pre-harvest interval and post-harvest culinary applications. Food Chem. 98:127–135. Cho S-K, Abd El-Aty AM, Musfiqur Rahman M, Choi J-H, Shim J-H. 2014. Simultaneous multi-determination and transfer of eight pesticide residues from green tea leaves to infusion using gas chromatography. Food Chem. 165:532–539. Dai F, Hong T, Zhang K, Hong Y. 2010. Nondestructive detection of pesticide residue on longan surface based on near infrared spectroscopy. Int Conf Intell Comput Technol Autom (ICICTA). 2:781–783. [FAO] Food and Agriculture Organization of the United Nations. 2012. FAOSTAT Database. Available from: http://faostat3. fao.org/download/Q/QC/E. [FAO/WHO] Food and Agriculture Organization of the United Nations and World Health Organization. 2013. Codex Pesticides Residues in Food Online Database. Available from: http://www.codexalimentarius.net/pestres/data/index.html. Gu C, Xiang B, Xu J. 2012. Direct detection of phoxim in water by two-dimensional correlation near-infrared spectroscopy combined with partial least squares discriminant analysis. Spectrochim Acta, Part A. 97:594–599. Heise HM, Winzen R. 2006. Chemometrics in near-infrared spectroscopy. In: Siesler HW, Ozaki Y, Kawata S, Heise HM, editors. Near-infrared spectroscopy: principles, instruments, applications. Germany: Wiley-VCH; p. 125–162. Jamshidi B, Minaei S, Mohajerani E, Ghassemian H. 2012. Reflectance Vis/NIR spectroscopy for nondestructive taste characterization of Valencia oranges. Comput Electron Agric. 85:64–69. Jamshidi B, Minaei S, Mohajerani E, Ghassemian H. 2014. Prediction of soluble solids in oranges using visible/nearinfrared spectroscopy: effect of peel. Int J Food Prop. 17:1460–1468. Juan-García A, Picó Y, Font G. 2005. Capillary electrophoresis for analyzing pesticides in fruits and vegetables using solidphase extraction and stir-bar sorptive extraction. J Chromatogr A. 1073:229–236. Kavdir I, Lu R, Ariana D, Ngouajio M. 2007. Visible and nearinfrared spectroscopy for nondestructive quality assessment of pickling cucumbers. Postharvest Biol Technol. 44:165–174. Li W, Sun M, Li M. 2013. A survey of determination for organophosphorus pesticide residue in agricultural products. Adv J Food Sci Technol. 5:381–386. Maniwara P, Nakano K, Boonyakiat D, Ohashi S, Hiroi M, Tohyama T. 2014. The use of visible and near infrared spectroscopy for evaluating passion fruit postharvest quality. J Food Eng. 143:33–43. Masiá A, Blasco C, Picó Y. 2014. Last trends in pesticide residue determination by liquid chromatography–mass spectrometry. Trends Environ Anal Chem. 2:11–24. Mercader JV, López-Moreno R, Esteve-Turrillas FA, Abad-Somovilla A, Abad-Fuentes A. 2014. Immunoassays
863
for trifloxystrobin analysis. Part II. Assay development and application to residue determination in food. Food Chem. 162:41–46. Nicolaï BM, Beullens K, Bobelyn E, Peirs A, Saeys W, Theron KI, Lammertyn J. 2007. Nondestructive measurement of fruit and vegetable quality by means of NIR spectroscopy: a review. Postharvest Biol Technol. 46:99–118. Pan L, Zhu Q, Lu R, Mitchell McGrath J. 2015. Determination of sucrose content in sugar beet by portable visible and nearinfrared spectroscopy. Food Chem. 167:264–271. Peng Y, Li Y, Chen J. 2012. Optical technologies for determination of pesticide residue. In: Theophile T, editor. Infrared spectroscopy - materials science, engineering and technology. Rijeka: InTech; p. 453–466. Salguero-Chaparro L, Gaitán-Jurado AJ, Ortiz-Somovilla V, Peña-Rodríguez F. 2013. Feasibility of using NIR spectroscopy to detect herbicide residues in intact olives. Food Control. 30:504–509. Sánchez M-T, De la Haba MJ, Benítez-López M, FernándezNovales J, Garrido-Varo A, Pérez-Marín D. 2012. Nondestructive characterization and quality control of intact strawberries based on NIR spectral data. J Food Eng. 110:102–108. Sánchez M-T, Flores-Rojas K, Guerrero JE, Garrido-Varo A, Pérez-Marín D. 2010. Measurement of pesticide residues in peppers by near-infrared reflectance spectroscopy. Pest Manag Sci. 66:580–586. Saranwong S, Kawano S. 2005. Rapid determination of fungicide contaminated on tomato surfaces using the DESIR-NIR: a system for ppm-order concentration. J Near Infrared Spectrosc. 13:169–175. Subedi PP, Walsh KB. 2011. Assessment of sugar and starch in intact banana and mango fruit by SWNIR spectroscopy. Postharvest Biol Technol. 62:238–245. Tao Y, Dong F, Xu J, Liu X, Cheng Y, Liu N, Chen Z, Zheng Y. 2014. Green and sensitive supercritical fluid chromatographic–tandem mass spectrometric method for the separation and determination of flutriafol enantiomers in vegetables, fruits, and soil. J Agric Food Chem. 62:11457–11464. Teye E, Huang X, Afoakwa N. 2013. Review on the potential use of near infrared spectroscopy (NIRS) for the measurement of chemical residues in food. Am J Food Sci Technol. 1:1–8. Tiwari G, Slaughter DC, Cantwell M. 2013. Nondestructive maturity determination in green tomatoes using a handheld visible and near infrared instrument. Postharvest Biol Technol. 86:221–229. Watanabe E, Kobara Y, Baba K, Eun H. 2015. Determination of seven neonicotinoid insecticides in cucumber and eggplant by water-based extraction and high-performance liquid chromatography. Anal Lett. 48:213–220. Xue L, Cai J, Li J, Liu M. 2012. Application of particle swarm optimization (PSO) algorithm to determine dichlorvos residue on the surface of navel orange with Vis-NIR spectroscopy. Procedia Eng. 29:4124–4128.
Food Additives & Contaminants: Part A Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tfac20
Non-destructive detection of pesticide residues in cucumber using visible/near-infrared spectroscopy a
b
c
d
a
Bahareh Jamshidi , Ezeddin Mohajerani , Jamshid Jamshidi , Saeid Minaei & Ahmad Sharifi a
Agricultural Engineering Research Institute, Agricultural Research Education and Extension Organization (AREEO), Karaj, Iran b
Laser and Plasma Research Institute, Shahid Beheshti University, Tehran, Iran
c
Zarif Mosavar Industrial Manufacturing Company, Isfahan, Iran
d
Agricultural Machinery Engineering Department, Faculty of Agriculture, Tarbiat Modares University, Tehran, Iran Accepted author version posted online: 19 Mar 2015.Published online: 14 Apr 2015.
Click for updates To cite this article: Bahareh Jamshidi, Ezeddin Mohajerani, Jamshid Jamshidi, Saeid Minaei & Ahmad Sharifi (2015) Non-destructive detection of pesticide residues in cucumber using visible/near-infrared spectroscopy, Food Additives & Contaminants: Part A, 32:6, 857-863, DOI: 10.1080/19440049.2015.1031192 To link to this article: http://dx.doi.org/10.1080/19440049.2015.1031192
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Food Additives & Contaminants: Part A, 2015 Vol. 32, No. 6, 857–863, http://dx.doi.org/10.1080/19440049.2015.1031192
Non-destructive detection of pesticide residues in cucumber using visible/near-infrared spectroscopy Bahareh Jamshidia*, Ezeddin Mohajeranib, Jamshid Jamshidic, Saeid Minaeid and Ahmad Sharifia a
Agricultural Engineering Research Institute, Agricultural Research Education and Extension Organization (AREEO), Karaj, Iran; Laser and Plasma Research Institute, Shahid Beheshti University, Tehran, Iran; cZarif Mosavar Industrial Manufacturing Company, Isfahan, Iran; dAgricultural Machinery Engineering Department, Faculty of Agriculture, Tarbiat Modares University, Tehran, Iran b
Downloaded by [New York University] at 10:46 08 May 2015
(Received 7 January 2015; accepted 14 March 2015) The feasibility of using visible/near-infrared (Vis/NIR) spectroscopy was assessed for non-destructive detection of diazinon residues in intact cucumbers. Vis/NIR spectra of diazinon solution and cucumber samples without and with different concentrations of diazinon residue were analysed at the range of 450–1000 nm. Partial least squares-discriminant analysis (PLS-DA) models were developed based on different spectral pre-processing techniques to classify cucumbers with contents of diazinon below and above the MRL as safe and unsafe samples, respectively. The best model was obtained using a first-derivative method with the lowest standard error of cross-validation (SECV = 0.366). Moreover, total percentages of correctly classified samples in calibration and prediction sets were 97.5% and 92.31%, respectively. It was concluded that Vis/NIR spectroscopy could be an appropriate, fast and non-destructive technology for safety control of intact cucumbers by the absence/presence of diazinon residues. Keywords: cucumber; diazinon; maximum residue level; near-infrared spectroscopy; non-destructive; pesticide residues
Introduction Fruit and vegetables are vital for a healthy diet because they contain vitamins and minerals. Pesticides such as fungicides, insecticides and herbicides are chemical treatments that are widely used during the growing stage of such agricultural crops to control plant pests and diseases, as well as to increase productivity. By using pesticides excessively, incorrectly or irrationally during production, ignoring the degradation period of pesticides and early harvesting, transferring to the market and selling fruit and vegetable products immediately or a few days after spraying, the pesticides will remain as chemical residues in these products. In most cases, failure to meet safety requirements when using these chemical treatments may mean that residues in fruit and vegetables exceed the MRLs established by current standards, thus indicating a potential risk to consumer health and safety (Sánchez et al. 2010; Peng et al. 2012). To ensure public health, it is essential to detect and control pesticide contamination of agricultural products (Saranwong & Kawano 2005). Typically, the determination of pesticides in fruit and vegetables involves a sample treatment using different techniques such as SPE, supercritical fluid extraction, microwave-assisted extraction and accelerated solvent extraction. Also, conventional techniques such as GC, HPLC, TLC, supercritical fluid chromatography, GC-MS, capillary electrophoresis, an enzyme inhibition method, an immunoassay method and a biosensor method are *Corresponding author. Email: [email protected] © 2015 Taylor & Francis
used to measure the concentration of pesticide residue (Juan-García et al. 2005; Li et al. 2013; Cho et al. 2014; Masiá et al. 2014; Mercader et al. 2014; Tao et al. 2014; Watanabe et al. 2015). However, these techniques are destructive, have a highly time-consuming sample preparation, are very expensive, and require well-trained personnel and an advanced laboratory for controlling individual products (Sánchez et al. 2010; Teye et al. 2013). Therefore, development of a non-destructive, fast, simple, low-cost, environmentally friendly and reliable detection technique of pesticide residue is imperative. Near-infrared (NIR) spectroscopy is one of the most promising non-destructive techniques that is flexible for both qualitative and quantitative analyses in chemistry, agriculture, medicine and other areas. It is rapid, less expensive than conventional methods and environmentally friendly, requires little or no sample preparation and can be used in processing lines (Sánchez et al. 2010; Xue et al. 2012). While visible/near-infrared (Vis/NIR) and NIR spectroscopy have been widely used for quality assessment and chemical compounds measurement of intact fruit and vegetables (Subedi & Walsh 2011; Jamshidi et al. 2012, 2014; Sánchez et al. 2012; Tiwari et al. 2013; Maniwara et al. 2014; Pan et al. 2015), few publications have addressed the use of this technique for the detection and determination of pesticide residues in such products. One of the first studies with this issue was carried out by Sarawong and Kawano (2005), who used a dry-extract
Downloaded by [New York University] at 10:46 08 May 2015
858
B. Jamshidi et al.
system for infrared analysis (DESIR) technique to detect dichlofluanid fungicide levels on the surface of tomato. Sánchez et al. (2010) assessed the feasibility of reflectance NIR spectroscopy to detect pesticide residues in peppers. Moreover, non-destructive detection of trichlorfon pesticide residues on the surface of longans was studied by Dai et al. (2010) based on a Vis/NIR spectroscopy method. Xue et al. (2012) used this technique for the determination of dichlorvos residues on the surface of navel oranges. Salguero-Chaparro et al. (2013) also assessed the viability of using NIR spectroscopy to detect diuron herbicide levels in intact olives. All these researches confirm the possibility and reasonability of detection of these pesticide residues using NIR spectroscopy. Cucumber is a popular vegetable in people’s diet and is cultivated and grown in most countries. According to the FAO, Iran is third in world ranking in terms of the production quantity of cucumbers (FAO 2012). In Iran, greenhouse production of cucumber is becoming more popular. While cucumber is susceptible to insect attacks, especially in greenhouse conditions, intensive use of pesticides is needed to control insects. Diazinon is a dangerous organophosphorus pesticide that is widely used to control insects and pests in greenhouses (Cengiz et al. 2006). Unfortunately, using diazinon excessively and incorrectly, even in the few hours before harvesting, has caused accumulation of pesticide residues in this product at levels higher than the MRL. This research aims to assess the feasibility of using a Vis/NIR spectroscopy technique combined with partial least squares regression-discriminant analysis (PLS-DA) for non-destructive detection of diazinon residues in cucumber that contain levels higher than the MRL (0.1 mg kg−1) established by the FAO/WHO Codex Alimentarius (FAO/WHO 2013).
length) were also measured by digital balance and Vernier caliper, respectively. Reference measurements were carried out 1 day after spectroscopy analysis. Therefore, cucumbers were frozen to be sent to the laboratory for reference data analysis.
Vis/NIR spectroscopy A USB2000 spectrometer (Oceanoptics Inc., Dunedin, FL, USA) with charge coupled device (CCD) detector and a tungsten halogen light source (LS-1; Oceanoptics) were used for Vis/NIR spectroscopy at a range of 450–1000 nm with 1.5 nm resolution. In order to identify the spectral regions at which diazinon absorbance took place and to determine the absorbance bands of the cucumber spectra that might correspond to diazinon, the pesticide solution was placed in a quartz cell and its transmittance spectrum (T) was acquired using two fibre optics of P400-2-Vis-NIR model (Oceanoptics) (Figure 1a). Then, the collected spectrum was converted to absorbance values (log 1/T). To collect the spectra of intact cucumbers including internal information of the sample, a fibre optic of P400-7-Vis-NIR model (Oceanoptics) was used for spectroscopy in interactance mode (Figure 1b). For each sample, interactance spectra at six positions along the cucumbers and around equatorial locations on the samples (Figure 1c) with five scans at each position were acquired by OOIBase32 software (Oceanoptics). Then, the mean spectrum was
Materials and methods Sample preparation A total of 120 cucumbers were harvested in greenhouses near Karaj city in Alborz province, Iran, and stored at 5°C until use. Some cucumbers with no treatment were used as a diazinon-free set. The commercial pesticide, diazinon (O,O-diethyl-O-(6-methyl-2-(1-methylethyl)-4pyrimidinyl)phosphorothioate), containing 60% diazinon (C12H21N2O3PS), was utilised to obtain a different pesticide residue content in other samples. The pesticide was diluted to 1/500 and sprayed on some of these cucumbers. Moreover, the remained samples were placed in the prepared solution for 1 h. All samples were stored at 5°C until Vis/NIR measurements. Prior to each measurement, they were left to reach equilibrium temperature in a laboratory environment. Morphological properties of each sample including weight and size (maximum diameter and
Figure 1. Spectra acquisition of diazinon solution (a) and intact cucumbers (b) in six positions on the samples (c).
Food Additives & Contaminants: Part A calculated from a total of 30 scans for each cucumber and converted to absorbance values (log 1/R).
Downloaded by [New York University] at 10:46 08 May 2015
Reference measurements After Vis/NIR spectroscopy, all cucumbers were sent to the Chemical Analysis Center at the Iranian Institute of R&D in Chemical Industries for diazinon measurement using a reference method. To this end, sample preparation was done based on British Standard BS EN 15662:2008 (British Standard 2008). Extracted sample (2 µl) was injected into the GC using an Agilent 7890A gas chromatograph (Agilent Technologies Inc., Santa Clara, CA, USA) fitted with a flame ionisation detector. It was equipped with an Agilent Technologies 7683B series injector, a split/splitless injector port operated in splitless mode and with a HP-5 capillary column (30 m × 0.32 mm i.d.), film thickness 0.25 µm. Helium was used as the carrier gas. Throughout the analysis, the injector and detector temperature were kept at 280 and 300°C, respectively. The column temperature was programmed from 50°C (held for 2 min) to 225°C at 10°C min−1 then to 300°C at 30°C min−1 (held for 2 min). The analyte (diazinon) was determined with a linearity range of 0.01–5 mg kg−1. Results were summarised as safe (all samples with or without diazinon concentrations ≤ 0.1 mg kg−1) or unsafe (all samples with diazinon concentrations > 0.1 mg kg−1) groups. Data analysis Before developing the classification models, outliers which are the samples containing interferences (sample contamination, measurement artefacts or temperature effects) having a negative influence on modelling (Heise & Winzen 2006) were detected and removed. To this end, the spectral variation of all the samples was analysed using PCA and the outlier samples were determined as the points outside the normal range of variability in the PCA scores plot (Nicolaï et al. 2007). Figure 2 shows the PCA scores plot for all the samples with the detected outliers. After removing the outliers (14 samples), all the
859
106 remained samples were arranged in order of their diazinon residue measured by the reference method of GC analysis. Then, after every three samples, one sample was selected for the external validation set. Therefore, 26 samples (approximately 25% of the samples), i.e. one out of every four samples in the overall set, were used as the external validation set. The remained 80 samples (approximately 75% of the samples) were selected for the calibration set. Partial least squares-discriminant analysis (PLS-DA) was performed for classification of the cucumbers in two groups defined as safe or unsafe samples. Briefly, PLS-DA is a supervised classification technique based on modelling the differences among several classes using a PLS calibration model to discriminate the unknown samples. In this approach, the PLS model relates the spectral variations (X) to the assigned dummy variables (Y) by maximising the covariance between two types of variable (Cen et al. 2007; Gu et al. 2012). When there are only two classes to discriminate, this model uses one response variable (Y) coding for each class as follows: −1 for membership of one class and +1 for membership of the other class. Each unknown sample takes a predicted value between −1 and +1 after predicting with the developed PLS model. A sample with predicted value close to −1 will belong to one class and if the predicted value is close to +1 the sample is identified as the other class. However, a value close to zero indicates that it is not easy to belong it to any of the classes, especially when the estimated deviation (uncertainty) around the predicted value includes zero. All PLS models were developed with full crossvalidation in the calibration set and investigating the maximum of 10 latent variables (LVs). Before modelling, multiplicative scatter correction (MSC) and standard normal variate (SNV) were used to correct both multiplicative and additive effects of the spectra. To increase the spectral resolution, first and second derivatives of the spectra (D1, D2) based on the Savitzky–Golay algorithm with five smoothing points and polynomial order of 2, were also performed. The precision of the developed PLS-DA models was evaluated based on having the lowest standard error of cross-validation (SECV) and the highest percentage of correctly classified samples for both calibration and external validation sets. All analyses were conducted using Unscrambler software X10.3 (CAMO Software, Oslo, Norway).
Results and discussion Statistics of the samples Figure 2. (colour online) PCA scores plot (PC1 × PC2 × PC3) with the detected outliers.
The statistics of morphological properties and diazinon contents of the samples are shown in Tables 1 and 2, respectively. According to Table 1, the weight, maximum
860
B. Jamshidi et al.
Table 1. Statistics of both calibration and prediction sets for morphological properties of the samples. Calibration set (N = 80)a
Weight (g) Diameter (mm) Length (mm)
Prediction set (N = 26)
Range
Mean
SDb
Range
Mean
SD
37.937–93.081 20.000–36.000 110.500–150.400
59.989 25.484 131.706
13.204 2.932 9.717
41.965–91.108 20.000–30.400 110.500–161.000
60.867 25.419 132.346
12.664 2.324 13.203
Notes: aN, number of the samples. b SD, standard deviation.
Table 2. Statistics of both calibration and prediction sets for diazinon contents in the samples.
Downloaded by [New York University] at 10:46 08 May 2015
Diazinon (mg kg−1)
Diazinon (mg kg−1)
N
Range
Mean
SD
Group
N
Range
Mean
SD
Calibration set
80
0.000–32.000
5.069
7.684
Prediction set
26
0.000–28.000
5.867
8.298
Safe Unsafe Safe Unsafe
16 64 5 21
0.000–0.070 0.650–32.000 0.000–0.100 0.200–28.000
0.008 6.335 0.020 7.260
0.021 8.118 0.450 8.689
equatorial diameter and length of the samples were in the range of 37.94–93.08 g, 20–36 mm and 110.5–161 mm, respectively. Therefore, the cucumbers were quite varied in terms of morphology. As can be seen in Table 2, there was a large variability in the diazinon concentrations of the samples ranging from 0 to 32 mg kg−1 including both safe (MRL ≤ 0.1 mg kg−1) and unsafe (MRL > 0.1 mg kg−1) cucumbers.
Spectra interpretation Figure 3 shows the chemical structure of diazinon, the Vis/ NIR absorbance spectrum (log 1/T) of diazinon solution and its first derivative in the ranges of 450–1000 nm. The curve had a decreasing trend in the visible region. A perceptible peak around 900 nm was found that could be due to the second overtone of O-H or the third overtone of C-H according to the distribution of overtones of organic bonds in NIR region (Cen & He 2007). After that there was an increasing trend up to 1000 nm because of the second overtone of O-H. According to the chemical structure of diazinon, the spectrum trend in NIR region could be more related to C-H absorbance. The first derivative of the spectrum also confirmed the absorbance peaks in NIR region. The absorbance spectra (log 1/R) of intact cucumbers with different pesticide concentrations and their first derivative are also presented in Figure 4. The absorbance in the visible region was mainly related to compounds responsible for cucumber colour. There was a strong peak around 680 nm due to the chlorophyll a content of
Figure 3. Absorbance (log 1/T) spectrum of diazinon solution and its first derivative.
Food Additives & Contaminants: Part A
861
NIR region, the increasing trend of the spectra could be due to the second overtone of O-H or the third overtone of C-H. Therefore, it was noted that diazinon measurements in the NIR region of the spectrum may be more related to C-H absorbance similar to that suggested by Saranwong and Kawano (2005) and Sánchez et al. (2010) for other pesticide residues.
Downloaded by [New York University] at 10:46 08 May 2015
PLS-DA analysis Table 3 shows the results of PLS-DA models developed with different pre-processing methods for discrimination of the cucumbers based on the presence/absence of diazinon residues. All the developed calibration models had an excellent ability to discriminate the samples into two groups of safe and unsafe cucumbers based on the presence of diazinon residues. After cross-validation in the calibration set, the total percentage of correctly classified samples was 97.5% for all models developed using the different pre-processing techniques. In calibration, there were 16 samples of safe group of which only two samples were misclassified. Therefore, the percentage of correctly classified samples of safe group was 87.5%. However, 100% of the samples of unsafe group in the calibration set (64 samples) were correctly classified. While all the models developed using various preprocessing techniques had similar predictive ability as reported by Sánchez et al. (2010), the best results were achieved using first derivative method with the lowest standard error of cross-validation (SECV = 0.366). The predicted versus reference values for this calibration model are presented in Figure 5. This plot shows how close to the ideal values of −1 and 1 (groups of safe and unsafe, respectively) are the predicted values. To investigate how the discriminant model will behave on the unknown samples and predict their class, external validation was performed on the samples of the prediction set not included in model development (26 samples). Results indicated that all PLS-DA models had very good
Figure 4. Absorbance (log 1/R) spectra of intact cucumbers with different pesticide concentrations and their first derivative.
the samples, similar to that described by Kavdir et al. (2007). According to the first derivative of the spectra, more details were also found in the visible region around 650 nm regarding the absorbance of chlorophyll b. In the
Table 3. Results of PLS-DA models for discrimination of safe and unsafe cucumbers based on the presence of diazinon residues. Calibration set
Prediction set
Correctly classified (%)
MSC SNV D1 D2 MSC + D1 MSC + D2 SNV + D1 SNV + D2
Correctly classified (%)
LVs
Total
Safe
Unsafe
SECV
Total
Safe
Unsafe
8 8 7 9 7 4 7 4
97.5 97.5 97.5 97.5 97.5 97.5 97.5 97.5
87.5 87.5 87.5 87.5 87.5 87.5 87.5 87.5
100 100 100 100 100 100 100 100
0.378 0.383 0.366 0.369 0.374 0.396 0.372 0.392
88.46 88.46 92.31 92.31 92.31 88.46 92.31 88.46
60 60 60 60 60 60 60 60
95.24 95.24 100 100 100 95.24 100 95.24
862
B. Jamshidi et al. 2 Unsafe
1.5
Predicted class
1 0.5 0 -1.5
-1
-0.5
0
0.5
1
1.5
-0.5 -1
destructively screening the cucumbers with a content of diazinon above/below the MRL (0.1 mg kg−1). This conclusion was in agreement with the results reported by Sánchez et al. (2010) and Salguero-Chaparro et al. (2013) who assessed the feasibility of reflectance NIR spectroscopy in the range of 1100–1650 nm to detect residues of some pesticide classes in intact peppers and in the range of 400–2500 nm to detect diuron herbicide levels in intact olives, respectively.
-1.5 Safe -2 Reference class
ability to classify the unknown samples. The total percentage of correctly classified samples of the prediction set using the developed models was above 88%. It was also 92.31% for the best calibration model developed based on the first derivative method. While all the samples of unsafe group were 100% correctly classified, the percentage of correctly classified samples of safe group was 60%. Figure 6 shows the predicted values of the unknown samples as the horizontal lines with the estimated uncertainties displayed as the vertical lines for the best calibration model. As it can be seen, all unsafe samples had predicted values close to 1 which assigns them to class ‘unsafe’. For safe group, while two samples were identified as the unsafe group because of having the predicted values close to 1, the remained samples had predicted values close to −1, which assigns them to class ‘safe’. Moreover, no sample had the predicted value close to zero and the deviations around the predicted value did not include zero. Therefore, the quality of the prediction for the unknown samples could be ensured. The results showed the potential of interactance Vis/ NIR spectroscopy in the range of 450–1000 nm for non-
Acknowledgements The authors are grateful to the Agricultural Engineering Research Institute and the Laser & Plasma Institute of Shahid Beheshti University for their suppport, and to the Chemical Analysis Center at the Iranian Institute of R&D in Chemical Industries for cooperation on reference measurements.
Disclosure statement No potential conflict of interest was reported by the authors.
Funding 2.5
This research was funded by the Iran National Science Foundation (INSF) [project number 91051000].
2 1.5 1
References
0.5 0 -0.5
Safe Safe Safe Safe Safe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe Unsafe
Predicted value
Downloaded by [New York University] at 10:46 08 May 2015
Figure 5. Predicted versus reference values plot obtained using the calibration model developed based on a first-derivative preprocessing method.
Conclusions The feasibility of utilising Vis/NIR spectroscopy combined with PLS-DA was investigated for nondestructive detection of diazinon residues in intact cucumbers that contain levels below/above the MRL. The results confirmed the ability of this technology for discrimination of intact cucumbers by the presence/ absence of diazinon residues. Therefore, Vis/NIR spectroscopy could be used for fast, low-cost and non-destructive preliminary safety and healthy control of the cucumbers before or after harvesting. Nevertheless, further work should be considered to adapt the Vis/NIR spectroscopy to detect diazinon residues in other varieties of cucumber at different regions which have been contaminated directly by pesticide spraying in greenhouse and field.
-1 -1.5
Samples
Figure 6. Predicted values with estimated deviation for the unknown samples using the best calibration model.
British Standard (BS EN 15662). 2008. Foods of plant origin – Determination of pesticide residues using GC-MS and/or LC-MS/MS following acetonitrile extraction/partitioning and clean-up by dispersive SPE – QuEChERS-method. Cen H, He Y. 2007. Theory and application of near infrared reflectance spectroscopy in determination of food quality. Trends Food Sci Technol. 18:72–83. Cen H, He Y, Huang M. 2007. Combination and comparison of multivariate analysis for the identification of orange varieties using visible and near infrared reflectance spectroscopy. Eur Food Res Technol. 225:699–705.
Downloaded by [New York University] at 10:46 08 May 2015
Food Additives & Contaminants: Part A Cengiz M, Certel M, Gocmen H. 2006. Residue contents of DDVP (Dichlorvos) and diazinon applied on cucumbers grown in greenhouses and their reduction by duration of a pre-harvest interval and post-harvest culinary applications. Food Chem. 98:127–135. Cho S-K, Abd El-Aty AM, Musfiqur Rahman M, Choi J-H, Shim J-H. 2014. Simultaneous multi-determination and transfer of eight pesticide residues from green tea leaves to infusion using gas chromatography. Food Chem. 165:532–539. Dai F, Hong T, Zhang K, Hong Y. 2010. Nondestructive detection of pesticide residue on longan surface based on near infrared spectroscopy. Int Conf Intell Comput Technol Autom (ICICTA). 2:781–783. [FAO] Food and Agriculture Organization of the United Nations. 2012. FAOSTAT Database. Available from: http://faostat3. fao.org/download/Q/QC/E. [FAO/WHO] Food and Agriculture Organization of the United Nations and World Health Organization. 2013. Codex Pesticides Residues in Food Online Database. Available from: http://www.codexalimentarius.net/pestres/data/index.html. Gu C, Xiang B, Xu J. 2012. Direct detection of phoxim in water by two-dimensional correlation near-infrared spectroscopy combined with partial least squares discriminant analysis. Spectrochim Acta, Part A. 97:594–599. Heise HM, Winzen R. 2006. Chemometrics in near-infrared spectroscopy. In: Siesler HW, Ozaki Y, Kawata S, Heise HM, editors. Near-infrared spectroscopy: principles, instruments, applications. Germany: Wiley-VCH; p. 125–162. Jamshidi B, Minaei S, Mohajerani E, Ghassemian H. 2012. Reflectance Vis/NIR spectroscopy for nondestructive taste characterization of Valencia oranges. Comput Electron Agric. 85:64–69. Jamshidi B, Minaei S, Mohajerani E, Ghassemian H. 2014. Prediction of soluble solids in oranges using visible/nearinfrared spectroscopy: effect of peel. Int J Food Prop. 17:1460–1468. Juan-García A, Picó Y, Font G. 2005. Capillary electrophoresis for analyzing pesticides in fruits and vegetables using solidphase extraction and stir-bar sorptive extraction. J Chromatogr A. 1073:229–236. Kavdir I, Lu R, Ariana D, Ngouajio M. 2007. Visible and nearinfrared spectroscopy for nondestructive quality assessment of pickling cucumbers. Postharvest Biol Technol. 44:165–174. Li W, Sun M, Li M. 2013. A survey of determination for organophosphorus pesticide residue in agricultural products. Adv J Food Sci Technol. 5:381–386. Maniwara P, Nakano K, Boonyakiat D, Ohashi S, Hiroi M, Tohyama T. 2014. The use of visible and near infrared spectroscopy for evaluating passion fruit postharvest quality. J Food Eng. 143:33–43. Masiá A, Blasco C, Picó Y. 2014. Last trends in pesticide residue determination by liquid chromatography–mass spectrometry. Trends Environ Anal Chem. 2:11–24. Mercader JV, López-Moreno R, Esteve-Turrillas FA, Abad-Somovilla A, Abad-Fuentes A. 2014. Immunoassays
863
for trifloxystrobin analysis. Part II. Assay development and application to residue determination in food. Food Chem. 162:41–46. Nicolaï BM, Beullens K, Bobelyn E, Peirs A, Saeys W, Theron KI, Lammertyn J. 2007. Nondestructive measurement of fruit and vegetable quality by means of NIR spectroscopy: a review. Postharvest Biol Technol. 46:99–118. Pan L, Zhu Q, Lu R, Mitchell McGrath J. 2015. Determination of sucrose content in sugar beet by portable visible and nearinfrared spectroscopy. Food Chem. 167:264–271. Peng Y, Li Y, Chen J. 2012. Optical technologies for determination of pesticide residue. In: Theophile T, editor. Infrared spectroscopy - materials science, engineering and technology. Rijeka: InTech; p. 453–466. Salguero-Chaparro L, Gaitán-Jurado AJ, Ortiz-Somovilla V, Peña-Rodríguez F. 2013. Feasibility of using NIR spectroscopy to detect herbicide residues in intact olives. Food Control. 30:504–509. Sánchez M-T, De la Haba MJ, Benítez-López M, FernándezNovales J, Garrido-Varo A, Pérez-Marín D. 2012. Nondestructive characterization and quality control of intact strawberries based on NIR spectral data. J Food Eng. 110:102–108. Sánchez M-T, Flores-Rojas K, Guerrero JE, Garrido-Varo A, Pérez-Marín D. 2010. Measurement of pesticide residues in peppers by near-infrared reflectance spectroscopy. Pest Manag Sci. 66:580–586. Saranwong S, Kawano S. 2005. Rapid determination of fungicide contaminated on tomato surfaces using the DESIR-NIR: a system for ppm-order concentration. J Near Infrared Spectrosc. 13:169–175. Subedi PP, Walsh KB. 2011. Assessment of sugar and starch in intact banana and mango fruit by SWNIR spectroscopy. Postharvest Biol Technol. 62:238–245. Tao Y, Dong F, Xu J, Liu X, Cheng Y, Liu N, Chen Z, Zheng Y. 2014. Green and sensitive supercritical fluid chromatographic–tandem mass spectrometric method for the separation and determination of flutriafol enantiomers in vegetables, fruits, and soil. J Agric Food Chem. 62:11457–11464. Teye E, Huang X, Afoakwa N. 2013. Review on the potential use of near infrared spectroscopy (NIRS) for the measurement of chemical residues in food. Am J Food Sci Technol. 1:1–8. Tiwari G, Slaughter DC, Cantwell M. 2013. Nondestructive maturity determination in green tomatoes using a handheld visible and near infrared instrument. Postharvest Biol Technol. 86:221–229. Watanabe E, Kobara Y, Baba K, Eun H. 2015. Determination of seven neonicotinoid insecticides in cucumber and eggplant by water-based extraction and high-performance liquid chromatography. Anal Lett. 48:213–220. Xue L, Cai J, Li J, Liu M. 2012. Application of particle swarm optimization (PSO) algorithm to determine dichlorvos residue on the surface of navel orange with Vis-NIR spectroscopy. Procedia Eng. 29:4124–4128.
