Assessment of pesticide residues in vegetables produced in central and eastern Ethiopia

Background In Ethiopia, pesticides are widely used in vegetable production. However, if used incorrectly they may harm consumers of vegetables contaminated with pesticide residues, as well as producers handling the pesticides and lead to ecological damage. We performed a cross-sectional survey to assess pesticide residues in vegetables produced in central and eastern Ethiopia. Methods A total of 232 vegetable samples (91 tomatoes, 106 cabbages and 35 Swiss chard) were collected from fields and retail markets, and were screened for 35 pesticides (16 organochlorine, 11 organophosphate, 3 pyrethroids, 2 carbamates and 3 other agrochemicals) using GC–MS analysis. Results Pesticides residues were detected in 60% of Swiss chard, 47% of cabbage and 45% of tomato samples. Two or more pesticides were detected in 20% of cabbages, 13% of tomatoes and over half of Swiss chard samples. Bendiocarb, diazinon, endrin, piperonyl butoxide, profenofos and propargite were detected, but only diazinon, propargite and profenofos had residual values above EU Maximum Residue Level (MRL), with diazinon commonly detected at relatively high levels. About 15% of the total analyzed samples, 10% of cabbages and tomatoes, and nearly half of Swiss chard samples had pesticide concentration exceeding EU MRL. However, none of the tested samples had residues exceeding Codex Alimentarius Commission (Codex) MRLs. Conclusion This study showed widespread contamination of vegetables with pesticides, mainly organophosphates. We recommend monitoring and regulation of pesticides usage, with promotion of good agricultural practices.


Introduction
Pesticides improve crop production by protecting from pests (Tarannum et al., 2019), but their inappropriate use creates health risks for vegetable producers and consumers, and impact on the wider ecology and environment (Sharma et al., 2012;Negatu et al., 2016).
Unsafe use of pesticides in agriculture increases the presence of their residues in produce such as vegetables after harvest (Kapeleka et al., 2020). Intake of foods contaminated with pesticide residues exposes consumers to pesticides with potential long-term health risks (Darko and Akoto, 2008;Ferré et al., 2018;Nougadère et al., 2020). Farm workers may also be exposed to pesticides during application on farm; as may local residents exposed through inhalation of pesticide drift and/or volatilization from the environment (Hanssen et al., 2015;Teysseire et al., 2020;Polledri et al., 2021).
Studies have demonstrated that exposure to pesticides such as organophosphates and organochlorines can result in increased risk of cancers (De Roos et al., 2003), disruption of circulating hormones, reproductive problems (Meeker et al., 2008) and neurological disorders (Eskenazi et al., 2007). In addition to affecting human health, pesticides can have negative impacts on the environment including, but not limited to, non-point source pollution (Mekonen et al., 2016;Teklu et al., 2018), loss of pollinators and biodiversity (Fikadu, 2020), and losses to birds and aquatic wildlife (Yohannes et al., 2014).
Good Agricultural Practices (GAP) and pesticide usage regulations have been put in place with MRLs for pesticide levels in foods specified in national and international standards (WHO, FAO, 2010). MRLs are set to avoid potentially toxic high pesticide exposures while still allowing sufficiently high exposures needed for effective pest control. The EU defines sampling strategies and MRLs for relevant foods sold within the EU with removal from market and potentially prosecution if these MRLs are exceeded. The Codex Alimentarius Commission (Codex) also publishes MRLs that are expected to be applied as minimum requirements for foods traded internationally. In Ethiopia, pesticide MRLs have not yet been established for vegetables but the country uses the Codex MRLs as a reference (WHO, FAO, 2010).
Monitoring of pesticide residue concentrations in vegetables should be performed routinely to assess MRL compliance and protect consumers. Demonstrable effective monitoring can also help build trust between exporting and importing countries by showing oversight and control of food safety (Mahugija et al., 2017;Ali et al., 2020;Omwenga et al., 2021).
The use of chemical pesticides in agriculture is rapidly increasing in Ethiopia (Negatu et al., 2016), but there has been inadequate awareness and enforcement of good practices when using pesticides (Mengistie, 2016), and misuse of pesticides is widespread in Ethiopia (Negatu et al., 2016(Negatu et al., , 2021. A report generated from a larger project, which encompass this pesticide assessment as one of its activities, assessing vegetable and chicken food safety in Harar and Dire Dawa, eastern Ethiopia 1 found high levels of consumer concern over contamination of foods with agricultural chemicals such as pesticides. In addition, experts reported anecdotes of inappropriate pesticide usage by farmers, repeated application of pesticides close to harvest (Amenu et al., 2021). Although these evidence shows poor GAP use of pesticides implying their higher presence on produce, there is little information on pesticide residue levels in vegetables in Ethiopia. To partially address this gap, we performed a cross-sectional screening survey to assess levels of pesticide residues in selected vegetables from fields and retail markets in central and eastern Ethiopia.
1 Pull-Push Project: urban food markets in Africa-incentivizing food safety using a pull-push approach.

Study areas
Vegetable samples were collected from markets, fields (farmers' vegetable farms) and small-scale vegetable producers between October 2020 and January 2021 at fields in the central Rift-Valley of Ethiopia and the Akaki Kality areas of Addis Ababa. Samples from Harar and Addis Ababa retail markets were also collected ( Figure 1).

Sampling and data collection 2.2.1. Sample collection in fields
A total of 113 vegetable samples [31 cabbages (Brassica oleracea var. capitata), 55 tomatoes (Solanum lycopersicum) and 27 Swiss chard (Beta vulgaris)] were collected from farmers' fields in major vegetable producing areas of East Shewa zone in Oromia and from small-scale producers' vegetable plots within Addis Ababa (Akaki Kality sub-city) during October-November 2020. Experts identified these vegetables as high risk because of the common use of pesticides in these vegetables, in addition tomatoes are a focus of the project within which this survey was performed (Amenu et al., 2021). Fields growing these vegetables were randomly selected and samples were collected from different corners of a particular field. Depending on the size of the field, two to four samples were collected, packed, labeled, and placed in a container in shade. Within a field, samples from up to five individual plants were collected to create a 1 kg sample. Samples were kept cold until they reached the laboratory.

Sample collection in food markets
A total of 119 vegetable samples were collected from food markets (75 cabbages, 36 tomatoes and 8 Swiss chard) between 28 December 2020 and 22 January 2021. Samples from five major vegetable markets (both wholesale and retail) were collected in Harar city (market names: Dakar, Shewa ber, Shankor, Arategna, and Bate). The number of retail and wholesale traders sampled from each market was calculated with probability proportional to size based on a recent census of vegetable outlets (unpublished project activity), sampling a fixed proportion within each market. The main market in the nearby town of Haramaya was also sampled systematically selecting stalls evenly spread-out across the market, as were major retail markets in Bole and Yeka sub-cities of Addis Ababa (market names: Yerer, Goro, Gurd shola, Summit, Meri, Hayat, and Kotebe 02).
Within each market, the first encountered vegetable stall on entering the market was sampled and then every Kth vegetable stalls encountered during a walk covering the market were sampled; where K was the number of vegetable stalls in the market divided by the number of predefined vegetable samples required from that market. At a stall, a 1 kg sample of firm but mature fresh vegetables were collected and bagged. The stall-keeper selected the vegetables as for a normal customer and was not aware that the samples were collected as part of a study. The vegetable samples were kept cool until transported to the laboratory (Bless Agri Food Laboratory Services, Legetafo Legedadi, Addis Ababa, Ethiopia) within 0-4 days.
All reagents and solvents were of analytical grade including acetonitrile, acetic acid, magnesium sulfate, sodium acetate and primary secondary amine (PSA) which were used for extraction and clean-up of the samples.

Sample preparation
Vegetable sample extraction and clean-up was performed using the QuEChERS (quick, easy, cheap, effective, rugged, and safe) method as indicated in the AOAC Official Method 2007.01 with slight modifications (Association of Official Analytical Chemists International, 2007). Briefly, a 1 kg of each vegetable sample was thoroughly chopped and homogenized using a high-speed multifunction comminutor. Fifteen grams (15 g) of the homogenized vegetable sample were weighed into a 50 mL Teflon tube. Then, 15 mL of 1% acetic acid in acetonitrile was added and the tube was sealed and vigorously shaken by hand for 1 min. Following this, a 1.5 g anhydrous sodium acetate and 6 g anhydrous MgSO 4 were added into the Teflon tube and vortexed for 1 min and then centrifuged at 4,000 rpm for 10 min. After this, 4 mL aliquot of the upper layer extract was transferred to a 10 mL Teflon tube containing 200 mg primary secondary amine (PSA) and 150 mg anhydrous MgSO 4 . The tubes were sealed and vortexed for 1 min and then centrifuged at 4,000 rpm for 10 min. Finally, a 1.5 mL aliquot of the cleaned extracts was transferred into GC vials and injected into Gas Chromatography-Mass Spectrometry (GC-MS) system for pesticide analysis.

Method validation
Pesticide standard stock solutions were prepared in acetonitrile at 2,000 mg/L and stored in the dark at −20°C. Standard working solutions were prepared by dissolving appropriate amounts of stock solution with a mixture of acetone (9:1, v/v). The obtained standard working solutions were used to plot calibration curves as a function of peak area vs. concentrations of the selected pesticides. The concentration range was from 5 to 50 μg/kg for organochlorines, while that of organophosphates, pyrethroids, carbamates and the other agrochemicals was from 2 to 40 μg/kg. Our analytical method was validated using linearity (expressed Frontiers in Sustainable Food Systems 04 frontiersin.org as correlation coefficient), precision (expressed as repeatability relative standard deviation), mean recovery/reliability (as a measure of trueness) and sensitivity (expressed as slope of the regression equation). Moreover, the Limit of Detection (LOD) and Limit of Quantification (LOQ) of each screened pesticide were evaluated using 3S/m and 10S/m, respectively (where S is the standard deviation of the intercept and m is the slope of the regression line). Representative samples from each commodity group (tomatoes, cabbages and Swiss chard) were spiked with known concentrations of each standard dilution of pesticide and recovery values were determined. A non-spiked sample was also analyzed and used as a control.

GC-MS analysis
This study was performed on an Agilent 7890B GC equipped with G4513A auto-sampler coupled to a 7000C GC/MS Triple Quadrupole mass detector system (Agilent Technologies, Inc., CA, United States). An Agilent J&W DB-5 ms Ultra Inert capillary GC column with 30 m length, 0.25 mm internal diameter, and 0.25 μm film thickness was used to provide analyte separation and a highly inert flow path into the detector. The carrier gas was helium (99.999%) at a column flow rate of 1.2 mL/ min. The interface and the injector were programmed at 280°C for splitless injection of 1 μL. The eluent from the GC column was transferred through a transfer line at a temperature of 280°C and fed into a 70 eV electron impact ionization source at a source temperature of 280°C. The analysis was done in the selected ion monitoring mode.

Statistical analysis
The pesticide residues (μg/kg) of samples were compared with MRLs from Codex Alimentarius Commission standards (Codex MRLs) and European Commission legislation and standards (EU MRLs;Codex, 2021;European Commission, 2021). For a specific pesticide, its concentration in a given vegetable sample was compared with its MRL in that vegetable in EU or Codex pesticide database but a default MRL of 10 μg/kg (0.01 mg/kg) for EU MRL and 2,000 μg/kg (2 mg/kg) for Codex MRL was used when a pesticide is not specifically mentioned (Codex, 2021;European Commission, 2021).
Pesticide residues concentration was log-transformed using log10 1 x + ( ) to improve the normality of our data distribution. The distribution of pesticide residues was described, stratifying by sample location (market, field) and vegetable type, with associations tested using Chi-square tests and One-way ANOVA. Analysis was done using R (version 4.1.0; R Core Team, 2021) and maps produced in ArcGIS (version 10.2, ESRI, California, United States; ESRI, 2020).

Method validation
Our results showed excellent linearity and reproducibility indicating good fitness of the model with R 2 ≥ 0.99 for all analytes. The good reliability of our method was expressed by recovery values obtained to be in the range from 71% to 112%, which are better comparing to the acceptable limits for individual recovery results that should normally be within the range of 60%-140%. The relative standard deviations showed a good repeatability and precision of the method, was under 5%, which is in the recommended range, RSD ≤ 20%. The LOD and LOQ values of the tested pesticides were very low in the order of picogram/kg. The obtained values of LOD and LOQ in this study were below the Codex MRLs (Table 1).

Discussion
The prevalence of pesticide contamination of vegetables in our study is comparable with pesticide residue prevalence reported in vegetables elsewhere in Sub-Saharan Africa: 46% in Tanzania (Kapeleka et al., 2020) and 54% in Kenya (Nakhungu et al., 2017), with higher values sometimes reported: 63% in Zambia (Mwanja et al., 2017), 92% in Sudan (Ali et al., 2020) and 96% in Tanzania (Mahugija et al., 2017). However, in Ethiopia, pesticide residues in vegetables has seldom been assessed (Loha et al., 2020), with a few studies assessing the residues in   Bendiocarb (  other commodities such as khat (Daba et al., 2011;Atnafie et al., 2021), wheat (Daba et al., 2011) and tea (Siraj et al., 2021). EU MRL exceedance of pesticide residues concentration in vegetables observed in this study was relatively lower compared with findings in other developing countries where 21%-28% were reported (Lozowicka et al., 2016;Jallow et al., 2017;Ramadan et al., 2020). Considering inadequate pesticide legislation enforcement in Ethiopia, higher pesticide concentration is anticipated in the vegetables than the current findings, though, various factors could contribute to pesticide concentration levels in vegetables. Fields samples are expected to have higher residue levels than market samples as pesticides evaporate and degrade after application, also traders may wash the vegetables especially cabbages and Swiss chard to keep them fresh, which could reduce pesticide concentration (Inonda et al., 2015;El-Saeid and Selim, 2016). Our current results also showed slightly higher pesticide residues concentration in fields samples than market samples, though, the evidence for the variation between the sample types was weak. In addition, in this study, more than half of the samples were market samples which might contribute to reduced concentration levels. Pesticide concentration in vegetables can also vary with seasons of study (Inonda et al., 2015), with our samples collected at the post-harvest stage which might contribute to reduced concentration as frequency of pesticide application is greatly reduced at this stage. Crop types can affect the levels of pesticide volatilization-a process by which pesticides Violin and boxplots showing comparison of pesticide residue levels (μg/kg) on log(x + 1) scale in field and market vegetable samples for diazinon, propargite, profenofos, endrin, bendiocarb, and piperonyl butoxide. Blue points show mean concentration of pesticide residues in vegetables and black points show outliers. Horizontal dashed and solid lines indicate EU and Codex MRLs (μg/kg) on log(x + 1) scale, respectively. NB, we used a general default MRL of 10 μg/kg for EU MRL and 2,000 μg/kg for Codex MRL where a pesticide is not specifically mentioned. Number of pesticides found in a single sample of cabbage (n = 106), tomato (n = 91) and Swiss chard (n = 35) and percentage of the samples with the detected number of pesticides.
Frontiers in Sustainable Food Systems 09 frontiersin.org dissipate-determining pesticide residue concentration in them (Gamalero et al., 2003;Inonda et al., 2015). Consistent with this, we found in our analysis that pesticide residue concentration varied among vegetable types. Lack of adherence to pre-harvest interval between application and harvest is also a common cause of high pesticide residues (Osei et al., 2015). Farmers' knowledge level and practices in the safe use of pesticides can further influence pesticide concentration levels in vegetables (Horna, 2008). Foods with residue levels below MRLs can be considered safe for consumers (WHO, FAO, 2010) but MRL is not a safety limit as foods with residues above MRL may still be safe for consumption (Keikotlhaile and Spanoghe, 2011). However, in addition to health and ecological impacts, MRL exceedance can restrict access to international export markets (Horna, 2008) with governing authorities required to monitor and enforce MRL compliance (WHO, FAO, 2010). Reflecting cultural differences in risk perception, the European Union has set more stringent MRLs (typically 0.01 mg/kg) than the Codex Alimentarius Commission (typically 2.0 mg/kg; Codex, 2021; European Commission, 2021; Food Business Africa, 2021), reflecting the need for exporting countries to establish and enforce national pesticide regulations to meet these standards.
Farmers in Ethiopia have poor knowledge about the safe use of pesticides, contributing to MRL exceedance (Gesesew et al., 2016;Negatu et al., 2016;Mergia et al., 2021). Although GAP allows productive farming with lower amounts of agrochemicals (Kılıç et al., 2020) contributing to reduced pesticide residues in produce (Inonda et al., 2015), it is rarely practiced in the country's agricultural activities. Ethiopia currently exports vegetables to only non-European countries. One of the constraints for EU market access is an inability to prove that pesticide use is compliant with GAP and that the MRLs are within the required limits (CBI, 2020).
This study found widespread presence of pesticide residues on vegetables with nearly half of the vegetable samples contaminated with one or more types of pesticide, mostly organophosphates. We also detected multiple pesticide residues in a single vegetable sample. Pesticide concentration varied between vegetables and, field and market samples representing the need to adhere to specific GAP while using pesticides for different vegetable types. About 15% of vegetable samples had pesticide residue concentrations above EU MRL but none were above Codex MRLs. But it is unlikely that we capture the margins of the distribution of pesticide contamination and rare, higher exposures are anticipated. Nonetheless, the study results still raise food safety and economic risk concerns needing improved monitoring and regulation of agricultural pesticides, with improved farmer education on the safe use of pesticides and the risks they pose to producers, consumers, and the environment.

Ethics statement
Ethics approval was granted for this study by ILRI's Institutional Research Ethics Committee (approval number: ILRI-IREC2019-36/1; ILRI, 2021). While collecting vegetable samples from retailer markets, we did not collect information on vegetable stalls that could identify them and raise privacy issues. We obtained oral consent from vegetable farmers to collect vegetable samples from their farms, analyzing the survey data anonymously.

Author contributions
TJDK designed, supervised and managed the project and received funding for the study. LG, SG, and WB collected the study data. GD curated the data and created data visualization. GD and SMF performed formal data analysis. GD wrote original draft of the manuscript. TJDK, WB, MFS, RS, RR, DG, and HG reviewed and edited the manuscript. All authors contributed to the article and approved the submitted version. License has already been assigned to the Author Accepted Manuscript version that might arise from this submission. The work was also supported by the German Federal Ministry for Economic Cooperation and Development (BMZ) through the One Health Research, Education and Outreach Centre in Africa (OHRECA) led by ILRI. Additional support was received from CGIAR Research Program on Agriculture for Nutrition and Health (A4NH). Support was also provided by United Kingdom aid from the UK Foreign, Commonwealth and Development Office.