1 Introduction

Cotton production systems are often subjected to several and diverse attacks from pests. Thus, more than 1,300 species of pests are known to affect cotton plants [1, 2]. The most fearsome pests are biting pests (leaves worms, grasshoppers, crickets and caterpillars) and sucking pests (bugs, flies, ants and aphids). Cotton yield losses from pest activities can reach 80% of the overall production [3]. In response to severe damages, the use of pesticides was recommended a few decades ago to fight pests and to improve agronomic performances of cotton trees and especially yields of cotton [4]. However, the overuse and misuse of synthetic pesticides by cotton producers could potentially harm environment and human health.

There is a growing focus on the potential harm synthetic pesticides can inflict on the environment and its various components [5, 6]. Consequently, research findings advocate for a shift back to organic farming, despite its somewhat lower productivity [7, 8].

Benin is one of the largest cotton production countries in Africa. With more than 600,000 tons of cotton yields for the 2017–2018 cotton season, Benin was the third highest cotton producer in Africa behind Mali and Burkina Faso. For the cotton season 2019–2020, Benin is the first cotton producer country in Africa. Cotton sector employs over two-thirds of the rural population and accounts for over 60% of the country's export resources [5]. However, previous research highlighted that cotton production systems are dominated by conventional agricultural practices characterized by unsustainable use of fertilisers and pesticides which are still maintained farmers in poverty and hindering the country economic development [6].

Conventional cotton production systems in Benin are characterized by the use of several tons of synthetic pesticides each year to ensure a given productivity [7]. Given the downward trend in yields, despite the use of pesticides, there is an increase of cotton production lands throughout the country. The overall area cultivated increased from 418,947 hectares during the 2016/2017 season to 530145 hectares in the 2017/2018 season [8]. Despite this increase in cotton production and subsequent use of pesticides, only few studies [9, 10] have addressed cotton pesticide residues induced contamination on soils and subsequent assessment of their impacts on the environment and human health. Comparative analysis of the level of soils contamination by pesticides residues between conventional and organic cotton productions and subsequent implications for soils fertility status is still undocumented.

The main objective of this study was to analyse the effect of long-term pesticide spraying along Benin cotton production system lands in west Africa. Especially, this study aimed to (i) assess soil physico-chemical properties of Benin cotton production systems and (ii) to quantify pesticide residue levels throughout cotton production systems in Benin. Research hypotheses are (i) the physico-chemical properties of soils under conventional cotton production systems are more degraded than soil physico-chemical properties of organic cotton production systems and control field and (ii) soils under conventional cotton production systems presents high contamination levels by pesticides residues than organic cotton production systems and control field.

2 Material and methods

2.1 Study area

This study was carried out in Benin’s main cotton production sites located in the northern and central part of the country. Study site covers six regions with a total area of 14642 km2 including districts of Malanville (11°52′00''N and 3° 23′00''E), Kandi (11°07′43''N and 2°56′13''E), Banikoara (11°18′00''N and 2°26′00''E), Pehunco (10°13′42''N and 2°00′07''E), Glazoue (7°58′25''N and 2°14′24''E) and Djidja (7°20′40''N and 1°56′00''E). The mean temperature ranges between 24.9 °C and 32.5 °C and rainfall between 700 and 1400 mm [11]. These regions account for more than half of Benin total annual cotton production. Conventional cotton production systems are widely adopted by the farmers in these areas. However, organic cotton production systems are also observed. Banikoara district is both the first cotton production district and the largest user of synthetic pesticides in Benin. In these regions, cotton is the main source of income for the population.

2.2 Experimental design and soil sampling and samples preparation

Soil samples were collected in the field before cotton harvesting, either 30 days after last pesticides spraying. In each district, soil samples were taken at 0–20 cm of depth in six cotton conventional systems, three organic cotton systems and in a natural forest except in Glazoue where organic cotton systems were marginal. A minimum distance of 1 km was observed between two fields. In each cotton production system and natural forest used as control, ten soil samples were collected at different level of soil depth and sands were mixed to form a composite soil sample. Two composite samples were taken per cotton production system. Overall, 114 soil samples were taken in six the major cotton regions of Benin (Malanville, Angaradebou, Pehunco, Bannikoara, Glazoue and Djidja) and 14 active ingredients (abamectin, acetamiprid, bifenthrin, cyfluthrin, cypermethrin, deltamethrin, emamectin benzoate, imdacloprid, indoxacarb, lambdacyhalothrin, novaluron, profenofos, teflubenzuron and triasophos) belonging to 6 different groups of pesticides (Neonicotinoid, pyrethroid, organophosphate, benzoylurea, oxadiazine and micro-organism derived) authorized in Benin were quantified.

After removing roots and other fragments, each composite sample was dried at room temperature (25 °C), sieved and stored at 4 °C. One part was used for soil physico-chemical characteristics assessment and the other parts for pesticide residues quantification. Composite samples from natural forests were used as a basis for comparison between fields and as blank samples for extraction and chromatographic methods development. From each composite sample, 5 g of soil was taken and put into a 50 ml falcon tube to initiate extraction procedures. Recoveries at two different levels of fortifications (10 μg / kg and 100 μg / kg) of pesticide concentration in blank soil were prepared by adding to 12 composite soil samples, 50 μl of a mixed solution of pesticides at concentrations of 1 μg / ml and 10 μg / ml, respectively for recovery experiments.

2.3 Chemicals and reagents

Standard pesticides compounds: abamectin, triasophos, bifenthrin, cypermethrin, lambdacyhalothrin, novaluron, deltamethrin, emamectin benzoate and imidacloprid were purchased from Sigma-Aldrich and acetamiprid, cyfluthrin, indoxacarb, profenofos and teflubenzuron from Dr Ehrenstorfer. Purity of compounds ranges from 98% to 99.99%. Deionized water was prepared using Milli-Q advantage A10 ultrapure water purification system. Acetonitrile and methanol were purchased from Ratburn Mikro Lab (Denmark), isopropanol and ammonium acetate from Fisher Chemical (Denmark), magnesium sulfate (MgSO4) and sodium chloride (NaCl) from VWR (Denmark). QuEChERS kits containing 150 mg of magnesium sulfate (MgSO4) and 25 mg of primary and secondary amine (PSA) were purchased from Phenomenex (Denmark).

2.4 Stock solutions

Stock solutions were prepared by accurately weighing and dissolving each pure compound in acetonitrile at 1000 mg / ml. Stock solutions were stored at −20 °C and were used to prepare a mixture of all compounds at 10 mg / ml for calibration standards by diluting in methanol and water (40: 60 v/v) and in soil blank matrix and water (50: 50 v/v). Concentrations of multistandard working solutions were 800, 400, 200, 100, 50, 25, 12.5, 6.25, 3.125, 1.56, 0.78 and 0.39 ng/ml.

2.5 Soil samples characterization

Four soil parameters were investigated basing on their importance in processes of pesticides mineralization. There are contents of clay, carbon content, organic matter and pH. The potentiometric method was used for pH determination, Walkley and Black method for total carbon and the international modified method using Robinson pipette was used for the granulometry [12]. The determination of organic matter content was carried out indirectly based on the determination of organic carbon content according to the international standard method [13]. Organic matter (OM) content is commonly calculated by multiplying the carbon content value by standard coefficient ranging between 1.72 and 2 for farming lands’ soils [14]. Conversion factor 2 was used in this study. Loss upon ignition at 375 °C tested for 10 samples showed that the organic matter content of soils in the study areas were greater than twice of carbon contents.

2.6 Sample extraction and clean-up

Homogenized soil (5g) was weighed into 50 ml flask. Milli Q water 5 ml was added and vortex shaken at 1300 rpm for 30 s. 10 ml acetonitrile was then added and the mixture was vortex shaken for 1 min and for 3 min in an ultrasonic bath at room temperature. Salting out extraction procedure were performed by adding a mixture of anhydrous salts NaCl (1g), MgSO4 (4g) and immediately and vigorously hand-shaking for 2 min. The resulting mixture was centrifuged at 4000 rpm for 5 min and 1 ml of the supernatant were transferred into 2 ml QuEChERS centrifuge tubes containing 150 mg of MgSO4 and 25 mg of PSA for the clean-up. The mixture was vortex shaken at 1300 rpm for 1 min and centrifuged at 4000 rpm for 3 min. The supernatant was diluted in water 1:1 (v/v) and filtered through a 0.45 µm PTFE syringe filter before injection in the LC–MS/MS system.

2.7 Chromatographic and mass spectrometer conditions

The LC–MS/MS chromatographic method was used for the separation, identification and quantification of pesticide compounds. Agilent HPLC 1260 infinity was used for analyte separation. The column was a Hypersil BDS C18 column (250 mm × 2.1, 5 µm particle size) supplied by Thermo Scientific (Denmark). Mobile phase A was composed of 1% isopropanol, 9% methanol 10 mM ammonium acetate and mobile phase B of 20% isopropanol, 80% methanol 10 mM ammonium acetate. The elution gradient started with 00% B, after one minute increased to 70% and then to 100% in 21 min was held constant for 5 min and decreased to 00% in one minute and then equilibrate for 11 min. The gradient was run at 200 µl/min and at 25 µl as injection volume. A 3200 QTRAP mass spectrometer was connected to the LC-system. The mass spectrometer was run in positive mode at 50 °C with ESI parameters set as follow: ion spray voltage 4500 V, source temperature 475 °C, curtain gas 20 psi. ESI–MS/MS was activated in scheduled multiple reaction monitoring mode (MRM) in positive polarity by scanning one precursor/product ion transition for each target compound. However, for the compounds with low differences in retention times, two precursor/product ion transitions were scanned but only the most intense was used for the quantification and the second as a qualifier (qualification and confirmation). Table 1 shows target compounds as well as optimized values of dependent parameters. Analyst software version 1.6.2. and Excel software were used for data acquisition and processing.

Table 1 Instrument conditions and MRM transitions of precursor/product ions of analytes
Full size table

2.8 Validation of procedure

Validation of method used for the multi-residue analysis of 14 pesticides assessed was based on assessment of accuracy and precision, matrix effects, limit of detection and limit of quantification following SANCO guidelines (SANCO/12571/2013). Recovery percentage and relative standard errors (RSEs) were evaluated twice in six soil samples repeatedly measured at two fortification levels (10 and 100 µg/kg) for each compound using the multistandard compound solution. Extraction recovery rates between 70 and 120% were acceptable (SANCO/12571/2013). Limit of detection (LOD) and limit of quantification (LOQ) were determined as the compound concentrations which yielded signal-to-noise (S/N) ratios of 3 and 10, respectively. Linearity was assessed using 12 calibration standards within the range of 0.39 ng/ml to 800 ng/ml. Weighted (1/X) least squares regression was used to fit the calibration curves. Only curves with a correlation coefficient (R2) > 0.997 were accepted. For each compound, more intense multiple reaction monitoring (MRM) transition were selected and used for quantification. However, for abamectin, cyfluthrin, cypermethrin, deltamethrin and emamectin benzoate, a second MRM transition were used as a qualifier ion to eliminate potential false positive results caused by close elution. Matrix effects (ME) were assessed by comparing slopes in calibration solutions prepared in soil matrix (Sm) and in solvents (Ss). The formula (Eq.) below was used to calculate matrix effects.

$${\text{ME(\% )}} = {(}{\raise0.7ex\hbox{${{\text{S}}_{{\text{m}}} }$} \!\mathord{\left/ {\vphantom {{{\text{S}}_{{\text{m}}} } {{\text{S}}_{{\text{s}}} }}}\right.\kern-0pt} \!\lower0.7ex\hbox{${{\text{S}}_{{\text{s}}} }$}} - {1)} \times {100}$$

Value of ME comprise between −20% and 20% indicates low and mild Matrix effects; ME between −20% and −50% or 20% and 50% indicate that matrix effects are medium whereas beyond 50% or −50% indicate that matrix effects are very strong [15]. When matrix effects are strong, matrix-matched standard calibration must be used in the quantification method [16].

3 Results

3.1 Current physico-chemical properties of cotton production system soils in Benin

Results of soil samples content analysis indicated moderate variability of organic matter content of different cotton production systems’ soils in Benin (Table 2). Soils under conventional cotton cultivation showed low organic matter (OM) content (0.95% ≤ OM ≤ 1.87%) than soils under organic cotton cultivation and natural forests where organic matter contents were respectively on average (1.11% ≤ OM ≤ 3.39%) and quite high (1.17% ≤ OM ≤ 7.99%). Soil samples from Banikoara cotton production systems, the largest conventional cotton production area with use of large amounts of pesticide showed the lowest and highest levels of organic matter respectively 0.95% ± 0.18 for conventional cotton production systems’ soils and 7.99% for soils of natural forests (Table 2). Samples were more acidic (pH = 5.7) and had poorest clay content (4.34% ± 0.96). However, all other soils sampled were less acidic (5.75 ≤ pH ≤ 6.97).

Table 2 Characteristics of soils of cotton production systems in Benin
Full size table

3.2 Salting out extraction and clean up procedures

Results showed an average recovery percentage ranging from 74.2 to 112.2 and 71.9 to 104.0 respectively at 10 µg/kg and 100 µg/kg fortification levels (Table 3). Relative standard errors (RSE) were generally low (RSE < 10%) for all analytes except for cyfluthrin where RSE = 14.00 at low fortification level.

Table 3 Percentage recoveries at two fortification levels
Full size table

3.3 Method validation

Results revealed that matrix effects assessment (ME) ranged between −0.3 and −43.7%. Abemectin, cypermethrin and triasophos, matrix effects were −10.27, −11.39 and −12.97%, respectively (Table 4). Only acetamiprid and bifenthrin had relatively high values of ME equal to −20.59 and −43.66% respectively.

Table 4 Retention time, linearity, matrix effects, LODs and LOQs obtained for LC/MSMS method
Full size table

Low values of limit of detection (LODs) and limit of quantification (LOQs) were obtained for all analytes in soil matrix extract. Triasophos had the lowest LOD and LOQ values equal to 0.26 and 0.88 μg / kg respectively and the highest values of these parameters were observed for cyfluthrin, 16.68 and 55.60 μg / kg (Table 4).

3.4 Trends of pesticide residue levels throughout cotton production system soils in Benin

Results from pesticides residues analysis revealed seven pesticide compounds (emamectin benzoate, Abamectin, Acetamiprid, deltamethrin, Imidacloprid, Profenofos, and Triasophos) were detected and quantified in the soil samples analysed. Concentrations measured for Cyfluthrin, lambdacyhalothrin, Teflubenzuron, cypermethrin, novaluron and indoxacarb were below to their respective limit of detection (LOD, 16.68 μg/kg, 7.03 μg/kg, 7.98 μg/kg, 5.02 μg/kg, 1.27 μg/kg and 0.47 μg/kg) (Table 5). Bifenthrin was detected in some areas but couldn’t be quantified. Concentrations measured between limit of detection (LOD = 1.01 μg/kg) and limit of quantification (LOQ = 3.39 μg/kg) represents more than 95% of samples analyzed (Table 5).

Table 5 Concentrations of pesticides in soil samples from Benin cotton production systems
Full size table

Fifty percent of soil samples did not show abamectin compounds. However, in samples from Banikoara and in some other conventional samples, concentrations measured above limit of detection (LOD = 3.07 μg/kg) and quantification (LOQ = 10.25 μg/kg) were observed for abamectin.

Like for cyfluthrin and lambdacyhalothrin, concentrations measured for deltamethrin were mostly below the limit of detection except in some conventional samples from Banikoara and Djidja where concentrations above the limit of quantification (LOQ = 7.22 μg/kg) were observed.

Emmamectin benzoate, profenofos and triasophos were quantified in several soil samples from fields under organic cotton cultivation. High measured concentrations of imidacloprid were observed in Banikoara (7.00 μg / kg), Djidja (116.34 μg/kg and 29.18 μg/kg) and Glazoue (19.31 μg/kg) (Tables 5 and 6). These values are well above the limit of detection or quantification of imidacloprid (2.3 and 7.68 μg/kg). High concentrations of triasophos (12.64 μg/kg) were observed in Glazoue whereas Banikoara concentrated high concentrations of acetamiprid and profenofos, 10.82 and 12.66 μg/kg respectively (Tables 5 and 6).

Table 6 Minimum and maximum values of the concentrations of pesticides quantified in soils of the cotton cultivation areas of Benin
Full size table

Finally, emamectin benzoate (LOD = 0.50 μg/kg and LOQ = 1.67 μg/kg) was detected and quantified in several soil samples including those from organic cotton fields and natural forests (Tables 5 and 7). Concentrations measured are much higher than LOD and LOQ in conventional, organic samples and sometimes even in natural forest samples. High contaminations value of this compound was observed in Banikoara (18.5 μg/kg) (Table 5).

Table 7 Concentrations of pesticides quantified in fields under organic cotton production and in natural forests
Full size table

4 Discussion

4.1 Current physico-chemical properties of cotton production systems’ soils in Benin

The main findings indicated that soils from conventional cotton production farming lands were a little acid with lower values of organic matter content and had poorest clay than those from organic farming lands and natural forests used as a control. Therefore, the use of pesticides in conventional cotton production systems has altered soil physico-chemical properties. These findings strengthen previous knowledge according to which organic production systems’ soils tend to have higher levels of soil health metrics than conventional systems’ soils [17]. Indeed, soil organic matter (SOM) for example plays an important role to soil fertility. SOM allows storage of nutrients, increase of the cation exchange capacity, improvement of the stability of the aggregation and creates higher water holding capacity [18] and microbial and enzymatic activities [19]. Organic matter is also the main source of energy and nutrients for soil organisms [20]. It is therefore very important for growth and development of soil microorganisms, which are in fact the main architects of the microbiological processes of pesticide degradation in soils [21]. Furthermore, continuous destruction of soil physico-chemical properties will reduce also carbon sequestration potential of soils under conventional cotton production systems as highlighted by previous researches [17]. These findings call for a revision of pesticides use policies and development of an agriculture in accordance with environment and health protection.

4.2 Salting out extraction and clean up procedures

The average recovery percentages and low relative standard errors obtained are acceptable according to the recommendations of SANCO/12571/2013 guideline. Salt extraction (without buffer) using 1g of NaCl and 4g of MgSO4 was part of this method. Results obtained are comparable to those obtained by Dankyi et al. [22], Shi et al. [23] and Dong et al. [24] who respectively reported recovery rates ranging from 76.3 to 96.8%.; 82.9 to 112.0%; 77.6 to 87.9% and 70 to 120% using QuEChERS extraction method on soil samples with 1 g of NaCl and 4 g of MgSO4 for salting out procedure. Throughout several comparison procedures, Dankyi et al. [22] reported that the use of 1 g of NaCl and 4 g of MgSO4 provides best percentages of recovery. However, Feng et al. [25] through a method of multiple analyses of 31 pesticides in soil samples, found out that although use of 1g of NaCl and 4 g of MgSO4 provides good percentages of recovery (60 at 132%), extraction with buffered solution based on citrate buffer with 61 to 140% recovery was more advisable. Indeed, buffered procedures allow a better extraction of pesticides sensitive to pH variation [26].

Original QuEChERS extraction method uses combinations of primary and secondary amines (PSA) and magnesium sulfate (MgSO4) to absorb water, organic acids and polysaccharides present in the organic extract for clean-up procedure [27]. Dispersive solid-phase extraction (d-SPE) clean up step in this study used QuEChERS kits containing 150 mg of MgSO4 and 25 mg of PSA. Good percentages of recovery indicates there are very high interactions between PSA and analytes. Also, electro spray ionization is performed in positive mode, suggesting use of PSA for better stabilization of analytes [28].

4.3 Method validation

Matrix effect (ME) ˃ -20% indicates indeed low and mild matrix effects [15] whereas ME < -20 indicates signal suppression and suggests the use of matrix-matched calibration curve for the quantification [16]. Nevertheless, in this study calibration curves in solvent were used. Linearity was assessed by using calibration curves in solvent which were linear in the studied working range from 0.39 to 200 ng / ml and the R2 coefficients of determination are greater than or equal to 0.998 for all analytes (Table 4).

Low values of limit of detection (LODs) and limit of quantification (LOQs) were obtained for all analytes in soil matrix extract. For imidacloprid, the LOD, 2.30 μg / kg is similar to that reported under same conditions for the same compound (2 μg / kg) by Dankyi et al. [22]. The same study reported LOD = 1 μg / kg for acetamiprid. Thus, the value is twice the value obtained here, LOD = 0.47 μg / kg. Throughout a multi-residue analysis of 36 soil pesticides using QuEChERS and LC–MS/MS, Feng et al. [25] found lower LODs values for imidacloprid, acetamiprid, profenofos and triasophos (0.2; 0.2; 0.2 and 0.1 μg/kg respectively). This can be explained by the difference in extraction methods. Indeed, buffered procedures based on citrate were used during extraction of these compounds. This means that these compounds could be sensitive to pH variation during extraction. Golge and Kabak [29, 30] have obtained more or less similar LODs for abamectin, acetamiprid, bifenthrin, deltamethrin, imidacloprid and teflubenzuron respectively in tomatoes and oranges matrix (4.7; 1; 1.9; 2.9; 2.7; 1.3 μg / kg and 5; 1; 2; 1; 1; 2 μg / kg).

4.4 Trends of pesticide residue levels throughout cotton production systems’ soils in Benin

Although actives ingredients Cyfluthrin, lambdacyhalothrin, Teflubenzuron, cypermethrin, novaluron, indoxacarb and bifenthrin were registered in Benin and therefore can be used, they were not quantified in the soil samples evaluated. While Teflubenzuron, novaluron and indoxacarb are not often found in the formulations of pesticides used in the cotton production systems of Benin, cyfluthrin, cypermethrin and lambdacyhalothrin are among the active ingredients found in pesticides commonly used in Benin [31]. Their absence in the soil samples could be explained by the fact that the soil samples were collected 30 days after the last pesticides application. In fact, the half-lives (DT50) are respectively 33 days, 22 days and 27 days for cyfluthrin, cypermethrin and lambdacyhalothrin (https://sitem.herts.ac.uk/aeru/ppdb/index.htm). Half of amounts of these active ingredients used could therefore disappear before soil samples were sampled. The work of Aikpo et al. [32] revealed that soils of Djidja district were contaminated with cypermethrin, profenofos, acetamiprid and glyphosate. Biaou et al. [33] also indicated contamination with endosulfan and triasophos in the department of Borgou.

Active ingredients abamectin, profenofos, deltametrine and acetamiprid were quantified, although their very short half-life times. This could mean that large quantities of these active ingredients were sprayed in the cotton production systems, especially at Banikoara and Djidja Districts. This misuse or overuse of these molecules did not however constitute a serious environmental and health risk from a toxicological point of view. In fact, concentrations measured and reported were lower and much lower than amounts necessary to induce a biological activity capable of causing death (LD50). However, there is a risk of bioaccumulation or bioconcentration from these active ingredients.

Active molecules emamectin benzoate and imidacloprid have the longest half-life times, respectively 300 days and 191 days. They are therefore more persistent in the environment and could represent a potential risk for human health, insects, birds and soil microorganisms. Contamination of water and aquatic living organisms is also to be feared.

Presence of Emmamectin benzoate, profenofos and triasophos in some organic cotton production farms is an indication that neighbouring chemical-based production systems have contaminated surrounding environments as recently found in other study [34].

These results show that high cotton production regions were all contaminated by pesticides used to control cotton pests in Benin. However, with regard to maximum concentrations measured and values reported, these contaminations do not really represent a threat for sanitarian risk from a toxicological point of view. Therefore, it is relevant and urgent that studies addressing environmental and health risks associated to the use and presence of these pesticide residues in soils from Benin cotton production regions should be addressed in order to help governments to take relevant measures to protect populations in general and especially those who are particularly exposed to pesticides. This study assessed the residues of parent active ingredients. Study addressing persistence and toxicity of key metabolites would be welcome to actually evaluate the impact of the use of pesticides in cotton production in Benin.

5 Conclusion

Benin cotton production system soils characterization revealed soil degradation and contamination to pesticides residues especially in conventional cotton production systems. An overall soil contamination mainly by emamectin benzoate and secondarily by imidacloprid, profenofos, acetamiprid, triasophos, abamectin and deltamethrin were observed in conventional cotton production farming lands. Banikoara and Djidja district were the most contaminated and presented the most degraded soils. Environmental risk assessment, pesticide residues and their key metabolites monitoring studies are needed to assist in making relevant decisions for safe environment cotton production systems in Benin. Development of diverse organic pesticides based on local ecological knowledge should be also a new direction for future cotton production practices in the country and in West Africa.