Professional Documents
Culture Documents
Chemosphere
journal homepage: www.elsevier.com/locate/chemosphere
h ig hl ig ht s g r a p h i c a l a b s t r a c t
articleinfo
abstract
Article history:
Received 2 June 2016 Honeybees and bee products are potential bioindicators of the presence of contaminants in the envi-
Received in revised form ronment, enabling monitoring of large areas due to the long distances travelled by bees. This work
2 August 2016 evaluates the use of bee pollen as a bioindicator of environmental contamination by pesticides. A GC-
Accepted 3 August 2016 MS/ MS analytical method for multiresidue determination of 26 different pesticides in pollen was
developed and validated in accordance with the recommendations of the European Union SANCO guide.
Environ- mental monitoring was conducted using the analysis of 145 pollen samples collected from ten
Handling Editor: I. Cousins
beehives
in the experimental apiary of Embrapa in Jaguariúna (Sa~o Paulo State, Brazil). Bioallethrin and
Keywords:
pendi-
Bee pollen
methalin were identified in four and eighteen samples, respectively, at concentrations below the LOQ of
Bioindicator
Pesticides the method (25 ng g—1). Passive sampling with polyurethane foam discs was used as a control, and no
Contamination pesticides were found. The detection of pesticide residues in seven samples (33%) from commercial
Multiresidues apiaries in Ribeira~o Preto (S~ao Paulo State) confirmed the efficiency of the analytical method and
GC-MS/MS the
need for environmental monitoring for the presence of pesticide residues. The results demonstrated the
potential of bee pollen as a bioindicator of environmental contamination by pesticides.
© 2016 Elsevier Ltd. All rights reserved.
1. Introduction
constant application of pesticides poses risks to human health, due
to exposure to contaminated water, soil, air, animals, and plants.
The use of pesticides is essential in agriculture as it provides pest
Furthermore, pesticide use can lead to the disappearance of bene-
control, increases productivity, and reduces costs. However,
ficial insects (bees and wild pollinators, for instance) and may
promote the appearance of new pests and the evolution of pesticide
resistance among the pest population (Ferna´ndez et al.,
* Corresponding author. 2001; Tomita and Beyruth, 2002; Koifman and Hatagima,
E-mail address: Susanne.rath@gmail.com (S. Rath). 2003;
Koifman and Koifman, 2003; Magalha~es, 2005; Veiga, 2007).
http://dx.doi.org/10.1016/j.chemosphere.2016.08.022
0045-6535/© 2016 Elsevier Ltd. All rights reserved.
R.C. de Oliveira et al. / Chemosphere 163 (2016) 525e534
52
Although honeybees (Apis mellifera) are not the target of pesti-
Acetonitrile (residue analysis grade) was purchased from Tedia
cides, they are highly vulnerable to contamination, since they are
(Brazil). Anhydrous magnesium sulfate, sodium chloride, sodium
exposed to these substances while collecting nectar, pollen, and
citrate, and sodium hydrogen citrate sesquihydrate (all analytical
water for maintenance of the colony. In cases of acute toxicity,
grade) were purchased from J. T. Baker (USA). Primary-secondary
honeybees can die rapidly after the application of pesticides, while
amine (PSA, 40 mm Bondesil) sorbent was purchased from Varian
chronic exposure to sub-lethal doses can injure their foraging
(USA).
behavior and affect colony health and development (Johnston et al.,
2010; Gregorc et al., 2012; Pettis et al., 2012; Di Prisco et al., 2013;
2.2. Standard solutions
Williamson and Wright, 2013). The declining number of pollinating
insects, usually due to the abuse of pesticides, is of concern in
For preliminary studies, standard stock solutions
Brazil and in many other countries, due to the damage it can cause
(1000 mg mL—1) of alachlor, aldrin, bifenthrin, bioallethrin, etrinfos,
to agricultural production (Bacandritsos et al., 2010; Pettis
fluazifop, heptachlor epoxide, malathion, oxyfluorfen, pendime-
and
thalin, terbufos, and trifluralin were prepared in acetone. DDT,
Delaplane, 2010; Burkle et al., 2013; Castilho and Gonza´lez, alpha endosulfan, beta endosulfan, phenthoate, ethyl parathion,
2013;
methyl parathion, permethrin, pirimiphos-ethyl, and fempropa-
Tylianakis, 2013a, b; Tomazela, 2014).
trim were prepared in hexane. Hexachlorobenzene was prepared in
Bees cover a wide area (up to 7 km2) in their search for nectar
chloroform. Abamectin, used in the sorption studies, was prepared
and pollen, and are also numerous and easily kept by humans (Celli
in methanol. All the stock standard solutions were stored in dark
and Maccagnani, 2003). For these reasons, the use of honeybees
glass vials at — 16 ○C. Individual intermediate standard solutions
and bee products (pollen and honey) as bioindicators of environ- containing 10.0 mg mL—1 of these pesticides and the internal stan-
mental contamination has been of great interest in recent years. dard lindane were prepared daily by diluting 0.10 mL of the stock
Studies have shown that the level of contamination of hives by standard solution in 10 mL of acetonitrile.
pesticides is closely related to the proximity of the contamination Method validation employed two working standard solutions.
source and the duration of exposure (García-Chao et al., 2010; Mixture I contained aldrin, bifenthrin, bioallethrin, chlorpyrifos-
Mullin et al., 2010; Chauzat et al., 2011; Panseri et al., 2014; Malhat methyl, disulfoton, alpha-endosulfan, beta-endosulfan, fempropa-
et al., 2015). Among bee matrices, pollen has been indi- cated as the trin, o'p'DDT, oxifluorfem, ethyl parathion, methyl parathion, pen-
best for assessment of the presence of environmental pesticide dimethalin, trifluralin, and the lindane internal standard. Mixture
residues, as it is easy to collect and is frequently contaminated II contained acetochlor, alachlor, etrinfos, phenthoate, fluazifop,
(Chauzat et al., 2006, 2011). Furthermore, Chiesa et al., 2016 fos- alone, heptachlor epoxide, hexachlorobenzene, malathion,
verified that the presence of persistent organic pollut- ants in permethrin, pirimiphos-ethyl, terbufos, and the lindane internal
organic honey, affected by the geographical area, confirms that standard. Mixture I, containing fifteen standards (including the
honey bee and beehive matrices are appropriate sentinels for internal standard), was analyzed using Method I, while Mixture II,
monitoring environmental contamination. containing thirteen standards (including the internal standard),
In order to use bee pollen as a bioindicator, it is necessary to was analyzed using Method II. All standard solutions were kept in a
develop a sensitive and selective analytical method with an freezer at —16 ○C for up to six months.
appropriate sample preparation step, as this matrix is highly
complex. Solid-liquid extraction and the QuEChERS (Quick, Easy, 2.3. Environmental monitoring
Cheap, Effective, Rugged, and Safe) approach have been widely
employed for this purpose. Regarding analytical techniques, gas An apiary containing 10 beehives was installed in a bushland
chromatography coupled to mass spectrometry (GC-MS) and liquid region of a protected natural reserve. This field station, in Tan-
chromatography coupled to tandem mass spectrometry (LC-MS/ quinho Velho (municipality of Jaguariúna, Sa~o Paulo State),
MS) are the techniques most commonly used for multiresidue is located between latitudes 22º420 4400 and 22º420 5500 S and
analysis of pesticides at low levels of detection (Berrada et al., 2010; longi- tudes 47º000 5300 and 47º0100500 W, near km 127.5 of the
García-Chao et al., 2010; Mullin et al., 2010; Wiest et al., 2011; SP340
Kasiotis et al., 2014). highway (Campinas-Mogi Mirim), between the Atibaia and Jaguari
The main objective of the present study was to evaluate the rivers, and has a total area of 131.0 ha.
potential of bee pollen as a bioindicator of environmental The main crops in and around the experimental field were sugar
contamination by pesticides. Sorption of pesticides on bee pollen cane (Saccharum officinarum L.) pasture hay (several grass
was performed in order to assess the affinity between the pesti- species), coffee (Coffea arabica L.), citrus (mainly Citrus sinensis),
cides and the matrix. The presence of 26 pesticides was monitored corn (Zea mays), sunflower (Helianthus annus), eucalyptus
by GC-MS/MS using bee pollen samples, with passive samplers as (Eucalyptus spp.), and green manure (several species). There were
controls. also areas of pro- tected native forest.
Collection of 145 pollen samples was made in the
2. Material and methods experimental apiary between May 2012 and May 2013. The
collection of the pollen samples was performed directly in the
2.1. Chemicals and reagents ten hives using front porch pollen traps. This type of collector
has an entrance screen with several 4.5 mm holes that allow the
Certified analytical standards of acetochlor, alachlor, disulfoton, passage of worker bees but are small enough to retain the pollen
alpha-endosulfan, beta-endosulfan, etrinfos, fempropatrim, phos- grains attached to their hind legs. Subsequently, the pollen falls
alone, heptachlor epoxide, hexachlorobenzene, oxifluorfem, ethyl into a collection drawer. The collection period with the screen
parathion, methyl parathion, pendimethalin, and terbufos were installed was continued for three successive days, after which
purchased from Dr. Ehrenstorfer (Augsburg, Germany). Aldrin, the bees were given free access to the hives for approximately 30
bifenthrin, bioallethrin, methyl chlorpyrifos, DDT (dichloro- days, in order to prevent any negative impacts on colony
diphenyl-trichloroethane), phenthoate, fluazifop, lindane, mala- development.
thion, permethrin, pirimiphos-ethyl, and trifluralin were obtained As a control, the presence of pesticides in the environment was
from Fluka (USA). Abamectin were purchased from Chem Service also monitored using passive samplers containing polyurethane
Inc. (USA). All these compounds had purity of at least 98%. foam (PUF) discs, as described by Shoeib and Harner (2002). The
PUF discs had the following specifications: 14 cm diameter x
The extracts were collected in round-bottom flasks and
1.35 cm thickness; density 0.018 g cm3; surface area 324 cm2; mass
3.3 g; volume 184.7 cm3. Before use, all discs were previously evapo- rated to dryness in a rotary evaporator at 40 ○C. The
cleaned by Soxhlet extraction with dichloromethane and acetone residues were resuspended in 5.0 mL of acetonitrile and
transferred to 12 mL polypropylene tubes. The extracts were
(60 ○C, 8 h with each solvent) to ensure the absence of contami-
evaporated to dryness and
nants and/or interferences. After cleaning, each disc was dried
the residues were resuspended in 1.0 mL of acetonitrile before GC-
under an extraction hood, wrapped in aluminum foil, and stored at
MS/MS analysis.
room temperature until use.
For environmental sampling, six samplers were installed in the
2.7. Gas chromatography analysis
experimental field: one in the apiary and the others in the sur-
rounding bushland. The PUFs were left in place for about 30 days,
The pesticides were analyzed using a Varian 3900 GC gas
and were exchanged and analyzed for every pollen collection
chromatograph equipped with a Saturn 2100 T ion trap mass
period.
spectrometer (Varian Inc., Palo Alto, CA, USA). The data were ac-
The samples were placed in glass vessels (bee pollen) or wrap-
quired and processed with MS Workstation v. 6.5 software (Varian
ped in aluminum foil (PUFs) and stored in a freezer at—16 ○C until Inc., Palo Alto, CA, USA). The analytes were separated on an Rxi ® 5
analysis.
Sil (30 m, 0.25 mm id., 0.25 mm) capillary column (Restek Corpo-
ration, Bellefonte, PA, USA). The column oven temperature was
2.4. Commercial bee pollen samples
programmed from 100 ○C (hold for 2 min) to 150 ○C (hold for
10 min) at 10 ○C min—1, and then to 250 ○C (hold for 6 min) at
Twenty-one commercial bee pollen samples were obtained from
beekeepers in the region of Ribeira~o Preto (Sa~o Paulo State) in 5 ○C min—1. The injection temperature and volume were 250 ○C and
the autumn of 2013. Two of these samples were reported to be from 1 mL, respectively, and the trap, manifold, and transfer line tem-
beehives with elevated bee mortality and suspected contamination peratures were set at 220, 60, and 280 ○C, respectively. The carrier
by pesticides. The samples were packed in glass vessels and stored gas was helium at a constant flow rate of 1 mL min—1. Single ion
monitoring (SIM) was used for confirmation and quantitation of the
in a freezer at —16 ○ C until analysis.
pesticides, as required by SANCO document 10684/2009 (SANCO/
10684/2009). The retention times, precursor ions, collision en-
2.5. Survey of pesticides to be included in the study
ergies, and the selected quantitation and confirmation ions are
shown in Table S1.
The pesticides considered for inclusion in this study were those
used in the region surrounding the experimental field, as well as
2.8. High performance liquid chromatography analysis
substances monitored by the National Residues and Contaminants
of abamectin
Control Program (PNCRC) of the Brazilian Ministry of Livestock
Supply and Agriculture (MAPA) and by the Program for Pesticide
Abamectin was determined in pollen using the method
Residues Analysis in Food (PARA) of the National Health Surveil-
described by Dionisio and Rath (2016). The chromatographic sys-
lance Agency (ANVISA). Banned substances were also included.
tem consisted of an Agilent Series 1200 HPLC (Agilent
Information on the use of pesticides and the current crops in
Technologies, USA) equipped with a G1311A quaternary pump, a
the properties surrounding the experimental field was obtained
G1321A fluo- rescence detector (FLD), a G1316A column oven, and a
by means of interviews with the farmers. The use of pesticides
G1329A autosampler. Abamectin was separated on a
in the experimental field was reported on a monthly basis by the
LiChroCART® 55-4
re- searchers responsible.
Purospher® STAR RP-18 column (4.0 mm × 50 mm, 3 mm) (Merck,
Germany) at 30 ○C. The mobile phase was 98:2 (v/v) methanol:-
2.6. Analytical method
water, at a flow rate of 0.8 mL min—1. The injection volume was 5
mL. The excitation and emission wavelengths were set at 365 nm
2.6.1. Bee pollen sample preparation
and 470 nm, respectively.
Bee pollen samples were defrosted in the refrigerator (5 ○C),
brought to room temperature, dried in a circulating air oven
2.9. Method development and validation
(40e42 ○C) for up to 12 h, ground, and homogenized.
Approximately 2 g of pollen and 15 mL of acetonitrile were Validation of the developed GC-MS/MS analytical method for
transferred to polypropylene tubes (50 mL) and agitated in a the determination of pesticide residues in bee pollen was carried
vortex. Addition was made of 4.0 g of anhydrous magnesium out based on SANCO guideline (SANCO/10684/2009).
sulfate, 1.0 g of sodium chloride, 1.0 g of sodium citrate, and 0.5 g of Calibration curves were constructed using solvent, fortified
sodium hydrogen citrate sesquihydrate, followed by further blank bee pollen samples, and fortified blank bee pollen extracts,
agitation and employing the ordinary least-squares method. The linearity and
centrifugation at 13,440 g (5 ○C) for 5 min. An aliquot (1 mL) of the linear range were determined from calibration curves obtained by
supernatant was transferred to a polypropylene tube (50 mL) analysis of fortified blank bee pollen samples at six concentration
containing 95 mg of PSA sorbent and 750 mg of anhydrous mag- levels (10.0, 25.0, 50.0, 100.0, 200.0, and 300.0 ng g—1), with
nesium sulfate, followed by shaking in a vortex for 1 min and lindane as a surrogate internal standard (at 200.0 ng g —1). Possible
centrifugation for 10 min at 13,440 g (5 ○C). Finally, the outliers were evaluated by the Jacknife standardized residual test.
supernatant was transferred to a round-bottom flask and The ho- mogeneity of variance was evaluated by Levene's test.
evaporated to dryness The selectivity of the method was evaluated by the analysis of
in a rotary evaporator at 40 ○C. The residue was resuspended in six different commercial pollen samples in order to identify any
1.0 mL of acetonitrile and analyzed by GC-MS/MS. interference of co-extractives derived from the sample matrix.
Matrix effects were evaluated by comparing the calibration
2.6.2. PUF sample preparation curves obtained using the fortified blank bee pollen extracts (10.0,
The PUF samples were defrosted in a refrigerator (5 ○C), 25.0, 50.0, 100.0, 200.0, and 300.0 ng mL—1) and the calibration
brought to room temperature, cut into small pieces, and then curves obtained at the same concentration levels using solvent.
Soxhlet extracted with dichloromethane at 60 ○C for 8 h. Lindane (200 ng mL—1) was used as a surrogate internal standard.
For each concentration level, the matrix effect was expressed as the
were carried out in duplicate. The equilibration time was estimated
percentage of signal reduction or enhancement in the extract,
by plotting the sorption percentage as a function of time. The data
relative to the standard prepared in solvent.
were fitted to pseudo-first order, pseudo-second order, and Elovich
The within-laboratory repeatability was evaluated using
models in order to establish the sorption kinetics mechanism, as
blank bee pollen samples fortified with the pesticides at three
described by Peruchi et al. (2015).
different concentration levels (25.0, 100.0, and 300.0 ng g—1).
For the sorption isotherms, suspensions containing 100 mg of
Six indepen- dent replicates of each sample were analyzed on
pollen and 10 mL of water were agitated for 12 h in glass (aldrin and
the same day by the same analyst and with the same equipment.
abamectin) or plastic (malathion) Falcon tubes (50 mL), using an
The results were expressed as the relative standard deviation.
Accuracy was assessed by analysis of blank bee pollen samples orbital shaker with temperature control (24 ○C). The individual
analytes were then added to give final concentrations of 0.5, 1.0,
fortified with the 26 pesticides at three concentration levels (25.0,
2.0, 5.0, 10.0, and 20.0 mg g—1. The mixtures were agitated for 20 h
100.0, and 300.0 ng g —1), using six independent replicates (n ¼ 6).
The accuracy was expressed as the percentage recovery, defined as in the orbital shaker (24 ○C), centrifuged (13,440 g, 15 min), and
the ratio of the measured values to the values predicted by the the su- pernatant was discarded. The pollen samples were analyzed
calibration graph regression. According to SANCO document as described in Section 2.10. The analyte concentrations in the
10684/ 2009, acceptable recovery values are between 70 and extracts
120%. were determined using the matrix-matched calibration curves
obtained by analysis of blank bee pollen samples fortified with the
The limits of detection (LODs) and quantitation (LOQs) were
respective analytical standards, followed by calculation of the
determined by analysis of blank bee pollen samples fortified at
amount of analyte sorbed by the pollen (Cpads, mg g—1 pollen). The
decreasing analyte concentrations (300.0, 200.0, 100.0, 50.0, 25.0,
concentrations of the analytes in the aqueous solutions (C ads, m
aqg mL
and 10.0 ng g—1), evaluating the accuracy and precision of the re- —1
) were calculated from the difference between the initial
sults. The limit of detection (LOD) was equal to the minimum
concentrations in the aqueous solutions and Cpads. All the proced-
detectable concentration of each analyte present in the fortified
ures were carried out in duplicate. The results were modeled by
sample, based on a signal-to-noise ratio of 3. The limit of quanti-
means of the Freundlich equation (Eq. (1)), as described by Peruchi
tation (LOQ) was equal to the minimum concentration of each
et al. (2015). This model proposes an empirical isotherm for a non-
analyte present in the spiked sample, based on a signal-to-noise
ideal system, considering a heterogeneous surface, non-equivalent
ratio of 10, which could be quantified by the analytical method
bonding sites, and a multilayer sorption process.
with adequate precision (lower than 20%) and accuracy (recovery
between 70 and 120%). 1
log Csads ðeq:Þ ¼ log Kf ads þ log Caadsðeq:Þ [1]
n q
2.10. Palynological where Cads and Cads are the concentrations of analyte in the pollen
analysis
p aq
Ten bee pollen samples collected in the region of Ribeir~ao tained (in mg g pollen) were used to calculate the amounts of
—1
Preto (Sa~o Paulo State) were sent for palynological analysis at the sorbed analyte for each equilibration time (qt). All the procedures
Center
for Research in Palynology of the Institute of Botany, Department
of the Environment of Sa~o Paulo. The method employed was the
Eu- ropean standard for honey samples described by Maurizio
and
Louveaux (1965), without acetolysis and with the modifications
described by Barth et al. (2009) and Modro et al. (2009).
Fig. 1. Extraction efficiencies using the QuEChERS approach and a solid-liquid extraction with acetonitrile. Sample amount: 2 g, concentration of pesticides: 200 ng g —1.
Fig. 2. Matrix effect using the QuEChERS approach or solid-liquid extraction with acetonitrile. Sample amount: 2 g.
pesticides to the solvent. The QuEChERS approach also
and this time was used to establish the sorption isotherms. After
significantly decreased matrix effects, compared to solid liquid
20 h, the sorption percentages of aldrin, malathion, and abamectin
extraction (SLE). This technique was therefore selected for sample
were 90%, 62%, and 82%, respectively.
preparation.
For all the analytes evaluated, the pseudo-second order model
Finally, the method was validated in-house. Due to the positive
provided better fits to the data than the other models (pseudo-first
matrix effects observed, calibration curves were constructed using
order and Elovich) (Table 2). This indicated that the sorption ca-
fortified blank bee pollen samples, compensating the matrix effects
pacity of the pollen mainly depended on the availability of active
and ensuring the reliability of the results.
sites, rather than the analyte concentrations in solution.
The selectivity of the method was confirmed by the absence of
Sorption processes are influenced by the sorbent properties
interfering substances appearing at the retention times of the
(polarity, presence of bonding groups, porosity, and surface area),
pesticides that could compromise the identification or quantitation
the solute properties (the nature of the compound and its chemical
of the analytes or the internal standard in six blank bee pollen
structure, considering ionic charges, molecular size, presence of
samples from different sources.
rings, and polar groups), and by external factors such as pH, tem-
The validation parameters considered were linear range, line-
perature, and agitation intensity (Marshall et al., 1999). In order to
arity, matrix effects, LOD, and LOQ (Table 1). The linearity was
estimate the amount of each substance sorbed per gram of pollen,
higher than 0.91 for all the pesticides and the LOQ varied from 10
as a function of analyte concentration in solution, sorption iso-
to 100 ng g—1. The most pronounced matrix effect was observed for
therms modeled using the logarithmic form of the Freundlich
pirimiphos-ethyl, while the lowest was for phosalone.
equation (Eq. (1)) were constructed. The results are shown in
Precision, expressed as the relative standard deviation (RSD),
Table 3.
was lower than 20%. Accuracy, expressed as recovery, was between
Aldrin, malathion, and abamectin presented Freundlich
70% and 120% for all the concentration levels of the 26 pesticides
sorption coefficients (KF) of 2.95, 0.52, and 2.06 mg1—1/n (cm3)1/n g—1,
tested (25.0, 100.0, and 300.0 ng g —1, n ¼ 6). respectively. These values are of the same order of magnitude as
those reported by Thio et al. (2011) for magnetized ragweed pollen
3.3. Sorption experiments interacting with the pesticides atrazine, diuron, and lindane
(KF ¼ 1.86, 1.14, and 1.22 mg1—1/n (cm3)1/n g—1, respectively). Based
In order to evaluate the potential of bee pollen as a bioindicator
of environmental contamination, it is essential to assess its affinity
for the substances investigated. For this purpose, three pesticides
Table 2
with different log Kow were selected: aldrin, malathion, and aba- Sorption kinetic parameters of aldrin, malathion and abamectin for bee pollen.
mectin. Determination of the sorption equilibration time is a key
Pesticide Pseudo-second-order model
step required by OECD 106 for undertaking sorption studies
(OECD, 2000). The equilibration times were obtained by plotting qe (mg g—1) k2 (10—4 g mg—1 min—1) r
the sorption percentage against time (0, 1, 2, 4, 5, 7, 20, and 24 h). Aldrin 0.86 934 0.998
The initial concentrations of the analytes in contact with pollen Malathion 0.62 501 0.999
Abamectin 0.78 482 0.997
were
1.0 mg g—1. The graphs indicated that 20 h was long enough to reach
the apparent sorption equilibrium for all the analytes evaluated,
Table 1
Validation parameters of the analytical method.
Pesticide Matrix effect (%) Linear range (ng g—1) LOD (ng g—1) LOQ (ng g—1) Linearity (r) Classification
Fig. 3. Pollen production of each beehive. From bottom to top: March 2012; May 2012; March 2013; April 2013 and May 2013.
Table 4
Pesticide residues (ng g—1) determined in bee pollen samples collected in commercial apiaries from Ribeir~ao Preto.