BEES AND SYSTEMIC INSECTICIDES (IMIDACLOPRID, FIPRONIL) IN
POLLEN: SUBNANO-QUANTIFICATION BY HPLC/MS/MS AND GC/MS.
BONMATIN J.M.(1), MARCHAND P.A.(1), COTTE J.F.(1), AAJOUD A.(1),
CASABIANCA H.(2), GOUTAILLER G.(2), COURTIADE M.(2)
(1)Centre de biophysique moléculaire, CNRS, UPR 4301, F-45071 Orléans Cedex 02,
(2)Service Central d’Analyses, CNRS, USR 59, BP 22, F-69390 Vernaison, France.
Imidacloprid and fipronil are two insecticides acting on the central nervous system. They are used
worldwide, from the mid nineties, especially for seed coating of crops. Concomitantly to their
introduction and their increasing use in French fields, honeybee populations decreased. Bee
problems are nowadays reported in other countries (also called disappearing disease, desplobación
de las colmenas, trouble des abeilles, deperimento degli apiari or colony collapse disorder). Bee
problems have certainly several origins. Contamination of pollen and nectar by such chemicals
appeared as one of the most probable cause, since subletal effects and chronic intoxications were
observed on bees, at very low concentrations. We developed fully validated methods to measure
contamination in pollen (sunflower and maize) for imidacloprid (HPLC/MS/MS) and for fipronil
and 3 of its derivatives (GC/MS). For imidacloprid LOD and LOQ are 0.3 and 1 ng/g, respectively.
For fipronil and each derivative (fiprole), LOD and LOQ are 0.07 and 0.2 ng/g, respectively. The
averaged level of imidacloprid in pollen was 2-3 ng/g, which is 20-30 times higher than the
concentration inducing significant mortality by chronic intoxication. Fiproles were detected in
48% of pollens issuing from treated crops. Fipronil and its sulfone derivative represented 77% and
17% of contaminants, respectively. The averaged fiprole sum Σf was 0.3 - 0.4 ng/g, which is 30-40
times higher than the concentration inducing significant mortality of bees by chronic intoxication.
KEY WORDS: imidacloprid, fipronil, toxicity, pollen, bee, colony collapse disorder.
The systemic insecticides imidacloprid and fipronil are commercialised by Bayer AG
(Gaucho®, Confidor®…) and BASF (Regent®, Schuss®…), respectively. Rapidly
they were suspected of harmful effects on honeybees, particularly at subletal doses or
by chronic intoxication. Used worldwide for treatment of numerous crops, theses
powerful neurotoxins are insecticides acting on the central nervous system (CNS).
Imidacloprid is a chlorinated neonicotinoid that binds acetylcholine receptors. Fipronil
belongs to the phenylpyrazole class and binds GABA receptors. Both affect the CNS by
interfering at the post synaptic level.
Far under lethal doses (LD50 is 4-6 ng/bee for both insecticides; CST 2003, CST 2004,
Tingle 2003, Cox 2005, Agritox 2007), vital functions of bees are affected by very low
concentrations inducing subletal effects or chronic intoxications. These concentrations
are in the range from 0.1 to 10 ng of imidacloprid per g of food supply (Suchail 2001,
Colin 2004) and the situation appeared somewhat similar for fipronil. Here, recent
studies (Decourtye 2005 and El-Hassani 2005) showed that the chronic exposure to
fipronil during 11 days, with amounts extending from 0.075 to 0.3 ng/bee, has still
lethal effects. Furthermore, a significant mortality was also observed, after 11 days, for
0.01 ng/g of fipronil in the food of bees (Belzunces 2003).
When applied as seed coating for crop protection (generally from 50 to 100 g/ha), the
insecticide is first solubilised in the water of the soil, around seeds. The chemical is
distributed between the two major components of the soil, namely: the soil water and
the clay-humus complex. Thus, a quantity of chemical remains in a soluble state and is
suitable to be absorbed by roots cells. In fact, imidacloprid or fipronil are absorbed by a
simple process of passive diffusion. Then, it is transferred by the xylem pathway into
the sap flow, this also depending on its relative solubility (imidacloprid: pKow = 2.8;
fipronil: pKow = 0.57). Such systemic properties have been depicted by Bromilow in
1990. During the cycle of growth, high concentrations of the insecticide are observed
into the first leaves, whereas higher parts display lower levels. At the time of flowering
of the plant, a minor quantity of the product can be also transferred from the older
leaves towards the inflorescence, by the flow of the elaborated sap (phloemic pathway).
That are the reasons why, nectar and pollen can be significantly contaminated by the
To evaluate properly the risk for bees foraging on treated crops (sunflower, maize), one
of the first steps is to measure contamination of pollen in fields. Moreover, it is also
necessary to take into account the first metabolites which sometimes are as much (or
more) toxic than the parent compound. We have developed a HPLC/MS-MS
methodology to measure and characterize the behaviour of imidacloprid in soils, plants
and pollens. We also developed a GC/MS method to detect and to quantify fipronil and
its 3 main metabolites in pollen. Pollens were sampled directly on flowers and also at
the beehive entrance (trapped pollen effectively harvested by bees). Our analytical
methods are particularly sensitive. They satisfied quality standards, European
Directives (Directives 96/23/EC and 2002/657/EC) and specific criteria required by a
French expert committee (Scientific and Technical Committee of the multifactorial
study of bee disorders). Measures were performed according to the whole set of quality
criteria and following GLPs.
MATERIALS AND METHODS
All samples were collected from 1998 to 2005 in the whole French territory and
especially in intensive agricultural areas. Obviously, samples from treated crops have
been collected before suspension of insecticides used in France (Gaucho®: 1999 for
sunflower, 2004 for maize; Regent TS®: 2004 for sunflower and maize). Pollens from
flowers (sunflower, maize, cistus, buckwheat…) were carefully sampled in the middle
of the fields to circumvent from edge effects. Samples of trapped pollens were mostly
collected from beehives located in -or very close to- the fields of interest. Sampling was
also performed in crops growing under insect-proof tunnels in which beehives were
setup. All pollens were sampled by a specialized company (TESTAPI, France)
according to strict protocol adapted to this case study. Samples were bagged (double
bag), kept safe from light and frozen at temperature <-20°C.
Extraction and purification
Concerning imidacloprid, the preparative procedures are fully described in our previous
publications (Bonmatin 2003, Bonmatin 2004 and Bonmatin 2005a). Briefly, pollen
(10 g in ethanol/water) was mixed and extract was centrifuged and evaporated. A pH 7
buffer and dichloromethane were added and the organic phase was extracted and
evaporated. The oily residue was diluted with hexane, ultra-sonicated and centrifuged
after adding acetonitrile/water. The upper phase was centrifuged and an aliquot of 25
µL was injected in HPLC.
Concerning fipronil, trapped pollens were dried whereas pollens directly sampled on
flowers did not need this preliminary step. Then, pollen (10 g) was extracted twice with
ACN. Extracted phases were evaporated and put in dichloromethane. This solution was
purified, first on C18 , then on florisil. The recovered solution was evaporated and
dissolved in ethyl acetate. 2 µl of the latter solution was injected in GC.
For imidacloprid, the LC system was a Perkin Elmer (Framingham, USA). It was fitted
with a C18 Supelcosil ABZ + (150 mm × 4.6 mm) from Supelco Park, PA, USA. The
MS system was constituted of a standard atmospheric-pressure-ionisation source
configured as APCI. The signal corresponding to imidacloprid (m/z: 256→209 and
209→175) in pollen is illustrated in Figure 1.
Figure 1: Structure of imidacloprid and chromatogram of a typical pollen containing
imidacloprid at 2 ng/g (m/z = 209). The limit of detection (LOD) was 0.3 ng/g, whereas
the limit of quantification (LOQ) was 1 ng/g.
Figure 2: Structure of fipronil (X = SO) and its 3 derivatives (sulfide: X = S; sulfone:
2; desulfinyl: X = none) and total ion chromatogram (TIC) of a pollen in which
levels are at 0.2 ng/g. The LOD was 0.07 ng/g, whereas the LOQ was 0.2 ng/g.
X = SO
For fipronil, the GC system
0.25 mm; 0.25 µm
pectrometer quadrupole (Agilent 5973). For the source, the energy (Electronic Impact) s
was an Agilent 6890N with a DB-XLB column (30 m;
). The injector was an Agilent 7683 and the detector was a mass
was 70 eV; temperature: 230°C. The 3 selected ions used for detection and calibration
were at m/z 390-388-333 for fipronil desulfinyl, 420-353-351 for fipronil sulfide, 369-
367-351 for fipronil and 452-385-383 for fipronil sulfone. An example of signal for
fipronil and its metabolites (each at 0.2 ng/g) in pollen is shown in Figure 2.
Flowers and pollens from organically farming crops were used as references. They did
imidacloprid signal and were the basis for calibration and comparisons.
ated crops, issuing from a field which had received imidacloprid treated Note that untre
crops the year before, were not free from the chemical and represented improper
Figure 3: Distribution of the samples (flowers and pollens) as function of imidacloprid
concentration ranges(x in ng/g). Top: sunflower; bottom: maize.
ns, imidacloprid was not detected (LOD = 0.3 ng/g). 25% of samples were positive
ount t exceeding 1 ng/g. Owith the am of imidacloprid no
imidacloprid with quantified concentrations, between 1 and 11 ng/g. The mean value on
this data set is 3 ng/g.
not display any
of 24 pollens from treated sunflower were analyzed (Figure 3). In only 17% of
ther pollens (58%
35 flowers (sunflower)
24 pollens (sunflower)
48 flowers (maize)
47 pollens (maize)
< 0.30.3 < x < 11 < x < 1010 < x
< 0.30.3 < x < 11 < x < 1010 < x
A similar situation was depicted from 47 maize pollens (Figure 3). Here, the mean
value on the data set was 2 ng/g. It is noteworthy that imidacloprid was not
accompanied by its main derivatives, whatever the crop (sunflower or maize). This
suggests that flowers, thus pollens, display such levels of imidacloprid from the xylemic
and/or the phloemic pathway without a fully efficient metabolism within the plant.
lts for 195 pollens. For each pollen sample, analysis gave 4 separate
tives (4 fiproles). In order to describe results corresponding to fipronil and its 3 deriva
the fiprole sum (Σf) for each sample, these 4 data were added according to the
following principles. When a level was found under the limit of detection
(LOD = 0.7 ng/g), a value of zero was assumed. When a level was positive (between
LOD and LOQ = 0.2 ng/g), a mean value of 0.14 ng/g was adopted. When a level was
higher than the LOQ, the quantified result was kept. For instance, the fiprole sum is
Σf = 0.64 ng/g in the case of a sample having fipronil at 0.5 ng/g, a positive sulfone
signal, and the two other derivatives not detected.
We report here resu
Pollens from flowers: On 5 samples of pollens from untreated flowers (sunflower, corn
and buckwheat), only one is positive (Figure 3) leading to an averaged Σf = 0.03 ng/g.
On the 83 samples of pollen of flower from seeds treated with Regent TS, 47 % were
positive. Among them, 28 % appeared contaminated higher than LOQ (Figure 4). Here
the maximum value was 8.3 ng/g. For this data set, the averaged Σf was 0.42 ng/g.
Figure 4: Distribution of pollens sampled on flowers, according to the
level of fiproles. LOD = 0.07 ng/g and LOQ = 0.2 ng/g. D: at least one
fiprole was detected; Q: at least one fiprole was quantified.
averaged Σf was 0.06 ng/g. When seeds were treated, 50% of the 66 pollen data set was
positive. Among them, 20% were higher than LOQ. The maximum value was 4.3 ng/g.
For this data set, the averaged Σf was 0.29 ng/g.
Figure 4 and 5 show that about 20% of both types of pollen from untreated fields are
ositive for at least one fiprole. This probably originp
result is in agreement with recent data from Kadar et al. (2006) who have also analyzed
untreated pollens collected in 2005. Such contamination of untreated pollen could have
nd buckwheat). On
Pollen of flowers
poll : We sampled 41 trapped pollens from untreated crops (sunflower, corn
ata set, 17% of the samples are positive (Figure 5). The this d
ates from diffuse pollution. This
a first effect on pollinators at all. Further, when comparing pollens from untreated crops
to that from treated crops, it is clear that a novel class of contaminated pollens appears
in treated crops. Here, levels are higher than LOQ = 0.2 ng/g. Such pollens represent 20
to 30% of each data set. Obviously, pollinators should be as much affected by the latter
pollens. In other terms, we demonstrated i) a slight contamination of 20% of pollens by
diffuse pollution in untreated fields and ii) the direct (and higher) additional
contamination of 20 – 30% of pollens because of the systemic properties of fipronil,
consecutively to seed dressing. As a matter of fact, fiproles were detected in 48% of all
pollens from treated seeds. Here, our results confirm and precise data from Chauzat et
al. (2006) who have shown that 3 pollen samples contained fipronil.
Figure 5: Distribution of pollens collected by bees, according to the
level of fiproles. LOD = 0.07 ng/g and LOQ = 0.2 ng/g. D: at least one
fiprole was detected; Q at least one fiprole was quantified.
bees). We also observed that results are very similar when comparing the 2 major crops
of interest: sunflower and maize. This has two consequences. The first is that the
systemic impact of fipronil is now well established up to pollen. The second is that we
can cumulate data issuing from these 2 crops, in order to describe more accurately the
contamination of pollens, especially in terms of metabolism.
ample (pollens directly sampled on flowers
hy that Figure 4 and 5 demonstrate a similar situation whatever the type of
or trapped pollens which are collected by
etabolism in pollen
ollens from treated crops (48% of pollens). This examination was performed in term of
occurrence and in term of quantity. In term of occurrence, fipronil was detected in 95%
of these pollen samples. Occurrence is 53% for sulfone fipronil, 24% for sulfide fipronil
and 13% for desulfinyl fipronil. This shows that fipronil and its sulfone derivative are
the 2 compounds that appeared the more often in pollens. Other derivatives were
detected more rarely and were generally associated to the previous ones. In term of
quantity, the mean level of fiproles is Σf = 0.76 ng/g in these pollens. The quantitative
repartition is 0.59 ng/g for fipronil (77%) and 0.13 ng/g for its sulfone derivative (17%),
as shown in Figure 6. This indicated that fipronil largely dominates (77%) whereas its
oxidized derivative occurs for 17%, its reduced derivative for 5% and its photo-
degradation product for only 1%.
: Examination of the metabolite content was done in contaminated
66 Regent TS
Figure 6 : Mean quantitative distribution of fiproles in contaminated
ollens from treated crops. 100% corresponds to the mean fiprole level:
Σf = 0.76 ng/g.
measured with a sensitive HPLC/MS-MS methodology with Imidacloprid was
LOD = 0.3 ng/g and LOQ = 1 ng/g (Bonmatin 2003). For a comprehensive approach,
this technique allowed to follow imidacloprid from seed to soil, in the plant and up to
pollen (Bonmatin 2005a, Bonmatin 2005b). As a matter of fact, pollen contained
principally imidacloprid with an averaged level at 2 ng/g (maize) and 3 ng/g
(sunflower). When compared to the level inducing significant bee mortality on 11 days
(by repeated doses at 0.1 ng/g), the risk appeared worrying (CST 2003, Rortais 2005).
by fipronil than by
For fipronil, we developed and validated a new method by GC/MS in order to detec
and to quantify this compound and 3 of its derivatives in pollen. Two types of pollen
were extensively studied: pollen directly sampled on crops (pollens from flower) and
pollen collected by foraging bees and sampled at the entrance of beehives (trapped
pollens). From treated crops, half of pollens (48%) were found positive
(i.e. > LOD = 0.07 ng/g). The averaged fiprole levels were Σf = 0.4 ng/g (pollens from
flowers) and Σf = 0.3 ng/g (pollens from traps). Considering only contaminated pollens
from treated crops, the averaged fiprole level was Σf = 0.76 ng/g. Here, fipronil was the
major component (77%) followed by its sulfone derivative (17%).
Globally, pollens appeared less contaminated (10 fold lower)
imidacloprid. This is in agreement with their respective pKow (0.57 and 2.8,
respectively). The case of fipronil differs from that of imidacloprid because half pollens
were not contaminated (this ratio is 17% for imidacloprid). However fipronil and its
derivatives are about 10 times more toxic for bees, especially in terms of chronic
intoxication leading to mortality (imidacloprid: 0.1 ng/g; fipronil: 0.01 ng/g). Both
insecticides appeared bio-available in fields. For each insecticide, there is no significant
difference, in term of spatial or temporal distribution, with respect to sampling on
treated crops. For both insecticides, mean concentration in pollen was, at least, an order
of magnitude higher than the lowest level which still induces mortality on bees.
According to the PEC/PNEC method of risk assessment in fields (CST 2003 and CST
2004, Halm 2006), the risk appears as much worrying for both systemic insecticides.
ed by the French Ministry of Agriculture and supported by This program was supervis
the Commission Regulation (EC) No 917/2004. We thank Dr L. Belzunces, Dr E.R.
Bengsch, Dr M.E.Colin, Dr A.L. Thomann and J.P. Faucon for useful discussions.
. In Base de données sur les substances actives phytopharmaceutiques. Agritox., 2006. AFSSA
(on line : http://www.dive.afssa.fr/agritox/php/sa.php?source=UE&sa=1134).
B 19-2 Impact de la contamination du
B archand, P. A., Charvet, R., Colin, M. E., 2004. Fate of systemic insecticides in fields
(imidacloprid and fipronil) and risks for pollinators. Eurbee 1, (on line: http://web.uniud.it/eurbee/).
elzunces, L., 2003. Programme Communautaire pour l'Apiculture; Convention 3
miel par le fipronil sur l'activité et la survie des abeilles: aspects physiologiques et analytiques.
elzunces, L.P., VIème programme communautaire pour l’apiculture. Projet 2106. Rapport d’étude
onmatin, J. M., M
Imidacloprid Uptake in Maize Crops. Journal of Agricultural and Food Chemistry 53: 5336-41.
, J. M., Marchand, P. A., Charvet, R., Moineau, I., Bengsch, E. R., Colin, M. E., 2005a. Quantification of
B E., Bengsch, E. R., 2003. A LC/APCI-MS/MS
C C., Lachaize, J., Cougoule, N., Aubert, M., 2006. A survey of pesticide
C method to quantify
C Etude Multifactorielle des Troubles des Abeilles.1-221.
onmatin, J.M., Moineau, I., Charvet, R., Colin, M.E., Fléché, C. & Bengsch, E.R. Behaviour of imidacloprid
fields. Toxicity for honey bees. (2005) in Environmental Chemistry, eds. Lichtfouse, E., Schwarzbauer, J. &
Robert, D. (Springer Berlin Heidelberg New york), pp. 483-494.
onmatin, J. M., Moineau, I., Charvet, R., Fleche, C., Colin, M.
Method for Analysis of Imidacloprid in Soils, in Plants, and in Pollens. Analytical Chemistry 75(9): 2027-2033.
romilow, R.H., Chamberlain, K., Evans, A. A., 1990. Physicochemical aspects of phloem translocation
herbicides. Weed Science. 38: 305-314.
hauzat, M. P., Faucon, J. P., Martel, A.
residues in pollen loads collected by honey bees in France. J. Econ. Entomol. 99(2): 253-262.
olin, M. E., Bonmatin, J. M., Moineau, I., Gaimon, C., Brun, J. P., Vermandere, J. P., 2004. A
and analyze the foraging activity of honey bees : relevance to the subletal effects induced by systemic insecticides.
Arch. Environ. Contam. Toxicol. 47, 387-395.
ST. 2003. Comité Scientifique et Technique de l'
(on line : http://www.agriculture.gouv.fr/spip/IMG/pdf/rapportfin.pdf).
C des Troubles des Abeilles : Fipronil utilise
D ., Pham-Delegue, M. H., 2005.
H ovel photochemical desulfinylation with retention of
Hent Approach for Systemic Insecticides: The Case
K y liquid
ST. 2004. Comité Scientifique et Technique de l'Etude Multifactorielle
en enrobage de semences (Régent TS) et troubles des abeilles. 21/12/2004, 1-104.
ecourtye, A., Deviliers, J., Genecque, E., Le Menach, K., Budzinski, H. , Cluzeau, S
Comparative sublethal toxicity of nine pesticides on olfactory learning performances of the honeybee Apis
mellifera. Arch. Environ. Contam. Toxicol. 48, 242-250.
ainzl, D., Casida, J. E., 1996. Fipronil insecticide: n
neurotoxicity. Proc. Natl. Acad. Sci. USA. 93: 12764-12767.
alm, M. P.; Rortais, A.; Tasei, J. N.; Rault, S. New Risk Assessm
of Honey Bees and Imidacloprid (Gaucho) 2006 Environmental Science & Technology 40 7 2448-2454
adar, A., Faucon, J. P., 2006. Determination of traces of fipronil and its metabolites in pollen b
chromatography with electrospray ionization-tandem mass spectrometry. J. Agric. Food. Chem. 54 (26): 9741-6.
ortais, A., G., Arnold, Halm, M. P., Touffet-Briens, F., 2005. Modes of honeybees exposure to systemic
insecticides: estimated amounts of contaminated pollen and nectar consumed by different categories of bees.
Apidologie 36: 71-83.
chail, S., Guez, D., Belzunces, L. P., 2001. Discrepancy between acute chronic toxicity induced by imidacloprid
and its metabolites in Apis mellifera. Environmental Toxicology and Chemistry 20(11): 2482-2486.
T nmental fate, ingle, C. C. D., Rother, J. A., Dewhurst, C. F., Lauer, S., King, W., 2003. Fipronil: enviro
ecotoxicology, and human health concerns. Rev. Environ. Contam. Toxicol. 176, 1-66.