Country-specific effects of neonicotinoid pesticides on honey bees and wild bees

Abstract

Damage confirmed
Early studies of the impacts of neonicotinoid insecticides on insect pollinators indicated considerable harm. However, lingering criticism was that the studies did not represent field-realistic levels of the chemicals or prevailing environmental conditions. Two studies, conducted on different crops and on two continents, now substantiate that neonicotinoids diminish bee health (see the Perspective by Kerr). Tsvetkov et al. find that bees near corn crops are exposed to neonicotinoids for 3 to 4 months via nontarget pollen, resulting in decreased survival and immune responses, especially when coexposed to a commonly used agrochemical fungicide. Woodcock et al. , in a multicounty experiment on rapeseed in Europe, find that neonicotinoid exposure from several nontarget sources reduces overwintering success and colony reproduction in both honeybees and wild bees. These field results confirm that neonicotinoids negatively affect pollinator health under realistic agricultural conditions.
Science , this issue p. 1395 , p. 1393 ; see also p. 1331

Full text

NEONICOTINOIDS
Country-specific effects of
neonicotinoid pesticides on
honey bees and wild bees
B. A. Woodcock,
1
*J. M. Bullock,
1
R. F. Shore,
2
M. S. Heard,
1
M. G. Pereira,
2
J. Redhead,
1
L. Ridding,
1
H. Dean,
1
D. Sleep,
2
P. Henrys,
2
J. Peyton,
1
S. Hulmes,
1
L. Hulmes,
1
M. Sárospataki,
3
C. Saure,
4
M. Edwards,
5
E. Genersch,
6
S. Knäbe,
7
R. F. Pywell
1
Neonicotinoid seed dressings have caused concern world-wide. We use large field
experiments to assess the effects of neonicotinoid-treated crops on three bee species
across three countries (Hungary, Germany, and the United Kingdom). Winter-sown oilseed
rape was grown commercially with either seed coatings containing neonicotinoids
(clothianidin or thiamethoxam) or no seed treatment (control). For honey bees, we found
both negative (Hungary and United Kingdom) and positive (Germany) effects during crop
flowering. In Hungary, negative effects on honey bees (associated with clothianidin)
persisted over winter and resulted in smaller colonies in the following spring (24% declines).
In wild bees (Bombus terrestris and Osmia bicornis), reproduction was negatively correlated
with neonicotinoid residues. These findings point to neonicotinoids causing a reduced capacity
of bee species to establish new populations in the year following exposure.
Global declines in honey bees and wild bees
have been linked to pathogens, climate
change, habitat fragmentation, and pes-
ticide use (1–3). The potential threat from
neonicotinoid seed coatings applied to
flowering crops has been the subject of consid-
erable debate (4–9). Neonicotinoids have been
shown to increase mortality in honey bees by
impairing their homing ability (4) and to reduce
the reproductive success of bumble bees (5,8,10)
and solitary bees (8,11); other studies have iden-
tified no effects (8,12,13). There is limited infor-
mation from replicated studies on longer-term
survival of honey bee colonies following exposure
[see (12)]. Landscape-scale experiments under real-
world agricultural conditions are needed to inte-
grate spatial, temporal, and species-specific variation
in order to understand the impacts of neonico-
tinoidsonbees(8,12,14–16). Such studies should
explore the impacts of different neonicotinoid
formulations, land use, and regional climate. In a
large-scale experiment spanning three European
countries, we tested the hypotheses that (i) expo-
sure to seed treatments containing neonicotinoids
affected the reproductive potential of managed
and wild bee species and (ii) whether such effects
differ between countries.
At each of 33 sites (Germany, 9; Hungary, 12;
and United Kingdom, 12) an average of 63.1 ha
(SE of ±2.8 ha) of winter-sown oilseed rape
(OSR) was established in 2014 (Fig. 1, fig. S1,
and table S1). We clustered sites into triplets
(>3.2 km between sites) and randomly allocated
sites to one of three treatments: (i) clothianidin
applied at 11.86 to 18.05 grams of active ingre-
dient per hectare (g a.i. ha
−1
) with a fungicide
(thriam and prochloraz) and nonsystemic pyre-
throid (beta-cyfluthrin) (trade name Modesto); (ii)
thiamethoxam applied at 10.07 to 11.14 g a.i. ha
−1
and combined with the fungicides fludioxonil and
metalaxyl-M (trade name Cruiser); and (iii) con-
trol OSR receiving a commercial fungicide (thriam
and dimethomorph in Germany and Hungary and
thriam and prochloraz in the United Kingdom)
but no neonicotinoid seed treatment. All treatments
received typical commercial inputs of pesticide
(e.g., lambda-cyhalothrin) and fertilizer, with these
standardized across a triplet. Standar diz ed c olo-
nies of honey bees (Apis mellifera)
and wild bees (bumble bee Bombus
terrestris and solitary bee Osmia
bicornis)wereintroducedtoeach
site. For honey bees, we quan-
tified the impacts of the treat-
ments on colony viability during
the crop flowering period and in
the year following exposure (hive
survival and overwintering worker,
brood, and storage cell numbers).
Overwintering fitness defines the
multiyear persistence of honey
bees. For B. terrestris,wemea-
sur ed imp acts o n within-year re-
productive output (colony weight
gain and worker, queen, and drone
production) and for O. bicornis
the number of reproductive cells
produced (table S2). Neonicoti-
noids can be persistent and wide-
spread in agroecosystems (17,18), so we quantified
residuesbothinthenestsofbeespeciesand
those expressed in the crop.
We found that neonicotinoid seed treatment
affected the interannual viability of honey bee
colonies following the winter period in a country-
specific manner. In Hungary, worker numbers
were 24% lower where clothianidin was com-
pared with the control [treatment × country:
c
2
(6) = 1.47, P= 0.01, explained variance = 59.4%]
(Fig. 2), with no significant effect of thiame-
thoxam. Clothianidin was more likely to be ex-
pressed in the crop where it was applied as a seed
treatment, which identified a mechanism of expo-
sure to the bees [c
2
(2) = 6.46, P=0.04],butthis
was not so for thiamethoxam (table S3). In the
United Kingdom, high hive mortality precluded
a formal statistical analysis of overwintering
worker numbers. However, median worker numbers
were zero for all four clothianidin-treated sites but
above zero for two of the control and one of the
thiamethoxam sites (table S2 and Fig. 2). Worker
numbers following the winter in Germany showed
no treatment effect (table S4). Overwintering honey-
bee brood, stored hive products (pollen and nectar),
and the likelihood of hives surviving the winter
were not affected by seed treatments (table S3).
Neither B. terrestris queen nor O. bicornis
egg cell production was directly affected by the
seed treatments or its interaction with country
(table S5). However, they were negatively corre-
lated with peak [c
2
(1) = 2.09, P=0.03,explained
variance = 13.5%] (Fig. 3A) and median [c
2
(1) =
4.34, P= 0.04, explained variance = 0.8%] (Fig. 3B)
neonicotinoid nest residues (combined clothianidin,
thiamethoxam, and imidacloprid). Imidacloprid
was not applied as part of the study, and its
presence is most likely a result of environmental
contamination from previous widespread agro-
nomic use (17,18). Residues of neonicotinoids
detected in stored hive products did not differ in
response to seed treatments for any bee species
(table S6). This may be due to the amalgamation
of stored hive products at the site level for resi-
due analysis, which may have obscured within-site
heterogeneityinresidues.Thenegativecorrelation
RESEARCH
Woodcock et al., Science 356, 1393–1395 (2017) 30 June 2017 1of3
1
Centre for Ecology and Hydrology, Natural Environment
Research Council, Oxfordshire OX10 8BB, UK.
2
Centre for
Ecology and Hydrology, Natural Environment Research Council,
Lancaster Environment Centre, Lancaster LA1 4AP, UK.
3
Szent
István University, 2103 Gödöllö, Hungary.
4
Am-Heidehof 44,
14163 Berlin, Germany.
5
Leaside, Carron Lane, West Sussex
GU29 9LB, UK.
6
Institute for Bee Research, 16540 Hohen-
Neuendorf, Germany.
7
Eurofins, Ecotox-GmbH, 75223 Niefern-
Öoschelbronn, Germany.
*Corresponding author. Email: bawood@ceh.ac.uk
Fig. 1. Location of the 33 experimental sites in the United
Kingdom, Hungary, and Germany. Seefig.S2foradiagrammatic
representation of the experimental setup.
on June 29, 2017 http://science.sciencemag.org/Downloaded from

for B. terrestris queen production remained sig-
nificant when we excluded sites with imidacloprid
residues [c
2
(1) = 2.14, P=0.02],althoughthiswas
not the case for O. bicornis [c
2
(1) = 0.05, P=0.81].
Country-specific responses to neonicotinoid seed
treatment were found for B. terrestris drone pro-
duction, with positive and negative effects from
exposure to thiamethoxam in Germany and the
United Kingdom, respectively [treatment × country:
c
2
(6) = 13.1, P= 0.04, explained variance = 13.6%]
(Fig. 2).
We also found seed treatment effects during the
crop flowering period that lasted between 3 and
6 weeks (tables S4 and S5). Significant interactions
between seed treatment and country were iden-
tified for peak worker [c
2
(6) = 16.6, P< 0.01, ex-
plained variance = 45.3%], egg cell [c
2
(6) = 4.13,
P= 0.01, explained variance = 49.9%], and com-
bined pollen and nectar storage cell [c
2
(6) = 40.5,
P< 0.001, explained variance = 53.6%] numbers.
These responses describe within-year colony per-
formance. Neonicotinoid exposure resulted in
both negative (Hungary and United Kingdom)
and positive (Germany) effects on colony size
(see Fig. 2; pairwise treatment comparison
given in tables S4 and S5). Bombus terrestris
worker and peak colony weight showed no
seed treatment response.
Our quantification of neonicotinoid effects on
the interannual viability of honey bees and wild
bee populations represents a fundamental advance
in our understanding of the impacts of these
pesticides. For solitary bees and bumble bees
(queen production), neonicotinoid impacts were
associated with the residues found in nests
rather than the experimental seed treatments.
For B. terrestris, the few treatment effects and
thepresenceofimidaclopridinstoredpollen
and nectar (tables S7 to S9) suggests that neg-
ative impacts of neonicotinoids may be driven
by the persistence of residues in the wider land-
scape rather than current management alone
(18,19). The European Union (EU) moratorium
meant that no neonicotinoids were applied to
oilseed in the surrounding landscapes during
the experiment, so such residues may originate
from previous agricultural use leading to ex-
pression in nontarget plants (17–19), guttation
fluids, or contaminated water (19,20). Although
the reproductive potential of O. bicornis was
also negatively affected by neonicotinoid resi-
dues in nests, the explained variation of these
effects was small. However, a failure to detect
small population changes may be due to limited
experimental replication restricting statistical
power. Our results suggest that even if their
use were to be restricted, as in the recent EU
moratorium, continued exposure to neonico-
tinoid residues resulting from their previous
widespread use has the potential to impact neg-
atively wild bee persistence in agricultural land-
scapes (14,18,19).
Taken together, our results suggest that expo-
sure to neonicotinoid seed treatments can have
negative effects on the interannual reproductive
potential of both wild and managed bees, but
that these effects are not consistent across coun-
tries. The country-specific responses of honey
bees and bumble bees strongly suggest that the
effects of neonicotinoids are a product of inter-
acting factors (20–23). This study has identified
between-country differences in the use of oilseed
rape crop as a forage resource for bees (affect-
ing exposure to crop residues) and incidence
of disease within hives. Both factors were high-
er for Hungarian and U.K. honey bees (tables
S10 and S11). Overall neonicotinoid residues
were detected infrequently and rarely exceeded
1.5 ng g
−1
(w/w). As such, direct mortality effects
caused by exposure to high concentrations of
neonicotinoidsarelikelytoberare(tableS12).
However, our results suggest that exposure to
low levels of neonicotinoids may cause reductions
in hive fitness that are influenced by a number of
interacting environmental factors. Such interact-
ing environmental factors can amplify the impact
of honey bee worker losses (e.g., through sublethal
Woodcock et al., Science 356, 1393–1395 (2017) 30 June 2017 2of3
Fig. 2. Summary effect sizes for the response of honey bees and wild bees to the neonicotinoid
seed treatments. An effect size represents the difference between the mean population response for a
given seed treatment and the control within a country, with this difference divided by the pooled
standard deviation, where asterisk (*) indicates a significant difference between the control and seed
treatment [either TMX (thiamethoxam) or CTD (clothianidin)] determined from the predicted marginal
means of the model “y ~ seed treatment × country + block/country.”Dagger (†) indicates where U.K.
colony survival was too low for a formal analysis. Note that effect sizes differ between countries.
Fig. 3. Wild bee reproductive success in response to neonicotinoid nest residues. Separate
graphs are shown for the response of B. terrestris queen production and O. bicornis reproductive cell
production to neonicotinoid residues found in nests. The significance of these relationships is based
on a likelihood ratio test comparison of H0: “y ~ country”and H1: “y ~ neonicotinoid +country.”
Neonicotinoid residues are based on summed concentrations of clothianidin, thiamethoxam, and
imidacloprid.
RESEARCH |REPORT
on June 29, 2017 http://science.sciencemag.org/Downloaded from

toxicity effects) and reduce longer-term colony
viability (4,16). Note that our common experimen-
tal approach applied across three countries revealed
varying impacts and may explain the inconsistent
results of previous studies conducted in single
countries or at few sites (4,5,8,12,13,15).
REFERENCES AND NOTES
1. A. J. Vanbergen, Front. Ecol. Environ 11, 251–259 (2013).
2. S. G. Potts et al., Trends Ecol. Evol. 25, 345–353 (2010).
3. R. Winfree, R. Aguilar, D. P. Vázquez, G. LeBuhn, M. A. Aizen,
Ecology 90, 2068–2076 (2009).
4. M. Henry et al., Science 336, 348–350 (2012).
5. P. R. Whitehorn, S. O’Connor, F. L. Wäckers, D. Goulson,
Science 336, 351–352 (2012).
6. J. E. Cresswell et al., Zoology 115, 365–371 (2012).
7. B. A. Woodcock et al., Nat. Commun. 7, 12459 (2016).
8. M. Rundlöf et al., Nature 521,77–80 (2015).
9. G. E. Budge et al., Sci. Rep. 5, 12574 (2015).
10. D. Goulson, PeerJ 3, e854 (2015).
11. C. Sandro ck et al., Agric. For. Entomol. 16,119–128 (2 014).
12. G. C. Cutler, C. D. Scott-Dupree, M. Sultan, A. D. McFarlane,
L. Brewer, PeerJ 2, e652 (2014).
13. G. Christopher Cutler, C. D. Scott-Dupree, Ecotoxicology 23,
1755–1763 (2014).
14. B. A. Woodcock et al., J. Appl. Ecol. 53, 1358–1362 (2016).
15. E. Pilling, P. Campbell, M. Coulson, N. Ruddle, I. Tornier,
PLOS ONE 8, e77193 (2013).
16. M. Henry et al., Proc. R. Soc. B Biol. Sci. 282, 2015.2110 (2015).
17. A. Jones, P. Harrington, G. Turnbull, Pest Manag. Sci. 70,
1780–1784 (2014).
18. C. Botías et al., Environ. Sci. Technol. 49, 12731–12740 (2015).
19. D. Goulson, J. Appl. Ecol. 50, 977–987 (2013).
20. A. Fairbrother, J. Purdy, T. Anderson, R. Fell, Environ. Toxicol.
Chem. 33, 719–731 (2014).
21. FERA, Neonicotinoid Pesticides and Bees. Report to Syngenta
Ltd. (The Food and Environment Research Agency, UK, 2013).
22. F. Sánchez-Bayo et al., Environ. Int. 89-90,7–11 (2016).
23. C. R. Archer, C. W. W. Pirk, G. A. Wright, S. W. Nicolson, Funct.
Ecol. 28, 913–923 (2014).
ACKNO WLED GMEN TS
Data are in supplementary materials. Funded by Syngenta Ltd.
and Bayer CropScience (P. Campbell, M. Miles, C. Maus, D. Holah,
M. Coulson). Wild pollinator work supported by NERC CEH
National Capability funding (NEC05829). Thanks to K. Jaekel,
P.Fisher,M.Nowakowski,R.Hails,P.Scrimshaw,N.Mitschunas,
P. Nuttall, M. McCracken, S. Ball, J. Webb, B. Sutherland,
R. Freckleton, T. Tscharntke, J. Memmott, K. Norris, B. Raffa,
and D. Vaskor.
SUPPLEMENTARY MATERIALS
www.sciencemag.org/content/356/6345/1393/suppl/DC1
Materials and Methods
Figs. S1 and S2
Tables S1 to 12
References (24–32)
7 December 2016; accepted 22 May 2017
10.1126/science.aaa1190
Woodcock et al., Science 356, 1393–1395 (2017) 30 June 2017 3of3
RESEARCH |REPORT
on June 29, 2017 http://science.sciencemag.org/Downloaded from

Country-specific effects of neonicotinoid pesticides on honey bees and wild bees
J. Peyton, S. Hulmes, L. Hulmes, M. Sárospataki, C. Saure, M. Edwards, E. Genersch, S. Knäbe and R. F. Pywell
B. A. Woodcock, J. M. Bullock, R. F. Shore, M. S. Heard, M. G. Pereira, J. Redhead, L. Ridding, H. Dean, D. Sleep, P. Henrys,
DOI: 10.1126/science.aaa1190
(6345), 1393-1395.356Science
, this issue p. 1395, p. 1393; see also p. 1331Science
affect pollinator health under realistic agricultural conditions.
and colony reproduction in both honeybees and wild bees. These field results confirm that neonicotinoids negatively
on rapeseed in Europe, find that neonicotinoid exposure from several nontarget sources reduces overwintering success
, in a multicounty experimentet al.especially when coexposed to a commonly used agrochemical fungicide. Woodcock
exposed to neonicotinoids for 3 to 4 months via nontarget pollen, resulting in decreased survival and immune responses,
find that bees near corn crops areet al.neonicotinoids diminish bee health (see the Perspective by Kerr). Tsvetkov
environmental conditions. Two studies, conducted on different crops and on two continents, now substantiate that
However, lingering criticism was that the studies did not represent field-realistic levels of the chemicals or prevailing
Early studies of the impacts of neonicotinoid insecticides on insect pollinators indicated considerable harm.
Damage confirmed
ARTICLE TOOLS http://science.sciencemag.org/content/356/6345/1393
PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions
Terms of ServiceUse of this article is subject to the
is a registered trademark of AAAS.Science
licensee American Association for the Advancement of Science. No claim to original U.S. Government Works. The title
Science, 1200 New York Avenue NW, Washington, DC 20005. 2017 © The Authors, some rights reserved; exclusive
(print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience
on June 29, 2017 http://science.sciencemag.org/Downloaded from