A restatement of recent advances in the natural science evidence base concerning neonicotinoid insecticides and insect pollinators

A summary is provided of recent advances in the natural science evidence base concerning the effects of neonicotinoid insecticides on insect pollinators in a format (a ‘restatement') intended to be accessible to informed but not expert policymakers and stakeholders. Important new studies have been published since our recent review of this field (Godfray et al. 2014 Proc. R. Soc. B 281, 20140558. (doi:10.1098/rspb.2014.0558)) and the subject continues to be an area of very active research and high policy relevance.


Introduction
Neonicotinoid insecticides were introduced in the 1990s and their market share quickly expanded to approximately a third of the global insecticide total by value. They are used in different ways, but particularly as seed treatments where the chemical is absorbed by the growing plant and is distributed through all tissues at concentrations that can kill insect herbivores. However, neonicotinoids are also translocated to nectar and pollen where they can be consumed by pollinating insects. Numbers of pollinators have declined in agricultural landscapes and there is concern that the introduction and widespread use of neonicotinoids is partly responsible.
In December 2013, the European Union (EU) instigated partial restrictions on the use of neonicotinoid insecticides on crops that might be used as food by pollinating insects. This move is strongly opposed by many in the farming community and there has been a vigorous debate focusing on the scientific evidence that neonicotinoids harm pollinators, as well as the environmental and economic costs and benefits of the restrictions.
To try to assist the debate we produced a 'restatement' of the underlying natural science evidence base in a form that was intended to be accessible to informed but not expert policymakers and stakeholders [1]. Our avowed aim was to be as policy-neutral as possible while acknowledging that perfect neutrality is never achievable. The restatement was published as an appendix to a short paper in this journal accompanied by an extensive annotated bibliography as the electronic supplementary material.
Since the restatement was published the debate about restricting neonicotinoid use has continued unabated. applied for '120-day derogations' from the restrictions in several European countries (see electronic supplementary material, paragraph A.2) on the grounds of lack of alternative pest-management options, moves that have been criticized by environmental non-governmental organizations. The EU is committed to review the restrictions in 2015-2016 and through the independent European Food Safety Authority opened a call for evidence (closing 30 September 2015; http://www.efsa. europa.eu/en/data/call/150522). Much new research has been published on the topic (we review over 80 studies here) including the largest replicated field study to date [2].
Despite the relatively short time since the restatement was published we provide here an update in the same format. We do this (i) because of the significant advances in the science; (ii) because of the continuing need for policyneutral evidence summaries in this highly contested area, especially in the run up to the review of the EU restrictions; and (iii) in response to a request to do so by the UK Government Chief Scientific Adviser.

Methods
The literature on pollinators and neonicotinoids published since our restatement was completed was reviewed and a first draft evidence summary produced by a subset of the authors. All authors reviewed and revised the document, and agreed on the categorizing of the different evidence components using the same scheme we adopted earlier, and which is explained in paragraph A2 of the restatement update (appendix A). The second draft was sent to a series of stakeholders or stakeholder groups including scientists involved in pollinator research, representatives of the farming and agrochemical industries, non-governmental organizations concerned with the environment and conservation, and UK government departments and statutory bodies responsible for pollinator policy. The document was revised in the light of much helpful feedback from over 40 stakeholders (see acknowledgements). Though many groups were consulted, the project was conducted completely independent of any stakeholder and was funded by the Oxford Martin School ( part of the University of Oxford).

Results
The update to the restatement of the natural science evidence base concerning neonicotinoid insecticides and insect pollinators is given in appendix A, with an annotated bibliography provided as the electronic supplementary material.

Discussion
The new evidence and evidence syntheses that have been published in the last 18 months (between February 2014 and August 2015) significantly advance our understanding of the effects of neonicotinoids on insect pollinators. Nevertheless, major gaps in our understanding remain, and different policy conclusions can be drawn depending on the weight one accords to important (but not definitive) science findings and the weightings given to the economic and other interests of different stakeholders. The natural science evidence base places constraints on policies that claim to be consistent with the science, but does not specify a single course of action.
We also raise an issue here that arises from our original study but is not directly relevant to the evidence base on the effects of neonicotinoids on pollinators. In introducing the subject we wrote 'Neonicotinoid insecticides are a highly effective tool to reduce crop yield losses due to insect pests', and in the restatement itself listed a small number of papers in the scientific literature to support this statement [1]. It has been pointed out that some of these papers were funded by industry and that there are other studies that have recorded no benefits of neonicotinoid use (e.g. [3]).
The efficacy of neonicotinoids is clearly an important issue, and we believe few would doubt that in some circumstances (combinations of crops, pests and locales) they are highly effective and in other circumstances they do not justify the costs of their purchase. We did not attempt to review this subject and should have been more careful to say we were not commenting on efficacy per se.
Though a meta-analysis of efficacy would be very informative it would also be very difficult. Efficacy studies are largely conducted by industry, the sector that benefits most from the data, and are not the type of science usually funded by public organizations. Typically, the studies are not published in the peer-reviewed literature (though they are often made available to regulators) and some are kept confidential for commercial reasons. Efficacy trials are expensive and it seems unlikely that they will ever be publicly funded at scale. It is an interesting topic for debate whether industry would benefit in the long run from placing more of its data in the public domain as well as putting in place measures to increase public confidence in studies they fund themselves. The recent movement in the pharmaceutical sector to set up trial registries (see https://clinicaltrials. gov/ct2/home and https://www.clinicaltrialsregister.eu) provides a model for how the latter might be achieved.  (a) Introduction and aims A1 This document is an update to our previous 'restatement' of the natural science evidence base concerning neonicotinoid insecticides and insect pollinators. It does not repeat evidence presented earlier and concentrates on material published between February 2014 and August 2015. It is arranged in the same six sections (a-g). Paragraphs are numbered A1, A2, etc. and the symbol § (e.g. §16) is used to indicate the paragraph number in the original document [1], where the same subject was treated. A2 ( §1) The restrictions on the use of certain neonicotinoids as seed coatings on crops attractive to pollinating bees will have been in place for two years in December 2015. The Commission has now mandated the European Food Safety Authority to collate relevant data as the first step in the review of these measures. Industry groups in a number of EU countries have successfully applied for '120-day' derogations to use restricted neonicotinoids in defined geographical areas on the grounds of the absence of viable alternatives (see also A33). The province of Ontario in Canada is introducing restrictions on neonicotinoid use on maize (corn) and soy from July 2015. We are not aware of other equivalent measures that have been introduced elsewhere in the world. A3 ( §2) As before the authors provide a consensus judgement on the nature of the different evidence components. We use the following descriptions, which explicitly are not a ranking, indicated by abbreviated codes. Statements are considered to be supported by:  (c) Exposure of pollinators to neonicotinoid insecticides A5 ( § §13-14) As in the first version of the restatement we consider concentrations of neonicotinoids in pollen and nectar of the order of 2 -6 ng g 21 to be typical of those that a pollinator might encounter when foraging on seed-treated crops. Statements about low or high concentrations are made relative to this benchmark, though we acknowledge there will be variation around these figures and that this benchmark involves an element of expert judgment. A wide-ranging review of how neonicotinoids, introduced as seed coatings, may move through and persist in the environment has been published. (e) Neonicotinoid residues observed in pollen, nectar and wax in the field A16 ( § §29-31) New data, data compilations and reanalyses of earlier data continue to show that neonicotinoid residues can be detected in pollen and nectar collected by pollinating bees. However, these data are highly variable, making general inference hard. [S upp_ev ] Incidences of high neonicotinoid residues that would almost certainly cause acute toxic effects in honeybees and bumblebees do occur, but not commonly. [E xp_op ] A17 ( §32) Summary (unchanged from earlier restatement). Neonicotinoids can be detected in wild pollinators as well as honeybee and bumblebee colonies but data are relatively few and restricted to a limited number of species. Studies to date have found low levels of residues in surveys of honeybees and honeybee products. Observed residues in pollinating bees and the products they collect will depend critically on details of spatial and temporal sampling relative to crop treatment and flowering. [E xp_op ] (f ) Experiments conducted in the field A18 ( §33) As before, we give separate, detailed treatment to 'semi-field' studies where insects are exposed by the experimenter to a known dose of insecticide and then allowed to forage in the environment, and 'true field' studies involving exposure to neonicotinoids as applied in actual farm landscapes. There is continuing debate about the relevance of the doses and application methods used in semi-field studies, and about the relevance of methodologies used in true field experiments. [E xp_op ] A19 Dively et al. [4] provided replicate colonies of honeybees over a 12-week period with supplemental pollen paste diets containing imidacloprid at three concentrations (5, 20 and 100 ng g 21 ) with a fourth control treatment. Experiments were conducted in 2009 (10 replicates per treatment) and 2010 (seven replicates). They found no effect on foraging performance or colony health in the short term but over a longer period, colonies exposed to neonicotinoids were more likely to lose queens, suffer higher overwintering mortality and have greater Varroa infestations, though these effects were only statistically significant at the high (20) and very high (100 ng g 21 ) concentrations. [D ata ] The authors concluded that their results did not suggest that neonicotinoids were a sole cause of colony collapse. [P rojns ] A20 Lu et al. [5]. Honeybee colonies were fed with syrup containing high concentrations of imidacloprid or clothianidin, or with no added insecticide, for a 13week period from July to September (in Massachusetts, USA  [6] reported how the day-to-day foraging patterns of 259 bumblebee (B. terrestris) workers from 40 colonies were affected by individual or combined exposure to the neonicotinoid imidacloprid and the pyrethroid l-cyhalothrin. These data, and results presented by Gill et al. [7], were collected in the same experiment conducted in 2011 (see §37). Exposure to imidacloprid concentrations (10 ng g 21 ) towards the high end of what is typically observed in the field led to acute and chronic effects on individual foraging behaviour (although actual imidacloprid consumption by individual workers will have been diluted by foraging from untreated floral sources in the field; see §37). Whereas individual bumblebee foraging efficiency normally improves with experience, this did not occur in individuals exposed to imidacloprid. [D ata ] Evidence was found that the insecticide affected the pollinators' preference for different flowers as sources of pollen.
[S upp_ev ] A22 Moffat et al. [8]. Bumblebee (B. terrestris) colonies were provided with syrup containing low doses (approx. 2 ng g 21 ) of imidacloprid and placed in the field in a non-intensive agricultural location for 43-48 days. By most measures, the neonicotinoid had a significantly negative effect on colony performance compared with controls. [D ata ] A23 ( §38) A true field experiment by Thompson et al. [9] was originally interpreted as showing no effects of two neonicotinoids on bumblebee (B. terrestris) colony performance. The experiment placed multiple colonies adjacent to oilseed rape fields that had received different insecticide treatments (but with no replication at the field level). A colony-level reanalysis of the data by Goulson [10] showed a significant relationship between neonicotinoid concentration and performance: colonies with higher concentrations of thiamethoxam or clothianidin in nectar, or thiamethoxam in pollen stores, produced significantly fewer new queens. Because rspb.royalsocietypublishing.org Proc. R. Soc. B 282: 20151821 exposure was not manipulated at the colony level, this study should be considered as correlational rather than experimental. [P rojns ] A24 Cutler et al. [11]. Ten 2-hectare plots in Southern Ontario, Canada, were planted with oilseed rape, half of which were planted with seed treated with the neonicotinoid clothianidin with the other half controls. During peak flowering, four honeybee hives were placed in the centre of each field for two weeks before being moved to a site away from insecticide treated crops. Pollen from hives in treated fields had higher concentrations of clothianidin (which were non-zero in controls) but no effects of the insecticide were found for a variety of honeybee colony growth or overwintering metrics. [D ata ] A25 Cutler & Scott-Dupree [12]. Bumblebee (Bombus impatiens) colonies were placed beside four fields planted with organic maize and four with maize grown from neonicotinoid-coated seeds in Southern Ontario, Canada. The study took place on commercial farms and organic and non-organic maize produced pollen at different times. No differences were found in ten measures of colony health, except that colonies by treated fields had significantly fewer workers (which the authors attributed to an effect of crop development time). Analysis of collected pollen showed maize was a very small component (0-2%) of these bumblebees' diets. [D ata ] A26 Rundlöf et al. [2]. In southern Sweden eight pairs of spring-sown oilseed rape fields were chosen with one of each pair grown from clothianidin coated seeds and the other from non-coated seeds. The seed treatment used, as recommended by the manufacturer, led to higher concentrations of clothianidin in pollen than is normally observed in this crop. Treated fields had lower densities of solitary bees and bumblebees, and poorer bumblebee (B. terrestris) colony growth and queen production (all comparisons statistically significant). Solitary bees (Osmia bicornis) placed adjacent to treated fields all disappeared while a small but significantly higher number nested beside control fields. The experiment detected no significant effects on measures of honeybee colony strength. Wildflowers, to which pollinators may also be exposed, had higher levels of clothianidin when growing in uncultivated land around treated compared to untreated crops. [D ata ] A27 ( §40) Summary. Evidence continues to accumulate from semi-field experiments that sublethal exposure to neonicotinoid insecticides, chiefly but not exclusively at the high end of what is likely to be experienced in the environment, can affect foraging and other behaviours in the field. Several true field studies have reported no effect of exposure to neonicotinoid-treated crops on honeybee colony performance, but the first large-scale study of the exposure of bumblebees (see A26) [13] collected data on honeybee colony in-season loss and neonicotinoid use from nine regions of the UK every other year from 2000 to 2010. Controlling for region (but not year) they find a weak but significant correlation between colony loss and imidacloprid use, but not total neonicotinoid use. We found that this effect was due to a correlation between annual average colony loss and imidacloprid use. Imidacloprid use peaked mid-decade (after which it was replaced by thiamexotham and clothianidin) and there was a tendency for honeybee losses to be higher at this time. Because other factors not included in the analysis may show similar annual patterns, and because of statistical issues with the analysis (see Annotated Bibliography), the correlation of honeybee colony losses with imidacloprid use, and the lack of correlation with total neonicotinoid use, should be treated with great caution.  (1921 -1950 versus 1983-2012) in species richness and composition of bees and wasps in the UK suggests land use and management changes are the most important historical drivers with major faunal losses occurring early in the twentieth century. Any effects of changes in pesticide use over recent decades are unlikely to be picked up by these analyses. An analysis of the historical shifts in the ranges of European and North American bumblebees showed that they have failed to track climate warming at their northern range limits, while southern range limits have contracted. These shifts were independent of changes in land use (both continents) and pesticides application, including neonicotinoids (in North America only; pesticide data was unavailable for Europe). This study only assessed changes in species range distributions, and so any impacts of pesticides on population density or diversity at finer habitat or landscape scales would not be identified. A34 ( §46) Summary. There still remain major gaps in our understanding of how pollinator colony-level (for social bees) and population processes may dampen or amplify the lethal or sublethal effects of neonicotinoid exposure and their effects on pollination services; as well as how farmers might change their agronomic practices in response to restrictions on neonicotinoid use and the resulting positive or negative effects on pollinators and pollination. While these areas continue to be researched there is still a limited evidence base to guide policymakers on how pollinator populations will be affected by neonicotinoid use or how agriculture will respond to neonicotinoid usage restrictions.