A horizon scan of future threats and opportunities for pollinators and pollination

Background. Pollinators, which provide the agriculturally and ecologically essential service of pollination, are under threat at a global scale. Habitat loss and homogenisation, pesticides, parasites and pathogens, invasive species, and climate change have been identified as past and current threats to pollinators. Actions to mitigate these threats, e.g., agri-environment schemes and pesticide-use moratoriums, exist, but have largely been applied post-hoc. However, future sustainability of pollinators and the service they provide requires anticipation of potential threats and opportunities before they occur, enabling timely implementation of policy and practice to prevent, rather than mitigate, further pollinator declines. Methods.Using a horizon scanning approach we identified issues that are likely to impact pollinators, either positively or negatively, over the coming three decades. Results.Our analysis highlights six high priority, and nine secondary issues. High priorities are: (1) corporate control of global agriculture, (2) novel systemic pesticides, (3) novel RNA viruses, (4) the development of new managed pollinators, (5) more frequent heatwaves and drought under climate change, and (6) the potential positive impact of reduced chemical use on pollinators in non-agricultural settings. Discussion. While current pollinator management approaches are largely driven by mitigating past impacts, we present opportunities for pre-emptive practice, legislation, and policy to sustainably manage pollinators for future generations.


INTRODUCTION
Pollinators provide the key ecosystem service of pollination to agricultural crops and wild plants, with 35% of global crop production relying to some degree on pollination (Klein et al., 2007), along with more than 85% of wild flowering plants (Ollerton, Winfree & Tarrant, 2011). Consequently, declines in pollinators, which are occurring across the globe (Potts et al., 2010), may pose a significant threat to human and natural well-being. A suite of drivers, including habitat loss and homogenization (Kennedy et al., 2013), pesticides (Godfray et al., 2015), parasites and pathogens (e.g., Fürst et al., 2014;McMahon et al., 2015;Wilfert et al., 2016), invasive species (Stout & Morales, 2009), and climate change (e.g., Kerr et al., 2015) have been identified as past and current threats to pollinators (Vanbergen & The Insect Pollinator Initiative, 2013). Some actions to mitigate these threats, e.g., agri-environment schemes that provide forage and nesting resources (Batáry et al., 2015) and pesticide-use moratoriums to mitigate the potential impact of pesticides (Dicks, 2013), exist, but they have largely been applied post-hoc. While there is some evidence that such approaches might be mitigating pollinator losses (e.g., Carvalheiro et al., 2013), future sustainability of pollinators and the service they provide requires anticipation of potential threats and opportunities before they occur, enabling timely implementation of policy and practice to prevent, rather than mitigate, further pollinator declines.
One approach that can be used to anticipate future threats and opportunities for pollinators is the process of horizon scanning. Horizon scanning, a systematic technique to identify future threats or opportunities, is an important policy tool used in government and business to manage and proactively respond to upcoming threats and opportunities (Cook et al., 2014). In the last decade, horizon scanning has increasingly been applied to support environmental decision-making and inform policy and research on specific issues such as invasive species risk (Roy et al., 2014), management of particular geographic regions (Kennicutt et al., 2014) or threats to particular taxa (Fox et al., 2015). Proactive responses that pre-empt environmental risks are likely to be cheaper in the long term than reactive responses (e.g., Drechsler, Eppink & Wätzold, 2011) and potentially enable avoidance of substantial costs (Hulme et al., 2009).
Pollinator decline is one of the highest profile global environmental issues of the 21st century, as demonstrated through its selection by the International Platform on Biodiversity and Ecosystem Services (IPBES) as the subject of its first major assessment report (Gilbert, 2014). With governments around the world focused on this issue, and several producing national policies which largely focus around past and current threats, it is timely to identify forthcoming impacts on pollinators, both positive and negative, which may not yet be fully recognised by policy or research. Here we used a global horizon scanning team to identify potential future threats and opportunities for pollinators.

METHODS
We followed a Horizon Scanning approach based on the Delphi method (Sutherland et al., 2016). The same approach has been used since 2010 to generate global horizon scans for conservation (Sutherland et al., 2016), and thus it provides a reliable and accepted methodology. The exercise was carried out by a core group of 17 pollinator experts (the authors), balanced across area of expertise and geographic knowledge. Experts were drawn from NGOs, research institutes, and universities. One member from the agrochemical industry accepted, but withdrew before the first stage of the process (see below) was completed. Table 1 shows how the group maps on to the two criteria of expertise and geography, and demonstrates strong coverage within the horizon-scanning group.

Selecting issues
Each person in the team consulted their networks and collected up to five potential horizon issues for consideration; 55 people (see 'Acknowledgements'), in addition to the 17 experts, were consulted during this process. We searched for issues that were poorly known and considered likely to have a substantial impact on wild or managed pollinators (including insects, birds, mammals, and reptiles), either positive or negative, during the next one to 30 years. A 'substantial' impact could have a high magnitude, or take place over a large area, or both.
A long list of 60 issues, with associated references, was compiled (Table 2, Table S1) and sent to all core participants for a first round of anonymous scoring. Where the same issues had been identified by more than one member of the core group, these issues were grouped as one. Participants scored each issue from 1 (well known, unlikely to have substantial impact on pollinators) to 1,000 (poorly known, very likely to have substantial impact on pollinators). From these scores, we produced a ranked list of topics for each participant (the highest scored issue was given a rank of 1), and calculated the median rank for each topic (Table 2). Each person also stated whether they had previously heard of each issue or not.

Refining to a shortlist of priorities
The 28 issues with the lowest median ranks were retained, and participants had a chance to retain others they felt strongly should not be dismissed at this stage (no issues were brought back). Two participants were assigned to each of the 28 retained issues to research its technical details, likelihood, and potential impacts. These were not the same people who had suggested the issue.
Ten of the participants convened in Paola, Malta, in November 2015. We discussed each of the 28 issues in turn, with the constraint that the individual who suggested an issue was not the first to contribute to its discussion. All participants could see the median ranks and the percentage of the group who had heard of each issue (given as 'originality value' in Table 2), from round 1. Some issues were modified during this discussion. After each issue was discussed, participants independently and privately scored between 1 and 1,000 as previously described. The 'originality value' was used as a guide for scoring, although we were aware that, as the participants were all pollinator experts, it was unlikely to represent familiarity with these issues in the wider policy and research communities.
The remaining seven participants unable to attend the meeting took part in the process remotely, by submitting their research notes for issues they had been assigned (these were provided to each participant in printed form), and re-scoring independently after reading a detailed written account of the issues discussed.
The list of 15 issues presented here comprises those with the highest median ranks from the second round of scoring (Table 3). They are divided into High Priority and Secondary Priority issues (HPI, and SPI, respectively) because there was a clear break in the rankings among the top 15 issues, between the top six and the following nine. One issue (''Sanitary and genetic issues raised by international trade and globalization'') was removed from the final priority list despite having been ranked joint 13th by its median rank. While clearly important, the group agreed in the final stage that this was a current, well-known issue, and not an emerging issue on the horizon.

RESULTS
Using a modified Delphi process, we identified 60 initial issues of interest (Table 2, Table  S1), which reduced to six high priority issues and nine secondary priority issues (Table 3). These issues can be partially mapped onto areas previously identified as being important causes of pollinator decline, e.g., agricultural practices ( Fig. 1, Table 4). However, the issues we identified are largely distinct from past and current drivers of pollinator abundance, and Table 2 The results of the first round of voting on the horizon-scanning issues. Each issue is listed with its median rank (low rank = most strongly voted for as a horizon issue) and its originality score (0 = not heard of, 1 = completely familiar)(see Methods for details). The number in the left column is simply the order in which issues were compiled.    require distinct policy and practices to minimize the threat and maximise the opportunities they present (Table 4). As is standard for a horizon scanning process, the identified issues are presented in rank order below, with the highest ranked issue first.

HPI-1: corporate control of agriculture at the global scale
Consolidation in agri-food industries has led to unprecedented control over land access, land use and agricultural practices by a small number of companies (Worldwatch Institute, 2013). A newer trend is transnational land deals for crop production, which now occupy over 40 million hectares (http://www.landmatrix.org/en/), including areas of Brazil for soybean export to China, and West Africa for rubber and palm oil. Agri-food industries operating at scale tend to promote homogeneous production systems, which is rapidly changing landscapes, especially in the southern hemisphere (Laurance et al., 2014) in a way that could substantially reduce the diversity and abundance of native pollinators. From an opportunity perspective, large-scale control over agricultural practices could, under appropriate management practices, enable sustainable pollinator management to optimize pollination with respect to consumer demands.

HPI-2: sulfoximine, a novel systemic class of insecticides
Sulfoximines are a new class of insecticide that resemble neonicotinoids in mode of action, yet differ sufficiently to prevent cross-resistance (Sparks et al., 2013). The first sulfoximine to be marketed is Sulfoxaflor. In spray formulation, it is rapidly being registered for widespread crop use in countries across the globe, to combat rising resistance to neonicotinoids (Bass et al., 2015). If, as is likely, sulfoximines are next registered as seed treatments, they may soon replace neonicotinoids over vast geographic areas (Simon-Delso et al., 2015). Neonicotinoids have sub-lethal effects on wild pollinators (e.g., Rundlöff et al., 2015), which may be generated through impacts on neural processes and immunity (e.g., Di Prisco et al., 2013), but those of sulfoximines have not been studied. Seed treatments are particularly likely to generate sub-lethal effects broadly, since they are applied prophylactically, rather Table 4 The relationship between horizon scanning issues, past problems and actions, and future responses. The relationship between responses to current or past issues (column 1), identified horizon issues grouped by overarching driver (column 2), and potential pro-active responses to these issues (column 3).   than sprayed at specific times (where usage may be modified to reduce or avoid impacts on pollinators). Thus, the rapid proliferation of a new systemic, neuroactive insecticide without sufficient testing for sub-lethal effects is a grave concern, particularly if new formulations such as seed treatments arise.

HPI-3: new emerging RNA viruses
Emerging infectious diseases-some transmitted by exotic ectoparasitic Varroa destructor mites-are considered major causes of colony decline for the most abundant commercial pollinator, the Western honey bee (Apis mellifera). Such diseases are shared with, and likely spill over into, wild pollinators (Fürst et al., 2014). Chief among them are RNA viruses, whose high mutation and recombination rates make them particularly likely to cross host backgrounds (Manley, Boots & Wilfert, 2015). There is substantial risk of novel viral diseases emerging with elevated virulence, more efficient transmission and broad host range. The threat to both wild and managed pollinators is exacerbated by transport of managed pollinators to new locations, which may bring RNA viruses into contact with novel vectors (Roberts, Anderson & Tay, 2015).

HPI-4: increased diversity of managed pollinator species
Managed pollinators can replace or augment wild pollinators, but currently very few species are employed-most commonly Apis mellifera and, to a lesser extent, some bumblebees, stingless bees, and solitary bees (Free, 1993;Delaplane & Mayer, 2000). Diversifying the species managed for pollination could enhance pollination in crops that either require specialist pollinators or do not receive optimal service from existing managed species; provide insurance against perturbations in the supply of existing species; and enable use of native species in regions where existing managed species are not native. It also represents a business opportunity. Developing alternative managed pollinators requires biological and technical knowledge about the focal species, to ensure reliable supplies for growers. Risks associated with deploying new species, including parasite transmission, competition with local pollinators, introgression with the local gene pool, and ecosystem level impacts (Stout & Morales, 2009), require proactive risk assessment and regulation.

HPI-5: effects of extreme weather events under climate change
Effects of gradually changing climate on pollinators are increasingly well characterised, while the impacts of extreme events are poorly understood. Projected increases in frequency, magnitude, or intensity of, e.g., heatwaves and droughts are very likely across substantial parts of the globe (IPCC Summary for Policymakers, 2013). Heatwaves and droughts can affect pollinators directly, or indirectly by generating resource bottlenecks (Takkis et al., 2015). There is evidence that such weather patterns can lead to local extinction of pollinators (Rasmont & Iserbyt, 2012;Oliver et al., 2015) potentially leading to the breakdown of plantpollinator relationships (Harrison, 2000). Greater knowledge of the relative importance of different extreme events is urgently needed to future-proof pollinator-friendly habitat management.

HPI-6: positive effects of reduced chemical use on pollinators in non-agricultural settings
Chemicals that have negative impacts on pollinators are widely used in urban and suburban areas, and in the wider landscape (e.g., golf courses). Recent recognition of the value of such areas for pollinators (Baldock et al., 2015) provides an opportunity to increase awareness of chemical use, and drive successful 'reduce and replace' campaigns. The potential for large-scale reduction in chemical use across ever-growing urban and suburban areas could have significant positive impacts on insect pollinators (Muratet & Fontaine, 2015).

SPI-1: potential non-target effects of nanoparticle pesticides on crop visiting insect pollinators
Nanoparticle pesticide use is rapidly expanding (Sekhon, 2014), yet non-target effects have not been evaluated, and this technology may evade existing pesticide regulatory processes. Though major knowledge gaps exist, nanoparticle pesticides may adversely affect crop-visiting pollinators.

SPI-2: increasing use of fungicides
Fungicide use is expected to increase with higher summer rainfall, which has been predicted for many regions under climate change scenarios (IPCC Summary for Policymakers, 2013).
Current risk assessments for fungicides fail to capture sub-lethal and indirect impacts (e.g., on bee gut flora and fungi in pollen stores, synergies between fungicides and insecticides, and elevated susceptibility to disease (Pettis et al., 2013)).

SPI-3: risks and opportunities of cutting pollinators out of food production
Plant breeding technology can produce crop varieties that do not require biotic pollination (Mazzucato et al., 2015). Wide uptake of this technology could stabilize yields and reduce costs, but could further entrench the pollinator crisis by removing the imperative for pollinator protection and threatening the viability of remaining pollinator-dependent crops.

SPI-4: impacts of IPBES pollinators assessment
The Intergovernmental Platform on Biodiversity and Ecosystem Services' 2016 global assessment ''Pollinators, Pollination and Food Production'' (IPBES, 2016) is a critical evaluation of evidence on the status, value and threats to pollinators and pollination worldwide. It could galvanise or inform substantial new actions by governments, practitioners and researchers.

SPI-5: pollinators as pathways for pathogens
While visiting flowers, pollinators can also transmit plant and pollinator diseases (McArt et al., 2014). Crop industries concerned about pollinator-mediated disease spread could enact restrictions on movements of managed pollinators, providing economic incentive to prioritise the use of local wild pollinators.

SPI-6: reductions in pollinator species richness may drive epidemics
Infectious disease transmission involves interactions among networks of species. The inverse relationship between host species diversity and disease transmission (Civitello et al., 2015) could drive disease epidemics as pollinator diversity declines.

SPI-7: the impact of non-native managed pollinators on native bee communities in Asia
The commercial importation of European Bombus terrestris (He et al., 2013) is very likely to negatively impact bumblebee communities in China, the global centre of bumblebee species diversity, as it has in other areas (e.g., Morales et al., 2013). The eight native honey bee species are increasingly likely to be negatively impacted by commercial import of A. mellifera and other managed bees.

SPI-8: climate change: altering pathogen epidemiology to the detriment of pollinators
In addition to direct and indirect impacts on pollinators, climate change may alter pollinator susceptibility to disease or enhance environmental transmission of pathogens (Natsopoulou et al., 2015). This may change pathogen range, prevalence, epidemiology, and the impact of emerging infectious disease agents on pollinators and pollination.

SPI-9: destruction of bat roosts worldwide
Globally, bats face increasing threats (Regan et al., 2015) due to habitat loss, roost destruction, hunting and persecution. As human activities expand into tropical forest areas, destruction of roost sites will increase, while culling is an increasing threat. Bats are important pollinators in tropical forests, savannas, deserts, and for cultivated plants (e.g., agave). The consequences of precipitous declines in bat pollination have not been assessed.

DISCUSSION
Here we have identified a series of horizon issues, both positive and negative, for pollinators. Interestingly, while some of these have connections to previous causes of pollinator declines, and can be linked to over-arching drivers, such as agriculture and climate change, the policy and practice needed to minimize future threats and maximise future opportunities are largely distinct from current best practice in pollinator conservation. In addition to their direct effects, the horizon issues identified in this study may also interact to positively and negatively impact pollinators. For example, extreme weather events driven by climate change are likely to influence corporate agriculture, its location, and its spread across the globe, whilst at the same time calling for agricultural practices that develop or support locally specialized pollinators. Such interactions deserve further investigation.
Horizon-scanning projects are, of necessity, limited by the panel make-up and the range of sources they can draw on. We specifically invited panel members from all major geographical regions, and across government research institutes, industry, NGOs, and universities, in order to maximise the breadth of knowledge and experience in our panel. To increase this breadth even further, panel members consulted a wide range of experts. Nevertheless, we acknowledge that an alternative panel make-up could have arrived at a different ordering, or selection of issues. In addition, our selection of issues should not be taken as static. Horizon scanning detects possible future changes about which there is little current evidence (sometimes known as 'weak signals ';Cook et al., 2014). As the future unfolds, new technologies and global change phenomena arise, and so the process should be repeated as an ongoing part of policy and research planning.
Future-proofing pollinators is urgently required, in a world where demand for pollination services is rising at the same time as threats are increasing (Lautenbach et al., 2012;Potts et al., 2010;Vanbergen & The Insect Pollinator Initiative, 2013). Many of the issues we identified are new developments relating to current problems for pollinators, but some are potential opportunities, or entirely new potential threats (Fig. 1). As indicated in Table 3, for some issues the appropriate policy responses or actions to mitigate negative impacts might be different from those currently discussed or enacted. For example, methods of pollinator management may be needed to control the spread of both plant and insect diseases in future, especially if the number of managed pollinator species, and the distances they are moved, increases. Legislation for pesticide development urgently needs to incorporate chronic and interactive impacts and proper field trials for future pesticides. Early identification of such issues provides the opportunity to develop policies and practices to limit negative impacts, or to take advantage of potential positive impacts (Table 3).
While all horizon-scanning exercises are limited in their outputs, we believe we have identified current key issues that should be the focus of conservation practitioners, industry, and policy-makers if we are to maintain and benefit from a functional pollinator assemblage at the global scale in the ensuing decades.