Bee declines driven by combined stress from parasites, pesticides, and lack of flowers

See allHide authors and affiliations

Science  27 Mar 2015:
Vol. 347, Issue 6229, 1255957
DOI: 10.1126/science.1255957

Conserving pollinator services for crops

If pollination fails, ecosystems are eroded and we will lose reliable sources of many critical foodstuffs. Focusing on the pollination services provided by bees, Goulson et al. review the stresses bees are experiencing from climate change, infectious diseases, and insecticides. We can mitigate some of the stress on bees by improving floral resources and adopting quarantine measures, and by surveillance of bee populations. Crucially, we need to resolve the controversy surrounding prophylactic use of pesticides.

Science, this issue 10.1126/science.1255957

Structured Abstract


The species richness of wild bees and other pollinators has declined over the past 50 years, with some species undergoing major declines and a few going extinct. Evidence of the causes of these losses is patchy and incomplete, owing to inadequate monitoring systems. Managed honey bee stocks have also declined in North America and many European countries, although they have increased substantially in China. During this same period, the demand for insect pollination of crops has approximately tripled, and the importance of wild pollinators in providing such services has become increasingly apparent, leading to concern that we may be nearing a “pollination crisis” in which crop yields begin to fall. This has stimulated much-needed research into the causes of bee declines. Habitat loss, which has reduced the abundance and diversity of floral resources and nesting opportunities, has undoubtedly been a major long-term driver through the 20th century and still continues today. In addition, both wild and managed bees have been exposed to a succession of emerging parasites and pathogens that have been accidentally moved around the world by human action. The intensification of agriculture and increasing reliance on pesticides means that pollinators are also chronically exposed to cocktails of agrochemicals. Predicted changes in global climate are likely to further exacerbate such problems in the future.


It has lately become clear that stressors do not act in isolation and that their interactions may be difficult to predict; for example, some pesticides act synergistically rather than additively. Both pesticide exposure and food stress can impair immune responses, rendering bees more susceptible to parasites. It seems certain that chronic exposure to multiple interacting stressors is driving honey bee colony losses and declines of wild pollinators, but the precise combination apparently differs from place to place.

Although the causes of pollinator decline may be complex and subject to disagreement, solutions need not be; taking steps to reduce or remove any of these stresses is likely to benefit pollinator health. Several techniques are available that have been demonstrated to effectively increase floral availability in farmland. Similarly, encouraging gardeners to grow appropriate bee-friendly flowers and to improve management of amenity grasslands can also reduce dietary stress. Retaining or restoring areas of seminatural habitat within farmland will improve nest site availability. A return to the principles of integrated pest management and avoidance of prophylactic use of agrochemicals could greatly decrease exposure of bees to pesticides.


Interactions among agrochemicals and stressors are not addressed by current regulatory procedures, which typically expose well-fed, parasite-free bees to a single pesticide for a short period of time. Devising approaches to study these interactions and incorporating them into the regulatory process poses a major challenge. In the meantime, providing support and advice for farmers in more sustainable farming methods with reduced pesticide use is likely to have broad benefits for farmland biodiversity. Enforcing effective quarantine measures on bee movements to prevent further spread of bee parasites is also vital. Finally, effective monitoring of wild pollinator populations is urgently needed to inform management strategies. Without this, we have no early warning system to tell us how close we may be to a pollination crisis. With a growing human population and rapid growth in global demand for pollination services, we cannot afford to see crop yields begin to fall, and we would be well advised to take preemptive action to ensure that we have adequate pollination services into the future.

Multiple interacting stressors drive bee declines.

Both wild and managed bees are subject to a number of important and interacting stressors. For example, exposure to some fungicides can greatly increase the toxicity of insecticides, whereas exposure to insecticides reduces resistance to diseases. Dietary stresses are likely to reduce the ability of bees to cope with both toxins and pathogens.

Photo credit: DAVE GOULSON


Bees are subject to numerous pressures in the modern world. The abundance and diversity of flowers has declined; bees are chronically exposed to cocktails of agrochemicals, and they are simultaneously exposed to novel parasites accidentally spread by humans. Climate change is likely to exacerbate these problems in the future. Stressors do not act in isolation; for example, pesticide exposure can impair both detoxification mechanisms and immune responses, rendering bees more susceptible to parasites. It seems certain that chronic exposure to multiple interacting stressors is driving honey bee colony losses and declines of wild pollinators, but such interactions are not addressed by current regulatory procedures, and studying these interactions experimentally poses a major challenge. In the meantime, taking steps to reduce stress on bees would seem prudent; incorporating flower-rich habitat into farmland, reducing pesticide use through adopting more sustainable farming methods, and enforcing effective quarantine measures on bee movements are all practical measures that should be adopted. Effective monitoring of wild pollinator populations is urgently needed to inform management strategies into the future.

Insect pollination is vitally important to terrestrial ecosystems and to crop production. The oft-quoted statistics are that 75% of our crop species benefit from insect pollinators (1), which provide a global service worth $215 billion to food production (2). Hence, the chance that we may be facing a “pollination crisis” (3, 4), in which crop yields begin to fall because of inadequate pollination, has generated understandable debate and concern and stimulated much research in recent decades. Nonetheless, knowledge gaps remain substantial with regard to both the extent and causes of pollinator declines. Indeed, for most regions of the globe and for most wild pollinator taxa, we have no data as to whether there have actually been declines. Our best estimates are for numbers of domesticated honey bee colonies, which can be obtained for many countries with varying reliability. These suggest that numbers of managed honey bee colonies have decreased in Europe [25% loss of colonies in central Europe between 1985 and 2005 (5)] and markedly in North America [59% loss of colonies between 1947 and 2005 (6, 7)]. However, overall global stocks actually increased by ~45% between 1961 and 2008 because of a major increase in numbers of hives in countries such as China and Argentina (8). Conversely, there are widespread reports of unusually high rates of honey bee colony loss from many parts of the world, sometimes ascribed to a syndrome known as colony collapse disorder (CCD) (9). It seems that socioeconomic factors such as increasing demand for pollination or honey (10) are at present sufficient to incentivize beekeepers to overcome problems with bee health when examined at a global scale, but not locally in North America and Europe.

If we turn to wild pollinators, the best data available are for bumblebees (11). In Europe, many species have undergone substantial range contractions and localized extinction, with four species going extinct throughout the continent (11, 12) (Fig. 1A). In North America, some formerly abundant and widespread species such as Bombus terricola and B. occidentalis underwent severe declines from the late 1990s onward and now occupy only a small fraction of their former range (Fig. 1B) (13, 14). Bombus franklini, a species formerly found in northern California and Oregon, has not been recorded since 2006 and is presumed extinct. In a study of the bumblebee fauna of Illinois over the past 100 years, Grixti et al. (15) described substantial declines in species diversity, particularly during the period 1940–1960, with the extirpation of four species during the 20th century. In South America, the recent invasion by the European species B. terrestris is causing precipitous declines in the native B. dahlbomii (16). There is some evidence of loss of species richness from lowland areas of Sichuan in China (17, 18) and a few reports of declines in Japan from the mid-1990s onward (19, 20), but elsewhere in the world, few data are available. For the remaining wild bees, data are exceedingly sparse, although they comprise the large majority of the world’s approximately 22,000 bee species. Analysis of historic records suggests that diversity of both bumblebees and other wild bees declined in the United Kingdom, the Netherlands, and Belgium during the 20th century, but that these declines have decelerated since 1990 (21, 22). In surveys in Illinois, 50% of wild bee species went extinct over a 120-year period to 2010 (23). Given that bee diversity has declined in both Europe and the Americas, it is probably reasonable to assume that declines are also occurring elsewhere across the globe.

Fig. 1 Some wild bee species have undergone major range contractions.

(A) The bumblebee Bombus distinguendus in the United Kingdom (data from the National Biodiversity Network, UK). (B) Bombus affinis in North America (map produced by the Xerces Society; list of data providers at [Photo: Dave Goulson, Johanna James-Heinz]

The biggest knowledge gap concerns bee abundance; although we have maps of past and present distributions of bees for some well-studied countries such as the United Kingdom, we have almost no data on how populations have changed over time. Hence, we do not know whether common species such as B. terrestris in Europe or B. impatiens in North America are less abundant than formerly, or whether they are currently in decline. Most pollination is delivered by a small number of these abundant species, which tend to have large distributions. Declines in their abundance would not be detected in distribution maps until they become extinct in parts of their range, which is rather late to introduce conservation measures.

Another way to examine the likelihood or proximity of a pollination crisis is to examine delivery of pollination services. Although global honey bee stocks have increased by ~45%, demand has risen more than supply; the fraction of global crops that require animal pollination has tripled over the same time period (8), making food production more dependent on pollinators than before. It has also emerged that the majority of crop pollination at a global scale is delivered by wild pollinators rather than honey bees. Yields correlate better with wild pollinator abundance than with abundance of honey bees (2426); hence, increasing honey bee numbers alone is unlikely to provide a complete solution to the increasing demand for pollination. Reliance on a single species is also a risky strategy (27). Whereas Aizen et al. (28) concluded from a global analysis of changing crop yields over time that there was not yet any clear evidence that a shortage of pollinators was reducing yield, a subsequent analysis of the same data set by Garibaldi et al. (29) showed that yields of pollinator-dependent crops are more variable, and have increased less, than crops that do not benefit from pollinators, to the extent that a shortage of pollinators is reducing the stability of agricultural food production. In a meta-analysis of 29 studies on diverse crops and contrasting biomes, Garibaldi et al. (30) found that wild pollinator visitation and yields generally drop with increasing distance from natural areas, which suggests that yields on some farms are already affected by inadequate pollination.

To summarize the changes during the past 50 years: Global honey bee stocks have increased while wild bees appear to have declined substantially, as evidenced by data for bumblebees and very scant data for other bee species. The demand for insect pollinators in farming has tripled during the same period. There is clearly no major pollination crisis yet, but there is evidence for localized limitation of crop yield as a result of inadequate pollination.

Drivers of wild bee declines and honey bee colony losses

Habitat loss

Bee declines have been attributed to many factors, some more plausible than others; however, the clear consensus is that loss of habitat has been a long-term contributor to declines (11, 3133). Bees require appropriate floral resources during the adult flight season, which may be short for some solitary species or year-long for social species in tropical environments. They also require undisturbed nest sites, with different species occupying diverse locations (e.g., cavities underground, hollow-stemmed twigs, burrows in the soil, even abandoned snail shells). The conversion of natural and seminatural flower-rich habitat to farmland has been a major driver of long-term declines in bees. For example, in the United Kingdom, approximately 97% of flower-rich grasslands were lost in the 20th century (34), and this has resulted in major range contractions of bee species associated with this habitat, particularly long-tongued bumblebees (11). Declines of similar magnitude have affected the grasslands of North America, although these began in the early 19th century (35).

Urbanization also contributes to loss of natural habitat, but the net effect on bees is less clear. Gardens can support high densities of wild bees, particularly bumblebees and some solitary bee species, but highly urbanized environments have few bees (36, 37), and the building of roads and other infrastructure undoubtedly contributes to the ongoing fragmentation and degradation of habitats. Increased traffic can also cause direct mortality through collisions (38), although the number of bees killed in this way is not known. The planting of road verges and traffic islands with wildflowers is often promoted as a means of boosting urban pollinator populations, but it might increase mortality by vehicle collisions.

Parasites and disease

Bees naturally suffer from a broad range of parasites, parasitoids, and pathogens, the latter including protozoans, fungi, bacteria, and viruses. By far the majority of research has focused on those associated with honey bees and to a lesser extent with bumblebees; very little is known about the pathogens of other wild bee species. Some bee disease agents, such as deformed wing virus (DWV) and Nosema ceranae, have broad host ranges and are able to infect both honey bees and bumblebees; others, such as Crithidia bombi or Paenibacillus larvae, seem to be more host-specific (3941). Although natural pathogens undoubtedly play an important but poorly understood role in influencing the population dynamics of their bee hosts, our focus here is on the impacts that non-native parasites and pathogens may have.

The spread of most honey bee parasites and pathogens has occurred inadvertently as a result of transporting honey bees long distances. Much of this happened in historic times, but it continues despite some improvements in quarantine procedures. The best-known example is the mite Varroa destructor, originally associated with the Asian honey bee Apis cerana. Varroa has since jumped hosts to the European honey bee Apis mellifera, which has little resistance to this pest. Since the 1960s, Varroa has spread from Asia to Europe, the Americas, and most recently to New Zealand. The mite acts as a vector for pathogens such as DWV, and the combined effect of the mite and the diseases it transmits is a major contributor to honey bee colony losses in North America and Europe (42, 43). Fortunately, the mite appears unable to survive on bees outside the genus Apis.

A strikingly similar series of events has also seen the microsporidian N. ceranae jump from A. cerana to A. mellifera, and in the past 20 years it has spread to Europe and the Americas, where it is now prevalent at high frequency (Fig. 2) (44, 45). It has also been detected in wild bumblebees in Europe, China, and South America (41, 46, 47) and in solitary bees in Europe (48). In the lab, N. ceranae appears to have higher virulence in bumblebees than it does in honey bees (41), although the impact it has had on wild populations is unknown. Asia is not the only source of non-native diseases; the African honey bee parasite Aethina tumida (small hive beetle) recently invaded North America, Egypt, Australia, and Europe, and causes considerable damage to B. impatiens colonies (49, 50). It seems highly likely that it also attacks other wild bumblebee species that are not so readily cultured and are therefore less well studied.

Fig. 2 World distribution of the microsporidian pathogen Nosema ceranae in different bee hosts.

This parasite was first isolated from the Eastern honeybee (A. cerana) collected in China in 1996 (163) and was subsequently found infecting Western honey bees (A. mellifera) in Europe in 2005 (164). Soon after, N. ceranae was detected in A. mellifera in many regions of the world, including Africa, Asia, the Americas, and Oceania (44, 45, 165168), and more recently in other bee species including several Asian Apis species (169, 170) and wild bumblebee species from Europe, China, and South America (41, 46, 47). N. ceranae has also now been detected in solitary bees from Europe (48), confirming a very wide range of hosts and high dispersal rate. Although the origins and primary host of N. ceranae are yet to be accurately established, the apparent late and gradual invasions of N. ceranae into different A. mellifera populations have led some authors to suggest that A. cerana may be the primary host of N. ceranae and that it may have only recently emerged as a parasite of Western bees (170). The mechanism by which N. ceranae broadened its host range from an Asian bee species to other bee species across the world is unknown, but there has been human-mediated contact between Asian and Western bees for at least a century. Note that regions in the figure where Nosema appears to be absent (white) may be due to a lack of sampling in these areas.

Bee diseases are also being redistributed around the globe by the commercial trade in bumblebee colonies, which are mainly used for pollination of greenhouse crops such as tomatoes. This trade began in the 1980s in Europe, and now more than 1 million nests of the European B. terrestris are reared each year and exported to various countries. In North America, the eastern American species B. impatiens is reared for this purpose. Unfortunately, it does not yet seem possible to rear colonies that are free of disease, not least because the bees are reared on honey bee–collected pollen, providing a route for exposure to many bee pathogens. Commercial colonies of B. terrestris are commonly infected with one or more parasites, including Nosema bombi, N. ceranae, Apicystis bombi, and DWV (41).

There is evidence that non-native pathogens or pathogen strains associated with these colonies are having devastating impacts on wild bumblebee populations. In North America, the accidental importation of a non-native strain of the parasite N. bombi via commercial bumblebees has been implicated in the marked decline of several bumblebee species, although convincing causal evidence remains elusive (51, 52). The evidence from South America is clearer; here, B. terrestris were deliberately introduced by the Chilean government despite the presence of native Bombus species, with terrestris spreading rapidly to occupy a vast area of southern South America. The arrival of B. terrestris appears to have led to the rapid local extinction of the native B. dahlbomii at a speed plausibly explained only by pathogen spillover (16). Although the parasite responsible has yet to be ascertained with certainty, both A. bombi and C. bombi have been shown to be highly prevalent in the invasive species (16, 53). There is a clear parallel with the devastating impact of European diseases on native Americans 500 years ago.

Even when commercial bees are free of disease upon arrival or are infected only with indigenous parasites, they can still affect native pollinators. High-density populations of managed bees may provide conditions for the rapid multiplication of parasites that then spill over into wild populations (5456). A combination of field observations and modeling suggests that waves of Crithidia bombi infection travel outward from greenhouses containing commercial bumblebees. Prediction indicates that waves can spread at ~2 km per week, with up to 100% of wild bees within the spreading radius becoming infected, although this is not yet well supported by direct evidence (55).

In general, we know little about the natural geographic range, host range, prevalence, or virulence of most bee pathogens, and so it would seem wise to take very careful precautions to prevent further introductions of bee pathogens from outside their native range, in addition to minimizing any spillover from commercial pollination operations (52, 57).


Pesticides are the most controversial and debated cause of bee declines. When appropriately used, pesticides provide a clear economic benefit, but they bring the welfare of bees into direct conflict with industrial agriculture. Herbicides are highly effective at minimizing weed problems in most cropping systems, enabling farmers to grow near-pure monocultures, but their use inevitably reduces the availability of flowers for pollinators and can contribute substantially to rendering farmland an inhospitable environment for bees (11, 58, 59). Understandably, most attention has been paid to the direct toxic effects of pesticides on bees, particularly the impacts of insecticides. Of the 161 different pesticides that have been detected in honey bee colonies (60, 61), Sanchez-Bayo and Goka (61) predicted that three neonicotinoids (thiamethoxam, imidacloprid, and clothianidin) and two organophosphates (phosmet and chlorpyrifos) pose the biggest risk to honey bees at a global scale, as determined from data on toxicity, frequency in hives, and concentrations detected. It is clear that bees are often chronically exposed to cocktails of pesticides throughout their development and adult life (6164), but the effects of this are poorly understood and are not examined by current regulatory procedures for risk assessment (65).

Neonicotinoids are the newest of the main insecticide classes and are the group most strongly implicated in bee declines (65, 66). They are neurotoxins that target the insect central nervous system, binding to postsynaptic nicotinic acetylcholine receptors and causing overstimulation, paralysis, and death (67). These insecticides are commonly applied as seed treatments and are systemic within plants, spreading through plant tissues and into the pollen and nectar of flowering crops such as canola. They are also water-soluble but highly persistent in soil and soil water, and as a result have been found at biologically relevant concentrations in the pollen and nectar of wildflowers near crops (63). Thus, there is a clear route for ingestion by bees. Oral toxicity is high, with the short-term LD50 for the most commonly used neonicotinoids in the range of 4 to 5 ng per honey bee (68) (LD50 = lethal dose 50%, the dose that kills 50% of test organisms). Long-term chronic exposure results in mortality in overwintering honey bees when feeding on food contaminated with concentrations as low as 0.25 ppb (69). Sublethal effects of neonicotinoid exposure have also been observed in both honey bees and bumblebees, including reductions in learning, foraging ability, and homing ability, all of which are essential to bee survival (7074). Yang et al. (75) recently showed that even low exposure during the larval stage (0.04 ng per larva, equating to less than 1% of the LC50 for adult bees) can have a lasting impact on learning in adult honey bees (LC50 = lethal concentration 50%, the concentration that kills 50% of test organisms).

It seems very likely that bees living in most arable farmland are routinely exposed to sufficient neonicotinoids to suffer both lethal and sublethal effects. However, whether this translates into a detrimental effect at the colony level remains disputed. In bumblebees, exposure of colonies to field-realistic concentrations of imidacloprid greatly impaired colony growth and reduced queen production by 85% (76). In contrast, field studies with honey bee colonies have proved more challenging to perform convincingly (77, 78), not least because of the huge areas over which honey bees forage, the lack of a clear endpoint to colony development, and their long-term storage of food reserves. This lack of clear evidence of harm in the field is often misinterpreted as evidence that toxicological studies on individual bees do not translate into colony losses in the field, rather than as the absence of evidence.

Monotonous diets

Intensively farmed areas provide few wildflowers but do provide spatially and temporally isolated gluts of flowers, in the form of mass-flowering crops such as sunflowers and canola. If a human were to consume nothing but sardines one month, chocolate the next, turnips the month after, and so on, one could reasonably expect that person to fall ill. This may seem a frivolous example, but it is a reasonable parallel to the experience of some honey bee colonies, particularly those in North America that are transported back and forth across the continent each year to provide pollination for major crops such as almonds in California, blueberries in Maine, and citrus in Florida. Where the nectar or pollen of crop flowers contain toxins, such as the glycoside amygdalin found in almonds (79), bees might potentially consume harmful concentrations of such substances.

More generally, it seems certain that bees inhabiting intensive farmland have a more monotonous diet than they would have experienced in their evolutionary past, but how this might affect their fitness remains unclear. The pollen of different plant species varies greatly in protein content, amino acid composition, lipid, starch, and vitamin and mineral content. Nectar commonly contains varying and low concentrations of a range of nutrients and other compounds of largely unknown importance (8084). Thus, we might expect the type and range of flowers available to affect individual bee health and colony fitness in multiple ways; for example, in honey bees, both pollen quality and diversity influence longevity, physiology, and resistance or tolerance to disease (8588). However, this topic has been little investigated, particularly for wild bees. The perception that honey bees may be receiving an inadequate diet has led to the development of protein supplements, but once again, there has been little research on the long-term effectiveness of such supplements on colony health (89, 90).

Interpreting the effects of availability of mass-flowering crops on bees and their colonies is further complicated because visiting such crops often exposes bees to pesticide residues, so that positive effects of increased food availability may be offset by negative effects of the pesticide. Some studies have found positive effects of proximity to canola on bumblebee colony growth and abundance (91, 92) and on numbers of nesting red mason bees Osmia bicornis (93, 94); others have found no relationship for bumblebees (58, 95) nor for solitary bees (96). Interestingly, none of these studies considered what role pesticides might have played in mediating the effect of the crop, or even reported which pesticides were applied to crops in the study area—an omission that now seems naïve, given the recent focus on impacts of neonicotinoid insecticides on bees.

Shipping fever

It seems reasonable to hypothesize that the long-distance transport of bees, as routinely occurs for honey bees in North America and for commercial bumblebee colonies, places stress on the colonies. For several days, they may be confined and subject to vibration, high temperatures, high levels of carbon dioxide, and irregular disturbance. It has long been known that such stress can activate bacterial and viral infections in vertebrate livestock (97), but this has not been investigated in bees, although Bakonyi et al. (98) suggested that shipping stress may have contributed to honey bee colony losses in Hungary. This is clearly an area where further research is needed.


The role of competition in determining the relative abundance of species is notoriously difficult to ascertain in mobile organisms such as bees, but it seems likely that competition for floral resources and perhaps also for nest sites does occur in natural communities, and that it can be exacerbated by the introduction of non-native species, particularly when the latter are present at high densities (57). For example, there is evidence that high concentrations of domestic honey bee hives can displace wild bumblebees from their preferred foodplants and from whole areas if hive densities are sufficiently high (99, 100). This can result in a reduction in the size of bumblebee workers (101) and reduced reproductive success of bumblebee colonies (102). Although in general, the interests of honey bee keepers and wild bee conservationists are aligned (all would agree on the benefits of increasing floral resources, reducing exposure to pesticides, and preventing invasions of alien pathogens), there may occasionally be conflict where beekeepers wish to place hives in areas with important populations of rare wild bees (57).

Climate change

Climate change is widely accepted to pose one of the largest threats to biodiversity worldwide, but likely impacts on pollinators and pollination are not well understood. One danger is that the phenology of pollinators may diverge from that of the plants they pollinate, with potentially disastrous consequences for both, but there is little evidence that this has happened so far (103). Advances in flowering and bee emergence are often broadly similar, and in any case few plants are dependent on a single pollinator, so that any mismatch with one pollinator is likely to be compensated by increased availability of another (103, 104).

Another potential effect of climate change is as a driver of range shifts, leading to a spatial mismatch between plants and pollinators. Range shifts in response to climate have been demonstrated in butterflies (105) and are to be expected in bees (13); for example, there is already evidence that the lower altitudinal limit of some montane bumblebees has shifted uphill in Spain (106). We would predict declines in bumblebees at the southern edge of their range because they tend to be poorly adapted to high temperatures.

Of course, climate change is not solely associated with warming; extreme weather events such as storms, floods, and droughts are predicted to increase, and we would expect these to have major impacts on local bee communities. For example, flooding is likely to be harmful to the many bee species that nest or hibernate underground.

Overall, although there is little strong evidence that climate change has yet had any great effect on bees, it is likely to provide a growing source of stress in the future that would exacerbate the impact of other factors such as habitat loss.

Interactions between stressors

Unfortunately, the public debate on bee health has often become polarized, with claims that, for example, Varroa or neonicotinoid insecticides are the sole or primary cause of bee declines or honey bee colony losses. If a middle-aged man who is overweight, does little exercise, and smokes and drinks heavily were to die of a heart attack, we would not be surprised and we might not spend too long arguing over which single risk factor was most important in bringing about his untimely demise. Similarly, wild bee declines and honey bee colony losses are clearly due to multiple, interacting, and sometimes synergistic factors, and the combination of factors involved no doubt varies in time and space.

In general, the combined effect of multiple stressors is likely to be more harmful than one stressor alone (107109) (Fig. 3). In the worst-case scenario, sublethal stressors that do not incur harmful effects in isolation could, in combination, result in lethal effects. As we have already seen, bees are often exposed chronically to mixtures of pesticides and other chemicals. Some, such as ergosterol biosynthesis inhibitor (EBI) fungicides, have very low toxicity in themselves but may increase the toxicity of some neonicotinoids and pyrethroids by as much as a factor of 1000 (110112). Piperonyl butoxide is often added to pesticide formulations and also acts synergistically with some neonicotinoids, increasing toxicity by a factor of up to 244 (111). Intriguingly, while imidacloprid alone has been shown to impair olfactory learning (113), combined exposure to imidacloprid and coumaphos has been shown to result in a slight increase in learning in honey bees (114). So although regulatory processes examine the effects on bees of exposure to a single pesticide at a time, in reality bees are simultaneously exposed to many pesticides, some of which have combined effects that cannot be predicted from studies of their effects when used in isolation.

Fig. 3 Both wild and managed bees are subject to a number of significant and interacting stressors.

For example, exposure to some fungicides can greatly increase toxicity of insecticides (110112), whereas exposure to insecticides reduces resistance to diseases (115123, 125, 126). Dietary stresses are likely to reduce the ability of bees to cope with both toxins and pathogens (127129). [Photo: Beth Nicholls; Flickr Commons, AJC1]

Several recent studies indicate that interactive effects between pesticides and pathogens could be especially harmful for bees (115121). For instance, developmental exposure to neonicotinoid insecticides renders honey bees more susceptible to the impact of the invasive pathogen N. ceranae (122). Imidacloprid can act synergistically with Nosema spp. by increasing the prevalence of Nosema infections in hives (116) and increasing Nosema-induced mortality (115). Similarly, Aufauvre et al. (118) showed that mortality of honey bees was greater when bees were exposed to the insecticide fipronil and infected by N. ceranae than when only a single stress factor was present. There is evidence that exposure to pesticides may impair the immune function of insects, which would explain these effects (43, 123125). For example, Di Prisco et al. (126) recently showed that exposure to neonicotinoids (clothianidin or imidacloprid) leads to immunosuppression in honey bees, which in turn promotes the replication of DWV in insects with covert infections. This effect was found at very low concentrations, well below those that bees are likely to encounter in the field.

Interactions between stressors are not confined to pesticides and pathogens. The ability of bees to survive parasite infections is compromised by nutritional stress. For example, Crithidia bombi causes little mortality in well-fed bumblebees but becomes virulent in bumblebees with a restricted diet (127). Activating the immune response has a metabolic cost; bumblebees increase their food consumption when immune responses are up-regulated (128), and artificially stimulating the immune response by injecting latex beads caused mortality in starving bumblebees but not in those that were well fed (129). Increased food consumption in infected bees could also increase exposure to pesticides. Activating immunity has been shown to impair learning in both honey bees (130) and bumblebees (131, 132), and impaired learning will reduce the bees’ ability to locate floral resources and extract rewards, thus exacerbating nutritional stresses.

Although to our knowledge this has not yet been examined, it seems highly likely that nutritional stress may also modulate the ability of bees to cope with pesticides, and this may explain in part why the observed LD50 of toxins in bees is highly variable across studies (65).

In summary, stressors do not act in isolation. Bees of all species are likely to encounter multiple stressors during their lives, and each is likely to reduce the ability of bees to cope with the others. A bee or bee colony that appears to have succumbed to a pathogen may not have died if it had not also been exposed to a sublethal dose of a pesticide and/or been subject to food stress (which might in turn be due to drought or heavy rain induced by climate change, or competition from a high density of honey bee hives placed nearby). Unfortunately, conducting well-replicated studies of the effects of multiple interacting stressors on bee colonies is exceedingly difficult. The number of stressor combinations rapidly becomes large, and exposure to stressors is hard or impossible to control with free-flying bees. Nonetheless, a strong argument can be made that it is the interaction among parasites, pesticides, and diet that lies at the heart of current bee health problems.

Sustainable pollination into the future

There is universal agreement that we must ensure adequate pollinator populations into the future if we wish to continue to grow a diversity of insect-pollinated crops and also ensure the integrity of natural ecosystems. It is also clear that moving toward heavy reliance on only a few species of managed pollinators, such as honey bees or one or two species of bumblebee, runs the risk of supply failure; for example, should honey bee stocks in North America fall much further, the viability of almond production in California would be threatened (133). Wild pollinators provide a service that is largely free and globally already contributes to the majority of crop pollination (24, 25, 134, 135). Maintaining a diversity of pollinator species improves crop success via functional complementarity; different species visit different parts of the crop or crop plant, at different times of the day or year, and respond differently to environmental perturbations (1, 136141). A diversity of pollinators can buffer impacts of climate change that might otherwise result in a mismatch in phenology of pollinators with the flowering of crops (104). It is thus essential that we take steps to conserve a broad community of pollinators in farmland.

Fortunately, although the causes of pollinator ill health may be complex and varied, conserving pollinators need not be difficult or expensive. If we accept that declines are due to interacting stressors, then it follows logically that removing or reducing any of the stressors we have described is likely to benefit bee populations. Measures can be taken that are likely to simultaneously benefit a broad suite of species, both domesticated and wild.

Increasing abundance, diversity, and continuity of floral resources

Schemes such as the sowing of flower-rich field margins or hedgerows, or retaining patches of seminatural habitat among or near farmland, provide clear benefits to bee diversity and abundance (30, 141148) (Fig. 4A). This in turn increases pollination to nearby crops and provides an economic incentive to farmers growing insect-pollinated crops (149). Many countries also offer financial incentives to farmers for taking measures to boost biodiversity that help to offset implementation and opportunity costs. However, take-up of schemes to boost pollinators remains low in most countries, perhaps reflecting a lack of understanding of the economic and environmental benefits or a lack of familiarity with implementation of such measures. Education and outreach in this area could pay great dividends for pollinator conservation.

Fig. 4 Increasing floral abundance in the landscape is very likely to benefit pollinator populations.

(A) Schemes to boost flower abundance in farmland, such as this wildflower strip along a field margin, have been demonstrated to provide clear benefits for wild bee populations [e.g., (140145)]. (B) Urban areas can support high populations of pollinators, which may spill over into neighboring farmland. Conversion of amenity grasslands in urban areas to wildflower patches has been shown to greatly boost numbers of wild pollinators (151).

Planting of appropriate flowers in gardens and amenity areas can also contribute to pollinator conservation (150, 151) (Fig. 4B). There is evidence that urban areas can support higher populations of some pollinators than farmland [e.g., (36)] and can boost bee numbers in adjacent farmland (58). Many lists of bee- or wildlife-friendly flowers are available on the Internet, but they tend to be based on anecdote rather than evidence, and there is a need to develop regionally appropriate, evidence-based advice as to the best plants to grow (152).

Providing nest sites

Wild bees use a diversity of habitats for nesting, including burrowing into bare soil and using existing cavities underground, holes in wood, or hollow plant stems. Seminatural habitats, hedgerows, and permanently uncropped field margins allow for many of these; hence, schemes to boost floral diversity are also likely to boost nesting opportunities (141). Additional nest sites can also be provided by providing bundles of hollow reeds or canes, or patches of bare soil (153).

Reducing exposure to pesticides

Bees are currently chronically exposed to a cocktail of pesticides, some of which act synergistically. Since the late 1990s, the cost of pesticides has fallen markedly relative to labor and fuel costs and the value of the crops (154). As a result, current levels of pesticide use are generally high and are not always justified by evidence that they are necessary to maintain yield (66, 155). The widespread prophylactic use of systemic insecticides (such as neonicotinoids) as seed dressings exposes bees and other nontarget wildlife, results in accumulation of pesticides in the environment, and places strong selection pressure on pests to evolve resistance. A return to the principles of integrated pest management (IPM) (156), which depends on preventive methods such as crop rotation and views the use of pesticides as a last resort in the battle against insect pests, could greatly reduce exposure of bees, benefit the environment, and improve farming profitability. Some European countries have independently developed national pesticide reduction programs (156), and the European Union Sustainable Use of Pesticides Directive 2009/128/EC required member states to implement national action plans to minimize pesticide use by January 2014. In most EU member states, this directive appears to have had little or no impact on farming practices.

Current risk assessment procedures, which examine the short-term impact of a single pesticide in isolation, are clearly not adequate to encapsulate the true scenario faced by bees living in farmland. Improvements are needed to make them more realistic while keeping the cost of regulatory tests affordable, posing a considerable challenge to the ingenuity of scientists and regulators.

The EU moratorium on the use of three neonicotinoids (which started in December 2013) is an attempt to use policy change to reduce exposure of bees to stressors, following a review by the European Food Standards Agency (157159) that declared neonicotinoids an “unacceptable risk” to bees. However, if this simply leads farmers to replace neonicotinoids with other pesticides, this may not be of great benefit to bees or the environment. Funding for research and for the provision of clear, independent advice for farmers with regard to how to reduce pesticide use generally by adopting IPM practices might provide a better and more sustainable long-term solution.

Preventing further introductions of non-native bees, parasites, and pathogens

The careless disregard with which we ship bees from country to country has resulted in the irreversible spread of many serious parasites and pathogens. Strict quarantine controls should be implemented on the movement of all commercial bees, and there is an urgent need to develop means of rearing commercial bumblebees that are free from disease. Deliberate introductions of non-native bee species (such as the recent introduction of the European B. terrestris to South America) should of course be prevented. The companies that rear commercial bees should bear some responsibility here and should refuse to sell bees to regions where they are not native. There is clear hypocrisy in the policies of countries that prevent importation of non-native species but allow exportation of species to places where they do not naturally occur.

Develop monitoring programs

We have good distribution maps for pollinators in some countries, particularly for bumblebees, and citizen science schemes such as Bumble Bee Watch in North America and Beewatch in the United Kingdom can help to track changes in these distributions. However, the lack of long-term data on pollinator abundance is a glaring knowledge gap that urgently needs to be filled. It will probably never be possible to monitor all pollinator species at a global scale, but it would be practical to systematically collect data on the abundance of a subset of the more abundant and economically important pollinators. Citizen science surveys can provide a cost-effective means for large-scale population monitoring; for example, the UK butterfly monitoring scheme uses volunteers to walk regular transects using a standard methodology to count the butterflies seen, and has generated a large and long-term data set that has provided powerful insights into insect population change (160, 161). Beewalks, a similar scheme to count bumblebees, has been launched by the Bumblebee Conservation Trust to obtain population data for bumblebees in the United Kingdom, although it is still in its infancy. In the United States, the Great Sunflower Project asks volunteers to count pollinators in flower patches in their local area. However, such schemes are limited by the taxonomic skills of volunteers, particularly for the many pollinator taxa that are hard or impossible to identify in the field. LeBuhn et al. (162) argue that a pan-trapping network, which could use citizen scientists to place out the traps but experts to identify the catch, would be the most cost-effective means for monitoring a large cross section of pollinator species on a large geographic scale. Indeed, for a relatively modest sum, it would be possible to set up an international pan-trapping network to monitor pollinators, following a standard methodology. Until good population data become available, we cannot identify species or regions under most threat and hence we cannot prioritize management.

In the absence of pollinator monitoring, we have no early warning system to tell us how close we may be to a pollination crisis. With a growing human population and rapid growth in global demand for pollination services, we cannot afford to see crop yields begin to fall, and we would be well advised to take preemptive action to ensure that we have adequate pollination services into the future.

References and Notes

  1. Acknowledgments: We thank R. Hatfield and S. Jepsen for their help with Fig. 1B.
View Abstract

Stay Connected to Science

Navigate This Article