Observations on the development of managed bee colonies
Numerous studies have reported a decline in honey bee health and numbers of colonies in recent years (e.g., Potts et al. 2010; Van Engelsdorp et al. 2007) but there are other voices which challenge the contention that overwintering losses are any worse now than they have been in the past (Doebler 2000; Borst). Bee keepers in a number of countries have reported a decline in the ability of colonies to successfully survive the winter. A very extreme form of bee decline was reported from the USA with a sudden disappearance of all but a few bees from managed bee colonies and this phenomenon was called Colony Collapse Disorder (Van Engelsdorp et al. 2008). In Europe, a similar phenomenon has not been observed and larger numbers of colony deaths have generally been locally confined (e.g., Genersch et al. 2010). In recent times, the most dramatic decline in managed colonies in Europe occurred in the 1990’s coinciding with the socio-political changes in Eastern Europe while at the same time colony numbers remained stable in Western Europe (Moritz et al. 2010). Here, colony losses have been more varied over time and between countries as shown by Hendrikx et al. (2012) when looking at overwintering losses from 2000 to 2009 in Denmark, Finland, Sweden, Germany, England and Wales. A baseline colony loss rate of around 10 % was identified but peaks of over 30 % were seen in some countries in certain years. In general, colony losses have been compensated for by beekeepers replacing colonies as particularly seen in the 70’s and 80’s in response to the appearance of Varroa destructor, resulting in a steady increase in the number of managed colonies overall (Moritz et al. 2010).
Many factors may have contributed to a decline in honey bee health and increases in bee or even colony losses, where they occur. The spread of parasites and pathogens is a possible cause, although no single agent has been identified and instead a virulent combination of parasites and pathogens has been suggested (Chen and Evans 2007; Johnson et al. 2009). Other possible factors include a reduction in available forage (Decoutye et al. 2010), beekeeping management practices (for example insufficient Varroa control and the development of resistance to treatments), movement of colonies, weather and climate change (Kluser et al. 2011). Exposure to pesticides is another factor that has been implicated in bee health decline (Mullin et al. 2010). Although the majority of cases where bees are killed by pesticides are caused by foliar-applied products (Fletcher and Barnett 2003; Barnett et al. 2007; Thompson and Thorbahn 2009), the current public discussion is focused on systemic seed treatment products. It is proposed by some researchers that adverse effects may occur when bees are feeding on seed-treated crops or dust drift-contaminated plants or if they collect contaminated pollen and/or nectar from these plants and return it to the hive. However, major incidents related to seed treatments with nitro-substituted neonicotinoid insecticides have only been related to the emission of dust during drilling of corn seeds which had been improperly treated (Pistorius et al. 2009). Dust emission during maize drilling has also been discussed as a potentially contributing factor of spring bee mortality in Italy (Greatti et al. 2002). However, some of the reported changes in colony numbers pre-date the agricultural use of neonicotinoids and so it is important to look at the overall picture as well as identifying specific information linking possible neonicotinoid exposure to observed colony losses.
Findings from monitoring studies and field surveys
Monitoring data from a number of countries are available to assess the presence of neonicotinoid residues in honey bee samples and possible impacts of these residues at the colony level. A review of winter losses of bee colonies up to 2008 under the auspices of the German Bee Monitoring Project is provided by Janke and Rosenkranz (2009). This was carried out by establishing a database of 120 apiaries and 1200 bee colonies over a period of four years. Data were collected for a range of parameters including for the apiary (e.g., site, nuclei, movement of colonies, Varroa treatment), strength of the colonies in autumn and spring, honey yields, residues in bee bread (stored pollen) and bee disease analysis. Overwintering losses of the monitored colonies ranged between an average of 7.9 and 15.9 % over the four project years. Of all apiaries participating during the four project years, nearly one-third had no losses while about 15 % had losses over 20 %. Bee bread samples collected during or after the flowering of oilseed rape in spring were analysed using a sensitive multi-residue method. In 215 analysed samples collected from 2005 to 2007, clothianidin was not found in any sample while imidacloprid was found in only one sample (3 μg/kg). Some samples contained no residues but the majority did and in most cases more than one active ingredient was found with at least 55 being identified overall (usually only in trace amounts). Nearly 4400 data sets were statistically analysed for the identification of triggers with negative influence on overwintering. The winter losses were significantly correlated with Varroa infestations and virus infections in autumn. It was concluded that no acute effects of pesticides on honey bee colonies were expected on the basis of the evaluated residue data but additional work was suggested to further assess long-term effects on e.g. colony development and over-wintering success.
In a follow-up paper, Genersch et al. (2010) provided a more detailed assessment of the German Bee Monitoring Project. All data were statistically analysed in respect to the overwintering mortality of the colonies. It was demonstrated for several factors that they are significantly related to the observed winter losses of the monitored honey bee colonies as previously described, i.e. high Varroa infestation level and virus infection (specifically deformed wing virus (DWV) and acute bee paralysis virus (ABPV)) in autumn. A clear significant effect was shown for the age of the queen: colonies that survived the winter had on average significantly younger queens compared to the colonies that collapsed during winter. A clear effect on overwintering success was also found for colony strength (number of bees) in October. There was no significant difference in overwintering success between apiaries with no pesticide residues in the bee bread and those with higher amounts of residues. In addition, there was no significant correlation between winter losses and the amount of oilseed rape pollen in the honey harvested in summer (in Germany, oilseed rape seed was almost 100 % treated with neonicotinoid insecticides during the monitoring period). Some residues were identified in bee bread although mainly substances which are considered non-toxic for bees and the observed amounts of the residues were quite low i.e., three orders of magnitude lower than the respective LD50 values. Accordingly, no relation between chemical residues in pollen and colony development or winter losses could be demonstrated even though particular emphasis was put on this aspect.
Foraging honey bee, honey and pollen trap samples collected from eighteen apiaries in Western France (Bretagne and Pays de la Loire) from four different landscape contexts during four different periods in 2008 and in 2009 were analysed to evaluate the presence of pesticides and veterinary drug residues (Lambert et al. 2013). A multi-residue analysis was developed to identify and quantify 80 pesticides (covering the majority of active ingredients used for plant protection) and veterinary drugs in the three beehive matrices. A total of 141 honey bee, 141 honey and 128 pollen samples were collected from the 18 apiaries during 2008 and 2009. Of these, 102 (72.3 %), 135 (95.7 %) and 75 (58.6 %) of the samples, respectively, contained at least one of the defined 80 chemicals. The frequency of detection was higher in the honey samples (28 compounds) than in the pollen (23 compounds) or honey bee (20 compounds) samples, but the highest concentrations were found in pollen. Although most compounds were rarely found, some of them reached high concentrations that might lead to adverse effects on bee health. The three most frequent residues in all three matrices were the widely used fungicide carbendazim and two acaricides, amitraz and coumaphos, that are used by beekeepers to control Varroa destructor. Clothianidin was not found at all while imidacloprid was only detected in 3 out of 141 honey samples (2.1 %) and in 1 of the 128 pollen samples (0.8 %) and was not found among the 141 honey bee samples. While it was recognised that lower levels established for this monitoring exercise could be present (e.g., <LOD for clothianidin in pollen of 1.4 µg/kg), the study authors noted that no adverse effects had been identified under laboratory conditions at the detection limits used in this study.
In Greece, sudden deaths, unusual behaviour and declines of adult honey bee populations were reported in summer 2009 by the beekeepers in the region of Peloponnese, Greece (Bacandritsos et al. 2010). A preliminary study was carried out to investigate these unexplained phenomena in this region and samples were collected from the affected colonies. The clinical symptoms of selected colonies were noted and 37 bee samples were collected from five different apiaries suffering similar symptoms. Symptomatic adult honey bees tested positive for Varroa destructor, Nosema ceranae, chronic bee paralysis virus (CBPV), acute paralysis virus (ABPV), deformed wing virus (DWV), sacbrood virus (SBV) and black queen cell virus (BQCV). Chemical analysis revealed that amitraz, thiamethoxam, clothianidin and acetamiprid were all absent although imidacloprid was present in three out of the five apiaries sampled. Residue levels were detected at an average concentration of 27 µg/kg tissue which is equivalent to a dose of about 3 ng/bee. This residue level is close to the lower threshold of the calculated acute oral LD50 dose which range between 3.7 and 40.9 ng/bee (Schmuck et al. 2001). However, the presence of multiple pathogens and pesticides made it impossible to associate a single specific cause to the depopulation phenomena observed in Greece. The authors identified five possible factors: (i) multiple virus infection by five different viruses along with infection by N. ceranae, (ii) imidacloprid residues in bee tissues, (iii) stress induced by transportation, (iv) temperature and humidity fluctuations and (v) the collection of fir honeydew (low water content).
In another study carried out in Greece (2011–2013), honey bee, bee pollen and honey samples from different areas across the country where incidents had been reported were analysed for the presence of pesticide residues (Kasiotis et al. 2014). In the majority of cases areas were rural with substantial agricultural activity. Honey bee samples were collected very near to or at the entrance of the hives, while bee pollen was collected from the hives. In total, 71 samples of honey bees, bee pollen and honey were collected by individual beekeepers or other authorities (e.g., veterinary institutes) usually after the report of honey bee death incidents during 2011, 2012 and 2013 (44 honey bees, 14 bee pollen and 13 honey samples). From the analysis of the samples the presence of 14 active substances was observed in all matrices, with 73 % of honey bees, 43 % of pollen and 0.1 % of honey samples being positive for at least one compound. A number of neonicotinoids were identified, including clothianidin, which had concentrations in honey bees ranging from 0.7 to 14.7 ng/bee in 2011, 2.7 to 39.9 ng/bee in 2012 and 6.1 to 6.8 ng/bee in 2013 i.e. around the acute lethal oral dose. Since the sample sizes in terms of numbers of analysed bees were very small, e.g., only 17 honey bees in 2011, 19 in 2012 and 8 in 2013, the available data tend to indicate that locally high levels of residues may occur but they do not appear to be widespread and cannot be responsible for any general trends.
In Italy, two regions in the North (Lombardy and Veneto) organised an institutional network whereby if beekeepers observed damage to their bees they were requested to report it to the local Veterinary Authority and fill in a questionnaire (Bortolotti et al. 2009). This was followed up with inspection of the apiaries and the collection of samples (dead bees and pollen from surrounding vegetation) for analysis of pathogens and neonicotinoid residues. Collected data indicate that the higher number of bee loss events occurred in intensively cultivated flat areas, located in the North of Italy, mainly during or after maize sowing. The chemical analyses of dead bees (105 samples) revealed the presence of three neonicotinoid residues: imidacloprid was found in 25.7 % of the samples, thiamethoxam in 2.8 %, clothianidin in 25.7 %, both imidacloprid and thiamethoxam in 4.7 %. The concentrations ranged from 1.01 to 240.6 ng imidacloprid/g, from 3.67 to 39.2 ng clothianidin/g and from 24.8 to 138 ng thiamethoxam/g. Following visual examination and virological analyses the authors did not consider pathological causes as being responsible for the reported damages. They concluded that the spatial and temporal correlation between reported bee mortality and maize sowing and the presence of residues of active ingredients used for seed dressing (imidacloprid, thiamethoxam and clothianidin) in almost half of the samples confirmed a correlation between bee mortality and the sowing of corn seed dressed with neonicotinoids. The use of these neonicotinoids for seed dressing of maize was therefore suspended in Italy in September 2008. Following this, monitoring was carried out from March to May 2009 in 60 hives belonging to 10 different apiaries spread across representative areas of the Friuli Venezia Giulia region (Frilli et al. 2009). It was concluded that no population decline or mortality occurred in the hives in 2009 at the time of maize-sowing.
In Canada, data obtained from the Canadian Pest Management Regulatory Agency (PMRA) was reviewed by Cutler et al. (2014a). Since 2007, 110 honey bee-pesticide incident reports were received by the PMRA but there were very few incidents (six) reported up to 2011. However, in 2012 a significant number of incidents were reported in the province of Ontario and Quebec, where exposure to neonicotinoid dust during planting of corn was suspected to have caused the incidents in about 70 % of cases (either alone or in combination with other pesticides). Most of these incidents were classified as ‘minor’ by the PMRA, and only six cases were considered ‘moderate’ or ‘major’, involving 402 colonies. In that same year, however, there were over three times as many moderate or major incidents due to older non-neonicotinoid pesticides (mainly dimethoate or chlorpyrifos), involving 3855 colonies, i.e., moderate or major incidents due to non-neonicotinoid pesticides affected nearly 10× more hives than incidents related to neonicotinoid pesticides. These data thus agree with the general conclusion with regards to seed dust, i.e., exposure of honey bees to neonicotinoid-containing dust during corn planting needs to continue to be mitigated. However, even in this specific case it identifies the need for the risk assessment to be carried out in a balanced manner for a sustainable agroecosystem, taking into account that other pesticides may pose an even greater risk.
In contrast to maize (corn), there has been no evidence that planting canola (oilseed rape) seed treated with neonicotinoid insecticides in Canada places pollinators at risk. Seed treatments used for canola have a very limited potential to release dust particles, and field studies investigating exposure of honeybees to systemic residues in pollen and nectar of seed-treated canola show no chronic or acute poisoning when analysed at field scale rates (Cutler and Scott-Dupree 2007; Cutler et al. 2014b). In the past decade, the number of honey bees in Canada has reached near-record levels, i.e., more than 700,000 colonies Canada-wide in 2012, up from 600,000 in 2000 (Statistics Canada Cansim Table 001–007). More than 70 per cent of these colonies are in Western Canada, where canola (which is largely seed-treated with nitrosubstituted neonicotinoids) has become one of the most important crops. Clothianidin was first registered in Canada as a seed treatment on canola at the end of 2003 and so its use has been contemporaneous with this increase in colony number i.e. there are no indications of any adverse effects.
Overall then, the available monitoring data indicates that there is no widespread correlation between neonicotinoids such as clothianidin (either in terms of use or residues present in bee matrices) and observed trends in honey bee performance, either in general or in terms of colony losses. The exceptions to this were the cases of an inadequate seed treatment quality of maize seed drilled in the Upper Rhine valley in Germany in 2008 and the correlation between mortality and maize sowing in Italy (together with the presence of residues of active ingredients used for seed dressing). Also, in spring 2011 a high number of bee poisoning incidents was recorded during the sowing of maize in the Pomurje region of Slovenia and the presence of clothianidin in dead bees and pollen was attributed to the sowing of maize treated with the insecticide Poncho Pro (Van der Geest 2012). However, the specific issue of maize seed treatment can be addressed by appropriate risk management measures relating to improvements in seed treatment quality and drilling technology. Apart from this, the only other occurrence of high levels of residues that may be linked to effects on honey bees is localised (related to specific circumstances, e.g., crop, application and local conditions) and does not appear to be responsible for any general trends.
This conclusion is further supported by preliminary results on a survey about bee colony mortality of the 2014/15 season which were published by the independent bee researchers’ network COLOSS. In total, 23,234 beekeepers contributed data to the survey, representing a total of 469,249 honey bee colonies. Respondents are from 31 countries (most EU and a few non-EU countries in Europe and the Mediterranean Region). Losses were particularly heterogeneously distributed between countries as well as between regions. High losses seem to have occurred in particular in the countries of the central region of Europe, but an obvious or consistent pattern is not recognizable: colony loss rates per country varied from 5 % (Norway) to 36 % (Belgium). Overall loss rate in the evaluated countries was 17.4 %, which is twice as high as in the previous year. More details can be found under http://www.coloss.org/announcements/losses-of-honey-bee-colonies-over-the-2014-15-winter-preliminary-results-from-an-international-study. There was a particularly low loss rate immediately after the 2013/14 season when the full spectrum of neonicotinoid products were still on the market and before the restriction came into force, but twice as high a loss rate in the next season (2014/15) when the restrictions were in place. This indicates that the restrictions of the neonicotinoids in the EU did not lead to an immediate improvement of bee health, and further suggests that there is no correlation between colony losses and the use of neonicotinoids. Nevertheless, there is not a tendency towards a generally improving bee health situation in Europe and loss rates are varying according to patterns which are not yet fully understood from year to year and from region to region.