German Bee Monitoring Project

A critique of a report investigating winter honey bee colony losses in Germany

The following provides a translation of the summary points of a German Report by Dr Peter P. Hoppe and Dr Anton Safer:  “Das Deutsche Bienenmonitoring-Projekt: Anspruch und Wirklichkeit. Eine kritische Bewertung" - A critical appraisal of the German Bee Monitoring Project which can be downloaded from a link at the end of this article.

About The German Bee Monitoring Project

The German Bee Monitoring Project (DeBiMo) was an observational, longitudinal, multi-center study which ostensibly aimed to investigate the various factors involved in mass bee-deaths in Germany.   

DeBiMo was directed by a project board including members of nine German Bee Institutes, two apiarist associations (DIB and DBIB), the German Farmers`Federation (DBV), the Federal Ministry of Nutrition, Agriculture and Consumer Protection (BMELV), and industry (BASF SE, BayerCropScience AG, Bayer HealthCare AG, and Syngenta). Until 2010, the pesticide industry provided 50% of the funding; as of 2010 the project was totally funded by BMELV.

Results for 2004-2008 were published by Genersch et al (2010).

Dr Peter P. Hoppe (formerly of the University of Munich) & Dr Anton Safer (University of Heidelberg) conducted a critical review of this publication (biographies of both scientists may be found at the end of this article).

Main points:

(Comments added in italics are my own):

• About 110 German experienced beekeepers were asked to participate and they provided 10 colonies each for monitoring during the study. Regional distribution appeared to be biased towards areas with less intensive agriculture, where pesticide-use is low.
  
  Overall, winter colony-loss among the study-hives ranged from 3 to 16% compared with the estimated norm of about 30% for all of Germany. For these reasons, the study cannot be considered as representative of all German beekeepers, nor of the true scale of bee-colony losses in Germany.
   
• Data collection was incomplete and inconsistent. While beekeeping conditions and varroa samples were collected with relatively few gaps, the data collection on Nosema and virus infestation and chemical residues shows serious gaps. Data collection for chemical residue analysis in bee bread was performed in about 5% of all bee hives.
  
  Comment:
  One of the characteristics of honey bee colonies is that their by-products help us to assess chemical exposure within the hive, such that we might then be able to monitor for colony effects.  It’s interesting, therefore, that residue analysis on bee bread was only carried out on 5% of hives.
  
  
• The statistical evaluation of results is inadequate. In particular, the authors avoided any multi-variate approach to clarify the respective contributions of crucial factors such as pesticide-exposure, varroa, and viral pathogens on winter death of colonies. Inappropriate statistical methods are used to twist evaluations suggesting significance close to proof of causality.
  
  Comment: 
  It seems the authors of the study avoided looking at the evidence to see whether (or the extent to which) the causes of honey bee colony deaths might also be related to such factors as pesticide contact, as well as exposure to viral diseases and Varroa.  Inappropriate statistical methods were used, which serve to twist  against conclusions that might otherwise have been drawn.  In this way, true causes of honey bee colony deaths (or suggested causal relationship) could be disguised.
  
  
• The authors claim that Varroa is “the dominant killer of honey bee colonies during winter”. However, the relationship between autumn Varroa infestation and winter-mortality is depicted by a misleading graph, without legend, that obscures the uncertainty of the relationship.
  
  Comment:
  It looks like the authors are determined to pin the blame to an isolated cause, namely Varroa (which would of course, deflect attention away from pesticides and associated diseases due to reduced immunity and harming of the proper functioning of the colony).  In order to support their argument, the authors present the information in a misleading way.
  
  
• The main flaw: The authors are either unaware of the most basic principles of epidemiology or appear to be deliberately confusing statistical association for causality in order to reach preconceived conclusions not supported by the underlying data or the methods used to collect them.
  
  Comment:
  In other words, data is being misused in such a way that it simply supports a preferred argument, and yet that argument is not in reality, supported by the data.
  
  
• Bee-pathogenic viruses were investigated qualitatively. The presence of Deformed Wing Virus (DWV) and Acute Bee Paralysis Virus (ABPV) was significantly associated with winter mortality. The authors misinterpret these observations as effects rather than statistical associations, resulting in the unsubstantiated claim that “pests and pathogens” (Varroa and viruses) are the main causes of winter losses.
  
  Comment: 
  By presenting the data as ‘effects’ (i.e. result/consequence/cause) rather than ‘associations’, the authors can imply that DWV and ABPV are the singular causes of winter honey bee colony death. This helps to deflect attention from the reason DWV and ABPV might be present in the first place. 
  However, if we refer to ‘associations’, we immediately look to common ‘partners’ within the association, and begin asking questions.  For example, were there any associations between pesticide contamination and diseases in dead honey bee colonies? 
  
  
• All aspects of the study which related to pesticides were handled superficially. In particular, with few exceptions, no numerical results for pesticides in bee bread were reported. Moreover, the pesticide-profiles (names and concentrations of pesticides in bee bread and pesticide-treated crops in the vicinity) of individual monitoring apiaries were not revealed. Assessment of potential negative effects was based solely on LD50 values, whereas synergistic and sub-lethal or chronic effects of pesticides were not discussed. This creates the impression that the study was designed to exclude pesticides from the picture and to exonerate the systemic neonicotinoids from all suspicion of involvement in the bee colony losses.
  
  Comment: 
  Bayer admit in their marketing brochures the power of sub-lethal and chronic effects of their pesticides on large colonies of insects (i.e. very tiny amounts of pesticides which do not kill immediately, but instead harm the insects in some way, such that they cannot function properly, thus ultimately hampering the  functioning of the colony in which the members play a vital role for its complete survival).  
  
  
• Several points suggest a conflict of interest within the project board. These include the number of non-scientists appointed to the board, the existence of a secret “working group on plant protection” within the board which is in charge of all aspects relating to pesticide analyses and reporting, delay in pesticide analyses, failure to report analytical results of pesticide analyses, the undefined plausibility check of data, and omission of data. Moreover, the publication (in Apidologie) does not reveil who was responsible for designing the study, for writing of the manuscript and forthe final version of the manuscript.  Last but not least, it is remarkable that the 3 reviewers of the DeBiMo publication failed to detect the main flaw and the numerous deficiencies.
  
  Comment: 
  I am reminded of the bizarrely named ‘Bee Protection’ group within the ICPBR of the EPPO, in which mostly industry representatives and a handful of civil servants formulated ideas about the best way to conduct pesticide tests on honey bees.  None of their ideas seemed to include anything which might provide a genuine and realistic picture of the risks to bees posed by their poisons, but then to have done so would hinder their product registrations.
  
  
  
• Taken together, it is clear that the study does not accord with ‘the scientific spirit’. For these reasons, DeBiMo has failed to deliver a thorough and credible scientific evaluation of the possible causes of bee-colony death in Germany. It has failed to prove that Varroa and viruses (“pests and pathogens”) are the primary causes and that “pesticides play no role”.

To download the full report in English of the  German Bee Monitoring Project  (opens a new window).

Scientist Biographies:

Peter P. Hoppe, Dr. med. vet. has held academic positions in animal physiology and nutrition at the University of Munich and University of Nairobi. He joined BASF Ludwigshafen in 1979 to become Head of Nutrition Research Station, Ludwigshafen until retirement in 2002. He has wide research experience in wild and domestic animal species and humans. He has published about 100 publications and has long-time experience as reviewer. He is a member of NABU (Nature Protection Federation Germany).

Anton Safer, Dr. rer. biol. hum.; Agricultural engineer from Stuttgart-Hohenheim University, graduated in Human Biology at Hannover Medical School. Worked 36 years as biometrician in pharmaceutical companies, clinical and preclinical studies (toxicology, pharmacology); currently project statistician at the Heidelberg University Institute of Public Health/Epidemiology. Member of the German Branch of Evidence Based Medicine Association (DN-EbM), affiliated with Cochrane.org, and Friends of the Earth (BUND).