A recent study by Woodcock et al. from the UK-based Centre for Ecology and Hydrology claims that wild bee population declines in England are caused by neonicotinoid seed treatments on oilseed rape. This conclusion is highly speculative as it is based on correlation, not causation, while ignoring other important factors for the occurrence of wild bees.
The study, which looked at the distribution records of wild bees in England from 1994 – 2011 and compared them with oilseed rape cultivation and the use of neonicotinoid seed treatment on this crop during the same time period, suggests that wild bee species foraging on treated oilseed rape are more likely to experience population declines than bees feeding on other plants.
The study concentrates on the potential impact of oilseed rape and neonicotinoids only, in a time period where tremendous change has taken place in agriculture. The authors argue that there was no substantial change in land use over the investigated period, because the total area of cropped land remained relatively constant. However, between 1994 and 2011, oilseed rape acreage increased from around 400,000 to 700,000 hectares in England, replacing many other crops. That such a major shift in crop production was not deemed important is puzzling, since it is very likely to have had an impact on bee populations. Moreover, beyond these significant structural changes, it is obvious that changes in the smaller landscape structures within certain types of arable land have taken place, as well. Such small-scale changes, like removal of hedgerows and field edges where weeds and wild flowers used to grow, can result in fewer nesting sites and sources of pollen and nectar for wild bees. While the authors recognize the importance of landscape structures in affecting the population persistence of bee species, they also acknowledge that there is a lack of data showing such changes over the 18-year period and instead refer only to macro trends in English agriculture – without stating how these general trends relate to the specific area of investigation.
Additionally, the report is silent on variations in climate, or changes in agricultural policies and farming practices during that time frame, all of which would impact the environment and ultimately have effects on bees.
By arbitrarily focusing on a single variable – the potential effects of neonicotinoids – and by failing to consider other important factors impacting the occurrence of wild bees, the study draws correlations that are based on a significant experimental bias and thereby oversimplifies the complex pollinator population dynamics by pointing the finger at one potential factor only.
Admittedly, it is not an easy task to link the impacts of all relevant variables – but it is essential if one is evaluating correlative relationships. This study’s failure to include an analysis of the development of relevant landscape structures is a critical omission. Instead of speculating that these population changes may be linked to the application of neonicotinoids, a far more likely explanation could be simply because some bee species had less nesting sites available.
“When we look at the study results we get the impression that they are, in fact, actually measuring the impact of intensive agriculture on wild bee species”, says Dr Julian Little, Communications & Government Affairs of Bayer CropScience in the UK. “For instance, a high percentage of neonicotinoid seed-treated oilseed rape per area unit can also be an indicator for an intensive growing practice.” It is therefore reasonable to find a similar correlation if a study were to be made investigating the impact of farming intensification on the occurrence of wild bees.
Conversely, an attempt to equate correlation with causation is a dangerous trap, especially in scientific studies conducted in highly complex and dynamic field environments. Or in non-scientific terms: we all know that it is not the storks which are bringing the babies; but nevertheless there were periods where the development of stork populations in Western and Central Europe was correlated with human birth rates. Looking at neonicotinoids as the only factor impacting wild bee populations is very likely to lead to erroneous conclusions.
In addition to excluding the role of other factors, the authors based their conclusions on questionable assumptions of exposure.
To define “exposure to neonicotinoids” simply according to the presence and the proportion of neonicotinoid seed-treated oilseed rape in a 25 km2 area appears highly speculative. Many wild bee species are capable of flying only very short distances in search for food. Because there are a number of treated oilseed rape fields within an area seven times the size of Central Park in New York does not automatically mean that the bees living in this area will collect pollen from those fields.
One the most revealing observations from this study was the fact that both populations of wild bees – those allegedly exposed to neonicotinoids (the treatment group) and those which were not exposed (the control group) – declined over the time period. In an attempt to explain this inconsistency, the authors speculate that neonicotinoid residues are present everywhere, or that species not known to forage on oilseed rape, must in fact be doing so. In either case the authors essentially undermine their own study methodology and ultimately, their conclusions.
In summary, insecticide use is not the only factor that has changed over a period of 18 years in the UK. Portraying neonicotinoids as the single cause of wild bee decline is misleading because it neglects the importance of other key factors, such as landscape structures and availability of suitable nesting habitats. While this study raises interesting questions, it by no means establishes a clear cause for a decline in wild bee populations. Weaknesses in the assumptions regarding exposure, as well as the confusion over correlation and causation, are compounded by the authors’ use of highly speculative theories to attempt to bridge the inconsistencies seen between control and treatment populations. A far better approach would have been to compare wild bee species in agricultural landscapes with those living in different habitats, or to include an analysis of habitat structures within the agricultural landscape to assess their importance to wild bees populations.
“The presence or absence of potential nesting sites and other habitat structures is a primary factor affecting the occurrence of wild bee species. Together with our partners, we are working on several projects aimed at a detailed investigation of the importance of suitable habitats and foraging sources to increase diversity of wild bees” says Dr Christian Maus, Global Pollinator Safety Manager at Bayer CropScience.