News - November 15, 2017

‘Alarming insect study unsound’

Roelof Kleis

The study in which researchers from Nijmegen conclude that three quarters of the German insects have disappeared is based on flimsy data. This was expressed by Wageningen entomologists.

Entomologist Kees Booij. © Guy Ackermans

In an article on Kennislink, the entomologists Kees Booij and Theodoor Heijerman mop the floor with the study that was published last month in PlosOne. The article was followed by many alarming articles in media in the Netherlands and across the borders. British newspaper The Guardian even wrote of an Armageddon that threatens all life on Earth. The Dutch House of Representatives wants to take measures to better protect insects. 

But Is that story true? Is there really just a quarter of the insects left compared to thirty years ago? Nonsense, say Booij and Heijerman on Kennislink. Both have fundamental points of criticism on the statistical substantiation of the article. ‘PlosOne should never have accepted this in the current form’, says Booij when asked. ‘I do not understand how it is possible. But unfortunately, these things do happen.’

The measurements were only done in a very small part of Germany, and many spots only saw a single or a few measurements over the 27-year period.
Kees Booij

The researchers of Radboud University used a long-term dataset of the Entomologischer Verein Krefeld (Entomological Association of Krefeld) which put insect traps in 63 German nature reserves. According to the researchers, the average daily catch dropped from 9 grams in 1989 to 2 grams in 2016. This is a decline of approximately three quarters. But according to the Wageningen researchers, this is an ungrounded jump to conclusions.

The primary point of criticism that Booij and Heijerman have is about the study itself. ‘This is not a monitoring study, as that would have required the yearly measuring of biomass in a large number of areas across Germany. The measurements were only done in a very small part of Germany, and many spots only saw a single or a few measurements over the 27-year period. This means the setup is fundamentally flawed and unfit to monitor a general trend.’

According to both entomologists, the dataset is unsuitable to perform trend analyses on. The data show significant fluctuations in biomass between the years. According to Booij, this is a known point in insects. The number of insects (biomass) can fluctuate up to a factor of ten in between weeks or years. The researchers are skating on thin ice by trying to deduce a trend from these data.

It is always an equilibrium between reliability and news value.
Kees Booij

The first year of measurements, 1989, was apparently a good year for insects in Germany. But according to Booij, that spike in the data is the results of merely six traps, which, to make matters more interesting still, were all located close to each other in the vicinity of Bonn. ‘Without that spike, not much remains of the diminishing trend.’ He does not understand why the reviewers did not see through that. Or perhaps he does: ‘It is always an equilibrium between reliability and news value.’

Upon the publication of the article by PlosOne, Booij immediately responded online. Neither the editorial office nor the authors of the article replied to his reaction. He is currently writing a scientifically substantiated response to the article. With his writing on Kennislink, he already throws the first cat among the pigeons.