We are allegedly losing “only” 9% of terrestrial insect populations globally every decade – far less than predicted in most local studies. As for freshwater insects, their situation would be substantially improving, with their numbers increasing by 11% per decade. Are things not so bad after all? The idea that false data can come from scientific publications and degrade the quality of public debate is a terrifying prospect for any researcher.

For four years, ecologist Laurence Gaume (French National Center for Scientific Research, CNRS) and economist Marion Desquilbet (National Research Institute for Agriculture, Food and the Environment, INRAE) invested substantial time and energy, in line with their concerns, to examine InsectChange, the largest database on the global evolution of insect populations, which is the source of these estimates.

The result of this painstaking work was published on October 8 in Peer Community Journal (PCJ) and warns of a spectacular accumulation of errors and systematic biases in this database. These discrepancies are likely to significantly underestimate the collapse of insect populations and create confusion about the causes of this decline while exonerating the dominant agricultural model.

What the work of the two French researchers reveals is not only alarming for the future of biodiversity, said ecologist François Massol (CNRS, Université de Lille), who supervised the publication. “It also calls into question the practices of the most prestigious scientific journals, which are sometimes reluctant to acknowledge errors even when they are serious and obvious.”

Four main categories

The critical work of the two researchers was accompanied and published by a young non-profit scientific journal, PCJ, created and run by researchers keen to break free from major publishers. According to Gaume, the work “stems from our determination to preserve the integrity of science, but also to ensure that exact and unbiased information is diffused to the general public and decision-makers.”

Presented in 2021 in the journal Ecology, the disputed database was already the basis of a meta-analysis published in 2020 in the journal Science, which drew widespread criticism but has since been cited in over 1,000 subsequent studies. Meta-analyses are large-scale studies that compile all available results on a given issue to provide theoretically more reliable answers than each individual study on its own.

In this case, the reality seems very different. The two researchers sifted through the 165 datasets in the InsectChange database and identified 553 problems, which they classified into four main categories: errors, inconsistencies, methodological problems and lack of information. Almost all data sets – 161 out of 165 – were affected by at least one problem. However, beyond the accumulated isolated errors, the entire methodology is being called into question.

You have 56.13% of this article left to read. The rest is for subscribers only.