The Beekeeper’s Companion Since 1861
icon of list

Science Insider

Is There Really an Insect Apocalypse?

- April 1, 2022 - Alison McAfee - (excerpt)


Questions over the real status of insect populations shouldn’t prevent conservation efforts

Most children can’t wait for summer to arrive, but I was not among them. I spent my childhood on a far-flung homestead on British Columbia’s coast, surrounded by rocky beaches, thick forests, fog, and choppy waters. It was beautiful, and comfortable, except in June. I dreaded June. Like clockwork, June was when the ants came out.

During the annual “ant hatch,” winged carpenter ants emerged from their nests in huge numbers to mate and establish new colonies. I would spray a moat of Raid around my bed, of little use, as the ants would drop from the ceiling. The dreadful tickling of ants in my sheets, or worse, the pinch of their mandibles, would keep me up most nights.

I feel differently about the ants today. I still visit my childhood home, but there is nowhere near the same volume of ants that there used to be, and the void is unsettling. In what has been popularly coined “the windshield effect,” referencing diminishing splats on the windshield while blasting down the freeway, many people have noticed similar trends of insect declines in their lifetime [see Rusty Burlew’s article in the September 2020 issue of ABJ: “Thinking like a scientist”]. But whether these trends extrapolate to a wider scale is hotly debated, and the more data come in, the muddier things get.

The German parks

Take the study of protected lands in Germany,1 which was one of the first to make headlines. Caspar Hallmann and colleagues analyzed changes in total insect biomass in nature protection areas over more than a quarter century. Their sampling protocol was systematic, but insect trapping sites were not consistent from year to year, so the authors had to aggregate data to get a sense of long-term trends across the landscape.

Still, what they found was not good. “More than 75 percent decline over 27 years in total flying insect biomass in protected areas,” the title of their paper reads, thus sparking the “insect apocalypse” media blitz, a narrative which persists today.

The threat of losing insects is worrying indeed. Insects are foundational for ecosystems, performing essential functions like accelerating decomposition, controlling plant abundance through both herbivory and pollination, and providing a huge source of food for predators. We often don’t notice insects going about their work, but we would surely notice if they were gone.

Hallmann’s results were particularly alarming because they described a loss of insects in protected areas — nature reserves, for example — not places where you would expect much high-intensity management, urbanization, or pollution. The survey also does not focus on a particular species: The style of trap the researchers used (“Malaise” traps, which look like gossamer tents), in theory, catches any insect that flies and is too big to crawl through the mesh. In practice, these are mainly ants, bees, and wasps (Hymenoptera), flies (Diptera), and a smattering of moths and butterflies (Lepidoptera).

Before Hallmann’s study was published in 2017, most, but not all, data on insect population fluxes focused on just one or a handful of species at a time, like honey bees, locusts, or monarch butterflies. With about a million insect species described by science, data on a few species is but one stroke in a mural. While some species declined, others increased, thus not causing widespread alarm. That’s why Hallmann’s work, which described total flying insect biomass — literal pounds of insect bodies — raised such concern.

Thus ensued the debate. Critics say that measuring biomass is a limitation, rather than an asset, because changes in total mass could be driven by one or more highly abundant species, and may not represent a dire impact on the community. Others worried that that sampling itself may have removed a significant number of insects from the environment, that aggregating the data obscured site-specific trends, and that perhaps the study sites were chosen for the very fact that insect abundance was already high and thus more likely to fall (notably, the latter two arguments can be, and have been, flipped on their heads to counter studies identifying no changes in abundance as well).

Some of this criticism subsided when a second study focusing on German forests and grasslands was published.2 The time scale was shorter (10 years) but methods were more rigorous. The researchers used different trapping techniques, sampled the same sites year after year, categorized insect taxa, and chose locations where initial abundances were both high and low.

Although not as dramatic, the overall picture was similar, with both biomass and species richness declining over time. Today, the best estimates are that insect biomass is declining at a rate of about 1% per year, which doesn’t sound like much, but it adds up to a substantial loss over the course of a person’s life.

“It’s hard to think of why what’s happening in Germany would be profoundly different than the rest of the developed world,” says Dave Goulson, professor of biology at the University of Sussex and author of “Silent Earth: Averting the insect apocalypse.” “In the absence of more data, it would be wise to assume that insects are declining elsewhere as well.” 

The United Kingdom

The problem is, “more data” does not always yield a clearer picture. In 2009, Chris Shortall and colleagues published a paper analysing flying insect biomass collected from suction traps (glorified vacuums) at four sites in the southern U.K.3 Similar to the German park study, these data were collected over 25 years, from 1973 to 2002. But the picture Shortall observed was not so obvious, with biomass appearing stable at three collection sites and declining at just one, mainly driven by a decline in flies and less-abundant large insects.

The trap data analyzed by Shortall are part of the Rothamsted Insect Survey, which is considered to be “the most comprehensive long-term insect monitoring programme in the world.”4 The survey also includes additional sites and sampling methods with a focus on different groups of insects, like aphids and moths. Shortall’s analysis revealed that moth abundance significantly declined at three of the four sites, indicating that some species are more widely affected than others.

In 2019, another paper, also utilizing Rothamsted Survey data, raised more questions.5 This dataset represented fifty years of moth collections using light traps — a method that attracts moths to the trap using a strong light bulb illuminated at night. The paper, published by independent researchers, showed that while moth biomass has been declining since the 1980s, before that, it was increasing. The previous gains mean that even with the recent declines, in 2017, moth biomass was still slightly higher than was recorded in 1967, when the study began.

This rise and fall of moth abundance illustrates that longer data do not necessarily yield a clearer picture. Does the fact that today’s U.K. moth abundance is similar to 1967 negate the fact that they have been declining for the last thirty years? Goulson doesn’t think so. As he says succinctly, “The direction of the trend matters.”

The United States

If data from Europe seem difficult to interpret, they have nothing on the U.S., which is where another debate has begun. In 2020, Michael Crossley and coauthors analyzed data from U.S. long-term ecological monitoring sites, and they found no evidence for widespread insect declines6 [see also the October 2020 issue of ABJ, “News and Events”].

This conclusion was met with backlash in a debate which was about as feisty as scientists get, on the record. A formal rebuttal to the Crossley paper,7 published in Nature Ecology and Evolution (the same journal that published the Crossley study), accuses Crossley and colleagues of using “unsuitable datasets” and “mistakes” in their data analysis, ultimately “urging scepticism regarding Crossley et al.’s general conclusion.” For example, some of the data that the researchers used were extracted from projects specifically aimed at supporting and restoring insect populations, where results would clearly be biased toward positive outcomes.

Crossley et al. did make some mistakes, and in turn published a response to the criticisms, many of which go into the weeds of their methods and rerunning tests with updated datasets and statistical models.8 They argue that even after accounting for the issues raised with as much rigor as the data will allow, their overall results are largely unchanged. However, the spat does reveal that large, patchy datasets require complicated, imperfect analysis methods, and some level of skepticism is not unwarranted, even when those analyses are performed correctly. 

The world

Part of why Crossley’s paper was so controversial was because it included datasets from the U.S. Midwest. That’s important because these data were also part of a global meta-analysis synthesized by van Klink et al., and published in the journal Science, which arrived at the conclusion that, globally, terrestrial insects are actually under significant decline.9 The hotspot of that decline was the Midwest — a finding not reflected in the Crossley paper, despite sharing some of the same data. What gives?

The discrepancy partially lies in the fact that Crossley included some Midwest sampling locations that van Klink did not, and those locations largely showed increases in insect abundances, counteracting the declines apparent at other sites. Crossley also notes that they analyzed not only abundance, but also species richness and evenness, two other common metrics for ecosystem health, and much of the alarm in van Klink’s Midwest data was rooted in a strong decline of two species of highly abundant grasshoppers — changes which had a huge influence on biomass, but small effect on the community structure.

Van Klink’s analysis is widely considered to be the most comprehensive meta-analysis on insect decline to date. The researchers analyzed data from thousands of sites spread amongst ….