The windscreen phenomenon: anecdata is not scientific evidence

The windscreen phenomenon refers to people’s perception that there are fewer insects being splattered on their windscreen than they used to see. It is one of the most common anecdotes presented as evidence of global insect decline in the Insectageddon stories. But anecdotes are not scientific evidence. Anecdotes describe local conditions, not globally-relevant facts.

People often claim the ‘windscreen phenomenon’ is established evidence and proven fact. But a search of academic journal databases returns only one published study that has used car windscreens to measure changes in local insect abundance. In that study, Anders Møller compared insect abundance (although it’s not clear from the Methods if he actually measured density) with breeding rates of insectivorous birds in an agricultural landscape in Denmark. Data was collected in the same way at the same location for 20 years, which is very impressive, and analysis showed an 80% decline in insects across the period.

It’s a good study and the parallel declines in birds and their insect food is an important result – it reminds us that ecosystems and our survival depend on cascades of interactions, not single species. But the data represent one location with a specific environmental context, i.e. climate change effects, agricultural land use, baseline insect populations etc. For the data collection, the author drove a total of 26.2 km every year, on the same two roads in the study region. This tells us about insect splatter on car windscreens in that location, not the world. There are more than 21 million km of roads across the world.

There are a lot more studies measuring broader-scale effects of roads on insects. But, as with all ecological questions, there is never one single factor influencing a dataset. A recent review found only 50 studies that had investigated the ecological impact of roads on insects (the review was published in 2015 and there have been a few more studies published since then). Overall the authors found generally negative effects on insect abundance and diversity. They also found there are lots of factors within the broader ‘road – insect’ interaction that affect results.

For example, during my PhD, we investigated edge effects on wild insect pollinators in almond orchards (the edge was a two-lane low-traffic road between monoculture almond orchards and native mallee woodland). The road wasn’t a barrier, but we found that the edge effect varied across time as the floral resource pulse of the orchards on one side peaked and declined.

Roads come in many shapes, sizes, and conditions, a fact that is usually ignored in the ‘windscreen phenomenon’ soundbite. Roads create edges in the landscape… and edge effects are an ecological phenomenon that are notoriously difficult to measure and generalise. Measuring on-road mortality vs. movement across the gap will yield different results. Studies that only sample the roadway don’t tell us how these results compare to insect assemblages away from the roads. Sampling method also influences the type of insects collected in the sample – measuring dead insects on the side of the road will yield mostly larger-bodied insects compared to measuring biomass of windscreen splatter, or sticky traps on cars. Sampling in the day time may yield different results to sampling at night.

Traffic volume, time of day, season, width of road corridor, land use on either side of the road, paved vs. unpaved, insect sex or life history traits, baseline insect populations, and pollution can all influence the observed effect a road has on an insect. Most studies that have tested effects of roads on insects have found variation depending on the environmental context. For example, this study on insect mortality on some Yorkshire roads in 1938 showed that the number of animals killed varied across taxonomic orders and was most numerous in summer months (not surprising, this is when insects are most active). The author also notes the influence of the sampling method (invertebrate carcase counts by bicycle) on results, saying that those insects “which have met their end on car radiators, windscreens, etc., or died off the road itself, cannot be included”.

And, if we’re talking about windscreens, what about the type of car? As far as I know, no one has tested this effect. But going by the number of registered patents for devices and designs that prevent insect splatter on windscreens, I think there is a scientific experiment waiting to happen if anyone can get the funding to do it.

Anecdotes are powerful storytelling tools and useful sources for scientific hypotheses. But on their own, they are not scientific evidence. For every person claiming they have seen fewer insects on their windscreen lately, there is someone claiming they have seen just as many as they used to.

We can’t take any of these anecdotes as evidence without knowing more about the observation: Have these people driven the same roads repeatedly at the same time of year? Did they drive during plagues and swarms? Did they drive during winter, when few insects are active, or summer when lots of insects are active? Some adult insects only live for a few days and emerge en masse – did the person drive before, during or after the mass emergence? What is happening in the surrounding landscape to influence local insect populations? Has the person’s perception been influenced by hearing other people talk about the windscreen phenomenon?

So the windscreen phenomenon is not proof that insects are declining. Yes, it’s clear that some aspects of roads and traffic can kill insects. But, taking into account the short duration of adult insect lives and their high reproduction rates, there is currently very little understanding of how this mortality scales up to affect populations and insect diversity in the long-term. All this proves is how little we know about the smallest animals on Earth.

An extreme case of windscreen insect splatter: driving in western New South Wales during a locust plague in April 2010.

© Manu Saunders 2019

8 thoughts on “The windscreen phenomenon: anecdata is not scientific evidence

  1. Alex SL July 7, 2019 / 10:46 PM

    I appreciate that an anecdote is not data, and I appreciate all the caveats one has to consider. But to a certain degree this can come across as a similar argument as those used in climate change denial. Yes, there aren’t perfect 50 year time series for 50 sites across the planet. But first, the data that do exist plus the undeniable facts of habitat loss and change in agricultural practices lead to an obvious conclusion, taken together. It is not as if half of the studies that are available show a massive increase in insect abundance, as far as I understand. Second, by the time there are perfect data it will be too late. Third, even if somebody presented 50 year time series for 50 sites across the planet, one could still argue that we don’t know enough until there are data for, say, an additional 200 sites covering this so far neglected road type and that type of landscape. And so on forever.

    I guess what I want to say is that we always base tentative conclusions on what we have instead of perennially saying that we can’t say anything because the data could still be better, be it in climate science, ecology, phylogenetics, etc.


    • Manu Saunders July 8, 2019 / 8:51 AM

      Disagree that this is a similar argument to climate change denial. I am concerned about the mis/dis-information being presented as fact in the insect apocalypse stories, precisely because of the implications for public understanding of science in the long term. In the case of the windscreen phenomenon, there is no data to show this is true and a lot of reasons why it may not be a valid generalisation to prove global declines. This is not denying that insects are under threat.


  2. The Snail of Happiness July 9, 2019 / 4:38 PM

    I wonder how improved aerodynamics of cars have affected windscreen spatter. My first car was rather like a box on wheels (as were many cars at the time), whilst my current one is much more sleek, with a much more sloping windscreen – I’m sure such changes in shape will have influenced impacts.

    Liked by 2 people

  3. Nildes February 13, 2020 / 8:06 PM

    Cars are wind tunnel tested and designed to be more “slippery” so that they use less fuel and meet other design targets. The change in slipstream effect over time will certainly reduce the number of insect impacts at some speeds so anecdotal evidence of observations over time doesn’t account for this change.


What do you think?

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s