Scientists released 600 beetles into 30 identical landscapes; after 13 generations, some traveled three times farther than others despite the same lab conditions

Ecology is often seen as a science in which, once you know the main drivers of a species’ movement, you can often make reasonable predictions. However, a 2009 lab-based experiment shows this may not be the case. In September 2009, Brett Melbourne and Alan Hastings published a study titled “Highly Variable Spread Rates in Replicated Biological Invasions: Fundamental Limits to Predictability,” in the journal Science, showing that even when conditions in the environment are the same, outcomes can be very different.

Under identical landscapes, beetles showed very different patterns of movement. Although this study was conducted over a decade ago, the predictions and questions it raised are still relevant today, particularly in the US, where, according to UC Davis, there were hundreds of damaging invasive species at the time of the study.

A beetle experiment with no room for excuses


In this study, 30 identical landscapes were created using identical habitat patches, as documented in a 2009 article from the National Science Foundation, the organisation which funded the work. Conditions were equal throughout, and each started with 20 red flour beetles (which are attracted to wheat flour). That’s 600 beetles in total, left to spread for 13 generations. The experiment lasted 15 months, during which the fastest-spreading populations advanced across as many as 31 patches, while the slowest-spreading ones covered roughly a third of that distance. Thus, the same threefold gap in outward reach referenced in the headline, measured here in the number of identical patches each population crossed rather than a raw distance figure.

There isn’t just one reason behind this. Instead, the study shows that the natural variability could be due to chance or genetics. This seems to have a large effect on how species spread. Melbourne compares it to weather forecasts. The National Science Foundation noted that ecological systems are inherently unpredictable because of stochasticity in environmental and biological processes, and ecologists have seldom quantified this unpredictability.

Why it is relevant to you

This isn’t just a problem for a far-away ecologist. Hastings notes that the results indicate that it will not be as simple as hoped to list all factors driving an invasion. If invasions are unpredictable, then they are also hard to manage. As mentioned, there are hundreds of damaging invasive species within the US, including kudzu, zebra mussels, and the light brown apple moth.

According to a 2003 U.S. Geological Survey report, the zebra mussel arrived in North America in 1988 via ballast water from a transatlantic freighter. It colonised parts of Lake St. Clair. Within less than a decade, it spread to all 5 Great Lakes. However, a separate USGS report published in 2013 points out that by 2010, zebra and quagga mussels had been documented in 131 river systems and 772 inland lakes, reservoirs, and impoundments within the US.

The price tag of being caught unaware

It costs a lot of money to be caught unaware. Worldwide, biological invasions have cost at least $1.288 trillion (in 2017 dollars) between 1970 and 2017, according to a 2021 research paper titled “High and rising economic costs of biological invasions worldwide,” published in Nature. The study found that North America had the highest costs overall. It also showed that the true figure is severely underestimated and that costs are increasing consistently by three times every decade.

What the beetlescan and can't teach us

Although these were beetles in a lab, ecologists will increasingly be asked to predict outcomes from climate change, habitat loss, and biodiversity loss. What the experiment shows ecologists can't do is treat an invasion's spread as something a complete list of factors could pin down to a precise number; chance and genetics alone were enough to send otherwise-identical populations down very different paths. What it shows they can do, and increasingly must, is build that uncertainty directly into their models and forecasts rather than treating it as noise to be ignored. More research into this topic is needed. However, there is something comforting about knowing that even with all variables controlled in a lab, there is no guarantee the same results will occur. It means you always need to allow for some surprises when planning.