Pollinators are declining worldwide, with pesticides a major culprit. It’s time to apply the big data approach of molecular medicine to the problem. The future of sustainable agriculture is at stake, says Turing Fellow Yannick Wurm, Reader in Bioinformatics at Queen Mary University of London.


First people noticed bees were in decline in the UK. Then we realised that other pollinators were also suffering. Now we know that all kinds of insects are affected. In addition to being of intrinsic value as part of the Earth’s rich ecosystems, insects are responsible for the pollination of about two-thirds of the UK’s crops.

There are many reasons for the disappearance of insects, including climate change and the rise of monoculture agriculture, but the effect of broad-spectrum pesticides is undoubtedly a big contributor. It has been too easy to get a pesticide approved: you need to demonstrate it is “safe” for a set of species, but the honeybee was the only pollinator species considered. A portion of a clearly poisonous dose was deemed safe. It’s an arbitrary criterion, and regulatory processes didn’t initially consider effects other than death, or the effects on the thousands of other bee species, butterflies, hoverflies, moths, or beetles that could also be exposed (regulatory evaluation processes are now already much improved).

Consider neonicotinoid pesticides. They are broad-spectrum neurotoxins that became popular since being introduced in the 1990s. The neonicotinoid is applied to a seed and the plant distributes it through its tissues as it grows. Neonicotinoids were thought to be perfect – only insects eating the plant would be exposed. Yet 20 years later, the most popular neonicotinoids were banned for outside use across the European Union. They have been found to reduce cognitive abilities, foraging performance, and ultimately the survival of insect pollinators.

A new, molecular approach to protect pollinators

To protect pollinators, we need a new way to evaluate pesticides, and modern medical approaches show how we can do this. Medicine has changed dramatically over the last 15 years. Nowadays, if you get a tumour, we sequence it to find out which changes in gene sequence or gene activity have occurred compared with healthy tissue, and this informs the treatment.

Take that idea and flip it around. If we have detailed information about how healthy pollinators use their genes, we can compare it with similar information from pollinators that have been exposed to a pesticide. If we expose a single pollinator to a pesticide, we can now obtain tens of thousands of datapoints on how it is reacting physiologically to that toxin. We are pursuing this type of work at an even larger scale.

To do this, of course, we need huge amounts of genomic information. Thankfully, the cost of genetic sequencing has dropped about 50,000-fold in the last 12 years. That’s a game-changer. (As another example, the genomes of all 60,000 species of animals, plants and fungi in the UK are now being sequenced at the Wellcome Sanger Institute.)

"If we expose a single pollinator to a pesticide, we can now obtain tens of thousands of datapoints on how it is reacting physiologically to that toxin. We are pursuing this type of work at an even larger scale."

Using our molecular approach, many different questions can be asked, such as:

  • What genes determine a species’ susceptibility?  Do some species have pesticide detoxification genes that others lack? 
  • Which detoxification processes are effective against which pesticides?
  • How much do effects differ between pesticides or between species? Or between different life stages?

Once we have identified susceptibility genes in a large number of insects and increased our understanding of how those genes affect responses to pesticides, a quantum leap becomes possible. We could create a predictive model, based solely on genomic data, to anticipate how sensitive a given pollinator will be to a particular pesticide.  

This type of high-resolution, big-data exploration is impossible with traditional approaches used to examine pollinator health, but it will eventually make it a great deal easier to develop pesticides that kill pests while minimising harm to the pollinators that we love and need for the stability of our ecosystems.

Solenopsis invicta ants on their genome. By Romain Libbrecht and Yannick Wurm
Solenopsis invicta ants on their genome. By Romain Libbrecht and Yannick Wurm

We need to build up this body of knowledge. Tests are needed on large numbers of species, across life stages, and large numbers of pesticides. My lab is pioneering some of this work thanks to funding from Natural Environment Research Council and Biotechnology and Biological Sciences Research Council, and with the help of collaborators in the UK and abroad. 

I believe our new, big data approach will eventually become standard practice, but we still need to champion it, because the development and evaluation of pesticides remains too blunt. To this end, I’m encouraged when I find younger biologists on-board with our direction: they understand the molecular approach in biomedicine, and many are acutely aware of the challenges facing the environment.

Edited by Sean O'Neill

Further Reading

Healthy pollinators: Evaluating pesticides with molecular medicine approaches.
F. López-Osorio, Y. Wurm. Trends in Ecology & Evolution.

Caste‐ and pesticide‐specific effects of neonicotinoid pesticide exposure on gene expression in bumblebees.
T.J. Colgan et al. Molecular Ecology.

Cover photo: B. terrestris, foraging on lavender, by Andres Arce