Discovering the secret networks between plants and fungi

Mycorrhizal fungi cannot survive on their own. They depend on the carbon produced by plants through photosynthesis, which in turn obtain their nitrogen and phosphorus from fungal networks. But how do these elements circulate in the mycelium filaments, which are thinner than a human hair?

For the first time, researchers have been able to observe the development of fungal networks and track the flow of nutrients within the filaments, from the plant to the fungi and vice versa.

This success is the result of close international collaboration between scientists: the Mosaic project team from the Inria Lyon Centre, the AMOLF institute and the Vrije Universiteit Amsterdam in the Netherlands, as well as Princeton University in the United States.

“This discovery is the result of several innovations in different disciplines,” explains Christophe Godin, director of research at Inria and head of the Mosaic project team.

How did the story begin? It began with Corentin Bisot, an engineering student at the time, studying for a Master’s degree in Soil Biology, at École Polytechnique in Paris and Wageningen University in the Netherlands. “After seeing a TED talk by biologist Toby Kiers from the Vrije Universiteit Amsterdam on mycorrhizal fungi, I got in touch with her,” he recalls. “These fungi are fascinating. They live in symbiosis with the majority of plants on the planet.” When he discovered a prototype developed by the researcher with the AMOLF Institute, mainly with Loreto Oyarte Galvez, he came up with a new approach: studying mycorrhizal fungi using machine learning on images taken by this robot.

Corentin Bisot then joined AMOLF’s “Physics of Behaviour” team for an internship, followed by a thesis. Here, he used the robot previously designed by Loreto Oyarte Galvez and the whole team to obtain a large number of images of unprecedented precision. The aim was to create a high-resolution time-lapse video of the growth of the entire symbiotic network at the scale of each mycelium thread.

“However, we were faced with major obstacles, in particular processing the gigantic mass of data collected”, explains the director of the “Physics of Behaviour” team, biophysicist Thomas Shimizu.

“The high-throughput imaging system can track more than 500,000 fungal nodes simultaneously and measure 100,000 nutrient flow trajectories within the networks. So, after three years of collecting images with the robot, we acquired as much data as in 100 years of microscopic observation by a human!”

Corentin Bisot suggested calling on Inria’s Mosaic project team, which specialises in mathematics and computer science for the morphogenesis of plants and other living organisms. “Corentin asked Thomas and myself to become his thesis supervisors and we helped him to explore and model this mass of data”, says Christophe Godin.

But before reaching this stage, the researchers had to overcome other challenges. “To make the growth of the mycelia visible to the imaging robot, it had to take place in a single layer in a transparent medium”, explains Loreto Oyarte Galvez. To achieve this, the fungi were grown in a gel in symbiosis with the root system of a plant.

Then, in order for the robot to capture usable images of the mycelium, some ingenuity was required. The team used Petri dishes separated in two by an acrylic and nylon barrier, with the root compartment on one side and the fungal compartment on the other. A cellophane membrane was also applied to the gel to restrict mycelial growth to a single layer in the fungal compartment. Without this trick, the filaments would have penetrated the gel and spread out in three dimensions.

The robot developed by the AMOLF Institute then records images 24 hours a day, 7 days a week, for weeks on end. The growth of the mycelium can be seen as black lines on a white background. In order to improve the contrast and visibility of the lines, the images first had to be mathematically transformed. And to obtain the resolution necessary for the study, the images were taken at a very local level. In order to expand to the scale of the entire network, Corentin Bisot developed a series of algorithms to reconstruct the dynamics of the entire mycelial network in the computer.

The mycelium is also a living network, where the filaments cross, merge and become entangled, adding to its complexity.

“The difficulty lies in being able to extract relevant data from the images obtained,” says Corentin Bisot. “Christophe Godin’s support has been decisive in solving this problem.”

The young researcher developed numerical tools, based on graph theory, which provided a better understanding of which pathways in the mycelial network were most involved in transporting nutrients. “The experiments carried out showed that the most central lanes ensured faster flows, which makes it possible to identify them and create a model that extrapolates to the global scale”, summarises Christophe Godin.

Moving forward and understanding the role of these flows required the use of models of the processes contributing to the growth of these networks. The idea of a model was inspired by a popular work on fungi that Corentin Bisot was reading.

“The book mentioned a 1989 publication proposing a mathematical model of growth that proved to be relevant for interpreting what we were seeing in the Petri dishes,” he explains. “We used it to analyse how fungal networks develop in response to various conditions. Mycorrhizal fungi appear to assess the carbon supply and nutrient demand of plants, which guides the way they build their networks, in the form of self-regulating waves.” 

By assessing supply and demand, the fungi optimise the mycorrhizal surface without using too much energy. “This ability of an organism that has no brain or central nervous system to assess its environment, make decisions and make choices on an ongoing basis to regulate its growth and optimise its nutrient supply network is remarkable,” says Thomas Shimizu. And its impacts do not end there, as Christophe Godin points out: “These discoveries could have implications for the study of other biological networks, such as the blood or neuronal systems.”

The researchers now have detailed maps of the fungal networks, showing where each filament begins, merges and ends. A new challenge now awaits them: using these maps to gain a better understanding of how fungi work and the learning tasks they can perform underground. Because they have yet to reveal all their secrets.