Mapping the world’s fungal networks with machine learning – Geographical

Life is underpinned by fungi. Although most conspicuous when mushrooms and other fruiting bodies pierce the soil surface, their true heft lurks below ground. Fungal filaments extend through the soil in networks of mycelia, but we know relatively little about them and questions abound, including where on Earth they are and in what diversity and abundance. Vast and powerful, mycelial networks sequester carbon, hold soils together and supply as much as 80 per cent of all nutrients to terrestrial plants. In just one hectare of grassland, the extent of fungi is equivalent to around 12 million times the length of the Amazon River. But just like the distributions of plant species, those of fungi are almost certainly shifting in response to climate change. They’re also being ripped up by deforestation and land-use change.
A team of scientists has embarked on a mission to fill our fungal knowledge gaps by mapping the world’s subterranean mycelia. ‘Protecting underground fungal networks begins with mapping them,’ says Toby Kiers, founder of the Society for the Protection of Underground Networks (SPUN), an organisation that has been set up to produce detailed information on mycelial whereabouts and make it available to governments and policymakers so that protection can be better targeted.
Kiers’s team will first draw upon data from the GlobalFungi database, which pulls together field data from a global network of fungal studies. Using this information, they will build algorithms to elucidate which ecosystems are likely to have high fungal diversity and which are understudied. Such areas – including Patagonia and the Russian taiga – will be prioritised in future sampling work, which will draw upon the help of citizen scientists. ‘We’re calling them myconauts,’ says Kiers, ‘and our aim is to collect 10,000 fungal DNA samples in the next 18 months across diverse ecosystems across the globe.’
The ultimate aim of the project is to utilise machine-learning techniques in order to both predict regions with high fungal diversity and characterise those most threatened by land-use changes. In the future, the work will also build scientific consensus on the ecological role of fungal networks across the globe. ‘Each observation in our fungal biodiversity dataset is geo-located, meaning it can be coupled with hundreds of layers of remote-sensing data on climate, vegetation and soils,’ says Kiers. Other organisations, such as the European Space Agency (ESA), are working on planetary models of the Earth’s ecosystems, but they are lacking sophisticated data and maps of fungi.
‘Machine-learning algorithms give us a level of prediction that was nearly impossible in the past,’ Kiers adds. SPUN scientists hope that their work will pave the way for targeted conservation policies. ‘Fungal networks have been a global blind spot in conservation and climate agendas, but they are an ancient life-support system and they need to be protected just like above-ground biodiversity.’

SUBSCRIBE TO OUR MONTHLY PRINT MAGAZINE!Subscribe to Geographical today for just £38 a year. Our monthly print magazine is packed full of cutting-edge stories and stunning photography, perfect for anyone fascinated by the world, its landscapes, people and cultures. From climate change and the environment, to scientific developments and global health, we cover a huge range of topics that span the globe. Plus, every issue includes book recommendations, infographics, maps and more!
Source: http://geographical.co.uk/nature/item/4278-mapping-the-world-s-fungal-networks-with-machine-learning