> To answer this, we used a combination of approaches. First, we conducted a global meta-analysis of studies in which researchers measured the length of fungal hyphae in local soils using microscopes. This length of hyphae per unit of soil is known as “hyphal density”, and it provides a way to quantify how much fungal network is present in a given amount of soil. We assembled data from 322 studies, representing more than 16,000 individual soil cores collected across 9 global biomes. This gave us one of the most comprehensive datasets of arbuscular mycorrhizal fungal hyphal density compiled to date, spanning ecosystems from forests and grasslands, to drylands and farms.

> We then used these observations to train a machine learning model. The model was built using dozens of geospatial environmental layers, including climate, vegetation, soils, and land use. By learning how measured hyphal density varied across these environmental conditions, the model allowed us to predict the density of AM fungal networks across Earth’s terrestrial ecosystems at a fine spatial resolution of approximately 1 km². We also used spatial uncertainty analyses, including bootstrapping, to understand where predictions were more or less certain.

This is absolutely mindblowing work! One of the best things I've ever seen on the internet. I only wish it was granular enough to help identify mother trees[0] in individual forests

[0] https://www.scientificamerican.com/article/mother-trees-are-...