Truffles are fruiting bodies of a subterranean ascomycete fungus common on the roots of oak, hazel, poplar, beech, and pine trees. They play a significant role in nutrient cycling and drought tolerance by providing nutrients to the tree and getting sugars in return, a relationship called ectomycorrhizal symbiosis.
Yet, the management of truffle plantations is challenging. In the wild, truffle spores are spread onto tree roots by animals like pigs, meerkats, grizzly bears, chacma baboons, and potoroo. In truffle plantations however, the process is more involved. After planting compatible trees, farmers must wait a few years until the trees have reached maturity, at which stage growers start spreading truffle spores around tree roots. If after a few years, a burnt-looking circle appears around the base of the tree, it means a strong mycorrhizal activity is taking place. Around 8-years after planting, truffles are ready for harvesting. Harvested in the winter only, a truffle plantation can be harvested every year for about 20 years. Since the fruit bodies grow 20cm underneath the ground, truffle growers have traditionally trained dogs to use their sense of smell to identify where truffles are ready to harvest. But what if there are other equally precise methods that don’t involve dogs? AGRON, an agricultural imagery analytics and drone services company in Hungary decided to test how precise multispectral imagery could be in identifying truffle growing trees for harvest.
The Technology
Building on the idea that a symbiotic relationship with mycorrhiza is beneficial for the tree, AGRON presented the hypothesis that trees with ready-to-harvest truffle fruit bodies might show a better physiological condition compared to other trees without mycorrhiza. Thus, they decided to use the MicaSense series RedEdge-MX multispectral sensor to monitor the physical and physiological conditions of planted trees.
The Project
Agron tested the technology in a young plantation near Martonvásár, Hungary, where the first ready-to-harvest spot appeared under a single hazel tree in 2018. They used this tree as a reference sample to find possible spots in the 2019 season.
After mapping the field, AGRON used sensor data to compare the multispectral profile of the sample tree with the rest of the plantation. The data collected was analyzed using their online platform AGRONmaps and the Principal Component Analysis (PCA) to evaluate and sort the trees (Figure 1). The PCA showed a group of trees similar to the reference sample concentrated in three spots on the map. Mature truffles were found on two of the three spots, with the third one being ready for harvest in 2020. The map below shows the hazel trees in light blue, and the oak trees in light pink.
The RedEdge-MX proved to be a valuable tool in differentiating spectral signatures of trees and subsequently identifying those with physiological characteristics that indicate they are ready for harvest. In the future, AGRON can use these models to help truffle farmers carry out future harvests without the extra cost of canines and their handlers.
