Hyperspectral insight provides early warning of threatening crop disease
Subtle shifts in the way light interacts with crops plus changes in plant temperature are just two of the early signs of infection used to detect some of the most threatening plant diseases for Australian agriculture.
Among these diseases is Xylella fastidiosa, a bacterial pathogen known to infect up to 500 plant species, including olive, almond and citrus crops and ornamental trees like oaks and elms.
Although not yet present in Australia, Xylella subspecies and the insects that help to spread these diseases are at the top of the National Priority Plant Pest list for Australia’s Department of Agriculture, Water and the Environment (DAWE).
At the University of Melbourne, Professor Pablo Zarco-Tejada is leading a biosecurity project for DAWE using hyperspectral and thermal imaging to develop models to detect Xylella in Australian olive and almond orchards to protect against an outbreak.
Camera-mounted aircraft quickly cover vast areas, producing images with information on the health of plants below. The first flight in the project covered 1000 hectares of almond trees in the Mildura region on the Victoria–NSW border.
The light and level of detail provided by hyperspectral imaging is beyond the range of human vision, and the combination of imagery plus computer algorithms allow Professor Zarco-Tejada and his team to produce maps identifying healthy and infected areas within crops. Plant pathologists check the results of the algorithms using in-crop evaluations of plant health on the ground, with both visual assessment and with laboratory testing.
Collecting images of healthy crops this way provides baseline data to allow this imaging process to deftly identify when stressors such as the deadly Xylella pathogen may be present.
Professor Zarco-Tejada has extensive experience in this area. He previously led time-critical research for the European Commission to combat the deadly bacterium. His work has also included projects with the Joint Research Centre in Italy and the Spanish National Research Council.
In Europe, Professor Zarco-Tejada helped develop the methodology that allows remote sensing to detect early signs of infection before it is visible to the human eye through these hyperspectral images, with up to 90 per cent accuracy.
Xylella has caused problems in the Americas for many years, especially affecting grape growers in California.
In 2013 the bacterium was identified in Europe, having likely travelled with tobacco plants from Costa Rica. It soon destroyed much of the olive industry in southern Italy. Professor Zarco-Tejada says some of the olive trees killed were up to 1000 years old. The infection led to a huge loss of heritage and left a lingering economic impact for growers.
As a cure or treatment is yet to be discovered, tree-culling is the only viable option for growers.
We need to have detection methods in place in case Xylella is found in Australia; that’s why we have been active in preparedness, and New Zealand as well, Professor Zarco-Tejada says.
We are using manned aircraft to collect images of thousands of hectares of almond, olive and table grape vines in Australia. We are using the University of Melbourne’s Airborne Remote Sensing Facility, which comprises high-resolution hyperspectral and thermal imaging sensors mounted onto a light aircraft.
The images of crops, which will include signs of common stressors such as lack of water or nutrients, will provide baseline data for the varieties grown here, and these will be added to the existing database which includes data from Xylella-infected plants. Then we will develop models to detect Xylella infection for varieties currently grown in Australia, which will be ready for use in case there is an outbreak here.
Professor Zarco-Tejada says a preventative approach backed by rapid detection is critical.
The earlier you detect it, the sooner you can stop the spread. This is a valid approach for any plant disease, he says.
Fast detection is critical because in woody crops such as olives or almonds, infected trees that appear to be asymptomatic may be already spreading the disease to other trees and crops for months before they appear infected.
Professor Zarco-Tejada says the technology and methodologies developed for Xylella detection could be extended to various crop diseases.
Other high-priority disease threats to Australian agriculture include the fatal Huanglongbing plant infection, previously known as citrus greening disease, and Panama disease (tropical race 4), a form of fusarium wilt that kills banana plants.
Professor Zarco-Tejada’s work combines expertise from Melbourne School of Engineering and Faculty of Veterinary and Agricultural Sciences at the University of Melbourne. The current research project will be complete in late 2021.