Yield hopes carried on the power of light
GroundCover™ Issue: 136 September – October 2018 | Author: Gio Braidotti
A looming food security crisis arising from the wheat yield ceiling has inadvertently inspired the development of a remarkable new breeding technology that may break this production impasse
An innovation boom with major crop production potential is underway in wheat pre-breeding as Australian scientists shift their grain-yield research in bread wheat (and rice) from the harvest index to photosynthesis traits.
By creating opportunities for plants to make more sugars from available sunlight, researchers have demonstrated that enhancing these photosynthesis traits can significantly increase a crop’s biomass. Computer modelling then makes it possible to understand their yield impact, in ways that also factor in realistic climatic conditions and agronomic practices common across Australia’s wheat-growing regions.
Already, commercial wheat-breeding companies are expressing keen interest in these ‘phenotyping’ advances.
However, these advances are merely the lead items within a research pipeline that is also radically enhancing gene manipulation technology to ‘rewire’ entire metabolic pathways – photosynthesis being just one.
A leading researcher in this field, Professor Robert Furbank of the Australian National University, says these new technology platforms would have been “inconceivable” five years ago.
Among the most advanced developments is a virtual counterpart of the fictional Star Trek tricorder – a multifunction, hand-held device used (cinematographically) as a sensor for scanning the environment and collecting, analysing and recording or transmitting data.
The real-world agronomic tricorder can rapidly measure photosynthesis rates in wheat leaves by scanning reflected light.
Called ‘hyperspectral reflectance’, it uses artificial intelligence (AI) to extract information about photosynthetic rates inside the plant from differences in the way light is reflected from leaves.
This new technology has made it possible to rank Australian wheat varieties for their photosynthetic efficiency, with a 30 per cent difference detected between the best and worst performers. The differences grow bigger within the broader gene pool when wilder genetic resources, such as landraces, are included.
The identification of genetic diversity behind photosynthesis rates is now driving genetic gain in breeding programs and also making it possible to detect the genetic basis of high-performing traits. Eventually, this technology platform could create prediction-based breeding capability that relies on genomic selection (see GroundCover™ issue 135, July – August 2018, ‘Computer ‘brain’ trawls vast datasets for crop advances’).
While Australia is proving an epicentre of photosynthesis-driven innovation, some of this highly collaborative work is taking place within global R&D initiatives charged with lifting stagnating wheat and rice yield gains.
Included are the $100 million International Wheat Yield Partnership (IWYP) and the C4 Rice program, which is funded by the Bill and Melinda Gates Foundation. Australian laboratories are playing important roles within both consortia, with the GRDC investing directly in IWYP projects.
In addition, the Australian Research Council (ARC) has funded the Centre of Excellence for Translational Photosynthesis, which is headed by Professor Furbank and is active in both the international wheat and rice consortia. This ability to work on both bread wheat and rice has important implications, given that important advances in manipulating the rice genome are being realised and will likely drive additional innovation in wheat breeding.
“In the first Green Revolution, Norman Borlaug’s pioneering work was all about selecting for harvest index and grain number,” Professor Furbank says. “Now we have run out of capacity to improve these traits, which is why we have seen this shift to photosynthesis. While many of these global initiatives only got underway in the past five years, we are already seeing impressive progress.”
Light an information carrier
Because light is how humans see the world, it is easy to forget that it is actually an electromagnetic phenomenon – the product of interactions between electric currents and magnetic fields.
It is the negatively charged electrons of atoms that are quintessential to electromagnetism, as it is these particles that absorb and emit packets of electromagnetic energy (called photons) during changes in energy states. In turn, these electromagnetic emissions can be used to fingerprint the identity of the emitting atoms (even in distant suns) using an optical spectrometer.
Besides information about atomic structure, humans have learnt to exploit electromagnetic fields to carry digital information (as occurs in wi-fi transmissions at the microwave range of the electromagnetic spectrum).
Now researchers at the ARC Centre of Excellence for Translational Photosynthesis have further exploited the information-savvy nature of the electromagnetic spectrum to better understand leaf physiology.
“It turns out we can extract all sorts of information from the spectrum of light reflected from leaves,” Professor Furbank says. “The information we harvest then allows us to rank genetically diverse wheat lines, select the best performers for photosynthesis and find the genes responsible.”
Besides measuring the rate of key photosynthetic enzymes, hyperspectral reflectance can measure how well different cultivars absorb light (their light-harvesting capacity) and even the nitrogen level in a crop.
“This new technology means we can measure in 20 seconds important photosynthetic traits, where previously these measurements took 20 minutes,” Professor Furbank says.
The hyperspectral reflectance method for measuring wheat photosynthesis was developed by Dr Viridiana Silva Perez during a PhD project co-supervised by Professor Furbank, Professor John Evans and CSIRO’s Dr Tony Condon. Dr Silva Perez travelled from Mexico to study in Australia as part of a $10 million endowment to wheat research made by the Mexican prime minister. Dr Silva Perez is now based at CSIRO as a researcher.
“Our next goal is to scale the technology from measuring leaves one at the time to something we can put on a drone to scan thousands of trial plots,” Professor Furbank says. “This would enable the technology to be integrated into wide-scale pre-breeding trials and commercial breeding programs, thereby accelerating the delivery of improved photosynthesis traits into future wheat cultivars.”
The role of AI
The machine intelligence needed to process light measurements is not in use solely in Australia – rather, it has been made available to researchers worldwide via a web-based portal. That means researchers anywhere in the world can use the site to process their light measurements, helping to drive additional genetic gain. Wheat breeders at the International Maize and Wheat Improvement Center (CIMMYT) in Mexico, especially, are tapping into this new resource.
“This area of machine learning is similar to the way Facebook can tag people in multiple images because it has figured out what you look like,” Professor Furbank says. “Our work uses the same principle of training software to pull information out of a dataset. But instead of processing images, it deals with light reflected from wheat leaves.”
The breadth of information extracted by the AI is astonishing. There is information about the key enzyme responsible for fixing atmospheric carbon dioxide into sugars, which is called rubisco (ribulose-1,5-bisphosphate carboxylase/oxygenase).
Included is information about how much rubisco is present in leaves, how efficiently this enzyme is working, and how efficiently leaves are absorbing and using light energy.
“Hyperspectral reflectance is useful on different fronts,” Professor Furbank says. “The same measurements used to profile photosynthesis rates can also provide data about respiration – meaning how the carbon is used by the plant after it is fixed by rubisco. The technology can also accurately detect how much nitrogen is in the crop and monitor how efficiently the crop is taking up nitrogen.”
To train the AI to extract this information first involved creating a ‘training dataset’ by measuring photosynthesis the old fashioned way. This data provided a reference for the AI to link various photosynthetic states with reflected light.
Parallel phenomics measurements made with lasers then provide data about plant growth rates and biomass made using CSIRO’s phenomobile (see GroundCover™ issue 118, September – October 2015, ‘World unites to crack the cereals yield ceiling’).
To predict the impact of these photosynthetic improvements on biomass and yield, Professor Graeme Hammer at the University of Queensland has extended the Agricultural Production Systems sIMulator (APSIM) plant growth model in order to predict likely yield given a certain amount of inputs and climatic conditions.
“A 30 per cent variation in photosynthesis results in a significant difference in biomass,” Professor Furbank says. “Combined with the yield impact modelling to guide our research, we are seeing rapid progress making photosynthesis the target of breeding programs. In addition, the work is bringing new technologies to bear on the yield problem and also helping to create international collaborations.”
the wheat physiology predictor tool can be accessed at
Was this page helpful?