Artificial intelligence helps make the world more food secure

10 July 2017

In the race to improve food security, a new QAAFI-developed intelligent crop modelling tool is helping researchers identify how improved plant growth efficiency will translate into better yields. Planet Earth’s population is expected to reach nine billion in 2050, but the production of important food staples – such as bread wheat and rice – are currently projected to fall short of growing demand. 

A major R&D push is underway globally to lift the yield potential of staple cereal crops. 

The target of these research efforts is photosynthesis. Yield gains are being sought by improving plants’ efficiency at capturing sunlight and converting it into plant growth, biomass and grain, with a target to increase grain yields by 50 per cent in the next 20 years.

These research efforts take many forms, one of which is the $100 million International Wheat Yield Partnership, which draws heavily on Australia’s world-leading photosynthesis research capability. 

Within Australia, leading laboratories are working collaboratively through the ARC Centre of Excellence (ARC) for Translational Photosynthesis.

At QAAFI, Professor Graeme Hammer leads the University of Queensland’s node of the ARC Centre. 

Crop simulation software

Professor Hammer’s team has developed crop simulation software that can radically accelerate the discovery process. The software system can be fed early-stage photosynthesis discoveries – discoveries that are being made at the molecular (or subcellular) scale – and extrapolate how this altered photosynthesis biochemistry might impact on crop performance in the form of virtual plants grown under realistic farming conditions affected by real-world weather, soil and rainfall data.

“There is this idea that if photosynthesis pathways in plant leaves can be made more efficient you will see an equivalent lift in yields, but that is simply not the case,” Professor Hammer says. “As you move up from the molecular scale to the crop level, you are going to lose effects.”

Without the crop simulation technology, it can take years of field trials to understand the relationship between enhanced photosynthesis efficiency and crop performance. Professor Hammer’s simulations can achieve something similar in just hours.

Within the simulation, sunlight strikes the crop’s canopy at different angles and with varying intensity as the virtual sun moves across a virtual sky whose cloud cover is drawn from real-world weather data. The effects of all these dynamic variations are captured and used to better understand the impact on crop productivity of changes to photosynthesis biochemistry.

Dr Alex Wu (QAAFI) with Viridiana Silva-Perez (CSIRO) and the CoETP

Agricultural Production Systems sIMulator

The bedrock to this new, more advanced crop simulation technology is APSIM – the Agricultural Production Systems sIMulator – that was developed in Australia (with input from Professor Hammer, among others) to support researchers and plant breeders while also assisting farmers with crop management decisions. 

APSIM was developed to simulate biophysical processes in farming systems, in particular where there is interest in the economic and ecological outcomes of management practice in the face of climatic and market risks. 

It is modular in structure. Different components deal with factors relating to farm management decisions, climate and the plant’s genetic makeup.
Essentially, APSIM can configure a virtual paddock where plant genetics can interact realistically with soil, climate and farm practices to explore the effect of changing these variables.

Professor Hammer’s team has now extended APSIM’s core capabilities to take into account variation in the plant’s photosynthetic efficiencies.

“The previous modelling platform included a simple mathematical relationship when dealing with photosynthesis,” Professor Hammer says. “For sorghum crops, for example, on average the plant creates 1.25 grams of dry mass per megajoule of light absorbed, given enough water and nitrogen. 

“What we have now added in are sub-routines that determine this light conversion efficiency based on more detailed models of photosynthesis biochemistry. That means we can now take into account variation in how efficiently plants capture carbon dioxide or how well the carbon dioxide is converted into sugar.”

The ability to integrate photosynthesis’ constitutive parts into a crop model draws heavily on the work of project collaborators Professor Graham Farquhar and Professor Susanne von Caemmerer at the Australian National University, who have pioneered the development of photosynthesis biochemistry models. 

The value of this modelling capability was highlighted recently when an initial study showed that a 25 per cent increase in efficiency of photosynthetic enzymes generated just a five per cent increase in crop growth.

“The modelling highlights that there are many interacting factors affecting growth at crop scale and the impact on yield can vary in different environments,” Professor Hammer says. “Serious play with these models helps us retain a sense of the broader reality in which the plant functions so that we can better focus on the best-bet options for different growing environments.”