A Novel Approach to Optimize Grain Filling

A novel systems model, in conjunction with genetic algorithms, identify ideotypes with high yield potential.

Rice plays a crucial role as the primary staple food, feeding over half of the world’s population. Global demand is escalating, driven by population growth, shifting dietary preferences, and rising incomes worldwide. This comes as yields are increasingly under threat from the adverse effects of climate change. Consequently, there is an urgent need to substantially enhance rice yields.

In an encouraging development, a recent study has uncovered a yield-enhancing process that could increase rice yield by more than 50%.

Tian-Gen Chang, research scientist in the Plant Systems Biology Group (lead by Prof. Xinguang Zhu) at the Center for Excellence in Molecular Plant Science, Chinese Academy of Sciences and colleagues developed the Whole-plAnt Carbon–Nitrogen Interaction (WACNI) model to simulate rice plant growth from flowering to harvest, offering new insights into grain filling – a key phase in rice development.

Rice geneticists and breeders have previously attempted to enhance rice yield by increasing the number of ears and spikelets. However, the potential increase in production resulting from these variations is not always fully realized as many grains often remain empty. The authors therefore directed their efforts towards optimizing grain filling, which promised to be a game-changer in yield enhancement.

The WACNI model differed from previous rice models because it simulated plant growth and senescence using a bottom-up approach rather than preset growth rules. “We developed a kinetic model for the basic carbon and nitrogen metabolism occurring in a plant. Consequently, the carbon and nitrogen budgets, along with major physiological dynamics during grain filling, naturally arise from this bottom-up modeling approach. Excitingly, the ‘in silico rice’ created by our model successfully predicted grain yield formation under various environmental, agronomic, and genetic perturbations,” explained Chang.

Think of WACNI as a virtual rice plant. Inside this digital world, the scientists were able to tweak 28 key factors, like how fast roots uptake nitrogen, how fast sugars export from leaves, and how fast grains use sugars and grow, to see how they affected plant growth and the final grain yield of the virtual plant. Then, they unleashed a “genetic algorithm” to find the superior combinations of those factors for maximum grain filling.

A genetic algorithm is a computational method inspired by the principles of natural selection and genetics. By iteratively applying the selection, reproduction, and evaluation steps, the genetic algorithm explores the parameter space, gradually refining the parameter values to find the combinations that resulted in the highest grain yield. These parameter values serve as the defining characteristics of the “elite ideotypes” for super-high yielding rice.

Key physiological traits of these elite in silico ideotypes were calculated, revealing a potential 54% yield increase over standard varieties. This improvement was achieved by optimizing resource allocation between plant organs and enzyme activity in various metabolic processes, albeit at a 37% reduction in final grain nitrogen content.

Notably, the elite ideotypes, regardless of whether they possessed higher or lower photosynthetic capacity or had a longer or shorter grain-filling season, shared some common defining features. Among these, the most remarkable is a stable grain-filling rate from flowering to harvest. Further analysis has shown that “stable grain filling” can serve as a marker of a balanced source-sink relationship during the grain-filling season, which is crucial for high yield.

But do these in silico predictions align with real-world outcomes? To confirm that the simulated traits translated to real-world yield gains, the researchers grew two of the newest super-high-yielding rice cultivars in one of the most ideal regions for rice cultivation, ultimately achieving a record high yield (21 tons per hectare of rough rice at 14% moisture content). They monitored the traits of these cultivars during the grain-filling phase. Upon comparing the physiological traits measured in the field with the model’s predictions during this phase, they observed significant alignment. Both in silico and super-high-yielding rice had increased root nitrogen uptake after flowering, decreased stem non-structural carbohydrate content at harvest, a gradual reduction in leaf area post-flowering, consistently low grain nitrogen content, and most notably, a remarkably stable grain-filling rate from flowering to harvest.

Is it possible to enhance the yield of typical rice cultivars by altering the key factors identified by the model that regulate grain filling? When the team applied their optimized grain-filling model to a typical rice cultivar under standard cultivation practices, they discovered it could enhance grain yield by approximately 30–40%. This significant increase was achievable solely through optimizing enzyme activities, without the need to alter plant organ sizes at the flowering stage. With advancements in genetics and genome editing technologies, such improvements could potentially be achieved through targeted genetic engineering in the near future.

One parameter that is generally focused on as target to increase yield is increased photosynthesis. The authors argue that if this is achieved coordination of the source–sink relationship must also be improved to ensure that the increase in carbohydrates can be utilized.

Chang concludes, “Our innovative computational framework for rice grain filling highlights the pivotal role of coordinating source-sink relationship during grain filling. This should be a key focus in crop breeding, alongside increasing leaf photosynthesis. The molecular and physiological markers identified in this study may guide the development of future high-yielding rice varieties.”

READ THE ARTICLE:

Tian-Gen Chang, Zhong-Wei Wei, Zai Shi, Yi Xiao, Honglong Zhao, Shuo-Qi Chang, Mingnan Qu, Qingfeng Song, Faming Chen, Fenfen Miao, Xin-Guang Zhu, Bridging photosynthesis and crop yield formation with a mechanistic model of whole-plant carbon–nitrogen interaction, in silico Plants, Volume 5, Issue 2, 2023, diad011, https://doi.org/10.1093/insilicoplants/diad011

The source code used for this study, along with the operational commands and user guide, is freely available for non-commercial use at https://github.com/rootchang/WACNI-rice.git.