Rachel ShekarSince 1991, the Agricultural Production Systems sIMulator (APSIM) has grown from a farming systems framework used by a small number of people, into a large collection of models used by many thousands of modelers internationally. Twenty years later, its modelling framework encompass over 35 plant species, ranging from legumes and grasses to tuber crops and trees. Thanks to the research initiative’s dedication to ongoing development, management, and use of APSIM, there are close to 1,000 research articles based on its simulations.
Platforms like APSIM help researchers explore the dynamics between the atmosphere, the crop, and the soil, assist in crop agronomy, pest management, breeding, and natural resource management, and assess the impact of climate change.
Junqi Zhu, researcher at The New Zealand Institute for Plant and Food Research Limited Marlborough Research Centre, lead a team that created the first perennial fruit crop model using the APSIM Next Generation framework. In the article published by in silico Plants, the authors used grapevine, one of the most economically important perennial fruit crops worldwide, as the plant model.
Perennial crops represent a long-term investment by landowners. Grapevines remain economically productive for 20 to 60 years. There are no opportunities to change location, genotype and plant configuration to adapt to climate compared with annual crops. As such, reliable models for evaluating options at establishment and for ongoing management are valuable aids for decision making.
“The main difference between annual and perennial crops is the yield formation processes. The reproduction cycle of a perennial crop lasts 15 to 18 months or more with potential carryover effects (e.g., carbohydrate reserves) from previous years,” explains Zhu. “We needed to adapt and add modules to represent the nature of fruit-bearing vines.”
The modules developed by the authors included:
- Phenology – Unlike annual crops, the perennial grapevines undergo dormancy followed by budding, flowering, fruit setting, berry development and leaf death phases. Perennial organs included the cane, trunk, and structural root.
- Light interception – Grapevines have unique architecture and are planted in rows with a wide alley between them. The model calculated light interception by the row canopy, expressed as a function of canopy width, canopy depth, the distance between two rows, leaf area of the row crops and light extinction coefficient
- Carbohydrate allocation – The trunk and root of perennial plants demand for carbohydrates for both growth and reserves are higher than for annual crops, which prioritize organ growth over reserves.
- Yield formation and berry composition – Grapes (berries) grow in bunches, with multiple bunches growing from each shoot with multiple shoots per vine. The model encapsulated bunch number per shoot, berry number per bunch, fresh weight, dry weight, total soluble solids (sugar concentration) and titratable acid. Titratable acidity plays a significant role in taste, colour, and microbial stability of the grape’s juice.
The authors then calibrated and validated their new model using existing datasets.
Simulations were run across 8 sites in New Zealand with varying node numbers to represent vine pruning. Some growers prune during the reproductive stage to enhance wine quality by reducing the carbohydrate competition between vegetative and reproductive growth.
The grapevine model captured the variations in phenological time across the sites for five different varieties over 15 growing seasons. Simulated dates for budburst, flowering and véraison (the onset of the ripening) correlated well with the observed dates.
The newly added row-planted crop radiation model captured the seasonal light interception pattern of a vertical shoot position training system and provided a framework for modelling the effects of vineyard configuration and alley management.
The model reproduced the dry matter dynamics of different organs caused by different seasonal weather conditions and pruning strategies (e.g., different retained node number) over two seasons.
Finally, the model captured the wide variations in yield components (berry fresh weight, total soluble solids, and titratable acid) over 10 seasons in five sites with five different retained cane numbers.
Says Zhu, “the grapevine model represents an important advance as it is the first perennial fruit crop fully implemented in APSIM, and it provides a useful template for the development of models for other perennial fruit crops. We hope other researchers will continue the development and testing of the model in other countries.”
RESEARCH ARTICLE:
Junqi Zhu, Amber Parker, Fang Gou, Rob Agnew, Linlin Yang, Marc Greven, Victoria Raw, Sue Neal, Damian Martin, Michael C T Trought, Neil Huth, Hamish Edward Brown, Developing perennial fruit crop models in APSIM Next Generation using grapevine as an example, in silico Plants, 2021;, diab021, https://doi.org/10.1093/insilicoplants/diab021
This manuscript is part of in silico Plant’s Functional Structural Plant Model special issue.
More about the APSIM Next Generation Framework:
APSIM Next Generation, uses a version control system to ensure model reliability and a modern distribution system to ensure users can easily access models and receive updates. It has auto-documentation and a user-friendly interface whereby developers can drag and drop different modules and functions to represent the physiological processes. The user interface is at the level of zero programming-skill requirement, enables more scientists to contribute to model development. All source code is available on the APSIM initiative repository under a research and development license.
The APSIM grapevine model, data used for model development, and source code are publicly available in the APSIM Next Generation git repository: https://github.com/APSIMInitiative/-ApsimX/tree/master/Tests/Validation/Grapevine. The R code for plotting and analysis are available from the authors upon request.
