By targeting the SlGA20ox genes—key regulators of plant height—the study successfully produced short-statured tomato cultivars that maintain normal yield and fruit quality. A deep learning–based volumetric model analyzing chlorophyll fluorescence data achieved over 84% classification accuracy in identifying gene-edited plants.
The global food system faces mounting pressure as climate change, extreme weather, and urbanization reduce available farmland and threaten productivity. Vertical farming has emerged as a promising solution, allowing dense crop cultivation in controlled indoor environments. However, most systems currently support only leafy vegetables due to spatial limitations. Fruit crops like tomato, an essential global food, often exhibit indeterminate growth, making them unsuitable for confined cultivation. Developing compact, high-yielding tomato varieties could revolutionize urban agriculture. Based on these challenges, researchers sought to manipulate growth-regulating genes and apply advanced AI-powered phenotyping tools to engineer new tomato genotypes optimized for vertical farming environments.
A study (DOI: 10.1016/j.plaphe.2025.100095) published in Plant Phenomics on 14 August 2025 by Dae-Hyun Jung’s & Choon-Tak Kwon’s team, Kyung Hee University, marks a significant step toward sustainable high-density agriculture, enabling efficient crop management in constrained environments such as urban vertical farms.
The study first combined targeted gene editing and AI-based phenotyping to engineer and evaluate compact tomato plants for vertical farming. Researchers began by screening 12 SlGA20ox genes, then used multiplex CRISPR-Cas9 to edit SlGA20ox2 and SlGA20ox4, generating single and double mutants in a triple-determinate tomato background. They characterized plant architecture, photosynthetic physiology, and yield across greenhouse and vertical farm trials, and then applied a 3D convolutional neural network (3D-CNN) to chlorophyll fluorescence imaging data to capture subtle physiological differences that conventional measurements might miss. This integrated approach produced several key results. All edited lines (slga20ox2, slga20ox4, and slga20ox2/4) showed strongly reduced height, shorter internodes, and tighter inflorescence spacing, yet maintained normal flowering timing, photosynthetic efficiency (Fv/Fm), chlorophyll and carotenoid levels, and importantly, unchanged fruit set, yield, fruit size, ripening behavior, and sugar content. These traits were stable in both greenhouse and vertical farm conditions, indicating that plant size was reduced without sacrificing productivity or fruit quality. The deep learning pipeline then converted time-resolved chlorophyll fluorescence features into volumetric inputs and classified genotypes with an accuracy of about 0.84, outperforming SVM, LSTM, and 1D-CNN models. The 3D-CNN not only generalized well in cross-validation but also revealed genotype-specific fluorescence dynamics, especially in non-photochemical quenching (NPQ), and cleanly separated mutant classes in t-SNE space. This demonstrates that CRISPR-driven modulation of SlGA20ox can generate compact, high-performing tomatoes, and that volumetric phenotyping can non-destructively distinguish elite lines for vertical farming.
This research establishes a new paradigm for intelligent crop breeding, merging gene editing with AI-based non-destructive phenotyping. The compact SlGA20ox-edited tomatoes offer ideal traits for vertical farming—short stature, stable yields, and high photosynthetic efficiency—making them promising candidates for sustainable urban agriculture. The volumetric deep learning model also provides a cost-effective and scalable platform for high-throughput phenotyping, eliminating the need for expensive multi-sensor systems. Beyond tomatoes, this analytical framework can be extended to other crops to accelerate selection of desirable genotypes, enabling data-driven breeding strategies that address global food security and resource efficiency challenges.
###
References
DOI
10.1016/j.plaphe.2025.100095
Original URL
https://doi.org/10.1016/j.plaphe.2025.100095
Funding information
This research was funded by the National Research Foundation of Korea (NRF) grant from the Ministry of Science and ICT (MSIT), Republic of Korea (RS-2024-00407469 and RS-2025-00517964), and partially funded by the BK21 FOUR program of Graduate School, Kyung Hee University, Republic of Korea (GS-5-JO-NON-20250783).
About Plant Phenomics
Plant Phenomics is dedicated to publishing novel research that will advance all aspects of plant phenotyping from the cell to the plant population levels using innovative combinations of sensor systems and data analytics. Plant Phenomics aims also to connect phenomics to other science domains, such as genomics, genetics, physiology, molecular biology, bioinformatics, statistics, mathematics, and computer sciences. Plant Phenomics should thus contribute to advance plant sciences and agriculture/forestry/horticulture by addressing key scientific challenges in the area of plant phenomics.