Tomato plant architecture plays a central role in determining yield, light interception, and resource-use efficiency in greenhouse production systems. Traits such as stem thickness, leaf spacing, and fruit distribution are controlled by complex interactions between genes and the environment and often change substantially during development. Although advances in genomics have accelerated the identification of candidate genes, progress has been constrained by the lack of efficient, large-scale phenotyping methods capable of capturing trait dynamics over time. Traditional manual measurements are labor-intensive, subjective, and typically limited to single growth stages. Based on these challenges, there is a clear need to conduct in-depth research on dynamic, high-throughput phenotyping approaches that can reveal temporal patterns of trait expression and heritability.
Researchers from Zhejiang University, in collaboration with the Hangzhou Academy of Agricultural Sciences and Cornell University, reported (DOI: 10.1093/hr/uhaf109) their findings on April 30, 2025, in Horticulture Research. The study introduces an unmanned ground vehicle–based phenotyping system designed to monitor tomato plants continuously throughout their growth cycle. By integrating multispectral, depth, and RGB (Red, Green, and Blue) imaging with advanced segmentation algorithms, the team extracted detailed architectural traits and quantified how their genetic heritability changes over time. The work provides a systematic framework for linking plant structural development with genetic stability under controlled greenhouse conditions.
The proposed phenotyping system uses an autonomous ground robot equipped with RGB-D and multispectral sensors to collect high-resolution images of tomato plants at multiple developmental stages. A deep-learning segmentation model fuses spectral and depth information to accurately separate stems, leaves, flowers, and fruits, even in dense greenhouse canopies. From reconstructed 3D point clouds, the system automatically extracts six architectural traits: stem thickness, leaf spacing, inflorescence height, fruit spacing, total leaf area, and leaf inclination angle.
Validation against manual measurements showed high accuracy, with strong agreement across most traits and low estimation errors. Importantly, the continuous data stream enabled the researchers to analyze not just trait values, but also how their broad-sense heritability evolved over time. Traits such as stem thickness, inflorescence height, and fruit spacing exhibited increasing heritability as plants matured, suggesting stronger genetic control at later stages. In contrast, leaf-related traits showed declining heritability, indicating greater environmental influence as canopy complexity increased. Total leaf area followed a non-linear pattern, with heritability fluctuating across growth stages. These results demonstrate that the genetic importance of architectural traits is highly time-dependent, emphasizing the need for stage-specific phenotyping in breeding studies.
“This study shows that when we measure plant traits can be just as important as what we measure,” said the corresponding author of the study. “By capturing architectural traits continuously, we can identify the developmental windows when genetic effects are strongest. This helps breeders focus on the most informative stages for selection and genetic analysis. The integration of robotics, multimodal imaging, and heritability analysis represents a significant step toward more precise and efficient crop improvement strategies.”
The findings have important implications for both plant breeding and greenhouse production management. By revealing when specific traits become more genetically stable, the approach can guide breeders to optimize sampling time points for quantitative trait locus mapping and selection decisions. In commercial production, continuous architectural monitoring could support precision management strategies, such as optimized pruning, spacing, and resource allocation. More broadly, the phenotyping framework can be adapted to other crops and controlled-environment systems, accelerating the integration of robotics and data-driven decision-making into modern agriculture. As breeding increasingly relies on dynamic trait information, such systems may become essential tools for aligning genetic potential with real-world growing conditions.
###
References
DOI
10.1093/hr/uhaf109
Original Source URL
https://doi.org/10.1093/hr/uhaf109
Funding information
This work was funded by the Science and Technology Project of the Ministry of Agriculture and Rural Affairs, the Fundamental Research Funds for the Central Universities (226-2022-00217) and the National Natural Science Foundation of China (32371985).
About Horticulture Research
Horticulture Research is an open access journal of Nanjing Agricultural University and ranked number one in the Horticulture category of the Journal Citation Reports ™ from Clarivate, 2023. The journal is committed to publishing original research articles, reviews, perspectives, comments, correspondence articles and letters to the editor related to all major horticultural plants and disciplines, including biotechnology, breeding, cellular and molecular biology, evolution, genetics, inter-species interactions, physiology, and the origination and domestication of crops.