AI-enabled simultaneous phenotyping of leaf vein and stomatal traits uncovers independent genetic control in maize
Sechi, M.; Porcedda, R.; Pallaoro, M.; Ferguson, J. N.; Dell'Acqua, M.; Vandin, A.; Caproni, L.
Show abstract
BackgroundLeaves maintain hydraulic homeostasis during photosynthesis through the coordinated action of stomata, which regulate gas exchange and transpiration, and veins, which supply water to the leaf lamina. While functional links between stomatal and vascular traits are known in dicots, their potential genetic coordination in C4 crops remains poorly understood. We investigated the genetic architecture of these traits in maize using a Multi-parent Advanced Generation Inter-Cross (MAGIC) population and a low-cost, high-throughput phenotyping platform integrating leaf clearing, digital microscopy, artificial intelligence, and image analysis ResultsWe phenotyped 285 recombinant inbred lines and the MAGIC founder lines, generating 8,072 images from 2,026 leaf samples taken from seedlings grown in controlled conditions. A YOLOv8-based model automatically detected stomata, while a custom and efficient image-processing pipeline quantified vein traits and stomatal spatial distribution patterns along cell bundles. This enabled simultaneous characterization of stomatal density, size, and distribution together with vein density, thickness, and bundle-associated spatial patterning. Substantial phenotypic variation was observed among genotypes, with strong correlations between abaxial and adaxial traits but no significant correlations between stomatal and vein traits. QTL mapping identified 37 genomic regions associated with stomatal and vein traits, including loci containing known developmental regulators such as stomatal density and distribution1 and stomagen1, as well as novel loci controlling stomatal spatial patterns, divergence between leaf surfaces and veins traits. ConclusionsThese results support independent genetic control of stomata and veins and decoupled contribution to water-use efficiency, providing a novel genetic framework to independently optimize leaf hydraulic capacity and gas exchange in target environments.
Matching journals
The top 4 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
- Machine learning enabled phenotyping for GWAS and TWAS of WUE traits in 869 field-grown sorghum accessions 94%
- Development of a mobile, high-throughput, and low-cost image-based plant growth phenotyping system 94%
- Paralog editing tunes rice stomatal density to maintain photosynthesis and improve drought tolerance 94%
Similar papers in this journal
- SUPPRESSOR OF MAX2 1-LIKE 5 promotes secondary phloem formation during radial stem growth 93%
- OPEN leaf: an open-source cloud-based phenotyping system for tracking dynamic changes at leaf-specific resolution in Arabidopsis 93%
- Rapid identification of an Arabidopsis NLR gene conferring susceptibility to Sclerotinia sclerotiorum using time-resolved automated phenotyping 93%
Similar papers in this journal
- Root anatomical gradients and cultivar differences underlie variation in root hydraulic properties in German winter wheat 95%
- Wild genes to the rescue: High-throughput genomics uncovers the wild source of broomrape resistance in sunflower 95%
- Characterizing grapevine (Vitis spp.) inflorescence architecture using X-ray imaging: implications for understanding cluster density 94%
Similar papers in this journal
- Increased and ectopic expression of Triticum polonicum VRT-A2 underlies elongated glumes and grains in hexaploid wheat in a dosage-dependent manner 94%
- Maize Brittle Stalk2-Like3, encoding a COBRA protein, functions in cell wall formation and carbohydrate partitioning 94%
- High-resolution 4D spatiotemporal analysis reveals the contributions of local growth dynamics to contrasting maize root system architectures 93%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.