Inefficient nitrogen use in agriculture drives excessive fertilizer application, redundant crop growth, greenhouse gas emissions, ecosystem toxicity and risks to human health. Wheat production depends heavily on synthetic nitrogen, yet nitrogen-use efficiency (NUE) and nitrogen responsiveness are controlled by many genes and shaped by environment, soil and management. Inconsistent field phenotyping across seasons and sites also makes it hard to compare traits such as nitrogen-uptake efficiency (NupE) and nitrogen harvest index (NHI). Against these challenges, deeper research is needed into reproducible dynamic phenotyping that can capture nitrogen-response changes and connect them with yield, NUE and underlying genetics.

Researchers at Nanjing Agricultural University, the National Institute of Agricultural Botany (NIAB) and collaborators published (DOI: 10.34133/plantphenomics.0128) the study in Plant Phenomics on 22 December 2023. The team used low-cost drones and the AirMeasurer platform to phenotype 54 winter wheat varieties under 0, 180 and 270 kg N ha⁻¹. They quantified plant height, canopy coverage, normalized difference yellowness index (NDYI), visible atmospherically resistant index (VARI), 3D canopy index (3DCI) and angular second moment (ASM). They then built dynamic profiles, trained a random forest-nitrogen response (RF-NRES) model and performed genome-wide association study (GWAS) with the Wheat55K single-nucleotide polymorphism (SNP) array.

The study combined static phenotypes from 10 drone flights with dynamic phenotypes derived from curve fitting. Gaussian, Fourier and Weibull functions produced profile curves for six traits across the season, and compound growth rates (CGRs) were calculated in five-day windows from 1 to 25 days after fertilization (DAF). Using 12 yield-production and N-utilization indices, principal component analysis (PCA) generated nitrogen efficiency comprehensive scores (NECS) and divided varieties into four groups: high-yield-high-N-utilization (HYHN), low-yield-high-N-utilization (LYHN), low-yield-low-N-utilization (LYLN) and high-yield-low-N-utilization (HYLN). A machine-learning model, RF-NRES, classified N-responsiveness groups from N-response phenotypes with 91.6% accuracy; cross-validation reached 93.0% for HYHN and 90.5% for LYLN varieties. GWAS using 44,215 SNPs showed that dynamic phenotypes associated with more significant SNPs than static traits. For example, under N0, 714 SNPs were linked to dynamic phenotypes versus 94 for static phenotypes. Candidate loci included SVP, SAR and AP2L5 under N0; NRT1.1A under N180/N270; and CAO and CA1Pase for height and VARI changes. These findings highlight dynamic phenotyping as a powerful bridge between field performance and genetic discovery.

The authors said the approach could change how breeders screen wheat for nitrogen responsiveness. They explained that instead of relying on a few manual measurements at limited time points, affordable drones can follow canopy changes across the season and turn those changes into quantitative traits. They added that combining dynamic phenotypes with machine learning offers a practical way to rank varieties, while genetic mapping shows where those responses may be controlled. The team said larger populations and multiple environments will be needed before these markers can be used routinely in breeding.

The work offers a scalable route for nitrogen management and wheat breeding. Drone-based dynamic phenotyping could help identify varieties that maintain yield with less nitrogen, support site-specific fertilizer decisions and reduce nitrogen leaching and greenhouse gas emissions. The RF-NRES model could accelerate large-scale screening, while the associated loci may inform marker-assisted selection once validated in larger panels. The authors also released source code and testing data to encourage reuse. Still, the study used 54 varieties and limited environments; broader trials, standardized calibration and multi-environment validation are needed to translate these findings into routine breeding and farm practice.

###

References

DOI

10.34133/plantphenomics.0128

Original Source URL

https://doi.org/10.34133/plantphenomics.0128

Funding information

This work was supported by the National Natural Science Foundation of China (32070400 to J.Z.). The drone-based phenotyping, field experiments, yield, and N measures were also supported by the Key Project of Modern Agriculture of Jiangsu Province (BE2019383). J.Z., R.J., and G. Deakin were partially supported by the Allan & Gill Gray Philanthropies' sustainable productivity for crops improvement (G118688 to the University of Cambridge and Q-20-0370 to NIAB). J.Z. and R.J. were supported by the One CGIAR's SeedEqual Initiative (5507-CGIA-07 to J.Z.), as well as the United Kingdom Research and Innovation's (UKRI) Biotechnology and Biological Sciences Research Council's (BBSRC) International Partnership Grant (BB/X511882/1).

About Plant Phenomics

Plant Phenomics publishes breakthrough research that integrates genomics, genetics, physiology, molecular biology, bioinformatics, statistics, mathematics and computer science to advance plant phenotyping—from the cellular level to entire populations. The journal tackles key scientific challenges in phenomics by showcasing cuttingedge technologies for data acquisition, management, interpretation and modelling. Its scope spans practical applications across all plant sciences, aiming to translate novel phenotyping tools into realworld solutions for breeding, crop management and fundamental plant biology. By fostering interdisciplinary innovation, Plant Phenomics serves as a hub for researchers developing highthroughput, accurate and scalable methods to decipher complex plant traits, ultimately accelerating progress in agriculture and plant science.