How to Use Drone Data to Build Growth Curves for Crop Breeding Success

Introduction

Understanding when and how long genetic effects influence crop growth is crucial for breeders aiming to develop high-performing varieties. Traditionally, this required painstaking manual measurements throughout the season. Now, drones equipped with multispectral sensors can capture detailed time-series data, allowing you to build growth curves that reveal the timing and duration of genetic effects—much like tracking growth spurts in siblings. This guide walks you through the process step by step.

What You Need

• Field trial setup: A plot with multiple genotypes (e.g., inbred lines or hybrids) replicated across blocks to account for environmental variability.
• Drone with multispectral camera: Capable of capturing red, green, blue, near-infrared (NIR), and red-edge bands for vegetation indices like NDVI.
• Automated flight planning software: To schedule regular flights (e.g., every 3–7 days) over the growing season.
• Image processing tools: Software like Pix4D, Agisoft Metashape, or open-source alternatives (OpenDroneMap) to stitch orthomosaics and extract plot-level data.
• Data analysis environment: R or Python with packages for time-series analysis, curve fitting, and statistical modeling (e.g., nlme, tidyverse in R; scipy, pandas in Python).
• Ground truth data: Destructive or non-destructive measurements (e.g., plant height, biomass) at key growth stages for validation.
• GPS/GNSS ground control points: For accurate georeferencing of drone imagery.

Step-by-Step Instructions

Step 1: Design and Establish a Field Trial with Diverse Genotypes

To capture genetic variation, include a representative set of genotypes—such as elite lines, landraces, or mapping populations—laid out in a randomized complete block design with at least three replicates. Ensure the trial covers a uniform field to minimize soil and microclimate effects. Mark plot boundaries clearly using GPS or physical flags, and place ground control points at known coordinates around the trial area.

Step 2: Conduct Regular Drone Flights Over the Growing Season

Plan flights at consistent intervals—every 5 to 7 days works well for most annual crops. Use automated flight planning software to maintain the same altitude (typically 30–60 m) and overlap (forward: 80%, side: 70%) across all missions. Fly under clear skies between 10 a.m. and 2 p.m. to minimize shadow and atmospheric effects. Record flight metadata (date, time, solar angle) for later adjustments.

Step 3: Process Drone Imagery to Extract Vegetation Indices

Stitch raw images into a single orthomosaic using photogrammetry software. Next, derive plot-level values of vegetation indices such as NDVI (Normalized Difference Vegetation Index) or NDRE (Red Edge) for each time point. Clip the orthomosaic to plot boundaries using shapefiles or polygon vectors, then calculate the mean index value per plot per flight. Export the data as a tidy table with columns: plot ID, genotype, replicate, flight date, and index value.

Step 4: Generate Growth Curves for Each Genotype

For each genotype, plot the vegetation index over time. Fit a smoothing curve (e.g., logistic, Gompertz, or cubic spline) to the time-series data to capture the temporal progression. Use curve-fitting algorithms that handle irregular time intervals—drone flights may be delayed by weather. Extract key parameters: onset (when growth accelerates), maximum rate (steepness of slope), peak value, and duration (width of the curve). Store these parameters for each genotype.

Step 5: Analyze Curves to Identify Timing and Duration of Genetic Effects

Compare growth parameters across genotypes using mixed models or ANOVA. For example, if Genotype A reaches its peak NDVI 10 days earlier than Genotype B, that difference indicates a genetic effect on the timing of flowering or canopy closure. Similarly, if the curves of two siblings have the same peak but different widths, the broader curve signifies a longer duration of active growth—an important trait for stress recovery or nutrient uptake. Use pairwise contrasts and plot confidence intervals to highlight significant differences.

Step 6: Validate Drone-Derived Curves with Ground Truth Measurements

Select a subset of plots covering the range of growth curves. At critical time points (e.g., maximum vegetation index, mid-season), manually measure plant height, leaf area index, or biomass. Correlate these ground measurements with drone-derived indices for the same dates. If correlations are strong (r > 0.8), the curves are reliable. If not, recalibrate the vegetation index or adjust for soil background effects using soil-adjusted indices like SAVI.

Step 7: Apply Findings to Breeding Decisions

Use the identified genetic effects to inform crossing strategies. For instance, if late timing of peak growth is correlated with higher yield in your environment, select parents that express that timing. If a short growth duration makes a genotype vulnerable to drought, breed for longer duration. Build a database of growth curve parameters for your germplasm and integrate it into genomic selection models to predict performance of untested lines.

Tips and Best Practices

• Standardize flight conditions: Changes in sun angle or cloud cover can bias indices. Use reflectance calibration panels before each flight to normalize data.
• Handle missing data gracefully: Sometimes drone flights are impossible due to weather. Use interpolation methods (e.g., linear or cubic) to fill gaps in time series, or apply models that handle irregular intervals.
• Consider phenology stages: Anchor your growth curves to specific growth stages (e.g., emergence, leafing, flowering) to make genetic effects more interpretable. You can use degree-day accumulation instead of calendar days for better physiological alignment.
• Beware of saturation: Vegetation indices can saturate at high biomass (e.g., full canopy). Use alternative indices (e.g., EVI or NDRE) that remain sensitive in dense canopies.
• Share and compare: Collaborate with other breeders to build a reference dataset for your crop. The more genotypes you evaluate, the more robust the insights into timing and duration of genetic effects become.
• Iterate and refine: After one season, review your flight schedule—maybe weekly was too sparse for a fast-growing crop? Adjust for the next year.

For deeper analysis, revisit Step 5 on analyzing curves and consider non-linear mixed models to partition variation into genetic and environmental components. With practice, these drone-derived growth curves will become a powerful, high-throughput tool in your breeding program.