Plant breeding with drones

Among other things, the future of breeding is in the air: KWS uses drones over its boards to produce new varieties even faster. Images taken with their cameras are analyzed on a computer to get accurate results on plant growth. This will help the company to be able to offer new seeds to farmers sooner – a big advantage for agriculture. The air vibrates. Sounds like a swarm of bees approaching. For a few seconds, the four rotors of Marius Burkhardt’s bright orange flying machine cause such strong turbulence on KWS’s experimental plot in Einbeck, Lower Saxony, that the plants in the ground bend and the airflow fluctuates.

The drone then clings high and begins to fly the mathematically accurate routes much more quietly over the area, where sensitive corn plants a few inches high grow. That’s why 40-year-old Burkhardt is flying a quadcopter over Einbeck this morning. There are a number of sensors hanging from it that scan the plants in the field to make sure they are growing. The drone systematically captures an area between three and eight acres with a full charge of the battery. That’s enough to stay in the air for almost half an hour, depending on the application and altitude, sensors, and flight plan. A multispectral camera and an imaging thermal camera measure the amount of light reflected and absorbed by plants at different wavelength ranges. Photos of the area are also used for evaluation.

Expert Marius Burkhardt is watching to see if the photos are taken from a constant height of 25 meters. This is the only way for the software to later determine the exact GPS coordinates and create a 3D model of each plant. Such a combination of technical innovation and breeding experience makes it possible to breed plant varieties resistant to new diseases in a shorter period of time. After all, the goal is for the seed to produce high-yielding crops, even in the most difficult farming conditions.

“Farmers want to use the seed that exactly meets their individual needs,” says Dr. Christoph Bauer, who at KWS is responsible for developing digital phenotyping technology and coordinating drone flights and evaluating the data from them. KWS’s in-house new digital tools support breeders in selecting the most suitable plants for their work from hundreds of thousands of options. “All of this helps our breeders identify the best plants,” says Bauer. “Drones and image analysis are increasingly valuable tools.”

The method behind it is as old as plant breeding itself: Phenotyping provides answers to questions such as whether the new varieties tolerate the environmental conditions of the site well. Or how and where the pests became infected. “Throughout KWS’s more than 160-year history, experts have surveyed field crops,” says Bauer. “It’s essential for the development of new varieties.” Breeders plant new plants each year and then evaluate how they develop in the area. They record the size, color, growth rate, number and shape of leaves, and other characteristics of the plants. Breeders can thus, in many small, often very laborious steps, ultimately deliver varieties perfectly adapted to the needs of the farmer. Aerial photography and software analysis significantly speed up this process. And it has additional benefits for farmers as well, explains technology expert Bauer. Fungal infection, dry areas, low chlorophyll content: Nothing goes unnoticed in front of digital sensors. Because measuring tools and software never get tired, the results of analyzing plant growth rates and leaf numbers and sizes are often more accurate and faster than without technical assistance.

As Bauer puts it bluntly: The human eye is not accurate enough. “If several people walk through a field to examine the development of the foliage, they usually get very similar results, but there are still small differences.” Technology is always objective and helps to eliminate these differences – and the advantage of speed is: “No one can assess such vast areas with such unquestionable objectivity.” Despite drones and software, plant breeders are still indispensable. Above all, the measurements need to be calibrated: “Of course, for us humans, it is easy to distinguish between two beet or corn plants,” says Bauer. “At the same time, transferring this knowledge to machines is a laborious task.” An example: The sunlight is particularly low in the morning and evening over the fields, and their light is then slightly redder than at noon. At 12 o’clock, however, the sun is very high in the sky and the proportion of blue light is higher. If there is dust on the leaves in the morning on dry days, but a shower in the afternoon washes it away, the appearance of the plants will change again within a day. Or the shape of the shadows differs in the morning and at noon. “We need to take all of this into account and make adjustments,” Bauer says. “Only then can we accurately recognize the condition of plants and leaf diseases.”

People are also vital to the evaluation and use of results. After returning from the field, drone pilot Burkhardt feeds the images into the analysis software. Combines hundreds of images into one large image. Delicate green, red, and blue lines run across the virtual field that appears on the screen. The group then discusses the data: What is the average leaf color? What area do the plants cover? And what does this mean for further breeding work? The goal here is to identify concrete successes and challenges. KWS’s interdisciplinary working group on drone images employs a number of image analysis experts. Physicist Christoph Bauer leads the group. Their staff and technologies are used at breeding stations around the world and have already helped gather a lot of positive experiences: “We have joined forces with the input of breeders, biologists and engineers,” says Bauer. “That was the reason for our joint success in analyzing the data.”

Plant breeders should continue to work in the field: They should check the conditions of field measurements and draw conclusions for breeding based on their personal observations. “We never make people redundant,” Christoph Bauer adds. “But the technology developed will help people make breeding decisions.” Drones are flying over more and more farms. Precision farming helps protect deer and moles from harvesters, detect impending drought at an early stage, and map mole hikes as uneven soil to draw conclusions about potential harvest losses. KWS continuously optimizes the complex system of drones, sensors, software, image recognition systems and servers used in plant breeding.