The availability of cheaper drones has propelled research, especially in fields like environment and ecology, where the subjects are often inaccessible. An example is the challenge scientists face studying Victoria amazonica, the giant Amazon water lily famous for its enormous, sturdy leaves and beautiful flowers. Named after Queen Victoria, these plants are fascinating, but getting up close to measure their growth manually is difficult and time-consuming due to the hostile, swampy regions they grow in. It can also be potentially dangerous due to the spikes and soft pond beds, and disturb the plant's natural development.
A recent study by an international collaboration of researchers from India and the Republic of Korea showcases just such a solution using unmanned aerial vehicles, or drones, to monitor the Amazon water lily aquatic plants. Researchers from Yeungnam University, Republic of Korea; SRM University AP; and Ulsan National Institute of Science, Republic of Korea embarked on a mission to see if readily available drone technology could accurately capture the physical characteristics, or phenotypic traits, of Victoria amazonica.
They used a standard commercial drone, specifically the DJI Phantom 4 Pro V2.0 equipped with a regular RGB camera, to fly over water lilies growing in an open pond. Over several months, they conducted weekly flights, capturing hundreds of high-resolution photos from different angles. The images were then fed into software that uses a technique called Structure from Motion (SfM).
SfM allows the software to use the features it sees in multiple photos to create a 3-dimensional model of the subject. Comparing how these features appear from slightly different viewpoints, the software can create a flat, map-like view called an orthomosaic, essentially stitching the aerial photos into a perfect, distortion-free overview, to build incredibly detailed, measurable 3D digital models of the plants.
Using these digital models, the researchers could measure key traits like leaf length and width, the angle between leaves, the height of the distinctive upturned leaf rim, the length and width of the submerged petiole (leaf stalk), and even flower width, all from their computer screens.
The results were impressive. When the researchers compared their drone-based digital measurements to traditional, hands-on measurements taken on an accessible plant, they found an almost perfect match. The accuracy was remarkable, with statistical tests specifically, an adjusted R-squared value, showing that the drone data could predict the manual measurements with over 98% accuracy for all traits. This means if the drone model measured a leaf as 50 cm wide, the actual leaf was almost certainly 50 cm wide, give or take a tiny fraction. The study successfully tracked the growth spurts of leaves and petioles, observed how young leaves were asymmetrical but grew into a more circular shape, and pinpointed traits like rim height and petiole dimensions as being highly variable during the plant's life cycle – information crucial for understanding plant development and breeding.
A drone-based method offers significant improvements over older techniques. It's contactless and non-destructive, meaning the plants aren't disturbed or damaged. It's much faster and safer than wading into a pond with measuring tapes and calipers. It also allows for frequent monitoring, which is essential for fast-growing species, and generates a rich dataset over time without the considerable investment in resources that is otherwise required.
While incredibly efficient, the researchers did face a few hurdles. Accurately measuring the entire length of the submerged petiole when leaves overlapped was a key challenge, as the digital model primarily captures the water's surface view. Shadows from nearby trees could also affect the 3D model generation, and drone battery life naturally limits flight time per session. However, the researchers suggest these are manageable issues that could be solved with technologies like underwater drones or advanced image filtering.
The research demonstrates a cost-effective, flexible, and highly accurate way to study giant water lilies and potentially other biodiversity in difficult-to-reach environments. The capability can also aid in precision agriculture and plant breeding programs by making it easier to monitor crop health in detail and identify desirable traits. More importantly, drones can allow for better ecosystem monitoring and conservation efforts, all while keeping the researchers and their study subjects safe and sound.
This research article was written with the help of generative AI and edited by an editor at Research Matters.