Improving accuracy of plant observation using AI

Artificial intelligence (AI) can help plant scientists collect and analyze unprecedented amounts of data that would not be possible using traditional methods. Researchers at the University of Zurich (UZH) are now using big data, machine learning and field observations in the university’s experimental garden to show how plants respond to changes in their environment.

With climate change, it is becoming increasingly important to know how plants can survive and thrive in a changing environment. Traditional laboratory experiments have shown that plants accumulate pigments in response to environmental factors. Previously, such measurements were made by taking samples, which required removing and damaging parts of the plant. “This labor-intensive method is not viable when thousands or millions of samples are required. Additionally, taking repeated samples can damage the plants, which can affect how plants respond to environmental factors. “There hasn’t been a good way to observe individual plants within an ecosystem over long periods of time,” said Reiko Akiyama, lead author of the study.

With support from UZH’s University Research Priority Program (URPP) ‘Evolution in Action’, a team of researchers has developed a method that allows scientists to observe plants in nature with great precision. PlantServation is a method that incorporates robust image acquisition hardware and deep learning-based software to analyze field images and works in any weather.

Millions of images support the robustness evolution hypothesis

Researchers used PlantServation to collect the following (top-down) images. Arabidopsis We observed plants over three seasons (5 months from autumn to spring) in an experimental plot on UZH’s Irchel campus and analyzed over 4 million images using machine learning. The data document species-specific accumulation of plant pigments called ‘anthocyanins’ in response to seasonal and annual variations in temperature, light intensity and precipitation.

PlantServation also allowed scientists to experimentally recreate what happens after natural speciation in hybrid polyploid species. These species arise from duplication of the entire ancestral genome. This is a common type of species diversification in plants. Many wild and cultivated plants, such as wheat and coffee, arose in this way.

In the current study, the anthocyanin content of hybrid polyploid species is A. Kamchatika From autumn to winter, its anthocyanin content was similar to that of the ancestral species originating from warm regions, and from winter to spring, its anthocyanin content was similar to other species originating from cold regions. “Our results confirm that these hybrid polyploids combine the environmental responses of their ancestors, supporting long-standing hypotheses about polyploid evolution,” said the study’s two corresponding authors. One of them, Tsugurie Shimizu, says:

From the Irchel campus to remote areas

PlantServation was developed in the experimental garden of UZH’s Irchel campus. “It was very important for us to be able to use the gardens on the Irchel campus to develop PlantServation’s hardware and software. But its application goes even further. When combined with solar power, that hardware Equipped with economical, robust hardware and open-source software, PlantServation is paving the way for future biodiversity research to explore other plants using AI. Open. Arabidopsis – from crops such as wheat to wild plants that play important roles in the environment,” said Kentaro Shimizu, corresponding author and co-director of URPP Evolution in Action.

The project is an interdisciplinary collaboration between LPIXEL, a company specializing in AI image analysis, and Japanese research institutes Kyoto University and the University of Tokyo, among others, under the UZH Global Affairs and International Leading global strategy and partnership funding scheme. It’s research. A research grant project of the Japan Society for the Promotion of Science (JSPS). The project was also funded by the Swiss National Science Foundation (SNSF).

Strategic partnership with Kyoto University

Kyoto University is one of UZH’s strategic partner universities. Strategic partnerships ensure that high-potential research collaborations receive the support they need to succeed, for example through UZH’s Global Strategic and Partnership Funding Scheme. In recent years, several joint research projects between Kyoto University and UZH have already received funding, among them “PlantServation”.