New AI Technology Enables Study of Plant-Climate Relationships
Scientists from the University of New South Wales (UNSW) and the Botanic Gardens of Sydney have harnessed the power of artificial intelligence (AI) to analyze data from millions of plant specimens housed in herbaria worldwide. This breakthrough allows researchers to better understand and combat the effects of climate change on plant life.
Herbaria, which are collections of plant specimens, provide a valuable record of plant biodiversity over time. However, as the number of specimens continues to grow rapidly, manually processing the data is no longer feasible. To address this challenge, the research team employed a new machine learning algorithm to analyze over 3,000 leaf samples.
The study, published in the American Journal of Botany, challenges the commonly observed interspecies pattern that suggests leaf size increases in warmer climates within a single species. By processing large amounts of data, the team discovered that factors other than climate have a significant impact on leaf size within a plant species. This finding highlights the potential of AI to transform static specimen collections into dynamic resources that can quickly document the effects of climate change.
Herbaria have been in existence since at least the 16th century and contain a vast amount of information about plant species. To facilitate scientific collaboration, efforts have been made in recent years to digitize these collections. The Botanic Gardens of Sydney undertook the largest herbarium imaging project, transforming over 1 million plant specimens into high-resolution digital images.
Dr. Jason Bragg of the Botanic Gardens of Sydney reached out to Associate Professor Will Cornwell from UNSW to explore how machine learning could be applied to the digitized herbarium specimens. Together with UNSW Honours student Brendan Wilde, they developed an automated algorithm capable of detecting and measuring the size of leaves from scanned herbarium samples belonging to the Syzygium and Ficus plant genera.
Using a convolutional neural network, a type of AI known as computer vision, the algorithm was trained to identify and measure leaves. This approach enables the efficient processing of specimens and the logging of individual characteristics. While previous studies have shown a consistent pattern of larger leaves in wetter climates and smaller leaves in drier climates, the team’s research found that this correlation does not hold within a single species across different regions. This discrepancy may be attributed to a process called gene flow, which weakens plant adaptation on a local scale.
The use of AI and machine learning in this study not only improves the speed and accuracy of leaf size measurements but also provides valuable insights into the relationship between leaf traits and climate. Additionally, these tools can aid in documenting and predicting the effects of climate change on plant species. By training the algorithms to recognize hidden trends, researchers can uncover new knowledge about plant evolution and adaptations.
The application of machine learning to herbarium specimens opens up new possibilities for studying plant-climate relationships on a global scale. The digitization of herbaria and the integration of AI technology enable scientists to unlock the potential of these valuable collections, contributing to a greater understanding of our changing natural environment.
