Study: Plant functional diversity varies greatly

“We wanted to find out how large the temporal changes in functional diversity are across different regions of the Earth, and whether simple data collection methods can provide the information we need. Thanks to new time series of images from the EnMAP satellite and the use of artificial intelligence for data analysis, this has now been made possible for the first time,” explains Mederer, who conducts research at the Institute for Earth System Science and Remote Sensing. A single snapshot, he notes, is not enough to capture and understand biodiversity in all its dimensions. By combining satellite image analysis with AI, the researchers gained insights into global biodiversity patterns that would not be possible using traditional methods. While the new technologies cannot replace established approaches, Mederer adds that they offer a valuable complement for gaining a better understanding of ecosystems.

The researchers used AI algorithms to derive plant traits from the satellite scenes. Based on these traits, they then calculated quantitative measures of functional diversity – that is, the different roles and characteristics of plants within a habitat. Their goal is to make maps of plant functional diversity as reliable as possible. This is important, among other things, for global models and for monitoring ecosystems in the context of climate change, as such products can be used as decision-making criteria, Mederer explains.

“Next, we plan to improve the spatial resolution of the EnMAP satellite scenes using an image-sharpening algorithm so that we can also capture small-scale differences,” he adds. That would allow us to represent changes in plant functional diversity with even greater accuracy,” emphasises Mederer.

So far the spatial resolution of the satellite images has been 30 metres. Because of this, the researchers observed changes in functional diversity at the landscape scale rather than for individual plants. In addition, data availability was not consistent across all regions of the world. For example, the team was unable to analyse the tundra and boreal forest biomes because there were not enough EnMAP scenes available for these areas. Clouds also proved problematic, as they often obscure vegetation during many seasons, Mederer notes. Moreover, the researchers were unable to obtain information about features of the understorey or about plant traits that are not visible in the vegetation canopy.