A University of Minnesota-led research team has developed new methods to assess how biodiversity loss impacts forest ecosystems by determining how sunlight reflects off the surface of the forest canopy using spectral images taken from an airplane. The research, published in the journal Nature Ecology & Evolution, lays the foundation for measuring the consequences of changes in biodiversity on ecosystem function remotely at a significant scale.
"We need to be able to determine how biodiversity is changing and impacting ecosystem functions in real time on large scales, from individual ecosystems to the biosphere," said study author Jeannine Cavender-Bares, a professor in the College of Biological Sciences and director of the newly established Biology Integration Institute (BII). "Right now, two in five plants are considered endangered. This study is an important step in learning how to detect where biodiversity is being lost, where efforts to slow its loss are succeeding, and how these changes in diversity are affecting our life support systems. Filling these knowledge gaps is important for making informed policy decisions that will impact future generations globally."
Many studies have shown how the functioning of ecosystems depends on biodiversity. Typically, to be able to assess the consequences of biodiversity change on ecosystem function in a given area, it requires small experiments where scientists manipulate the number and types of plants and assess how ecosystem functions respond. Often, this research begins with examining how much plants take up carbon and grow in plots of one species compared with mixtures of species. This is often a slow and difficult process that is limited to a small area of an ecosystem. However, the methods developed by researchers allow for a quicker analysis that could be applied over large areas via spectral imaging.
The first method was developed:
- by collecting remote spectral images across a tree diversity experiment at the University of Minnesota Cloquet Forestry Center outside of Duluth, Minnesota;
- through machine learning approaches, trained models to assess the tree biomass production on different plots from their spectral reflectance;
- comparing the spectral signatures and predicted biomass in plots of single species and species mixtures that researchers had already identified to determine how much mixing species enhances biomass.
The second method, built upon the first, determined spectral fingerprints of the 12 species present at the research site, and used these fingerprints to predict species diversity and abundance in different plots - showing that species and diversity itself may be remotely detected alongside the consequences of diversity for forest growth.
Laura Williams - a scientist at the BII, a postdoctoral associate in the College of Food, Agricultural, and Natural Resource Sciences and a graduate of the College of Biological Sciences - and colleagues established both methods that can be used with the spectral fingerprints of many species of trees in a temperate forest system.
"These models can be adapted to a variety of ecosystems around the world, as long as scientists have an idea of what types of plants are in a given area and the plant's spectral information has been collected and analyzed," said Williams.
In the future, methods such as those developed by Williams will inform analysis from NASA's global Surface Biology and Geology satellite mission, which is set to launch in 2027 and will focus - in part - in creating a baseline measurement of global biodiversity.
Additionally, using these methods, the study found that:
- nitrogen in the leaves of trees - which affects how much plants photosynthesize - can be mapped from spectral images;
- increased canopy nitrogen increased ecosystem productivity; and
- trees, when growing together in highly biodiverse areas, were able to pull more nitrogen out of the ground and into the canopy, which helps explain why more diverse forests may be more productive.
Study co-authors include: Peter Reich and Artur Stefanski with the University of Minnesota; Philip Townsend and Zhihui Wang, with the University of Wisconsin-Madison; John Couture, with Purdue University; and Christian Messier, with the Université du Québec à Montréal.
This research was funded by the National Science Foundation and the National Aeronautic and Space Administration (NASA), Hubachek Wilderness Research endowment, the Canada Research Chairs program, and the National Science Foundation's Biology Integration Institutes program.
