For decades, scientists have generally thought that rivers emit more carbon dioxide, a greenhouse gas, than they take in. But a new analysis of every river network in the contiguous United States — including underrepresented rivers in deserts and shrublands — challenges this assumption, uncovering hints that many Western waterways may be soaking up carbon dioxide from the atmosphere. The findings about rivers' carbon uptake were published in Science and led by Taylor Maavara, an aquatic biogeochemist at Cary Institute of Ecosystem Studies.

“Rivers are one of the most uncertain parts of the global carbon cycle,” explained Maavara. “So in terms of balancing the global carbon budgets, figuring out where the carbon in rivers is coming from and where it's going is essential.”

One of the biggest sources of uncertainty in rivers is metabolism — the balance between how much carbon dioxide rivers soak up through photosynthesis, and how much they emit through the respiration of plants, animals, and microbes. Historically, monitoring data on river metabolism has been biased toward forested rivers in temperate areas. Using machine learning, Maavara and colleagues upscaled large observational datasets for a more holistic view. The study is the largest analysis of river metabolism to date, including monthly and annual photosynthesis and respiration rates for all the streams and rivers in the US.

“Estimating stream metabolism at large scales has been an elusive problem despite its importance to understanding the food webs of these unique and biodiverse systems,” said co-author Pete Raymond of Yale University, regarding rivers' carbon uptake. “This work has advanced our understanding of how streams function, which will allow for better stewardship of these important ecosystems.”

The team used US Geological Survey data to establish photosynthesis and respiration rates at hundreds of sites across the country. Then they used those data to train a machine learning algorithm to rank the factors driving photosynthesis and respiration rates. Factors included light availability, water temperature, nutrient and organic matter, and river flow rates. The model was then able to estimate photosynthesis and respiration rates for river reaches where data hadn’t been collected.

Read the full article about rivers' carbon uptake at Cary Institute of Ecosystem Studies.