Forest communities change over time and so does the amount of carbon being sequestered and stored in those forests. Linking forest communities to carbon cycling can improve knowledge about the feedbacks between forest dynamics and climate change.

Forest communities—tree species that commonly co-occur—are constantly changing as trees die and new trees replace them. At the same time, carbon moves from the live to the dead carbon pool. A challenge is to determine the amount of carbon being sequestered and stored by these forest communities as they change over time. Northern Research Station scientists and collaborators borrowed an approach from text mining research, an artificial intelligence (AI)-based discipline focused on extracting information from written data, to determine how commonly co-occurring words (species) in sentences (forest inventory plots) form topics (communities). Scientists used over 20,000 remeasurements of around 5,600 Forest Inventory and Analysis (FIA) plots in the Great Lakes region to evaluate how forest communities change over time.

The findings quantified how much carbon was stored in each of six unique forest communities identified by the text mining algorithm. For example, the sugar maple-basswood community had the most live carbon, but the fir-cedar-birch community had the most standing dead carbon. The findings also showed that a combination of both the forest community and the forest structure in the FIA plots best predicted changes to the carbon pools. These improved knowledge areas can inform forest managers how shifts in their forests’ community composition and structure may impact their forests’ ability to capture and store carbon to further combat climate change.