Site: Harvard Forest LTER
The flux tower rising above the forest canopy at the Harvard Forest LTER has been observing increasing CO2 uptake by the forest for nearly 2 decades. These typical New England forests with mixture of deciduous and coniferous species are important for keeping CO2 out of the atmosphere
John Budney

Terrestrial ecosystems, especially mid-latitude forests, are accumulating carbon (2.6 Pg-C y-1) that would otherwise contribute to increasing atmospheric CO2. CO2 fertilization, forest regrowth, climate variability, and increased nutrient inputs are potential factors. Ecological theory suggests that forests older than about a century should approach equilibrium and no longer be significant carbon sinks. Many forests in North America are approaching this age making it critical to understand what controls the net carbon balance in forests over time, and how long they can continue to absorb more CO2 than they release.

Observations of CO2 exchange between the forest and atmosphere using eddy covariance were established at Harvard Forest LTER in the early 1990’s. Automation was key to achieving consistently calibrated around-the-clock, year-round measurements. Simultaneous observations of temperature, light, and humidity provided insights into the dependence of whole-ecosystem carbon exchange on primary environmental drivers. Biomass inventories from plot-based measurements were consistent with cumulative fluxes. Since this pioneering study, hundreds of sites have begun similar measurements, leading to AmeriFlux, a U.S. network of carbon flux sites, and Fluxnet, a global network of networks that spans the globe.

In the forest where the HFR tower is located, the initial average annual carbon uptake was 2 Mg-C yr-1` (1 Mg = 1 metric ton), with some interannual variability. However, as the record extended past a decade it became apparent that net carbon uptake was accelerating rather than slowing as the forest matured. Analysis showed that simple temperature and light functions along with knowledge of phenology could predict 80% of the hour-to-hour variation in net carbon exchange, but those relationships combined with observed weather could not account for interannual variability or the observed trend. Carbon accumulation in this forest is dominated by growth of red oaks, while the biomass of the second most important tree species, red maple, has held nearly constant. Oaks have a higher photosynthetic efficiency and as their contribution to the forest canopy increases, more light is used to fix carbon. Annual temperatures have increased by nearly 1.5°C and the length of active growing season has increased by nearly 30 days. Partly the extension of carbon uptake comes from increasing contributions by conifer species in the spring and fall when the deciduous canopy is bare, showing the importance of increasingly complex forest structure over time. Wood in live trees accounts for most of the carbon in the forest, while accumulating dead wood comprises only about 10% of total carbon. Stimulation by increasing CO2 concentrations or changes in nutrient availability may play a role in the increasing carbon uptake as well.

These results demonstrate the importance of long-term measurement to observe forest dynamics on the time scales relevant to climate change and succession. The growth rate of this forest cannot increase indefinitely; continued observations will detect its eventual decline or reset by disturbance. The increasing rate of carbon uptake at Harvard Forest implies that middle aged, mixed deciduous forests that are typical across the northeastern U.S. continue to be important carbon sinks as long as they remain undisturbed. The dominant species have life expectancies over a century, and decay of dead trees is slow.

Graph for
Annual sums of Net carbon exchange (NEE) since 1992 at the Harvard Forest LTER show an increasing trend. The measured NEE is partitioned into ecosystem respiration (R) and gross ecosystem exchange (GEE = photosynthesis) based on temperature and light dependence of NEE.
Adapted and updated from Urbanski et al. 2007.