The Pacala, Medvigy, Hedin, and Sarmiento groups are working to understand whether forest storage of carbon will be enhanced by increasing CO2 levels, what might limit “CO2 fertilization,” and the history and future of this component of the terrestrial carbon sink.


History of the terrestrial sink

The most important accomplishment of the Pacala climate modeling group in 2012 was an analysis of the value of the historical carbon sink in slowing the rise of atmospheric CO2 and warming. During 2012, Elena Shevliakova and Sergei Malyshev participated in development, evaluation, and analysis of the GFDL comprehensive Earth System Models ESM2M and ESM2G. In collaboration with the GFDL and CMI scientists, they explored interactions between historic land use and increased atmospheric CO2 concentrations and their implications for carbon cycle and climate. The researchers estimated for the first time that in the absence of historical CO2 fertilization, the concentration of atmospheric CO2 would have been 80% greater than observed, and the warming would have been 40% larger than observed. This work will appear in Proceedings of the National Academy of Sciences in 2013.


Nutrient limitation and the CO2 fertilization sink

Because elevated CO2 generally increases photosynthetic production, enhanced forest growth could scrub anthropogenic CO2 from the atmosphere and provide a negative feedback on climate change. It is commonly suggested that enhancement of the land sink by CO2 fertilization should be limited by the availability of nitrogen, yet forest free-air CO2 enhancement experiments (FACE) have shown continued CO2 fertilization despite nitrogen limitation.

Ray Dybzinski and Caroline Farrior have developed a new forest model that includes a previously unrecognized mechanism and explains the continuing CO2 fertilization in the FACE experiments. In their simulations, carbon allocation patterns are determined by competition. Enhanced carbon fixation under elevated CO2 resulted in elevated wood growth and height, but constant fractional allocation to wood, constant allocation to leaves, and elevated fractional and absolute fine root growth. This is positive news for carbon mitigation, as the new model predicts that the CO2 fertilization component of the land sink will continue for decades. The researchers are working to implement this model in the GFDL land model, LM3 to improve predictions of carbon cycling.

In a related collaboration with Steve Pacala, Ensheng Weng of the U.S. Forest Service developed a working version of a new model of the terrestrial biosphere and has written a manuscript that will be submitted soon. This model has a revolutionary structure, in that it models realistic competition among plant types and so should be able to predict the CO2 fertilization effects described above.


CO2 fertilization of recovering tropical forests

The Hedin/ Medvigy group is also focusing on understanding the competitive processes that affect the forest carbon sink, specifically the role of changes in resource availability and competitive dynamics among individuals following disturbances in the tropics. Tropical forests contribute a significant portion of the land carbon sink, but their future ability to sequester CO2 likely depends on how nutrients interact with forest recovery from cutting, agricultural land use, or natural disturbances. Recent observations place particular importance on this as yet unresolved interaction; first, it is becoming clear that a large fraction of tropical forests worldwide are recovering from some form of disturbance. Second, there is increasing evidence for exceptionally strong constraints by nutrients on carbon accumulation, but only at specific “bottleneck” periods during forest recovery. Third, results from models and empirical studies imply that nitrogen-fixing trees may act to alleviate nitrogen limitation on plant growth during particular periods of forest recovery.

Although existing models of the tropical land carbon sink are exceptionally sensitive to potential interactions between carbon recovery and nutrient cycles, they have not been constructed to resolve some of the spatial and temporal scales that are fundamental to nutrient-driven processes. To address this problem, Jennifer Levy has developed a new framework for understanding nutrient cycling on the level of individual trees, and has successfully incorporated this framework into a terrestrial biosphere model, the Ecosystem Demography model 2 (ED2).

Unlike conventional ecosystem models, ED2 resolves (1) heterogeneity in resource environments and (2) resource competition between trees of different sizes and functional types. Because of these two factors, it is possible to scale understanding of nitrogen fixation and nutrient limitation from individual trees to ecosystem-level properties. Furthermore, because field studies often measure properties of individual trees, there is a wealth of data that can be used to challenge and evaluate the new model. The researchers are currently using measurements of nitrogen fixation and the results of nutrient fertilization experiments for this purpose. The resulting validated model will be an important tool for assessing the capacity of tropical forests to act as carbon sinks.

The Medvigy and Hedin labs are in the process of expanding this model by adding a phosphorus algorithm to enable simulation of carbon, nitrogen, and phosphorus interactions. Once this algorithm is ready, it will be easily transferrable into LM3 and other similar models in use Princeton.


North-South variation of nitrogen fixation and terrestrial carbon uptake

This year Duncan Menge used forest inventory data from the USA and Mexico to show that nitrogen-fixing plants comprise ~10% of trees south of 35 degrees latitude but only ~1% of trees north of 35 degrees. Furthermore, the dominant type of nitrogen-fixing tree switches at this same threshold. Menge’s research also showed that this transition from 10% to 1% can be explained by a concomitant transition in the nitrogen-fixing “strategy,” from rapid tuning of nitrogen fixation in the south to slow or no tuning in the north. The dominant types in the north versus south are thought to have these different strategies, lending support to the hypothesis that these different strategies explain the transition. This is important for climate change because plants that tune nitrogen fixation rapidly (the southern type) remove more carbon dioxide from the atmosphere, whereas plants that do not tune nitrogen fixation (the northern type) remove less carbon dioxide.


Understanding temporal shifts in terrestrial uptake of atmospheric CO2

In 2011, Jorge Sarmiento and colleagues reported that an abrupt shift in the net land carbon sink, estimated as the residual between fossil fuel emissions, the growth rate of atmospheric CO2 at Mauna Loa, and modeled ocean carbon uptake, occurred in the late 1980s. The land carbon uptake appears to have remained relatively constant for three decades and to have increased rapidly after 1988/1989.

In collaboration with researchers at UCLA, NASA and the Medvigy group as part of a study supported by a NASA Carbon Cycle Science grant, this year the Sarmiento group analyzed a suite of simulations of primary productivity from the terrestrial biogeochemical model CASA and upscaled FluxNet data to identify independently regions/ecosystems in which carbon uptake is consistent with the timing and magnitude of carbon sinks derived from previous studies. Results show that globally, the net primary productivity (NPP) increased by about 1 Pg C/yr (or 1 billion metric tons of carbon per year) after 1989. The gross primary productivity also increased of approximately 2 Pg C/yr at the same time. These estimates are consistent with the shift in net land carbon uptake detected in previous work. Results further suggest that three key regions are contributing to the abrupt increase in productivity in the late 1980s: Northern Eurasia, Tropical Africa and Tropical South America (see Figure 1).

Figure 1: NPP differences between 1990-2008 and 1982-1989. The NPP is estimated using the terrestrial biogeochemical model CASA. Only significant differences are shown (t-test, 5% critical level). The green rectangles are highlighting three regions exhibiting a very strong signal: a) Northern Eurasia, b) Tropical Africa and c) Tropical South America.

The group found that these changes may be climate-constrained. Results showed that the productivity changes observed in the three regions seem influenced by different climatic factors: a) warming in Northern Eurasia (from late 1980s onwards), b) increased precipitation in Tropical Africa (from late 1980s/early 1990s onwards) and c) increased solar radiation over Tropical South America (from mid-1990s onwards).

Claudie Beaulieu in the Sarmiento group is currently analyzing additional CASA simulations to study sensitivity of the results to different forcing data sets. Furthermore, the group is analyzing net ecosystem exchange runs to verify whether the key regions have actually gained carbon or if this increase is due to changes in respiration.