Lead Authors:
Deborah N. Huntzinger, Northern Arizona University
Abhishek Chatterjee, Universities Space Research Association and NASA Global Modeling and Assimilation Office
Contributing Authors:
David J. P. Moore, University of Arizona
Sara Ohrel, U.S. Environmental Protection Agency
Tristram O. West, DOE Office of Science
Benjamin Poulter, NASA Goddard Space Flight Center
Anthony P. Walker, Oak Ridge National Laboratory
John Dunne, NOAA Geophysical Fluid Dynamics Laboratory
Sarah R. Cooley, Ocean Conservancy
Anna M. Michalak, Carnegie Institution for Science and Stanford University
Maria Tzortziou, City University of New York
Lori Bruhwiler, NOAA Earth System Research Laboratory
Adam Rosenblatt, University of North Florida
Yiqi Luo, Northern Arizona University
Peter J. Marcotullio, Hunter College, City University of New York
Joellen Russell, University of Arizona
Science Lead:
Melanie A. Mayes, Oak Ridge National Laboratory
Review Editor:
Tara Hudiburg, University of Idaho
Federal Liaisons:
Elisabeth Larson, North American Carbon Program and NASA Goddard Space Flight Center, Science Systems and Applications Inc.
John Schade, National Science Foundation
Karina V. R. Schäfer, National Science Foundation

Future of the North American Carbon Cycle

By absorbing atmospheric CO2, the land and ocean play an important role in slowing the buildup of GHGs in the atmosphere, thereby slowing the pace of climate change. As mentioned at the outset of this chapter, an important question in carbon cycle science is whether ocean and land systems will continue to provide this service or whether the strength of the ocean and land carbon sink will decrease under changing climate conditions (Michalak et al., 2011). Numerous vulnerabilities are associated with assessing current and projected carbon cycle conditions. Taking into account the magnitude, timing, and likelihood of projected carbon cycle changes discussed in this chapter, this section synthesizes current understanding, highlighting critical carbon cycle vulnerabilities, knowledge gaps, and key research needs related to the co-evolution of carbon cycle dynamics in a changing climate.

19.7.1 CO2 Fertilization

Crucial to projecting future changes in the North American carbon cycle is the ability to project the response of land ecosystems to increasing atmospheric CO2. As discussed in Section 19.4.1, three areas of incomplete understanding limit current efforts to project forest and terrestrial ecosystem responses to increasing CO2: 1) age distribution of forests, 2) nutrient interactions (particularly nitrogen), and 3) soil carbon responses. These three areas are interrelated because of a lack of understanding about carbon-nitrogen coupling. More research is needed to understand what constitutes plant nitrogen demand, carbon-allocation strategies used by plants to respond to nutrient demand, the carbon cost of nitrogen acquisition, factors that determine the capacity of soils to supply nitrogen, and soil carbon losses associated with increased soil nitrogen mineralization.

19.7.2 Permafrost Carbon–Climate Feedback

A primary uncertainty in carbon-climate feedback projections stems from limited understanding of the responses of carbon stocks in the northern high latitudes (≥60°N) to a changing climate. Estimates show that, globally, surface permafrost (0 to 3 m) contains about 33% of the overall surface soil carbon pool (1,035 ± 150 Pg C; Hugelius et al., 2014). Along with carbon deposits deeper than 3 m (including those within the Yedoma region) and subsea permafrost carbon, the total estimate of terrestrial permafrost carbon in the northern permafrost zone is 1,330 to 1,580 Pg C (Schuur et al., 2015). More recent simulations (McGuire et al., 2018) estimate that between 2010 and 2299, losses of permafrost between 3 and 5 million km2 for the RCP4.5 climate and between 6 and 16 million km2 for the RCP8.5 climate may be possible.

The permafrost zone’s overall carbon budget is determined by the soil carbon as well as vegetation carbon dynamics and their interactions. For example, increased vegetation growth due to warming leads to greater soil carbon inputs, whereas permafrost thawing accelerates carbon release (see Ch. 11: Arctic and Boreal Carbon). The presence of large carbon stocks in a rapidly warming region raises concern about increased carbon emissions, as well as changes in global albedo, the hydrological cycle, and thermohaline circulation (Hinzman et al., 2013).

The primary challenge in projecting the trajectory of permafrost thawing is that the physical and biogeochemical properties of permafrost vary widely depending on the characteristics of the parent material, ice and liquid water content, topography, biota, and climate (Jorgenson et al., 2010). With continued warming and large-scale losses of near-surface permafrost, almost all terrestrial carbon cycle models indicate that by the end of this century, the Arctic could shift from a net sink to a source of carbon (Cox et al., 2000; Fisher et al., 2014b). Considerable debate remains, however, on the amplitude, timing, and form of the carbon release (e.g., Lenton et al., 2008; Schuur et al., 2015; Slater and Lawrence 2013). This disagreement is directly related to a lack of understanding of three key factors that determine the potential climate feedback of the permafrost carbon pool: 1) area and depth of permafrost vulnerable to release, 2) the speed with which carbon will be released from thawing soils, and 3) the form of carbon (e.g., CO2 or CH4) that will be released (NRC 2014). Similar to land permafrost, questions have emerged about the stability of organic carbon sequestered in the marine permafrost of Alaska and Canada amid climate warming (see Section 19.7.4). Combined, these limitations in understanding result in considerable uncertainty in how future climate change will affect northern high latitudes and reshape traditional ways of life. Ongoing research efforts led by U.S., Canadian, and international partners have highlighted the need for long-term empirical observations to capture soil carbon dynamics to improve understanding of land carbon–climate feedbacks and evaluate model performance, thereby constraining future projections.

19.7.3 Disturbance

Fire and Disease

Natural and human-driven disturbances will influence future vegetation carbon storage. Forest disturbance is a fundamental driver of terrestrial carbon cycle dynamics (Hicke et al., 2012), and harvesting, fire, wind throw, storms, pathogen and pest outbreaks, and drought collectively lead to the removal of 200 Tg C from U.S. forests annually (Williams et al., 2016). Initially, most disturbances shift an ecosystem to a carbon source, while recovery from disturbance is commonly associated with greater net ecosystem carbon storage (Magnani et al., 2007; Odum 1969). Hence, disturbance effects on carbon balance in forests are both immediate and lagged and potentially long lasting. Given current management practices, climate change is likely to increase the frequency and intensity of ecological disturbances across multiple spatial and temporal scales (Running 2008). For example, reduced water availability resulting from decreased precipitation and snowpack probably will increase forest susceptibility to fire and insect attack (Allen and Breshears 1998; Breshears et al., 2009; Westerling et al., 2006).

Fire activity is largely expected to increase (Sommers et al., 2014; Westerling et al., 2006) in many regions, with fire seasons starting earlier and ending later compared to previous decades (Jolly et al., 2015). Uncertain, however, is whether regional fire severity will decrease or increase (Collins 2014; Fried et al., 2004; Parks et al., 2016; Stavros et al., 2014) by midcentury. In the western United States specifically, projected increases in fire activity (Westerling et al., 2006) imply a decrease in biomass accumulation between successive fires, resulting in less biomass available for combustion and, thus, a reduction in fire severity. A recent study by Parks et al. (2016) also points out that projected increases in water stress will decrease productivity in the generally water-limited western United States, which may also feedback to further reduce the amount of biomass available to burn. However, since changes in fire–carbon cycle linkages are highly ecosystem specific, temperature-limited forests (e.g., northern high latitudes)—unlike the water-limited forests of the western United States—will likely experience increased fire frequency and severity under a warmer climate (Kasischke et al., 2010).

The extent and severity of forest insect disturbances has increased with changing climate conditions (Kurz et al., 2008). As climate warms, the range of insects (e.g., mountain pine beetle) has expanded into higher elevations and latitudes, putting previously unaffected forests at risk (Bentz et al., 2010; Kurz et al., 2008). Combined, these changes in disturbance regime and severity may result in significant loss of forest carbon sinks, particularly in North America as live carbon stocks transition to dead (Hicke et al., 2012; Kurz et al., 2008). However, the timing of carbon release associated with forest insect disturbances is unclear because of uncertainty surrounding respiration suppression or enhancement (Borkhuu et al., 2015; Levy-Varon et al., 2014; Moore et al., 2013); specific biogeochemical, microbial, and hydrological responses (Edburg et al., 2012; Maurer et al., 2016; Trahan et al., 2015); and the overall ecosystem carbon balance (Ghimire et al., 2015). Losses of carbon stocks caused by disturbance are mediated by interactions among climate, vegetation type, and productivity, with changing forest management practices resulting in reduced potential fuel loads and thus reductions in fire severity (Parks et al., 2016).


Similar to fire and insect infestations, droughts can trigger immediate and time-lagged effects on carbon stocks and flows (van der Molen et al., 2011). Both seasonal short-term observations and modeling studies have documented the effects of drought on ecosystem carbon fluxes (Anderegg et al., 2012, 2015; Ciais et al., 2005; Doughty et al., 2015; Keenan et al., 2009; Zeng et al., 2005). Over the last decade, midlatitudes in the United States have experienced frequent drought events, and similar events are expected to increase in area, frequency, intensity, and duration (e.g., Blunden et al., 2011; Kogan et al., 2013; USGCRP 2017a). Although early prediction and detection of water-induced vegetation stress are critical for agribusiness and food security (Jones et al., 2011), the exact coupling between the carbon and hydrological cycles remains unclear, as does the response of different vegetation types to short-term water stress. For example, the impact of the 2012 summer drought in the United States was compensated by increased spring carbon uptake due to earlier vegetation activity (Wolf et al., 2016); these two opposing effects mitigated the impact on the net annual carbon uptake for 2012. Is the response observed in 2012 representative of what can be expected under future climate change? The answer to this question remains highly uncertain. Climate projections from the CMIP5 ensemble of model simulations show warmer spring and drier summer mean conditions across the United States similar to those observed in 2012. Additionally, drought-induced near-term changes in plant water content can have a longer-term impact by increasing an ecosystem’s vulnerability to other disturbances, such as wildfire and insect outbreaks (Arnone et al., 2008; Reichstein et al., 2013; van Mantgem et al., 2009). Thus, future projections of carbon cycle vulnerability due to drought need to adopt a holistic modeling framework to assess the full range of responses to climate extremes.

Land-Use and Land-Cover Changes

Understanding the carbon cycle effects of changes in land-use and land-cover (LULC) management requires insights into diverse issues and processes. These include the socioeconomic factors (e.g., technological change and market incentives) driving human use of land, as well as the biophysical (e.g., albedo, evaporation, and heat flux), biogeochemical (e.g., carbon and nutrient cycling), and biogeographical processes (e.g., location and movement of species) affected by land-use choices. For example, intensive agriculture in the western United States appears to have caused abrupt losses of Arctic ecosystem structure and soil erosion (carbon cycling) due to increased populations of migrating snow geese supported by agricultural food supplies (Jefferies et al., 2006; MacDonald et al., 2014). Such dynamic interconnectivity and coupling between natural and human-driven activities at different space-time regimes demonstrate the challenge in projecting long-term feedbacks between the carbon cycle and land use.

As discussed in Section 19.3.2, generating estimates of future potential LULC management and change is challenging because of the difficulty in projecting not only dynamics within and between complex terrestrial ecosystems, but also future potential climate, macroeconomic, and social conditions. Moreover, many of these conditions can vary significantly, depending on location and the temporal and spatial scales of the analysis. Policies and programs can significantly affect land use, especially on public lands, whereas market signals can have a large impact on how private lands are used. For example, the role of markets is important as landowners make decisions affecting LULC management, which in turn affects GHG emission levels, ensuing climate change, and thus carbon cycles. As a result, there is relatively high variability in projected estimates of land-cover change and associated impacts on carbon stocks and net emissions (Buchholz et al., 2014). Additional research is needed to model existing trends in land management and to develop scenarios of future land management and associated changes in carbon stocks and emissions (USGCRP 2017b).

19.7.4 Ocean and Coastal Carbon Cycles

Key uncertainties in processes that affect carbon cycling in the ocean and coastal zones limit the ability to project future system responses. Often highly populated, coastal zones have diverse uses as residential, urban, industrial, shipping, and recreational areas, resulting in a complex interplay of management drivers. Management of coastal wetlands, mangroves, and seagrass beds amid sea level rise, in particular, will have important carbon cycle consequences because these systems sequester carbon with extremely high efficiency and would be replaced by other systems whose sequestration efficiency is much lower. Natural disturbances commonly responsible for the loss of carbon-intensive ecosystems include hurricanes, earthquakes, disease, and herbivore grazing. The human activities most affecting these coastal ocean ecosystems are nutrient and sediment loading from runoff and sewage disposal, dredging and filling, pollution, upland development, and certain fishing practices such as trawling (Short and Wyllie-Echeverria 1996). Although activities such as dredging of shipping channels and erosion-control measures can have locally strong implications, more regionally expansive activities such as bottom trawling may have important coastal carbon cycle effects, depending on trawling intensity and bottom biogeography (e.g., Duplisea et al., 2001).

Changes in sedimentary carbon processing due to warming, acidification, or deoxygenation will alter the source and sink status of coastal zones, which already are insufficiently understood. Continued human disturbance of coastal zones represents an added perturbation to biological production and respiration both in the water column and in sediments, with the potential to substantially alter existing and also poorly understood coastal carbon cycling. Microbial regeneration of organic matter under warming, deoxygenation, and acidification may change as well, altering the timing, magnitude, or locations of CO2 release back into seawater. Vertical export of carbon via the creation of sinking material such as fecal pellets and marine snow (Alldredge and Silver 1988) is still poorly understood and parameterized in many models. In addition, the physiological and ecosystem impacts previously outlined (e.g., changes in grazing or recycling) also may influence how much carbon is sequestered to the deep ocean by vertical export (Marsay et al., 2015). Finally, compared to terrestrial systems, there is only rudimentary understanding of ocean and coastal system resilience to climate- or carbon-driven perturbations and the speed with which they may recover from short-term disturbances under climate change.

High-latitude coastal ecosystems are among those most likely to experience an amplification of global change (e.g., Serreze and Francis 2006). Along with significant increases in river discharges in the past century, most of the coastline in the northern high latitudes is receding at an unprecedented rate due to coastal erosion, mobilizing large quantities of sediments and carbon. Estimates of the biogeochemical processes, interactions, and exchanges across the land-ocean interface in this region are still poorly constrained. Detailed studies have examined specific aspects of individual northern, high-latitude rivers including the Yukon (Dornblaser and Striegl 2009; Spencer et al., 2008) and Mackenzie (e.g., Emmerton et al., 2008). However, only a few studies have assessed how these riverine fluxes directly affect the coastal ecosystems from river deltas to estuaries on larger regional scales (e.g., Dittmar and Kattner 2003) and longer-term decadal timescales (e.g., Overeem and Syvitski 2010).

19.7.5 Freshwater Carbon Cycle

Freshwater ecosystems are particularly vulnerable to anthropogenic disturbances and are considered to be among the most threatened ecosystems on the planet (Vorosmarty et al., 2010). Human activities such as water management, river fragmentation by dams, alteration of natural flow, construction of water impoundments, and changes in land use have a major impact on freshwater ecology, biology, and carbon cycling. There is high confidence that direct human impacts will continue to dominate the threats to most freshwater ecosystems globally over the next three decades as urbanization increases, irrigated agriculture expands, and human demand for water resources grows (Settele et al., 2014). The high connectivity between lakes and their catchments suggests that future CO2 concentrations in lakes and exchanges with the atmosphere will be highly sensitive to altered catchment management and effects of climate change on catchment characteristics (Maberly et al., 2012). Projected increases in human-driven nutrient inputs, from either watershed or airshed processes (Rabalais et al., 2009), are expected to enhance inland water primary production and biological uptake of atmospheric CO2 (Pacheco et al., 2014). Acidification may put additional ecological pressure on freshwaters (Hasler et al., 2016; Phillips et al., 2015; Weiss et al., 2018), thus further confounding the impacts. Similarly, concomitant increases in organic carbon inputs and intensification of mineralization could offset increased CO2 uptake in many of these systems (Jansson et al., 2008).

Projecting the response of freshwater systems to future environmental change will require accounting for differences across systems and climatic regimes. Also needed are projections that include the complex interactions between climate change and the many natural and human-driven stressors that affect inland ecosystems. Key uncertainties exist in the mechanistic understanding of carbon sources, lability, and transformations taking place in inland waters. To better predict freshwater systems, improved coupled ­hydrodynamic-biogeochemical models are needed, along with new remote-sensing tools and sensors with high spatial and spectral resolution for capturing the broad spatiotemporal variability that characterizes freshwater carbon fluxes.

Finally, it is worth underscoring that significant knowledge gaps remain in current understanding of the future trajectory of North American carbon storage in terrestrial and aquatic ecosystems, permafrost carbon-climate linkages, and the role of natural and human-driven disturbance on carbon cycling dynamics. These and other impacts, vulnerabilities, and risks are recognized as meriting attention and research. For all these emerging research areas, a combination of observational, experimental, synthesis, and modeling activities is needed to gain a predictive understanding of these processes (see Box 19.2, Improving Model Projections of Future Carbon Cycle Changes), and thereby better constrain the future of the North American (and global) carbon cycle.

See Full Chapter & References