The atmosphere and the earth’s ecosystems are parts of a coupled system. For a large variety of processes, forcing from one partner in the interaction elicits one or more responses from the other partner, which in turn elicits other responses from the first. This bidirectional coupling gives atmosphere-ecosystem interactions the potential to be among the most complex in the natural world. The increasing involvement of human actions as important drivers introduces a broad new suite of responses and interactions. Historically, most of the study of drivers and responses in atmosphere-ecosystem interactions has started with single-factor investigations, building on the infrastructure, concepts, and tools of particular disciplines. Over the last several decades new knowledge has continued to accumulate in the traditional disciplines, but more and more of the breakthroughs are at the borders of traditional disciplines. Climate dynamics, hydrology, atmospheric chemistry, ecology, oceanography, and geomorphology function increasingly as a single superdiscipline, often called earth system science. In the future continued progress in this new superdiscipline is likely to require effective collaboration with or integration of a wide range of human sciences, from agronomy and civil engineering to economics and government.
The potential importance of bidirectional interactions is long acknowledged but relatively little studied, at least until recently. For example, Ahhrenius’s calculations (1896) of climate forcing from coal combustion identified key components in anthropogenic warming, and in the 19th century the claim that rain follows the plow was a powerful inducement for agricultural expansion in the western United States. Following the introduction of climate models, insights on bidirectional coupling began to emerge. Studies by Charney et al. (1975, 1977)
on the role of vegetation (or lack of vegetation) in modulating the climate of the Sahara are classic foundations for the modern science of atmosphere-ecosystem interactions. Later studies on the role of vegetation in the climate of the Amazon basin (Salati and Vose, 1984; Shukla et al., 1990; Dickinson and Henderson-Sellers, 1988; Lean and Warrilow, 1989) began to bring human actions into the science of atmosphere-ecosystem interactions. At about the same time, analyses of deforestation indicated its potentially large contribution to climate forcing through the carbon cycle (Woodwell et al., 1983). Also around this time a series of breakthroughs established the role of chemicals released from plants and from human processes in modulating the chemistry of the atmosphere (ozone hole, biogenic volatile organic components).
Since these early discoveries, understanding the nature and implications of atmosphere-ecosystem interactions has been one of the central goals in earth system science. It is also increasingly clear that understanding atmosphere-ecosystem interactions is one of the fundamental prerequisites for designing a path to a sustainable future.
CONTEXT AND IMPACTS
The earth, oceans, atmosphere, and human actions need to be considered as a single, coupled system for a thorough understanding of climate, ecosystems, hydrology, or atmospheric chemistry. At small spatial and temporal scales, the coupling ceases to be of first-order importance. But at larger scales of space and time, the coupling between the atmosphere, land ecosystems, and oceans is always relevant and often dominant.
In the coupled earth system, components respond differently to different forcings. Responses are often nonlinear and often have threshold-type characteristics, with little response over a wide range of forcing, followed by a transition to a fundamentally new state over a short time or a narrow range of forcing. Understanding the locations of these thresholds and the mechanisms controlling them is among the most important challenges in earth system science. The lack of an obvious response to initial forcing can lead to the incorrect conclusion that a component of the system is insensitive to the altered environment.
Many of the behaviors of parts of the earth system have clear threshold responses. Wildfires, for example, almost never occur until temperature, humidity, fuel load, and fuel moisture enter the permissive range. But when all the environmental conditions are compatible with sustaining a wildfire, risks increase rapidly. This wildfire threshold could have important implications for Amazon rainforests if the future is warmer and drier (Nepstad et al., 1999). It could also interact in an important way with anthropogenic burning given the recent evidence that aerosols from Amazon fires can decrease rainfall (Andreae et al., 2004). And in a clear example of a feedback, reduced rainfall over tropical land masses during El Niño events has been shown to encourage more biomass burn-
ing in the tropics, which in turn yields higher annual concentrations of carbon dioxide (CO2) in the atmosphere; knowing this could lead resource managers to greater policing in El Niño years in an attempt to reduce the extent of burning by agriculturalists and reduce carbon emission.
Other important examples of threshold come from the response of temperate forest ecosystems to warming or the deposition of atmospheric nitrogen. In controlled ecosystem experiments nitrogen inputs produce little change over several years, but the nitrogen excess eventually reaches a point where the system collapses.1 In response to warming the initial response is a large increase in soil warming, followed by a sudden decline when the ecosystem runs out of easily decomposable material.2
Some of the important thresholds in earth system responses can operate in more than one direction. One good example of this is the relationship of atmospheric ozone to levels of volatile organic compounds (VOCs) in the atmosphere. Depending on the ratio of VOCs to nitrogen oxides (NOx), an increase in VOCs could lead to a large decrease or a large increase in ozone production.3 Changes in land use can behave as thresholds.4 Often, dramatic changes in land use follow changes in policy, price supports, or transportation infrastructure. If consequent changes in local climate make the changes in land use difficult to reverse (Dickinson and Henderson-Sellers, 1988; Lean and Warrilow, 1989; Shukla et al., 1990), the changes that occur across a narrow threshold can be locked in place.
Interactions between the earth, oceans, and atmosphere often involve the simultaneous action of diverse mechanisms. Terrestrial and ocean carbon balance provide beautiful examples of the overall fluxes controlled by a large number of individual mechanisms. In the oceans, temperature interacts with alkalinity, salinity, and dissolved inorganic carbon to control CO2 solubility (Sabine et al., 2004). Biological processes are also important contributors to the carbon balance of the oceans, with potentially subtle changes in the composition of the producer and consumer communities leading to substantial effects on the downward transport of particulate carbon (Sabine et al., 2004). On longer timescales the delivery of mineral nutrients from upwelling or from the delivery of windborne dust plays an important role.
On land, diverse processes contribute to the overall carbon balance (Pacala et al., 2001). The current carbon balance of the United States has large influences due to land use change, CO2 fertilization, nitrogen deposition, ozone, and climate.5 The early optimism that future terrestrial carbon dynamics might be modeled as a simple response to atmospheric CO2 (Bacastow and Keeling, 1973) has been replaced by an appreciation that drivers from human actions,
atmospheric composition, climate, and ecological processes all interact, with contrasting relative importances in different settings.
Climate change is expected to influence the capacities of the land and oceans to act as repositories for anthropogenic carbon dioxide and in turn provide a feedback that affects climate further. Modeling experiments show that carbon sink strengths vary with the rate of fossil fuel emissions, so carbon storage capacities of both land and oceans decrease and climate warming increases with faster emissions (Fung et al., 2005).
The bidirectional nature of earth-atmosphere interactions has important implications for a wide range of earth system processes. Coupling plays a central role in both carbon-climate and climate-albedo feedbacks. Many experiments and simulations indicate that, depending on the starting point, a warmer climate can lead to either a loss or gain of ecosystem carbon6,7 (Mack et al., 2004; Shaver, et al., 2006). If in response to warming, ecosystems lose carbon, then atmospheric carbon increases, producing a positive feedback on the initial warming. If warming leads to an increase in ecosystem carbon (with more carbon in plants and soils), then the feedback is negative. A number of model experiments now explore the role of land and ocean feedbacks in modulating the climate forcing from atmospheric CO2. The general conclusion from these studies is that the terrestrial feedback is positive (in the direction of exaggerating warming)8 (Cox et al., 2000; Friedlingstein, 2004; Fung et al., 2005), although the magnitude of the feedback is uncertain. The CO2 sensitivity of the climate model used in these simulations plays an important role in determining the strength of the feedback, as does the tendency of the ocean to take up the carbon released from the land. The magnitude of the current uncertainty is large. With comparable forcing from anthropogenic CO2 emissions, equally credible models end the 21st century with atmospheric CO2 levels differing by more than 200 ppm, a quantity of CO2 greater than the total released by fossil fuel combustion to date.
Albedo-climate feedbacks may be equally important. New evidence indicates that warming in the Arctic is already leading to increased abundances of shrubs, which lead to an increase in the absorption of solar radiation, especially in the spring, and reinforce the warming (Chapin et al., 2005). Simulation results indicate that historical land use in the central United States has cooled the climate. The lack of historical cooling reflects the combined effects of this albedo effect, plus other processes that have counteracted it. Combining effects on albedo and carbon storage, increasing forest biomass in the temperate latitudes tends to produce a net warming, while reforestation or afforestation in the tropics tends to produce a net cooling (Gibbard et al., 2005).
Humans exert and respond to a wide range of stresses in the coupled earth-ocean-atmosphere system. Almost all studies of natural science components of global change considered human drivers as a fixed set of boundary conditions, and analyses of human responses viewed changes in climate or air quality as givens. While these are clearly simplifications, research teams simply did not have the breadth of expertise or the technical tools to tackle truly integrated approaches. A few teams have recently made bold attempts to integrate human actions and the natural sciences in an interactive framework. For example, the scenarios developed for the Millennium Ecosystem Assessment use a modeling framework that attempts to integrate changes in agricultural demand with changes in climate, leading to, among other things, projections of deforestation and prices of major agricultural crops (MEA, 2005). Recent Massachusetts Institute of Technology Emissions Prediction and Policy Analysis simulations address interactions among climate, ozone, crops, and the economy in a coupled framework.9 They have also looked at air pollution, human health, and the economy as a coupled system. Consistent with the early stage of this research, many of the potentially most important drivers of change in patterns of human action have not been explored with coupled models. Specifically, the impacts of HIV/AIDS and other major epidemics could have major impacts on future human activity.10 The fundamentally important distribution of wealth, opportunity, and independent decision making11 was a focus of the Millennium Ecosystem Assessment,12 but its exploration with coupled models is just beginning.
Clearly, atmosphere-ecosystem interactions unfold through diverse processes, across a range of scales, and with nonlinearities. We have some understanding of a variety of the mechanisms involved, but there are many uncertainties. Much uncertainty relates to the impacts of global change on humans, ecosystems, and economies; interactive effects among these sectors have the potential to amplify or suppress the initial effects, sometimes by a large multiplier. As with the drought case study, this is a scenario where varied impacts can accumulate and expand in scope, extent, and intensity. From one impact there can be cascading impacts.
IMPACTS ON HUMANS, ECOSYSTEMS, AND ECONOMIES
Atmosphere-ecosystem interactions unfold through diverse processes. Climate, air pollution, droughts, and fires are all sensitive to controlling mechanisms that have atmospheric components, ecosystem components, and components that arise specifically from the interactions between them. Though most of these
mechanisms are understood in outline form, many of the details are unknown. Much of the reason that the range of uncertainty related to impacts of global changes on humans, ecosystems, and the economy is so large is that the interactive effects have the potential to amplify or suppress the initial effects, sometimes by a large multiplier.
Several kinds of human factors can exaggerate vulnerability to the impacts modulated by atmosphere-ecosystem interactions. Poverty, lack of control over one’s destiny, and an extremely unequal distribution of wealth all tend to decrease coping capacity, increase vulnerability, degrade ecosystem services, and increase the challenge of finding effective paths toward solutions.13 In contrast, human factors that stimulate technical innovation, distribute control, and encourage local decision making can decrease vulnerability while increasing ecosystem services.14
Atmosphere-ecosystem interactions introduce potentially important uncertainties into a large suite of future global changes. Characterizing these uncertainties and, where possible, reducing them, is one of the central challenges of global change research. Still, it is important to recognize that unknowns in the realm of human actions increase the uncertainties even further.15 For a truly useful understanding of the range of global change processes, we need to develop useful ways to more effectively integrate earth, atmospheric, and human processes.16
POLICY OPTIONS: ADAPTATION AND MITIGATION
Atmosphere-ecosystem interactions have important impacts not because they result in new phenomena but because they modulate a wide range of earth and atmospheric processes. Especially in a context with multiple, simultaneous interaction drivers, this modulation can be of primary importance. Atmosphere-ecosystem interactions have the potential to amplify the impacts of minor processes or suppress the impacts of major ones. They are the dominant sources of uncertainty in some processes and a major source of uncertainty in others. Increased understanding of these interactions is a major theme in global change research.
In many settings increased understanding of atmosphere-ecosystem interaction can play a central role in designing effective strategies for adapting to or mitigating impacts of global change. A central need is a thorough enough understanding of these interactions to map the locations of thresholds, especially those that cause positive feedbacks in global change responses. Examples of threshold responses in these interactions are increasingly well developed. For instance, it
is well known that warming that leads to increases in the abundance of Arctic shrubs, which when the shrubs become common enough, decreases local albedo and amplifies warming (Chapin et al., 2005). Another example is the effect of nitrogen fertilization. While increasing deposition of reactive nitrogen (typically as NO3 and NH4+) can lead to increased uptake of CO2, nitrogen fertilization typically also results in emissions of nitrogen gasses (e.g., NH3 and N2O, an even more powerful greenhouse gas than CO2) (e.g., Vitousek et al., 1997). Also, nitrogen fertilization leading to nitrogen saturation of terrestrial ecosystems results in the loss of nutrient cations, causing reduced productivity locally and eutrophication of aquatic systems downstream (Vitousek et al., 1997). Interactions can also be multistep. For example, the effect of biogenic volatile organic carbon on tropospheric ozone can lead to an indirect impact of high temperatures on crop yields and forest growth.17 The negative effects of elevated ozone are, somewhat surprisingly, not suppressed by growing trees in elevated CO2 (Karnosky et al., 1999).
In some cases a thorough understanding of atmosphere-ecosystem interactions can provide an insurance policy against adaptations that fail to accomplish their goals or that have undesirable side effects. For example, recent evidence shows that especially in midlatitude forests the warming caused by decreased albedo can be larger than cooling from carbon storage, providing an important caveat in the motivation for broad reforestation efforts in the midlatitudes (Gibbard et al., 2005; Feddema et al., 2005).
In other cases an atmosphere-ecosystem interaction can serve as an effective foundation for successful mitigations even if they are unintentional. Increases in plant growth and ecosystem carbon in response to elevated atmospheric CO2 (Prentice et al., 2001) provide a classical example of negative feedback on atmospheric carbon (see Figure 3-1). Another example concerns the possibility that an ocean acidified enough, in response to high atmospheric CO2, to start dissolving carbonate may dramatically increase its rate of CO2 uptake (Sabine et al., 2004).
In sum, atmosphere-ecosystem interactions do not establish a single set of issues for adaptation and mitigation. Instead, they appear almost as a large suite of risks and opportunities. Positive feedbacks have the potential to increase vulnerability, especially when responses cross thresholds. Negative responses have the potential to amplify the utility of adaptation and mitigation measures. In general, thorough understanding is critical, as the nature, direction, and magnitude of likely feedbacks are rarely clear.
The literature is increasingly rich with examples of important atmosphere-ecosystem interactions, but few, if any, are thoroughly understood. The community with the expertise to address questions in atmosphere-ecosystem interactions is small. Investigators in this area need to combine a research-level understanding of atmospheric processes with a sophisticated knowledge of terrestrial and marine ecosystems. For investigators not equipped to tackle the coupled system, collaborations are an essential tool, though interdisciplinary collaborations are often difficult and complex. Because some of the interactions unfold only on long timescales or large spatial scales, we need experimental, observational, and simulation techniques to explore the range of possibilities. This kind of work will require multifactor experiments.
From the workshop discussions participants identified the following interactions that could benefit from increased multifactor research. Participants noted that for most of these questions, key elements of a comprehensive understanding are in place, but resources for thorough study have been lacking. These items
are not prioritized or necessarily similar in scope but rather reflect the workshop participants’ on-site thinking.
How do CO2 and a warming climate interact to affect soil moisture, ocean acidification, and the carbon balance of ecosystems (both terrestrial and marine)?
What is the role of biogenic VOCs in generating ozone, and what is the role of ozone in degrading crop yields?
What is the role of changes in surface albedo in changing climate? What are the carbon-cycle implications of this?
What is the role of air pollutants in degrading crop yields? How do these effects change with warming?
How might human pressures for increased food production lead to an expansion of agricultural land, and what are the costs in ecosystem services for the converted land?
How do human decisions about cropping, land use or cover change, and urbanization influence atmosphere-ecosystem interactions?
How do greenhouse gas increases and associated warming, land use, and air pollution interact with biodiversity?
What is the relative role of extreme events and average conditions in establishing the impacts of atmosphere-ecosystem interactions? Under what conditions do atmosphere-ecosystem interactions enhance the value of investments in adaptation and mitigation?
How do drivers relate with stresses to produce certain vulnerabilities/ adaptive capacities?
How do socioeconomic, institutional, and environmental processes influence environmental change and adaptive capacity (livelihoods, migration)?
How does information get to those who need it? What kinds of information are most useful to decision makers, resource managers, and others who could benefit?
How can we develop improved tools and strategies for addressing multiple environmental stresses, such as improved observational and modeling capabilities, integrated sensors, regional information systems, and predictive capabilities?
How can societal resilience to multiple environmental stresses be improved? How can adaptive management approaches be developed and implemented?