The structure and function of the terrestrial biosphere are closely coupled to the atmosphere through multiple interactions involving physical changes to the land surface (biogeophysical feedback) and changes in the radiatively active gas composition of the atmosphere (biogeochemical feedback). Human activities are forcing large and pervasive changes in these interactions, making it more important than ever to understand the "natural** regulation of the atmosphere-terrestrial, biosphere-ocean system as it acted in the past as well as the consequences of human perturbations for the state and stability of the system in the future. Karth System Models are being developed for these purposes, although no model includes the full range of interactions that are known to occur. Study of paleoenvironmental records has yielded insights and information thai can be used as a test of Earth System Models and as a stimulus to their improvement. For example, biogeophysical (vegetation-atmosphere) interactions as well as ocean-atmosphere interactions seem to be necessary for the correct simulation of major changes in vegetation and climate in response to orbital forcing during the Holocene- There is also considerable scope, as yet unexplored, to use paleorecords to test models with coupled climatic and hiogeochemieal components. The I-ast Glacial Maximum, in particular, poses fundamental challenges concerning the causation and radiative effects of low CO? and high eolian dust concentrations, and the biological effects and consequent feedbacks engendered b> low CO>. Fully prognostic models of the terrestrial biosphere (dynamical global vegetation models, or IXjVMs) have been developed, partly in response to these challenges posed by paleorecords. DGYMs should form part of the design of Earth System Model experiments to test our ability to simulate changes in the system over annual to millenial time scales. Early applications of IXjVMs to future climate scenarios predict a decline in the capacity of the terrestrial biosphere to take up the excess CO? released by human activities during the coming century. I his predicted decline is a consequence of saturation of the biochemical response to CO^ combined with enhanced release of soil carbon in a warmer world. Continued deforestation is also likely to reduce this capacity. These considerations suggest that it is important for CO? "stabilization" calculations to explicitly include the effects of changing climate and land use on terrestrial ecosystem carbon metabolism; however, there are still considerable differences among the results of DGYMs, and these differences must be resolved with independent data. Overall, our understanding of whole Ivarth System function is tantalizing!) incomplete and is characterized by an array of interactions among different subsystems that include positive as well as negative feedbacks. Some particular aspects, however, are now relatively well understood, thanks to a combination of explicit numerical modeling and the systematic use of observational constraints in the contemporary and paleo domains.

11.1 Processes

Terrestrial ecosystems and climate interact through numerous biogeochemical and biogeophysical processes (Melillo et al., 1996). The term "interact" here means a two-way exchange, always including an impact of atmospheric conditions on the structure and function of ecosystems and a feedback of the structure and function of ecosystems on the atmosphere, Biogeochemical feedbacks influence climate by influencing the composition of the atmosphere, especially concentrations of greenhouse gases such as carbon dioxide (CO>) and methane (QI4). Biogeophysical feedbacks influence climate more directly, by altering the exchanges of energy, momentum, and water vapor betw een atmosphere and land - exchanges that are important in controlling the atmospheric circulation. This short overview begins by listing some of the atmosphere-biosphere interactions that are generally considered to be of potential importance, although the list is by no means exhaustive and many interactions arc still poorly quantified.

• Exchanges of CO? and water vapor between the soil-vegetation system and the atmosphere are strongly coupled through stomatal behavior, and they are sensitive to variations in atmospheric conditions, including atmospheric CO2 concentration itself. These exchanges are large enough that relatively small fluctuations in the annual balance of upward and downward fluxes can (a) have major consequences for the surface energy and water balance (Sellers et al., 19%, 1997), and (b) influence atmospheric CO2 concentration measurably (Heimann, 1997). The particular effect of CO2 concentration on the efficiency of photosynthesis results in a negative feedback, through which terrestrial ecosystems take up more carbon during periods of rapidly rising atmospheric CO? (as today) and would be expected to release carbon during periods of declining CX)> (Taylor and Lloyd, 1992; Thompson et al., 1996). Interannual variability in climate also causes substantial year-to-year fluctuations in the annual rate of increase of CO? in the atmosphere. These fluctuations could in principle be explained by variations in ocean uptake or terrestrial uptake of COj. The relative magnitude of these contributions is not well established, but modeling and analytical studies indicate that a substantial fraction, at least, is attributable to variations in the carbon balance of the terrestrial biosphere (Bacastow, 1976; kindermann et al., 1996; Heimann, 1997; Dettinger and Ghil, 1998). It therefore seems likely that the partitioning of carbon between the atmosphere and terrestrial ecosystems is also sensitive to climatic effects on ecosystem function on longer time scales.

• Methane (CH4) production in wetland soils constitutes a component of heterotrophic respiration (HR) that occurs under waterlogged and therefore anaerobic conditions. Although a relatively small component of HR 2-5% in boreal w etlands (Christensen et al., 1996), this release of CH4 is important from an Earth System point of view because CH4 is a far more potent greenhouse gas than the

C02 that would otherwise be emitted under aerobic conditions (Schimel et al., 1996). Over periods of a year or longer, HR approximately balances net primary production (NPP), which is the integral over time of photosynthesis (CO2 fixation by plants) minus autotrophic respiration (GO? release by plants). The annual imbalance of ±2 Pg C yr ' alluded to above is small compared with a global annual NPP that Pg C yr K Thus, atmospheric factors (GOg concentration and climate) that affect NPP, and all factors affecting the wetness of soils, can in principle affect the strength of the terrestrial CH4 source and therefore the additional radiative forcing due to CH4 (e,g., Hutchin et al., 1995; Walter et al,, 1996; Potter, 1997),

• Some other carbon-containing reactive gases, originating mainly in terrestrial ecosystems, interact indirectly with CH4 by affecting the oxidative capacity of the atmosphere, that is, the ability of HO, (HO2 and OH) radicals in the atmosphere to destroy a wide variety of gases, including CH4, These key reactive gases include nonmcthanc hydrocarbons (NMHC) and carbon monoxide (CO), both gases whose natural sources are closely tied to ecosystem function (e.g , Jacob and Wofsy, 1988; Lerdau et al, 1997). NMHC is a side product of photosynthesis. CO is generated in fires, which are an integral component of the natural function of many ecosystem types, including savannas, grasslands, and coniferous forests (Bergamaschi et al., 1998)+

• Cycling of carbon in ecosystems is naturally tied to the rate of nitrogen cycling, which in turn is dependent on atmospheric and soil conditions (e.g., Schimel et al., 1994). The rate of nitrogen cycling determines the emission rate of the biogenic trace gases nitrous oxide (N2O) and NO, (a shorthand for NO, NO), and other reactive compounds of N and O) from soils, and soil wetness determines which of these products is dominant (Firestone and Dav idson, 1989; Khalil and Rasmussen, 1992; Bouwman et aL, 1995; Davidson, 1995; Nevison et al., 1996). NO* is also produced in biomass burning. NjO is a potent and long-liv ed greenhouse gas (Schimel and Sulzman, 1995), whereas NO, affects the oxidative capacity of the atmosphere by facilitating photochemical reactions involv ing CO and NMHC (e.g., Derwcnt, 1996).

• Because different types of plants, animals, and microorganisms are adapted to different climates, any climate change sustained over decades to centuries must bring about changes in the biological composition and structure of ecosystems (Woodward, 1987; Prentice, 1992; Webb and Bartlein, 1992). Changes in ecosystem structure (for example, between forest and nonforest ecosystems, or between evergreen and deciduous vegetation) influence physical properties of the land surface, most importantly albedo, surface roughness, canopy conductance, and rooting depth (Dickinson, 1992; Henderson-Sellers et al., 1993), By influencing net radiation and heat flux partitioning at the surface, these properties exert a strong control over the surface climate. In particular, warming at high northern latitudes shifts the forest limit poleward, reducing surface albedo (especially in late winter and spring before ihe snow melts) and thus generating a positive climate feedback of hemispheric extent (Bonan et aL, 1992; Foley et aL, 1994).

All these processes are increasingly being interfered with by human actions, primarily fossil fuel burning and land use changes. It is important to understand the "natural" operation of these processes if we are to do a better job of understanding what may happen as a consequence of human actions today. Changes in the Earth System, documented in palcorecords of climate, vegetation, and atmospheric composition, provide a source of information about the natural operation of Earth System processes. We need to understand how the Earth System behaves in steady state and how it functions during periods of rapid climate change as documented in the paleorecord, in order to have confidence in our ability to project what will happen as human activities drive the system ever further from steady state (Broecker, 1997).

The following is a noncxhaustive summary of the changes being wrought by human activities on the atmosphere-biosphere processes discussed in the preceding section.

• The cycling of {Ah between the atmosphere and the biosphere is being altered by deforestation, which returns additional CO2 to the atmosphere (Houghton et al., 1987; Houghton, 1991; Schimcl et aL, 1995). On the other hand, forest regrowth on abandoned agricultural land and forest management changes, especially in the North Temperate Zone, may be contributing to the removal of CO? from the atmosphere (Kauppi et aL, 1992; Mcltllo et aL, 1995; Schimcl et aL, 1995), Fossil fuel emissions are still the major contributor to the rising atmospheric CO? content. The rising CO? itself is believed to be inducing a sink of CO? in terrestrial ecosystems because of disequilibrium between NPP and HR (GifTord, 1993; Lloyd and Farquhar, 1996; Tian et aL, 1998; Kicklighter et aL, in press). Other factors such as land use changes may also be contributing to the current sink. Analysis of recent CO? and 0?:%2 ratio measurements suggests that extant ecosystems in the Tropics have been taking up about enough carbon to balance that released by deforestation, and an additional terrestrial carbon sink exists in the mid- to high latitudes of the Northern Hemisphere (Battle et aL, 1996; Keeling et aL, 1996; Heimann, 1997). Although the locations and magnitudes of terrestrial COj sinks are difficult to estimate precisely and are controversial (e.g., compare Fan et al., 1998, with Rayner et al., in press), there is clearly a potential for both climate change and land use change to alter the terrestrial biosphere's capacity to act as a sink for CO? released by human activities.

• While additional CH4 is released to the atmosphere from ruminant herds, landfills, and natural gas leakage, natural CH4 sources are being diminished because of draining of wetlands for forestry or agriculture (Schimcl et aL, 1996). Atmospheric CH4 now stands at an extremely high level relative to measurements for any preindustrial time and contributes almost one-fifth of the total greenhouse gas forcing; on the other hand, the growth rate of CH4 in the atmosphere has slowed in recent vears.

• Human activities strongly alter fire frequencies through deliberate burning or, in other regions, deliberate fire prevention. This has implications first of all for the CO2 balance but also for the atmospheric budget of CO, NOx, and other trace constituents that are released during fires, The net effect of current human actions with regard to fire is diffiult to estimate because tire affects virtually every aspect of biosphere-atmosphere interaction and because it is difficult to classify individual fire events unambiguously as natural or human-caused.

* The tie between carbon and nitrogen cycling is partially broken (a) because of the widespread use of nitrogen fertilizer (required to produce adequate protein to feed the human population), leading to release of both N^O and NOt and (b) because of NO* emissions associated with fossil fuel combustion. Nitrogen deposition from the atmosphere (Holland et al., 1997) is believed to be a major cause of increased tree growth rates downstream from industrialized areas. This effect may be allowing additional carbon storage by these ecosystems and possibly contributing significantly to the terrestrial carbon sink (e.g., Townsend et al., 1994; MeliUo et al., 1995), However, long-term effects of this nitrogen addition include damage symptoms, and the most affected ecosystems appear to be already N-saturated (Schulzc et al., 1989). At present, far too little is known about the amount, type, and pattern of atmospheric nitrogen deposition, and about the ecosystem impacts.

* Deforestation is drastically modifying ecosystem structure in some regions, with potentially important consequences for the atmospheric circulation and regional climate (Henderson-Sellers et al., 1993). Because of the spatial complexity of human-caused deforestation, however, the feedbacks to climate mav be very differ-


ent from the simple ones predicted in the first sensitivity experiments on this topic.

* Agriculture and construction are contributing additional dust to the atmosphere, leading to a net negative radiative forcing that may be comparable in magnitude (though not in spatial pattern) to the effect of industrially generated sulphate aerosol (Tegen and Fung, 1995).

Current understanding of these impacts relics on a combination of geophysical observations (including remotely sensed measurements) and modeling. The need for numerical models, mentioned earlier in the context of understanding the "natural" operation of the gcobiosphere, applies a fortiori to understanding the consequences of massive human involvement.

11,3 Models

Progress is being made toward the goal of building Earth System Models that include the main physical, chemical, and biological processes that interact in determining the slate and change of the atmosphere, terrestrial biosphere, and oceans. Such modeling builds on the existence of three-dimensional physical models of the atmospheric and oceanic circulation, and on the existence of "fuzzy" laws governing the aggregate behavior of organisms (Prentice, 1998}.

Earth System Models could be used as an aid to understanding the patterns seen in records of past environments and recent historical observations (including atmospheric gas measurements, remote-sensing measurements of photosynthetic activity, etc.). Today we are far from having models that include all the processes listed in Section ILL

Howev er, we do have a basic toolkit of models that describe the most basie ecosvstem

MaTepi/tan, 3aLHwmeHHbifr aBTopcKUM npaBOM

functions of CQa, water, and energy exchange by terrestrial ecosystems and illustrate how these influence and are influenced by ecosystem structure and climate.

The recent evolution of global terrestrial biosphere models has taken place in two overlapping phases. In the first phase (about 1985-1995), two types of model evolved in parallel:

• Biogcography models (e.g., BIOME of Prentice et al, 1992) invoke plant-physiological controls on the distribution of plant forms as a way to predict the large-scale geographic distributions of biomes (tropical forests, savannas, boreal forests, deserts, grasslands, etc.).

• Biogeochemistry models (TEM of Melillo et aL, 1993; Century of Parton et al., 1995; CARAIB of Warnant et al,, 1994), and tens of others, invoke the mechanisms of photosynthesis, respiration, évapotranspiration, carbon and nitrogen allocation, and so on to predict the CO2, water, and nitrogen metabolism of ecosystems.

The original biogeography models did not simulate ecosystem function, whereas biogeochemistry models did not predict where different types of ecosystem are found. Biogeochemistry models therefore have required a prescribed distribution of ecosystem types (VEMAP, 1995).

The second phase (about 1990 to present) has been characterized by a more comprehensive approach and has led to the dev elopment of the following types of models.

• Coupled biogeography-biogeochemistry models (e.g., BIOME3 of Haxeltine and Prentice, 1996; MAPSS of Neilson and Marks, 1994) predict the structure and function of ecosystems assumed to be in equilibrium with a given climate and soil regime, based on comparing the performance of different plant functional types in each environment.

• Dynamic global vegetation models (DGVMs, e.g., IBIS of Foley et aL, 1996; LPJ of S. Sitch, B. Smith, and I. C Prentice, unpublished) mimic the same processes but include the time-varying component by which ecosystem structure and function respond to changes in climate.

• At least one coupled DGVM-atmosphere model (IBIS-Genesis of Foley et al., 1998) fully incorporates ecosystem dynamics in the framework of an atmospheric general circulation model.

• Coupled DG YM-atmosphere-ocean dynamics models of intermediate complexity (Earth System Models of Intermediate Complexity, or EMICs, such as CLIMBER of Ganopolski et al., 1998a; Ganopolski et al,, 1998b) incorporate simplified ecosystem dynamics at low spatial resolution w ith parameterized atmospheric and ocean models. This design allows long simulations, including the physical couplings among all three components.

This second phase of model development has also seen the first attempts to include the source strengths of various biogenic trace gases (nitrogen oxides, CR», NMHC, etc.) in the outputs of terrestrial biosphere models (e.g., Nevison et al., 1996; Potter, 1997). Progress is also being made toward the inclusion of biogeochemical processes into coupled I)GVM-atmosphere models and EMICs, variations arc paced by the orbital variations, there is a phase lag; the LGM ice sheets were shaped by antecedent orbital conditions and were strongly out of equilibrium with the contemporary orbit.

• Concentrations of greenhouse gases COj, CI L*, and N?G all at historic lows: CO2 less than 20(1 ppm compared with 280 ppm in the late preindustrial Holocene (Barnola et al., 1987), CH4 at about 40(1 ppb compared with 750 ppb (Blunier ct aE, 1995), N?0 at about 190 ppb compared with 270 ppb (Leuenberger and Siegenthalcr, 1992).

• Like 6000 yr BP, the subject of an internationally standardized comparison of atmospheric models within PMIP (Pinot et aL, in press),

• Evidence for a biosphere far more radically changed even than during the mid-Holocene; with greatly reduced or displaced forests in all latitudes, extensive tundra and steppes in mid-latitudes, and expanded deserts (Wright et aL, 1993; Prentice et aL, in press).

• A negative anomaly of &UC. in the deep ocean (Duplessy et aE, 1991), suggesting that total biosphere carbon content was on the order of 25% less than in the Holocene (Prentice and Sarnthein, 1993; Bird et aE, 1994).

• Ice core, marine sediments, and terrestrial loess indicating many times greater atmospheric dust concentrations and fluxes than in the Holocene, especially at high latitudes (Petit et aE, 1990; Steffensen, 1997; Mahowald et aL, in press).

• Marine and terrestrial paleodata from the Tropics, remote from the ice sheet, showing widespread cooling. The tropical average was probably 2-3 C less than present at modern sea level, but colder still in certain regions (Sonzogni et aL, 1997; Farrera et aL, in press). Terrestrial vegetation and snowline records from high elevations indicate a stronger cooling aloft - that is, a steeper than present lapse rate (Farrera et aL, in press). There was much stronger cooling in the northern continents and high-latitude oceans, especially in winter (e.g., Webb et aL, 1993; Weinelt et aL, 1996; Peyron et aL, 1998),

• An extraordinary level of "sub-Milankovitch" climate variability, on century

* » r w to millenial time scales, expressed in the form of Dansgaard-Oeschger events,

Heinrich events, and Bond cycles (Bond ct aL, 1992; Hammer et aL, 1995; Broecker, 1997).

We do not yet know with any precision how the very large observed changes in the land surface might have contributed to determining the LGM climate through biogeophysical mechanisms of the kind discussed for 6000 yr BP, although a sensitiv ity experiment by Crowley and Baum (1997) suggested a significant and rather complex effect. Some attention, however, has been paid to the causes and effects of the high atmospheric dust loading at LGM. Possible causes include enhanced winds (Rea, 1994), reduced removal of dust in precipitation (Yung er aL, 1996), and increased unvegetated source areas (Broecker and Henderson, 1998), The first two mechanisms are insufficient to fully simulate the enhanced dust loading at high latitudes (e.g., Joussaume, 1990; Andersen et aL, 1998). However, the observed dust distribution and loading can be approached in an atmospheric transport model when all three mechanisms are allowed and "future" climate were obtained from a transient simulation of the coupled ocean-atmosphere general circulation model of t he Hadley Centre, United Kingdom (Mitchell et al.t 1995). This simulation assumes a continuing, steady exponential increase of atmospheric CO? concentration (or equivalent radiative forcing due to other greenhouse gases) and takes into account an associated scenario of changes in the global amount and distribution of sulphate aerosol. The DGVMs deliberately and artificially simulated a "natural" biosphere, dependent only on climate, soils, atmospheric CO?, and its own recent history. The CO^ scenario itself w as a pure assumption, and there was no feedback from the CO2 balance simulated by the DGVMs to the assumed time course of CO2 concentration in the atmosphere. At the end of the experiment, the climate was instantaneously stabilized - a device to explore how far the simulated ecosystem state might be from equilibrium with the new climate.

Studies of the recent (observational) period help to constrain the response of terrestrial biosphere models to climate, on the time scales of interannual and to some extent interdecadal variability. On longer time scales, other processes may come into play. For example, changes in vegetation structure brought about by climate change may increase the potential for long-term carbon storage (over centuries) while transiently adding carbon to the atmosphere because of diehack of extant vegetation in some regions (the "carbon pulse" hypothesis of King and Neilson, 1992, and Smith et al., 1992). The processes involved are all explicitly simulated by DGVMs. However, the simulated behavior is essentially unconstrained by observations at the whole-system level. It may be possible to constrain such behav ior by analysis of the paleorccord (especially by attempts to simulate the small-amplitude variability of CO? during the 1 Iolocenc, as shown by recent high-resolution ice core records from Antarctica: Inder-muehle et al., 1999). For the time being, the lack of constraint is an important caveat concerning DGVMs. Nevertheless, the DGVMs (which were generated from differing assumptions and perspectives) agreed in broad features. The carbon pulse hypothesis was not supported, but in all models the terrestrial sink for CO? (itself caused by the negative feedback mechanisms alluded to previously) began to level off or decline at around 50 years from present because of a combination of transient vegetation changes, the asymptotic nature of CO? fertilization, and the effect of warmer temperatures in stimulating HR (Cao and Woodward, 1998). The models differ in the magnitude and rate of the decline because of differences in the climate response ofNPP (Cramer et at., in press).

One lesson from this admittedly stylized experiment was to draw attention to the heavy reliance currently placed on the continuation of a carbon sink in nonagricultural ecosystems. The simulations indicate that this sink is potentially vulnerable to climate change. A moment's reflection also leads to the conclusion that at least the tropical component of the terrestrial carbon sink probably cannot persist if conversion of forest to agriculture persists at its present rate. Thus, climate effects and potential land use changes must be considered in future attempts to estimate theCO? emissions reductions required, for example, to meet the objectives of the Kyoto Protocol. On the other hand, DGVMs must be subjected to more-rigorous tests that hopefully w ill reduce the uncertainty due to differences among models.

11.7 Concluding Remarks

We are presented with a tantalizing pieiure of the Earth System. On the one hand, interdisciplinary research involving a combination of geophysical observations and modeling appears to be providing firm explanations for some aspects of Earth System behavior as shown in recent observational records and paleo data for selected time periods, On the other hand, our understanding of the events shown in the paleo-record is seriously incomplete. Terrestrial biosphere processes have been shown to be an important component of Earth System dynamics, but only the most basic processes are as yet fully integrated into Earth System Models. "Predictions'1 of future climate and atmospheric CO2 concentrations, made with various assumptions about future economic, technological, and environmental developments, continue to routinely ignore most of these processes. Thus, there is much to be done before some of the narratives pursued here can be translated into quantitative and testable statements about past global changes and into more-comprehensive tools for understanding the consequences of human activ ities.

Recognition of the significance of terrestrial biosphere in the climate system has further implications for future research priorities, and for our view of how the Earth System functions as a whole, if one generalization emerges from the bewildering variety of feedbacks that link the land to the atmosphere and ocean, it is the absence of any general rule favoring negative feedbacks that would stabilize the system. Such feedbacks exist, but there are positive, potentially destabilizing feedbacks in the ocean-atmosphere system (e.g., Stocker and Schrmttner, 1997) as well as in the terrestrial hiosphere-atmosphere system, Thus, no generalization can substitute for a thorough analysis of the processes, their incorporation into explicit numerical models, and the vigorous pursuit of observational constraints against which the models can be progressively honed.

0 0

Post a comment