Amazonia and the modern carbon cycle: lessons learned
In this paper, we review some critical issues regarding carbon cycling in Amazonia, as revealed by several studies conducted in the Large Scale Biosphere Atmosphere Experiment in Amazonia (LBA). We evaluate both the contribution of this magnificent biome for the global net primary productivity/net ecosystem exchange (NPP/NEE) and the feedbacks of climate change on the dynamics of Amazonia. In order to place Amazonia in a global perspective and make the carbon flux obtained through the LBA project comparable with global carbon budgets, we extrapolated NPP/NEE values found by LBA studies to the entire area of the Brazilian Amazon covered by rainforest. The carbon emissions due to land use changes for the tropical regions of the world produced values from 0.96 to 2.4 Pg C year−1, while atmospheric CO2 inversion models have recently indicated that tropical lands in the Americas could be exchanging a net 0.62±1.15 Pg C year−1 with the atmosphere. The difference calculated from these two methods would imply a local sink of approximately 1.6–1.7 Pg C year−1, or a source of 0.85 ton C ha−1 year−1. Using our crude extrapolation of LBA values for the Amazon forests (5 million km2) we estimate a range for the C flux in the region of −3.0 to 0.75 Pg C year−1. The exercise here does not account for environmental variability across the region, but it is an important driver for present and future studies linking local process (i.e. nutrient availability, photosynthetic capacity, and so forth) to global and regional dynamic approaches.
KeywordsAmazonia Carbon cycle Deforestation Climate changes
The sources and magnitudes of the major drivers of the global carbon cycle have changed since the beginning of last century, when the atmospheric carbon dioxide concentration started to rise consistently due to global emissions from fossil fuel and biomass burning (Schimel 1995). In the last few decades, a series of studies started to lay out the processes that could account for the fate of this additional carbon introduced into the atmosphere. However, the sum of fossil fuel and biomass burning emissions is higher than the balance of the carbon dioxide accumulated in the atmosphere and absorbed by the ocean and terrestrial biomes (Houghton et al. 1998). This budget mismatch between global sources and sinks, initially called “the missing carbon sink”, led to years of arguments about where the unaccounted sink could lie. Some hypotheses considered that the oceans could take more carbon dioxide than originally thought (Bousquet et al. 2000; Tans et al. 1990; Conway et al. 1994; Ciais et al. 1995). Others considered that the missing carbon was taken up by terrestrial ecosystems, via photosynthesis, but there were considerable difficulties in calculating precisely how much (Fan et al. 1998). Additionally, global modeling studies have also calculated an increase in the rate of atmospheric CO2 accumulation throughout the past century (Cox et al. 2000; Itoand Oikawa 2000; White et al. 2000).
The tropical rain forest biome plays an important role in the global carbon budget (Melillo et al. 1993; Dixon et al. 1994; Field et al. 1998), but there is great uncertainty about the ecosystem’s current and future impact on atmospheric CO2 accumulation (Tian et al. 2000; Schimel et al. 2001; Cox et al. 2000). There is no clear understanding of how the carbon balance of tropical forests responds to rapid, on-going changes in climate and atmospheric composition (Gash et al. 2004). In addition, more studies relating the role of nutrient limitation, such as nitrogen and phosphorus, to ecosystem carbon uptake (Vitousek 1984; Vitousek et al. 1986; Chadwick et al. 1999) are needed. Based on the records of atmospheric CO2 concentrations and carbon isotope ratios (13CO2), Townsend et al. (2002) suggested a sizable terrestrial carbon sink in tropical latitudes for a period from 1991 to 2000.
Climate forcing, such as changes in precipitation (Marengo et al. 2001) and radiation (Procópio et al. 2004) also play a key role on future carbon exchange scenarios. With continuing global warming, significant changes in water regimes could influence carbon budget in tropical forests (Victoria et al. 1998; Malhi and Wright 2004). A better understanding of the nonlinear and complex integrated effects of nutrients, climate, and hydrology on carbon cycling is needed to improve global process models being used to project future atmospheric CO2 levels and climate (Prentice et al. 2001; Cox et al. 2000).
Since the 1980s, there was a large increase in tropical deforestation followed by biomass burning. The carbon dioxide released from this burning has become critical for the global carbon balance equation. The role of the Amazon region, the largest continuous area of tropical forest in the world, came to the foreground in the global carbon quest. Several studies began estimating the area and rate of deforestation. The use of different remote sensing techniques to calculate the extent of the deforested areas produced heated controversy (Fearnside 1996), but considerably improved our understanding of the magnitude and rates of this process in this region (Fearnside 1997; Houghton et al. 2000). The average deforestation rate for the nineties was 17,000 km2·year−1, increasing to approximately 25,000 km2 in 2002 and 2003 (INPE 2004). By 2004, an estimated of 16.32% of the total Brazilian Amazon area (5.8 million km2) will be deforested (INPE 2004). Most of this process is concentrated in the southern and eastern part of Amazonia, while the central, less accessible areas, are relatively protected. However, recent government plans for paving roads and developing infrastructure (Laurance et al. 2001), associated with increasing presence of highly capitalized agribusiness companies, threat central areas in a very near future (Soares-Filho et al. 2004).
Our understanding of the probable fate of the “missing” carbon had progressed considerably towards the end of the 1990s, with evidence indicating that secondary forest regrowth in temperate forests was contributing to the terrestrial C sink. Also, extensive field studies, mainly in temperate terrestrial ecosystems, and large scale atmospheric surveys, associated with modeling efforts, indicated that the actual atmospheric carbon dioxide concentration, combined with excess nitrogen deposition, could be fertilizing ecosystems and increasing the carbon uptake through photosynthesis (Schimel 1995; Tans et al. 1998). However, the precise location and magnitude of these sinks remained unknown. This led to several studies focusing on the potential distribution of terrestrial sinks (Friedlingstein et al. 1995; Melillo et al. 1996; Thompson et al. 1996; McGuire et al. 1997). Due to their high biomass density, tropical forests were a natural candidate for a large sink component. Several modeling studies clearly supported this hypothesis (Friedlingstein et al. 1995; Melillo et al. 1996; Thompson et al.1996; McGuire et al.1997). Therefore, Amazonia was considered to be a potentially significant player in the global carbon cycle as either a sink or source, mainly due to the large amount of carbon stored in its biomass and large annual fluxes from photosynthesis, respiration, decomposition and fire.
The Brazilian scientific community realized that it was time to develop a large integrated program devoted to studying ecological functioning of the Amazon region, especially considering its role in the modern carbon cycle. This was promptly supported by American and European colleagues. In 1997, the Large Scale Biosphere–Atmosphere Experiment in Amazonia (LBA), a Brazilian-led, multinational study of the interactions between the Amazonian region and the global climate system, was officially launched. Two major questions drove LBA investigations: (1) how does Amazonia currently function as a regional entity? (2) how will changes in land use and climate affect the biological, chemical, and physical functions of Amazonia, including the sustainability of development in the region and the influence of Amazonia on Global climate? (http://lba.inpa.gov.br/lba)
The long-lasting ecological paradigm, that mature tropical forests should be at climax, i.e. over time the net gain of carbon should near zero, presumed the existence of a dynamic mass balance in old growth forests. However, the first conclusive direct measurements of ecosystem CO2 exchange conducted in Amazonia using eddy covariance technique (Grace et al. 1995a, b; 1996; Malhi et al. 1998), suggested a large net carbon uptake of 1–6 t C ha−1 year−1. Therefore, the presumption of mass balance in primary forests was under question. These first measurements led to the speculation of whether pristine Amazon forests could be functioning as a giant carbon sink. An unforeseen sink in the tropical forests of the world, and in Amazonia in particular, could resolve the geographical mystery of the missing carbon sink. However, many researchers were skeptical about the magnitude of the sink predicted by these early eddy flux data. As a result, an intense review of the eddy covariance method, including the applicability of scaling from the measuring site to the region, was initiated. This ongoing review has already pointed to difficulties in measuring CO2 emissions in the very stable atmosphere that develops in tropical forests (Kruijt et al. 2004; Baldocchi et al. 2000), and shown large local heterogeneity of ecosystems (Araujo et al. 2002; Saleska et al. 2003). In parallel, a large effort in compiling long-term biomass changes in forest stands across the Amazon and around the tropical world sought to ascertain the sink or source status of carbon in primary forest vegetation (Phillips et al. 1998). These efforts were not yet clarified beyond doubt the overall role of Amazonia as a sink or source.
The main objective of this paper is to provide a critical overview of facts and uncertainties, and raise potential implications of what we have learned so far regarding the role of Amazonia in the modern carbon cycle. We will start with a review of the extent of deforestation in Amazonia, including estimates of how much carbon was released into the atmosphere through land use changes. This will be followed by a review of methods used to estimate the net ecosystem exchange (NEE) in the region, and ecological implications of the estimates found thus far.
2 Amazonia: the modern carbon budget and deforestation
Emissions mainly from fossil fuel burning: +6.3±0.4 Pg C year−1
Emissions due to land-use change: +1.4–3.0 Pg C year−1
Atmospheric increase: +3.2±0.1 Pg C year−1
Ocean–atmosphere flux: −2.1±0.7 Pg C year−1
Land–atmosphere flux: −1.0±0.8 Pg C year−1
Residual terrestrial sink: −1.6 Pg to −4.8 Pg C year−1
As mentioned earlier, tropical forests are probably the only major biome that may play an important role in both sides of the global carbon balance budget shown above. For instance, the Amazon forest vegetation in Brazil alone contains about 70 Pg of carbon, which amounts to between 10% and 15% of global terrestrial biomass (Houghton et al. 2001). Overall, causes for deforestation in tropical areas are numerous and complex, most of them linked to economical and social processes (see Hecht 1993; Margulis 2003; Nepstad et al. 2002; Walker and Moran 2000; Andersen et al. 2002). Road construction in pristine areas, cattle ranching, logging and agriculture are the main economic activities leading to deforestation (Nepstad et al. 2001). Recently, Hirsch et al. (2004) proposed the carbon and land-use change (CARLUC) model and calculated a cumulative carbon release to the atmosphere due to deforestation of ~7 Pg C for the period from 1970 to 1998 in the Brazilian Amazon. This expansion of deforested areas reflected a large population increase in the region, from about 100,000 in 1840 to 1.5 million in 1940, up to about 18 million nowadays (Instituto Brasilieiro de Geografia e Estatística, IBGE; http://www.ibge.gov.br, Brazil), with more than 50% of people living in villages and cities (Browder and Godfrey 1997).
According to measurements based on satellites images (http://www.inpe.br) the deforested area in the Brazilian Amazon was approximately 600,000 km2 in August 2002. The Instituto Nacional de Pesquisas Espaciais (INPE) defines deforestation as “the conversion of primary forest areas by human activities for the development of agriculture and/or cattle ranching activities, as detected by orbiting satellites”. Another way to measure the deforested area is through ground-based land surveys done by IBGE Agricultural Census in Brazil. Andersen et al. (2002) presented an interesting comparison between these two methods. Estimations of deforested areas were generally larger based on surveys than when estimated by satellite data, probably because satellites detect gross deforestation and whereas surveys record the net deforestation, taking into account the re-growth of secondary forests (Andersen et al. 2002; Margulis 2003). Most of the deforestation occurs in primary forests, with grasses as the main replacing vegetation. Margulis (2003) suggested that planted pastures covered approximately 70% of the deforested area in Amazonia. Considering that fallow areas are used also for pasture, this estimate increase to almost 90%. It is also a consensus that most of the deforestation occurs along roads (Nepstad et al. 2001; Carvalho et al. 2001). Chomovitz and Thomas (2001) have calculated that 75% of all deforestation occurred within a 25 km-wide swath along the roads.
Estimates of deforestation provide only one component to assess carbon emissions. One also needs to know the standing biomass and the fraction of biomass released through burning immediately and also slowly over time. The difficulties and uncertainties of biomass estimates in tropical forests will be discussed in a later section of this article. Here, we will only summarize the available information in order to illustrate our lack of knowledge on quantifying this parameter, and how it affects estimates of carbon emissions.
Considering the rates of deforestation between 1988 and 1998, the total biomass of Brazilian Amazonian forest, and the amount of carbon stored in the re-growing vegetation, it was possible to estimate the annual emission of carbon to the atmosphere. The average flux for this period was 0.18±0.08 Pg C year−1 (Houghton et al. 2000). If one accounts for logging activities and natural fires, the range of emissions can be calculated as 0.10–0.40 Pg C year−1 (Houghton et al. 2000). This range would account for approximately 15–65% of the higher estimates of carbon emission for Tropical Latin America in the 1990s (0.62 Pg C year−1), and for 7–30% of the upper estimates of carbon emission for tropical areas globally (1.62 Pg C year−1) for the same period (DeFries et al. 2002). The global emissions due to land-use changes varies from 1.4 Pg C year−1 to 3.0 Pg C year−1, considering the last estimate of 2.18 Pg C year−1 made by Houghton (2003), the land-use change in Amazonia would be responsible for approximately 5–20% of global emissions. The Amazon region (~7.0 million km2) covers an area equivalent to 4.7% of the terrestrial area of the planet (150 million km2). Assuming that deforestation has already burned approximately 9% (0.6 million km2) of the region implies that an area equivalent to approximately 0.40% of the planet should be responsible for a significant share of the global carbon emissions due to land-use changes.
3 Indirect effects of Amazonian deforestation on atmosphere composition and land–atmosphere carbon exchange
3.1 Changes in atmospheric composition—radiation and precipitation
Deforestation affects the atmosphere in several ways. First, there is a change in the energy and water balance when forest is replaced by pasture and this change has the potential to alter the atmospheric water content and precipitation patterns (Silva Dias et al. 2002). Second, when forests are cut and burned, large amount of particles and reactive trace gases are released into the atmosphere. This can lead to profound changes in the atmospheric composition (Andreae et al. 2002).
Most global circulation models simulated a decrease in the precipitation over Amazonia as a result of deforestation (Marengo and Nobre 2001), but these models have a very coarse resolution, and local processes and interaction between the land and atmosphere due to heterogeneity of the landscape cannot be properly assessed. New high-resolution mesoscale models were used to investigate the impact of land use change in Amazonia (Avissar et al. 2002), and they also simulated a decrease in precipitation due to deforestation. However, one recent study suggested that deforestation in portions of the region can lead to locally increased precipitation (Baidya Roy and Avissar 2002). This raises the hypothesis that there is a threshold of extent and distribution deforestation that leads to a precipitation decline (Avissar et al. 2002).
In addition to changes in the overall amount of precipitation, models have demonstrated that the patterns of precipitation over Amazonia are changing due to deforestation. Under forested conditions, low concentrations of aerosol particles prevail in Amazonia and a maritime-type cloud regime is formed in the region (Williams et al. 2002). This phenomena inspired atmospheric scientists to nickname Amazonia as the ‘green ocean’ (Andreae et al. 2004). Due to biomass burning, the aerosol concentration increased abruptly in some regions of Amazonia (Artaxo et al. 2002).
It is becoming clear that the Amazonian ecosystem is very sensitive to global climate forcing agents such as El Niño, regional changes in precipitation patterns and global warming (Marengo and Nobre 2001). Fire risk is among those sensitive to climate, but fire is also affected by management (Nepstad et al. 2001). The potential for fire to spread from deforested areas into fragmented forests represents a threat to long-term ecosystem health and sustainability (Cochrane et al. 1999; Cochrane and Laurance 2002). Logging thins the canopy, increases sunlight penetration to the understory and facilitates burning of the forest floor. After an Amazonian forest stand burns once, the reduced canopy cover and increased penetration of solar radiation increases the probability that fire will be able to spread through the stand again (Cochrane et al. 1999). The interactions of climate, management and fire susceptibility produce several mechanisms of positive feedback that may lead to further forest degradation, fire risk, and negative economic and human health consequences (Nepstad et al. 2001).
3.2 Emissions of volatile organic carbon
Tropical forests release substantial quantities of biogenic volatile organic compounds (VOCs) to the atmosphere (Guenther et al. 1995). VOC release is important to the ecosystem carbon balance (Crutzen et al. 1999; Kesselmeier et al. 2002a). A full accounting of the influence of VOCs on ecosystem carbon budgets must consider both sources and sinks for these compounds and the reaction products of VOCs including atmospheric particulates. One of the key issues here is the large biodiversity of Amazonia, because VOC emissions (such as isoprene and α-pinene emissions) are specie-dependent (Guenther et al. 1995). Greenberg et al. (2004) and Harley et al. (2004) employed tethered balloon and leaf measurements approach to study the distribution of VOCs in Amazonia. Similar to the threefold variation that Malhi et al. (2004) reported for woody biomass production across the basin, these authors found a threefold variation in emission rates of VOCs across different regions, and they attributed this variation to diverse species composition. Also significant seasonal variability is found, with Kuhn et al. (2002, 2004) reporting up to a twofold difference in isoprene emissions between wet and dry seasons in Rondônia. Due to diversity of environments and species richness, we can expect to find similar variability in other regions in Amazonia.
Canopy emissions of VOCs amount to a few percent of net primary productivity (Guenther 2002; Kesselmeier et al. 2002a; Wiedinmyer et al. 2004), but their effects on atmospheric chemical processes and physical climate processes far exceed their modest contribution to the C cycle. Recently, Clayes et al. (2004) showed that isoprene emissions in Amazonia could make up a significant fraction of cloud condensation nuclei and thus have a role in precipitation patterns. Biogenic hydrocarbons are critically important in the regulation of concentrations of ozone, hydroxyl radical and other important trace gases and radicals in tropical atmospheric chemistry (Andreae et al. 2001). Terpenes are precursors of radiation forcing aerosol particles, which also act as cloud condensation nuclei, but terpene concentrations amount to only a few parts per billion in several sites in Amazonia (Kesselmeier et al. 2002b). Still, part of the precipitation formation mechanism could be controlled by the vegetation itself through isoprene and terpene emissions (Artaxo et al. 2002; Clayes et al. 2004). Hence, links between the biosphere and the atmosphere appear to be extremely complex, with climate affecting VOC emissions and vice versa in a highly non linear process (Kesselmeier et al. 2000, 2002a, b; Wiedinmyer et al. 2004).
4 Measurements of carbon exchange at the ecosystem level
4.1 Eddy covariance studies
Atmospheric CO2 fluxes showed seasonal variability, linked to basin scale spatial variability of climate and local scale biogeochemistry.
Surface energy partition and evapotranspiration are often tied to the CO2 flux variability.
Structural differences between the tropical forests and Cerrado (savanna) physiognomies, or pasturelands, were corroborated by measured fluxes.
Wetlands function substantially different from dry lands. Wetland contribution might be related to the carbon budget of the entire basin, and dependent on the flood-pulse variability.
Site-specific characteristics (e.g. geomorphology, past disturbances) constrain the measured CO2 flux variability.
The combination of annual accumulation of CO2 fluxes with biometric measurements can help infer the sink or source role of a plant community, regardless of its transient daily character.
Although eddy covariance technique has provided reliable measurements of bulk photosynthesis and respiration of temperate forest sites (Baldocchi et al. 1997; Jarvis et al. 1997), there are situations where its capacity to provide accurate integrated measurements of net carbon exchange is uncertain. The uncertainties stemmed from difficulties related to (1) night time fluxes, when respired CO2 tend to accumulate within the forest canopy and intermittent updraft events transfer it to the atmosphere in a complex spatial pattern not consistently caught by the flux tower sensors (Staebler and Fitzjarrald 2004), and (2) an irregular topography, where lateral CO2 advection to lower positions on the landscape, can result in CO2 draining out from the eddy covariance tower footprint (Araujo et al. 2002).
Since last decade, the continuous use of eddy covariance technique and microclimate measurements has amassed an impressive body of information on the CO2 and water vapor cycle within Amazonian boundaries (Fig. 3). Variability in seasonal CO2 flux of terra firme forests was noticed among sites of eastern (Caxiuanã, Tapajós), central (Cuieiras) and southern (Jarú) regions of the Amazon basin. At Tapajós and Caxiuanã, there was no evidence of drought stress during the dry season as evapotranspiration peaked concurrently with decreasing cloud cover and increasing surface net radiation (Carswell et al. 2002; Rocha et al. 2004). Minimum NEE peaked before the wet season (nearly October) at the eastern sites, where carbon was taken up in the dry season and released in the wet season, based on 3 years of observation (2000–2003) (Saleska et al. 2003; Carswell et al. 2002; Goulden et al. 2004)). This pattern was explained by an increase in soil respiration (and wood biomass) in the wet season, and a substantial reduction in the dry season, while gross ecosystem productivity (GEP) responded weakly on a seasonal basis. GEP was lowest in July, when the incoming solar radiation is also lowest. The net carbon uptake peak late in the dry season and the minimum GEP are partly explained by the flushing of new leaves and the innate seasonal rhythms of the vegetation (Goulden et al. 2004). Events like the reduction of precipitation due to ENSO warm episodes, can cause a reduction in net primary productivity (NPP) as evidences shown by Townsend et al. (2002), enhancing the seasonality on CO2 fluxes in Amazonia.
Among the sites previously mentioned, at Cuieiras the average NEE shows a weak seasonal variability, suggesting a lower average carbon uptake in the dry season (−17 μmol CO2 m−2 s−1) relative to the wet season (−21 μmol CO2 m−2 s−1) (Malhi et al. 1998; Araujo et al. 2002). As opposed to Tapajós, at Cuieiras, there appeared to be limited reduction of soil respiration in the dry season. This could be related to the shorter dry season length in the Manaus region (~ July–October) than in Tapajos (~July–December). Given that, observations at Cuieiras were taken partly over poorly drained valleys (contrasting to the observations over flatter and well-drained plateaus at Tapajos), where soils saturate more often (Chambers et al. 2004), a shortest dry season could prevent large moisture depletion (Davidson et al. 2000). This combined with small but significant increases in air and soil temperatures, could prevent a large soil respiration reduction. Possibly also related to different dry season lengths, the seasonal amplitude of NEE appeared larger in Tapajós (~700 kg C ha−1 month−1) than in Cuieiras (~150 kg C ha−1 month−1) (Saleska et al. 2003; Araujo et al. 2002).
At Jarú site, net carbon uptake is reduced continuously throughout the dry season (June–August), as opposed to Tapajós, and with larger seasonal amplitude than Cuieiras (Kruijt et al. 2004; von Randow et al. 2001). This pattern, observed at Jarú transitional tropical forest, is influenced by recurrent cooler temperatures (~13°C) in the winter, and responds similarly to semi-deciduous forests, where leaf senescence is pronounced in the dry season. Culf et al. (1996) reports increasing solar albedo during May–August at the Jarú site, concurrent with soil moisture depletion, which partly explains the changes in phenology and the CO2 fluxes. The authors observed a similar variation in albedo in the Manaus region (Reserva Ducke forest), although seasonality in CO2 fluxes was less evident in Cuieiras so far.
Over the Cerrado sites (savannas to the South), a net carbon uptake clearly occurs during the wet season, whereas strong positive NEE occurs in the late dry season. This pattern is markedly shown in the transitional mature forest (Cerradão) (11.5°S) (Vourlitis et al. 2001) and over a Cerrado restrito site in Brasilia (16°S) (Miranda et al. 1996). Vourlitis et al. (2004) stress the remarkable reduction in evapotranspiration during the dry season over Cerradão. Vourlitis et al. (2001) showed that the NEE of Cerradão was similar to that of tropical rainforest during the wet season (a net sink between −0.6 μmol and −1.2 μmol CO2 m−2 s−1, during February and April), but during the dry season it was more similar to that reported for tropical savanna, being in balance during August–September, and a net source of 0.6 μmol and 1.7 μmol CO2 m−2 s−1 during October–November. A comparable variability was observed over a Cerrado restrito site down in southeast Brazil (18°S) (Rocha et al. 2002). This is very suggestive that despite the latitudinal and climate differences among sites, adaptive mechanisms of these species can exert a large and common control on the carbon exchanges for the Cerrado biome.
Over dry forests, plant gross primary productivity (GPP) increases during the wet season, as seasonal drought results in a negative water balance in the trees and a reduction in stem growth. In seasonally flooded forests, however, tree growth occurs during the terrestrial (dry) phase (Dezzeo et al. 2003; Schongart et al. 2002), as flooding causes leaf shedding. It is predicted that GPP starts to increase at the end of the aquatic phase, when the leaves flush. Interannual variability also increases plant GPP, in these forests, as a reduction in precipitation by ENSO warm episodes in eastern and central Amazonia reduce the period of inundation (Schongart et al. 2004). Atmospheric NEE observations in the Bananal Island (characterized by a seasonally flooded ecotone Cerradão-Cerrado-campo) show a reduction in ecosystem nocturnal respiration during the aquatic phase (February–April) and consequently increasing ecosystem carbon uptake (H. Rocha in preparation).
Conversion of tropical forests to pasturelands in Amazonia reduces evapotranspiration, mostly in the dry season (Wright et al. 1992), and the moisture status in the root zone (Hodnett et al. 1995). The control of carbon exchanges become variable and largely dependent of the crop physiology and management. Sakai et al. (2004) measured the turbulent CO2 fluxes over agricultural fields in Santarém, in a succession of pasture, plowed bare soil and a rice plantation. They observed a larger average carbon uptake over the rice plantation (non-irrigated) (−7.0±0.5 μmol CO2 m−2 s−1) compared to the pasture (−2.3±0.5 μmol CO2 m−2 s−1 in the wet season and −1.6±0.2 μmol CO2 m−2 s−1 in the dry season). Plowed bare soil in between these two plantations showed an average loss of 1.8±0.5 μmol CO2 m−2 s−1.
While it seems unlikely that as large a carbon uptake as 6 Mg C ha−1 year−1 reflects a steady ecosystem state (Chambers et al. 2001b), it is also not obvious that the sum of instantaneous CO2 fluxes always provides a realistic estimate of average net ecosystem production. Lack of air turbulence and insufficient mixing of CO2 from canopy respiration during the night often is corrected with the measurement of a storage term. These corrected fluxes are intended to achieve better accuracy in estimating the amplitude of diurnal and seasonal cycles and the variability among sites in Amazonia.
However, the generally presumed conditions of flat and drainage-free surfaces along the tower fetch are rarely found. Thus, local topography and air-mass circulation are crucial (Araujo et al. 2002). Accounting for these site-specific problems is not trivial (Staebler and Fitzjarrald 2004). The hypothesis that low turbulence during nighttime is correlated proportionally to drainage losses, suggests that the use of thresholds based on turbulence scales (such as u* used in deciduous forests; Goulden et al. 1996) may be effective for some sites in Amazonia. This would allow the replacement of nighttime low-turbulent events with well-mixed ones (Miller et al. 2004; Araujo et al. 2002; Saleska et al. 2003). This may improve the realism of the NEE annual sum (Saleska et al. 2003). However, the accurate application of turbulence filters to a given site’s data requires parameters from other studies, such as using radon as proxy for CO2 transport within the canopy (Martens et al. 2004) or from direct measurement of in-canopy advection (Staebler and Fitzjarrald 2004). Until the problem of topography-related CO2 loss is properly resolved, with the development of widely accepted eddyflux correction methodology, site-dependent estimates of NEE annual sums for Amazonian forests remain uncertain.
5 Biometry studies
The annual allocation of photosynthetic products to wood growth is estimated from the annual biomass increment in a forest stand. This is achieved either through repeated surveys of trees (Baker et al. 2004a, b) or through the use of dendrometer bands assigned to representative trees within the stand (Rice et al. 2004; Vieira et al. 2004). Such forest inventory plots were used to infer average net storage in live biomass of undisturbed tropical forests. However, these measurements only account for inter-annual and stand level variations over a relatively short time scales and limited spatial coverage. The evaluation of residence time for carbon in vegetation is critical for carbon storage calculations, thus growth rates, stand and vegetation ages are essential measurements for carbon dynamic studies. Houghton et al. (2000) observed that the previous biomass of deforested areas was the major uncertainty in the estimation of carbon flux to the atmosphere. The uncertainties in biomass estimates have several causes. Maybe one of the most important and difficult to solve is the natural heterogeneity of tropical forests. According to Thomaz (2000) there may be up to 35,000 species of angiosperms in the Amazon basin. But the extent of the potential diversity still to be found is unknown (Tuomisto et al. 1995), and may render this last estimate a conservative one. Still, this number represents almost 14% of all known angiosperms species in the globe, encompassed in an area equivalent to 4.7% of the land area on Earth. While on one hand, such high biodiversity is a joy of life, on the other hand, it makes allometric biomass measurement a logistical nightmare. As there are hundreds of different individuals and species per hectare, it is expected high variability on form, height, wood density and structure within a site (Clark 2004). Besides this in-site variability there is also important variability among different regions due to differences in soil composition and climatic differences (Malhi et al. 1999; Marengo and Nobre 2001).
Thus, the mega complexity of this ecosystem engendered several questions related to above ground biomass estimation (Keller et al. 2001; Brown et al. 1995). The most common methods to tackle these questions are: (1) direct estimates by cutting and weighting all trees above 10 cm diameter at breast height (DBH) over a relatively small area, generally smaller than 0.5 ha (Phillips et al. 2004) or (2) indirect estimates by using allometric equations (Chambers 2001; Gerwing and Farias 2000; Brown et al. 1995). These equations are obtained by cutting and weighting a determined number of trees and relating their biomass with DBH and height. Estimates of biomass in Amazonia may involve scattered measurements of DBH and height over large areas, like the measurements made by the RADAMBRASIL project or over small areas (generally smaller than 5 ha) (Houghton et al. 2001). Although it seems simple, measurement of parameters like DBH can be difficult in tropical forests, resulting in large errors in biomass estimation (Clark 2002). In addition, it seems that the choice of the right allometric equation is of fundamental importance to reduce those errors (Chave et al. 2004). Most of the live biomass, growth, mortality and recruitment measurements are taken on permanent forest transects, located in more representative parts of the studied forest. Normally, longer transects aims to incorporate spatial heterogeneity avoiding bias associated with small-scattered plots that can be disproportionately influenced by emergent trees (Rice et al. 2004). Brown et al. (1995) support the general idea that in tropical forests, the aboveground biomass is dominated by few large trees, showing as example a forest in western Amazonia where trees with DBH above 60 cm account for 50% of the biomass. In this transect, trees are identified and dendrometers are placed in a random sub-sample within the plot, but distributed across taxonomic families and sizes. The dendrometers normally used are stainless-steel bands (Rice et al. 2004; Vieira et al. 2004) or automated systems (Miranda 2002; Silva et al. 2002). Baker et al. (2004a, b) produced three estimates based on above-ground measurements in plots. Their results were clustered in three groups: east and central plots; western plots and floodplain plots. The east and central plots showed the smallest carbon gain (−0.37 ton C ha−1 year−1) and the floodplain forest plots the largest carbon gain (−1.2 ton C ha−1 year−1).
Number of individuals with diameter at breast height higher than 10 cm/ha followed by the number of botanical families and species in Manaus, Rio Branco and Santarém (Vieira et al. 2004)
An important aspect of aboveground biomass, especially wood, is the residence time of carbon in this pool. This is related to functional group, forest stand, nutrient availability and environmental conditions. Condit et al. (1995) have shown large increases in tree mortality associated with extremely dry conditions in tropical forests. The difference in adaptation to extreme conditions potentially leads to changes in how biomass is distributed among different functional types (Philips and Gentry 1994). Also, the accumulation of C in living wood may be offset by tree mortality; hence the degree to which a forest is acting as a source or sink of C to the atmosphere is not solely dependent on the growth rate of living wood. Understanding mortality, its spatial and temporal distribution, and its relationship to climate and disturbance (such as fire, for instance) are critical points to be considered when attributing mean life time for trees on global models of climatic change. Chambers et al. (1998) found that the ages of emergent trees (DBH varying from 80 cm to 240 cm) sampled near Manaus, central Amazonia, varied from 200 years to 1,400 years, although such millenary trees were not documented by other investigations anywhere else in the tropics (Worbes 2002).
Besides the high variability in tree age reported for a single site, radiocarbon ages determined in several trees in Manaus, Rio Branco and Santarém showed also a large within site variability, as well as a high variability among sites (Vieira et al. 2004). For instance, the mean radiocarbon age in Manaus was 306 years, while in Santarém the mean radiocarbon age was much younger, approximately 160 years (Simone Vieira and Plínio Camargo, personal communication). Overall, the few results shown so far indicate that mean tree age in Amazonia appears to be higher than 42 years as used in some models.
6 NEE of Amazonia
As most of the early estimates of NEE pointed towards a carbon gain by the forest, several studies attempted to locate the potential extra carbon storage in the Amazonian ecosystems. Some claimed upfront that very large uptake by these forests was ecologically unrealistic (Keller et al. 2001), mainly because there would be nowhere to store the implied large amount of carbon. Malhi and Grace (2000) considered the hypothesis that the majority of the extra carbon was accumulating in the soil organic matter at a rate of 3.5 Mg C ha−1 year−1. However, Telles et al. (2003) have estimated carbon stocks from 102 Mg to 151 Mg C ha−1 year−1 with storage rates of 0.09–0.13 Mg C ha−1 year−1 (they considered a 0.5% year−1 increase in ecosystem productivity), and with additional potential storage of 0.18–0.27 Mg C ha−1 year−1 in surface litter and roots. These authors have reported turnover rates in the soil organic matter pool of less than 10 years; however, for the last 20 years, a stable carbon stock was identified in primary forests of the Manaus region, suggesting a low possibility of these soils being potential carbon sinks for the considered period (Telles et al. 2003). Residence time and fluxes of carbon in or out of the soil and the time of residence of this carbon in the soil profile are key factors in estimating the importance of this compartment to the carbon balance in a forest system. Studies by Camargo et al. (1999) and Bernoux et al. (2002) have shown that the majority of carbon stocks in the soil are associated with recalcitrant compounds (coal, lignin, etc.) that do not turnover on the timescales of decades or centuries. Thus, production and decomposition of wood predominates the dynamic carbon stock of these forests (Chambers et al. 2000; Phillips et al. 1998). Recent data from three sites in Amazonia (Chambers et al. 2004; Viera et al. 2004; Silva et al. 2002) have shown that tree growth and mortality rates are too slow to account for the amount of carbon similar to that indicated by early eddy covariance studies. Chambers et al. (2004) using stochastic-empiric modeling of carbon dynamics of live and dead trees for one site central Amazonia indicated that the forest can support only a 0.5% increase in NPP over a long period. However, these authors warn that their “...predictions about whether or not old-growth tropical forests will act as carbon sources, sinks, or remain in overall balance, depends on assumptions for which reliable field data are often not available.” (Chambers et al. 2004).
The site-specific question was addressed in a recent work by Saleska et al. (2003). The authors presented eddyflux NEE estimates at Tapajós, with biometric measurements suggesting that the site was under the effects of transient disturbance. These could include recent past ENSO warm episodes, which may have caused anomalously large mortality rates and have potentially offset carbon uptake during the recovery. Keller et al. (1996) have shown that the fall of a giant emergent tree can result in a loss of about 750 kg C ha−1 over 13 ha within a few years. These authors also pointed out that recovery from long-term climate stress could otherwise lead to net carbon uptake in the present. These uncertainties suggest that there is a large spatial variability over terra firme tropical forests in Amazonia, and that average annual NEE estimates should combine biometric measurements with climate interannual variability to provide more accurate assessments.
The aquatic systems are another important component of the Amazonian ecosystem that need to be considered for a regional carbon balance. Richey et al. (2002) estimated that up to 1.2 Mg C ha−1 year−1 could be lost to the atmosphere in a large swath of central Amazonia. Extrapolation across the entire flooded area of the basin produced an emission of approximately 0.5 Pg C year−1, which is of the same magnitude as the carbon emission caused by land-use changes in Tropical Latin America (DeFries et al. 2002). Contrastingly, Waterloo et al. (in preparation) have recently concluded, from a continuous long-term measurement in a monitored black water catchment near Manaus, that typical terra-firme forest export a significantly smaller load of dissolved organic compounds (<0.3 Mg C ha−1 year−1) than speculated by Richey et al. (2002). This transfers the potential source of the estimated riverine emissions to productive ecosystems within whitewater floodplains (Junk 1997; Piedade et al. 2000). Shedding a little more light on this controversy, eddyflux measurements were carried out for a complete seasonal cycle in the Pantanal wetland area, near the border of Brazil and Bolivia, to the South of Amazonia (Manzi et al. 2002). These measurements have indicated a marked source during the flood period, similar to the estimate of Richey et al. (2002), but have also shown a strong sink in the ebb season (Manzi et al. 2002). The reversal in carbon balance is probably due to vigorous revegetation by C4 grasses and other floodplain plants (Piedade et al. 2000), confirming the potential importance of productivity within the floodplain in balancing the total exchange of these biomes with the atmosphere. In addition, radiocarbon measurements (Emilio Mayorga and Anthony Aufdenkampe, personal communication) show that most emitted C is relatively recent in origin—aquatic systems lose C to the atmosphere rapidly enough so that adjustments to change would be instantaneous. Therefore, the role of aquatic systems in local to regional C budgets will depend primarily on the temporal balance of CO2 sources in the flood season and CO2 sinks in the ebb season. However, the origin of the CO2 being lost to the atmosphere (e.g. in-stream or riparian processes vs. export from uplands) and how long it takes for C to transit to the aquatic system (without significant—e.g. decadal—time lags, there can be no net storage or loss) also need to be considered. The importance of aquatic systems will also depend on their potential for change with land use or climate, a matter still under investigation.
More recently, a series of articles called for caution in interpreting potential gains of carbon by biomass in the Amazonia (Chambers and Silver 2004; Clark 2004; Körner 2004). In short, the main message of these studies may be summarized by two sentences from Christian Körner works: (1) “a stimulation of assimilation does not necessarily mean that there is more growth”, and (2) “a stimulation of growth does not necessarily mean enhanced carbon sequestration”. Thus the C storage under a NEE perspective is also dependent on changes in the dynamic structure, or regrowth, of forests. A change from emergent trees to lianas and small trees (Laurance et al.1997; Phillips et al. 2002b) can reflect a net carbon loss to the atmosphere. Another key issue, illustrated by LBA project, is that the Amazonian ecosystems have very large capacity for interannual shifts in C balance (i.e. the interannual variation in the wood storage term shown by Vieira et al. 2004).
An recent comparison of atmospheric CO2 inversion models indicated that tropical lands in the Americas could be a net source of 0.62±1.15 Pg C year−1 (Gurney et al. 2002). This contrasts with estimates by Schimel et al. (2001) where the tropics appear as a neutral source or sink for carbon. Both, Gurney et al. (2002) and Schimel et al. (2001), stressed that uncertainties in the tropical regions of the world are much greater than elsewhere due to the lack of an adequate and accurate CO2 concentration measurement network.
The four last estimates of carbon flux in the tropics due to land use changes produced different values. Fearnside (2000) and Houghton (2003) using deforestation rates from FAO estimated 2.4 Pg and 2.2 Pg C year−1, respectively, for the net flux due to land use changes in the tropics. Achard et al. (2002) and DeFries et al. (2002), based on deforestation inferred directly from satellite data, produced lower estimates for the tropics of 0.96 Pg C year−1, respectively. If the estimates of Fearnside (2000) and Houghton (2003) of tropical deforestation are correct, the emission term in the tropics would be approximately 2.3 Pg C year−1. However, the inversion models are capturing in average total emissions of only 0.62 Pg C year−1. Forgetting for a moment the uncertainty of the inversion models, this would imply that there must be a local sink of approximately 1.6–1.7 Pg C year−1, which would be equivalent to a net gain of 2.7 ton C ha−1 year−1 in areas of pristine rainforest. If the estimates by Achard et al. (2002) and DeFries et al. (2002) are correct, conversion of tropical forests would be releasing to the atmosphere an average of 0.93 Pg C year−1. In this scenario, a smaller C sink of approximately 0.31 Pg C year−1 would be needed then to satisfy the inversion models, resulting in a potential sink strength for pristine forests of 0.85 ton C ha−1 year−1. In this case, the tropical forests would act as source of carbon to the atmosphere equivalent to 0.85 ton C ha−1 year−1. Using our crude extrapolation of LBA values for the Amazon forests (5 million km2) we come out with a carbon exchange estimate varying from −3.0 Pg to 0.75 Pg C year−1 (Fig. 7).
There is a large uncertainty in carbon fluxes estimates for the tropics as a whole and in Amazonia in particular bring the awareness that we have not enough information to decide whether Amazonia is a carbon source or sink, or if it is both, depending on climatic variation and the rates of forest conversion. The accumulation of sufficient information for a huge system such as the Amazon region has to advance step by step.
The future of the carbon balance, especially in the Brazilian Amazonia, is closely related to the economy. As the Brazilian economy grows, deforestation rates may increase if the government does not provide environmentally sustainable regulation. Increasing deforestation rates might overwhelm a potential sink by primary and regrowth forests.
The complex sink/source duality of Amazonia was discussed extensively in the literature and during several scientific meetings of the LBA project; however, no strong conclusions were drawn, other than the ones presented in this work. The valuable data emerging from the pioneering network of flux towers and forest plots in Amazonia bring, for the first time, the unique chance to compare and understand diverse and contrasting ecophysiological processes. Process models have yet to capture this immense complexity in a fine resolution. Certainly, it is an important and vital question for Brazil and for the world from many perspectives. However, by paying too much attention to the issue of uniqueness of a sink or source character and finding its magnitude, we may loose a great opportunity to discuss the basic functioning and the complexities of the ecosystems in Amazonia.
Jean Pierre H.B.Ometto, Luiz Martinelli and Humberto Rocha acknowledge FAPESP and NASA, through LBA, for financial support. Antonio D. Nobre acknowledges support from PPG7-FINEP, European Commission and CNPq. Paulo Artaxo acknowledges the Millennium Institute from CNPq and FAPESP for financial support. The authors acknowledge Adam West, Suzan Trumbore and an anonymous reviewer for great and critical contribution that improved the accuracy and clarity of this paper.
- Andersen LE, Granger CWJ, Reis EJ, Weinhold D, Wunder S (2002) The dynamic of deforestation and economic growth in Brazilian Amazonia. Cambridge University Press, UKGoogle Scholar
- Andreae MO, Artaxo P, Fischer H, Freitas SR, Gregoire JM, Hansel A, Hoor P, Kormann R, Krejci R, Lange L, Lelieveld J, Lindinger W, Longo K, Peters W, de Reus M, Scheeren B, Dias MAFS, Strom J, van Velthoven PFJ, Williams J (2001) Transport of biomass burning smoke to the upper troposphere by deep convection in the equatorial region. Geophys Res Lett 28(6):951–954CrossRefGoogle Scholar
- Andreae MO, Almeida SS, Artaxo P, Brandão C, Carswell FE, Ciccioli P, Culf A, Esteves JL, Gash J, Grace J, Kabat P, Lelieveld J, Malhi Y, Manzi AO, Meixner FX, Nobre AD, Nobre C, Lourdes Ruivo MA, Silva-Dias MA, Stefani P, Valentini R, von Jouanne J, Waterloo M (2002) Biogeochemical cycling of carbon, water, energy, trace gases and aerosols in Amazonia: the LBA-EUSTACH experiments. J Geophys Res 107(D20):8066–8091. DOI 10.1029/2001JD000524Google Scholar
- Araujo AC, Nobre AD, Kruijt B, Culf AD, Stefani P, Elbers J, Dallarosa R, Randow C, Manzi AO, Valentini R, Gash JHC, Kabat P (2002) Dual tower longterm study of carbon dioxide fluxes for a central Amazonian rain forest: the Manaus LBA site. J Geophys Res Atmos 107(D20):8090. DOI 10.1029/2001JD000676Google Scholar
- Artaxo P, Martins JV, Yamasoe MA, Procopio AS, Pauliquevis TM, Andreae MO, Guyon P, Gatti LV, Leal AMC (2002) Physical and chemical properties of aerosols in the wet and dry seasons in Rondonia, Amazonia. J Geophys Res Atmos 107(D20):8081Google Scholar
- Avissar R, Silva Dias PL, Silva Dias MAF, Nobre C (2002) The large-scale biosphere-atmosphere experiment in amazonia (LBA): insights and future research needs. J Geophys Res Special LBA Issue 107. DOI 10.1029/2002JD002704Google Scholar
- Baidya Roy S, Avissar R (2002) Impact of land use/land cover change on regional hydrometeorology in Amazonia. J Geophys Res 107(D20). DOI 10.1029/2000JD000266Google Scholar
- Baker TR, Phillips OL, Malhi Y, Almeida S, Arroyo L, Di Fiore A, Killen T, Laurance SG, Laurance WL, Lewis SL, Lloyd J, Monteagudo A, Neill DA, Patino S, Pitman NCA, Silva N, Vasquez MR (2004a) Variation in wood density determines spatial patterns in Amazonian forest biomass. Global Change Biol 10(5):545–562CrossRefGoogle Scholar
- Baker TR, Phillips OL, Malhi Y, Almeida S, Arroyo L, Di Fiore A, Erwin T, Higuchi N, Killeen TJ, Laurance SG, Laurance WF, Lewis SL, Monteagudo A, Neill DA, Vargas PN, Pitman NCA, Silva JNM, Martınez RV (2004b) Increasing biomass in Amazonian forest plots. Phil Trans R Soc Lond B. DOI 10.1098/rstb.2003.1422Google Scholar
- Baldocchi D, Falge E, Gu LH, Olson R, Hollinger D, Running S, Anthoni P, Bernhofer C, Davis K, Evans R, Fuentes J, Goldstein A, Katul G, Law B, Lee XH, Malhi Y, Meyers T, Munger W, Oechel W, U KTP, Pilegaard K, Schmid HP, Valentini R, Verma S, Vesala T, Wilson K, Wofsy S (2001) FLUXNET: A new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide, water vapor, and energy flux densities. Bull Am Meteorol Soc 82:2415–2434CrossRefGoogle Scholar
- Browder J, Godfrey B (eds) (1997) Rainforest Cities—Urbanization, Development, and Globalization of the Brazilian Amazon. Columbia University Press, New YorkGoogle Scholar
- Carswell FE, Costa AL, Palheta M, Malhi Y, Meir PW, Costa J de PR, Ruivo M de L, Leal LSM, Costa JMN, Clement RJ, Grace J (2002) Seasonality in CO2 and H2O flux at an Eastern Amazonian rain forest. J Geophys Res Atmos 107(D20):8076. DOI 10.1029/2000JD000284Google Scholar
- Chomovitz KM, Thomas TS (2001) Geographical patterns of land use and land intensity in the Brazilian Amazon. Policy Research Working Paper 2687, The World BankGoogle Scholar
- Ciais P, Tans PP, White JWC, Trolier M, Francey RJ, Berry JA, Randall DR, Sellers PJ, Collatz JG, Schimel DS (1995) Partitioning of ocean and land uptake of CO2 as inferred by δ13C measurements from the NOAA climate monitoring and diagnostics laboratory global air sampling network. J Geophys Res 100:5051–5070CrossRefGoogle Scholar
- Clark DA (2004) Sources or sinks? The responses of tropical forests to current and future climate and atmospheric composition. Phil Trans R Soc Lond B. DOI 10.1098/rstb.2003.1426Google Scholar
- Conway TJ, Tans PP, Waterman LS, Thoning KW, Kitzis DR, Masarie KA, Zhang N (1994) Evidence for interannual variability of the carbon cycle from the National Oceanic and Atmospheric Administration/Climate Monitoring and Diagnostics Laboratory global air sampling network. J Geophys Res 99:22831–22855CrossRefGoogle Scholar
- Culf AD, Esteves JL, Marques Filho AO, Rocha HR (1996) Radiation, temperature and humidity over forest and pasture in Amazonia. In: Gash JHC, Nobre CA, Roberts JM, Victoria RL (eds) Amazonian climate and deforestation. Wiley, New York, pp 175–192Google Scholar
- Gash JHC, Huntingford C, Marengo JA, Betts RA, Cox PM, Fisch G, Fu R, Gandu AW, Harris PP, Machado LAT, von Randow C, Silva Dias MA (2004) Amazonian climate: results and future research. Theor Appl Climatol (LBA Special Issue) 78:187–193Google Scholar
- Gurney KR, Law RM, Denning AS, Rayner PJ, Baker D, Bousquet P, Bruhwiler L, Chen YH, Ciais P, Fan S, Fung IY, Gloor M, Heimann M, Higuchi K, John J, Maki T, Maksyutov S, Masarie K, Peylin P, Prather M, Pak BC, Randerson J, Sarmiento J, Taguchi S, Takahashi T, Yuen CW (2002) Towards robust regional estimates of CO2 sources and sinks using atmospheric transport models. Nature 415(6872):626–630PubMedCrossRefGoogle Scholar
- Hirsch AI, Little WS, Houghton RA, Scott NA, White JD (2004) The net carbon flux due to deforestation and forest re-growth in the Brazilian Amazon: analysis using a process-based model. Global Change Biol 10(5):908–924. DOI 10.1111/j.1529–8817.2003.00765.xGoogle Scholar
- Houghton RA (2003) Revised estimates of the annual net flux of carbon to the atmosphere from changes in land use and land management 1850–2000. Tellus 55B:378–390Google Scholar
- House JI, Prentice IC, Ramankutty N, Houghton RA, Heimann M (2003) Reconciling apparent inconistencies in estimates of terrestrial CO2 sources and sinks. Tellus 55B:345–363Google Scholar
- INPE (Instituto Nacional de Pesquisas Espaciais) (2004) Desflorestamento 1998–2004. (http://www.inpe.gov.br) INPE, São José dos Campos, SãoPaulo, Brazil
- Junk WJ (1997) The central Amazon floodplain: ecology of pulsing system. Springer, Berlin Heidelberg, p 525Google Scholar
- Kesselmeier J, Kuhn U, Wolf A, Andreae MO, Ciccioli P, Brancaleoni E, Frattoni M, Guenther A, Greenberg J, Vasconcellos PD, De Oliva ST, Tavares T, Artaxo P (2000) Atmospheric volatile organic compounds (VOC) at a remote tropical forest site in central Amazonia. Atmos Environ 34:4063–4072CrossRefGoogle Scholar
- Kesselmeier J, Ciccioli P, Kuhn U, Stefani P, Biesenthal T, Rottenberger S, Wolf A, Vitullo M, Valentini R, Nobre A, Kabat P, Andreae MO (2002a) Volatile organic compound emissions in relation to plant carbon fixation and the terrestrial carbon budget. Global Biogeochem Cycles 16:1126. DOI 10.1029/2001GB001813Google Scholar
- Kesselmeier J, Kuhn U, Rottenberger S, Biesenthal T, Wolf A, Schebeske G, Andreae MO, Ciccioli P, Brancaleoni E, Frattoni M, Oliva ST, Botelho ML, Silva CMA, Tavares TM (2002b) Concentrations and species composition of atmospheric volatile organic compounds (VOCs) as observed during the wet and dry season in Rondonia (Amazonia) J Geoph Res Atmos 107(D20):8053Google Scholar
- Körner C (2004) CO2 enrichment may cause tropical forests to become carbon sources. Phil Trans R Soc Lond B. DOI 10.1098/rstb.2003.1429Google Scholar
- Kuhn U, Rottenberger S, Biesenthal T, Wolf A, Schebeske G, Ciccioli P, Brancaleoni E, Frattoni M, Tavares T, Kesselmeier J (2002) Isoprene and monoterpene emission of Amazonian tree species during the wet season: Direct and indirect investigations of controlling functions. J Geophys Res Atmos 107. DOI 10.1029/2001JD000978Google Scholar
- Laurance WF (2004) Forest–climate interactions in fragmented tropical landscapes. Phil Trans R Soc Lond B. DOI 10.1098/rstb.2003.1430Google Scholar
- Lloyd J, Bird MI, Veenendaal EM, Kruijt B (2001) Should phosphorus availability be constraining moist tropic al forest responses to increasing CO2 concentrations? In: Schulze ED (ed) Global biogeochemical cycles in the climate system. Academic, San Diego, pp 95–114Google Scholar
- Malhi Y, Baker T, Phillips OL, Almeida S, Alvarez E, Arroyo L, Chave J, Czimezik CI, DiFiore A, Higuchi N, Killeen TJ, Laurance SG, Laurance WF, Lewis SL, Montoya LMM, Monteagudo A, Neill DA, Vargas PN, Patina S, Pitman NCA, Quesada CA, Silva JNM, Lezama AT, Martinez RV, Terborgh J, Vinceti B, Lloyd J (2004) The above-ground wood productivity and net primary productivity of 100 neotropical forest plots. Global Change Biol 10(5):563–591CrossRefGoogle Scholar
- Malhi Y, Wright J (2004) Spatial patterns and recent trends in the climate of tropical rainforest regions. Phil Trans R Soc Lond B. DOI 10.1098/rstb.2003.1433Google Scholar
- Manzi AO et al (2002) NEE over a floodable ecosystem in Pantanal, Brazil, II LBA Scientific Conference, ManausGoogle Scholar
- Marengo JA, Nobre CA (2001) General characteristics and variability of climate in the Amazon basin and its links to the global climate system. In: McClain ME, Victoria RL, Richey JE (eds) The biogeochemistry of the Amazon basin. Oxford University Press, New YorkGoogle Scholar
- Margulis S (2003) Causes of Deforestation of the Brazilian Amazon. World Bank working paper no. 22Google Scholar
- McGuire AD, Melillo JM, Kicklighter DW, Pan YD, Xiao XM, Helfrich J, Moore B, Vorosmarty CJ, Schloss AL (1997) Equilibrium responses of global net primary production and carbon storage to doubled atmospheric carbon dioxide: sensitivity to changes in vegetation nitrogen concentration. Global Biogeochem Cycles 11(2):173–189CrossRefGoogle Scholar
- Miranda AC, Miranda HS, Lloyd J, Grace J, McIntyre JA, Meir P, Riggan P, Lockwood R, Brass J (1996) Carbon dioxide fluxes over a cerrado sensu stricto in central Brazil. In: Gash JHC, Nobre CA, Roberts JM, Victoria RL (eds) Amazonian climate and deforestation. Wiley, New York, pp 353–364Google Scholar
- Miranda EV (2002) Padrão de desenvolvimento radial horario de tres especies arboreas da Amazonia utilizando dendrometros automaticos. MSc Dissertation, INPA, ManausGoogle Scholar
- Martens CS, Shay TJ, Mendlovitz HP, Matross DM, Saleska SR, Wofsy SC, Woodward WS, Menton MC, De Moura JMS, Crill PM, De Moraes OLL, Lima RL (2004) Radon fluxes in tropical forest ecosystems of Brazilian Amazonia: night-time CO2 net ecosystem exchange derived from radon and eddy covariance methods. Global Change Biol 10(5):618. DOI 10.1111/j.1365–2486.2004.00764.xGoogle Scholar
- Phillips OL, Martinez RV, Arroyo L, Baker TR, Killeen T, Lewis SL, Malhi Y, Mendoza AM, Neill D, Vargas PN, Alexiades M, Ceron C, Di Fiore A, Erwin T, Jardim A, Paiacios W, Saidias M, Vinceti B (2002b) Increasing dominance of large lianas in Amazonian forests. Nature 418:770–774PubMedCrossRefGoogle Scholar
- Phillips O, Baker T, Arroyo L, Higuchi N, Killeen T, Laurance WF, Lewis1 SL, Lloyd J, Malhi Y, Monteagudo A, Neill D, Nunez Vargas P, Silva N, Terborgh J, Vasquez Martınez R, Alexiades M, Almeida S, Brown S, Chave J, Comiskey JA, Czimczik CI, Di Fiore A, Erwin T, Kuebler C, Laurance SG, Nascimento HEM, Olivier J, Palacios W, Patino S, Pitman N, Quesada CA, Saldias M, Torres Lezama A, Vinceti B (2004) Pattern and process in Amazon tree turnover, 1976–2001. Phil Trans R Soc Lond B. DOI 10.1098/rstb.2003.1438Google Scholar
- Piedade MTF, Junk WJ, Parolin P (2000) The flood and photosynthetic response of trees in a White water floodplain (várzea) of the Central Amazon, Brazil. Verh Int Verein Limmol 27:1734–1739Google Scholar
- Prentice IC, Farquhar GD, Fasham MJR, Goulden ML, Heimann M, Jaramillo VJ, Kheshgi HS, Le Quere C, Scholes RJ, Wallace DWR (2001) The carbon cycle and atmospheric carbon dioxide. In: Houghton JT (ed) Climate change 2001: the scientific basis. Cambridge University Press, Cambridge, pp 183–237Google Scholar
- Procópio AS, Artaxo P, Kaufman YJ, Remer LA, Schafer J S, Holben BN (2004) Multiyear analysis of Amazonian biomass burning smoke radiative forcing of climate. Geophys Res Lett 31:L03108. DOI 10.1029/2003GL018646Google Scholar
- Rocha HR, Freitas H, Rosolem R, Juarez R, Tannus RN, Ligo MV, Cabral OMR, Silva Dias MAF (2002) Measurements of CO2 exchange over a woodland savanna (Cerrado Sensu stricto) in southeast Brasil. Biota Neotropica 2(1):Google Scholar
- Saleska SR, Miller SD, Matross DM, Goulden ML, Wofsy SC, da Roacha HR, de Camargo PB, Crill P, Daube BC, de Freitas HC, Hutyra L, Keller M, Kirchoff V, Menton M, Munger JW, Pyle EH, Rice AH, Silva H (2003) Carbon in Amazon forests: unexpected seasonal fluxes and disturbance-induced losses. Science 302:1554–1557PubMedCrossRefGoogle Scholar
- Schimel DS, House JI, Hibbard KA, Bousquet P, Ciais P, Peylin P, Braswell BH, Apps MJ, Baker D, Bondeau A, Canadell J, Churkina G, Cramer W, Denning AS, Field CB, Friedlingstein P, Goodale C, Heimann M, Houghton RA, Melillo JM, Moore III B, Murdiyarso D, Noble I, Pacala SW, Prentice IC, Raupach MR, Rayner PJ, Scholes RJ, Steffen WL, Wirth C (2001) Recent patterns and mechanisms of carbon exchange by terrestrial ecosystems. Nature 414:169–172PubMedCrossRefGoogle Scholar
- Silva Dias MAF, Rutledge S, Kabat P, Silva Dias PL, Nobre C, Fisch G, Dolman AJ, Zipser E, Garstang M, Manzi A, Fuentes JD, Rocha H, Marengo J, Plana-Fattori A, Sá L, Alvalá R, Andreae M O, Artaxo P, Gielow R, Gatti L (2002) Clouds and rain processes in a biosphere atmosphere interaction context in the Amazon region. J Geophys Res. DOI 10.1029/2001JD000335Google Scholar
- Telles ECC, de Camargo PB, Martinelli LA, Trumbore SE, da Costa ES, Santos J, Higuchi N, Oliveira RO Jr (2003) Influence of soil texture on carbon dynamics and storage potential in tropical forest soils of Amazonia. Global Biogeochemical Cycles 17:1040. DOI 10.1029/2002GB001953Google Scholar
- Vieira S, Camargo PB, Selhorst D, Silva R, Hutyra L, Chambers JQ, Brown IF, Higuchi N, Santos J, Wofsy S, Trumbore SE, Martinelli LA (2004) Forest structure and carbon dynamics in Amazonian tropical rain forest. Oecologia. DOI 10.1007/s00442–004–1598-zGoogle Scholar
- von Randow C, Sá LDA, Prasad GSSD, Manzi A, Kruijt B (2001) Scale variability of surface fluxes of energy and carbon over a tropical rain forest in South-West Amazonia. I. Diurnal Conditions. J Geoph Res 107. DOI 10.1029/2001JD000379Google Scholar
- Williams E, Madden N, Rosenfeld D, Gerlach J, Atkinson L, Dunnemann N, Frostrom G, Gears N, Antonio M, Biazon B, Camargo R, Franoa H, Gomes A, Lima M, Machado R, Manhaes S, Nachtigall L, Piva H, Quintiliano W, Machado L, Artaxo P, Roberts G, Renno N, Blakeslee R, Bailey J, Boccippio D, Betts A, Wolff D, Roy B, Halverson J, Rickenbach T, Fuentes J, Avelino E (2002) Contrasting convective regimes over the Amazon: implications for cloud electrification. J Geophys Res 107(D20):50.1–50.19Google Scholar
- Wiedinmyer C, Guenther A, Harley P, Hewitt N, Geron C, Artaxo P, Steinbrecher R, Rasmussen R (2004) Global organic emissions from vegetation. In: Granier C, Artaxo P, Reeves C (eds) Emissions of trace gases and aerosols into the atmosphere. Kluwer, DordrechtGoogle Scholar