Aldose, amino acid, and elemental compositions were determined for flux-weighted samples of coarse (> 63 pm) and fine (< 63 pm) particulate organic material and ultrafiltered (> 1,000 Daltons) dissolved organic matter collected at three sites along the Brazilian Amazon River and six of its major tributaries. Concentrations of total organic C (TOC) were relatively uniform (55Ok 100 PM) at all sites, with DOC comprising the major (50-100%) component. An average of 77% of the total DOC was isolated by ultrafiltration.The greatest compositional differences observed in the Amazon River system were among the coarse, fine, and dissolved organic fractions. All coarse particulate fractions were nitrogen-poor (atomic C : N = 21) and exhibited stable carbon isotope, aldose, and amino acid compositions similar to those of angiosperm tree leaves. Coarse particulate organic materials, although the least degraded of the three fractions, had lost appreciable carbohydrate and had immobilized excess nitrogen of apparent bacterial origin. Fine particulate materials were more nitrogen-rich (C : N = 9) than coarse counterparts and had lower total aldose yields and glucose percentages. Fine particles gave greater total yields of amino acids, characterized by high ratios of basic vs. acidic components. Coexisting dissolved organic materials recovered by ultrafiltration were nitrogen-poor (C: N = 27-52) and yielded the lowest amounts of aldoses, among which deoxy sugars were concentrated. Dissolved fractions gave extremely low yields of amino acids in mixtures that were enriched in nonprotein components and in acidic vs. basic molecules. These yield and composition patterns are consistent with a "regional chromatography" model in which highly degraded leaf material is solubilized and then partitioned between soil minerals and water during transport to the river, resulting in suspended fine particulate organic materials of soil origin that are nitrogen-rich and coexisting dissolved organic substances that are nitrogen-poor. manuscript.

Depth-integrated, discharge-weighted water samples were collected over 1,800km ofthe Amazon River on eight cruises at different stages of the hydrograph, 1982-1984 and coarse (CPOC, > 63 rm) particulate organic carbon as weight percentage of suspended sediment varied between 0.9-1.5% for FPOC and 0.5-3.49/o for CPOC. Concentrations of FPOC ranged from 5 mg liter-' upriver to 2 mg liter-' downriver in the mainstem and from 6 mg liter-' inthe Rio Madeira to i 1 in the Rio Negro. CPOC had similar distribution patterns. but with concentrations 15.-30% those of FPOC. Dissolved organic carbon (DOC) averaged 4-6 mg liter-r in the mainstem and up to 12 ml; liter -I in the Rio Negro. Upriver dissolved inorganic carbon (DIC) concentrations of about 1,200 @LM were diluted by tributaries and floodplain drainage to 600 PM at the most downriver site. Evasion ofCO,, invasion of O,, and in situ oxidation were of comparable magnitude, 3-8 pmol m-2 s-r.The average export of total organic carbon (TOC) was 36.1 Tg yr-i (8.5 g m-2 yr-I), of which 62% was DOC, 34% was FPOC, and 4% was CPOC. TOC inputs were insufficient to support in situ oxidation by a factor of at least two. A relatively small, rapidly cycling pool of labile organic matter may coexist with a much larger pool of more refractory material.A central problem in riverine ecology is to determine how the dynamics of carbon vary as flowing waters increase in size from first-order springs and seeps to the world's great rivers. With the advent of the energy r To whom correspondence should be addressed.

Wetlands of the Amazon River basin are globally significant sources of atmospheric methane. Satellite remote sensing (passive and active microwave) of the temporally varying extent of inundation and vegetation was combined with field measurements to calculate regional rates of methane emission for Amazonian wetlands. Monthly inundation areas for the fringing floodplains of the mainstem Solimõ es/Amazon River were derived from analysis of the 37 GHz polarization difference observed by the Scanning Multichannel Microwave Radiometer from 1979 to 1987. L-band synthetic aperture radar data (Japanese Earth Resources Satellite-1) were used to determine inundation and wetland vegetation for the Amazon basin (o500 m elevation) at high (May-June 1996) and low water (October 1995). An extensive set of measurements of methane emission is available from the literature for the fringing floodplains of the central Amazon, segregated into open water, flooded forest and floating macrophyte habitats. Uncertainties in the regional emission rates were determined by Monte Carlo error analyses that combined error estimates for the measurements of emission and for calculations of inundation and habitat areas. The mainstem Solimõ es/Amazon floodplain (54-701W) emitted methane at a mean annual rate of 1.3 Tg C yr À1 , with a standard deviation (SD) of the mean of 0.3 Tg C yr À1 ; 67% of this range in uncertainty is owed to the range in rates of methane emission and 33% is owed to uncertainty in the areal estimates of inundation and vegetative cover. Methane emission from a 1.77 million square kilometers area in the central basin had a mean of 6.8 Tg C yr À1 with a SD of 1.3 Tg C yr À1 . If extrapolated to the whole basin below the 500 m contour, approximately 22 Tg C yr À1 is emitted; this mean flux has a greenhouse warming potential of about 0.5 Pg C as CO 2 . Improvement of these regional estimates will require many more field measurements of methane emission, further examination of remotely sensed data for types of wetlands not represented in the central basin, and process-based models of methane production and emission.

[1] This paper presents the first application and validation of a 2D hydrodynamic model of the Amazon at a large spatial scale. The simulation results suggest that a significantly higher proportion of total flow is routed through the floodplain than previously thought. We use the hydrodynamic model LISFLOOD-FP with topographic data from the Shuttle Radar Topography Mission to predict floodplain inundation for a 240 Â 125 km section of the central Amazon floodplain in Brazil and compare our results to satellite-derived estimates of inundation extent, existing gauged data and satellite altimetry. We find that model accuracy is good at high water (72% spatial fit; 0.99 m root mean square error in water stage heights), while accuracy drops at low water (23%; 3.17 m) due to incomplete drainage of the floodplain resulting from errors in topographic data and omission of floodplain hydrologic processes from this initial model. Citation: Wilson, M.,

Tropical reservoirs upstream from hydroelectric dams are known to release significant amounts of methane to the atmosphere. Here we demonstrate that methane emissions downstream from hydroelectric dams can also be large. Emissions of CH4 downstream of Balbina reservoir in the central Amazon basin (Brazil) were calculated from regular measurements of degassing in the outflow of the turbines and downstream diffusive losses. Annual emissions from the reservoir surface and downstream from the dam were 34 and 39 Gg C, respectively. The downstream emission alone represented the equivalent of 3% of all methane released from central Amazon floodplain.