Methane emissions from Amazonian Rivers and their contribution to the global methane budget Henrique O. Sawakuchi, David Bastviken, Andre O. Sawakuchi, Alex V. Krusche, Maria V. R. Ballester and Jeffrey E. Richey Linköping University Post Print N.B.: When citing this work, cite the original article. Original Publication: Henrique O. Sawakuchi, David Bastviken, Andre O. Sawakuchi, Alex V. Krusche, Maria V. R. Ballester and Jeffrey E. Richey, Methane emissions from Amazonian Rivers and their contribution to the global methane budget, 2014, Global Change Biology, (20), 9, 2829-2840. http://dx.doi.org/10.1111/gcb.12646 Copyright: Wiley. http://eu.wiley.com/WileyCDA/ Postprint available at: Linköping University Electronic Press http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-110267 Title: Methane emissions from Amazonian Rivers and their contribution to the global methane budget Running head: Methane emissions from Amazonian Rivers Henrique O. Sawakuchi1*, David Bastviken2, André O. Sawakuchi3, Alex V. Krusche1, Maria Victoria R. Ballester1, Jeffrey E. Richey4 1Environmental Analysis and Geoprocessing Laboratory, Center for Nuclear Energy in Agriculture, University of São Paulo, Av. Centenário, 303, Piracicaba, SP 13400-970, Brazil;2Department of Thematic Studies – Water and Environmental Studies, Linköping University, Linköping, SE-581 83, Sweden;3Departament of Sedimentary and Environmental Geology, Institute of Geosciences, University of São Paulo, Rua do Lago, 562, São Paulo, SP 05508-080, Brazil;4School of Oceanography, University of Washington, Seattle, WA 98195-7940, USA. *Corresponding author: Henrique O. Sawakuchi Phone: +55 19 34294076; Fax: +55 19 34294059 e-mail: [email protected] Keywords: methane flux, CH , ebullition, tropical rivers, Amazon, greenhouse gas, natural 4 emission. Type of paper: Primary Research Articles 1 Abstract Methane (CH ) fluxes from world rivers are still poorly constrained, with measurements 4 restricted mainly to temperate climates. Additional river flux measurements, including spatio-temporal studies, are important to refine extrapolations. Here we assess the spatio- temporal variability of CH fluxes from the Amazon and its main tributaries, the Negro, 4 Solimões, Madeira, Tapajós, Xingu, and Pará Rivers, based on direct measurements using floating chambers. Sixteen out of 34 sites were measured during low and high water seasons. Significant differences were observed within sites in the same river and among different rivers, types of rivers, and seasons. Ebullition contributed to more than 50% of total emissions for some rivers. Considering only river channels, our data indicate that large rivers in the Amazon Basin release between 0.40 and 0.58 Tg CH yr-1. Thus, our estimates 4 of CH flux from all tropical rivers and rivers globally were, respectively, 19-51% to 31- 4 84% higher than previous estimates, with large rivers of the Amazon accounting for 22- 28% of global river CH emissions. 4 2 Introduction Despite their small areal extent inland waters play an important role in regional and global carbon balances as sources of both CO (Battin et al., 2009, Cole et al., 2007, 2 Richey et al., 2002, Tranvik et al., 2009) and CH (Bastviken et al., 2011). Recent 4 estimates show that inland waters outgas around 2.1 Pg C yr-1 as CO (Raymond et al., 2 2013) and 0.65 Pg C yr-1 as CH (Bastviken et al., 2011). 4 Global estimates of CH release from rivers are on the order of 1.5 Tg CH yr-1 4 4 (Bastviken et al., 2011). However, due to the scarcity of river CH data this estimate was 4 based on a small number of studies, largely from temperate areas. The lack of data from tropical rivers is particularly important given their large area and higher rate of emissions per unit area compared to temperate ecosystems (Bastviken et al., 2011). Most of the previous CH flux measurements in the Amazon were done in the 4 adjacent areas of the river channel, such as the floodplains locally called “varzeas”, lakes and flooded forest (Bartlett et al., 1988, Belger et al., 2011, Crill et al., 1988, Devol et al., 1988, Devol et al., 1990, Rosenqvist et al., 2002) or in hydroelectric (Abril et al., 2005, Kemenes et al., 2007, Lima, 2005). However, the large concentration of CH found in the Amazon river channel 4 (Richey et al., 1988) have shown the potential importance of this river itself as a source of CH to the atmosphere. Significant CH fluxes from three other tropical rivers were 4 4 recently estimated in Africa (Kone et al., 2010). However, these studies focused only on the diffusive component of CH fluxes, calculated from water-air CH concentration gradient 4 4 and piston velocity, whereas recent studies have shown that ebullition can also be important 3 in running waters (Baulch et al., 2011). Therefore, studies on CH emissions demand the 4 evaluation of both ebullition and diffusive components. Here we present the results from total flux measurements separated into diffusive and ebullitive components in the Amazon River and most of its main tributaries (Solimões, Negro, Madeira, Tapajós and Xingu Rivers), as well as their general spatial and temporal distribution. Our data points to a more significant role of the Amazon basin in the global CH budget than previously estimated. 4 4 Methods Sites description and sampling scheme The Amazon river basin stands out as the largest river system on Earth (Archer, 2005), formed by an extensive network of tributaries draining approximately 6 million km² of Andean and lowland basins that feed the 6,700 km long main river channel (Richey et al., 1988). In general, the weather is characterized by high temperatures with low variations throughout the year and is divided into well defined wet and dry seasons. Precipitation has strong seasonality modulated by shifts in the Intertropical Convergence Zone (ITCZ). The southward shift of the ITCZ during austral summer brings a large amounts of moisture to the basin, generating a monsoon precipitation system (Grimm et al., 2005, Vera et al., 2006), which results in large variations in river water levels. The Amazon river tributaries have distinct characteristics related to their water types and channel morphology. A general classification by water color is frequently used to separate rivers in the Amazon Basin (Sioli, 1985). Black water rivers such as Negro River usually drain lowland areas with heavily weathered rocks and sandy soils and have high dissolved organic matter content, low amounts of suspended sediments, median turbidity, low ionic strength, and high acidity (Mayorga & Aufdenkampe, 2002, Sioli, 1985). White water rivers such as Solimões and Madeira Rivers have their upstream catchment draining Andean areas and have high suspended sediment loads and dissolved solids concentrations, with neutral to alkaline pH (Mayorga & Aufdenkampe, 2002, Sioli, 1985). Clear water rivers such as Tapajós and Xingu drain the Brazilian shield and have low suspended 5 sediment loads, intermediary ionic content and slightly alkaline pH (Mayorga & Aufdenkampe, 2002, Sioli, 1985). Amazonian rivers have different types of depositional systems with varied sedimentary dynamics and sediment distribution (Archer, 2005, Latrubesse et al., 2005). This heterogeneity in sedimentary dynamics is seen in the occurrence of channel areas with higher deposition of organic rich sediment where CH production is favored. Great 4 differences in channel morphology and sediment deposition occur downstream from our studied site in some tributaries, as observed during field trips for CH measurements. The 4 mouths of the Negro, Xingu and Tapajós Rivers are blocked by sediment bars from the Amazon main channel. This damming of the lower portion of these tributaries generates wide channels (up to 19 km wide) with low water flow and regular wave action, promoting lake-like sedimentary dynamics in which deposition of organic rich mud in the central portion of the channel is common. Rivers draining highlands in the Andes such as the Solimões and Madeira have high suspended sediment load and their lowermost reaches are characterized by relatively narrow (1.5-3.5 km wide) and straight channels dominated by sand deposition with mud deposition occurring mainly over adjacent floodplains. The Amazon main channel has these same characteristics upstream from the Xingu River mouth. The Xingu River has an unique channel morphology. Its upstream sectors drain bedrock from an incised valley and have relatively low sediment deposition rates due to high water flow. This is in contrast with its depositional lake-like river mouth. Concentrations and fluxes (total flux, diffusive flux, and ebullition) of CH to the 4 atmosphere were measured on 52 occasions at 34 sites at the Negro, Solimões, Preto da Eva, Madeira, Tapajós, Xingu, Pará and Amazon Rivers and at a white water lake (Lake Curuai) in the Amazon River floodplain. Sixteen of these sites were measured during both 6 high (May 2012) and low (November 2012) water seasons and one site at Tapajós River was also measured in the falling water season (July 2012) (site number 14 in Figure 1). The remaining sites were visited only once during low, high or falling water season (see Table 1 for details). Sites in the Amazon River, near Óbidos, (numbered 27, 28 and 29 in Figure 1 and Table 1) and sites in the Pará River near Belém (numbered 23, 24 and 25) represents two cross-section profiles where measurements were made at three locations equally spaced across the channel of those rivers. Figure 1 and Table 1 show details about sampling periods and additional information about the sites. CH flux measurements 4 Surface water samples were collected simultaneously with flux measurements. CH 4 concentrations in water were determined after headspace extraction according to the methods of Bastviken et al. (2010). Dissolved CH concentration was calculated using 4 Henry's Law adjusted for temperature according to Wiesenburg and Guinasso (1979) following analysis in a Shimadzu GC17A gas chromatograph, modified to contain an online methanizer coupled to a FID detector. Chamber deployments for CH total flux at all sites were performed in the center of 4 the river channels using floating chambers as described by Bastviken et al. (2010). Measurements were made for approximately one hour at each site while drifting, using 7 to 15 chambers separated 1.5 m from each other. The chambers used were of the same type as previously tested and shown to produce non-biased flux values relative to other flux measurement methods (Cole et al., 2010, Galfalk et al., 2013). Using many chambers simultaneously increases the probability of capturing ebullition and allows for the 7 calculation of diffusive flux and ebullition. Total flux and the contribution from diffusive and ebullitive emissions were calculated according to Bastviken et al. (2004, 2010). Samples from chambers were withdrawn using syringes and immediately transferred to 20 ml glass vials filled with salt solution to prevent solubility and capped with a 10 mm thick butyl rubber stopper and an aluminum crimp seal. Gas concentrations were measured by gas chromatography as above. Air temperature, atmospheric pressure and wind speed were measured with a weather station (HOBO; Onset Computer Corporation, Bourne, MA, USA) installed on the boat and water temperature was measured with a pH meter (Orion 290APlus; Thermo Fisher Scientific Inc., Waltham, MA, USA). Flux measurements were done with wind speed ranging from 0.36 to 6 m/s. Diffusive flux calculations Diffusive flux across the water surface into the floating chamber can be described by the equation: ( ), (1) where F is flux (mol m-2d-1), k the piston velocity (m d-1), C is the concentration of CH w 4 measured in the water (mol m-3), and C is the CH concentration in the water at fc 4 equilibrium with the CH partial pressure in the floating chamber (Cole & Caraco, 1998). 4 However, in equation (1) the flux is partially driven by the change in concentration, which will decrease with time in the chambers as the internal concentration increase. Therefore, this simple calculation will underestimate the instantaneous flux rate. In order to reduce this error, we solved for k to estimate instantaneous flux (e.g. the flux for each time step; here time zero (0) to time “t”, F . First, F is expressed as 0-t 8 (2), where n and n are the number of moles in the chamber at time zero and time “t” and A is 0 t the chamber area. Then, the moles are expressed as P and Pt given conversion according 0 to the common gas law (PV=nRT). Finally, the concentration numbers are also expressed as corresponding gas pressure following Henrys Law (C=K P). Hence, by making this h equation continuous, instead of having discrete time steps (e.g. dP/dt instead of P-P ), t 0 Equation (1) could be rewritten as: ( ) ( ) , (3) where dP/dt is the slope of CH accumulation in the chamber (Pa d-1), V is the chamber 4 volume (m3), R is the gas constant (8.314 m3 Pa K-1mol-1), T is temperature (K), P is the w partial pressure of CH in the chamber at equilibrium with C (Pa), P is the partial pressure 4 w 0 of CH in the chamber at time 0 (approximately the same than atmosphere), and K is the 4 h Henry’s Law constant for CH (mol m-3 Pa-1). Thus, 4 ( ) (4) After solving for k using equations 4 the instantaneous flux was calculated using equation (1). The temperature dependence of K was calculated from the Bunsen coefficients given h by Wiesenburg and Guinasso (1979). Ebullition calculations To determine which chambers captured ebullition we used the distributions and variance of the apparent piston velocities as described in Bastviken et al. (2004, 2010). First the calculated (apparent) k values for each chamber were transformed into k values, 600 9