Methane and nitrous oxide emissions from a subtropical estuary (the Brisbane River estuary, Australia)

Science of the Total Environment 472 (2014) 719–729

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Methane and nitrous oxide emissions from a subtropical estuary (the Brisbane River estuary, Australia) Ronald S. Musenze a, Ursula Werner a, Alistair Grinham a,b, James Udy c, Zhiguo Yuan a,⁎ a b c

Advanced Water Management Centre (AWMC), the University of Queensland, Brisbane, Qld 4072, Australia School of Civil Engineering, the University of Queensland, Brisbane, Qld 4072, Australia Healthy Waterways Ltd, P.O. Box 13086, George Street, Brisbane, Qld 4003, Australia

H I G H L I G H T S • • • • •

The estuary is a strong source of atmospheric methane and nitrous oxide. Emissions had strong spatial-temporal variability with unclear seasonal patterns. Dissolved gas saturation comparable to that in tropical rivers and polluted estuaries. Emissions are dominated by N2O, which positively correlated with NOx concentrations. Currently existing models contribute to uncertainty in emission estimates.

a r t i c l e

i n f o

Article history: Received 13 August 2013 Received in revised form 21 October 2013 Accepted 17 November 2013 Available online 10 December 2013 Keywords: Greenhouse gas Methane Nitrous oxide Estuary Subtropical Aquatic emissions

a b s t r a c t Methane (CH4) and nitrous oxide (N2O) are two key greenhouse gases. Their global atmospheric budgeting is, however, flout with challenges partly due to lack of adequate field studies determining the source strengths. Knowledge and data limitations exist for subtropical and tropical regions especially in the southern latitudes. Surface water methane and nitrous oxide concentrations were measured in a subtropical estuarine system in the southern latitudes in an extensive field study from 2010 to 2012 and water–air fluxes estimated using models considering the effects of both wind and flow induced turbulence. The estuary was found to be a strong net source of both CH4 and N2O all-year-round. Dissolved N2O concentrations ranged between 9.1 ± 0.4 to 45.3 ± 1.3 nM or 135 to 435% of atmospheric saturation level, while CH4 concentrations varied between 31.1 ± 3.7 to 578.4 ± 58.8 nM or 1210 to 26,430% of atmospheric saturation level. These results compare well with measurements from tropical estuarine systems. There was strong spatial variability with both CH4 and N2O concentrations increasing upstream the estuary. Strong temporal variability was also observed but there were no clear seasonal patterns. The degree of N2O saturation significantly increased with NOx concentrations (r2 = 0.55). The estimated water–air fluxes varied between 0.1 and 3.4 mg N2O m−2 d−1 and 0.3 to 27.9 mg CH4 m−2 d−1. Total emissions (CO2-e) were N2O (64%) dominated, highlighting the need for reduced nitrogen inputs into the estuary. Choice of the model(s) for estimation of the gas transfer velocity had a big bearing on the estimated total emissions. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Methane (CH4) and nitrous oxide (N2O) are two atmospheric trace gases that have attracted great scientific attention. They are potent greenhouse gases (GHG) with respective global warming potentials of around 25 and 300 times that of carbon dioxide (CO2), on a 100-year horizon (Ehhalt et al., 2001; IPCC, 1995, 2007). Nitrous oxide is also a strong Ozone depleting substance (Ravishankara et al., 2009). Together, these long-lived greenhouse gases contribute nearly 30% of the total warming due to greenhouse gases resulting from anthropogenic ⁎ Corresponding author. Tel.: +61 7 3365 4374; fax: +61 7 3365 4726. E-mail address: [email protected] (Z. Yuan). 0048-9697/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.scitotenv.2013.11.085

influences (Cicerone and Oremland, 1988; IPCC, 2007). The current atmospheric CH4 concentration is nearly triple its pre-industrial level and N2O is also around 20% higher than its pre-industrial level. Moreover, their concentrations are still on the rise (IPCC, 1995; Rigby et al., 2008). Aquatic systems are likely significant sources for both atmospheric CH4 and N2O (IPCC, 2007; Seitzinger et al., 2000). Coastal systems (and specifically estuaries) are presumed to be a strong aquatic source of emissions. They are estimated to contribute up to 60% and 75% of the respective global oceanic N2O and CH4 emissions (Bange et al., 1994, 1996; Seitzinger and Kroeze, 1998). The reported concentrations and fluxes for both CH4 and N2O are, however, widely ranging (Bange et al., 1996; Upstill-Goddard et al., 2000) especially along climatic

720

R.S. Musenze et al. / Science of the Total Environment 472 (2014) 719–729

regional divides (Seitzinger et al., 2000). Both dissolved gas concentrations and fluxes (especially for CH4) have also been found to have high spatial and temporal variability in coastal systems (Bange, 2006). For estuaries, spatial variability has been specially linked to the activities within the surroundings drained by these systems. For instance, elevated dissolved GHG concentrations and fluxes have been reported in rivers and estuarine segments draining farmlands, within vicinities of industrial and domestic wastewater effluent discharges, and generally high organic matter input terrestrial systems (Beaulieu et al., 2010; Richey et al., 1988; Law et al., 1992; Machefert et al., 2004; Toyoda et al., 2009). On a temporal basis, both concentrations and fluxes of CH4 and N2O are generally higher in the warmer summer season and low in the cold winter season (Beaulieu et al., 2010; Clough et al., 2007; de Angelis and Scranton, 1993). However, the absence of clear seasonal patterns has also been reported (Beaulieu et al., 2008; Stow et al., 2005). This highlights a high degree of systems diversity and the importance of region specific measurements. Attempts have been made at establishing regional and global aquatic GHG emission budgets (Bange, 2006; Cicerone and Oremland, 1988). However, uncertainties and immense challenges still surround these estimates (Cicerone and Oremland, 1988; Nevison et al., 1995). One such major challenge is that there are limited emission measurements (Bange, 2006; Beaulieu et al., 2010; US EPA, 2010) from tropical and subtropical aquatic systems, especially in the southern latitudes, despite the fact that aquatic systems in tropical and subtropical regions have been reported to be strong sources of GHGs (Bastviken, 2009; Richey et al., 1988; Koné et al., 2010). Additionally, many emission estimates are often biased by limited spatial and temporal coverage (Nevison et al., 1995). Given the big range in source strengths with the apparent spatial and temporal variability, it is now evident that the knowledge gap due to lack of a good understanding of regional emissions hampers effective global GHG budgeting. Emission studies from the currently understudied systems will help bridge this gap and improve GHG accountability. This study is aimed at quantifying CH4 and N2O emissions from a subtropical estuarine system — the Brisbane river estuary. Special attention was paid to assessment of both temporal and spatial variability. Measurements were undertaken at 18 monitoring stations along the

estuary over a period of two years (October 2010 (spring) to August 2012 (winter)). CH4 and N2O fluxes were estimated using the thin boundary layer approach with the gas transfer velocity estimated on the basis of both wind speed and bottom flow turbulence.

2. Materials and methods 2.1. Physical setting and monitoring stations The Brisbane River (344 km) is the longest river in Southeast Queensland, Australia. It is dammed in the upper reaches forming lake Wivenhoe and meanders through the city of Brisbane before discharging into Moreton Bay. Heavy dredging for bottom sand extraction has led to increased riverbed and bank erosion, high turbidity, increased sedimentation in Moreton Bay, and changes in tidal hydraulics. Dredging extended the river's tidal influence limit from 16 km to 85 km upstream (O'Brien et al., 2001). The river has a catchment area of 13,100 km2 (Allen et al., 2011). The studied estuarine section (86 km) (Fig. 1) has a stream network length of 2475 km formed by 19 tributaries and creeks (HWPL, 2010). The estuary also receives effluents from seven wastewater treatment plants (wwtp), all of which perform biological chemical oxygen demand (COD) and nitrogen removal with the discharges of 40–100 mg COD L−1 (Law, unpublished data) and 2.2–50 mg N L− 1 (Farre et al., 2010), respectively. The estuary has also been at the centre of major flooding events with the January 2011 flood being the most recent devastating one. For the purpose of this work, we have divided the estuary into three sections: lower section (0–33 km), middle section (33–60 km) and upper section (60–86 km). Measurements were done from 18 stations within the estuary (Fig. 1). Sixteen of these stations are part of the bigger monitoring network established under the environmental health monitoring programme (EHMP) – Department of Environment and Resources Management (DERM) – Queensland Government, being used for monthly estuarine water quality monitoring. The 2 nonEHMP stations were established to monitor the impact of the Bremer River at the confluence with the Brisbane River estuary. All stations are described by their location (as distance) from the sea.

Fig. 1. Map of the Brisbane River estuary. Solid dots are stations that were used for field sampling. Numbers are relative distances (km) from the estuary's mouth/sea. Inset is the map of Australia showing location of the study area.

R.S. Musenze et al. / Science of the Total Environment 472 (2014) 719–729

2.2. Sampling campaign design for measuring dissolved CH4 and N2O concentrations and other physicochemical factors To capture temporal variability of CH4 and N2O, we conducted 7 fieldtrips covering different seasons within the study period. Monitoring campaigns were done in October 2010 (spring-10), February 2011 (summer-11), August 2011 (winter-11), November 2011 (spring-11), February 2012 (summer-12), May 2012 (autumn-12) and August 2012 (winter-12). All sampling started 1 h after the high tide from the estuary's mouth continuing upstreams. At all EHMP stations, we measured dissolved CH4 and N2O concentrations (Cobs) and a suite of physico-chemical parameters including temperature, pH, turbidity, dissolved oxygen (DO), electro-conductivity (EC), nitrogen (total, ammonia and oxidised (NOx)), dissolved organic carbon (DOC) and phosphorous. All these parameters, except for nutrients, were also measured at the 2 non-EHMP stations. All physicochemical parameters monitored at the EHMP stations were measured monthly from October 2010 for 23 months as to obtain their temporal trends. Additional sampling campaigns were conducted to assess the potential influence of tides on the dissolved CH4 and N2O concentrations (Cobs) and the uncertainty associated with the use of spot (once in a season) measurements for seasonal flux estimations. While investigating the influence of tides, measurements were done on a full tidal cycle on 18th July 2011 at a station 30 km from the sea. The tidal level was locally referenced at the station with a baseline at the Brisbane Bar using the bureau of meteorology's tide reference schedules for Queensland (http://www.bom.gov.au/cgi-bin/oceanography/tides/ tide_predications.cgi). Sampling was done every 30 min for 13 h. Assessment of the uncertainty associated with spot measurements was done at one station by sampling twice (taking triplicate measurements on each sampling) every 1–3 days for six weeks (October– December 2011). First set of samples was taken at a given time of the day (10:00 am) and the second 1 h after the high tide. We used this study's findings to estimate uncertainty associated with spot measurements in estimating seasonal fluxes. 2.3. Determination of dissolved CH4 and N2O concentration (Cobs) and other water quality parameters Surface water for determination of dissolved CH4 and N2O concentrations was collected from depths not exceeding 0.2 m. Samples were transferred with a syringe and needle into Exetainers® (Labco, Wycombe, UK) for dissolved gas measurement, in triplicates (n = 1 for February 12). Details of the sampling procedure and handling are given in supporting information S1. Dissolved gas extraction was done by single-phase headspace equilibration with ultra high purity (UHP) nitrogen of the same quality used as the GC make-up gas, in an approach similar to that described by Upstill-Goddard et al. (1996). An inflatable glove bag was used for exetainer headspace pressure balancing to atmospheric level. Exetainer headspace gas analysis was done with a 7890A GC system complete with an autosampler (Agilent technologies, Santa Clara, CA, USA). The machine was configured for simultaneous analysis of CH4 and N2O on separate Flame Ionisation Detector (FID) and microElectron Capture Detector (ECD), respectively. Details of the system configuration, calibration standards and measurement procedures are included in SI–S2. Henry's law was used to determine the dissolved gas concentration using the determined headspace concentration and sample volume in the exetainer as described elsewhere (e.g. Alberto et al., 2000; Wang et al., 2009). Physicochemical water quality parameters (pH, DO, EC, temperature and turbidity) were measured using a YSI 6920 multi parameter probe (YSI incorporated, Ohio, USA). The probe was calibrated on the morning of the measurements and corrected for, if any, drifting at the end of the day. Water depth was measured with sonar depth sounders installed on the research vessels and verified with the YSI probe while taking the

721

physicochemical profiles. Salinity was determined from the relationship with conductivity and temperature using algorithms of the practical salinity scale and the international state of seawater (UNESCO, 1983). Nutrient samples (P, N and C) were collected with 60 ml syringes after rigorously rinsing the syringes at least 3 times. All nutrient samples, except for total N, were filtered with 0.45 μm PES filters (Merck Millipore, Victoria, Australia) into collection bottles. All bottles were rinsed with either unfiltered or filtered samples accordingly, prior to filling. All samples were kept on ice and transported in the dark to the lab for analysis or storage (at − 20 °C) before testing. Phosphorus − + (dissolved reactive), NO− 3 , NO2 and NH4 were analysed with an automated LACHAT 8000QC flow injection analyser (FIA) (Lachat instruments, Colorado, USA). Total nitrogen samples were analysed in accordance with APHA (1998). Non-purgeable DOC was analysed by high temperature oxidation and infrared desorption detection. 2.4. Flux estimation using the Thin Boundary Layer Equation Throughout the manuscript the term ‘flux’ has been used to refer to the areal water–air gas exchange rate of a gas of concern (mg m − 2 d − 1 ), while the term ‘emission’ is used to refer to the weighted product of the CH4, N2O or both CH4 and N2O given off from the water surface with due consideration of their respective global warming potentials (GWP). In its use, ‘emission’ carries the conventional unit of carbon dioxide equivalents (CO2-e). To determine the water–air fluxes, we used the Thin Boundary Layer Equation (TBLE) (Eq. (1)) where the flux is estimated using a gas transfer velocity coefficient, k, and the concentrations gradient between the surface water and the overlying atmospheric air. This approach has been widely used and description of the key underlying principles has been previously made (e.g. Wanninkhof, 1992). F ¼ kðC obs −C sat Þ −2

ð1Þ −1

−1

where F (μg m d ) is the flux and k (m d ) is the normalized gas transfer velocity; Cobs (μg m−3) is surface water dissolved gas concentration and Csat (μg m−3) is atmospheric equilibrium/saturation concentration at the prevailing in situ temperature and salinity. Cobs was measured, while Csat was calculated using Henry's Law by assuming the atmospheric concentration of CH4 and N2O. For the calculation of Csat, N2O solubility was determined using the approach of (Weiss and Price, 1980), while CH4 solubility was determined using the coefficients of Yamamoto et al. (1976). Atmospheric N2O mixing ratio was assumed to be the 2011 projected global average (325 ppb) using a growth rate of 0.25% pa since 1990, while the atmospheric CH4 concentration was assumed to be 1800 ppb (Dlugokencky et al., 2003; IPCC, 2007). This assumption was also informed by atmospheric air measurements done in spring 2010, summer 2011, winter 2011 and spring 2011 at the various measurement stations, which produced mean values of 345.3 ppb (standard error = 2.8 ppb) and 1756.9 ppb (standard error = 24.4 ppb), for N2O and CH4, respectively, which are very close to the respective global averages. For the estimation of k, we used wind and currents-based models. k was then estimated as a sum of kwind and kcurrents (details below). The degree of saturation of N2O and CH4 in surface waters (in percentage) relative to the atmospheric equilibrium concentrations was determined as in Eq. (2): % N2 O or CH 4 saturation ¼

C obs  100%: C sat

ð2Þ

2.5. Estimation of the gas transfer velocity (k) and flux (F) For the estimation of the gas transfer velocity due to wind, kwind, five wind-based models as described by Wanninkhof (1992) (hereafter Wan 92) (Eqs. (4) & (5)), Liss and Merlivat (1986) (hereafter LM 86)

722

R.S. Musenze et al. / Science of the Total Environment 472 (2014) 719–729

(Eq. S1–S3), Ro and Hunt (2006) (hereafter RH 06) (Eq. S4), Nightingale et al. (2000) (hereafter N 00) (Eq. S5) and Raymond and Cole (2001) (hereafter RC 01) (Eq. S6) were used. Wan 92 is the most widely used of these models (Zhang et al., 2010). All approaches require wind speed from the study area measured over the study period. Wind speed data was obtained from the Bureau of Meteorology (BoM) as recorded by the Brisbane automatic weather and climate station (Brisbane: 040913). Wind data was logged every 10 min and averaged at 30 min intervals throughout the study period. The 30 min averages were then integrated over the entire measurement season as to cater wind speed variability. This approach gives seasonally representative wind-based flux estimates (Koné et al., 2010). Wind speed was normalized at 10m above the ground using the logarithmic wind profile relationship (Eq. (3)) (e.g. Clark et al., 1994): U 10 ¼

lnð10=Z o Þ U lnðZ=Z o Þ z

ð3Þ

where U10 is the normalized wind speed at 10 m (above the water surface) of the measured wind speed Uz (m s− 1) at height Z (m) above the roughness height Zo (m). We assumed neutral conditions and stability of the boundary layer for which the typical value for Zo is 2 × 10− 4 m (Clark et al., 1994; Large and Pond, 1981). The use of the Wan 92 model (Eq. (4)) for flux estimation is explained below as an example, the use of other models is further described in SI: 2

−0:5

k ¼ a:U 10 ðSc =660Þ

ð4Þ

where a is a constant dependent on the duration of wind speed measurement with values of 0.39 and 0.31 for climatological (long-term) and spot wind speeds, respectively (Wanninkhof, 1992). SC (Schmidt number) for CH4 and N2O, which are dependent on the in-situ temperatures and salinity, were determined from the predictive expressions of Wanninkhof (1992) (Eq. S7–S9). From Eqs. (1), (3) and (4), by assuming that the measured surface water dissolved gas concentrations are representative, within acceptable limits over the flux estimations period (T), the total flux due to wind F(T)wind, estimated by the Wan 92 model (Eq. (5)), over the season would then be: −0:5

F ðT Þwind ¼ 0:39ðSc =660Þ


 Z lnð10=Z o Þ 2 T 2 ðC obs −C sat Þ U z dt: lnðZ=Z o Þ 0

ð5Þ

The flux component due to bottom flow turbulence was estimated using the expression of O'connor and Dobbins (1958) normalized to Sc600 with compensation of in-situ conditions of temperature and salinity (Eq. (6)). This relationship has been widely used and is generally accepted as giving an adequate estimation of the gas transfer velocity induced by currents (Zappa et al., 2003). Three other models (e.g. Langbein and Durum, 1967; Owens et al., 1964; Wilcock, 1984), were also used in estimating fluxes (see SI–S3 for details). −0:5

F ðT Þcurrents ¼ 1:829ðSc =600Þ

Z T 0:5 −0:5 ðC obs −C sat Þ w h dt 0

ð6Þ

Current velocity w (cm s−1) and depth, h (m) at 20 minute resolutions at the respective monitoring stations were obtained using the CSIRO RWQM3 real time model (CSIRO, 2010). Current data was available for the period 6th July 2011 to 29th February 2012. The data in this period were also used to cover corresponding seasons of the remaining campaigns for which current data were unavailable. The summer and winter data was averaged to represent current velocities for autumn 2012. Sc was calculated using the formulations of (Wanninkhof, 1992) as in Eq. S7–S9.

Total flux was then determined as in Eq. (7). This is synonymous with kwind + kcurrents (e.g. Chu and Jirka, 2003; Upstill-Goddard, 2006). F ðT Þ ¼ F ðT Þwind þ F ðT Þcurrents

ð7Þ

To estimate total seasonal fluxes, the estuary was divided into 17 segments around the 18 monitoring stations to obtain an assumed area of influence for the respective stations. Segmentation was made according to factors likely to impact on water quality (e.g. location of point sources, creeks and wwtp outfalls), station locations and inter-station length (e.g. Rosamond et al., 2012). Segment areas were determined with ImageJ 1.45 s (Wayne Rashband National Institute of health, USA). The estimated fluxes at the respective stations (Eq. (7)) were then multiplied with their corresponding segment areas to obtain the respective segment's seasonal emissions, the summation of which would give the estimated total estuarine emissions. The whole estuarine reach average seasonal fluxes (area weighted) were obtained by dividing the total seasonal emissions by the total area and number of days in the season. To determine the average annual spatial fluxes variability, we averaged corresponding seasonal fluxes at the respective monitoring stations before averaging across the seasons still at the respective monitoring stations. 2.6. Statistical analyses We tested for differences between the respective CH4 and N2O data from the different measurement campaigns. The CH4 and N2O data were pooled into separate groups for the different measurement campaigns. Homogeneity of variance tests (Bartlett Chi-square) were performed showing heterogeneity of variance even after square root and log transformations. We then performed non-parametric Kruskal–Wallis ANOVA. To understand the influence of the measured physicochemical factors on the observed degree of CH4 and N2O saturation, measured CH4 and N2O percent saturation from the different measurement campaigns was pooled together with the corresponding physicochemical parameters (surface water salinity, temperature, DO, pH, turbidity, total N, NOx and NH+ 4 ) as the independent variables and a series of correlation and simple linear regressions analysed. Only physicochemical data measured at the time of dissolved gas sampling were used during these analyses. Correlation of CH4 and N2O fluxes with the degree of CH4 and N2O saturation, respectively, was analysed by pooling estimated station fluxes from the different measurement campaigns with their corresponding CH4 and N2O saturation levels. All statistics were performed on mean station data (e.g. Koné et al., 2010). All analyses were performed with StatPlus®mac:2009 5.8.3.8 (AnalystSoft Inc., USA). 3. Results 3.1. Dissolved N2O and CH4 concentrations and degree of saturation The estuary was supersaturated with both N2O and CH4 with respect to atmospheric equilibrium concentrations during all monitoring campaigns (Fig. 2). Dissolved N2O concentrations varied between 9.1 ± 0.4 and 45.3 ± 1.3 nM while the percentage saturation varied between 135 and 435%. CH4 concentrations varied between 31.1 ± 3.7 and 578.4 ± 58.8 nM while the percentage saturation varied between 1210 to 26,430%. In most measurement campaigns the degree of saturation for both N2O and CH4 increased from the estuary's mouth to the upper sections of the estuary. CH4 had a higher upstream gradient than N2O, except for summer 2011 (Feb-11). Elevated percent N2O saturation was often found at ca. 64 km from the estuary's mouth. Further upstream, a decrease of percent N2O saturation was always observed. The Bremer River (ca. 73 km) always had elevated N2O concentrations compared to the Brisbane River at the confluence of the two rivers. It was also found with slightly higher CH4 concentrations compared with the dissolved CH4 concentration in the surface water of the Brisbane River at the confluence of the two rivers.

R.S. Musenze et al. / Science of the Total Environment 472 (2014) 719–729

(a)

(b)

(c)

(d)

723

Fig. 2. Spatial and temporal variability of N2O and CH4 in the Brisbane estuary. (a) N2O in 2010–2011, (b) N2O in 2012, (c) CH4 in 2010–2011, and (d) CH4 in 2012. (Error bars are standard deviation, n = 3, n = 1 for Feb 11. Dashed line is 100% saturation — atmospheric equilibrium saturation).

There was no clear seasonal pattern for both N2O and CH4 concentrations over the investigations period. However, significant differences existed for the respective degrees of saturation for both N2O (Hdf,n = 54.396,124, p b 0.001) and CH4 (Hdf,n = 54.056,124, p b 0.001) amongst the different measurement campaigns. Overall, measurements in winter 2011 (Aug-11) found higher dissolved N2O in the estuary compared to all monitoring campaigns, except for summer 2011 (Feb-11) in the middle sections (28.7–57.3 km) where percent N2O saturation formed a distinct peak (Fig. 2a). Spring 2011 (Nov-11) had the highest spatial variability of percent N2O saturation while the least spatial variability was in spring 2010 (Oct-10) when N2O had a generally flatter profile. Variability between the different measurement campaigns was also higher in 2011 (Fig. 2a) compared to 2012 (Fig. 2b) when respective station percent N2O saturation for the three different seasonal monitoring campaigns were closely similar. Dissolved CH4 (Fig. 2c,d) showed a slightly different pattern from that of N2O along the estuary. Percent CH4 saturation generally had a higher upstream gradient than N2O in all the monitoring campaigns except in summer 2011 (Feb-11) (Fig. 2c), when it had a near levelled out profile. Although there were no clear seasonal patterns for the measured percent CH4 saturation, significant differences (Hdf,n = 54.056,124, p b 0.001) amongst the different measurement campaigns were eminent. Winter 2011 (Aug-11) had the highest CH4 saturation values in all the monitoring campaigns of 2010–11 up to 64 km where percent saturation for spring 2011 (Nov-11) exceeded it. The winter 2011 percent saturation values were also higher than most 2012 measurements except for summer 2012 in the upper sections and close to the Bremer River confluence in winter 2012 (Fig. 2d). Other than in summer 2011, percent CH4 saturation peaks were observed in all the measurement campaigns, but at different points in the middle and upper sections of the estuary. The percent CH4 saturation gradients along the estuary were generally lower in 2011 (Fig. 2c) compared to 2012 (Fig. 2d).

3.2. Water–air CH4 and N2O fluxes To assess if the spot measurements of CH4 and N2O concentrations measured in the respective sampling campaigns are representative for a specific time period (e.g. day and season) we conducted two studies.

In the first study, the variability of CH4 and N2O over the tidal cycle was assessed at a station 30 km from the sea. Both dissolved N2O and CH4 showed minimal variability over the tidal cycle at the investigated station (Fig. S1) indicating that tides did not appreciably affect the dissolved gas concentrations at the monitored station. In the second study at the same station, variability of CH4 and N2O was investigated over a 6-week period across three months in spring 2011. Respective CH4 and N2O concentrations at the two sampling time points from the 6-weeks measurements were closely similar. Though variations in concentrations were observed amongst the different monitoring days, there was no particular pattern over this period. The N2 O concentration at the measurement station was 13.0 ± 2.2 nM while CH4 was 112.4 ± 37.6 nM (average ± stdev, n = 45). The standard errors on the averaged concentrations for both CH4 and N2O were below 5.0%. These results suggest that; (1) tides did not contribute to pronounced changes in the dissolved CH4 and N2O concentrations, and (2) both CH4 and N2O did not show a pronounced variation over the measurement period. This limited variability of the dissolved CH4 and N2O concentrations over the tidal cycle and over the 6-week period indicated that spot measured surface water concentrations could be used for estimation of daily and seasonal fluxes, within uncertainty revealed by the 6-week data. We used 5 wind-based models and 4 currents-based models to estimate the gas transfer velocities and eventually fluxes. Fig. 3 is a comparison of gas transfer velocities estimated using the selected models. There were wide disparities (2.0 to 8.0 folds) in the estimated transfer velocities (Fig. 3a). LM 86 always gave the lowest estimates while RC 01 gave the highest estimates. Using wind speeds from across all seasons during the study period, gas transfer velocity estimates with different wind models would differ by 2.0 to 11.6 folds. Though disparities also existed between gas transfer velocity estimates with different currents models (Fig. 3b), they were much smaller (1.2 to 2.2 folds) than those for the wind based models. LD 67 always gave the lowest estimates while MOD 84 always gave the highest estimates. Based on this assessment, we present the k and subsequently flux estimates obtained using Wan 92 and OD 58 models. These two models give intermediate gas transfer velocity estimates relative to the other selected models (Fig. 3a,b) and are also the most widely used. We also

724

R.S. Musenze et al. / Science of the Total Environment 472 (2014) 719–729

(a)

(b)

(c)

Fig. 3. Comparison of gas transfer velocities estimated using; (a) different wind-based models, (b) currents-based models and (c) wind and currents models — spatial variability of the total gas transfer velocity (kwind + kcurrents) along the estuary based on Wan 92 and OD 58. One day wind, currents, and dissolved gas concentrations on 21st February 2012 at 41.9 km (summer 2012) were used. Error bars are kmin to kmax values. kmin was estimated with a combination of LM86 and LD 67. kmax was estimated with RC 01 and MOD 84 (wind models: LM 86 — Liss and Merlivat (1986), Wan 92 — Wanninkhof (1992), RH 06 — Ro and Hunt (2006), N 00 — Nightingale et al (2000), RC 01 — Raymond and Cole (2001). Current models: OD 58 — O'connor and Dobbins (1958), LD 67 — Langbein and Durum (1967), OW 64 — Owens et al. (1964), MOD 84 — Modified O'Connor and Dobbins model - Wilcock (1984)).

present the expected minimum and maximum flux estimates, obtained through a combination of the LM 86 and LD 67, and the RC 01 and MOD 84 models respectively, to provide a broad range of these estimates. To calculate k, we used wind data measured close to the middle sections. There was minimal inter-seasonal wind speed variability over the investigation period (results not included). Current velocities varied with location in the estuary and with time but integrated current velocities in the respective estuarine segments on different days and amongst measurement seasons were closely similar. Oneday currents measurements would also closely estimate seasonal fluxes (Fig. S2). The total gas transfer velocity was calculated as the sum of kwind and kcurrents. Due to wind and currents variability along the estuary, the gas transfer velocity from the different estuarine sections also spatially varied with the highest transfer velocities being in the upper sections (Fig. 3c), due to the higher impact of bottom flow induced turbulence. Generally, water–air fluxes showed a similar spatial pattern to that of the observed degree of saturation of the dissolved CH4 (p b 0.001, r2 = 0.88) and N2O (p b 0.001, r2 = 0.79). Over the measurements period, station fluxes varied from 0.33 to 27.89 mg CH4 m−2 d−1 and 0.10 to 3.37 mg N2O m−2 d−1. There was a general increase of emissions from the estuary's mouth upstreams with fluxes being highest in the upper sections (Figs. 4, 5). In summer 2012, for instance, fluxes from some of the upper section stations were nearly 7 times the fluxes from some of the lower section stations (Fig. 5f, see also Fig. S5). Consequently, 25% of the total emissions (t CO2-e) were from the upper section, which is only 10% of the estuarine area. There were recognisable temporal patterns in the estimated seasonal fluxes of CH4. Weighted average seasonal fluxes varied between 1.75 and 5.63 mg CH4 m−2 d−1. Fluxes increased from spring 2010 (Fig. 4b) until peaking in summer 2012 (Fig. 4f, see also Fig. S4). Whole estuary N2O area weighted average seasonal fluxes were between 0.32 and 0.70 mg N2O m−2 d−1. No specific seasonal patterns in N2O fluxes were observed (Fig. 5). N2O fluxes were nonetheless highest in summer 2011 (Fig. 5c) and generally higher in 2011 (Fig. 5c–e) compared to 2012 (Fig. 5f–h, see also Fig. S4).

Though N2O fluxes were always smaller (up to 10-fold) than CH4 fluxes, weighted total emissions for the study period were N2O dominated (64% (40–80%)). It is only in summer 2012 that CH4 emissions (59%) were higher than N2O emissions (Fig. S3a) and only then did N2O contribute below 60% of the total emissions. 3.3. Emissions from the estuary Total emissions (CO2-e) for the estuary over the study period totalled 3400 t CO2-e and were N2O (64%) dominated. The selection of the models used to estimate k would substantially affect these estimates. Using different combinations of all the selected models, estimated emissions for the period would vary between 2160 to 5090 t CO2-e. Using the models Wan 92 and OD 58, seasonal emissions varied between 340 and 670 t CO2-e (Fig. S3). By averaging across seasons over the entire study period, we estimated annual emissions to be 1914 t CO2-e. If emission estimates from a single season were upscaled to derive annual emissions, the estimated annual emissions would vary between 30% (underestimation) to 140% (overestimation) of the annual average emissions. 3.4. Physicochemical parameters and correlation analysis Some physicochemical parameters e.g. total nitrogen, oxidised nitrogen (NOx) and ammonium (Fig. 6), and salinity (Fig. S6) had strong spatial and temporal variability. Both total nitrogen and NOx increased upstream the estuary up to around 64 km forming a distinct plateau between 40 and 70 km. Some seasonal measurements found nitrogen peaks around 40, 64 and 70 km stations (Fig. S7). Ammonium also peaked around 12.9 and 70 km, and was mainly higher in the lower sections. Overall, nitrogen concentrations in the estuary were highest in summer 2011 following the floods (Jan and Feb-11, Fig. S8). There is a strong salinity gradient along the estuary. The estuary's mouth is around 35 psu while the upper sections have fresh water (Fig. S6). Generally, the wet summer seasons have low salinities suggesting dilution by the inflows. The estuary was generally of low turbidity especially in the lower sections. However, turbidity increased in summer 2011

R.S. Musenze et al. / Science of the Total Environment 472 (2014) 719–729

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

725

Fig. 4. Whole estuary CH4 average annual and seasonal fluxes. n = 3, n = 1 for Feb 11. Annual average estimates were obtained by first averaging corresponding seasons before eventually averaging all seasons. Flux estimates are based on Wan 92 and OD 58. Error bars are lowest-highest flux estimates based on combination of LM 86 and LD 67 (lowest) and RC 01 and MOD 84 (highest). Spring is Sept–Oct, summer is Dec–Feb, autumn is Mar–May and winter is Jun–Aug.

following the floods with a peak value of 670 NTU at 52 km from the estuary's mouth (Table 1). DOC measured at two stations, one in the lower sections (12.9 km) and one in the middle sections (46.6 km), was also higher in the middle sections than in the lower sections (Fig. S9). Other measured parameters (temperature, DO and pH) showed minimal spatial variability (Table 1). Percent CH4 saturation negatively related with salinity (p b 0.001, r2 = 0.23), and positively related with turbidity (p b 0.001, r2 = 0.1). Percent N2O saturation also negatively related with salinity (p b 0.001, r2 = 0.14) and positively with turbidity (p b 0.001, r2 = 0.1). Percentage N2O saturation also increased with both NOx (p b 0.01, r2 = 0.55) and total nitrogen concentrations (p b 0.01, r2 = 0.53).

4. Discussion 4.1. The estuary is a strong source of both N2O and CH4 The estuary was supersaturated with both N2O and CH4 during all the monitoring campaigns. The observed degree of N2O saturation is higher than previously observed in a number of estuarine systems (see compiled lists in Bange et al. (1996) and Zhang et al. (2010)) and in range but on the lower tail end of some of the world's polluted river systems (Table S1). CH4 concentrations in estuarine systems have been reported with a broad range of concentrations (5 − 1.2 × 104 nM) in surface waters (e.g. Upstill-Goddard et al., 2000). The upper tail end is

726

R.S. Musenze et al. / Science of the Total Environment 472 (2014) 719–729

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

Fig. 5. Whole estuary N2O average annual and seasonal fluxes. n = 3, n = 1 for summer 2011. Annual average estimates were obtained by first averaging corresponding seasons before averaging all seasons. Flux estimates are based on Wan 92 and OD 58. Error bars are lowest-highest flux estimates based on combination of LM 86 and LD 67 (lowest), and RC 01 and MOD 84 (highest). Spring is Sept–Oct, summer is Dec–Feb, autumn is Mar–May and winter is Jun–Aug.

considered exceptional and associated with organic-rich systems such as tributaries in the Amazon. Most measurements from different climatic regions have reported a concentrations range of 10–190 nM (e.g. Biswas et al., 2007; Ferrón et al., 2007). A study by (Richey et al., 1988) in the Amazon main stem also found concentrations within this range (180 ± 30 nM (excl. flood plain)). Our results (31–600 nM CH4 ; 1210–26,430% saturation) are on the higher end of these typical measurements. They are more comparable to measurements from tropical systems (e.g. 48–870 nM; 2221–38,719% saturation; Koné et al., 2010) with a high organic matter input and others from some polluted estuarine systems (Table S1). Our measurements found higher DOC in the middle sections than in the lower sections. Likely major nutrient sources in this estuarine section include wastewater treatment plants (wwtp) effluent discharges and inputs from some creeks. It is

also likely that the estuary receives a lot of organic matter input especially from the catchments in the upper sections, which drain livestock and agricultural farmlands. Labile organic matter availability in this estuarine section conceivably supports high conversion rates resulting into the high CH4 concentrations and degree of saturation measured, especially in the upper and middle sections. Our N2O flux estimates are comparable to fluxes from comparably large rivers such as the Ohio River (mean = 0.3–0.4 mg N2 O m − 2 d− 1 ) (Beaulieu et al., 2010), Hudson River (mean = 0.2 mg N2O m − 2 d− 1 ) (Cole and Caraco, 2001), eutrophic river systems such as San Joaquin River (0.6–5.0 mg N2 O m− 2 d− 1 ) (e.g. Hinshaw and Dahlgren, 2012) and the Shanghai river network (1.1–2.71 mg N2O m− 2 d− 1, e.g. Yu et al., 2013). The relatively high nutrient (especially nitrogen and phosphorous) levels (e.g. Fig. S15, see also HWPL, 2012) is likely

R.S. Musenze et al. / Science of the Total Environment 472 (2014) 719–729

Fig. 6. Trends of total N (closed circles), NOx-N (triangles) and NH4-N (open squares) in the Brisbane estuary measured monthly from spring 2010 to winter 2012. Values are averages (±SE, n = 23).

one of the key factors for the found high concentrations and percentage saturation values for N2O, given that the degree of N2O saturation positively correlated with the estuarine nitrogen concentrations (e.g. Yu et al., 2013). For both measured percentage saturations and water–air fluxes, our results show that subtropical estuaries may match (and in certain cases surpass) estuaries from other climatic regions as sources of GHGs. Our flux estimates are well within the range for wetlands, which are a recognised strong source of GHGs. CH4 fluxes from Australian wetlands and mangroves have been estimated to be between 0.05 and 290 mg m−2 d−1 while N2O fluxes range between 0.08 and 0.55 mg m− 2 d−1 (Dalal and Allen, 2008). A number of other mangrove and wetland systems also fall within this broader range (e.g. Biswas et al., 2007). This comparability therefore suggests that subtropical estuaries could be a significant but neglected source of GHGs that ought to be considered especially for effective state and regional greenhouse gas budgeting. 4.2. There is strong temporal and spatial variability of emissions from the estuary There were significant differences between and amongst measurements for the different monitoring campaigns with observable strong spatial variability for both CH4 and N2O percentage saturations and fluxes. Seasonal patterns were, however, unclear (e.g. Yu et al., 2013) contrary to the findings made in some tropical and temperate systems (e.g. Bange et al., 1998; Ferrón et al., 2007; Koné et al., 2010;). Since the estimated fluxes positively correlated with the degree of dissolved gas saturation, and seasonal wind speed and currents velocity variability was minimal, we attribute the observed estimated flux variability to, majorly, variability of dissolved CH4 and N2O concentrations. Dissolved CH4 and N2O percentage saturation related with several measured factors differently. Salinity varied most and may have contributed to the observed patterns as observed from its relationship Table 1 Physicochemical water quality in the Brisbane estuary. Values are average ± standard deviation (n = 18). Based on the indicated sampling date, when measurements for dissolved gas were also done. October is spring, February is summer, May is autumn and August is winter. Date of Monitoring

Temperature (°C)

DO (%)

6 Oct 2010 15 Feb 2011 10 Aug 2011 9 Nov 2011 21Feb 2012 18 May 2012 14 Aug 2012

22.1 27.5 16.6 25.9 27.8 20.4 16.4

85.6 68.2 99.2 83.8 79.5 73.2 88.1

± ± ± ± ± ± ±

0.9 0.3 0.2 0.8 0.2 0.5 0.5

± ± ± ± ± ± ±

6.1 16.4 6.5 13.2 14.4 26.7 6.3

DO (mg/L)

pH

6.9 5.2 9.1 6.4 6.0 7.4 8.3

7.8 7.5 8.2 7.8 8.0 7.9 8.0

± ± ± ± ± ± ±

0.3 1.2 0.6 0.6 1.0 1.2 0.7

Turbidity (NTU) ± ± ± ± ± ± ±

0.1 0.3 0.1 0.2 0.2 0.1 0.2

60.6 116.8 22.7 44.7 35.9 30.9 10.6

± ± ± ± ± ± ±

40.3 71.0 13.1 34.7 26.3 19.1 13.0

727

with both GHGs. Our study did not, however, specifically investigate the influence of salinity on the production and consumption processes. It thus remains unclear as to whether the observed decrease of concentrations with salinity is a reflection on dilution with seawater or a result of salinity's impact on the processes. Discussion of the impact of salinity on the physiology and biochemistry of the mediating organisms and its general impact on the production and consumption processes' rates is out of scope of this work. These topics have, however, been previously explored and discussed in relative detail (e.g. Abril and Iversen, 2002; Amouroux et al., 2000; Hanson and Hanson, 1996; Rysgaard et al., 1999; Wang et al., 2009; Zhang et al., 2006). Another factor behind the observed variability is the variability in nutrient concentrations. The impact of nutrient availability on the production of both N2O and CH4 has been previously discussed. Increased nitrogen loading and labile carbon availability increase the production of N 2O and CH4 (e.g. Borges and Abril, 2011; Laursen and Seitzinger, 2004; Stow et al., 2005; Usui et al., 2001; Yu et al., 2013). This would, on one hand, explain the differences between the observed seasonal dissolved gas concentrations in 2011 for instance where there were more nutrients throughout the estuary, especially following the floods, than in 2012. On the other hand, it would also likely explain the observed spatial variability. The upper and middle sections of the estuary have high nutrient concentrations likely from inputs from farmlands (e.g. Bremer river) and point sources such as wwtp effluent outfalls. The high NH+ 4 concentrations in the lower sections also suggest: (1) strong NH+ 4 inputs in the lower sections (e.g. Yu et al., 2013) or (2) reduced nitrification rates due to high salinity (e.g. Rysgaard et al., 1999). Other sources such as creeks or even groundwater (e.g. Clough et al., 2007) could, however, also be making a contribution to the observed CH4 and N2O concentrations within the different estuarine sections. Our measurements found the water column always oxygenated indicating limited possibilities for water column methanogenesis and denitrification. The observed CH4 is thus likely to be of mainly sedimentary origin (Borges and Abril, 2011) but N2O could be a result of both water column nitrification and sedimentary denitrification (Seitzinger, 1988). Our measurements did not, however, decouple the different sources and the responsible processes for the production of the observed CH4 and N2O in the surface waters. Further investigations with the use of isotopes may be undertaken to identify, elucidate and quantify the contribution of the different sources of the observed GHGs (e.g. Khalil and Baggs, 2005; Toyoda et al., 2009; Whiticar et al., 1986). In summer 2011, the estuary was at the centre of a major flooding event. These floods are likely to have distorted the estuary's spatial– temporal patterns. The low and flattened CH4 percentage saturation profile together with the low estuarine salinity in the immediate aftermath of flooding is evidence for dilution due to the high inflow of floodwaters. Recently, heavy dilution of CH4 concentrations has also been reported in tropical rivers and lagoons following heavy rainfall events (Koné et al., 2010). Our results also showed increased nutrient concentrations in the estuary following the floods and generally during the wet summer months. The low DO levels during the same periods are a likely outcome of this increased nutrient loading. Koné et al. (2010) and Grinham et al. (2012) have also reported increased nutrient loads in tropical rivers and lagoons, and subtropical storages following heavy rainfall and catchment runoff events. Our observations of increased turbidity following the floods are also in unison with those of Brown et al. (2011) who reported increased sediment transport during and in the immediate aftermath of the floods. This combination of increased nutrients and fresh sediments is likely to have sustained high nutrient cycling rates that led to the observed high CH4 and N2O concentrations (and percentage saturation) say even in winter (2011) where it would otherwise have been expected to be low (e.g. Beaulieu et al., 2010). Studies investigating the impact of flooding on CH4 and N2O emissions from aquatic systems are currently scarce. Monitoring both CH4 and N2O emissions at high temporal resolutions following flooding would provide valuable scientific information on

728

R.S. Musenze et al. / Science of the Total Environment 472 (2014) 719–729

the impact of floods on GHG emissions from aquatic systems and presumably help improve the IPCC emission factors (e.g. IPCC, 2006).

would be a more suitable approach than use of a single model to generate a single flux estimate value. 5. Conclusions

4.3. Dealing with uncertainties in emission estimations There are two main uncertainties in aquatic emission estimations; (1) representativeness of the measured surface water dissolved gas concentrations for the emission estimation period and (2) accurate estimation of the gas transfer velocity, k. Our study has shown an approach to follow to resolve uncertainty associated with representativeness of the measured gas concentrations. However, due to existence of both spatial and temporal variability along the estuarine profile, accurate quantification of this uncertainty would require more repeated measurements from a number of stations in the different sections of the estuary. Though seemingly laborious, it is a feasible way of resolving this uncertainty component. However, uncertainty due to estimated gas transfer velocities remains a key issue (e.g. Nevison et al., 1995). k estimation can be done directly or indirectly. The challenges associated with direct estimation approaches using for example floating chambers, amongst others, have been discussed elsewhere (e.g. Duchemin et al., 1999). Under rough and flowing water conditions indirect approaches, the use of models, is preferred. There are several wind-based models (e.g. Liss and Merlivat, 1986; Nightingale et al., 2000; Raymond and Cole, 2001; Ro and Hunt, 2006; Wanninkhof, 1992) and currentsbased models for k estimation (e.g. Langbein and Durum, 1967; O'connor and Dobbins, 1958; Owens et al., 1964; Wilcock, 1984). A simple linear additive formulation of ktotal = kwind + kcurrents (e.g. A. Borges et al., 2004; A.V. Borges et al., 2004; Chu and Jirka, 2003; UpstillGoddard, 2006) has been suggested to give a good estimate of the resultant gas transfer velocity even though the actual relationship is thought to be more complex (A. Borges et al., 2004; Chu and Jirka, 2003). k estimations based on wind have two uncertainties; (1) due to wind speed variability and (2) inherent weaknesses with used model. We have demonstrated a way of dealing with wind speed variability by integrating wind speed over the entire k estimation period (season). However, our results demonstrate that differences due to the used kwind estimation models can cause up to an order of magnitude difference in the estimated fluxes. The difference between these models' estimates highlights the complexity of the influence of wind on k, the currently existing limited understanding of k parameterisation and generally the uncertainty still surrounding flux estimations using indirect methods. In fact, it is recognised that none of the current wind-based models for the estimation of gas transfer velocities from water surfaces satisfies all field conditions (Liss and Merlivat, 1986; Livingstone and Imboden, 1993; Wanninkhof, 1992). Using kwind alone from the expression of Wanninkhof (1992), our estimated total emissions for the study period were only 30% of the total emissions obtained with the use of kwind + kcurrents. This indicates that currents can be a key factor in the parameterisation of k. To date, only few studies have used the approach of ktotal = kwind + kcurrents to estimate the resultant gas transfer velocity (e.g. A. Borges et al., 2004; A.V. Borges et al., 2004; Hinshaw and Dahlgren, 2012). It is therefore likely that previous emission estimations that used wind only to estimate k in rivers and estuaries may have grossly underestimated emissions. We have also shown that kcurrents is less sensitive to the models used, in comparison to kwind. Unfortunately, no consensus exists on the quality or representativeness of k estimates from the currently existing models. The differences in the model estimates are likely because these models were developed in different environment settings such as laboratory wind tunnels, flowing rivers, streams and estuaries. Similarly, their calibration was with datasets obtained using different methods such as tracers and floating chambers, which may themselves have different accuracy. Uncertainty due to used models thus remains a key issue that requires resolving for effective GHG budgeting. Until this is resolved, it seems the use of multiple models to generate a range of flux estimates

This study has shown that subtropical estuarine systems are strong sources of atmospheric methane and nitrous oxide. Emissions had strong spatial and temporal variability. Fluxes increased from the estuary's mouth upstreams towards the upper reaches. There were, however, no clear seasonal patterns. To adequately estimate total emissions from such an estuarine system would thus need measurements from a number of locations along the estuary covering different seasons. Our results have shown that currents contribute significantly to the gas transfer velocity in the estuary. Studies that use wind speed to estimate emissions in estuaries and rivers may thus be grossly underestimating aquatic emissions and their contribution to the global atmospheric greenhouse gas budgets. Lack of agreement between the currently used models in estimating fluxes creates a cloud of uncertainty around the estimated emissions. It is thus imperative that future efforts are directed towards parameterisation of the gas transfer velocity in aquatic systems. For now, providing a range for the emissions by using a more than one model may be a more reliable approach than use of a single model to generate a single flux value. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.scitotenv.2013.11.085. Conflict of interest This work was funded by the Australian Research Council (ARC), Healthy Waterways Ltd and Seqwater through an industry linkage grant (ARC Linkage project # LP100100325). The study was done with support of technical staff from the Department of Environment and Resources Management — Queensland government. Except where acknowledged within the manuscript, there is no other conflict of interest. Acknowledgements This work was funded by the Australian Research Council (ARC), Healthy Waterways Ltd and Seqwater through an industry linkage grant (ARC Linkage project # LP100100325). Dr Keshab Sharma participated in data processing. We thank Ms Katrin Sturm for the support during the tidal studies. The authors greatly appreciate the support of EHMP–DERM management, tech staff and crews of Yellow Fin, Seratta and Seriops cruises (2010–2012). Additional water quality data was obtained from DERM under the DSITIA single supply licence. References Abril G, Iversen N. Methane dynamics in a shallow non-tidal estuary (Randers Fjord, Denmark). Mar Ecol Prog Ser 2002;230:171–81. Alberto MCR, Arah JRM, Neue HU, Wassmann R, Lantin RS, Aduna JB, et al. A sampling technique for the determination of dissolved methane in soil solution. Chemosphere Global Change Sci 2000;2:57–63. Allen D, Dalal RC, Rennenberg H, Schmidt S. Seasonal variation in nitrous oxide and methane emissions from subtropical estuary and coastal mangrove sediments. Australia. Plant Biol 2011;13:126–33. Amouroux D, Roberts G, Rapsomanikis S, Andreae MO. Biogenic gas (CH4, N2O, DMS) emission to the atmosphere from near-shore and shelf waters of the North-western Black Sea. Estuar Coast Shelf Sci 2000;54:575–87. APHA. Standard methods for the examination of water and wastewater. 20th ed. Washington, DC: American Public Health Association; 1998. Bange HW. Nitrous oxide and methane in European coastal waters. Estuar Coast Shelf Sci 2006;70:361–74. Bange HW, Bartell UH, Rapsomanikis S, Andreae MO. Methane in the Baltic and North Seas and a reassessment of the marine emissions of methane. Global Biogeochem Cycles 1994;8:465–80. Bange HW, Rapsomanikis S, Andreae MO. Nitrous oxide in coastal waters. Global Biogeochem Cycles 1996;10:197–207. Bange HW, Dahlke S, Ramesh R, Meyer-Reil LA, Rapsomanikis S, Andreae MO. Seasonal study of methane and nitrous oxide in the coastal waters of the Southern Baltic Sea. Estuar Coast Shelf Sci 1998;47:807–17. Bastviken D. Methane. In: Gene EL, editor. Encyclopedia of inland waters. Oxford: Academic Press; 2009. p. 783–805.

R.S. Musenze et al. / Science of the Total Environment 472 (2014) 719–729 Beaulieu JJ, Arango CP, Hamilton SK, Tank JL. The production and emission of nitrous oxide from headwater streams in the Midwestern United States. Glob Chang Biol 2008;14:878–94. Beaulieu JJ, Shuster WD, Rebholz JA. Nitrous oxide emissions from a large, impounded river: the Ohio River. Environ Sci Technol 2010;44:7527–33. Biswas H, Mukhopadhyay SK, Sen S, Jana TK. Spatial and temporal patterns of methane dynamics in the tropical mangrove dominated estuary, NE coast of Bay of Bengal, India. J Mar Syst 2007;68:55–64. Borges AV, Abril G. Carbon dioxide and methane dynamics in estuaries. In: Eric W, Donald M, editors. Treatise on estuarine and coastal science. Waltham: Academic Press; 2011. p. 119–61. Borges A, Vanderborght J-P, Schiettecatte L-S, Gazeau F, Ferron-Smith S, Delille B, et al. Variability of the gas transfer velocity of CO2 in a macrotidal estuary (the Scheldt). Estuaries 2004;27:593–603. Borges AV, Delille B, Schiettecatte L-S, Gazeau Fdr, Abril Gl, Frankignoulle M. Gas transfer velocities of CO2 in three European Estuaries (Randers Fjord, Scheldt, and Thames). Limnol Oceanogr 2004;49:1630–41. Brown RJ, Chanson H, McIntosh D, Madhani J. Turbulent velocity and suspended sediment concentration measurements in an urban environment of the Brisbane River Flood Plain at Gardens Point on 12–13 January 2011. The University of Queensland; 2011. Chu C, Jirka G. Wind and stream flow induced reaeration. J Environ Eng 2003;129:1129–36. Cicerone RJ, Oremland RS. Biogeochemical aspects of atmospheric methane. Global Biogeochem Cycles 1988;2:299–327. Clark JF, Wanninkhof RIK, Schlosser P, Simpson HJ. Gas exchange rates in the tidal Hudson river using a dual tracer technique. Tellus B 1994;46:274–85. Clough TJ, Buckthought LE, Kelliher FM, Sherlock RR. Diurnal fluctuations of dissolved nitrous oxide (N2O) concentrations and estimates of N2O emissions from a spring-fed river: implications for IPCC methodology. Glob Chang Biol 2007;13:1016–27. Cole JJ, Caraco NF. Emissions of nitrous oxide (N2O) from a tidal, freshwater river, the Hudson River, New York. Environ Sci Technol 2001;35:991–6. CSIRO. South East Queensland receiving water quality model V3 (RWQM3). Brisbane: CSIRO; 2010. Dalal RC, Allen DE. TURNER REVIEW No. 18. Greenhouse gas fluxes from natural ecosystems. Aust J Bot 2008;56:369–407. de Angelis MA, Scranton MI. Fate of methane in the Hudson River and Estuary. Global Biogeochem Cycles 1993;7:509–23. Dlugokencky EJ, Houweling S, Bruhwiler L, Masarie KA, Lang PM, Miller JB, et al. Atmospheric methane levels off: temporary pause or a new steady-state? Geophys Res Lett 2003;30:1992. Duchemin E, Lucotte M, Canuel R. Comparison of static chamber and thin boundary layer equation methods for measuring greenhouse gas emissions from large water bodies. Environ Sci Technol 1999;33:350–7. Ehhalt D, Prather M, Dentener F, Derwent R, Dlugokencky E, Holland E, et al. Atmospheric chemistry and greenhouse gases. In: Joos MM F, editor. IPCC, 2001: climate change 2001: the scientific basis. Contribution of working group I to the third assessment report of the intergovernmental panel on climate change. Cambridge, UK and New York: Cambridge University Press; 2001. p. 239–87. Farré MJ, Döderer K, Keller J, Poussade Y, M.J.W.G. N-nitrosodimethylamine (NDMA) in purified recycled water. Urban water security research alliance technical report No. 34. Urban Water Security Research Alliance (UWSRA); 2010. Ferrón S, Ortega T, Gómez-Parra A, Forja JM. Seasonal study of dissolved CH4, CO2 and N2O in a shallow tidal system of the bay of Cadiz (SW Spain). J Mar Syst 2007;66:244–57. Grinham A, Gibbes B, Gale D, Watkinson A, Bartkow M. Extreme rainfall and drinking water quality: a regional perspective. WIT Transactions on Ecology and the Environment, 164. WIT Press; 2012183–94. Hanson R, Hanson T. Methanotrophic bacteria. Microbiol Rev 1996;60:439–71. Hinshaw SE, Dahlgren RA. Dissolved nitrous oxide concentrations and fluxes from the eutrophic San Joaquin River, California. Environ Sci Technol 2012;47:1313–22. HWPL. Health of the freshwater, estuarine and marine waterways of South East Queensland. Ecosystem Health Monitoring Program. Annual technical report 2008–09. Brisbane: South East Queensland Healthy Waterways Partnership; 2010. HWPL. South East Queensland Health-e-Waterways 2012 report cards. Brisbane: Healthy Water Ways Limited; 2012. IPCC. Climate change 1995: the science of climate change. Contribution of WGI to the second assessment report of the intergovernmental panel on climate change. In: Houghton JT, L.G.M.F., Callander BA, Harris N, Kattenberg A, Maskell K, Lakeman PEJA, editors. Cambridge, UK: IPCC; 1995. p. 588. IPCC. 2006 IPCC guidelines for national greenhouse gas inventories. In: Egglesto HSBL, Miwa K, Ngara T, Tanabe K, editors. Intergovernmental panel on climate change. Japan: IGES; 2006. IPCC. Climate change 2007: the physical science basis. In: Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller HL, editors. Contribution of working group I to the fourth assessment report of the intergovernmental panel on climate change. Cambridge, United Kingdom and New York, NY, USA: IPCC; 2007. Khalil MI, Baggs EM. CH4 oxidation and N2O emissions at varied soil water-filled pore spaces and headspace CH4 concentrations. Soil Biol Biochem 2005;37:1785–94. Koné Y, Abril G, Delille B, Borges A. Seasonal variability of methane in the rivers and lagoons of Ivory Coast (West Africa). Biogeochemistry 2010;100:21–37. Langbein WB, Durum WH. The aeration capacity of streams. Geol Surv Circ 1967;542. [6 P]. Large WG, Pond S. Open ocean momentum flux measurements in moderate to strong winds. J Phys Oceanogr 1981;11:324–36. Laursen AE, Seitzinger SP. Diurnal patterns of denitrification, oxygen consumption and nitrous oxide production in rivers measured at the whole-reach scale. Freshw Biol 2004;49:1448–58.

729

Law CS, Rees AP, Owens NJP. Nitrous oxide: estuarine sources and atmospheric flux. Estuar Coast Shelf Sci 1992;35:301–14. Liss PS, Merlivat L. Air–sea gas exchange rates: introduction and synthesis. In: Buat-Ménard P, editor. The role of Air–Sea exchange in geochemical cycling. Dordrecht, The Netherlands: Reidel; 1986. p. 113–29. Livingstone DM, Imboden DM. The non-linear influence of wind-speed variability on gas transfer in lakes. Tellus B 1993;45:275–95. Machefert SE, Dise NB, Goulding KWT, Whitehead PG. Nitrous oxide emissions from two riparian ecosystems: key controlling variables. Water Air Soil Pollut Focus 2004;4:427–36. Nevison CD, Weiss RF, Erickson III DJ. Global oceanic emissions of nitrous oxide. J Geophys Res 1995;100:15809–20. Nightingale PD, Malin G, Law CS, Watson AJ, Liss PS, Liddicoat MI, et al. In situ evaluation of air–sea gas exchange parameterizations using novel conservative and volatile tracers. Global Biogeochem Cycles 2000;14:373–87. O'Brien LPM, Holland Ivan, Angela Grice, Tracey D. Tidal Brisbane River. State of South-east Queensland Waterways; 200175. O'connor DJ, Dobbins WE. Mechanism of reaeration in natural streams. Trans Am Soc Civ Eng 1958;123:44. Owens M, Edwards R, Gibbs J. Some reaeration studies in streams. Air Water Pollut 1964;8:469. Ravishankara AR, Daniel JS, Portmann RW. Nitrous oxide (N2O): the dominant ozone-depleting substance emitted in the 21st century. Science 2009;326:123–5. Raymond P, Cole J. Gas exchange in rivers and estuaries: choosing a gas transfer velocity. Estuar Coasts 2001;24:312–7. Richey JE, Devol AH, Wofsy SC, Reynaldo Victoria, Riberio MNG. Biogenic gases and the oxidation and reduction of carbon in Amazon River and floodplain waters. Limnol Oceanogr 1988;33:10. Rigby M, Prinn RG, Fraser PJ, Simmonds PG, Langenfelds RL, Huang J, et al. Renewed growth of atmospheric methane. Geophys Res Lett 2008:35. Ro K, Hunt P. A new unified equation for wind-driven surficial oxygen transfer into stationary water bodies. Trans ASABE 2006;49:1615–22. Rosamond MS, Thuss SJ, Schiff SL. Dependence of riverine nitrous oxide emissions on dissolved oxygen levels. Nature Geosci 2012;5:715–8. Rysgaard S, Thastum P, Dalsgaard T, Christensen P, Sloth N. Effects of salinity on NH+ 4 adsorption capacity, nitrification, and denitrification in Danish Estuarine Sediments. Estuar Coasts 1999;22:21–30. Seitzinger SP. Denitrification in freshwater and coastal marine ecosystems: ecological and geochemical significance. Limnol Oceanogr 1988;33:702–24. Seitzinger SP, Kroeze C. Global distribution of nitrous oxide production and N inputs in freshwater and coastal marine ecosystems. Global Biogeochem Cycles 1998;12:93–113. Seitzinger SP, Kroeze C, Styles RV. Global distribution of N2O emissions from aquatic systems: natural emissions and anthropogenic effects. Chemosphere Global Change Sci 2000;2:267–79. Stow CA, Walker JT, Cardoch L, Spence P, Geron C. N2O emissions from streams in the Neuse River Watershed, North Carolina. Environ Sci Technol 2005;39:6999–7004. Toyoda S, Iwai H, Koba K, Yoshida N. Isotopomeric analysis of N2O dissolved in a river in the Tokyo metropolitan area. Rapid Commun Mass Spectrom 2009;23:809–21. UNESCO. Algorithms for computation of fundamental properties of seawater. Unesco technical papers in marine science; 1983. p. 44. Upstill-Goddard RC, Rees AP, Owens NJP. Simultaneous high-precision measurements of methane and nitrous oxide in water and seawater by single phase equilibration gas chromatography. Deep Sea Res. Part I 1996;43:1669–82. Upstill-Goddard RC. Air–sea gas exchange in the coastal zone. Estuar Coast Shelf Sci 2006;70:388–404. Upstill-Goddard RC, Barnes J, Frost T, Punshon S, Owens NJP. Methane in the southern North Sea: low-salinity inputs, estuarine removal, and atmospheric flux. Global Biogeochem Cycles 2000;14:1205–17. US EPA. Methane and nitrous oxide emissions from natural sources. Washington DC: United States Environmental Protection Agency; 2010. Usui T, Koike I, Ogura N. N2O production, nitrification and denitrification in an estuarine sediment. Estuar Coast Shelf Sci 2001;52:769–81. Wang D, Chen Z, Sun W, Hu B, Xu S. Methane and nitrous oxide concentration and emission flux of Yangtze Delta plain river net. Sci China B Chem 2009;52:652–61. Wanninkhof R. Relationship between wind speed and gas exchange over the ocean. J Geophys Res 1992;97:7373–82. Weiss RF, Price BA. Nitrous oxide solubility in water and seawater. Mar Chem 1980;8: 347–59. Whiticar MJ, Faber E, Schoell M. Biogenic methane formation in marine and freshwater environments: CO2 reduction vs. acetate fermentation-Isotope evidence. Geochim Cosmochim Acta 1986;50:693–709. Wilcock RJ. Methyl chloride as a gas-tracer for measuring stream reaeration coefficients-II: stream studies. Water Res 1984;18:53–7. Yamamoto S, Alcauskas JB, Crozier TE. Solubility of methane in distilled water and seawater. J Chem Eng Data 1976;21:78–80. Yu Z, Deng H, Wang D, Ye M, Tan Y, Li Y, et al. Nitrous oxide emissions in the Shanghai river network: implications for the effects of urban sewage and IPCC methodology. Glob Chang Biol 2013;19:2999–3010. Zappa C, Raymond P, Terray E, McGillis W. Variation in surface turbulence and the gas transfer velocity over a tidal cycle in a macro-tidal estuary. Estuar Coasts 2003;26:1401–15. Zhang G, Zhang J, Xu J, Zhang F. Distributions, sources and atmospheric fluxes of nitrous oxide in Jiaozhou Bay. Estuar Coast Shelf Sci 2006;68:557–66. Zhang GL, Zhang J, Liu SM, Ren JL, Zhao YC. Nitrous oxide in the Changjiang (Yangtze River) Estuary and its adjacent marine area: Riverine input, sediment release and atmospheric fluxes. Biogeosciences 2010;7:3505–16.