Article pubs.acs.org/est
Novel Use of Cavity Ring-down Spectroscopy to Investigate Aquatic Carbon Cycling from Microbial to Ecosystem Scales Damien T. Maher,*,† Isaac R. Santos,† Jasper R. F. W. Leuven,‡ Joanne M. Oakes,† Dirk V. Erler,† Matheus C. Carvalho, and Bradley D. Eyre† †
Centre for Coastal Biogeochemistry, School of Environment, Science and Engineering, Southern Cross University, Lismore, New South Wales 2480, Australia ‡ Faculty of Geosciences, Utrecht University, Utrecht, The Netherlands S Supporting Information *
ABSTRACT: Development of cavity ring-down spectroscopy (CRDS) has enabled real-time monitoring of carbon stable isotope ratios of carbon dioxide and methane in air. Here we demonstrate that CRDS can be adapted to assess aquatic carbon cycling processes from microbial to ecosystem scales. We first measured in situ isotopologue concentrations of dissolved CO2 (12CO2 and 13CO2) and CH4 (12CH4 and 13 CH4) with CRDS via a closed loop gas equilibration device during a survey along an estuary and during a 40 h time series in a mangrove creek (ecosystem scale). A similar system was also connected to an in situ benthic chamber in a seagrass bed (community scale). Finally, a pulse-chase isotope enrichment experiment was conducted by measuring real-time release of 13 CO2 after addition of 13C enriched phytoplankton to exposed intertidal sediments (microbial scale). Miller-Tans plots revealed complex transformation pathways and distinct isotopic source values of CO2 and CH4. Calculations of δ13C-DIC based on CRDS measured δ13C-CO2 and published fractionation factors were in excellent agreement with measured δ13C-DIC using isotope ratio mass spectroscopy (IRMS). The portable CRDS instrumentation used here can obtain real-time, high precision, continuous greenhouse gas data in lakes, rivers, estuaries and marine waters with less effort than conventional laboratory-based techniques.
■
INTRODUCTION Coastal aquatic ecosystems play an important role in the global carbon cycle. It has been estimated that global CO2 emissions from estuaries are equivalent to ∼30% of CO2 uptake in the open ocean and roughly equal to the total continental shelf uptake, even though estuaries account for only ∼0.3% of the combined area.1 Carbon cycling within aquatic systems is driven by complex biogeochemical processes, which in turn control the concentration and partitioning of carbon forms. Elucidating aquatic carbon cycling process rates and pathways has been a key area of research for decades initially driven by ecological studies looking at primary productivity and foodwebs2 and more recently by studies of feedback mechanisms between atmospheric CO2 and aquatic systems.3−5 Development of portable systems for measuring in situ partial pressure of CO2 (pCO2) and more recently CH4 have seen a rapid accumulation of high temporal and spatial resolution data.6−8 This has led to a greater understanding of the magnitude of aquatic CO2 sources and sinks, although the exact mechanisms and pathways of carbon transformation are still poorly constrained. Insights into carbon pathways and drivers are often obtained from carbon isotope analysis.9−12 © 2013 American Chemical Society
Traditionally this has involved collection of discrete samples, which are then analyzed using large laboratory-based isotope ratio mass spectrometers (IRMS), limiting the temporal and spatial resolution of previous studies. With advances in laserbased spectroscopy, it is now possible to collect high quality near-continuous in situ data, enabling collection of stable isotope date with improved spatial and temporal resolution. Cavity ring down spectroscopy (CRDS) is a laser-based spectroscopy technique that uses the optical absorbance characteristics of individual gas species (including isotopologues and isotopomers) to measure concentrations.13 CRDS has been used in a number of atmospheric chemistry applications.14−16 Several recent studies have used a CRDS to measure dissolved inorganic carbon (DIC) and the carbon stable isotope ratio of DIC (δ13C-DIC) in aquatic systems. These studies have revealed high temporal variability in tropical streams associated with precipitation and discharge events17 and diurnal variability in floodplains driven by shifts in the balance of Received: Revised: Accepted: Published: 12938
June 24, 2013 October 17, 2013 October 17, 2013 October 17, 2013 dx.doi.org/10.1021/es4027776 | Environ. Sci. Technol. 2013, 47, 12938−12945
Environmental Science & Technology
Article
borosilicate vials containing 50 μL of saturated HgCl2 solution, with no headspace. Vials were sealed with Teflon-lined septa screw-caps and stored at ∼4 °C in the dark until analysis (within a week). δ13C-DIC was analyzed using an OI 1030W TOC analyzer interfaced with a Thermo Delta V Plus IRMS.23,24 Experiment 2: Time Series. The same air−water equilibrator CRDS instrumentation as described for experiment 1 was deployed at the mouth of a small mangrove tidal creek at Evans Head, Northern NSW Australia (S29.120934°, E153.428208°) for ∼30 h during winter in 2012. The system was identical to the one described for the surveys, with a submersible pump fixed ∼20 cm above the sediment surface, and a length of hose delivering the water to the shower head GED located on the creek bank. A Hydrolab DS5X sonde at the water intake location logged temperature, pH, depth, and salinity. Samples were collected at hourly intervals for δ13C-DIC analysis as described in the previous section. CO2 and CH4 standards were run at the start, middle and end of the time series. Experiment 3: Benthic Chamber. A benthic chamber (see ref 25 for details) was deployed over a dense seagrass bed (Zostera muelleri) at Shaws Bay, Northern NSW, Australia (S28.864601, E153.585920) for 10 h from 6 pm on the 18th December 2012. Water was pumped in a closed loop from the chamber, through a LiquiCel membrane contactor GED, and back to the chamber. Air was pumped in a counter-direction to the waterflow through the GED, through a Drierite column, through the CRDS, and back to the GED in a closed loop. A SAMI-II submersible spectrophotometric pH instrument was placed next to the chamber, with the inlet tube drawing water from within the chamber at 10 min intervals. Temperature and salinity were measured with a temperature conductivity meter (YSI EC300A) at ∼1 h intervals by taking discrete samples from the chamber. Experiment 4: Pulse Chase Enrichment. A sediment core was collected from an intertidal sand shoal adjacent to the chamber experiment described above. The overlying water of the core was drained, and a visual inspection revealed no visible burrows or macrofauna at the surface. A layer of commercially available 13C-labeled lyophilized algal cells (Agmenellum quadruplicatum, ≥98% 13C, Cambridge Isotope Laboratories), which were grown axenically (i.e., without bacteria), was added quickly across the surface of the core, resulting in addition of ∼1000 mg 13 C m−2. This method was intended to create immediate contact between bacteria and the organic matter (OM). After addition of OM, the core was capped with the sediment exposed to the air, simulating low tide exposed conditions. Air flowed in a closed loop from the headspace of the closed core through a Drierite column, through the CRDS and back into the headspace of the core. The experiment was terminated ∼5 min after a detectable 13C enrichment of the CO2 in the headspace was measured. Data Analysis. The data file from the CRDS contains time stamped measurements at ∼1 Hz, with concentrations of individual isotopologues of CO2 and CH4, δ13C values of CO2 and CH4, and 30 s, 2 and 5 min averages of the δ13C values of CO2 and CH4 (along with instrument diagnostic data). For the survey and chamber experiment we used a 1 min running average for all parameters, for the time series we used a 5 min average, and for the pulse-chase experiment we used the raw data. These times were used as a compromise between the measurement error (which is a function of measurement time) and the changes in actual concentrations. The concentrations of CO2 and CH4 measured by the CRDS need to be corrected for the drying of the gas stream prior to analysis. Dry concentrations of CO2 were converted to in situ partial pressure (pCO2) using
autotrophy and heterotrophy.18 In addition, several open ocean studies have used a CRDS to measure pCO2 and the δ13C of CO2 (δ13C-CO2) by measuring equilibrated air from the headspace of a gas exchange device (GED).19,20 In this paper we demonstrate how CRDS may provide insights into aquatic carbon cycling rates and pathways. We present the first quasi-continuous in situ measurements of both CO2 and CH4 partial pressures and their respective carbon stable isotope ratios (δ13C) in a range of aquatic settings. CO2 and CH4 dynamics were investigated in seagrass beds (community scale), a mangrove system and whole estuary (ecosystem scale) and through a pulse chase experiment following 13C-enriched phytoplankton addition to intertidal sediments (microbial scale). In addition we demonstrate in situ δ13C-DIC values can be back calculated from continuously measured δ13C-CO2 values from a GED.
■
EXPERIMENTAL SECTION Instrumentation. A commercially available CRDS (G2201-I Picarro Inc., Santa Clara, CA) was used to determine quasicontinuous concentrations of dissolved 12CO2, 13CO2, 12CH4, and 13 CH4 in a number of experimental setups. The manufacturer guaranteed specifications for δ13C-CO2 and δ13C-CH4 precision (1σ, 5 min average) are 0.16‰ and 0.55‰, respectively, for the concentration range of our experiments, and CO2 and CH4 concentration precision (1σ, 30 s average) is 210 ppb (+0.05% of reading) and 60 ppb (+0.05% of reading), respectively. Four calibration gases of differing CO2 and CH4 concentrations (and δ13C-CO2 and δ13C-CH4 values) were run prior to and following each experiment (see Supporting Information (SI) for more details). To extract the dissolved CO2 and CH4 we used two different gas equilibration devices (GED). The first was a shower-head equilibrator and the second a Liqui-Cel membrane contactor as described elsewhere.21 A second shower-head equilibrator with the headspace open to the atmosphere was connected to the vent of the first to ensure no pressure differential between the atmosphere and the equilibrator (see SI for more details). The showerhead was chosen over the Liqui-Cel GED for the survey and time series experiments to minimize any potential blockage or microbially mediated CO2 production within the GED.21 A laboratory experiment showed no differences in concentrations or δ13C values measured using the two GED systems. Experiment 1: Surveys. The CRDS was placed aboard a small research vessel and was powered by 6 × 100 AH 12 V deep cycle batteries connected to a 1000 W inverter. The vessel was driven at ∼6 km h−1 from the mouth of North Creek estuary to ∼15 km upstream (see ref 22 for study area details). During the survey, water was continuously pumped from a submerged bilge pump (∼50 cm depth) to the shower-head GED and a flow-through chamber containing a calibrated Hydrolab DS5 sonde, which logged salinity, temperature, and pH. The equilibrated air headspace of the GED was pumped through polyethylene lined tubing (1/8 in. i.d., 1/4 in. o.d. Bev-A-Line IV) to a Drierite column, then to the CRDS and returned to the GED in a closed loop. Dissolved 12CO2, 13CO2, 12 CH4, and 13CH4 in the equilibrated air were measured at ∼1 s intervals by the CRDS. A calibrated LiCor Li-7000 was run in line with the CRDS to simultaneously determine CO2 concentration. During the survey 28 grab samples of water for δ13C analysis of the dissolved inorganic carbon (δ13C-DIC) pool were also collected from ∼50 cm deep with sample rinsed 40 mL polypropylene syringes during the survey. Samples were filtered through Whatman GFF syringe filters into precombusted 12939
dx.doi.org/10.1021/es4027776 | Environ. Sci. Technol. 2013, 47, 12938−12945
Environmental Science & Technology
Article
Figure 1. Survey results for A. Dissolved CH4 concentrations; B. δ13C-CH4 of dissolved CH4; C. Miller-Tans Plots of dissolved CH4 for the four delineated sections; D. pCO2; E. δ13C-CO2 of dissolved CO2; and F. Miller-Tans Plots for CO2. Note Section 4 is the most downstream section.
standard equations.26 Data are presented as μatm, with the exception of the chamber experiment where data are presented as μmol CO2(aq). The same procedure was used for CH4, however data are presented in nmol using Bunsen solubility coefficient equations.27 A comparison of the δ13C-CO2 from the CRDS and the 13 δ C-DIC values measured by IRMS was made by calculating the theoretical δ13C-DIC from the δ13C-CO2 measured with the CRDS using temperature and carbonate fraction dependent εDIC‑g fractionation factors28 (see SI). Unfortunately we have no
alternative method for measuring CH4 concentrations or isotopes to compare that data against, however CH4 standards showed negligible drift in both concentration and isotopes over the time scales used for these experiments. To estimate the isotope value of the CH4 and CO2 source during each experiment we used the linear model developed for atmospheric source characterization in the form of δobsCobs = δsCobs − C bg(δ bg −δs) 12940
dx.doi.org/10.1021/es4027776 | Environ. Sci. Technol. 2013, 47, 12938−12945
Environmental Science & Technology
Article
where C and δ are concentration and δ13C, respectively, and the subscripts obs, bg and s refer to the observed, background and source values.29 By plotting δobsCobs (y) against Cobs (x) (MillerTans plots) the slope of the line is equal to δs. This method was preferred over traditional keeling plots as it is less sensitive to changes in background concentration and δ13C.29 Here we use the ordinary least-squares regression method for estimating the slope.
Time Series in a Mangrove Tidal Creek. The sensitivity of the CRDS technique to tidal variation in the concentration and δ13C of CH4 and CO2 is shown in Figure 2. This type of high resolution time series measurement provides a remarkable insight into carbon cycling that has previously been unavailable. While other studies have also described similar trends, for example, refs 30 and 31, the CRDS technique provides a massive improvement in measurement resolution compared to previous methods (i.e., minute versus hourly resolution). CO2 partial pressure ranged from near equilibrium values around high tide (i.e., ∼400 μatm) to ∼5000 μatm during the second low tide (Figure 2B). δ13C-CO2 values displayed an inverse trend to concentration with the lowest values occurring during low tide (down to ∼−16‰), and highest values at high tide
■
RESULTS/DISCUSSION Survey along an Estuarine Gradient. Coastal waters are often highly dynamic and high resolution temporal observations are often needed to resolve patterns and fluxes of greenhouse gases to the atmosphere.22,30,31 CH4 concentration and δ13C values ranged from 2 to 74 nmol and −61.07 to −48.62‰, respectively. There was a distinct trend of decreasing concentration and increasing δ13C in the downstream direction (i.e., from section 1 to section 4; Figure 1A and B). The Miller-Tans plots show isotopically identical sources of methane in the upper two sections (sections 1 and 2; Figure 1C) with slope values of −63.84 ‰. Generally there is a shift in biogenic CH4 formation pathways from acetate fermentation (leading to a δ13C range of −65 to −50 ‰) in freshwaters, to CO2 reduction (leading to a δ13C range of −110 to −60 ‰) in marine waters.32 If in situ CH4 production was the dominant process we should expect to see differences in the δ13C source values between section 1 (salinity 0−10) and section 2 (salinity 10−30) with decreasing δ13C source values with increasing salinity. However the source value was identical in these two sections. This suggests that a common lateral input of CH4, most likely from groundwater input, rather than in situ production, may be the dominant driver of dissolved CH4 in these areas of the estuary. Previous research in this estuary has found that groundwater inputs drive surface water pCO2.22 During downstream transport this CH4 undergoes oxidation leading to the isotopically heavier pool observed in sections 3 (δ13C source −60.68‰) and 4 (δ13C source −50.42‰). In addition there was a flux to the atmosphere due to the higher than equilibrium concentrations (up to ∼3500% saturation), leading to a kinetic fractionation (13C enrichment) of the residual CH4 pool (although this effect is small 33). It is unlikely that the δ13C source values for CH4 in sections 3 and 4 reflect significant inputs from CO2 reduction because low, near equilibrium concentrations were observed throughout the lower estuary (i.e., concentrations were
